Metal-insulator transition characteristics of VO2 thin films grownon Ge„100… single crystals

Z. Yang,a� C. Ko, and S. RamanathanHarvard School of Engineering and Applied Sciences, Harvard University, Cambridge,Massachusetts 02138, USA

�Received 25 July 2010; accepted 18 August 2010; published online 11 October 2010�

Phase transitions exhibited by correlated oxides could be of potential relevance to the emerging fieldof oxide electronics. We report on the synthesis of high-quality VO2 thin films grown on singlecrystal Ge�100� substrates by physical vapor deposition and their metal-insulator transition �MIT�properties. Thermally triggered MIT is demonstrated with nearly three orders of magnituderesistance change across the MIT with transition temperatures of 67 °C �heating� and 61 °C�cooling�. Voltage-triggered hysteretic MIT is observed at room temperature at threshold voltage of�2.1 V for �100 nm thickness VO2 films. Activation energies for electron transport in theinsulating and conducting states are obtained from variable temperature resistance measurements.We further compare the properties of VO2 thin films grown under identical conditions on Si�100�single crystals. The VO2 thin films grown on Ge substrate show higher degree of crystallinity,slightly reduced compressive strain, larger resistance change across MIT compared to those grownon Si. Depth-dependent x-ray photoelectron spectroscopy measurements were performed to provideinformation on compositional variation trends in the two cases. These results suggest Ge could bea suitable substrate for further explorations of switching phenomena and devices for thin filmfunctional oxides. © 2010 American Institute of Physics. �doi:10.1063/1.3492716�

I. INTRODUCTION

Vanadium dioxide �VO2� undergoes a sharp metal-insulator transition �MIT� at �340 K.1 Recently, it has beenobserved that the MIT in VO2 can be triggered electrically2–6

in addition to the well-studied thermally induced MIT. Dueto the near-room temperature transition and voltage-triggeredMIT, a great deal of novel logic, memory, and sensor devicesare being proposed utilizing VO2.7–11 Particularly, the elec-trically triggered MIT in VO2 is promising for potential ap-plications as ultrafast switches,12 Mott field effecttransistor,7,8,13 two-terminal memristive devices14 whereinone may use the phase transition-induced resistance changeor nonlinear current-voltage �I-V� characteristics as a switch.A vertical device geometry is preferred among some of thesenovel VO2 devices, particularly for using the phase transitionelement to control current flow in a device or circuit.10,12 Todate, the most frequently used conducting substrate in verti-cal VO2 devices is heavily doped Si.4,5,15 Comparing to theTiO2 �Refs. 16 and 17� and sapphire18 substrates, the crys-tallinity of VO2 thin films on Si substrate is compromiseddue to larger lattice mismatch and formation of interfacialoxides, however, TiO2 and sapphire are insulating and maynot be applicable as bottom contacts in VO2 based out-of-plane devices. Heavily doped Ge substrates are conducting,and have a slightly smaller lattice mismatch with VO2 thanSi,19 but the VO2 growth on Ge substrates has not been stud-ied so far to the best of our knowledge. Further, Ge is a highmobility semiconductor that is attracting a great deal of in-terest in the transistor research community20–23 and hence iswell-suited for this study. In this paper, we report on the

synthesis of high-quality VO2 thin films grown on Ge sub-strates with the observation of both thermally triggered MITand room temperature voltage-triggered MIT. To the best ofour knowledge, this is the first report on growth and func-tional properties of VO2 films on single crystal Ge.

II. EXPERIMENTS

VO2 thin films were grown on heavily doped n+-Ge�100� single-crystalline substrates ��0.008 � cm resistivity�and reference n+-Si �100� single-crystalline substrates�0.002–0.005 � cm resistivity� simultaneously with identi-cal growth conditions using radio frequency �rf� sputteringtechnique with a V2O5 target. Numerous growth experimentswere carried out to identify optimal growth conditions forgood quality films. The VO2 thin film growth is quite sensi-tive to the oxygen gas flow rate. Figure 1 shows an exampleof flow rate effect on VO2 thin films grown on sapphiresubstrates using V2O5 target at growth temperature and pres-sure of 550 °C and 5.0 mTorr. The total Ar and O2 gas flowrate is 100 SCCM �SCCM denotes cubic centimeter perminute at STP�. Figures 1�a�–1�d� show the resistance ratiobetween 25 and 100 °C for VO2 thin films of 413 �green�,968 �blue�, 1051 �red�, and 14 �black� for O2 gas flow ratesof 0 SCCM, 0.5 SCCM, 0.8 SCCM, and 1.0 SCCM, respec-tively. The O2 gas flow rates and resistance ratios are sum-marized in Fig. 1�e�. The sensitivity of VO2 properties onsynthesis conditions has been noted in earlier studies aswell24–31 and this is due to the fact that vanadium can readilyexist in multiple valence states,24–31 leading to several Mag-neli VnO2n−1 phases.32 Hence, great care is needed in deter-mining optimal growth conditions that include control overthe oxygen pressure during growth to obtain VO2 films witha�Electronic mail: zyang@seas.harvard.edu.

JOURNAL OF APPLIED PHYSICS 108, 073708 �2010�

0021-8979/2010/108�7�/073708/6/$30.00 © 2010 American Institute of Physics108, 073708-1

good MIT properties. Specifically, the growth temperature,growth pressure, gas flow rate, target gun plasma power, andgrowth time were 550 °C, 5.0 mTorr, 99.2 SCCM Ar plus0.8 SCCM O2, 120 W, and 4 h, respectively, for the studyreported here. X-ray diffraction �XRD� was performed on theVO2 thin films in �-2� geometry using Cu K� of 0.1548 nmas the x-ray source. X-ray photoelectron spectroscopy �XPS�was performed using Al K� of 1.4866 keV as x-ray source.Depth-dependent XPS studies were assisted with Ar ion mill-ing. Au�200 nm�/Ti�20 nm� metal contacts with a size of100�100 �m2 were deposited using a Sharon e-beamevaporator with shadow masks. The I-V characteristics weremeasured by a Keithley 236 source meter integrated in atemperature controllable probe station.

III. RESULTS AND DISCUSSION

Figure 2�a� shows XRD patterns of VO2 thin filmsgrown on Ge and Si substrates as red and green curves, re-spectively. The �011� peak dominates in the XRD patterns ofboth the VO2 thin films on Ge �VO2-on-Ge� and the VO2

thin films on Si �VO2-on-Si�. Si and Ge possess the samecrystal structure of diamond but an �4% lattice constantmismatch.19 Hence, it is reasonable that the growth directionof both VO2 thin films are along the same orientation �011�,indicated by the XRD patterns. Several minor orientationsappear in the XRD pattern of VO2-on-Si but not VO2-on-Ge,such as �200�, �002�, and �210�, indicating qualitatively thatthe VO2-on-Si is less textured than VO2-on-Ge. Figure 2�b�shows the Gaussian fits �blue solid lines� to the �011� XRD

peaks of VO2-on-Ge �red symbols� and VO2-on-Si �greensymbols�. The full-width-at-half-maximum �FWHM� of the�011� XRD peaks of VO2-on-Ge and VO2-on-Si is 0.50° and0.70°, respectively, indicating VO2-on-Ge has slightly bettercrystallinity. The average grain size of VO2-on-Ge andVO2-on-Si is estimated to be �17 and �12 nm, based onScherrer equation �grain size d��0.9� /FWHM cos �� with� and � of the x-ray wavelength and diffraction angle in theXRD measurements33�. The slight �011� XRD peak shift inVO2-on-Ge comparing to VO2-on-Si �2�=27.78° versus27.95°�, arises from the slightly smaller compressive strainin VO2-on-Ge than that in VO2-on-Si due to the larger latticeconstant of Ge than Si.19 The VO2-on-Ge shows a slightlyless compressive strain �by �0.6%� than VO2-on-Si.

Figures 3�a� and 3�b� show depth-dependent XPS andhigh-resolution XPS spectra of VO2-on-Ge. Figures 4�a� and4�b� show similar XPS spectra taken from VO2-on-Si. Thecurves are intentionally vertically shifted for clarity. The bot-tom XPS curves are taken from near the VO2 thin film sur-face, while the top XPS curves are near the VO2–Ge �orVO2–Si� interface. In the high-resolution XPS spectra, wefocus on O 1s, V 2p1/2, and V 2p3/2 peaks. The peak posi-tions and shapes of V 2p3/2 are used to analyze the valencestates for vanadium.24–31 Initially, the dominating valence ofvanadium near the VO2 thin film surfaces on both Ge and Sisubstrates is +4 �V 2p3/2 at 515.8�0.2 eV for +4valence�,24–31 indicating stoichiometric VO2. With the sur-face VO2 sputtered off by ion milling, a shoulder peak showsup at lower energy side of the +4 V 2p3/2 peak, indicating

FIG. 1. �Color online� ��a�–�d�� The temperature dependence of normalizedresistance in VO2 thin films with O2 gas flow rate of 0 �green�, 0.5 �blue�,0.8 �red�, and 1.0 �black� SCCM using V2O5 target at 550 °C and 5.0mTorr. The total Ar and O2 gas flow rate is 100 SCCM. �e� The plot ofresistance ratio between 25 and 100 °C of the VO2 thin films as a functionof O2 gas flow rate.

FIG. 2. �Color online� �a� XRD patterns of VO2 thin films grown on Ge�red� and Si �green� substrates. �b� Gaussian fits �blue solid lines� of �011�XRD peaks of VO2-on-Ge �red symbols� and VO2-on-Si �green symbols�.The FWHM of the �011� XRD peaks of VO2-on-Ge and VO2-on-Si is 0.50°and 0.70°, respectively.

073708-2 Yang, Ko, and Ramanathan J. Appl. Phys. 108, 073708 �2010�

the existence of a slight mixture valence of vanadium in thefilm, which is very close to the +2 V valence position�V 2p3/2 at 513.7�0.2 eV for +2 valence�.24–31 In theVO2-on-Ge, this shoulder peak persists until the interfacebetween VO2 and Ge �Fig. 3�b��, however, the shoulder peakeventually dominates and becomes a separate peak near theinterface between VO2 and Si �Fig. 4�b��, representing alower valence state of vanadium. This indicates that theVO2-on-Si shows a more evident off-stoichiometry thanVO2-on-Ge near the interface region. Figure 5�a� shows anexample of how the relative fraction of each valence state ofvanadium is extracted from the fitting analyses. The V 2p3/2XPS peak �in the high-resolution XPS spectra� was deconvo-luted to two V 2p3/2 peaks with +4 and +2 valences. Theratio of the areas underneath the two deconvoluted peaks�after subtraction of the Shirley baseline background� givesthe relative fraction of the mixture. Figure 5�b� shows de-tailed analyses results of the relative fraction of +4 valencevanadium in the VO2-on-Ge �red� and VO2-on-Si �green�.The VO2-on-Ge shows a high fraction of +4 near the surface�80%�. Both VO2-on-Ge and VO2-on-Si show good sto-ichiometry ��70%� in center region of the films ��10 nmbelow the surface and �20 nm above the interface�, whilethe VO2-on-Ge shows a slightly higher fraction of +4 va-lence. When the VO2 thin films reach the near-interface re-gions, the fraction of +4 valence in the VO2-on-Si decreasesmuch faster than that of the VO2-on-Ge. This could poten-tially arise from interfacial reaction with the silicon substrateowing to its propensity to form silicon oxides while germa-

nium oxide is less stable from thermodynamicconsiderations.34,35 The error bars in Fig. 5�b� arise from thefitting uncertainties of the inaccuracy. As noted earlier, sto-ichiometry control is particularly difficult for VO2 growthdue to presence of Magneli phases32 in this system and easeof vanadium to take on several possible oxidation states.Slight off-stoichiometry with mixture of several valencestates of vanadium is common, especially when the Si orglass substrates are used instead of TiO2 or sapphiresubstrates.24–31 Our results are consistent with these priorreports and a slight degradation in resistance ratio across theMIT is to be expected when compared to single-crystallinefilms on lattice-matched insulating substrates or bulk singlecrystals.

Figure 6�a� shows the temperature dependence of thenormalized in-plane resistance �R�T� /R�20 °C�� of theVO2-on-Ge substrate �red circles� and VO2-on-Si �greencircles�. The resistance of the VO2 thin film at each tempera-ture point is determined by the reciprocal slope of an in-plane linear I-V scan between two probes. �The linear I-Vcurves indicate Ohmic contacts36 between the probes and theVO2 thin film.� The solid and open circles represent tempera-ture ramping up and down, respectively. Nearly three ordersof magnitude resistance change is observed from 20 to100 °C in the VO2-on-Ge, but less than two orders change inVO2-on-Si, indicating a higher quality VO2-on-Ge thanVO2-on-Si, consistent with the XPS and XRD results. UsingGaussian fittings on the derivative logarithmic plot of

FIG. 3. �Color online� �a� Depth-dependent XPS spectra of VO2 thin film onGe substrate. �b� High-resolution depth-dependent XPS spectra of VO2 thinfilm around O 1s and V 2p regime. The curves are vertical shifted forclarity. The bottom XPS curves are from near the VO2 thin film surface,while the top XPS curves are near the VO2-Ge substrate interface.

FIG. 4. �Color online� �a� Depth-dependent XPS spectra of VO2 thin film onSi substrate. �b� High-resolution depth-dependent XPS spectra of VO2 thinfilm focusing on O 1s and V 2p regime. The curves are intentionally ver-tical shifted for clarity. The bottom XPS curves are from near the VO2 thinfilm surface, while the top XPS curves are near the VO2-Si substrateinterface.

073708-3 Yang, Ko, and Ramanathan J. Appl. Phys. 108, 073708 �2010�

R-T,15,37,38 the MIT temperature TC of VO2-on-Ge is deter-mined to be 67.3 °C and 60.8 °C during temperature ramp-ing up and ramping down, respectively, as shown in Fig.6�b�, while 67.3 °C and 58.6 °C for VO2-on-Si during tem-perature ramping up and ramping down, respectively, asshown in Fig. 6�c�. The VO2-on-Ge shows 2.2 °C narrowerMIT width �6.5 °C versus 8.7 °C� than the VO2 thin film onSi. The minor difference in the transition temperature andMIT width between VO2-on-Ge and VO2-on-Si is possiblydue to variation in the faction of +4 valence �other valencevanadium oxide phases can distort the MIT �Ref. 31�� andstrain39 within the films as discussed earlier. However, theobserved thermal-MIT hysteresis window width and transi-tion temperatures in both VO2 thin films show consistenttrends with previous reported values.4,5,15,38

Figure 7�a� shows the plot of logarithmic resistance�Ln�R�� versus reciprocal temperature �1 /T�, which is em-ployed for the activation energy �Ea� analyses. The activationenergy is estimated by the slope of a linear fitting of Ln�R�over �1 /kBT�, as Ln�R�=b+Ea /kBT, where R, T, kB, and bare resistance, temperature �in unit K�, Boltzmann constant,and intercept. Figures 7�b� and 7�c� show the activation en-ergy of VO2-on-Ge �red� and VO2-on-Si �green� in low-temperature “insulating” and high-temperature “metallic”states. The activation energy of VO2-on-Ge is 193�7 meVand 149�4 meV for low- and high-temperature states, re-spectively. The reason that the high-temperature activationenergy is not significantly smaller than that of the low-temperature is likely due to the mixed valence states of theVO2 thin films. Once the temperature ramps up across theMIT transition temperature, only the VO2 phase in the filmundergoes transition into the metallic state. In other words,even at high temperatures the film is not entirely metallic;

FIG. 5. �Color online� �a� An example of fitting on the V 2p3/2 XPS peakand deconvolution into two peaks representing vanadium +4 and +2 va-lences, showing how the relative fraction of each valence state of vanadiumis estimated. �b� Fraction of the +4 valence vanadium atoms in VO2-on-Ge�red� and VO2-on-Si �green� along the film thickness direction.

FIG. 6. �Color online� �a� Temperature dependence of the normalized resis-tance of the VO2-on-Ge �red circles� and VO2-on-Si �green circles�. Thesolid and open circles represent temperature ramping up and down, respec-tively. �b� The MIT temperature TC of the VO2-on-Ge is determined to be67.3 °C and 60.8 °C during temperature ramping up and ramping down,respectively, using Gaussian fits on the derivative logarithmic plots. �c� TheMIT temperature TC of the VO2-on-Si is determined to be 67.3 °C and58.6 °C during temperature ramping up and ramping down, respectively,using Gaussian fits on the derivative logarithmic plots.

FIG. 7. �Color online� �a� Plot of logarithmic resistance �Ln�R�� vs recipro-cal temperature �1 /T� of VO2-on-Ge �red� and VO2-on-Si �green�. �b� Theactivation energy Ea of VO2-on-Ge �red� and VO2-on-Si �green� at high-temperature conducting state. �c� The activation energy Ea of VO2-on-Ge�red� and VO2-on-Si �green� at low-temperature insulating state.

073708-4 Yang, Ko, and Ramanathan J. Appl. Phys. 108, 073708 �2010�

hence the activation energy is not significantly smaller thanits low-temperature insulating state. In a previous study,4 forexample, a VO2 thin film with nearly four orders of magni-tude resistance change in thermal-MIT, which is closer to theideal VO2, shows a smaller high-temperature activation en-ergy than the low-temperature one �80–90 meV versus 230–250 meV�. For the VO2-on-Si sample, again due to mixtureof valences and smaller order of resistance change acrossMIT than VO2-on-Ge, the high-temperature activation en-ergy is not distinct from the low-temperature one, consider-ing the error bars. This likely indicates that the high-temperature state is a mixture of metallic VO2 regions andsome insulating phases �from other valence vanadium oxidephases�. Similar results on uniform temperature dependenceof resistance on both sides of the MIT were also observedpreviously in off-stoichiometric VO2 films grown byelectron-beam evaporation.40 We do note here that the esti-mated activation energies are comparable to previousstudies.41–43 Kristensen41 prepared VO2 using powder sinter-ing method and found an activation energy of 240 meV fromtemperature dependence of the electrical conductivity analy-ses. Sahana et al.42 grew VO2 thin films on glass substratesby chemical vapor deposition and the temperature-dependentresistance measurements indicated that the activation energyof the VO2 thin films ranged from 180–260 meV dependingon growth conditions.

Figure 8�a� shows the I-V characteristics of VO2-on-Gemeasured in vertical geometry between 0 and 5 V at 25 °C�red�, 60 °C �green�, and 90 °C �blue�. Voltage-triggeredMIT is observed at 25 °C with threshold voltage at �2.1 V.The inset in Fig. 8�a� shows the magnified region of theresistive hysteresis window of the I-V curves at 25 °C, withscan directions marked with arrows. The width of the voltagetriggered MIT window is �1 V. Considering the �100 nmthickness of the VO2 thin film, the threshold electric-field fortriggering MIT is approximately 2.1�107 V /m at 25 °C,also in good agreement with studies on both VO2 planarjunctions on sapphire and capacitor-type devices onsilicon.2–7 In the hysteresis curves, some “staircase” discon-tinuity shapes are observed, which is likely due to grainboundary or domain related contributions to MIT.44 Similarphenomena were also observed in nanoscale VO2 planarjunctions by Sharoni et al.,45 who concluded that thethermal-MIT in VO2 occurs through a series of staircaseshape avalanches induced by impact ionization due to grainboundary effects. The effects were seen primarily in nano-scale junctions and hence reasonable to expect in our mea-surements since we are studying voltage-triggered phasetransition across an �100 nm film. Each staircase step maycorrespond to a grain/domain boundary associated change�slight differences in the insulating state conductivity lead toshifts in the critical voltage needed to initiate the transitionand hence a step-like shape emerges�. Electrically triggeringMIT studies by conducting atomic force microscopy but inregions smaller than a single grain size, shows free of stair-case steps in the I-V curves,46 which supports the analyseshere and those in Ref. 45.

Figure 8�b� shows the Poole–Frenkel �PF� plot �Ln�I /V�versus Sqrt �V�� of the I-V curves shown in Fig. 8�a�. At

large voltages ��3.1 V�, the VO2 thin film shows metallicbehavior over a broad range of temperature. At low voltages��0.7 V�, the VO2 thin film only shows metallic behaviorat high temperatures, but behaves as insulator at lower tem-peratures. At intermediate bias voltages and below thermal-MIT temperature, the PF mechanism dominates the conduc-tion, indicated by the linear dependence of Ln�I /V� over Sqrt�V�. The two bold blue lines in the PF region is used forvisual guidance for the linear dependences. Three regions,namely metallic, “Ohmic” and “PF dominating” are denotedin Fig. 8�b�, showing the dominant conduction mechanism ineach region. At either high-temperature or large applied volt-ages, the VO2 enters metallic state due to the thermally orelectrically triggered MIT. At low temperatures and smallbias voltage, Ohmic conduction behavior is predominant inVO2. At low-temperature and intermediate bias voltage, thePF mechanism dominates. Nonstoichiometric defects, suchas oxygen vacancies in VO2, may serve as localized trapsrequired for the PF emission. Prior to onset of metallic state,the localized states assist the PF conduction under electric-field bias.4,5,47

IV. SUMMARY

VO2 thin films were grown on Ge�100� single crystalsubstrates using rf sputtering. Thermally triggered MIT withnearly three orders of magnitude resistance change with an

FIG. 8. �Color online� �a� I-V characteristics of the VO2-on-Ge measured invertical geometry between 0 to 5 V at 25 °C �red�, 60 °C �green�, and90 °C �blue�. The electric-field-triggered MIT is observed at 25 °C. Theinset shows the magnified region of the resistive hysteresis window of theI-V curves at 25 °C, with scanning directions marked with arrows. �b� PFplot of the I-V curves �Ln�I /V� vs Sqrt �V��. Three regions, metallic, Ohmic,and PF dominating are defined. The two blue bold lines in the PF region arefor visual guidance, indicating the dominating of PF conduction mechanismin that region.

073708-5 Yang, Ko, and Ramanathan J. Appl. Phys. 108, 073708 �2010�

MIT width of �6 °C is observed in VO2 thin films grown onGe. Hysteretic voltage-triggered MIT is observed at roomtemperature at a critical voltage of �2.1 V with a hysteresiswindow of �1 V in VO2 thin films grown on Ge. The struc-tural and electrical measurements suggest that Ge may be asuitable substrate for further explorations of phase transition-based oxide electronics utilizing MITs.

ACKNOWLEDGMENTS

This work was supported by ONR Grant N00014-10-1-0131 and AFOSR Grant FA9550-08-1-0203.

1F. J. Morin, Phys. Rev. Lett. 3, 34 �1959�.2G. Stefanovich, A. Pergament, and D. Stefanovich, J. Phys.: Condens.Matter 12, 8837 �2000�.

3B. G. Chae, H. T. Kim, D. H. Youn, and K. Y. Kang, Physica B 369, 76�2005�.

4C. Ko and S. Ramanathan, Appl. Phys. Lett. 93, 252101 �2008�.5D. Ruzmetov, G. Gopalakrishnan, J. Deng, V. Narayanamurti, and S. Ra-manathan, J. Appl. Phys. 106, 083702 �2009�.

6D. Ruzmetov, G. Gopalakrishnan, C. Ko, V. Narayanamurti, and S. Ra-manathan, J. Appl. Phys. 107, 114516 �2010�.

7H. T. Kim, B. G. Chae, D. H. Youn, S. L. Maeng, G. Kim, K. Y. Kang, andY. S. Lim, New J. Phys. 6, 52 �2004�.

8S. Hormoz and S. Ramanathan, Solid-State Electron. 54, 654 �2010�.9M. J. Lee, Y. Park, D. Suh, E. H. Lee, S. Seo, D. C. Kim, B. Jung, B. S.Kang, S. E. Ahn, C. B. Lee, D. H. Seo, Y. K. Cha, I. K. Yoo, J. S. Kim,and B. H. Park, Adv. Mater. �Weinheim, Ger.� 19, 3919 �2007�.

10T. H. Yang, R. Aggarwal, A. Gupta, H. H. Zhou, R. J. Narayan, and J.Narayan, J. Appl. Phys. 107, 053514 �2010�.

11A. Gupta, R. Aggarwal, P. Gupta, T. Dutta, R. J. Narayan, and J. Narayan,Appl. Phys. Lett. 95, 111915 �2009�.

12A. Cavalleri, C. Tóth, C. W. Siders, J. A. Squier, F. Ráksi, P. Forget, andJ. C. Kieffer, Phys. Rev. Lett. 87, 237401 �2001�.

13C. Zhou, D. M. Newns, J. A. Misewich, and P. C. Pattnaik, Appl. Phys.Lett. 70, 598 �1997�.

14T. Driscoll, H. T. Kim, B. G. Chae, M. Di Ventra, and D. N. Basov, Appl.Phys. Lett. 95, 043503 �2009�.

15C. Ko and S. Ramanathan, J. Appl. Phys. 106, 034101 �2009�.16Y. Muraoka and Z. Hiroi, Appl. Phys. Lett. 80, 583 �2002�.17J. Lu, K. G. West, and S. A. Wolf, Appl. Phys. Lett. 93, 262107 �2008�.18K. Okimura, J. Sakai, and S. Ramanathan, J. Appl. Phys. 107, 063503

�2010�.19In monoclinic VO2, lattice constant a=0.5753 nm. The lattice constants

of Si and Ge are 0.5431 nm and 0.5646 nm, respectively.20S. Verdonckt-Vandebroek, E. F. Crabbe, B. S. Meyerson, D. L. Harame, P.

J. Restle, J. M. C. Stork, and J. B. Johnson, IEEE Trans. Electron Devices41, 90 �1994�.

21M. Ieong, B. Doris, J. Kedzierski, K. Rim, and M. Yang, Science 306,2057 �2004�.

22M. L. Lee, E. A. Fitzgerald, M. T. Bulsara, M. T. Currie, and A.Lochtefeld, J. Appl. Phys. 97, 011101 �2005�.

23T. Maeda, Y. Morita, and S. Takagi, Appl. Phys. Express 3, 061301�2010�.

24G. A. Sawatzky and D. Post, Phys. Rev. B 20, 1546 �1979�.25J. Mendialdua, R. Casanova, and Y. Barbaux, J. Electron Spectrosc. Relat.

Phenom. 71, 249 �1995�.26M. Ghanashyam Krishna, Y. Debauge, and A. K. Bhattacharya, Thin Solid

Films 312, 116 �1998�.27M. Demeter, M. Neumann, and W. Reichelt, Surf. Sci. 454–456, 41

�2000�.28G. Silversmit, D. Depla, H. Poelman, G. B. Marin, and R. De Gryse, J.

Electron Spectrosc. Relat. Phenom. 135, 167 �2004�.29N. Alov, D. Kutsko, I. Spirovová, and Z. Bastl, Surf. Sci. 600, 1628

�2006�.30A. Romanyuk and P. Oelhafen, Thin Solid Films 515, 6544 �2007�.31D. H. Youn, H. T. Kim, B. G. Chae, Y. J. Hwang, J. W. Lee, S. L. Maeng,

and K. Y. Kang, J. Vac. Sci. Technol. A 22, 719 �2004�.32J. B. Goodenough, J. Solid State Chem. 3, 490 �1971�.33A. Patterson, Phys. Rev. 56, 978 �1939�.34W. Kaiser, P. H. Keck, and C. F. Lange, Phys. Rev. 101, 1264 �1956�.35L. M. Terman, Solid-State Electron. 5, 285 �1962�.36L. J. Mandalapu, Z. Yang, and J. L. Liu, Appl. Phys. Lett. 90, 252103

�2007�.37D. Brassard, S. Fourmaux, M. Jean-Jacques, J. C. Kieffer, and M. A. E.

Khakani, Appl. Phys. Lett. 87, 051910 �2005�.38D. Ruzmetov, K. T. Zawilski, V. Narayanamurti, and S. Ramanathan, J.

Appl. Phys. 102, 113715 �2007�.39T. Kikuzuki and M. Lippmaa, Appl. Phys. Lett. 96, 132107 �2010�; B.

Lazarovits, K. Kim, K. Haule, and G. Kotliar, Phys. Rev. B 81, 115117�2010�.

40R. G. Mani and S. Ramanathan, Appl. Phys. Lett. 91, 062104 �2007�.41I. K. Kristensen, Mater. Res. Bull. 9, 1677 �1974�.42M. B. Sahana, M. S. Dharmaprakash, and S. A. Shivashankar, J. Mater.

Chem. 12, 333 �2002�.43J. Li and N. Yuan, Appl. Surf. Sci. 233, 252 �2004�.44A. Frenzel, M. M. Qazilbash, M. Brehm, B. G. Chae, B. J. Kim, H. T.

Kim, A. V. Balatsky, F. Keilmann, and D. N. Basov, Phys. Rev. B 80,115115 �2009�.

45A. Sharoni, J. G. Ramírez, and I. K. Schuller, Phys. Rev. Lett. 101,026404 �2008�.

46J. Kim, C. Ko, A. Frenzel, S. Ramanathan, and J. E. Hoffman, Appl. Phys.Lett. 96, 213106 �2010�.

47P. P. Boriskov, A. A. Velichko, A. L. Pergament, G. B. Stefanovich, and D.G. Stefanovich, Tech. Phys. Lett. 28, 406 �2002�.

073708-6 Yang, Ko, and Ramanathan J. Appl. Phys. 108, 073708 �2010�