Letter pubs.acs.org/NanoLett
Effect of Interface Atomic Structure on the Electronic Properties of Nano-Sized Metal−Oxide Interfaces Wei Qin,†,‡ Jiechang Hou,‡ and Dawn A. Bonnell* Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States S Supporting Information *
ABSTRACT: We report that the size dependence of electronic properties at nanosized metal−semiconducting oxide interfaces is significantly affected by the interface atomic structure. The properties of interfaces with two orientations are compared over size range of 20−200 nm. The difference in interface atomic structure leads to electronic structure differences that alter electron transfer paths. Specifically, interfaces with a higher concentration of undercoordinated Ti result in enhanced tunneling due to the presence of defect states or locally reduced tunnel barrier widths. This effect is superimposed on the mechanisms of size dependent properties at such small scales. KEYWORDS: metal−semiconductor interface, electronic properties, interface structure, AFM, TEM, metal oxide
N
tronic properties of individual interfaces are quantified as a function of size and interface orientation using conductive atomic force microscopy (C-AFM).5,19,20 The electronic properties are compared to the interface atomic structure determined from high resolution transmission electron microscopy (HRTEM) and aspects of the electronic structure determined from electron energy loss spectroscopy (EELS).21 EELS has previously been used to characterize the valence states along SrTiO3/LaTiO3 superlattices,22 the electronic structures at SiOx/Si interfaces,23 as well as dislocation cores and grain boundaries.24,25 This combination of measurements enables the correlation between the interface structure resulting from the two ORs, the consequent differences in local atomic structural configurations, and the interface properties at sizes ranging from 20 to 150 nm. Experimental Procedures. SrTiO3 single crystalline substrates with Nb dopant concentration of 0.02 at% (Princeton Scientific, 5 × 5 × 0.5 mm3) were annealed at 1000 °C for 1 h after cleaning by acetone and ethanol. Nanosized Au/SrTiO3 contacts were produced by drop-casting gold nanoparticles (citrate stabilized in H2O solution, British Biocell) with various diameters (20, 40, 60, 100, 150 nm) on the atomically flat surfaces. After the solution was evaporated, samples with dispersed nanoparticles were annealed in ambient for an hour at 950 °C to remove residual organic compounds and obtain intimate interface contacts. An aluminum thin film (∼100 nm) was thermally evaporated on the back side of the samples to form Ohmic contacts, after which Au/STO/Al
anosized metal-oxide interfaces have become substantial and critical components in technologies, including electronic devices,1,2 energy harvesting,3 and catalysis.4 Continuous downscaling raises opportunities for new functionalities, as well as challenges in reliable performance. Many factors that are negligible when averaged at the macroscopic scale and even at the micron-scale, such as interfacial orientation,5−7 polarization,8 and local inhomogeneity,9,10 become crucial to the electronic properties of systems with nanoscale interfaces. Among these effects, interfacial atomic arrangement or orientation relations are nontrivial and can modulate interfacial electronic states. Because of the complex nature of metal− semiconductor interfaces, the orientation dependence of electronic properties and the associated mechanisms remain open issues. Some researchers argue that little or no orientation dependence is present, for example, on Au, Al/Ge,11 and Au/ Si12 interfaces, whereas others demonstrate orientation-based variations at CoSi2/Si,13 Au/SrTiO3,14 Au/GaAs15 and Au/ carbon nanotube interfaces.6 Possible mechanisms have been proposed for specific systems, generally based on metal-induced gap states or Fermi level pinning at the Schottky contact, resulting from different bonding interactions at the interfaces.6,11,16 However, the majority of this research focuses only on the semiconducting substrate orientation. The metal crystal orientations could also affect the interface properties. Here, we address the issue of interface structure dependence of electronic properties using Au nanoparticle/atomically flat SrTiO3 (STO) single crystalline contacts as a model system.17,18 Taking advantage of the behavior of nanoparticles, we produce two distinct orientation relations (ORs): Au(100)//STO(100) and Au(111)//STO(100). The elec© XXXX American Chemical Society
Received: September 3, 2014
A
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 1. AFM topographic images for the Au(100)//STO(100) interface (a) and the Au(111)//STO(100) interface (d). The substrate orientations are labeled by white arrows and the Au nanoparticle orientations are labeled by black arrows. The inset in (a) illustrates the square cap shape of the labeled octagonal particle under a larger height scale. The corresponding cross section TEM images for Au(100)//STO(100) (b) and Au(111)//STO(100) (e), with arrows displaying crystal orientations for Au and STO. (c) and (f) Illustration of the possible interfacial atomic arrangements for Au(100)//STO(100) and Au(111)//STO(100), respectively.
Images and spectra were processed using Digital Micrograph. All EELS spectra were background-subtracted by a power law and normalized by a thickness correction (about 20 eV above the threshold). The energy resolution is 0.8 eV and signals are averaged over the beam spot. Results and Discussion. The nanosized Au/Nb-doped STO contacts are dispersed on the single crystalline STO(100) substrate surfaces. The annealed substrates present atomically flat surfaces with steps along [100] directions and a height of ∼4 Å, consistent with the STO lattice constant.27 After nanoparticle deposition, the subsequent annealing facilitates particle faceting into truncated octahedra, a process driven by surface energy reduction.28 At thermal equilibrium, the truncated octahedra consist of six (100) and eight (111) surfaces. Because the Au(100)//STO(100) and Au(111)// STO(100) interface energies are comparable, these two orientations coexist experimentally29 and have been observed in TEM, AFM, and STM.5,30,31 Figure 1a displays the topographic image of particles with Au(100)//STO(100) interfaces, in which the particles resemble octagons with square caps (inset) in AFM images. The cross-sectional TEM image [Figure 1b] exhibits a clean intimate contact between the Au nanoparticle and STO substrate, without organic residue from surface ligands. Electron diffraction and lattice planes in the images show that Au[001] is parallel to STO[001], perpendicular to the interface plane as indicated by arrows in the figure, consistent with Wang et al.’s observation.31 In the plane of the interface Au [100] is also parallel to STO[100], as determined by comparison of the STO surface step edge directions and the Au surface plane directions. Similarly, the particles with Au(111)//STO(100)
stacked samples were silver-pasted on an atomic force microscopy (AFM) mount. Topographic images of samples were acquired in AC tapping mode (Asylum MFP-3D AFM). Aspects of the interfacial orientation relations were determined from the 2D lateral shape and the known substrate orientation, and the interface sizes were estimated from the height profiles. In contact mode (ORCA), a Pt-coated conductive tip (Olympus AC240TM) was engaged on individual nanoparticles. Only completely faceted particles with distinguishable shapes in topographic images were measured. Interface electronic properties were characterized from current−voltage (I−V) responses collected from −10 to 10 V for at least 8 cycles (0.2 Hz). Tip loading forces on the particle were optimized and limited to less than 10 nN.26 Following electrical measurement, the particles were rescanned and any particle exhibiting damage was not considered in the analysis. Over 100 interfaces were characterized. Orientations of some Au nanoparticles were identified using scanning electron microscopy (SEM). Cross-sectional TEM samples were extracted with the focused ion beam (FIB) liftout method using an FEI Strata DB235. A 5 kV low voltage ion beam was applied for final thinning to minimize the beam damage. The thicknesses of TEM samples were less than 50 nm at the Au/STO interface when measured by SEM. High resolution TEM bright field images were obtained in a JEOL 2100 and EELS data were collected in a JEOL 2010F equipped with an electron energy loss spectrometer (Gatan, GIF) under scanning transmission electron microscope (STEM) mode. Both TEMs were operated under 200 kV accelerating voltage and a vacuum of less than 3 × 10−5 Pa at room temperature. B
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 2. (a) Representative I−V curves to compare Au(100)//STO(100) and Au(111)//STO(100) interfaces. Arrows show the voltage ramping directions. The transition region of the HRS is circled. (b) Schematic illustration of sample structure and electronic characterization, where the conductive tip, sample, and built-in voltage source form a circuit. Comparison of the resistances of HRS (c) and LRS (d) for the two orientations Au(100)//STO(100) and Au(111)//STO(100) at −2.5 V. Error bars represent the deviation of measurement cycles. Interfacial size is represented by particle height.
2a shows representative I−V responses for the two orientations with 40 nm interfaces. Note, the plateaus in the high voltage regimes are set to 2 nA by amplifier protection, whereas those near 0 V are limited by current resolution (0.1 pA). The current loops divide the I−V curves into two branches: a high resistance state (HRS) with lower current and a low resistance state (LRS) with higher current. In our samples, positive bias switches the system from the HRS to the LRS, whereas negative bias switches from the LRS to the HRS. In the positive bias regime, the I−V curves exhibit no significant interface orientation dependence. Correspondingly, the Schottky properties and resistive switching responses of the two orientations are similar, consistent with our previous work.9,34 In the negative bias regime, however, the Au(100)//STO(100) interfaces are more conductive than are the Au(111)//STO(100) interfaces for both the LRS and the HRS branches. The Schottky model applies to this system and thermionic emission dominates the interfacial transport, as expressed by eq 135
interfaces resemble hexagons in the AFM image, as shown in Figure 1b. The cross-sectional TEM image [Figure 1e] specifies that Au[111] is perpendicular to the interface. In the plane of the interface, Au[110] is parallel to STO[100]. On all particles with sizes larger than that of the tip, the top surface plane was parallel to the STO surface plane, indicated negligible tilt. (See Supporting Information Figure S1) Though other possible orientation relations may coexist,30 we only discuss and compare these two orientations. The atomic structures of interfaces with these two orientations would necessarily differ, resulting in possible orientation-dependent electronic properties. Figure 1c and f schematically illustrate the different atomic alignments at the interfaces that would develop in the absence of relaxation and defects. The lateral alignment of the atoms at the interface has been predicted by first-principles to be that the Au sits above the O atom or bridges between two O atoms.32,33 Regardless, as illustrated in Figure 1c and f, alignment with either of these positions would be more efficient on the Au(100)//STO(100) interface than on the Au(111)//STO(100) interface. The implications of this are discussed below. The electronic properties of nanosized Au/STO Schottky contacts were characterized by conductive AFM for interface diameters ranging from 20 to 150 nm. In experiments, only the nanoparticle height D can be precisely measured. Because the interfacial diameter is proportional to the particle height, all interfacial sizes are represented by the measured particle height D.9 Specifically, properties were characterized for interface sizes 40, 60, 70, 80, 100, and 150 nm for the two orientations. Figure
I = AJ0 = AA*T 2e−qΦB / kBT (e qV / nkBT − 1)
(1)
where J0 is the current density, A is the interfacial area, T is the temperature, A* is the Richardson constant, ΦB is the Schottky barrier, and n is the ideality factor. Consequently, the reverse current saturates at a value of −J0. The absence of saturation has been observed in the macroscopic Schottky contacts, especially at a large reverse bias.35,36 The large leakage current is similar to the “breakdown” phenomenon, attributed to image C
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 3. (a) Band structure of the intrinsic junction of Au (yellow) and STO (purple). Positive charges accumulate at the STO surface, whereas the metal surface is negatively charged. The top arrow indicates the direction of the built-in electric field. Oxygen vacancies migrate from STO to interface on annealing. (b) Band structure under negative bias, with reduced STO Fermi level. (c) In-gap electron states associated with undercoordinated Ti enhance tunneling. (d) Sufficiently high concentration of undercoordinated Ti reduces the depletion width.
Figure 4. EELS spectra of Ti-L and O−K edges for Au(100)//STO(100) and Au(111)//STO(100) interfaces at different distances from the interface. (a) Ti-L edge at Au(100)//STO(100), (b) O−K edge at Au(100)//STO(100), (c) Ti-L edge at Au(111)//STO(100), and (d) O−K edge at Au(111)//STO(100). The distances from the interface are labeled in (a). Each color represents the same distance from the interface.
force lowering of Schottky barrier or edge-induced tunneling. Given that the reverse current enhancement is much larger than that caused by image force lowering (ΔΦB is less than 0.01 eV)
for our system, the edge-induced tunneling dominates the leakage current in the reverse bias. The edge effect is governed by the development of depletion regions beneath the particles. D
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
At small interface sizes, the depletion region is size dependent and decreases with decreasing interface size, that is, with a larger edge-to-area ratio.37 For intermediate sized interfaces the presence of the edge reduces the depletion width, locally favoring tunneling. This explains the size-related trend in Figure 2c and d. The electronic structure of the interface is shown in Figure 3. The difference in the work functions of Au and STO drives positive charges (oxygen vacancies or electron holes) in the STO toward the interface. Under negative bias, Figure 3b, the STO Fermi level shifts down and the effective tunneling barrier width decreases. Due to the reduced tunneling depletion width, the edge effect is more prominent under the reverse bias; therefore, the current under negative bias is more susceptible to tunneling than that in the positive regimes. Figure 2c and d compares the resistances for the Au(100)// STO(100) and Au(111)//STO(100) interfaces at a voltage −2.5 V, to demonstrate the variation of the electronic properties with orientation. Additional comparisons are included in Supporting Information Figure S2. For both the LRS and the HRS, the Au(111)//STO(100) interface exhibits higher resistances over the size range of 40−150 nm. The slope change [circled in Figure 2a] is apparently due to interface charging since it is a function of voltage ramping rate (see Supporting Information Figure S3). Regardless of this effect, the overall HRS resistance of the Au(111)//STO(100) is higher than that of Au(100)//STO(100). Charging is not observed for LRS branches, and the LRS resistance of the Au(100)//STO(100) interfaces is smaller than that of the Au(111)//STO(100) for all sizes. As Figure 1 suggests, a difference in interface atomic structure can result in differences in the local electronic structure, which in turn can influence interface properties. For the case of STO, the expected variations concern the local concentration of O and the associated coordination of the Ti. In the stoichiometric STO Ti is fully oxidized in the nominal 4+ state. At surfaces and in the presence of O vacancies, it is reduced. To compare the possible variations in local Ti coordination at and near the interfaces, EELS gradient profiles were collected. Figure 4a−d compares the EELS profiles of Ti L2,3 (Ti 2p → 3d) edges and O K (O 1s → 2p) edges for the two interface orientations as a function of distance from the interfaces. Spectra were obtained on interfaces with similar sizes (∼400 nm) and identical thermal history. For both samples, the spectra furthest from the interface exhibit 4 peaks and the trend is for the peaks to broaden and merge close to the interfaces, indicating significant concentration of reduced Ti. The Ti is more oxidized at and near the (111) than at the (100) interface. A concomitant “damping-out” of O K peaks is also observed from bulk to interface as indicated by reduced relative intensities of the O K edges. This is consistent with the presence of oxygen vacancies associated with undercoordinated Ti atoms.22,38,39 The peak intensity of the O edge is much weaker than that of the Ti edge, so we primarily rely on the Ti edge differences to discuss the chemical variation within the interface region. As labeled in Figure 4a, the two peaks in Ti L3 correspond to eg (∼457 eV) and t2g (∼459 eV) orbitals with the separation determined by the crystal field due to the local atomic structure. The broadened peaks closer to the interface are consistent with the crystal field variation induced by structural distortion of the TiO6 octahedra.38−41 Table 1 lists the eg and t2g peak positions measured from the EELS spectra as a function of distance from the interface. The separation of
Table 1. Ti L3 eg and t2g Peak Positions for Au(100)// STO(100) and Au(111)//STO(100) Orientation Relationsa Au(100)//STO(100) depth (nm) 1 3 5 10 15 20 25 30 a
eg (eV)
t2g (eV)
458.2 457.9 457.8 456.7 457.1 459 457.5 459.3 458 460 457.9 460.1
Au(111)//STO(100)
separation (eV)
eg (eV)
t2g (eV)
separation (eV)
0 0 0 0 1.9 1.8 2 2.2
457 457.6 457 456.9 456.5 457.1 457.5 457.4
458.9 459.8 459 459 458.5 459.2 459.6 459.7
1.9 2.2 2 2.1 2 2.1 2.1 2.3
Error of peak positions is ±0.1 eV.
the eg and t2g peaks decreases from ∼2.2 eV (@30 nm) to almost 0 eV at the interface for Au(100)//STO(100), whereas less change is observed for Au(111)//STO(100). To semiquantitatively compare the chemical defect gradients for the two orientations, the Ti3+ intensity ratios from the EELS profiles were extracted. Assuming that the spectra are a linear combination of reference spectra for the 4+ and 3+ states, the best fit for the combination of the references with the experimental data were determined.22 The ratios were calculated as f = af1 (Ti 3 +) + bf2 (Ti4 +)
η=
a a+b
(2)
where f is the experimental EELS spectrum; a and b are the fractional contribution of Ti3+ and Ti4+ valence states, respectively; and η is the Ti3+ fractional ratio. The analysis in Figure 4 used experimental end point spectra as internal references (f1 and f 2) of the Ti3+ and Ti4+ states, that is, one at 0 nm and one at 150 nm from the Au(100)//STO(100) interface. This was done to avoid ambiguity associated with the fact that the published reference is based on LaTiO3. However, the same analysis using the published reference is presented in the Supporting Information Figure S4. The depth profile of Ti3+ ratio η is shown in Figure 5. Figure 5 illustrates the Ti3+ concentration gradients near both interfaces, implying a ∼60% higher vacancy concentration at the Au(100)//STO(100) interface. The Ti3+ concentration gradients for the two interfaces decay with nearly the same rate and approach zero at 150 nm into the STO substrate. The measurements of Au(100)//STO(100) interface (Figure 5) beyond 100 nm in depth include multiple inelastic effects due to sample thickness. A comparison of Figure 2c and d with Figure 5 shows that a higher concentration of Ti3+ is correlated with a lower resistance. In the STO lattice, undercoordinated Ti is associated with a state 0.2 to 1 eV below the conduction band edge depending on the local atomic configuration.42 Related states exist at surfaces and grain boundaries.43 At the Au/STO interfaces, these states would provide an additional transport path via resonant tunneling, reducing the effective tunneling barrier width and the measured resistance (Figure 3c). In addition, at sufficiently high concentration, the electrons associated with these states within the depletion region locally increase the carrier concentration which would reduce the depletion width, also decreasing resistance (Figure 3d). Both of E
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Present Address †
Department of Chemical Engineering and Materials Science, University of California, Davis, California 95616, United States. Author Contributions ‡
W.Q. and J.H. contributed equally to this work. W.Q. was responsible for the TEM sample preparation and EELS characterization. J.H. conducted the topographic and electronic characterization by C-AFM. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the Department of Energy, Office of Basic Energy Science (DE-FG02-00ER45813). The Nano/ Bio Interface Center (grant DMR08-32802), the Laboratory for Research on the Structure of Matter (grant DMR11-20901), and the Nanoscale Characterization Facility are acknowledged for access to C-AFM, SEM, and STEM. We are grateful for the assistance of Douglas M. Yates in STEM and EELS characterization, as well as the discussion with Xiang Yang and Pavan Nukala about EELS analysis.
Figure 5. Depth profile of the Ti3+ fraction ratio for Au(100)// STO(100) (black) and Au(111)//STO(100) (red) interfaces. Bars indicate least-squares fitting errors.
these mechanisms are consistent with the experimental observations. Figure 5 clearly shows a large difference between the Ti3+ concentration at the two interfaces. Note that these interfaces are on the same substrate crystal and, therefore, experienced identical thermochemical histories. Hence, the differences in concentration cannot be attributed to variations in processing. The schematic illustration in Figure 1c and f indicates the atomic alignment at the interface. Regardless of the details of resulting structures, it is clear that a higher fraction of Au at the interface will be in closer proximity to O on the Au (100)// STO (100) interface than on the Au (111)//STO(100). Although Au does not hybridize easily with O to form competing bonds, the large electronegativity of Au will act to withdraw electron density from the Ti. If this effect were a factor, the Au(100)//STO(100) would exhibit a higher degree of Ti reduction at the interface, consistent with the observations. It has also been predicted that Au will couple with oxygen vacancies, which would also be consistent with the experimental results. Conclusion. The effect of interface structure on the orientation dependence of electronic properties at nanosized Au/STO interfaces is demonstrated, with reduced resistance at Au(100)//STO(100) compared to Au(111)//STO(100). These results show that the details of the atomic configuration of the metal play a strong role in the development of electronic defects near the interface. In conjunction with size dependent edge related tunneling, this effect provides an explanation for property variations that can be used in device design and optimization.
■
■
(1) Jin, S.; Whang, D.; McAlpine, M. C.; Friedman, R. S.; Wu, Y.; Lieber, C. M. Nano Lett. 2004, 4 (5), 915−919. (2) Waser, R.; Dittmann, R.; Staikov, G.; Szot, K. Adv. Mater. 2009, 21, 2632−2663. (3) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205−213. (4) Zhang, X.; Chen, Y. L.; Liu, R.-S.; Tsai, D. P. Rep. Prog. Phys. 2013, 76, 046401. (5) Kraya, R.; Kraya, L. Y.; Bonnell, D. A. Nano Lett. 2010, 10, 1224− 1228. (6) Shan, B.; Cho, K. Phys. Rev. B 2004, 70, 233405. (7) Yan, J.-A.; Yang, L.; Chou, M. Y. Phys. Rev. B 2007, 76, 115319. (8) Liu, X.; Wang, Y.; Burton, J. D.; Tsymbal, E. Y. Phys. Rev. B 2013, 88, 165139. (9) Hou, J.; Nonnenmann, S. S.; Qin, W.; Bonnell, D. A. Appl. Phys. Lett. 2013, 103, 252106. (10) Yang, J. J.; Pickett, M. D.; Li, X.; Ohlberg, D. A. A.; Stewart, D. R.; Williams, R. S. Nat. Nanotechnol. 2008, 3, 429−433. (11) Nishimura, T.; Kita, K.; Toriumi, A. Appl. Phys. Lett. 2007, 91, 123123. (12) Schowalter, L. J.; Lee, E. Y. Phys. Rev. B 1991, 43, 9308−9311. (13) Zhu, S.; Meirhaeghe, R. L. V.; Detavernier, C.; Cardon, F.; Ru, G.-P.; Qu, X.-P.; Li, B.-Z. Solid-State Electron. 2000, 44, 663−671. (14) Shimizu, T.; Usui, Y.; Nakagawa, T.; Okushi, H. J. Electroceram. 2000, 4, 299−303. (15) Chang, S.; Brillson, L. J.; Kime, Y. J.; Rioux, D. S.; Kirchner, P. D.; Pettit, G. D.; Woodall, J. M. Phys. Rev. Lett. 1990, 64, 2551−2554. (16) Gutknecht, P.; Strutt, M. J. O. Appl. Phys. Lett. 1972, 21, 405− 407. (17) Shimizu, T.; Gotoh, N.; Shinozaki, N.; Okushi, H. Appl. Surf. Sci. 1997, 117, 400−405. (18) Shimizu, T.; Okushi, H. J. Appl. Phys. 1999, 85, 7244−7251. (19) Tedesco, J. L.; Rowe, J. E.; Nemanich, R. J. J. Appl. Phys. 2009, 105, 083721. (20) Sullaphen, J.; Bogle, K.; Cheng, X.; Gregg, J. M.; Valanoor, N. Appl. Phys. Lett. 2012, 100, 203115. (21) Williams, D. B.; Carter, C. B. Transmission Electron Microscopy: A Textbook for Materials Science; Springer: New York, 2009. (22) Ohtomo, A.; Muller, D. A.; Grazul, J. L.; Hwang, H. Y. Nature 2002, 419, 378−380. (23) Muller, D. A.; Sorsch, T.; Moccio, S.; Baumann, F. H.; EvansLutterodt, K.; Timp, G. Nature 1999, 399, 758−761.
ASSOCIATED CONTENT
S Supporting Information *
Cross-section height profile of nanoparticles; orientation dependence of electronic response; Ti3+ fractional ratio. This material is available free of charge via the Internet at http:// pubs.acs.org.
■
REFERENCES
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. F
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(24) Fitting, L.; Thiel, S.; Schmehl, A.; Mannhart, J.; Muller, D. A. Ultramicroscopy 2006, 106, 1053−1061. (25) Browning, N. D.; Buban, J. P.; Moltaji, H. O.; Pennycook, S. J.; Duscher, G.; Johnson, K. D.; Rodrigues, R. P.; Dravid, V. P. Appl. Phys. Lett. 1999, 74, 2638−2640. (26) Hou, J.; Rouxel, B.; Qin, W.; Nonnenmann, S. S.; Bonnell, D. A. Nanotechnology 2013, 24, 395703. (27) Kawasaki, M.; Takahashi, K.; Maeda, T.; Tsuchiya, R.; Shinohara, M.; Ishiyama, O.; Yonezawa, T.; Yoshimoto, M.; Koinuma, H. Science 1994, 266, 1540−1542. (28) Shim, J.-H.; Lee, B.-J.; Cho, Y. W. Surf. Sci. 2002, 512, 262−268. (29) Francis, A. J.; Salvador, P. A. J. Mater. Res. 2007, 22, 89−102. (30) Silly, F.; Castell, M. R. Phys. Rev. Lett. 2006, 96, 086104. (31) Wang, C. M.; Shutthanandan, V.; Zhang, Y.; Thomas, L. E.; Baer, D. R.; Thevuthasan, S.; Duscher, G. J. Appl. Phys. 2004, 95, 5060−5068. (32) Wei, W.; Dai, Y.; Guo, M.; Huang, B. Appl. Surf. Sci. 2011, 257, 6607−6611. (33) Zhao, K. L.; Chen, D.; Li, D. X. Comput. Mater. Sci. 2010, 50, 98−104. (34) Hou, J.; Nonnenmann, S. S.; Qin, W.; Bonnell, D. A. Adv. Funct. Mater. 2014, 24, 4113−4118. (35) Rhoderick, E. H.; Williams, R. H. Metal-Semiconductor Contacts, 2nd ed.; Oxford University Press: New York, 1988; pp 89−116. (36) Brillson, L. J. Contacts to Semiconductors, Fundamentals and Technology; Noyes: Park Ridge, NJ, 1993; pp 216−218. (37) Smit, G. D. J.; Rogge, S.; Klapwijk, T. M. Appl. Phys. Lett. 2002, 81, 3852−3854. (38) Muller, D. A.; Nakagawa, N.; Ohtomo, A.; Grazul, J. L.; Hwang, H. Y. Nature 2004, 430, 657−661. (39) Zhang, Z.; Sigle, W.; Rühle, M. Phys. Rev. B 2002, 66, 094108. (40) Browning, N. D.; James, E. M.; Kishida, K.; Arslan, I.; Buban, J. P.; Zaborac, J. A.; Pennycook, S. J.; Xin, Y.; Duscher, G. Rev. Adv. Mater. Sci. 2000, 1, 1−26. (41) Tanaka, I.; Nakajima, T.; Kawai, J.; Adachi, H.; Gub, H.; Ruhleb, M. Philos. Mag. Lett. 1997, 75, 21−28. (42) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides; Cambridge University Press: New York, 1996; pp 200−208. (43) Bonnell, D. A.; Garra, J. Rep. Prog. Phys. 2008, 71, 044501.
G
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Effect of Interface Atomic Structure on the Electronic Properties of Nano-Sized Metal−Oxide Interfaces Wei Qin,†,‡ Jiechang Hou,‡ and Dawn A. Bonnell* Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States S Supporting Information *
ABSTRACT: We report that the size dependence of electronic properties at nanosized metal−semiconducting oxide interfaces is significantly affected by the interface atomic structure. The properties of interfaces with two orientations are compared over size range of 20−200 nm. The difference in interface atomic structure leads to electronic structure differences that alter electron transfer paths. Specifically, interfaces with a higher concentration of undercoordinated Ti result in enhanced tunneling due to the presence of defect states or locally reduced tunnel barrier widths. This effect is superimposed on the mechanisms of size dependent properties at such small scales. KEYWORDS: metal−semiconductor interface, electronic properties, interface structure, AFM, TEM, metal oxide
N
tronic properties of individual interfaces are quantified as a function of size and interface orientation using conductive atomic force microscopy (C-AFM).5,19,20 The electronic properties are compared to the interface atomic structure determined from high resolution transmission electron microscopy (HRTEM) and aspects of the electronic structure determined from electron energy loss spectroscopy (EELS).21 EELS has previously been used to characterize the valence states along SrTiO3/LaTiO3 superlattices,22 the electronic structures at SiOx/Si interfaces,23 as well as dislocation cores and grain boundaries.24,25 This combination of measurements enables the correlation between the interface structure resulting from the two ORs, the consequent differences in local atomic structural configurations, and the interface properties at sizes ranging from 20 to 150 nm. Experimental Procedures. SrTiO3 single crystalline substrates with Nb dopant concentration of 0.02 at% (Princeton Scientific, 5 × 5 × 0.5 mm3) were annealed at 1000 °C for 1 h after cleaning by acetone and ethanol. Nanosized Au/SrTiO3 contacts were produced by drop-casting gold nanoparticles (citrate stabilized in H2O solution, British Biocell) with various diameters (20, 40, 60, 100, 150 nm) on the atomically flat surfaces. After the solution was evaporated, samples with dispersed nanoparticles were annealed in ambient for an hour at 950 °C to remove residual organic compounds and obtain intimate interface contacts. An aluminum thin film (∼100 nm) was thermally evaporated on the back side of the samples to form Ohmic contacts, after which Au/STO/Al
anosized metal-oxide interfaces have become substantial and critical components in technologies, including electronic devices,1,2 energy harvesting,3 and catalysis.4 Continuous downscaling raises opportunities for new functionalities, as well as challenges in reliable performance. Many factors that are negligible when averaged at the macroscopic scale and even at the micron-scale, such as interfacial orientation,5−7 polarization,8 and local inhomogeneity,9,10 become crucial to the electronic properties of systems with nanoscale interfaces. Among these effects, interfacial atomic arrangement or orientation relations are nontrivial and can modulate interfacial electronic states. Because of the complex nature of metal− semiconductor interfaces, the orientation dependence of electronic properties and the associated mechanisms remain open issues. Some researchers argue that little or no orientation dependence is present, for example, on Au, Al/Ge,11 and Au/ Si12 interfaces, whereas others demonstrate orientation-based variations at CoSi2/Si,13 Au/SrTiO3,14 Au/GaAs15 and Au/ carbon nanotube interfaces.6 Possible mechanisms have been proposed for specific systems, generally based on metal-induced gap states or Fermi level pinning at the Schottky contact, resulting from different bonding interactions at the interfaces.6,11,16 However, the majority of this research focuses only on the semiconducting substrate orientation. The metal crystal orientations could also affect the interface properties. Here, we address the issue of interface structure dependence of electronic properties using Au nanoparticle/atomically flat SrTiO3 (STO) single crystalline contacts as a model system.17,18 Taking advantage of the behavior of nanoparticles, we produce two distinct orientation relations (ORs): Au(100)//STO(100) and Au(111)//STO(100). The elec© XXXX American Chemical Society
Received: September 3, 2014
A
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 1. AFM topographic images for the Au(100)//STO(100) interface (a) and the Au(111)//STO(100) interface (d). The substrate orientations are labeled by white arrows and the Au nanoparticle orientations are labeled by black arrows. The inset in (a) illustrates the square cap shape of the labeled octagonal particle under a larger height scale. The corresponding cross section TEM images for Au(100)//STO(100) (b) and Au(111)//STO(100) (e), with arrows displaying crystal orientations for Au and STO. (c) and (f) Illustration of the possible interfacial atomic arrangements for Au(100)//STO(100) and Au(111)//STO(100), respectively.
Images and spectra were processed using Digital Micrograph. All EELS spectra were background-subtracted by a power law and normalized by a thickness correction (about 20 eV above the threshold). The energy resolution is 0.8 eV and signals are averaged over the beam spot. Results and Discussion. The nanosized Au/Nb-doped STO contacts are dispersed on the single crystalline STO(100) substrate surfaces. The annealed substrates present atomically flat surfaces with steps along [100] directions and a height of ∼4 Å, consistent with the STO lattice constant.27 After nanoparticle deposition, the subsequent annealing facilitates particle faceting into truncated octahedra, a process driven by surface energy reduction.28 At thermal equilibrium, the truncated octahedra consist of six (100) and eight (111) surfaces. Because the Au(100)//STO(100) and Au(111)// STO(100) interface energies are comparable, these two orientations coexist experimentally29 and have been observed in TEM, AFM, and STM.5,30,31 Figure 1a displays the topographic image of particles with Au(100)//STO(100) interfaces, in which the particles resemble octagons with square caps (inset) in AFM images. The cross-sectional TEM image [Figure 1b] exhibits a clean intimate contact between the Au nanoparticle and STO substrate, without organic residue from surface ligands. Electron diffraction and lattice planes in the images show that Au[001] is parallel to STO[001], perpendicular to the interface plane as indicated by arrows in the figure, consistent with Wang et al.’s observation.31 In the plane of the interface Au [100] is also parallel to STO[100], as determined by comparison of the STO surface step edge directions and the Au surface plane directions. Similarly, the particles with Au(111)//STO(100)
stacked samples were silver-pasted on an atomic force microscopy (AFM) mount. Topographic images of samples were acquired in AC tapping mode (Asylum MFP-3D AFM). Aspects of the interfacial orientation relations were determined from the 2D lateral shape and the known substrate orientation, and the interface sizes were estimated from the height profiles. In contact mode (ORCA), a Pt-coated conductive tip (Olympus AC240TM) was engaged on individual nanoparticles. Only completely faceted particles with distinguishable shapes in topographic images were measured. Interface electronic properties were characterized from current−voltage (I−V) responses collected from −10 to 10 V for at least 8 cycles (0.2 Hz). Tip loading forces on the particle were optimized and limited to less than 10 nN.26 Following electrical measurement, the particles were rescanned and any particle exhibiting damage was not considered in the analysis. Over 100 interfaces were characterized. Orientations of some Au nanoparticles were identified using scanning electron microscopy (SEM). Cross-sectional TEM samples were extracted with the focused ion beam (FIB) liftout method using an FEI Strata DB235. A 5 kV low voltage ion beam was applied for final thinning to minimize the beam damage. The thicknesses of TEM samples were less than 50 nm at the Au/STO interface when measured by SEM. High resolution TEM bright field images were obtained in a JEOL 2100 and EELS data were collected in a JEOL 2010F equipped with an electron energy loss spectrometer (Gatan, GIF) under scanning transmission electron microscope (STEM) mode. Both TEMs were operated under 200 kV accelerating voltage and a vacuum of less than 3 × 10−5 Pa at room temperature. B
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 2. (a) Representative I−V curves to compare Au(100)//STO(100) and Au(111)//STO(100) interfaces. Arrows show the voltage ramping directions. The transition region of the HRS is circled. (b) Schematic illustration of sample structure and electronic characterization, where the conductive tip, sample, and built-in voltage source form a circuit. Comparison of the resistances of HRS (c) and LRS (d) for the two orientations Au(100)//STO(100) and Au(111)//STO(100) at −2.5 V. Error bars represent the deviation of measurement cycles. Interfacial size is represented by particle height.
2a shows representative I−V responses for the two orientations with 40 nm interfaces. Note, the plateaus in the high voltage regimes are set to 2 nA by amplifier protection, whereas those near 0 V are limited by current resolution (0.1 pA). The current loops divide the I−V curves into two branches: a high resistance state (HRS) with lower current and a low resistance state (LRS) with higher current. In our samples, positive bias switches the system from the HRS to the LRS, whereas negative bias switches from the LRS to the HRS. In the positive bias regime, the I−V curves exhibit no significant interface orientation dependence. Correspondingly, the Schottky properties and resistive switching responses of the two orientations are similar, consistent with our previous work.9,34 In the negative bias regime, however, the Au(100)//STO(100) interfaces are more conductive than are the Au(111)//STO(100) interfaces for both the LRS and the HRS branches. The Schottky model applies to this system and thermionic emission dominates the interfacial transport, as expressed by eq 135
interfaces resemble hexagons in the AFM image, as shown in Figure 1b. The cross-sectional TEM image [Figure 1e] specifies that Au[111] is perpendicular to the interface. In the plane of the interface, Au[110] is parallel to STO[100]. On all particles with sizes larger than that of the tip, the top surface plane was parallel to the STO surface plane, indicated negligible tilt. (See Supporting Information Figure S1) Though other possible orientation relations may coexist,30 we only discuss and compare these two orientations. The atomic structures of interfaces with these two orientations would necessarily differ, resulting in possible orientation-dependent electronic properties. Figure 1c and f schematically illustrate the different atomic alignments at the interfaces that would develop in the absence of relaxation and defects. The lateral alignment of the atoms at the interface has been predicted by first-principles to be that the Au sits above the O atom or bridges between two O atoms.32,33 Regardless, as illustrated in Figure 1c and f, alignment with either of these positions would be more efficient on the Au(100)//STO(100) interface than on the Au(111)//STO(100) interface. The implications of this are discussed below. The electronic properties of nanosized Au/STO Schottky contacts were characterized by conductive AFM for interface diameters ranging from 20 to 150 nm. In experiments, only the nanoparticle height D can be precisely measured. Because the interfacial diameter is proportional to the particle height, all interfacial sizes are represented by the measured particle height D.9 Specifically, properties were characterized for interface sizes 40, 60, 70, 80, 100, and 150 nm for the two orientations. Figure
I = AJ0 = AA*T 2e−qΦB / kBT (e qV / nkBT − 1)
(1)
where J0 is the current density, A is the interfacial area, T is the temperature, A* is the Richardson constant, ΦB is the Schottky barrier, and n is the ideality factor. Consequently, the reverse current saturates at a value of −J0. The absence of saturation has been observed in the macroscopic Schottky contacts, especially at a large reverse bias.35,36 The large leakage current is similar to the “breakdown” phenomenon, attributed to image C
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 3. (a) Band structure of the intrinsic junction of Au (yellow) and STO (purple). Positive charges accumulate at the STO surface, whereas the metal surface is negatively charged. The top arrow indicates the direction of the built-in electric field. Oxygen vacancies migrate from STO to interface on annealing. (b) Band structure under negative bias, with reduced STO Fermi level. (c) In-gap electron states associated with undercoordinated Ti enhance tunneling. (d) Sufficiently high concentration of undercoordinated Ti reduces the depletion width.
Figure 4. EELS spectra of Ti-L and O−K edges for Au(100)//STO(100) and Au(111)//STO(100) interfaces at different distances from the interface. (a) Ti-L edge at Au(100)//STO(100), (b) O−K edge at Au(100)//STO(100), (c) Ti-L edge at Au(111)//STO(100), and (d) O−K edge at Au(111)//STO(100). The distances from the interface are labeled in (a). Each color represents the same distance from the interface.
force lowering of Schottky barrier or edge-induced tunneling. Given that the reverse current enhancement is much larger than that caused by image force lowering (ΔΦB is less than 0.01 eV)
for our system, the edge-induced tunneling dominates the leakage current in the reverse bias. The edge effect is governed by the development of depletion regions beneath the particles. D
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
At small interface sizes, the depletion region is size dependent and decreases with decreasing interface size, that is, with a larger edge-to-area ratio.37 For intermediate sized interfaces the presence of the edge reduces the depletion width, locally favoring tunneling. This explains the size-related trend in Figure 2c and d. The electronic structure of the interface is shown in Figure 3. The difference in the work functions of Au and STO drives positive charges (oxygen vacancies or electron holes) in the STO toward the interface. Under negative bias, Figure 3b, the STO Fermi level shifts down and the effective tunneling barrier width decreases. Due to the reduced tunneling depletion width, the edge effect is more prominent under the reverse bias; therefore, the current under negative bias is more susceptible to tunneling than that in the positive regimes. Figure 2c and d compares the resistances for the Au(100)// STO(100) and Au(111)//STO(100) interfaces at a voltage −2.5 V, to demonstrate the variation of the electronic properties with orientation. Additional comparisons are included in Supporting Information Figure S2. For both the LRS and the HRS, the Au(111)//STO(100) interface exhibits higher resistances over the size range of 40−150 nm. The slope change [circled in Figure 2a] is apparently due to interface charging since it is a function of voltage ramping rate (see Supporting Information Figure S3). Regardless of this effect, the overall HRS resistance of the Au(111)//STO(100) is higher than that of Au(100)//STO(100). Charging is not observed for LRS branches, and the LRS resistance of the Au(100)//STO(100) interfaces is smaller than that of the Au(111)//STO(100) for all sizes. As Figure 1 suggests, a difference in interface atomic structure can result in differences in the local electronic structure, which in turn can influence interface properties. For the case of STO, the expected variations concern the local concentration of O and the associated coordination of the Ti. In the stoichiometric STO Ti is fully oxidized in the nominal 4+ state. At surfaces and in the presence of O vacancies, it is reduced. To compare the possible variations in local Ti coordination at and near the interfaces, EELS gradient profiles were collected. Figure 4a−d compares the EELS profiles of Ti L2,3 (Ti 2p → 3d) edges and O K (O 1s → 2p) edges for the two interface orientations as a function of distance from the interfaces. Spectra were obtained on interfaces with similar sizes (∼400 nm) and identical thermal history. For both samples, the spectra furthest from the interface exhibit 4 peaks and the trend is for the peaks to broaden and merge close to the interfaces, indicating significant concentration of reduced Ti. The Ti is more oxidized at and near the (111) than at the (100) interface. A concomitant “damping-out” of O K peaks is also observed from bulk to interface as indicated by reduced relative intensities of the O K edges. This is consistent with the presence of oxygen vacancies associated with undercoordinated Ti atoms.22,38,39 The peak intensity of the O edge is much weaker than that of the Ti edge, so we primarily rely on the Ti edge differences to discuss the chemical variation within the interface region. As labeled in Figure 4a, the two peaks in Ti L3 correspond to eg (∼457 eV) and t2g (∼459 eV) orbitals with the separation determined by the crystal field due to the local atomic structure. The broadened peaks closer to the interface are consistent with the crystal field variation induced by structural distortion of the TiO6 octahedra.38−41 Table 1 lists the eg and t2g peak positions measured from the EELS spectra as a function of distance from the interface. The separation of
Table 1. Ti L3 eg and t2g Peak Positions for Au(100)// STO(100) and Au(111)//STO(100) Orientation Relationsa Au(100)//STO(100) depth (nm) 1 3 5 10 15 20 25 30 a
eg (eV)
t2g (eV)
458.2 457.9 457.8 456.7 457.1 459 457.5 459.3 458 460 457.9 460.1
Au(111)//STO(100)
separation (eV)
eg (eV)
t2g (eV)
separation (eV)
0 0 0 0 1.9 1.8 2 2.2
457 457.6 457 456.9 456.5 457.1 457.5 457.4
458.9 459.8 459 459 458.5 459.2 459.6 459.7
1.9 2.2 2 2.1 2 2.1 2.1 2.3
Error of peak positions is ±0.1 eV.
the eg and t2g peaks decreases from ∼2.2 eV (@30 nm) to almost 0 eV at the interface for Au(100)//STO(100), whereas less change is observed for Au(111)//STO(100). To semiquantitatively compare the chemical defect gradients for the two orientations, the Ti3+ intensity ratios from the EELS profiles were extracted. Assuming that the spectra are a linear combination of reference spectra for the 4+ and 3+ states, the best fit for the combination of the references with the experimental data were determined.22 The ratios were calculated as f = af1 (Ti 3 +) + bf2 (Ti4 +)
η=
a a+b
(2)
where f is the experimental EELS spectrum; a and b are the fractional contribution of Ti3+ and Ti4+ valence states, respectively; and η is the Ti3+ fractional ratio. The analysis in Figure 4 used experimental end point spectra as internal references (f1 and f 2) of the Ti3+ and Ti4+ states, that is, one at 0 nm and one at 150 nm from the Au(100)//STO(100) interface. This was done to avoid ambiguity associated with the fact that the published reference is based on LaTiO3. However, the same analysis using the published reference is presented in the Supporting Information Figure S4. The depth profile of Ti3+ ratio η is shown in Figure 5. Figure 5 illustrates the Ti3+ concentration gradients near both interfaces, implying a ∼60% higher vacancy concentration at the Au(100)//STO(100) interface. The Ti3+ concentration gradients for the two interfaces decay with nearly the same rate and approach zero at 150 nm into the STO substrate. The measurements of Au(100)//STO(100) interface (Figure 5) beyond 100 nm in depth include multiple inelastic effects due to sample thickness. A comparison of Figure 2c and d with Figure 5 shows that a higher concentration of Ti3+ is correlated with a lower resistance. In the STO lattice, undercoordinated Ti is associated with a state 0.2 to 1 eV below the conduction band edge depending on the local atomic configuration.42 Related states exist at surfaces and grain boundaries.43 At the Au/STO interfaces, these states would provide an additional transport path via resonant tunneling, reducing the effective tunneling barrier width and the measured resistance (Figure 3c). In addition, at sufficiently high concentration, the electrons associated with these states within the depletion region locally increase the carrier concentration which would reduce the depletion width, also decreasing resistance (Figure 3d). Both of E
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Present Address †
Department of Chemical Engineering and Materials Science, University of California, Davis, California 95616, United States. Author Contributions ‡
W.Q. and J.H. contributed equally to this work. W.Q. was responsible for the TEM sample preparation and EELS characterization. J.H. conducted the topographic and electronic characterization by C-AFM. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the Department of Energy, Office of Basic Energy Science (DE-FG02-00ER45813). The Nano/ Bio Interface Center (grant DMR08-32802), the Laboratory for Research on the Structure of Matter (grant DMR11-20901), and the Nanoscale Characterization Facility are acknowledged for access to C-AFM, SEM, and STEM. We are grateful for the assistance of Douglas M. Yates in STEM and EELS characterization, as well as the discussion with Xiang Yang and Pavan Nukala about EELS analysis.
Figure 5. Depth profile of the Ti3+ fraction ratio for Au(100)// STO(100) (black) and Au(111)//STO(100) (red) interfaces. Bars indicate least-squares fitting errors.
these mechanisms are consistent with the experimental observations. Figure 5 clearly shows a large difference between the Ti3+ concentration at the two interfaces. Note that these interfaces are on the same substrate crystal and, therefore, experienced identical thermochemical histories. Hence, the differences in concentration cannot be attributed to variations in processing. The schematic illustration in Figure 1c and f indicates the atomic alignment at the interface. Regardless of the details of resulting structures, it is clear that a higher fraction of Au at the interface will be in closer proximity to O on the Au (100)// STO (100) interface than on the Au (111)//STO(100). Although Au does not hybridize easily with O to form competing bonds, the large electronegativity of Au will act to withdraw electron density from the Ti. If this effect were a factor, the Au(100)//STO(100) would exhibit a higher degree of Ti reduction at the interface, consistent with the observations. It has also been predicted that Au will couple with oxygen vacancies, which would also be consistent with the experimental results. Conclusion. The effect of interface structure on the orientation dependence of electronic properties at nanosized Au/STO interfaces is demonstrated, with reduced resistance at Au(100)//STO(100) compared to Au(111)//STO(100). These results show that the details of the atomic configuration of the metal play a strong role in the development of electronic defects near the interface. In conjunction with size dependent edge related tunneling, this effect provides an explanation for property variations that can be used in device design and optimization.
■
■
(1) Jin, S.; Whang, D.; McAlpine, M. C.; Friedman, R. S.; Wu, Y.; Lieber, C. M. Nano Lett. 2004, 4 (5), 915−919. (2) Waser, R.; Dittmann, R.; Staikov, G.; Szot, K. Adv. Mater. 2009, 21, 2632−2663. (3) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205−213. (4) Zhang, X.; Chen, Y. L.; Liu, R.-S.; Tsai, D. P. Rep. Prog. Phys. 2013, 76, 046401. (5) Kraya, R.; Kraya, L. Y.; Bonnell, D. A. Nano Lett. 2010, 10, 1224− 1228. (6) Shan, B.; Cho, K. Phys. Rev. B 2004, 70, 233405. (7) Yan, J.-A.; Yang, L.; Chou, M. Y. Phys. Rev. B 2007, 76, 115319. (8) Liu, X.; Wang, Y.; Burton, J. D.; Tsymbal, E. Y. Phys. Rev. B 2013, 88, 165139. (9) Hou, J.; Nonnenmann, S. S.; Qin, W.; Bonnell, D. A. Appl. Phys. Lett. 2013, 103, 252106. (10) Yang, J. J.; Pickett, M. D.; Li, X.; Ohlberg, D. A. A.; Stewart, D. R.; Williams, R. S. Nat. Nanotechnol. 2008, 3, 429−433. (11) Nishimura, T.; Kita, K.; Toriumi, A. Appl. Phys. Lett. 2007, 91, 123123. (12) Schowalter, L. J.; Lee, E. Y. Phys. Rev. B 1991, 43, 9308−9311. (13) Zhu, S.; Meirhaeghe, R. L. V.; Detavernier, C.; Cardon, F.; Ru, G.-P.; Qu, X.-P.; Li, B.-Z. Solid-State Electron. 2000, 44, 663−671. (14) Shimizu, T.; Usui, Y.; Nakagawa, T.; Okushi, H. J. Electroceram. 2000, 4, 299−303. (15) Chang, S.; Brillson, L. J.; Kime, Y. J.; Rioux, D. S.; Kirchner, P. D.; Pettit, G. D.; Woodall, J. M. Phys. Rev. Lett. 1990, 64, 2551−2554. (16) Gutknecht, P.; Strutt, M. J. O. Appl. Phys. Lett. 1972, 21, 405− 407. (17) Shimizu, T.; Gotoh, N.; Shinozaki, N.; Okushi, H. Appl. Surf. Sci. 1997, 117, 400−405. (18) Shimizu, T.; Okushi, H. J. Appl. Phys. 1999, 85, 7244−7251. (19) Tedesco, J. L.; Rowe, J. E.; Nemanich, R. J. J. Appl. Phys. 2009, 105, 083721. (20) Sullaphen, J.; Bogle, K.; Cheng, X.; Gregg, J. M.; Valanoor, N. Appl. Phys. Lett. 2012, 100, 203115. (21) Williams, D. B.; Carter, C. B. Transmission Electron Microscopy: A Textbook for Materials Science; Springer: New York, 2009. (22) Ohtomo, A.; Muller, D. A.; Grazul, J. L.; Hwang, H. Y. Nature 2002, 419, 378−380. (23) Muller, D. A.; Sorsch, T.; Moccio, S.; Baumann, F. H.; EvansLutterodt, K.; Timp, G. Nature 1999, 399, 758−761.
ASSOCIATED CONTENT
S Supporting Information *
Cross-section height profile of nanoparticles; orientation dependence of electronic response; Ti3+ fractional ratio. This material is available free of charge via the Internet at http:// pubs.acs.org.
■
REFERENCES
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. F
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(24) Fitting, L.; Thiel, S.; Schmehl, A.; Mannhart, J.; Muller, D. A. Ultramicroscopy 2006, 106, 1053−1061. (25) Browning, N. D.; Buban, J. P.; Moltaji, H. O.; Pennycook, S. J.; Duscher, G.; Johnson, K. D.; Rodrigues, R. P.; Dravid, V. P. Appl. Phys. Lett. 1999, 74, 2638−2640. (26) Hou, J.; Rouxel, B.; Qin, W.; Nonnenmann, S. S.; Bonnell, D. A. Nanotechnology 2013, 24, 395703. (27) Kawasaki, M.; Takahashi, K.; Maeda, T.; Tsuchiya, R.; Shinohara, M.; Ishiyama, O.; Yonezawa, T.; Yoshimoto, M.; Koinuma, H. Science 1994, 266, 1540−1542. (28) Shim, J.-H.; Lee, B.-J.; Cho, Y. W. Surf. Sci. 2002, 512, 262−268. (29) Francis, A. J.; Salvador, P. A. J. Mater. Res. 2007, 22, 89−102. (30) Silly, F.; Castell, M. R. Phys. Rev. Lett. 2006, 96, 086104. (31) Wang, C. M.; Shutthanandan, V.; Zhang, Y.; Thomas, L. E.; Baer, D. R.; Thevuthasan, S.; Duscher, G. J. Appl. Phys. 2004, 95, 5060−5068. (32) Wei, W.; Dai, Y.; Guo, M.; Huang, B. Appl. Surf. Sci. 2011, 257, 6607−6611. (33) Zhao, K. L.; Chen, D.; Li, D. X. Comput. Mater. Sci. 2010, 50, 98−104. (34) Hou, J.; Nonnenmann, S. S.; Qin, W.; Bonnell, D. A. Adv. Funct. Mater. 2014, 24, 4113−4118. (35) Rhoderick, E. H.; Williams, R. H. Metal-Semiconductor Contacts, 2nd ed.; Oxford University Press: New York, 1988; pp 89−116. (36) Brillson, L. J. Contacts to Semiconductors, Fundamentals and Technology; Noyes: Park Ridge, NJ, 1993; pp 216−218. (37) Smit, G. D. J.; Rogge, S.; Klapwijk, T. M. Appl. Phys. Lett. 2002, 81, 3852−3854. (38) Muller, D. A.; Nakagawa, N.; Ohtomo, A.; Grazul, J. L.; Hwang, H. Y. Nature 2004, 430, 657−661. (39) Zhang, Z.; Sigle, W.; Rühle, M. Phys. Rev. B 2002, 66, 094108. (40) Browning, N. D.; James, E. M.; Kishida, K.; Arslan, I.; Buban, J. P.; Zaborac, J. A.; Pennycook, S. J.; Xin, Y.; Duscher, G. Rev. Adv. Mater. Sci. 2000, 1, 1−26. (41) Tanaka, I.; Nakajima, T.; Kawai, J.; Adachi, H.; Gub, H.; Ruhleb, M. Philos. Mag. Lett. 1997, 75, 21−28. (42) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides; Cambridge University Press: New York, 1996; pp 200−208. (43) Bonnell, D. A.; Garra, J. Rep. Prog. Phys. 2008, 71, 044501.
G
dx.doi.org/10.1021/nl503389b | Nano Lett. XXXX, XXX, XXX−XXX
