The lectures series of the 33rd Professor Harry Messel International Science School 3-16 July 2005

The Science Foundation for Physics within The University of Sydney

“In the Pursuit of Excellence” EDITORS Dr Chris Stewart Executive Officer, The Science Foundation for Physics, The University of Sydney, Australia

Associate Professor Robert Hewitt Director The Science Foundation for Physics The University of Sydney, Australia

Editorial assistance from Alison Thorn and Alex Viglienzone

A course of lectures given at the 33nd Professor Harry Messel International Science School for High School Students, organised by The Science Foundation for Physics within The University of Sydney, at the University of Sydney

3-16 July 2005

The Science Foundation for Physics The University of Sydney NSW 2006 Australia http://www.physics.usyd.edu.au/foundation

© Copyright Science Foundation for Physics June 2005

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or otherwise, without permission in writing from the Science Foundation for Physics, The University of Sydney.

Designed and produced by Peter Thorn Design, Sydney

Waves of the Future ISBN: 1 86487 725 1

The Science Foundation for Physics gratefully acknowledges the Telstra Foundation's Community Development Fund for their generous support in the production of this book. Radio telemetry in the study of wildlife 09 Dr Frank Seebacher Catch, Move and Twist with Optical Tweezers: Biophotonics at work 23 Professor Halina Rubinsztein-Dunlop The Treatment of Cancer Using Ionising Radiation 39 Dr Clive Baldock The psychophysics of real and virtual auditory spaces 51 ContentsAssociate Professor Simon Carlile Telecommunications: the here and now67 Professor Martijn de Sterke Telecommunications: looking to the future 79 Professor Martijn de Sterke The Science of the Aerosols we breathe 91 Professor Lidia Morawska Creating and overcoming invisibility: scrutably personalised ubiquitous computing 107 Associate Professor Judy Kay Seeing in the Nanoworld 119 Professor David Cockayne Building in the Nanoworld 133 Professor David Cockayne Understanding Brain Dynamics 147 Professor Peter Robinson Wind, Waves and Beaches 157 Professor Andrew D Short The ever changing life of galaxies 173 Dr Raffaella Morganti Monsters lurking in the centre of galaxies 191 Dr Raffaella Morganti Quantum Mechanics: The Wild Heart of the Universe 207 Dr Joseph Hope Einstein and the Quantum Spooks 221 Professor Huw Price The Messel Endowment 235 3 n 1999 the Science Foundation for Physics within the University of Sydney established the Messel Endowment to honour Professor Harry Messel and to fund the International Science School I in perpetuity. The Messel Endowment is managed to preserve the real value of all its donations. The surplus interest is used to support the Professor Harry Messel International Science School (ISS). If income exceeds the requirements of the ISS, the Foundation will use the funds to support other initiatives within the School of Physics. Such activities will be named to honour Professor Harry Messel.

The Science Foundation for Physics sincerely thanks all supporters of the Messel Endowment. All contributions to the Endowment are important to its success and the Foundation acknowledges the following for their generosity. For more information on the Messel Endowment please contact The Science Foundation for Physics on +61 2 9351 3622, email [email protected] or visit www.physics.usyd.edu.au/foundation/

MESSEL ENDOWMENT DONORS

Extra-Galactic Donors: A$1 000 000 and over Emeritus Professor Richard Collins The Department of Education, Science and Training and Mrs Marilyn Collins Mulpha Australia Limited Emeritus Professor Maxwell G Brennan, AO, and Mrs Ionie M Brennan Galactic Donors: A$100 000 to A$999 999 Associate Professor Robert G Hewitt The Science Foundation for Physics and Mrs Helen Hewitt Nell & Hermon Slade Trust Hermon Slade Foundation Planetary Donors: A$1 000 to A$9 999 Mrr Reginald J Lamble, AO Stellar Donors: A$10 000 to A$99 999 Mr Anthony M Johnston Mr Terrey P.Arcus IBM Australia Limited Mr Michael Messel Mr Trevor E Danos James N. Kirby Foundation Mr David B Herrman OneSteel Limited Dr Jenny A Nicholls Cochlear Limited Mr Basil Sellers AM through Sellers Pty Ltd Australian Business Limited Dr Emery Severin & Mrs Sharman Severin Cecil & Ida Green Foundation Dr Brian J O’Brien Macquarie Charitable Foundation Limited Mr Nicholas Manettas through Nick’s Seafood Mr Robert Arnott Bar & Grill Mr John A L Hooke, CBE Mr Raymond Walton & Mrs Margit Walton USA Foundation Ms Valma G Steward Mr Jim O’Connor Mrs Kathy Manettas 4 Westpac Banking Corporation Mrs Georgina Donaldson Lahili Pty Limited Dr Jennifer J Turner Emeritus Professor John Davis Mr Tim M Smyth Mr Steven K Eckowitz Dr David R V Wood Ms Danielle M Landy Mrs June Papadopoulos Mr Thomas M F Yim on behalf of Alex Yim Mr Frank Papadopoulos Dr Bruce McAdam & Mrs Janice McAdam Toni R Kesby Pty Ltd Dr Joseph A Beunen Professor David R Fraser Mr John A Vipond Ms Jennifer H F Wanless Professor Lawrence E Cram Mrs R Lambert Mr Peter Manettas Mrs Mary Moore Mr Graham H Hall Mr Spiro J Pandelakis Mr Christopher C Vonwiller Mr Robert R B Murphy Kenneth Coles Foundation Mr Peter C Perry Dr Stephen D Segal Mr Ian G Dennis Dr Robin B Fitzsimons Mr Ian A Dyson Dr Robert L Every Mr Harry J Pemble Dr David Malin Mr Geoffrey D Pople Mr David C Davidson Mr Alan K Milston, OAM through David C Davidson Pty Limited Dr Xian Zhou Southcorp Limited Dr Kevin C Allman ASA ITF Foundation for the Advancement of Astronomy Dr David Z Robinson Dr Christopher J E Phillips Asteroidal Donors: up to $999 Associate Professor Donald D Millar Mrs Iona S Dougherty Dr David G Blair Professor Roger V Short Ms Tomoko Kikuchi Ms Yvonne Pitsikas Mr Steven Kambouris Mrs Irene P Gibson Mr Jeff Close Mr Wen W Ma Fr Mervyn J F Ziesing Mr John H Valder Dr Hugh S Murdoch Mr Enrico Piccioli Dr Claire E Cupitt Mr Arun Abey Ms Anne Woods Dr P E Rolan Ms Elana Bont Barker College Mr George Papadopoulos Dr Robert H Masterman The Australian Association of Phi Beta Kappa Dr George F Brand Sir Walter Bodmer Ms Margaret A Desgrand Ms Julie K Ellinas Ms Belinda H Allen Mrs Chrissie Athis Mr Julian J Dryden Mr John W L Rawson Mr John Paterson Mr George Athis Mr Gavin M Thomson Mr Arthur J Buchan Mr Allan F Rainbird Mrs Helen Bell Mr Thomas M F Yim on behalf of Jerome Yim Emeritus Professor Louis C Birch 5 Supporters The Science Foundation for Physics warmly thanks the supporters of the 2005 ISS: Waves of the Future

The Messel Endowment IBM Asia Pacific NSW Department of Education and Training College of Sciences and Technology, The University of Sydney Faculty of Science, The University of Sydney Scientific Services Pty Ltd Telstra Foundation Community Development Fund Ralph’s Café The Kirby Foundation

Australian students were selected with the support of the Science Teachers Associations in Victoria, Tasmania, South Australia, the Northern Territory, the ACT and Western Australia, and the NSW Department of Education and Training.

The following institutions assisted in the selection and travel of the overseas students:

The Affiliated Middle School of Beijing University, China MONBUSHO, Japan Ministry of Education Malaysia The Royal Society of New Zealand Ministry of Education, Singapore Ministry of Education, Thailand The Association for Science Education, UK The Royal Institution of Great Britain NESTA (The National Endowment for Science, Technology and the Arts) Department of Energy, USA

Webcasting is made possible with generous gifts from:

Emeritus Professor Harry Messel IBM Asia Pacific NSW Department of Education and Training Associate Professor Bob Hewitt Dr Jenny Nicholls

6 Preface

HE SCIENCE Foundation for Physics within the University of Sydney is delighted to present the 33rd Professor Harry Messel International Science School for High School Students, named after the man T who had the vision to initiate them in 1958. The United Nations has designated 2005 as the International Year of Physics since it is the centenary of the year in which Albert Einstein published his groundbreaking papers in three areas of physics which are the foundation for modern science. Therefore the 2005 Science School, titled “Waves of the Future”, will give scholars the opportunity to encounter and experience research in some of the wide range of scientific fields that have developed rapidly as a result of Einstein’s contributions.

The primary aim of all of the Science Schools is to acknowledge the excellence of the scholars, who have been selected on the basis of their academic abilities. The presence of gifted young people from many countries will allow the scholars to experience the values of many cultures and to learn new ways of doing things. The Science Foundation stands for the Pursuit of Excellence, and is always pleased to have an opportunity to acknowledge this spirit in young people.

The International Science School can only be held because of the generous financial contributions of the Supporters and the Donors to the Messel Endowment, and because of the time and energy given by the Lecturers. Like the Science Foundation itself, the Supporters, Donors and Lecturers are committed to promoting science education at the very highest level of excellence. On behalf of the Foundation, I express our grateful thanks to all three classes of benefactors.

Robert G Hewitt Sydney, June 2005

7 What got you interested in science in the first place? I was a little field biologist as a kid, partly for the adventure of exploring and partly because of my interest in the natural environment. I DR FRANK SEEBACHER grew up with a great interest in School of Biological animals and plants, and as a kid Sciences, University of spent most of my free time roaming Sydney. After completing around collecting lizards, frogs, and a Ph.D. at the University insects, breeding tadpoles and of Queensland on the planting a garden. This interest turned thermal biology of the into inquiry to find out how animals Australian freshwater and natural systems work, and I crocodile, Dr Seebacher decided to study science at university. worked as a post- doctoral researcher at What’s the best thing about being James Cook University a researcher in your field? and The University of Biology is a diverse field and the Queensland. He has research can be really varied. For been at the University of Sydney example, my research covers since 2001. His research is everything from gene expression, to concerned with the response of fieldwork studying animal behaviour. animals to changing environments, It is this multidisciplinary approach with a resolution spanning from gene that reveals how natural systems expression to behaviour in the wild. work, and that makes being a Frank is a member of the IUCN Biologist very compelling. Crocodile Specialist Group. Who inspires you – either in science or in other areas of your life? I get great inspiration from books, and one of my favourites is the 16th century thinker Michel de Montaigne.

If you could go back and specialise in a different field, what would it be and why? I always wanted to be a mathematician because of the logical structure of mathematics, and its ubiquitous importance in explaining the world.

What’s the ‘next big thing’ in science, in your opinion? What’s coming up in the next decade or so? The ongoing advances in molecular biology will change the way humans interpret their environment, and molecular biology will also impact on 8 our personal lives by its secondary technological applications. Radio telemetry in the study of wildlife

Dr Frank Seebacher

Introduction THE WORLD IS becoming an increasingly dire place for wildlife. Increases in human population around the globe and the increasing demand for natural resources means that there is less and less room for animals and plants to live in their natural environment. Australia has a unique assemblage of wild animals and plants that have evolved in isolation from other continents for many millions of years. Like most countries in the world, there is an urgent need for Australia to manage its wildlife to ensure that the biodiversity of the continent is preserved into the future. A prerequisite for effective wildlife management is an understanding of how animals work in their natural environment. Stunning technological advances in physics over the past 30 years have made radio telemetry an essential tool for wildlife research allowing the monitoring of undisturbed animals in the wild.

9 The cane toad Bufo marinus n this chapter I will give a broad outline of how The rate at which chemical reactions proceed living organisms interact with their environment, from a set of reagents to a product is dependent I and of the importance of environmental change on many physical factors such as pH, pressure, on the functioning of organisms. This background and particularly temperature. Temperature is a will help us understand why it is important to measure of heat energy, and the Laws of study animals in their natural environment. I will Thermodynamics state that the more energy is then introduce radio telemetry and its uses in contained within a chemical system, the more wildlife research. The second half of the chapter readily the reactions within that system will occur. will review three case studies in which telemetry Considering that living organisms may be defined has been used to address important wildlife as a complex system of biochemical reactions, it management and scientific questions. becomes obvious that the functioning, or fitness, of living beings is dependent on temperature and on other factors of the environment. The Earth is not a Stable Place: Evolution is Driven by Environmental Most of the daily life of animals is preoccupied Change with interactions with the environment. Think Living systems are made up of thousands of about your own life: many of your decisions biochemical reactions that must work together for every day depend on the climate where you live. the whole organisms to function. Even if a single For example, your choice of what to wear will one of those reactions is disrupted, the depend on the weather outside. Mammals are of functioning of the whole organism may be course a special case because we must maintain compromised. There are many examples from our internal temperature more or less constant, medicine where a single enzyme is lacking so and that is achieved by producing metabolic that a reaction cannot proceed to produce the heat. In other words, we breakdown food – some required product. In humans, such disruptions are of this is used as building blocks and nutrients, often treated as medical conditions, but variations but most of it is dissipated as heat for us to in biochemical function are found in all organisms regulate our body temperature. As mentioned and they are not necessarily bad – in fact, above, mammals, and birds as well, are a special variation is essential for organisms to be able to case: most animals do not regulate their internal respond to changes in their environment. temperature by producing metabolic heat, but choose suitable environments in which their In nature, the ultimate goal of animals is to bodies will adopt a temperature that is within a reproduce so that a particular set of genes is range which will permit the individual to maintain passed on to the next generation. The capacity of its fitness. Of course, all living organisms will an organism to achieve reproductive success is produce some metabolic heat, but in the case of referred to as its ‘fitness’. Fitness is not only ectotherms – that is, most invertebrates, fish, determined by reproduction in its immediate sense, frogs, lizards and snakes – metabolic rates are but also by the general well-being or performance too low to produce sufficient heat for of an individual in its environment. For example, if thermoregulation. However, regardless of an individual lizard is a very slow runner, it may not whether an animal is ectothermic or endothermic be able to escape from a bird that intends to eat it (those that produce metabolic heat for as well as a slightly faster lizard may. If the former thermoregulation) the environment, and lizard gets eaten it cannot reproduce any more so particularly climate, will dictate much of its that its fitness is compromised compared to the behaviour and physiology. latter lizard because of its slow running speed. To refer back to the example above, a disruption in a The climate on the Earth is in constant flux. biochemical pathway may similarly decrease the Climate fluctuations occur at different temporal fitness of an animal so that, in the worst case, it scales from millions of years to days, and at all may not be able to reproduce and its heritable scales these are important for the fitness of WAVES OF THE FUTURE OF WAVES material will disappear from the gene pool of that animals. For example, climate during the 10 species or population. Mesozoic, the time when dinosaurs dominated terrestrial ecosystems from 220 to 65 million Before it is possible to assess the molecular years ago, was much warmer than today. mechanisms that underlie the evolutionary response to changing environments, we must know the ecology of species and how they interact with their environment. Good ecological knowledge is not only at the basis of [FIGURE 1 HERE] a scientific understanding, but it is also crucial for management. The natural fluctuations in climate and the environment are exacerbated by human induced changes. Human activity may change environments rapidly and severely, so that the pressures on organisms may be far greater than those originating form naturally occurring fluctuations alone. For example, weekly activity of humans perceptibly changes the climate in the USA so that there are short- Figure 1: Mean global climate has fluctuated considerably during the earth’s history. term weekday and weekend cycles. Human induced climate change in combination with pollution and habitat destruction means that The Cretaceous period (120 to 65 million years there are immense pressures on wildlife. ago) in particular was characterised by very warm Sophisticated and effective wildlife conditions, and even at very high arctic latitudes management will be the challenge for this the climate was more similar to present day century, and at its core lies a good Melbourne than to the permafrost of today’s understanding of the movement, habitat use, arctic regions. Similarly, the Antarctic continent, at thermal requirements, and physiological present covered by permanent ice up to 3 km responses of focal species under natural thick, was once very equable with extensive conditions. vegetation cover and diverse terrestrial wildlife.

Given the instability of the earth’s climate, and Telemetry the dependence of organisms’ fitness on The challenge for wildlife scientists and environmental parameters, it becomes clear that managers is to collect data from animals that the environment represents a major selection are undisturbed in their natural environment. pressure for organismal evolution. A common Radio telemetry is a technique that allows perception is that animals have evolved to do monitoring of animals from a distance, thereby best in the environment they currently live in. By permitting data collection from animals that are and large this is true, but how do we know what unaware of the researcher. Telemetry the response of organisms will be as the climate encompasses a transmitter that sends signals changes at sudden short time scales, and also to a remote receiver. The most basic gradually over thousands and millions of years? application of telemetry is monitoring the It is crucial for our understanding of ‘life’ to know locality of a transmitter. Transmitters emit the mechanisms that are at work to help animals regular signals, and they broadcast at a known maintain their fitness in changing environments. frequency, so that a transmitter attached to an These mechanisms are ultimately molecular, animal will not only tell where the animal is, based on the expression of genes within but it will also be uniquely identified by the individuals, and the recombination of genomes transmitter frequency. The signal is received by between generations. These genetic changes an investigator with a specialised telemetry may manifest themselves at a physiological receiver up to several kilometers away, and the level, for example in the functioning of muscles use of a unidirectional antenna (one that RADIO TELEMETRY IN THE STUDY OF WILDLIFE or the cardiovascular system, or at a receives the signal over a narrow range of behavioural level. incident angles) will allow the tracking of the 11 transmitter. In recent years, biological applications In the early 1970s when the telemetry work was of telemetry have become increasingly started, crocodiles were nearly extinct in Australia sophisticated, and it is possible now to track as a result of hunting for their skins. Australia has animals with satellites, and to measure body two species of crocodile: the endemic freshwater temperature, heart rate and blood flow. A crocodile, Crocodylus johnstoni, and the estuarine expansion of radio telemetry are data loggers that crocodile, Crocodylus porosus. Freshwater are deployed on animals to measure similar crocodiles are the smaller of the two species and variables - but instead of emitting a signal, live primarily in inland billabongs and rivers, but dataloggers store the information and must be they may also enter saltwater. retrieved so that the information can be downloaded onto a computer. Crocodylus porosus is the largest living crocodile species with reputable records of maximum Wildlife telemetry was first developed in the length of seven metres! Populations of both 1950s in studies on birds. The particular species, and particularly of C. johnstoni,have Australian perspective, and one that is certainly recovered very well and they are again found in relevant for this Science School, is the work of good number in the Northern Territory, Western Professor Harry Messel starting in the 1960s. In Australia and Queensland. Increasing populations 1968 Harry was commissioned by one of the of estuarine crocodiles or ‘salties’ also increases leading space electronics industries to design a adverse encounters between people and small, long-range, and long-lasting transmitter crocodiles, particulalry around populated areas that could be used under extreme environmental such as Cairns in North Queensland. Effective conditions. The purpose of this transmitter was management of populations is therefore essential not so much wildlife tracking as tracking humans, for the wellbeing of humans and to conserve in rescue operations for example. Nonetheless, crocodile populations into the future. Research wildlife provided a good test system. The first and management of crocodiles continues at a prototype was tested on polar bears in the Arctic much more professional level than in the hunting – and failed. After several years, a second days before 1970, and two examples of current attempt was made, this time using crocodiles in work on crocodiles are given below. Northern Australia as the guinea pigs. This started a major research program on the development of Biotelemetry, encompassing both radio-telemetry wildlife tracking devices, and also on crocodile as well as data loggers (also known as archival management and conservation. telemetry), is of course not restricted to crocodiles

Figure 2: Freshwater crocodile, Crocodylus johnstoni. WAVES OF THE FUTURE OF WAVES

12 but it is widely used in wildlife research, as well as return repeatedly to the same shelter, breeding or in laboratory monitoring. The examples below are feeding sites. It may be advantageous for toads to taken from the recent scientific literature and will return to familiar sites, which they know to provide hopefully provide a good introduction to the food shelter or mates, rather than incurring the potential uses and importance of telemetry in unknown and potentially more expensive cost of science and management. locating alternative sites. For management purposes, the difference between random long- distance movements and restricted site-specific Telemetry in Wildlife Studies activity is crucial. (a) Tracking cane toads Cane toads (Bufo marinus) are a large terrestrial Regardless of their pattern of movement, terrestrial amphibian, originally from South and Central amphibians are always faced with the threat of America. They were introduced into many Pacific dehydration. This is particularly true for toads areas as a control agent for sugar cane pests. because, unlike many native species of frogs (there Cane toads were introduced into Australia in the are no native Australian toads) that secrete a waxy 1930s as a ‘natural’ pest control coating on their skin to reduce measure for the cane beetle that water loss, toads have no skin infested the extensive sugar “... cane toads have just resistance to evaporative water cane crops in North Queensland. loss – in other words, water can In the event, the toads ate arrived at the World evaporate through their skin the anything but the cane beetle, Heritage listed Kakadu same as from an open bowl of and they have spread throughout water or a puddle. Queensland, into New South National Park in the Wales, and cane toads have just Northern Territory...” Radio-telemetry is a perfect tool arrived at the World Heritage for collecting data that will listed Kakadu National Park in resolve whether toads move the Northern Territory. As a randomly or whether they result, there is a strong desire in the general and display homing behaviour within a restricted scientific communities to control populations of B. activity range. In combination with environmental marinus in Australia. measurements, such as soil moisture or air temperature, tracking data will also reveal the What makes matters much worse is that cane best conditions for movement. In a study I toads are potentially lethal for native predators conducted on Orpheus Island with Ross Alford owing to their venom glands situated just behind from James Cook University in Townsville, we the head. Although the direct impact of cane toads on native species is not well documented, there are considerable management efforts aimed at controlling toad infestations. As a first step for any effective management, it is essential to understand the ecology of toads, particularly their movements and the environmental conditions that favour their dispersal.

It is not clear whether toads in northern Australia are nomadic, with restricted movements during dry periods but long and unpredictable migrations during wet seasons. In their native habitat in South America, toads move only short distances each night and remain within limited activity areas even RADIO TELEMETRY IN THE STUDY OF WILDLIFE during the wet season. A number of toad species Figure 3: Orpheus Island in North Queensland where the show quite an impressive ability to navigate and study on movement of cane toads (insert) was conducted. 13 surgically implanted radio transmitters into cane toads to determine whether the toads moved randomly in their environment, and whether their movements were limited by environmental factors.

We found that movement of B. marinus depended on the moisture level of the ground. Toads never moved when the ground was dry (less than 0.1 ml of water per gram of dry soil) which, of course, is not surprising given the high water loss rates of terrestrial amphibians. Interestingly, rate of movement increased as the soil became Figure 5: Patterns of movement by cane toads. Measured movement data obtained with radio telemetry (red line) wetter, but there was also a maximum (i.e. an were not different from movement patterns generated by a asymptote) that probably indicates the physical random walk model (blue line). limit to how far toads could move per night (Figure 4). daytime shelter site after moving at night, they did not return more often than would be expected from completely random movement.

(b) Satellite tracking of crocodiles Tracking toads was a relatively easy thing to do. Toads are abundant and the animals are easily caught by hand, and their small size means that the distances traveled are not too great for tracking on foot. The situation becomes more difficult when dealing with large crocodiles: the animals are potentially dangerous to deal with and catching even a two to three metre long crocodile requires considerable logistic effort to Figure 4: Movement of cane toads. Toads did not move at set traps and handle the animals. To make all when the soil was dry, and there was a maximum rate matters worse, crocodiles travel in water, and of movement. they travel long distances in a relatively short time period. To resolve the question of whether toads moved randomly, as opposed to staying within a well- One of the shortcomings of radio telemetry is that defined home range, we compared the movement the signal emitted by the transmitter is attenuated data collected by radio tracking animals in the by freshwater reducing the range over which field to a random walk. A random walk in physics signals may be received; for example, signals is a process that is defined as a sequence of from the transmitters inside the cane toads could discrete steps in random directions. Random be received from 1-2 km away, but if animals walks vary greatly depending on the dimension of were in freshwater, this range would be reduced the steps. Using distances and angles randomly to about 300-500 m. Radio signals do not travel determined within the range of measured values at all in salt water so that when radiotracking from toads, random movement patterns can be marine animals, the investigator must rely on the constructed, and these can be compared to animal being at the water surface with the actual movement over a given period of time antenna protruding. An alternative to radio (Figure 5). transmitters are sonar transmitters. Sonar works

WAVES OF THE FUTURE OF WAVES by sound waves that do travel in water, and As it turned out, the toads moved randomly, and which are emitted by the transmitter. In fact, 14 although they did frequently return to the same some marine animals, such as whales and dolphins, use sonar to communicate and to locate operational satellites, signals from each predators and prey. Sonar tracking is usually transmitter on earth will be received for about ten more involved than radiotracking on land, minutes on each of 20 to 30 occasions per day. because several listening devices, or The satellites send data to ground stations that hydrophones, must be employed to locate the forward them by e-mail to the researcher. transmitter under water, and it is not really useful for animals that move long distances. To date, Mark has deployed satellite transmitters on 15 animals at different locations on Cape York Clearly, tracking estuarine crocodiles is quite a Peninsula in North Queensland, Australia. The challenge. Nonetheless, obtaining movement data transmitters are sutured to an area of protruding from C. porosus is absolutely essential for bony scutes on the crocodile’s neck called the management, and it is of great scientific interest nuchal shield. with respect to species radiation and interactions between populations of crocodiles living at different parts of the world. From a management perspective, crocodiles that could potentially become a threat to humans are often removed to more remote areas. It is not unusual for those animals to return to their original residency. These behaviour patterns raise the question of how far and how often crocodiles move naturally, and how much site fidelity different sized animals and animals of different sexes display.

Knowing movement patterns also becomes important in conservation of endangered crocodiles. Although C. porosus appear to be safe Figure 6: Estuarine crocodile, Crocodylus porosus, with a in Australia, they are endangered or threatened satellite transmitter attached to its nuchal shield. over most of their natural range, which includes India and Southeast Asia, the Pacific Islands, The data from the satellite trackers provide a Indonesia, Papua New Guinea, and Australia. Is it fascinating snapshot of crocodile life that would possible that animals from Australia supplement not have been possible without the development depleted populations in New Guinea and of this technology. The first example of a tracking Indonesia? record (Figure 7) comes from a small (2.5 metre) male crocodile that was released 50 to 60 Mark Read from the Queensland Environmental kilometres further south from its capture site. Protection Agency in Cairns recently initiated a program of tracking C. porosus via satellite. Soon after release, within 2.5 weeks, the animal Satellite tracking works similarly to ‘ordinary’ moved back north to its original capture site, with radiotracking, and most wildlife tracking uses the the occasional foray back south. Clearly, Argos Satellite System. Several satellites that movement of 50 kilometres or so is no problem carry the necessary reception equipment orbit the even for relatively small crocodiles, and the earth at any time. The satellites receive the October records (blue triangles) in particular show ‘message’ from the transmitter on the animal and that the animal moved between initial capture and relay the data to the ground in real time or as release sites within the same month. Nonetheless, stored information. At any point in time, a most of the time, the crocodile moved within the particular satellite can receive signals from fairly restricted area of its original river system. transmitters on earth over a ground area of 5000 Crocodiles are territorial animals and although km2, and owing to the earth’s rotation the they will travel about occasionally, they remain in RADIO TELEMETRY IN THE STUDY OF WILDLIFE reception area moves in swathes around the their territories for many years and will defend globe. This means that if there are two them against other, intruding animals, particularly 15 during the breeding season between October/ November and February. Most crocodile attacks on humans are not stimulated by the crocodile’s intention to feed but by territorial defense.

The second example (Figure 8) stems from a 3.9 metre animal caught in the Nesbit River on Cape York Peninsula. The map is on a finer scale than in the previous example, and the tracking data clearly emphasise that adult crocodiles stay within well-defined territories most of the time, but that they do travel occasionally.

The crocodile traveled quite a long way upstream into the freshwater reaches of the river on one occasion. Although they are named estuarine or saltwater crocodiles, C. porosus often enter freshwater, and they build their nests only in Figure 7: Satellite tracking data (example 1) from a small adult estuarine crocodile that was released 60 km from its freshwater. The other striking feature about this capture site. tracking record is the animal’s movement out to sea. Saltwater crocodiles – this time as their name suggests – routinely swim out into the ocean and are well able to navigate their way back to their home river. In fact, crocodiles are sighted quite frequently on tourist beaches in Figure 8: Satellite tracking data of a 3.9 metre saltwater Cairns and Darwin. crocodile (example 2) on Cape York Peninsula, Queensland. WAVES OF THE FUTURE OF WAVES

16 On a different note, a recent report by the BBC In addition to body temperature, another (September 2004) in the United Kingdom important physiological parameter is heart rate. describes quite a different sort of tracking using Heart rate is routinely monitored by veterinarians satellites. In three pilot schemes initiated late in to assess the wellbeing of their patients. It 2004, satellite tracking technology is used to provides insight not only into their cardiovascular track convicted burglars, robbers and car thieves. physiology, but also into an animal’s metabolic For the first time in Europe, offenders will be status. Each time the heart beats, a very weak fitted with transmitters as part of a community electric current is generated. Specialised heart sentence, or as a condition of their release from rate transmitters can detect that current, and the prison, so that they can be tracked by satellites electric pulse generated by the heart triggers the 24 hours a day. Transmitters are monitored by a transmitter to emit a signal. Hence, heart rate can control station that records the location of the be detected from a distance by the signals person to within a few metres, and if the received from a heart rate transmitter. I have offender strays into a prohibited area, the police used this technology frequently to study the are alerted. cardiovascular response to different thermal environments by free-ranging reptiles. In a recent (c) Physiological monitoring study at Lakefield National Park on Cape York In addition to tracking the movement of animals, Peninsula with colleagues from the University of technology has been developed to measure Queensland and the Queensland Environmental physiological responses of animals by radio Protection Agency, we were interested in telemetry. A common application is measuring measuring physiological responses – body body temperatures of animals behaving naturally in temperature regulation and heart rate – in the wild. In temperature telemetry, a temperature- relation to diving in wild freshwater crocodiles. sensing device, or thermistor, is incorporated into the transmitter. The thermistor causes the Many lineages of terrestrial vertebrates, for transmitter to emit signals at a rate that depends example crocodiles, seals, and otters, have on the temperature of the transmitter. By secondarily recolonised aquatic environments (that calibrating the interval between two signals against is, they have evolved from ancestors that lived on temperature before deploying the transmitter, it is land, and they have since adapted back to the possible to obtain the body temperature of an water again). They have presumably gained animal after surgically implanting a transmitter into selective advantages by adopting a semi-aquatic the animal’s body cavity (Figure 9). lifestyle. The pronounced physical differences between air and water impose different challenges on anatomical and physiological characteristics of semi-aquatic animals. For example, many semi- aquatic vertebrates have independently evolved anatomical features that assist movement in water (webbed feet, fin-like appendages, dorso-ventrally flattened tails, etc), although behavioural patterns, such as foraging behaviour and avoidance of predation often encompass movement between water and land. Behavioural responses are supported by physiological functions, particularly metabolic and cardiovascular, that must respond to the unique characteristics of both aquatic and terrestrial environments. Figure 9: Heart rate transmitters were attached to crocodiles in the field (A), and body temperature Crocodiles and alligators evolved from ancestral transmitters were surgically implanted into the animals (B). terrestrial animals in the Late Triassic and

The interval between two signals from a temperature RADIO TELEMETRY IN THE STUDY OF WILDLIFE transmitter depends on the temperature of the transmitter secondarily became aquatic. Although modern – a typical relationship is shown in (C). crocodiles are proficient in terrestrial locomotion, 17 much of their ecology is geared towards an water depth the animal has been at, and aquatic lifestyle. Crocodiles possess a number of therefore obtain a record of natural diving cardiovascular characteristics that make them behaviour (Figure 10). well disposed for aquatic behaviour and diving. In laboratory trials, crocodiles can slow their heart rate dramatically to only 4 or 5 beats per minute in response to submergence in water. This slowing of heart rate, or bradycardia, is thought to be an adaptation to diving because it results in decreased blood flow to the tissues and, therefore, less use of oxygen. The problem with laboratory experiments is that the animals are often stressed, and stress may elicit similar response of the cardiovascular system. This is where telemetry provides the solution, enabling measurements of heart rate in undisturbed crocodiles. Figure 10: A typical example of a daily dive profile from a freshwater crocodile. Additionally, ectotherms such as crocodiles must also reconcile essential terrestrial thermoregulatory behaviour such as basking in Interestingly, we found that animals were most the sun on land with reproduction, social active when their body temperature was low and interactions, and feeding that are almost entirely before they basked in the sun. It was always restricted to water. In many reptiles believed that reptiles and other ectotherms must thermoregulation is closely tied to the warm up in the sun before they can become cardiovascular system. Numerous laboratory active. Our data clearly showed that this is not studies have shown that heart rates during the case (Figure 11). heating are significantly higher than during cooling. Animals gain an advantage from this heart rate pattern (known as heart rate hysteresis) by being able to control rates of heating and cooling, thereby increasing the time spent within a preferred thermal range during the day. Given the reliance of crocodilians on water for thermoregulation, we were particularly interested in determining the interaction between diving ecology and physiological thermoregulation.

A further challenge, beyond successfully deploying body temperature and heart rate transmitters, was to measure diving behaviour of Figure 11: Diving activity of freshwater crocodiles in wild crocodiles. To achieve this, we used a relation to body temperature. Activity, expressed as number of vertical movements per hour (blue area), was greatest recently developed device called a time-depth before the animals warmed up in the late morning. recorder. These are archival devices; that is, they store data in their internal memory for later downloading to a computer, rather than sending Crocodiles perform most behaviours, such as signals to a receiver. The recorders sense feeding and mating, under water and they mainly changes in pressure with time, which can be do so in the early morning. Thermal sensitivity of WAVES OF THE FUTURE OF WAVES related to water depth. By attaching a time-depth locomotor performance originates primarily from 18 recorder to an animal it is possible to infer the the temperature-induced constraints on muscle performance and metabolic potential. It may be Waves of the Future that metabolic and muscular demands of crocodiles during their normal behaviour never New physiological applications of radio telemetry reach their full potential and that the reduced are constantly being developed. A radio potential at the lower body temperature does not transmitter that measures blood flow is being pose a physiological constraint on activity. tested at the moment. Applications of such technology are not restricted to wildlife studies. Additionally, metabolic processes may be Monitoring of cardiovascular parameters and relatively temperature insensitive so that activity body temperature are routine procedures in may not be significantly curtailed as a result of medical and biological research laboratories, and lower body temperature in the morning. Warming increasing numbers of telemetry systems are up later in the day, while not necessary for being developed to cater for that demand. activity, may be important for other physiological functions such as digestion of food, and recovery The multidisciplinary approach of applying from exercise. innovative research on the physics of signal transmission and environmental sensing to the Regardless of the asynchrony in the timing of study of natural systems has clearly led to major activity and peak body temperatures, the animals advances in biological research in recent years. did regulate their body temperature by emerging The continuing development of new telemetry to bask, and by displaying the typical devices will ensure the progress of wildlife cardiovascular changes in response to heat. research and management into the future. Faster heart rates during heating than during cooling significantly increases the efficacy of References and Further Reading thermoregulation by conferring control over Seebacher, F. and Alford, R. A., 1999, Movement and heating and cooling rates. Interestingly, the microhabitat use of a terrestrial amphibian (Bufo marinus) decrease of heart rates during diving would on a tropical island: seasonal variation and environmental augment the temperature induced decrease in correlates, Journal of Herpetology, 33, pp 208-214. heart rate in animals that dive after basking, although the two mechanisms operate Seebacher, F., Franklin, C. E. and Read, M., 2005, Diving independently from each other (Figure 12). behaviour of a reptile (Crocodylus johnstoni) in the wild: interactions with heart rate and body temperature, Physiological and Biochemical Zoology, in press.

Cooke, S. J., Hinch, S. G., Wikelski, M., Andrews, R.D., Kuchel, L. J., Wolcott, T. G. and Butler, P. J., 2004, Biotelemetry: a mechanistic approach to ecology, Trends in Ecology and Evolution 19, pp 334-343.

Grigg, G. C., Seebacher, F. and Franklin, C. E. (eds.), 2001, Crocodilian Biology and Evolution, (Surrey Beatty, Chipping Norton), ISBN 0 949324 89 2.

Figure 12: Heart rate and body temperature in relation to diving behaviour. The red shading indicates times when the crocodile basked in the sun. Both heart rate and body temperature increased during basking. When the crocodile entered the water (blue shading), heart rate decreased immediately, thereby slowing cooling of the body. RADIO TELEMETRY IN THE STUDY OF WILDLIFE

19 History of the ISS

THE PROFESSOR HARRY Messel International Science School has a long and distinguished history. The 140 students attending Waves of the Future are the 33rd group to gather at the University of Sydney for the Science School – in all, almost 4000 have attended a Science School since they began in 1958.

Initially the Schools were annual The ISS2005 has students Between 1960 and 1979 the ISS events, and the first four Schools, attending from nine countries in lectures were shown on television held between 1958 and 1961, were total: Singapore, Malaysia, Thailand, – in fact, many people recall for teachers. In 1962 Professor Japan, China, the USA, the UK, waking up early on Sundays to Harry Messel, the founder of the New Zealand and, of course, every make sure they didn’t miss the ISS, changed the focus to honour state and territory of Australia. telecast! One member of the excellence in senior high school School of Physics here at the students and to encourage them to The Great Lecturers University of Sydney is adamant consider careers in science. One of the features of the that the lectures shown on TV were International Schools is the lecture a key part of her decision to A Truly International Science series. Past ISS lecturers include become an astronomer. School James Watson, who won a Nobel One student from New Zealand Prize for discovering the structure of Today, the ISS is no longer a attended the very first Science DNA, and Jerome Friedman, also a feature of the television schedule School in 1962, and overseas Nobel laureate his for work on – but we have moved on to students have been a feature of the particle physics. Sir Hermann Bondi embrace new technology. In 2003 ISS ever since. In 1967, ten (physicist and astronomer at part of the lecture series was students traveled from the USA to Cambridge University), Margaret broadcast on the internet as a trial attend the School; the following Burbidge (astronomer with the run, and in 2005 the entire series year they were joined by five from Hubble Space Telescope), Carl Sagan will be webcast. Which means the the United Kingdom and five from (famous astronomer and science ISS has once again moved out of Japan. South-East Asia joined the broadcaster) and Lord May the lecture halls and out into ISS in 1985 when students (President of the Royal Society in the people’s homes. attended from Singapore, Malaysia, UK) have all given talks at the ISS. Thailand and the Philippines – sadly, that was the only time the And of course, who could forget the Philippines has participated. China brilliant science demonstrations of has sent five students to every ISS Professor Julius ‘Why is it so?’ since 1999, except for 2003 when Sumner Miller, which were such a the SARS epidemic restricted travel popular feature of the ISS that they in the region and they reluctantly spawned a television show! These withdrew. days, Dr Karl Kruszelnicki – the Foundation’s Julius Sumner Miller

WAVES OF THE FUTURE OF WAVES Fellow – entertains and enthuses the ISS Scholars with his famous 20 Great Moments in Science. Science Schools for High School Teachers Year Teachers Theme 1958 123 Selected Lectures in Modern Physics and the Astronomer’s Universe 1959 123 Lecture notes on an introductory course in modern physics and nuclear power and radioisotopes 1960 123 From Nucleus to Universe 1961 123 Space and the Atom TOTAL 492

International Science Schools For High School Students Year Boys Girls Total Theme 1962 108 45 153 A Journey through Space and the Atom 1963 104 51 155 The universe of Time and Space 1964 106 53 159 Light and Life in the Universe 1965 114 42 156 Time (and Relativity) 1966 104 52 156 Atoms to Andromeda 1967 101 57 158 Apollo and the Universe 1968 109 20 129 Man in Inner and Outer Space 1969 118 21 139 Nuclear Energy Today and Tomorrow 1970 99 33 132 Pioneering in Outer space 1971 87 35 122 Molecules to Man 1972 95 28 123 Brain Mechanisms and the Control of Behaviour 1973 93 29 122 Focus on the Stars 1974 90 33 123 Solar Energy 1975 76 43 119 Our Earth 1977 54 50 104 Australian Animals and their Environment 1979 63 52 115 Energy for Survival 1981 50 65 115 Biological Manipulation of Life 1983 67 51 118 Science Update 1983 1985 71 59 130 The Study of Populations 1987 70 56 126 Highlights in Science 1989 69 58 127 Today’s Science Tomorrow’s Technology 1991 61 70 131 Living with the Environment 1993 60 72 132 Carbon: Element of Energy and Life 1995 55 80 135 Breakthrough! Creativity and Progress in Science 1997 72 65 137 Light 1999 73 66 139 Millennium Science 2001 70 71 141 Impact Science 2003 54 85 139 From Zero to Infinity TOTALS 2293 1442 3735 HISTORY OF THE ISS 2005 ?? ?? 140 Waves of the Future 21 PROFESSOR HALINA RUBINSZTEIN-DUNLOP is Head of Physics and Director of the Centre for Biophotonics and Laser Science at the University of Queensland. She is a program manager of one of the scientific programs of the Centre of Excellence in Quantum Computer technology. Professor Rubinsztein-Dunlop is also a Research Director of the Faculty of Engineering, Physical Sciences and Architecture at the University of Queensland. After completing her PhD at the University of Gothenberg and Chalmers University of Technology in Sweden Halina worked on the development of laser based methods for ultra- sensitive trace element analysis and established a strong research group in this area. She moved to the University of Queensland in 1989. Halina’s research interests are in laser physics, laser micromanipulation, atom optics, quantum computing, linear and nonlinear high resolution laser spectroscopy, and nano-optics. She is one of the originators of laser enhanced ionisation spectroscopy, and is known for her work in laser micromanipulation and atom optics. Professor Rubinsztein-Dunlop has more than 130 international journal publications, 3 book chapters, and a large number of international conference contributions.

22 Catch, Move and Twist with Optical Tweezers: Biophotonics at work

Professor Halina Rubinsztein-Dunlop

Introduction THE PAST CENTURY HAS brought about an unprecedented number of technological breakthroughs, one of which is photonics. Photonics uses photons instead of electrons to transmit, process and store information, providing a great gain in capacity and speed on information technology. This all-encompassing light-based optical technology is predicted to become the dominant technology for this new millennium. The invention of lasers, which represent a concentrated source of monochromatic and highly directional light, has had a tremendous impact on photonics. Since the demonstration of the first laser in 1960 and its very first application in the correction of the retinal detachment, lasers have illuminated all aspects of our lives, from barcode scanners in the supermarkets and home entertainment, through high capacity information storage, to fibre optics communications, thus opening up numerous opportunities for photonics.

23 new extension of photonics is biophotonics, also know that such waves can transport linear the science of generating and harnessing momentum. If light is absorbed or reflected by an A light to image, detect and manipulate object, momentum is transferred to this object. biological materials. Biophotonics is used in This transfer of momentum from light to an biology to probe for molecular mechanisms, object creates a force on the object. It is like a function and structure. It is used in medicine to collision between two objects, only this collision study tissue and blood at the macro (large-scale) is slightly unusual in that one of the objects is and micro (very small scale) organism level to actually light! That is, it is possible to exert a detect, diagnose and treat diseases in a way pressure, called radiation pressure, on an object that are non-invasive to the body. by shining light on it.

Nature has used biophotonics as a basic principle Does it mean that we have to worry about of life from the beginning. Harnessing photons to opening the door on a bright day and being achieve photosynthesis and the conversion of knocked over by the light pouring in? Do we feel photons through a series of complex steps to a recoil force when we turn on a flashlight? create vision are the best examples of Obviously not. The recoil force involved is too biophotonics at work. Biophotonics offers great small in relation to the forces of our daily hope for the early detection of diseases and the experience for us to feel it. So how strong are the development of novel techniques for light-guided forces exerted by light? How can we describe and light-activated therapies. Lasers have already these forces? made a great impact on general, neural and plastic surgeries. Laser technology allows the How strong are the forces exerted by light? administration of a burst of ultrashort pulses that Assume that a parallel beam of light falls on an can be used for improved imaging and for tissue object for some time t, and is entirely absorbed engineering. Furthermore, biophotonics may be by the object. Maxwell showed that, if an amount used to produce retinal implants for restoring of energy E is absorbed during this time interval, vision. New and exciting applications of the magnitude of the momentum change, p,of biophotonics are emerging very quickly and many the object due to the absorption will be given by laboratories around the world are involved in p = E/c (where c is the speed of light). If the these rapidly expanding field. radiation is totally reflected back along its original path, the magnitude of the momentum Imagine focusing a laser beam specifically on to an change of the object is twice the amount we organelle, a structure within a living cell. Consider estimated for the absorption case – so in this further that the beam can actually grasp that tiny case the momentum change will be p = 2E/c. entity and hold it in place. Now imagine that while this microbeam This is similar to the situation acts as tweezers, a second beam “... can we catch, move or where you throw a ball at an serves as scalpel or scissors to object, like a milk bottle. When conduct a delicate surgery on the rotate microscopic you bounce a perfectly elastic organelle. Is this possible, or just a objects without touching tennis ball off the bottle, it figment of our imagination? In receives twice as much other words, we are asking: can them? Is it possible to momentum as when it is we catch, move or rotate use light to do all that?” struck by a perfectly inelastic microscopic objects without ball (a lump of putty, for touching them? Is it possible to example, which would stick use light to do all that? and not bounce off) of the same mass and velocity. If the incident radiation is partially absorbed and partially Radiation pressure reflected, the momentum change of the object WAVES OF THE FUTURE OF WAVES It is well known and easily verifiable that is somewhere between the two of the above 24 electromagnetic waves carry energy. Perhaps you estimated values. then be4.4x10 ihsfiin nest,andthereforesufficient with sufficientintensity, question iswhetheritpossible tocreatelight The object suchasabiologicalcellaround. that itisnotevenenoughtomove amicroscopic W/m find thatweneedanintensityofabout11.7 wewill needed tomovearoundsuchanobject, question ofwhatsortintensitylightwouldbe about 8x10 force onanarea is completelyabsorbedthenwecanseethatthe intercepted bythearea theenergy t, Inatimeinterval of theradiation. on aflatarea intensity, by radiationpressureintermsofthelight To findexpressionsfortheforceexerted F =p/t. change inmomentumisrelatedtoaforcegiveby From Newton’s secondlawweknowthata needed tomovea If weinsteadaskhowmuchforcewouldbe impossible tousethelightforcedothistask! soitwouldbe is twelveordersofmagnitude, Thedifferencebetweenthesetwoforces 588 N. the forcerequiredtoovercomegravitywouldbe ifwewanttoliftapersonweighing60kg around: the forcethatisneededtomoveanobject gravity willbe7.84x10 Earth’s surfaceisabout 0.2W/m Average lightintensityfromsunatthe such aforcewouldbe7.84x10 is 3x10 around 0.2 W/m power perunitareadeliveredbytheSunis andthataverage is 40cmwideby1.65mtall, we canassumethattheareaexposedtosun Forsimplicity we openthedooronabrightday. magnitude oftheforceexertedonourbodywhen the forexample, wecanestimate, Knowing this, force intheabsorbingcase. theforcewillbeequaltotwice reflected, Ifinsteadtheradiationistotally F =IA/c. thespeedoflight–thatis, whichisc, constant, times theareaA volume 2x10 that wecanapproximatethecellbyacubeof Ifweassume different. situation wouldbevery arough estimateshowsthatthe biological cell, 2 andthemomentum transfertoproduce , 8 wehavetoconsiderradiationfalling I, /,theforceexertedonourbodywill m/s, -15 A -6 g Sotheforceonit dueto kg. that isperpendiculartothepath m 2 -10 Knowingthatthespeedoflight . A divided bytheproportionality 3 microscopic thenitwillhaveamass of , is givenbytheintensity .We couldcomparethiswith N. A -14 is .Ifwegobacktothe N. Iftheenergy E =IAt. bet suchasa object, 2 -14 whichsuggests , kg.m/s. I iue1 Tail fortheSun. ofacometpointsaway Figure 1. around insomeway? formovingmicroscopicobjects momentum, radiation pressure. radiation proposed byNicholsandHullforameasurement of Thearrangement oftorsionbalancetechniqueas Figure 2. pressure wasmade between1901and1903by The firstexperimentalmeasurement ofradiation [FIGURE 2HERE] surfaces heatedbylightratherthandirectly. the forcesofmolecularbombardmenton butinfactitsactionisdueto radiation pressure, 1873 wasthoughtforsometimetodemonstrate radiometer inventedbySir William Crookesin The mundane causessuchasconvection. forces couldalwaysbeexplainedbymore these earlyattemptsfailedasthemeasured butallof measure theradiationpressureoflight, to was theinspirationforearlyexperimentstotry In latertimesNewton’s oflight corpusculartheory comet tailstoalwayspointawayfromtheSun. proposed thatlightpressureiswhatcauses Keppler Intheseventeenthcentury new one. The conceptthatlightcanexertforcesisnota micromanipulation Optical 25 CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK Nichols and Hull1 and by Lebedev2, thirty years the particle of 5x104 m/s2, more than 1000 times after Maxwell’s theoretical predication of this the acceleration due to gravity! effect. In order to measure the radiation pressure force they used a torsion balance technique. A Such calculations led to many experiments in the mirror (a perfect reflector) and a black disk (a early 1970s that used laser beams in various perfect absorber) are connected by a horizontal configurations to move microscopic particles rod suspended from a fine fibre. Normal- around. The particles used in these experiments incidence light striking the black disk is were transparent, as the heating effects on an completely absorbed, so all of the momentum is absorbing particle were thought likely to obscure transferred to the disk. Normal-incidence light radiation pressure effects similar to the way that striking the mirror is totally reflected, and hence heating and convection foiled earlier attempts to the momentum transferred to the mirror is twice measure radiation pressure. Ashkin’s paper from as great as that transferred to the disk. The 1972 entitled ‘The Pressure of Light’ (see radiation pressure is determined by measuring reference 3) discusses some of these the angle through which the horizontal connecting experiments in detail. One of the experiments rod rotates. The measured radiation pressures described there considers small transparent are very small (about 5x10-6 N/m2). How can we plastic spheres dispersed in water, placed in a increase the radiation pressure force that is glass cell under a microscope and illuminated available to us? Maybe one way of doing it is to from below by a focused laser beam. In his focus the light beam. observation Ashkin noted that not only was there a force propelling transparent particles in the The invention of the laser in 1960 enabled an direction of beam propagation, but spheres near unprecedented development within the optical the edge of the beam experienced a force pulling research and applications. With researchers now them into the centre of the beam. being able to focus very strong beams of coherent light on to objects, the effects of radiation pressure were no longer limited to a [FIGURE 3 HERE] feeble force observable only under high vacuum conditions. Early demonstration of the radiation pressure by Ashkin3 involved optical levitation of a small particle by means of vertically directed laser beam. In this famous experiment, which appears in many undergraduate physics textbooks today, a transparent glass sphere of about 20 microns in Figure 3. Refraction of a pair of rays by a high refractive diameter is lifted in air about 1 cm above a glass index sphere. A pair of rays from a Gaussian beam are plate. The scattered laser light makes the refracted as the pass through a sphere with refractive levitated spheres visible to the naked eye, a index higher than the surrounding medium. Since ray a is striking display of laser light force. stronger than ray b, the force on the sphere due to refraction of ray a, Fa, is larger than Fb, and the sphere is pulled into the centre of the beam propelled in the direction We can do a simple calculation of the magnitude of beam propagation by radiation pressure. of the radiation pressure force. Assume we have a 10 mW Helium-Neon (He-Ne) laser beam, focused to a spot one wavelength (λ) across, This effect can be explained by considering a pair onto an object of the same diameter with density of rays of a Gaussian-shaped laser beam a and b ρ = 1 g/cm3. If the particle were 10% reflective situated symmetrically with respect to the centre and acted as a flat mirror, then the momentum of a sphere having refractive index higher than change of the laser light each second would be that of the medium. (A Gaussian beam is one 2x10-3/(3x108) kg.m/s. By conservation of where the cross-section intensity profile is a momentum, this results in a force on the particle Gaussian, or ‘normal function’, shape – bright in WAVES OF THE FUTURE OF WAVES of approximately 6.7x10-12 N. Estimating the mass the centre, becoming less intense radially ρπ 3 26 of the particle by r /6 gives an acceleration on outwards.) For a sphere situated off the axis of a Gaussian beam, ray a is stronger than ray b. On strongly focusing the beam. This gradient force transmission through the sphere, rays a and b has both radial and axial components, and the are bent, giving rise to forces Fa and Fb in the axial component can be made strong enough to opposite directions to the momentum change of overcome the gravitational and scattering forces the rays. Since ray a is stronger than ray b, the on a small dielectric particle, and thus confining it force Fa is greater than force Fb. The resultant to the most intense region, the beam waist. This force F = Fa + Fb has components both in the trap does not rely on the balance of the direction of the beam propagation and towards scattering force (radiation pressure) and gravity. the centre of the beam, so one expects a net transverse force pulling the high index sphere Another, simple way of treating forces acting on a into the centre of the beam where the intensity is microscopic particle exposed to a highly focused maximum (see figure 3). light is by analysing the rays a and b of light incident on the particle as shown in figure 4. That means that a particle that is mostly transparent and has refractive index greater than the surrounding medium experiences a gradient [FIGURE 4 HERE] force, which is greater than the absorption force, when irradiated by a tightly focused beam of light, and is thus trapped in the region of highest intensity. Under favourable circumstances this will even mean that such particles will move back along the beam axis to a beam waist, giving three dimensional trapping.

A highly absorbing particle, however, will experience a much greater absorption force than gradient force and will be ‘pushed’ away from the region of high intensity. These particles cannot Figure 4. Ray optics model of optical trapping of therefore be trapped three-dimensionally using transparent spheres. Restoring force on a sphere when it is displaced form the focus. Gaussian laser beams, the most common profile of laser light. They can, however, be trapped using a ring of light, a so called ‘doughnut beam’. When the light ray entering a particle with Absorbing particles are pushed away from the refractive index greater than that of the intense region of the beam, either away from the surrounding medium is bent toward the normal, beam or, preferably, into the central dark region the change in momentum of the light results in a where they will be trapped. force on the particle. If we follow the rays a and b, as shown in the figure 4, we can see the With the ordinary Gaussian beam, in 1978 forces produced by change of the momentum. If Ashkin4 showed theoretically that a gradient force the centre of the particle is below the focal spot could be produced in the direction opposite to of the beam, the particle will be moved upwards beam propagation, and that small dielectric to the most intense part of the beam. If the particles could be three-dimensionally trapped particle is above the centre of the beam it will be using a single laser beam. In 1986 Ashkin et al.5 pushed down, and if it is situated to the side of reported the observation of the first single beam the waist of the beam it will be pulled sideways gradient force optical trap, also known as into the centre. The total effect will give a three- optical tweezers. dimensional trap.

The distinguishing feature of this trap was that it The essential elements of the single-beam was the first all-optical single-beam trap. A gradient optical trap, or optical tweezers, is a high CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK gradient force, proportional to and in the direction numerical aperture lens, necessary to bring the of the beam intensity gradient, is produced by trapping beam to the tightest possible focus. This 27 use – once a particle is trapped at the beam waist it can be manipulated relative to its image surroundings by either changing the angle at which the beam enters the back pupil of the Laser light objective (that is, by moving the laser beam) or by moving the cell containing the sample. Both of Microscope these are relatively uncomplicated, making optical objective tweezers the first truly practicable N.A.~1.3 micromanipulation tool. Following the discovery in 1987 that live biological specimens could be optically trapped for quite long periods of time (from minutes to hours) and still remain viable, the single-beam gradient trap has found many applications in biologically related fields, and its popularity as a Figure 5. The essential elements of a single beam optical biophysics and biology tool is steadily increasing. tweezers trap. The same objective is used for both In response to the demand of researchers, there producing the trap and imaging. are several commercially produced optical tweezers instruments on the market. tight focussing is responsible for the axial gradient force and a microscope objective is Early application of optical micromanipulation typically used for this purpose. The typical set up mostly made use of the advantage that live for optical tweezers is shown below, where the specimens can be manipulated in a closed same objective is used for both producing the environment in a very controlled way, leading to trap and imaging. the use of optical tweezers in applications which previously employed micropippetes. With the In figure 6 we show a photograph of a simple additional tool of a cutting beam, which can be optical tweezers set-up. It consists of a high achieved with the same arrangement as tweezers numerical aperture objective, a piezo-driven stage but with the laser light delivered in short pulses with microscope slide and cover slip containing and the wavelength chosen so that it is absorbed particles dispersed in water, the laser source and by the specimen, optical tweezers have found some optics to shape the beam entering the their way to wide variety of research. objective. The same objective is used here for observation of the trapped particles. Optical tweezers have been used to trap dielectric spheres, viruses, bacteria, living cells, organelles, small metal particles, and even strands of DNA. Applications include confinement and organization (for example, for cell sorting), tracking of movement (of bacteria, for example), application and measurement of small forces, and altering of larger structures (such as cell membranes). Other uses for optical traps have been the study of molecular motors and the physical properties of DNA. In both areas, a biological specimen is biochemically attached to a Figure 6. Photograph of the microscope constructed for micron-sized glass or polystyrene bead that is optical tweezers. The set-up is versatile with respect to then trapped. By attaching a single molecular adding and removing optics and photodetectors. motor (such as kinesin, myosin or RNA WAVES OF THE FUTURE OF WAVES Because it consists only of a single laser beam, polymerase) to such a bead, researchers have 28 the optical tweezers trap is extremely simple to been able to probe molecular motor properties. The studies have started to answer questions Another use of optical tweezers and scissors is such as: Does the motor take individual steps? for cell fusion. In this case the optical trap is What is the step size? How much force can the combined with a pulsed UV laser micro-beam. motor produce? Similarly, by attaching the beads The two selected cells are brought into close to the ends of single pieces of DNA, experiments contact by the optical tweezers. Once inside the have measured the elasticity of the DNA, as well trap, the two cells can be fused by applying as the forces under which DNA breaks or several pulses of the UV laser micro-beam. With undergoes a phase transition. this technique a selective fusion of two cells is done without critical chemical or electrical When two trapping beams are introduced (a treatment. Laser induced cell fusion should double-beam optical tweezers), the microscope provide an increased selectivity and efficiency in studies can be performed on objects that have to generating viable hybrid cells in the future8. be stretched, aligned or turned. Using this technique, Chu et al.6 studied recoil behaviour and viscoelastic properties of DNA molecules. The molecules were stretched out and fixed with the tweezers and then viewed with scanning tunnelling or atomic force microscopes.

(a) (b) (c) (d) When optical tweezers are combined with a laser micro-beam (also called an optical scalpel) Figure 8. Trapping and cutting beam for the cell fusion. Two 7 human lymphocyte cells (No. 1 and 2) were brought controlled cell fusion can be carried out . The together by means of the optical tweezers (a). Laser pulses basic idea of the use of a laser scalpel for of the cutting beam (dark spot in b denoted by the arrow) intracellular microsurgery on biological objects is perforated the outer cell membrane and both cells fused that the laser cutting beam is strongly focused together (c, d). 40 seconds later the cells start to fuse (c). onto the object from a large numerical aperture 160s after the wall perforation (d). microscope objective (like the beam of the optical tweezers). This ensures that the laser scalpel has An important application of combined usage of very short effective depth of field, implying that optical tweezers and UV-laser micro-beam is there will be enough power density for the manipulation of gametes and early embryos. desired effect only at very limited depth in the Using these techniques the fertilization processes object (1-2 µm). Above and below this depth, the can be studied in more detail leading to light intensity will cause no harm to the tissue. increased efficiency of in-vitro fertilization. The This ensures that the micro-beam surgery will be combination of a UV-laser micro-beam and an carried out in the interior of an unperforated cell optical tweezers was first suggested by Ng et al.9. with no damage to cell walls or membranes. Subsequently Schütze et al.10 successfully drilled a hole into the zone pellucid and inserted a single sperm through the laser drilled hole into the [FIGURE 7 HERE] pervitelline space using these combined techniques.

Figure 9 below shows another example of the use of a micro-beam. Here sperm motility is stopped with a few laser pulses placed close to the waving tail (a reversible process). Cutting the sperm tail is possible by focusing the laser onto the tail and performing a single laser shot (an Figure 7. Laser microbeams is used as scissors to cut irreversible process). chromosomes. In the visible range (400-700 nm), these

beams can be coupled with imaging methods and optical CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK tweezers. The arrows indicate where the cutting beam was We can conclude that a combined system of focused on mitotic chromosomes in a living cell. optical tweezers, laser micro-beam and laser 29 fictional nanobots, microrobots have been fabricated with ‘elbows’ and ‘fingers’ that are capable of manipulating micrometer-sized objects11. Production of working micromachines has motivated research into microdevice production methods, surface engineering of the substrates and driving mechanisms for the machines. Microscopic electromagnetic motors and piezoelectric and electrostatic actuators have been incorporated into MEMS. More novel driving mechanisms that avoid contact with the macro world suggested for micromachines include a dielectric fluid motor based on convection, an opto-micro-engine based on the same principles as a Crookes’ radiometer, and the use of optical torque from strong sources of laser light. Figure 9. Trapping of sperm for in vitro fertilization. Stopping sperm motility is done with a few laser shots placed close to the waving tail (reversible process). Cutting Microdevices are getting smaller and are actually the sperm tail is possible by focusing the laser onto the tail becoming nanodevices, and laser light has played and performing a single laser shot (irreversible process). a role in many of the advances that are making this possible. Together with other modern technologies, lasers have been used in almost all induced fluorescence detection converts the aspects of micro and nanodevice research, ordinary light microscope from being a passive including their fabrication, construction, and as a analytical tool into a preparative and manipulative source of torque to provide a driving mechanism instrument that allows micro-manipulation of for the devices. Laser light has been used to biological objects without any mechanical contact. polymerize resins to produce structures with nanometer-sized features12 and to bring together Optically driven micromachines parts of a two-element moving microsystem13.In both these and other experiments light was also It isn’t uncommon for scientists to be inspired by used to drive the rotation of the elements. science fiction, and good science fiction has a strong basis in real science. Modern science fiction is well populated with ‘nanobots’ and Light as a source of torque ‘microbots’, tiny robots that are injected into the There are two basic ways that light can be used to bloodstream where they perform tasks ranging drive the rotation of an object. In the first case, the from repairing damage and curing diseases to torque originates from the light itself carrying controlling human thoughts and actions. The angular momentum, which can then be transferred miniature robot concept may seem far-fetched to the object by processes such as transmission, today, yet technological advances in the area of reflection and absorption. Types of light carrying micromachine research are bringing the angular momentum include elliptically polarized possibility of devices like these much closer to beams and beams with helical phase structure. In reality than fiction. the second category, the torque originates from the shape of the object. Radiation pressure can act on Micromachines have potential advantages over asymmetries in the object’s shape in a similar way macromachines in their mobility, information to wind on the blades of a windmill: light deflected transfer rates and energy efficiency. The best- by the particle exerts torque to drive rotation. known micromachines are the microelectromechanical systems (MEMS) that Angular momentum due to polarization WAVES OF THE FUTURE OF WAVES incorporate mechanics together with electronics Scientists have known for a long time that certain on a miniature scale. More along the lines of the 30 types of light carry angular momentum, and so in principle that light could be used to exert torque is transferred to the plate. In Beth’s experiment on an object. In practice though, the effects of circularly polarized light was passed twice optical angular momentum are hard to observe, as through a wave plate suspended on a torsion they represent very small quantities. For example, fibre, and the resulting period of oscillation the angular momentum flux carried by a circularly measured. In his experiments he was able to polarized 10 mW He-Ne laser beam is of the order confirm that the sign and magnitude of the effect of 10-18 Nm, which is millions of times smaller than agreed with theory – the light did indeed possess the torque driving the balance wheel of a and transfer angular momentum. mechanical wrist watch. Each photon of circularly polarized light carries +h of angular momentum A modern version of this experiment (figure 10b) (where h =h/2π, and h is Planck’s constant). uses laser light, optical tweezers and microscopic waveplates to observe the same torque, and is In 1936 Richard Beth carried out the first the basic idea behind the use of circularly measurement of torque produced by light14,in polarized light to drive the rotation of microscopic which he used a series of waveplates suspended elements15. A laser beam is focussed to a very on a torsion fiber (see figure 10a). When linearly small spot, providing an extremely intense light polarized light is passed through a birefringent source, and is passed through spherical material of the correct thickness (λ/4), it birefringent calcium carbonate crystal. The calcite becomes circularly polarized. If that circularly polarized light is passed through a linear polariser Figure 11. 11(a) The electric field vector of circularly (λ/2 plate), the handedness of the polarization is polarized light. Each cycle the electric field vector rotates reversed and, in the process, angular momentum 2π radians. The rotation of the electric field vector at he optical frequency is associated with the spin angular momentum of a circularly polarized photon. 11(b) The phase fronts of a helical beam of charge l = 3, Torsion and the intensity pattern when interfered with a plane wave. The intensity pattern when interfered with a plane fibre Aluminium wave is the configuration used in the propeller beam layer experiments. slow fast λ/4

fast slow λ/2

fast slow λ/4

(a)(a)

(b) Figure 10. Schematic diagram for measurement of torque produced by polarized light as in the original Beth experiment, (a), and as performed when using

spherical birefringent crystals CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK in optical tweezers with polarized laser beams, (b). 31 particles can act as wave-plates. On passage Depending on the type of phase plate used, light through a crystal, different components of the with many intertwined helices can be created. incident light will undergo different phase shifts Orbital angular momentum can be transferred to and the electric field vector will rotate (figure an object by absorption or reflection, and can 11a). This will introduce a change in the angular also be used to spin tiny particles trapped using momentum carried by the light, and there will be optical tweezers 16. When light absorption is the a corresponding torque on the material. cause of the angular momentum transfer, both Depending on the polarization of the incident spin and orbital angular momentum can be beam, the particles either become aligned with transferred to the material at the same time. It is the plane of polarization (and thus can be rotated then possible to change the rotation rates of the through specified angles) or spin with constant spinning objects by a simple rotation of the wave rotation frequency. The light transmitted through plates, or by reversal of the helicity of the wave. the crystal will become elliptically polarised. Effectively, the measurement of the change in These helical wavefronts can be interfered with a polarisation state of light on transmission though plane wave so that a spiral interference pattern the birefringent crystal will allow determination of with three arms is produced within an optical the rotation rate of the crystal and also an optical tweezers trap (see figure 11b). The arms of the measurement of torque. pattern rotate when the path length of the interferometer is changed, and so particles that Polarized light can also exert torque on an object are trapped in the bright regions of the arms are if the material is absorbing – in that case if the rotated too17. These ‘propeller beams’ offer a light carries both linear and angular momentum, technique that does not depend on intrinsic both will be absorbed by the material. In this properties of the particles (e.g. birefringence) and case, not only does the rotating object feel force also avoid absorption of the field – the method in the direction of light propagation, it is also requires that the particles be transparent and heated as it absorbs energy from the field. This have a higher refractive index than their means that high rotation speeds cannot be surrounding medium. achieved, since the increase in laser power required to increase the spinning rate will burn the rotating element. So the phenomenon is quite Optical tweezers and light-driven interesting scientifically but due to these extra machines effects is less useful as a possible driving Microscopic particles trapped in the tightly focused mechanism for rotating microscopic objects. beam of an optical tweezers trap often tend to rotate, either due to their own shape or through Helical light and propeller beams interaction with light carrying angular momentum. The source of the polarization torque is the Both these effects have been used to drive the rotation in time of the electric field vector of the rotation of microscopic machine elements. light field. Light can also exert torque if the wavefront associated with the field is rotating in Recently, a method to build microscopic light- time. Here the wave has a helical phase driven rotors was reported18 where optical structure (figure 11b) while its electric field may tweezers were used in both the production and be rotating or not. We call the angular manipulation of the rotors. Microscopic particles momentum from polarization spin angular of arbitrary shape were produced by a two- momentum, and the angular momentum due to photon polymerization method. In this experiment the helical structure orbital angular momentum. the researchers used a resin that, when Just as passing light through a wave plate can polymerized, results in a glass-like material with produce circularly polarized light, helical light can refractive index n=1.56, which is ideal for be produced when light is passed through a handling using optical tweezers. The light source phase plate. Combinations of the two are of for the polymerization process was at the focus WAVES OF THE FUTURE OF WAVES course possible if the light is prepared using intense enough to initiate two-photon excitation. 32 both wave and phase plates. To build the structures, the beam focus is moved along a preprogrammed trajectory and a three- microfabricated elements were cog-like shapes dimensional shape is built up from the line along 10µm in diameter and 0.5µm thick, made of SiO2 which the resin is hardened. Arbitrary shapes using a photolithographic double-liftoff technique. could be constructed using this method with features of about 0.5 µm in size. The researchers Once trapped, the rotor was moved next to a experimented with different shapes including trapped cog (see figure 13). Then, when the light helices, sprinklers and propellers, and found that was made circularly polarized through rotation of the most efficient shape for rotation was the a λ/4 plate, the calcite spun, inducing the nearby sprinkler shape, with an added central linear axis cog to rotate also. In this experiment the torque is to improve stability in the optical tweezers trap. transferred from the rotor to the machine element Using 20 mW in the optical trap produced via the fluid between them. For both particles, the rotation rates of several Hertz. optical tweezers acts as an axle for them to rotate about.

[FIGURE 12 HERE]

spinning CaCO3 crystal Microfabricated SiO2 “cog” Figure12. Light-driven micromachinery produced and driven by light. The arrows point to the three interconnected machine cogs. Figure 13. Optically driven and assembled micromachine. Dual fully steerable optical tweezers are used. The first trap Using this method, a complex micro machine is used to trap a birefringent crystal, the other one is used consisting of two engaged cogwheels rotated by a to trap the “cog”. The light is circularly polarized and the light drive rotor was constructed, where the rotor crystal rotates. The other trap brins the “cog” close to the was held and driven by optical tweezers and the spinning crystal. The rotation of the cog is induced. The torque here is transferred from the rotor to the machine cogwheels rotated on axes fixed to the glass element via the fluid surroundings. The dual optical surface (see figure 12). The same techniques could tweezers provide axles for the rotor and the machine be used to produce much more complicated element. arrangements, offering a promising method for constructing micron sized light driven machines. This proof of principle experiment has at least one obvious drawback: energy is being lost to the Micromachine elements have also been driven by fluid that could be used to drive the element. If light by transferring the angular momentum from the machine elements could be made from a photons to a microscopic particle, which was birefringent material, then light could be used to then used to drive rotation of a microfabricated drive the elements directly, avoiding the need for element19. Optical tweezers again played an the calcite. Simple birefringent structures of a important role in the experiment. Two fully similar size to these cogs have already been steerable traps were used to hold and manipulate produced20 that show the same behaviour in the light driven rotor and the microfabricated polarized light as calcite, so production of more element. The rotor consisted of a microscopic complicated shapes may soon follow. fragment of birefringent calcite, which could be CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK induced to spin at hundreds of Hertz when The advantages of light driven microscopic rotors trapped in circularly polarized light. The and machine elements are obvious: the non 33 contact nature of the driving mechanism means allows highly localized measurements to be made, that these micromachines can be operated in any since the probe particle does not move in the microscopic systems that are accessible by laser surrounding medium. At the same time the use of light. Micro-rotors may find application as spherical probe particles greatly simplifies the instruments for measuring properties of biological theoretical analysis of the fluid flow. Also, since the systems such as torsional elasticity of biological probe particle is rotationally driven by the optical polymers or microscopic viscosity. Spinning trap, the rotation rate can be readily controlled. birefringent particles have already been used to turn biological specimens around to view them Using this technique the viscosity of water and from different angles21 and for studies of cell some other liquids in extremely small volumes of membranes. sample have been measured, demonstrating further possible applicability of this method. By monitoring the change in circular polarization of light passing through an object, the reaction torque Recently we have combined the techniques of on the object can be found. This idea was optical scalpel and rotating tweezers to measure exploited as the basis for an optically driven the properties of liquid inside a biological cell. In viscometer, an instrument that can measure this experiment a biological cell was exposed to viscosity of liquids. The viscosity of a liquid can be very sharply focused pulsed laser with the pulses determined by measuring the torque required to of the order of femtosecond (that is, 10-12 of a rotate a sphere immersed in the liquid at a second) in length for a short period of time. These constant angular velocity – this concept has been laser pulses created a hole in the membrane of implemented at the micrometer scale using a the cell. Subsequently optical tweezers were used system based on optical tweezers. Birefringent to insert a birefringent spherical crystal into the spheres of synthetically grown material (see figure cell, and then rotating tweezers were used to 14) were trapped three dimensionally in a measure the viscosity of the liquid inside the cell. circularly polarized optical tweezers trap and the Figure 15 shows the cell with the crystal inside it. frequency of rotation as well as the polarization change of light passing through the particle were With further advances in the fabrication materials measured to determine the viscosity of the and methods, the manufacture of microscopic surrounding fluid. This method, which is based on fluid pumps for extremely localized delivery of direct optical measurement of the rotation of a chemicals will become possible. Optical tweezers spherical probe particle in a stationary position, and optical scissors have been joined by the

Figure 14. (a) Optical microscope image of a typical vaterite crystal used for viscosity measurements. (b) Scanning electron microscope image of a vaterite crystal. WAVES OF THE FUTURE OF WAVES

34 Figure15. Biological cell with a birefringent spherical crystal inside it.

optical spanner and the propeller beam, providing 19 MEJ. Friese, TA. Nieminen, NR. Heckenberg and H. tools to trap, manipulate, cut, align, turn over and Rubinsztein-Dunlop, Nature 394, 348 (1998). rotate a wide range of micro- and nano-objects. It 20 E. Higurashi, R. Sawadaand T. Ito, J. Michromech Microeng seems certain that these laser tools will continue 11, 140 (2001) to make their mark in the world of micro 21 S. Bayoudh et al., J Microsc 203, 214 (2001) engineering.

References 1 E. Nichols and G. Hall, Phys. Rev. 13:307 (1901). 2 P. Lebedev. Ann. Phys. 6:433 (1901). 3 A. Ashkin, Scientific American 226,(2): 62 (1972). 4 A. Ashkin, Phys. Rev. Lett. 19: 283 (1978). 5 A. Ashkin, JM. Dziedzic, JE Bjorkholm and S. Chu, Opt. Lett. 11(5 : 288 (1986). 6 TT Perkins, De. Smith and S. Chu, Science 253, 861 (1991). 7 K. Schütze and A. Clement_Sengewald, Nature 368, 667 (1994). 8 WH. Wright, GH. Sonek, Y. Tadir and MW. Berns, IEEE Journal of Quantum Electronics 26, 2149 (1990). 9 Y. Ng et al., J. Assist. Reprod. Genet. 9, 191 (1992) 10 K. Schutze, A. Clement-Sengewald and A. Ashkin, Fert. Steril. 61, 783 (1994) 11 E.W.H. Jager, O. Inganäs and I. Lundström, Science 288, 2335 (2000) 12 P. Galajda and P. Ormos, Appl. Phys. Lett. 78, 249 (2001) 13 M.E.J. Friese et al., Appl. Phys. Lett. 78, 1 (2001) 14 R.A. Beth, Phys. Rev. 50, 115 (1936) 15 M. Friese, T.A.Nieminen, N.R. Heckenberg and H. Rubinsztein-Dunlop, Nature 394, 348 (1998) 16 H. He, M.E.J. Friese, N.R. Heckenberg and H. Rubunsztein- Dunlop, Phys. Rev. Lett. 75, 826 (1995) CATCH, MOVE AND TWIST WITH OPTICAL TWEEZERS: BIOPHOTONICS AT WORK 17 L. Paterson et al, Science 292, 912 (2001) 18 P. Galajda and P. Ormos. Appl. Phys. Lett. 78, 1 (2001). 35 Einstein Failed School By Dr Karl Kruszelnicki

AT THE END OF the 20th Century, Time magazine voted Albert Einstein to be the Man of The Century. Albert was the dude who came up with all that really weird Relativity stuff – and he was your genuine certified Mega Brain. After all, we are told that he even won the Nobel Prize for his work in Relativity. On the other hand, generations of school kids have consoled themselves over their poor school marks with the belief that Einstein failed at school. Some motivational speakers also make this claim – but this claim is as wrong as the claim about the Nobel Prize.

First, Einstein did not win the 1921 the unglamorous Photoelectric prestigious Federal Polytechnic Nobel Prize in Physics for his work Effect that got him the Nobel Prize. School (or Academy) in Zurich, on Relativity. Let’s back up a little. Well, that’s one myth out of Switzerland. He was 16, two years Back in 1905, Einstein had the the way. younger than his fellow applicants. biggest year of his life. He wrote, He did outstandingly well in physics with the help of his wife, Mileva, Second, Einstein definitely did not and mathematics, but failed the five ground-breaking papers that, fail at high school. Einstein was non-science subjects, doing according to the Encyclopaedia born on 14 March in Ulm, in especially badly in French – so he Britannica “forever changed Man’s Germany, in 1879. The next year, was not accepted. So in that same view of the Universe”. Any scientist his family moved to Munich. At the year, he continued his studies at would have been proud to write age of 7, he started school in the Canton school in Aargau (also even one of these magnificent Munich. At the age of 9, he entered called Aarau). He studied well, and papers – but Albert published five the Luitpold-Gymnasium. By the age this time, he passed the entry of them in one year! of 12 he was studying calculus. exams into the Federal Polytechnic Now this was very advanced, School. One paper, of course, dealt with because the students would Relativity – what happens to normally study calculus when they So the next year, he finally started objects as they move relative to were 15 years old. He was very studying at the Federal Polytechnic other objects. Another paper proved good at the sciences. But, because in Zurich (even though he was now that atoms and molecules had to the 19th-century German education one year younger than most of his exist, based on the fact that you system was very harsh and fellow students). Also in the year could see tiny particles jigging regimented, he didn’t really develop 1896, even though he was only around when you looked at a drop his non-mathematical skills (such as 16 years old, he wrote a brilliant of water through a microscope. A history, languages, music and essay that led directly to his later third paper looked at a strange geography). In fact, it was his work in relativity. property of light – the Photoelectric mother, not his school, who Effect. Plants and solar cells do the encouraged him to study the violin So he definitely did not fail his high Photoelectric Effect, when they turn – and he did quite well at that school, and definitely was not a light into electricity. His paper as well. poor student.

WAVES OF THE FUTURE OF WAVES explained the Photoelectric Effect. Relativity may have captured the In 1895, he sat the entrance So how did the myth that he failed 36 public’s consciousness, but it was examinations to get into the high school start? Easy. In 1896, which was Einstein’s And so, anybody looking up last year at the school in Aargau, Einstein’s grades would see that he the school’s system of marking had scored lots of grades around was reversed. “1” – which under the new marking scheme, meant a “fail”. A grading of “6”, which had previously been the lowest mark, And that means that schoolkids was now the highest mark. And so, can’t use that mythconception as a

a grading of “1”, which had been crutch any more – they’ll just have EINSTEIN FAILED SCHOOL the highest mark, was now the to work harder... FROM Dr Karl’s book Mythconceptions lowest mark. (Harper Collins Publishers) 37 DR CLIVE BALDOCK Before attending university, Clive was employed as a medical physics technician in the Nuclear Medicine Department of St. Paul’s Hospital, London and the Institute of Urology, University of London. He subsequently obtained his BSc (Hons) in Physics from the University of In June 2003 Clive was appointed as Sussex, Brighton in Senior Lecturer and Director of the 1987. He then worked as a Basic Institute of Medical Physics within the Grade Medical Physicist in the School of Physics at the University of Department of Clinical Physics and Sydney. The Institute acts as an Bioengineering, Guy’s Hospital, umbrella organization for medical London and the United Medical and physics activities including research, Dental Schools (UMDS), University postgraduate supervision and of London. undergraduate teaching within the School of Physics. The Institute has In September 1993, Clive moved to strong links to University of Sydney the Medical Physics and Nuclear Hospitals as well other hospitals in Medicine Departments, Royal Sussex New South Wales. County Hospital, Brighton Health Care NHS Trust, as part of the team Clive’s current research interests providing the scientific service include radiotherapy gel dosimetry, supporting the Nuclear Medicine radionuclide dosimetry, motion Department and Magnetic Resonance correction in medical imaging and Imaging (MRI) scanner. From January radiotherpy, electronic portal imaging, 1997, he was a Lecturer, and kilovoltage dosimetry Monte Carlo subsequently Senior Lecturer, in calculation and applications of Medical Physics in the Centre for SPECT/CT and PET/CT in Medical, Health and Environmental radiotherapy. Physics, School of Physical Sciences, Queensland University of Technology (QUT) in Brisbane, Australia.

38 The Treatment of Cancer Using Ionising Radiation

Dr Clive Baldock

Radiation and what it does to cancer CANCER IS CHARACTERIZED by a prolific and uninhibited replication of cells, which can interfere with the function of normal cells and organs, thereby endangering the life of the patient. The aim of radiation treatment is to deliver a sufficient radiation dose to sterilize cancer cells while limiting accidental damage to adjacent healthy tissue.

39 2 ut what is ‘dose’, and what does it mean to Sfx = exp(–αD – βD ) sterilize a cell? Radiation dose is the energy Babsorbed by cells from an incident radiation where α is the coefficient of non-repairable beam, per unit mass of tissue, and is a measure damage and β is the coefficient of repairable of molecular damage inflicted on the cells. It is damage. A plot of log Sfx against dose is called a measured in joules per kilogram (Jkg-1). Many cell-survival curve – an example is shown in different types of radiation have been investigated figure 1. for cancer-cell killing efficacy with varying results: these include beams of photons, electrons, neutrons, protons, pions and even heavy atoms like carbon-12. The majority of modern clinics use photon and/or electron beams to treat cancer, and these beams disrupt normal cellular function principally by breaking chemical bonds through ionizing interactions. These ionizations can corrupt key molecules required for cellular replication and metabolism. A cell is sterilized when sufficient molecular damage has been inflicted that the cell can no longer replicate. Note that while cancer cell-death may be preferable, cell sterilization is often sufficient to preserve the life of the patient. Figure 1: Cell survival curve illustrating the percentage of Radiation damage to DNA molecules in the cell surviving cells after doses of radiation. nucleus is the primary mechanism for cell sterilization. DNA contains coded instructions for Empirical tests determine that DNA repair the entire functionality of the cell, and when a cell mechanisms of tumour cells are significantly less replicates the DNA divides and an identical copy effective than those of normal healthy cells (that is made for the offspring cell. Radiation damage is, tumour cells exhibit low values of β). This to DNA can prevent cell replication and the limit effect is exploited by ‘fractionating’ the radiation the viability of the offspring cell. treatment into a succession of small doses, typically one dose-fraction a day for six to seven Radiation biologists have developed the linear- weeks. After each fraction the normal healthy quadratic (LQ) mathematical model to describe cells are able to repair some of the damage radiation sterilization of cancer cells. The model caused by the radiation, whereas the tumour cells assumes that a cell is sterilized when both are unable to repair this damage, leading to strands of a DNA double helix molecule are compounded decimation with each new fraction. broken; this can occur either by a single particle The fractionation is optimized when maximum that breaks both strands at the same time, or by cancer cell sterilization is achieved with minimal two particles that each break one of the strands damage to normal tissues. Determining the with a short time interval between breakages. The optimal trade-offs between dose per fraction, significant difference between this ‘single’ or time interval between fractions and total ‘double’ DNA hit is that normal cellular repair treatment dose is the subject of a great deal of mechanisms can repair some or all of the double current research. hit damage (the first broken strand may be repaired before the second strand is broken) but these mechanisms are unable to repair the How do we generate radiation in a substantial damage when both strands are hospital setting? broken at the same time. According to the LQ Photon beams are used for both the localization WAVES OF THE FUTURE OF WAVES model, the surviving fraction of cells Sfx after a and treatment of human tumours. Localization is 40 dose of radiation D is given by the process of defining the physical extent and location of the tumour and is performed using low Bremsstrahlung, or braking radiation, is the energy ‘diagnostic’ x-rays, typically 0.03 MeV phenomenon whereby a fast moving electron is mean energy, which give relatively high contrast suddenly decelerated by interaction with a heavy, between soft-tissue and bone. Low energy positively charged target nucleus (a coulombic diagnostic beams are ineffective for therapy of interaction). During deceleration the electron most cancers, however, as they have poor radiates away some of its energy in the form of a penetration (especially through bone) and a photon. In general, faster moving electrons significant fraction of radiation scatters out of the radiate a greater fraction of their energy as treatment region. Radiotherapy therefore uses bremsstrahlung photons and deposit less heat to high energy photon and electron beams, typically the heavy target atoms. In a low-energy x-ray with a mean energy of 2 to 10 MeV. The problem generator (Figure 2) electrons emitted from a of generating radiation therefore falls into two heated filament via thermionic emission are divisions: the production of low and high energy accelerated across an evacuated tube to strike a x-rays. Interestingly, although the two divisions tungsten target where the bremsstrahlung differ dramatically in technical implementation, interaction occurs. Notice how the radiation they both employ the same physical mechanism emerges almost at right angles to the direction of of radiation generation: the bremsstrahlung the incident electrons. interaction. The Linear Accelerator: at the heart of radiation treatment! The low energy x-ray tube of figure 2 is only required to accelerate electrons to a few tenths of an MeV, which can be achieved with this simple design. Therapy machines are required to accelerate electrons up to 25 MeV, and this requires very high electromagnetic (EM) fields Figure 2: Low-energy diagnostic and superficial therapy and a substantial number of high technology x-ray tube. components.

Figure 3: The linear accelerator. (a) Treatment-room view of a travelling wave accelerator; (b) Schematic diagram of a standing wave accelerator. THE TREATMENT OF CANCER USING IONISING RADIATION

41 The Linear Accelerator is a technological wonder which become progressively more distinct as they at the heart of the radiotherapy department, and travel along the drift tube. The second resonant generates both the high-energy photon and ‘catcher’ microwave cavity has intense electric electron beams used in cancer therapy. It is fields induced on it by the electron bunches, which usually located in a thick concrete underground are consequently decelerated. Energy is thus bunker to minimize exposure to hospital staff and transferred from the kinetic energy of the electron the general public. A treatment-room view and a bunches to EM microwave power, which then exits schematic design are shown in Figure 3. Echoes the klystron and is transported to the main of the basic x-ray circuit are seen in the electron accelerator waveguide. gun (corresponding to the filament), the main accelerator waveguide (where electrons are accelerated, necessarily longer than in an x-ray Treatment with photon beams tube) and the high-Z (high atomic number) Photons are regarded as indirectly-ionizing tungsten target. Intense EM fields generated by a radiation – that is to say, photon interactions in complex resonant microwave cavity combination tissue generate fast moving electrons that called a klystron accelerate electrons almost to propagate through tissue, directly damaging cells the speed of light along the main via multiple ionizations along the waveguide of the accelerator. A electron’s track. These variety of beam modifiers are interactions are predominantly placed in the beam to adapt it for The Linear Accelerator is with orbital electrons of atoms clinical use. The most important a technological wonder ... and molecules in the tissue. modifiers are the flattening filter, Photons have greater penetrability which achieves a flat, uniform and generates both the into tissue than electrons (see beam, and the multi-leaf high-energy photon and Figure 6) and therefore create collimator, which shapes the significant electron fluence at cross section of the beam to electron beams used in depth, without giving excessive match the projection of the cancer therapy. dose to the intervening healthy tumour. tissue. Although photons penetrate tissue quite well, they The klystron still deposit a significant dose A cross-sectional drawing of an elementary two- superficially (that is, at the surface). To maintain the cavity klystron is shown in Figure 4. The klystron is surface dose below acceptable levels several able to massively amplify low-power input photon beams are generally used in crossfire microwaves using two coupled resonant microwave techniques. All the beams in the treatment plan cavities. Thermionic emission from a heated contribute dose to the tumour, at the point of cathode introduces electrons into the first cavity, crossfire, but surface tissue generally only receives where they are ‘bunched’ together by the low- dose from one or two beams. power input microwaves. The first cavity also exerts ‘velocity modulation’ on the electron bunches, Physical interactions There are three main physical interaction mechanisms by which a therapeutic energy photon can interact with human tissue: the photoelectric effect, Compton scattering and pair production. The relative probabilities of each interaction are simple functions of the energy of the incident photon and the atomic mass number of the target atom. More complicated relations connect interaction probability to parameters like the scattering angle and energy distribution WAVES OF THE FUTURE OF WAVES amongst particles. Over the therapeutic range of 42 Figure 4: The klystron. energies considered here, Compton scattering is generally the dominant mechanism, although A typical depth-dose curve for a clinical photon photoelectric absorption becomes important at beam is illustrated in figure 6. lower energies (< 100 keV) and pair production becomes important at high energies (> 5 MeV).

A schematic diagram of the Compton scattering interaction is shown in Figure 5. An incident photon scatters off an outer orbital electron with reduced energy; the electron leaves the atom carrying off the energy difference between the incident and scattered photons. In the photoelectric effect, the incident photon interacts with an inner bound orbital electron. The incident photon is completely absorbed in the interaction Figure 6: Photon depth-dose curves. (there is no scattered photon) and the electron leaves the atom carrying off the difference energy, accounting for the binding energy of the Clinical examples of photon beam electron to the atom. treatment Breast cancer Breast cancer is the most common malignant disease of women in the western world – the lifetime probability of a woman developing breast cancer is estimated to be about 11%. Breast cancer is a particularly dangerous disease because of its tendency to break through basement membranes of small tissue-ducts, leading to metastatic spread through the Figure 5: Schematic diagram of inelastic Compton scattering. An outer orbital ‘Compton’ electron is ejected lymphatic system. Typical treatment will involve from the atom by a high-energy incident photon, which lumpectomy of the gross tumour followed by loses energy in the interaction and is scattered. post-operative irradiation of the entire breast tissue. The entire breast is treated because of the In pair production, the incoming photon is once microscopic invasive potential of breast cancer to again completely absorbed, this time in an the surrounding tissue to uniform dose. interaction with the field of the nucleus, resulting in the production of an electron-positron pair. The A typical photon beam treatment arrangement incident photon therefore has to have energy and dose distribution are illustrated in Figure 7. greater than the rest masses of the electron and Two opposed tangential 6 MV radiation beams positron. The positron causes ionization similar to deliver a near-uniform dose to the breast and an electron until it is brought to rest, when it lumpectomy cavity. Dose uniformity is enhanced annihilates with an electron producing a by placing metal wedge filters in the beam, which characteristic back-to-back 511 keV photon pair. reduce the intensity of radiation progressively

Figure 7: (a) Transaxial view through a CT scan illustrating a typical radiation treatment for breast cancer. Two opposed wedged tangential 6 MV radiation beams (only one shown) deliver a near-uniform dose to the entire breast. Isodoses illustrate dose uniformity across the breast. (b) The radiation fields are shaped to THE TREATMENT OF CANCER USING IONISING RADIATION minimize dose to lung and heart tissue. 43 towards the thick end of the wedge. The wedge distribution to the shape of the prostate and to compensates for the ‘missing tissue’ towards the avoid excessive damage to these critical organs. apex of the breast. Although both beams are wedged, only one is shown in Figure 7. A typical An example of a four-field prostate treatment is radiation treatment course is a dose of 180 cGy shown in Figure 8. The treatment plan consists of per day, five days a week, for a total dose of 45 two pairs of beams: an anterior-posterior pair and Gy (the Gray, or Gy, is the SI unit of radiation a right-left lateral pair. The beams crossfire on dose, equal to 1 joule of energy deposited in 1 kg the prostate to give a uniform therapeutic dose. of tissue or other material). A boost dose to Often higher energy photon beams (> 18 MeV) microscopic tumour residue around the surgical are used to take advantage of their greater scar is sometimes given with an electron beam. penetrability to the prostate, which can lie at Recent advances in breast radiotherapy (see depths of 18 cm or more. In a typical fractionated Kestin et al 2000) use intensity-modulated course of radiotherapy the patient will receive a radiation therapy (IMRT), which leads to improved daily dose of 180 cGy, five days a week, for eight dose homogeneity and better cosmetic outcome. weeks. Areas of current research interest include incorporating organ motion into treatment Prostate cancer planning (see Yan and Lockman 2001, Yan et al Prostate cancer is the most common malignant 2000), and online image-guided therapy, where disease of men in the US and Europe, with a anatomical information is obtained at the time of lifetime probability of development of about 12%. treatment (see Jaffray and Siewerdsen 2000). Radiotherapy has been found to be as effective as radical prostatectomy for tumours limited to the prostate, and has considerably less toxicity. Treatment with electron beams The prostate is located close to the rectum and Electron beams are obtained from the linear bladder, and sophisticated treatment planning accelerator by simply moving the tungsten target techniques are employed to shape the dose out of the path of the accelerated electrons

Figure 8: Typical four-field radiation treatment for cancer of the prostate. Isodose lines show a uniform dose delivered to the whole prostate organ. The dose-limiting healthy structures are the rectum, bladder and femoral heads. WAVES OF THE FUTURE OF WAVES

44 travelling along the waveguide of the accelerator. when a patient is irradiated with a photon or an The electrons exit the head of the treatment electron beam. The patient, however, does not machine and are incident on the patient after feel anything at all at the time of treatment. All passing through a scattering foil which widens the millions of interactions that take place in the beam and increases uniformity. Electron tissue are completely undetectable to the human beams are useful for treating superficial lesions nervous system at the time of treatment. Patients within 6 cm of the skin surface. The depth-dose only feel the effects of their radiation treatment curve illustrated in Figure 6 shows that, unlike when significant numbers of targeted cells have photons, the dose decreases rapidly with depth been sterilized and removed by the immune after a maximum value. Both the depth of the system. This time interval depends on many maximum and the gradient of dose fall-off vary issues, like cell-cycle time and the structure and with the incident energy of the electrons. sensitivity of the irradiated tissue.

Electrons lose energy in tissue through coulombic Patients undergoing palliative treatment, when interactions with atomic electrons and atomic the aim is to improve quality of life rather than to nuclei (bremsstrahlung). In low-Z materials (e.g. cure the patient, often receive pain relief within a tissue) energy loss is predominantly by inelastic few days. Negative symptoms from radiation ionizing events with atomic electrons. The mean treatment normally occur after 2-3 weeks if at all. rate of energy loss of electrons in tissue is about Side effects that may occur are red and tender 2 MeVcm-1. This means that a 10 MeV beam will patches of skin, mild nausea and vomiting for have a maximum range of 5 cm. The dose abdominal treatments, diarrhoea and rectal distribution from electron beams can be difficult discomfort for pelvic irradiation. to predict, especially for small fields, due to electron scattering and build-up effects. Electrons The medical physicist and the radiation oncology ‘backscatter’ at an interface to a high-Z material treatment team (e.g. metal dental implant or hip prosthesis), The patient interacts with a highly trained team of causing a significant increase in dose upstream specialists who form the radiation oncology team. from the interface. In low-density regions like the The primary roles in the team are the physician who lungs, electrons can travel three times further determines the nature of the treatment, the physicist than in normal tissue. In such cases it may be who advises on technical aspects, the dosimetrist necessary to modify the angle of incidence of the who formulates the computer treatment plan, the treatment beam or the treatment dose to therapist who delivers the treatment to the patient, minimize the dose to healthy tissue. and the nurse who is the primary care-giver.

A clinical example of the application of an The medical physicist makes critically important electron beam would be to boost the dose to the contributions at many stages of the treatment surgical scar after lumpectomy of the breast. The process. The physicist is responsible for correct surgical scar may be contaminated with sub- functioning of all aspects of radiation equipment, clinical tumour deposits during the surgical and for the purchasing, clinical acceptance and procedure. Electrons are also often used to boost commissioning of all new equipment. Physicists the dose to superficial lymph nodes after a are responsible for analysing treatment efficacy predetermined photon dose has been delivered. and for developing and implementing This ‘mixed-beam’ approach achieves dose at improvements, whether from technological depth with the photon component, with a boosted advances or new possibilities in treatment superficial dose from the electron component. technique. An example is gel dosimetry, a new technique for obtaining high resolution 3D images of complex dose distributions. What’s it like to have radiation treatment? Measuring three-dimensional absorbed dose in THE TREATMENT OF CANCER USING IONISING RADIATION We have seen that a tremendous amount of radiotherapy using gel dosimetry activity occurs at the atomic and subatomic level As long ago as the 1950s, radiation-induced 45 colour change in dyes was used to investigate spectroscopy (Rintoul 2003) and ultrasound doses in radiation sensitive gels (Day 1950, (Mather 2003). Numerous clinical applications of Andrews 1957). Subsequently, nuclear magnetic these radiologically tissue equivalent (Keall 1999) resonance (NMR) relaxation properties of gel dosimeters have been reported in the irradiated gels infused with conventional Fricke scientific literature and the international DOSGEL dosimetry solutions were measured (Fricke 1927, conference series on radiotherapy gel dosimetry Gore 1984). In Fricke gels, Fe2+ ions in ferrous (DOSGEL 1999, 2001, 2004). sulphate solutions are usually dispersed throughout a gel matrix. Fe2+ ions are converted to Fe3+ ions with a corresponding change in In conclusion paramagnetic properties that may be quantified The role of the medical physicists is often a very using NMR relaxation measurements (Gore 1984) interesting and challenging one, given the rapid or optical measurement techniques development of computing and technical (Appleyby 1991). hardware and software. The physicist also plays a critical role as the department trouble-shooter, Due to predominantly diffusion-related limitations available at immediate notice to solve problems (Baldock 2001), alternative polymer gel as they arise at treatment time and during the dosimeters were subsequently suggested treatment planning process. Such a role often (Maryanski 1993, 1994). In these polymer gels, demands quick thinking in a stressful situation. now commonly known as BANG-type (Maryanski The role of the medical physicist demands high 1994) or PAG-type (Baldock 1998), monomers professional standards; at the same time it can are usually dispersed in an aqueous gel matrix. be an exciting and rewarding profession, which The monomers undergo a polymerisation reaction makes a real difference to the quality of life of as a function of absorbed dose resulting in a 3D patients treated in the Radiotherapy department. polymer gel matrix. The radiation-induced formation of polymer influences NMR relaxation References properties and results in other physical changes Andrews HL, Murphy RE and LeBrun EJ 1957. Gel dosimeter that may be used to quantify absorbed radiation for depth dose measurements. Review of Scientific dose with the potential for true 3D dosimetry (De Instruments 28 329-332. Deene 1998, 2000). Appleby A and Leghrouz A 1991. Imaging of radiation dose by As the polymerization is inhibited by oxygen visible color development in ferrous-agarose-xylenol orange (Maryanski 1994, Baldock 1998) in BANG-type or gels. Med. Phys. 18 309-312. PAG-type polymer gel dosimeters, all free oxygen has to be removed from the gels. For many years Audet C, Hilts M, Jirasek A and Duzenli C 2002. CT gel this was achieved by bubbling nitrogen through dosimetry technique: Comparison of a planned and the gel solutions and by filling the phantoms in a measured 3D stereotactic dose volume. J. Appl. Clin. Med. glove box that is perfused with nitrogen. An Phys. 3 110-118. alternative polymer gel dosimeter formulation, known as MAGIC gel (Fong 2001) was proposed Baldock C, Burford RP, Billingham NC, Wagner GS, Patval S, in which oxygen is bound in a metallo-organic Badawi RD and Keevil SF 1998. Experimental procedure complex thus removing the problem of oxygen for the manufacture of polyacrylamide gel (PAG) for inhibition and enabling polymer gels to be magnetic resonance imaging (MRI) radiation dosimetry. manufactured on the bench-top of the laboratory Phys. Med. Biol. 43 695-702. with the potential for true 3D dosimetry (Gustavsson 2003). Baldock C, Harris PJ, Piercy AR and Healy B 2001. Experimental determination of the diffusion coefficient in As well as MRI (Vergote 2004), other quantitative two-dimensions in ferrous sulphate gels using the finite techniques for measuring dose distributions element method. Australasian Physical & Engineering WAVES OF THE FUTURE OF WAVES include optical (Oldham 2003) and x-ray Sciences in Medicine 24 19-30. 46 computer tomography (Audet 2002), vibrational Day MJ and Stein G 1950. Chemical effects of ionizing Keall PJ and Baldock C 1999. A theoretical study of the radiation in some gels. Nature 166 146-147 radiological properties and water equivalence of Fricke and polymer gels used for radiation dosimetry. Australasian De Deene Y, De Wagter C, Van Duyse B, Derycke S, De Neve Physical & Engineering Sciences in Medicine 22 85-91. W and Achten E 1998. Three-dimensional dosimetry using polymer gel and magnetic resonance imaging applied to Kestin L L, Sharpe M B, Frazier R C, Vicini F A, Yan D, Matter the verification of conformal radiation therapy in head-and- R C, Martinez A A and Wong J W 2000. Intensity neck cancer. Radiother. Oncol. 48 283-291. modulation to improve dose uniformity with tangential breast radiotherapy: initial clinical experience Int. J. Radiat. De Deene Y, De Wagter C, Van Duyse B, Derycke S, Oncol.Biol. Phys. 48 1559-68 Mersseman B, De Gersem W, Voet T, Achten E and De Neve W 2000. Validation of MR-based polymer gel Maryanski MJ, Gore JC, Kennan RP and Schulz RJ 1993. dosimetry as a preclinical three-dimensional verification NMR relaxation enhancement in gels polymerized and tool in conformal radiotherapy. Magnetic Resonance in cross-linked by ionizing radiation: a new approach to 3D Medicine 43 116-125. dosimetry by MRI. Magn. Reson. Imaging 11 253-258.

DOSGEL 1999. Proceedings of the 1st International Workshop Maryanski MJ, Schulz RJ, Ibbott GS, Gatenby JC, Xie J, Horton on Radiation Therapy Gel Dosimetry. (Canadian D and Gore JC 1994. Magnetic resonance imaging of Organization of Medical Physicists, Edmonton) radiation dose distributions using a polymer-gel dosimeter. Eds. LJ Schreiner and C Audet. Phys. Med. Biol. 39 1437-1455.

DOSGEL 2001. Proceedings of the 2nd International Mather ML and Baldock C 2003. Ultrasound tomography Conference on Radiotherapy Gel Dosimetry. (Queensland imaging of radiation dose distributions in polymer gel University of Technology, Brisbane, Australia) Eds. C dosimeters. Med. Phys. 30 2140-2148. Baldock and Y De Deene. Oldham M, Siewerdsen JH, Kumar S, Wong J and Jaffray DA DOSGEL 2004 Proceedings of the 3rd International 2003. Optical-CT gel-dosimetry I: Basic investigations. Conference on Radiation Therapy Gel Dosimetry eds De Med. Phys. 30 623-634. Deene, Y and Baldock, C (Ghent University, Ghent, Belgium) Oldham M, Siewerdsen J H, Shetty A and Jaffray D A 2001. Fong P M, Keil D C, Does M D and Gore J C 2001. Polymer High resolution gel-dosimetry by optical-CT or MR gels for magnetic resonance imaging of radiation dose scanning Med. Phys. 28 1436-45 distributions at normal room atmosphere. Phys. Med. Biol. 46 3105-3113. Rintoul L, Lepage M and Baldock C 2003. Radiation Dose Distribution in Polymer Gels by Raman Spectroscopy. Appl. Fricke H and Morse S 1927. The chemical action of roentgen Spectrosc. 57 51-57. rays on dilute ferrous sulfate solutions as a measure of radiation dose. Am. J. Roentgenol. Radium Therapy Nucl. Vergote K, De DeeneY, Duthoy W, De Gersem W, De Neve W, Med. 18 430-432. Achten E and De Wagter C 2004. Validation and application of polymer gel dosimetry for the dose verification of an Gore JC, Kang YS and Schulz RJ 1984. Measurement of intensity-modulated arc therapy (IMAT) treatment. Phys. radiation dose distributions by nuclear magnetic resonance Med. Biol. 49 287-305. (NMR) imaging. Phys. Med. Biol. 29 1189-1197. Yan D and Lockman D 2001. Organ/patient geometric Gustavsson H, Karlsson A, Back SAJ, Olsson LE, Haraldsson P, variation in external beam radiotherapy and its effects Med. Engstrom P and Nystrom H, 2003. MAGIC-type polymer gel Phys. 28 593-602 for three-dimensional dosimetry: Intensity-modulated radiation therapy verification. Med. Phys. 30 1264-1271. Yan D, Lockman D, Brabbins D, Tyburski L and Martinez A 2000. An off-line strategy for constructing a patient-specific THE TREATMENT OF CANCER USING IONISING RADIATION Jaffray D A and Siewerdsen J H 2000. Cone-beam computed planning target volume in adaptive treatment process for tomography with a flat-panel imager: initial performance prostate cancer Int. J. Radiat. Oncol. Biol. Phys. 48 289-302 47 characterization Med. Phys. 27 1311-23 2005: The International Year of ...

EACH YEAR THE United Nations declares that this is officially the International Year of Something Important. With the UN’s backing, an announcement like that draws the world’s attention to a cause, helps raise awareness and galvanises support and action.

For example, 2004 was the But most importantly (for this book, institutions, museums and other International Year of Rice – and at any rate) 2005 is the public organizations are running given the number of people who International Year of Physics! events to showcase the excitement depend on that grain for their daily and achievements of the physical sustenance and survival, it’s a worthy sciences, to get the collective mind recipient of its own year. 2003 was working and to honour the legacy the Year of Fresh Water, for similar of Einstein. To see what events are reasons. Ecotourism got its year in happening near you, visit 2002; that was also the www.wyp2005.org. International Year of Mountains, and the International Year of Cultural Throughout this book, tucked Heritage – 2002 was a big year between the chapters, you will find indeed. And levering the symbolism short articles about Einstein’s of a new millennium, the United discoveries of a hundred years ago. Nations declared 2001 to be the We hope you enjoy them, and that International Year of Mobilization Why physics? What makes this you make the most of this, the against Racism, Racial year more physics-y than any International Year of Physics. Discrimination, Xenophobia and other? This year we’re celebrating Related Intolerance (or IYoMRRDXRI one hundred years since a young for short). Swiss patent clerk wrote a series of research papers that changed the So what has the UN decided way we view the world around us. for 2005? His insights into the machinery of This is the year of Microcredit the physical universe changed how (www.yearofmicrocredit.org) we understand light, and matter, focussing on the practice of and energy, and even space and delivering small loans to needy time – all in this one incredible year individuals or groups. of 1905.

It’s also the International Year of A century later scientists around Sport and Physical Education the globe got together and (www.un.org/sport2005), aiming to convinced the UN to declare 2005 promote the value of sport to health the International Year of Physics

WAVES OF THE FUTURE OF WAVES and culture worldwide. (IYOP) in honour of that great man – Albert Einstein. All over the world, 48 groups of scientists, educational 01 nentoa ero oiiainaantRcs,Rca iciiain Xenophobia RacialDiscrimination, International Year ofMobilizationagainstRacism, 2001: International Year ofEcotourism International Year ofFreshwater 2002: International Year andits toCommemoratetheStruggleagainstSlavery Abolition 2003: 2004: International Year ofPhysics International Year ofDesertsandDesertification 2005: 2006: Official UnitedNationsInternationalYears 91 International Healthand MedicalResearch Year World Refugee Year InternationalCooperation Year 1959/1960: InternationalTourist Year 1961: International Year forHumanRights 1965: InternationalEducation Year 1967: International Year for Action toCombatRacism andRacialDiscrimination 1968: World Population Year 1970: InternationalWomen’s Year 1971: 1974: International Year oftheChild 1975: InternationalAnti-ApartheidYear International Year forDisabledPersons 1978/1979: International Year ofMobilizationforSanctionsagainstSouth Africa 1979: World Communications Year 1981: International Youth Year 1982: Year oftheUnitedNations 1983: International Year ofPeace 1985: International Year ofShelterfortheHomeless 1985: International Literacy Year 1986: International Space Year 1987: International Year ofthe World’s IndigenousPeople 1990: 1992: International Year ofSportandtheOlympicIdeal 1993: World Year ofPeoples’Commemoration Victims oftheSecond World War 1994: International Year fortheEradicationofPoverty International Year oftheOcean 1995: International Year ofOlderPersons 1996: 1998: InternationalYearof Thanksgiving 1999: 2000: United Nations Year forCulturalHeritage International Year ofMountains International Year ofRice International Year ofMicrocredit International Year forSportandPhysicalEducation International Year oftheFamily Nations Yearfor ToleranceUnited International Year fortheCultureofPeace International Yearof Volunteers United Nations Year ofDialogueamongCivilizations and RelatedIntolerance 49 2005: THE INTERNATIONAL YEAR OF ... SIMON CARLILE did his undergraduate and graduate training at The University of Sydney. His PhD thesis work examined the bioacoustic and physiological basis of the representation of auditory space in the mammalian auditory nervous system. He spent five years in Oxford as a postdoctoral fellow and then a Beit Memorial Fellow and Junior Research Fellow of Green College Oxford. During this time he worked in the multidisciplinary sensory neuroscience group led by Colin Blakemore. In 1993 he moved back to The University of Sydney as a Lecturer in Neuroscience in the Department of Physiology and has established the Auditory Neuroscience Laboratory. The Laboratory has a broad focus with current work ranging from the bioacoustics of outer ear, the psychophysics of real and virtual auditory space as well as the neurophysiological mechanisms that result in neural representations of auditory space. He also has interests in the applications of information technology to medical education and has lectured and tutored in History and Philosophy of Science. The psychophysics of real and virtual auditory spaces

Simon Carlile

Introduction IN THIS CHAPTER we will be considering how we perceive the space around us using our sense of hearing. The environments in which we normally live are full of many different sounds, often occurring at the same time. Some of these sounds are of interest to us, but many are not. Take for example a noisy club or party where you are concentrating on what your friend is saying (the foreground sound of interest) and trying to ignore the concurrent background music and conversations (the background noise). This issue has been called the ‘cocktail party problem’ by E. C. Cherry in the 1950s (Cherry 1953, see Figure 1) – his ideas about how the nervous systems solves this problem have been very influential in guiding our development of communications devices such as telephones as well as more advanced virtual reality displays. The way in which our auditory system sorts out the different information from each of these concurrent sources using only two ears is a magnificent feat of information processing. As we shall see, our ability to determine where the different sources are located in space plays an important role in this process.

51 multiple different sound sources but also the passive acoustics of the environment such as reverberation. Our perception of auditory space can be described along a number of different dimensions. Three of the principal dimensions are direction, distance and spaciousness. While we can objectively characterise direction and distance using the standard three dimensions (also known as a Euclidian description), this does not necessarily mean that our perception of auditory space maps onto this coordinate system.

For instance, spatial coordinates could be described using spherical coordinates that indicate the direction to a sound source and its distance from the listener. However, in our qualitative description of a sound’s location in everyday life we generally talk about the horizontal direction and distance, and the height Figure 1: The cocktail party problem – how does this above or below the audio-visual horizon. This person manage to hear the conversation next to him description implies that we use a more cylindrical through all the background noise? view of space when talking about sound. In addition to the perception of location (direction and distance), the extent or ‘spaciousness’ of the efore proceeding, it is probably worth space inhabited by the listener is also a powerful spending a little time exploring the meaning element of perception. Bof the title of this chapter, as the terms are not ones used in everyday conversation. Firstly, Thirdly, we need also to consider what is meant ‘psychophysics’ is the study of human perception by ‘real’ and ‘virtual’ spaces. A real space is the that is concerned with establishing quantitative sort of space where you find yourself now. The relations between physical sounds are coming from stimulation and perceptual individual sources that occupy events – in other words, “In general, properly different locations in the world attaching sounds we hear to around you. They may be moving events we see or feel. matching the display or static, and the passive Psychophysics has a quite long and the sensory acoustic environment (the history as a science and arose in objects and spaces surrounding the early part of the twentieth system requires an you) may also be shaping and century. In the context of this understanding of the modifying the sounds that reach chapter, we will be examining your ears. By contrast, in a the performance of human perceptual limits of the virtual auditory space the sounds subjects undergoing different auditory system.” usually come from a limited tests of hearing. The results of range of sources – such as a such experiments allow pair of headphones. However, inferences about what the the sounds are processed in a relevant physical stimuli are and how they are way that leads you to perceive different sources processed by the auditory system. at different locations in particular acoustic environments. This is very different to the normal Secondly, the term ‘auditory space’ refers to the experience of listening to sounds over WAVES OF THE FUTURE OF WAVES acoustic environment that surrounds the listener headphones. 52 at any particular time. This will not only include (a)

For instance, when you play music or speech Figure 2: (a) Sound from stereo speakers seems to come over headphones, the sound image is heard from somewhere outside your head, typically between the inside or very close to the head, rather than from speakers; (b) in contrast, sound from a pair of headphones seems to originate from somewhere inside or very close to a location away from the head, as is the case your head. when you are listening to the same or speech over stereo speakers in your lounge room (see (b) Figure 2). We will see later just how we can process the sound to generate the illusion of sounds outside the head in ‘virtual auditory space’. In this context the perception of spaciousness plays an important role in the generation of the sense of ‘presence’ enjoyed by the listener (see Durlach 1991) – that is, the feeling of actually being in a virtual environment can be generated using auditory and visual displays.

One of the interests of some of the researchers whose work I will describe in this chapter is how perception can be degraded – in the case of to implement virtual auditory displays that match generating virtual auditory space, listeners would the capacity of the human listener. From a not get a compelling or accurate illusion of an theoretical point of view, the fidelity of any external world when listening over headphones. reproduction is a key issue in the generation of On the other hand, where fidelity of the the representation of an object. In a practical representation is higher than that of the sensory sense, the fidelity of reproduction is often system, then the effort of producing such a high determined by the use to which the reproduction fidelity reproduction is wasted. In the context of is to be put. auditory virtual displays, significant computing power is needed to generate these displays so For instance, where the objective is to produce a ‘over-engineering’ the fidelity places a serious representation that is as close as possible to a overhead on performance. perfect copy of the original object, then, from a perceptual point of view, the limiting factor is the In general, properly matching the display and the fidelity of the sensory system encoding the

sensory system requires an understanding of the THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES representation. On the one hand, when the fidelity perceptual limits of the auditory system. One way of reproduction is low the quality of the resulting in which this can be determined is by making 53 Figure 3: The interaction of the head and ears with a sound field. For a sound located away from the mid-line plane (dashed line) there is a path length difference between the sound source and each ear which gives rise to a difference in the time of arrival of the sound at each ear. Likewise, the head is an effective obstacle for the sound field and shadows the ear farthest from the source producing a difference in the level at each ear.

making objective measures of the user’s psycho- sound’s location is called the binaural cues. physical performance on tasks in real space These arise as a consequence of the ears being compared with virtual space – an approach that separated by the acoustically dense head and we will examine in more detail below. the fact that each ear is located at slightly different locations in space. In the next sections we will consider the physical acoustic properties of a sound that the auditory What this means is that each ear simultaneously system can use as cues to the location and samples the sound field under slightly different nature of the sound. We will also consider how conditions (Figure 3). For a sound located off the well we can locate sounds in space and the role midline, there is a difference in the length of the that the differences in the locations of the sound path from the sound source to each ear, which sources plays in sorting out and understanding results in a difference in the arrival time of the speech in acoustically complex or ‘cluttered’ sound at each ear. This difference is called the environments. We will then look at how the interaural time difference (or ITD) and is sounds can be processed before they are dependent on the horizontal location of the presented over headphones so that they give rise source of the sound with respect to the head. For to the illusion of virtual auditory space. In the final a sound located directly ahead there is no section we will look at a few of the more difference in the path lengths from the source to interesting applications of virtual auditory space. each ear. By contrast, when the sound source is located opposite one ear then the interaural difference is at a maximum. As sound travels at Physical cues to the perception of around 330 m/s, this difference is generally very auditory space small – the maximum difference is less than 1 ms. The acoustic cues used by the auditory system in generating our perceptions of space are based However the auditory system is able to extract on the interactions of the sound with the two this information from the sound encoded at each ears, the head and torso as well as with the ear and compute the horizontal position of the WAVES OF THE FUTURE OF WAVES reflecting surfaces in the immediate sound. In addition, as the head is relatively large environment. A very powerful set of cues to a 54 with respect to the wavelengths of the sound, lower frequency range where the wavelengths are long compared to the dimensions of the outer ear.

A critical feature of these spectral cues is that when considered across the whole range of frequency sensitivity of the listener, the filtering is unique for each possible location in space (Figure 5). Of course, a limitation of this cue is that for it to be unambiguous there needs to be a wide range of frequencies in the source sound and that the original spectrum of the source is also known by the auditory system. Figure 4: The outer ear has a very convoluted shape that helps us to Fortunately, this is the case for sharp transient locate sounds. (Thanks sounds, which contain a very wide range of to Natalie for posing for frequencies and are generally spectrally flat – the photograph.) that is, there is a reasonable, even distribution of energy across the wide frequency range of the sound. Given that there is probably significant the ear furthest from the source will be evolutionary pressure to locate the transient acoustically shadowed giving rise to an interaural sounds of approach of a predator (say, the snap level difference in the level of the sound in of a breaking twig) then encoding and processing each ear. this type of cue is likely to have played an important role in the survival of the species. On While these binaural cues to sound location are the other hand, sounds that have a more limited powerful, it is also well known that these cues spectral range are not going to provide much of a are ambiguous. That is, because of the symmetry monaural or spectral cue and indeed will not then of the placement of the ears on the side of the help in resolving the ambiguities in the binaural head, any particular interaural interval will specify cues. As such it is no surprise that spectrally the surface of a cone centred on the interaural limited sounds are very hard to localise and, in axis – the so called ‘cones of confusion’. For fact, many animals produce warning calls with a instance, a sound in front of the listener will very narrow spectrum to sound an alert while generate zero differences in the binaural cues but avoiding detection! then so will a source located directly above or directly behind the listener. Likewise, a sound at A range of physiological studies have 45° to the left of the midline will generate the demonstrated that neural representations of same interaural intervals as the same source auditory space in the mammalian midbrain are located behind the listener at 135° to the left of dependent on the integration of these binaural the midline. As we don’t usually confuse the and monaural cues (King and Carlile 1994). This location of a sound in front with one behind, the convergence and integration also appears to lead auditory system must have some other way of to a form of spatial channel processing that may resolving the ambiguities in the binaural cues. underlie the perception of auditory space (Carlile, Hyams et al. 2001) although a more complete This third set of cues, in addition to the two discussion of this very interesting processing by binaural cues, is produced by the asymmetrical the nervous system is beyond the scope of and highly convoluted shape of the outer ear this chapter. (Figure 4). This complex physical structure gives rise to a location-dependent filtering of the sounds and produces the so called spectral or monaural Cues for Direction THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES cues to location. Reflections from the shoulder and Coding of the direction of the source of a sound torso may also contribute to the filtering of the arises principally as a result of the interaction of 55 Figure 5: The variation in filter functions of the outer ear are shown as a function of the location of the source on the audiovisual horizon where 0° is directly ahead and 90° is on the interaural axis opposite the ear. The gain of the filter functions is indicated by the contour colour. The filter function for each location is unique. Note in particular that the filter functions for the frontal portion of space (0 - 90°) are very different for the back portion of space.

the sound with the auditory periphery. By both of these cues, there is a confounding of contrast, the coding of the distance of the sound source characteristics (intensity and spectrum) source is dependent on the detection of the with distance. That is, the overall level of the interactions of the sound with objects in the sound at the ears or the level of the high listening environment. Four acoustic cues to frequency content will only provide a reliable cue distance have been identified for stationary if the level at the source is known. Consequently, sources (Mershon and King 1975). Under ideal these cues may only be reliable for particular conditions the intensity of a sound decreases with classes of familiar sounds. Interestingly, we are distance according to the inverse square – this generally very good a judging the distance of leads to a 6 dB loss with a doubling of distance. someone speaking – probably because if we However, in practice this relationship is substantially decrease (whisper) or increase dependent on the reflective characteristics of the (scream) the level of the voice the spectrum of environment. the sound also changes.

A second and similar cue for distance results A third cue to distance is the ratio of the direct to from the transmission characteristics of the air – reverberant energy – that is, the proportion of the that is, high frequencies (> 4 kHz) are absorbed, energy reaching the listener directly decreases WAVES OF THE FUTURE OF WAVES leading to a reduction of around 1.6 dB per with the distance of the source from the listener 56 doubling of distance (Zahorik 2002). Notably for (i.e. is subject to the inverse square law of distance) while the level of reverberation in a cognitive factors can depend on expectations that room is determined principally by the the listener may have about the nature of the characteristics of the room. This is a particularly sounds or the listening environment. Indeed, powerful cue for distance but obviously is recent listening experience in a particular space dependent on the reverberant characteristics of can also subsequently affect the perception of the listening environment. sound in that space. A most straightforward example is the well-known ventriloquist effect For sound sources close to the head (say within a where the source location of a talker is captured 1 metre radius), recent work has confirmed the by the image of the talker (as in say the cinema observations in the 1920s (Hartley and Fry 1921) or TV). that substantial variation in the interaural level differences can occur with variation in the distance of the source from the head (see for The fidelity of the perception of instance Shinn-Cunningham 2000). As the source auditory space gets closer to the head, the wave front becomes In this section we will consider how well we are more spherical and will bend around the head able to determine where a sound is located in and interact at the far ear – this is very different space. There are a large number of from the case when the wave front is effectively psychophysical studies of the resolution and parallel, as occurs for a source some distance accuracy of the auditory system relating to from the head. In contrast to the effects on level, auditory space (for reviews see Middlebrooks and the distance effect on interaural time difference Green 1991; Carlile 1996). The resolution of the appears to be much less salient (Brungart and auditory system has been examined by Rabinowitz 1999). This is what would be measuring how well the auditory system can predicted acoustically, given that the path lengths resolve differences in the locations of a from the source to each ear would not vary sequentially presented sound source (Mills 1958). substantially as a function of the distance from the centre of the head. This is referred to as a minimum audible angle (MAA) detection task and provides information about the just-noticeable differences in the cues Cues for Spaciousness to a sound’s location in space. MAA studies have The final cues considered here are those that result demonstrated that the resolution of the auditory in our perception of spaciousness (see for example system is highly dependent on the range of Blauert and Lindermann 1986, Potter, Raatgever et frequency in the stimulus and the absolute spatial al. 1995). This has been characterized by (i) location about which the change in location is ‘apparent source width’, which is related to early being determined (see Grantham 1995). Subjects lateral reflections and the level of low frequency demonstrate the smallest MAA for sounds with a sound, and (ii) ‘listener envelopment’, which is broad range of frequencies (1-2° for sound related to the reverberant sound field, particularly locations directly in front of the listener) with for sounds arriving more than 80 ms after the significant increases in the MAA for narrow band direct sound (see Okano, Beranek et al. 1998). stimuli and for locations well off to the side. Interestingly, it is these sorts of reflection patterns that architects have sought in designing the best More recent work has also examined the ability of opera and concert halls in the world, as the subjects to discriminate concurrent sounds as qualities of apparent width and envelopment originating from different locations (Best, Schaik contribute significantly to the appreciation of et al. 2004). This study demonstrates that the musical performance in these spaces. ability to separate out two concurrent broadband stimuli is dependent on the magnitude of the In addition to the range of purely acoustic cues differences in the interaural cues generated by discussed above it is also important to note that a the two stimuli rather than differences in the THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES range of other factors can play a role in the spectral cues generated by the filtering of the perception of space. Such so-called top-down or outer ear. 57 The accuracy in determining the absolute location of the trials result in ‘front-back’ errors. Of note, of a sound source has been assessed by allowing localisation performance is strongly related to the subjects to indicate by pointing or calling out characteristics of the stimulus; in particular, the coordinates of the perceived location of a sound performance levels noted above are for short source (e.g. Makous and Middlebrooks 1990; duration (150 ms) broadband stimuli presented Carlile, Leong et al. 1997). Absolute localisation under anechoic conditions. Where sounds are of accuracy has generally been determined under a longer duration then performance can improve anechoic conditions – that is, in special environ- as the auditory system has the opportunity to ments where there are little to no echoes or take multiple samples of the sound source and reverberation that are known to degrade this ability. hence the cues to the location are more reliable. However, with narrowband stimuli or less controlled listening conditions localisation performance can degrade substantially.

The perception of distance has been studied much less exhaustively than the perception of the relative direction of a sound source. In general, listeners tend to underestimate the source distance for far distances and overestimate for near distances, with a cross over point around Figure 6: Pooled localisation responses from 19 subjects 1.5 metres. This is called the specific distance shown for front and back hemispheres of space. The actual target locations are shown by the small ‘*’ and the mean of tendency (Mershon and King 1975). Recent work the pooled localisation responses for each location is has indicated that the distance cues that shown by the small filled circle. The ellipse surrounding dominate in a particular distance judgment task each mean response indicates the standard deviation. vary according to the reliability of the set of distance cues. Depending on the spectral content In general, results from such studies indicate that of the sound source and the nature of the there are two broad classes of localisation errors: acoustic environment, different cues will be more (i) large so called ‘front-back’ confusions or ‘cone or less reliable indicators of distance (Zahorik of confusion’ errors where the perceived location 2002). It appears that the auditory system is in quadrant different from the actual sound dynamically weights its reliance on different cues source, and perceived to come from a source at depending on how well the different cues an angle reflected about the interaural aural axis; contribute to the overall assessment of the and (ii) local errors where the sound location is distance of the sound source. perceived to be located in the vicinity of the actual target. The average localisation errors for 19 In addition to the static cues to a sound’s location, subjects are illustrated on spherical plots in Figure there are also a range of dynamic cues to both 6 (Carlile, Leong et al. 1997). These plots illustrate static sound sources and moving sound sources. both the systematic errors in the localisation Dynamic cues to the location of static sources are (indicated by the difference between the actual produced by moving the head to allow more location and the average of the localisation samples of the sound field. This provides binaural estimates) and a dispersion of the responses information that can be used to resolve the about the average perceived location (indicated ambiguities in the binaural cues and can result in by the ellipses surrounding each average). significant reduction in the cone-of-confusion or front-back error (Wallach 1940; Lambert 1974). In general, localisation accuracy decreases for locations from the anterior to posterior midline However, some other findings indicate that head and as the elevation of the target deviates from movement (self-induced or otherwise) only leads the audio-visual horizon (see also Makous and to a reduction in local errors when the sound is WAVES OF THE FUTURE OF WAVES Middlebrooks 1990). When stimuli contain a relatively narrow band or the spectral cues to 58 broad range of frequencies only about 3% to 6% location are degraded in some way (see Pollak and Rose 1967, Fisher and Freedman 1968, and The generation of virtual auditory see also Wightman and Kistler 1999). The space availability of such cues of course requires that As discussed above, an understanding of the the sound of interest is of sufficient duration to fidelity of the perceptual encoding of auditory allow multiple sampling. space can be used to guide the development of auditory displays that exploit the spatial nature of The second class of dynamic cues are those the auditory perception. Quite simply, to create produced by moving stimuli. The minimum the illusion of external auditory space using audible movement angle (MAMA) has been headphones, the principal processing goal is to defined as the smallest angle a sound must travel reproduce at the ear drums the pattern of sound before its direction is correctly discriminated. The waves that would have occurred had the sound ability to detect the movement of a sound source sources actually been in the free field. All things is dependent on the same sorts of cues used in being equal, the auditory system should then be locating a stationary stimulus, with the advantage able to extract the necessary acoustic cues to that multiple sampling is possible with a relatively identify the sources and their relative locations long-duration stimuli. In addition there may also along with the characteristics of the auditory be a Doppler shift of the frequencies of the sound environment. dependent on the movement trajectory of the source with respect to the listener. The top panel in Figure 7 illustrates the situation when normally listening to, for example, a click It has been consistently found that MAMAs are sound over headphones. However, imagine that larger than the static MAA (generally 2 to 4 we have also inserted small microphones close to degrees of arc) using both noise bursts and tones the ear drums of the listener so we can record emitted from a moving speaker (Harris and the sound waves at the ear drums – the outputs Sergeant 1971). It has also been reported that of these microphones are shown on the right and increased source velocity results in a larger MAMA left of each panel. By varying the binaural cues of (Perrott and Musicant 1977; Grantham 1986). interaural time difference or the interaural level This apparent loss in spatial acuity with rapidly difference (that is, the timing or amplitude of the moving sources might indicate a minimum clicks presented to each headphone) the sound integration time required to perform these tasks. image heard by the listener is perceived to move However, there is much debate about the duration closer to the ear where the click arrives first of that integration time. Recent work exploiting and/or is loudest. virtual auditory space stimulation techniques (see below) has also indicated that human listeners are Despite the fact that this manipulation is varying relatively insensitive to changes in the velocity of the cues to sound location, the listener still hears moving sound sources and that absolute sensitivity the sound inside the head. The middle panel is velocity dependent (Carlile and Best 2002). illustrates the situation for a sounds presented from a single loudspeaker located away from the listener. Again we can see the differences in the interaural level and time cues, but in this case the click sound is filtered by the outer ears and the sound source is heard outside the head – so if we simply take the recordings of sounds at the ear made in the middle panel, and play them back over headphones, then, following any correction for distortions by the headphones, the sound waves at the eardrums should be the Figure 7: The differences between free field and headphone same as in the middle panel. Doing this, we find listening are illustrated in the top two panels. When the that the listener also hears the sound outside the filtering functions of the outer ear are properly accounted THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES for with headphone listening, the sound image is heard head and at the correct location in space. This is outside the head in virtual auditory space (bottom panel). the basis of Virtual Auditory Space (VAS). 59 Figure 8: The anechoic chamber at the University of Sydney’s Auditory Neuroscience Laboratory. The walls are specially designed to reflect as little sound as possible. The chamber is equipped with a robot arm carrying a small speaker that can be placed at almost any location on the surface of an imaginary sphere surrounding a test subject located in the middle of the chamber.

Of course, if a virtual environment is to be useful see Figure 8. A speaker, mounted on the robot we want to be able to present any sound at any arm, delivers the measurement stimuli location around the listener and not simply play automatically from up to 400 locations evenly back what has been previously recorded as in our distributed on an imaginary sphere surrounding description above. Lets begin by describing for the subject. The resulting HRTFs are stored the simplest scenario: a single source placed at a digitally and can then be used to filter any sound particular location with respect to the head of the stimuli before presentation to the left and right listener. As we have seen above, the sound from ears using high quality headphones. In many a particular source will be transformed or filtered cases the transfer function of the headphones to by the outer ears by the time it reaches the the microphones in the ears are also measured eardrum. Acoustically, this is be characterised as so that any frequency distortion due to the the head related transfer function (HRTF). As headphones can also be accounted for. mentioned above, there is a unique HRTF for each ear and for every direction in space The fidelity of such an auditory display is surrounding the listener (Carlile and dependent on how well the HRTFs used to Pralong 1994). generate the VAS match the actual HRTFs of the listener. However, the HRTFs are highly The HRTFs are measured in the laboratory by individualised because the filtering is dependent placing small recording microphones in an on the precise shape of the outer ear, and earplug secured flush with the outer end of the everybody’s ears are slightly different (Jin, Carlile ear canal for each ear (Møller, Sorensen et al. et al. 2000). Therefore, if we want to generate 1995, see also Pralong and Carlile 1994). The high fidelity VAS, the specific HRTFs of the WAVES OF THE FUTURE OF WAVES measurements are performed inside an anechoic listener need to be accounted for. 60 chamber with the subject placed at the centre – One way of doing this is to record the HRTFs for sense of up and down. Obviously, in low altitude each user of an auditory display and use these to situations this can produce high risks for the generate the listener’s own personalised VAS. Of pilot and the aircraft. One experimental course this is not practical for more general uses application of 3D audio has been to map the of this technology because the facilities and aircraft’s horizon indicator into an audio ‘icon’ expertise required to record the HRTFs are not presented over headphones to the pilot. When widespread. On the other hand, methods are the plane is flying level and upright then the pilot being developed for predicting the HRTFs of hears the ‘icon’ above the head of the pilot, but individual listeners by making physical if the plane has banked heavily to say the left measurements of the shapes of the outer ears then the ‘icon’ will appear to be to the right of and their placement on the head (e.g. Jin, Carlile the pilot – that is, the sound presented in VAS et al. 2000). indicates which way is up for the pilot.

Alternatively, a listener could select from a This is an example of a man-machine interface range of HRTF libraries and select those that which has mapped data from the gravitational best match their own by using some form of domain (usually signalled by the organs of performance task – indeed, this is the approach balance) into the auditory domain and that that is often taken in assessing the fidelity of the provides a very natural means for the pilot of VAS rendered by a particular display. In these determining which was is up. More widespread assessments, the ability of a subject to applications of this technology awaits the determine exactly where a sound is coming from implementation of binaural sound systems into is compared for sounds presented in real space the cockpits of high performance aircraft – with sounds presented in virtual space. In most something that is a target of the next generation studies where the recordings have been of aircraft being developed in North America and carefully controlled, performance is nearly Europe. identical under both conditions (Wightman and Kistler 1989; Bronkhorst 1995; Carlile 1996; More recently, there has been growing interest in Carlile, Leong et al. 1996; Langendijk and the use of 3D audio to enhance speech Bronkhorst 2000). discrimination in multi-talker communication systems to support teleconferencing and command and control activities. For instance, in Examples of the uses of virtual air traffic control the ground control personnel auditory space technologies and the pilots often have to monitor multiple As the popularity of so-called virtual environments conversations over their headphones. Anyone who has blossomed, the uses for 3D audio, as VAS is has participated in a multi-person teleconference also sometimes referred to, have become over the phone will know that it is very difficult to increasingly more widespread (see Shinn- follow individual conversations or to interact Cunningham and Kulkarni 1996; Shinn-Cunningham, naturally when people talk over each other in a Lehnert et al. 1997). In addition to the obvious situation where all of the talkers appear to be games and entertainment applications of so coming from one place (i.e. the telephone called ‘virtual realities’, some of the early speaker). experimental applications of 3D audio have been in navigational and collision avoidance systems However, if the individual talkers are spatialised by the air force (see for example Begault and so that they appear at different locations around Pittman 1994). the listener then the ability to distinguish the individual conversations is greatly enhanced. For instance, in a high performance aircraft Under normal listening conditions, we naturally cockpit, there are numerous visual displays that use the positions of different talkers to assist us need to be monitored almost simultaneously. In in focussing our attention on what is being said THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES high-G manoeuvres, vision can often be and to ignore other talkers or to shift our compromised along with the pilot’s gravitational attention from talker to talker. In fact such uses of 61 VAS technologies in multi-talker environments Carlile, S. and D. Pralong (1994). ‘The location-dependent help simulate normal communication nature of perceptually salient features of the human head- environments to allow us to naturally solve the related transfer function.’ J. Acoust. Soc. Am. 95(6): 3445- ‘cocktail party’ problem discussed at the start of 3459. this chapter. Cherry, E. C. (1953). ‘Some experiments on the recognition of References speech with one and two ears.’ J. Acoust. Soc. Am. 25: Begault, D. R. and M. T. Pittman (1994). 3-D audio versus 975-979. head down TCAS displays. San Jose, San Jose State University. Durlach, N. (1991). ‘Auditory localization in teleoperator and virtual environment systems: ideas, issues and problems.’ Best, V., A. v. Schaik, et al. (2004). ‘Separation of concurrent Perception 20: 543-554. broadband sounds sources by human listeners.’ J. Acoust. Soc. Am. 115: 324-337. Fisher, H. G. and S. J. Freedman (1968). ‘The role of the pinna in auditory localization.’ Journal of Auditory Research. 8: Blauert, J. and W. Lindermann (1986). ‘Auditory spaciousness: 15-26. Some further psychoacoustic analyses.’ J. Acoust. Soc. Am. 80(2): 533-542. Grantham, D. W. (1986). ‘Detection and discrimination of simulated motion of auditory targets in the horizotal plane.’ Bronkhorst, A. W. (1995). ‘Localization of real and virtual J. Acoust. Soc. Am. 79(6): 1939-1949. sound sources.’ J. Acoust. Soc. Am. 98(5(Pt 1): 25422553. Grantham, D. W. (1995). Spatial hearing and related Brungart, D. S. and W. M. Rabinowitz (1999). ‘Auditory phenomena. Handbook of Perception and Cognition. B. C. localisation of nearby sources. Head related transfer J. Moore. San Diego, Academic. functions.’ J. Acoust. Soc. Am. 106: 1465-1479. Harris, J. D. and R. L. Sergeant (1971). ‘Monaural/binaural Carlile, S. (1996). The physical and psychophysical basis of minimum audible angles for a moving sound source.’ J. sound localization. Virtual auditory space: Generation and Spch. Hear. Res. 14: 618-629. applications. S. Carlile. Austin, Landes: Ch 2. Hartley, R. V. L. and T. C. Fry (1921). ‘The binaural location of Carlile, S., Ed. (1996). Virtual auditory space: Generation and pure tones.’ Physics Rev. 18(6): 431-42. applications. Austin, Landes. Jin, C., S. Carlile, et al. (2000). Enabling individualised virtual Carlile, S. and V. Best (2002). ‘Discrimination of sound source auditory space using morphological measurements. Int velocity by human listeners.’ J. Acoust. Soc. Am. 111(2): Symp on Multimedia Info Processing, Sydney, Australia. 1026-1035. King, A. J. and S. Carlile (1994). Neural coding for auditory Carlile, S., S. Hyams, et al. (2001). ‘Systematic distortions of space. The Cognitive Neurosciences. M. S. Gazzaniga. auditory space perception following prolonged exposure to Boston, MIT Press: 279-293. broadband noise.’ J. Acoust. Soc. Am. 110: 416-425. Lambert, R. M. (1974). ‘Dynamic theory of sound-source Carlile, S., P. Leong, et al. (1997). ‘The nature and distribution localization.’ J. Acoust. Soc. Am. 56(1): 165-171. of errors in the localization of sounds by humans.’ Hearing Res. 114: 179-196. Langendijk, E. H. A. and A. W. Bronkhorst (2000). ‘Fidelity of three-dimensional-sound reproduction using a virtual Carlile, S., P. H. W. Leong, et al. (1996). High fidelity virtual auditory display.’ J. Acoust. Soc. Am. 107(1): 528-537. auditory space: an operational definition. SimTecT, Melbourne, Australia, DSTO. Makous, J. and J. C. Middlebrooks (1990). ‘Two-dimensional sound localization by human listeners.’ J. Acoust. Soc. Am. WAVES OF THE FUTURE OF WAVES 87(5): 2188-2200. 62 Mershon, D. H. and L. E. King (1975). ‘Intensity and Wallach, H. (1940). ‘The role of head movements and reverberation as factors in the auditory perception of vestibular and visual cues in sound localization.’ Journal of egocentric distance.’ Percept Psychophys 18(6): 409-415. Experimental Psychology 27(4): 339-368.

Middlebrooks, J. C. and D. M. Green (1991). ‘Sound Wightman, F. L. and D. J. Kistler (1989). ‘Headphone localization by human listeners.’ Annu. Rev. Psychol. 42: simulation of free field listening. II: Psychophysical 135-159. validation.’ J. Acoust. Soc. Am. 85(2): 868-878.

Mills, A. W. (1958). ‘On the minimum audible angle.’ J. Wightman, F. L. and D. J. Kistler (1999). ‘Resolution of front- Acoust. Soc. Am. 30(4): 237-246. back ambiguity in spatial hearing by listener and source movement.’ J. Acoust. Soc. Am. 105(5): 2841-53. Møller, H., M. F. Sorensen, et al. (1995). ‘Head-related transfer functions of human subjects.’ J. Audio Eng. Soc. 43(5): Zahorik, P. (2002). ‘Assessing auditory distance perception 300-321. using virtual acoustics.’ J. Acoust. Soc. Am. 111(4): 1832- 1846. Okano, T., L. L. Beranek, et al. (1998). ‘Relations among interaural cross-correlation coefficient (IACC(E), lateral fraction (LFE), and apparent source width (ASW) in concert halls.’ J. Acoust. Soc. of Am 104(1): 255-265.

Perrott, D. R. and A. D. Musicant (1977). ‘Minimum audible movement angle: Binaural localization moving sound sources.’ J. Acoust. Soc. Am. 62(6): 1463-1466.

Pollak, I. and M. Rose (1967). ‘Effect of head movement on the localization of sounds in the equatorial plane.’ Perception and Psychophysics 2(12): 591-596.

Potter, J. M., J. Raatgever, et al. (1995). ‘Measures for Spaciousness in Room Acoustics Based on a Binaural Strategy.’ Acta Acustica 3(5): 429-443.

Pralong, D. and S. Carlile (1994). ‘Measuring the human head-related transfer functions: A novel method for the construction and calibration of a miniature ‘in-ear’ recording system.’ J. Acoust. Soc. Am. 95(6): 3435-3444.

Shinn-Cunningham, B. and A. Kulkarni (1996). Recent developments in virtual auditory space. Virtual auditory space: Generation and applications. S. Carlile. Austin, Landes: Ch 6.

Shinn-Cunningham, B. G. (2000). Distance cues for virtual auditory space. IEEE 2000 International Symposium on Multimedia Information Processing, Sydney, Australia.

Shinn-Cunningham, B. G., H. Lehnert, et al. (1997). Auditory displays. Spatial and Biaural Hearing. R. Gilkey and T. Anderson. New Jersey, Lawrence Erlbaum Associates, THE PSYCHOPHYSICS OF REAL AND VIRTUAL AUDITORY SPACES Publishers: 611 - 664. 63 Use Your Brain By Dr Karl Kruszelnicki

THE HUMAN BRAIN is one of the most complicated devices that you could think of. It’s a very expensive organ – from a metabolic point of view. It takes a lot of energy to run the brain. Even though it weighs only about 2% of our body weight, it uses about 20% of our blood supply and 20% of our energy – and it generates about 20% of our heat. There are many myths about this mysterious organ. One persistent myth is that we really use only 10% of our brain – and that if we could use the remaining 90% we could each win a Nobel Prize or a gold medal at the Olympics, or even unleash our supposed psychic powers.

This myth has been going for trained rats to run through mazes, conscious part of our brain, with nearly a century, and it keeps re- and then measured how well they the remaining 90% the emerging. Over the last decade, did as he removed more and more subconscious part. (In reality, there some motivational speakers have of the cortex of their brains. He is no such neat division.) shamelessly recycled this myth, found that memory is not stored in and they claim that if you take their one single place, but exists In the 1980s, Yorkshire TV in the expensive course, you will suddenly throughout the entire cortex, and UK showed a documentary called, be able to use all of your probably a few other places as Is Your Brain Really Necessary? It brainpower. well. In fact, his results showed that described the work of the late removal of any of the cortex caused British neurologist, Professor John One of the earliest popular memory problems. Karl Lashley’s Lorber. He was a kids’ doctor, and mentions of this myth is in Dale fairly-straightforward result was he saw many cases of Carnegie’s 1936 book, How To Win somehow changed to read that rats hydrocephalus. Hydrocephalus Friends and Influence People.He did fine until they had only 10% of happens to about one of two kids wanted to back up his claim that if their brains left. First, he never out of every 1,000 kids that are you worked your brain just a little claimed that. Second, he never born alive. There is a constant harder, you could improve your life removed as much as 90% of the circulation of cerebro-spinal fluid enormously. Without any brain. around the brain and spinal cord. If neurological proof whatsoever, he too much fluid is produced, or if boldly claimed that most people This myth has been constantly there is a blockage to its outflow used only 15% of their brains. His reinvented every decade or so. So from the brain, then it can build up book sold very well indeed, so that one version might have that inside the skull. This excess fluid helped push the myth. certified mega-brain, Albert usually makes the skull grow Einstein, saying (guess what?) that bigger, but sometimes it just makes Dale Carnegie might have got his “we use only 10% of our brain”. the brain meat get thinner as the misinformation by wrongly But I have also heard the version meat gets squashed up against the interpreting the experiments of a where some anonymous scientist bony skull. certain Karl Lashley, back in the (who is never named) supposedly 1920s. He was trying to find out discovered that we use only 10% of Professor Lorber discussed many WAVES OF THE FUTURE OF WAVES just where in the brain this strange our brain. Another version is that cases where young people had not 64 thing called “memory” is stored. He 10% of the mass of the brain is the much brain, but normal intelligence. In one extraordinary case, a young In fact, if you did use all of your man had only one millimeter brain at the same time, you would thickness of gray matter in his probably have a Grand Mal brain, instead of the average 45 epileptic fit. And finally, have you mm. Even so, he had an IQ of 126 ever heard a doctor say, “Luckily, (the average is 100) and had gained he had a stroke in that 90% of the an honours degree in mathematics! brain we never use, so I think he’ll be alright.”? But this does not prove that most of your brain is useless. Instead, it FROM Dr Karl’s book Mythconceptions shows that in some cases, the (Harper Collins Publishers) brain can recover from, or compensate for, quite major injuries.

The myth that we use only 10% of our brain is finally being proved untrue, because over the last few decades, we have invented new technologies (such as Positron Emission Tomography and Functional Magnetic Resonance Imaging) that can show the metabolism of the brain. In any one single activity (talking, reading, walking, laughing, eating, looking, hearing, etc) we use only a few per cent of our brain – but over a 24- hour day, all the brain will light up USE YOUR BRAIN on the scan. 65 What got you interested in science in the first place? I did not see suddenly see the light, so to speak. I did physics in high school and thought it was really neat. I also saw a number of TV programs PROFESSOR MARTIJN on particle physics for TV in which the DE STERKE studied discovery of the J/psi particle was Engineering and Applied described – that also got me interested. Physics at Delft University in the What were you like as a kid? Were Netherlands before you curious, pulling apart stuff to receiving his doctorate see how it worked? in optics from the I never pulled things apart – which University of Rochester might explain why I evolved into a in the USA in 1988. He theorist. held a position as a research fellow at the What’s the best thing about being University of Toronto, a researcher in your field? Canada until 1990, This is a good time to be a before moving to Sydney in 1991. researcher in optics. Many of the problems are very fundamental, yet Martijn is currently a Reader with the the applications are real and not too School of Physics’s Centre for far in the future. Ultrahigh-bandwidth Devices for Optical Systems – CUDOS – doing Who inspires you – either in science research into the next generation of or in other areas of your life? communication systems that will use Early in my career I had the great light, not electricity, to carry and privilege to work with one of the process information. He is a greats in the field. Someone who theoretical physicist, whose made time for everyone in his (very approach to his research is large) group on a weekly basis, characterized by actively seeking worked in half a dozen different collaborations with experimentalists. areas, and is one of nicest guys in He has authored papers in the fields the universe. of optics and photonics, solid-state physics, and acoustics, and these If you could go back and papers have appeared both in the specialise in a different field, what physics and in the engineering would it be and why? literature. During high school biochemistry was coming up strongly, but it did not have the rigorous basis of physics for example. In another life I might have done biochemistry.

What’s the ‘next big thing’ in science, in your opinion? What’s coming up in the next decade or so? Essentially unlimited bandwidth for all, artificial life forms, and mobile phones that dispense coffee in the 66 Fig 1. Representation of the links that are powering the internet. morning. Telecommunications: the here and now

Martijn de Sterke

PEOPLE COMMUNICATE VIA many different means: speaking to each other in person, over the phone, mobile or fixed line, or increasingly through the world-wide web, sending each other letters or email messages, pictures and movies – Figure 1 shows a schematic of the main internet connections in the world. Here I will discuss aspects of communications science and engineering, and some of the issues involved in creating the high-speed communications that allowed the development of the internet. I also want you to get some appreciation of the relevant numbers related to quantities of information that come up in this area. In this lecture I will predominantly look to the present: how are things done at present and why. In the next chapter I will discuss the challenges coming up if we want to achieve even larger data transfer rates than now, and some of the ideas that have appeared on the horizon to deal with some of these issues.

67 s I mentioned, there are many means of the dice is unbiased then pn=1/6 for each n = 1, communications; let us first use long- ..., 6. In contrast, if the dice is biased then not all A distance telephone as an example. Any pn are the same – but the probabilities must still long-distance phone call makes use of an optical add up to unity. The question is now: if you are fibre. First I want to discuss why fibres are used given just one outcome and nothing else, what is and why they have displaced other means of the expected information that you have been communication. To understand the unique given? advantages of optical fibres we need to understand some communication theory, an area that Let us first see if we can find some general straddles mathematics and electrical engineering. properties that the expected information should Communication theory was developed after World satisfy. First, negative information makes no War II by, among others, Claude Shannon in the sense, and so the expected information must be US and Norbert Wiener in Europe. Here we will non-negative.1 Now, when N=1 (which would be follow some of the ideas developed by Shannon like tossing a coin with heads on both sides) then (his picture is shown in Figure 2). I will start with no information at all is given, because the answer some of the elementary ideas and then fast- was known in advance (since the coin would forward to the result that we require. always come down heads!).

Further, if N=6 and pn=1/6 for all n (an unbiased dice) then the expected information per outcome is larger than when N=2, corresponding to throwing an unbiased coin – larger, because there were more possible outcomes for the roll of [FIGURE 2 HERE] the dice than the toss of the coin, so the actual outcome of the dice roll carries, in a sense, more information than a toss of a coin.

Let us now consider not N=1, but a situation close to it: we take N=2, but p1=1-ρ and p2=ρ. When ρ Y0, or when ρ Y1, one of the outcomes is certain and the other impossible – it becomes the same as the N=1 case above. For this particular case of N=2, then, the expected information is zero when ρ is zero or one, and it presumably peaks somewhere in between. Fig 2. Claude Shannon. Arguments of this type can be continued for a while, but let us not further beat around the bush Defining Information – the expected information H per outcome is Since communication is about taking information given by from one place to another, we first need a good definition of information. The definition we adopt H= -∑ pn log pn. (1) is perhaps initially somewhat counterintuitive: it essentially measures the amount of uncertainty Note that this function satisfies the requirements that can be resolved. Let us initially take digital discussed earlier: since all pn are between 0 and communications, where we communicate via 1, H cannot be negative.2 The logarithm ensures signals that can only have discrete values, as an that knowing two outcomes gives twice the example. Suppose the variable that we measure expected information as that of a single known can take one of N values, each with a probability outcome, which can be checked for example by WAVES OF THE FUTURE OF WAVES pn, such that, of course, ∑ pn =1. As a simple comparing H for the tossing of an unbiased coin 68 example, if the variable is a dice then N=6, and if or dice once and twice. In communication theory we’re talking about input into an information channel is indicated by transferring messages of some kind, and Z, which signifies all possible inputs zi each with messages come as a string of symbols. In this probability pi. For a binary signal, the zi can take case, the equivalent of seeing the outcome of a the values ‘0’ and ‘1’, each of which is likely to roll of the dice is seeing what symbols appear in have a probability of 50% (though we could easily our message – so in communications we call generalise to cases where z is not 50%). The outcomes ‘symbols’. output of the communication link is similarly indicated by W, with the symbols now indicated The only thing left to discuss now is the base of by wj. the logarithm: the obvious ones are base 10 logs or natural logarithms with base e. However, in communication theory one uses base 2 Message sent, message received logarithms, and the unit of information is then a We now address the following question, which bit. Let us consider an unbiased coin: in that case goes to the heart of communications: if we know N=2 and p1=p2=1/2. Since log2(1/2) = -1 we what message is received, what does that say find that H=1 bit per symbol. While Equation (1) about the message that was sent? That is, given gives the expected information per symbol, that we know the output W, what can we say multiplication by the number of symbols per about the input Z? To start answering this second gives, of course, the expected information question, we introduce the idea of Conditional per second. Entropy, which is defined as

H(Z / W) = -∑ p(z,w) log(p(zi / wj)). (2) Information or Entropy? ij i j Some of you may think that Equation (1) looks Here p(a/b) is the conditional probability familiar – indeed, apart from a factor p(a/b)=p(a,b)/p(b).3 The Conditional Entropy can k=1.38x10-23 J/K (the Boltzmann constant, which be interpreted to be the uncertainty in our plays a central role in thermodynamics) and a knowledge of Z if W is known. As an extreme factor 1/ln 2 (the difference between base 2 logs case, consider when Z and W are independent, and natural logs), essentially the same expression so the output of the channel bears no relation to is used in physics for the definition of entropy.In the input (which would not be a very useful fact, the expected information is often referred to communications channel, it’s true). Then as the entropy, a fundamental function in H(Z/W)=H(Z,W)/H(W)=H(Z)H(W)/H(W)=H(Z), and thermodynamics. In this definition the N can refer knowing W has not cleared up anything about Z. to the number of particles (atoms) in the system Here the first equality is the definition of H(Z/W) under study, and the pn to the probability for the and the second uses the definition of system to be in a particular state. For an atom independent events from Footnote 3. this could refer to the energy levels that the electrons have access to. This relation between The other extreme case occurs when W gives the expected information and the entropy is no perfect information about Z – in that case there is coincidence, as the entropy can be thought of as no uncertainty left when W is known and measuring the degree of disorder of a physical H(Z/W)=0, since every term in Equation (2) is system – a large degree of disorder means that a proportional to log(p(z/w)) = log(1) = 0. lot of information is required to describe the system. Indeed, many of the more advanced Having defined the conditional entropy allows us concepts in communication theory are couched in finally to introduce the Mutual Information, which terms of the entropy. corresponds to the information regarding the input W that is obtained by knowing the output Z. It is now easy to make the link between the The mutual information is defined to be entropy and the way we can send information TELECOMMUNICATIONS: THERE HERE AND NOW through an information channel, the details of I(W;Z) = H(Z) – H(Z/W). (3) which do not matter at this point. Suppose the 69 This is not surprising if you think about it: if H(Z) information between input and output is is the uncertainty that would be cleared up if Z maximized when p(0)=p(1)=1/2. were known, and H(Z/W) is the uncertainty that is left regarding Z if W is known, then the difference Using a bit of thought you can find from Equation between these two is the information that W (3) that, in this case, C=1+(1-p)log(1-p)+plog(p) provides about Z. Considering the two extreme and so C=1 bit when p=0, but drops to C=0.92 cases discussed in the previous paragraph again: bit when p=1%. When p=0.5 – that is, when in a very bad channel where the output does not each bit has a 50% chance of being wrong – depend on the input, H(Z/W)=H(Z) and thus then C=0 and all information is lost in the I(W;Z)=0 – no information whatsoever is channel. (When the symmetry between 0s and 1s provided. In a perfect channel, though, H(Z/W)=0, is dropped, that is when the noise in the 0s and and so I(W;Z)=H(Z), and so the Mutual 1s are different, then the capacity is not achieved Information equals the Expected Information. In for p(0)=p(1)=1/2.) practice, of course, the properties of a channel are somewhere between these extremes. Capacity and Bandwidth With the formal definition of the Capacity of a Capacity of the Information Channel channel we are now in a position to discuss one Now that we know a little about communication of the most celebrated theorems in theory, let us concentrate on a channel that takes communication theory, derived by Shannon. A information from one place to another. We like to well known form of this theorem does not apply know the Capacity of this channel, the maximum to the digital signals that we have been amount of information that can possibly be considering up to now, but with analogue signals, squeezed through this channel given its physical but this does not matter much. The theorem limitations. This does not necessarily mean that refers to an analogue signal of power S that is information is actually lost. Using clever coding affected by noise. The noise is assumed to be techniques, for example error correcting code, it white noise, which means that the value of the is possible to correct the errors that were made noise can change arbitrarily quickly in time, and in transmitting the information.4 However, coding to have average power N. theory is well outside the scope of this lecture and we will not discuss it further; here we are At each particular time, the noise is assumed to interested in the physical characteristics of the have a Gaussian distribution, which indicates the information. Using the concepts introduced earlier probability that the noise has some particular the Capacity of a channel is defined as value at a given time.5 These assumptions are idealizations, but this can be turned into a C = max | P(Z) I (z, w); (4) positive: the situation that is analysed is, in some sense, the best possible one, and the situation to Here max|P(Z) means the maximum value that can aim for in practice. Under these conditions, the be reached by adjusting the P(Z). This means that capacity was proven by Shannon to have the the number of input symbols and their probabil- value ities are chosen such that the mutual information of the input and the output is maximized. C = B log ( 1 + S/N ), (5) As an example, consider a system with binary where B is the bandwidth. Equation (5) has units information encoded in 0s and 1s as discussed in of inverse seconds, and thus refers to the Footnote 3. During the travelling through the capacity per unit time. It is important to channel the signal is subject to noise and other appreciate the conceptual difference between deleterious effects. As a consequence, there is a Equations (4) and (5): the former is simply a small chance p of an error, so that a 0 is received definition of the Capacity, whereas the second WAVES OF THE FUTURE OF WAVES as 1 or vice versa. Since the error probability is gives the value of the actual capacity in a 70 symmetric between 0s and 1s, the mutual communication system under the conditions specifically discussed above. It is clear that, as the same thing. Then the same information is expected, when the noise power increases, the carried through the channel in half the time; thus capacity goes down. However, since the signal- the information is being transferred at a rate to-noise ratio S/N enters in logarithmic form, in twice as large as before. In the language that we practice it does not vary by very much – I return have developed here this means that the entropy to this below. H that can be transferred per unit time has doubled too. There are now two things to discuss: what is meant by the bandwidth, and what exactly does This procedure would imply that arbitrarily high Equation (5) mean? Since most of us have at rates can be achieved, but this is of course least a vague notion of bandwidth, let us start impossible – the bandwidth of the channel limits with the latter. The meaning of Equation (5) is that the rate at which the signal can change its value, if a data stream with HC then no such code exists and the channel. One way of understanding the transmission errors inevitably occur. occurrence of the bandwidth in Shannon’s result (5) is to imagine a scheme in which the The theorem is obviously very powerful but it information is transferred using short pulses. The shares the infuriating characteristics of all presence of a pulse would indicate a 1, while the ‘existence theorems’: proving that something can absence would indicate a 0. Clearly, the shorter be done does not necessarily mean that you each pulse, the more tightly the pulses can be know how to do it. In fact, codes that can achieve packed, and the higher the information transfer. the theoretical limit have not yet been devised! By definition, a short pulse needs to have steep sides, and a steep side can only be generated by Let us now return to the bandwidth B that high frequencies. appears in Shannon’s result: it basically tells us how fast the signal can change with time. If the Let us use Equation (5) to see how we can bandwidth is low, this means that the signal optimize the capacity of a channel. You might first varies only slowly with time and not much want to imagine why you would want to increase information can be transferred. Suppose, for the capacity in the first place. Take as an example example, that we have some signal that carries the internet, and the time it takes to download information, say that in Figure 3. It is carried by various files. Movies are massive files, and yet an information channel at some rate. Now you do not want to wait more than a few minutes compress the time axis by a factor of 2 and do to download one of them. Future applications of the internet will require even larger bandwidths. In fact, because new telecommunications technology always has unforseen applications, the historical record of predicting demand for capacity is poor and has always been much lower than the actual outcome, as illustrated in Figure 4.

Returning now to the question as to how the capacity may be increased, note that this can be done by making the signal-to-noise ratio S/N as large as possible. However, this is not a particularly good way to go about this, since the TELECOMMUNICATIONS: THERE HERE AND NOW Fig 3. Example of a signal with Gaussian noise as a ratio appears in the argument of a logarithm, a function of time. function that varies very slowly as a function of 71 go through all kinds of materials, so the channel would be transparent.

The problem is that we simply cannot manipulate X-rays particularly well to encode them with any decent amount of information. Other high- frequency waves are in the Ultraviolet part of the spectrum: but now the problem is that UV radiation is easily scattered or absorbed, so the channel isn’t transparent anymore, and, as with Fig 4. Evolution of installed bandwidth with time, compared X-rays, we do not know how to manipulate UV to various predictions. The latter have always been severe underestimates. radiation well enough. The class of waves with the next lower frequency range is visible light. We can easily generate these, we understand them the argument. For example, to make the capacity well, and some materials are very transparent. ten times larger, the signal to noise needs to We also know how to modulate light and so we increase by roughly a factor 210=1024. Instead, it may consider this option in some more detail. is more efficient to try and increase the capacity via the bandwidth, as it enters into Shannon’s Though light propagates through air quite well result linearly: a ten-fold increase in the Capacity (after all, we can see each other!), over long can be achieved by a tenfold increase in W. Thus, distances the light tends to diffract, which means the way to increase the capacity is to increase that it spreads while it propagates. Light the bandwidth.6 propagation through air also needs a clear line of sight at all times. Both of these problems can be overcome when light propagates through an Optimising Bandwidth optical fibre. In an optical fibre the light is The question now thus is: how can we optimize confined in the fibre core and cannot escape the bandwidth of a communications channel? Let from it, and it therefore does not diffract. us think about AM radio waves: these have a wavelength of about λ≈1 km and thus a This can briefly be understood using a schematic frequency f=c/λ≈300 kHz, where c is the speed of an optical fibre, shown in Figure 5(a). It shows of light in vacuum. Clearly, the maximum a small core with a diameter of 8.3 µm, bandwidth that can be achieved using surrounded by a much larger cladding with a microwaves is a few times 105 Hz, and the diameter of 125 µm. The cladding is made of capacity is thus roughly similar. To make the pure silica and has a refractive index around bandwidth any larger you would need waves with 1.45. The cladding has a refractive index that is higher frequencies, which would no longer be AM slightly higher since it is doped with a small radio waves. This is a general result: the amount of germanium. Germanium is chemically maximum bandwidth that can be achieved is similar to silicon, since they appear in the same roughly equal to the highest frequency that can column in the periodic table; the addition of be used to carry the information. germanium increases the refractive index since a germanium atom has more electrons than a The result from the previous paragraph now silicon atom. In the standard SMF28 optical fibre clearly shows how to optimize the capacity: we this difference in refractive index amounts to only want to use waves with the highest possible 0.36%, i.e., the refractive index difference frequency. Of course the communication channel between 1.45 and 1.455. needs to be transparent to these waves, and we must be able, somehow, to encode information Now remember Snell’s law: on (or ‘modulate’) this wave. So if we want high WAVES OF THE FUTURE OF WAVES frequencies, let’s consider X-rays: they have n1 sin θ 1 = n2 sin θ 2 (6) 72 wavelengths λ≈1 nm, and so f=3x1017 Hz. X-rays Fig 5. (a) Schematic of the cross section of an optical fibre. (b) Illustration of Snell’s law and total internal reflection. Here n1 is taken to be larger than n2.

It applies when light, travelling through a medium Here the losses are 0.2 dB/km, where dB stands with refractive index n1, is incident at angle (1 on for decibels – to understand what this means you a medium with refractive index n2 (see Figure need to know that 10 dB corresponds to a factor 5(b)). Angle θ2 then gives the direction in which 10, that 20 dB corresponds to a factor 100, etc. the light propagates through medium 2. Suppose In general, the value in dB of a factor F is 0.3 now that light is in the core, which has a slightly 10log10(F). Since 10 ≈2, that means 3 dB higher refractive index than the cladding. Then if corresponds to a factor of 2. Coming back to the angle θ1 is sufficiently large the left-hand side of loss of 0.2 dB/km, this corresponds to 3 dB per Equation (6) can be larger than n2 – and so we 15 km – and so the (infrared) light loses half of cannot find an angle θ2 that satisfies the its strength after travelling an astonishing 15 km. equation. As a consequence, the light is totally Once the losses get too large the radiation cannot internally reflected off the interface between the be used, so the bandwidth of infrared light in an core and the cladding, and must stay inside the optical fibre corresponds to roughly 200 nm, core. This can only work if the light is initially in which is approximately 25 THz.7 Assuming the high-index medium and is the reason why the S/N=100, corresponding to 20 dB, Equation (5) fibre has a slightly higher refractive index than then gives a capacity of 166 Tb/s. the cladding. As per our previous discussion, this bandwidth is The light remains confined to the fibre core, even orders of magnitude higher than can be achieved when the fibre is bent (by a modest amount, a with, for example, AM radio waves, since these few centimetres or so), and so there is no need have much lower frequencies. All optical fibres for a clear line of sight. Figure 6 shows the loss made of silica have their minimum loss around in a number of optical fibres as a function of λ=1.55 µm, and so all long distance optical wavelength. Note that it reaches its lowest value communication uses light in this part of the around λ=1.55 µm – not in the visible part of the spectrum. spectrum, but the near-infrared. Bandwidth in Perspective Let us put the number obtained in the previous paragraph in some sort of perspective. Have a look at Figure 7, showing what physical form the information in the world takes. Note that this graph has a logarithmic vertical scale and that the units are Petabytes (1015 bytes=8x1015 bits). Here the length of one character of information is 1 byte = 8 bits.

The Library of the US Congress contains a total amount of information of a few Petabits. Which TELECOMMUNICATIONS: THERE HERE AND NOW Fig 6. Propagation losses in a number of different sounds like a lot of information – until you realise optical fibres. that an optical fibre can transfer all this 73 well does it work in practice? The theoretical limit given by Shannon’s theorem has not yet been reached, but the capacity of an optical fibre is nonetheless incredibly high.

Below I discuss some of the ways in which information is sent through optical fibres. The key word to know here is that of Wavelength Division Multiplexing, or WDM for short. Before explaining this let us use an analogy. Earlier we discussed AM radio and found that the capacity Fig 7. Estimates of amounts of information available stored is quite low because of the low frequencies that in various ways. are involved.

information in 10-15 seconds or so. The total However, what we did not discuss is that the total memory of all people has been estimated to be AM band carries roughly 10 stations, each of 1000 Pb, which, with 6 billion people on this which has a different centre frequency. Each planet, gives a modest 100-200 Mb each. radio station needs a certain bandwidth of tens of Though this is more than a floppy disc, it roughly kHz, and so the frequencies of the different corresponds to the amount of information that stations in a particular geographical area must be can be stored on a flash memory device, and is at least this far apart. Different radio stations can dwarfed by any type of memory disc – a sobering be picked up by changing the centre frequency insight for members of the genus homo sapiens! that the radio receives. Wavelength division multiplexing works similarly. Note from the table that the total digital storage capacity available, mostly disc space, is another We saw in the discussion of Figure 6 that the few orders of magnitude larger. Let us also total bandwidth available in a fibre is about 25 consider the total amount of information that is THz. In practice this bandwidth is chopped up into available in the entire universe. Lloyd has narrow bands, the analogue of a radio station, 3/2 estimated this to be (t/tp) bits, where t is the with a width of typically 50 or 100 GHz. Each of age of the universe (about 1010 years), and these bands is, confusingly, referred to as a 5 -11 tp=√(Gh/c ) – in this equation G=6.67x10 “channel”. Each channel corresponds to a slightly Nm2/kg2 is the gravitational constant that enters different colour than the other bands, and is Newton’s theory of gravity, and h=1.05x10-34 Js meant to transfer information independently of is Planck’s constant, the constant that governs the others. A schematic of the setup is shown in quantum processes. The time tp is the Planck Figure 8. A set of different lasers, each operating time, which enters into research to understand at slightly different frequencies, encodes the nature at its most basic level. Substituting the information for each of the channels. These are numbers we find the resulting storage capacity then combined in a multiplexer and sent through of the universe is something like 1091 bits.8 This an optical fibre. To account for the losses, even puts the results in Figure 7 somewhat in though these are very small, a number of perspective. amplifiers need to be included in the fibre link, but the details do not matter here.

Bandwidth: theory vs. reality At the far end of the fibre link the different Much of the above (in this context particularly channels are separated again and fall on different Equation (5), but also the estimated number of receivers. In this case the data rate of each WDM bits in the universe) was based on existence channel is about 100 Gb/s, and there are roughly theorems. This means something can be done in 100 channels, so the total aggregate data rate is WAVES OF THE FUTURE OF WAVES principle, but does not say how it is to be 10 Tb/s – not that close to the theoretical 74 achieved. So the natural question to ask is: how maximum of Shannon. Notes 1 Let us here not consider certain reality shows on TV which may violate this assumption. 2 Remember that, for a logarithm of any base, log(x)>0 for x>1 and log(x)<0 for 0moon is full and the date being 25 December are hour? The entire library of the US congress; the uncorrelated. An often used example in a communications memory of about 10 million context is a binary channel in which the people; the total genomic code probability of the bit arriving incorrectly of 1 million people. However, it What can be transferred is P(0/1)=P(1/0)=p and the probability corresponds to only about 10-17 that they arrive correctly is thus of the information needed to in one hour? The entire P(0/0)=P(1/1)=1-p. teleport a human. library of the US 4 A simple type of error correcting code is to send every symbol 3 times, say. If In the next chapter we will see a congress; the memory of one of the three symbols arrives number of effects that limit the about 10 million people; incorrectly than this can easily be data rates described here and detected. This method assumes that the that limit us increasing the data the total genomic code probability of receiving the wrong signal rates even more, which would of 1 million people. is small, so that it is very unlikely that allow applications such as two of the three symbols are received telemedicine and other However, it corresponds incorrectly, since then the wrong applications that link to remote to only about 10-17 of the conclusion would be drawn. In practice locations. My personal favourite the error correcting codes are much application, though somewhat information needed to more sophisticated, but outside the remote, is for people to touch teleport a human. scope of the discussion here. each other at a distance – 5 This means that the probability that the wouldn’t it be great if noise has a particular value s is grandparents could hold their proportional to exp(-s2/σ2) where σ is a new grandchild at a distance and have it feel as if characteristic width that here is related to S. they had the child in their arms? In lecture two 6 Amongst all this discussion about high Capacity and we will also consider possible solutions to these bandwidth, we may want to pause for a minute and think problems. about channel with very low Capacity. We may want to discuss this during the lecture, and perhaps a question to think about is the following: what is the lowest capacity I am grateful to Dr Richart Slusher from Bell channel that you can think off? Labs, Lucent Technologies, for the use if some of 7 Here ‘T’ is short for Tera, 1012 his figures. I am also grateful to my colleagues at 8 S. Lloyd, “Computational capacity of the universe,” Phys. TELECOMMUNICATIONS: THERE HERE AND NOW CUDOS, particularly Ben Eggleton for discussions Rev. Lett. 88, 237901 (2002). Note that there is a typo in regarding some of the issues raised here. this paper: the 3/4 exponent should be 3/2. 75 Albert Einstein: Scientist or Superstar?

I WANT TO know how God created this world. I am not interested in this or that phenomenon, in the spectrum of this or that element. I want to know His thoughts; the rest are details.

If there is such a thing as a business, spent some time events into dynamic leading players superstar scientist, Albert Einstein travelling and eventually enrolled in in their own right. His were truly must be it. His shock of wild hair, the Federal Institute of Technology in groundbreaking ideas that paved his great nose and bushy Zurich. He graduated in 1900 with a the way for the era of modern moustache, his craggy features – fairly ordinary academic record. physics. Einstein’s image gazes wisely out from posters, t-shirts, coffee mugs Afterwards, he did what many General Relativity ... even cuddly toys and action graduates do: he looked for work, The ideas of 1905 were just the figurines! taking various odd jobs before preamble to what was arguably landing a position as a patent Einstein’s greatest achievement: his What is it about him that captures examiner with the Swiss patent theory of General Relativity. In this the public’s imagination this way? office. And there he worked, for most beautiful of works he showed Perhaps it’s that he seems so sage several years ... how space and time, and energy and yet so gentle, he’s the and matter, act and react to intellectual giant but also the kindly The Miraculous Year produce the effect we know as grandfather-figure. Perhaps it’s the Prior to 1905, nothing in Einstein’s gravity. The theory’s consequences way he tried – naively some would academic career foreshadowed the – warped space, stretched time, say – to influence politics the way events to come. In just one year, he bent light beams and black holes – he influenced global science. wrote a series of five papers that seem very strange and non- Perhaps it’s that he’s the archetypal changed our view of the universe intuitive, yet today the theory is a absent-minded professor, too forever and cemented his vital part of science and technology. preoccupied with the secrets of the reputation as one of the finest For example, Global Positioning universe to be concerned with minds of his or any other Satellite (GPS) systems rely on haircuts or matching socks. generation. General Relativity for their incredible accuracy. An inauspicious beginning In 1905, Einsein wrote about Albert Einstein was born in 1879 at Brownian Motion and demonstrated The questions left open by the Ulm in Württemberg, Germany, the conclusively that atoms really do General Theory of Relativity are still son of middle-class Jewish exist; he took the notion of tantalizing scientists today, as they parents. He was a quiet, curious quantised energy, previously an search for evidence of gravity child; his parents were concerned abstract mathematical trick, and waves, attempt to figure out if the that he didn’t start speaking until showed how it explained perfectly Universe is open or closed, and try he was three years old. At school the interactions of light and matter; to find ways to reconcile he didn’t exactly stand out amongst and he merged space and time, fundamental difficulties with mixing his peers, and left at the age of transforming them from a General Relativity with the Quantum WAVES OF THE FUTURE OF WAVES fifteen. He worked in his parents’ background arena for physical Theory. 76 Einstein the Political Scientist Einstein’s political reputation at times matched his scientific reputation. When the First World War broke out in August 1914, Einstein was vilified as a traitor by some in Germany for his pacifistic ideas and his open criticism of the government’s ambitions. Though he remained in Germany throughout the war, life as a Jewish scientist became increasingly difficult and, eventually, untenable. In 1933 he Nations and the need for nuclear renounced his German citizenship disarmament. Throughout his life and moved to America. For many he wrote many essays on political years the Nazi government banned philosophy, alongside his physics even teaching Einstein’s ideas, research. along with those of other Jewish scientists. Einstein died in 1955 at the age of 76. He was approached to become President of Israel (though he In January 2000, TIME magazine declined the offer), and he wrote to nominated Albert Einstein as the the American President, F.D. Person of the Century. It’s hard to Roosevelt, warning him about think of another individual who has Germany’s plans to build nuclear gained so much attention, who is weapons. He encouraged the USA so well recognised across borders to develop on their own nuclear and generations ... and who, in the research programme – his earlier end, is so deserving of the honour. belief in total pacifism was tempered by his belief that Germany One thing I have learned in a the aggressor would not hesitate to long life: that all our science, use their own atomic weapons. measured against reality, is

primitive and childlike – and yet ALBERT EINSTEIN: SCIENTIST OR SUPERSTAR? In his later life, Einstein was a it is the most precious thing strong supporter of the United we have. 77 78 Telecommunications: looking to the future

Martijn de Sterke

IN THE PREVIOUS chapter I reviewed how the capacity of a channel is defined, and Shannon’s expression for the capacity for a channel with Gaussian noise. We saw that the main contribution to the capacity comes from the bandwidth of the channel, but that the capacity also depends more weakly on the signal to noise ratio. I also showed how modern telecommunications systems make use of wavelength division multiplexing (WDM), in which the total available bandwidth is chopped up in narrow channels, each of which, using separate laser sources and detectors, carries a signal independent of the other channels. For reasons that we will discuss shortly, the properties of the fibre cause Shannon’s result not apply to this case, and we thus require a generalization – we’ll tackle this in the first part of this chapter. In the second part we discuss how to deal with these effects.

79 Linear and nonlinear described by its refractive index and its In the previous chapter we made a number of absorption coefficient (though, as we shall see, tacit assumptions about the optical fibres that we these quantities generally depend on frequency). are dealing with; one of the most important of Note that if the proportionality constant between these is the assumption of linearity.To d and E is small, then the refractive index is close understand what this means we need to to unity; in contrast, if it is easy to move the understand something about the physics of how positive and negative charges with respect to light interacts with matter. The model people each other, then the refractive index is large. In often use to describe this interaction is the order to prepare us for the nonlinear effects we Lorentz model, developed by the famous Dutch discuss below let me point out that one of the physicist Hendrik Lorentz who lived from hallmarks of linear systems is that no new 1853-1928. frequencies are generated: if the light has a frequency f or a series of frequencies f1,f2, ..., Consider the atom as in Figure 1(a). It has a then light with other frequencies is not generated nucleus in the centre, and an electron, with some by the medium. probability distribution, around it. We now apply an electric field as indicated in the figure, and as We now consider applied fields that are stronger, a consequence the nucleus and the electron are but not so strong that the atom is destroyed (see subject to a force. However since the nucleus and Footnote 1). In this case the linear result no the electron are oppositely charged, the forces longer applies and can be generalized to d ∝ E+ 2 3 point in opposite directions. Let us now consider α1 E + α2E , where α1 and α2 are small the situation in which the force is small.1 Then, constants; the terms that are nonlinear in the the effect of the electric field is that the nucleus electric field are then thus small corrections to and ‘centre of charge’ of the electron no longer the linear relation between d and E. Notice that coincide (see Figure 1(b)). Provided the force is the linear relation d ∝ E has the property that small, the distance d between these two is when E flips sign then so does d; it is thus left- proportional to the applied field and so d ∝ E. right symmetric. The cubic term in the nonlinear generalization has the same property, but the quadratic term does not; its presence causes electric fields that point to the left and to right to lead to different values of d. Here we are interested in materials that are left-right symmetric and we therefore drop the quadratic term; the first correction to linearity is then cubic.

The question we now wish to answer is the following: what is the effect of the additional cubic Figure 1: (a) Schematic of an atom, consisting of an term on the optical properties of the material? The electron, indicated by its probability distribution, around the nucleus. (b) In an applied electric field the electron and the answer is that, unlike the linear response nucleus are subject to forces that point in opposite discussed earlier, the nonlinear response does directions so that the centres of the positive and negative lead to the generation of new frequencies. These charges no longer coincide. new frequencies can be found as follows: consider a light field with frequencies f 1,f2 and f 3; then In this model, the light is considered to be a wave the new frequencies are those of2 with an associated electric field (the effect of the 3 magnetic field is weak and is neglected here); (cos(2πf1t) + cos(2πf2t) + cos(2πf3t)) (1) thus, for the purposes of our discussion, the light is modelled as an electric field with a frequency In order to find these frequencies you first need of roughly 1015 Hz. Under this condition of to expand to third power, leading to 10 different WAVES OF THE FUTURE OF WAVES linearity (d ∝ E), the properties of a medium that terms, all of the form cos(2πfit) cos(2πfjt) π 80 consists of a collection of such atoms can be cos(2 fkt), where i,j,k can be any of 1,2,3. The next step is to use the trigonometric identities3, the probabilities of the processes (a) and (b)? One which show that expression (1) can be written as might expect that the probability of one of the a sum of terms of the type cos(2π(fi±fj±fk)t). processes occurring depends on the density of photons that act as ‘input’ into the FWM process. Therefore, since there are more photons at f1 Frequency Mixing than at f2, we expect process (a) to be more likely Let us look at some of the consequences of this than process (b) – and indeed this is so. In fact, by first taking i = j = k =1, say. Then we find the probability of the process is proportional to cos(6πfjt), i.e., the medium generates a wave the square of the intensity, or, using photons with frequency 3f1 (third-harmonic generation, or again, proportional to the square of the number THG). More generally, the nonlinear medium of photons. generates the new frequencies |fi ±fj ±fk |. This general process is referred to as frequency mixing; since in total there are four frequencies Frequency Mixing and WDM involved, in the optical context it is referred to as Now we know about frequency mixing, what is its four-wave mixing (FWM). Though we identified relation to the WDM that we discussed earlier? the new frequencies that are generated by the The key thing here is to recall that the channels medium, we have not discussed the amplitudes are equally spaced. Therefore, in a generalization of each of these. This is a more difficult problem the processes (a) and (b) from the previous that we do not consider here.4 paragraph, any of the newly generated frequencies also corresponds to one of the The example of THG is actually not a very good channels. Thus, the effect of FWM is expected to one. For reasons that are beyond the scope of be particularly detrimental in WDM systems. The our discussion here, THG is very difficult to details of this have been considered by a number accomplish in practice. In practice the three of people, starting with Mitra and Stark. Here, incoming frequencies f ÿ k are quite similar, and however, we quote a result derived by Stark, two of them may be identical, and the newly Mitra, and Sengupta5: they show that for a WDM generated frequency is similar again. An system, in the presence of FWM, Shannon’s important example here is that of two incident result from Lecture 1 is generalized to frequencies f1 and f2, which generate the new β2 frequencies (2f1 – f2) and (2f2 – f1), leading to C = Blog 1 + (1- )S . (2) four equally spaced frequencies. These ( N + β2 S) frequencies can be considered to follow from energy conservation: the first process can be Note that this reduces to Shannon’s result when considered be of the type β = 0. Parameter β corresponds to the rate at which energy is transferred between different (a) (f1 + f1) Y (f2 + (2f1 – f2)), channels; note that in the numerator of the second term in the argument of the logarithm, and the second β2S corresponds to signal power taken out of a channel by FWM, whereas in the denominator, it (b) (f2 + f2) Y (f1 + (2f2 – f1)). corresponds to this energy being deposited in the other channels and adding to the noise. The Since each photon has an energy E = hf, where h general expression for β is quite complicated, but is Plank’s constant, in each of these processes in one regime it reduces to β = ψSLe. Here ψ is energy is conserved. a coefficient that measures the degree of non- linearity of the fibre and is proportional to α2, and Suppose now that f1 is much stronger than f2 – Le is an effective length that corrects the actual in terms of electric fields this means that the field length of the fibre for losses. at f is much stronger than that at f ; in terms of

1 2 TELECOMMUNICATIONS: LOOKING TO THE FUTURE photons it means that there are many more photons at f1 than at f2. What can be said about 81 Dispersion frequency, the amplitude of the mass is small (it Below I describe ways in which one can deal with is as if the mass really does not want to oscillate FWM – as we will see this is rather subtle. at this frequency). Therefore, before doing so, I will quickly discuss dispersion, another way in which the signal in a Let us now translate these properties of mass- fibre deteriorates, and how to fix it. The contrast spring systems to the properties of atoms when between these two is quite revealing. To irradiated by light. Just as for a mass-spring understand dispersion we first need to remember system, the light, with frequency f, imposes its that the information in a fibre is encoded in short frequency onto the atom; in addition, if f is close optical pulses. The precise number depends on to fa then the amplitude of the atomic motion is the type of system, but a rough pulse length is large. In contrast, when these two frequencies 100 picoseconds (ps). Now recall from the differ significantly, then the amplitude is small. previous lecture that in optical fibres the light is encoded in short light pulses, and that these We also saw that the linear effect leads to a light pulses consist of a number of different refractive index of the medium. Now we combine frequencies. what we have learned: since (a) the applied field forces a sinusoidal oscillation onto the atom, and Let us now return to the Lorentz model that we (b) this oscillation leads to the refractive index considered earlier in this lecture: the atom when considering a collection of atoms, then (c) if consists of a nucleus that is bound to electrons the amplitude of the oscillation depends on that surround it. Earlier we argued that for small frequency, then so must the refractive index – fields the atom behaves linearly: d ∝ F, where d this is referred to as dispersion. is the distance between the electrons and F is the force (which, in turn, is proportional to the applied Let us now combine this with our earlier finding field). Now it so happens that systems in which that a light pulse consists of different the deviation from equilibrium is proportional to frequencies. You may remember that the velocity the applied force are well known: such systems of light in a medium is v=c/n, where c is the give rise to simple harmonic motion (SHM). A velocity of light in a vacuum. So it follows that if good example of a system that exhibits SHM is a different frequencies have different refractive mass hanging off a spring (a ‘mass-spring’ indices in a medium, they must travel at different system), and so let us look at this example for a velocities. The consequence of this is that minute. If we give the mass a small bump then it dispersion has the effect of broadening a pulse, starts to oscillate backwards and forwards. since, upon propagation, the different frequencies Because this oscillation is SHM, it is known that start to march out of step. Once the pulses start (1) the position of the mass varies sinusoidally to broaden so that adjacent pulses begin to with position at a characteristic frequency (fa, overlap, the information would seem to be lost. say), and that the frequency depends on the spring’s strength and the mass, but that (2) the frequency is independent of the amplitude. If, on Overcoming FWM and Dispersion the other hand, we impose a sinusoidal motion on So we have seen two different physical effects the mass, for example by a small motor, then the that affect pulses when they propagate through mass oscillates at this frequency (recall that an optical fibre: dispersion, a linear effect that linear systems such as this one cannot generate causes light pulses to broaden, and FWM, a its own frequencies). nonlinear effect that causes the quality of an optical pulse to decrease more generally.6 If However, the system does have a characteristic untreated, both lead to the loss of information – frequency, fa. The effect of this is that if the this may be OK when you talk to friend over the imposed frequency is close to the characteristic phone, but is unacceptable when your bank frequency, then the mass oscillates with large statement information, for example, is carried by WAVES OF THE FUTURE OF WAVES amplitude. In contrast, for an imposed frequency the fibre. Researchers have therefore studied 82 that differs significantly from the characteristic countermeasures, and I will discuss some of these. We will see that compensation of the fibre dispersion accurately over the entire dispersion is straightforward, at least in principle, transparency bandwidth shown in Figure 6 of the but that the compensation of FWM is much previous chapter.8 Note that, as required, this harder. method can be applied to very short pulses.

Let us first, however, look at the simplest As mentioned earlier, dealing with FWM is much methods for overcoming dispersion and FWM. We more difficult, essentially because it is a nonlinear know roughly the strength of the dispersion and process. The first step in dealing with FWM is to the FWM, and we thus know over what distance have some idea how badly the pulses have they destroy the pulses. Suppose we now include deteriorated. This issue of monitoring is one of in our fibre a detector at some position before the most important since otherwise any this happens. The detector essentially turns the correction is a stab in the dark. The aim of the light into a current; we can then use any method monitoring process can be described as finding from electronics to fix up the information. We now out to what degree the energy in the fibre is let the current associated with the fixed-up signal concentrated in pulses. This is illustrated in drive a laser (as in Figure 8 of the previous Figure 2, which shows (a) a high-quality signal chapter), and a clean pulse once again with well-defined pulses, and (b) a badly propagates through the fibre. This ploy may deteriorated signal with the same average energy. sound too good to be true, but at some level it is To be able to distinguish between these two not: until a few years ago fibre systems worked (without using electronics!) requires a process like this! that depends nonlinearly on the intensity.

There is a problem though: the optical pulses last about 100 ps, and though may sound short, it corresponds to about 100,000 periods of the light wave, and so for the light it could be considered to be quite long! In contrast, for electronics, which as we saw in the previous chapter operates at much lower frequencies, 100 ps is very short, and it is at these time scales that electronics starts to run out of steam. Though Figure 2: (a) Low-noise pulse train, and (b) a pulse train electronics as fast enough for the present with noise, after propagation through an optical fibre. generation of optical networks, it will not be sufficiently to deal with future systems. Therefore We earlier encountered such a process that another approach is needed.7 depends nonlinearly on intensity: it is FWM! Recall that the process (a) f1 + f1 Y f2 + (2f1 – Our task is to negate the effects of dispersion f2) depends quadratically on the intensity of f1. without the use of electronics – we require all- The monitoring process that we require is optical solutions. Fortunately, this is easy, in therefore straightforward, at least in principle. We principle. In a standard fibre the dispersion is start with the signal frequency fs and add another such that the blue frequencies (short frequency fp, and we then monitor the intensity of wavelengths) travel faster than the red ones (long light at the frequency (2fs – fp), the intensity of wavelengths). What is needed to undo the effect which depends quadratically on the intensity of fs. of dispersion is to let the pulses propagate In turn, this tells us whether the pulses in the through a fibre with the opposite dispersion: the signal are narrow (so that they can be red frequencies that lag behind the blue distinguished) or broad (so that distinguishing frequencies in standard fibre would be able to them is difficult). catch up again. The effect is that the pulses narrow down again and that the information can Such an experiment was done by students and TELECOMMUNICATIONS: LOOKING TO THE FUTURE be read out. Though this may sound easy, in colleagues in CUDOS, here in the School of 9 practice it is tricky since it is difficult to negate Physics at the University of Sydney . The crucial 83 result is shown in Figure 3, showing the intensity power of the output signal versus the power of of frequency (2fs – fp) versus a control parameter the input signal. Suppose that we have a device that determines how narrow the signal pulses with a transfer function as shown in Figure 4(a), are. At the zero setting the pulses attain their which has the property that all input signals minimum value, as in Figure 2(a), and the below the switching power Ps leads to no output, intensity of the frequency (2fs – fp) peaks. At whereas all input signal above Ps lead to the other settings, the pulses are much wider, as in output Pt. Consider now an input consisting of a Figure 2(b), and the intensity at frequency (2fs – series of pulses, possibly of poor quality. Then in fp) is significantly lower. the output all input values below Ps disappear, while the values above Ps get the value Pt. Provided that Ps is sensibly chosen, this can regenerate the input signal, as shown schematically in Figure 4(b).

A practical problem with this approach is that no optical systems exist that have a transfer function as in Figure 4(a), and one therefore settles for systems that have a transfer function as in Figure 5(a). Clearly, when the width of the transition region vanishes, it turns into the ideal transfer function in Figure 4(a). A transfer function as in Figure 5(a) can in fact be achieved in FWM, as 10 Figure 3: Intensity at frequency (2fs - fp) (solid line) versus a was first pointed by Ciaramella and Trillo .We control parameter that determines the widths of the pulses. saw earlier that in FWM the transfer function of With the control parameter set to zero, the pulses are the input signal at frequency fs, to the output at narrow, as in Figure 2(a), and the frequency (2fs - fp) is intense. At other settings, the pulses are wider, as in Figure (2fs – fp) is quadratic – this takes care of the low- 2(b), and the frequency (2fs - fp) is much weaker. The input part of the transfer function in Figure 5(a). dotted line gives the pulse width. The data is given by the dots, with the curve following from theory.

Regenerating the signal Let us now take this a step further and discuss how the signal can be regenerated, rather than monitored. It is well known that this can be achieved through the use of a suitable transfer function. The transfer function of a device is the

Fig. 4: (a) Ideal transfer function, leading to no output Fig. 5: Schematic of an approximation to the ideal transfer

WAVES OF THE FUTURE OF WAVES unless the input power exceeds Ps, in which case the function in Figure 4(a). (b) Actual transfer function that is of output power is Pt. (b) Operation of the transfer function in the form in (a). Powers are expressed in dBm, which is a 84 (a) to the noisy signal from Figure 2(a). dB scale in which 1 mW corresponds to 0 dB. To get the tailing-off at high-power inputs, recall Now we saw in the previous chapter that the what the FWM actually does: two photons at essence of a fibre is that the light propagates in frequency fP disappear, generating photons at the core, and this cannot be done using just air frequency fs and at (2fP - fs). Clearly this process alone – we need refractive index jumps between must slow down once a significant fraction of the some outer layer and the core. The solution to photons at fP have disappeared. This is indeed so, this in the context of optical fibres was provided and it leads to the desired saturation effect at by Russell in the 1990s11. His solution does not high input powers. eliminate the glass altogether, but at least lets the light travel predominantly through air. Figure 7, Let us now turn to some practical implementation shows a cross-section of a photonic crystal fibre of this idea, which is shown in Figure 5(b)10, (PCF), and is thus the equivalent of Figure 5(a) of which shows a measured transfer function in a the previous chapter. The hole in the centre of FWM experiment. The measurement is similar to these figures is the fibre core, and the region with that described earlier, but not completely the the smaller holes in it acts as the cladding. same – however, that does not really matter here. It may not look very much like Figure 5(a), but that is because both axes are logarithmic (this tends to hide some of the differences in the curves). Nonetheless, at high input intensities the output clearly saturates, whereas at the lowest input powers, the output increases only slowly. Let us now take this transfer function and apply it to a noisy pulse as in Figure 6(a), which leads to the output in Figure 6(b). Clearly it has less noise both at low and at high intensities, as desired.

[FIGURE 6 HERE]

Fig 6: (a) Noisy optical pulse, and (b), regenerated optical pulse, using transfer functions as in Figs 5, with less noise at both high and low power levels.

Eliminating FWM Of course the best way to deal with FWM is to eliminate it completely, and the last topic I want to discuss deals with just that. We saw earlier that FWM is a nonlinear process, caused by the nonlinear response of the medium. It is Fig. 7: (a) Electron micrograph of a PCF. (b) Close up around the core region of the fibre in (a). interesting to point out that the strength of the nonlinear effects in silica glass is much smaller than in just about any other material; thus, trying Notice that, according to our argument in the last to reduce FWM by using a material other than chapter, such fibres could never guide light; total silica glass is doomed. The best way to do so, in internal reflection only works if the light travels TELECOMMUNICATIONS: LOOKING TO THE FUTURE fact, is to eliminate the glass altogether and let form a high-refractive index to a low-refractive the light propagate in air. index material. In a PCF the core has the lowest 85 refractive index, and total internal reflection thus the wavelength λ (or, actually, a multiple thereof, does not occur. Though this argument is perfectly but we’ll ignore this). We thus require correct, PCFs still guide light – to understand how it works recall that for a fibre to work it λ = 2¯n Λ. (3) needs some form of ‘mirror’ that prevents the light from escaping. In a conventional fibre the But if the light reflected off two adjacent periods mirror comes from total internal reflection; in interferes constructively, then the light reflected PCFs the mirror comes from Bragg reflection, off all periods does so. A periodic structure made which does not require the core to have the to the Bragg condition, equation (3), will therefore highest refractive index. reflect light very strongly. Though our argument applies to one-dimensional structures, it is Bragg reflection is a ubiquitous phenomenon that perfectly general: a wave propagating through relies on periodicity (as in the PCF cladding in any type of medium that is periodic undergoes Figure 7). It is most easily explained in a one- Bragg reflection – the condition for this to occur dimensional periodic structure, as in Figure 8, differs from (3) in general, but is essentially of the which shows the refractive index versus position same form. with period Λ. Let us assume that the refractive index jump ∆n is much smaller than the average Let us now return to PCFs, with their periodic refractive index n . This is not essential, but it cladding (Figure 7). It is now clear how it can makes the argument slightly simpler. Suppose we confine the light to the core. The cladding has to have a wave coming from the left, as indicated by be designed such that Bragg reflection occurs at the arrow. Now each time the light encounters a the wavelengths of interest. In fact, this is jump in refractive index, some of the light is somewhat trickier than the one-dimensional reflected. In the case we are considering here the treatment above might suggest – in a PCF you light reflected at each jump is minute, but that require Bragg reflection not just in one direction, does not matter. but in all directions: if Bragg reflection occurs in all direction but one, then the light can still seep out of the core, and confinement is lost. In some of the most recent work in this area, the cladding consists almost entirely of air, with some very thin glass membranes making up the fibre (as in Figure 7(b)). In these fibres around 99% of the light propagates in air, rather than in glass, which just about completely eliminates FWM and other nonlinear optical effects. For this and other Fig. 8: Schematic of Bragg reflection in a one-dimensional reasons, many groups around the world are doing periodic medium. The symbols are discussed in the text. research on the properties and fabrication of PCF.

Now consider the light that is reflected off two Conclusion jumps that are one period apart, and let’s work This completes my overview of some of the out under what condition these two contributions issues that are coming up in telecommunications, interfere constructively. To answer this, note that and some of the ways to deal with these. In these two contributions are the same in every closing I should give a few general comments. way, except for one difference: one of them The first is that I obviously only have been able to travels further, back-and-forth, by one period. scratch the surface and that many challenges Using the assumption that ∆n is small, we find exist that need to be solved (and they are being that the two waves have an optical path length solved!). Following from this, the general area of differing by 2¯n Λ. For these two waves to interfere telecommunications is not ‘finished’, and much WAVES OF THE FUTURE OF WAVES constructively, they must be in phase, which more research needs to be done. The overarching 86 means that the optical path lengths must differ by comment is the issues that we are dealing with here belong to general physics and mathematics underground, varying temperatures during the day lead to – telecommunications is brought to you by the small variations in the dispersion. basic sciences. 9 T.T. Ng, J.L. Blows, J.T. Mok, P. Hu, J.A. Bolger, P. Hambley, and B.J. Eggleton, ‘Simultaneous residual chromatic I am grateful to my colleagues Justin Blows and dispersion monitoring and frequency conversion with gain Ben Eggleton for discussions regarding some of using a parametric amplifier’, Optics Express 11, 3122- the issues discussed here, and I thank Trina Ng 3127 (2003). In fact, in this work the broadening of the for providing me with Figure 3. pulses is due to dispersion, rather than to FWM, however, this does not matter for us. Notes 10 Some relevant papers include E. Ciaramella and S. Trillo, 1 This means that the applied electric field should be small ‘All-Optical Signal Reshaping via Four-Wave Mixing in compared to the internal field of the atom. The latter can Optical Fibers’, Photonic Technology Letters 12, 849-851 be estimated roughly from Coulomb’s law for the hydrogen (2000), and S. Radic, C.J. McKinstrie, R.M. Jopson, J.C.

2 atom: E = 1/(4πε0)(q/a ). Here ε0 is the permittivity of Centanni, and A.R. Chraplyvy, ‘All-optical regeneration in vacuum, q is the elementary charge, and a is the Bohr one- and two-pump parametric amplifiers using highly radius, the average distance between the nucleus and the nonlinear optical fiber’, Photonic Technology Letters 15, electron in hydrogen. An applied field of this strength would 957-959 (2003). rip the atom apart. Though this is interesting, we are not 11 Such a fibre was first demonstrated in R.F. Cregan, B.J. interested in such fields here. Mangan, J.C. Knight, T.A. Birks, P.St.J. Russell, P.J. Roberts 2 In Equation (2) the factors 2π are included in the argument and D.C. Allan, ‘Single-mode photonic band gap guidance to make sure that the time T=1/f corresponds to a period of light in air’, Science 285, 1537-1539 (1999). of the wave. 3 The particular trigonometric identity that we require, a generalization of those that you may have seen before, is cos(a+b+c) = (cos(a+b+c) + cos(a+b–c) + cos(a–b+c) + cos(a–b–c))/4 4 Note that the discussion here is not the entire story. Apart from the frequency mixing we get other effects which can be gleaned by taking i = j = k = 1 and the signs such that

fi ±fj ±fk = f1. This is interesting: the nonlinear effect leads, indirectly, to one of the input frequencies. This nonlinear contribution has the effect of making the refractive index depend on the intensity of the light. 5 The references are: P.P. Mitra and J.B. Stark, ‘Nonlinear limits to the information capacity of optical fibre communications’, Nature 411, 1027-1030 (2001); J.B. Stark, P. Mitra, and A. Sengupta, ‘Information capacity of nonlinear wavelength division multiplexing fiber optic transmission line’, Optical Fiber Technology 7, 275-288 (2001). 6 Many other things of course also happen when light travels through a fibre. For example, even though the losses are small (see Figure 6 of the previous chapter), they do not vanish. However, as we saw in the last chapter, an amplifier can be used to boost the power. 7 In addition to this, electronics processing starts to become very cumbersome – the required equipment becomes very large, expensive and requires a lot of power. 8 In other strategies the dispersion is compensated in each WDM channel separately. But in this case the dispersion is TELECOMMUNICATIONS: LOOKING TO THE FUTURE required to be tunable since the fibre dispersion can vary with time. For example, even though the fibre is buried 87 Y-Chromosome: Waste- or Wonder-Land? By Dr Karl Kruszelnicki

IF YOU’VE FOLLOWED the news over the last few years, you’ll know that we humans have just about mapped our own DNA. Now the big difference between Boy- and Girl-DNA is that one of the chromosomes is different – boys have a Y-chromosome, while girls don’t. For a while the scientists thought that the tiny Y-chromosome was relentlessly headed for oblivion, and that it would be gone in a few million years. But the latest research shows that for all its faults, the Y-chromosome (and its nasty byproduct, boys) is probably here to stay.

Just about every cell in your body boys it’s tiny, and carries only 78 Chromosomes 45 and 46 provide a has the entire DNA needed to make genes. So guys, next time you go spare set of plans for each other. another you. Our DNA is lots of thinking that you are the Lord of The fertilized egg uses these plans things at once – it’s a blueprint for Creation, just remember that your as a blueprint as it grows itself into making a human being, it’s a pulsating masculinity is made a baby human over 9 months. history book of our ancestry, it’s a possible by just 78 genes. In fact, giant medical book, and it’s a you can think of women as the If one X-chromosome has a whole lot more. luxury model with all the options, damaged section, the growing while guys are the cheap economy embryo in the uterus can use an When a cell gets ready to split into model without the air-con, power undamaged section from the other two, the DNA also prepares to split steering and cruise control. X-chromosome as it builds itself. itself into two. The 2 to 3-metre length of the DNA bunches itself up Women have an advantage with But boys can’t do this. After all, into 46 little packages. We call their chromosomes 45 and 46 while their chromosome 45 is a big them “chromosomes” because as being big bold matching boofy X-chromosome, their early as 1848, the geneticists X-chromosomes. chromosome 46 is a minuscule would colour them with various Y-chromosome. The Y-chromosome dyes, to show up various features You can understand this advantage doesn’t have a matching partner. If – and “chromosome” means if you think about constructing a the Y-chromosome is damaged, it “coloured body”. building from the architect’s plans. can’t read the X-chromosome next Now suppose that the plans have door to find an undamaged section. If you look at these 46 been made harder-to-read by chromosomes with a microscope, smudges, burger stains and spilt So the Y-chromosome looks like a you’ll see that 45 of them are the coffee. It would be very hard to disaster waiting to happen. And same in boys and in girls, but that complete your building with these indeed, boys suffer quite a few the 46th chromosome is different. damaged plans. It would be very diseases (such as red-green WAVES OF THE FUTURE OF WAVES It’s quite large in girls where it handy to have a spare undamaged colour-blindness) that girls 88 carries about 1,500 genes, but in set of plans. don’t get. But the news gets worse for the pathetic little Y-chromosome.

It mutates very rapidly, about 1,000 times faster than any other chromosome. It seems that it was virtually identical to the X-chromosome about 300 million years ago. In fact, there are a few short sections on the very ends of the Y-chromosome that are identical to the very ends of the X-chromosome – but over 95% of the Y-chromosome is very different from the X-chromosome. It has mutated massively. Over the last 300 million years, the once-proud Y-chromosome has shrunk from about 1,500 genes to its current miserly 78 genes. And if it keeps shrinking and mutating at the present rate, it’ll be totally use-less in just a few million years.

Yup, for a while there, the scientists really thought that the Y-chromosome was a regular wasteland.

But in June 2003, Nature published some reassuring research by Dr. Page from the Massachusetts Institute of Technology, which It is very exciting that sections of turned the Y-chromosome from the Y-chromosome are WasteLand to WonderLand. He and palindromes. It means that if one his team worked out, after a lot of section of the Y-chromosome is very hard work, that the damaged, sometimes it can find an Y-chromosome has a special trick, undamaged version of the same which means it doesn’t need a section somewhere else in the Y- matching partner to cover for any chromosome. It can ignore the damaged sections (or bits). damaged section, and use the undamaged section somewhere up They discovered that large sections at the other end of the Y- of the Y-chromosome are chromosome. “palindromes”. You might remember from your high-school But while this is good news, it is English days that a “palindrome” is also a little disturbing. Yes, the Y- a word, or phrase, that reads the chromosome can fix itself same in each direction. The word up – but only by having sex Y-CHROMOSOME:WONDER-LAND? OR WASTE- “radar” is a palindrome, as is the with itself... FROM Dr Karl’s book Bum Breath, Botox word “level”. and Bubbles (Harper Collins Publishers) 89 Illustration courtesy Adam Yazxhi LIDIA MORAWSKA is a Professor at the School of Physical and Chemical Sciences, Queensland University of Technology (QUT) in Brisbane, Australia, and the Director of the International Laboratory for Air Quality and Health at QUT (ILAQH, which is a WHO Collaborating Centre of the World Health Organization on Research and Training in the field of Global Burden of Disease due to Air Pollution). She conducts fundamental and applied research in the interdis- ciplinary field of air quality and its impact on human health and the environment, with a specific focus on science of airborne particulate mat- ter. Professor Morawska is a physi- cist and received her doctorate at the Jagiellonian University, Krakow, Poland for research on radon and its progeny. Prior to joining QUT she spent several years in Canada con- ducting research first at McMaster University in Hamilton as a Fellow of the International Atomic Energy Agency, and later at the University of Toronto. Dr Morawska is an author of over hundred fifty journal papers, book chapters and conference papers. She has also been involved at the executive level with a number of relevant national and international professional bodies and has been acting as an adviser to the World Health Organization. She is the immediate Past President of the International Society of the Indoor Air Quality and Climate. 90 The Science of the Aerosols we breathe

Professor Lidia Morawska

Introduction WAITING FOR A CHANGE of traffic lights at a busy intersection in any large city we can smell vehicle exhaust; standing on a hill above the city we can often see brown haze blanketing the urban site at our feet; travelling on a plane we pass through pollution plumes from an urban metropolis, industrial chimneys or forest fires. In situations like these we may contemplate whether air pollution is a new phenomena of the industrialised world, whether nations are gaining control over the problem of the pollution – or whether in fact it is getting worse.

91 irborne pollution was reported long ago as viruses, bacteria or pollens, when in the air, a nuisance problem in ancient Roman belong to the world of airborne particles. The A cities, and has continued through the scientific complexity of the world of atmospheric development of modern civilization ever since. Yet aerosol particles is depicted by the many different the impact of pollution, nor the necessity to terms that are used to describe or identify them. combat it, was not realised until recent times; Some of the terms identify particles by their it is only in the last fifty years that it has became sizes, others by the processes that led to their obvious that pollution introduced to the air by generation and some by a particle’s ability to humans has a significant negative impact on enter human respiratory tract. human health and the environment. Since then, the subject has became the focus of an Considering even the basic parameter of particle increasing number of scientific studies conducted size, there are various classifications and worldwide, and with new knowledge emerging it terminologies used to define particle size ranges has also became evident how complex, and how for the purpose of scientific discussion or scientifically and technically challenging this practical applications. The division most interdisciplinary field is. commonly used is between fine and coarse particles; however, the boundary between these Over the last decade, one specific aspect of two ranges is somewhat arbitrary and has been airborne pollution has been increasingly defined differently by various authors in relation preoccupying scientists as well as politicians: the to different aerosols and applications. The division study of airborne particles, particulate matter, and line often used in aerosol science and technology more specifically very small particles. The is somewhere between 1 and 2 µm. This is the physics, chemistry and biology of these particles range of the natural division between smaller – which can vary in size by up to a hundred particles, which are generated mainly from thousand times between the smallest and the combustion and other process leading to gas to largest and are involved in multitudes of reactions particle conversion, and larger particles generated and interactions – are not only fascinating topics from mechanical processes. Obviously any such for scientific studies, but ones with important division is somewhat arbitrary, as nature itself practical implications. This chapter aims to does not provide a perfect division. introduce the scientific world of airborne particles and explain how such particles affect human The terminology that has been used in the heath and the environment. wording of the ambient air quality standards, and also in characterization of indoor and outdoor particle mass concentrations includes PM2.5 and The Complex World of Atmospheric PM10 fractions. PM2.5 (or fine particles) is the Aerosol Particles mass concentration of particles with aerodynamic In scientific terminology, airborne particles are diameters smaller than 2.5 µm, while PM10 refers called aerosols – these are defined as an to the mass concentration of particles with assembly of liquid or solid particles suspended in aerodynamic diameters smaller than 10 µm a gaseous medium long enough to enable (more precisely, the definitions specify the inlet observation or measurement. Airborne particles cut-offs for which 50% efficiency is obtained for range in diameter from about 0.001 µm, to about these sizes). Total suspended particulate (TSP) is 100 µm, with the lower limit describing the size the mass concentration of all particles suspended of large molecules and the upper limit the size of in the air. Other size classifications of particles particles that deposit very quickly due to may be into submicrometer and supermicrometer gravitational force. particles, which are particles smaller than and larger than 1 µm, respectively, and ultrafine Particles can be characterized by their physical particles, which are smaller than 0.1 µm properties, such as size, shape or size WAVES OF THE FUTURE OF WAVES distribution, but also by their chemical 92 composition or biological nature. For example Where do Aerosols come from? Smoke – a solid or liquid aerosol, the result of While the above classification of particles incomplete combustion or condensation of considers only their sizes, usually a particle’s size supersaturated vapour; usually considered the is a consequence of the process that led to its precursor of smog generation. Particles in the submicrometer range (smaller than 1 µm) are generated mainly by The above classifications are an indication of the combustion processes such as burning timber, many different processes leading to generation of smoking a cigarette or combusting fossil fuel in particles with different physical and chemical vehicle engines, and by processes which result in properties. It is interesting to note that while in gas or vapour molecules combining to form a scientific language these terms clearly identify the new particle (gas-to-particle conversion, mechanisms by which the particles were nucleation processes or photochemical generated, in common everyday language they processes). Particles in this size range typically are used loosely to identify the presence of contain a mixture of components including soot, particle matter in the air without, however, acid condensates, sulfates and nitrates, as well attaching to them their scientific meaning. as trace metals and other toxins. Larger particles, Physical Properties of Aerosols in the supermicrometer range, result mainly from mechanical processes such as cutting, grinding, The most important physical properties of aerosol breaking and wear of material and dust particles include: number and number size resuspension, and contain largely earth crustal distribution, mass and mass size distribution, elements and compounds. surface area, shape, and electrical charge. To a large extent these are the physical properties of From the above it can be concluded that some of particles that determine particle behaviour in the the particles in the air originate from direct air air and, ultimately, removal from the atmospheric emissions, while others are formed in the systems. The efficiency of various forces acting atmosphere by the chemical reactions of gases, on particles and processes to which they are particularly sulphur dioxide, nitrogen oxides, subjected in the air depends strongly on the ammonia and volatile organic compounds. The particles’ physical properties, of which size is one former are called primary, and the latter are of the most important. secondary particles. Health and environmental effects of particles are Particle sources are also indicated in the strongly linked to particle size, as it is the size commonly used descriptive terms for airborne that is a predictor of the region in the lung where particles, and in particular: the particles would deposit, or the outdoor and indoor locations to which the particles can Dust – solid particles formed by crushing or penetrate or be transported. Also sampling of other mechanical breakage of a parent particles and choice of an appropriate material, larger than about 0.5 µm (see below instrumentation and methodology is primarily for definition of house dust) based on a particle’s physical properties. Next we’ll Fog – liquid particle aerosol formed by discuss aerosol shape and size in more detail. condensation or atomization Fume – particles that are usually the result of Aerosol Shape and Equivalent Diameter vapour condensation in low-temperature combustion processes, with subsequent Particles vary significantly in shape, with some of agglomeration, usually smaller than 0.05 µm the shapes fairly simple and regular; however the Smog – an aerosol consisting of solid and majority display a varying degree of irregularity or liquid particles created, at least in part, by the complexity. In general, shape relates to a action of sunlight on vapour triggering particle’s formation process or its origin. Particles resulting from coagulation and agglomeration of photochemical reactions and generating THE SCIENCE OF THE AEROSOLS WE BREATHE secondary pollutants smaller solid particles are usually highly irregular. Examples of these are particles resulting from 93 combustion process such as vehicle emissions This process is schematically presented in Figure (which are agglomerates of carbonaceous 2. Particle equivalent diameter is the diameter of spherical particles), or tobacco smoke. Similarly, a sphere having the same value of a physical dust particles and particles resulting from property as the irregularly or complex shaped mechanical processes typically have irregular particle being measured. Equivalent diameter shapes. In contrast, liquid aerosol particles are relates to particle behaviour (such as inertia, usually spherical, while simple fibres are rod electrical or magnetic mobility, light scattering, shaped. Biological particles are complex in shape radioactivity or Brownian motion) or to particle and differ significantly between various types. In properties (such as chemical or elemental Figure 1, microscopic images of different types of concentration, cross-sectional area or volume to particles are presented. surface ratio). Therefore the particle diameter determined experimentally depends on the choice of particle properties or behaviour measured.

Figure 1: Particles collected in residential houses in Brisbane, Australia, and examined with an energy- dispersive X-ray analyser attached to a transmission electron microscope. (a) The three types of particles in this Figure 2: Representation of particles of irregular shape by picture are: a square particle (NaCl crystal), many big means of particle equivalent particle diameter. fibrous particles (dominant elements: Mg, Cl, S, O), and many fine fibrous particles (dominant elements: Ca, S). (b) Particles of the same type with the dominant elements: Mg, Application of different methods for measurement Cl, S, K, Na. (c) Small NaCl particles (crystals) and a long of particle diameter usually results in somewhat particle with dominant elements: C, O, Ca, Mg, S, P different values of the diameter obtained. (possibly a fragment of an insect or a plant). (c) Dry drop. Therefore understanding of which particle The dominant elements of the big particles are Cl, K, Na and of the small ones Mg, K, Cl, S, O, Na, Si. (Images from diameter is actually determined in a particular Morawska and Salthammer, 2003.) study is important for interpretation of the results and for comparison between different studies. Particles of complex or irregular shape can be The most commonly used equivalent diameter is characterized by many parameters, but for the aerodynamic diameter, the diameter of a unit- practical applications these are usually reduced density sphere having the same gravitational to one or two parameters that can be measured. settling velocity as the particle being measured. Most commonly these are particle diameter, or for fibres, their length and width. Diameter is a characteristic of spherical objects; however, as Aerosol Size and Size Distribution explained above only a small fraction of airborne As explained above, particles in ambient air are particles are spherical. Therefore a way of mixtures generated by a large number of sources representing particles of irregular shape has been including motor vehicles, power plants, WAVES OF THE FUTURE OF WAVES introduced by means of a particle’s equivalent windblown dust, photochemical processes, 94 diameter. cigarette smoking, nearby quarry operation, and so on – and particle size is dependant on particle particles in the range of 5 to 10 µm. Outdoors, formation processes. For example the diameter of processes such as wear on vehicle tires, road particles produced by various indoor and dust resuspension or coal processing result in combustion sources is very small and typically airborne particles typically in the size range of may lie in the ranges (according to source): for several micrometers. Figure 3 presents the large gas burners, gas ovens, and toasters in the order variation in the sizes of airborne particles. of 0.01 to 0.02 µm; for frying and grilling in the range from 0.05 to 0.2 µm; natural gas, propane, Almost all sources generate particles with some and candle flames generate particles between distribution of particle sizes (so called 0.01 and 0.1 µm; for cigarette smoking in the polydisperse aerosol) rather than particles of a range from 0.01 to 0.1 µm; and incense burning single size (monodisperse aerosol). A very about 0.1 µm. interesting phenomenon related to this distribution of sizes is that particles generated by Particles generated by outdoor combustion most sources have a log-normal size distribution sources are also generally small. A significant – this means that the curve of particle proportion of diesel emission particles have concentration versus particle size is ‘normal’ (bell diameters smaller than 0.1 µm. Gasoline particles shaped) when the particles are plotted on a are mostly carbonaceous spherical logarithmic scale. When a single pollution source submicrometer agglomerates ranging from 10 to is investigated and when it operates under steady 80 nm. Particles from CNG emissions are smaller conditions (for example steady parameters of the than those from diesel or even petrol emissions combustion process), the size distribution of and range from 0.01-0.7 µm, with the majority particles obtained is likely to have one distinctive being between 0.020 and 0.060 µm. The peak, and sometimes additional, usually much majority of particles emitted from biomass smaller peaks. These peaks are called modes of burning, which includes controlled burning and the distribution. uncontrolled fires (as well as burning for the purpose of heating and cooking) are ultra fine, Different emission sources are characterised by with only a small fraction in the larger size range different size distributions, and while these and with most of the mass present in particles distributions are not unique to these particle less than 2.5 (m in aerodynamic diameter. sources alone, the information on the size distribution can help to identify their contribution In contrast, particles generated by mechanical to particle concentrations in ambient air, and also processes are larger. For example in indoor serve as a source signature. environments, walking, moving one’s arms, and even sitting in front of a computer produces Figure 4 presents examples of size distributions of particles generated by different combustion processes. Figure 5a presents an example of a typical, urban, air particle number size distribution measured in Brisbane, Australia.

Particle distributions can be presented either in terms of number or mass distributions. In terms of number, the vast majority of airborne particles are in the ultrafine range. For example, in urban outdoor air where motor vehicle emissions are a dominant pollution source, over 80% of particulate matter in terms of number is in the ultrafine size range. The total mass of the ultrafine particles is, however, often insignificant THE SCIENCE OF THE AEROSOLS WE BREATHE in comparison with the mass of a small number Figure 3: Large variation in the sizes of airborne particles. of large particles, with which most of the mass of 95 Figure 4: Examples of particle size distribution spectra of environmental tobacco smoking (ETS), coal, petrol and, diesel. WAVES OF THE FUTURE OF WAVES Figure 5: Typical measured urban air particle number size distribution (a), and mass distribution (b) calculated from the 96 number distribution. airborne particles is associated. Particle surface of organic and inorganic compounds associated area in turn is largest for particles somewhat with the particle ‘body’ (adsorption/desorption, above the ultrafine size range. This relationship water solubility, extractability) is of interest. The between particle number and volume is most important chemical properties of particles presented in Figure 5a and b – the relationship include: was derived using measured particle number size Elemental composition distribution (Figure 5a) and calculating particle Inorganic ions volume distribution assuming their sphericity Carbonaceous compounds (organic and (Figure 5b). Particle mass can be calculated from elemental carbon) the volume when particle density is known or can be assumed. It can be seen from Figures 5a and In general, interest in the elemental composition b that the peak in the number distribution derives from the potential health effects of heavy spectrum appears in the area where there is elements like lead, arsenic, mercury and almost no volume in the volume distribution cadmium, and the possibility of using the spectrum, and vice versa, the peak in the volume elements as source tracers. Water-soluble ions distribution spectrum is where the particle such as potassium, sodium, calcium, phosphates, number is very low. sulfates, ammonium and nitrates associate themselves with liquid water in indoor The form of presentation of particle size environments and can also be used for source distribution used in Figures 4 and 5 is quite apportionment. common, but it is somewhat simplistic, as it does not properly reflect the logarithmic nature of the Carbonaceous compounds are composed of distribution. The proper and most common way of organic and elemental carbon. The former can presenting particle distributions is by plotting (in contain a wide range of compounds such as logarithmic scale) particle number, surface area polycyclic aromatic hydrocarbons, pesticides, and volume respectively, per logarithmic interval phthalates, flame retardants and carboxylic acids, of size. some of which are tracers for certain sources, while the latter is sometimes termed ‘soot’, ‘black carbon’ and ‘graphitic carbon’. Chemical and Biological Properties of Aerosols Another important aspect of airborne particles is The chemical composition of particles is multi- their biological properties. Individual bacteria or factorial and depends, as discussed already, on pollens suspended in the air are examples of particle sources as well as post-formation biological particles. But bacteria or pollen spores processes. For example, some types of particles can be attached to other, non-viable particles, like asbestos and glass fibres consist of inorganic such as smoke or dust, which become carriers of materials, while other types like cellulose fibres biological particles. are purely organic. In many cases, the behaviour

Table 1: Sources of biological particles

Outdoor sources Indoor sources Surfaces of living and dead plants (fungal spores, Occupational environment where organic materials bacteria), soil are handled Natural and anthropogenic waters such as sewage lagoons or Agriculture (processing of products) cooling towers (bacteria) Microbial growth in buildings (heating, ventilation and air Aerosolation of water conditioning systems, building structure) Building exhaust and sanitary vents Ornamental fountains, showers Humans Pets THE SCIENCE OF THE AEROSOLS WE BREATHE Hospital procedures Indoor plants 97 There are different classes of biological particles, almost immediately after becoming airborne), which include many different species, varying not while smaller particles are removed by only in terms of their biological composition but precipitation or diffusional deposition. also size. In general, viruses range from 0.02 to 0.3 µm, bacteria from 0.5 to 10 µm, fungi from 0.5 to 30 µm, pollen 10 to 100 µm and house Aerosol Detection Methods dust mites are about 10 µm. Biological particles Measurement is a means of detecting or, in other can originate from a number of indoor or outdoor words, seeing the airborne particles and learning sources that can be broadly classified as about their composition. Depending on which presented in Table 1. measuring technique is used, the particles are seen somewhat differently. For example, as explained above, an irregular particle can be Aerosol Dynamics described by a number of diameters, called Following formation or resuspension, airborne equivalent diameters, which relate to the physical particles undergo a range of physical and method used for approximating the irregular chemical processes that change their chemical shape as a regular, spherical object. There are composition, physical characteristics and many different measurement and sampling concentration in the air. The most important of techniques for airborne particles and these processes include: coagulation, which understanding the principles of operation of the results from Brownian motion and collision of various techniques is of importance not only to particles, mainly of similar sizes; deposition of those who design and conduct experimental smaller particles on the surface of bigger studies, but also to anyone involved with particles; changes to particle size due to changes analysing and interpreting the data generated in moisture content (the latter could be through the experiments. hygroscopic growth or shrinking by evaporation); sedimentation and deposition on surfaces. The In general, the experimental techniques for complexity of assessing the role of individual measurements of particle characteristics are processes or quantifying their rates relates to the complicated and expensive. Such measurements fact that all of them take place simultaneously, should cover a broad range of particle sizes from affect differently particles of different size ranges nanometers up to at least several micrometers. and are dependent on a large number of factors However, due to the different physical properties and characteristics of indoor environments. of particles in this size range, various methods and different instruments have to be applied for Some emission products, for example those that comprehensive measurements of particle are combustion related, are highly dynamic characteristics. Investigations of the particle mixtures of hot gases and particles undergoing number size spectrum yield information about the rapid changes; others, like mechanical dust, are relationship between particle number and less so. Particles measured away from the diameter. This information is usually lost in emission site, or particles generated indoors and commonly applied measurement methods measured some time after emission, will have targeted at the determination of the TSP (Total different characteristics to those measured Suspended Particles), PM10 or even PM2.5 immediately after formation. fractions, especially for submicrometer particles. As explained above, the larger particles contribute The residence time of emission products in the strongly to the total mass of the air particles, but air depends on the nature of the processes they often the number of such particles is several are involved in, and varies in outdoor air from orders of magnitude lower than the number of seconds or minutes to days or weeks. Larger small particles. particles (of a micrometer size range in aerodynamic diameter and more) are removed Measurements of particles can be conducted by WAVES OF THE FUTURE OF WAVES from the atmosphere mainly through gravitational active or passive sampling followed by application 98 settling (with particles above 100 µm settling of appropriate analytical techniques for analyses of the material collected through sampling. distribution of viable and non-viable particles Passive sampling means that the particles measured in a piggery. It took only two minutes to deposit on the sampling medium through conduct this measurement and obtain this graph, gravitation, diffusion or other natural processes. which is significantly faster than using classical By contrast, active sampling means application of microbiological techniques. pumps or other devices for drawing a certain volume of air through the sampling medium. Passive sampling is less commonly used [FIGURE 6 HERE] nowadays for particles, as its outcomes are associated with large uncertainties.

Measurements of airborne particles can also be performed using in-situ methods where the sample is temporarily captured by the instrument, which measures some particle characteristics in real time. In this case there are no further analytical techniques required. Many characteristics of particles can be measured in real time, including particle mass or number concentration, particle number size distribution and surface area. Some of these methods provide Figure 6: Size distribution of viable and non-viable particles a direct measure of the property investigated; measured in a piggery. others measure a parameter, which is then translated into the property of interest. For example, optical counters directly count particles Aerosols in Outdoor Air crossing the sensing area of the instrument, while Particle concentration levels in environments microbalances measure changes in the oscillating which are not influenced by human activities frequency of a crystal on which particles are (clean environments) are usually of the order of a sampled, and translate the change of the few hundreds particles per cm3. In urban frequency into the mass collected. Not all particle environments, background particle number properties can be measured in real time, and in concentrations range from a few thousands to general, while there are readily available methods about twenty thousand particles per cm3. for real time measurements of many physical Background concentrations refer to the properties of the particles, there are only very few concentrations measured at monitoring stations, such methods for measurements of chemical or which are not influenced by a nearby emission biological properties. source. Near roads and in tunnels, vehicular traffic constitutes the most significant urban However, there are new methods being developed source of particle numbers. Here particle number and new instruments becoming available, so it is concentrations can be more than ten times likely that in the future it will be possible to higher than in other urban environments and can measure more particle properties in real time. For reach or exceed levels of 105 particles per cm3. example, the time-of-flight spectrometer enables This is in contrast to PM10 and PM2.5 mass measurements of the size and chemical concentrations, which at such roads have been composition of individual aerosol particles in near shown to be no more than 25 to 30% above real time, while an ultraviolet aerodynamic background levels (calculated as the difference particle sizer (UV-APS) measures the fluorescence between the maximum at the road and the characteristics of individual particles in an aerosol background levels). sample, which makes possible real time identification of biological aerosol particles, as When considering emissions of particles into the THE SCIENCE OF THE AEROSOLS WE BREATHE distinct from inanimate particles. As an example, air, the formation of new particles in that air, the Figure 6 presents a diagram of the size resulting airborne particle concentration, and 99 ultimately the impact that these particles have on local and global impacts of emissions. In most human health and the environment, different urban environments, motor vehicle emissions are spatial scales have to be examined. In particular, the main anthropogenic source of air pollution it is important to distinguish between local air and specifically particles, significantly contributing pollution in magnitude and impact, and global air to the deterioration of urban air quality. Since pollution and its effects. Global air pollution population density is higher in urban than in rural impacts are related to the total emissions on a areas, the number of people exposed world wide continental or worldwide scale and include the to elevated levels of vehicle emissions is greenhouse effect, effects on climate and enormous. The impact of vehicle emissions on ozone depletion. ambient urban concentrations of particles and gaseous pollutants could be considered in terms The effects of local air pollution are local of two spatial scales: environmental problems (for example visibility Large scale, which is the total urban air reduction), but most important are the impacts on shed, and which means contribution of human health. Figure 7 presents a situation of vehicles to the background urban dense brown smog over a central European city concentration levels of pollutants, and located in a valley surrounded by mountains Small scale, which is the close proximity to where, in addition to vehicle emissions, coal and a road, where the concentrations are wood burning for heating purposes during winter elevated above the urban background result in significant local air pollution. The global levels scale of pollution resulting from combustion processes (vehicles, domestic heating and Estimates of the total vehicle emission levels, cooking, burning of fossil fuels for industrial and which include the vehicle emission inventory and energy generation processes) can be appreciated the total contribution of vehicles to the urban when flying over continents and seeing large background concentration of pollutants, are areas under similar brown smog, extending to the derived using transport and traffic models with areas beyond urban metropolises. the relevant vehicles’ emission factors. The small spatial scale of vehicle emission impacts is concerned with the areas adjacent to the roads carrying considerable amounts of traffic, or road intersections and other traffic congestion areas. In order to assess the impact of the emissions from a nearby road on air quality in the neighbouring buildings, the following aspects need to be taken into account: Total vehicle flow on the road and the speciation of the flow into individual vehicle classes in terms of vehicle size and the fuel on which they operate Emission factors of individual classes of vehicles Variation of vehicle flow with time of the day

Figure 7: Dense brown smog over a central European city caused by vehicle emissions, and coal and wood burning It has been shown that while spatial distribution for heating purposes during winter. of pollutants in the urban environment appears to be highly homogenous based on the data from air While it is evident there are many significant quality monitoring stations, closer analyses of sources of ambient airborne particles, vehicle pollutants’ dispersion and transport reveals that WAVES OF THE FUTURE OF WAVES emissions in particular will be discussed here in for gaseous and particle number concentrations 100 more detail to illustrate the differences between there is a high level of heterogeneity displayed, with significantly elevated concentrations in the 1) The physico-chemistry of aerosols immediate vicinity of roads. Particle number 2) The anatomy of the respiratory tract concentration, like the concentration of gaseous 3) The airflow patterns in the lung airways pollutants and other surrogates for very small particles, decreases significantly with the distance In relation to the physico-chemistry of aerosols, from a road. the forces acting on a particle and its physical and chemical properties such as size or size Decay in particle concentration was approximated distribution, density, shape, hygroscopic or by exponential curves in a number of studies and hydrophobic character and chemical reactions of it was shown that the impact of a road on particle the particle will affect its deposition. With respect number concentration, while significant in the to the anatomy of the respiratory tract, important immediate vicinity of the road, is not parameters are the diameters, the lengths, and distinguishable past about 300 metres from the the branching angles of airway segments, which road. It was shown that dispersion is the main determine deposition. Physiological factors factor responsible for the decrease in particle include airflow and breathing patterns that further concentration with distance from the road. influence particle deposition.

A practical implication from these findings is that Particle penetration and deposition in the the exposure to gaseous pollutants and number respiratory tract is, to a large extent, governed by concentration of particles emitted by vehicle particle size. Large-size particles deposit mainly traffic on roads is significantly increased within in the upper part of the respiratory tract due to the distance of the first 100 to 200 meters from impaction, interception, gravitational the road, compared to the urban average sedimentation, as well as turbulent dispersion. exposure levels, and reduces to the urban Very small particles that can follow the gas background level at distances larger than about streamlines, such as those generated through 300 to 400 metres from the road. On this basis, combustion, can penetrate to deeper parts of the it is reasonable to assume that people living and respiratory tract processes, and have a high working in close proximity to an urban arterial probability of deposition there due to their high road will likely be exposed to levels of ultrafine diffusivities. and submicrometer particles beyond what could be considered ‘normal’ ambient levels. The understanding of the mechanisms of particle deposition in the human respiratory tract and the This is a finding that needs to be taken into ability to quantify the deposition in individual parts consideration in future urban land and transport of the respiratory tract is of principal importance planning. The situation is somewhat different in for assessment of human intake of particles due relation to particle mass, which is not so strongly to inhalation, which can then be used for risk elevated close to roads compared to the assessment. Over the last three decades or so, a background level. The reason for this is that large number of studies have been conducted to newer vehicle technologies result in lower investigate particle deposition in the human emissions of particle mass and in addition, there respiratory tract, with a somewhat larger number is little dust generated from modern, sealed focused on theoretical modelling than on the roads. experimental determination of the deposition. Figure 8 presents the results of recent experimental studies conducted at the Aerosol Inhalation and Deposition in the International Laboratory of Air Quality and Health, Lung Queensland University of Technology, Brisbane, The main mechanism for intake of airborne showing the dependence of deposition of diesel particles by the human body is inhalation of particles in the human respiratory tract on particles and deposition in the respiratory tract. particle size, and also individual differences THE SCIENCE OF THE AEROSOLS WE BREATHE Factors influencing the deposition of inhaled between people. particles can be classified into three main groups: 101 Figure 8: Size dependent deposition of diesel smoke particles in the lung of fourteen volunteers for the study.

Health Effects due to Exposure exposure to airborne particulate matter, to Aerosols particularly of small sizes and at lower Inhalation of airborne particles has been shown concentrations, and the health effects they to be detrimental to human heath due to their cause. Both fine and ultrafine particles appear to toxic or carcinogenic properties, but also affect health outcomes such as mortality, and because they act as irritants respiratory and cardiovascular causing discomfort and affecting morbidity, and appear to do so general human well being. A independently of each other. large number of epidemiological A large number of However, the database at studies conducted in cities in epidemiological studies present is too limited (both in different parts of the world have numbers of studies and linked daily mortality statistics conducted in cities in numbers of subjects) and with increased particle different parts of the geographically restricted to concentrations measured allow clear conclusions on the outdoors (epidemiology is the world have linked daily mode of action or generalization science concerned with the mortality statistics with to other settings. All of the study of disease in a general studies demonstrate that the population, and determination of increased particle primary determinant of the the incidence or rate of concentrations ... effect of ultrafine particles is occurrence and distribution of a their number and their surface particular disease by age, sex, area and not the weight of or occupation, which may provide information particles present. This means that the traditional about the cause of the disease). An increase of 1 use of particle mass weight measures is to 8% in deaths per 50 (g µm-3 increases in inappropriate in evaluation of the likely biological outdoor air particle mass concentrations has effects of ultrafine particulates. been a common conclusion from these studies.

WAVES OF THE FUTURE OF WAVES The mechanisms by which the particles induce There is still, however, only limited information the range of health effects are still hypothesised, 102 available in terms of quantitative links between but not fully proven. The size of airborne particles determines in which parts of the respiratory tract Future Directions in Aerosol Research the particles are deposited: small airborne In summary, despite the significant increase in particles less than one micrometer in diameter national and international efforts towards (submicrometer particles) have a high probability extending our understanding of various aspects of deposition deeper in the respiratory tract and related to airborne particles, there is still only are likely to trigger or exacerbate respiratory limited understanding of the production rate, disease. Small particles also have higher burdens airborne concentrations and composition in of toxins which, when absorbed in the body, can different environments, the theory of particle result in health consequences other than dynamics, the fate of very small particles in the respiratory health effects. air, on the exposure-response relationship and on the mechanisms by which particles induce a Effects on Climate Change range of health effects. Global warming has been linked to the One important reason for the current difficulties introduction into the air of substances that result relates to the scientific complexity of in trapping heat in the atmosphere, the so-called investigations on airborne particulate matter at all greenhouse effect. Different substances have levels of approaches, including: instrumentation, different effects on global warming, expressed in measurement and modelling, model validation, terms of global warming potential. For example, data interpretation, exposure assessment and risk over a 100-year time frame, nitrous oxide is 310 quantification. All these aspects require times more effective than carbon dioxide at considerably more scientific knowledge and trapping heat in the atmosphere. understanding, and will constitute the future directions for research in this broad, There is growing recognition that the very small interdisciplinary field. particles also contribute to the greenhouse effect, but the degree of contribution has not yet References and Further Reading been established. There are many challenges in Baron, P.A and Willeke, K., 2001, Aerosol Measurement: quantification of particle contribution to the Principles, Techniques, and Applications, (Wiley: New York, greenhouse effect. For example, calculation of USA), ISBN 0471356360 (cloth). the potential of atmospheric aerosols resulting Morawska, L and Salthammer, T., 2003, Indoor Environment: from motor vehicle emissions for causing earth’s Airborne Particles and Settled Dust, (Wiley-Vch: Weinheim, climatic change through radiative forcing cannot Germany), ISBN 3-527-30525-4. be done in the same way as for greenhouse IPCC, 1995, Summary for Policymakers: The Science of gases. This forcing is comparable, but of Climate Change – IPCC Working Group I, opposite sign to, the radiative forcing due to http://www.ipcc.ch/pub/sarsum1.htm. greenhouses gases (IPCC, 1995). However, unlike greenhouse gases, the atmospheric Web-based resources aerosol is not uniformly distributed about Aerosol Related Home Pages: http://www.aaar.org/hplinks.htm the globe.

In summary, as stressed in the report of the Scientific Committee on Problems of the Environment 2000: Measurements of tropospheric and stratospheric aerosol particles, the vast majority of those of anthropogenic origin coming from combustion sources, is thus of critical importance for developing an understanding of climatic effects of particles THE SCIENCE OF THE AEROSOLS WE BREATHE and establishing effective ways for minimising the effects. 103 Einstein’s Miraculous Year /Part 1

Albert and the Electrons – The Photoelectric Effect THE MORE YOU think about it, the less likely it seems: a patent clerk, an unknown in science, submits his first academic paper to the Annalen der Physik (Annals of Physics). For a first effort from an unheard-of scientist, you might expect something simple, something naïve, something mundane and forgettable.

But no – Albert Einstein’s first But there were experiments that the wave’s frequency was, as long scientific paper in March 1905, A couldn’t be explained this way. In as it was big enough – in other heuristic point of view concerning particular, there was the words, the photoelectric current the production and transformation photoelectric effect. The idea is should depend on amplitude of the of light (Annalen der Physik,17 this: shine light onto a polished light (that is, how bright the light pages 132-148), earned him the metal surface and you can cause is), but not on the frequency (that Nobel Prize for Physics in 1921. electrons to jump off the surface. is, the light’s colour). Set things up in the right way and At the heart of this paper lie the you can use this to produce an This all seems straightforward ... very nature of light and matter, and electric current – this is the except Nature had other ideas. If the interactions between them. principle behind some solar cells you do the experiment, it is easy to Prior to 1905, the debate about and light-detectors. show that, first, the number of whether light was a wave or a stream of particles seemed to have been decided. Isaac Newton, two centuries earlier, had proposed that light was really comprised of tiny particles, like pellets from an air gun. However, physicists such as Hooke, Huygens, Euler, Poisson, Fresnel and, later, Faraday and Maxwell showed that light had various properties that were better explained if light is considered a wave. In particular, when light passes a fine edge, or through a thin slit, it forms a complex pattern Now, light waves could knock electrons knocked off does depend of light and dark bands called an electrons off a metal surface – on the frequency – the higher the interference pattern. Interference is certainly ocean waves are capable frequency of the incident light, the a property easily demonstrated with of bashing things around. So what more electrons you get off the other kinds of waves, like water would you expect if light waves metal surface. But this is only true waves or sound waves – and so were responsible for the above a certain cut-off frequency; light, it seemed, was also, photoelectric effect? You’d expect below this cut-off, no electrons are

WAVES OF THE FUTURE OF WAVES conclusively, a kind of wave too, that the larger the wave, the more ejected at all. And here’s the weird because that explained many of the electrons would be knocked off the thing – this is true no matter how 104 experimental results. surface. It shouldn’t matter what big the wave is. It’s a bit like a huge ocean wave amounts of energy, and so the smashing into a sand castle, yet electron will never have enough when the water washes away, the energy to break free – no matter castle is intact. how many photons there are!

Einstein thought about this and Einstein used the photon idea to decided there was a better way to explain the photoelectric effect. So look at this problem – you have to light is made of particles, right? Not stop thinking of light as a wave, so fast, there – remember, other and think of it as a particle instead. properties of light, like interference, If light consists of little bundles, or are explained by the wave model, photons, of energy, he said, and if not the particle model. So it seems

that energy increases in proportion light isn’t a wave, or a particle ... to the frequency of the light, then it somehow, strangely, it’s both. all makes sense. An electron will This first paper of Einstein’s in only ever absorb one photon at a 1905 paved the way for time – if that photon has enough developments in the Quantum energy (that is, if the light is high Theory, which has become one of

enough frequency), it will knock the modern science’s most spectacular EINSTEIN'S MARVELLOUS YEAR, 1 PART electron off its atom in the metal achievements. The Nobel Academy surface. But if the frequency is too thought the effort worthy of the low, the photons each have small Nobel Prize for Physics in 1921. 105 What got you interested in IT in the first place? Near the end of high school, I began hearing about computers and decided that they were going to an important part of our future world, JUDY KAY is an and that I wanted to be part of that. Associate Professor at the School of Information What were you like as a kid? Were Technologies at the you curious, pulling apart stuff to University of Sydney. see how it worked? Were you She is a principal in always interested in science/IT, or the Smart Internet did this come on later? Technology Research I delighted in learning about most Group, which conducts things. I recall a few bad experiences both fundamental and in taking something apart and being applied research in unable to reassemble so steered user-adapted systems. away from that sort of thing. The core of her work is to represent and manage What’s the best thing about being personalisation ensuring the user a researcher in your field? maintains control while also being A chance to take a small part in able to scrutinise and control the inventing the future, or influencing it. whole process: the user can determine what is modelled about Who inspires you - either in IT or them, how this is managed and how in other areas of your life? it is used. People who come up with elegant, She applies this in ubiquitous, simple and powerful ideas. pervasive computing as well as intelligent teaching systems. If you could go back and specialise in a different field, what would it be and why? Ah! If I’m allowed to replay my life once, well why not twice or more? In that case, I would have loved more time for music, art, history, ... I would work my way through all of them systematically.

What’s the ‘next big thing’ in IT, in your opinion? What’s coming up in the next decade or so? Lots of computation throughout our lives, in every endeavour, making real changes to the way we do and see things.

106 Creating and overcoming invisibility: scrutably personalised ubiquitous computing

Associate Professor Judy Kay

Introduction MOORE’S LAW DESCRIBES the observed exponential growth in computing power, with the capacity of new chips doubling every 18 to 24 months for more than forty years. This applies to processing power, main memory and discs: computers have become dramatically more powerful and cheaper, and this trend appears likely to continue. This has meant that we now have many special purpose computers embedded in cars and home appliances.

107

Figure 1: SharPic coffee table interfaces his trend will continue, with increasing 2-sigma improvement in learning that Bloom numbers of invisible computers in every observed: the result that a learner who is at the T venue from the home to workplace, working 50th percentile in typical classroom settings can quietly and unobtrusively. We are just beginning achieve at the 98th percentile with expert to explore the meaning of invisible computing, personal tuition. To move closer to this, we need also described by Weiser, the much quoted father better, personalised learning support from of this emerging area, as ubiquitous or calm computers; they should be within our normal technology. environment, so that when and where we have a learning need, we can consult our personalised One part of the vision for invisible computing was teacher. described by Nicholas Negroponte, who envisaged butler-like agents within our There is a huge amount of computer science environment: research needed to achieve these visions and much of the technical progress needs to be ... computer surrogates that possess a body of integrated with human, social and legal demands. knowledge both about something (a process, a This chapter explores two sides of invisibility: field of interest, a way of doing) and about you some of the current research exploring how to in relation to that something (your taste, your create invisibility and, in anticipation of achieving inclinations, your acquaintances). Namely, the that, explorations of ways to overcome computer should have dual expertise, like a invisibility’s problems. cook, gardener, and chauffeur using their skills to fit your tastes and needs in food, planting, and driving. (Negroponte 1995, p151). Creating Invisibility One invisibility goal is for computational Such computers should look after your needs resources to fit so well into our environment that without bothering you unduly, just as a good they feel like a natural part of it: we will not even butler should. Negroponte also described a think of them as computers. This contrasts with, personal newspaper: and complements, current personal computers with their screen, keyboards and mice. Although Imagine a future in which your interface agent these are becoming increasingly common in can read every newswire and newspaper and workplaces and homes, they are quite different catch every TV and radio broadcast on the from the computers in cars, washing machines, planet, and then construct a personalised ovens and the like. We are barely aware of summary. This kind of newspaper is printed in computers integrated into appliances. They an edition of one. (Negroponte 1995, p153) simply make those appliances increasingly convenient and useful (at least when everything The envisioned invisible computing will provide works properly). The challenge of this invisibility that personal news report in very flexible modes; goal is to build systems so that they are useful, it should be delivered as audio if you are driving work well and do not require any effort to and need to watch where you are going, or it may learn anything. be in almost conventional print that appears on your coffee table. Of course, this all relies on a A somewhat different, and conflicting, invisibility computer learning what you like, want and know goal is that the computers in our environment are as well as how you want information presented. invisible in the sense that we can do the things we want to without needing to think about the Another part of the vision is described by the computer. An expert user of a personal computer Computer Research Association in their 2002 does this when they type an essay, being able to Grand Research Challenges in Computer Science focus on the task of writing, barely thinking about and Engineering over the next twenty years. They the computer at all – they can be in the flow of WAVES OF THE FUTURE OF WAVES identified the goal of ‘A Teacher for every writing the essay and, in a sense, the computer is 108 Learner’. This strives to achieve the famous almost invisible. One problem is that the expert only gets to this level after making a real effort to Another part of the same project is the prototype learn and practice using those computing tools. digital scrapbook being used by the person in the The pay-off for this work is that the power user middle back of Figure 2 – this uses the Anoto can achieve so much more. pen and paper systems (www.anoto.com). The research explores natural pen and paper-based interfaces for a scrapbook. It allows people to The coffee table interface create a physical scrapbook with photographs This is a prototype for a future coffee table where and annotations written with them, rather like any people interact with digital images naturally other scrapbook. However, it also enables the without conventional computer interfaces. Figure user to record audio that is associated with the 1 shows two people using this coffee table’s pages of the book and, on revisiting the SharePic application to collaboratively work with scrapbook, the user can replay it. The system photographs. At the top left of the table you can also synchronises the physical scrapbook with an see a triangular red area beneath the images. online version. So, for example, I could create a This is the personal space of the user at the right scrapbook with pictures and associated text and side of the table. He has just dragged a picture audio. Then it appears on the web, available for into that personal space: he simply placed his relatives overseas. finger near the middle of the image and then slid it along toward the red personal space. Once an image is in his personal space, only he can drag it out.

At the right middle of the table, the swirling image, the black hole, swallows any picture that comes near it. Near the black hole is a small blue rectangle, ‘the frame’. It is like any other image on the table: you can enlarge or shrink it by putting a finger at a corner and moving outwards or inward. You can rotate it by putting a finger in a corner and sliding around an axis through the centre of the image. (And you can do both at once, in a rosize gesture.) However, the frame is also special in that if you put it over any other images and allow your finger to dwell in the Figure 2: The scrapbook pen interface is used by the middle, it creates a new picture of whatever is person at the middle back, while other people use SharePic. In the background is a large 3-D wall, which is under it. So, for example, if it is over a corner of used for fly through visualisations and other activities, another image, you get a cropped part of that like games. PERSONALISED UBIQUITOUS COMPUTING image. If it is over a set of three images, the frame creates a composite of those images. The Anoto pen works with conventional paper SharePic works by projecting onto the special that has a special pattern on it. The pen has a table, which can detect where the user has placed camera that enables it to capture what the user their finger or fingers. It can distinguish which has written. It uses Bluetooth wireless person is touching the screen. SharePic was part communication to send this to the processing of a research project to support remembrance part of the intelligent environment. So, for activities. An important motivation for the work is example, if the user draws the audio recording the potentially important role of reminiscence shape, they can then record audio for that point activities, based on the huge amounts of digital in the scrapbook and later, by ticking that memorabilia that people will own. The research location, they can replay the audio. Of course, CREATING AND OVERCOMING INVISIBILITY: SCRUTABLY aims to help people build personal histories and this requires a microphone and sound system share memories with others. within the intelligent environment. A similar 109 process can be used to associate video with parts of the scrapbook. Figure 3 shows an example of a notebook page with audio and video.

Figure 3: Example of a page from the scrapbook.

Magic Mirror and Keep-in-Touch (KiT) The KiT project supports families to Keep in Touch. The research is multi-disciplinary and Figure 5: Magic mirror display when a family member activates it. aims to help distributed members of cross- generational families to easily maintain contact. It provides message-based asynchronous Family members can interact with KiT by using communication between the members of a hand gestures to play messages and make new family. The prototype normally looks like a mirror. ones. For example, the gesture of moving your For strangers, it blends into the environment, as a hand down selects the grandmother’s image and normal mirror in a typical house, as in Figure 4. then another gesture starts the recording of a However, if a family member approaches the message. The gestures are intended to be easy mirror, the screen behind the half-silvered mirror to learn and to customise (and they work without becomes active. anyone putting finger prints on the mirror). KiT delivers messages to a similar appliance in the Figure 5 shows one KiT prototype, with a grandmother’s Queensland home. She can tell distributed family – most of the family lives in when there is message as a red dot appears on Sydney, one child is in England and a the screen for each message she has not played. grandparent lives in Queensland. Other prototypes use different displays that look like There are many other possibilities for useful framed pictures, and KiT also operates on ambient displays, like the orb shown in Figure 6. conventional computers. This glows in different colours and light patterns to present digital information, like changes in the [FIGURE 4 HERE] stock market. For example, the family that uses the KiT also has an orb in their Sydney home; this glows green when their child in England is online, with the orb signal enabled – that way, they know when to try to get in touch. Other examples include art works and even subtle light dots. WAVES OF THE FUTURE OF WAVES

110 Figure 4: Magic mirror in mirror state. you have forgotten the magic words? The augmented reality phone (AR phone) is a prototype that tackles this problem. At the same time, it might be useful for much more mundane invisibility problems – for example, if you are in a room with a printer and it seems not to be working, the AR phone could deliver additional information to help you overcome the problem.

It needs special so-called fiducial markers like the large symbol at the right of Figure 7 – Figure 8 shows what happens when the phone recognises one of these. The AR-phone has a camera, like Figure 6: Ambient display of information with the glowing many current phones; the user points this at the orb. fiducial marker. The phone sends this image to a system, which processes the image and The examples above have focused on ways to recognises the marker. It then looks into its interact with machines that might be integrated collection of information and images associated into homes. This is the user’s view of research with this marker, and displays the scene with the systems that will help create the invisible future. additional retrieved information. In the case of There is considerable technological challenge Figure 7, this additional information is a picture of under the hood, with research needed on a teapot. networks and operating systems. One of the real challenges is to make appliances that are ‘zero- configuration’ devices (or Zeroconf – see [FIGURE 7 HERE] www.zeroconf.org), meaning that when you buy your magic mirror, you can install it at your home as easily as current consumer electrical devices, like toasters or TVs.

Overcoming Invisibility – building cloaks of visibility In parallel with work toward invisible computing, there is exploration of the expected problems that will come with it. For example, you might come to a new place and wonder what services are PERSONALISED UBIQUITOUS COMPUTING available there. Or, you might be surprised by something that the intelligent environment does, and wonder why that happened. You might also observe that the same environment treats your brother differently, and you might wonder why [FIGURE 8 HERE] that happens and what you can do about it. We now describe two examples of research towards these challenges of invisibility.

Augmented Reality Phone Suppose you are in a room that has a hidden

Figure 7: Augmented reality phone and the fiducial marker CREATING AND OVERCOMING INVISIBILITY: SCRUTABLY door. Suppose that you are allowed to open that it recognises, with the image associated with this marker door by saying the magic words. What happens if (the teapot). 111 Figure 8: Architecture of the AR-phone. Figure 9: User models, users, programmers and the world.

With a device like the AR phone, it is easy to input to the user model: for example, the user personalise the information to be delivered. So, for may answer questions about their preferences. example, if you are not supposed to know about a An important thing to notice is that the user hidden door in the wall in front of the camera, the model is an artificial model, intended to information presented may be a description of the effectively capture a tidy, useful but dramatically painting on the wall that is there as well. On the simplified view of the real world, which is other hand, if you are allowed to use the door, the exceedingly messy. phone might give you a clue to the magic words – for example, if the magic word is teapot, the A personalised system is driven by the user model image in Figure 7 should help you remember the that holds the system’s beliefs about the user. words to open the door. There are many reasons for enabling people to scrutinise personalisation and the user model itself. Essentially, this is a matter of ensuring that when Scrutably personalised museum tour we build future personalised ubiquitous computing At the core of any of personalised systems is a environments, we want to ensure that people are user model in which the computer holds still in control. This is a fine balancing act: on the information about the user. Figure 9 shows how a one hand, the whole invisibility enterprise aims to user model is related to the ensure that people need not world, the person it models and bother with technical details and the programmer(s) who build the the technology will simply do the systems. The top of the figure “... on the one hand, the right thing, as generally happens shows the real world; in the case whole invisibility with the many existing appliances of our emerging invisible with embedded computers. On computing environment, this enterprise aims to the other hand, it is immensely world has many sensors that can ensure that people need irritating when systems are too collect information about a smart and make mistakes: it is person. Some of the information not bother with technical even worse if the user cannot from these can flow straight into details and the work out just what is happening, the user model. Most goes along and why it is happening, and how the path shown at the right, technology will simply do to get back in control. In the case being interpreted by programs, the right thing ... On the of user model in the intelligent which use it to draw conclusions environment, there are also about the user. As the figure other hand, it is legislative requirements related to indicates, the programmer’s own immensely irritating the use of personal information. understandings, goals and prejudices are inherently built when systems are too Figure 10 shows an example of into their code. The figure shows what a user can see when they WAVES OF THE FUTURE OF WAVES smart and make that the user also interacts with mistakes ...” decide to scrutinise the 112 the world and may provide direct personalisation provided by a museum guide for the Nicholson Museum at the To the right of the display, there is a description University of Sydney, Australia’s first museum. This of the parts of the user model – if the user places display enables this user to see information that their mouse over one of personalised parts of the was not presented to them in the normal screen, they see the ‘mouse-over’, as in this operation of the system: this has a yellow figure, explaining why the material was omitted. background. The information with a grey Of course, in the normal mode, the information is background was presented because of this user’s simply presented as in any personalised web model. Other parts, with no background, are page – and, then it looks rather like the many presented to everyone. personalised sites on the internet. These use cookies to connect each visit to a web site from the same computer. The web site can build up user models or profiles: people are generally unaware of what is personalised, and often cannot easily find out.

Your own avatar guide to a personalised museum The PEACH project is exploring ways to personalise museum tours. Figure 11 shows one part of PEACH that is working towards helping help people figure out where they look in a rich environment. The female figure at the left is projected onto the wall Figure 10: Scrutably personalised information delivery. of the museum and moves around, leading the user’s eye towards relevant parts of the environment. In this case, she is standing next to a large display that presents information. The figure Figure 11: Personal tour guide avatar at left and personal information delivery avatar at right. at the right is an artist: he presents information PERSONALISED UBIQUITOUS COMPUTING CREATING AND OVERCOMING INVISIBILITY: SCRUTABLY

113 Figure 12: Personalised information delivery that exploits only to those visitors who want an artistic emphasis large displays for common information and individual in their tour. displays for personalised extras.

An interesting part of the PEACH work is to (SIV) helps with a visualisation like that in Figure explore ways to deal with groups of people – 13. This user model could well be derived from people do tend to visit museums in groups and, one of the many movie-recommender web sites for that matter, they spend much of their time at that build up models of what a person likes or home, work and elsewhere in groups as well. does not like, based on collections of information Figure 12 shows three people at a large screen about other people’s preferences. Such sites are display with audio information delivered by the an example of another group of inscrutable, but artist figure at the right of the display. PEACH uses personalised, systems that are widely available. the large display for common information; this is for all members of this group. In addition, each If a recommender were enhanced with a SIV person has a personal device, with personalised display, like that in Figure 13, you could easily see information. In this case, an invisibility problem that most of the titles displayed are green, occurs since people may not know when and indicating the user model represents them as where to look. The PEACH project uses animated ‘liked’ by this user. The red titles are ones modelled avatars to help with this too. as ‘not liked’, and Great Catherine stands out in red. The movie that is currently selected is Lawrence of Arabia – all the titles that are in larger Scrutinising large user models print are more similar to this film than those shown Suppose you wanted to know about your user in smaller print. The visualisation was designed to model, as used in a complex system. You need to give a quick overview of a user model so the user WAVES OF THE FUTURE OF WAVES deal with the problem of seeing a large amount can then select parts of interest to scrutinise in 114 of information. Scrutable Inference Visualisation more detail. Conclusion Negroponte, N. (1995). Being Digital, Hodder and Stoughton. This paper has presented a user’s view of a selection of research projects in the area of Donald Norman, The invisible computer, why good products personalised ubiquitous computing. Some involve can fail, the personal computer is so complex, and hardware that is still novel but is likely to be a information appliances are the solution, Cambridge, Mass. precursor of appliances of the future. In many of MIT Press 1998. the projects, much of the research explores a range of deep technical issues as well as the Weiser, M. (1991) The Computer for the Twenty-First Century, pragmatics of how these systems will fit into Scientific American, 94-10. people’s lives. In this brief overview, it has not been possible to describe these; nor has there Web sites to explore been the space to analyse the impact this will www.ambientdevices.com/cat/index.html have on every aspects of home, work and leisure Examples of current products that are ambient devices life. There are extraordinary possibilities for invisible computers, as well as the conventional www.media.mit.edu visible ones, to support activities in all professions MIT Media Lab and other aspects of our lives. oxygen.lcs.mit.edu Acknowledgements MIT Project Oxygen: pervasive human-centred computing Several of the projects described in this paper were funded by the Smart Internet Technology home.cc.gatech.edu/ubicomp Cooperative Research Centre. Georgia Tech Ubiquitous Computing Research Group

www.ubiq.com/hypertext/weiser/UbiHome.html References A summary of foundation work on ubiquitous computing with Assad M, D J Carmichael, D Cutting, A Hudson. (2003). links to major papers by Mark Weiser. Accessible Augmented Reality in the Intelligent Environment. Proceedings of OZCHI, pp 268-287. www.doc.gov/ecommerce/eudir.htm

Bloom, B. S. (1984). The 2-sigma problem: The search for peach.itc.it/consortium.html methods of group instruction as effective as one-to-one PEACH Personal Experience with Active Cultural Heritage tutoring. Educational Researcher, 13(6), 4-16. Consortium PERSONALISED UBIQUITOUS COMPUTING

Figure 13. Visualisation of a CREATING AND OVERCOMING INVISIBILITY: SCRUTABLY large user model for a person’s movie preferences 115 Mexican Wave By Dr Karl Kruszelnicki

IF YOU’VE EVER been to, or watched a major sporting event, you’ll probably have seen the famous Mexican Wave. This wave sweeps around the audience in a stadium or sports arena, as firstly one group of people leap to their feet with their arms up and then sit down and then the group of people next to them does the same thing, and so on. If you’re on the other side of the stadium looking across, you can see this beautifully rhythmic and synchronised movement rolling through the audience. On a good night, you can see multiple waves winding their way around the terraces. Some scientists have studied this strange phenomenon, and not only do they now understand it, they can probably do something useful with this knowledge.

This so-called “Mexican Wave” first Now mathematicians have made But it was a bit of a surprise to find became famous during the 1986 mathematical models of much of that this theory of “excitable media” Soccer World Cup in Mexico. In the world around us, and some of could deal with something as fact, that’s how the Mexican Wave them did mathematical models of complicated as human behaviour – got its name, because it got its first what are called “excitable media”. but then again, we are talking world-wide exposure at this event about a Sports Event. – and soon enough, there was a One example of “excitable media” swell of interest in far-away Europe. is the dry trees and dry litter in a You can think of a person as being forest. This particular theory of an excitable unit – and that’s not Scientific work was done by Támas “excitable media” describes how a so unusual. This simply means that Vicsek and his colleagues from the forest fire starts, and then spreads. a person can be prompted into Eötvös University in Budapest in action by some sort of external Hungary, and the University of Another example of “excitable stimulus. In general, the closer and Technology at Dresden in Germany. media” is the set of muscles that more powerful the stimulus is, the They video taped, and then make up your heart tissue. When more the human being, as an analysed, 14 separate Mexican they’re given a little jolt of excitable unit, would respond. So to Waves in football stadia each electricity, the muscles in your explain how a Mexican Wave can holding more than 50,000 people. heart will contract. It turns out that get started, and keep rolling, each They noticed that the wave there are a quite a few electrical person needs only three internal generally went in a clockwise abnormalities of the heart that rules that they obey, one after the direction – that it spread from one involve problems with the other. First, they wait in their resting person to the next person on their excitability of the heart tissue – state, ready to be excited by the left. The Mexican Wave usually either it’s too excitable, or it’s not right stimulus. Second, when moved at around 12 metres (or excitable enough. In fact, the theory stimulated, they go through the 20 seats) per second, and was of “excitable media” can also active phase where they stand up WAVES OF THE FUTURE OF WAVES about 9 metres wide (about 15 predict that some people will have and wave their arms. Third, they go 116 seats wide). extra heart beats. to the refractory phase, which is FROM Dr Karl’s book Bum Breath, Botox and Bubbles (Harper Collins Publishers) Illustration courtesy Adam Yazxhi

what you and I would call “sitting small group of agitators tried to So if you do see, or are involved in, down again”. get a large crowd over-excited. a Mexican Wave, this research also tells you that you’re not getting You now might say that this The scientists did find a good your money’s worth – because the research is incredibly useless. But relationship between their theory, game is boring, and the crowd are it is important with regard to and what they actually saw in real entertaining themselves... crowd control. When you have life. Basically, you need a minimum 100,000 people at a sports event, critical mass of two or three dozen simply moving them in and out of people to get the wave going. And the stadium can be dangerous if even that’s not enough – you need it’s not done wisely. Even a lull happening in the sports something as simple as putting a event. After all, if the football game hand rail along the centre of a or athletics event is incredibly corridor can speed up the riveting, the audience is not going movement of people in or out of a to pay attention to the person next MEXICAN WAVE stadium. So knowing this theory of to them suddenly jumping up. excitable media could help if a 117 David, what got you interested in science in the first place? I was always interested in ‘how things worked’, and the encouragement of teachers at PROFESSOR DAVID school was important in fostering COCKAYNE FRS this interest. has been a Professor in the Department of What’s the best thing about being Material at the University a researcher in your field? of Oxford since 2000. You get paid to pursue your hobby. Before that he was a Professor at the Who inspires you – either in University of Sydney. science or in other areas of He studied your life? undergraduate physics When I was younger, I was inspired at the University of by older eminent scientists who Melbourne, and he knew so much more than I did. Now carried our research at I’m older, I’m inspired by young the University of Oxford for his D Phil. students who know so much more In 1999 he was elected to the Royal than I do. Society for his pioneeering work in electron microscopy. If you could go back and specialise in a different field, what would it be? Today it would be aboriginal medicine; yesterday it was international law; tomorrow it might be legal aid; and then there’s how the memory works, and carpentry and school teaching and ... Why those fields? Because of the enthusiasm that people I come in touch with have for these topics (my children, in three of the cases).

What’s the ‘next big thing’ in science, in your opinion? What’s coming up in the next decade or so? The next big thing in science needs to be a significant development in renewable energy sources.

118 Seeing in the Nanoworld

Professor David Cockayne FRS

Introduction ADVENTUROUS YOUNG people in the 18th and 19th centuries travelled the world, exploring unknown countries, and seeing wonderful and strange sights. When they first landed in Sydney where we are today, they discovered not only a new country, but also an amazing number of animals never before seen by Europeans. When they wrote of their discoveries, many of their countrymen did not believe them. Indeed, when the skin of a platypus was sent back to England, scientists thought it was a fraud.

119 owadays every corner of the earth has This close relationship between diffraction and been ‘discovered’, and we might think that imaging has been understood since the work of N there’s nowhere left to explore. But there Abbe in the late 19th century. He showed that if is – scientists are the explorers of the 21st light of wavelength λ is scattered by an object, century and the nanoworld is waiting to be then the information carried by that scattered explored. And just as Cook and Magellan light can tell us about the details of that object, depended upon specially designed ships for their but only down to a level of detail d, where adventures, so scientists rely upon advanced instrumentation and techniques to explore the d = 0.5λ [1] nanoworld. And just as the early explorers saw strange and unfamiliar animals, plants and land This means that light of wavelength 500 nm, say, formations, so the scientist-explorer is finding a will carry no information about any feature of an wealth of unfamiliar objects – from bucky balls to object that is less than 250 nm in size. At the end nanotubes, and carbon nanotrees to of the 19th century, this was of great concern to nanomatches. In this lecture, we follow the scientists (and indeed to at least one poet (Belloc development of the most important tool for 1912)), because it meant that much of the exploring the nanoworld – the electron microscopic world would never be visible. It microscope – and we discuss the close wasn’t that scientists of those days didn’t have relationship between diffraction and imaging. good light microscopes – they had perfect microscopes – the problem was that the information simply wasn’t there in the scattered The relationship between diffraction light. and imaging To understand why we can’t see the nanoworld This question of how much information can be with our naked eyes, or even with a light carried by a wave is easily understood if you think microscope, we must consider the relationship about the following experiment. You are standing between diffraction and imaging. We see diffraction on the sea shore, looking at the waves coming in in various forms all the time – in the colours of smoothly from far out at sea. Suddenly the waves butterfly wings and of opals, and in some aspects start to arrive with froth in them. Seeing this of a rainbow. So diffraction is not unfamiliar to us. disturbance, you realise that there must be some And we can clearly see the relationship between object over the horizon that is causing this diffraction and imaging if we carry out a simple disturbance. Could it be a small rowing boat? No, experiment of firing a laser through a grid of wires. because a small rowing boat would just bob up On a distant wall, the pattern we see is an array of and down on the wave, without causing any froth. spots; a hexagon if the grid of wires is hexagonal, It must be something bigger – perhaps a whale and a line of spots if the grid is a set of parallel or a ship. wires. This array of spots (a line or a hexagon) carries information about the wires, displayed in a So it is only if the object disturbs the wave that particular way – the diffraction pattern, as we call you will know it is there – and you can imagine the pattern on the wall. If we now place a lens that something smaller than about half the between the grid of wires and the wall, the same distance between the wave tops (for example, a information can be displayed in a different way – rowing boat or a small fish) won’t disturb the an image. This image can be in focus or out of wave. Of course, whether or not you can focus, depending upon exactly where we place the differentiate between the froth from a whale and lens. But the important point to realise is that it is the froth from a ship is a question of how the same information in the scattered light which intelligent you are – it isn’t a problem that the results in these patterns, the images and the waves can solve for you! And it’s the same with diffraction pattern. As we shall see, sometimes it is light. If the object doesn’t disturb the light, the more useful to study the information as an image, image or the diffraction pattern won’t show WAVES OF THE FUTURE OF WAVES and sometimes it is more useful to study it as a anything about it. 120 diffraction pattern. The situation improved with the discovery of X- can be used to determine the shape of the rays (Roentgen, 1895), since they can have a repeating object (the grid wires in our case). So if wavelength of 1 nm or less. Equation 1 shows we use X-rays or electrons, the diffraction pattern that the scattered X-rays would then carry can be used to study the arrangements of atoms information at the level of interatomic spacings, in crystals. about 0.3 nm. As a result, X-ray diffraction (the study of the scattered X-ray intensity as a But now we should reflect on the next step. As function of the scattering angle) has been a we have seen, X-ray diffraction existed, but (until powerful tool for investigating the structure of very recently) there were no X-ray lenses and so materials down to the atomic level for over half a no images with X-rays – you could not see century. It is a major tool in chemistry, materials objects much smaller than the wavelength of science mineralogy and molecular biology. light. There are no lenses for water waves. None for gamma rays. It was not inevitable that there But the discovery of X-rays did not lead to seeing would be lenses for electrons – not at all. But it objects in the nanoworld because, although the was known that electrons can be deflected by a scattered X-rays carried information at the magnetic field, and in 1926, Busch showed that nanoscale, no lens was available to form an the path of electrons through a magnetic field is image. similar to that of light through a glass lens, i.e. a magnetic field has a lens action. Things might have remained there – resolution in imaging at 0.25 micron with light, and at the Very quickly electron lenses were made, and put atomic level with X-ray diffraction. But then de together to construct an electron microscope. For Broglie (1924) showed that electrons could be this, Ruska received the Nobel Prize (in 1986, considered as waves, with their wavelength some 50 years after his invention!). The given by arrangement of the lenses is identical to that for a light microscope, with magnetic fields replacing λ = h/(mv) [2] the glass lenses, and fast electrons replacing light. How good were these electron lenses? (h is the Planck constant, and m and v the mass Absolutely terrible – about as good a lens for and velocity of the electron respectively). For electrons as the bottom of a Coca Cola bottle electrons travelling at 75% the speed of light, this would be for light. In fact electron lenses are gives a wavelength shorter than that of X-rays. such terrible lenses that they hardly deserve to The experimental proof of this hypothesis resulted be called lenses at all. But the reason they are from experiments by Davisson (with Germer and used is because if they were perfect, we would Kunsmann) in 1927 (and by Thomson at the be able to see detail about half the wavelength of same time, but with a different kind of the electron – about 0.002 nm, or 1/100th of the experiment), which demonstrated that periodic size of an atom. So with terrible lenses, perhaps objects (crystals) give rise to diffraction patterns we can see an atom. And if we didn’t have for which the distribution of diffraction spots is electron lenses – and as I’ve said, there’s no given by the Bragg equation inevitability that they would have existed – then we would not have seen the structure of cells, or 2d sin(ϑ/2) = nλ [3] their organelles, we would not have seen bucky balls or carbon nanotubes, and certainly not where d is the repeat pattern spacing and n is an atoms. integer. (This same equation can be used to describe the distribution of light spots caused by So at this stage, scientists had electron diffraction, the scattering of the laser by the grid of wires we needing no lenses, and electron lenses (even considered earlier.) If we know the light or X-ray though they were terrible) to form images at or electron wavelength λ, and we measure the resolutions previously unachievable. And so they SEEING IN THE NANOWORLD angles θ, then we can determine the spacing d. had the possibility of both imaging and diffraction In addition, the intensities of the diffraction spots with electrons. Just like excited explorers given 121 new kinds of ships, what new explorations could One of the most important advantages of electron they carry out with these new microscopes? We diffraction over X-ray or neutron diffraction is that can certainly see objects below the limit of the it can be carried out in an electron microscope light microscope such as viruses and molecules, from a region of sample chosen from within the small catalyst particles and the components of image. That is, we can compare the diffraction modern computers and DVDs. And with the best pattern from small regions of the same sample, microscopes, we can even see atoms. and so use selected area diffraction as a means for characterising differences in structure within a sample. Because X-rays and neutrons have no Diffraction with electrons lenses (although X-ray lenses are now under One of the first proofs that electrons had a wave development), X-ray and neutron diffraction nature (as given by Equation 1) was in a series of patterns are necessarily average patterns from experiments by Davisson and his coworkers, who the entire sample. scattered electrons off a nickel target, and looked at the diffraction pattern. The pattern (Figure 1) was very uninteresting! But one night the Different ways of seeing apparatus exploded, and the entire experiment Transmission Electron Microscopy was destroyed. They put it back together again, Since most of you will know the construction of and heated the sample to a high temperature the light microscope, the best way to explain the because it had itself become destroyed. And construction of a transmission electron when they repeated the experiment, the results microscope (TEM) is by analogy with the were completely different – the diffraction pattern transmission light microscope (LM). If we turn the now showed maxima of intensity at angles given light microscope upside down (see Figure 2), then by the Bragg equation (Equation 3). And because the Bragg equation describes scattering by i. At the top we have a source of light – a waves, this provided proof of the wave nature of heated wire. This wire emits both light and electrons. This kind of luck is called serendipity, heat, and perhaps a few electrons. If the wire and is an essential attribute for successful is made of tungsten (W) it will give off heat and scientists! (The word serendipity was made up by light, but also a lot of electrons (because of its the writer Horace Walpole in 1754, from a fairy low work function). So we use a W wire in a tale in which three princes from a country TEM as the source of electrons. (Increasingly, Serendip – an old name for Sri Lanka – made other brighter sources, such as lanthanum accidental discoveries about things that they were hexaboride crystals and field emission sources, not looking for.) are being used.)

ii. In the LM, the light travels at the speed of light (!) towards the sample. In the electron microscope, the electrons that come off the W wire are ‘thermal’ electrons, and are not travelling very fast. We want them to travel fast, because we want them to have a short wavelength so that they will carry information about the small details of objects (see Equation 1). And since the wavelength of an electron is related to its velocity (Equation 2), we accelerate it across a large voltage difference (30,000 to 1 million volts). The electrons, like Figure 1: The scattering curves from nickel as reported by the light in the LM, then travel rapidly towards Davisson and Germer before and after their apparatus the sample. WAVES OF THE FUTURE OF WAVES exploded. 122 iii. In the LM, a glass lens (the condenser lens) concentrates the light and directs it onto the sample. As we have seen, glass does not act as a lens for electrons, but a magnetic field does. So for electrons we use a magnetic field as the condenser lens, to concentrate the electrons onto the sample.

iv. The sample scatters the light (electrons), which is then collected by the objective glass (magnetic) lens to form an image (always a real image in the case of the TEM). In the case of the LM, this image is a variation in light Figure 2: A comparison of a modern TEM and a transmitted intensity across the field of the image. In the light microscope. case of the TEM, it is a variation in electron intensity across the field of the image. lens (moving it in the case of the light v. This image is then treated as an object, and microscope, and changing the current through imaged once more into a final image by a the windings in the case of the TEM). series of lenses (the eyepiece or projector lenses) onto a photographic film in the case of a LM. Similarly, for a TEM, the image is formed Scanning electron microscopy (SEM) on photographic film or onto the screen of a There is a second way of forming images that camera. For the LM, the final projector lens can does not depend upon having a physical imaging be our eye, and the final screen our retina. If lens after the sample. If we use a lens to we use our eye and retina for the final lens and concentrate the electrons into a very small spot screen for the TEM, they will be destroyed! (say 3 nm in size) on the surface of a solid sample, they can be scattered backwards from vi. Even though the electrons are travelling very the sample surface and collected on a detector (a fast, they are not massive, and so they are back scatter detector). Their number is used to stopped by about a centimetre of air. control the brightness of a dot on a cathode ray Consequently we must evacuate the electron tube (CRT). The spot is then scanned across the microscope. Inserting the sample requires sample surface in a raster, and the dot on the passing it through an air lock, to maintain the CRT is scanned in synchronism. vacuum. If there are variations in the sample surface (e.g. vii. Because the electrons are so easily different elements at different positions, different stopped, the sample for an electron topology) then the number of back-scattered microscope must be very thin, typically less electrons will vary with position, and so the than 1 micron in thickness. intensity on the CRT will vary. In this way a backscattered (BS) image is formed. At the same The lens action in the electron microscope comes time, some of the electrons penetrate the from the magnetic field of the lenses, which is surface, and electrons (secondary electrons, SE) formed by passing electric current through a wire are ejected from the sample. A different detector (windings) which are wrapped around the can count these secondary electrons, as a cylindrical metal lens. Consequently the strength function of the spot position, and so form a SE of the electron lens (its ability to magnify) can be image on the CRT. varied by adjusting the current through the windings. This has obvious advantages over a The SEM has a much greater depth of field (that light microscope, where each lens has a fixed is, the depth over which the object remains in SEEING IN THE NANOWORLD magnifying power. Just as in a light microscope, focus) than the TEM or the LM, and consequently the image is focussed by adjusting the objective the SEM is very useful for investigating the 123 surface of objects which have a complicated and our brain puts these together into a 3D topography, as in Figure 3. Because the electrons image. So the question is, how can we obtain two in a transmission EM (TEM) pass through the images in the electron microscope? sample, the TEM is used to look at the internal structure of the sample, while the scanning EM It is not difficult – we take two pictures with the (SEM) looks at the surface (or the near-surface). sample in slightly different orientations, and then The resolution of the SEM is limited by the size of show one image to our left eye and the other to the incident electron spot, and how much it our right eye, simultaneously. We can view them spreads out inside the sample when it generates with a special viewer (a stereo-viewer), or we can the detected signal. cross our eyes, or we can print one image in green and the other in red, and use filtering glasses. We can go further than this: we can Seeing in 3D present the two images to a computer, and let it The electron microscope techniques we have reconstruct the 3D object for us. Then on the discussed so far give us two-dimensional images; computer screen we can rotate it, view it from a but since many objects are three-dimensional, we chosen direction, and analyse it in full detail. would like to be able to see in 3D. How do we normally see in 3D? By using two eyes, with each To improve resolution, we can collect not two but eye viewing the object from a slightly different tens, or tens of thousands, of different views – as direction (interestingly, this means that beyond a though we had tens of thousands of eyes – and few metres, you can’t see in 3D!). Each eye so build up a detailed 3D structure of the object, receives a different image from the one object, at a resolution approaching 1 nm. An example of WAVES OF THE FUTURE OF WAVES

124 Figure 3: A scanning electron microscope image of a radiolarian. such a procedure for a magnetite nanoparticle could be made by spending a lot of money and a from a meteorite from Mars is shown in Figure 4, lot of time. Round magnetic lenses necessarily from various viewing directions. This technique is have aberrations. Initially aberrations were known as electron tomography. The collection of reduced by improving the lens design to minimise the many images can be automated, and their effect, but as improvements occurred, automated electron tomography is now an further improvements proved more and more immensely powerful tool for revealing the difficult to achieve. structure of objects, and especially in biology for investigating the structure of viruses and Recently, two approaches have been developed macromolecules. which completely overcome the major aberrations. The first approach was to accept that every image is distorted by the aberrations of the lens, and to realise that the form of this distortion changes if the image is taken slightly out of focus. So, by taking a series of images each with a different focus – a through focal series – the total set of images can be used to reconstruct what the image would have looked like if the lens had no aberrations.

An example is seen in Figure 5 for a boundary between two crystalline grains of gold. Any single image is unclear (see the first two in Figure 5) because the lens aberrations have distorted it,

Figure 4: A nanocrystal of magnetite reconstructed from a series of tomographic electron microscope images. The sample is from a meteorite from Mars.

Atomic resolution Since the development of the first electron microscope, the goal has been to see atoms. The importance of being able to see atoms to understand the physical world cannot be overemphasised. According to Feynman, if he were to have a choice of leaving only one brief statement to future inhabitants of the world if a cataclysmic event were to occur, it would be that all things are made of atoms. And the fact that we can see atoms, and how they interact to make crystals and molecules, makes the electron microscope perhaps the most important scientific tool to have been developed in the 20th century.

To achieve this, the major problem of lens Figure 5: Two high resolution TEM images taken at aberrations, referred to earlier, had to be different settings of focus, and the image reconstructed SEEING IN THE NANOWORLD overcome. And unlike glass lenses for light, there from a set of such images. The images are of a boundary was no possibility that a perfect electron lens in Au. 125 but the final image, which is reconstructed from What kind of atoms? the full set of images, allows us to see the Seeing atoms is one thing – but can we tell what boundary clearly (the last image in the Figure). atoms they are? We can answer this question by referring to the Bohr model of the atom, in which The second approach is similar to what we do the atomic electrons circle around the nucleus in when we have poor eyesight: we wear corrector well-defined orbitals, with each orbital having a lenses (glasses). In a similar way, correctors have well-defined energy. Different atoms have been developed for electron lenses to different electronic shell structures and different compensate for the aberrations. In the orbital energies. If we fire a fast electron at an microscope shown in Figure 2, these correctors atom, we can knock out one of the inner orbital are located about half way down the column. electrons (Figure 7) and make the atom unstable. It can regain its stability by having an electron Using these techniques, it is now relatively from a higher orbital ‘fall’ into the vacant orbital; straightforward to form images showing detail at in doing so, the electron loses energy equal to the atomic level. So we can see crystalline the energy difference between the two shells – structures and their defects, we can see for example, 8 keV for Cu and 6 keV for Fe. individual atoms on surfaces, and even watch This lost energy can express itself as an X-ray of them move, and we can investigate atomic the same energy. If this X-ray is collected by a interactions at the atomic scale. detector that can measure the energy, then we have a means of knowing whether the electron Atomic resolution can also be achieved using has hit a Cu atom or a Au atom. Figure 8 shows scanning electron microscopy. As we have seen, a spectrum of X-ray energies collected by a solid SEM involves scanning a small probe of electrons state X-ray detector from a metal alloy. We see a across the surface of a sample and recording the large number of X-rays corresponding to a scattered signal as a function of probe position. If number of different elements. This spectrum can the sample is very thick, the electron beam is be converted to a compositional analysis if we scattered within the sample, and resolution is know about the physics of X-ray generation lost. But if the sample is very thin (say 10 nm) (which we do!). then it passes through the sample scattering in all directions.

Electrons that have passed close to the nucleus are scattered to high angles, and can be collected on an annular detector. Their intensity is proportional to Z2 (Z is the atomic number) and so the scanning image shows heavy elements (large Z) brightly. We call this technique scanning transmission electron microscopy (STEM). In this Figure 7: A fast electron (1) ejects a core electron (2) from way we can investigate the arrangements of the inner shell of an atom (left) and loses energy. Then an heavy elements within a matrix of lighter electron from another shell drops into the inner shell, and elements, and the example of Figure 6 shows Lu emits an X-ray (right). atoms that have segregated to the surface of silicon nitride crystals in a ceramic used for automotive bearings. In an electron microscope, we can focus all the electrons into a very small region of the sample, and so collect an X-ray spectrum from just that Figure 6: The interface between a area. By scanning the electron spot across the silicon nitride crystal (bottom) and a surface, and recording the number of X-rays of a glass (top) in a silicon nitride particular energy (the X-ray spectrum) while the

WAVES OF THE FUTURE OF WAVES ceramic. The bright atoms at the interface are Lu, used to engineer scan proceeds, we can generate an elemental 126 the strength of the interface. map of the sample. Figure 9 shows an example Figure 8: An X-ray spectrum from a multi-element alloy. of the Nb and Cu elemental distributions in a It is clear from the X-ray spectrum and EELS of multilayered material, where the layers are 1.5 Figures 8 and 10 that the more there is of a nm in thickness. particular element, the larger will be the signal at the energy corresponding to that element. The Of course, the incident electron in Figure 7 will relative signal strengths can be used to give the lose energy when it knocks the orbiting electron relative amounts of each element, if we know from the atom, and the amount of energy it loses how easy it is to generate signals of each energy will be equal to the energy required to knock out (the cross section). The ability to determine the the orbiting electron. This energy will depend cross sections for experiment or theory is now upon the atomic species. By placing a very sophisticated, so that quantitative X-ray and spectrometer after the sample, we can display EELS microanalysis is widely used for the number of electrons that have lost different microstructural elemental characterisation. amounts of energy (the electron energy loss spectrum or EELS) as in Figure 10. Just as we can use particular X-ray energies to give Lorentz Microscopy compositional maps, so we can use the energy Magnetic materials are among some of the most loss signal in a similar way. important for technological applications e.g. for recording data or for acting as fast switches. These materials are generally composed of local regions (domains) in which the direction of magnetisation is well defined and is different in each domain. The information is stored by these domains by their direction of magnetisation, and the aim of the materials scientist is to develop Figure 9: An elemental map of SEEING IN THE NANOWORLD Cu (red)/Nb (green) mutilayers materials in which this information is stable, and (1.5 nm period). can easily be written and read. 127 scattered electrons give signals that are sensitive to the atomic number, Z, and because of this, we can determine the local composition, almost to the atomic level.

In the next chapter, we will see how this wonderful instrument can be used to explore the nanoworld.

Reference H Belloc H 1912 “The Microbe” from More Beasts For Worse Children. Duckworth 1912 Figure 10: An energy loss spectrum from BN. Website For a clear description of the works of Davisson and of de In order to visualise these domains, we can use Broglie and of Ruska and other Nobel Laureates, see the fact that electrons are deflected by magnetic http://nobelprize.org/physics/laureates/ fields. An electron beam passing through a domain will be deflected by the domain in a direction that depends upon the direction of the domain’s magnetic field. Adjacent domains, with different directions of magnetic polarisation, deflect the electrons in different directions. If the image is formed from only those electrons that are scattered into a particular direction, then domains which scatter in that direction will appear in the image, and domains which don’t, won’t. Figure 11 shows an example for small magnetic domains in a FePd alloy, where alternating domains appear bright and dark, depending upon whether or not the electrons are scattered into the direction for which the electrons are allowed to reach the image.

Conclusion The realisation that electrons have a wave nature, and the invention of electron lenses, opened up the possibility of seeing objects smaller than could ever be seen with light microscopes. The limitation of the resolution that can be achieved is set, in the first instance, by the aberrations of the electron lenses.

Over the past 60 years, these aberrations have gradually been reduced, but very recently, methods for eliminating the effects of the major aberrations have been developed. It is now possible to see individual atoms, and how they WAVES OF THE FUTURE OF WAVES come together to form structures. At the same Figure 11: Lorentz microscopy (Foucault) magnetic domain image of an FePd alloy. 128 time, the X-ray and energy loss signals from SEEING IN THE NANOWORLD

129 Einstein’s Miraculous Year /Part 2

Albert and the Atoms – Brownian Motion THE CONCEPT OF the atom is so commonplace today, it’s hard to imagine not believing in these tiny bundles of matter. Yet believe we do, and the belief is founded on a some degree of faith, because unless you use something like a powerful electron microscope, you can’t actually see atoms at all. Instead, from our earliest lessons in science at school, we’re assured that atoms exist in all their different elemental varieties, and we believe in the evidence built over years of scientific research that overwhelmingly supports their existence.

But back in 1905, the case for the constant, random, zig-zagging suspended particles, their Brownian atom wasn’t quite so strong – in motion. No obvious reason for this motion, could be explained if you fact, you could say that the whole was seen at the time, and so the imagine what’s going on at an even field of science focused on the phenomenon was dubbed Brownian smaller scale, down at the level of properties of matter was in a bit of motion and it remained a curiosity water molecules. a bother. Boltzmann had a for over 75 years. statistical theory of Einstein argued that the water thermodynamics (the study of In May 1905, Einstein published his molecules would be in constant temperature and the flow of energy second paper of that remarkable thermal motion, the random kinetic within physical systems) that stated year: On the movement of small energy associated with that irreversible changes, like the particles suspended in stationary temperature, all jiggling about and melting of an ice cube, could be liquids required by the molecular- colliding with the much larger explained statistically by the kinetic theory of heat (Annalen der particles floating in the water. And reversible motions of tiny, unseen, Physik 17, pages 549-560). In this those many collisions, averaged hypothetical atoms. But there were paper, Einstein showed that the over time, could explain the many other physicists who doubted mysterious haphazard wanderings of meandering paths of the larger the idea (including the great Max Planck, at first). So the atomic hypothesis was very much up for grabs in 1905.

In another, very different corner of scientific inquiry – more down the biological end of things – there was a seemingly unrelated phenomenon searching for an explanation. In 1828 a botanist named Robert Brown had been observing pollen grains suspended in water. During his observations, Brown noticed WAVES OF THE FUTURE OF WAVES that the tiny particles of dust and pollen in the water seemed to be in 130 suspended particles. His paper accurately, and instead concentrated began: on the average displacement of the ... according to the molecular- particle from a starting position over kinetic theory of heat, bodies of time. This quantity is much easier to a microscopically visible size observe through a microscope than suspended in liquids must, as a trying to track the incessant zig- result of thermal molecular zagging itself. motions, perform motions of such magnitudes that they can Einstein’s theoretical work, and be easily observed with a subsequent experimental microscope. It is possible that measurements of Brownian motion the motions to be discussed here that confirmed the theory, provided are identical with so-called some of the first real evidence for Brownian molecular motion; the existence of atoms and however, the data available to molecules. As a bonus, because the me on the latter are so imprecise mathematical treatment related that I could not form a judgment Brownian motion to physical on the question. parameters and constants, Einstein’s theory provided a new But more than offering a way to measure Avogadro’s mechanism, Einstein calculated the number, too! magnitude of the effect – he showed with rigorous statistics that the So Albert Einstein’s second paper of molecular thermal motion gave just 1905 gave a real boost to the the right amount of kick to result in statistical treatment of thermo- the observed Brownian motion. dynamics and, through the lens of a Einstein’s great insight came from microscope, showed the world some the way he tackled the problem: he of the first real evidence for the ignored the actual random existence of atoms and molecules – EINSTEIN'S MIRACULOUS YEAR, 2 PART wanderings of the suspended their microscopic thermal jiggling particles, which are hard to define manifest as the macroscopic meanderings of Brownian motion. and even harder to observe 131 Opposite: (Fig 2) Five different ways of 132 arranging a repeating pattern (five 2-D Bravais lattices). Building in the Nanoworld

Professor David Cockayne FRS

Introduction NANOTECHNOLOGY IS about building on the nanometre scale. And just as it is important to be able to see what you are building if you are a bricklayer (would you hire a blind bricklayer?), so it is important to be able to see if we want to build in the nanoworld. And one of the main reasons that the nanoworld is of such great research interest nowadays is because of the many new instruments we have for seeing in the nanoworld, such as atom probes and atomic force microscopes and the electron microscopes that we discussed in the earlier chapter.

133 ut nature has been building at this level for an aeon, and so the best approach is to Blearn from what can be found there. At the simplest level, this involves understanding how atoms pack together, both in the crystalline and in the amorphous state. From there we can consider how systems assemble at the atomic and molecular level. And, because the surfaces of objects take on an even more important role at the nanoscale than they do in larger systems, we consider the structure of interfaces and how to build them to control properties. From these kinds of observations, we can use the knowledge gained to build artificial structures, such as Fig 1: High resolution image of crystalline gold showing the atomic structure. quantum dot arrays, nanowires and quantum computers. later). Next time you are visiting a shop that sells wallpaper, look to see how many different designs Atomic packing there are. I don’t mean whether the paper has How is the material world constructed? We will pictures of roses or horses or cars, but how the start with a simple system and work towards the pattern repeats itself. It will be very surprising if more complex. Figure 1 is an electron micrograph you find any more than the five different pattern of gold, in which we see the atoms tightly packed repeats shown in Figure 2 (shown at the start of together in a regular array. The atoms are like this chapter). Each of these patterns is composed people – they like to snuggle up to each other. of a set of objects (the motif), repeated on a That’s because atoms interact through the forces regular grid of points called the lattice.You can that they exert on each other. In the simplest have an infinite number of different motifs – terms, these are (i) an attractive force due to the horses, flowers, houses, squiggles, and in red, electrostatic attraction of the negative charge blue, green – so that there are an infinite number distribution, and (ii) a repulsive force between the of different wallpapers (which is why it takes so nuclei. The opposing action of these two forces long to choose), but I doubt that you will find any results in the atoms having a preferred distance other lattice than one of the five shown. We write apart at which these forces balance, and this this arrangement mathematically as (motif*lattice) distance depends upon the atoms involved. For where * is a mathematical operation called gold, the preferred distance is 0.29 nm. Cu and convolution. You can see that the pattern of Pt and Al pack in exactly the same way as Au as Figure 1 is (oblong) * (one atom). We can refer to in Figure 1, but because these atoms have the repeating unit as the unit cell. different atomic structures (different atomic numbers Z), they have different interatomic forces Atoms pack in a regular array not only in two, but and so they have different preferred distances to also in three dimensions; and in three dimensions their nearest neighbour – 0.29 nm for Cu, 0.27 there are not 5 but 14 ways of packing identical nm for Pt, and 0.28 nm for Al. objects together in a repeating pattern – these are called the 14 Bravais lattices. There is an Because the atoms have a preferred distance of infinite number of 3D motifs, so that the possible separation, when a large number come together number of different crystals = (motif )*(Bravais they arrange themselves in a repeating pattern, lattice) is infinite. as we see in Figure 1. This raises the question of how many possible repeating patterns there are. In 3D, gold has the ‘cubic F’ Bravais lattice with a motif of one atom (Figure 3), as do Cu and Pt and WAVES OF THE FUTURE OF WAVES Let us first consider this question in two Ag, although the sizes of their unit cells are 134 dimensions (we will consider three dimensions slightly different because the atoms have different interatomic forces. And since they all have this Nowadays, because we can see atoms, we can same structure, it is not too difficult to replace an investigate the structure of these materials by Au atom by a Pt atom without causing too much forming images with a variety of microscopes, stress for the system. (There will be some stress, including scanning tunnelling microscopes, because the Pt atom will not be at its optimum electron microscopes and atom probes. But distance from the Au atoms, since it is forced to because they are crystals, we can also have the Au-Au rather than the Au-Pt distance.) investigate their structure by scattering waves So Pt and Cu are commonly used in Au for (such as neutrons, electrons, X-rays) from them jewellery (wedding rings, for example), since Au is and forming diffraction patterns, as we discussed too soft on its own. in the previous chapter. Because the crystals have a repeating pattern, the diffraction pattern will be an array of spots, as we saw in the first chapter. Electron or X-ray or neutron diffraction studies involve the analysis of these diffraction patterns to determine both the motif and the lattice of the structure.

The diffraction pattern we obtain depends upon the direction of the incident ray relative to the orientation of the crystal, and so patterns collected with different incident beam directions carry different information about the crystal. Consequently we can use a set of different diffraction patterns, collected for different incident Fig 3: The cubic F Bravais lattice. ray directions, to increase the information and so obtain a more detailed analysis of the crystal So crystalline materials are very common in structure. Figure 5 shows the resulting nature – the sand on the sea shore is crystalline convergent beam electron diffraction pattern,in SiO2, common salt (NaCl) is crystalline, as are which each diffraction spot is converted into a almost all the metals you know, and as are most disc, equivalent points in each disk corresponding semiconductors in your cell phones and watches. to a different direction of the incident electron And all of them have one of the 14 Bravais beam. lattices as the building block, repeated over and over to form a crystal. If a crystal grows slowly, starting at one point and gradually expanding in volume, then very large single crystals can be grown (for instance, precious diamonds grown over a long time in the earth’s crust). But usually structures grow from many sites at once, forming small crystalline grains that meet at grain boundaries. Consequently a large block of the material might be composed of many small crystal grains (an example is shown in Figure 4).

Fig 5: Convergent beam electron diffraction pattern.

Fig 4: Crystalline grains BUILDING IN THE NANOWORLD in silicon showing the atomic structure. 135 Crystal deformation Atoms arrange themselves into a crystal because it is the lowest energy configuration they can assume. (Indeed we can try to predict their structure by studying which model system has the lowest energy.) Consequently to change the crystal from this state requires us to do some work – we must stretch or even break the interatomic bonds. To move the crystal from one energy state to another requires work to overcome an energy barrier. Consequently the perfect array of atoms in a crystal makes it Fig 6: The motion of a dislocation through a crystal. difficult to deform (do you think a leaf or your skin are crystalline?). at the two cases of a copper grid and a crystal), So we might wonder how it is that we can, for then scattering occurs in well-defined directions example, bend or stretch a piece of metal. Since (given by the Bragg equation). If we shine the structure of a metal is repeated layers of electrons onto a crystal which contains a atoms like those seen in Figure 1, it might be dislocation, then the amount of scattering that thought that the deformation could occur by occurs into a particular direction will be different sliding some of the layers of atoms over one for the region where there is a dislocation, another, like sliding the cards in a pack over each compared to the region where there isn’t a other. But this would require that all the bonds dislocation (Figure 7), because of the atom between the atoms on the two layers break at the disorganisation near the dislocation line. same time, and an estimate of the force to do that shows that it is far greater than what is required in practice.

Figure 6 shows what actually happens: the bonds break along a line (into the page) in the crystal, with all the deformation concentrated along that line. All the atoms on the left hand side have slipped sideways by one repeat unit, and, as the dislocation line moves to the right, the region which has slipped gradually increases, until eventually the deformed region reaches the right hand side, and the entire set of atoms above the slip plane has moved to the right relative to the atoms below the slip plane. We call this localised region of deformation the dislocation line, and deformation proceeds by the dislocation line moving across the crystal from one side to the other. Since each dislocation causes a movement of only about one interatomic distance (about 0.2 nm), a deformation of 1 mm involves the motion Fig 7: The formation of an image of a dislocation by the scattering of electrons. of about 106 dislocations.

We can see these dislocations by using a If we then prevent all electrons from reaching the combination of diffraction and imaging in the image except for those which are scattered into WAVES OF THE FUTURE OF WAVES electron microscope. As we saw in the first this particular direction, the dislocation will 136 chapter, if we have a periodic object (we looked appear as a dark line, as in Figure 8. We can deliberately hinder the movement of the Allotropes dislocations by putting ‘boulders’ in their way, so The most well-known crystal structure of carbon that it then would become more difficult to is diamond, which can be described as (cubic deform the crystal. This is the principle of F)*(C-C). That is, it has the same Bravais lattice precipitation hardening, in which the ‘boulders’ as Au (cubic F), but the motif is a pair of C atoms are nano-sized precipitates. On the other hand, if instead of one Au atom. Carbon crystallises in there are sufficient dislocations travelling in several other forms – one is graphite, in the form different directions through the crystal, they can of sheets of C atoms arranged in a hexagonal become entangled and restrict each other’s array, with these sheets held together by weak movement, and the result is called work (Van der Waals) forces. If we take a sheet of hardening. Understanding the role that graphite, it can be rolled into a tube and joined, dislocations play in controlling the strength and with no discontinuities. These carbon nanotubes properties of materials has been one of the major were first discovered by Iijima in the electron advances in materials science over the past microscope in 1991. They are very strong, and so 50 years. can be used as strengthening fibres within composites; and they can be used as containers [FIGURE 8 HERE] inside which thin crystal wires can be grown, with properties which are different to the bulk, or within which individual molecules can be packed. If a piece with 60 carbon atoms is cut from a single graphite sheet, it can be bent into a football shaped molecule, C60, known as a bucky ball (after the architect Buckminster Fuller who had previously built large structures with this shape). Indeed the bucky ball is small enough to pack inside a nanotube (see Figure 9), and an atom is small enough to place inside a bucky ball. So individual atoms can be held inside bucky balls, which in turn are packed within a nanotube. Scientists are investigating whether this structure might be the basis for the next generation of (quantum) computers. Fig 8: An electron microscope image of dislocations in silicon.

Fig 9: Models of a carbon nanotube and a bucky ball, and

electron microscope image of BUILDING IN THE NANOWORLD bucky balls inside a single walled carbon nanotube. 137 Quantum dots chapter, so to when the electron in the quantum Quantum dots are formed by depositing a layer of dot makes a transition between the quantised one material (e.g. Ge) onto the surface of another energy levels, it emits light of a well-defined material with a very similar structure (e.g. Si). wavelength. In this way intense, efficient sources These two materials have the same Bravais of light can be produced for use not only as laser lattice and the same motif. Their only difference pointers and fluorescent dies, but as replacements is that they have different atoms (Si or Ge), and for brake lights on cars and traffic lights around this difference results in them having different cities of the world. The structure of these quantum unit cell sizes. dots – their shape and composition profile – determines their quantised states, and so it is When Ge is deposited as a thin film onto a Si important not only to be able to investigate what crystal surface, this size difference can be structures are obtained for a given method of accommodated by the film of Ge being growth, but also to be able to predict what constrained into adopting the unit cell size of Si. structure will be obtained from a particular set of We call this accommodation epitaxial growth.As experimental conditions. the thickness of the Ge film is increased, eventually the strain in the film becomes too Modelling and simulation great to support, and the surface film collapses. One way it can collapse is for the Ge to form into This predictive aspect of materials science is one small islands or dots, an example of which is of the most important tools for progress in seen in Figure 10 – in this case, an InAs dot developing new materials, because it allows us to on GaAs. design materials with specific properties without having to first make samples of them. To do this requires a close interaction between modelling and experimentation. For example, in the case of quantum dots, a model atomic structure can be set up in the computer, with the computer programme describing the forces acting between the atoms, either using quantum mechanics or as empirical equations. The total energy of the structure can be calculated, and then an iterative process of moving atoms one by one can be carried out, at each step recalculating the energy of the structure to determine whether or not the Fig 10: An InAs quantum dot on a GaAs surface. move has taken the system to a lower energy. On the assumption that the physical structure will be When crystals have such a small volume (say stable when the energy is a minimum, the 100x100x100 atoms), they can have unusual predicted stable structure can be obtained. properties not found in the bulk. For example, if we have a small crystal of Ge into which we inject an electron, the electron can have certain well- defined energies that depend upon the size and composition of the crystal. These energies are called quantised energies (from the Latin word quantus, ‘how much’), and the object is called a quantum dot. Because these quantised energies remind us of the quantised energies of electrons in the shells of an atom, quantum dots are sometimes referred to as giant atoms. And just as WAVES OF THE FUTURE OF WAVES atoms emit X-rays of a given wavelength when Fig 11: A model of the distribution of Ge in a Ge-Si 138 they become unstable, as we saw in the first quantum dot. Figure 11 shows the result of such a procedure image or diffraction pattern. Figure 12 shows an for a quantum dot composed of Ge and Si, in example of a comparison of experimental image which the Ge and Si atoms were initially randomly of a dislocation with images calculated for distributed. The modelling has resulted in a different models of the dislocation. redistribution of Ge and Si atoms, with Ge (blue colour) concentrating near the surface. This result Amorphous materials is in agreement with the experimental results As we have seen, atoms like to come together as obtained from energy-loss and X-ray mapping crystals. However there are many situations (discussed in the previous chapter), and the where the conditions for forming crystals do not agreement gives us confidence in the modelling. exist, or where the crystalline structure is These calculations are extremely time consuming, destroyed. For example, in DVDs, a laser pulse is even for the most powerful computers, because used to blast the surface of a crystal, creating a of the large number of atoms involved in nanovolume of amorphous material, which then modelling a realistic structure. serves as the information bit. Other examples are thin amorphous films used as hard coatings on A second important role of computers is as an surfaces, for example on high speed drills. aid in the interpretation of images and diffraction patterns obtained from the electron microscope. It might be thought that, with microscopes that Although we have not mentioned it to this point, a can resolve atoms, we could study these major difference between electron scattering and materials atom by atom. But there are two neutron and X-ray scattering is that electrons are shortcomings to this approach. Firstly, because of scattered orders of magnitude more strongly than the disorganised arrangement of the atoms, an X-rays or neutrons (we say that atoms have a atom-by-atom description is of little use – a larger cross section for electron scattering). This statistical description is needed. And secondly, it has the advantage for electrons that, when is very difficult to interpret the images of these studying small objects, the scattered signal is materials because the individual images of all the greater for electrons; but it has the disadvantage atoms superimpose on each other and obscure that the electrons, in passing the sample, will any detail. usually be scattered many times. This multiple scattering often makes it difficult to interpret So let us consider what we might do with images and diffraction patterns straightforwardly. diffraction. We start by considering scattering of Fortunately we have a thorough understanding of electrons by an individual atom. Many of the how this multiple scattering occurs, and so we electrons are scattered in the forward direction, can describe it mathematically and write computer programmes to simulate it. In this way, model structures can be set up in the computer, and simulated images and diffraction patterns can be obtained for comparison with experimental images. Then a choice between alternative model structures can be made by comparing the simulated image or diffraction pattern from each of them with the experimental BUILDING IN THE NANOWORLD Fig 12: Experimental (left) and simulated (right) images of a dislocation in brass. Fig 13: The electron diffraction pattern of amorphous carbon. 139 while some are scattered through large angles as in Figure 13. If we have a large number of atoms scattering independently, the diffraction pattern looks very much the same, but stronger because of the larger number of atoms. But if we look closely at the diffraction pattern from this array of atoms, we see (Figure 14) that there are oscillations which we might not expect from the diffraction from an array of randomly distributed atoms. These oscillations are telling us that the carbon atoms are not arranged randomly, but have some degree of organisation, arising from the interatomic forces that gave rise to the Fig 16: The tetrahedral arrangement of atoms in diamond, crystals discussed earlier. A mathematical showing the rotation around bonds that occurs in analysis of the diffraction pattern (a Fourier amorphous diamond. transformation) allows us to display this atomic arrangement as a ‘distribution function’ (Figure 15). We see that even in this amorphous material, and the second-nearest neighbours are at the carbon atoms like to be 0.17 nm apart and 0.25 nm. But why are these same distances 0.25 nm apart, but none like to be 0.20nm apart. observed in the amorphous state? Why is this? We can understand it if the atoms are in a crystal – specifically, diamond. For in The reason is that in diamond, each carbon atom diamond the atoms are arranged in a very regular is at the centre of a tetrahedron of four other array, as we have discussed earlier. And in carbon atoms, and this arrangement is diamond, the nearest neighbours are at 0.17 nm, exceedingly strong. And this tetrahedron of atoms survives, with some distortion, into the amorphous state of carbon. Adjacent connected tetrahedra can rotate about their common bond (2-5 in Figure 16), and this random rotation, when repeated for all bonds, results in the amorphous structure. During this rotation, the first nearest distance (1-2 in Figure 16), and the second nearest distance (1-5) of the tetrahedron are retained, and appear in the distribution function of the amorphous state, but the third Fig 14: The intensity profile of Figure 13, showing nearest neighbour distance (1-8) is not retained oscillations due to structure in the amorphous carbon. and does not appear.

Fig 15: The radial distribution function of amorphous The amorphous state is of increasing carbon; the vertical lines mark the nearest neighbour distances in crystalline diamond. technological importance as it plays a central role WAVES OF THE FUTURE OF WAVES

140 in materials such as optical fibres, DVDs, solar collectors and ceramics, and experiments of the kind described allow us to study these materials in detail.

An interesting case results from throwing bucky balls into a heap and scattering electrons from them. The distribution function that results from the diffraction pattern (Figure 17) has numerous maxima on the length scale of 0.2 to 1 nm, which arise from the various distances between atoms on the surface of the bucky ball. Using this data about the interatomic distances within the ball, we can build a model structure to agree with the data. But in Figure 17 there is also a very Fig 18: The path of a crack in a silicon nitride ceramic. large, broad maximum at 1.1nm which is larger than the size of a single bucky ball; so it can’t The ceramic is made up of a high density of small arise from interatomic distances within a ball. needle-like silicon nitride crystals, separated by a This feature shows us that the balls themselves thin layer of amorphous glass. By a method that like to arrange themselves into one of the Bravais we shall discuss, the interface between the glass lattices – a cubic lattice – with a repeat distance and the crystal is made weaker than both the of 1.1nm. So again we see nanostructures self- glass and the crystal. So, if a crack forms at the organising into periodic structures. edge of the glass, it will follow a path such as that shown by the yellow line. The longer the path the better, because, to advance, the crack must break the interatomic bonds between the glass and the crystal, and each broken bond absorbs energy. Eventually, if the crack doesn’t have enough energy, it comes to a halt. So the aim is to produce long crystals with a relatively weak

Fig 19: The atomic structure of the interface of silicon nitride grain and an intergranular film; the bright dots are Lu atoms which act as a “zipper” to guide the path of a crack.

Fig 17: The experimental radial distribution function of bucky balls. The meeting of the crystalline and amorphous nanoworlds As we can understand from Figure 4, interfaces between crystals occur frequently in technological materials, and we consider the case of silicon nitride as an example. Silicon nitride ceramics are used in many important applications including brake linings and bearings in cars, because of their strength and toughness. One of their most important features is that cracks do not easily BUILDING IN THE NANOWORLD pass through the material. The way this is achieved can be understood from Figure 18. 141 interface with the glass. Rare earth elements such as lanthanum or lutetium are introduced for the purpose, and the high-resolution image of Figure 19 shows how they act. They ‘paint’ the surface of the silicon nitride grain, firstly stopping it growing sideways, so that it forms a long crystal, and then acting as a ‘zipper’, guiding any crack along the crystal surface.

Building with atoms The ability to manipulate the arrangement of atoms using a variety of tools is becoming increasingly sophisticated. One of the most amazing demonstrations is the use of a scanning tunnelling Fig 20: Ge quantum dots on a Si substrate. The quantum microscope to manipulate single atoms to form dots have grown in rows above dislocations which are in geometric shapes such as quantum corrals and the Si. letters of the alphabet. While such manipulations are fascinating, they are likely to The assembly of individual be technologically useful only if atoms into small halide crystals they can be replicated on a large “The assembly of has recently been achieved by scale. Obvious targets are the inserting them into carbon growth of bone and skin, so that individual atoms into nanotubes, which act as tiny test repair can be carried out following small halide crystals has tubes. An example is shown in damage to the human body. But Figure 21, where a wire of KI, others include the self-assembly recently been achieved by one unit cell wide, has been of nanoparticles to form arrays for inserting them into assembled within the tube. KI is quantum computing or for normally cubic in the bulk, but memory and recording devices. carbon nanotubes, which from the figure we see that the act as tiny test tubes ...” unit cell has different sizes in the An example is the growth of directions along and across the arrays of quantum dots. We have wire. This is an example of how already discussed the growth of quantum dots the structure and properties of materials are and their ability to produce quantised light changed when they form at the nano scale. emissions. If a large number of dots can be arranged to pack tightly, so that their density is high, then bright light sources could result. One way to pre-determine where the quantum dots might grow on a surface is to bury dislocations (which we discussed earlier) just below the surface of a crystal. Each dislocation results in a slight disturbance to the positions of the atoms at the crystal surface, and this disturbance causes the quantum dot to form above the dislocation rather than elsewhere on the surface. By having a regular array of dislocations beneath the surface, a regular array of quantum dots can be formed. Fig 21: A KI crystal grown within a carbon nanotube. Frances Ross and her colleagues at IBM have successfully demonstrated this approach to the Nature itself has developed many methods for WAVES OF THE FUTURE OF WAVES self assembly of quantum dots, and an example self-assembly, and we can often learn techniques 142 of their work is seen in Figure 20. from the natural world. One example of nano Fig 22: Magnetic particles arranged as a backbone within a single cell of Magnetospirillum magnetotacticum.

objects serving a purpose in nature is the small magnetite crystals found in certain bacterium. These small magnetic crystals form a chain as a backbone to the bacterium, as seen in Figure 22. These crystals are referred to as magnetosomes. They are usually arranged in a linear chain inside each bacterium and they result in a permanent magnetic moment to the cell, which results in it having the ability to align itself and move parallel to the local geomagnetic field lines. It is thought that this benefits the bacteria by making them able to more efficiently locate the optimum oxygen concentrations in the water columns of the sea where they are found.

Conclusion There are many fascinating objects in the nanoworld which occur naturally, and which can be seen only with powerful microscopes. This is a real part of the physical world, so far relatively unknown to us because we have only recently developed the tools needed to explore in this world. Many of the objects we find, such as carbon nanotrees and nano-onions observed to occur naturally under certain carbon growth conditions, have no obvious use to date. But whether or not a discovery proves immediately useful, the important point to remember is that, for the nanoworld, we are in the age of exploration and you can be the explorers. BUILDING IN THE NANOWORLD

143 Radioactive Boy Scout By Dr Karl Kruszelnicki

MOST KIDS HAVE some kind of hobby – a sport, collecting stamps or computer games. But David Hahn, who lived in Commerce Township in Michigan, about 40 km out of Detroit, had a scientific hobby – chemistry. And so he tried to build a nuclear reactor.

At the age of ten he was given a hits this core, it splits into two neutrons, the thorium-232 would book called The Golden Book of smaller atoms and a neutron or two turn into uranium-233. So he Chemistry Experiments. Something and also gives off a huge amount bought thousands of gas mantles clicked, and by the age of 12 he of energy. and turned them into thorium ash had mastered his father’s with a very hot gas flame. How did University-level chemistry books, So David started off by making a he purify the thorium? Simple – he and by 14, he had made neutron gun. He pretended to be a bought a few thousand dollars nitroglycerine. His father thought Physics Lecturer, and got lots of worth of lithium batteries, cut them that David needed a stabilising professional help from industrial open, and did some simple influence, so he advised him to try companies, the American Nuclear chemistry to concentrate the for the goal of Eagle Scout – which Society and the Nuclear Regulatory thorium. But alas, the effort was needs a total of 21 Merit Badges. Commission. He found out that he wasted. His neutron gun didn’t Some Merit Badges (like could get radioactive Americium- have enough grunt to turn thorium- citizenship, first aid and personal 241 from household smoke 232 into uranium-233. management) are compulsory, detectors – so he bought 100 while some (from American broken ones at a dollar each. The Time for Plan B. Radium delivers Business to Woodchuck) are friendly customer-services heaps of ∝-particles, and he had chosen by the scout. David chose representative told him exactly been told if you blast these ∝- the Merit Badge in Atomic Energy. where the Americium-241 was, and particles onto beryllium, you get To get this badge, you have to how to remove it from its inert gold enormous numbers of neutrons. know about nuclear fission, know matrix. He then put his tiny pile of But how could he get some who the important people in the Americium-241 inside a hollow radium? Well, until the late 1960s, history of Atomic Energy were, and lead block, and drilled a small hole the glow-in-the-dark faces of make a few models of some atoms, in it. As Americium decays, it gives clocks, and car and airplane and other stuff. David built a model off a-particles. When a-particles hit dashboard instruments glowed of a nuclear reactor with some tin aluminium, the aluminium gives off because they were painted with cans, drinking straws and rubber neutrons. So he put a strip of radium. So he started the slow bands, and earned his Merit Badge aluminium in front of the hole in the process of haunting junk and in Atomic Energy on May 10, 1991 lead block where the a-particles antique shops, surreptitiously (when he was 14 years and came out, and bingo – he had a chipping off the glowing radium. 7 months old). Then he decided to neutron gun. But one day, he got lucky when his aim higher. Geiger Counter went off its brain. He had found out that the cloth He bought the clock for $20, and Atoms have a core of positively- mantles in gas lanterns are covered inside, found a complete vial of charged protons, and neutral with thorium-232 (because thorium radium paint conveniently left neutrons. Some of the bigger is very resistant to high behind. WAVES OF THE FUTURE OF WAVES atoms (like uranium, for example) temperatures). He also knew that if 144 have unstable cores. If a neutron you hit thorium-232 with enough So he rigged up a more powerful neutron gun with a hollow lead block with a hole, his precious radium inside, and some beryllium to get hit by the ∝-particles and give off neutrons. What did he use for a target? Some uranium ore he got from a friendly supplier. But failure again. The neutrons were moving too fast (about 27 million kilometres per hour) and just zipped through the uranium. So he slowed them down to about 8,000 km/h by running them through tritium (which he painstakingly scraped off modern glow-in-the-dark gun and bow sights) – and the uranium ore got more radioactive.

By this time, David Hahn was 17, and he decided to stop fooling around. He mixed his radium with his americium and aluminium, wrapped it in aluminium foil, and then wrapped the whole mess in his thorium and uranium – of course, all held together with gaffer tape. Finally he had success – the bizarre ball got more radioactive every day. Perhaps too much success – he could pick up the radioactivity 5 houses away. He panicked, and began to dismantle And David? Well, while he was a his creation. whiz at science, he never was much good with maths and English. At 2.40 am on the 31st of August, So today, he’s a junior 1994, the local police were called sailor/deckhand on the aircraft because a young man was doing carrier, USS Enterprise, which has something suspicious near a car. 8 nuclear reactors. David told the police to be careful of his toolbox, because it was And if George Bush ever needs to radioactive. Soon some men in call in the Heavy Artillery, maybe he ventilated white moon suits were should forget the SEALs and the chopping up his radioactive shed SAS, and call in the Boy Scouts to with chainsaws, and stuffing the do their Bob-A-Job... parts into thirty nine 200-litre sealed drums which they took away to a nuclear waste repository. The clean-up cost about $120,000 FROM Dr Karl’s book Bum Breath, Botox RADIOACTIVE BOY SCOUT – but it did protect the 40,000 and Bubbles (Harper Collins Publishers) nearby inhabitants from harm. Illustration courtesy Adam Yazxhi 145 PROFESSOR PETER ROBINSON received his PhD in theoretical physics from the University of Sydney in 1987, then held a postdoc at the University of Colorado at Boulder until 1990. He then returned to Australia, joining the permanent staff of the School of Physics at the University of Sydney in 1994, and obtaining a chair in 2000. He is currently an Australian Research Council Federation Fellow working on topics including brain dynamics, space plasma physics, wave theory and self-organized criticality.

146 Understanding Brain Dynamics

Professor Peter Robinson

Introduction ONE OF THE KEY questions of neuroscience is how to relate brain activity to what that activity is accomplishing. In other words, when brain cells (neurons) are active, what are they doing? Before one can attempt to answer this question, one must be able to measure activity, and to relate activity to the stimuli that cause it.

Figure 1: The brain sectioned along a vertical plane running approximately from ear to ear. The gray matter is seen as a thin layer on the outside, immediately surrounding the white matter. The location of the thalamus is also shown.

147 he work described here attempts to relate absence of a proper theory that enables stimuli to brain activity, and brain activity to measurements to be predicted (i.e. calculated) T measurements of that activity, in ways that starting from the cellular level, all that can be enable measurements to pin down details of the found is a (very large!) set of associations activity and underlying stimuli. These insights will between experimental situations and the resulting help in understanding brain function (information measurements. The primary aim of the University processing, etc.) and malfunction (e.g., disease). of Sydney group is to enable detailed predictions to be made and compared with In this chapter, I will first briefly describe some of EEG experiments. the ways in which brain activity can be measured. Then I will discuss how the Another type of measurement of brain activity, connections between stimuli, activity, and one that has become available in the last 20 measurements can be modelled theoretically. years, is provided by functional magnetic Some of the key predictions, and their resonance imaging, or fMRI. This technique relationships to experiment, will be reviewed, measures tiny differences in the magnetic followed by a short discussion of commercial properties of oxygen atoms, between those within applications currently under way. the oxygen-carrying blood chemical haemoglobin, and those that are detached from haemoglobin. Because of the slight differences in their Measuring Brain Activity environment, oxygen atoms behave slightly One measure of brain activity is electrical signals differently in these two situations, thereby of neurons. When neurons become active, small enabling the amounts of haemoglobin with and currents flow briefly between their interior and without oxygen to be measured in various parts exterior. These currents are associated with of the brain. The amount of deoxygenated electric potential (voltage) changes, which take haemoglobin is highest in areas that are active, the form of spikes on a voltage-time plot. When where oxygen is being used, thereby enabling averaged over many neurons these voltages can brain activity to be mapped. be detected by electrodes placed at the surface of the brain or on the scalp. Again, in the case of fMRI, our aim was to make the connection between the stimuli, activity, and The first detection of brain electrical activity was measurements of activity sufficiently precise that reported in 1875 by a British physiologist, they can be used to obtain new insights into Richard Caton, who used a galvanometer to brain function. measure electrical activity in the brains of rabbits and monkeys. Little more was done until the mid-1920s when Berger showed that similar Modeling The Brain activity was present in humans, and depended on The largest part of the human brain is the mental activity. Since then, recordings of electrical cerebral cortex, seen in Figure 1. It is composed activity from the scalp – electroencephalograms, of a thin folded layer (2 to 3 mm thick) of ‘grey or EEGs – have been widely used to measure matter’, which is involved in high-level brain activity and its changes with state of information processing. This is underlaid by arousal (e.g. awake, asleep), brain function ‘white matter’, which comprises mainly bundles (information processing, attention, type of sensory of neurons that interconnect areas of grey input involved), and disease (e.g. Alzheimer’s, matter. Because it is closest to the skull, cortical epilepsy, attention deficit disorder). grey matter gives rise to most of the observable electroencephalographic signals seen at It has long been recognized that the electrical the scalp. signals seen at the scalp are ultimately produced by brain activity at the neural level. What has Another key brain structure is the thalamus (see WAVES OF THE FUTURE OF WAVES been lacking, however, is a quantitative Figure 1), which relays all sensory information 148 connection between these two levels. In the except smell to the cortex. It also receives EEG Spectra Small departures from steady states can be studied by assuming that they are ‘linear’ – that is, twice the stimulus produces twice the change in the brain. If this is done, a large number of properties of brain activity can be calculated in terms of the ‘linear transfer function’, which is a measure of the ratio of activity change to stimulus.

Because the inputs to the brain are extremely complex functions of space and time, we approximate them as having an equal mix of all possible spatial and temporal scales, at least to a first approximation. If this is done, we can use the transfer function to calculate the spatial and Figure 3: Schematic of connections (indicated by arrows) temporal scales of the brain’s response to them. between the cortex and thalamus, showing the relay and reticular nuclei of the thalamus separately. External stimuli In particular, we can calculate the amount of each enter at the bottom. frequency of response is present – the ‘power spectrum.’ Figure 4 shows the EEG power Overall, we obtain a set of equations that can be spectrum obtained from a resting adult subject, solved to predict brain properties in terms of a overlaid with a predicted curve from our model. modest number of physiologically measurable We see prominent peaks near 10 and 20 Hz, parameters, such as ranges and signal speeds of which are the so-called ‘alpha’ and ‘beta’ axons, numbers, speeds, and strengths of rhythms, first discovered by Berger in the 1920s. synapses, and time delays in travelling between At high frequencies, there is a rapid fall-off in neural populations. power, while power climbs steeply at low frequencies, before levelling off below 1 Hz. Results Once equations for all the above features are written down, standard analytic and computational techniques can be used to solve them and extract predictions for the types of behaviour that the brain should exhibit when stimulated. Some of these are discussed in the following subsections. Steady Brain Activity The first question we ask is whether the brain can show stable, steady-state activity on average if it receives a constant average level of external stimulation. Solution of our equations shows that there are three possible steady states, two of which are stable. Of these two, one corresponds to all the neurons firing (spiking) flat-out, which relates to some kind of epileptic seizure. The other has average firing rates of a few spikes per

WAVES OF THE FUTURE OF WAVES second per neuron, which turns out to be in good Figure 4: Example power spectrum (solid) from a typical agreement with what is actually observed when adult subject in the relaxed, eyes-closed state. The dotted 150 electrodes are inserted into living brains. curve is a fit of our theoretical predictions to these data. The good match between model predictions and comparison between experimentally measured data seen in Figure 4 supports the validity of the and theoretically predicted evoked responses to model and means that we can try to interpret the an auditory tone. Again, there is good above features in terms of the physics in the agreement, and the parameters for which model. In particular, we note the following: agreement are found turn out to be very similar (i) The alpha and beta peaks are due to to those found for EEG spectra. This is not resonances in the corticothalamic closed loops, surprising, since both phenomena are produced mentioned above. These are similar to violin by the same brains. What is perhaps more strings, in having only certain frequencies at surprising is that these two types of phenomena which they can oscillate. have been studied almost independently for many (ii) The fast high-frequency fall-off is due to the years, in the absence of a theory that links them relatively slow rate at which dendrites and together. synapses can respond to incoming spikes – which means higher frequencies are blocked. States of Arousal (iii) The rapid rise at low frequencies is a sign that It has long been known that the EEG shows major the brain is close to instability, where it would differences between sleep, wake, and other cross into a seizure of some sort. The states of arousal. Figure 6 compares some of our frequency at which the spectrum levels off is model predictions for EEG time variations with a measure of the degree of stability; the lower corresponding experimental results. Again, good this frequency, the less stable the brain. If our agreement is found, based on the transfer brains were too stable, we would not be able function. to be flexible in our responses to complex inputs – but they mustn’t become unstable. Hence, they operate close to the edge of stability. Evoked Responses to Stimuli The transfer function also enables us to calculate the response of the brain to a short stimulus, such as a flash of light or a sudden tone –- the so-called ‘evoked response.’ Figure 5 shows a

Figure 6: Left column: Model time series representative of (a) alert, eyes-open, (b) relaxed, eyes-closed, (c) normal sleep, and (d) deep sleep. Right Column: Corresponding time series from human subjects (Nunez, 1995). Note that the vertical scales differ between the various frames.

Epileptic Seizures If we allow disturbances to become large, the behaviour of the model is no longer linear – doubling a stimulus could result in increase in the response by more (or less) than a factor of two, and may even send the brain off into an oscillatory or chaotic state. We have found that Figure 5: Experimental (solid) and theoretical (dotted) BRAIN DYNAMICS UNDERSTANDING evoked response potentials in response to an auditory the stability of the brain can be plotted stimulus. approximately in a three-dimensional space, as 151 shown in Figure 7. The x, y and z axes measure the strengths of cortical feedback on itself, corticothalamic feedback, and feedback within the thalamus, respectively.

Figure 8: Sample time series from the model in a regime corresponding to a petit mal seizure. This strongly resembles experimental time series from such seizures.

similarly realistic results for 10 Hz ‘grand mal’ seizures that do involve convulsions. New Measures of Brain Parameters If we make a prediction from our theory and compare it with experiment, we usually have to adjust the parameters of the theory slightly to get the best possible match to data from each Figure 7: Brain stability zone. Stable states lie inside the individual subject. Used in the reverse direction, tent-shaped surface. Approximate locations are shown of alert eyes-open (EO), relaxed eyes-closed (EC), normal this provides a new way to measure those sleep (NS), and deep sleep (DS) states, with each state parameters for that subject. We have used this located at the top of its bar. The quantities x, y, and z are approach to estimate a number of parameters, discussed in the text. such as axonal ranges and speeds, and synaptic properties, for the population at large and for All stable states of the brain lie within the tent- individuals. These estimates have proved to be in shaped region in Figure 7, with several specific agreement with independent ones obtained by examples shown. If the boundaries of the tent are physiologists and anatomists, thereby further crossed, the brain either zooms strengthening the basis of off to a sustained high-firing rate our theory. state (some kind of extreme EEGs are known to be epileptic seizure) or it enters an very different in One powerful potential oscillatory mode. Brain oscillations application of this approach is to are predicted to occur at different arousal and use it to measure parameter frequencies between roughly 3 disease states, and our changes between states of and 10 Hz, and can correspond to arousal, or between different seizures if their amplitudes method promises to clinical conditions. EEGs are become large enough. Figure 8 enable these differences known to be very different in shows an example of a roughly 3 different arousal and disease Hz oscillation obtained on to be interpreted in states, and our method promises crossing the lower left boundary terms of changes in the to enable these differences to be in Figure 7. The time course of interpreted in terms of changes this oscillation is very similar to underlying physiology and in the underlying physiology and that seen in ‘petit mal’ seizures, anatomy. anatomy. Initial work shows that which affect about 1% of children this approach has great promise at some point. During such a for opening up a new, non- seizure the subject loses invasive window on these subtle consciousness, but does not collapse or undergo brain changes. Corrective changes induced by convulsions. On return of consciousness, after 20 medications can also be monitored objectively to WAVES OF THE FUTURE OF WAVES to 30 seconds, they may even continue what they optimise drug dose and get the best outcomes 152 were doing beforehand. We have also found while avoiding overdosing. Functional Magnetic Resonance Imaging anatomists. The predictions of the model provide So far, I have mentioned only EEG-based a successful, unified description of a wide range measurements. Recently, we have extended the of phenomena relating to brain activity, including theory to calculate the change in blood oxygen the existence of steady states, EEG spectra, level caused by neural activity. This enables us to evoked responses, time dependence of EEGs, estimate the response measured in functional seizures, arousal changes, fMRI, and other magnetic resonance imaging (fMRI) experiments phenomena. that are sensitive to blood oxygenation levels. We are currently comparing the predictions in detail Fitting the model’s predictions to data provides a with experimental measurements. non-invasive probe of the physiology and anatomy, yielding parameter values that are consistent with independent measures, and which Commercial Applications will enable new objective measures of differences The ability to measure brain parameters using our in brain state due to arousal and disease, for methods is already being commercialised via a example. spinoff company, the Brain Resource Company, which was set up in 2001. A database of brain This work represents an extension of biological function and structure measures on over 4000 physics into a new field where quantitative normal and clinical subjects has been amassed, methods have been lacking, despite over 130 against which individual subjects’ measurements years of effort. A host of questions can be can be standardized. This enables deviations from addressed with the methods now available, and it normality to be rapidly detected and quantified, is possible to adopt a unified approach to and provides data that serve to test and calibrate previously unconnected sub-areas of theoretical efforts such as our own. neuroscience relating to brain activity.

Already, a number of significant features of the References and Further Reading data have been predicted and confirmed via Kandel, ER, Schwartz, JH, and Jessell, TM, Principles of modelling efforts. For example, it has long been Neural Science, 3rd Ed, Appleton and Lange, Norwalk, known that the frequency of the alpha rhythm Connecticut, 1991, ISBN 0-8385-8034-3. increases in childhood, peaks at an age of around 20 years, and then declines slightly thereafter. In Nunez, PL, Neocortical Dynamics and EEG Rhythms, Oxford our model, this is predicted to be connected with Univ. Press, Oxford, 1995, ISBN 0-19-505728-7. a speeding up, then gradual slowing of conduction velocities in axons, particularly those linking cortex Robinson, PA, Rennie CJ, Rowe, DL, and O’Connor, SC, and thalamus. The physics of the situation Estimation of Multiscale Neurophysiologic Parameters by immediately implies that there should be a parallel Electroencephalographic Means, Hum. Brain Mapp. 23, speeding up, then slowing down, of evoked 53-72, 2004. responses, and this proves to be borne out by the data. Interestingly, this effect is only discernable in enormous databases, such as the one operated by the BRC, because individual variations obscure it if only a few subjects are examined.

Conclusion Recent physiologically-based investigations of brain dynamics have resulted in a model that incorporates the main relevant features of corticothalamic physiology and anatomy in UNDERSTANDING BRAIN DYNAMICS UNDERSTANDING relatively few parameters, which can all be independently estimated by physiologists and 153 Einstein’s Miraculous Year /Part 3

2 Special Relativity – Einstein’s =magnificent c reation THIS IS THE one everyone knows about: Einstein’s theory of relativity. So it might come as a surprise that the actual paper, published in June 1905, was titled On the electrodynamics of moving bodies (Annalen der Physik 17, pages 891-921).

Electrodynamics? Moving bodies? waves through water. If the medium This led to another significant result What about the famous cosmic is in motion relative to an observer, in the lead-up to 1905: the result speed limit, c, the velocity of light in a this will affect the perceived speed of the Michelson-Moreley vacuum? What about bendable space of the wave – think of the circular experiments. If light travelled and warpable time? What about the ripples from a rock thrown in a through the aether, and the aether mind-bending twin paradox? river, travelling downstream with formed a background reference the current. frame, then the earth must also be It makes sense once you travelling through the aether as it understand the context in which But electromagnetic theory predicts orbits the Sun each year. Michelson Einstein produced this remarkable the speed of light is c, about and Moreley reasoned that this piece of theoretical physics. Two 300,000 km/s – but there’s no would mean light travels at physical results set the stage for explicit medium, no reference different speeds relative to an the revolution in 1905: one from frame. It was clear, then, that since observer on earth, depending on experiment, the other theoretical. light must travel through something, whether the light was going in the Both had to do with the nature therefore this something – called direction of the Earth’s passage of light. the aether – must form a through the aether, or at right background, an absolute reference angles to it. The first grew out of Maxwell’s work frame against which all other on the laws of electromagnetism, motion could be compared. which were still fresh, and their astounding success in unifying a wide range of diverse phenomena was still being celebrated across the scientific community. But one consequence of the electro- magnetic equations caused some concern: they predicted electro- magnetic waves with a precisely defined velocity in the vacuum. The trouble was, the theory didn’t specify what you were supposed to measure the velocity relative to.

With every other kind of wave, the speed is relative to the medium

WAVES OF THE FUTURE OF WAVES through which the wave travels: soundwaves through air, ocean 154 So they performed experiments to and Poincare were working from a detect the effect of our motion classical Newtonian point of view, through the aether – and despite and saw the constancy of the some really clever experiments, speed of light as an observed fact they found nothing: their results that needed to be explained. showed no significant change in Einstein’s great leap of insight was the speed of light when measured to assume the speed of light is in different directions. Which came constant in all frames of reference as a bit of a surprise at the time from the start, as a principle – and and led many people to propose all then to see what happens as a result. sorts of reasons for the null result. The insights in Einstein’s first paper Lorentz and Poincare in particular on relativity are contained within pushed Newtonian physics as far two very simple postulates: the as it could go with their laws of physics are the same in all mathematical insights. Using the inertial reference frames, and the usual ideas of Newtonian dynamics, speed of light (through a vacuum) Lorentz came up with the equations has the same value, c, in any of time dilation and length reference frame, no matter how the contraction, the classic equations light source is moving relative to usually attributed to special the observer. relativity, years before Einstein ever From these two ideas, the rest wrote them down. Poincare showed flows: time slowing down and that the ideas helped to explain speeding up, space shrinking and why the effects of our motion stretching – in fact, the classical through the aether weren’t seen by notions of space and time as an the Michelson Moreley experiments arena for physical events went out – in effect, the idea was that the window: space and time Newton was still correct, but Nature became players in the events of was being sneaky in ‘hiding’ the Einstein’s universe. Space could effects of the aether from us. become time, time could become space, and physics could never be So if Lorentz and Poincare were the same. there first, what’s the big deal about Einstein’s work in 1905? The EINSTEIN'S MIRACULOUS YEAR, 3 PART difference is in the underlying assumptions, and what those imply about the physical universe. Lorentz 155 What got you interested in science in the first place? I was good at geography in high school and decided to study it at University, which once I found PROFESSOR out I could study the coast, lead ANDREW D SHORT, to coastal geomorphology and School of Geosciences, marine science. University of Sydney is a Sydneysider who What were you like as a kid? developed an attraction Were you curious, pulling for beaches and surf at apart stuff to see how it an early age. He found worked? Were you always as an undergraduate at interested in science, or did Sydney in the 1960s you this come on later? could actually study I was always interested in the them, and did an surrounding environment – as a Honours thesis on child I loved going for long McMasters Beach. He walks, exploring the bush and then took his surfboard and headed harbour foreshore. When I was a to Hawaii when he completed his teenager I hitchhiked every Masters at the University of Hawaii, possible day to the beach to go followed by a PhD in Marine surfing. Once I was 17 I bought Sciences at Louisiana State a car and started exploring University where he was sent to Australia and its coast, which I study the usually frozen beaches of am still doing. the north slope of Alaska. He has since conducted beach research in What’s the best thing about the USA, Brazil, Europe, New Zealand being a researcher in your and around the entire Australian field? coast. After a postdoctoral position at I get paid to go to the beach. Macquarie University he has been at the University of Sydney since 1977. Who inspires you – either in His research and publications relate science or in other areas of to the evolution, nature and your life? dynamics of coastal systems, People who lead and do things, particularly beaches, as well as its they don’t wait to be told. application to managing our coast. What’s the ‘next big thing’ in science, in your opinion? What’s coming up in the next decade or so? A better understanding of our marine environment – we might finally get to know as much about our oceans and seabed as we do about Mars and the Moon.

156 Wind, Waves and Beaches

Professor Andrew D Short

Introduction HAVE YOU EVER sat on sandy beach and looked at the waves, perhaps breaking well out to sea before finally expending their energy as they reach the shore, collapsing and running up the beach as a thin layer of swash? That final thrust of energy and water, some of which runs back down the beach, while some soaks into the sand, may have had a beginning many thousands of kilometres away and many days ago.

Figure 1: As a wind streamline flows over the surface of a wave it pushes (stresses) harder the exposed windward face, while the pressure and stress is less on the leeward side of the crest. This produces shear stress on the face, as well as pressure differences between the windward and leeward sides. Both the shear stress and pressure gradient combine to increase wave height. 157 he energy that propels waves across the will then look at what happens to waves as they oceans and is finally expended at the shore reach the shallow waters of the coast and finally T is derived from the atmosphere, from wind break on a sand beach, and then what happens blowing across the ocean surface. The energy to form the typical wave and tide-dominated transferred to the ocean surface as waves can sandy beaches that surround Australia and much then transport this energy across vast ocean of the world’s coastline. distances with surprisingly little loss of energy. Once these streamlined, energy-efficient waves Wind and waves reach the coast, they rapidly transform as they Waves are disturbances of a fluid medium interact with the shallow seabed. The energy through which energy is moved. released in shoaling and breaking sets off a whole new chain of events that ultimately shapes The lower atmosphere contains four major belts the coastline and shore. Within the surf zone, of pressure which encircle the earth: the polar ocean waves and secondary long high pressure, the subpolar low waves, edge waves, standing pressure, the subtropical highs waves and shear waves, and The beach, and in and, around the equator, the associated currents, travel in all equatorial lows. Air is directions and, in doing so, particular the surf, may continuously flowing from the arrange every grain of sand into a look a little confusing and areas of high pressure towards predictable suite of bars, troughs, and into the areas of low rip channels, sloping beaches, daunting, but it is pressure and in doing so berms, cusps, megacusps ... The behaving in a very generates the worlds great beach, and in particular the surf, wind systems: the low velocity may look a little confusing and predictable way governed polar easterlies, the high daunting, but it is behaving in a by the laws of physics. velocity westerly of the roaring very predictable way governed by 40s, and the moderate velocity the laws of physics. trade winds which flow over the huge expanse of the subtropics. As 71% of the Sandy beaches make up 50% of Australia’s earth’s surface is ocean, most of these winds are 30,000 km long coastline. In fact, there are blowing across large expanses of the Pacific, 10,685 of them and each and every one has Atlantic, Indian, Southern and Arctic oceans, as been classified into one of 15 different beach well as the many smaller seas. As the winds types: six are wave-dominated beaches that interact with the ocean surface they generate occupy much of the high wave energy southern waves. half of Australia; three are tide-modified and four are tide-dominated beaches, both of which are Wind-wave interaction most prevalent across northern Australia where Wind energy is transferred to the ocean surface tides are high and waves lower; and the final two by both pressure fluctuations and tangential are fronted by intertidal rocks or fringing coral stress exerted by the streamlines (Figure 1). reefs. The type of beach produced is a function of Pressure fluctuations are expected to dominate the incoming waves, the tide range and the type because in nature wind is always turbulent at the of sand that forms the beach, acting within the surface, which permits pressure differences to local coastal environment. It is the waves, initiate and enhance wave motion. Once seas however, that supply the bulk of the energy to our begin to grow they increase surface roughness; beach systems. that is, the undulations in the surface manifest as wave crests and troughs. The increased In this chapter we begin where waves begin, roughness in turn increases the shear stress where the wind blows over the ocean surface, between the wind and ocean surfaces, which transferring energy to form waves – or, more builds larger waves. It also increases the pressure WAVES OF THE FUTURE OF WAVES correctly, sea. The sea then is transformed into difference through the sheltering effect, as the 158 swell and sets off across the ocean surface. We increasingly high waves shelter the backing trough, which generates additional pressure time the wind blows – the longer the duration the differences. Since the sheltering effect is bigger the waves. Third is the wind direction dependent on wave height there will be a which will determine the direction of wave travel. constantly changing balance between the two as Fourth is the fetch, or length of sea or ocean over waves grow to a fully aroused sea. which the wind can blow unimpeded – the longer the fetch the higher the waves. And fifth is water There are, however, two factors that limit the height depth, as shallow water (less than half the length) waves can reach, or to which they can be aroused. will mean wave interaction with the seabed and First is gravity, which, as wind is building the cause the wave to break. waves, attempts to restore the water to its original horizontal state. As crests of waves are pulled A fully arisen sea is therefore the maximum wave down by gravity the momentum continues below conditions that can be produced by a given wind the flat-water surface, developing a trough, and so velocity, duration and fetch blowing over a deep gravity waves are developed – another name for ocean. For light winds the waves will be very low, ocean waves. Second is wave steepness, the ratio while only very strong winds, blowing in a between the wave height H and length L. When H/L constant direction, for a low period (days) over a is about 1/7 the wave will form white caps and long fetch of deep water can generate the bigger break, thereby limiting its height. waves. Figure 2 illustrates the spectral energy transferred to a fully arisen sea for a range of Fully arisen seas wind velocities. All winds flowing over a water surface will generate waves, however the size of the waves the wind can build is dependent on a number of Deepwater waves factors. First is the wind velocity – the higher the While wind is blowing over the ocean surface it velocity the higher the waves, and as wind energy generates waves called a sea. By nature, sea is transferring to waves at an exponential rate, consists of waves that are relatively high but very high waves require very strong winds. short in length, and therefore steep. They are Second is the duration of the wind, the period of prone to breaking and travel in a broad range of

Figure 2: The amount of wave energy generated by winds from 20 to 40 knots. Wave energy increases exponentially with wind velocity. (From Neuman and Pierson, 1966.) WIND, WAVES AND BEACHES

159 directions either side of the wind direction. Once Wavelength, as illustrated in Figure 3, is the the wind stops blowing, or the seas leave the distance between successive wave crests or area of wind action, they rapidly transform into troughs. It is also related to the wave period T – what we call swell – swell waves are lower, the time between successive crests – such that longer and therefore flatter, and tend to travel in a 2 more uniform direction. Once formed into swell, L = g/2π T tanh (2πho/L) (1) waves can theoretically travel around the globe, though in reality they always run into something, where g is the gravitational constant and tanh such as an island or the coast. When both sea refers to the hyperbolic cotangent. and swell are travelling in wave depths (ho) where ho/L > 0.5 they are called deepwater waves, The speed at which the wave travels, C, is called the meaning the water is too deep for them to wave phase velocity and is given by the ratio L/T interact with the seabed (Figure 3). In reality most waves have lengths less than 350 m, while the C =L/T = g/2π T tanh (2πh/L) (2) oceans average over 4 km deep, meaning in the deep ocean all waves are deepwater waves. Both equations can be approximated in deepwater when h becomes large, causing tanh(h) ≈ 1, which enables equations 1 and 2 to reduce to

2 Lo = gT /2π (3)

and

Co = gT/2π (4)

2 Since g and π are constant, Lo = 1.56 T and Co = 1.56 T, meaning L increases at the square of the period, and C is directly related to T. It also means that waves with a longer T and L travel faster than shorter waves. Typical deepwater waves arriving around the southern Australian coast have T between 10 and 14 s, and therefore lengths between 156 and 306 metres and speeds Figure 3: Orbital motion beneath a deepwater wave. So between 15.6 and 21.8 m/s, or 56 and 78 km/hr. long as the depth is greater then L/2 the wave will not interact with the seabed. (From Sager, 1998.) Figure 4: Seas (to the left of the crossover) are relatively high and short in period and length. As they Deepwater waves have a number of predictable transform to swell (right of characteristics, which are illustrated in Figure 3. crossover) they become lower First, while the waveform is moving in the and longer in period and length, direction of wave travel, the water in the wave and as such can theoretically undergoes an orbital motion – that is, the water travel for thousands of kilometres. (From Davies, 1980.) moves around and around, as the wave crests and troughs pass overhead. This is why when you are in a boat at sea the wave passes underneath, while the boat just bobs up and down. The orbital motions extend down into the ocean, but decrease exponentially in size and orbit with WAVES OF THE FUTURE OF WAVES depth, so at depths greater that half the 160 wavelength (L/2) no orbital motion remains. Figure 4 illustrates the transformation of the higher strong westerly winds that encircle the southern but shorter and slower seas into lower but longer hemisphere at those latitudes. They blow year and faster swell, which can then travel great round across long ocean fetches to generate the distances with minimal additional loss of energy. worlds biggest wave factory. Because of the Coriolis effect (which is due to the rotation of the Earth) these westerly waves are deflected to the Wave climates left towards the equator, and so can travel from Because waves are dependent on the global wind south of Australia to Hawaii and California. regimes they can be classified according to the Likewise in the northern hemisphere similar part of the ocean, or wind regime, that produces waves are generated, but only in the northern them. Figure 5 illustrates the global distribution of winter, as the low pressure regions reside over waves that occur 10%, 50% and 90% of the the landmasses during summer. time. Note how the world’s biggest waves are located year round in the Southern Ocean The waves are termed storm wave environments, between 40ºS and 60ºS, where they are and the swell emanating from their stormy wave generated by the subpolar lows and factories are called west coast swell, as they tend accompanying ‘roaring 40s’ and ‘raging 50s’, the to arrive on the western side of the oceans’ basins, as well as southern Australia. The swell that has to bend and travel up the eastern side of Figure 5: Global distribution of wave heights occurring the continents, like the east coast of Australia, is 10%, 50% and 90% of the time. The biggest waves are generated year round in the southern oceans. (From Short, called east coast swell. The mighty though 1999, based on Young and Holland, 1996.) moderate velocity trade winds, which dominate the world’s wind systems, generate only low to moderate waves in the subtropics. The tropics are the site of the ‘doldrums’, calm balmy conditions which produce few waves; consequently the tropics only receive larger waves when they are generated outside the tropics by the systems mentioned above. Likewise in polar regions the polar easterlies are not only low in velocity for much of the year, they also blow over a frozen sea, so only low waves are generated in summer.

The world’s deepwater wave climates are therefore very closely related to latitude and the prevailing wind systems, as well as location and orientation of the coastline. Those facing towards the west coast swell receive these energetic waves even though they may be located thousands of kilometres distant. The deepwater wave height is, however, only half the story – to have beaches those waves have to get to the shore, and in crossing the shallow waters of the continental shelf a range of factors act to hinder their arrival.

Shallow water waves All waves eventually run into shallow water and in WIND, WAVES AND BEACHES the process transform into shallow waves. Shallow water occurs when water depth h < L/2, 161 since at this depth the wave orbits begin to depth decreases – in other words, waves shorten increasingly interact with the seabed, and in the and slow down as they approach the shore. process undergo a range of predictable transformations based on the depth and This leads to a range of shallow water impacts. configuration of the seabed. In shallow water First, waves shoal, meaning L shortens, which in tanh(h) becomes very small, enabling the turn concentrates the energy over a shorter following shallow water approximations to length, thereby increasing wave height. Waves equations 1 and 2: also attenuate, meaning they transfer some of their energy to the sea bed, for example by Ls = T √gh (5) Cs = √gh (6)

Both shallow water wavelength Ls and speed Cs are dependent on depth, and both decrease as

Figure 6: Waves shoaling, refracting and breaking along sections of the Tasmanian coast. WAVES OF THE FUTURE OF WAVES Figure 7: Wave breaking - a) surging, b) plunging and c) 162 spilling waves. moving sand, which will lead to a loss of wave In the deep ocean tsunamis travel very fast (up to energy and a decrease in wave height. Over 800 km/hr), are spaced approximately 200 km variable seabed, the waves move faster in deeper apart and have a period of about 15 minutes. water and slower in shallower water, causing the Tsunamis have a crest and a trough like all wave crests to bend and change direction, a waves. The trough usually arrives first, resulting in process known as wave refraction (Figure 6). a draw-down in the water and sea level, which Finally, if the water becomes too shallow the unfortunately can attract people onto the exposed trough slows faster than the backing crest, which seafloor. However within five to ten minutes the attempts to overtake the trough and in the first crest will arrive. While in the deep ocean process breaks. tsunamis may only be a few decimetres high, because they are so long and contain so much Breaking waves water they build in height as they approach the Wave tend to break when h = 0.78Hb, where Hb shallow coastline and slow down to 40-50 km/hr. is the breaker wave height. However the type of breaker will depend on the slope of the seabed or Once it hits the coast the wave not only increases sand bar over which the wave is breaking. When in height, up to ten metres and more in Sumatra the slope is gentle waves gradually peak and the and Sri Lanka, but because of its great length it tip of the crest spills forwards producing spilling just keeps on coming for several minutes, raising breakers. Moderate slopes result in a plunging the water level and flowing inland until it reaches breaker where the crest thrusts forward and curls ground higher than the tsunami. It then retreats over the face of the wave, producing what surfers for several minutes before the second and usually call a tubing wave. On steeper slopes the front largest wave arrives. This is followed by a series face and crest of the wave remain smooth and of increasingly smaller waves. the wave slides directly up the beach without breaking as a surging breaker (Figure 7). Tsunamis occur relatively frequently, particularly in the Pacific Basin, and major waves the size of the Aceh tsunami occur on average every one Tsunami hundred years. The devastation and lost of life The tsunami on 26 December 2004 focused caused by the Aceh tsunami was a result of a worldwide attention to this type of wave and its large (8.9 Richter scale) undersea earthquake potentially deadly impact. While winds have and the associated uplift of the seafloor occurring nothing to do with tsunamis, a brief description is close to a densely populated, low-lying coast. The warranted for those interested in their science. earthquake generated a large tsunami, which arrived at the Sumatra coast within 10 to 15 Tsunamis are waves generated by a sudden minutes as a ten metre high wall of water that impulse or impact in the sea. The source of the rapidly flooded the densely populated low-lying impulse is usually an earthquake that displaces coastline. The tsunami wreaked havoc further the sea floor up or down. In the case of Aceh, a afield in Thailand, Sri Lanka, India and east Africa, 1000 km long by 200 km wide area of seafloor with smaller remnants circulating right around the was suddenly lifted by ten metres. Tsunamis are globe within 24 hours. An early warning system also generated by large undersea and coastal could have warned many of the locations in time landslides (usually triggered by an earthquake), for evacuation of the local population. volcanic explosions (Krakatoa near Java generated a massive tsunami in 1883) and, more rarely, meteorite-comet impacts, which can Waves and beaches generate mega-tsunamis. The single impulse Once waves break they undergo a rapid almost generates several waves as the ocean surface instantaneous transformation, from a progressive gradually returns to normal, in much the same wave composed of potential energy that may way as the waves generated by a rock thrown have travelled thousands of kilometres, to a WIND, WAVES AND BEACHES into a pond. translatory wave full of kinetic energy that can do work in the surf zone. A translatory wave is one 163 where the water moves with the waveform, and waves in the deep ocean related to the groups or its energy is translated shoreward as a broken sets of higher and lower waves, called wave wave, also called a wave bore or white water. The groups. These groups cause the sea level to be surf zone is the area between the wave breaking slightly depressed under the high waves, and and the shoreline – it’s the most energetic and elevated under the lower waves, forming a long dynamic part of the Earth’s surface. wave with the same period as the wave group (which is typically several minutes). These long The energy released as the wave breaks is in the waves enter the surf zone unbroken. As waves form of turbulence, sound (the roar of the surf) travel shoreward, break and transform they and even heat. The turbulence moving towards transfer energy and water from the shorter gravity the shore stirs sand into suspension and carries it waves to these longer waveforms. This transfer shoreward with the wave bore. The wave bore reaches a maximum at the shore where the long decreases in height shoreward, eventually waves reach their maximum size and amplitude. collapsing into swash as it reaches the shoreline. As a rule of thumb the height of the long wave at the shore is one-third to one-half of the height of Breaking waves, wave bores and swash, together the breaker waves. with unbroken and reformed waves, all contribute to a shoreward momentum in the surf zone. As Long waves can also be manifest as standing this energy moves shoreward it is transferred into and/or edge waves. Standing waves are formed other forms of surf zone currents, namely when an incoming long wave interacts with an longshore, rip feeder and rip currents, as well as outgoing (reflected) long wave in the surf zone, so long waves and associated currents, all of which as to form a wave whose crest simply moves up can move shoreward, alongshore or seaward. and down, but does not progress – it just stands Eventually, all this water has to return seaward; it in place. The water under the crest, however, does so through three related mechanisms: it must move sidewards and return to the crest with may simply reflect off the beach face and travel each oscillation, thereby generating horizontal seaward as a reflected wave; it may flow currents associated with the standing waves. sideways and then turn and pulse seaward as a What we have, therefore, is a standing wave, relatively strong, narrow rip current; or it may periodically moving up (set-up) and down (set- contribute to the growth of standing (long) waves down), extending across the surf zone and a few against the shoreline, which also pulses water wavelengths out to sea. seaward via bed return flow, which is water flowing seaward underneath the breaking waves Standing waves are important because, on higher and wave bores. energy beaches with bars and surf, they determine the location of the outer bar and the So while waves deliver potential energy to the spacing of multibars. The bar crests form under shore, it is the transformation of the wave energy the antinode or crest of the wave, while the from potential to kinetic energy and a range of deeper troughs of the bar form under the node or secondary wave and currents processes that do trough of the wave. The horizontal currents the work in the surf zone to shape the beaches of moving between the standing crest and trough, the world. The most important secondary waves as mentioned above, are responsible for moving can be grouped under the general term of long sediment towards the antinodes and forming the waves, with subdivision into standing waves and bar crests. edge waves. Long waves are by definition long in period, usually greater than 20 seconds and up to Standing waves increase in spacing exponentially several minutes. They are also known as offshore, so when there are two or more bars infragravity waves – that is, beyond or longer their spacing likewise increases exponentially than ocean gravity waves. (Figure 8). The wavelength of a standing wave is WAVES OF THE FUTURE OF WAVES 2 Long waves form in the surf zone due to a Ls = g/2π Ts (2n+1) tanβ (7) 164 combination of factors. First, there are long Figure 8: Standing waves form in the surf zone in association with breaking waves, with the horizontal currents under the waves (arrows) leading to the formation of bars crests under the antinodes.

where Ts is the standing wave period, n standing wave mode and β the slope of the surf zone. We can therefore predict the standing wave length and bar spacing: the longer the period and lower the beach slope, the greater the wave length and therefore bar spacing.

Another form of long wave motion is edge waves. Edge waves are standing waves that are trapped in the surf zone and also propagate along shore as a series of standing crests and troughs. They therefore stand and oscillate both perpendicular and parallel to the beach. Like standing waves, edge waves are smallest and shortest at the shoreline and increase exponentially in size offshore. At the shoreline they generate a cellular circulation in the upper swash zone, which can lead to the formation of beach cusps. On ocean coasts cusps are usually spaced every 20 to 40 metres, which matches the edge wave spacing (Figure 9a). In the surf zone the edge waves have wavelengths ranging from 100 to 500 metres and are responsible for the cellular circulation that leads to the formation of rip currents spaced at similar distances, and the associated intervening crescentic bars (Figure 9b).

Therefore, while it is the shorter period gravity waves (sea and swell) that drive sand shoreward and form the general beach profile, it is the

Figure 9: a) Well developed beach cusps on Pearl Beach, WIND, WAVES AND BEACHES NSW; b) Crescentic bars and regular rip currents along Forster Beach, NSW. 165 longer period standing and edge waves that are shoreward, usually at a depth equal to a quarter responsible for rearranging this sand into shore of the ocean wave length – between depths of 20 parallel bars, and segregating the swash to form to 30 metres along the New South Wales coast. beach cusps and the surf zone currents to form The swash limit is as far as the swash reaches rip currents and crescentic bars. up the beach, usually an elevation of about 3 metres. Therefore to have a beach you must have sand and waves – and as mentioned Sand and beaches previously this combination forms half the Waves are only half the equation when it comes Australian coastline. to beaches. A beach is a wave-deposited accumulation of sediment lying between wave Just as the size of waves will vary around the base and the swash limit. They are usually coast, depending on the deepwater wave climate composed of sand but can form from cobbles and shoaling processes, so too the sediment that and boulders. Wave base is the depth at which composes the beaches ranges from fine sand to waves can pick up and move sediment coarse sand, cobbles and boulders. The impact of

Figure 10: Distribution of wave-dominated (WD), tide- modified (TM), tide-dominated (TD) and beaches fronted by rock or reef flats (RF) beaches around Australia. (From Short, 2005.) WAVES OF THE FUTURE OF WAVES

166 sediment size is to change the gradient or slope generate a reworking of the shoreline to form of the beach, from a low 1º gradient on a fine megacusps, regular undulations in the shoreline sand beach to a few degrees on medium sand, to the same spacing as the rip currents. 10º or more with coarse sand, and up to 15º on cobbles and boulder beaches. All this also There are three types of wave-dominated controls the slope and width of the surf zone, beaches. Under low waves (less than one metre) with the widest surf zones composed of fine and particularly if the sand is coarse, the beaches sand, while coarse sand, cobble and boulders will be reflective. Reflective beaches have no bars form a slope so steep that no bar or surf zone or surf zone; the waves arrive at the base of the can form. beach, surge up a relatively steep beach face and reflect back out to sea – hence the name It is therefore the variation in waves and sediment ‘reflective’. Beach cusps are usually present, as that produces the seemingly wide range of illustrated in Figure 9a. They represent the low- beaches present along the coast, ranging from energy end of the beach spectrum. the steep, narrow, protected beaches to the broad, low gradient beaches with wide surf When waves are between 1 and 2.5 metres they zones, large rips and massive breakers. Yet every tend to produce intermediate beaches, which are beach follows a predictable pattern of response, characterised by bars, breaking waves, surf and largely governed by its sediment size and cellular rip circulation. The rip and bar spacing is prevailing wave height and length. The next controlled by the edge waves and, in Australia, section discusses the types of beaches that can range from a small 100 metres in the north to up be produced by waves and sand around to 500 metres on high-energy southern Australia Australia, followed by the impact of increasing beaches (Figure 11a). This is the most common tide range, as is typical of northern Australia.

Australian beaches Around southern Australia the coast is exposed to the persistent high swell originating in the Southern Ocean. This generates a west coast swell that dominates from North West Cape in the west around to Tasmania in the east, and an east coast swell running up the southeast coast to Fraser Island. This energetic wave climate combined with a low (less than two metre) tide range results in an overwhelming occurrence of wave-dominated beach systems in the south. This contrasts with northern Australia where lower subtropical and tropical wave climates and tides up to ten metres produce tide-modified and tide-dominated beach systems (Figure 10). Wave-dominated beaches Wave-dominated beaches by definition have low tide ranges, as a result of which the breaker zone, surf zone and shoreline are relatively stationary. This enables all the wave and long-wave processes to imprint themselves upon the surf zone and beach morphology – bars, rip currents and beach Figure 11: a) Rip dominated beach with alternating bars WIND, WAVES AND BEACHES cusps all form in a relatively stationary surf zone and rip channels, Victoria; b) Dissipative beach with shore area. In addition, once rips are formed they usually parallel bars and trough, South Australia. 167 beach type around southern Australia and result in the formation of 13,500 beach rips around the coast.

Waves greater than 2.5 metres, when breaking over fine sand, produce the highest energy beach type called a dissipative beach, because the high waves dissipate their energy over a wide, low gradient multibar surf zone (Figure 11b). Large standing waves and vertically-segregated bed return flow dominates the surf zone circulation. Tide-modified beaches Tide-modified beaches occur when the tide range is between three and twelve times the breaker wave height. As shown in Figure 10, they occur in the higher tide range areas of northern Australia as well as in the South Australian gulfs and parts of Tasmania. The major difference between wave- and tide-modified beaches is the periodic oscillation of the tide, which shifts the shoreline, (a) surf zone and breakpoint backwards and Figure 12: a) Tide-modified forwards across the intertidal zone, over a beach with low tide rip distance of hundreds of metres. As a result, while highlighted by dye, all the breaking, standing and edge wave Queensland; b) tide-dominated beach with 500 m wide very processes can occur on tide-modified beaches, low gradient intertidal zone, because they aren’t stationary they have difficulty north Western Australia. imprinting themselves upon the shoreline and

(b) WAVES OF THE FUTURE OF WAVES

168 intertidal-surf zone. Consequently, while cusps This transformation both releases and transforms are commonly found on the high tide part of the the wave energy into a range of longer-period beach, bars and rip channels are usually absent secondary waves and currents, which are from the intertidal-surf zone (Figure 12a). There is primarily responsible for shaping the finer detail insufficient time to form them because of the of the surf zone and shoreline. The interaction of continual shift in the processes, and any incipient waves, sand and tides are responsible for fifteen form is reworked and smeared by the shifting different types of beaches, and within this system shoreline and breaker zone. of categorization we can classify all 10,685 Australian beach systems. Tide-modified beaches do however have a steep reflective beach at high tide typically containing References cusps, and a wide low gradient intertidal zone; Papers and books only at low tide do some form rips and bars as Davies, J L, 1980, Geographical Variation in Coastal the tide slows and turns. Development. 2nd Ed (Longman, London). Neuman, G and Pierson, W J, Jr, 1966, Principles of Physical Tide-dominated beaches Oceanography (Prentice Hall, Englewood Cliffs, NJ). Tide-dominated beaches are the most common Sager, D A, 1998, Introduction to Ocean Sciences (Wadsworth beach type across northern Australia (Figure 9). Publishing Co., Belmont, CA.). They require areas of very low waves and high Short, A D (ed), 1999, Beach and Shoreface Morphodynamics tide, such that the tide range is between 12-50 (John Wiley and Sons, Chichester). times the wave height. These conditions occur on Short, A D, 2005, Australia beach systems – nature and a third of the Australia’s beaches. These beaches distribution. Journal of Coastal Research. are typified by a small sandy high tide beach, Young, I R and Holland, G J, 1996, Atlas of the Oceans, Wind fronted by a very wide, low gradient sandy and Wave Climate (Elsevier, UK). intertidal zone, which may have subdued sand ridges. They may be flat and featureless (Figure Websites 12b), have the imprint of tidal currents and, NSW wave buoys finally, they may be formed of mud. Beyond, the http://www.mhl.nsw.gov.au/www/real_quick.htmlx shoreline grades into true tidal flats. Beach video images http://www.wrl.unsw.edu.au/coastalimaging/public/tweed/ Summary index.html Most energy is transferred in some form of wave motion, whether it be electromagnetic waves emanating from the Sun to warm our planet, or the sound waves you hear when listening to someone talking. Wind energy is derived from differences in atmospheric pressure, which are due to the differential heating of our planet – and so is ultimately derived from the Sun’s rays. The wind is able to transfer some of this energy into the surface of the ocean in the form of waves and ocean currents, both of which can continue to travel long after the wind has stopped blowing. This energy is therefore stored in the ocean surface. The wave energy is able to rapidly arrange itself into energy-efficient swell waves, and can potentially travel great distances to far shores and coasts. The potential energy carried WIND, WAVES AND BEACHES by the waveform, is converted into kinetic energy as the waves shoals and break across the shore. 169 Right Way to Kiss By Dr Karl Kruszelnicki

THERE ARE MANY different types of kisses – the soft fluttery kisses between a child and parent, the chaste closed-mouth kiss given to a grandparent or aunt, and of course, the wild rollicking hungry open-mouthed kisses between two people who love each other in a very special way. In an average lifetime, we spend about two weeks in kissing. So how come, if we do it so often, one third of us kiss the wrong way?

We humans have a long history whomever you are kissing, so brain in various birds and of kissing. Early Christians you can work out how to handle animals. One day, he happened kissed each other whenever them. And a third theory says to be stuck for 5 hours in they met. The bride and groom that we used to believe that Chicago Airport. He noticed that kiss after the marriage your soul lived in your breath, when a particular couple kissed, ceremony. Eskimos and and that kissing would marry they each tilted their head to Polynesians kiss by rubbing the breaths together, and fuse the right – so that each noses together, while the your souls for all eternity. This person’s nose was to their right inhabitants of Southeastern last theory is really cute, but of the other person’s nose. He India do their version of the kiss like all the other theories, is suddenly realized that he had by each pressing the nose totally unprovable. just seen a left-right preference, against the cheek, with the and that public places were a active person inhaling deeply. On the other hand, some good place to collect data. There’s deep (or French) kissing peoples have thought it was where one person’s tongue wrong to kiss. The state of Now being a good scientist (as goes a’roaming in the other Indiana in the USA has a law opposed to a perve in a brown person’s mouth – and we’ll stop making it illegal for a man with raincoat) he immediately set up right there. And as a complete a moustache to “habitually kiss some criteria. There had to be contrast, kissing was not very human beings”. In Hartford in lip-to-lip contact, the faces had visible in the old days in Asia, the state of Connecticut, it is to be aimed at each other, there because the bow was the all- still illegal for a man to kiss his had to be an obvious direction purpose greeting, and people wife on a Sunday. And in 16th of head-tilting, and finally, they would kiss only in private – so century Naples, in Italy, kissing couldn’t carry any luggage of course, you wouldn’t see it. was an offence that carried the because this could influence the death penalty. direction in which they tilted There are a few theories on their head. Over the next 2.5 how this habit came to be. One Even so, most of us today will years, he collected data on 124 theory claims that it all began in still kiss – so how can there be scientifically-valid “kissing ancient times with mothers a “wrong way to kiss”? Well, pairs” at airports, parks, chewing up food to pass it that’s the opinion of Onur beaches and railway stations. directly into the mouth of their Gunturkun from the Ruhr The results were clear-cut – baby. A second theory reckons University in Bochum, Germany. two-thirds of people tilt their

WAVES OF THE FUTURE OF WAVES that kissing gets you close His speciality is studying heads to the right. This, of enough to smell the mood, food differences between the left and course, means that one-third of 170 and recent adventures of right sides of the body and us kiss the wrong way. What’s going on? years, so that when we kiss, we Firstly, about 90% of us are will tilt our heads to the right. right-handed. About two-thirds of us (about 60-70%) prefer to Maybe for once this is too much use our right foot, right eye or science, and we would be right ear. So on average, we better off listening to the words humans tend to use our of the 18th-century Scottish right side. poet, Robert Burns:

Secondly, we humans tend to “Honeyed Seal of soft affections, turn our heads to the right Tenderest pledge of future bliss, (rather than the left) for our Dearest tie of young connections, last weeks in the uterus, and Love’s first snowdrop, virgin kiss”, our first six months after being born. and go right ahead and kiss in whatever way seems natural... So Dr. Gunturkun reckons that we humans start off with a preference to look to the right, and pay more attention to

events on our right. He also FROM Dr Karl’s book Bum Breath, Botox RIGHT WAY TO KISS reckons that this preference and Bubbles (Harper Collins Publishers) carries right on into our fertile Illustration courtesy Adam Yazxhi 171 What got you interested in science in the first place? Perhaps the enthusiasm of my father for astronomy (and science in general). He passed this on to me. I still remember when a total solar DR RAFFAELLA eclipse was visible in Italy, when I MORGANTI has worked was only a few years old, and how for the last few years at he tried very hard to explain to me the Netherlands how unique this phenomenon was Foundation for Research and make sure it would remain in my in Astronomy (ASTRON) memory! in Dwingeloo, as a member of the What is the best thing about being Westerbork Radio a researcher in your field? Observatory group. She I think doing scientific research is was born in Italy and great, regardless of the field! Of obtained her PhD in astronomy, I like the fact that the Astronomy at the community is so international, one University of Bologna. works with people and instruments Afterwards she enjoyed several years that are spread all over the world in Munich at the European Southern (and beyond!). Some of the Observatory (ESO) and at the telescopes are located in very Australia Telescope National Facility remote and fascinating places! in Sydney. Before moving to The Netherlands, she worked as staff If you could go back and astronomer at the institute of Radio specialize in a different field, what Astronomy in Bologna. Raffaella’s would it be? current research interests include the When I had to decide which course study of Active Galactic Nuclei (radio to take at the University, I was galaxies in particular), as well as the actually uncertain between geology, study of the interstellar medium in biology and astronomy. In my next normal and active galaxies. life, I would probably like to try geology!

What ‘s the ‘next big thing’ in science? It is going to be a great time to be an astronomer. The astronomical community is busy planning and building the new generation of telescopes. They will represent such a major step forward that will make possible to almost look at the beginning of the Universe, something that now we can only predict through numerical simulations.

172 The ever changing life of galaxies

Raffaella Morganti

ONE NIGHT I WAS, as usual, observing the sky with my telescope. I noticed that a sign was hanging from a galaxy a hundred million light- years away. On it was written: I SAW YOU. I made a quick calculation: the galaxy’s light had taken a hundred million years to reach me, and since they saw up there what was taking place here a hundred million years later, the moment when they had seen me must date back two hundred million years. Even before I checked my diary to see what I had been doing that day, I was seized by a ghastly presentiment [...] Italo Calvino, Cosmicomics

Figure 1 – The Milky Way and the galactic centre as seen from La Silla ESO observatory in Chile.

Figure 2 (opposite, background image) The Hubble Deep Field (HDF) was taken with the Hubble Telescope by staring at one tiny piece of sky for ten days. The image reveals more than 1500 galaxies.

173 imescales and distances in the Universe are Way actually went on for centuries, but the final amazingly large. Not only are the distances answer only came when astronomers were able to Tbetween objects enormous, the objects measure with reasonable accuracy the distances themselves are also huge. Most of the astronom- to these systems, thanks to Edwin Hubble in ical objects appear immutable in the sky but this 1925. Owing to the very deep images (see for is only because the timescales of changes are example Figure 2) that can be taken by the very usually much too long to be directly measurable powerful telescopes available at present, we can by humans. There are, however, other ways to observe galaxies so far away that they must have find out how these objects change during their formed very soon after the Universe started. life. In this chapter we will see how astronomers think some of the largest astronomical objects, the Galaxies are big! The number of stars that form a galaxies, are evolving as a result of the spectacular galaxy can go from 108, for a ‘dwarf’ galaxy, up to encounters – that can happen once or more in 1012 (a one followed by twelve zeroes!) for a large their lifetime – with other neighbouring galaxies. one. A galaxy can be so large that, for the stars at the periphery, it takes more than 100 million Galaxies are among the largest objects in the sky. years to make a full orbit. We live inside one of them, the Milky Way, and the great majority of what we see in the sky with Given the huge distances, the only way to study the naked eye are stars that belong to our galaxy. galaxies is through the information carried to us The Milky Way is quite an ordinary spiral galaxy by the light that they emit. The term ‘light’ in that, nevertheless, contains about 100 billion astronomy has a very broad meaning, as it does stars. The bright band that crosses the sky – not represent only the ‘visible light’ to which our from which the name Milky Way (see Figure 1) is eyes are sensitive (and that is only a tiny bit of derived – comes from the myriad of stars that the electromagnetic spectrum), but also the form its disk. To travel from its centre to the emission with wavelengths ranging from radio to location of the Sun would take almost 30 X-rays and gamma rays (see Figure 3). thousand million years ... that is, if one could travel at the speed of light. Not all wavelengths emitted by astronomical objects can reach the earth (see Figure 4) The Milky Way is just one of the many galaxies in because the atmosphere absorbs some of them. the Universe. It is quite amazing to realize that it Thus, the telescopes capable of observing these has only been only 80 since years we discovered wavelengths have to be located outside the that outside our Milky Way many similar worlds atmosphere using satellites or building them at exist (Figure 2). The discussion about the high elevation. Compared to only 50 years ago, existence of ‘island universes’ similar to the Milky astronomers now have instruments – both

Figure 3 - Light consists of electromagnetic waves that can have very different wavelengths. The figure shows what the different portions of the electromagnetic spectrum are called. The visible light we see with our eyes has wavelengths ranging from about 400 nm (1 nanometer is one billionth of a meter) at the blue end to about 700 nm at the red end. Light with somewhat longer wavelengths than red light is called infrared. Radio waves are the longest-wavelength of light – they can be

WAVES OF THE FUTURE OF WAVES a meter long! On the other end of the spectrum (with wavelength shorter than the blue light) is the ultraviolet and, with even shorter wavelengths, the X-ray and the gamma rays. Different bands of the electromagnetic spectrum can be observed using 174 different instruments. discovery that these objects can look very different when observed in different bands of the electromagnetic spectrum! Figure 5 shows the example of our Milky Way, but there are cases even more spectacular that we will see in this and the following chapter.

These differences are because the emission in the different wavebands is generated by different physical mechanisms, or by the same mechanism Figure 4 - Diagram showing the approximate depths to but under different conditions. For example, the which different wavelengths of light can penetrate Earth’s atmosphere. A large part of the electromagnetic spectrum emission from electrons moving at relativistic – except for visible light, a small portion of the infrared, speed in a magnetic field can be responsible for and radio – can be observed only from very high altitudes part of the emission at radio wavelengths. Gas at or from space. different temperatures emits in different wavebands: the higher the temperature, the shorter ground based and in orbit around the earth – to the wavelength emitted. Stars are responsible for observe the sky in all these different wavelength most of the emission at the optical wavelengths. bands. This has opened completely new possibilities to understand astronomical objects (like galaxies). In particular, it has allowed the How to make a galaxy One of the most outstanding problems in extragalactic astronomy, and one of the big Figure 5 - The Milky Way seen in different wavebands from radio to X-ray and observed using different instruments. puzzles that astronomers are trying to solve, is The different morphologies in the different wavebands are how galaxies form. Galaxies are believed to clearly visible. emerge from some small perturbations in the THE EVERCHANGING LIFE OF GALAXIES

175 Figure 6 - A schematic view of the two competing scenarios of galaxy formation.

(otherwise very smooth) Universe soon after the Two major scenarios can describe how galaxies Big Bang. Computer simulations show that these actually formed out of these condensations (see perturbations can grow to produce filamentary Figure 6). The first, considered by a larger structures, and from the higher density regions in community of astronomers as the most likely, is these filaments, galaxies form. The distribution of the so-called hierarchical scenario. In this picture, the regions of galaxy formation (if we let the there is no single epoch of galaxy assembly; simulations continue on!) would look remarkably rather, galaxies form and evolve continuously. The similar to the distribution of galaxies (and galaxy building up of galaxies is done by merging clusters) that we observe in the present smaller clumps that, over time, will form much day Universe. more massive structures. The second scenario is

Figure 7 - Examples of spiral and elliptical galaxies. On the left is NGC 2997, a spiral galaxy that contains a lot of gas and young stars, while on the right is the elliptical galaxy M87 (also known as NGC 4486, or Virgo A). The latter contains relatively little gas but has a powerful active nucleus in the centre associated with a super-

WAVES OF THE FUTURE OF WAVES massive black hole (see more about this galaxy in 176 the next chapter). the monolithic scenario, with an early and rapid a larger galaxy may completely tear apart a collapse of matter when the Universe was still smaller one. In other extreme cases, the galaxies young, followed by passive evolution of the get so close that they cannot avoid merging and galaxies thereafter. It is not easy to distinguish forming one object. Although less frequent, these between these two scenarios and there are pros events also occur to galaxies that are located in and cons in both cases. less dense environments (so called field galaxies).

Galaxies are mainly classified in three groups: elliptical galaxies that seemingly have little or no structure and little or no star formation or gas; spiral galaxies that are gas-rich and that have ongoing star formation; and a third group of so- called ‘irregular’ galaxies. Examples of a spiral and an elliptical galaxy are given in Figure 7.

If both elliptical and spiral galaxies have formed at the early stage of the Universe, as predicted by the monolithic scenario, how can we explain such a major difference in their appearance, or many other of their characteristics like, for example, the type of stars – young vs old – that dominates Figure 8 - Three impressive examples of interacting systems caught by the Hubble Telescope. (Top) The so- these systems? A possibility is that in the two called Tadpole Galaxy with the impressive 280 thousand systems the speed at which stars form is light-year tail likely formed by a small intruder galaxy that different. If the initial star formation happened very crossed in front of galaxy and was slung around by their fast and the gas was used up in an initial burst, gravitational attraction. (Middle) A rare and spectacular the galaxy will have a round shape and very little head-on collision between two galaxies – the Cartwheel Galaxy. The striking ring-like feature is a direct result of a gas left – and we would classify it as elliptical smaller intruder galaxy that passed through the core of the galaxy. Otherwise, if the star formation is slower, host galaxy. The collision sent a ripple of energy into space, the gas has time to settle and form a rotating blowing gas and dust in front of it. The Cartwheel Galaxy disk, and therefore the result is a spiral galaxy. was probably just a normal spiral galaxy before the collision. (Bottom) Two colliding galaxies nicknamed ‘The Mice’ because of their long tails. These galaxies will However, as mentioned above, the scenario eventually merge in one single galaxy. considered more likely by the astronomers is the hierarchical one, in which small condensations form first and then, through interaction and merging, they form larger structures like elliptical galaxies. One of the reasons why this is considered to be the way galaxies form is that observations are increasingly showing that galaxies are indeed still forming and evolving in the local Universe under the effect of interaction and/or merging.

What does this mean? Galaxies seldom live in a completely isolated environment. Very often, they live in clusters or groups, regions where the distances between galaxies is, on average, only a few times the size of a typical galaxy. Because of their motion inside these structures, galaxies are THE EVERCHANGING LIFE OF GALAXIES likely to interact with each other. This means that they feel each other’s gravity up to the point that 177 Figure 8 shows a few examples caught by the because of this, it is less gravitationally bound to Hubble Space Telescope; in these cases, the the galaxy. Thus, atomic hydrogen is much more distortion is clearly visible from the distribution of sensitive to disturbances produced by the passage the stars. In less extreme cases, the stars do not of another galaxy and this is shown by the show distortions and one has to look at the presence of tails, plumes and, in general, by the distribution of the gas to find some more subtle strange distribution of the gas. Figure 11 shows effects. Spectacular cases of galaxies distorted the example of NGC 4631 where the effects of by the effect of a collision or by interaction have interaction are only visible in the neutral hydrogen been known for a long time, but they were wavelengths; observing in the optical band, the previously regarded as kind of ‘weird’ cases. galaxies appear more or less undisturbed. Deeper and deeper observations are showing that these are probably just the ‘tip of the iceberg’, Neutral hydrogen has another interesting and interactions and merging – albeit in a less property. Because it can extend to much larger spectacular way – could affect the life of a radii compared to the main stellar distribution, it considerable fraction of the galaxies in the sky. can also provide a better measure of how much

Evolving galaxies It then follows from the above that even huge systems like galaxies can have a ‘private life’, evolving and changing exactly like a human being. The times involved in these changes are enormous compared to, for example, the human lifespan or even timescales of human civilization. However, from their morphology and physical characteristics it is possible to derive information about their evolutionary stage and, in particular, Figura 9 - Observations of the neutral hydrogen 21-cm line whether their quiet life has been disturbed by a are done using radio telescopes. In order to reach the close encounter. required resolution (in other words, to be able to produce sharp enough images that can reveal as many as possible As shown in Figure 8, interaction with a nearby details in the structure of the radio sources) many radio telescopes are working as interferometers. This means that companion can produce tail-like structures. The they are formed by a large number of single antennas and tails are the result of stars and gas pulled out of the final image is made by combining the signal from all the objects during the gravitational interaction. these antennas. The Figure shows five of the six dishes of Before looking in more detail at which kind of the Australia Telescope Compact Array located about 500 interactions can affect the life of a galaxy, it is km from Sydney. important to understand that these tails and the presence of interaction are not always so evident from the optical images, because they trace only the distribution of stars. A much more powerful way is to observe the distribution of atomic hydrogen. The most common element in the Universe, atomic hydrogen is believed to be associated with the very origin of the galaxies.

Atomic hydrogen (or ‘neutral’ hydrogen) emits a spectral line with a wavelength of 21 cm, and is Figure 10 - The almost face-on spiral galaxy NGC 6946. On therefore observable in the radio band using radio the left, a true-colour optical image (based on images from the Digital Sky Survey) is shown while on the right a deep telescopes (see Figure 9). In galaxies, the atomic

WAVES OF THE FUTURE OF WAVES image of the neutral hydrogen obtained with the hydrogen usually extends out to much larger radii Westerbork Synthesis Radio Telescope reveals how much 178 than the stellar component (see Figure 10) and extended the gas is compared with the stars. Figure 11 - An example of galaxy interaction detected by destroyed. In any case the large galaxy can get observing the neutral hydrogen: NGC 4631 (upper galaxy) and NGC 4656 (lower galaxy). On the left, an optical image a new input of gas that can be used, for of the field where no obvious sign of interaction is seen, on example, to form new stars. the right the same field but with superimposed contours If the interaction is much closer, the large representing the emission from the neutral hydrogen galaxy swallows a small satellite that hits or obtained with the Westerbork Synthesis Radio Telescope passes very close to it (a so-called ‘minor (regions with more contours correspond to a stronger emission). The large tails of gas between the interacting merger’). This process will not only bring new galaxies are clearly visible. gas but may also have a major impact on the kinematics (and perhaps morphology) of the main galaxy. mass is in the galaxy. Essentially, the faster the A ‘major merger’ happens when two galaxies, neutral gas rotates (in regions as far as possible of about equal mass, get close enough to from the centre), the more mass is present inside collide and merge. This gives rise to a ‘new’ that radius. This can be compared with the mass galaxy that can be quite different from the of the visible matter (stars or gas) to estimate progenitors! whether the mass of the galaxy is all due to this Finally, we should not forget that the normal (visible) matter, or whether there is more hiding cycle (from life to death) of stars affects the somewhere. The rotation velocity of the neutral interstellar medium of the galaxy. For example, hydrogen is often observed to stay almost the supernovae phase is crucial in ‘polluting’ constant outside the region where the stars are the medium. This happens in the life of every observed. But if the matter in the galaxy were galaxy. In addition to this, some galaxies may due predominantly to the presence of stars, the experience a phase of intense star formation velocity of the gas would have been expected to (the ‘starburst phase’). decrease with distance from the edge of the optical body of the galaxy. The fact that this does not happen means that there is some other The nearest interacting system: our matter – the so called dark matter, as it has not Milky Way yet been observed directly, only indirectly – that Ideally, one would like to study the theory of keeps the velocity of the gas from decreasing. galaxy formation and evolution by observing the The presence of dark matter around galaxies is a distant Universe. However, this does not always crucial ingredient in the theoretical models of give all the physical details needed by the galaxy formation. astronomers. Thus, the objects in the nearby Universe become important targets; the extreme Coming back to the evolution of a galaxy, there example is the study of stars and gas in our own are different ways in which a galaxy can change: Milky Way. Slow and quiet accretion of gas due to gravitational attraction can happen if a small A giant stellar stream surrounds the Milky Way THE EVERCHANGING LIFE OF GALAXIES galaxy passes near a large one. The small Galaxy, originated by a companion dwarf galaxy companion may or may not be completely called the Sagittarius dwarf. Furthermore, a 179 supports the idea that the epoch of galaxy formation is continuing even now, although at a much slower rate. These streams are perhaps a generic feature of almost all galaxies.

While it is very difficult to see the stellar streams in the visible spectrum in galaxies outside the region of our Local Group, it is much easier to detect them in neutral hydrogen – as we shall see shortly.

Figure 12 - Image of the distribution of the neutral Stealing from your neighbour hydrogen in the As our telescopes become increasingly sensitive Magellanic Stream. thanks to improved technology, we discover that Darker regions represent many galaxies have very likely experienced (at zones where the signal is more intense. The two least once in their lifetime) some kind of dark clouds correspond interaction. Some examples have been already to the Magellanic Clouds shown in Figure 8 and a few more will be connected by a bridge of described below – but in fact, there is an endless neutral hydrogen. The tail list of cases. The web links at this end of this going up from these Clouds is the Magellanic lecture will guide you to some of the sites where Stream. The Milky Way is you can see more examples of them. just outside the figure at the bottom.

narrow tail of neutral hydrogen, called the Magellanic Stream, has been known for many years. This stream trails the (Small and Large) Magellanic Clouds – the two closest companions of the Milky Way – in an orbit around the Milky Way.

More recent studies with the Parkes radio telescope in Australia have shown that this stream of neutral hydrogen is much more extended than previously thought, and it extends in the opposite direction to what astronomers expected (see Figure 12). Astronomers consider this a further confirmation of tidal interactions between the Magellanic Clouds and the Milky Way. This means that the Magellanic Clouds will be slowly torn apart and destroyed, and their gas and stars will become part of our Galaxy. Figure 13 - An optical image of the galaxy NGC 3359 (grey A stellar stream has been also found around the scale) with superimposed contours representing the Andromeda Galaxy (or M31), the nearest large intensity of the emission from the neutral hydrogen (denser contours represent regions of more intense emission). The galaxy to the Milky Way. The stream found there, WAVES OF THE FUTURE OF WAVES data were obtained with the Westerbork Synthesis Radio coming from a small companion of this galaxy, Telescope. The faint tail-like emission connecting the big 180 supplies Andromeda with gas and stars. This galaxy NGC 3359 with the faint companion can be seen. As mentioned above, galactic interactions can be lanes are thought to be the left over from a small, quite gentle. A large galaxy can gravitationally pull gas-rich companion that was swallowed by the gas out of a smaller companion. Figure 13 shows big galaxy. This galaxy, only about 10 million a case where a faint connection of neutral light-years away from us, is the nearest large hydrogen is visible between a large galaxy, NGC galaxy of this type (despite the large dust lane, it 3359, and its distorted companion (almost can be considered in many respects an elliptical invisible in the optical image). The neutral hydrogen transferred from this small companion is less than 3% of the atomic hydrogen mass of the large galaxy. This kind of interaction is very much reminiscent of that seen in the case of the Milky Way and M31.

This interaction is not producing any major change in the morphology and characteristics of the large galaxy. The effect of this slow accretion (that in some cases can go on for billions of years) is mainly to stimulate new star formation in (a) the large galaxy, both through the new gas supplied by the unlucky companion and from the compression of the gas produced by the interaction itself. This is one of the ways to continuously supply gas to a galaxy and keep it growing.

Swallow a fly The small companion can, however, be much less fortunate. Depending on how the encounter (b) happens (the so-called initial conditions, like the relative direction of the encounter, the relative rotation of the two objects, etc.) the effects for the main galaxy can be very different. The middle image in Figure 8 shows one of the nicer examples of head-on collision, with a small galaxy hitting a larger one – the famous ‘Cartwheel’ galaxy. The collision has produced a shock wave that is now propagating at a speed of 300,000 km/h through the surrounding gas. The compression of this shock wave in the gas is making the attractive ring-like feature around the galaxy. (c)

Finally, the small companion can end up Figure 14 - (a) Optical image of the galaxy Centaurus A completely captured and swallowed by the large (also known as NGC 5128) obtained with the 8-m VLT ESO telescopes in Paranal (Chile); (b) an infrared image of the galaxy. This must have happened in one of the same galaxy obtained by the Spitzer Space Telescope; (c) most famous galaxies in the sky, Centaurus A. optical image of Centaurus A (taken from the Digital Sky The main body of this elliptical galaxy is divided Survey) with superimposed contours representing the in two by a spectacular lane of dust. At the emission from the neutral hydrogen obtained with the Very optical wavelengths, the dust appears as dark Large Array (more contours indicate stronger emission). Most of the neutral hydrogen follows the dust lane but THE EVERCHANGING LIFE OF GALAXIES patches where the light from the stars has been some ‘left-overs’ can be seen in a kind of half-ring absorbed (see Figure 14a). Structures like dust structure. 181 galaxy) – it also hosts at its core an active black Violent Collisions hole emitting copious amounts of radio waves! The most impressive galaxy mergers are those between two galaxies of similar size (known as It represents, therefore, a great laboratory for major mergers). The system known as the many phenomena. We will meet this galaxy again Antennae represents one of the most amazing in the next lecture. Figure 14b shows how cases; Figure 15 shows the extremely complex different the galaxy looks when observed in the structure in detail. The system consists of the infrared. At these wavelengths, the dust collision between two galaxies, and the striking reradiates the light absorbed from the stars. features of the picture are the two huge Astronomers believe that the peculiar geometrical ‘antennae’ that represent gas and stars pulled shape of the dust emission is the result of the out of the galaxies during the encounter. twisting and warping of the infalling spiral galaxy Numerical simulations show that these structures as it was captured by the large elliptical. are indeed the result of such an encounter if the galaxies passed by each other in the same sense Another reason why this galaxy is considered the as the rotation of each disk of stars (so-called result of a merger with a small companion is that prograde encounter). In this case, the outer rings elliptical galaxies are usually found to be very of stars and gas are ripped off from each of the poor or lacking in gas. However, when observed galaxies, resulting in the formation of the in the 21-cm line, Centaurus A shows emission spectacular tails. from the neutral hydrogen (see Figure 14c) indicating that a relatively large amount of this gas is present in this galaxy. This can only happen if the gas has been recently ‘donated’ to this galaxy by a merger with a gas-rich object. Figure 14c shows that the neutral hydrogen is mainly concentrated along the dust lane, and this is consistent with the idea that this structure (as well as the neutral hydrogen) has an external origin. However, some gas is also found at even larger distances from the centre (60,000 light- years or more) and is considered to be ‘left overs’ from the merger.

Both in the case of the Cartwheel and in the case of Centaurus A, the merger has had a clear effect also on the large galaxy – that is, the galaxy that Figure 15 - The two colliding galaxies known as Antennae has swallowed the fly! In particular, after the (NGC 4038 on the top and NGC 4039 at the bottom). The figure on the left shows the neutral hydrogen in blue (from merger the stars and gas in the galaxy can display observations with the NRAO Very Large Array) super- some very peculiar kinematics. Astronomers have imposed on an optical image from the CTIO 0.9m in green found amazing cases of stars rotating in one and white. This image clearly shows the long tails created direction while gas rotates in the opposite direction! by the interaction. The image on the right shows the centre In other cases, stars and gas rotate around of the two interacting galaxies as observed by HST. completely different (often perpendicular) axes. It is very difficult to imagine how such complex kinematics can be possible, and the only way There are many cases known of major mergers and appears to be if the gas has been acquired later by studying their characteristics it has been possible (once the stars in the galaxy were already well to construct a sort of evolutionary sequence. formed) through an encounter. In this case, the For example, the Antennae still show the two kinematics of the gas depends on the geometry of separate bodies of the merging galaxies and WAVES OF THE FUTURE OF WAVES the encounter, and therefore can be completely prominent tails of stars and gas. Large quantities 182 different from the kinematics of the stars. of neutral hydrogen (many times bigger than our Figure 16 - A composite image of the merger remnant NGC 7252 showing the optical light (green), star forming regions (yellow and pink) and the cold atomic hydrogen gas (blue) observed with the Very Large Array. This system is the result of two spiral galaxies, which collided and merged into a single object.

Figure 17 - The optical image of the elliptical galaxy NGC 5266 (yellow) superimposed with the emission from the neutral hydrogen (red) as observed with the Australia Telescope Compact Array. More than 10 billion solar masses of neutral hydrogen are observed around the galaxy, believed to be the result of a merger between two large and gas-rich spiral galaxies. Note how much more extended is the emission of the neutral hydrogen compared to the stellar light. THE EVERCHANGING LIFE OF GALAXIES

183 still be rich in atomic hydrogen and this characteristic allows galaxies that have formed through major mergers to be recognised. Indeed, NGC 5266 (the galaxy shown in Figure 17) has more than 10 billion solar masses of atomic hydrogen, one of the largest amounts found in an elliptical galaxy.

Astronomers have put this scenario together based on observations. However, computer simulations also show us that, under certain initial conditions (admittedly some of them quite uncertain), the characteristics of the observed Figure 18 - A computer simulation showing the evolution mergers can be nicely reproduced. In Figure 18 with time of a merger between two gas rich spiral galaxies. we can see one example of how a numerical The time interval (time increasing from left to right) simulation reproduces different steps of the between frames is roughly 100 Million years. The merger. The similarities with the images of real simulation shows clearly the creation of large tidal tails (similar to those observed in the Antennae) that fade away objects in Figures 16, 17, 18 are remarkable. as time goes by. At the end only the central object (merger of the two progenitors) will remain. A final important remark: it is interesting to notice that the gas brought by the merger will be used Galaxy) are observed from the centre of the not only to form stars but it will also provide fuel merger to the end of the tails. This merger likely to the central regions of the galaxy and, in started a few hundred million years ago. particular, to the central black hole. In this way, as we will see in the next chapter, the black hole Figure 16 shows instead the case of NGC 7552, can become active and produce enormous also the result of a major merger, but one that is amounts of energy at different wavelengths and much more evolved. The two progenitor galaxies in different forms! are already so well amalgamated that now they are just one object. This is extremely interesting in our quest to understand how galaxies form New gas, new stars! because the resulting object now resembles an One of the effects of accreting new gas either in elliptical galaxy (although the long tails still clearly a quiet way or through a violent merger, is to indicate its merging origin). The merger in NGC stimulate new star formation. In fact, mergers not 7252 probably started about a billion years ago. only bring new gas but also induce compression Interestingly, a large amount of neutral hydrogen of the gas that then can trigger star formation – is still present but it is mainly concentrated along in the high density, condensed gas, gravity can the tails. By now, the atomic hydrogen in the start to dominate, giving rise to the subsequent central region has been converted into stars; the production of stars. This process, if taken to the merger has gone through an intense phase of extreme as in the case of the so-called starburst star formation as a result of the violent galaxies, can have a major impact on the compression of the gas during the initial phase of structure, morphology and evolution of the galaxy. the merger. There are many spectacular cases of galaxies going If we are patient enough to wait few more billion through a starburst phase. Figure 19 shows the years (!) the merger will finally produce a real impressive case of NGC 3079. Some of the elliptical galaxy as shown in Figure 17. By now, observed starburst galaxies have a star formation the long tails have faded away and the atomic rate exceeding 100 stars per year, more than 100 hydrogen (once that the starburst phase is over) times the rate in our galaxy. It is clear that this WAVES OF THE FUTURE OF WAVES has slowly moved back to the galaxy under the extreme rate has a major influence on the 184 effect of gravity. The resulting elliptical galaxy will gaseous medium in the galaxy, and on the supernovae explode at about the same time, the overall effect is the creation of a super bubble, a bubble of hot gas (heated by the shock waves) so large that it will travel through the galaxy until it breaks free in the intergalactic medium. It then becomes more of a wind of hot gas that can push aside everything in its path. The consequence for the galaxy can be enormous. Figure 19 shows how spectacular these winds and their associated gas outflow can be.

Figure 19 - The spectacular starburst galaxy NGC 3079. A Large amounts of dust, the first by-product of Chandra’s X-ray image (blue) has been combined with an stellar formation, characterise star forming optical image from the Hubble Space Telescope (red and regions. The important characteristic of dust is green). The filaments consist of warm (about ten thousand that it can obscure these regions when they are degrees Celsius) and hot (ten million degrees Celsius) observed at optical wavelengths: dust absorbs formed by a super wind produced by a burst of supernova activity or from the super massive black hole. optical radiation and re-emits it at longer wavelengths, in particular in the infrared. Observations in this band of the electromagnetic galaxy’s structure. At such high rates of star spectrum can only be done either from outside formation, the number of supernovae that will the atmosphere or from telescopes located at explode (when the stars reach the end of their life) high altitude. is also about 100 times the rate in our galaxy.

Figure 20 - The distribution of morphology of galaxies Each supernova that explodes will produce a extracted from the Hubble Deep Field (see Figure 2) but shock wave through the gas medium around it, grouped in order of increasing redshift, corresponding to which creates a kind of bubble. If many increasing distance from us. THE EVERCHANGING LIFE OF GALAXIES

185 Only in recent years have technical improvements Universe. In this phase, the Universe was much made possible the study of starburst galaxies, more dense than the one of today, because it which are otherwise mostly obscured at the was much smaller. It is, therefore, not surprising optical wavelengths. These capabilities are that a lot of effort from the astronomical particularly crucial for the study of this community is now concentrated on the study of phenomenon in the distant Universe. distant objects. Because of their distance, these objects are small and faint and making these observations, even with state-of-the-art Galaxies in the distant Universe instrumentation, is extremely challenging. The objects considered so far are part of the local Universe – this means that what we observe One could write an entire lecture (even a whole corresponds to how they look today. The series!) on the recent studies of distant galaxies, distances from Earth to the galaxies we have but let’s just mention here two of most seen so far are not more than a billion light years, interesting recent results. Extremely remote and therefore the light we see now left these objects have been observed that are so far away galaxies a billion years ago. This is less than a that it corresponds to when the Universe was tenth of the estimated age of the Universe, which only about one-tenth of its current age. The is about 13 billion years. morphology of these distant galaxies appears, as expected, much more distorted than in the On one hand, it is very important for astronomers galaxies that populate the local Universe (see to study these objects because of the detailed Figure 20). Almost all the observed distant information that can be obtained and then galaxies show some kind of distortions and the compared with theoretical models. On the other separation in spiral and elliptical galaxies hand, if we want to know how galaxies form and becomes much more complicated, if almost evolved, the distant universe is the place where impossible. This indicates that interaction and the main ‘action’ is happening. Interaction merging do seem to play a major role in this happens more often in dense environments, and phase of the Universe. However, some of the so interactions between galaxies were probably objects observed seem to have at least the WAVES OF THE FUTURE OF WAVES extremely frequent (and therefore extremely central bright part, looking very much like an 186 important in shaping the galaxies) in the early elliptical galaxy. This would suggest that elliptical galaxies can form quite early in the life of the In conclusion, many open questions still remain Universe – but to make a nice, proper spiral for astronomers over the formation and evolution galaxy could take much longer. of galaxies. Nevertheless, they will continue on this endeavour using both the detailed studies of Another remarkable discovery of the last few the nearby spectacular cases and the study of years is the existence of a population of distant the mysterious objects at the edge of the galaxies that appear to be forming stars at very Universe. Improvements in technology that will high rate. These galaxies are among the most produce the next generation of telescopes will be luminous objects in the Universe and are powered crucial in making this difficult task possible. by a star burst (and probably also by a nuclear black hole), with star formation rates up to hundred times higher than the typical starburst Some interesting Web links: galaxy observed in the present-day Universe (so Hubble Telescope Picture Gallery and Outreach: up to 10,000 times the star formation rate of our http://hubblesite.org/gallery and galaxy). These galaxies were discovered using http://www.stsci.edu/outreach sub-millimetre observations (wavelengths ESO outreach page: http://eso.org/outreach between radio and infrared). Now called ‘sub-mm Astronomical photographs from David Malin images: galaxies’, these galaxies are mostly invisible (or http://www.davidmalin.com/ very faint) in the optical band while they are very NRAO image gallery: http://www.nrao.edu/imagegallery/ bright in sub-mm – this is thought to be the A living HI Rogues Gallery: result of the large amount of dust originating http://www.nrao.edu/astrores/HIrogues/RoguesLiving.shtml from the rapid star formation going on in Numerical simulation of interacting galaxies in the home page these objects. of Joshua Barnes http://www.ifa.hawaii.edu/faculty/barnes/barnes.html The few of these galaxies that have been studied Web links to see how radio telescopes work: in detail almost exclusively exhibit disturbed and http://www.astron.nl/p/astronomy2, unusual morphologies, in line with the idea that http://outreach.atnf.csiro.au, http://www.nrao.edu/students the large rate of star formation is the result of THE EVERCHANGING LIFE OF GALAXIES mergers, as expected in the hierarchical structure formation scenario. 187 Einstein’s Miraculous Year /Part 4

The Final Paper of 1905: that Famous Equation ASK ANYONE ON the street, What equation do you associate with Albert Einstein?, and most will answer, E = mc 2. They may not know what the equation means, but they know it’s classic Einstein.

This famous equation didn’t appear In this quote from the paper, the But when you add up the mass of in Einstein’s first paper on Special square brackets indicate where the these two atoms and the extra Relativity, which was described in modern symbols for mass, energy neutrons, the total is less than the Part 3 of this series. In September and the speed of light are used original Uranium atom’s mass. of 1905, he published yet another instead of Einstein’s original Where did the extra mass go? paper, titled Does the Inertia of a symbols. Body depend on its Energy When the Uranium splits, the Content? (Annerlen der Physik 19, Moving around this derivation, we of neutrons that escape have a lot of page 639). In this paper, Einstein course wind up at E = mc2. Not energy – this energy is the considers a body emitting a only does mass seem to be “missing mass”. When you use the quantify of energy in the form of equivalent to a form of energy, but famous E = mc2 equation, and radiation; he concludes that: energy ‘transmits inertia’ between factor in the energy carried by the bodies. We’re used to thinking of neutrons, you find everything If a body releases the energy [E] mass as something fixed and solid, adds up. in the form of radiation, its mass so Einstein’s conclusion that the decreases by [E/c2]. Since mass of an object depends on how obviously here it is inessential much energy it has came as a bit that the energy withdrawn from of a revelation. In particular, if an the body happens to turn into object is moving, it has kinetic energy of radiation rather than energy – and so an object’s mass into some other kind of energy, increases the faster it moves! we are led to the more general conclusion: The mass of a body But this isn’t just about the energy is a measure of its energy an object has through motion. content; if the energy changes Other kinds of energy – by [E], the mass changes in the gravitational potential energy, The quantity of energy you can get same sense by [E/c2]. electric potential energy and so on in this way is apparent when you – change the mass of an object as consider the energy output of If the theory [this derivation of m well. One particularly interesting nuclear power stations and, even = E/c2] agrees with the facts, application of this equation occurs more frighteningly, nuclear then radiation transmits inertia in nuclear physics, when an atom weapons. between emitting and absorbing of Uranium is split, for example. bodies. The Uranium atom decays into two lighter atoms – Barium and Krypton – and several stray neutrons. WAVES OF THE FUTURE OF WAVES

188 EINSTEIN'S MIRACULOUS YEAR, 4 PART

189 190 Monsters lurking in the centre of galaxies

Raffaella Morganti

THE UNIVERSE IS FULL with strange objects, but Active Galactic Nuclei (or AGN) are among the most spectacular of them all. When we look at galaxies in the sky using an optical telescope, most of them may appear like relatively normal elliptical or spiral galaxies. However, some of them hide a monster in their very centre. In these galaxies, the nucleus alone – that means a region not more than a few light-years in size – can be up to 10,000 times more luminous than the rest of the entire galaxy!

Figure 1 - Four examples of Active Nuclei: (below left) the Seyfert galaxy NGC 1068; (below centre) the quasar 3C273 observed by Hubble Space Telescope and two radio galaxies Centaurus A (opposite) and 3C31 (below right). The image of 3C31 is the superposition of a radio image (red) and an optical (blue). The image of NGC 1068 is a composite X-ray (blue and green) and optical (red) image. The X-ray shows gas blowing away in a high-speed wind from the vicinity of the central supermassive black hole. A composite image (opposite) is shown for Centaurus A: Chandra X-ray image (blue); radio 21 cm image and continuum (pink and green); optical (yellow).

191 or a while this incredible amount of energy that emit a lot of their energy in the radio band, released by these nuclei puzzled while others emit more in the X-ray band. The F astronomers. The energy from stars (even reason for these differences is still not completely considering a large number of them squeezed in understood; it could be related to some subtle this tiny space) is not enough to explain it. A differences in the structure of the various AGN or machine capable of doing this amazing job has to in the way they have formed. This makes, of be exceedingly massive and the only candidate is course, the job of classifying and understanding a super-massive black hole. Thus, we can AGN very complicated. Despite these diversities, describe an AGN as a ‘very compact region at the astronomers think that all AGN have some centre of a galaxy emitting a large amount of characteristics in common and therefore they can energy that cannot be produced only by stellar be grouped under only few categories. We shall emission’. The galaxy (either elliptical or spiral) see why shortly. hosting this active nucleus is then called an active galaxy. How many galaxies host an AGN? The presence of a central black hole in galaxies could be The phenomenon of AGN can manifest itself in relatively common. Even our galaxy, the Milky many different ways and it is impossible to Way, hosts one, although it isn’t so very big. summarize all in one chapter of this book. Figure However, active galaxies are only a small fraction 1 shows some examples of different types of of the total number of galaxies in the Universe. Of AGN: a so-called Seyfert galaxy, a quasi-stellar the relatively nearby galaxies, only at most a few object (quasar) and two galaxies particularly percent of them host an AGN. strong at radio wavelengths (radio galaxies). Some of the effects of the active nucleus in these Thus, the presence of a very massive and objects are shown in the pictures: compact object, the black hole, in the centre of a the strong X-ray emission coming from the galaxy is not enough to turn a quiet nucleus in an active nucleus of the Seyfert galaxy, that would AGN. The black hole must be active – into other otherwise be a normal spiral galaxy; words it should be supplied with fuel that allows the very bright nucleus of a quasar – which it to produce the huge amount of energy that we makes this object look like an ordinary star, but observe. Although they are relatively rare, the it is in fact several billion light-years away. In study of this phenomenon is extremely important: this object nothing stops us from looking the phenomenon is so extreme that we cannot directly at the powerful active nucleus; avoid being curious and making an effort to the striking difference between the emission understand the physics behind it. Moreover, it can detected at radio and X-ray wavelengths, have a major influence in the life of the galaxy. compared to the morphology of the optical The AGN phase is likely to be very destructive for emission in the Centaurus A galaxy (an object the galaxy – the release of so much energy can that we have met already in the previous have a major impact on, for example, the chapter); and interstellar medium and therefore even affect the the different morphology, again, but also the galaxy’s subsequent evolution. So the study of the strikingly different size of the radio emission AGN phase is necessary for understanding the compared to the optical emission in the radio life of a galaxy. galaxy 3C31. It is clear that AGN are exotic and extremely AGN emit not only in the optical but also in every interesting objects. Despite being one of the ‘hot other waveband. In fact, Active Galactic Nuclei topics’ in astronomy for many years, AGN still emit an enormous amount of energy over almost keep astronomers very busy trying to answer the the entire electromagnetic spectrum. The many open questions about this complex radiation observed at different wavelengths phenomenon. Because the subject is so vast, this comes from different regions inside the active chapter can only cover a very limited part. In WAVES OF THE FUTURE OF WAVES nucleus, and from different mechanisms. To make particular, we will see what astronomers think the 192 everything even more complicated, there are AGN structure of an AGN looks like, how the energy may be produced and released, and some How does an AGN work? examples of these incredible objects, in particular If stars alone are not sufficient to produce the the radio galaxies. energy released by an AGN, another mechanism that is more efficient needs to be found. Astronomers believe that the radiation produced Just a little bit of history by an AGN comes from converting matter into It is interesting to see how AGN were discovered. energy. This is made possible by the presence of In fact, this is a relatively recent discovery that a very massive black hole. Such a massive body happened when the astronomers finally were able works like a cosmic sink: all the stars and gas to look at the sky at wavelengths different from that happen to be close enough are ‘sucked in’ the optical. by the black hole’s gravity field. This is the only mechanism that can give the huge amount of Radio astronomers were the first to realize that energy we observe. The emitted radiation is something very strange was happening in some produced because gravity converts the potential galaxies. Starting in 1946 with very basic energy of the infalling matter into kinetic energy. instruments, Australian radio astronomers were Collisions between infalling particles convert the able to detect at least two strong radio sources kinetic energy into thermal energy, and photons that were perhaps associated with extragalactic carry this energy away. objects. This seemed unbelievable, as it meant an incredible amount of energy was being released. The matter falls toward the black hole in an The first radio telescopes, because of their small organized way: angular momentum causes matter size, could only produce very fuzzy images, so it from the stars and gas to circle around the black was very difficult to identify with certainty the hole in a so-called accretion disk (see Figure 2). By ‘optical counterpart’ of the strong radio signal – definition, a black hole is an object with such a high in other words, to figure out which optical object density (we are talking about 100 million solar (star or galaxy) was producing the radio signal. masses compressed in a volume about the size of our solar system!) that its gravitational field is It took some effort and creativity (even using the incredibly strong – at a certain point, the infalling ocean as a mirror!) before the radio images matter passes through the black hole’s event horizon became sharp enough to be able to solve the and, at that point, the gravity becomes so strong that dilemma. When finally the distances of these not even light can escape. For this reason, a black objects from Earth were estimated (at that time hole cannot be observed directly. Even the region thought to be up to 100 million light years), it was around it is so small that today’s telescopes cannot clear that the amount of energy that they were see its structure. emitting was far beyond any human imagination! The fact that such a distant radio source could be detected made it clear that radio astronomy could extend the boundary of the observed Universe far beyond what was accessible to optical astronomers.

Since then, a zoo of AGN has been discovered – we now know that AGN indeed come in different types and flavours, and can live in the centre of different types of galaxies. As it turned out, AGN that are also strong radio sources are only a fraction of the active nuclei, but they still form an important and fascinating group. MONSTERS LURKING IN THE CENTRE OF GALAXIES Figure 2 - An artistic illustration of the central regions of an AGN. 193 So how do we know that black holes really exist? the energy radiated across the electromagnetic The existence of a massive object in the centre of spectrum by these nuclear regions. For example, galaxies has been derived mainly through because of their high temperature, the accretion observations of the kinematics of the gas in the regions are mainly observed at the X-ray inner regions of the galaxy – at least, those wavelengths. On the other hand, at radio regions that we can observe. The gas in these wavelengths, the energy is ejected through very regions has been seen to rotate: on one side of collimated (narrow) jets of radio waves. How does the orbit the gas comes toward us, on the other this happen? side it moves away from us. From the amplitude of the rotation velocity, the amount of mass within The radio emissions from AGN are due to that region can be derived. Similar measurements synchrotron radiation, which occurs when were done using maser emission from water relativistic electrons (electrons accelerated to molecules (see Figure 3 – maser stands for velocity close to the speed of light) are present in ‘microwave amplification by stimulated emission of a magnetic field. The structure of the magnetic radiation’, a mechanism similar to the laser beam, field (see Figure 4) probably channels the but with microwaves instead of visible light). electrons into the collimated structure that we see, called jets. The super-massive black holes in the centre of powerful AGN are estimated to have a mass To the ‘first order’ (which is a scientist’s way of between 10 million and few billion times the saying that in reality it can be much more mass of our Sun. complicated than this) the direction of these collimated structures is along the rotation axis of the accretion disk. This is illustrated in Figure 5 in the case of the radio galaxy NGC 4261.

Figure 4 - An artistic concept of the formation of a radio jet.

Figure 3 - The detection of fast rotating gas in the nuclear regions (0.2 parsec in size corresponding to only about 0.6 The active nucleus also emits ultraviolet (UV) light year) of the galaxy NGC 4258 indicating the presence of a black hole of 40 million solar masses. radiation, which can profoundly influence the gas around the AGN by ionising it. This means it can provide the energy to extract electrons from the Radio Jets atoms and give rise to emission lines that are So far, we have seen that material goes into the observed mainly at optical wavelengths. The black hole, and this infalling material is converted study of these emission lines has played, since WAVES OF THE FUTURE OF WAVES in energy. What happens then to the energy the discovery of AGN, a crucial role in our 194 produced? A variety of mechanisms can explain understanding of these objects. Two examples of optical spectra observed in AGN are shown in If we can detect the broad lines, we can be sure Figure 6. The obvious difference between the two that we are looking directly at the nucleus of the is that in one case the lines are very broad. These object. When we do not observe such broad lines broad lines have been interpreted as coming from it indicates that part of the very central region is gas that is located only at most few light years perhaps obscured from view because of the away from the active nucleus, and it is therefore torus. Optical and UV radiation can be emitted, in moving quickly under the effect of the gravity of principle, in all directions, but the presence of the black hole. structures around the black hole, like the torus, can force this emission to be confined to a cone- Galactic Donuts like structure. Evidence of this comes from the At this point, it is also important to note that detection of just such a cone structure in the astronomers think that outside the accretion disk, Seyfert galaxy NGC 1068 (see Figure 7). We will a thicker, donut-shaped structure surrounds the see later how all this influences our knowledge AGN (called the torus). This is illustrated, in an and classification of AGN. artistic way, in Figure 2. A thick structure like this can obscure the very nuclear regions around the black hole. This is a crucial element because it means that, depending how this structure is oriented, one can or cannot detect the emission coming from the very inner part of the AGN (for example the broad emission lines). Astronomers think that orientation effects can explain some of the differences observed between AGN.

The exact size of the torus is not really known as its inner edge is supposed to be quite close to the black hole, while the outer edge could extend a thousand light years or more from the centre. Indeed the high resolution provided by the optical observations of the Hubble Space Telescope (which, because it orbits the earth in space, does not suffer from the effect of the atmosphere in blurring the images) has shown the existence of these structures in many galaxies (see Figure 5).

Figure 6 -Two examples of optical spectra of AGN. In both emission lines from many different elements are detected. However, in the spectrum at top some of the emission lines appear very broad while on the spectrum above all the emission lines are quite narrow. The units in the horizontal axis are Ångstroms (1 Å = 10-8 cm).

Figure 5 - Panel showing (left) a ground based image with MONSTERS LURKING IN THE CENTRE OF GALAXIES superimposed a radio image and (right) an image from the Hubble Space Telescope of the radio galaxy NGC 4261. 195 have luminosity of the order of 1039 watts, corresponding to 1012 times our Sun’s energy output – or a hundred times more powerful than our entire galaxy.

Seyfert galaxies (from the name of the astronomer that classified them) are nearby AGN, mainly hosted by spiral galaxies. They were originally identified as unusual because of the very bright point-like nucleus. Their characteristics are, therefore, similar to those of quasars although the AGN in Seyfert galaxies have a much lower power. As with quasars, Seyferts also show very strong emission lines.

It is impossible in the short space of this chapter to give a full overview of all the different AGN Figure 7 - Cone-like region of emission from gaseous clouds ionised by the intense radiation from the nucleus of types, so we will focus mainly on one kind: the the Seyfert galaxy NGC 1068. The nucleus is located near galaxies that are strong radio emitters, known the base of the cone. This region is only few hundred light simply as radio galaxies. It is important to years in size, therefore many times smaller than the region remember, though, that the fact that they are so shown in Fig.1 for the same galaxy. This image has been impressive at the radio wavelengths does not obtained by using a “narrow-band” filter that lets through only the light from the emission lines. necessarily mean that the majority of the energy that they emit ends up in the radio band. The radio photons are not very energetic, as the One can already imagine that the injection of all energy carried out by photons is inversely this radiation and particles from the AGN into the proportional to their wavelength (the longer the surrounding medium of the host galaxy can have wavelength, the lower the energy). Thus, while a major impact on this medium. Fast gaseous ‘radio-loud’ objects can look extremely impressive outflows reaching speeds of many thousands of when observed with radio telescopes, the bulk of kilometres per second have been observed in their energy is not at radio wavelengths. many AGN. Astronomers are beginning to realize the presence of an active nucleus can have a major impact in the life of the galaxy: for The spectacular radio galaxies example, it may blow away gas from the central The first thing one notes when looking at a radio regions and therefore prevent the formation of galaxy with a radio telescope is the strikingly stars, or it may even create a self-regulating different morphology of the radio emission mechanism that after a while stops the feeding of compared to the optical (stellar) morphology. the AGN. Figure 8 shows the superposition of the radio emission of Fornax A (red) and surrounding In the zoo of AGN optical field (blue-white). The radio source is Galaxies emitting strong radio signals were the dominated by two large lobes well outside the first AGN discovered – however, after that a optical galaxy, the large elliptical galaxy visible in number of other types of AGN were found. The the centre of the image. Figure 9 shows a much discovery of quasi-stellar radio source (or quasars more extreme radio source and illustrates how for short) was particularly surprising as these great the difference can be between the size of objects appear to be associated with star-like the radio emission and that of the optical galaxy. objects – but in fact they represent very distant WAVES OF THE FUTURE OF WAVES extragalactic objects. One of the first quasars for This difference clearly points to two completely 196 which the distance was estimated turned out to separate mechanisms responsible for the origin Figure 8 - The superposition of the radio emission of Figure 9 - Panel showing (left) the image of a radio galaxy Fornax A (red) and surrounding optical field (blue-white) taken at the wavelength of 20 cm with the Very Large The radio source consists of two large and complex radio Array; (middle) the radio image (red) superimposed onto an lobes. At the centre of the optical field is the elliptical optical image (blue, from the Digital Sky Survey) and (right) galaxy from where the extended radio emission originates. the optical image marked with the galaxy responsible for the huge radio emission. This last image is to illustrate the amazing difference in size between the optical galaxy and the radio emission.

for these emissions. Electrons and protons spiralling around magnetic field lines at nearly the speed of light (synchrotron emission) produce the radio waves from these galaxies.

Figure 10 shows one of the most spectacular and well-known radio galaxies, Cygnus A, as observed with radio telescopes at different frequencies and, more importantly, with increasing resolution, in order to explore the regions closer to the nucleus.

The host (optical) galaxy would be only about a tenth of the size of the full radio emission, shown in the top image in Figure 10. The structure of the radio emission is very interesting and has been crucial in giving clues on the structure of the AGN that we have described above. The typical radio structure shows a nucleus

Figure 10 - The radio source Cygnus A. This radio source is produced in a galaxy some 600 million light-years away. The images show the radio structure seen at different frequencies and different resolution. Red (blue) colours represent stronger (weaker) intensity. The two remaining images have been obtained with high resolution observations (with the Very Long Baseline Interferometry). These images allow investigating in more detail the inner

part of the jet, the region closer to the nucleus. The image MONSTERS LURKING IN THE CENTRE OF GALAXIES at the bottom can resolve details of only 0.1pc in size Figure 11 - The radio galaxy Cygnus A with the various (about 0.3 light years). components the radio emission indicated. 197 (coincident with the region of the black hole, accretion disk and torus) and two very narrow jets emerging from it. Two large lobes are seen where the jets end. The presence of the jets clearly indicates that the energy (at least the radio plasma) is emitted, not in every direction, but in collimated structures. Typically, these jets are also perpendicular to the nuclear dusty disks found by HST as seen in Figure 5.

The radio jets connect the nucleus to the lobes, so although the radio lobes are far away they are indeed fed by the active nucleus. Jets are thought to be the channels along which the accelerated electrons travel to very large distances from the nucleus. If the jet is powerful and fast enough, the strong interaction with the medium around it will produce a ‘shocked’ and bright region – which shows up as ‘hot spots’. Figure 11 illustrates all these different components in the case of Cygnus A.

In other cases, the jet will just fade away and Figure 12 - The panel shows view of M87 (Virgo A) at produce diffuse regions (lobes) that are formed different spatial scales. The bottom image (obtained using by ‘old’ electrons, not so energetic anymore the Very Long Baseline Array) shows the very central because they have been decelerated. This regions (red represents the regions of brighter emission, blue of fainter emission). The white bar indicates distances morphology can be seen in the radio galaxy M87 of about 0.03 light years or about only 2 times the distance (also known as Virgo A) illustrated in Figure 12. from the Earth to the Sun! The top left image is obtained This radio source is hosted by an elliptical galaxy using the Very Large Array and shows the radio jet and that we met in the previous Chapter. The radio lobes at much lower spatial resolution. The distortion emission here is much more diffuse and no hot observed in the radio emission on the large scale is amazing and it is likely due to the effect of the medium spots are seen at the edge of the lobes, around the radio source. The top right image shows the indicating that the jets do not impact strongly optical jet observed by the Hubble Space Telescope. with the outside medium. Note that here we can also see the presence of only one jet, which is Figure 13 - Example of compact source observed with the Very Long Baseline Interferometer. The size of this source is straight at first but then bends dramatically. only of the order of 100 pc (300 light years) and its age is These characteristics are often observed and can estimated to be only about 1000 years. be partly due to the effect of the medium around WAVES OF THE FUTURE OF WAVES

198 the radio source, or to projection effects due to Very high resolution observations became the orientation of the jets with respect to possible when radio astronomy developed the the observer. Very Long Baseline Interferometry technique. This consists of performing radio observations The radio emission can reach amazingly large with telescopes that are very far apart (thousand distances from the nucleus, many times the size of kilometres) – the signal from the radio of the stellar body of the galaxy. In some cases, sources detected by the different antennas is it can reach a few million light-years from recorded separately, and later on combined to the nucleus. produce the image. This system essentially creates a radio telescope as large as the Earth – From the characteristics of the radio emission in actually, even bigger if you use antennas in orbit these regions, astronomers can estimate how old around the Earth, as was done recently with a they are and therefore the age of the radio Japanese telescope! Consequently, the images source. Typical ages are a few tens of millions to can reach resolutions that would allow you to hundreds of millions of years – if you compare see a football on the moon (if there was one, and these with the time-scales of merging galaxies, if it could emit at radio wavelengths). The you can see that the radio source seems to live resolution that astronomers can reach using for a much shorter time. radio telescopes is still unsurpassed at other wavelengths. Not all the galaxies are so extended – baby radio sources also do exist! These are tiny radio For the study of AGN, this high resolution is sources, with a size of only few tens of light- crucial, as we want to explore really tiny regions. years but with morphology already essentially The jet observed in M87 (Figure 12, bottom identical to that of the grown-up sources; they image) shows the structure of the jet in the can be considered miniature radio galaxies. An region very close to the black hole where the jet example is shown in Figure 13: without the linear is forming. The formation of the jet appears to scale, it would be impossible to distinguish this occur within a few tenths of a light-year of the from a grown-up radio galaxy. galaxy’s core. While the jet seems to start as a wide structure, it then quickly becomes very The study of these sources allows investigations collimated: this suggests that something very like of how the radio plasma expands through the the magnetic field mechanism (see Figure 4) is medium in the initial phase of the life of a radio needed to make this happen. source. In some cases, it looks like it is quite difficult for the baby radio source to make its way The ability to observe such small features has through the dense gas that surrounds it. Many of allowed astronomers to track the motion of the them have quite distorted morphologies indicating plasma in the jet, confirming that indeed the jets that they are forced to interact with a dense are the channels of supply for the radio lobes. By medium around them. The gas that surrounds observing the region of radio jets very close to these sources is likely to be the same material the nucleus every few years, astronomers have that feeds the black hole and keeps the detected the change in the jets and the shift of nucleus active. some features due to their motion at very high speed, in some cases close to the speed of light. Figure 14 shows high-resolution observations, Narrow and powerful jets carried out monthly for 16 months, of the jet in Through the technique of interferometry the radio source called 3C120 – the nucleus is developed some years ago, radio astronomy has believed to be the bright spot on the left. First, it reached the ability to observe objects with very is clear that the jet is not really a continuous high spatial resolution (that is, being able to structure, rather it is formed by a series of bright distinguish two very close objects from far away). blobs. Every now and then the nucleus emits one MONSTERS LURKING IN THE CENTRE OF GALAXIES Today’s radio telescopes can resolve the most of these blobs that then starts moving away to inner structures of the AGN jets. reach the radio lobes. 199 A very exciting discovery (made more than 20 years ago) is that jets observed in powerful radio sources can have an apparent velocity higher than the speed of light, so-called superluminal velocities. This is clearly not possible, according to Albert Einstein’s theory of Relativity, and the explanation for this phenomenon is that it is due to a projection effect (that is, it is due to the way we observe the jet, not the jet itself). The objects that show this effect are sources where the radio jet is ejected in a direction very close to our line of sight, in other words almost straight in the direction of the observer. If such an ejection happens with velocity close to the speed of light (and some jets have been found to have a velocity more than 90% of light speed) then relativistic effects make it appear as if the velocity is greater than the speed of light.

One more effect comes from the high speed of the jet, combined with its projection or direction. As noted above, the jets can be very asymmetric in their luminosity: one jet is often much brighter than the other. This characteristic has been again interpreted to be due to projection effects. Because the jets are very narrow, the way we see them depends very much on whether they are on the plane of the sky or they are ejected by the nucleus in a direction close to the observer’s line of sight. When we observe them close to the line of sight, and when they travel at very high speeds, relativistic effects act in a way that the approaching jet looks much brighter than the receding jet. Thus, the asymmetry observed has been used as critical parameter to work out the direction (relative to the observer) in which the jet has been ejected.

Are all AGN different? For many astronomical objects, like stars, it does not matter very much from which direction they are viewed. For the AGN this is not the case. As

Figure 14 - Images of the radio source 3C120 obtained with the VLBA at 22 GHz (0.3 milliarsecond resolution), at monthly intervals. The 3C120 likes to eject components with apparent superluminal velocities of about 5 times the

WAVES OF THE FUTURE OF WAVES speed of light. The images also show rapid variations (scale of months) in total intensity, that we interpreted as 200 interaction with the external medium. Figure 15 - Effect of orientation on the way AGN look. The filled and empty circles represent the region of gas producing emission lines (respectively broad or narrow). In both cases, the astronomers are happy! we have seen, most of the energy produced by and galaxies might be simply because the an AGN is radiated through collimated jets or amazingly bright nucleus is making us blind to all cones, and therefore it is radiated only in a the rest! certain direction. Thus, one can picture AGNs a bit like a lighthouse: we can see the light only However, if we are not looking straight into the when it is emitted in (or close to) our direction. nucleus, then the presence of the torus around For the same reason, the direction in which the the accretion disk and black hole can obscure energy from an AGN is emitted becomes what is happening in the very inner regions. Thus, extremely important in determining the way AGN we do not see the broad lines and many other looks when observed from the Earth. signs, but the fact that the nucleus does not overshine the rest allows the astronomers to view Figure 15 illustrates these effects. When the this more easily than other phenomena. emission from the AGN is radiated almost in our direction, we can see directly into the nucleus. As result of these orientation effects, astronomers This means that the nucleus is so bright that it think that many different types of AGN (for overshines almost all the rest of the galaxy – in example quasars and radio galaxies) may in fact this case, what we observe is a quasar object. be the same kinds of objects only seen from a The broad emission lines that are detected in the different direction! optical spectra of quasars support the idea that we are looking directly into the nucleus. Well-fed and starving black holes Figure 16 shows what happens if we observe a The question is, are active nuclei located only in a quasar but we block (with a special instrument) small fraction of galaxies (that can therefore be the light coming from the bright nucleus. We can considered as special, lucky objects) or do they MONSTERS LURKING IN THE CENTRE OF GALAXIES now see the underlying galaxy, suggesting that represent a phase in the life of every (or almost indeed the apparent difference between quasars every) galaxy? 201 Figure 16 - Images of the nearby quasar 3C273 taken with the Hubble Space telescope. With the higher resolution of the right image, and by using a device to block the light from the central AGN, the host galaxy can be seen.

Figure 17 - Giant radio galaxy (called B1545-321, based on The main consensus now is for the second its position on the sky) showing indication of a recent hypothesis. As we have seen above, at least the restarted activity. This image has been obtained using the radio emission seems to have a relatively short Australia Telescope Compact Array. time scale compared to the life of a galaxy. But there’s more – there could even be more than one active period in a galaxies lifespan!

Evidence of this can be seen in the morphology of some giant radio galaxies. The radio galaxy in Figure 17 has two giant (and old) radio lobes about 1.5 billion light years in size. The period of radio activity that created those lobes is over and the giant lobes are now relics. The image shows, however, that the nucleus has recently become active again – bright new jets are being emitted by the black hole and new lobes are now seen advancing through the old ones. For some reason, new fuel has reached the back hole and this galaxy could happily start its nuclear activity all over again.

The idea that almost every galaxy could, at least once in its lifetime, go through an AGN phase is also supported by another fact. Astronomers now think that almost every galaxy may actually host a WAVES OF THE FUTURE OF WAVES massive black hole in its centre. The presence of this massive body can be seen from the 202 kinematics of the gas in the regions around it. Thus, from all we have seen in these two This means that the presence of a central black chapters, the life of a galaxy it is very hole does not automatically complicated indeed. Through ensure the production of the huge interaction and merging, not amount of energy detected from “Astronomers now think only can galaxies change their an AGN. There are, therefore, that almost every galaxy structure during their lifetimes starving as well as well-fed black ... but the same phenomena can holes and only the latter can may actually host a also help turn their nuclei into produce the amazing amount of massive black hole in its fascinating monsters! energy observed in AGN. centre.” If we calculate how much matter a black hole needs to eat in order to produce the energy necessary to shine as a bright AGN, it turns out to be Some interesting web sites: something like few tens of suns every year. This Alan Bridle’s page with impressive pictures of radio galaxies is assuming that the black hole can convert mass (and links to other interesting sites): in energy with a 10% efficiency. http://www.cv.nrao.edu/~abridle/image.htm ; more explanations about radio galaxies in This amount of food for a black hole can be http://www.cv.nrao.edu/~abridle/dragnparts.htm easily available in a galaxy, but the main problem Radio images of AGN obtained with Very Long Baseline is actually to make sure that the matter reaches Interferometry techniques can be seen also at the the very central nuclear regions to be able to European VLBI Network site: enter the gravitational field of the black hole. http://www/evlbi.org/gallery/images.html Interaction between galaxies, as we have seen in Nice pictures of AGN from the X-ray Chandra mission’s Photo the previous chapter, seem to be an ideal way to Album site: http://chandra.harvard.edu/photo do this job by ‘pushing’ some of the material NASA site: http://legacy.gsfc.nasa/ with link to astronomy for down to the central regions. Due to this, the students presence of interaction and the nuclear activity Nice images of radio, X-ray and optical jets can be found at: are thought to be closely related. http://hea-www.harvard.edu/XJET/index.cgi MONSTERS LURKING IN THE CENTRE OF GALAXIES

203 Rock, Paper, Scissors By Dr Karl Kruszelnicki

IF YOU EVER study anatomy, they’ll teach you that the hand is that incredibly important prehensile (or grasping) organ at the far end of the multi-jointed lever called the “upper limb” – the rest of us call it an “arm”. The human mind lets us think about the world, but the hand makes our dreams come true. It’s very necessary for medical doctors to know which nerves control which muscles of the hand – and now a simple game has come to help their memory.

What makes the hand very and-match and get all crossed muscles of the hand. He gave a special is the thumb, which is over to give just three nerves series of simple clinical probably as important as all the that control the hand – the questions about the three other fingers put together. If median nerve, the radial nerve nerves that run the hand you activate one set of and the ulnar nerve. (median, radial and ulnar muscles, you form your hand nerves) to 20 junior doctors. into the “static hook” – where As an example of this mixing- They could get any mark you grab a briefcase. Another and-matching thingie, the ulnar between 0 and 10. The average set of muscles gives you the nerve is made up of C8 and T1 mark was 3, with no doctor “pinch grip”, where the index (and sometimes, from C7) getting higher than 5. finger and thumb combine to – but no nerves from C5 and let you make precision C6. This crossing-over gives So Davidson wrote a paper in movements, such as passing a some protection in case of the surgical journal called Injury, thread through the eye of a injury – a finger or two might entitled “Rock, Paper, Scissors”. needle. You make the “power get weaker, but probably won’t grip” when you grab the handle go totally floppy. Rock, Paper, Scissors is a very of a hammer. And all the ancient game. Back before different muscles are controlled The arrangement of these scissors were invented there by nerves. nerves and the muscles they was a similar game called control is quite complicated, “Earwig, man, elephant”. Today, It all begins in the spine in the and memorising them has Rock, Paper, Scissors is often neck, where five nerves run out worried medical students for the used to decide matters between from the four cervical vertebrae last century. It’s important for two people in much the same (C5, C6, C7, and C8) and the the junior doctors to know way they might toss a coin. On first thoracic vertebra (T1). If which nerves run which the count of three, each player each of the five nerves ran one muscles of the hand, because has to select a hand position. of the five fingers of the hand, it they are often the first to see would be easy to remember. patients with injuries to the Now rock (which is a clenched But if you damaged one of arm. Dr. A. W. Davidson, from fist) will break scissors. But these nerves, you would lose all the Department of Trauma and scissors (which has the index function to one of your fingers. Orthopaedics at the Royal and second fingers open and WAVES OF THE FUTURE OF WAVES So in an area near your London Hospital, reckons that outstretched, and the little and shoulder called the Brachial junior medical doctors don’t ring fingers tucked in) will cut 204 Plexus, the five nerves all mix- really know the nerves and paper. And paper (the open the fingers, which gives you the rock position. The radial nerve will extend or stretch out fingers from the closed position to the open position – so that gives you the paper position. And the ulnar nerve does two things. Firstly it makes the little finger and the finger next to it, the ring finger, tuck in. And secondly, it spreads the index and middle finger. So overall, your hand looks like a pair of scissors with the ring and little fingers clawed up and the index and middle fingers opening and closing. hand held flat with the fingers random moves. You should together and outstretched) will either use psychology to So with a simple game, more cover the rock. anticipate your opponent’s junior doctors will be able to moves, or use a certain make a decent fist of So rock will break scissors, sequence of moves to influence diagnosing hand injuries. scissors will cut paper, and your opponent’s responses. paper will cover the rock – and around and around it goes. So And getting back to the hand you’ll win, lose or make a draw. itself, you can work out most injuries to the nerves that Now it turns out that there is a control the muscles to the hand World RPS or Rock, Paper, by getting the patient to make Scissors Society. They give you either a rock, a paper or a on their homepage various scissors. ROCK, PAPER, SCISSORS basic rules to the game, as well FROM Dr Karl’s book Bum Breath, Botox as advanced gambits. They Now it turns out that the and Bubbles (Harper Collins Publishers) 205 reckon that you should not play median nerve clenches all of Illustration courtesy Adam Yazxhi What got you interested in science in the first place? I’m curious about nearly everything, and science covers a very large part of ‘everything’. I think my fascination grew as I discovered DR JOE HOPE was that the huge number of experiences awarded his PhD from in the world were based on such a the ANU in 1997, and small number of fundamental took up a brief research principles. position at the University of Queensland before What’s the best thing about being beginning a lecturing a researcher in your field? position at the Auckland It’s a very fast-moving field at the University. He has convergence of many traditional worked in many areas of areas of physics. It’s a chance to quantum and atom mix the fundamental with the optics, and is currently applied. researching methods of detecting atoms non- Who inspires you – either in science destructively, controlling the quantum or in other areas of your life? state of atomic samples, and I’m inspired by people who keep producing a high quality atom laser. growing and learning throughout He has won awards for their lives – in all areas of their life. communicating science, including Such people have infectious the 2003 Dialogica Award from the enthusiasm, and they always make Academy of Science. In 2003, he me want to engage. became a founding member of the Australian Research Council Centre If you could go back and of Excellence for Quantum-Atom specialise in a different field, Optics, in which he leads the atom what would it be and why? laser theory group located at Probably writing. My artistic the ANU. energies are packed into the corners of my life, and I’d relish the chance to see what I could manage if I gave myself the time and space to concentrate on them.

What’s the ‘next big thing’ in science, in your opinion? In my own area, I’m hoping the recent development of systems in which we can test quantum field theories will lead to applied quantum technology of a dramatically different kind than has previously been possible – from quantum computers to usable superconductors. On the more applied side, I think there will be a massive expansion in the areas 206 of complex systems in general and biotechnology in particular. Quantum Mechanics: The Wild Heart of the Universe

Joseph Hope

Introduction QUANTUM MECHANICS IS one of the two fundamental theories underlying our current understanding of the universe, the other being Einstein’s theory of relativity. In terms of accurately predicting nature, it is by far the most successful theory of all time, with some experiments agreeing with quantum mechanical predictions to a dozen significant digits. Whenever quantum mechanics has made a preposterous claim about the way things behave that goes counter to intuition, physicists have always found that the prediction is correct and that it is our intuition that has been at fault.

Figure 1: The Poisson Spot - (a) simulated image formed from calculations using Fresnel’s theory (thanks to Dauger Research’s Fresnel Explorer simulator), (b) experimental image using red laser light.

207 ESPITE SUCH OUTLANDISH successes, no different colours led him to postulate the one really believes the theory. In the existence of different kinds of light particle. Dstandard description of the quantum theory, there is a little flaw that is easy (and quite Between 1805 and 1815 Laplace and Biot, from profitable) to ignore, but that has nevertheless the French Societe d’Arcueil in Paris, created an troubled physicists for the last eighty years. extensive theory of ‘neo-Newtonian’ optics based Worse than that, the nature of this flaw is that it on Newton’s ideas. This theory explained all the makes quantum mechanics incompatible with observed optical phenomena at the time. As it another cherished physical theory: relativity. At was nearing completion in 1817, Laplace and least one of them must be wrong, but no Biot arranged for the French Academy of Science experiment to date has managed to demonstrate to offer a prize for the best work on the theme of a failure of either quantum mechanics or ‘diffraction’, the apparent bending of light rays at relativity. With that in mind, there are very good the boundaries between different media. In 1818 reasons for the current form of quantum Fresnel submitted his thesis for the prize, mechanics, and we can definitely show that it is suggesting that light was a wave with an superior to other (some would say ‘more sane’) amplitude and phase at every point. This wavelike theories that have been proposed. picture of light explained diffraction and refraction with ease, and without the complexity of the neo- Newtonian theories. Wave/particle duality One of the best-known ideas in quantum Fresnel’s thesis produced a hot debate when he mechanics is that objects can behave like waves defended it in front of the judging panel. Critics and particles at the same time. Where did this noted that the wave theory of light had a harder idea come from, and how can it possibly be true? time explaining the tendency of light to travel in To turn that question around, what does it mean straight lines, and form sharp shadows. Fresnel to behave like a particle or a wave? The mental countered by pointing out that waves with a very image of a particle is something small wavelength would have a that has a definite position and correspondingly small blurring velocity; it travels in straight lines One of the best-known on the edge of shadows, and unless it interacts with a force; it ideas in quantum therefore they would seem to carries energy, mass, and travel in straight lines. In the sometimes charge. A wave can mechanics is that bulk of the shadow, the phase of also carry energy1, but rather objects can behave like the wave from different than having a definite position, it unblocked areas would add up is spread out over the medium waves and particles at randomly, and on average that is waving. The fact that it the same time. cancel out, making a dark area. does not have a single position means that it also tends not to Poisson, a mathematician have a single velocity. For working with Laplace, dealt what example, different parts of a water wave at the he thought was the killer blow: if Fresnel’s theory beach often travel in different directions at were correct, then his argument worked different speeds. everywhere except in the exact centre of the shadow formed by a circular disk or a sphere. At The struggle to use these ideas to classify a the centre of such a shadow, the waves would particular phenomenon as particle-like or wave- have to add constructively: there would be a like is epitomised in the history of our under- bright spot right in the middle of the shadow. This standing of the nature of light. In Newton’s early absurd consequence of Fresnel’s theory caused investigations into light, he saw how it travelled in him to leave in apparent defeat. straight lines, and decided that it must be made WAVES OF THE FUTURE OF WAVES of particles. The famous experiment in which he Soon afterwards, one of the judges, Francois 208 diffracted sunlight with a prism to show the Arago, actually performed the experiment – and there, in the centre of the shadow, he observed a coming from the two slits add together to give no bright spot. Fresnel went on to win the prize, and light, is the feature that most convinces us that the spot was ironically named Poisson’s spot, after light is indeed a wave - this ‘destructive the man who had ‘predicted’ it. The moral from interference’ is certainly a property of waves. this story is that any simple tests should always be performed rather than relying on intuition. After So what about things other than light? Is a pair of these events, light came to be known as a wave, sunglasses a particle or a wave? It is easiest to because while some waves can move in straight answer this by ripping up the sunglasses into lines like particles, particles have a very difficult pieces and doing experiments on the bits. Let us time mimicking the constructive and destructive assume that we are all comfortable with the idea interference of waves. that the resulting pieces are electrons, protons and neutrons. The very language implies that The two-slit experiment these things are particles, but we do not have to The phenomenon of interference is simpler when take this on faith – we can test it. It turns out to there are only two sources of waves. This can be be quite easy to perform a two-slit experiment achieved by making the waves pass through two with electrons, as the inside of every television small gaps in a blocking material that are very contains an electron gun. If very thin slits are small compared to the wavelength. This means made2, then we can set up a phosphorus screen that the waves on the other side will tend to that glows to show where the electrons arrive, radiate outwards as though from two separate and perform the same experiment that we did points. If the initial waves have flat wave-fronts, with light. then an equivalent experiment in two dimensions is to pass the waves through two parallel slits. Taken separately, the electron beam coming through each slit will produce a diffuse glow on the This experiment is easy to perform with a laser screen as they spread out. If the slits are almost and shows a pattern of bright and dark lines, next to each other, the resulting spots can be in corresponding to places where the light from the almost identical positions. However, when both slits two slits is interfering constructively and are uncovered at once, the electrons form a familiar destructively. Covering either of the slits causes pattern of bright and dark lines. This is proof that the bright and dark lines to be replaced by a the electrons must be a wave. But that is not the diffuse blob of light. The dark lines, where light whole story. When the intensity of the electron gun

Figure 2: The two-slit experiment for electrons, when only one electron is in the apparatus at the same time. The left column shows the pattern produced when the right slit is covered, and the second shows the pattern when the left slit is covered. Notice that these two patterns are broad enough that they overlap strongly. The third column shows the pattern produced when the electrons can go through both slits. The first image in each column shows the arrival of the first electron as a small spot on the screen. The next four images show the screen after 2, 10, 500 and 3000 electrons respectively. While each electron arrives randomly as a single spot, when both slits are uncovered, the overall pattern of their arrival shows the interference pattern of a wave going through both of the slits.

Left slit only Right slit only Both slits QUANTUM MECHANICS: THE WILD HEART OF THE UNIVERSE

209 is turned down, the smooth interference pattern the mystery go away by explaining how it resolves into single, tiny dots appearing on the works . . . In telling you how it works we will screen. In other words, we see what appear to be have told you about the basic peculiarities of particles arriving individually. We cannot just change all quantum mechanics.’ our minds and decide that electrons are particles The Feynman Lectures, Vol III without explaining the interference pattern. If we wait a long time, then the arrivals of these little We can’t even explain the interference pattern in blobs eventually average out to that interference terms of some complicated interactions between pattern that was visible at high intensity. the electrons. We can turn the electron gun down so low that only one electron is present in the Analysis of this experiment quickly leads to the system at a time, and the dots and the conclusion that neither the wave nor the particle interference pattern occur regardless. model of electrons is suitable. The electrons cannot be travelling as the little blobs that we see Can electrons really be going through both slits at on the screen, because to make the interference the same time? We can modify our original pattern they must be travelling through both slits experiment to try to detect the electrons as they at the same time. The pattern we would see if go through the slits. We can do this carefully so they were passing through each slit randomly that we don’t disturb them too much, and then would be two overlapping blobs, not a series of we’ll know which slit they go through, and try to bright and dark lines. It sounds ridiculous for a understand how this interference pattern can form. particle to travel through both slits at the same The amazing result of this experiment is that the time, but it is a very natural thing for a wave to interference pattern no longer forms. If the do, and the wave description describes the shape detectors are removed, the interference pattern of the interference pattern exactly. On the other reappears. Not only are the electrons trying to hand, a spread-out wave passing through two confuse us, they can tell when we are looking! slits to make an interference pattern is also spread out when it reaches the screen, and The same results can be found, with carefully would not cause a single, point-like dot. constructed apparatus, for protons, neutrons, entire atoms and even complex molecules. The famous American physicist Richard Feynman Turning down the intensity of our light source will had this to say about the two-slit experiment: also show the appearance of little, localised ‘We choose to examine a phenomenon which packets of light, called photons. It seems that is impossible, absolutely impossible, to everything has this duality between wave and explain in any classical way, and which has in particle behaviour. it the heart of quantum mechanics. In reality, it contains the only mystery. We cannot make As always, apparent paradoxes like this exist because of our preconceptions. As with all theories that challenge our preconceptions, it took a lot of evidence to change them. It was almost half a century of debate before anyone seriously argued that quantum mechanics was a description of reality rather than just a good calculational tool.

The axioms of quantum mechanics Leaving aside the details, we can write down the whole theory of quantum mechanics very simply: WAVES OF THE FUTURE OF WAVES Figure 3: Richard Feynman, Nobel Prize-winning physicist 210 and infamous bongo player. environment. Mathematically it is not very complicated, so once the concepts of quantum mechanics have been accepted it is not terribly hard to learn the calculational details.

The third and fourth points are required to understand the appearance of little ‘blobs’ in the two-slit experiment. After the electron wavefunction has travelled through the slits and has formed interference fringes, the screen performs a measurement on the electron position. The wavefunction then suddenly jumps to a single position randomly, with a probability that depends on the amplitude of the wavefunction. This explains the appearance of the small spots on the screen (the wavefunction collapsed) and also explains why they form an interference pattern over time, as more and more electrons are measured (the bright fringes would appear where the wavefunction’s amplitide is Figture 4: Erwin Schrödinger, Nobel Prize-winning physicist large, so there’s a greater chance for the dot to and creator of the Schrödinger Equation. appear in those regions).

1. Every object is described by a wavefunction. One common misconception about this random 2. These wavefunctions evolve according to ‘wavefunction collapse’ is that it represents some Schrödinger’s equation. kind of ignorance about where the electron 3. When a measurement is made on the ‘really’ was just before the measurement. This is wavefunction, it collapses to give a definite a natural idea, as that is what probabilities result. usually mean in physics, or in any other field 4. Immediately after the measurement, the new – probabilities normally indicate that you are wavefunction is consistent with the result of missing some information, not that reality itself is the measurement. uncertain. For example, if you know the secret formula then the computer-generated random Let’s examine three of these points in a little numbers in a gambling machine are actually more detail. The first point says that every object, predictable. whether it is an electron, a beam of light or an iPod, is described by a wavefunction. This means But the case of the electron position is different, that everything acts like a wave, and it is this as it really does have to go through both slits at axiom that explains the interference pattern in the the same time in order for the wave to make an two-slit experiment. These waves carry energy interference pattern. The randomness in quantum like all waves, but they can also have mass, mechanics is fundamental randomness. This idea electrical charge and other physical properties, explains all the phenomena that we can see, but depending on the kind of object they are many people have felt that there must be a describing. The thing that they all have in deeper theory that describes these effects common is that they have amplitude and a deterministically: phase, and can form constructive and destructive The theory yields a lot, but it hardly brings interference. us any closer to the secret of the Old One. In any case I am convinced that He does not The main detail left out of these axioms is the throw dice. QUANTUM MECHANICS: THE WILD HEART OF THE UNIVERSE form of Schrödinger’s equation, which also Albert Einstein to Max Born, depends on the particular object, as well as its 4 December 1926 211 Measurement has a very special effect on electron field’. This is imagined to exist throughout wavefunctions in quantum mechanics, and this is all space, much like the electromagnetic field. the explanation for the surprising ability of There is also a proton field and a neutron field. electrons to know when they are being observed. These fields sound very abstract, but the reason When the measurement devices look at the that we believe they exist is fundamentally the electrons before they reach the slits, the same as the reason we think walls, air and wavefunction collapses to go through either one electricity exist – the world behaves as if they do. slit or the other – but only one slit – with equal The difference is that we need only postulate an chance for either to occur. This means that the electron field after a careful examination of the interference pattern disappears. When the world, whereas walls are a little easier to believe. measurement is not made at the slits, the wave Electricity would be hard to explain to someone gets a chance to make an interference pattern born a thousand years ago, but an attempt would before it is measured by the final screen. be a little like this article: a description of a series of experiments and their conclusions. Taking a definition of ‘reality’ as ‘things that are there even when you’re not looking’, and noting The electron field has, at each point in space, an that quantum mechanics says that when you’re amplitude and a phase. All of these amplitudes not looking, there are no particles, some people and phases together define a function that we have seriously suggested that the act of call the wavefunction of the electron. measurement actually creates reality: Mathematically, we define this function to be [The] moon is demonstrably not there when complex-valued, as complex numbers also have nobody looks. an amplitude and a phase. Prof. David Mermin Some consequences of quantum What is waving? mechanics One of the first questions that arise when Quantum mechanics can be used to explain deciding that all matter is made of waves is this: many things about our everyday world. It also what is waving? The answer is obvious when you describes strange behaviours that we do not see a water wave at the beach, or a standing normally see in our everyday world, and one of wave in a guitar string, but what is waving when the larger mysteries of the theory is how the the wave is an electron? The answer is ‘an world can look so classical in everyday life when these strange things are happening all the time at a microscopic level. The existence of atoms Quantum mechanics explains why atoms, and therefore everything made of them, can exist. If electrons really were negatively charged particles circling a positively charged nucleus, then classically they would emit electromagnetic waves and spiral inwards towards destruction. As electrons are actually waves, it is fairly easy to show that for waves with a well-defined energy, only certain shapes fit around the nuclei. This means that there are certain stable configurations for the electrons in atoms, and that they have Figure 5: Visualisation of a field, making up a wavefunction. At each point, the field has an amplitude and phase, certain very specific energies, including a lowest allowable energy. The energy of electrons

WAVES OF THE FUTURE OF WAVES represented here by the size of a circle and the angle of the interior line respectively. A continuous series of these attached to atoms can only be changed in 212 points make up the wavefunction. discrete amounts, called quanta. This is where the theory gets its name. Ironically, a ‘quantum Heisenberg’s uncertainty principle jump’, which in ordinary usage means a large One of the best-known results in quantum and sudden change, is actually the smallest mechanics is Heisenberg’s Uncertainty Principle. possible change that is allowed. It has many forms, but the most common is this: The uncertainty in the position x of an object Quantum tunnelling multiplied by the uncertainty in the momentum p Quantum tunnelling is a phenomenon that was of that object must be greater than a constant, or first used to describe radioactivity, and is ∆x.∆p > h extremely hard to explain without quantum 2π mechanics. Some nuclei contain alpha particles In this equation, h is Planck’s constant. This that lack the energy to overcome the short-range equation means that it is impossible for an object attractive force attracting them to the centre. This to have a well-defined position at the same time can be described as an energy barrier, like a hill as having a well-defined momentum. This limit is that is too high for them to climb. If they ever often ascribed to the uncertainty caused by the found themselves on the other side, however, measurement process, but it is more correct to they would be able to escape the nucleus. If they think of it as an intrinsic property of the wave- were particles, they could only do this if they function. It is true that measuring the position of could somehow get enough energy to get over a wavefunction requires interaction with the the hill. Quantum mechanically, however, they are measurement device and that measuring the waves, and while most of the wave is located position with increasing accuracy means that the inside the nucleus, a small part of the wave is on object must be given an increasing spread in the outside. Measuring the position of the alpha momentum, but this is closer to an explanation of particle will cause the wavefunction to collapse, the dramatic effects of making a measurement in and there will always be a probability that this will quantum mechanics than a statement of the occur well outside the nucleus. The alpha particle uncertainty principle. The uncertainty principle says can therefore ‘tunnel’ through the energy barrier that not only is it impossible to measure a precise and appear on the outside without ever having position and momentum simultaneously, but that an enough energy to be in between. object cannot have them simultaneously.

This seems outrageous enough for a tiny particle, but why don’t we see it all the time? If every component of our bodies is a wave, then we are all made of waves. When we walk towards a door, some part of us is already on the other side. Plenty of people walk into doors every day, so why don’t they occasionally appear on the other side without opening it? The answer to this question is simply a matter of the numbers. Each Figure 6: wave comprising our body does indeed extend Werner Heisenberg, throughout the whole universe, but it is exponent- Nobel Prize- ially less likely to be found somewhere unusual, winning with a natural length scale of about a nanometre. physicist and It is therefore extraordinarily unlikely for one of source of the the many electrons in our bodies to be a whole Uncertainty Principle. millimetre out of place, never mind all of them.

While the chance of walking through a wall is Like all surprises in quantum mechanics, this is unsurprisingly low, one can imagine conducting much easier to understand if you think of things experiments that deliberately amplify the as wavefunctions rather than particles. We have QUANTUM MECHANICS: THE WILD HEART OF THE UNIVERSE quantum world. already discussed the fact that waves don’t normally have a single position but are spread 213 Figure 7: A simple version of Heisenberg’s uncertainty principle. A wave with a single momentum is different to a wave with well-defined position. The compromise between the uncertainty in each variable is quantified in Heisenberg’s equation.

out. The same is true for momentum. To look at laser cooling and old-fashioned evaporation. When the momentum of a wave, we look at the an atom absorbs a photon from a laser beam, it derivative of the wavefunction. A state that does also absorbs the momentum of the photon, and it not change if we apply the derivative is the gets a kick in the direction of the laser. This exponential function, so a wavefunction that has process would normally heat a cloud of atoms, a well-defined momentum p is given by making them go faster.

ψ(x) = e (2π p.x)/h If we make the frequency of the laser just a little less than the natural absorption frequency of the But the amplitude of this wavefunction is constant atoms, then the atoms will be more likely to everywhere, and the phase rotates linearly as we absorb a photon when they are moving towards move in the direction of travel. This is clearly very the laser beam than they are when moving away different to a wave with an amplitude that is from it. So the laser is more likely to slow the nearly zero everywhere except for a well-defined position – having a definite momentum isn’t σσ++ compatible with being in a definite place.

The reason we do not see this uncertainty behaviour all the time is that Planck’s constant is I a very small number indeed. A car travelling at a σ+ σ− speed known to the nearest nanometre per year I can have a wavefunction localised to less than a billionth of the size of an atom – which isn’t very uncertain at all. Heisenberg’s uncertainty principle σ−−σ is vastly more important for very small things such as electrons or atoms. An electron in an Figure 8: The laser cooling and magnetic trapping is atom has a size of about an Angstrom (10-10 m), performed in high vacuum in the glass cell shown. Red- so it must have an uncertainty in speed detuned lasers are incident from all sides, and currents in corresponding to about 1000 km/s! the coils allow the atoms to be trapped magnetically for the final evaporative stage. Image courtesy of Nick Robins.

Experiments on ultracold atoms and BEC atoms down than speed them up. Shining red- In the laboratories at the ANU in Canberra, as well detuned laser light from all directions can WAVES OF THE FUTURE OF WAVES as many others around the world, we cool atoms therefore cool a cloud of gaseous atoms from 214 down using a combination of techniques including nearly a thousand degrees Kelvin to a milliKelvin in just a few milliseconds. With the lasers then cannot have two particles in the same quantum turned off and the atoms captured in a magnetic state. This means that they start to repel each trap, they can be cooled even further by allowing other. Other atoms are bosons (like photons), and the highest energy ones to escape and waiting they can happily occupy the same state. In fact, if for the gas to reach thermal equilibrium. we cool a cloud of bosons so that the individual wavefunctions overlap, then they tend to all go This evaporation can cool the atoms to less than into the same quantum state. This is known as a microKelvin, or an average atomic speed of Bose-Einstein condensation (BEC), and it allows approximately a tenth of a millimetre per second. us to enter a world where we can manipulate This means that the wavefunction of these atoms large wavefunctions in the laboratory. This new is of the order of tens of microns across1, almost ability should allow us to test some of the strange as wide as a human hair, and virtually detectable claims in quantum mechanics; build ultra- to the naked eye. It stands to reason that a lot of sensitive measurement devices based on atom strange things occur when atomic wavefunctions interference; and model previously inaccessible are big enough to be seen by the human eye, quantum systems like superconductors, black and examining some of these effects is a major holes and neutron stars in systems where we can motivation for producing them. If you look at control most of the physical parameters. them, the wavefunctions collapse; they form interference fringes when split up and Current work on these ultracold atoms is recombined; they can tunnel through barriers focussed on exploring the possibilities of making without having enough energy to get over them; a useful atom laser. The formation of a Bose- and they can reflect off attractive potentials. Einstein condensate in an atom trap is directly analogous to the stimulated emission that These ultracold atoms do something else quite causes a laser, as both processes use the same extraordinary. When the atom cloud gets so cold property of bosons (photons in the laser case, and dense that the individual wavefunctions of and bosonic atoms in the other). Letting a the atoms overlap significantly, then the fact that condensate out of a trap therefore produces a all fundamental particles are indistinguishable3 coherent beam of atoms with similar properties causes them to behave very strangely. Some to the coherent beam of light emitted by a laser. atoms, like electrons, are Fermions: that is, they Atom lasers are still in their infancy, however, and there is much to learn about controlling these new atomic sources. Figure 9: The momentum spread of a cloud of atoms as we make it colder. The leftmost picture shows a thermal distribution. In the next picture, as we make the atoms only slightly cooler, we see a new, sharp distribution sticking up Teleportation in the middle. This is the formation of the Bose-Einstein Ever hear the joke about the office worker who condensate. The last picture, slightly cooler again, shows ran out of fax paper and asked a friend to fax the pure BEC. This is a single wavefunction, large enough them some more? It is a surprisingly common to see, containing a hundred thousand atoms. Image courtesy of Nick Robins. misconception that fax machines are actually teleporting paper. What they really do, of course, is make as accurate a copy of a piece of paper as they can by measuring the paper and then sending the information via the phone line. In principle, therefore, we could build a machine that did the same for a human, building her up from vats of chemicals. This is a little more like copying the human rather than teleporting her, but so long as we destroy the original, no one will be the wiser. QUANTUM MECHANICS: THE WILD HEART OF THE UNIVERSE

215 Figure 10: Schematic of the process of teleportation.

People who have recently read half an article on give a very noisy signal, but the noise on one side quantum mechanics may notice a little snag, would be the same as the noise on the other. however. A combination of the process of measurement and Heisenberg’s uncertainty Taking a laser beam and mixing it with one of the principle means that it will never be possible to noisy beams allowed measurements to be made, measure the position and momentum of each and these could then be used to recreate the first wave in the human’s body. After a position noisy beam’s state out of the second, correlated measurement, the momentum spread will be beam – in effect, the first’s state has been large, and unrelated to the original momentum teleported to the second. This seems to spread of the wave. After a momentum contradict our earlier argument, but the measurement, the wave will be spread out, measurements made on the original beam gave irrespective of where it may have initially been no useful information about it, because it was localised. In general terms, after any number of drowned out by the noise on the first correlated measurements, it is only possible to ever obtain beam. While this will probably never be used as a half of the information about a given initial method of transportation, the technology required wavefunction. This would make teleporting even to do quantum teleportation is extremely useful a single, microscopic object impossible by simply for manipulating quantum states. measuring it, unless we were happy with losing half of the information about it. We are more than happy to lose information about the precise Quantum computers pattern of paper and ink molecules in a letter to The most ambitious reason to want to manipulate be faxed, which is why fax machines have never quantum states would be to build a quantum had to confront this problem. However, computer. Any computation can be boiled down teleporting a quantum state or a brain (which to taking a number as an input and calculating may rely on particular quantum states) some arbitrarily complicated function of that appears doomed. number. Physically, this number is represented by some quantity such as the alignment of various Using another deep property of quantum systems, tiny magnetic domains in a magnetic material – called entanglement, it is possible to teleport a the ones and zeros on a hard-drive. This assumes quantum state without learning anything about it. that these quantities exist, of course, which is not This was first achieved in CalTech, but has since necessarily true when you realise that the been repeated in two other laboratories, including magnetic material is made of waves, and can be one at ANU. Researchers demonstrated these both ‘one’ and ‘zero’ at the same time. This quantum effects using light instead of atoms. superposition of states must be carefully avoided They generated pairs of photons that were quite in normal computers or else they get confused. WAVES OF THE FUTURE OF WAVES random, but strongly correlated. This meant that As the components of computers get smaller, and 216 measuring any property of the two beams would their wave-like behaviour become increasingly important, it is hard to keep the machine The future of quantum mechanics behaving classically. A quantum computer tries to Quantum mechanics already underpins most of do the exact opposite and keep the quantum our technology, including electronics and other nature of a machine intact as the machine is engineering standards such as lasers. These tools made larger. require a practical understanding of quantum mechanics to build. Usually, though, we tend to A quantum computer is a computer that performs use only the more mundane properties of these all of its calculations so carefully that if a devices. The actual control and manipulation of superposition of different inputs is put into the wavefunctions themselves is only just becoming machine, then the result is a superposition of all feasible. Although the future is notoriously hard to the possible corresponding results. It is easy to predict, we might find that our current century show that when the input number has a few becomes known as the time for the birth of thousand digits, a classical computer the size quantum engineering. of the known universe could not calculate all of those possible combinations in a reasonable time; Before quantum engineering can begin in a quantum computer, though, could do it earnest, it is important to put the finishing almost instantly. touches on quantum science. The conflict with relativity and the strange behaviour of Of course, having a superposition of all of the measurement are particularly worrying because possible outputs does not obviously help, as some of the ways people have attempted to measuring the output will collapse the resolve these problems have suggested that there wavefunction to one of the individual results are fundamental physical processes that destroy anyway, wasting the hard work that went into large superpositions. It is important to know performing all of the calculations at once. whether this is true. If not, and if we are careful However, there are some algorithms that can enough, we just might manage to out-compute obtain important results from testing such the universe and walk through walls after all. superpositions, the most famous of which involves factorising large numbers. This is Notes extremely interesting to anyone with a secret, or 1 Technically, it is possible to invent waves that carry mass an interest in other people’s secrets, as most and charge as well, but we will ignore this issue for the major codes base their security on the difficulty of moment for the sake of simplicity. that factorisation problem. It may be possible to 2 ‘Very thin’ in this case means much smaller than a human use a quantum computer to solve many other hair, but much larger than the wires on a computer chip. problems that, like code-breaking, were once 3 An extra axiom of many-body quantum mechanics is that considered permanently out of reach. all particles are indistinguishable. For example, if two people swap all of their electrons they are unchanged, as The real question with quantum computers is electrons are all the same. whether they can actually be made. In practice, any interaction with the environment messes up large superpositions, and that makes it very hard to build a quantum computer of any significant size. A microscopic quantum computer has already been used to factorise 15 into prime numbers, but the real problem seems to be scaling up the system without letting it become buried in the noisy world of macroscopic objects. QUANTUM MECHANICS: THE WILD HEART OF THE UNIVERSE

217 Tibet Cooled The World By Dr Karl Kruszelnicki

TIBET IS A spiritual place. It sits on the roof of the world – the 5 km high Tibetan plateau. Some researchers now believe that this plateau cooled the whole planet, and maybe helped the evolution of the human brain.

Now the climate of the world It could be just a coincidence, of the minerals that reach has been fairly predictable over but Big Brains do need a lot of the ocean. most of the last few hundred cooling. After all, we humans million years. Until recently, it really need our big brains. We The very heavy rains combined was warm and wet, like the can’t see very clearly, we can’t with enormously steep slopes tropics. Back then, the level of run very fast, our skin won’t cause huge erosion. The carbon carbon dioxide was twice the even stand up to a rose bush dioxide that is dissolved in the level that it is today. The and our nails are pathetic as rain drops forms a weak acid – dinosaurs, who lived from 200 claws. Compared to the other carbonic acid. This carbonic million to 65 million years ago, animals on the planet, our big acid combines with granite and enjoyed a temperature about brain is our only worthwhile limestone which come from the 8-11oC warmer, and swam in asset. But while our brain massive erosion. The seas about half a metre higher weighs only 2% of our body combination of carbon dioxide than we do today. weight, it takes 20% of our and granite/limestone makes blood supply, and so 20% of minerals which wash downhill But all this changed 50 million our waste heat gets dumped towards the ocean. These years ago when India collided from our head. minerals are very rich in with Asia at the frightening carbon. So Tibet takes carbon speed of 20 centimetres per Now a new theory claims that dioxide out of the atmosphere, year (roughly 4 times faster the Tibetan Plateau is and shoves it not into trees, but than your fingernails grow). As a responsible for cooling the into minerals. result of this slow but gigantic world by taking carbon dioxide collision, the Himalayas and out of the air. The Tibetan So, according to this theory, the Tibet relentlessly and gradually Plateau is a huge area, roughly Tibetan plateau is really a huge rose above sea-level as India half the size of Australia, and pump that takes carbon dioxide ploughed northwards another mostly higher than 5 km above out of the atmosphere, and 2,000 km. India slowed its sea level. Clouds run into this deposits it on the ocean floor northward speed to a more plateau, and dump their water where it stays locked away for sedate 5 cm per year. as rain. In fact, the Tibetan millions of year. plateau causes the annual Asian During this enormous collision, monsoons. As a result, eight The Tibet theory was created by the Antarctic began to ice up huge rivers, which include the oceanographer Maureen Raymo and the world cooled down. The Ganges, Mekong, Indus and the from MIT and her colleague Bill world’s temperature kept on Yangtze, drain from the Tibetan Ruddiman, a paleoclimatologist dropping. About 2-3 million Plateau and its approaches. of the University of Virginia. years ago, our human brain These rivers drain a total area They claim that chemical WAVES OF THE FUTURE OF WAVES began to double in size from of less than 5% of our Earth’s reactions caused by the Tibetan 218 600 ml to about 1,200 ml. land area, but they dump 25% plateau have removed so much carbon dioxide from the and non-Ice Ages. And at the atmosphere, that the end of that drop, our brains temperature has dropped – not began to evolve larger. So the Greenhouse Effect but the maybe Tibet not only chilled out Tibetan Ice Block Effect. the world, it also gave us swollen heads. Now the theory is in its early days, and it’s not rock solid, but FROM Dr Karl’s book Bum Breath, Botox we do know that after about 50 and Bubbles (Harper Collins Publishers) million years of a steady downward drop in both temperature and carbon dioxide levels, the Earth’s climate seems to have stabilised into an TIBET COOLED THE WORLD oscillating series of Ice Ages 219 Huw, what got you interested in science in the first place? I’ve always been very interested in anything that starts with “s”. “sc” is even better. I was quite good at all PROFESSOR HUW my school subjects, starting with PRICE was born in home economics (making scones), Oxford, and came to and by adolescence I was into sports Australia on an Italian science (Scalectrix, sculling and liner when he was 13. scuba diving). So when I looked He attended the ISS in around for career opportunities, 1969, and went on to science seemed like the go. Physics study Mathematics, is a particularly fruitful areas for Physics and Philosophy “sc”s: we have plenty of schisms, for at ANU, Oxford and example, several of them involving Cambridge. He has Schrödinger. worked at the University of Sydney since 1989, What were you like as a kid? Were except for a two-year you curious, pulling apart stuff to break early this century, when he see how it worked? was Professor of Logic and I often tried to pull other kids apart to Metaphysics in the Department of see how they worked. Philosophy at the University of Edinburgh. He is now ARC Federation What’s the best thing about being Fellow and Challis Professor of a researcher in your field? Philosophy at Sydney, and heads the I think I’d have to say the adulation. Centre for Time in the Department The adulation and the parking spot. of Philosophy. Who inspires you – either in Huw has written several books science or in other areas of your including Facts and the Function of life? Truth (Blackwell, 1988), Time’s Arrow I think it’s fair to say that most and Archimedes’ Point (Oxford scientists are inspired by fairly University Press, 1996), as well as a ethereal things. Science is not range of articles in academic journals something you plan, like climbing a such as The Journal of Philosophy, mountain: it’s a field in which you Mind, The British Journal for the have to be patient, and willing to Philosophy of Science and Nature.he follow your curiosity wherever it is a Fellow of the Australian Academy goes. And in both those things, I feel of the Humanities, and a Past — like most scientists, probably — President of the Australasian that reality TV is the best inspiration Association of Philosophy. He is also one could ask for. The way in which a consulting editor for the Stanford perfect strangers form strong Encyclopedia of Philosophy,an attachments and even stronger associate editor of The Australasian hatreds within minutes of meeting Journal of Philosophy, and on the each other has inspired me, for editorial board of The Philosophical example, to investigate the Quarterly and Logic and Philosophy mathematical symmetry between of Science. waves which converge on a point in 220 space and those which diverge from continues page 232 Einstein and the Quantum Spooks

Huw Price

Introduction THE INTERNATIONAL YEAR OF Physics celebrates the centenary of Einstein’s amazing debut: the three groundbreaking papers he published in 1905. The most famous paper introduced the first of the two great revolutions in twentieth century physics, the special theory of relativity. Another paper, studying the statistics of Brownian motion, provided crucial new support for the (then still controversial) hypothesis that matter was made of atoms. And the third, proposing a new understanding of something called the photoelectric effect, was one of the important steps towards the century’s second great revolution, the theory of quantum mechanics. So Einstein is not only the father of the theory of relativity. He’s also one of the grandparents of quantum theory that, after gestating for about a generation, was born into the world in another remarkable twelve months for physics, between June 1925 and June 1926.

221 Albert Einstein e know that grandparents tend to be This chapter is about why Einstein was unhappy more indulgent than parents. In the case with quantum mechanics, and about the amazing W of Einstein and his famous theoretical sequel to his objections that later came to light. offspring, however, it was the other way round. This sequel, unearthed by an Irish physicist called Far from being a doting grandparent to quantum John Bell (1928-1990) in 1965, is still one of the mechanics, Einstein always disliked it, or at least most puzzling things in contemporary physics. the interpretation of what it meant that became Nobody really knows what it means. Worse still, it widely accepted in physics during his lifetime. reveals a deep tension between quantum This popular view of the meaning of quantum mechanics and Einstein’s own special theory of mechanics was called the Copenhagen relativity. As the special theory reaches its one- Interpretation, because it was developed and hundredth birthday, in other words, we still don’t championed by the ‘Great Dane’ of twentieth know how to reconcile it with its illustrious eighty- century physics, Niels Bohr (1885-1962). year-old cousin. It is as if these two great theories have lived side by side for eighty years, never properly speaking the same language.

It is true that for many purposes this conflict doesn’t matter very much. Working physicists know how to deal with one or the other theory, as necessary. But the tension is still there, and it is one of the deepest mysteries that Einstein’s century has bestowed on the one we now call ours. It is impossible to say whether the resolution of this mystery will one day lead to new revolutions in physics, in the way that Einstein himself developed relativity as a solution to tensions in nineteenth century physics. But I think we can be sure that there is something important we don’t understand about the physical world, until we find a better understanding of these quantum mysteries.

Fuzzy Pictures versus Fuzzy Reality Quantum mechanics had several parents and grandparents, but the two with best claim to be Niels Bohr fathers of the new theory were young German physicist, Werner Heisenberg (1901-1976), and Bohr was another grandfather of quantum theory, the Austrian physicist, Erwin Schrödinger (1887- and his disagreement with Einstein about the 1961). Heisenberg and Schrödinger discovered meaning of the theory was very much like a bitter what turned out to be different but equivalent family feud. It led to a personal rift between these forms of the new theory in 1925 and 1926, two former friends, two of the giants of twentieth respectively. You’ve probably come across these century physics, which persisted until Einstein’s names already. You’ve heard about Heisenberg’s death in 1955. Even more like a family feud, Uncertainty Principle, and the idea that quantum perhaps, Einstein’s unhappiness with quantum mechanics shows that properties often don’t have mechanics had a lot to do with tensions between sharp values in the quantum world – that a quantum mechanics and his own brainchild, the particle can’t have both a sharply defined position theory of relativity, although the full extent of that and a sharply defined momentum, for example. WAVES OF THE FUTURE OF WAVES tension didn’t become clear until at least a You may have also heard, at least briefly, about 222 decade after Einstein’s death. Schrödinger’s unlucky cat. I said earlier that Einstein wasn’t at all a doting grandfather to quantum mechanics. And Schrödinger tended to side with Einstein on these important family matters. Certainly, he wasn’t at all happy with the interpretation that was soon being placed on quantum mechanics by people such as Bohr. His famous cat first turns up in 1935, as an objection to the Copenhagen view. This is what Schrödinger says: One can even set up quite ridiculous cases. A cat is penned up in a steel chamber, along with the following diabolical device ... In a Geiger counter there is a tiny bit of radioactive substance, so small that perhaps in the course of one hour one of the atoms decays, but also, with equal probability, perhaps none; if it happens, the counter tube discharges and through a relay releases a hammer which shatters a small flask of hydrocyanic acid.

If one has left this entire system to itself for an hour, one would say that the cat still lives if meanwhile no atom has decayed. The first atomic decay would have poisoned it. The ψ Erwin Schrödinger function for the entire system would express this by having in it the living and the dead cat example. If that’s really absurd, as Schrödinger (pardon the expression) mixed or smeared out thought it was, then it follows that the fuzzy in equal parts. reality interpretation of quantum mechanics must be wrong. The fuzziness of quantum mechanics It is typical of these cases that an must be in the picture, not in the world. indeterminacy originally restricted to the atomic domain becomes transformed into macroscopic Schrödinger’s Copenhagen opponents tended to indeterminacy, which can then be resolved by say that in quantum mechanics, reality stopped direct observation. That prevents us from so being fuzzy when we make a measurement – naively accepting as valid a “blurred model” for when we decide to measure either the position or representing reality. In itself it would not the momentum of an electron, for example, and embody anything unclear or contradictory. thereby make it the case that it has a definite There is a difference between a shaky or out- value for one or the other. However if that’s right, of-focus photograph and a snapshot of clouds what about the poor cat? Does it only stop being and fog banks. (Schrödinger 1935) neither alive nor dead when we open the box, and make a measurement – when we look to see Schrödinger’s basic point is if we understand how it is fairing, inside the ‘diabolical device’? quantum mechanics as saying (in the way that Bohr’s Copenhagen school recommended) that The usual answer was that the cat itself is reality is ‘fuzzy’, like a cloud or a fog-bank, then it perfectly capable of making a measurement. is easy to think of cases in which large things, Bohr and his supporters said that ordinary like cats, would have to be fuzzy too. And we’re classical physics applied to big or ‘macroscopic’ not talking about the usual kind of feline things, like measuring devices, and that surely fuzziness, of course: in the experiment as included cats. However, this answer just raises a EINSTEIN AND THE QUANTUM SP0OKS described, the cat would have to be in some further question. What does ‘big’ mean here? indeterminate state, neither alive nor dead, for How big does something have to be to count 223 as a measuring device and to stop the world the box is opened? ‘What counts as a being fuzzy? measurement?’ turns out to be one of the hardest problems to answer in quantum One way to make this problem vivid is to imagine mechanics. Today, it is called the Quantum a range of variants on the cat experiment, in Measurement Problem. Many people think that it which we replace the cat with progressively still doesn’t have a satisfactory solution (although simpler ‘detectors’, all the way down to this is a controversial issue). We don’t have time microscopic objects, such as an amoeba, or a to explore this debate here. Before we move on I virus, or a molecule, or an atom. A few of these want to emphasise two points. possible variations are shown in Figure 1. The notation ψyes + ψno is just a way of writing the First, I want to stress the main reason why this is possibility that quantum mechanics describes, in a problem: according to the Copenhagen view of which it is not yet a determinate matter whether quantum mechanics that Schrödinger was the radioactive atom in the source has decayed criticising (which remains popular today), nothing (‘yes’) or not (‘no’). The quantum state or wave definite happens in the quantum world until a function thus contains two components, one measurement is made. If that’s true, then it is a corresponding to each possibility. very important matter what counts as a measurement. Until we know that, we haven’t For good measure, I’ve also included a variant in understood why the world isn’t just ‘fuzz’, all the which the cat is replaced by something a little way up to the level of our experience. more complex. This version of the experiment has a name. It is called the ‘Wigner’s Friend’ thought The second thing I want to emphasise is that it is experiment, after the physicist Eugene Wigner, Schrödinger’s famous feline, seventy-years-old who suggested replacing Schrödinger’s cat with a this year, which first puts her paw on this crucial human observer. Wigner thought that only issue. That’s why she’s so important, and that’s consciousness could stop the world being fuzzy. why, as far as we can tell, she’ll have a permanent place in the mythology of physics, Here’s the problem, in this new form. In which of along with Archimedes’ bath, Galileo’s feather these various experiments does a measurement and Newton’s apple. take place, to stop the world being fuzzy, before

Figure 1 - variants on the Schrödinger cat experiment, with different 'detectors'. WAVES OF THE FUTURE OF WAVES

224 Einstein and the Completeness of This was the feature of these cases that interested Quantum Mechanics Einstein. In philosophical terms, Einstein was a As we’ve just seen, Schrödinger favoured the realist – in other words, he believed that the world view that quantum mechanics gives us a fuzzy exists independently of minds and observations. picture of a sharper reality. In other words, he He had no time for Bohr’s view that reality thought that quantum mechanics is an depends on what we humans choose to observe. incomplete description – a description that leaves And he thought that the features of quantum out some of the details. But Schrödinger isn’t the mechanics that Bohr and others took as evidence most famous opponent of the view that quantum of deep entanglement between observation and mechanics is a complete description. That honour reality were really a result of the fact that the goes to Einstein, and his strongest argument is in theory gives only a fuzzy description of reality. As a famous paper written with two of his Princeton he saw it, then, the crucial question is therefore colleagues, Boris Podolsky and Nathan Rosen, whether the quantum mechanical description of that appeared in the same year as Schrödinger’s reality can be considered to be complete. Does it Cat. (In fact, Schrödinger’s paper was a response say all there is to be said about a physical system, to the Einstein-Podolsky-Rosen paper, in which or are there further facts about the physical world Schrödinger was offering further arguments for not captured by quantum mechanics? the same conclusion.) The two-particle systems seemed to provide the The Einstein-Podolsky-Rosen (EPR) paper decisive argument that Einstein was looking for. introduces a class of experiments that turn out to He argued like so. First of all, he laid down what involve some of the strangest consequences of he called a criterion of reality – in other words, a quantum mechanics. Now known collectively as principle that tells us when there is something EPR experiments, the crucial feature of these real, ‘out there’ in the world. This criterion says cases is that they involve a pair of particles that that if we can predict with certainty what the interact and then move apart. Provided the result of some measurement would be then there interaction is set up in the right way, quantum must be an element of reality responsible for that mechanics shows that the results of measurement. Let’s write this down explicitly. measurements on one particle enable us to predict the results of corresponding Criterion of Reality: If we can predict with measurements on the other particle. For example certainty the result of a measurement of we might predict the result of measuring the some physical quantity F, then there is position of particle 1 by measuring the position something in reality corresponding to F. of particle 2, or predict the result of measuring the momentum of particle 1 by measuring the This is how Einstein, Podolsky and Rosen express momentum of particle 2 (see Figure 2). this criterion: ‘If, without in any way disturbing a

Figure 2 - An EPR experiment EINSTEIN AND THE QUANTUM SP0OKS

225 system, we can predict with certainty (i.e., with corresponding to the position of Particle 2. probability equal to unity) the value of a physical (This follows from A, by the Criterion of quantity, then there exists an element of reality Reality). corresponding to that quantity.’ (Einstein, C. But by the Assumption of Locality, what we Podolsky and Rosen, 1935) (You might like to do to Particle 1 doesn’t affect Particle 2. So think about whether you agree with this principle. it follows from B that there must be an If not, why not?) element of reality corresponding to the position of Particle 2, regardless of whether The second important ingredient in the EPR we measure the position of Particle 1. argument is an assumption. It says that so long D. Similarly, going through the same three as the two particles are sufficiently far apart, steps for momentum instead of position, what we do to one of them doesn’t affect the that there must be an element of reality other one. Another way to put this is to say that corresponding to the momentum of Particle anything we do to one particle only has effects 2, regardless of whether we measure the locally – and that is why this assumption is called momentum of Particle 1. the assumption of locality. E. So, even if we don’t measure anything on Particle 1, there must be elements of reality Assumption of Locality: There is no action at corresponding to both the position and the a distance. Or as EPR put it: ‘If at the time of momentum of Particle 2. Or in other words, measurement ... two systems no longer there is a sharp reality out there after all, interact, no real change can take place in the and quantum mechanics is just a fuzzy second system in consequence of anything picture of it. that may be done to the first system.’ (Einstein, Podolsky and Rosen, 1935) Thus, Einstein believed that he had given a conclusive argument that quantum mechanics At this point it’s important to keep in mind only gives us an incomplete description of reality. Einstein’s strongest reason for believing in this Accordingly, he thought that there must be extra locality principle. One of the fundamental ‘hidden variables’, of which quantum mechanics principles of his theory of relativity is that nothing didn’t provide us with any account. He seems to – no particle, signal, or causal influence – can have thought that quantum mechanics was just a travel faster than the speed of light. So if two statistical theory, describing the average particles are a long way apart – say, as far apart behaviour of large collections of particles, a bit as the Earth and the Sun – then no change in like the theory of gases provided in statistical one particle can affect the other particle for at mechanics (where properties such as temperature least eight minutes (the time it takes light to and pressure are just averages, not fundamental travel this distance). So from the EPR point of properties of the real constituents of gases). view, the locality assumption seemed to be guaranteed by the biggest Swiss bank in town, The response to this argument from Bohr and his the theory that Einstein himself had thought up Copenhagen Interpretation followers isn’t easy to while working in the patent office in Berne, thirty describe. In fact, many people say that they years earlier. simply don’t understand it. However, because Einstein’s argument is so simple, and obviously Let’s see how the EPR argument goes, given correct if you accept the two principles of the these two principles: Criterion of Reality and the Assumption of Locality, Bohr could only challenge it by rejecting A. If we measure the position of Particle 1, we one of these principles, or both. In fact, he seems can infer with certainty the result of a to have been committed to rejecting both: to position measurement on Particle 2. rejecting Locality, on the grounds that until a (Quantum mechanics tells us this.) measurement is made, the two particles are not WAVES OF THE FUTURE OF WAVES B. So, if we measure the position of Particle 1, genuinely independent; and to rejecting the 226 then there is an element of reality Criterion of Reality, on the grounds that there isn’t a definite element of reality, until a measurement Bohr Institute (NBI) in Copenhagen, founded is actually made. exactly a century earlier, to make way for several new swimming pools. Einstein was aware that his opponents would try to evade the argument in this way, but he From all over Scandinavia, hopeful pairs of wouldn’t have any of it, and for a very good swimmers arrive in Copenhagen, hoping to be reason: as I noted earlier, his own theory of chosen for the elite training squad. Of course, special relativity provided the strongest argument there’s a rigorous selection procedure. Only the in favour of Locality. His attitude is nicely most committed and synchronised teams are summed-up in a famous remark in a letter to his wanted. Here’s how it works. The two candidates old friend and colleague, Max Born (1882-1970) are separated, and isolated in different interview in 1947. (Max Born was another of the fathers, or rooms. Each of them is then asked just one grandfathers, of quantum theory. He was also the question, chosen from a list of three: maternal grandfather of the Australian singer and 1. Would you like to swim for Sweden? actress, Olivia Newton-John - which means that 2. Would you like to swim for Norway? she’s a kind of second cousin of quantum 3. Would you like to swim for Denmark? mechanics!) Einstein writes to Born that he can’t accept quantum theory in its current form, The two questions are chosen at random, because ‘the theory cannot be reconciled with the independently, in the two rooms, so sometimes idea that physics should represent a reality in they’re the same (three times out of nine, in fact, space and time, free from spooky actions at a on average) and sometimes they’re different. And distance.’ (Letter to Born, 3 March 1947, my neither candidate knows what question the other emphasis.) candidate is being asked.

However, the spooks were going to turn out to be Of course, if the two candidates are asked the much harder to eradicate than Einstein had same question and give different answers, then thought and, ironically, he himself had got half that’s the end of the matter. They’re not way to showing why by focusing attention on the sufficiently synchronised, and they’re both shown kinds of experiments involved in the EPR politely to the door. So consistency is absolutely argument. The Irish physicist John Bell made the vital in this case. It’s better if they both say No to next crucial step in 1965, about ten years after the same question than if one says Yes and the Einstein’s death. The crucial part of Bell’s other says No. argument is remarkably straightforward, as we can see by examining a case that has nothing to do But what if they’re asked different questions? In with quantum mechanics, at least on the surface. this case, it’s no use if they both say Yes. That would show that they wanted to swim for different countries, and again, they’d be shown The Scandinavian Institute of the door. And it’s not much better if they both say Synchronised Swimming No. That reduces their chances of getting picked Imagine it is the year 2021. There’s turmoil in the for any of the national teams, even if they do get world of sport. Many countries are still reeling into the Institute. So they have to try to give from their dismal performance at the Auckland different answers in this case: have one say Yes Olympics the previous year where, for the first and the other say No. time in Olympic history, the host nation won every single medal! The three Scandinavian countries – So far, this is just simple sport psychology. Now Sweden, Norway and Denmark – decide to join let’s introduce a little bit of mathematics. Let’s forces, and to concentrate on improving a single figure out the maximum possible success rate for sport. They choose synchronised swimming. At the second part of the strategy – the part that great expense, the three governments establish applies if they are asked different questions – EINSTEIN AND THE QUANTUM SP0OKS the Scandinavian Institute of Synchronised given that they need to guarantee that they Swimming (SISS) – demolishing the famous Neils always give the same answer when they’re asked 227 the same question. Obviously, the only way to It’s not hard to see that the same applies to any guarantee that they give the same answer if of the other five strategies we just listed: YYN, they’re asked the same question is for them to YNY, YNN, NYN or NNY. In each case, there’ll be agree in advance what they’ll say, in response to two out of six possible combinations of questions each of the three possible questions. We can in the table for which the strategy doesn’t give write down their possible policies in this form: different answers – so still a failure rate of 33%. YYN, YNY, etc. Thus YYN means that they would answer Yes to Questions 1 and 2, and No to Let’s summarise these conclusions. We’ve Question 3. YNY means that they would answer deduced that there is no strategy for making sure Yes to Questions 1 and 3, and No to Question 2. that the two candidates give the same answer And so on. when they’re asked the same question, which also has a success rate higher than about 66% It is easy to see that there are just eight possible when they are asked different questions (where strategies of this kind: YYY, YYN, YNY, NYY, YNN, ‘success’ means giving different answers, in NYN, NNY and NNN. Of these eight strategies, these cases). To put it another way, if they make two (YYY and NNN) ensure that the candidates sure they give the same answers to the same give the same answer, no matter what two questions, then they’ll also give the same questions they are asked. So these two strategies answers to different questions, at least about are bad strategies. Remember, the candidates are 33% of the time. As we’ve seen, it’s just a matter trying to maximise their chances of giving different of arithmetic. answers when they are asked different questions. That leaves just six possible strategies: YYN, YNY, Is there any way to cheat the arithmetic? To do NYY, YNN, NYN, and NNY. Let’s pick one of these, better than a 66% success rate, in the different say NYY, as an example, and think about what it question cases? Notice that it would be easy to implies about the chances of a pair of candidates do better if the two swimmers could who chose that strategy giving different answers, communicate, and tell each other what their own when they are asked different questions. question is – if they were telepathic, for example, and could flash a message such as ‘I’ve got the There are six possible ways the two candidates Sweden question!’ to their partner in the other can be asked different questions, as in the interview room. But we’ve assumed that they’re following table; this also shows, for each genuinely isolated. In other words, we’ve combination of questions, whether the NYY assumed that the question asked in one room candidates manage to give different answers. can’t make any difference to the answer the other swimmer gives in the other room. Candidate A Candidate B Different answers? Q1 Q2 Yes Does this assumption sound familiar? It should, Q1 Q3 Yes because effectively it is the Assumption of Q2 Q1 Yes Locality, as used by Einstein, in his argument that Q2 Q3 No quantum mechanics is incomplete. In other Q3 Q1 Yes words, we should really put our conclusion like Q3 Q2 No this:

So with the strategy NYY, pairs of candidates can SISS Theorem: If the Assumption of Locality expect a success rate of about 66% in their is true for candidates interviewed for the attempt to give different answers when they’re SISS, then the maximum success rate in the asked different questions. Some will be lucky, different-question cases is 66%. some won’t, but on average, if the questions are chosen at random, there’ll still be a failure rate of If the Assumption of Locality is true, in other around 33%. words, then there’s no way for the candidates to WAVES OF THE FUTURE OF WAVES cheat the arithmetic. 228 Quantum Mechanics to the Rescue? other one is always blocked, and vice versa. (We What has this got to do with quantum theory? can turn it into a perfect correlation by rotating It’s simple. Somehow, quantum mechanics does one polariser through 90°.) manage to cheat the arithmetic. In fact, if the founders of SISS hadn’t demolished the Niels Another version of the EPR experiment uses Bohr Institute so hastily, they would have electrons. In this case we measure a property of discovered that there’s a way to use quantum the electrons called ‘spin’, which is related to mechanics to get a higher success rate than the angular momentum. Like polarisation, it is arithmetic seems to allow. And the bit of measured in a chosen direction perpendicular to quantum mechanics we need is very similar to the direction of travel of the particles. Whatever the kind of experiments discussed by Einstein, direction we choose, electrons always turn out to in the EPR argument of 1935. Like that have a spin of either +1/2 or -1/2 (don’t worry argument, it involves physical systems in which too much about what the numbers mean) in the we have two particles produced in some chosen direction; and if a pair of electrons is common source, which can then be measured produced in the right way, the total spin must be in different locations. zero, and so there must be one of each. So a spin measurement on one particle enables us to In the original EPR experiment, we had a choice predict the result of a corresponding of two measurements on each particle, either measurement on the other particle (i.e., a spin position or momentum. But there are similar measurement at the same angle, perpendicular to experiments in which we can find three possible the line of flight of the electron), just as in the measurements we can perform on each particle, original EPR case. each of them mutually exclusive with the other two. In other words, just as we can measure In the spin case, the physics gives us a perfect either the position or the momentum but not both anti-correlation. In other words, if we get a result in the original experiment (remember, this is of +1/2 on one side we get -1/2 on the other. But Heisenberg’s Uncertainty Principle at work), so we by making the measurement on one side reveal can measure just one of these three new ‘minus-spin’ (that is, making the device display properties in the new EPR experiments. +1/2 when it measures -1/2, and vice versa) we easily turn this into a perfect correlation. Better One example is provided by the polarisation of still, we can set things up so that on one side the photons, or particles of light. We can measure the measurement device shows YES when it records polarisation of a photon by putting a polarising +1/2 and NO when it records -1/2, and on the lense in front of it, and detecting whether it other side, the same in reverse. passes through. And we can rotate the lense, and thus measure the polarisation in different We then have something with exactly the same directions. What we can’t do is measure the form as the SISS case, if we let the three polarisation in more than one direction at the different questions in the SISS case correspond to same time. measuring the spin in three directions spaced at 120° with respect to each other, perpendicular to Thus versions of the EPR experiment with the line of flight. In effect, there are three polarisation measurements work just as well as different ‘questions’ we can ask each electron, Einstein’s original, for the purposes of Einstein’s and the measuring device produces either a ‘Yes’ argument. If we choose our pairs of photons in or a ‘No’. And if we ask the same question, we the right way, we can find out the polarisation of get the same answer, on both sides. one photon, for a particular orientation of the lense, by measuring the polarisation of the other The simple arithmetic we used in the SISS case photon in the same orientation. Again, there’s a proves that if the questions are chosen at random perfect correlation between the two results – or on each particle, then when the two particles are EINSTEIN AND THE QUANTUM SP0OKS more exactly, an anti-correlation, in the sense asked different questions they will produce the that if one photon goes through the polariser, the same answer at least 33% of the time. As in the 229 SISS case, the argument depends on the Each candidate would take one box, and a three- Assumption of Locality – but as long as that setting measurement device. They’d be instructed holds, it is just arithmetic. to base the measurement setting on which of the three possible questions they were asked, and to This piece of simple arithmetic now has a name base the answer to the question on the result of in quantum theory. It’s called Bell’s Inequality, the measurement on the particle in the box. With after the physicist who first saw its importance. (It careful experimental design, they could certainly is called an ‘inequality’ because it says that the do better than the theoretical limit of 33% – in correlation has to be at least 33%.) Why is it principle, according to quantum mechanics, they important? Because, as Bell realised, quantum could reduce the number of occasions on which mechanics predicts something different. they gave the same answer to different questions Depending on how we set up the experiment, to around 25%. quantum mechanics predicts a correlation as low as 25%, in these cases in which the two particles In other words, quantum mechanics enables our are subject to different measurements. swimmers to do something that is mathematically impossible, if the Assumption of Locality is true. So quantum mechanics must imply that the Assumption of Locality is false! That was John Bell’s great discovery.

Synchronised Spookiness Thus Einstein had assumed Locality, and used it, in an ingenious argument based on these two- particle EPR experiments, to argue that quantum mechanics is incomplete – that quantum theory must be a fuzzy picture of a sharper reality. But Bell showed that those same EPR experiments could be used to show that the predictions of quantum mechanics were inconsistent with the Assumption of Locality. If quantum mechanics is right, then the Assumption of Locality is wrong anyway, and Einstein’s argument for the fuzzy interpretation collapses.

For this reason John Bell is sometimes called the man who proved Einstein wrong. But it is important to be clear what Bell actually proved Einstein to be wrong about. Bell did show that John Bell Einstein must be wrong about the Assumption of Locality (at least if quantum mechanics is true). One way to see how surprising this is is to notice But he didn’t show, as people often wrongly that if SISS hadn’t demolished the Niels Bohr assume, that Einstein was wrong about quantum Institute, they could have used a real-life device, mechanics being incomplete. It could still be true, based on quantum mechanics, to cheat the as Einstein thought, that there are extra ‘hidden arithmetic we derived above. In principle, it could variables’, not described by quantum mechanics. work like this. Electrons or photons would be It is just that they couldn’t be local hidden produced in the right kind of pairs, and directed variables, satisfying the Assumption of Locality. into mirrored boxes, where they could be stored Somehow, the measurement made on one WAVES OF THE FUTURE OF WAVES until needed. particle would have to affect the hidden variables 230 of the other. (There are some well-developed extensions of quantum mechanics of this kind. As we saw earlier, however, the strongest reason The best-known was invented by the physicist for believing in Locality is Einstein’s own special David Bohm (1917-1992), who also invented the theory of relativity. Accordingly, the real version of the EPR experiment described above, importance of Bell’s results is that they expose a on which Bell’s analysis was based.) very deep tension between the two most important theories in twentieth century physics. Roughly speaking, quantum mechanics does something that simply shouldn’t be possible, according to special relativity. Quantum systems can be ‘entangled’ in some strange way, even when they are very long distances apart. (In principle, we could arrange an EPR experiment in which the two particles had travelled light years apart. Almost everybody in physics now believes that even in this case, the bizarre effects of quantum entanglement would still apply.)

It is true that there are some subtleties here, which soften the blow a little. It looks as if the strange non-local correlations that Bell noticed in EPR experiments can’t be used to send faster- than-light messages, for example – there’s no prospect of an instantaneous ‘Bell telephone’, as someone once put it. But the conflict is there all the same, and Bell himself thought that it implied that Einstein’s own understanding of the meaning of special relativity was wrong – that we had to David Bohm go back to the ‘pre-revolutionary’ ideas that physicists such as Lorentz had developed Where do we stand, then, in the light of Bell’s before Einstein. result? First, Einstein’s best argument for the fuzzy picture view of quantum mechanics has been seriously undermined. Secondly, and more Back From the Future? importantly, Bell has put his finger on a simple To end this chapter, I want to describe another and basic fact: if quantum mechanics is true, curious idea, sometimes suggested as a way of then the Assumption of Locality is false, and vice resolving the tension between quantum versa. If quantum mechanics is true, in other mechanics and relativity. I’ll introduce it by going words, there really is spooky action at a distance. back to our synchronised swimmers in 2021, trying to ensure that they give At the time of Bell’s original work, different answers when they’re in the 1960s, experiments to test asked different questions at the the predictions of quantum “...the strongest reason SISS admission interviews. Think mechanics were not technically for believing in Locality is how easy it would be for the feasible. But by the 1970s, various swimmers if they had experimenters were devising ways Einstein’s own special precognition – if they could just to do it, and since then the theory of relativity.” ‘see’ in advance what question predictions of quantum mechanics they were going to be asked. If have been confirmed many times. they knew this before they left (The best-known results are those of a team led by each other’s company, then it would be easy for EINSTEIN AND THE QUANTUM SP0OKS the French physicist, Alain Aspect.) So very few them to collaborate, to make sure they gave people doubt that ‘Non-locality’ is here to stay. different answers. (For example, suppose the two 231 swimmers foresee that they are going to be in the future? But it’s worth examining this issue asked questions 1 and 2 from the list of three. a little more closely. After all, we don’t find They could then adopt a strategy such as anything controversial about the idea that YNY, which gives different answers to these photons and electrons know something about two questions). what happened to them in the past; or in other words, less metaphorically, that their properties This possibility reveals another hidden depend on what happened to them in the past. assumption in the mathematical argument we So why not the future, too? used to show that the maximum possible success rate was around 66%. We were assuming, At this point, we get to a fascinating tug of implicitly, that the swimmers didn’t know the intuitions. On the one hand, it seems just obvious questions in advance – that their strategy had to that causation only works one-way, from past to be independent of the choice of question in future – the past can affect the future, but the the future. future can’t affect the past. On the other hand, however, the basic laws of physics seem to make In the case of quantum particles, this amounts to no distinction between the past and the future. At the assumption that the properties of the two the fundamental level, physics is almost entirely particles in an EPR experiment cannot be affected time-symmetric, in the sense that if it allows a by the kind of measurement they’re going to process to happen then it also allows the reverse encounter in the future. (If they’re affected by the process to happen (roughly, what we would see if future measurement, then they ‘know about it’, at we reversed a video of the first process). So least metaphorically speaking; and again, there’s where does the past-future bias of causation a possible way of cheating the arithmetic – come from, if it isn’t in the fundamental physics? getting a success rate higher than 66%.) We don’t have time to explore these issues here. Of course, this assumption seems uncontroversial. They would take us deep into philosophy, as well How could photons and electrons possibly know as physics. But to finish up with, let’s think about anything about what is going to happen to them what it would mean for the conflict between Bell’s

continued from a point, such as when you throw a that would be one option. It is page 220 stone into a still pond or spa bath. possible to make a bit of money in These symmetries turn out to be science, but the really big bucks go greater than many have previously to the less honest among us. Another thought, and from this idea one can thing I might have liked to specialise get almost a whole theory of in would have been philosophy. That backwards-in-time causation ... might have been fun. something which might come in very handy for contestants on Survivor. What’s the ‘next big thing’ in science, in your opinion? What’s If you could go back and coming up in the next decade specialise in a different field, or so? what would it be and why? As someone at the very theoretical When I was appointed to a Personal end of the spectrum it’s hard for me Chair at the University of Sydney, I to know what will be next. In actually had the opportunity to name theoretical science you can never say the chair, and I thought of calling with much confidence what the next myself Professor of Management big thing will be, because if you WAVES OF THE FUTURE OF WAVES Consulting, so that I could make a lot could you’d already have it. But it’s 232 of money on Friday afternoons. So fun to guess. result and relativity, if we allowed ‘backward of two component actions, each of which is causation’. Suppose the properties of our two thoroughly compatible with relativity. (In other electrons are affected by the measurements we words, the proposal shows how we could have a choose to perform on them in the future. Then kind of pseudo-non-locality, that isn’t really in when we choose what measurement to make on tension with relativity. Apparent action at a Particle 1, we affect its properties, all the way distance gets resolved into a kind of zig-zag back to the source. However, at the source, effect, where the ‘zig’ goes backwards in time.) Particle 1 interacts with Particle 2. So by affecting the properties of Particle 1, it is I stress that at this stage, this is just an intriguing possible, at least in theory, that we could affect idea. It hasn’t been developed very far, and most Particle 2, as well; and Particle 2 could then carry physicists and philosophers seem to think that this those effects into the future, to the time at which kind of backward causation is even more spooky its properties are measured, on the other side of than the actions at a distance that Einstein hated the experiment. so much. But as I mentioned a moment ago, the temporal bias of ordinary forward causation is itself But this means, the choice of measurement on a bit spooky, or at least mysterious, in the light of one side of the EPR experiment could affect the the apparent time-symmetry of fundamental results on the other side – and all without any physics. So it’s just possible that this strange idea spooky actions at a distance! All the actions are will turn out to rid physics of two spooks, though ordinary local actions, and the only novelty is that admittedly at the cost of some considerable one of them works backwards in time. damage to naive ideas of cause and effect. If so, then it will have turned out that Einstein was right So here’s a possible resolution of the mystery, a after all, in two ways: first, in thinking that there is resolution which ought to make Einstein happy, at no genuine action at a distance; and second, in least in one sense, because it avoids spooky believing that quantum mechanics is incomplete action at a distance. On this account, the ‘non- (for if this proposal is right, then ordinary quantum local’ effects that look like action at a distance mechanics leaves out the mechanisms responsible actually turn out to be the result of a combination for these zig-zag causes).

I’d like to speculate that time- same thing will happen to the new symmetric theories of physics, theoreticians of computing. incorporating backwards causation, The third big thing I predict is will be one big thing in theoretical something like a cross between old- physics. Also probably esoteric theories skool rap and progressive house, of computation: that field’s been a bit only with a bit of an acid jazz static since shortly after Turing backbeat. Either that or yet another founded the field in the 1930s, but if blatant Christmas song, perhaps with quantum computing works then Rolf Harris. Turing’s whole theory will have to be reworked. Turing was an interesting Q&A with Professor Huw Price character. He is perhaps most famous (with a little help from Jason for his saying, “It’s amazing what Grossman) people will believe if they read it on the web, even though they wouldn’t believe it if they read it anywhere else.” Widely recognised as a genius at an early age, he was driven to suicide EINSTEIN AND THE QUANTUM SP0OKS in the 1950s by the homophobia of his society. I can’t predict whether the 233 Chapter 7 Figure 2: R. Williams and the Hubble Deep Image credits Figure 1: Morawska, L and Salthammer, T., Field Team (STScI, NASA). 2003, Indoor Environment: Airborne Particles Figure 3, 4: Bennet et al. All images are © the original owners, or as and Settled Dust, (Wiley-Vch: Weinheim, Figure 7: Anglo-Australian Observatory, stated below. Germany), ISBN 3-527-30525-4. photograph by David Malin. Figure 8: (a) ACS Science & Engineering Chapter 1 Chapter 9 Team, NASA; (b) Kirk Borne (STScI) and Figure 1: C. R. Scotese, Paleomap Project, Figure 1: Davisson C and Germer L H, NASA; (c) NASA, H. Ford (JHU), G. Hillingworth UniversityTexas, Arlington, U.S.A. Physical Review 30(6), (1927) 705 (UCS/LO), M. Clampin (STScI), G. Hartig Figure 4: Seebacher, F. and Alford, R. A., Figure 2, 3, 7: Cockayne DJH, Physics (STScI), the ACS Science Team and ESA. 1999, Movement and microhabitat usea Education 40 (2) (2005) 134 Figure 9: R. Boonsma & T. Oosterloo, terrestrial amphibian (Bufo marinus) on a Figure 4: P. R. Buseck, et al., Proceedings of RUG/ASTRON. tropical island: seasonal variation and the National Academy of Science 98 (24) Figura 10: CSIRO-ATNF. environmental correlates, Journal of (2001) 13491 Figure 11: T.A. Oosterloo and K. Kovac, Herpetology, 33, pp 208-214. Figure 5: C. Hetherington and A. Kirkland, Oxford. ASTRON/RUG. Figure 6, 7, 8: Mark Read, EPA, QLD. Figure 6: G. Winkelman and C. Dwyer, Oxford. Figure 12: Putman et al. 2003 ApJ 586, 170. Figures 10, 11, 12: Seebacher, F., Franklin, C. Figure 8: copyright Materials Evaluation and Figure 13: J.M. van de Hulst, RUG. E. and Read, M., 2005, Diving behavioura Engineering, Inc, Plymouth. Figure 14: (a) M. Rejkuba (ESO) et al. (b) reptile (Crocodylus johnstoni) in the wild: Figure 9: V. Keast, Sydney. J.Keene (SSC/Caltech) et al. interactions with heart rate and body Figure 10: from TEM: A textbook of materials (JPL,Caltech,NASA), (c) from Schiminovich D. temperature, Physiological and Biochemical science, Plenum Press (ISBN 03064532XX, van Gorkom J.H., van der Hulst J.M., Kasow Zoology, 78(1), pages 1-8, 2005, with 1996). S. 1994, ApJ 423, L101. permission from Publisher: University of Figure15: NRAO/AUI and STScI/NASA, Chicago Press. Chapter 10 investigators Hibbard J.E., van der Hulst J.M., Figure 1: Brian P. Gorman, Department of Barnes J.E., Rich-Whitmode B. & Schweizer F. Chapter 2 Materials Science and Engineering, University Figure 16: NRAO/AUI and J. Hibbard & J. van Figure 4: adapted from A. Ashkin, Scientific of North Texas. Gorkom. American 226 (2), page 62 (1972). Figure 5: Tanaka, Terauchi and Kaneyama, Figure 18: J. Barnes. Figure 7: Michael Berns, Irvine, California. JEOL Ltd, Tokyo. Figure 19: NASA/CXC/STScI/U.North Figure 8, 12: P. Ormos, Institute of Figure 9: J. Brook, J. Sloan, A. Briggs, Carolina/G. Cecil. Biophysics, Biological Research Centre of the Department of Materials, Oxford University. Figure 20: Simon Driver & Alberto Fernandez- the Hungarian Academy of Sciences. Figure 10: K. Jacobi, Max Planck Institute. Soto, UNSW. Figure 9: K. Schütze, PALM Corporation, Figure 11: C. Lang, Department of Materials, Germany. Oxford University. Chapter 14 Figure 11b: K. Dholokia, School of Physics and Figure 18: G. Winkelman, Department of Figure 1: X-ray - NASA/CXC/MIT/UCSB/P.Ogle Astronomy, St Andrews University, Scotland. Materials, Oxford University. et al; Optical - NASA/STScI/A.Capetti et al.; Figure 19: G. Winkelman, C. Dwyer, John Bahcall (IAS), NASA); Alan Bridle; Chapter 3 Department of Materials, Oxford University. NASA/CXC/M. Karovska et al.); NRAO/VLA Figures 1, 2, 4, 5, 6: from Biomedical Uses of Figure 20: from F Ross, IBM J. Res. Develop. van Gorkom/Schiminovich et al.; J.Condon et Radiation, William R. Hendee, (ed.), Wiley- 44(4) July 2000. al.; Digital Sky Survey U.K. Schmidt/STScI. VCH: Brisbane (1999) ISBN 3-527-29668-9. Figure 21: J. Sloan and A. Kirkland, Figure 2: CXC/M.Weiss. Department of Materials, Oxford University Figure 3: NRAO/AUI Miyoshi et al. and Gerald Chapter 5 Figure 22: P. Midgely, Department of Materials Cecil and Holland Ford. Figure 1: Stephen G. Eick, from the book Atlas Science and Metallurgy, University of Cambridge. Figure 4: NASA and Ann Field (Space of Cyberspace, M Dodge and R Kitchin Telescope Science Institute). (Addisom-Wesley), 2001 ISBN 0-201-74575-5. Chapter 11 Figure 5: W. Jaffe (Leiden), H. Ford Figure 2: Lucent Technologies Inc. Figure 1, 2: Kandel, ER, Schwartz, JH, and (JHU/STScI) and NASA). Figure 5, 6, 7, 8: Dr. R Slusher, Bell Labs, Jessell, TM, PrinciplesNeural Science, 3rd Ed, Figure 8: NRAO/AUI, Fomalont et al. Lucent Technologies. Appleton and Lange, Norwalk, Connecticut, Figure 10: R. Perley, C. Carilli & J. Draher, 1991, ISBN 0-8385-8034-3. NRAO/AUI; Krichbaum. Chapter 6 Figure 6: Nunez, PL, Neocortical Dynamics Figure 12: NRAO/AUI. Figure 3: T.T. Ng, J.L. Blows, J.T. Mok, P. Hu, and EEG Rhythms, Oxford Univ. Press, Oxford, Figure 13: Polatidis et al. J.A. Bolger, P. Hambley, and B.J. Eggleton, 1995, ISBN 0-19-505728-7. Figure 14: Gomez Jose1-Luis et al. Simultaneous residual chromatic dispersion Figure 16: NASA and J. Bachall (AS) - NASA, monitoring and frequency conversion with Chapter 12 Martel et al., the ACS Science Team and ESA. gain using a parametric amplifier, Optics Figure 1, 3: Sager, D A, Introduction to Ocean Figure 17: L. Saripalli, R. Subrahmanyan, Express 11, 3122-3127 (2003). Sciences (Wadsworth: Belmont, CA) 1998. Udaya Shankar (ATCA). Figure 5: S. Radic, C. J. McKinstrie, R. M. Figure 2: Neuman, G and Pierson, W J, Jr, Jopson, J. C. Centanni, and A. R. Chraplyvy, Principles of Physical Oceanography (Prentice Chapter 15 All-Optical Regeneration in One- and Two- Hall: Englewood Cliffs, NJ) 1966. Figure 4: AIP Emilio Segré Visual Archives Pump Parametric Amplifiers Using Highly Figure 4: Davies, J L, Geographical Variation Figure 6: photograph by Francis Simon, Nonlinear Optical Fiber, Photonic Technology in Coastal Development. 2nd Ed (Longman, courtesy AIP Emilio Segré Visual Archives, Letters 15, 957-959 (2003). London) 1980. Francis Simon Collection. Figure 6: S. Radic, C. J. McKinstrie, R. M. Figure 5: Short, A D (ed), Beach and Figure 8: Nick Robins, ANU. Jopson, J. C. Centanni, and A. R. Chraplyvy, Shoreface Morphodynamics (John Wiley and Figure 9: Nick Robins, ANU. All-Optical Regeneration in One- and Two- Sons: Chichester) 1999. Reprinted from Atlas Pump Parametric Amplifiers Using Highly of the Oceans, Wind and Wave Climate, Chapter 16 Nonlinear Optical Fiber,” Photonic Technology Young, I R and Holland, G J, page Figure 1: (b) Alan Henderson, copyright Letters 15, 957-959 (2003). 241,Copyright 1996, with permission from Museum Victoria, Australia Figure 7: (a) from R.F. Cregan, B.J. Mangan, Elsevier. Erwin Schrodinger: photograph by Francis J.C. Knight, T.A. Birks, P.St.J. Russell, P.J. Figure 10: Short, A D, Australia beach Simon, courtesy AIP Emilio Segré Visual Roberts and D.C. Allan, “Single-mode systems – nature and distribution. Journal of Archives, Francis Simon Collection.

WAVES OF THE FUTURE OF WAVES photonic band gap guidancelight in air, Coastal Research, in press, 2005. Neils Bohr: Neils Bohr Archive, Copenhagen. Science 285, 1537-1539 (1999). (c) Photonics and Photonic Materials Group, Chapter 13 Einstein on a bicycle, page 36: the Archives, 234 University of Bath. Figure 1: Nico Housen, European Southern California Institute of Technology. Observatory. The Messel Endowment

THE INTERNATIONAL Science Schools were established by 2005. There is information in the Professor Harry Messel in 1962 to recognise and reward booklet given to you on how to talented senior high school students, and to encourage them apply. Secondly, you could approach someone in your to pursue careers in science. Alumni of the International neighbourhood, for example a Science Schools can be found in senior positions in all walks company or philanthropic institution of life, with many of them acknowledging that ‘their’ Science or even a member of your family School was responsible for changing their lives, and recalling with the capacity to contribute, to its two weeks as an exciting developmental experience. seek a donation for the Messel Endowment. A personal approach is The Science Schools, renamed Foundation, be used to support always the best method of the Professor Harry Messel other initiatives names to honour obtaining such support and a International Science Schools in Professor Harry Messel. Examples person such as yourself, who has 1999 to honour Harry for his of such initiatives include a just benefited from an International foresight, have continued Professor Harry Messel Visiting Science School, makes the best uninterrupted since then. In order Chair and a Professor Harry Messel ambassador. Finally, we realise that to ensure the future of the Lectureship. you may not be in a position, at International Science Schools, so present, to give to the Endowment that future students may also As acknowledged with gratitude in yourself – however, at some stage benefit, the Science Foundation for the front of the book, many people in the future, if you are in a position Physics has established the Messel and companies have already to do so, we ask that you consider Endowment. The Foundation aims contributed to the Messel it then. A donation form has been to raise $4,500,000 in 2005 Endowment, and the 2005 included on the next page to assist Australian dollars. At the time of Professor Harry Messel you if you choose to proceed with printing, the fundraising campaign International Science School has some of these suggestions. More has come a long way: in 2005, the received considerable benefit from donation forms are available on Endowment sits at just over half of the amount already raised. Donors www.physics.usyd.edu.au/ the amount needed. Despite this are also honoured on a web page, messel/index.htm. success, for which we are www.physics.usyd.edu.au/messel/ enormously grateful, the index.htm; they will be A contribution of A$25,000 to the Endowment still has a long way to acknowledged in all books of the Messel Endowment will ensure the go, and the Foundation is lectures presented at future participation of one student in determined to achieve the target. International Science Schools, and perpetuity. on a permanent display in the The Messel Endowment will be School of Physics. Please join us today in our vision managed so that the real value of for the young scientists of the capital is preserved. Bequests We hope you have enjoyed your tomorrow through the Messel will also be sought to ensure the time at the International Science Endowment. growth of the capital in the years to School. Many people, after come. The primary purpose of the experiencing the Science Associate Professor Endowment is to support the Schools, wish to give something Robert G Hewitt International Science Schools. If back, but are not sure how to do Director, income from the Endowment this. Here are some suggestions. Science Foundation for Physics

exceeds the requirements of the First, you could consider coming June 2005 THE MESSEL ENDOWMENT Science Schools the funds may, back as staff for the next with the approval of the Science International Science School in 235 THE MESSEL ENDOWMENT DONATION FORM

You can contribute to the Messel Endowment in one of a number of ways. All donations to the Messel Endowment by Australian residents are fully tax deductible. Donations by cheque must be posted to the address given below. Donations by credit card may be faxed to the number given below.

NAME COMPANY ADDRESS STATE POSTCODE COUNTRY TELEPHONE (BH)(AH)

PLEASE FILL IN THIS SECTION FOR A ONCE-ONLY DONATION

I enclose a cheque, made out to the University of Sydney for A$ Please charge my credit card nominated below for the sum of A$ ❍ Bankcard ❍ Visa ❍ Mastercard CARD NUMBER /// EXPIRY DATE / CARDHOLDER NAME DATE / CARDHOLDER SIGNATURE

OR PLEASE FILL IN THIS SECTION TO GIVE A YEARLY DONATION (AN ANNUAL REMINDER WILL BE ISSUED).

I undertake to give A$ once a year for ❍ 2, ❍ 3, ❍ 4 or ❍ 5 years and enclose a cheque, made out to the University of Sydney as a first installment.

OR Please charge the credit card nominated below for the sum of A$ once a year from / (insert month & year) for ❍ 2, ❍ 3, ❍ 4 or ❍ 5 years. ❍ Bankcard ❍ Visa ❍ Mastercard CARD NUMBER /// EXPIRY DATE / CARDHOLDER NAME DATE / CARDHOLDER SIGNATURE

The Science Foundation for Physics Residents of the USA should contact School of Physics A28 The Secretary The University of Sydney NSW 2006 The University of Sydney USA Foundation Australia Facsimile: +612 9351 7726 Telephone: (607) 277 1346 Facsimile: (607) 277 0078 Email: [email protected] WAVES OF THE FUTURE OF WAVES

Thank you for your investment in the future of young people through The Messel Endowment. 236