Tin dioxide (SnO2) has various applications due to its unique surface and electronic properties. These properties are strongly influenced by Sn oxidation states and associated defect chemistries. Recently, the oxidation of volatile organic compounds (VOCs) into less harmful molecules has been demonstrated using SnO2 catalysts. A common VOC, 2-propanol (isopropyl alcohol, IPA), has been used as a model compound to better understand SnO2 reaction kinetics. We have used ambient-pressure x-ray photoelectron spectroscopy (AP-XPS) to characterize the surface chemistry of IPA and O2 mixtures on stoichiometric, unreconstructed SnO2(110)-(1 × 1) surfaces. AP-XPS experiments were performed for IPA pressures ≤3 mbar, various IPA/O2 ratios, and several reaction temperatures. These measurements allowed us to determine the chemical states of adsorbed species on SnO2(110)-(1 × 1) under numerous experimental conditions. We found that both the IPA/O2 ratio and sample temperature strongly influence reaction chemistries. AP-XPS valence-band spectra indicate that the surface was partially reduced from Sn4+ to Sn2+ during reactions with IPA. In situ mass spectrometry and gas-phase AP-XPS results indicate that the main reaction product was acetone under these conditions. For O2 and IPA mixtures, the reaction kinetics substantially increased and the surface remained solely Sn4+. We believe that O2 replenished surface oxygen vacancies and that SnO2 bridging and in-plane oxygen are likely the active oxygen species. Moreover, addition of O2 to the reaction results in a reduction in formation of acetone and an increase in formation of CO2 and H2O. Based on these studies, we have developed a reaction model that describes the catalytic oxidation of IPA on stoichiometric SnO2(110)-(1 × 1) surfaces.

Tin dioxide (SnO2) is utilized for a wide range of applications, including transparent conductors, sensors, and oxidation catalysts.1–5 These applications are based on the ability to control the oxidation state of the tin ions and the formation of oxygen vacancies, both of which change the materials’ conductivity and chemistries. SnO2 oxidation catalysts are of interest due to their low cost, physical and chemical stability, and the low oxygen vacancy formation energy.1,6 A prototypical reaction to evaluate oxidation catalysts is the CO oxidation reaction.7 Early studies demonstrated that SnO2 can catalytically oxidize CO at moderately low temperatures (<423 K).8 During these studies, it was observed that the SnO2 catalyst deactivated over time and the catalyst surface became partially reduced. A Mars–van Krevelen mechanism has been proposed for CO oxidation on SnO2, where CO reacts with lattice oxygen resulting in the formation of CO2 and an oxygen vacancy followed by gas phase O2 reacting with the oxygen vacancy to reoxidize the SnO2 surface.1 

Atomically thin SnO2 sheets were recently found to effectively oxidize CO at significantly lower temperatures than SnO2 nanoparticles.3 The authors proposed that the high fraction of undercoordinated surface atoms accounted for the improved catalytic activity by the optimization of the adsorption, dissociation, and diffusion steps. Further studies have investigated the effect of the surface Sn2+/Sn4+ ratio for SnO2 clusters on CO oxidation rates.9 These studies found that clusters with higher Sn2+/Sn4+ ratios were more active for CO oxidation, and it was proposed that Sn2+ was the active reaction site. CO oxidation properties can also be optimized by controlling the shape of SnO2 nanomaterials.10 It was found that SnO2 nanorods with exposed (110) planes had much higher CO oxidation activities compared to other SnO2 nanomaterials. The authors proposed that the (110) planes have more exposed Sn atoms which could provide more unsaturated-coordination sites.

SnO2 has also been studied for the oxidation of other small molecules, including 2-propanol (isopropyl alcohol, IPA).2,11,12 Studies have indicated that during the catalytic oxidation of IPA by metal oxides, the formation of acetone or propene can be correlated with redox or acidic reaction sites, respectively.11,13 For SnO2, it was found that at 473 K, the acetone turnover frequency was >500× than that for propene.12 These results suggest that the redox mechanism was dominant for SnO2, although details of the reaction sites were not studied in detail. Recently, SnO2 has been studied for the oxidation of volatile organic compounds (VOCs), where IPA was used as a model VOC.2 The complete oxidation of IPA to CO2 and H2O was carried out to characterize the activity of the SnO2 catalyst. These studies found that the complete oxidation of IPA occurred at ∼550 K, which was much lower than that observed on a bare stainless-steel support.

In order to gain a fundamental understanding of alcohol decomposition, studies using well-defined surfaces and controlled alcohol exposures are essential. As mentioned above, nanorods with exposed (110) planes had much higher CO oxidation activities compared to other SnO2 nanomaterials.10 The ideal stoichiometric SnO2(110) surface contains rows of bridging O ions than can be removed simply by heating.4,14,15 When the bridging O atoms are removed, the oxidation state of the surface changes from exclusively Sn4+ to a mixture of Sn2+ and Sn4+.1,4,14 SnO2(110) has several stable surface reconstructions, which allows control of the tin oxidation state and the concentration of oxygen vacancies.14,16–18 This tunability makes SnO2(110) an excellent surface to study oxidation reactions and will allow us to further understand the alcohol adsorption, decomposition, and oxidation, which will allow for improvements in catalyst design.19 

In this study, we have used ambient pressure x-ray photoelectron spectroscopy (AP-XPS)20 to further understand the surface chemistry of IPA on a well-defined SnO2(110) surface. The stoichiometric surface was characterized using x-ray and ultra-violet photoelectron spectroscopy (XPS and UPS), as well as low energy electron diffraction (LEED). AP-XPS was performed to determine the effect of IPA pressure and sample temperature on the adsorbed species and to study the oxidation state of the Sn during IPA reactions. Mass spectrometry was used to measure gas phase products during surface reactions. Finally, we propose a reaction mechanism that describes the oxidation of IPA on stoichiometric SnO2(110).

Experiments were performed in a custom-built SPECS ambient pressure x-ray photoelectron spectroscopy system.20 The system has an ultrahigh vacuum (UHV) manipulator and an ambient pressure (AP) cell which can be transferred into the analysis chamber. The AP cell can be pressurized up to ∼10 mbar while maintaining a base pressure of 1 × 10−9 mbar in the surrounding chamber. XPS and UPS experiments will be defined as being performed on the UHV manipulator, while AP-XPS experiments will be defined as being performed in the AP cell regardless of the pressure. A connected sample preparation chamber has a LEED system and a four pocket e-beam evaporator. Samples can be introduced into the system using a load lock. A third chamber has a variable temperature and pressure scanning tunneling microscope. The analysis, preparation, and microscopy chambers have ion pumps, turbo molecular pumps, and titanium sublimation pumps and achieve base pressures <2 × 10−10 mbar.

The gas manifolds for the system are connected to both an Edwards nXDS6i scroll pump and a HiCube 80 Pfeiffer Turbo. A base pressure of <1 × 10−5 mbar was achieved for the manifolds. The IPA (Absolute AR ACS Grade, Macron Fine Chemicals) was loaded into a Schlenk vial (Wilmad Lab Glass) and connected to one manifold. Several freeze-pump-thaw cycles were used to degas the IPA before each experiment. A second manifold was connected to an O2 lecture bottle (99.997% Matheson Purity, Matheson). Two high-precision variable leak valves were used to control the pressure of each gas in the ambient pressure cell. A residual gas analyzer (RGA200, Stanford Research Systems) was located in the differentially pumped portion of the electron analyzer lens system and was used to obtain mass spectra from the AP cell and verify the purity of the IPA and O2.

A 10 × 10 mm2 natural SnO2(110) (±0.5°) single crystal was purchased from SurfaceNet GmbH. The crystal was sonicated in IPA (ACS grade, >99.5%) for 5 min and mounted on a Ta sample holder. The sample holder had a type K thermocouple spot welded next to the sample. The sample was loaded into the SPECS AP-XPS system through the load lock and mounted onto the UHV manipulator. Sample cleaning and preparation was performed based on literature studies.1,14 Numerous cycles of Ar+ sputtering at 1 kV with ∼1.5 µA sample current for 600 s were performed with the sample at room temperature (RT). This procedure was followed by flashing to 700 K. These sputtering cycles were performed until no impurities and adventitious carbon were detected with XPS. The stoichiometric oxidized surface used for experiments was prepared by Ar+ sputtering using a 10 × 10 mm2 raster and ∼1.5 µA sample current for 600 s at RT. The sample was then transferred to the AP cell and annealed to 580 K under UHV (base pressure <3 × 10−9 inside the cell). During annealing, O2 was added to the cell until the O2 partial pressure (PO2) was 1–3 mbar, and the sample was held at 580 K for 10 min under these conditions. The cell was then pumped to UHV while the sample was still at 580 K. Once the pressure in the AP cell was below 1 × 10−8 mbar, the sample was flashed to 750 K. The PO2 was increased to 1–3 mbar after the sample cooled to 600 K. This second oxidation step was performed for 10 min. The sample was then cooled to 380 K while keeping PO2 ∼ 1–3 mbar to prevent the formation of defects on the stoichiometric oxide surface.4 The AP cell was pumped to UHV once the sample was below 380 K, and the sample was cooled to RT. During the sample preparation, the RGA was initialized and background mass spectra were obtained prior to the introduction of any gas into the AP cell. The RGA continuously collected mass spectra during the duration of the oxidation procedure.

AP-XPS and XPS were performed on the C 1s, O 1s, and Sn 3d core-levels; Sn MNN Auger; and valence-levels using monochromatized Al Kα radiation (hν = 1486.6 eV, 50 W, and 15 kV). The electron analyzer pass energy was set to 35 eV, and normal emission was used. The x-ray beam diameter on the substrate was ∼300 µm. Our AP-XPS and XPS data were charge corrected to the C 1s aliphatic carbon binding energy (Eb) at 284.8 eV.21 The spectra were fit using Gaussian–Lorentzian line shapes with a linear background. Ultra-violet photoelectron spectra were obtained using a SPECS UVS 10/35 operating at 1 × 10−7 mbar of He. Ultra-violet photoelectron spectra were collected with He I (21.2 eV) both with a grounded sample and +10 V bias applied to determine the ionization potential. A SPECS ErLEED 150 was used to obtain diffraction patterns of the prepared surface. A cathode current of 1.3 A and screen voltage of 5 kV were used. The beam energy was set to 90 eV, yielding a beam current of ∼1.1 μA. Images were collected across the sample surface to ensure uniformity, which was performed by the observation of a consistent LEED pattern. Beam energies of 60 eV and 135 eV were additionally used. After the oxidation process, a (1 × 1) LEED pattern was obtained, suggesting an unreconstructed SnO2(110)-(1 × 1) surface. LEED patterns and valence-level spectra are available in the supplementary material. All ball-and-stick model images were made using the Avogadro 1.2.0 software.

Pressure dependent IPA experiments were performed at RT. AP-XPS data were obtained for the C 1s, O 1s, and Sn 3d core-levels; Sn MNN Auger; and valence-level at each IPA pressure. Temperature dependent experiments were performed immediately after the sample cooled to 380 K following oxidation. This was done to save time and avoid potential contamination from exposure to low vacuum. After obtaining the desired IPA partial pressure (PIPA) or O2/IPA ratio, the temperature was raised to 400 K. AP-XPS measurements were obtained for all the previously mentioned transitions. AP-XPS measurements of the gas phase C 1s and O 1s were then obtained by moving the sample away from the analysis cone (∼1 mm to 2 mm). Experiments were also performed at 500 K and 600 K. The RGA was constantly collecting partial pressure vs temperature mass spectra (P vs T) from the beginning of the oxidation throughout the entire AP-XPS experiment.

Since an e-beam is used to heat the sample, background experiments were performed without the SnO2(110) sample mounted on the sample holder. For each temperature and pressure condition, both gas phase AP-XPS and mass spectra were obtained. To illustrate the chemistry due to the SnO2 sample, we have subtracted these background mass spectra from those obtained when the SnO2 sample was present for experiments performed under identical conditions. The ratios of each mass to m/z 45 (IPA) were calculated for each temperature and averaged over three measurements with no SnO2 sample. These ratios were then multiplied by the m/z 45 intensity in the main experiment with the SnO2 sample, with respect to each temperature used, to obtain a background value that was subtracted from each respective mass.

AP-XPS spectra were additionally normalized to the low binding energy background for each transition to account for changes in intensity due to pressure changes and attenuation through different gas molecules in the AP cell during measurements. This was done by taking the average area under the spectra for a 1 eV binding energy window ∼5 eV below the main transition and then dividing all the intensities by this value. This method sets the lower binding energy background to the same intensity.

To evaluate the pressure dependence of IPA adsorption on the stoichiometric SnO2(110) surface, we have varied the PIPA between 0.1 mbar and 3 mbar for a sample at RT. AP-XPS was performed for the C 1s, O 1s, and Sn 3d core-levels; Sn MNN Auger; and valence-level at each pressure. AP-XPS was also conducted after exposure while evacuating IPA from the AP cell, and XPS was performed after transferring the sample to the UHV manipulator. Figure 1(a) shows the C 1s spectra for the indicated sample conditions. The bottom most AP-XPS data are for the clean surface after evacuating O2 from the AP cell. These data show that the surface does not have detectable carbon contamination, and the surface is a good starting point for these studies. During IPA exposure, adsorbed species were observed on the surface at the lowest pressure studied (PIPA = 0.1 mbar). We found that the adsorbed species were stable on the surface, both during exposure and after pumping out the IPA (spectrum labeled Pumpdown). Even for relatively low IPA pressures (PIPA = 0.1 mbar), gas phase C 1s peaks for IPA become apparent and overlap with the adsorbed IPA carbon peaks. We deconvoluted the spectra by constraining the area ratio of C—H:C—O to be 2:1 and fixing the full width at half maximum (FWHM) to be identical for both peaks. Gas phase spectra were obtained with the sample moved back from the analysis position to determine baseline gas phase peak fitting parameters. The AP-XPS data obtained for PIPA = 0.1 mbar require four peaks to fit the spectra, where aliphatic carbon (C—H) has Eb = 284.8 eV (blue), carbon atoms singly bound to oxygen (C—O) have Eb = 286.3 eV (red), gas phase C—H has Eb = 286.0 eV (green), and gas phase C—O has Eb = 287.6 eV (purple).21–23 As IPA pressure is increased, the adsorbate intensities stay roughly constant, while the gas phase peak intensities increase significantly. To estimate the IPA coverage on the SnO2(110) surface, we have used a standard nonattenuating overlayer model.24 This model requires the C 1s adsorbate and Sn 3d substrate peak intensities (Nx), values for the differential orbital photoionization cross sections (dσx/dΩ), the electron attenuation lengths in the substrate (Λe), and the SnO2 interlayer distance (d). Differential photoionization cross sections were calculated using Eq. (89) from Ref. 24, published photoionization cross sections,25 a 60° x-ray source-to-analyzer-angle, and normal electron emission (θ = 90°). The attenuation length was calculated using Eq. (15) from Ref. 26. Rewriting Eq. (120b) from Ref. 24 and assuming minimal changes in the system detection efficiency, the solid angle, and the attenuation of the electrons in the gas phase, we can calculate the IPA fractional coverage using the following equation:

ss=Nc(dσSn/dΩ)ΛesinθNSn(dσc/dΩ)d
(1)

where s′/s is the fractional monolayer coverage of IPA with respect to Sn cation sites at the (110) surface. The coverage can be determined for all three pressures using the C 1s (C—O) and Sn 3d peak intensities. The saturation coverage was estimated to be 0.19 ± 0.01 monolayers (ML), which suggests that at PIPA = 0.1 mbar, the surface is saturated with approximately a 1/5 monolayer of IPA. After evacuating the AP cell, we find that the adsorbed species corresponding to IPA remains, although there is a slight increase in the C—H/C—O intensity ratio likely due to an increase in aliphatic carbon from the adsorption of adventitious carbon. After transferring the sample to the UHV manipulator, a carbon doubly bound to oxygen (C=O) peak is observed at Eb = 289.2 eV.21 We believe that this is most likely due to adventitious contamination that occurs during the time it takes to fully pump down the AP cell before transferring to the UHV manipulator.

FIG. 1.

AP-XPS and XPS for (a) C 1s and (b) O 1s from stoichiometric SnO2(110)-(1 × 1) before and after exposures to IPA. AP-XPS was performed for PIPA = 0 mbar, 0.1 mbar, 1 mbar, and 3 mbar at room temperature and after evacuation of the AP cell.

FIG. 1.

AP-XPS and XPS for (a) C 1s and (b) O 1s from stoichiometric SnO2(110)-(1 × 1) before and after exposures to IPA. AP-XPS was performed for PIPA = 0 mbar, 0.1 mbar, 1 mbar, and 3 mbar at room temperature and after evacuation of the AP cell.

Close modal

Figure 1(b) shows the O 1s spectra for the indicated sample conditions. The bottom most AP-XPS data are for the clean surface after evacuating O2 from the AP cell. This spectrum can be fit with two peaks. The most intense peak, with Eb = 530.3 eV (blue), corresponds to lattice oxygen27,28 and the high binding energy shoulder, with Eb = 532.4 eV (red), is likely due to surface hydroxyls (O—H) or adsorbed O2 [O2(ad)].27–30 No significant changes in the O 1s spectra take place as the IPA pressure is increased to PIPA = 0.1 mbar, other than a slight increase in intensity for the high binding energy shoulder. This is likely due to the adsorption of IPA onto the surface, and the high binding energy shoulder is due to the oxygen singly bound to carbon (O—C).21,22 The gas phase oxygen singly bound to carbon (O—C) peak is not visible until PIPA = 1 mbar and corresponds to Eb = 534 eV (gray).22 Increasing the IPA pressure to PIPA = 3 mbar results in an increase in intensity for the gas phase IPA peaks. No significant changes were observed after evacuating the AP cell (Pumpdown) or after transferring to the UHV manipulator. Due to the limited changes in the O 1s spectra, we have focused primarily on the C 1s spectra in the following experiments. Furthermore, no significant changes were observed after exposure to IPA at RT for the Sn 3d and Sn MNN spectra or the valence-level spectra. This suggests that IPA exposures at RT do not significantly change the surface oxidation state of SnO2.

To evaluate the temperature dependence of IPA adsorption on the stoichiometric SnO2(110)-(1 × 1) surface, we have set PIPA = 0.5 mbar and varied the sample temperature. As mentioned earlier, studies have shown that SnO2 thin films can achieve 100% IPA conversion for a reaction stream containing 1000 ppm IPA/20% O2/9.9% He/70% N2.2 The temperatures for our experiments were chosen based on the reported conversion rates vs temperature, where 0% conversion occurred at 400 K, ∼60% occurred at 500 K, and 100% occurred at 600 K.2 Our initial set of experiments focused on exposure to IPA only. The AP-XPS C 1s results for PIPA = 0.5 mbar and T = 400 K, 500 K, and 600 K are shown in Fig. 2(a). PIPA = 0.5 mbar was chosen since the C 1s intensities from the gas phase IPA peaks were high enough to distinguish them from the adsorbed IPA. This allowed a relatively straight forward deconvolution of the contribution from adsorbates. For the C 1s spectra, the C—H:C—O ratio was constrained to 2:1 for both adsorbed and gas phase species, and the FWHM were fixed independently for the adsorbed and gas phases. No significant difference in the AP-XPS data was observed between the 400 K spectrum compared to the RT spectrum shown in Fig. 1(a). However, at 500 K, a new higher energy peak (brown) at Eb = 289.0 eV becomes evident, which is consistent with a gas phase —C=O species.22,31 At 600 K, another peak (teal) at Eb = 287.9 eV was observed, while the peaks associated with adsorbed IPA (C—H, blue and C—O, red) significantly decreased in intensity. This new peak is consistent with a doubly bound carbon (—C=O) adsorbed to the surface. Based on the relative binding energy shift between the —C=O and —C—H peaks and their relative intensities, we believe that this peak is due to adsorbed acetone.21,22,31 After evacuating the cell, only aliphatic carbon remains on the surface. Figure 2(b) shows AP-XPS data for the valence-level spectra at each temperature. As the temperature is increased, there is an increase in intensity in the bandgap region at the top of the valence band maximum. Dashed lines are provided to help illustrate the relative increase in intensity as the temperature increases. This increase in the density of states in the bandgap can be correlated with an increase in Sn2+ character at the surface.14–16 This is a result of oxygen vacancies being formed through the reaction of oxygen anions with IPA.14–16,32,33 With the substrate as the only source of oxygen, this results in the partial oxidation of IPA to acetone.12,33 LEED patterns after the reaction were still (1 × 1), but more diffuse than before the reaction suggesting a more disordered surface. This can be due to oxygen vacancy formation at the surface and/or adsorbed species at the surface.

FIG. 2.

AP-XPS of (a) the C 1s core-level and (b) valence-level obtained at PIPA = 0.5 mbar for 400 K, 500 K, and 600 K, as well as after exposure under UHV.

FIG. 2.

AP-XPS of (a) the C 1s core-level and (b) valence-level obtained at PIPA = 0.5 mbar for 400 K, 500 K, and 600 K, as well as after exposure under UHV.

Close modal

Mass spectrometry was used to measure the partial pressures of the reaction products to obtain further information on the catalytic reactions. Gas phase AP-XPS and mass spectra were used to correlate the peak binding energies with specific molecules. Procedures and results are described in the supplementary material. Figure 3 tracks the mass/charge (m/z) ratios at several reaction temperatures for several different molecules to better understand the reaction chemistries. These molecules include acetone (m/z = 58, 43), propene (m/z = 41), CO or N2 (m/z = 28), methanol (m/z = 31), CO2 (m/z = 44), water (m/z = 18), hydrogen (m/z = 2), oxygen (m/z = 32), and IPA (m/z = 45). Although m/z = 43 could be related to other carboxyl species, including acetic acid, which could be a potential intermediate, our analysis suggests that this peak is only representative of gas phase acetone for these experiments. A discussion of our analysis is provided in the supplementary material. As the sample temperature is increased to 400 K and 500 K, the IPA signal (gray) is the most intense, while propene (orange), CO/N2 (green), and CO2 (blue) are also observed. The CO/N2 and CO2 intensities are two orders of magnitude below IPA with low signal to noise ratios. This is likely due to slight deviations in background intensities for the experiments. At these sample temperatures, the propene m/z 41 intensity is very low compared to the IPA (m/z = 45) signal and also expected due to fragmentation in the mass spectra.34 Although a gas phase —C=O species was initially seen in the AP-XPS data at 500 K, this species was not detected in the mass spectra. This is likely due to the low sensitivity of the mass spectrometer due to the large distance from the AP cell. Once the reaction temperature increases to 600 K, there is a significant change in the mass spectra. For example, the IPA partial pressure decreased substantially, while the acetone, hydrogen, propene, CO/N2, CO2, and water partial pressures increased significantly. Based on signal intensity, the primary products at 600 K are H2 (pink), propene (orange), water (teal, overlapped with propene), and acetone (yellow). Upon completion of the reaction at 600 K, the reaction temperature was reduced to 400 K. As shown in Fig. 3, the IPA signal returned to its original baseline, suggesting that the changes in partial pressures were solely related to reactions between IPA and the SnO2(110)-(1 × 1) surface.

FIG. 3.

Mass spectra after subtracting off background signals for acetone (m/z 43, yellow and m/z 58, red), propene (m/z 41, orange), CO or N2 (m/z, 28 green), methanol (m/z 31, lime), water (m/z 18, teal), CO2 (m/z 44, blue), hydrogen (m/z 2, pink), O2 (m/z 32, black), and IPA (m/z 45, gray) at PIPA = 0.5 mbar.

FIG. 3.

Mass spectra after subtracting off background signals for acetone (m/z 43, yellow and m/z 58, red), propene (m/z 41, orange), CO or N2 (m/z, 28 green), methanol (m/z 31, lime), water (m/z 18, teal), CO2 (m/z 44, blue), hydrogen (m/z 2, pink), O2 (m/z 32, black), and IPA (m/z 45, gray) at PIPA = 0.5 mbar.

Close modal

As mentioned earlier in the section titled Experimental, the background was subtracted from the mass spectra by conducting identical experiments with no sample. This subtraction of the background mass spectra does not work well for dynamic changes in temperature, specifically when ramping between temperatures. This is likely due to large variations in the filament temperature depending on the sample holder during the sample temperature ramp. Therefore, the data obtained during temperature changes (signals close to black bars on graph) are not used to understand the reactions. Additionally, moving the sample away from the cone aperture for gas phase measurements (done toward the end of each temperature) affects the flow rate of gas through the AP cell. This leads to a slight change in IPA partial pressure, which is why there is a slight increase just before each temperature change. Ultimately, the mass spectra indicate that acetone, hydrogen, propene, and water are formed through reactions on SnO2. General dehydrogenation [reaction (R1)] and dehydration [reaction (R2)] are shown below and are consistent with prior experiments performed on SnO2 catalysts,12 

CH3CHOHCH3gCH3COCH3g+H2g,
(R1)
CH3CHOCH3gCH3CH=CH2g+H2Og.
(R2)

Our mass spectrometry results appear to suggest that the reaction of IPA at 600 K on SnO2 is selective toward the formation of propene over acetone. However, the relative intensities for each mass for our experiments have not been calibrated to the quantity of the species. To better understand the relative intensities of propene and acetone, we use Fig. S3a from the supplementary material, which measures the C 1s AP-XPS spectra for the gas phase during reaction. Three peaks are observed at EB = 286.7 eV, 288.1 eV, and 289.5 eV, which are correlated with gas phase C—H, C—O, and C=O, respectively. The two lower energy peaks correspond to the IPA doublet, the lowest and highest energy peaks correspond to the acetone doublet, and the lowest energy peak correspond to the propene. Knowing that the relative intensity for the C—H peak should be twice the intensity of the C—H peak for both acetone and IPA and that the C—H intensity can also be due to propene, we can estimate the amount of propene formed by peak fitting the spectra and holding the C—H to C—O and C—H to C=O to 2:1. Any remaining intensity for the C—H component can then be related to propene. Using this approach, we find that the amount of propene formed is negligible. As such, the reaction of IPA with SnO2 results primarily in acetone.

We can write the detailed mechanism for IPA decomposition to acetone as follows, where (g) represents gas phase, (l) lattice oxygen, (b) bridging oxygen, (a) adsorbed species, and VO oxygen vacancies:

CH3CHOHCH3g+ObCH3CHOCH3a+OH(a),
(R3)
CH3CHOCH3a+ObCH3COCH3a+OH(a).
(R4)

After heating to 600 K,

OHa+OHaH2g+2O(b),
(R5)
OHa+OHaH2Oa+O(b),
(R6)
H2OaH2Og+VO.
(R7)

The hydroxyl groups in reactions (R3)–(R6) are likely correlated with the bridging oxygens, although in-plane oxygen anions could contribute as well.35,36 The SnO2 surface was reduced during reactions with IPA due to the formation of oxygen vacancies during the desorption of water [reaction (R7)]. We propose that the desorbed water forms an oxygen vacancy. However, it should be noted that adsorbed oxygen species may remain on the surface after the high pressure oxidation process and these may result in more complicated interpretation of the chemistries.30 The stoichiometric reaction for the complete oxidation of IPA by O2 is given by the following reaction:

2CH3CHOHCH3g+9O2g6CO2g+8H2O(g).
(R8)

For the next round of experiments, we used a 9:2 ratio of O2:IPA to further our understanding of the catalytic oxidation of IPA by the SnO2(110)-(1 × 1) surface. In Figs. 4(a) and 4(b), we show AP-XPS results for the C 1s and valence-level data, respectively, with PIPA = 0.2 mbar and PO2 = 0.9 mbar. The total pressure for the AP cell was optimized so that we could get comparable signal to noise ratios for the same data acquisition time used for the PIPA = 0.5 mbar experiments. This corresponds to a lower signal intensity for gas phase carbon species compared to Fig. 2(a). In Fig. 4(a), we use the same fitted peak colors as in Fig. 2(a) to distinguish each component. At 400 K, an additional C 1s peak is observed at Eb = 289.0 eV (brown) which is due to —C=O gas phase species. Increasing the reaction temperature to 500 K, we find evidence of carboxyl species [O—C=O, Eb = 289.2 eV (gold)]31 while the adsorbed IPA intensity is reduced. Further increasing the reaction temperature to 600 K, we see the appearance of the adsorbed —C=O peak [Eb = 288.0 eV (teal)] also seen in Fig. 2(a). Once the cell is evacuated to UHV, there is O—C=O (gold) and —C=O (teal) on the surface. No evidence of adsorbed O—C=O was seen when reacting in only IPA, as shown in Fig. 2(a). Additionally, only adsorbed alcohol was seen with UHV XPS after reacting solely in IPA, while adsorbed —C=O and O—C=O are seen following the reaction in 9:2 O2:IPA. The valence-level spectra in Fig. 4(b) also indicate differences for reactions performed in mixtures of IPA/O2 compared to IPA alone. With O2 present, there was little to no reduction of the SnO2 surface, which suggests that the oxygen vacancies formed during reactions with IPA are being replenished by reactions with O2.

FIG. 4.

AP-XPS for (a) C 1s and (b) valence-level at 9:2 PO2:PIPA for 400 K, 500 K, and 600 K, as well as after exposure under UHV.

FIG. 4.

AP-XPS for (a) C 1s and (b) valence-level at 9:2 PO2:PIPA for 400 K, 500 K, and 600 K, as well as after exposure under UHV.

Close modal

Mass spectrometry was also used for the 9:2 PO2:PIPA exposure to monitor reaction products. Figure 5 shows the partial pressures for reactions performed in 9:2 PO2:PIPA while tracking the same masses as in Fig. 3. No reaction products were observed for reactions at 400 K. Again, at this temperature, the propene (orange) is a fraction of the IPA signal, as is the methanol (lime) which is a fraction of the oxygen signal (m/z 32) detected by our spectrometer. However, increasing the reaction temperature to 500 K, there was a decrease in IPA and an increase in water (teal), hydrogen (pink), propene (orange), and acetone (yellow) intensities. At 500 K, there was only a trace amount of CO2 produced. Increasing the reaction temperature to 600 K resulted in the further decrease in the IPA and an increase in the water intensity. At the same time, the acetone intensity significantly decreased, while there was a significant increase in CO2 formation. This time, again based on signal intensity, the primary products at 600 K are CO2 (blue), H2 (pink), propene (orange), water (teal, overlapped with propene), and CO (green). During the experiment, the O2 partial pressure decreased continuously. At the end of the experiment, the reaction temperature was lowered to 400 K and the IPA partial pressure returned to its original baseline; however, the O2 partial pressure only reached about 2/3 its original baseline. This suggests that O2 was consumed marginally faster than the delivery rate of O2 into the AP cell during the reaction. This may have unintentionally changed the PO2:PIPA ratio slightly during the experiment. These results indicate that the SnO2 is effective in the oxidation of IPA to CO2 when using IPA/O2 mixtures. LEED patterns after the reaction were still (1 × 1) but again more diffuse than before the reaction suggesting a more disordered surface. This can be due to adsorbed species at the surface.

FIG. 5.

Mass spectra after subtracting off background signals for acetone (m/z 43, yellow and m/z 58, red), propene (m/z 41, orange), CO or N2 (m/z, 28 green), methanol (m/z 31, lime), water (m/z 18, teal), CO2 (m/z 44, blue), hydrogen (m/z 2, pink), O2 (m/z 32, black), and IPA (m/z 45, gray) at 9:2 PO2:PIPA.

FIG. 5.

Mass spectra after subtracting off background signals for acetone (m/z 43, yellow and m/z 58, red), propene (m/z 41, orange), CO or N2 (m/z, 28 green), methanol (m/z 31, lime), water (m/z 18, teal), CO2 (m/z 44, blue), hydrogen (m/z 2, pink), O2 (m/z 32, black), and IPA (m/z 45, gray) at 9:2 PO2:PIPA.

Close modal

These studies suggest that the oxygen anions take part in the reaction and that the resulting vacancies are filled by O2, which is consistent with a Mars–van Krevelen mechanism.37,38 This has also been seen on TiO2(110),32,33 which shares the same surface structure as SnO2(110). The first step in the proposed reaction mechanism is the adsorption of IPA onto Sn sites.39 The hydrogen from the alcohol group reacts with bridging or in-plane oxygen, which results in the formation of a bridging or terminal hydroxyls, respectively, and isopropoxide.12,22,32,33 The next step in the reaction is the transfer of the hydrogen bonded to the alpha carbon in the IPA to another oxygen anion, which forms another bridging or terminal hydroxyl.33 This reaction results in the formation of acetone that can desorb from the surface. The surface hydroxyls can react to form water and an oxygen vacancy,22,33,39 or two hydrogens can combine to form molecular hydrogen.12 Either water or molecular hydrogen can then desorb from the surface.12,22,39 During the reaction, the oxygen anions are constantly being consumed through the adsorption and reaction with IPA and the resulting desorption of water. Due to the limited amount of oxygen available on the SnO2(110) surface, only the partial oxidation of IPA to acetone occurs if O2 is not present.33 In the case where O2 is present, the oxygen vacancies can be replenished which allows acetone production at lower temperatures.33 Additionally, adsorbed IPA can continue to react with the replenished oxygen vacancies resulting in the formation of acetic acid, CO intermediates, and ultimately CO2.22 The proposed reaction pathway can occur through the methoxylation of the bridging oxygen ions,

CH3COCH3a+2O(b)CH3COOa+OCH3(a)+VO,
(R9)
CH3COOa+ObOCH3a+CO2g+VO,
(R10)
OCH3a+3Ob3OHa+CO(a)+VO,
(R11)
COaCOg,
(R12)
COa+ObCO2g+VO,
(R13)
O2g+2VO2Ob.
(R14)

Reaction (R14) represents the ambient oxygen replenishing the surface through a Mars–van Krevelen mechanism. Figure 6 shows a schematic for the proposed reaction mechanism for IPA oxidation on stoichiometric SnO2(110)-(1 × 1).

FIG. 6.

Schematic of the mechanism for IPA adsorption and decomposition on a stoichiometric SnO2(110)-(1 × 1) surface with and without O2.

FIG. 6.

Schematic of the mechanism for IPA adsorption and decomposition on a stoichiometric SnO2(110)-(1 × 1) surface with and without O2.

Close modal

Understanding the reaction mechanisms for the oxidation of IPA provides further details on the SnO2 surface during catalytic reactions. AP-XPS studies, for a range of experimental conditions, provided significant insight into the oxidation of IPA to either acetone or CO2 on the stoichiometric SnO2(110)-(1 × 1) surface. We propose that the reaction occurs through a Mars–van Krevelen mechanism due to the reduction of the SnO2 surface without O2 as a reactant. The addition of O2 as a reactant results in a fully oxidized surface during the IPA oxidation reaction. Our results agree with the literature, which suggest that the thermal decomposition of IPA on SnO2 should occur through a dehydration reaction that produces water.11,12,40 However, prior studies have suggested that propene formation can be favored over acetone formation for reactions of IPA on some metal oxide surfaces.11,12,40 SnO2 has been shown to be active toward both IPA decomposition routes.12 Kinetic data have shown that metal oxide surfaces exhibit high selectivity between acetone or propene based on the availability of redox or acidic sites.12 Redox sites show preferential conversion to acetone, while acidic sites yield propene.12 The adsorption and desorption of bridging oxygen provide available redox sites on the stoichiometric Sn(110) surface.41 One must also consider the impact of photon exposure leading to excited oxygen species from interaction with photoelectrons. Acetone has been shown to be the primary pathway favored over propene formation for photocatalytic dehydrogenation of IPA on TiO2.33 Further studies where we prepare surfaces with varying numbers of oxygen vacancies will allow us to further assess the reaction mechanism. By extension, these studies will enable us to further develop a molecular-level understanding of IPA reactions on well-defined tin dioxide surfaces.

We have used AP-XPS and in situ mass spectrometry to further the understanding of IPA oxidation reactions on the SnO2(110)-(1 × 1) surface. Our results show that IPA adsorbs to a stoichiometrically prepared SnO2(110) following exposures to PIPA = 0.5 mbar at 400 K. Increasing the temperature to 600 K leads to the selective conversion of IPA to acetone. During these reactions, the SnO2(110) surface is reduced, where Sn2+ states are observed in the bandgap. Using 9:2 O2:IPA ratios during reactions results in the formation of acetone at 500 K. At 600 K, there is a significant reduction in IPA and O2 along with the formation of CO2 and water. Using 9:2 O2:IPA ratios during these reactions, the SnO2(110) surface remains oxidized. These results suggest a Mars–van Krevelen mechanism for the oxidation of IPA on stoichiometric SnO2(110)-(1 × 1).

See the supplementary material for more information on the surface structure and oxidation state characterization of the stoichiometric SnO2(110)-(1 × 1) surface, gas phase AP-XPS during reaction experiments, and a description of the mass spectra analysis.

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

This research was funded by the Center for Sustainable Materials Chemistry (CSMC) supported by the U.S. National Science Foundation under Grant No. CHE-1606982. This work was performed, in part, at the Northwest Nanotechnology Infrastructure, a member of the National Nanotechnology Coordinated Infrastructure, which is supported by the National Science Foundation under Grant No. NNCI-1542101. Acquisition of the ambient-pressure x-ray photoelectron spectroscopy/ambient-pressure scanning tunneling microscopy system was supported by the National Science Foundation-Major Research Instrumentation program (Grant No. DMR-1429765), the M. J. Murdock Charitable Trust, Oregon BEST, Oregon Nanoscience and Microtechnologies Institute, and Oregon State University.

1.
M.
Batzill
and
U.
Diebold
,
Prog. Surf. Sci.
79
,
47
(
2005
).
2.
J.
Leclercq
,
F.
Giraud
,
D.
Bianchi
,
K.
Fiaty
, and
F.
Gaillard
,
Catal. Commun.
46
,
192
(
2014
).
3.
Y.
Sun
,
F.
Lei
,
S.
Gao
,
B.
Pan
,
J.
Zhou
, and
Y.
Xie
,
Angew. Chem., Int. Ed.
52
,
10569
(
2013
).
4.
V. E.
Henrich
and
P. A.
Cox
,
The Surface Science of Metal Oxides
(
Cambridge University Press
,
Cambridge, UK
,
1994
), pp.
1
3
;
149
157
.
5.
Y. F.
Sun
,
S. B.
Liu
,
F. L.
Meng
,
J. Y.
Liu
,
Z.
Jin
,
L. T.
Kong
, and
J. H.
Liu
,
Sensors
12
,
2610
(
2012
).
6.
J.
Oviedo
and
M. J.
Gillan
,
Surf. Sci.
467
,
35
(
2000
).
7.
H.-J.
Freund
,
G.
Meijer
,
M.
Schefler
,
R.
Schlögl
, and
M.
Wolf
,
Angew. Chem., Int. Ed.
50
,
10064
(
2011
).
8.
M. J.
Fuller
and
M. E.
Warwick
,
J. Catal.
29
,
441
(
1973
).
9.
Y.
Inomata
,
K.
Albrecht
, and
K.
Yamamoto
,
ACS Catal.
8
,
451
(
2018
).
10.
X.
Wang
,
L.
Xiao
,
H.
Peng
,
W.
Liu
, and
X.
Xu
,
J. Mater. Chem. A
2
,
5616
(
2014
).
11.
P. A.
Christensen
,
Z. T. A. W.
Mashhadani
, and
A. H. B.
Md Ali
,
Phys. Chem. Chem. Phys.
20
,
9053
(
2018
).
12.
D.
Kulkarni
and
I. E.
Wachs
,
Appl. Catal., A
237
,
121
(
2002
).
13.
D.
Haffad
,
A.
Chambellan
, and
J. C.
Lavalley
,
J. Mol. Catal. A: Chem.
168
,
153
(
2001
).
14.
D. F.
Cox
,
T. B.
Fryberger
, and
S.
Semancik
,
Phys. Rev. B
38
(
3
),
2072
(
1988
).
15.
S.
Das
and
V.
Jayaraman
,
Prog. Mater. Sci.
66
,
112
(
2014
).
16.
M.
Batzill
,
K.
Katsiev
, and
U.
Diebold
,
Surf. Sci.
529
(
3
),
295
(
2003
).
17.
P.
Ágoston
and
K.
Albe
,
Surf. Sci.
605
,
714
(
2011
).
18.
C. L.
Pang
,
S. A.
Haycock
,
H.
Raza
,
P. J.
Møller
, and
G.
Thornton
,
Phys. Rev. B
62
,
R7775
(
2000
).
19.
C.
Arble
,
M.
Jia
, and
J. T.
Newberg
,
Surf. Sci. Rep.
73
,
37
(
2018
).
20.
See nnci.oregonstate.edu/apscl for the Ambient Pressure Surface Characterization Laboratory.
21.
J. F.
Moulder
,
W. F.
Stickle
,
P. E.
Sobol
, and
K. D.
Bomben
,
Handbook of X-Ray Photoelectron Spectroscopy
(
Perkin-Elmer Corporation: Physical Electronics Division
,
Eden Prairie, MN
,
1992
).
22.
Y.
Zhang
,
A.
Savara
, and
D. R.
Mullins
,
J. Phys. Chem. C
121
,
23436
(
2017
).
23.
A.
Jürgensen
,
N.
Heutz
,
H.
Raschke
,
K.
Merz
, and
R.
Hergenröder
,
Anal. Chem.
87
,
7848
(
2015
).
24.
C. S.
Fadley
, in
Electron Spectroscopy: Theory, Techniques and Applications
, edited by
C. R.
Brundle
and
A. D.
Baker
(
Academic Press
,
London, England, UK
,
1978
), pp.
23
75
.
25.
J. J.
Yeh
and
I.
Lindau
,
At. Data Nucl. Data Tables
32
,
1
(
1985
).
26.
P. J.
Cumpson
and
M. P.
Seah
,
Surf. Interface Anal.
25
,
430
(
1997
).
27.
O. A.
Chuvenkova
,
E. P.
Domashevskaya
,
S. V.
Ryabtsev
,
Y. A.
Yurakov
,
A. E.
Popov
,
D. A.
Koyuda
,
D. N.
Nesterov
,
D. E.
Spirin
,
R. Y.
Ovsyannikov
, and
S. Y.
Turishchev
,
Phys. Solid State
57
,
153
(
2015
).
28.
É. P.
Domashevskaya
,
S. V.
Ryabtsev
,
S. Y.
Turishchev
,
V. M.
Kashkarov
,
Y. A.
Yurakov
,
O. A.
Chuvenkova
, and
A. V.
Shchukarev
,
J. Struct. Chem.
49
,
80
(
2008
).
29.
K.
Tabata
,
T.
Kawabe
,
Y.
Yamaguchi
, and
Y.
Nagasawa
,
Catal. Surv. Asia
7
,
251
(
2003
).
30.
K.
Shimanoe
,
K.
Ikari
,
Y.
Shimizu
, and
N.
Yamazoe
,
Sens. Actuators, B
118
,
90
(
2006
).
31.
D. E.
Starr
,
D.
Pan
,
J. T.
Newberg
,
M.
Ammann
,
E. G.
Wang
,
A.
Michaelides
, and
H.
Bluhm
,
Phys. Chem. Chem. Phys.
13
,
19988
(
2011
).
32.
Y. K.
Kim
,
B. D.
Kay
,
J. M.
White
, and
Z.
Dohnálek
,
Surf. Sci.
602
,
511
(
2008
).
33.
D.
Brinkley
and
T.
Engel
,
J. Phys. Chem. B
102
,
7596
(
1998
).
34.
NIST Mass Spectrometry Data Center, Collection (C) 2014 copyright by the U.S. Secretary of Commerce, https://webbook.nist.gov/cgi/cbook.cgi?ID=C67630&Mask=200#Notes; accessed November 13, 2019.
35.
V. A.
Gercher
,
D. F.
Cox
, and
J.-M.
Themlin
,
Surf. Sci.
306
,
279
(
1994
).
36.
V. A.
Gercher
and
D. F.
Cox
,
Surf. Sci.
312
,
106
(
1994
).
37.
38.
K.
Ramesh
,
L.
Chen
,
F.
Chen
,
Y.
Liu
,
Z.
Wang
, and
Y. F.
Han
,
Catal. Today
131
,
477
(
2008
).
39.
E.
Farfan-Arribas
and
R. J.
Madix
,
J. Phys. Chem. B
106
,
10680
(
2002
).
40.
B. H.
Bui
,
R. S.
Zhu
, and
M. C.
Lin
,
J. Chem. Phys.
117
,
11188
(
2002
).
41.
P. G.
Choi
,
N.
Izu
,
N.
Shirahata
, and
Y.
Masuda
,
Sens. Actuators, B
296
,
126655
(
2019
).

Supplementary Material