G N I T U P M O C

30

THE ART A RACE IS ON to build a different kind of computer that will exploit the peculiarities of OF QUANTUM the quantum world to accomplish number- crunching feats that are currently impossible. The principles by which a ‘quantum com- COMPUTING puter’ would work are quite different from those governing today’s ‘classical’ computers. Can ‘spooky action at a distance’ be Despite current processor and memory designs already having to take quantum- harnessed to build a new class of computers? mechanical effects into account, quantum computing is something far stranger.

by Miya Knights BITS AND QUBITS Classical computers work by manipulating binary digits, or bits. Quantum computers work by manipulating qubits. A classical bit

Engineering & Technology | January 2007 | www.theiet.org/engtechmag G N I T U P M O C

31

can take the values 1 or 0. A qubit can be 1 or 0, or both at By exploiting entanglement, the state of the same time. This ‘superposition’ of states is possible one qubit can be linked to the state of another, because of the ambiguity inherent in quantum so that setting one qubit to register the value mechanics, in which an observable property of an object 1 or 0 sets the other to the same value, despite can possess more than one value at once. the physical separation. This ability of qubits Classical computers encode their bits in discrete to become entangled enables quantum gates voltages, representing a binary 1 as a voltage above 2.4V, to be built. for example, and a binary 0 as a voltage below 0.8V. To be useful, a quantum computer must Quantum computers encode their qubits in quantum also give an answer to a computation. Getting states, such as the spin (up or down) of an atom, or the the qubits from their ambiguous, superposed polarisation (horizontal or vertical) of a photon of light. state to an unambiguous state is done by a Unlike classical computing, in which small voltage errors process called ‘quantum measurement’, do not matter, quantum computers are very sensitive to which turns each ambiguous qubit state into slight errors in their qubit values. a definite 1 or 0. The answer to the compu- Once a quantum computer has more than one qubit, tation is given by the sequence of 1s and 0s it’s possible to exploit another aspect of quantum that results from the process of quantum mechanics called entanglement, which Albert Einstein Above, left to right: measurement. described as “spooky action at a distance”. According to Two barium ions in a

s trap (University of e quantum mechanics, two entangled quantum states have THE POWER OF UNCERTAINTY

m Washington); a a J to be described by reference to each other, even though the surface electrode ion Why bother with all this strangeness? The t t a atoms, ions or fundamental particles that hold the states trap made at NIST; part answer is that computations performed by M

: may be physically separated. This is counter-intuitive, but of an optical bench; a quantum computers could be incredibly n o

i two-layer alumina ion t has become a well-accepted part of modern science. As powerful. A quantum computer could test a

r trap; part of t s Niels Bohr once said: “Anyone who is not shocked by Christopher Monroe’s every possible combination of input values u l l I quantum theory has not understood it.” ion trap experiment for its qubits simultaneously. Each time a

Engineering & Technology | January 2007 | www.theiet.org/engtechmag G N I T U P M O C 32

qubit is added to the quantum computer, its computational Above, left to right: However, progress is being made. In 1995 power doubles. If a quantum computer has 1000 qubits, it a multizone ion trap, Christopher Monroe and David Wineland, which can confine ions can test 2 1000 combinations of inputs at once. There are working at the US National Institute of in any of 10 zones: 300 about 2 fundamental particles in the universe, so even if experiment detail; the Standards and Technology (NIST) built the every one of them were part of a cosmic classical laser system used in first two-qubit logic gate using trapped computer, it would fall far short of being able to store the trapped barium-ion beryllium ions. In 2005, a team led by quantum computing information held in 1000 qubits. experiments at the Professor Rainer Blatt at the Institut für Quantum computers could therefore be fundamentally University of Experimentalphysik in Austria built a more powerful than ordinary computers. The best Washington prototype of a quantum computer using computers of today cannot factorise numbers with more several calcium ions. His group has entangled than 512 bits. In 1994, computer scientist Peter Shor, up to eight calcium ions, the most yet. working at AT&T’s Bell Labs, showed that quantum However, a useful quantum computer would computers would always finish the factorisation of a large need hundreds or thousands of qubits. number in fewer steps than a classical computer. Even if Prototype quantum computers have the classical computer has a much faster clock, for already proved themselves against classical numbers with a large sequence of digits, the quantum computers, although not in real-life computer will finish more quickly. Since the protection situations. Professor Christopher Monroe, afforded us by modern cryptography is based on the now at the University of Michigan, says: difficulty of factorising large numbers, there’s plenty of “An example from our laboratory has been a interest in developing workable quantum computers. database search algorithm, where the number of queries before finding an BUILDING A QUANTUM COMPUTER element in the database is smaller (on The trouble is, no one has yet managed to build one. Each average) than is possible with a classical time physicists add another qubit to a quantum computer, computer. But the database only stored four it makes the engineering task more difficult. Adding elements, so that’s not useful. Furthermore, qubits makes the quantum computer much more the clock speed on existing tiny quantum vulnerable to decoherence, in which the state of the qubit computers is very slow.” degrades. Decoherence can be caused by interactions with Monroe’s approach to developing quantum the external world, and so far it has proved such a problem computers has been to use ion traps. By that experimenters find it difficult to maintain coherence carefully arranging electric and magnetic in their systems for more than a few seconds. fields, an ion can be held at a point in space.

Engineering & Technology | January 2007 | www.theiet.org/engtechmag G N I T U P M O C

33

Trapping the ion isolates it, which makes it is easier to Above, left to right: electrodes that trap these atoms. We have avoid decoherence. The problem with ion traps as a basis ion trap detail; lab set- learned that this noise, while not well up; alumina ion trap for quantum computers, however, is that adding more understood, can be suppressed by simply detail; monolithic ion than a few ion qubits poses tough engineering problems. trap with cadmium ion cooling the electrodes with liquid nitrogen. In Monroe’s group has performed ion-trapping experi- fluorescing under laser the future, we may need to cool to [much ments on a semiconductor chip. He says: “We and other illumination lower] liquid helium temperatures to get rid groups have only trapped a couple of ion qubits in these of this noise almost completely.” chip traps, and they have problems that still need to be worked out before they’re ready to do a quantum SOLID STATE computation. But I’m confident there will be great “The consensus is that ion traps are progress on this front.” particularly promising, mainly due to their Building an ion chip trap is a laborious process. “The history of being the only system to have trap was designed and fabricated by a student in the demonstrated all the necessary ingredients, Michigan group, Dan Stick,” Monroe says. “He spent two albeit with only a few quantum bits,” says years devising ways to lithographically carve and etch Monroe. “We know exactly what we need to do three-dimensional electrodes out of a chunk of gallium to scale up the ion trap, with no fundamental arsenide semiconductor material. More recently, other roadblocks in the way – but it will be groups – particularly the groups of Dave Wineland at technically difficult.” NIST and Richart Slusher at Lucent Technologies – have Another approach would be to use shown how to do this much more easily in a different superconducting circuits, which Monroe says geometry.” have been improving greatly in the past five David Wineland, group leader of ion storage at NIST, years. But he adds: “Many feel that some type says: “As with Chris Monroe’s group, we want to fabricate of solid-state structure will ultimately rule a device in such a way that making many traps involves due to their ‘easy’ scalability.” the same number of steps as making one. So we turned to Ion traps and solid-state approaches each lithography, micro-electromechanical systems and related have their advantages, according to Dr Boris techniques. We found a way to make traps where all the Blinov, who leads the trapped-ion quantum- electrodes lie in a single plane. Such 2D geometries are computing group at the University of easier to fabricate than 3D structures.” Washington. But Monroe adds: “Problems have surfaced concerning “Solid-state methods are very attractive the reduction of electrical noise emanating from the from the point of view of fabrication,” he

Engineering & Technology | January 2007 | www.theiet.org/engtechmag 34 COMPUTING al be Qu le h ma e c un N on mi tr abo c c succ h i th sys Eng sa de on ry pt omp uter omp uter x ig an IC HE aps ss g at ys. PAI tre cau se coh erenc T imp a r a b a e ine gh t iqu W or an eali kin g tem s h s ut t r bs nd h u d, m ee e a dec ist t in e m it hm og pr u tu m tr ri are re o et re n se el w “ U m be enj OR d act sef N ng e ely ta r o wh oba Ho ra n ha y la o d a aybe w c ha we TI iversty heren ce & in nt th a urr s oy u ph c n s. tha physic lar en WO wil l t we c fro I Te urr cha u bil it l we s d e, om NG a am t par y , chno sef t bee n ca th ent ge ext r I ve m e for a qua ntum ca RL D-C H I we’re en t was or use ven ll en gin g. nd pu te rs, probably n e y , u ca pla r , proce ssor , t of ut i l o ly og he l HYS P eme ly . f pro ces si ng conepts n be ly at sh the prg o nig on y to n’s Washing every qubits ‘Fau 14 Tr see n my on s fo pai st ar t | or in le ad ing Ja wrong . lea st , apped r A try ing b nu nt li chosing lt-T o t : if NG ING ? ulk rammi fe, th e com puter by ary to tcaer low “I ’d It physic. te day . ICS loo k li ne. ” th ey le but for 20 7 s m. r do but in many of mi le ad ; ee ra ion s Ion in to But ever y dec oheren ce nt quite s in more, resea r until gh t not hi ng | sol ay the a ca n physic w .theior Boris und ersta a Q A I’d trapp ers ho weve r , as str aight forward and ma rath on if u id app be al mo st wil l recntly w a techn ical a tr be pi be nt it ng su the ch ith Blinov while. appe state xel um licat fair atoms, works, ch inst loo k n deve lo ped , like in mos t nd thinkg as eg ner g/ in C ead. d-i sol id-st ate are ions it to an a tec eng The o say: rates. ‘Hg a the and m like an ra ce , folwed go od on proble ms re y it lik on sa y puti enc oura ged Naturly, , e al-p works.” hm EDM’ im age, pr edi ct io sol id-st ate nt e “I overcome wil l, pron q about the Ab ‘C ly such ng ag Trap ped but anglem th at uant an al oyg narowly ma y path a p ur o ’. way dmiu ” ve it ot qub and wit h e on ly s creating her um bu ose io n are a own a be as t to to nd t r it n o ‘Wavefunctio t nt escapd has Ye l eft: path. was low rem paintgs Blinov first That Entagled’ ained b e on com is and Cola base th an un in e til g lis ps d I an t, e’ o , n h harves Realit co theore ha wh an pl man b bg e co av be co co i co mac si an quant simple f th mo ex classica M qu ta qua adde li c its pro bra t te per c un r c p or t h e a o o a r k u mu c sk onr i su l e mp mp mm mm mp i ai p am m m i ra nr d an el iver s a m ces D W B “ “ “ hno s en de o ck e fmoer cess ins . ab an wor nt It Qua In la s hin arly pt y u p l p po li u avid aged ts. l d e es la t tu t ut ut ut pl ll l oe s n nder s.” bl l um p um n b y’ t an er un el is i e an , tical e ses. ma ed ss lyg o ting h s wh m t o m e et to r t in er er x ld, io n ing . es. e e : e l n at c v qu he An o ve i . ic i is r d. of I te d “ i u u fr d bl tum g, s u m ef g n t co y B w a s ru en machines Deutsch Ho o a s a t stan this h nt th computers ati n ti To “ s, ry r antum ene r fa n om l i e A fi u o – rg e h to h wr e a mpu h T iv n 20 computat do c natu th nn l l a app ther er d t q ci t th s s we av ner on h wi k ha an do un e is e ua n c c t s g l qu th e cy , co rgie at mest be e it e o uma d, o e e a r oll in e l n o i o p e roundwork l n s t l v o l s: es n s d s t ed w w ist te li hy ri l mp n o m a th ssin ant ex re er s tl k , sy to e g t is th i c abor a pe tu ke il he e b s r ra e a bu n i y s “I . mo , , entu in is by ll si systems ng pe as p , d n st i e l ew a w utat w m of um rfmor [ o cat ly I whe wer e ows s c e m who b c g h t t t ems I an na n c hi l ct re within ai bei di il l s in re hey th e ions er be c t as fu a e i som a g ec that th phys c l p m o e e c ry, n t t s co l a t st n tura d i g en ion, o ty a a r io la si li n s be n com s t n c i w at a b a m t t inc u v c i ev ] – g e ca had he f gs ss pe of f n ed laid n d oo .” t co tools h th l i e o t s a se ul n tr ai w pu the h i a o v w l ics r e e nfor ic will 1 l l b l of is e bl r whi ni v f e h t k an hi that e m put e es l- ful and nat 0 f e t r a cr i outsi et laws ap x n te for m 10 al of i u ed i thi f sc e to e . ‘ y t po n nt le utr h b si ee the The ce d e ar we n a m r a n in ac ea pl e t al e co n i u ate is ch s m s mu a t powere d o o n t u n n a f r outperfor y ly h hin technol r vent as ic n e hi or rs lot sma en s t c a cann s r e b m ati s mpu s n o i t no ai m t fr o d de t ears. e th en go mat aring 10 e imal e h t o er d wou l i a quantum qua Fa o q q q la ks . ne xt s w th e t t k I vneir n t u ua ua o par io b pot ge he ti t y hu s no re as ste of ion br sg n i e b ll f n , an e ec ab erials e w ” te ut o ot ns ntum t - nt nt s ner a s me o o d n i w ti man ts r i f o all t o r f f O sa tu p wa c c ur out, r a a yg o f e ir s ir st um um the the uld s.” me me fro ale s for dn nd ne b ry , be be be fo of of of in in m m i l al el e, d y y a a g s t ,