MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

A

CXXXX American Chemical Society

Tunneling Desorption of SingleHydrogen on the Surfaceof Titanium DioxideTaketoshi Minato,*,†,‡,O Seiji Kajita,*,§,) Chi-Lun Pang, ) Naoki Asao,^ Yoshinori Yamamoto,^

Takashi Nakayama,§ Maki Kawai,*,# and Yousoo Kim*,‡

†International Advanced Research and Education Organization, Tohoku University, Sendai 980-8578, Japan, ‡Surface and Interface Science Laboratory, RIKEN,2-1 Hirosawa, Saitama 351-0198, Japan, §Department of Physics, Chiba University, 1-33 Yayoi, Inage, Chiba 263-0022, Japan, )Department of Chemistry, UniversityCollege London, London WC1H 0AJ, United Kingdom, ^WPI Advanced Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Sendai 980-8577, Japan,and #Department of Advanced Materials Science, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan. OPresent address: T.M.: Office ofSociety-Academia Collaboration for Innovation, Kyoto University, Uji-shi, Kyoto 611-0011, Japan. )Present address: Toyota Central R&D Labs, Inc., 41-1 Yokomichi,Nagakute Aichi 480-1192, Japan.

Electric conductivity, photophysicalproperties, magnetic properties, andcatalytic activity can be produced

by introduction of defects onto inertmaterials.1�8 Clarifying the fundamentalproperties of such defects is therefore ofextraordinary interest in developing newmaterials. Typical defects on titanium diox-ide (TiO2) include hydrogen (H) adatoms,oxygen vacancies (Ovac), and Ti interstitials,each of which confer new properties absentin the perfect material. TiO2 is a transitionmetal oxide showing a number of charac-teristic functions that can be exploited inprocesses such as heterogeneous catalysis,photocatalysis, sensing, and light-inducedswitching, among others.1�10

Defects have been created at TiO2 sur-faces by thermal annealing11 as well as byphoto- and electron-stimulated desorption(PSD and ESD).1,12,13 The principles of thesemethods are based on thermal reaction toproduceO2 in the former case and interionicAuger decay processes in the latter twoexamples.1,11�13 While these methods have

been successfully used to introduce defectson the TiO2 surface, precise control of thedefect arrangement is not possible. Thecontrol of the defect arrangement on TiO2

has been reported by applying specificvoltages with a scanning tunneling micro-scope (STM). For example, Suzuki et al.found that hydrogen atoms on TiO2 canbe removed by scanning at a raised bias,14

and we have recently reported the intro-duction of an individual Ovac by applying asingle voltage pulse.15 While manipulationof defects by STM methods allows for idealspatial control of the defect, the mecha-nisms are still unclear.In the manipulation of atoms/molecules

by STM, the mechanisms can be catego-rized into electron excitation and nonelec-tron excitation processes. In electron excita-tion reactions, tunneling electrons exciteelectronic states that are related to chemicalbonds [1, electronic state excitation]16 orvibrational states (including translational, ro-tational, and conformational change states)to overcome reaction barriers by exciting

* Address correspondence tominato.taketoshi.5x@kyoto-u.ac.jp,fine-controller@mosk.tytlabs.co.jp,maki@k.u-tokyo.ac.jp,ykim@riken.jp.

Received for review December 5, 2014and accepted June 2, 2015.

Published online10.1021/acsnano.5b01607

ABSTRACT We investigated the reaction mechanism of the desorption of single hydrogen from a

titanium dioxide surface excited by the tip of a scanning tunneling microscope (STM). Analysis of the

desorption yield, in combination with theoretical calculations, indicates the crucial role played by the

applied electric field. Instead of facilitating desorption by reducing the barrier height, the applied electric

field causes a reduction in the barrier width, which, when coupled with the electron excitation induced by

the STM tip, leads to the tunneling desorption of the hydrogen. A significant reduction in the desorption

yield was observed when deuterium was used instead of hydrogen, providing further support for the

tunneling-desorption mechanism.

KEYWORDS: defect . manipulation . scanning tunneling microscopy . titanium dioxide

ARTIC

LE

MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

B

reaction coordinate modes [2, vibrational stateexcitation].17�19 Also, heatingbyelectrons causes reactionin electron excitation [3, manipulation by local heating].20

In nonelectron excitation reactions, direct interactionbetween the STM tip and atoms/molecules by means ofvan der Waals/chemical forces [4, direct manipulation]21

or electric field to reduce the height of the reaction barrier[5, electric field excitation]22�24 have been utilizedto manipulate atoms/molecules. Here, we provide insightinto such reaction mechanisms for manipulation of de-fects on TiO2 by using action spectroscopy measuredby STM (STM-AS), whereby the response of a moleculeis recorded as a function of applied bias voltage.16�19 Hdesorption on TiO2 was chosen as a model systembecause it is the simplest reaction. We found that Hdesorption induced by the STM is a tunneling reactionfacilitated by the reduction of the barrier width by theapplied electric field together with excitation by thetunneling electrons. Although tunneling reactions ofatoms/molecules induced by STM tips are reported,25,26

this reaction mechanism whereby the barrier width isreduced by the applied electric field has not been ob-servedpreviously. Apart frombeing an important reactionmechanism in other H/TiO2 systems, such reaction me-chanisms are also likely to be important in other atomicscale reactions as well.

RESULTS AND DISCUSSION

Figure 1a shows a typical STM image of H on TiO2(110)obtained with sample bias (Vs) =þ1.5 V. The bright rowsand the spot at the center of the image are the Ti ionsand H on bridging oxygen (Ob), respectively.

1�8,14,15

After applying a Vs = þ1.7 V pulse to the H for 1 s, thespot corresponding to H disappeared due to desorptionof H (Figure 1b). In contrast, no change occurred whenvoltage pulses were applied with negative sample bias(up to Vs = �5.0 V). These observations are in line withprevious reports.1�8,14,15,27�29

In order to investigate the reaction mechanism forH desorption, STM-ASwasmeasured.When the tip wasfixed at Vs =þ1.0 V and It = 0.02 nA (with no additionaltip displacement (TD = 0 nm)), the obtained spectrumshowed two clear threshold energies at Vs = þ1.4 andþ1.7 V (blue filled squares and curve in Figure 2a).Displacing the tip closer to the TiO2 surface leads toa shift of the thresholds toward lower energy. Forexample, when the tip was moved toward the surface(TD = �0.04 nm), the threshold energies were shiftedto Vs =þ1.3 andþ1.6 V (green filled triangles and curvein Figure 2a), and with further displacement of the tipto �0.10 nm, the threshold energies were also furthershifted to Vs =þ1.2 andþ1.5 V (purple filled circles andcurve in Figure 2a). Further shifts of the thresholdenergies were not observed for displacements of�0.14 and �0.20 nm (orange filled squares and blackfilled triangles in Figure 2a). In contrast, displacementof the tip away from the surface to þ0.1 nm led to a

shift of the threshold energy originally at Vs =þ1.7 V toVs = þ2.0 V (red filled circles and curve).30 Figure 2bsummarizes the threshold energy shifts as a functionof tip displacement. Clearly, the shift of the thresholdenergies depends on the tip displacement (TD be-tween þ0.1 and �0.1 nm) with saturation of the shiftin the closer region (TD less than �0.1 nm).Based on our STM-AS results, we consider possible

mechanisms for the voltage pulse-induced desorptionof H: [1] electronic state excitation of reaction coordi-nate modes,16 [2] multiple vibrational state excita-tion for overcoming the reaction barrier,17�19,25,26 [3]thermal reaction caused by local heating,20 [4] directmanipulation by means of van der Waals/chemicalforces,21 and [5] electric field excitation for reductionof the reaction barrier.22�24 To assess the possibility ofelectronic state excitation of antibonding states, dI/dVmeasurements were performed on H/TiO2(110) as afunction of TD. No electronic states were observed indI/dV curves that correspond with the threshold en-ergies observed in STM-AS (Figure S1). Also, the shift ofthe threshold energies (≈3.0 V/nm in�0.1 toþ0.1 nmof TD) in STM-AS did not match with the shift in dI/dVspectra (≈6.0 V/nm) that is caused by tip induced bandbending (TIBB) (Figure S1). This implies that the reac-tion cannot be caused by electronic state excitation.Next, we consider the possibility of multiple vibrationalexcitation to overcome the reaction barrier. Fromdensity functional theory (DFT) calculations, the energybarrier of H desorption was obtained as 3.7 eV (opencircles in Figure 3). To overcome the reaction barrierby vibrational state excitation, multiple excitationis necessary (for example, the vibrational energy ofstretching mode of OH on TiO2(110) is 0.457 eV(3690 cm�1)31). However, the tunneling current depen-dence of the reaction yield indicates a one-electronprocess (Figure S2). Therefore, this reaction cannot beexplained by multiple vibrational state excitation.32 Inaddition, the possibility of local heating and directmanipulation by means of van der Waals/chemical

Figure 1. STM images of TiO2(110) (a) before and (b) after Hdesorption (2.7 nm � 3.0 nm, Vs = þ1.5 V, It = 0.3 nA). (c, d)Schematic images of H desorption on TiO2(110).

ARTIC

LE

MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

C

forces can be eliminated because the thermal energy isnot changed by tip position and the van der Waals/chemical forces are not dependent on the appliedelectric field. Finally, the possibility of electric fieldexcitation was considered. We estimated the electricfield generated in our experimental condition by carry-ing out simulations using a tip�sample distance of0.4 nm at Vs = þ1.0 V,35 with the SEMTIP programdeveloped by Feenstra.36 The electric field for thethreshold energies of the shifted region (�0.1 < TD <þ0.1 nm) was estimated to be 3.25�4.00 V/nm. How-ever, the reaction barrier obtained by the DFT calcula-tion under the electric field (3.6 eV at 4.5 V/nm; blackfilled circles in Figure 3) is not lower than the thermalactivation energy estimated from the Arrhenius equa-tion using the experimental condition (0.15�0.19 eVat 78 K assuming that reaction rates and frequencyfactor are 1�100 s�1 and 1 � 1012 s�1, respectively37).We conclude therefore that the reaction cannot becaused by electric field excitation. Based on the abovediscussion, we conclude that the reaction cannot becaused by any one of these excitation processes.Then, how does this reaction proceed? From the

calculated potential curve under an electric field, it wasfound that while the height of the reaction barrier does

not change significantly, either side of the maximumpoint (i.e., thepoints indicatedby redarrows in Figure3)decreases as a result of the applied field. The energygain in the potential curve is caused by ionization ofthe desorbed H and variability of the valence state andbond distance of Ti and O in the applied electric field(Supporting Information section 4). This means thebarrier width is reduced by the electric field and thismay open up a tunneling channel for the desorption ofH. In addition to this, the tunneling electrons injected bythe STM tip also excite the adsorbed hydrogen intoelectronic or vibrational states part way up the energybarrier at which point the barrier width is narrower,thereby further increasing the tunneling probability.From the calculation of potential curves with and with-out applied electric field, therefore it is suggested thatSTM tip-inducedHdesorption is causedbyaH tunnelingreaction (for the effects of the tip on the reaction barrier,see Supporting Information section 5).To test this tunneling desorption mechanism

further, we used STM-AS to measure any isotope effecton the reaction yield. STM-AS of D desorption fromTiO2(110) showed a significant decrease (≈10�4) in thereaction yield at Vs = þ1.8 V (black filled square andcurve in Figure 4a) compared with H desorption (bluefilled square and curve in Figure 4a). A theoreticalmodel was used to quantitatively explain the isotopicshift of the tunneling desorption. H on the Ob of TiO2 ispositively charged, with the hydrogen being attachedto the O site through a bond with some ionic char-acter.37,38 The hydrogen feels an electric potential�ReEx from the STM tip when the sample voltage ispositive, where R is a constant showing the valency ofhydrogen, e is the elementary charge, E is the electricfield, and x is the position of the hydrogen. Wemodeled the system as an energy-potential well re-presented by the potential curve shown in Figure S4.We assume that the hydrogen receives energy, ε, fromthe tunneling electrons flowing from the STM tip to the

Figure 2. (a) STM-AS of H desorption on TiO2(110). Spectrum with blue filled squares is obtained with start parametersVs = þ1.0 V, It = 0.02 nA and with no tip displacement (i.e., TD = 0 nm). All other spectra have the same start parameters butwith different TD. Red filled circles: TD =þ0.1 nm. Green filled triangles: TD =�0.04 nm. Orange filled circles: TD =�0.10 nm.Purple filled squares: TD = �0.14 nm. Black filled triangles: TD = �0.20 nm. (b) Threshold energy observed in STM-ASdepending on TD.

Figure 3. Potential energy of H as a function of the dis-placement from the surface of TiO2(110) under an electricfield as calculated using DFT. Open circles, 0.0 V/nm; filledcircles, 4.5 V/nm.

ARTIC

LE

MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

D

sample. Using the Wentzel�Kramers�Brillouin (WKB)approximation,39 the tunneling probability for thehydrogen desorption denoted by T is expressed by

T ¼ exp �8πffiffiffiffiffiffiffi2m

p

3heE(V0 � ε)3=2

� �(1)

where h is the Planck constant, m is the mass ofhydrogen, V0 is the reaction barrier (for detail, seeSupporting Information section 6). Equation 1 gives thedependence of T onm, so that with the mass of D beingtwice that of H, the multiplier factor of the exponentialwill increase by

√2 given that the isotopic element does

not change electronic properties such as V0. Figure 4ashows the H desorption yield spectrum together withthe same spectrum replotted with the electric fieldmultiplied by

√2. The replotted spectrum (blue open

squares and curve in Figure 4a) appears to reproducethat obtained for D desorption (black filled squares andcurve in Figure 4a), thus confirming that STM tip-inducedHdesorption on TiO2(110) is indeeddue to the proposedtunneling reaction which results from the narrowing ofthe barrier width by the electric field together withexcitation by the tunneling electron (Figure 4b).Previously, Acharya et al. reported a vibrational

excitation mechanism for H desorption on TiO2(110)based on their STM-AS measurements.40 They ob-served (A) no threshold energy, (B) no isotope effectson the reaction yield, (C) no tip position dependence ofthe reaction yield, and (D) a reaction order≈1.7, whichwe did not observe in our experiments. Observation (A)might be caused by different energy resolutions be-tween the different experiments (current detectioncircuit, vibration and scanner noise, etc.), whereas

observations (B)�(D) are probably caused by differenttip positions in our experiments. The STM-AS experi-ments in ref 40 were measured with Vs = þ1.7 V andIt = 0.5�5 nA. In our experiments, the tunnelingreaction was observed with Vs = þ1.0 V andIt = 0.001�0.5 nA (TD = þ0.1 to �0.1 nm). Thus,in the earlier work,40 a much closer tip sample dis-tance was used and in that regime, H desorption iscaused by vibrational state excitation leading to ob-servations (B)�(D) in ref 40. We note that when wedisplace the tip toward the sample by more than�0.1 nm (TD = �0.1 to �0.2 nm, Figure 2), we alsodid not observe any further shift of STM-AS.The saturation of the shift as the tip approaches

closer to the surface and the origin of the two thresholdenergies in STM-AS (Figure 2) are not completelyunderstood so far. The former is probably related tothe formation of a chemical bond of the tip with H onTiO2(110) and the latter probably related to the electricfield that narrows the width of the desorption barrier(see Supporting Information section 7).

CONCLUSION

We have investigated the reaction mechanism ofthe desorption of single hydrogen on the surfaceof TiO2(110) by STM-AS and DFT calculation underan electric field. Our results clearly shows the newreaction mechanism of defect control on TiO2(110)by a narrowing of the desorption barrier due to theelectric field, together with electron excitation. Under-standing the role of the tunneling reactions of H maybe important for the application of TiO2 and relatedmaterials in electronic devices.1�8,41�43

METHODSThe experiments were performed with a low temperature

STM (Omicron GmbH) housed in an ultrahigh vacuum chamber

(base pressure: 4� 10�9 Pa). The TiO2(110) samples (Shinkosha,

Co. Ltd.) were cleaned by cycles of Arþ sputtering (1 keV, 10 μA

for 10min) and annealing (900 K for 10 min). The density of Ovac

Figure 4. (a) STM-AS measurements of the desorption of single H (blue filled squares) and single D (black filled squares)from TiO2(110). STM-AS results for H are also replotted with the applied electric field multiplied by

√2 (blue open squares).

Initial tip position was Vs = þ1.0 V, It = 0.02 nA. (b) Schematic energy diagram for H desorption on TiO2(110).

ARTIC

LE

MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

E

on pristine TiO2(110) was 9%. After cleaning, a small amount(1� 10�8 Pa for 3 s) of H2O (or D2O) was dosed in the chamberto form H (or D) on the surface. All STM results were obtained at78 K with electrochemically etched tungsten tips.To analyze the reactionmechanism, we usedDFT calculations

of the charged surface under an electric field. The DFT calcula-tions were performed using the Tokyo ab initio programpackage (TAPP).44 We employed the Perdew�Burke�Ernzefhof(PBE) functional as the exchange correlation term,45 ultrasoftpseudopotentials,46 and a plane wave basis set. In order toexamine the effect of an electric field, the field-induced charge-sheets method was implemented in the DFT calculations.47 Inthis method, electrons are subtracted from the slab to realizethe positively charged surfaces, and negatively charged countersheets are inserted in the vacuum regions to maintain overallcharge neutrality. The TiO2(110) surface is modeled by a re-peated (1� 2) slab made of five TiO2 trilayers. Single H adsorbson the surface O site of the slab.

Conflict of Interest: The authors declare no competingfinancial interest.

Acknowledgment. This work was supported by Grant No.21750002 from Ministry of Education, Culture, Sports, Scienceand Technology of Japan, the Sasakawa Scientific ResearchGrant from The Japan Science Society, Kurata Grants of Hitachifoundation, and Grant for Exploratory Research for YoungScientists of Tohoku University. C.-L.P.'s visit to RIKEN wassupported by a JSPS BRIDGE Fellowship. T.M. and C.-L.P. per-formed STM experiments with support by N.A., Y.Y., and M.K.,and Y.K. S.K. and T.N. performed theoretical calculations. T.M.,C.-L.P., and S.K. wrote the manuscript.

Supporting Information Available: Experimental detailsand theoretical methods, electronic structure of H/TiO2(110),reaction order of H desorption on TiO2(110), reduction of thebarrier width for H desorption on TiO2(110), effects of theSTM tip on the desorption of H from TiO2(110), tunnelingprobability of H desorption and saturation of the energy shift,and the origin of the threshold energies in the STM-AS mea-surements. The Supporting Information is available free ofcharge on the ACS Publications website at DOI: 10.1021/acsnano.5b01607.

REFERENCES AND NOTES1. Henrich, V. E. The Surface Science of Metal Oxide; Cam-

bridge University Press: Cambridge, 1994.2. Diebold, U. The Surface Science of Titanium Dioxide. Surf.

Sci. Rep. 2003, 48, 53–229.3. Thompson, T.; Yates, J. T., Jr. Surface Science Studies of the

Photoactivation of TiO2-New Photochemical Processes.Chem. Rev. 2006, 106, 4428–4453.

4. Pang, C. L.; Lindsay, R.; Thornton, G. Structure of Clean andAdsorbate-Covered Single-Crystal Rutile TiO2 surfaces.Chem. Rev. 2013, 113, 3887–3948.

5. Dohnálek, Z.; Lyubinetsky, I.; Rousseau, R. Thermally-Driven Processes on Rutile TiO2(110)-(1� 1): A Direct Viewat the Atomic Scale. Prog. Surf. Sci. 2010, 85, 161–205.

6. Petek, H.; Zhao, J. Ultrafast Interfacial Proton-CoupledElectron Transfer. Chem. Rev. 2010, 110, 7082–7099.

7. Henderson, M. A. A Surface Science Perspective on TiO2

Photocatalysis. Surf. Sci. Rep. 2011, 66, 185–297.8. Minato, T. Atomic Defects in Titanium Dioxide. Chem. Rec.

2014, 14, 923–934.9. Minato, T.; Susaki, T.; Shiraki, S.; Kato, H. S.; Kawai, M.; Aika,

K.-i. Investigation of the Electronic Interaction betweenTiO2(110) Surfaces and Au Clusters by PES and STM. Surf.Sci. 2004, 566–568, 1012–1017.

10. Minato, T.; Izumi, Y.; Aika, K.-i.; Ishiguro, A.; Nakajima, T.;Wakatsuki, Y. Nitric Oxide Reduction by Carbon Monoxideover Supported Hexaruthenium Cluster Catalysts. 1. TheActive Site Structure That Depends on Supporting MetalOxide and Catalytic Reaction Conditions. J. Phys. Chem. B2003, 107, 9022–9028.

11. Henrich, V. E.; Dresselhaus, G.; Zeiger, H. J. Observation of2-Dimensinal Phases Associated with Defect States onSurface of TiO2. Phys. Rev. Lett. 1976, 36, 1335–1339.

12. Knotek, M. L.; Feibelman, P. J. Ion Desorption by Core-HoleAuger Decay. Phys. Rev. Lett. 1978, 40, 964–967.

13. Zhang, Z.; Cao, K.; Yates, J. T., Jr. Defect-Electron Spreadingon the TiO2(110) Semiconductor Surface byWater Adsorp-tion. J. Phys. Chem. Lett. 2013, 4, 674–679.

14. Suzuki, S.; Fukui, K.; Onishi, H.; Iwasawa, Y. HydrogenAdatoms on TiO2(110)-(1 � 1) Characterized by ScanningTunneling Microscopy and Electron Stimulated Desorp-tion. Phys. Rev. Lett. 2000, 84, 2156–2159.

15. Minato, T.; Kawai, M.; Kim, Y. Creation of Single OxygenVacancy on Titanium Dioxide Surface. J. Mater. Res. 2011,27, 2237–2240.

16. Lastapis, M.; Martin, M.; Riedel, D.; Hellner, L.; Comtet, G.;Dujardin, G. Picometer-Scale Electronic Control of Molec-ular Dynamics inside a Single Molecule. Science 2005, 308,1000–1003.

17. Stipe, B. C.; Rezaei, M. A.; Ho, W. Inducing and Viewing theRotational Motion of a Single Molecule. Science 1998, 279,1907–1909.

18. Shen, T. C.; Wang, C.; Abeln, G. C.; Tucker, J. R.; Lyding, J. W.;Avouris, P.; Walkup, R. E. Atomic-Scale Desorption throughElectronic and Vibrational-Excitation Mechanisms. Science1995, 268, 1590–1592.

19. Komeda, T.; Kim, Y.; Kawai, M.; Persson, B. N. J.; Ueba, H.Lateral Hopping of Molecules induced by Excitation ofInternal Vibration Mode. Science 2002, 295, 2055–2058.

20. Eigler, D. M.; Lutz, C. P.; Rudge, W. E. An Atomic SwitchRealized with the Scanning Tunneling Microscope. Nature1991, 352, 600–603.

21. Gimzewski, J. K.; Joachim, C. Nanoscale Science of SingleMolecules using Local Probes. Science 1999, 283, 1683–1688.

22. Carpinelli, J. M.; Swartzentruber, B. S. Direct Measurementof Field Effects on Surface Diffusion. Phys. Rev. B 1998, 58,R13423–13425.

23. Mo, Y. W. Reversible Rotation of Antimony Dimers on theSilicon (001) Surface with a Scanning Tunneling Micro-scope. Science 1993, 261, 886–888.

24. Alemani, M.; Peters, M. V.; Hecht, S.; Rieder, K.-H.;Moresco, F.; Grill, L. Electric Field-induced Isomerizationof Azobenzene by STM. J. Am. Chem. Soc. 2006, 128,14446–14447.

25. Stroscio, J. A.; Celotta, R. J. Controlling the Dynamics of aSingle Atom in Lateral Atom Manipulation. Science 2004,306, 242–247.

26. Kumagai, T.; Kaizu, M.; Hatta, S.; Okuyama, H.; Aruga, T.Direct Observation of Hydrogen-Bond Exchange within aSingle Water Dimer. Phys. Rev. Lett. 2008, 100, 166101.

27. Wendt, S.; Schaub, R.; Matthiesen, J.; Vestergaard, E. K.;Wahlstrom, E.; Rasmussen, M. D.; Thostrup, P.; Molina, L. M.;Laegsgaard, E.; Stensgaard, I.; Hammer, B.; Besenbacher, F.Oxygen Vacancies on TiO2(110) and their Interaction withH2O and O2: A Combined High-Resolution STM and DFTStudy. Surf. Sci. 2005, 598, 226–245.

28. Lackinger, M.; Janson, M. S.; Ho, W. Localized Interactionof Single Porphyrin Molecules with Oxygen Vacancies onTiO2(110). J. Chem. Phys. 2012, 137, 234707.

29. Minato, T.; Asao, N.; Yamamoto, Y.; Kawai, M.; Kim, Y.Photoresponse on the Desorption of an Atomic Hydrogenon Titanium Dioxide Surface Induced by a Tip of ScanningTunneling Microscope. Chem. Lett. 2013, 42, 942–943.

30. An additional threshold at lower energy could not beobserved because of the low tunneling current.

31. Henderson, M. A. An HREELS and TPD study of water onTiO2(110): The Extent of Molecular versus DissociativeAdsorption. Surf. Sci. 1996, 355, 151–166.

32. DFT calculations using the GGA functional underestimatethe energy barrier.33,34 The true energy barrier should behigher than 3.5 eV.

33. Lindan, P. J. D.; Zhang, C. Exothermic Water Dissociationon the rutile TiO2(110) Surface. Phys. Rev. B 2005, 72,075439.

ARTIC

LE

MINATO ET AL. VOL. XXX ’ NO. XX ’ 000–000 ’ XXXX

www.acsnano.org

F

34. Jursic, B. S. Density Functional Theory Study of RadicalHydrogen Abstraction with Hydrogen and Hydroxyl Radi-cals. Chem. Phys. Lett. 1996, 256, 603–608.

35. Minato, T.; Sainoo, Y.; Kim, Y.; Kato, H. S.; Aika, K.-i.; Kawai,M.; Zhao, J.; Petek, H.; Huang, T.; He, W.; Wang, B.; Wang, Z.;Zhao, Y.; Yang, J.; Hou, J. G. The Electronic Structure ofOxygen Atom Vacancy and Hydroxyl Impurity Defects onTitanium Dioxide (110) Surface. J. Chem. Phys. 2009, 130,124502.

36. Feenstra, R. M. Electrostatic Potential for a HyperbolicProbe Tip near a Semiconductor. J. Vac. Sci. Technol., B2003, 21, 2080–2088.

37. Kajita, S.; Minato, T.; Kato, H. S.; Kawai, M.; Nakayama, T.First-Principles Calculations of Hydrogen Diffusionon Rutile TiO2(110) Surfaces. J. Chem. Phys. 2007, 127,104709.

38. Zhao, J.; Li, B.; Jordan, K. D.; Yang, J.; Petek, H. InterplayBetween Hydrogen Bonding and Electron Solvation onHydrated TiO2(110). Phys. Rev. B 2006, 73, 195309 1–10.

39. Bohm, D. Quantum Theory; Dover: New York, 1989.40. Acharya, D. P.; Ciobanu, C. V.; Camillone, N.; Sutter, P.

Mechanism of Electron-Induced Hydrogen Desorptionfrom Hydroxylated Rutile TiO2(110). J. Phys. Chem. C2010, 114, 21510–21515.

41. Spahr, E. J.; Wen, L.; Stavola, M.; Boatner, L. A.; Feldman,L. C.; Tolk, N. H.; Lupke, G. Giant Enhancement of HydrogenTransport in Rutile TiO2 at Low Temperatures. Phys. Rev.Lett. 2010, 104, 205901.

42. Varghese, O. K.; Gong, D.; Paulose, M.; Ong, K. G.; Grimes,C. A. Hydrogen Sensing using Titania Nanotubes. Sens.Actuators, B 2003, 93, 338.

43. Gopel, W.; Rocker, G.; Feierabend, R. Intrinsic defects ofTiO2(110): Interaction with chemisorbed O2, H2, CO, andCO2. Phys. Rev. B 1983, 28, 3427–3438.

44. Yamauchi, J.; Tsukada, M.; Watanabe, S.; Sugino, O. First-Principles Study on Energetics of c-BN(001) ReconstructedSurfaces. Phys. Rev. B 1996, 54, 5586–5603.

45. Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized GradientApproximation made Simple. Phys. Rev. Lett. 1996, 77,3865–3868.

46. Vanderbilt, D. Soft Self-Consistent Pseudopotentials in aGeneralized Eigenvalue Formalism. Phys. Rev. B 1990, 41,7892–7895.

47. Kajita, S.; Nakayama, T.; Kawai, M. Ab-initio CalculationMethod for Electronic Structures of Charged Surfacesusing Repeated Slab and Density-Variable Charge Sheets.J. Phys. Soc. Jpn. 2007, 76, 044701.

ARTIC

LE

1

Supplemental Information for

Tunneling Desorption of Single Hydrogen on

the Surface of Titanium Dioxide

Taketoshi Minato1, 2

, Seiji Kajita3 Chi-Lun Pang

4, Naoki Asao

5,

Yoshinori Yamamoto5, Takashi Nakayama

3, Maki Kawai

6, Yousoo Kim

2

1International Advanced Research and Education Organization, Tohoku University, Sendai

980-8578, Japan. 2Surface and Interface Science Laboratory, RIKEN, 2-1 Hirosawa, Saitama, 351-0198, Japan.

3Department of Physics, Chiba University, 1-33 Yayoi, Inage, Chiba 263-8522, Japan.

4Department of Chemistry, University College London,WC1H 0AJ, United Kingdom.

5WPI Advanced Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Sendai

980-8577, Japan. 6Department of Advanced Materials Science, The University of Tokyo, 5-1-5 Kashiwanoha,

Kashiwa, Chiba 277-8561, Japan.

2

§1. Experimental details and theoretical methods

The experiments were performed using a low-temperature STM system (Omicron GmbH)

consisting of two chambers separated by a gate valve. TiO2(110) single crystals (Shinkosha Co.

Ltd.) were transferred to the preparation chamber without any pre-treatment. The TiO2(110)

samples were cleaned in the preparation chamber (base pressure: 7 × 10−7

Pa) via cycles of Ar+

sputtering (at 1 keV and 10 μA for 10 min) and annealing (at 900 K for 10 min). After cleaning,

the TiO2(110) was transferred to the STM chamber (base pressure: 4 × 10-9

Pa), where it was

placed on a gas-dosing stage. A small amount (1 × 10-8

Pa for 3 sec) of H2O (or D2O) was

introduced to the dosing stage at room temperature to allow H (or D) to form on the surface. After

H2O (or D2O) exposure, the TiO2(110) was transferred to the STM stage, which was maintained at

78 K. All STM results were obtained using electrochemically etched tungsten tips.

To record the action spectroscopy measured via STM (STM-AS), the STM tip was placed

above the H (or D) on the TiO2(110) surface while an electrical bias was applied. The tunneling

current (I) was monitored using an oscilloscope while the bias was applied. The I signal decreased

rapidly when desorption of H (or D) from the TiO2(110) surface occurred. The reaction yields (Y)

and reaction rates (R) were calculated as follows:

Y = 𝑒

I × t

R = 1

t

where e is the elementary charge and t is the duration of the applied electric bias, calculated from

the onset of the applied bias until desorption. The error bars of Y (Yerror), R (Rerror) and I (Ierror) in

Figure 2 and Figure S2 were calculated as follows:

Yerror = ±�1

N Yi − Yav .

2

N

i=1

Rerror = ±�1

N Ri − Rav .

2

N

i=1

Ierror = ±�1

N Ii − Iav .

2

N

i=1

where N is the total number of data; Yi, Ri and Ii are the values of Y, R, and I in each

measurement; and Yav., Rav., and Iav. are the average values of Y, R, and I.

DFT calculations were performed using the Tokyo Ab-initio Program Package (TAPP)1.

We employed the Perdew-Burke-Ernzefhof (PBE) functional as the exchange-correlation term2,

ultrasoft pseudopotentials to describe the ionic cores3 and a plane-wave basis set with a cut-off

energy of 25 Ryd to expand the wave functions of the valence electrons. The TiO2(110) surface

was modeled using a repeated (1 × 2) slab composed of five TiO2 trilayers separated by a vacuum

space equivalent to 1.3 nm. A single H was adsorbed onto the surface O site of the slab. A (4 × 4 ×

3

1) Monkhorst-Pack mesh was used for k-point sampling in the Brillouin zone4. The slab was

relaxed until the force on each atom was below 0.005 Hartree/a.u. with the center trilayer of the

slab held frozen in the optimized geometry. The reliability of the parameters was confirmed by

evaluating the lattice parameters, and the discrepancy with respect to the experimental values was

found to be less than 3%. The field-induced charge-sheets (FICS) method was implemented in the

DFT calculations to examine the effect of an applied electric field5. In this method, electrons are

subtracted from the slab to model positively charged surfaces, and negatively charged counter

sheets are inserted into the vacuum regions to maintain overall charge neutrality. Unrealistic

interactions between the repeated charged slabs are effectively removed by artificially inserting

counter sheets into vacuum regions because the density of the counter sheets varies to screen the

electric field of the charged slab. Thus, one can accurately calculate the total energy and electronic

states of charged surfaces. The FICS method offers several advantages in terms of the stability of

the convergence in the electronic states, its ease of implementation and its accuracy compared

with other methods of charged-surface calculation6-8

. We set a parameter of width of the counter

sheets at 0.5 Bohr.

4

§2. Electronic structure of H/TiO2(110)

Figure S1 presents scanning tunneling spectra (STS) of H/TiO2(110) for a range of different

sample-tip distances. On TiO2(110), which is an n-type semiconductor, STS spectra acquired with

positive sample bias exhibit an energy shift as a function of the sample-tip distance because of

tip-induced band bending9-12

. No electronic states corresponding to the threshold energies

observed in the STM-AS results (Fig. 2) are observable. This finding demonstrates that the

STM-tip-induced H desorption from TiO2(110) is not caused by the excitation of an anti-bonding

state.

(dI/

dV

)/(I

/V)

[arb

. u

nit

]

1.61.20.80.4

sample bias [V]

+ + + +

Figure S1. Scanning tunneling spectra obtained on H/TiO2(110) for various tip-sample

distances. Blue curve: tip displacement (TD) = 0 nm (Vs = +1.0 V and It = 0.02 nA); green

curve: TD = -0.04 nm; yellow curve: TD = -0.10 nm; gray curve: TD = -0.14 nm; purple curve:

TD = -0.20 nm. All spectra were normalized by dividing by I/V.

5

§3. Reaction order of H desorption on TiO2(110)

The reaction barrier for H desorption on TiO2(110) is 3.7 eV (Fig. 3). Multiple excitations are

necessary to overcome the reaction barrier via vibrational excitation because no single vibrational

excitation is higher than 3.7 eV (e.g., the OH stretching vibration of H/TiO2(110) is only 0.457

eV12

). Figure S2 presents the reaction rates and reaction yields of hydrogen desorption from the

TiO2(110) surface at Vs = +1.6 V. From these results, the reaction order (N) of this reaction was

found to be 1.0571. This means that the reaction occurs via single-electron excitation and that the

desorption of H from TiO2(110) is not induced by multiple vibrational excitations.

102

2

4

6

103

2

4

6

104

2

Rea

ctio

n r

ate

[s-1

]

4 6 8

12 4 6 8

102

Tunneling current [nA]

N = 1.0571

(a)

10-8

2

4

6

810

-7

2

4

Rea

ctio

n y

ield

[p

er e

lect

ron

]

4 6 8

12 4 6 8

102

Tunneling curent [nA]

(b)

Figure S2. (a) Reaction rates and (b) reaction yields of H desorption from the TiO2(110) surface

obtained at Vs = +1.6 eV.

6

§4. Reduction of the barrier width for H desorption on TiO2(110)

When an electric field (positive sample bias) is applied to TiO2(110), the width of the

reaction barrier for H desorption is reduced, as indicated by the arrows in Fig. 3. These features

can be explained by the following mechanisms due to the two effects of the electronic field on the

potential curve.

In the neutral state of the hydrogenated TiO2(110) surface, the H on the Ob site forms

(H-Ob)-, which indicates an excess electron associated with the H is provided to the surface. Based

on charge-distribution analysis (Fig. S3a), this excess electron redistributes onto Ti sites beneath

the Ob sites, and consequently it weakens the ionic bonds between Ti4+

and Ob2-

. In the process of

the H desorption, the Ob is also displaced toward the vacuum to keep the form (H-Ob) unit.

When a positive electric field is applied on the surface, in the region around the +0.30 nm

(or larger than +0.30 nm) displacement of the H in Fig. 3, the protonated H is completely

separated from the Ob2-

. The desorption H takes an energy gain by the positive electrostatic filed.

This energy gain accounts for the reduction of the barrier width in this region. In the region around

the +0.15 nm (or shorter than +0.15 nm) displacement of the H, the positively charged surface

decreases the amount of the excess electron on the Ti sites, because the excess electron lies nearest

to the Fermi level of the H/TiO2(110) system (as a typical example, in case of no displacement of

H is indicated in Fig. S3b). This decrease of the excess electrons causes a recovery in the stiffness

of the weakened Ob-Ti ion bond; that is, the Ti sites are able to stretch more to follow the

desorbing (H-Ob) unit. Indeed, Fig. S4 presents that the elongation of bond length between Ti and

Ob of ObH in the H desorption process is suppressed by the positively charged effect. The

flexibility of the Ti sites in the desorption process enhances the electrostatic energy gain because

the Ti4+

sites is close to the counter STM tip that is charged up negatively. Thus, the barrier width

is reduced by applying a positive charge on the TiO2 surface.

Fig. S3. Distribution of excess electron (a) in the neutral state and (b) in applying 4.5 V/nm

electric field on H/TiO2(110). In (b), the density of the excess electron is decreased. These figures

are obtained by subtraction of the electron distribution of the clean TiO2 surface from that of the

H/TiO2 surface. The cross section in (a) and (b) go along the Ob line in the surface normal

direction, that is denoted by the horizontal line in (c).

7

Fig. S4. Variations of the bond length between Ti and Ob of ObH in the H desorption process.

Open circles: 0.0 V/nm; filled circles: 4.5 V/nm.

8

§5. Effects of the STM tip on the desorption of H from TiO2(110)

The H that desorbs from the TiO2(110) adsorbs on the STM tip. Therefore, the effects of the

tip on the potential curve for H desorption must be considered. As the tip approaches the TiO2(110)

surface, the potential curves for H on TiO2 and on the tip begin to cross. The crossing of the

potential curves decreases the height of the reaction barrier for H desorption, thus increasing the

tunneling probability for H desorption. However, if the effect of the tip on the potential curve only

determines the reaction barrier, then the threshold energies observed in the STM-AS

measurements (Fig. 2) at a fixed tip-sample distance should be triggered by the excitation of

electronic or vibrational states by the tunneling electrons. As shown in Fig. S1, no electronic states

were observed in the STS results at the STM-AS threshold energies. As for vibrational excitation,

the observed threshold energies (1.2 – 2.0 eV = 9679 – 16131 cm-1

) are too large for the

vibrational states of H/TiO2(110). These results indicate that the reaction barrier is not only

determined by the effects of STM tip on the potential curve of the H desorption on TiO2(110).

9

§6. Tunneling probability of H desorption

The model from which Eq. 1 is derived is explained in detail here. For an isotopic analysis of

the tunneling desorption, we used a simple energy-potential well model under an applied electric

field, as illustrated in Fig. S5. In this model, the hydrogen on TiO2(110) is a positive ion. We

assume that the protonated hydrogen has a charge of +αe and is bound within a potential well, in

which the potential inside the TiO2(110) is represented by an infinite wall and there is an energy

barrier of V0, representing the desorption energy, at x = 0. In the region of x > 0, an electric field,

E, is applied between the TiO2(110) and the STM tip. We assume that the hydrogen receives

energy, ε, from the tunneling electrons from the STM tip and that the received energy is converted

into the excitation of hydrogen vibrations. Here, we use a classical limit of quantum theory called

the Wentzel-Kramers-Brillouin (WKB) approximation13

. The probability of the hydrogen

tunneling toward x > 0 can be written as follows:

T = exp �−4π

ℎ 𝑑𝑥 2𝑚 −𝛼𝑒𝐸𝑥 + V0 − ε

𝑎

0

− (𝑆1)

where h is the Planck constant, m is the mass of hydrogen, and a is its position after the tunneling

reaction. Equation S1 can be solved analytically as follows:

T = exp�−8π 2𝑚

3ℎ𝛼𝑒𝐸(𝑉0 − ε)

32 .

This equation is the same as Eq. 1 in the main text.

Figure S5. Schematic potential-energy diagram used in the estimation of the tunneling probability

for hydrogen desorption. The yellow ball represents the terminal hydrogen on TiO2(110).

10

§7. Saturation of the energy shift and the origin of the threshold energies in the

STM-AS measurements

The saturation of the shift in the STM-AS threshold energies as the tip approaches the surface

(Fig. 2) cannot be completely understood thus far. One possible explanation is that as the STM tip

(tungsten (W)) approaches the sample, the tip makes a chemical bond with the H on the TiO2(110).

As the tip continues to draw closer, the W-H-O bond distorts. This distortion may not affect the H

desorption behavior. Therefore, the STM-AS measurements do not shift as the tip continues to

approach.

Moreover, the origins of the two threshold energies observed in the STM-AS results also

remain unclear. The threshold energies observed at a fixed tip-sample distance (Fig. 2) are not

directly related to electronic or vibrational states. The reduction of the width of the reaction barrier

by electric field would be related to the two threshold energies.

11

References 1Yamauchi, J.; Tsukada, M.; Watanabe, S.; Sugino, O.; First-Principles Study on Energetics of

c-BN(001) Reconstructed Surfaces. Phys. Rev. B 1996, 54, 5586-5602. 2 Perdew, J. P.; Burke, K.; Ernzerhof, M.; Generalized Gradient Approximation made Simple.

Phys. Rev. Lett. 1996, 77, 3865-3868. 3 Vanderbilt, D.; Soft Self-Consistent Pseudopotentials in a Generalized Eigenvalue Formalism.

Phys. Rev. B 1990, 41, 7892-7895. 4 Monkhorst, H. J.; Pack, J. D.; Special Points for Brillouin-Zone Integrations. Phys. Rev. B 1976,

13, 5188-5192. 5 Kajita, S.; Nakayama, T.; Kawai, M.; Ab-initio Calculation Method for Electronic Structures of

Charged Surfaces using Repeated Slab and Density-Variable Charge Sheets. J. Phys. Soc. Jpn.

2007, 76, 044701. 6 Fu, C. L.; Ho, K. M.; External-Charge-Induced Surface Reconstruction on Ag(110). Phys. Rev.

Lett. 1989, 63, 1617-1620. 7 Kajita, S.; Nakayama, T.; Yamauchi, J.; Density Functional Calculation of Work Function using

Charged Slab Systems. J. Phys.: Conf. Ser. 2006, 29, 120-123. 8 Otani, M.; Sugino, O.; First-Principles Calculations of Charged Surfaces and Interfaces: A

Plane-Wave Nonrepeated Slab Approach. Phys. Rev. B 2006, 73, 115407. 9 Batzill, M.; Katsiev, K.; Gaspar, D. J.; Diebold, U.; Variations of the Local Electronic Surface

Properties of TiO2(110) induced by Intrinsic and Extrinsic Defects. Phys. Rev. B 2002, 66, 235401. 10

Yoshida, S.; Kanitani, Y.; Oshima, R.; Okada, Y.; Takeuchi, O.; Shigekawa, H.; Microscopic

Basis for the Mechanism of Carrier Dynamics in an Operating p-n Junction Examined by Using

Light-Modulated Scanning Tunneling Spectroscopy. Phys. Rev. Lett. 2007, 98, 026802. 11

Minato, T.;�Sainoo, Y.; Kim, Y.; Kato, H. S.; Aika, K.-i.; Kawai, M.; Zhao, J.; Petek, H.;

Huang, T.; He, W.; Wang, B.; Wang, Z.; Zhao, Y.; Yang, J.; and Hou, J. G.; The Electronic

Structure of Oxygen Atom Vacancy and Hydroxyl Impurity Defects on Titanium Dioxide (110)

Surface. J. Chem. Phys. 2009, 130, 124502. 12

Henderson, M. A.; An HREELS and TPD Study of Water on TiO2(110): The Extent of

Molecular versus Dissociative Adsorption. Surf. Sci. 1996, 355, 151-166. 13

Bohm, D.; Quantum Theory, Dover New York, 1989.