Letter pubs.acs.org/NanoLett
Real-Time Observation of Reconstruction Dynamics on TiO2(001) Surface under Oxygen via an Environmental Transmission Electron Microscope Wentao Yuan,† Yong Wang,*,† Hengbo Li,† Hanglong Wu,† Ze Zhang,*,† Annabella Selloni,‡ and Chenghua Sun*,§ †
Center of Electron Microscopy and State Key Laboratory of Silicon Materials, School of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, China ‡ Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States § ARC Centre for Electromaterials Science, School of Chemistry, Monash University, Clayton, Victoria 3800, Australia S Supporting Information *
ABSTRACT: The surface atomic structure has a remarkable impact on the physical and chemical properties of metal oxides and has been studied extensively by scanning tunneling microscopy. However, acquiring real-time information on the formation and evolution of the surface structure remains a great challenge. Here we use environmental transmission electron microscopy to directly observe the stress-induced reconstruction dynamics on the (001) surface of anatase TiO2. Our in situ results unravel for the first time how the (1 × 4) reconstruction forms and how the metastable (1 × 3) and (1 × 5) patterns transform into the (1 × 4) surface stable structure. With the support of first-principles calculations, we find that the surface evolution is driven by both low coordinated atoms and surface stress. This work provides a complete picture of the structural evolution of TiO2(001) under oxygen atmosphere and paves the way for future studies of the reconstruction dynamics of other solid surfaces. KEYWORDS: Oxide surface, surface reconstruction, surface evolution, surface dynamics, in situ TEM, environmental transmission electron microscopy (ETEM)
W
resolve this issue in situ side-view of the reconstructed surface could provide the most convincing evidence as well as critical information regarding surface strain from the subsurface layers but unfortunately this cannot be obtained from STM and conventional transmission electron microscopy TEM.10,20−23 Furthermore, the detailed formation process and dynamics are poorly understood because the real-time evolution is difficult to record. For instance, (1 × 3) and (1 × 5) reconstructions, both of which have slightly lower stability with respect to the most favorable (1 × 4) one,18 have been observed experimentally14 but whether and how those different patterns finally transform to the (1 × 4) is not known. Therefore, techniques that can record the real-time evolution of surface and subsurface atoms are highly desirable when the dynamics of surface reconstruction is required. Environmental transmission electron microscopy (ETEM) can be a powerful tool for the study of surface reconstruction. Technically, the reconstruction can be monitored in situ by both side- and top-view ETEM images. This offers the
hen a crystal is cleaved to create a surface, a substantial rearrangement of surface atoms (reconstruction) often occurs in order to achieve higher stability. After reconstruction, surfaces may show remarkably different physical and chemical properties with respect to the bulk-truncated ones. Therefore, correctly determining reconstructed geometries and understanding how they form are of paramount importance in surface chemistry.1−3 Over the last decades, scanning tunneling microscopy (STM) has been successfully employed to characterize various reconstructions with atomic resolution.4−9 Even so, however, STM provides only a top view and cannot “see” the subsurface layers to acquire crucial information on the surface stress; more importantly, it remains a great challenge to secure a real-time observation of how the reconstruction forms by STM.10−12 A typical example is anatase TiO2. Because of its high surface energy, the bulk-truncated (001) surface is not stable and prefers to form a (1 × 4) reconstruction, as widely observed in STM studies.1,13−16 To describe the atomic geometry of the reconstructed structure, several models have been proposed, all of which are broadly consistent with the STM images.13,15,17,18 Among these, the “ad-molecule” model (ADM),18 originally proposed on the basis of first-principles calculations, is the most widely accepted one, even though some debate is still present among researchers.15,19 To fully © 2015 American Chemical Society
Received: August 16, 2015 Revised: November 25, 2015 Published: December 11, 2015 132
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
Letter
Nano Letters
Figure 1. (1 × 4) reconstructed (001) surface of the anatase TiO2 nanosheet, viewed along the [010] axis. (a) HRTEM images of unreconstructed TiO2 (001); (b) HRTEM image of the (1 × 4) reconstructed surface. (c) ADM model of the reconstructed surface with the TiOx row highlighted by the red circle. (d) Experimental HRTEM compared with simulated image based on ADM. (e) Enlarged side view for the first five layers. (f) The color-enhanced image of (e). (g) Intensity profiles along the colored dashed lines in (e).
Figure 2. Tracking the formation process of the (1 × 4) reconstruction, viewed from [010] direction. (a−d) Sequential HRTEM images of the anatase TiO2(001) surface during the reconstruction, acquired at 0, 34.9, 56.2, and 257.8 s, collected from Movie S2. The enlarged images of the dotted rectangles are shown in the lower panels of (a−d), respectively. (e,f) Intensity profiles along the dashed lines in the lower panels of (a−d) (matching colors). The orange and green lines are acquired from the reconstructed layer and top-surface layer, respectively.
capability to “see” the subsurface layers below the reconstruction at the atomic scale, thus providing valuable information regarding the surface strain and stress in these layers. An important challenge of these studies is, however, that surface structures can be easily damaged by the electron beam (ebeam) of the TEM. To overcome this difficulty, we have here introduced a protection gas (oxygen) in the ETEM chamber. We were then able to record the dynamic evolution of TiO2(001) as TEM movies. These movies allow us to observe not only the real-time transition between different patterns but also record the intermediates. Combined with computational calculations, the mechanism for the surface reconstruction can be investigated in great detail. Our study starts from “dirty” nanocrystals prepared by wet chemistry methods,24,25 rather than expensive single crystals typically used for STM measurements. TiO2 nanosheets, dominated by (001) surfaces, are synthesized with the use of hydrofluoride as the controlling agent; additional details are given in the Supporting Information and Figures S1 and S2. Initially, the TiO2 nanosheets are covered by an amorphous
layer introduced during the synthesis, as seen in Figure 1a. To clean the surface, a heat treatment is carried out and in our case the temperature is gradually increased to 500 °C in oxygen environment (5 × 10−2 Pa). Under these conditions, surface pollutants are removed with the aid of e-beam irradiation26 (the detailed mechanism is discussed in the Supporting Information), and the (1 × 4) reconstruction shows up gradually on the bare surface (Figure 1b and Movie S1). It should also be noted that early investigations reported that the (1 × 4) reconstruction can form on micrometer scale crystal surfaces under ultrahigh vacuum conditions, whereas the surface structure of nanometer scale crystals or in a gas environment have seldom been explored.1,16,27 Our experimental results confirm that such reconstruction is stable also on nanosized crystals in an oxygen environment. Several structural models have been proposed to explain the (1 × 4) reconstruction of the anatase (001) surface (Supporting Information Figure S3).13,15,17,18 To identify which model is most energetically favorable under our experimental conditions, we carried out density functional 133
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
Letter
Nano Letters
Figure 3. Atomic evolution of the (1 × n) reconstructions. (a) Sequential HRTEM images of the dynamic structural evolution, viewed from [010] direction, with the red arrows indicating the unstable states. (b) The statistical diagram of the locations of the TiOx rows in Movie S4 with green and red lines indicating the stable states and the unstable states. (c) Side view of the proposed model for the unstable two-row state with the TiOx row shown as ball-and-stick (Ti, gray; O, red) on the TiO2 stick framework. The green and red arrows indicate the stable single-row and instable doublerow structures, respectively. (d,e) Experimental HRTEM image compared with the simulated image based on the model in (c).
theory (DFT) calculations of the surface energies of the five most popular ones as a function of the oxygen chemical potential.28 As shown in Figure S4 of the Supporting Information, the ADM model (see Figure 1c) is the most stable one under our experimental conditions. This result is also supported by our simulated TEM image for the ADM, which agrees well with our experimental data (see Figure 1d). To better analyze the surface structure, in Figure 1e we show an enlarged TEM side-view image of the top layers of the nanosheet, with the first three layers indicated as X (orange line), Y (blue line), and Z (green line). A color-enhanced version of the same image is shown in Figure 1f, from which the distance between specific atoms can be measured readily, while the intensity profiles along the X-, Y- and Z-lines are reported in Figure 1g. From the X-profile, one can see that along the [100] (horizontal) direction in Figure 1e the distance between adjacent adsorbed TiOx rows is ∼1.52 nm, which agrees well with the distance of four times the (1 × 1) unit cell (∼1.51 nm).13 Compared to the perfect (1 × 1) surface or bulk structure (Supporting Information Figure S5), the three units between two ridges show a significant compression of ∼10.5% along the [100] direction (bulk structure, 1.14 nm; reconstructed structure, 1.02 nm), as indicated in Figure 1g. It has been reported that the bulk-truncated (001) surface has a high surface energy and large stress,18 which makes it unstable energetically. To release the surface stress, the surface reconstruction takes place and finally the (1 × 4) structure is formed. Therefore, the large compression that we observe is
indeed the result of surface stress relaxation, which provides direct experimental support for the identification of local stress as a driving force for the reconstruction of TiO2 (001).18,29 This periodical shrinkage and rearrangement can be clearly seen in the color-enhanced image of the same area (Figure 1f). It is of interest to note that only the outermost layer shows remarkable lattice distortion and no significant variation is observed in the underlying layers (see Y- and Z-lines in Figure 1g), indicating that the stress associated with the reconstruction is confined to the outmost layers. Figure 2a−d are snapshots from a TEM video (Movie S2; another example is shown in Supporting Information Figure S6 and Movie S3), displaying the dynamic process of the reconstruction. At the beginning, adsorbed layers are observed, covering the (001) surface. This adsorbed layer may consist of a mixture of amorphous organics and TiOx species, which are not distinguishable at this stage. After a few seconds of e-beam irradiation, amorphous organics are gradually removed (see Figure 2a, here set as t = 0 s). At t = 34.9 s, the amorphous organic species are almost completely removed, as indicated by TEM and confirmed by electron energy loss spectroscopy (see Supporting Information Figures S7 and S8), and a crystalline layer appears on the (001) surface, corresponding to adsorbed Ti or TiOx species (Figure 2b). Here, it should be noted that it is rather difficult to distinguish the Ti or TiOx in the TEM images based on phase contrast. We note that because the titanium atoms can be rapidly oxidized and the surface oxidation of pure metal titanium can even occur below −100 134
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
Letter
Nano Letters
Figure 4. Mechanism of the full process of surface reconstruction. (a) Proposed five-step reconstruction mechanism, starting with (1) randomly adsorbed TiOx species, to (2) forming small row/islands, and (3) forming regular patterns (3d as an example), (4) partially transformed state (PTS), and (5) completely transformed to 4d patterns. Adsorbed TiO2 molecules and a part of surface oxygen are shown as balls (O, red; Ti, gray), and the major TiO2 substrate is shown as lines. (b) Schematic energy changes associated with surface evolution with DBs and surface stress identified as the driving force for two stages. (c) Side-view of the computational model used to investigate the 3d-to-4d transition with the reaction coordinator, doo, labeled. (d) Calculated energy profile for the 3d-to-4d transition with doo changing from 1.28 to 1.57 nm.
rightmost 3d transforming into 4d at t = 178.47 s, finally leading to the 4d−4d−4d pattern. With respect to (1), the rightmost TiOx row has been shifted to the right by one unit with continuous contrast change. Interestingly, the leftmost TiOx row in (1) consists of a double-row structure, a feature that has been frequently observed in the sequential TEM images (indicated by a red arrow). Such double-row structures often change to single rows rapidly, indicating they are unstable. It is worthwhile to mention that these double-row structures show lighter contrast in the TEM images while the contrast of the joint single row becomes much darker (refer to panels (7− 15) in Figure 3), which is similar to what observed in Figure 2. This result further confirms that the outmost crystalline layer in Figure 2b is a TiOx layer. Such information is crucial for the development of the model discussed below. In Figure 3b, we report unstable double-rows (double red lines) and stable single-rows (single green lines) as a function of time. We can see that the unstable double-row structure is an intermediate state appearing during the transformation from (1 × n) to (1 × 4) reconstruction. A model for such double-row structure is presented in Figure 3c (side view). Its characteristic feature is that the adsorbed TiOx species are split to partially occupy two rows. This yields a simulated TEM image (Figure 3e), which is in good agreement with the experimental data (Figure 3d), thus validating the proposed structure. Such double-row structures have also been frequently observed during the transformation from (1 × 5) to (1 × 4) patterns (Supporting Information Figure S10, Movie S5) and confirmed by top view images of the reconstructed surface (Figure S11 and S12, Movie S6 and S7 in the Supporting Information). On the basis of the above in situ observations, a schematic mechanism for the surface reconstruction can be proposed, as illustrated in Figure 4a: (1) TiOx species are initially adsorbed on the surface in a random manner with many dangling bonds (DBs) (also refer to Figure 2b for the side-view image); (2) TiOx species tend to form islands or small rows to reduce the number of DBs; (3) ordered rows are formed with 3d, 4d, and 5d patterns and all DBs removed; (4) various patterns tend to transform to the most favorable 4d pattern but the trans-
°C,30 it is reasonable to expect the outmost layer is TiOx at our experimental conditions (500 °C and oxygen environment). Furthermore, we performed DFT-based molecular dynamics calculations of single Ti atom on anatase TiO2 (001) surface in oxygen environment and confirmed that the outmost Ti could be easily oxidized even at 0 K, which indicates that the outermost layer in our experiments should be TiOx (see Supporting Information Figure S9). More discussion can be found in the Supporting Information. From the side view, bright spots are also evident (see red arrow), indicating a distortion within the top layers. After t = 56.2 s, adsorbed species begin to exhibit nonuniform contrast; for example, the left-most gray dot in Figure 2b disappears and its adjacent dot turns darker, which suggests that the TiOx in the leftmost row migrated to its adjacent row. Therefore, under the same imaging conditions the lattice dot (projection of a single row along the zone axis) will become darker if any TiOx is added into the row. At the same time, the bright spots in the second layer of Figure 2b also become darker, indicating that the surface layer is actively involved during the surface relaxation. At t = 257.8 s, adsorbed species finally form the 4× periodicity (Figure 2d), which remains stable during the following reaction. The above evolution is reflected in the intensity profiles for the first two surface layers shown in Figure 2e,f, respectively. We can see that the uniform contrast at reconstructing locations shows a gradual change to strong contrast, which can be utilized to analyze the intermediate states involved in the formation process. Such crucial information is unique for in situ TEM and is hardly achievable through other surface analytic methods. Previous studies have mainly focused on characterizing the atomic structure of reconstructed surfaces without considering the early stages of the reconstruction process. Now real time ETEM can record in situ information from the initial stages of the process to the final reconstructed surface. A sequence of snapshots from Movie S4 is shown in Figure 3a. Initially, there are four TiOx rows (or ridges) on the surface, forming a 3d− 3d−4d pattern (“d” indicating the periodicity of the 1 × 1 bulkterminated surface). Figure 3a(1−7) show the evolution from 3d−3d−4d to 4d−4d−3d, while panels (7−15) show the 135
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
Letter
Nano Letters formation may be incomplete, resulting in a “partially transformed state” (PTS) (also refer to Figure 3a for the side-view images and Supporting Information Figure S11 for the top-view images); and (5) (1 × 4) reconstruction finally forms. Key steps of the above mechanism are summarized in Figure 4b. These can be divided in two stages, steps (1−3) dominated by the tendency to reduce the number of DBs, and (3−5), driven by surface stress.18 In terms of the dynamics, removing DBs can greatly reduce the total energy and stabilize the systems, so steps (1−3) are fast. Starting from (3), the surface evolution is driven by the difference of surface stress between the (1 × n) and (1 × 4) reconstructions. This effect is significantly weaker than that of DBs, and thus steps (3−5) are much slower. The PTS, as shown in Figure 4a(4), is a typical state during the slow stage of the transformation. As demonstrated by our experiments, the PTS can survive long enough to be observed by TEM. PTS and completed (1 × n) reconstruction are featured as double- and single-row, respectively, which can be easily identified from TEM images (side view). To roughly estimate the barrier for the transformation from 3d to 4d patterns, we calculated the energy profile as a function of the distance doo between the oxygen atoms of two neighboring adsorbed TiOx rows (Figure 4c). According to our tests, the energy profiles are affected by the length of the TiO2 rows, which experimentally can vary from one row to another. Here we considered the simplest case of infinitely long rows (see details in the Supporting Information, Figure S13), which gives a barrier of 0.6 eV for the 3d-to-4d transition without PTS (Figure 4d). This barrier should be considered as an upper limit. If larger models are employed so that the PTS (Figure 4a(4)) can be incorporated, most likely a smaller barrier will be obtained. Overall, transition barriers between different (1 × n) reconstructions are expected to be relatively small, indicating that such transitions can happen readily, which can explain why PTSs have been frequently observed in our experiments. To achieve high quality TEM movies for the surface reconstruction, a protection gas, oxygen in this work, is essential. When nitrogen or vacuum is employed, the surface cannot be maintained for long enough time (Figure S14 and S15, and Movie S8 in the Supporting Information). However, we should note that the reconstruction is not induced by oxygen. Indeed, such reconstruction can also be observed in nitrogen or vacuum at elevated temperature (Figure S15 and S16, Movie S8 in the Supporting Information). In addition, more accurate real-time information can be secured with higher-time resolution camera, which may turn ETEM as a powerful tool to investigate dynamic phase transition and even surface reactions of nanocrystals. It is worth to mention that the ETEM can be used with easy-to-prepare (“dirty”) samples, and different surfaces can be investigated using one sample (e.g., TiO2 (001) and (101) in Figure 1a). Therefore, it is particularly valuable for studying those highly reactive surfaces, which often diminish so rapidly that they only exist with a small percentage in the crystals dominated by majority surfaces. In summary, we have presented in situ atomic scale ETEM observations of the formation and evolution of the (1 × 4) reconstruction on the anatase TiO2 (001) surface. Through analysis of top and side views, not only has the real-time dynamics for the transition from metastable (1 × 3) and (1 × 5) to (1 × 4) been revealed but also unstable intermediate states have been observed and identified. Combined with DFT calculations, both lowly coordinated atoms and surface stress
have been identified as the driving force for the surface evolution. We anticipate that this in situ real-time technique can be readily applied to study the dynamic formation and evolution of surface reconstructions on other systems.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b03277. Materials and methods, detailed information about the surface reconstruction and DFT calculation, and effect of the e-beam and oxygen in the experiments. (PDF) The in-situ formation of the (1×4) reconstruction on anatase TiO2 (001) surface observed from the [010] direction. (AVI) The in-situ formation of the (1×4) reconstruction on anatase TiO2 (001) surface observed from the [010] direction. (AVI) The in-situ formation and evolution of the reconstruction on anatase TiO2 (001) surface observed from the [010] direction. (AVI) The in-situ observation of the reconstructing anatase TiO2 (001) surface from the [010] direction (3d to 4d). (AVI) The in-situ observation of the reconstructing anatase TiO2 (001) surface from the [010] direction (5d to 4d). (AVI) The in-situ observation of the reconstructing anatase TiO2 (001) surface from a top view. (AVI) The in-situ observation of the reconstructing anatase TiO2 (001) surface from a top view. (AVI) In situ observation of the dissolution of anatase TiO2 nanosheet in N2 atmosphere. (AVI)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected]. *E-mail:
[email protected]. *E-mail:
[email protected]. Author Contributions
Y.W. and Z.Z. proposed the project. Y.W. conceived and designed the experiments. C.S. was responsible for the theoretical calculations and model development. Y.W., Z.Z., and C.S. supervised the overall research. W.Y., H.L., and H.W. performed the in situ ETEM experiments. A.S. contributed to the data analysis. W.Y., Y.W., and C.S. wrote the manuscript. All the authors participated in discussions of the research and the revision of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge the support of National Natural Science Foundation of China (51390474, 11234011, 11327901), the Ministry of Education of China (IRT13037) and National Young 1000 Talents Program of China. Y.W. thanks Shengbai Zhang at RPI for helpful discussion. C.S. acknowledges the financial support from ARC Discover Project (DP130100268) and Future Fellowship (FT130100076). A.S. thanks the support of DoE-BES, Chemical Sciences, Geosciences and Biosciences Division, Contract No. DE-FG02-12ER16286. C.S. 136
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
Letter
Nano Letters also appreciates the generous grants of CPU time from Australian National Computational Infrastructure.
■
REFERENCES
(1) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (2) Lin, Y.; Wu, Z.; Wen, J.; Overbury, S. H.; Poeppelmeier, K. R.; Ellis, D. E.; Marks, L. D. Nano Lett. 2014, 14, 191. (3) Diebold, U.; Li, S. C.; Schmid, M. Annu. Rev. Phys. Chem. 2010, 61, 129. (4) Besenbacher, F.; Lægsgaard, E.; Stensgaard, I. Mater. Today 2005, 8, 26. (5) Dulub, O.; Batzill, M.; Solovev, S.; Loginova, E.; Alchagirov, A.; et al. Science 2007, 317, 1052. (6) Merte, L. R.; Peng, G.; Bechstein, R.; Rieboldt, F.; Farberow, C. A.; et al. Science 2012, 336, 889. (7) Setvin, M.; Aschauer, U.; Scheiber, P.; Li, Y. F.; Hou, W. Y.; et al. Science 2013, 341, 988. (8) Wahlstrom, E.; Vestergaard, E. K.; Schaub, R.; Ronnau; Vestergaard, A. M.; et al. Science 2004, 303, 511. (9) Monig, H.; Todorovic, M.; Baykara, M. Z.; Schwendemann, T. C.; Rodrigo, L.; et al. ACS Nano 2013, 7, 10233. (10) Subramanian, A.; Marks, L. D. Ultramicroscopy 2004, 98, 151. (11) Diebold, U. Nat. Mater. 2010, 9, 185. (12) Chiaramonti, A. N.; Marks, L. D. J. Mater. Res. 2005, 20, 1619. (13) Liang, Y.; Gan, S. P.; Chambers, S. A.; Altman, E. I. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 235402. (14) Tanner, R. E.; Sasahara, A.; Liang, Y.; Altman, E. I.; Onishi, H. J. Phys. Chem. B 2002, 106, 8211. (15) Wang, Y.; Sun, H. J.; Tan, S. J.; Feng, H.; Cheng, Z. W.; et al. Nat. Commun. 2013, 4, 2214. (16) Liu, G.; Yang, H. G.; Pan, J.; Yang, Y. Q.; Lu, G. Q.; et al. Chem. Rev. 2014, 114, 9559. (17) Herman, G. S.; Sievers, M. R.; Gao, Y. Phys. Rev. Lett. 2000, 84, 3354. (18) Lazzeri, M.; Selloni, A. Phys. Rev. Lett. 2001, 87, 266105. (19) Ohsawa, T.; Lyubinetsky, I. V.; Henderson, M. A.; Chambers, S. A. J. Phys. Chem. C 2008, 112, 20050. (20) He, M. R.; Yu, R.; Zhu, J. Nano Lett. 2012, 12, 704. (21) Zhu, G. Z.; Radtke, G.; Botton, G. A. Nature 2012, 490, 384. (22) Enterkin, J. A.; Subramanian, A. K.; Russell, B. C.; Castell, M. R.; Poeppelmeier, K. R.; et al. Nat. Mater. 2010, 9, 245. (23) Shibata, N.; Goto, A.; Choi, S.; Mizoguchi, T.; Findlay, S. D.; et al. Science 2008, 322, 570. (24) Wen, C. Z.; Jiang, H. B.; Qiao, S. Z.; Yang, H. G.; Lu, G. Q. J. Mater. Chem. 2011, 21, 7052. (25) Wu, X.; Chen, Z. G.; Lu, G. Q.; Wang, L. Z. Adv. Funct. Mater. 2011, 21, 4167. (26) Egerton, R. F.; Li, P.; Malac, M. Micron 2004, 35, 399. (27) Selcuk, S.; Selloni, A. J. Phys. Chem. C 2013, 117, 6358. (28) Reuter, K.; Scheffler, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 65, 035406. (29) Ibach, H. Surf. Sci. Rep. 1997, 29, 195. (30) Lu, G.; Bernasek, S. L.; Schwartz, J. Surf. Sci. 2000, 458, 80.
137
DOI: 10.1021/acs.nanolett.5b03277 Nano Lett. 2016, 16, 132−137
