Superconductivity in one-atomic-layer metal films grown on Si(111)

Transcription

1 LETTERS PUBLISHED ONLINE: 10 JANUARY 2010 DOI: /NPHYS1499 Superconductivity in one-atomic-layer metal films grown on Si(111) Tong Zhang 1,2, Peng Cheng 1, Wen-Juan Li 1,2, Yu-Jie Sun 1, Guang Wang 1, Xie-Gang Zhu 1, Ke He 2, Lili Wang 2, Xucun Ma 2, Xi Chen 1 *, Yayu Wang 1, Ying Liu 3, Hai-Qing Lin 4, Jin-Feng Jia 1 and Qi-Kun Xue 1,2 * The two-dimensional (2D) superconducting state is a fragile state of matter susceptible to quantum phase fluctuations. Although superconductivity has been observed in ultrathin metal films down to a few layers 1 10, it is still not known whether a single layer of ordered metal atoms, which represents the ultimate 2D limit of a crystalline film, could be superconducting. Here we report scanning tunnelling microscopy measurements on single atomic layers of Pb and In grown epitaxially on Si(111) substrate, and demonstrate unambiguously that superconductivity does exist at such a 2D extreme. The film shows a superconducting transition temperature of 1.83 K for an atom areal density n = Pb atoms nm 2, 1.52 K for n = 9.40 Pb atoms nm 2 and 3.18 K for n = 9.40 In atoms nm 2, respectively. We confirm the occurrence of superconductivity by the presence of superconducting vortices under magnetic field. In situ angle-resolved photoemission spectroscopy measurements reveal that the observed superconductivity is due to the interplay between the Pb Pb (In In) metallic and the Pb Si (In Si) covalent bondings. The one-atomic-layer films of Pb and In studied here were grown with atomic precision on bulk-terminated Si(111) substrate using molecular beam epitaxy. The one-atomic-layer films of Pb have two different structural phases depending on the coverage (for sample preparation, see the Methods section). Figure 1a,d shows the schematic structure and scanning tunnelling microscopy (STM) topograph of the so-called striped incommensurate (SIC) phase, which has a Pb coverage of 4/3 monolayers (ML; refs 11 14). Here 1 ML is defined as the surface atomic density of the Si(111) with areal density n=7.84 atoms nm 2. In a unit cell of the SIC-Pb phase, there are four Pb atoms per three surface Si atoms. Three of the four Pb atoms each form a covalent bond with an underlying Si atom, leaving one Pb atom without bonding to the Si substrate. Besides the covalent bonds with the Si substrate, the metal atoms also form metallic bonds within the metal overlayer. As all Pb atoms are located exactly in the same atomic-layer sheet (see the large-scale STM image and cross-section height profiles in Supplementary Fig. S1), the resulting areal density of Pb atoms is nm 2. Compared with the bulk Pb(111) plane, the lattice of the SIC phase is compressed by 5%. Ultralow-temperature (down to 0.40 K) scanning tunnelling spectroscopy (STS) on the SIC phase reveals a clear signature of superconductivity. Figure 2a shows the tunnelling spectra taken on the SIC phase using a superconducting Nb tip. At 0.42 K, the differential conductance shows a zero conductance region near the Fermi level (E F ) and two sharp peaks at bias voltages V ±1.8 mev. This is the classical hallmark of a superconductor insulator superconductor (SIS) tunnelling junction, which has a zero quasiparticle density of states (DOS) at E F and two Bardeen Cooper Schrieffer (BCS)-like DOS peaks at V = ±( tip + SIC )/e. Here tip and SIC are the superconducting gaps of the Nb tip and the SIC phase, respectively; e is the electron charge. Starting from 1.05 K, two extra peaks at V = ±( tip SIC )/e become evident. This is caused by thermally excited holes/electrons 15. The observation also demonstrates the high energy resolution of our STS measurements 16,17. With increasing temperature, the spectra are broadened and the coherence peaks are suppressed. The energy gap SIC becomes nearly invisible at 1.82 K, suggesting that the superconducting transition occurs around this temperature. The superconducting characteristics are further demonstrated by using a non-superconducting PtIr tip (Supplementary Fig. S2). Using a superconducting gap tip = 1.46 mev (T C = 9 K) for the Nb tip, which was calibrated by a separate measurement on Ag islands (Supplementary Fig. S3), we obtain the temperature evolution of SIC, as shown in Fig. 2b. Fitting the data using the BCS gap function 18 yields SIC (0)=0.35 mev, T C =1.83 K and the BCS ratio 2 SIC /k B T C = 4.4 (k B is the Boltzmann constant). The BCS ratio is very close to the value of 4.3 in bulk Pb, suggesting that the SIC phase is a strongly coupled BCS superconductor. Another one-atomic-layer film of Pb has the same fundamental 7 3 building block as SIC but a smaller coverage of 6/5 ML (refs 12 14, 19 21). Hereafter this is called the 7 3-Pb phase (Fig. 1b,e). In a unit cell of the 7 3-Pb phase, there are six Pb atoms per five surface Si atoms. Five of the six Pb atoms each form a covalent bond with an underlying Si atom, leaving one Pb atom without bonding to the Si substrate. The areal atom density is nearly the same as that of the bulk Pb(111) plane, with a slight lattice expansion of 0.1%. Figure 2c shows the tunnelling spectra of the SIS junction in the 7 3-Pb phase, which is qualitatively the same as that of the SIC phase, except for a reduction of the energy gap. The superconducting gap, T C and the 2 /k B T C ratio deduced from the BCS fitting of the experiment are 0.27 mev, 1.52 K and 4.12, respectively (Fig. 2d). Indium atoms grow on Si(111) with the same 7 3 structure as that of Pb (refs 22, 23) (Fig. 1c,f). The In coverage is 6/5 ML, and the areal atom density is 9.40 nm 2. Thus, its lattice is 0.3% expanded compared with that of the bulk In(001) plane. Figure 2e 1 Key Lab for Atomic and Molecular Nanoscience, Department of Physics, Tsinghua University, Beijing , China, 2 Institute of Physics, Chinese Academy of Sciences, Beijing , China, 3 Department of Physics and Material Research Institute, The Pennsylvania State University, Pennsylvania 16802, USA, 4 Department of Physics, The Chinese University of Hong Kong, Hong Kong, China. * xc@mail.tsinghua.edu.cn; qkxue@mail.tsinghua.edu.cn. 104 NATURE PHYSICS VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved.

2 NATURE PHYSICS DOI: /NPHYS1499 a LETTERS b c a First layer Si Pb First layer Si Pb Second layer Si d First layer Si In Second layer Si e Second layer Si f 2 nm 2 nm 2 nm 11 14,19,22 (a c) and high-resolution STM images Figure 1 One-atomic-layer metal superconducting films of Pb and In. a f, Schematic structure models (d f) of the SIC-Pb (a,d), 7 3-Pb (b,e) and 7 3-In (c,f) phases grown on Si(111) substrate. Imaging conditions are V = 0.1 V and I = 0.05 na. c 3.10 K 1.78 K 1.43 K 3.07 K 1.70 K 1.41 K 3.01 K 1.58 K 1.38 K 2.81 K 1.50 K 1.34 K 2.61 K 1.42 K 1.30 K 1.15 K di/dv (a.u.) 1.46 K 1.31 K 1.25 K 1.21 K 2.36 K 2.07 K 1.75 K 1.10 K 1.51 K 0.82 K 1.01 K 1.21 K 0.65 K 0.84 K 0.81 K 0.42 K 0.42 K 0.41 K Bias (mv) Bias (mv) d Bias (mv) f 0.25 SIC Pb Data BCS theory Tc = 1.83 K T (K) Δ (mv) Δ (mv) 3.14 K 1.82 K 1.05 K b e 1.49 K di/dv (a.u.) di/dv (a.u.) 2.00 K Pb Data BCS theory Tc = 1.52 K T (K) Δ (mv) a In Data BCS theory Tc = 3.18 K T (K) Figure 2 Superconductivity in one-atomic-layer metal films. a, The differential conductance of single-particle tunnelling measured with a Nb tip on the SIC-Pb phase as a function of temperature. The tunnelling junction (also applies to c and e) was set at V = 0.01 mv and I = 0.2 na. The curves at different temperatures are offset vertically for clarity. b, The superconducting gap of the SIC-Pb phase as a function of temperature. The open circles show the measured gap, the fitting by the BCS gap function (see the text). c f, The tunnelling conductance (c,e) and superconducting gap and the solid curve shows (d,f) for the 7 3-Pb (c,d) and 7 3-In (e,f) phases. shows the tunnelling spectra of the 7 3-In phase. The values of, TC and 2 /kb TC extracted from the BCS fit (Supplementary Fig. S4) are 0.57 mev, 3.18 K and 4.16, respectively (Fig. 2f). Unlike the Pb films, where C is strongly suppressed compared with the T bulk value, the 7 3-In phase has a surprisingly high TC close to the bulk value 3.4 K, as predicted previously23. More significantly, NATURE PHYSICS VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved. 105

3 LETTERS NATURE PHYSICS DOI: /NPHYS1499 a 50 nm b 50 nm e c 0.03 T 0.06 T d 50 nm 50 nm Normalized ZBC T 0.12 T Magnetic field (kg) Figure 3 Superconducting vortices in the SIC-Pb phase. a d, The vortices at different magnetic fields ( T) measured at 0.42 K using a PtIr tip. In each vortex image, the superconducting regions of low conductance are shown in maroon colour and the gapless normal state inside the vortex cores are yellow coloured. The tunnelling junction was set at V = 0.01 mv and I = 0.1 na. The bias voltage was ramped to 0 V with feedback off at each pixel to record the ZBC. e, Normalized ZBC measured at locations between two vortices under different magnetic fields. The upper critical field of the SIC phase is estimated to be 1450 G. the enhanced 2 /k B T C ratio compared with the bulk (3.6) implies that it is transformed into a strongly coupled superconductor by reducing the dimension. The occurrence of superconductivity in the one-atomic-layer films is further confirmed by the presence of superconducting vortices under magnetic field. The magnetic measurement on the SIC phase is shown in Fig. 3 and Supplementary Fig. S5. As seen in the zero bias conductance (ZBC) spectra mapping, the vortices start to form at a field of around 0.03 T (Fig. 3a), and increase in density with the field (Fig. 3b,c). The vortex image becomes severely blurred at B 0.12 T (Fig. 3d), suggesting that the superconducting state is almost completely suppressed around this field. From a systematic field-dependence study of ZBC (ref. 24), an upper critical field H C2 of 1,450 G is obtained (Fig. 3e). In a type II superconductor, H C2 (T) is determined by the coherence length ξ GL through H C2 (T) = Φ 0 /2πξGL 2 (T), where Φ 0 is the magnetic flux quanta. The 1,450 G upper critical field renders a coherence length ξ GL = 49 nm, which tells us the size of the Cooper pair and vortex core in this phase. To understand the mechanism of the superconductivity in these one-atomic-layer superconductors, we have carried out highresolution angle-resolved photoemission spectroscopy (ARPES) measurements 3,25. The electronic structure revealed by ARPES shows that despite the extreme 2D geometry and the coupling with the Si substrate, the one-atomic-layer films show well-developed 2D free-electron-like metallic bands 21,23. Figure 4a shows an ARPES spectrum of the SIC phase along the Ɣ M( Ɣ K M ) direction at T =30 K. Ɣ, M, K and Ɣ, M, K denote the symmetry points of 1 1 and 3 3 surface Brillouin zones (SBZs) of Si(111), respectively. Two parabolic bands (red solid lines in Fig. 4a) crossing E F are centred at 0.62 and 1.26 Å 1 (the K points), which are attributed to the surface states of SIC because they are situated above the edge of Si bulk bands 21. The ARPES data along the Ɣ K( Ɣ M Ɣ ) direction (not shown) reveal other parabolic surface bands with larger Fermi circles centred on Ɣ and K( Ɣ ). Therefore, the surface states of SIC reflect the 3 3 periodicity with a large electron pocket at Ɣ and a small one at K (Fig. 4b). The total sheet carrier density estimated by the sizes of the two kinds of electron pocket is about cm 2. The ARPES data for 7 3-Pb are shown in Fig. 4c,d, and agree with the previous study 21. Despite a different Fermi-surface shape, its electronic structure (as well as that of the 7 3-In; ref. 23) could also be understood on the basis of a 2D free-electron-like band. The Si substrates used for the growth of the one-atomiclayer films shown above are heavily doped with boron (hole concentration cm 3 ). To clarify the possible role of the dopants on superconductivity, we have studied the SIC-Pb phases prepared on both heavily p-type ( cm 3 ) and normally n-type ( cm 3 ) doped Si substrates. The SIS spectra of the two films are shown together in Supplementary Fig. S6. The nearly identical results exclude any significant role of the dopants in Si, and suggest that the electronic property of the metal overlayer is dominated only by the intra-overlay metallic and the Pb Si interface covalent bondings. In this sense, the superconducting films seem to be electronically isolated from the underlying Si lattice. Thus, only the metal overlayer and the metal Si interface are responsible for the occurrence of superconductivity. We now discuss the possible roles of the Si substrate. Through the formation of covalent bonds at the interface 11 14,19 23, the Si substrate serves as a structural support stabilizing the crystal structure of the metal overlay. This is important because freestanding monolayer films of metals such as Pb or In are unstable, unlike graphene. To investigate whether Si takes part in the electron pairing, we have carried out a variable-temperature ARPES measurement where the dimensionless electron phonon coupling parameter λ can be evaluated by the temperaturedependent photoemission spectra 26. The result for the SIC phase is λ = 1.07 ± 0.13 (for details, see Supplementary Fig. S7), which is greatly enhanced compared with the λ < 0.9 of the much thicker Pb films 25,27, apparently owing to the presence of the metal Si interface. The strongly enhanced λ ( 1) in the 7 3-In film 23 is particularly significant, because now its surprisingly high T C 3.18 K can be explained by the strong-coupling McMillan formula 28. The similar λ values in the Pb and In films suggest that the phonon modes from interface bonding must contribute to electron phonon coupling in a significant way NATURE PHYSICS VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved.

4 NATURE PHYSICS DOI: /NPHYS1499 LETTERS a M Γ b E F Binding energy (ev) Γ K M M K x 1 SBZ SIC SBZ Wavevector (Å 1 ) c M Γ d E F 0.2 K Binding energy (ev) Γ M x 1 SBZ 7 x 3 SBZ Wavevector (Å 1 ) Figure 4 The ARPES spectra of one-atomic-layer Pb films on Si. a, The energy band measured along the Ɣ M direction for the SIC phase on p-type Si at 30 K. The photon energy was ev (He Iα). Free-electron-like bands (indicated by parabola) are observed around k = 0.6 and 1.26 Å 1. The parabolic feature centred at the Ɣ point near E F and dispersing to higher binding energy is the edge of the Si bulk valence band. b, The SBZ of the SIC-Pb phase. c, The energy band measured along the Ɣ M direction for the 7 3-Pb phase on p-type Si at 30 K. d, The SBZ of the 7 3-Pb phase. As for the intra-overlay metallic bonding, the development of nearly free-electron-like metallic bands seen in Fig. 4 implies that the supercurrent should be carried by the carriers in the metal overlay. The difference between SIC and 7 3-Pb further highlights the crucial role of the overlayer structure. The denser packing in the SIC phase leads to a stronger superconductivity, which remains an interesting subject to be further investigated. These observations suggest a new route towards 2D superconductivity: the metal overlay serves as a charge reservoir and the electron phonon interactions that glue the electrons to form pairs are provided both by the intralayer metallic and more importantly the interface bonds. The situation is rather different from that of the 2 ML Pb film superconductivity reported recently 9, which is essentially a property of the film itself with the same mechanism of superconductivity as that of thicker films. Methods To investigate the electronic structure and superconductivity of the one-atomic-layer films, we have used two ultrahigh-vacuum systems, both equipped with molecular beam epitaxy for film growth. The first system features an in situ ultralow-temperature STM system (Unisoku and RHK) operated at a base temperature of 0.40 K and under a magnetic field up to 11 T (refs 16, 17). The energy resolution is better than 0.1 mev at 0.40 K when a Nb tip is used 16. This proves critical because the superconducting gap of the samples is well below 1 mev. The second system features in situ low-temperature STM (Omicron) and high-energy-resolution (10 mev) ARPES with a Scienta R4000 analyser 3,25. To prepare high-quality samples, we used pure Pb ( %) and In ( %), which were respectively evaporated from two Knudsen cells for molecular beam epitaxy (Omicron). The clean Si(111)-7 7 surface was used as substrate for preparing the one-atomic-layer films of Pb and In. The Si(111)-7 7 surface was prepared by cycles of annealing and flashing up to 1,480 K of the as-supplied Si wafer in the ultrahigh-vacuum chamber. The deposition rate of Pb or In and thus the metal coverage and film thickness was carefully calibrated by the Pb (In)-induced 3 3 phase on the clean Si(111) surface, which is known to have a precise surface coverage of 1/3 ML. The Pb-SIC phase was prepared by depositing 1.5 ML Pb atoms on Si(111)-7 7 surface at room temperature and subsequently annealing at 550 K for 30 s. The deposition rate of Pb/In was 0.1 ML min 1. The 7 3-Pb phase was prepared by annealing the SIC phase at 550 K for two minutes to reduce the Pb coverage. The 7 3-In phase was prepared by depositing 1.5 ML In atoms on Si(111)-7 7 surface at room temperature and subsequently annealing at 700 K for 4 min. To ensure that the same phase is studied in two separate systems, in each experimental run the structure of the film was first checked by the atomically resolved STM images (Fig. 1d f), followed by STS measurement of the characteristic electronic states near E F (Supplementary Fig. S1d,e). Received 29 June 2009; accepted 26 November 2009; published online 10 January 2010 References 1. Shal nikov, A. I. Superconducting thin films. Nature 142, 74 (1938). 2. Haviland, D. B., Liu, Y. & Goldman, A. M. Onset of superconductivity in the two-dimensional limit. Phys. Rev. Lett. 62, (1989). 3. Guo, Y. et al. Superconductivity modulated by quantum size effects. Science 306, (2004). 4. Özer, M. M., Thompson, J. R. & Weitering, H. H. Hard superconductivity of a soft metal in the quantum regime. Nature Phys. 2, (2006). 5. Eom, D., Qin, S., Chou, M.-Y. & Shih, C. K. Persistent superconductivity in ultrathin Pb films: a scanning tunnelling spectroscopy study. Phys. Rev. Lett. 96, (2006). 6. Nishio, T., Ono, M., Eguchi, T., Sakata, H. & Hasegawa, Y. Superconductivity of nanometer-size Pb islands studied by low-temperature scanning tunnelling microscopy. Appl. Phys. Lett. 88, (2006). NATURE PHYSICS VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved.

5 LETTERS 7. Shanenko, A. A., Croitoru, M. D. & Peeters, F. M. Oscillations of the superconducting temperature induced by quantum well states in thin metallic films: Numerical solution of the Bogoliubov de Gennes equations. Phys. Rev. B 75, (2007). 8. Nishio, T. et al. Superconducting Pb island nanostructures studied by scanning tunnelling microscopy and spectroscopy. Phys. Rev. Lett. 101, (2008). 9. Qin, S. Y., Kim, J. D., Niu, Q. & Shih, C. K. Superconductivity at the two-dimensional limit. Science 324, (2009). 10. Brun, C. et al. Reduction of the superconducting gap of ultrathin Pb islands grown on Si(111). Phys. Rev. Lett. 102, (2009). 11. Seehofer, L., Falkenberg, G., Daboul, D. & Johnson, R. L. Structural study of the close-packed two-dimensional phases of Pb on Ge(111) and Si(111). Phys. Rev. B 51, (1995). 12. Horikoshi, K., Tong, X., Nagao, T. & Hasegawa, S. Structural phase transitions of Pb-adsorbed Si(111) surfaces at low temperatures. Phys. Rev. B 60, (1999). 13. Kumpf, C. et al. Structural study of the commensurate incommensurate low-temperature phase transition of Pb on Si(111). Surf. Sci. 448, L213 L219 (2000). 14. Chan, T. L. et al. First-principles studies of structures and stabilities of Pb/Si(111). Phys. Rev. B 68, (2003). 15. Giaever, I. Electron tunnelling between two superconductors. Phys. Rev. Lett. 5, (1960). 16. Ji, S. H. et al. High-resolution scanning tunnelling spectroscopy of magnetic impurity induced bound states in the superconducting gap of Pb thin films. Phys. Rev. Lett. 100, (2008). 17. Chen, X. et al. Probing superexchange interaction in molecular magnets by spin-flip spectroscopy and microscopy. Phys. Rev. Lett. 101, (2008). 18. Bardeen, J., Cooper, L. N. & Schrieffer, J. R. Theory of superconductivity. Phys. Rev. 108, (1957). 19. Brochard, S. et al. Ab initio calculations and scanning tunnelling microscopy experiments of the Si(111)- 7 3-Pb surface. Phys. Rev. B 66, (2002). 20. Hupalo, M., Schmalian, J. & Tringides, M. C. Devil staircase in Pb/Si(111) ordered phases. Phys. Rev. Lett. 90, (2003). NATURE PHYSICS DOI: /NPHYS Choi, W. H., Koh, H., Rotenberg, E. & Yeom, H. W. Electronic structure of dense Pb overlayers on Si(111) investigated using angle-resolved photoemission. Phys. Rev. B 75, (2007). 22. Kraft, J., Surnev, S. L. & Netzer, F. P. The structure of the indium-si(111) ( 7 3) monolayer surface. Surf. Sci. 340, (1995). 23. Rotenberg, E. et al. Indium 7 3 on Si(111): A nearly free electron metal in two dimensions. Phys. Rev. Lett. 91, (2003). 24. Bergeal, N. et al. Scanning tunnelling spectroscopy on the novel superconductor CaC 6. Phys. Rev. Lett. 97, (2006). 25. Zhang, Y. F. et al. Band structure and oscillatory electron phonon coupling of Pb thin films determined by atomic-layer-resolved quantum-well states. Phys. Rev. Lett. 95, (2005). 26. McDougall, B. A., Balasubramanian, T. & Jensen, E. Phonon contribution to quasiparticle lifetimes in Cu measured by angle-resolved photoemission. Phys. Rev. B 51, (1995). 27. Luh, D.-A., Miller, T., Paggel, J. J. & Chiang, T.-C. Large electron phonon coupling at an interface. Phys. Rev. Lett. 88, (2002). 28. McMillan, W. L. Transition temperature of strong-coupled superconductors. Phys. Rev. 167, (1968). Acknowledgements The work at Tsinghua and IOP was supported by NSFC and MOST of China. Y.L. was supported in part by DOE under Grant DE-FG02-04ER46159 and by NSFC under Grant No Author contributions T.Z., P.C., W.-J.L., Y.-J.S., G.W. and X.-G.Z. carried out the experiments; K.H., L.W., X.M., Y.W., Y.L. and H.-Q.L. analysed the data; X.C., J.-F.J. and Q.-K.X. designed and coordinated the experiments; Q.-K.X. wrote the paper. All authors discussed the results and commented on the manuscript. Additional information The authors declare no competing financial interests. Supplementary information accompanies this paper on Reprints and permissions information is available online at Correspondence and requests for materials should be addressed to X.C. or Q.-K.X. 108 NATURE PHYSICS VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved.

6 SUPPLEMENTARY INFORMATION doi: /nPHYS1499 Superconductivity in one-atomic-layer metal films grown on Si(111) substrate Tong Zhang 1, 2, Peng Cheng 1, Wen-Juan Li 1,2, Yu-Jie Sun 1, Guang Wang 1, Xie-Gang Zhu 1, Ke He 2, Lili Wang 2, Xucun Ma 2, Xi Chen 1, Yayu Wang 1, Ying Liu 3, Hai-Qing Lin 4, Jin-Feng Jia 1, and Qi-Kun Xue 1, 2 1 Department of Physics, Tsinghua University, Beijing , P. R. China 2 Institute of Physics, Chinese Academy of Sciences, Beijing , P. R. China 3 Department of Physics and Material Research Institute, The Pennsylvania State University, Pennsylvania 16802, USA 4 Department of Physics, The Chinese University of Hong Kong, Hong Kong, P. R. China Corresponding authors: Q.K.X. (qkxue@mail.tsinghua.edu.cn) and X.C. (xc@mail.tsinghua.edu.cn). This file contains Supplementary Figures 1-7 with Legends. nature physics 1

7 supplementary information doi: /nphys1499 Supplementary Figure S1. The large scale STM images and STS spectra of the one-atomic-layer Pb and In films. a, The SIC-Pb phase. b, The 7 3 -Pb phase. c, The 7 3 -In phase. d, The STS of SIC-Pb and 7 3 -Pb from -1.5 V to V. e, The STS of 7 3 -In from 0 V to +2 V. The cross-section height profiles taken along the line marked in the STM images reveal that all three films are atomically flat, well-ordered, and have a thickness of one-atomic layer. The distinct peaks in d and e show the characteristic electronic structure near the Fermi level of the films. 2 nature physics

8 doi: /nphys1499 supplementary information Supplementary Figure S2. The differential conductance of the Pb films using nonsuperconducting PtIr tips. a, The STS spectra of the SIC-Pb (pink curve) and 7 3 -Pb phase (blue curve) grown on p-type Si. For comparison, the data of 20 ML Pb film (black curve) using a nonsuperconducting PtIr tip is also shown. The measurements were conducted at T=0.42 K. The tunnelling junction was set at V=0.01 mv, I=0.1 na. b, Temperature-dependent STS spectra of the SIC-Pb phase in a using a PtIr tip. c, BCS fitting (red curve) of the superconducting gap (0.31 mev) of the SIC-Pb phase using the spectra in b at T=0.42 K (blue curve). The gap is smaller than 0.35 mev determined using the STS spectra in Fig. 1a with an Nb tip. In fitting, a broadening factor (Dyne s lifetime papameter) of 0.03 mev was used. nature physics 3

9 supplementary information doi: /nphys1499 Supplementary Figure S3. The tunnelling conductance on an Ag island on Si(111) acquired with the same Nb tip used for acquiring Fig. 2a. a, Temperature-dependent STS from 0.42 K to 4.21 K. b, The density of states for the superconductive tip and the gap (1.46 mev) were determined by fitting the spectra at T=0.42 K (blue dots) with BCS theory (pink curve). Because the energy gap Δ of the tip changes very little (<0.01 mev) particularly when T<3.5 K, we consider it as a constant when fitting the experimental data. 4 nature physics

10 doi: /nphys1499 supplementary information Supplementary Figure S4. BCS fitting of the energy gap Δ of the 7 3 -In phase at different temperature. a, T=0.41 K. b, T=1.75 K. c, T=2.61 K. d, T=3.10 K. The spectra were acquired using an Nb tip. In the BCS fitting, we used a constant energy gap (Δ=1.46 mev) for the Nb tip, which was determined in Fig. S3. Γ is a fitting parameter for finite quasiparticle lifetime as described in Dynes, R. C., Narayanamurti, V. & Garno, J. P. Phys. Rev. Lett. 41, 1509 (1978). The experimental data are indicated by the blue dots, while the fitting ones by the pink curves. nature physics 5

11 supplementary information doi: /nphys1499 Supplementary Figure S5. The coherent length of the SIC-Pb phase. a, The STS at various locations around a vortex acquired with a PtIr tip. b, Spatial mapping of ZBC around a vortex. c, The fitting of ZBC as a function of distance from the centre of the vortex core using Ginzburg-Landau formula σ ( r,0) = σ0 + (1 σ0) (1 tanh( r/( 2 ξ))) 24 where σ 0 is the normalized ZBC far from an isolated vortex and r is the distance to the vortex core. The coherent length is given by the full width at half minimum of the vortex profile curve. Fitting the ZBC profile over a vortex yields ξ GL = 45 nm at T = 0.42 K, in good agreement with the value estimated from H C2. 6 nature physics

12 doi: /nphys1499 supplementary information Supplementary Figure S6. The differential conductance of the SIC-Pb phase grown on n-type (pink curve) and p-type (blue curve) Si. The measurement condition is the same as that used in Fig. 2a. nature physics 7

13 supplementary information doi: /nphys1499 Supplementary Figure S7. The electron-phonon coupling measured by variable temperature photoemission. a, The temperature-dependent photoemission spectra (78 K-200 K) of the SIC-Pb phase. b, The peak width of band of the SIC phase (red dots), as well as that of the 4ML Pb film (blue dots) for comparison, as a function of temperature. The linewidth (ΔE), deduced from Lorentzian lineshape fitting, increases linearly with temperature due to the shorter lifetime of quasi-particles. The electron-phonon coupling parameter λ was evaluated by fitting the temperature-dependent linewidth using 1 dδe λ = 25-27, where k B is π k dt 2 B the Boltzmann constant. To avoid the strong background from the valence band of Si, the photoemission spectra was taken at k // = 0.38 Å nature physics

14 news & views the optical and the effective centrifugal potential and are insensitive to Casimir and van der Waals forces, which are important in the case of atoms on a chip 8. As these neutrons feel the wall of the whispering gallery over a long distance and a long time span they are sensitive to any roughness and impurity effects of the wall, which may be used for precision measurements. This also extends the field of neutron spectroscopy into the nanoelectronvolt region and demonstrates how the classical behaviour of neutrons in a neutron guide tube with a typical radius of 100 m changes to quantum behaviour in a whispering-gallery geometry in which the radius is 2.5 cm. The massive neutron now follows the rules of quantum optics. Helmut Rauch is at the Atominstitut, TU-Wien, Stadionallee 2, 1020 Vienna, Austria, and at the Institut Laue-Langevin, BP 156, F Grenoble, France. rauch@ati.ac.at References 1. Rayleigh, I. Phil. Mag. 27, (1914). 2. Oraevsky, A. N. Quantum Electron. 32, (2002). 3. Nesvizhevsky, V. V., Voronin, A. Y., Cubitt, R. & Protosov, K. V. Nature Phys. 6, (2010). 4. Maier-Leibnitz, H. & Springer, T. J. Nucl. Energy 17, (1963). 5. Nesvizhevsky, V. V. et al. Nature 415, (2002). 6. Della Valle, G. et al. Phys. Rev. Lett. 102, (2009). 7. Folman, R., Krüger, P., Schmiedmayer, J., Denschlag, J. & Henkel, C. Adv. Atom. Mol. Opt. Phys. 48, (2002). 8. Rauch, H., Lemmel, H., Baron, M. & Loidl, R. Nature 417, (2002). ultrathin films the thinnest superconductor How thin can superconducting materials be and still retain their superconductivity? A recent study of thin films grown on silicon substrates reveals that one atomic layer is the limit. Yukio Hasegawa the electrical resistance of some types of metal disappears completely when they are cooled below certain temperatures. This peculiar phenomenon, called superconductivity, is one of the hottest topics in physics. One of the issues discussed is the size and dimension of superconductors: superconductivity is suppressed, for instance, when the film thickness is thinner than the size of the electron pairs that form the superconducting state. In the case of ultrathin films, in which electrons are mobile only in the planar direction, thermal and/or quantum fluctuations may disturb the coherent motion of the electron pairs and break the superconductivity. Under these conditions, the question to be answered is how thin can the films be and still retain superconductivity? Lead (Pb), an elemental superconductor whose superconducting transition temperature is 7.2 K, is known to form high-quality thin films on a clean silicon wafer, which does not show superconductivity in ambient conditions, owing to good crystalline and lattice matching between the two materials. A research group led by Qi-Kun Xue 1 first successfully grew millimetre-sized thin films of Pb, which were uniform in thickness with atomic-layer precision, and observed curious even odd atomiclayer dependence of the superconducting transition temperature, which is caused by the confinement of electronic waves in the thin films. Then Shengyong Qin et al. 2 were able to create Pb films just two-atomiclayers thick and found the ultrathin film to be still superconducting. It was thought that this was the extreme limit because the two-atomic-layer thin film is the thinnest film in which electrons can still keep their three-dimensional nature that is, remain mobile both in the planar and thickness directions. This record did not last long, however. Writing in Nature Physics, Zhang et al. 3 now report that one atomic layer of Pb or indium (In) is enough to retain superconductivity. The authors used scanning tunnelling microscopy (STM) to verify the superconductivity of the ultrathin films formed on a clean silicon wafer. Although STM is often used for recording atomically resolved images of sample surfaces using a sharp needle scanning over the surface, this function did not play a significant part in this study. Owing to the electron pairing, a superconductor has an energy gap in the electronic energy levels, whose separation corresponds to the energy required to break the pair. By measuring the energy level sequence using a spectroscopic mode of STM known as scanning tunnelling spectroscopy (STS), Zhang et al. detected the superconducting gap on the oneatomic-layer thin films, measured their temperature dependence, and found the dependence to be well explained by the Bardeen Cooper Schrieffer theory, which is an established model for describing the superconductivity of Pb and In. The response of the thin films to magnetic fields provides further evidence of superconductivity. Essentially, a magnetic field tends to break apart the superconducting electron pairs, and the superconductor repels the magnetic field. For instance, magnetic fields applied on superconducting bulk Pb cannot penetrate into the bulk until the field becomes strong enough for its energy to overcome the superconductor s condensation energy and destroy it. In the case of a thin film, however, a perpendicularly applied magnetic field penetrates into the superconductor in a quantized manner; bundled magnetic fluxes whose number is fixed enter one by one as the magnetic field is increased. The superconductivity is locally broken inside the quantized bundle, called a vortex, whose name comes from a current circulating the bundle to confine the magnetic flux from the surrounding superconducting area, and thus the superconducting gap disappears inside. The vortices can, therefore, be visualized by spatial mapping of the gap, measured in STS mode. Zhang et al. demonstrate visually that the number of vortices gradually increases with the applied magnetic field on the oneatomic-layer Pb thin film until the entire area is covered with vortices and the superconductivity is completely suppressed, as often shown schematically in textbooks on superconductivity. The superconductivity of the thin films can also be regarded as a new property of surfaces. The surfaces studied by Zhang et al. two types of Pd-deposited surface and an In-induced one have been previously characterized by surface 80 nature physics VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved

15 news & views analytical tools; their atomic structures and electronic energy levels have been well revealed. In general, atomic structures of clean surfaces or adsorbate-covered surfaces are often reconstructed from those of the corresponding bulk to reduce excess energy generated by surface formation, and the reconstructed surfaces have electronic properties different to those of the bulk. For instance, a metallic surface can exist on a semiconducting substrate, and spin-split surface states appear on a non-magnetic substrate. Although the one-atomic-layer thin films still keep freeelectron-like metallic properties, the metal layer has a significant interaction with the substrate, implying that the layers would not be the same as the corresponding self-standing layers, if such layers would exist. Nonetheless, these one-atomiclayer decorated surfaces formed on a semiconducting substrate are the first example of surface superconductivity, adding a new item in the list of peculiar surface-localized properties. The authors present unique superconductors whose contributing electrons are ultimately confined in a two-dimensional space. As physicists have already developed various methods for manipulating surface structures down to the atomic level with the aid of nanotechnology intriguing physical phenomena will be explored on these newly discovered systems. Yukio Hasegawa is at the Institute for Solid State Physics, University of Tokyo, Kashiwa-no-ha, Kashiwa, Chiba , Japan. hasegawa@issp.u-tokyo.ac.jp References 1. Guo, Y. et al. Science 306, (2004). 2. Qin, S. Y. et al. Science 324, (2009). 3. Zhang, T. et al. Nature Phys. 6, (2010). topological insulators a hex on protection Three-dimensional topological insulators are bulk insulators with charged surface states that are distinct from the bulk. The surface states are gapless (metallic) with linear excitation energy that in three dimensions traces out a cone, as does the Dirac energy spectrum for massless fermions. Robust against perturbations and scattering, such surface states are topologically protected the surface contains an odd number of Dirac cones as a result of the topological index (Chern number) that characterizes the electronic band structure. Besides being mathematical curiosities, these states are also under consideration for fault-tolerant quantum computation and low-power spintronic devices. In 2008, Bi 2 Te 3 was predicted to be the simplest three-dimensional topological insulator (Q. Xiao-Liang et al. Phys. Rev. B 78, ; 2008). Soon thereafter, angle-resolved photoemission spectroscopy experiments on Sn-doped Bi 2 Te 3 doping shifts the bulk states away from the Fermi energy, leaving only the surface states accessible to the charge carriers confirmed the presence of a single Dirac cone at the centre of the unit cell in momentum space (Y. L. Chen et al. Science 325, ; 2009). The Fermi surface associated with the surface states changes volume with doping concentration and varies from a hexagram at zero doping to a hexagon. Using scanning tunnelling microscopy (STM), Zhanybek Alpichshev and co-workers have now imaged the surface of Bi 2 Te 3 and studied the protected nature of the surface states (Z. Alpichshev et al. Phys. Rev. Lett. 104, ; 2010). Constant-energy contours of the surface-state bands of pure Bi 2 Te 3 are pictured, showing the change from the Dirac point at 330 mev (hexagon; upper left) to 0 mev (hexagram; lower right), where red denotes the occupied states. From about 100 mev, the surface becomes warped. Oscillations resulting from defect-induced interference patterns of electron waves detected by STM abruptly cease below 100 mev: backscattering is suppressed. So far, these angle-resolved photoemission spectroscopy and STM data are consistent with protected surface states. But the oscillations in STM spectra show a linear energy dependence where Alpichshev et al. believe the Fermi surface changes from convex hexagon to concave hexagram. The presence of hexagonal warping may indicate extra scattering channels along nesting wave vectors that connect parts of the Fermi surface together (L. Fu Phys. Rev. Lett. 103, ; 2009). Thus surface-band deformations could lead to competing ground states that would otherwise have been forbidden. may chiao 2010 Aps nature physics VOL 6 FEBRUARY Macmillan Publishers Limited. All rights reserved