GaN

Thermoelectric current was observed in the presence of O2 or CO gas in the ...... Young Keun Lee , Chan Ho Jung , Jonghyurk Park , Hyungtak Seo , Gabo...
20 downloads 0 Views 249KB Size
Download PDF
J. Phys. Chem. C 2007, 111, 15331-15336

15331

Interfacial and Chemical Properties of Pt/TiO2, Pd/TiO2, and Pt/GaN Catalytic Nanodiodes Influencing Hot Electron Flow Jeong Young Park,†,‡ J. R. Renzas,† Bryan B. Hsu,† and Gabor A. Somorjai*,†,‡ Department of Chemistry, UniVersity of California, Berkeley, California 94720, and Materials Sciences DiVision and Chemical Sciences DiVision, Lawrence Berkeley National Laboratory, Berkeley, California 94720 ReceiVed: June 13, 2007; In Final Form: July 25, 2007

The influence of physical and chemical properties of Pt/TiO2, Pd/TiO2, and Pt/GaN metal-semiconductor Schottky diodes on the yield of collected hot electron flow (number of hot electrons per product molecule) was investigated. We measured both the chemicurrent (electron flow) and chemical turnover rate during oxidation of carbon monoxide (at pressures of 100 Torr of O2 and 40 Torr of CO in the 373-513 K range) using reaction systems equipped for simultaneous reaction rate and current measurements. The chemicurrent was found to be correlated with the turnover rate and can be used to detect the turnover rate for the three diodes. Thermoelectric current was observed in the presence of O2 or CO gas in the absence of catalytic reaction. The chemicurrent was observed only under catalytic reaction condition. The chemicurrent yield of Pt/GaN ((3.5 ( 0.8) × 10-3) was higher than that of Pt/TiO2 or Pd/TiO2 ((2-3) × 10-4) by 1 order of magnitude. We found that the metal-semiconductor interface structure (roughness, grain size, and stepterrace) is important in controlling the magnitude of chemicurrent yield.

I. Introduction Understanding the mechanisms of energy transfer in exothermic chemical processes at surfaces has been a longstanding question and is of fundamental and practical importance. Recent experimental1-6 and theoretical7,8 studies have demonstrated the electronic excitations created during chemisorption and physisorption of gases at surfaces,1-3,9 and by chemical reactions at surfaces.10-15 Hellberg et al.4 proposed that an exothermic surface reaction (K + Cl2 in their case) induces electron transfer from the metal surface to the molecular orbital of the adsorbed diatomic molecule. The role of electronic excitation in catalysis at the oxidemetal interface has been explored. The enhancement of catalytic activity due to the presence of oxide-metal interfaces was first suggested by Schwab and others,16,17 who performed oxidation of carbon monoxide on Ag/NiO. The concept of “strong metalsupport interaction” was invoked recently for gold-oxide interfaces that exhibit unusual catalytic behavior for small gold particles deposited on titanium oxide or cerium oxide.18 Hayek and others19 showed that the reaction rate in the oxide-metal model system depends on the oxidation state of the supporting oxide, the free metal surface area, and the number of sites at the interface between the metal and the support.19-21 This effect was also investigated by Boffa and others22 using rhodium deposited on a large number of oxides. They observed a remarkable 14-fold increase in turnover rates for CO2 hydrogenation, especially in the presence of three different oxides, TiOx, NbOx, and TaOx. The activity was at a maximum when the oxide-metal interface area was at a maximum, which occurred at about 1/2 monolayer of oxide coverage. Interestingly, the enhancement of chemical reactivity was not quite as * To whom correspondence should be addressed. E-mail: somorjai@ berkeley.edu. † University of California, Berkeley. ‡ Lawrence Berkeley National Laboratory.

prominent for ZrOx, VOx, WOx, and FeOx. These experiments suggest that the metal-oxide interface plays an important role in determining chemical reactivity for some classes of oxides (for example, NiO, CeO, TiOx, NbOx, and TaOx). These oxides are semiconductors with bulk band gaps of 3-4 eV. Therefore, the interface of metal catalyst and oxides can form a Schottky barrier with a barrier height that is the energy difference between the work function of the metal and the electron affinity of the semiconductor.23 The origin of the oxide-metal support interaction has been attributed to either geometric or electronic effects.19,24 The geometric effect assumes that the active surface area of the noble metal is changed during the reduction process. The electronic effect involves charge transfer through a Schottky barrier formed at the metal-oxide interface. Elucidation of the origin of the metal-support interaction requires measurement of the charge transfer through the oxide-metal interface simultaneously with the turnover rate. As we will show later, catalytic nanodiodes that are composed of a thin metal catalyst and an oxide can be used to measure the hot electron flow through the metal-oxide interface. In this article, we present chemical reactivity and chemicurrent measured on Pt/TiO2, Pd/TiO2, and Pt/GaN diodes during CO oxidation. The chemicurrent yield, which is the number of hot electrons per CO2 molecule produced in the CO oxidation reaction for diodes, was obtained based on measurements of chemicurrent and turnover rate. We show that interfacial properties of metal-oxide Schottky diodes influence the yield of collected hot electron flow. II. Concept of the Catalytic Nanodiode Electronic excitation in exothermic catalytic reactions involves the flow of hot electrons with energy of 1-3 eV,7 assuming that most of the chemical energy is converted to electron flow. Once the electrons on the metal surface become energetic (or

10.1021/jp074562h CCC: $37.00 © 2007 American Chemical Society Published on Web 10/03/2007

15332 J. Phys. Chem. C, Vol. 111, No. 42, 2007

Park et al.

Figure 1. (a) Schematic and (b) energy diagram of the nonadiabatic electronic excitation during exothermic reaction on a metallic surface. Hot electrons are generated on the surface and dissipate energy and turn into low-energy electrons within the length scale of electron mean free path. (c) Schematic and (d) energy diagram of hot electron generation in metal-semiconductor nanodiodes. Schottky barrier is the energy barrier between the metal and the semiconductor. The barrier height (ESB) is the energy difference between the metal work function (φm) and the electron affinity of semiconductor (χ). The hot electrons overcome the Schottky barrier and turn into low-energy electrons in the semiconductor when the excess energy is larger than the Schottky barrier height.

hot) through the nonadiabatic electronic excitation, these hot electrons move into the bulk of the metal catalyst, as shown in Figure 1a, and eventually dissipate energy and turn into lowenergy electrons through lattice atom relaxation within the length scale of the electron mean free path, which is in the 3-7 nm range. These low-energy electrons can move back to the surface to fill vacancies left by the departed hot electrons. The flow of electrons is analogous to circulation of hot water upon heating, as shown in Figure 1a. This flow of hot electrons takes place only at the near-surface region, and is spatially limited by the electron mean free path. Figure 1b shows the energy diagram of the flow of hot electrons in the metal catalyst. The role of hot electrons in the metal catalyst is therefore limited, since hot electrons are localized at the near-surface region, and have short lifetimes (∼10 fs).25 However, since most metal catalysts are nanoparticles in the 1-10 nm range,26,27 hot electrons can reach the oxide-metal interface and overcome the potential barrier to be transported into the oxide. In a metal-oxide catalyst, the Schottky barrier is formed at the interface of metal and oxide (or wide band gap semiconductor). If the size of the metal nanoparticle or thickness of metal film is smaller than the mean free path, electrons are transported across the metal without collision as illustrated in Figure 1c. If the excess energy of electrons Eex ) |E - EF| is larger than the effective Schottky barrier (ESB), which is the difference between the conduction band minimum and the Fermi energy, EF, at the interface, electrons go over the Schottky barrier and transport to the semiconductor as shown in the energy diagram of Figure 1d. These electrons dissipate energy and turn into low-energy electrons in the semiconductor. No reverse transmission to the metal is allowed because of the energy barrier from the

conduction band of the semiconductor to the metal, which has a height of ESB - (Ec - EF). Instead, the low-energy electrons will be supplied to the metal through the electron circuit to fill the vacancies of hot electrons. This irreversibility of hot electron transport through the metal-semiconductor barrier causes the one-way transport of electrical carriers during the catalytic reactions, which makes it possible for us to measure hot electron flow. III. Experimental Details III.A. Fabrication of Nanodiodes. Details on fabrication of Schottky diodes are described elsewhere.10,13,14 Vertically oriented Pt/TiO2 or Pd/TiO2 Schottky diodes were fabricated on an insulating p-type Si(100) wafer covered by 100 nm SiO2. The back electrode was formed by an electron beam evaporating a 100 nm Au contact pad on top of a 30 nm titanium adhesion layer. A portion of the Au contact pad was then covered with 30 nm titanium, forming an ohmic contact between the Au pad and the titanium oxide semiconductor, due to the low work function of titanium. Reactive direct current (dc) magnetron sputtering was used to deposit approximately 150 nm of titanium oxide. During sputtering, the bias voltage was 430 V, O2 pressure was 11-12 sccm, and Ar pressure was 37 sccm. The resistivity of titanium oxide is 0.3-0.4 ohm cm, as measured using a four-point probe. A portion of the titanium oxide layer was covered by 200 nm of silicon nitride deposited by plasmaenhanced chemical vapor deposition to prevent electrical shorting between the titanium oxide and the top contact layer. The Schottky contact was then formed by electron-beam evaporation of 5 nm of Pt or Pd in a 3 mm diameter circle, one-third of which contacted the titanium oxide and two-thirds of which contacted the silicon nitride insulating layer. The top

Properties of Pt/TiO2, Pd/TiO2, and Pt/GaN Diodes

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15333

electrical contact was then formed on top of the Pt or Pd on silicon nitride by an electron beam depositing again a 100 nm Au contact pad. Two types of Pt/GaN diodes were used in this study. One type of Pt/GaN diode was fabricated with a photolithography process.12 The other type of Pt/GaN diode was fabricated on 2 in. n-type GaN/sapphire substrates purchased from Kyma Technologies. The GaN crystal has the wurtzite structure and the orientation of (0001) c-axis Ga face within 0.5°. The thickness of the GaN epitaxial layer is 5 µm, and the resistivity is 1-5 ohm cm. The wafers were sonicated successively in trichloroethylene, acetone, and methanol to remove contaminants and boiled in 1.0 N KOH for 10 min to modify the surface topography.28 A 200 nm SiO2 insulating layer was formed on the GaN surface via electron-beam evaporation. This layer prevents the low work function contact pads from shortcircuiting the devices. Then 5 nm of Pt was deposited, also by use of electron-beam evaporation, to form the Schottky contact. The Pt/GaN contact area is approximately 1 mm2. Finally, Ti/ Al ohmic contacts were deposited to complete the devices. Diffusion of Al and Ti toward GaN and formation of AlN or TiN alloys result in ohmic contact to the GaN layer.29 III.B. Characterization of Schottky Metal-Semiconductor Diodes. The formation of a Schottky barrier between Pt and TiO2 or GaN has been previously reported. Ali et al. fabricated Pt/GaN gas sensors and measured a Schottky barrier height of 0.92 eV30 based on thermionic emission theory. Wang et al. reported barrier heights of 1.1 and 0.96 eV based on I-V measurements on Pt/GaN and Pd/GaN diodes, respectively.31 Dittrich et al. reported Schottky barrier heights of 1.2-1.3 eV on Pt/TiO2 diodes.32 This energy barrier is high enough to suppress contribution of electron flow from the thermal excitation of electrons. In order to determine barrier heights and ideality factors for the nanodiodes, we fit the I-V curves of our devices to the thermionic emission equation. For thermionic emission over the barrier, the current density of Schottky contacts as a function of applied voltage is given by23

( )[ (

I ) FA*T2 exp -

) ]

Φn e0(Va - RsI) exp -1 kBT ηkBT

(1)

where F ) area, A* ) effective Richardson constant, Φn ) Schottky barrier height, η ) ideality factor, and Rs ) series resistance, respectively. The effective mass of the conduction electrons in GaN, m* ) 0.22m0, gives an effective Richardson constant A*(GaN) ) 2.64 × 104 A/cm2 K. The effective Richardson constant for TiO2 is A* ) 24 A/cm2 K. Figure 2 shows the I-V curve measured on Pt/TiO2 diodes. Based on fitting the I-V curve to the thermionic equation (1), we obtained a barrier height of 1.0 eV, an ideality factor of 1.9, and a series resistance of 780 ohm for the Pt/TiO2 diode. The Pt/GaN diode showed a barrier height of 1.19 eV, higher than that of the Pt/TiO2 diode, and an ideality factor of 1.9, similar to that of the Pt/TiO2 diode. The Pd/TiO2 diode showed the lowest barrier height, 0.85 eV. For detection of hot electron flow, it is necessary to have excess energy larger than the Schottky barrier height. Under CO oxidation conditions, the adsorption energies of O2 and CO are 2.4 and 1.5 eV, respectively.33 Both are higher than the energy barrier of metal-semiconductor diodes (0.9-1.2 eV). This makes the combination of high work function metal (Pt or Pd) and high band gap semiconductor (GaN or TiO2) suitable for the main component of the nanodiode.

Figure 2. I-V curve measured on Pt/TiOx diode and fitting to thermionic emission equation. The fitting gives rise to Φn (Schottky barrier height) of 1.0 eV, η (ideality factor) of 1.9, and Rs (series resistance) of 780 ohm.

IV. Results IV.A. Measurement of Chemicurrent and Thermoelectric Current. A batch reaction system combined with electrical measurement was built to carry out the gas-phase reaction. The design of the reaction cell has been described elsewhere.13,14 The reaction cell was evacuated down to 5 × 10-8 Torr by a turbomolecular pump. A ceramic heater was used to change the temperature at the sample. The temperature controller provided feedback to the current applied to the heater that kept the fluctuations of the temperature below 0.5 °C. A sampling loop, including a gas chromatograph and a circulation pump, continuously measured reaction rates from reactant and product concentrations. Figure 3a shows current signal measured between two gold contacts of Pt/TiO2 diodes under reaction conditions and under pure He. Zero bias is applied between two Au contacts during the current measurement. When the diode is under 1 atm of He, the thermoelectric current due to the elevated temperature is observed. The thermoelectric current is caused by the difference of electrical potential between two electrodes because of the Seebeck effect. We found that the thermoelectric current is mainly influenced by the Seebeck coefficient of each layer of diodes. In the case of Pt/TiO2 and Pd/TiO2 diodes, the thermoelectric current was in the opposite direction from the chemicurrent. Figure 3b shows the thermoelectric current and chemicurrent measured on Pt/GaN diode. While the thermoelectric current was smaller than that of Pt/TiO2, the chemicurrent increased at a lower temperature than Pt/TiOx. Interestingly, the thermoelectric current in the Pt/GaN diode is in the same direction with the chemicurrent and the current value was much smaller than that of the Pt/TiO2 and Pd/TiO2 diodes. This suggests that the thermoelectric current is crucially influenced by the thermoelectric property of the semiconductor layer. The higher thermoelectric current of the TiOx-based diode could be associated with the extraordinarily high Seebeck coefficient of TiO2, which is 0.4 mV/K at 300 °C. The smaller value and different polarity Seebeck coefficient of GaN, -0.05 mV/K at 300 K, is consistent with the variation of thermoelectric currents measured on Pt/GaN. Heat transfer from the exothermic chemical reaction to the Pt surface could also increase the local sample temperature independent of the heater. We found the increase of temperature caused by the exothermic reaction in our experimental range (up to turnover rate of 103 molecules per Pt site per second) based on the thermal transport equation is ignorable (less than 10-3 K).35 The small increase of local temperature is due to

15334 J. Phys. Chem. C, Vol. 111, No. 42, 2007

Park et al.

Figure 4. Current measured under various gas conditions as a function of temperature. Chemicurrent was measured under 100 Torr of O2, 40 Torr of CO, and 620 Torr of He. Black squares refer to the thermoelectric current measured in 760 Torr of He. Blue triangles represent the current measured in 100 Torr of O2 and 660 Torr of He, and red circles show the current in 40 Torr of CO and 720 Torr of He.

Figure 3. (a) Thermoelectric current and chemicurrent measured on Pt/TiO2 diode during CO oxidation as a function of the temperature. (b) Thermoelectric current and chemicurrent measured on Pt/GaN diode during CO oxidation. Chemicurrent was measured under 100 Torr of O2, 40 Torr of CO, and 620 Torr of He.

the structure of the diode (nanometer-scale thickness of layers and millimeter-scale spatial size of layers). IV.B. Measurement of Thermoelectric Current in He, O2, and CO in the Absence of Catalytic Reaction. We found that thermoelectric current was observed in the presence of O2 or CO gas in the absence of catalytic reaction, and the chemicurrent was observed only under catalytic reaction conditions. Figure 4 shows the plot of currents as a function of temperature under various gas conditions. The current measured under O2 (100 Torr) and He (660 Torr) was quite close to the thermoelectric current as measured under only He. Likewise, the current under CO (40 Torr) and He (720 Torr) was the same as the thermoelectric current, within experimental error. Only when we introduced the mixture of O2 (100 Torr) and CO (40 Torr), and CO oxidation took place on the Pt surface, was significant current signal with a large deviation from measured thermoelectric current detected. This measurement suggests that the presence of O2 or CO in the absence of catalytic reaction is not responsible for the hot electron generation, at least under highpressure conditions. Nienhaus et al. observed the flow of hot electrons through Ag/Si(111) or Cu/Si(111) Schottky barriers upon adsorption of hydrogen or deuterium atoms. The chemicurrent decreases exponentially with increasing exposure due to a decrease in the number of free adsorption sites.1,2 Since our experiment takes place in the high-pressure range, the surface would be covered with the adsorbate molecules immediately after releasing the gas, making the measurement of the transient change of chemicurrent caused by the adsorption of molecules quite challenging. Again, this experiment shows

Figure 5. (a) Arrhenius plots of chemicurrent and turnover rate measured on Pd/TiO2 diode and (b) Pt/GaN diode.

that the steady-state current detected on our diodes is directly associated with the nature of catalytic reactions which involve the continuous processes of adsorption of molecules, chemical rearrangement, and desorption of products. IV.C. Chemicurrent and Turnover Rates under Reaction Conditions of CO Oxidation. Production of CO2 molecules

Properties of Pt/TiO2, Pd/TiO2, and Pt/GaN Diodes

Figure 6. 1 µm × 1 µm contact mode AFM images on (a) TiO2 surface, (b) Pt on TiO2 surface, (c) GaN surface, and (d) Pt/GaN surface.

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15335 contact mode. Parts a and b of Figure 6 show 1 µm × 1 µm AFM images of TiO2 and Pt/TiO2 surfaces, respectively. The surface roughness (defined by the standard deviation of height distributions) of the TiO2 surface was found to be 0.48 ( 0.08 nm. Also, the TiO2 surface exhibited grains with an average size of 50 nm. After deposition of the Pt thin film (with thickness of 5 nm), the roughness increased to 0.59 ( 0.10 nm as shown in Figure 6b. Unlike the TiO2 surface, the GaN surface exhibits stepterrace structures due to the epitaxial growth of GaN layers, as shown in Figure 6c. The surface is composed of terraces with an average width of 100 nm, separated by atomic steps with a height of 0.27 ( 0.02 nm. This step height agrees well with the 2.6 Å step height expected for one (0001) GaN bilayer (one GaN layer thickness along the c-axis). On the terrace, the roughness of the GaN surface is less than 0.02 nm. After deposition of the Pt 5 nm layer, the surface exhibits Pt grains with a characteristic size of 10 nm and roughness of 0.5 nm, as shown in Figure 6d. Assuming that roughness of TiO2 or GaN surface reflects the interface roughness of Pt/TiO2 and Pt/GaN, the roughness of the interface of Pt/GaN is 1 order of magnitude smaller than that of Pt/TiO2. The roughness of the interface for the three diodes is also included in Table 1. V. Discussion

was measured during carbon monoxide oxidation using gas chromatography simultaneously with chemicurrent measurements. We found that the chemicurrent measured on Pt/TiO2 diode is well correlated with the turnover rate. The activation energy (Ea) obtained with the turnover rate on Pt/TiO2 was 22 kcal/mol, quite close to Ea (21 kcal/mol) obtained from the chemicurrent measurement.14 Parts a and b of Figure 5 show the Arrhenius plots of chemicurrent (nanoamperes) and turnover rate (molecules per Pt site per second) for Pd/TiO2 and Pt/GaN, respectively. From the slope of the Arrhenius plots of chemicurrent and turnover rate, the activation energy of the carbon monoxide reaction on the Pd/TiO2 and Pt/GaN nanodiodes was determined. Pd/TiO2 diode exhibits Ea of 20.4 ( 1.1 kcal/mol (chemicurrent measurement) and 21.1 ( 0.9 kcal/mol (turnover measurement), and Pt/GaN shows Ea of 20.6 ( 1.1 kcal/mol (chemicurrent measurement) and 20.3 ( 1.3 kcal/mol (turnover measurement). For three types of diodes, the activation energy obtained with chemicurrent measurement remains similar to that from the turnover measurement, suggesting excellent correlation between the hot electron flow and chemical turnover rate. Table 1 shows the summary of turnover rate, chemicurrent, and activation energy measured on three diodes. The chemicurrent yield (number of electrons per number of produced CO2 molecules) was obtained for three nanodiodes with the 5 nm thick metal thin film. Since the chemicurrent and the turnover rate have a similar dependence on temperature, the chemicurrent yield is independent of reaction temperature. IV.D. Interfacial Properties of Metal-Semiconductor Interface. Atomic force microscopy (AFM) is a useful tool to characterize the structural, mechanical, and electrical properties of surfaces.34 The diode surface was imaged with AFM in

As shown in Table 1, turnover rate and chemicurrent of Pd/ TiO2 is similar to Pt/TiO2. Therefore, the chemicurrent yield of Pd/TiO2, (3 ( 1.2) × 10-4, is similar to that of Pt/TiO2, within the error of the measurement. Interestingly, the chemicurrent yield of Pt/GaN, (3.5 ( 0.8) × 10-3, is higher than that of Pt/ TiO2 or Pd/TiO2 by 1 order of magnitude.35 The difference of chemicurrent yield between Pt/GaN and Pt/TiO2 could be associated with the interfacial properties, such as smoothness of the metal-semiconductor interfaces. The smooth metalsemiconductor interface can give rise to the lower number density of scattering centers and enhanced transmission probability. Another factor that can influence the chemicurrent yield is the size of the grains. While the average size of the grains on TiO2 film is 50 nm, the GaN surface exhibits atomically flat terraces with an average width of 100 nm. The irregularity at the grain boundary could serve as the scattering center during hot electron transport, decreasing the transmission probability. The thickness of the metal layer is an important factor in the efficiency of hot electron transport. The dependence of the chemicurrent on the Pt thickness in Pt/TiOx diodes was already shown in a previous publication.10 Ji et al. showed that chemicurrent decreased significantly as the thickness of the Pt layer increased. This result confirms that the current signal we measured on the Schottky diode attenuates within the mean free path of the Pt layer. Interestingly, the activation energy of these diodes obtained with chemicurrent and turnover rate measurements falls between 20 and 22 kcal/mol, as shown in Table 1. The excellent correlation between hot electron flux and turnover rate suggests that current measurement can indeed be used for monitoring

TABLE 1: Chemicurrent, Turnover Rate, Interface Roughness, and Activation Energy of Pt/TiO2, Pd/TiO2, and Pt/GaN turnover rate at 260 °C (per Pt site/s) chemicurrent at 200 °C (nA) yield (electron per reaction event) interface roughness (nm) activation energy (kcal/mol)

Pt/TiO2 diode

Pd/TiO2 diode

Pt/GaN diode

150 60 ((20) 2.3 ((0.9) × 10-4 0.48 ( 0.08 21.3 ( 0.7 (chemicurrent) 22.4 ( 1.3 (turnover rate)

100 70 ((20) 3.0 ((1.2) × 10-4 0.48 ( 0.08 20.4 ( 1.1 (chemicurrent) 21.1 ( 0.9 (turnover rate)

20 50 ((20) 3.5 ((0.8) × 10-3 0.02 ( 0.01 20.6 ( 1.1 (chemicurrent) 21.3 ( 1.3 (turnover rate)

15336 J. Phys. Chem. C, Vol. 111, No. 42, 2007 the chemical reaction in a quantitative manner. We note that the activation energy of CO oxidation depends upon the thickness of the Pt film. The activation energy of CO oxidation for a 20 nm thick metal film on silicon was found to be 26 kcal/mol.36 The activation energy of a single crystal Pt surface was found to be 30-40 kcal/mol below ignition temperature.37 This suggests that the lower activation energy in a 5 nm thick metal-semiconductor diode is related to the increased role of the metal-oxide interface. A new model catalyst system, the catalytic nanodiode, allows us to monitor the hot electron flow generated by exothermic chemical reactions. The transport of hot electrons through the metal-oxide interface could be an important ingredient that influences catalytic reactivity and selectivity. An intriguing experiment to test this idea is to measure the reaction rate under various electrical configurations such as open loop, closed loop, or closed loop under external voltage bias. Electrical measurement and manipulation of hot electron flow could provide us with a new tool to bridge the material gap38 between wellcharacterized model systems and industrially relevant catalysts. Hot electron flow can be utilized in various pressure ranges: from adsorption in ultrahigh vacuum1 to chemical catalysis in high pressure,14,39 which allows us to close the pressure gap. VI. Conclusion Hot electron flow through a catalytic nanodiode provides new insights into the role of electronic excitation in catalysis and suggests new energy conversion processes. The good correlation between the hot electron flux and the turnover of the chemical reaction suggests that the detection of hot electron flow can be used to study the role of the oxide-metal interface in heterogeneous catalysis and detect the catalytic turnover rate. We found that chemicurrent was observed only under catalytic reaction conditions, and thermoelectric current was observed in the presence of O2 or CO gas in the absence of catalytic reaction. The interfacial roughness characterized with AFM was found be correlated with the chemicurrent yield. Acknowledgment. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. References and Notes (1) Nienhaus, H. Surf. Sci. Rep. 2002, 45, 3-78. (2) Nienhaus, H.; Bergh, H. S.; Gergen, B.; Majumdar, A.; Weinberg, W. H.; McFarland, E. W. Phys. ReV. Lett. 1999, 82, 446-449. (3) Nienhaus, H.; Gergen, B.; Weinberg, W. H.; McFarland, E. W. Surf. Sci. 2002, 514, 172-181. (4) Hellberg, L.; Stromquist, J.; Kasemo, B.; Lundqvist, B. I. Phys. ReV. Lett. 1995, 74, 4742-4745. (5) Huang, Y. H.; Rettner, C. T.; Auerbach, D. J.; Wodtke, A. M. Science 2000, 290, 111-114.

Park et al. (6) Denzler, D. N.; Frischkorn, C.; Hess, C.; Wolf, M.; Ertl, G. Phys. ReV. Lett. 2003, 91, 226102. (7) Gadzuk, J. W. J. Phys. Chem. B 2002, 106, 8265-8270. (8) Mizielinski, M. S.; Bird, D. M.; Persson, M.; Holloway, S. J. Chem. Phys. 2005, 122, 084710. (9) Nienhaus, H.; Weyers, S. J.; Gergen, B.; McFarland, E. W. Sens. Actuators, B: Chem. 2002, 87, 421-424. (10) Ji, X. Z.; Somorjai, G. A. J. Phys. Chem. B 2005, 109, 2253022535. (11) Ji, X. Z.; Zuppero, A.; Gidwani, J. M.; Somorjai, G. A. Nano Lett. 2005, 5, 753-756. (12) Ji, X. Z.; Zuppero, A.; Gidwani, J. M.; Somorjai, G. A. J. Am. Chem. Soc. 2005, 127, 5792-5793. (13) Park, J. Y.; Somorjai, G. A. J. Vac. Sci. Technol., B 2006, 24, 1967-1971. (14) Park, J. Y.; Somorjai, G. A. ChemPhysChem 2006, 7, 1409-1413. (15) Somorjai, G. A. Catal. Lett. 2005, 101, 1-3. (16) Schwab, G. M. Angew. Chem., Int. Ed. 1967, 6, 375. (17) Schwab, G. M.; Siegert, R. Z. Phys. Chem. (Frankfurt) 1966, 50, 191. (18) Goodman, D. W. Catal. Lett. 2005, 99, 1-4. (19) Hayek, K.; Fuchs, M.; Klotzer, B.; Reichl, W.; Rupprechter, G. Top. Catal. 2000, 13, 55-66. (20) Penner, S.; Wang, D.; Podloucky, R.; Schlogl, R.; Hayek, K. Phys. Chem. Chem. Phys. 2004, 6, 5244-5249. (21) Rupprechter, G.; Seeber, G.; Goller, H.; Hayek, K. J. Catal. 1999, 186, 201-213. (22) Boffa, A.; Lin, C.; Bell, A. T.; Somorjai, G. A. J. Catal. 1994, 149, 149-158. (23) Sze, S. M. Physics of Semiconductor DeVices; Wiley: New York, 1981. (24) Haller, G. L.; Resasco, D. E. AdV. Catal. 1989, 36, 173-235. (25) Ogawa, S. Nagano, H.; Petek, H. Phys. ReV. B 1997, 55, 1086910877. (26) Niesz, K.; Grass, M.; Somorjai, G. A. Nano Lett. 2005, 5, 22382240. (27) Song, H.; Rioux, R. M.; Hoefelmeyer, J. D.; Komor, R.; Niesz, K.; Grass, M.; Yang, P. D.; Somorjai, G. A. J. Am. Chem. Soc. 2006, 128, 3027-3037. (28) Kim, H. G.; Kim, S. H.; Deb, P.; Sands, T. J. Electron. Mater. 2006, 35, 107-112. (29) Tan, L. S.; Prakash, S.; Ng, K. M.; Ramam, A.; Chua, S. J.; Wee, A. T. S.; Lim, S. L. Semicond. Sci. Technol. 2000, 15, 585-588. (30) Ali, M.; Cimalla, V.; Lebedev, V.; Romanus, H.; Tilak, V.; Merfeld, D. Sandvik, P.; Ambacher, O. Sens. Actuators, B: Chem. 2006, 113, 797804. (31) Wang, L.; Nathan, M. I.; Lim, T. H.; Khan, M. A.; Chen, Q. Appl. Phys. Lett. 1996, 68, 1267-1269. (32) Dittrich, T.; Zinchuk, V.; Skryshevskyy, V.; Urban, I.; Hilt, O. J. Appl. Phys. 2005, 98, 104501. (33) Kwong, D. W. J.; Deleon, N.; Haller, G. L. Chem. Phys. Lett. 1988, 144, 533-540. (34) Park, J. Y.; Ogletree, D. F.; Thiel, P. A.; Salmeron, M. Science 2006, 313, 186-186. (35) Park, J. Y.; Renzas, J. R.; Contreras, A. M.; Somorjai, G. A. Top. Catal. 2007, 46, 217-222. (36) Contreras, A. M.; Yan, X. M.; Kwon, S.; Bokor, J.; Somorjai, G. A. Catal. Lett. 2006, 111, 5-13. (37) Su, X. C.; Cremer, P. S.; Shen, Y. R.; Somorjai, G. A. J. Am. Chem. Soc. 1997, 119, 3994-4000. (38) Somorjai, G. A.; York, R. L.; Butcher, D.; Park, J. Y. Phys. Chem. Chem. Phys. 2007, 9, 3500-3513. (39) Somorjai, G. A.; Bratlie, K. M.; Montano, M. O.; Park, J. Y. J. Phys. Chem. B 2006, 110, 20014-20022.