Article pubs.acs.org/Langmuir
Photochemistry of Carboxylate on TiO2(110) Studied with Synchrotron Radiation Photoelectron Spectroscopy A. Sandell,*,† D. Ragazzon,† A. Schaefer,‡ M. H. Farstad,§ and A. Borg§ †
Department of Physics and Astronomy, Uppsala University, P.O. Box 516, SE-75120 Uppsala, Sweden Division of Synchrotron Radiation Research, Lund University, Box 118, SE-221 00 Lund, Sweden § Department of Physics, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, Norway ‡
ABSTRACT: We present a dedicated synchrotron radiation photoelectron spectroscopy (SR-PES) study of a photochemical reaction on the surface of rutile TiO2(110). The photoreaction kinetics of carboxylate species (trimethyl acetate, TMA) upon irradiation by UV and soft X-rays were monitored, and we show that it is possible to control the reaction rates from UV light and soft X-rays independently. We directly observe Ti4+ → Ti3+ conversion upon irradiation, attributed to electron trapping at Ti sites close to surface OH groups formed by deprotonation of the parent molecule, trimethylacetic acid (TMAA). TMA photolysis on two surface preparations with different oxygen vacancy densities shows that the vacancy-related charge quenches the amount of charge that can be trapped at hydroxyls upon irradiation. During the initial stages of reaction the correlation between the amount of photodepleted TMA and the amount of charge trapped in the Ti 3d band gap state is nearly 1:1. A first-order kinetics analysis reveals that the reaction rate decreases with decreasing TMA coverage. There is also a coverage-dependent difference in the electronic structure of TMA moieties, primarily involving the carboxyl anchor group. These changes are consistent with a decreased hole affinity of the adsorbed TMA and hence a decreased reaction rate. This discovery adds to the previously presented picture of a reactivity that is inversely proportional to the number of surface hydroxyls, suggesting that the balance between the amounts of TMA, OH, and trapped charge needs to be considered.
1. INTRODUCTION Photocatalysis represents one of the most important ways to harvest energy. In the photocatalytic process, light is used to induce redox reactions of adsorbates by photoexcited carriers (electrons and holes).1,2 The outcome is that the energy from the light is stored in chemical bonds. The most prominent example is splitting of water, generating H2 and O2.3−5 Photocatalysts are also extensively used for pollution control and are also considered to be of use for capture of CO2.6−8 The photocatalytic process is however not fully understood, and detailed studies addressing mechanistic aspects on a molecular level are highly motivated. A most important parameter in this context is how the reactant properties vary with coverage. Photoinduced surface reactions occur through electronic excitations of the system, which leads to chemical changes. This makes photoelectron spectroscopy (PES) an ideal tool to follow the evolution of the surface and the adsorbed species via changes in the electronic structure. However, PES has been very scarcely used in studies of surface photochemical reactions, and up to now there has been no study where the process is followed in detail by core and valence level photoemission spectra. A possible reason for the limited use of PES is the inherent complication that photoreactions can be induced by the measurement itself. Therefore, one important aim with the present study was to find an approach with which the dominating light source driving the reaction can be selected, setting the scene for PES to be used for studies of photocatalysis. Specifically, we explore the potential of SR© XXXX American Chemical Society
PES to monitor the photoreaction of an organic molecule on the surface of titanium dioxide (TiO2). TiO2 has for a long time been recognized as an extremely useful and versatile material for photochemical applications. Photodriven processes that occur on the surface of TiO2 are exploited in catalysis, environmental remediation (especially water purification), and solar energy harvesting.9−13 TiO2 in single crystalline form has for many years served as substrate in model studies of transition metal oxide surface chemistries under ultrahigh vacuum (UHV). The vast majority of the work has been devoted to the thermodynamically stable (110) face of rutile TiO2.14,15 The TiO2(110) surface exposes alternating rows of 5-fold coordinated Ti atoms [Ti(5)] and 2-fold coordinated bridging oxygen [O(2)] atoms. After cleaning the surface in UHV using the traditional method of sputtering and annealing cycles, the sample becomes reduced due to the loss of oxygen atoms. This makes the crystal more conducting, thereby allowing for studies using methods based on charged information carriers such as photoelectron spectroscopy. The reduction of TiO2(110) furthermore results in a surface where a number of O(2) atoms are missing; typical densities of such oxygen vacancies are 0.05−0.15 ML, where 1 ML corresponds to the density of Ti(5) atoms on the surface (= 5.2 × 1014 Received: August 10, 2016 Revised: October 7, 2016 Published: October 18, 2016 A
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 1. (a) Schematic illustration of the TMA adsorption geometry. The TMA aliphatic C atoms are labeled C1, and the carboxylic carbon is labeled C2. The substrate O atoms are labeled O1, and the adsorbate related O atoms (from the carboxyl and hydroxyl groups) are labeled O2. (b) Photoemission spectra of the O 1s and C 1s core level regions before and after adsorption and deprotonation of TMAA on the reduced rutile TiO2(110) surface [r-TiO2(110)], yielding a layer comprising TMA and hydroxyls. The peaks are labeled according to (a).
cm−2). This surface is commonly denoted r-TiO2(110) where “r” refers to the reduced character of the surface. Trimethyl acetate [TMA, (CH3)3CCOO−] has emerged as a reference adsorbate for surface science studies of the photochemical activity of TiO2. A series of previous studies have addressed the adsorption and photolysis of TMA on the rutile TiO2(110) and anatase TiO2(001) surfaces, revealing fundamental properties crucial for a mechanistic understanding.16−24 TMA is formed upon deposition of trimethylacetic acid [TMAA, (CH3)3CCOOH]. TMAA undergoes acid dissociation on TiO2 surfaces at temperatures ≥150 K whereby a substrate bridging O2− ion receives the proton, forming OH−. The TMA species form a densely packed (2 × 1) adlayer at or slightly below room temperature (RT), corresponding to a coverage of 0.5 ML using the definition given above.18,19 It has been concluded that each TMA anion bonds in a bidentate fashion across two Ti4+ cations with the OH group adjacent to it (see Figure 1a).18,19 Upon irradiation with photons having a photon energy exceeding the TiO2 band gap energy (3.2−3.3 eV) adsorbed TMA species undergo hole-mediated photodecomposition, generating CO2 and a tert-butyl radical, (CH3)3C. At RT the reaction can be monitored by the depletion of adsorbed TMA since all products rapidly desorb, a process that can schematically be described in two steps:18
essentially inactive. TMA adsorbed adjacent to bridging oxygen vacancies were possible to remove but at a significantly lower rate as compared to TMA on regular Ti sites.22 The reduced photoreaction probability is presumably caused by prompt recombination of photogenerated holes with the Ti 3d electrons located in the vicinity of the vacancies. That is, a reduced activity is associated with the presence of Ti3+ species. A recent study explored the coverage dependence of the reaction rate on stoichiometric parts, i.e., away from the oxygen vacancies. Trapping of charge was demonstrated using UPS. Further support for the inhibitive effect on the hole-mediated TMA photoreaction by electrons trapped at hydroxyls was provided by making a comparison to the reactivity after hydroxylation of the vacancies with water.24 In this paper, we present the first dedicated synchrotron radiation-based PES study of the photochemistry of TMA on TiO2(110). We demonstrate that it is possible to control the influence from UV light and soft X-rays on the photochemical reaction. We address the reaction kinetics and follow the process of charge trapping. A key result is the finding of coverage-dependent changes in the TMA electronic structuresomething that must be taken into account when discussing the photoreactivity of the system.
2. EXPERIMENTAL DETAILS
(CH3)3 CCOO(ads) + h+ → CO2 (g) + (CH3)3 C(ads)
The work was carried out on D1011 (30 ≤ hν ≤ 1600 eV), a bending magnet soft X-ray beamline at MAX-lab in Lund, Sweden.25,26 The D1011 endstation was equipped with a SCIENTA SES200 200 mm mean-radius hemispherical electron energy analyzer, a microchannel plate (MCP) NEXAFS detector, low-energy electron diffraction (LEED) camera, and residual gas analyzer (RGA). The analysis chamber is separated from the sample preparation chamber by means of a gate valve. The rutile TiO2(110) single crystal (Princeton Scientific) was prepared by several cycles of Ar ion sputtering at 1−2 keV and annealing to 870 K. The cleaning cycles were repeated until no contaminants (most notably C, K, and Si) were observed in surface sensitive photoemission spectra. The sputtered and annealed surface (denoted r-TiO2) features a certain density of bridging oxygen vacancies (10−15%). Reduction in the number of oxygen vacancies
(1)
2(CH3)3 C(ads) → (CH3)3 CH(g) + (CH3)2 CCH 2(g) (2)
In an early study STM and EELS were used to establish that trapping of charge occurs during UV photodecomposition of TMA on an oxidized TiO2(110) surface.16 The trapping presumably occurs at the OH groups and was detected through the formation of Ti3+; that is, the band gap state (Ti 3d) becomes populated. In a more recent STM study of TMA/rTiO2(110) it was clearly demonstrated that TMA groups adsorbed at regular Ti sites are readily removed by photooxidation while TMA adsorbed on oxygen vacancy sites are B
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir was accomplished by dosing of 10 langmuirs of O2 at RT. This surface is denoted o-TiO2. Trimethylacetic acid (TMAA, Aldrich 99%) was stored in a glass tube attached to a standard UHV leak valve. The TMAA was cleaned by freeze−pump−thaw cycles, and the dosing volume was baked for several hours. The TMAA doses are given in langmuirs (1 langmuir = 10−6 Torr·s). The photon energies used to record the C 1s and O 1s spectra were 390 and 610 eV, respectively. The valence spectra were recorded using a photon energy of 130 eV. The binding energy (BE) scale was referenced to the Fermi level of a Pt foil mounted on the sample holder. UV illumination was achieved with an Oriel Research Arc Lamp Source with a 100 W Hg bulb. The source was connected to a liquid light guide with a focusing assembly.
3. RESULTS 3.1. PES of the TMA Layer. The adsorbate coverage on TiO2 surfaces is commonly defined in terms of the number of available Ti sites to which the adsorbate binds. The coverage of the saturated first layer of TMA on rutile (110) is 0.5 ML, forming a (2 × 1) structure. Adsorption at RT saturates at coverages close to this value, and we have defined the value as 0.5 ML for the maximum coverage attained. Figure 1b shows C 1s spectra obtained following TMAA exposure (5 langmuirs at RT) on r-TiO2(110). The spectra were measured at a take-off angle 60° from the surface normal. The C 1s spectrum consists of two peaks characteristic of aliphatic carbon (low BE component, labeled C1 in Figure 1a) and carboxylic carbon (high BE component, labeled C2 in Figure 1a). The C 1s BE of the aliphatic peak is 285.20 eV, and the BE of the carboxylic peak is 288.95 eV. The previously reported C 1s binding energies for TMA on rutile (110) are 285.8 eV/289.1 eV and 285.3 eV/288.9 eV.17,21 The corresponding O 1s spectra are shown in Figure 1, left panel. The single layer O 1s spectra consist of two peaks. The low BE peak (at 530.80 eV) is attributed to the substrate oxygen ions (labeled O1 in Figure 1a,b). The high BE component (at 532.05 eV) contains contributions from the TMA oxygen atoms forming the bidentate bond and the oxygens of hydroxyls formed through deprotonation of TMAA. These are labeled O2 in Figure 1a,b. The individual spectral contributions from these types of oxygen atoms are not possible to resolve, but the expected ratio of the relative contributions from TMA and OH is 2:1. The absence of peaks at higher binding energies confirms the bidentate bonding configuration of TMA illustrated in Figure 1.27 3.2. Photodesorption Dynamics: UV vs X-rays. Figure 2 illustrates the effect of UV illumination of the saturated layer as monitored by measurements of the C 1s spectrum. The top spectrum (black line) is that of the as-deposited TMA layer. The C 1s spectrum in the middle (“before UV”, red line) was measured a few minutes later, and it exhibits a small decrease in the intensity due to photoreaction caused by the soft X-rays. The bottom spectrum (“after UV”, purple line) was measured after 156 min of UV illumination without any exposure to Xrays, apart from the quickly measured C 1s spectrum. The coverage has decreased from 0.50 to 0.13 ML. Noteworthy is also that the BE of the C 1s component from carboxylic carbon has shifted by +0.34 eV (BE = 289.29 eV) while the shift of the aliphatic carbon C 1s peak is much smaller (+0.03 eV). The change in the split of the two components indicates a change in the molecular electronic structure as discussed further below (section 3.4).
Figure 2. C 1s core level spectra of TMA on r-TiO2(110) before and after photon irradiation. Three spectra were measured with minimal Xray exposure: the top spectrum (black line) is that of the as-deposited TMA layer, the middle spectrum (“before UV”, red line) was measured a few minutes later, and the bottom spectrum (“after UV”, purple line) was measured after 156 min of UV illumination. For comparison, a spectrum measured after exposure to both X-rays and UV is also shown (dashed line).
Thus, a complication when characterizing the TMA layer with PES is that illumination with both UV light and synchrotron light leads to the removal of TMA species. It turned out not to be possible to distinguish between UV and Xray induced reactions from the PES spectra. This is illustrated in Figure 2, where a spectrum measured after extensive exposure to both X-rays and UV light (dashed black line) is compared to the spectrum measured after exposure to UV light. The spectra are identical, having the same peak widths and BE separation of the two components. However, the reaction rates from the two light sources can be separated. Assuming firstorder kinetics the following equations apply:20
−dΘ/dt = k Θ
(3)
ln Θ = −kt + C
(4)
Here Θ denotes the TMA coverage, k is the photodesorption rate constant, and C corresponds to ln(Θ0). Figure 3a shows ln(Θ) plotted vs t based on C 1s data for TMA photolysis on rTiO2 using four different light exposures: (1) soft X-rays with a 44 μm monochromator slit, (2) soft X-rays (16 μm slit) + UV light, (3) only UV light (except for the soft X-ray exposure for the two fast scans), and (4) soft X-rays (44 μm slit) + UV light. Focusing on the initial rate constants, the two data points for “only UV light” correspond to a rate constant of 9.0 × 10−3 min−1. This value should be regarded as a lower limit of the initial reaction rate since the coverage after irradiation is located after the point at which the rate constant changes into a lower value. With only soft X-rays (44 μm slit), the time constant is 6.7 × 10−3 min−1. When using soft X-rays (44 μm slit) + UV, the rate constant 1.6 × 10−2 min−1. Thus, the two light sources result in a total rate constant that is the sum of the two individually measured rate constants. Moreover, when decreasing the slit width, the rate for soft X-ray induced reaction C
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
UV light display a rate reduction from 1.3 × 10−2 to 4.3 × 10−3 min−1 after about 95 min, a point at which the coverage is 0.15 ML. At times longer than 200 min (Θ ≤ 0.10 ML), the rate has decreased even further, down to 1.9 × 10−3 min−1. These results are in line with previous work in which species with different TMA reaction rates are found on the r-TiO2 surface, with lower reaction rates found for TMA adsorbed at or adjacent to oxygen vacancies. Figure 3b compares the photoinduced decay of the TMA coverage on the r-TiO2 and o-TiO2 surfaces, that is, two surfaces with different oxygen vacancy densities. The plots are derived from C 1s PES spectra measured with soft X-rays (44 μm) + UV light. The decay rates on the two surfaces are in all essentials identical, including the rate change at about 75 min. Thus, for both r-TiO2 and o-TiO2 a similar decrease in the decay rate occurs when reaching 0.15 ML. Noteworthy is also that the coverage at which the rate changes is the same for the two slits used for TMA/r-TiO2; i.e., this property is not sensitive to the X-ray fluxes used here. 3.3. Charge Trapping and Defect States. Trapping of electrons is assumed to involve reduction of Ti4+ to Ti3+ at sites adjacent to the hydroxyls formed upon deprotonation of TMAA.16,24 When discussing the amount of charge trapped on the r-TiO2 and o-TiO2 surfaces it is therefore vital to relate the OH coverages on the two surfaces. Figure 4 shows the O 1s
Figure 3. (a) Plots of ln(Θ) vs time where Θ denotes the TMA coverage on r-TiO2(110) (in ML). Four different types of photon irradiation are compared, and numerical values of the rate constants obtained from the slopes of the lines are included. (b) A direct comparison of ln(Θ) vs time for TMA on r-TiO2(110) and oTiO2(110) when irradiating with both X-rays and UV light.
should decrease correspondingly. For soft X-rays (16 μm) + UV the initial rate constant is 1.3 × 10−2 min−1. With a rate constant of 9.0 × 10−3 min−1 for UV light only, this gives a rate constant of 4.0 × 10−3 min−1 for soft X-rays with 16 μm slit. Measurements comparing the intensity from the two slits yields that soft X-rays with 44 μm slit should give a rate constant of about 3 × 10−3 min−1, which is smaller than what we derived from the combined measurement at 16 μm but in good agreement if we consider that the rate value for UV light only is a lower limit. Consequently, the reaction rates from UV light and soft X-rays are close to additive and can as such be controlled: When using a slit of 16 μm, the UV process is clearly dominating over the X-ray-induced process, as shown by finding that the coverage after 156 min illumination by UV only is very close to that attained when using UV and X-rays with 16 μm slit. It is evident that the rate changes to lower values as the TMA coverage decreases (see Figure 3a). For TMA on the r-TiO2 surface the C 1s data measured with soft X-rays (16 μm) and
Figure 4. O 1s core level photoemission spectra showing, first, the clean spectra (black) for the two surfaces, second, the effect of adsorption of TMAA, which results in TMA plus hydroxyls (red), and, third, after subsequent photolysis of most of the formed TMA layer, which leaves mostly hydroxyls on the surface (purple). The substrate component is at lower binding energy while the adsorbate-induced component (TMA + OH) is at higher binding energy.
spectra before and after photolysis on the r-TiO2 and o-TiO2 surfaces (red and purple, respectively). The spectra for the clean surfaces are shown at the bottom for comparison (black). The clean spectra both have a weak feature on the high BE side of the main substrate peak. The structure is always present, even in the case of meticulously clean surfaces and possibly due to satellite states inherent to the O 1s line or structural imperfections in the sample.28,29 However, we note that the spectrum for the clean o-TiO2 surface has a slightly more D
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir intense high BE shoulder, which might be due to adsorption of OH (≤0.02 ML) during the preparation procedure. As noted above, the adsorbate-related peak (the high BE component) after TMAA saturation and before irradiation has a major contribution from the oxygen atoms of the TMA carboxyl group and a minor contribution from OH (expected stoichiometric ratio 2:1). After irradiation with soft X-rays (44 μm) + UV light, the TMA coverage on both surfaces is about 0.1 ML based on the C 1s data, which corresponds to 0.2 ML of carboxyl oxygen atoms. The OH coverage can be assumed to be close to the initial TMA coverage (0.4−0.5 ML). Thus, the contribution from the TMA carboxyl group after irradiation is minor, and the peak is mainly associated with OH. Since the O 1s spectra after adsorption and subsequent reaction on r-TiO2 and o-TiO2 are very similar, we infer that the number of OH groups formed by TMAA deprotonation does not vary significantly with the number of defects. Previous work has demonstrated that it is possible to study charge trapping by monitoring the band gap state intensity in the valence PES spectrum.24 As noted above, trapping of electrons involves reduction of Ti4+ to Ti3+ which leads to increased intensity of the largely Ti 3d derived band gap (BG) state at about 1 eV BE. Spectra of the band gap region for reduced and oxidized surfaces are shown in Figure 5. The spectra are normalized to the number of scans and the current in the synchrotron storage ring. It is instructive to start with rTiO2 (Figure 5a): The band gap state of the clean, reduced surface (1, black line) can be assumed to contain contributions from surplus electrons due to oxygen vacancies and from subsurface Ti interstitials.30−32 Previous experiments have however shown that the contribution from Ti interstitials is minor for samples that have been subjected to only a few sputtering cycles, which is the case here.31 The intensity of the BG state was obtained after subtraction of a Shirley background, indicated by the dashed red line in the figure, which to some extent can be viewed as taking care of the intensity from interstitials and bulk oxygen vacancies. Adsorption of TMA takes place on Ti(5) sites as well as oxygen vacancy sites. Hence, the addition of TMA on vacancy sites leads to an attenuation of the BG state intensity (2, red line). Upon photolysis the BG state intensity increases (3, purple line). The increase from 2 to 3 reflects trapping of electrons at Ti sites close to hydroxyls. It is not caused by a reduced attenuation of the BG state since (1) the hydroxyls are not removed in the photolysis process and (2) TMA from the defect sites react very slowly and most of the TMA adsorbed at vacancy sites are still left at the situation shown.22 That the increased BG intensity is associated with a reduction of Ti from +4 to +3 is verified in the Ti 2p spectra (not shown). For the oxidized surface (o-TiO2), prepared by exposing the r-TiO2 surface to O2, we note a substantial decrease in the intensity of the BG state; see Figure 5b (1, black line). This is attributed to partial healing of the oxygen vacancies and some oxidation of Ti interstitials.22 Exposure of the o-TiO2 surface to TMA results in a slight decrease of the intensity, verifying that very few TMA species are formed at oxygen vacancy sites on the o-TiO2 surface (2, red line). Photolysis of TMA by UV and soft X-ray irradiation again leads to an increase of the BG state intensity (3, purple line). The amount of trapped charge (from 2 to 3) is now larger, resulting in a saturation level of the BG state that is nearly identical to that of the r-TiO2 surface after photolysis. Thus, there appears to be a critical maximum BG state intensity. Since the OH coverages are basically the same
Figure 5. Band gap state (Ti 3d) spectra recorded for the r-TiO2 and o-TiO2 surfaces. Three situations are compared: as prepared surface (labeled 1, black line), after formation of the TMA + OH layer upon TMAA exposure (labeled 2, red line), and after subsequent photolysis using X-rays and UV light (labeled 3, purple line). For situation 2, a Shirley background is indicated (red dashed line).
on the two surfaces, it follows that the charge associated with oxygen vacancies inhibits photochemical charge trapping close to the hydroxyls. The next step is to correlate the BG state intensity to the amount of desorbed TMA. Figure 6 shows the yield of Ti3+ (in ML) derived from the BG intensity as a function of the amount of desorbed TMA (in ML) for the r-TiO2 (Figure 6a) and oTiO2 (Figure 6b) surfaces. The Ti3+ yield in ML was obtained on the basis of previous experiments relating the band gap state intensity to the oxygen vacancy density by the coverage of hydroxyls formed.31 The quantification is based on having no change in the attenuation of the BG state after TMAA dosing for reasons outlined above. We find a near 1:1 correlation between trapped charge and photodepleted TMA, consistent with the expected balance between holes and electrons.12,24 Interestingly, we note that the correlation is less than 1:1 for the oxidized surface in the beginning but reaches a 1:1 ratio after desorption of about 0.15 ML of TMA. This suggests a delay in the trapping during the initial stages. The Ti3+ yield on the r-TiO2 saturates at 0.15 ML desorbed TMA while the Ti3+ yield on o-TiO2 saturates at 0.25 ML desorbed TMA. This is E
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 7. Binding energy separation of the two C 1s components (C1 and C2 in Figure 1a) as a function of TMA coverage. The structure labeled “A” correlates to the undercompensation in the initial charge trapping shown in Figure 6.
Figure 8 compares spectra of the TiO2 valence band edge (VBE), the O 1s core level, and TMA-related valence states for TMA/o-TiO2 before and after photoreaction. Figures 8a and 8b show that the TiO2 VBE and the TiO2 related O 1s peak shift by +0.1 eV upon photolysis, which is attributed to the electron accumulation associated with trapping. However, as noted above, the C 1s peak from carboxylic carbon shifts by +0.34 eV. In an initial state picture, this suggests that the electronic states of the carboxylic group shifts with respect to the substrate states. To verify this, PES spectra of the valence region were analyzed and difference spectra extracted to highlight TMArelated states. The results for TMA/o-TiO2 are shown in Figure 8c. The situations correspond to TMA coverages of 0.42 ML (before irradiation) and 0.14 ML (after irradiation). The TMA related peaks are assigned on the basis of previous work.23 It is found that the states associated with C−O interaction shift by 0.25−0.30 eV to higher BE upon irradiation and photodesorption. In contrast, the states associated with C−H interaction do not exhibit a shift of this magnitude, consistent with the very small shift of the C−H related C 1s peak. We therefore conclude that the BE changes of the C−H and C−O related valence states follow the binding energies of the corresponding C−H and C−O related C 1s core level peaks. [The results for r-TiO2 are very similar (not shown).] The changes in the TMA electronic structure when decreasing the coverage thus involve a BE upshift of the states related to the carboxyl group (HOMO) with respect to the TiO2 levels.
Figure 6. Ti3+ yield (in ML) derived from the BG intensity as a function of the amount of desorbed TMA (in ML) for the r-TiO2 (a) and o-TiO2 (b) surfaces. Note that the initial stages for TMA/o-TiO2 display a charge trapping correlation less than 1:1.
expected given that the reduced surface comprises a higher number of unpaired electrons associated with oxygen vacancies, preventing charge trapping at hydroxyls formed at or close to these defects sites. When adding the amount of Ti3+ present prior to TMA adsorption and photolysis (0.21 ML for the rTiO2 surface and 0.12 ML for the o-TiO2 surface), we find that the maximum total amount of Ti3+ for both surfaces is approximately 0.4 ML. This is close to the saturation coverage of OH formed upon TMAA deprotonation, and we therefore infer that the saturation level of Ti3+ is determined by the amount of OH on the surface. 3.4. TMA Electronic Structure Changes. Both C 1s peaks shift toward higher BE when the TMA coverage decreases by way of photolysis. However, the C 1s peak from carboxylic carbon shifts significantly more than the peak from aliphatic carbon, clearly seen as an increased BE separation between the two peaks. Figure 7 compares the separation of the two C 1s components as a function of coverage for TMA/r-TiO2 and TMA/o-TiO2. Both series show the same trend; that is, the C 1s split increases with decreasing coverage. The increase in the separation of the two C 1s peaks when decreasing the TMA coverage from 0.4 to 0.1 ML amounts to 0.3 eV.
4. DISCUSSION The central topic for the discussion is how the changes in the electronic properties of the studied TMA/TiO2(110) systems relate to the reaction rate. Upon electron−hole pair creation, the TMA reacts by absorption of a hole while the electron can be trapped at a Ti site close to the hydroxyl group. The trapping leads to reduction of Ti4+ to Ti3+, which leads to population of the BG state. However, in the case of r-TiO2, the BG state is populated from the beginning due to oxygen vacancies (and Ti interstitials). This means that adsorption and deprotonation of TMAA on r-TiO2 leads to formation of TMA and two types of OH: nominally neutral OH (with empty trap F
DOI: 10.1021/acs.langmuir.6b02989 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
coverage from 0.10 to 0.08 ML. It is therefore very likely that such a slow reaction was not clearly visible in the STM images. The intermediate rate, which would then be assigned to TMA close to defects, is about 50% of that of TMA at regular sites. This is comparable to the rate decrease by 30% found with STM. In this context, we note that the reaction driven by X-rays only shows a kink at an earlier stage (Figure 3). This is qualitatively consistent with the observation of a higher amount of Ti3+ species for the clean surface in this case. However, as outlined in more detail below, there are other factors that also have to be taken into account when analyzing the reaction rates. Previous work has convincingly demonstrated that the BG state population is decisive for the ensuing photoreactivity, ascribed to the presence of surplus electrons that can annihilate the holes.22,24 Within this model the highest rate is found as long as there are empty trap states near the TMA−hydroxyl configuration. This behavior can be found in Figure 6, which shows that the Ti3+ yield has a near 1:1 correlation with the amount of desorbed TMA. However, on the r-TiO2 surface the Ti3+ yield saturates when there is 0.26 ML of TMA left, which is well before the coverage at which the first rate change is observed (0.15 ML). That is after all the trap states have been populated the reaction can continue for a while at undiminished rate. This is not consistent with a model where the most efficient hole-transfer reaction is strictly that which allows for concomitant charge trapping at nominally neutral OH. Saturation of the BG state intensity defines the point at which no further trapping is possible and where the TMA left is either close to or at defect sites. This is a critical point, which would in principle entail a decreased reaction rate due to the probability for electron−hole recombination at these sites. However, no rate change is observed at this stage, from which follows that the possibility for charge trapping is not the only factor that governs the reaction rate. The o-TiO2 surface shows exactly the same rate change at 0.15 ML as for the r-TiO2 surface. The BG state intensity for the clean o-TiO2 surface is significantly lower than that for the clean r-TiO2 surface. This means that adsorption and deprotonation of TMAA on o-TiO2 yields more neutral OH that can trap charge, as shown in Figure 6. In the case of photolysis of TMA on o-TiO2 the rate is decreased close to the point at which trapping has saturated (0.15 ML of TMA left). This implies that the reaction occurring at TMA coverages
