of e and g and therefore the proportion ing the population of a metastable state, New low pricing of atoms detected on those states. Set- they could infer jumps between two ting the detuning at just the right value other states.3 And in 1999 Steven Peil ensures that the second Ramsey cavity and Gerald Gabrielse trapped a single will rotate the superposition to e when- electron and watched it jump between ever a photon is present. Thus, detect- quantized cyclotron levels.4 4K ing an atom in g means no photon; de- When analyzed in theoretical detail, tecting an atom in e means one photon. jumps turn out to spring from the ap- ARS Displex Figure 2 shows the result of an ex- propriate Hamiltonian and the time- perimental run that lasted 2.5 s. Each dependent Schrödinger equation. Far Pneumatic vertical bar corresponds to the detec- from being awkwardly instantaneous or tion of one of the 2000 or so atoms sent unprovoked, jumps are an integral fea- Drive through the cavity. Most of the time, ture of orthodox quantum mechanics. barring noise, the atoms revealed an More recently, Wojciech Zurek of Cryocoolers empty cavity. But 1.05 s into the run, the Los Alamos National Laboratory ana- .8w @ 4.2K DE-210S cavity field jumped to its 1-photon lyzed quantum jumps by following the state, stayed there for 0.48 seconds, then progress of information from the sys- .25w @ 4.2K DE-204S jumped back down. tem to the apparatus and the environ- .1w @ 4.2K DE-202S At the cavity’s 0.8-K operating tem- ment. According to Zurek, that flow of perature, one expects the mirrors’ ther- information, along with basic assump- Research & OEM mal fluctuations to fill the cavity with tions of quantum theory, inevitably Cryostats one photon for about 5% of the time. breaks the unitary symmetry of Hilbert Averaging 500 runs, the ENS re- space and picks out the quantized states Optical & Non Optical searchers found a slightly higher value, that the system jumps to.5 True UHV (10 -11 torr) 6%, which they attribute to residual Brune, Haroche, Raimond, and their heating. team are now working on a new series Low Vibration Interface of experiments in which they fill the (5 nanometers) The meaning of jumps cavity with multiple photons. They Niels Bohr introduced the concept of hope to investigate another fundamen- quantum jumps in his model of the hy- tal quantum question: the transition to drogen atom. Their probabilistic na- classical behavior. ture—states suddenly changing with- Charles Day out cause—troubled some physicists. Quantum mechanics, they argued, is in- References trinsically statistical. Individual elec- 1. S. Gleyzes et al., Nature 446, 297 (2007). trons in atoms don’t jump. 2. M. Brune et al., Phys. Rev. Lett. 65, 976 In the 1980s physicists succeeded in (1990). 3. W. Nagourney, J. Sandberg, H. Dehmelt, trapping individual ions and looked for Phys. Rev. Lett. 56, 2797 (1986). electrons making jumps between states. 4. S. Peil, G. Gabrielse, Phys. Rev. Lett. 83, Warren Nagourney, Jon Sandberg, and 1287 (1999). Hans Dehmelt found the jumps in sin- 5. W. H. Zurek, http://arxiv.org/abs/quant- gle barium ions. By optically monitor- ph/0703160. Interface between nonmagnetic Compare insulators may be ferromagnetic our product specifications and conducting and our pricing! A ferromagnetic metal would be perhaps the first of many ARS novel electronic states that might emerge from the marriage manufactures of complex oxides. the complete cryocooler &

In the past few decades, re- materials may yield even richer behav- www.arscryo.com cryostat searchers have focused much attention ior than is found in bulk. Perhaps the in- on complex oxides—oxide compounds terfaces exhibit phases that don’t exist in which the electrons interact strongly in either of the constituent compounds. with one another and with the lattice. Just look at the wide variety of elec- Such compounds exhibit a fascinating tronic devices that arise from the merg- range of properties: , ing of semiconductors: One striking ex- Advanced Research ferroelectricity, high-temperature su- ample is the quantum Hall effect seen perconductivity, colossal magnetoresis- in the 2D electron gas formed at a gal- Systems, Inc. tance, and spin-glass behaviors. Not lium arsenide–aluminum gallium ar- surprisingly, these compounds are senide heterojunction. Tel 610 967 2120 being explored for myriad applications. Like semiconductors, many complex Fax 610 967 2395 e-mail; [email protected] Interfaces between complex-oxide oxides are closely lattice-matched to www.physicstoday.org June 2007 Physics Today 23 See www.pt.ims.ca/12307-14 one another and lend them- lem in different ways. The sur- selves to epitaxial growth. face may roughen as pyramids (AlO )– Only in the past decade, how- 2 e form there, so that other crystal ever, have experimenters /2 facets than the problematic (LaO)+ honed the techniques to grow e (001) plane present themselves high-quality oxide crystals and /2 to the outside. Or the ions might fabricate atomically abrupt (AlO )– rearrange themselves on the 2 e and uniform surfaces. /2 surface to compensate for the (LaO)+ One particular complex- e charge imbalance. / oxide interface has recently 0 2 What happens at the inter- shown some promise of fulfill- (TiO2 ) face when you try to grow a ing the high expectations. That positively charged plane of LaO (SrO)0 is the interface between lan- on top of the neutral TiO2 layer thanum aluminate (LaAlO3 or 0 of STO? To screen the charge LAO) and (TiO2 ) from the LaO layer, some of the

(SrTiO3 or STO). Although 0 cations might shift position, or both are insulators in their (SrO) oxygen vacancies might form at bulk form, Akira Ohtomo and the interface. It’s also possible, Harold Hwang found two Figure 1. Layered structure of the interface however, for the electrons to re- years ago that the interface be- between lanthanum aluminate and strontium arrange themselves there. For tween them was conducting. titanate, showing the alternating planes of atoms. example, electrons might be The electron carriers had a La, Al, Sr, Ti, and O atoms are shown as orange, transferred from the LaO layer 2 very high mobility, 1 m /(V-s), yellow, purple, green, and gray, respectively. to the TiO2 layer, with a net ad- with a scattering rate compara- Arrows labeled e/2 indicate the charge transfer dition of half an electron, e/2, ble to that of gallium ar- per unit cell. Some of these elec- between LaO and AlO2 layers because of ionic senide.1 Ohtomo is at Tohoku bonding. When the LaO layer is grown atop the trons might enter the previ- University and Hwang is at the ously unoccupied d shell of the originally neutral TiO2 layer, there is a net charge University of Tokyo. transfer of e/2 per unit cell to that layer. (Adapted Ti atoms, changing their va- Now it seems that the from refs. 3 and 10.) lence from 4+ to 3+ and giving LAO/STO heterojunction might them local moments because of be ferromagnetic as well as the unpaired d-shell electrons. metallic, even though neither of the important depending on growth (Ti, like most transition metals, can constituent compounds is magnetic. circumstances. have several different valences.) This This month a team from the University charge transfer is what gives rise to the of Twente in the Netherlands reported Electronic reconstruction possibility of both conductive and fer- evidence for such magnetic behavior as How can the electrons, which are local- romagnetic behavior. a large negative magnetoresistance at ized in the constituent insulators, gain To grow the LAO/STO interfaces, the LAO/STO junction and, below 0.3 mobility at the interface? A key factor is experimenters use single crystals of K, hysteresis in its sheet resistance— the electric field that is created when STO as a substrate and etch atomically that is, resistance in a 2D plane.2 you terminate a crystal flat terraces on top of that, terminating

Twente’s Alexander Brinkman ac- along a (001) plane. To understand the in a TiO2 layer. The LAO layers are de- knowledges that more evidence may be origin of this field, consider the struc- posited on this substrate by pulsed needed to nail down the existence and ture of the compounds that make up the laser deposition in a low-pressure nature of ferromagnetism. Still, ob- apparent metallic ferromagnet: STO background of oxygen. Atomic force serves Andrew Millis of Columbia Uni- and LAO. Each one is a type of oxide microscopy can confirm that the step- versity, it’s exciting just to see some type known as a , having the gen- and-terrace structure of the substrate of ferromagnetic signature at the inter- eral form ABO3, where A and B are has been maintained as the LAO film is face when none is seen in the bulk. cations. In perovskites, (001) planes of deposited.

The findings of conductivity and AO alternate with planes of BO2. As il- In their 2004 experiment, Ohtomo ferromagnetism are not without some lustrated in figure 1, the layers of STO and Hwang found that the charge car- reservations. The largest question con- are both neutral. But in LAO, positively riers at the LaO/TiO2 interface were cerns the origin of the charge and spin charged planes of LaO alternate with negative, as expected from either the carriers. Does the conductivity arise negatively charged planes of AlO2. The polar discontinuity or the presence of from an intrinsic effect, specifically lanthanum ions are sharing their elec- oxygen vacancies, and had high mobil- electronic rearrangement at the sur- trons with the aluminum ions in the ad- ity. By contrast, the reverse interface, faces? That’s the most interesting pos- jacent layers. between AlO2 and SrO layers, was in- sibility. Or does it stem more mun- If you try to terminate the LAO crys- sulating. No experiment has yet found danely from an extrinsic effect, the tal with a plane of LaO on top, an elec- any conductance at this interface, al- most likely being the appearance of tric field will point out of this plane. Na- though it is notoriously difficult to oxygen vacancies in the crystals? By al- ture won’t tolerate such a polar grow films with SrO as the top layer of tering the growth conditions, experi- discontinuity.3 For an isolated crystal of the STO crystal. menters are learning how to minimize LAO, one can in theory solve the prob- Hwang and his colleagues have also the oxygen vacancies and still see the lem by removing half an electron per found conductivity at the interface of effects of electronic rearrangements. unit cell from the top LaO layer and STO with a Mott , LaTiO3 4 Hwang asserts that most researchers adding that charge to the AlO2 layer on (LTO). Mott insulators are compounds now believe that both effects can be in the bottom of the crystal. In practice, that might be conductors except that play, with one or the other being more however, the material solves the prob- the Coulomb repulsion between elec-

24 June 2007 Physics Today www.physicstoday.org Pickett (University of California, Davis) 42 modeled the LAO/STO interface.5 Their Hey! results are consistent with the experi- mental findings of conductivity and fer- You need a low-noise amplifier? 40 romagnetism at the interface between Go to www.femto.de and find handy

LaO and TiO2 layers and with insulat- high performance signal recovery STANCE ing behavior at the AlO /SrO junction. tools, like … 38 2 Using a different approach, Satoshi Okamoto (Columbia), Millis, and … Variable-Gain Low-Noise

Ω Nicola Spaldin (University of Califor- 36

(k per unit square) Current Amplifier – DLPCA-200 SHEET RESI nia, Santa Barbara) studied the interface of STO with the Mott insulator, LTO.6 34 They predict the interface to be ferro- –4 –2 0 2 4 magnetic and conducting, with an elec- MAGNETIC FIELD (tesla) tron density distribution that agrees well with measurements.4 Figure 2. Magnetic dependence of the electrical resistance seen at an Oxygen vacancies? interface between lanthanum alumi- n Variable Gain over 9 Decades for Measuring The debate over oxygen vacancies has nate and strontium titanate. Arrows sub-pA to mA indicate data taken as the magnetic raged ever since the first report of a n Max. Bandwidth 500 kHz field is being raised or lowered. high-mobility electron gas at the n Low Input Noise down to 4 fA/Hz Measurements continued to higher LAO/STO interface. If the electric dis- n Ideal for Photodiodes and STMs fields are not shown, as suggested continuity at the interface can be largely by the dashed circle. The butterfly- screened by the electrons arising from shaped hysteresis loop suggests the oxygen vacancies, there may be no need … Wideband Voltage Amplifier – presence of ferromagnetism. for electrons to rearrange themselves. Series HVA (Adapted from ref. 2.) Before this concern arose, experi- menters had been growing interfacial films under relatively low partial pres- trons prevents them from hopping. sures of oxygen. Now, however, they In their study of magnetic effects in are increasing oxygen pressures to min- LAO/STO, the University of Twente imize the contributions of oxygen va- group, led by David Blank and Hans cancies to any observed conductance at Hilgenkamp, chose samples grown the interface. n Bandwidth DC up to 500 MHz under conditions that minimize the oc- Brinkman and his Twente collabora- n Switchable Gain up to 60 dB (x 1,000) currence of oxygen vacancies. The mag- tors found that the sheet resistance in- n For Pulse Measurements and as netoresistance and the hysteresis meas- creased by about seven orders of mag- Preamplifier for Oscilloscopes ured in the experiment are shown in nitude when the partial pressure was figure 2. Although one expects such a raised from 10−6 mbar to above 10−3 mbar. butterfly shape for ferromagnetic hys- The interface was still a conductor but … Lock-In Amplifier Module – teresis, lead author Brinkman admits not a great one. The Twente team chose Series LIA-MV-150 that he and his colleagues do not un- to look for magnetic effects in the derstand why this curve develops a higher-resistance samples to maximize local minimum just on either side of the likelihood that the behavior they ob- zero magnetic field. Moreover, they can served was due to the polar discontinu- only speculate on why the peaks in the ity and not to oxygen vacancies. hysteresis occur on the wrong side of Gertjan Koster and colleagues at zero: You’d expect to see the resistance Stanford University, the University of drop after, rather than before, the sign Twente, and North Carolina State Uni- n Manual and Remote Control of the field changes. versity point out that the intrinsic dop- n Ultra Compact Package James Eckstein of the University of ing caused by the polar discontinuity n Low Cost Illinois notes that while the shape of the cannot give rise to a charge density any n For Spectroscopy, Surface Analysis magnetoresistive hysteresis is unusual, higher than that equivalent to e/2 per and Laser Stabilization the fact that it exists suggests the unit cell.7 When Koster and his team an- switching of the direction of some mag- nealed their interfaces at increasingly netic order. This raises several ques- higher temperatures, the sheet carrier tions: Where are the oriented spins, density converged to a fixed value, are they itinerant or localized, and below that corresponding to e/2. what role might the interface play in Hwang certainly agrees that any Sophisticated Tools for Signal Recovery localization? carrier density higher than e/2 per unit Two theoretical groups have mod- cell implies another mechanism be- FEMTO® Messtechnik GmbH eled the electronic structure of complex sides the polar discontinuity. On the Paul-Lincke-Ufer 34 oxide interface with methods that ac- D-10999 Berlin/Germany other hand, he points out that if oxygen Tel.: +49-30-4 46 93 86 count for the strong interactions be- vacancies dominated, one should see Fax: +49-30-4 46 93 88 e-mail: [email protected] tween electrons. Rossitza Pentcheva some conductance at the interface www.femto.de (University of Munich) and Warren (continued on page 27) www.physicstoday.org June 2007 Physics Today 25 See www.pt.ims.ca/12307-15 9 (continued from page 25) tantalate (KTaO3 or KTO). In particular, Perhaps the new interest in oxide inter- between SrO and AlO2 planes. Instead, they looked at an interface that was ex- faces gives a brighter shine to its future. all measurements to date have shown pected to have positive carriers if the Barbara Goss Levi it to be insulating. dominant mechanism was the polar dis- Jochen Mannhart and coworkers continuity. Hall measurements indicate References from the University of Augsburg and the that the charge carriers are negative; the 1. A. Ohtomo, H. Y. Hwang, Nature 427, Pennsylvania State University, using result suggests the presence of oxygen 423 (2004). samples grown at relatively high oxygen vacancies. 2. A. Brinkman, M. Huijben, M. van Zalk, J. pressures, found that one needs a mini- Better characterization of the het- Huijben, U. Zeitler, J. C. Maan, W. G. van der Wiel, G. Rijnders, D. H. A. Blank, H. mum thickness of three unit cells of LAO erostructures can certainly help re- Hilgenkamp, Nat. Mater. (in press). deposited on the STO substrate before searchers understand what is going on 3. H. Y. Hwang, Science 313, 1895 (2006). the interface will conduct electricity.8 at the interface, and a number of groups 4. A. Ohtomo, D. A. Muller, J. L. Grazul, Such a steplike dependence of interface are working on such measurements. H. Y. Hwang, Nature 419, 378 (2002). conductance on thickness of the LAO Techniques such as scanning transmis- 5. R. Pentcheva, W. E. Pickett, Phys. Rev. B film is inconsistent with the presence of sion-electron microscopy can give clear 74, 035112 (2006). oxygen vacancies in significant numbers. pictures of the positions of the cations. 6. S. Okamoto, A. J. Millis, N. A. Spaldin, Phys. Rev. Lett. 97, 056802 (2006). Mannhart and coworkers were making a As for the oxygen vacancies, current 7. W. Siemons, G. Koster, H. Yamamoto, high-mobility transistor from the 2D techniques to detect them are indirect W. A. Harrison, G. Lucovsky, T. H. electron gas at the STO/LAO interface. and not yet capable of seeing vacancies Geballe, D. H. A. Blank, M. R. Beasley, By applying a gate voltage to an initially below a certain density threshold.10 It Phys. Rev. Lett. 98, 196802 (2007). insulating interface with just three unit would also be nice to measure the 8. S. Thiel, G. Hammerl, A. Schmehl, C. W. cells of LAO, they could switch to a con- charge distribution around the inter- Schneider, J. Mannhart, Science 313, 1942 ducting state and back again. face and, perhaps, to determine where (2006). See also M. Huijben et al., Nat. For a different slant on the oxygen va- the spins are. Are they indeed associ- Mater. 5, 556 (2006). 3+ 9. A. Kalabukhov, R. Gunnarsson, T. Clae- cancy question, Alex Kalabukhov and his ated with the Ti ions? son, D. Winkler, http://arxiv.org/abs/ coworkers at Chalmers University of Strontium titanate was once touted as 0704.1050. Technology in Göteborg, Sweden, stud- a substitute for diamond in jewelry, but it 10. N. Nakagawa, H. Y. Hwang, D. A. ied the interface of STO with potassium eventually lost out to other competitors. Muller, Nat. Mater. 5, 204 (2006).

Supplementary material that can arise when light related to these items impinges on a metal par- can be found at ticle whose size is on the www.physicstoday.org. order of the light’s wave- length. Strong scattering Observing electron tunneling in real time. To escape an atom, occurs, and the SPs cou- an electron usually has to gain enough energy to climb out of ple to the waveguide the atom’s Coulomb well. However, if the atom is bathed in an modes of Si; the light is intense laser pulse, the pulse’s strong electric field can lower the effectively trapped. The confining potential enough to allow the electron to tunnel to free- researchers used this phenomenon in a PV solar cell to increase dom. That hypothesis, proposed in 1965 by Leonid Keldysh, has the cell’s absorption efficiency. To create the particles, they now been verified, thanks to physicists’ recent ability to control deposited thicknesses of silver that ranged from 10 nm to more light on the time scale of electronic motion in atoms: atto- than 20 nm; during annealing, the silver coalesced into islands seconds. Matthias Uiberacker of the Ludwig Maximilians Uni- larger than 100 nm across. For 1.25-μm thin-film PV cells the versity of Munich, Germany, and his collaborators blasted a gas group found a 16-fold absorption enhancement for light at of neon atoms with an attosecond UV pulse followed by an 1050 nm, where Si normally absorbs poorly, and a 33% intense femtosecond red pulse. The attosecond pulse ripped off increase across all wavelengths of light (see the figure). For an inner electron and excited a second electron to the atom’s 300-μm-thick wafer PV cells, plasmon trapping increased full- periphery. Then the red pulse, consisting of just a few wave spectrum absorption by 19%. According to Supriya Pillai, opti- cycles with precisely controlled phase, could free the outlying mizing the nanoparticle size should bring additional improve- electron by lowering the confining potential. When the three ments. The scientists also used plasmons to increase emission most intense peaks at the center of the red pulse coursed through from thin-film LEDs. (S. Pillai et al., J. Appl. Phys. 101, the atoms, the tunneling probability neared 100%, and the elec- 093105, 2007.) —PFS trons escaped in three steps. Each step took less than 400 atto- seconds. (M. Uiberacker et al., Nature 446, 627, 2007.) —BPS Magnetic resonance force microscopy has reached 90-nm reso- lution. MRFM maps the spins in a sample that is mounted on an Plasmon-assisted solar cells. Because of its ubiquity in electron- ultrasensitive cantilever hanging vertically over a sharp ics, silicon is the favorite semiconductor used in solar photo- magnetic tip. If the tip’s magnetic field is highly inhomogeneous, voltaic cells. But Si is a poor light emitter and absorber, and the an applied radio-frequency field will resonate with the spins in a efficiency of thin-film Si PV cells is even poorer than that of highly localized region of the sample, and the cantilever will wafer-based thick cells. An important goal is to make the cells deflect due to the resulting forces. To achieve the nanoscale reso- inexpensive by using thin films, but also to make them nicely lution, John Mamin, Dan Rugar, and their colleagues at the IBM absorptive. Scientists at the University of New South Wales in Almaden Research Center in San Jose, California, made tips with Australia have now enhanced the absorption of sunlight using magnetic gradients of more than a million teslas per meter and surface plasmons generated on silver nanoparticles deposited on held the samples about 45 nm away. The test objects being PV cells. SPs are collective oscillations of conduction electrons imaged consisted of tiny islands of calcium fluoride evaporated www.physicstoday.org June 2007 Physics Today 27