In Situ Electronic Probing of Photoconductive Trap States for the

Clicking on the donut icon will load a page at altmetric.com with additional details about the ... The Supporting Information is available free of cha...
1 downloads 0 Views 3MB Size
Subscriber access provided by Iowa State University | Library

Surfaces, Interfaces, and Catalysis; Physical Properties of Nanomaterials and Materials

In-Situ Electronic Probing of Photoconductive Trap States for the Catalytic Reduction of CO by InO OH Nanorods 2

2

3-x

y

Joel Y. Y. Loh, and Nazir P. Kherani J. Phys. Chem. Lett., Just Accepted Manuscript • DOI: 10.1021/acs.jpclett.8b03433 • Publication Date (Web): 17 Jan 2019 Downloaded from http://pubs.acs.org on January 18, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

In-situ Electronic Probing of Photoconductive Trap States for the Catalytic Reduction of CO2 by In2O3-xOHy Nanorods Joel Y.Y Loh a, Nazir P. Kherania,b a Electrical b Material

and Computing Engineering, University of Toronto, Toronto, Ontario M5S 3G4, Canada.

Science and Engineering, University of Toronto, Toronto, Ontario M5S 3E4, Canada.

Corresponding author: [email protected]

Abstract: We use photoconductivity-time, optical absorption and electron quantum efficiency measurements under in-situ reactant CO2+H2 atmospheres to determine the role of surface trap states during photo reduction of CO2 to CO using In2O3-xOHy nanorods of varied annealing times. Photocurrent decay trends show an asymmetric energy distribution of surface barrier potentials with increased asymmetry from vacuum to CO2+H2. Urbach analysis shows crystalline disorder parameter of 0.35 - 0.40 under vacuum, 0.45 - 0.6 under CO2+H2. Quantum efficiency spectra show under H2+CO2, average tail state energies are similar with vacuum but with increased densities of photo-conductive gap states. Photo-electro-paramagneticresonance measurements show the creation of new paramagnetic centres. Overall, enhanced activity is associated with lower maximum barrier potential of 0.39eV than that of 0.41eV, lower average trap energies of 2.58eV than 2.69eV, with higher disorder due to increased surface state densities. These techniques pave the way for facile in-situ probing of gas-phase photocatalysts providing simple macrolevel understanding of adsorbed reactants on surface band bending thus correlating to catalytic efficacy.

_________________________________________________________________________

The quest for sustainable fuels and for methods to mitigate the amount of CO2 release has led to the investigation of various photo-thermal catalysts that can perform reverse water gas shift reactions (H2+CO2 -> CO+H2O) at mild temperatures and ideally accelerated under photoillumination conditions. A

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

suitable material candidate is In2O3-xOHy nanorod1 systems which were shown to produce at least 1.2 µmol/g/hr of CO under mild photo-thermal conditions of approximately 100°C - 200°C and a xenon/mercury lamp illumination of 50W/cm2. The most catalytic efficient composition was found by annealing In(OH)3 nanorods at 250°C for 12 hours in contrast to annealing for 8 hours (Figure 1a) , which lies at the morphotropic phase boundary2 between In(OH)3 and In2O3 . The meta-stable composition was found to be oxygen vacancy defect rich with terminal OH groups, which indicate that such defects are key to the mechanism of the reverse water gas shift reaction3 . This catalyst has the potential for solar-thermal scaling approaches, with an estimated reduction in the CO production activation energy from 107 kJ/mol to 86 kJ/mol 4, and an estimated quantum efficiency yield of 0.19%. Notwithstanding photo-geometric limitations it is possible to achieve quantum efficiencies of 10% under white light conditions via suitable photo-reactors such as fiber optic reactors5,6 and annular reactors7,8. While maintaining the same average nanorod lengths of 1700 ± 200nm and (OH + Ovac ) peak area ratios derived from XPS O1s binding energy plots where Ovac sub-peaks and OH sub-peaks are centered at 531.28 - 531.4eV and 532.23 - 532.25eV, respectively, the nanorods aged for 8 hours (labeled S1) have a CO production rate of ~0.8 µmol/g/hr whereas the nanorods aged for 12 hours (labeled S2) have a CO production rate of nearly 1.2 µmol/g/hr. Furthermore, the same nanorod material systems were shown to have 50-60% selectivity for 60 and 90 µmol/g/hr of methanol production9 with a similar aging period of 9 to 14 hours under photoillumination at 250°C. This indicates that the macroscopic defect system within the nanorods may play a key role in enhancing the kinetics of the H2+CO2 photothermal reduction. However, the defect trap energy spectrum was not investigated in this material system and in particular under reactor conditions. If gaseous reactants chemisorb or ionize on the In2O3-xOHy surface the induced surface states will extract or introduce charge carriers within the bulk, thus creating additional photoexcited free electron or hole carriers. Hence, the photo-activation of the gas absorption can thus be wavelength and temperature dependent, observed in the varied energetics of excitation charge that arises from the absorption or ionization of the gas atom on the charged In2O3-xOHy surface. For example, H2 dissociation on metal oxide surfaces is known, with homolytic dissociation10,11 of H2 into two H+ ions being more common than heterolytic dissociation of H2 into H+ and H- ions on highly basic metal oxides that have a high surface charge polarization. Various types of carbonate formations occur on metal oxide surfaces, with monodentate configurations on surfaces with high basicity and bidentate or tridentate configurations on surfaces with low basicity. Charge distribution would be enhanced with a more coordinated configuration than a monodentate configuration. Hence, electronic based measurements are a sensitive way of probing the surface. In this study, we investigate the changes in photo-active defect

ACS Paragon Plus Environment

Page 2 of 22

Page 3 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

energy states by conducting photoconductivity rise/decay time studies and measurements of changes in optical absorption at cryogenic temperatures under in-situ vacuum and at room temperature under H2+CO2 atmospheres. To more fully understand the underlying changes during the transition from an inactive atmosphere to an active atmosphere, we carry out electron quantum efficiency measurements under vacuum, H2, CO2, and H2+CO2 atmospheres.

Figure 1. (a) The material parameters (length of the nanorods and XPS peak ratio) and CO production rate from the reaction of H2+CO2 on catalytic In2O3-xOHy nanorods as a function of the annealing time of In(OH)3 nanorods (for 3, 8 and 12 hours) at 250°C. Data reproduced from reference (1); additional details are provided in supplementary information. (b) SEM images of In2O3-xOHy nanorods with average lengths of 1700±200nm.

An in-situ photoconductivity measurement under high vacuum and 1:1 H2+CO2 ambient atmospheric pressure is shown herein to elucidate the effects of gaseous reactants on the surface barrier potential on In2O3-xOHy nanorods aged at 8 and 12 hours. Figure 2 (a-b) shows the photocurrent rise and decay during and after photoillumination of 1 hour. For the material system under vacuum the rise and decay profiles are relatively similar, requiring ~45 minutes to reach photo-saturation and 60 minutes to return to the dark current value. S2 has a higher photocurrent than S1 (see SI Figure 2), which can be related to a higher charge carrier mobility with an increased aging period. Under CO2 + H2, the profiles are extremely different, with sub-linear photocurrent rise and decays. The photocurrent decays are extremely slow, indicating the presence of either trap states that do not offer recombination of charge carriers by being charge selective, or trap states that have large capture cross sections and large hopping activation energies thus delaying their recombination. In order to understand the distribution of the surface barrier potentials, we calculate the photocurrent rise and decay profiles and determine the parameters that best fit the experimental profiles. The

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 22

photocurrent rise can be expressed simply as N e  t   N sn  N sn exp   sn N pht  where N e , N sn , N ph represent respectively the number of excited states, surface states with a similar energy as the band gap, and N ph the incident intensity of photons with a similar energy as the band gap. We find a similar photocurrent rise time constant of 18-20 minutes for both S1 and S2 in vacuum, despite having a different increase between the maximum photocurrent and dark current. The photocurrent decay G ph  t  , after turning off the photoillumination, can be expressed not as a simple exponential decay profile but rather





as G ph  t   G ph  Gd exp    t sat



 sat  str   where G ph is the saturated photocurrent, Gd is the dark 

current, τstr represents the effective decay time constant, while 0 < β < 1 represents the stretched exponential power factor that determines the gradient of the decay slope. The closer β is to 1, the more similar the decay profile to that of an exponential profile, and if β is closer to 0, the decay profile is that of a sub-linear slope. We find a β of 0.7 and 0.8 for S1 and S2 in vacuum, respectively, and a β of 0.58 and 0.5 for S1 and S2 in CO2+H2, respectively. The decay time constant  str is 6.9 × 103 seconds and 2.3 × 103 seconds for S1 and S2 in vacuum, respectively, and 90 × 103 and 32 × 103 seconds in CO2 + H2 for S1 and S2, respectively. The probability density function distribution of the surface grain potential barriers (normalized by area of the peak) of S1 and S2 under vacuum and CO2+H2 can be used to determine the amount of disorder within the crystalline grains by using the probability distribution12 given below (which is a function of the potential energy barrier Us and β):

p U s ,   

1







 

1



 cos  /2  cos  str exp U s qT   rs s nd   u  u  sin  / 2  du  exp u 0

where фT is the temperature constant kT, q is the Coulombic charge, nd the carrier density, rs is the surface recombination probability, γ is the carrier capture cross section, and νs the surface carrier velocity. As Figure 2g shows, in vacuum, with increased aging and an increased β of 0.85, the distribution of the potential barrier energy of the nanorod grains of the S2 In2O3-xOHy system is fairly similar to a Gaussian peak and is centred at a lower energy of ~0.36eV, whereas S1 shows a larger distribution of low energy barrier potentials from below 0.39 eV which represent defect states below the conduction band that are partially filled and hence reduces the electron screening effect. Under CO2 + H2 (Figure 2h) both samples show a significantly lower energy surface barrier distribution, with S1 having a slightly higher potential of 0.41eV compared to 0.39eV and less asymmetry than that of S2. The reversal of the asymmetric Us

ACS Paragon Plus Environment

Page 5 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

distribution between vacuum and CO2+H2 indicates that the increased adsorption of both reactant gases creates a distribution of tail states in the sub-band gap that are lower in energy and induces a distribution of lower potential barriers. The change in the peak centres is ~0.05eV for S2 and ~0.02eV for S1, suggesting that the initial distribution of low energy potential barriers in S1 induces pinning of the adsorbed gas surface states.

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 22

Figure 2. (a-b) show the photocurrent rise and decay profiles of S1 and S2 in vacuum and in H2+CO2 atmosphere, respectively. (c-d) show the photocurrent rise (solid) along with the calculated photocurrent (dashed) increase as an exponential function with time constants of 20 / 30 minutes and 18 / 33 minutes for S1 and S2 in vacuum and H2+CO2, respectively. (e-f) show the photocurrent decay along with the calculated decay profile of a stretched exponential function with a decay time constant of 115 / 1500 minutes and 40 / 538 minutes for S1 and S2 in vacuum and H2+CO2, respectively. Macroscopic distributions for the grain surface potential barrier for S1 and S2 in vacuum (g), and CO2+H2 (h). The potential barriers are asymmetrically distributed, giving rise to a sub-linear decay profile for the photocurrent. Associated error bars for β and τD uncertainties of ±0.2 and ±10 mins are also shown.

To further confirm the aforementioned results which show an increased presence of surface defects under CO2 + H2 atmosphere, the optical absorption – temperature plots of S1 and S2 in vacuum and CO2 + H2 from 180K to 300K were obtained in order to determine the effects on the Urbach tail states. According to the Urbach theory13,14,15, the optical absorption edge can be described by an exciton peak described by the expression:    0 exp   h  E0  / kT  where E0 is the Urbach focus of the extrapolated lines from the experimental absorption edge, and  is known as the steepness parameter which can be









described by a phonon temperature dependence:    0 2kT h phn tanh h phn 2kT . h phn is the characteristic phonon energy (0.072eV)16 of In2O3 and kT  is the Urbach energy Eu which can be separated into thermal disorder and static disorder components, representing crystalline phonon modes and

disorder

phonon



modes.

Eu T , P   K    0  1  P  2  N exp   T   1

Eu 1

can

 where



be

described

by:

is the Einstein temperature

corresponding to the mean lattice phonon frequency (the Debye temperature of In2O3 is 700K and  is ¾ of the Debye temperature), P represents the degree of disorder, where P = 0 represents full crystallinity and no defects, P = 1 represents a fully amorphous structure, N represents the degree of crystallinity with N=1 indicating a Cody type high limit of crystallinity. We find that the Urbach focus increases from 4.13eV and 4.24eV for S1 and S2, respectively, to 4.14eV and 4.26eV, respectively, under CO2 + H2 atmosphere. The associated Urbach energy in vacuum is approximately 0.04 - 0.05eV, which is fairly similar to the strongest Raman scattering peak of 0.037eV (310cm-1) for crystalline In2O3. By adjusting the parameters P and N to fit the Eu – temperature plot (Figure 3a-b), we determine that the disorder parameter P of S1 and S2 under vacuum are similar at 0.35 and 0.40, respectively, and that both of them have the same N value of 0.9. However, the P parameter increases to 0.45 and 0.60, respectively, under CO2 + H2, which indicates an increase in disorder representing additional defect states that increase optical absorption.

ACS Paragon Plus Environment

Page 7 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 3. (a-b) Urbach analysis of S1 and S2 optical absorption parameters with increasing temperature. The optical adsorption edge can be described by a modified exponential function, the steepness parameter σ with temperature. The Urbach energy Eu values (b-i and b-ii) determined by kT/σ can be fitted with disorder parameters P of 0.35 and 0.40 for S1 and S2, respectively, in vacuum, and with P of 0.45 and 0.60 for S1 and S2, respectively, in CO2+H2.

In order to further our understanding of the effects of the various reactant gases on the photoactive trap states, we use the in-situ gas phase EQE spectrum (Figure 4) as a way of determining the tail state distribution at the various monochromatic wavelength points. After vacuum pumping down the chamber which contained the probe holder, we introduced the reactant gases at ambient pressure where for H2+CO2 the gas pressure ratio of 1:1 was achieved by adjusting the pressure regulator. The sample in the probe holder was then illuminated by monochromatic light and a photocurrent measurement was taken after 20 minutes at each wavelength interval. Both S1 and S2 show a smaller photo-excitation threshold of 2.6eV in vacuum atmosphere, which is in line with the literature17. Because the EQE spectra collection starts at the lower energy wavelength of 1.4eV the photoexcited free carriers should generally be initially excited from the lower energy trap states. By deconvolving the EQE spectra into Gaussian peaks with linewidths of 0.4-0.5eV such that the envelope of peaks fit the EQE spectra, and averaging the energy of the trap states (defined as the states leading from the absorption edge Gaussian peak at ~3.1eV) with the normalized density of trap states as a ratio of all conduction band (CB) tail states (by summing up the areas of the Gaussian sub-peaks - see supplementary

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

information) shows that the S1 and S2 average CB tail state energies are 2.64eV and 2.51eV, respectively. The presence of lower energy states that enable a photocurrent increase indicates that there are more traps that enable electron hopping than traps that enable recombination, thus increasing charge carrier lifetime. In vacuum, S2 has a smaller tail correlating with decreased crystalline disorder. Even though the photoactive traps of S1 have a higher average energy in vacuum than that of S2, after H2 introduction the activation energies increased by 0.11-0.26eV, indicating that higher energy states are formed by In-H+ and H+-OH formation after heterolytic or homolytic dissociation. Although there are additional low level states centred at 2.26-2.33eV, the higher trap energies in H2 are due to the presence of a conduction band tail state at ~2.9eV, which was determined by the inability to fit a single Gaussian peak within the conduction band edge. Hence, we assign the high energy state as the H+-OH state, and the low energy state as indium-hydride sites. The higher energetic increase in S2 from vacuum to H2 can be correlated to its higher CO2 reducing activity, which indicates that increased aging in S2 leads to more activated surfaces under H2. The amplitude of the tail states for S1 in H2 is smaller than that for S1 under vacuum, which indicates that the density of electron trap states are significantly decreased, showing that dissociated H+ adatoms react with hydroxyl groups to form desorbed H2O or reformed H2 with oxygen vacancy healing. In the CO2 only atmosphere, the tail state densities are also significantly greater, although the average trap energies are similar to that of the vacuum condition. This indicates that both oxygen vacancy filling to form carbonyl groups are present in addition to carbonate formation. In S2 the density of trap states are higher than that of S1, which in turn indicates that S2 has greater carbonate formation than S1. Noticeably, the conduction band tail state energies of S1 and S2 in H2+CO2 are 2.69eV and 2.58eV, respectively, due to the introduction of a low energy state at 2.28eV for S2 and the removal of a ~2.4eV state for S1, leading to somewhat similar average energies with a slight increase of 0.05-0.07eV for the vacuum atmosphere. Since H2 only atmospheres lead to an increase in trap energies, this indicates that the combination of H2+CO2 atmosphere lowers the surface energies from the activated H2 only state, whereby adsorbed H2 reacts with CO2 to form H2O and CO. Furthermore, the density of trap states available for photoconduction is significantly higher. Although S1 has a higher photocurrent density at the incident 2.7eV photon wavelength, S2 has a more evenly distributed density of states. Thus, S2 is likely to have higher carrier mobility under incident white light spectrum than S1. The higher charge mobility is thus linked to the CO2 reduction rate, since electronhole pairs can be separated at a coupled indium-hydride - oxygen vacancy supersite where electrical polarization is high, hence a higher carrier mobility will enable more efficient charge separation for CO2 reduction.

ACS Paragon Plus Environment

Page 8 of 22

Page 9 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 4. EQE spectra of S1 and S2 in vacuum (a-b), H2 (c-d), CO2 (e-f), H2+CO2 (g-h) atmospheres. The EQE tails were deconvolved with Gaussian peaks with linewidths of 0.4-0.5eV and the subpeak energy values were used to determine the average defect energies within the band gap.

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 5. EPR spectra of S1 (a) and S2 (b) in air before and during photoillumination by a 405nm wavelength laser after 3 minutes. The blue laser illumination was aligned with the sample cavity during the EPR measurement. A secondary peak (inset figure) develops especially in the S2 spectra with a corresponding decrease in peak signal intensities. (c) Simulated spectra with the assumption that the isotropic g-tensor of a spin centre changes to an axial tensor under light (grey dashed line), and spin concentration ratio of 0.21 : 1.12 for two distinct centres of isotropic g-tensor each (red dashed line) - for best fit of experimental spectrum. Isotropic g = [ 2.171 ] for a one-spin system fits the simulated spectra for S2 in dark (inset figure).

To further understand photo-effects on the photocatalyst system, photo-EPR measurements were carried out at below the optical absorption edge with a blue 405nm 1mW laser aligned with the microwave cavity and quartz capillary tube containing the powder samples. Paramagnetic centres in In2O3 are typically associated with singly ionized oxygen vacancies18,19,20 but also In2+ defects21. In Figure 5b, it is immediately obvious that a well-defined secondary peak develops in the shoulder of the upper bound peak of S2, as well as a significant peak intensity decrease, and less so in the lower bound peak. After turning off the laser and a wait of 10 minutes, the subsequent scan in dark showed a recovery of the primary signal peaks’ intensities and no presence of the secondary peak. A green 535nm laser photo-EPR was then carried out but no secondary peaks were seen during green photo-illumination, albeit a small drop in signal peak intensity was seen. A hint of the secondary peak and peak intensity decrease is seen in S1 during blue photoillumination, which suggests that photo-active sub band gap defects are relaxing photo-excited unpaired electrons back into the valence band, and thus any photoinduced paramagnetic centres are unstable at room temperature. In S2, the photo-induced

ACS Paragon Plus Environment

Page 10 of 22

Page 11 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

paramagnetic centres at 295-305 mT are likely valence band holes after various photo-transitions22. The elementary photo-ionization transitions are shown in Table 1: Gain/loss of

Gain/loss of

secondary centres

primary centres

[O]+ + hv -> [O] + h+VB ~ (I)

gain

loss

[O]2+ + hv -> [O]+ + h+VB ~ (II)

gain

gain

[O]0 + hv -> [O]+ + e-CB ~ (III)

---

gain

In3+ + hv -> In2+ + h+VB ~ (IV)

gain

gain

In2+ + hv -> In3+ + e-CB ~ (V)

---

loss

gain

---

---

---

loss

---

O- + hv -> O2- + h+VB

~ (VI)

(OH)- + hv -> OH + e-CB ~ (VII) e-CB + h+VB

~ (VIII)

Table 1. Possible photo-transitions for In2O3-xOHy system. Secondary centres are associated with valence holes while primary centres are mostly associated with [O]+ and In2+ defects.

where [O]+ represent single ionized oxygen vacancies, [O]2+ double ionized oxygen vacancies, [O]0 neutral oxygen vacancies, hv the incident photon with 3.062eV energy, h+VB the valence band hole, e-CB the conduction band electron, O the lattice oxygen and OH the hydroxyl group. It is clear that transition (I) would generate the change in EPR spectrum from dark to blue photo-illumination by itself, while any combination of transitions likely involving transition (V) and other transitions (II-IV) and (VI) can generate the same qualitative changes. Additionally, transition (V) by itself would be sufficient for the peak decrease but no additional secondary centres under green photo-illumination. We turn to some simple simulations of the S2 EPR spectra with Easyspin. Agnostically it is possible to recreate the doublet of the upper bound peak by setting the isotropic g-tensor of a single spin system from g= [2.171] to an axial tensor of g = [2.16, 2.264] representing the g factors of the secondary and primary peak inflections respectively, with an associated linewidth of 7 mT FWHM that roughly approximates the EPR spectra of S2 (Figure 5c). However, the proportionality of the secondary and primary peaks do not match unless the input parameters are of a two-spin system with spin concentration ratio of 0.21 : 1.12 with associated linewidths of 7.8 mT and 11.2 mT FWHM and two

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

isotropic g tensors of g1 = [2.173] and g2= [2.241] respectively and under the assumption that the two spins of g1 and g2 are both ½ spin each with minimal coupling interaction. The superior fitting of a twospin system with the experimental spectrum helps validate the assumption that photo-illumination generates a distinct paramagnetic centre with its own isotropic g-tensor. Since electron-electron coupling approximation of an exciton would typically generate doublet peaks on the upper and low bound of the spectrum even with anisotropic interaction, it is likely that the photo-generated electronhole pair have large interaction distances. The concentration ratio also implies that 0.16 of the total spin system concentration are photo-induced paramagnetic centres, which suggests a high rate of electronhole recombination. With these results taken together, we can sketch out the energy band diagram associated with the grain surface under vacuum and under a similar reactor atmosphere condition (Figure 6). We use S2 as the primary example. From the photocurrent decay trends, the majority of the grains have a surface barrier potential Us of ~0.36eV and some grains with Us of 0.31-0.35eV. The reason for the range of surface potential barrier values is due to the presence of surface defects, which can be assigned to OH or defective lattice O- or neutral, singly ionized and doubly ionized oxygen vacancy defects. Charge carrier hopping within the continuum of defects enables sub-band gap photon energies to excite charge carriers into the conduction band. We also assign the presence of indium defects Ind which are close to the valence band edge as a source of charge carrier contribution. Under CO2 + H2, the surface potential distribution widens to a range of 0.30 - 0.39eV. As shown in PC decay, Urbach and EQE trends, there are increased imperfections in the crystallinity of the In2O3-xOHy system such that we can attribute the increased density of defects to the surface reactions with gaseous reactants. Dissociated H atoms create H+OH, OH+ and In-H- defects23, while carbonate formation on the lattice oxygen sites and terminal OH groups as well as filling in of the oxygen vacancies, which is commonly observed in various metal oxides to create carbonyl groups24, 25. The increase in surface band bending in this situation can be analogized to interaction of a redox couple (which is represented by acidic Indium sites and basic OH and O sites) on a semiconductor support with reactant gases to create a new equilibrium. The filling in of surface states injects electrons into the bulk causing increased surface charge screening to compensate thus the increasing band bending. The dynamic charge transfer due to the adsorption and desorption of various gas products induces a distribution in the range of surface barrier potentials.

ACS Paragon Plus Environment

Page 12 of 22

Page 13 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 6. A sketch of the grain surface band diagram using the parameters determined for S2 in vacuum (a) and CO2+H2 (b) atmosphere. The surface potential barriers have a distribution of energies over the macroscopic nanorod grain system, approximating 0.3 - 0.39 eV, due to the presence of a range of defects. There are at least 5 defects comprising the sub-band gap defect system as determined by EQE. They were assigned to typical metal oxide defects of undercoordinated indium defects, ionized and neutral oxygen vacancies, under coordinated oxygen defects, and in the case of the In2O3-xOHy system an additional set of OH defects is introduced. The orange lines represent the photoexcitation rate (N(t)) of negative charge carriers from the valence band and neutral indium defects into the conduction band. After turning off illumination, negative charge carriers then relax into the valence band through the set of defects as sketched by the red line that results in a net photocurrent decay (ΔGph(t)). Under CO2+H2, an increased set of defects due to heterolytic dissociation of H2 and carbonate formation via oxygen vacancy filling and coordination with the oxide surface. The diagrams are not drawn to scale.

In this study we show that the catalytic performance of a metal oxide-metal hydroxide system, such as In2O3-xOHy nanorods, can be studied with in-situ photo-electrical measurements. The increase in the activity of the photocatalytic reverse water gas shift reaction can be attributed to the sample with the lower potential barrier height and greater increase in disorder / greater decrease in crystallinity yet having a lower average defect energy. The interplay between gaseous reactions and the changes to the band energy system of nanomaterials is an under-investigated topic and can yield fascinating insights into band gap engineering of highly active photo-catalysts. Supporting Information Available: The supporting information provides contextual information on Figure 1a, experimental details of the modified cryostat current-voltage measurements and further details on the Urbach analysis, EQE results and EPR measurements.

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Acknowledgements The authors gratefully acknowledge the support of the Natural Sciences and Engineering Research Council of Canada. The authors also acknowledge the kind assistance of Dr. L. He with the preparation of the nanorods as well as the discussions with Professor G. Ozin. The authors are also grateful to Dr Darcy Brown and Dr Karl Demmans for assistance with photo-EPR. References (1)

He, L.; Wood, T. E.; Wu, B.; Dong, Y. C.; Hoch, L. B.; Reyes, L. M.; Wang, D.; Kübel, C.; Qian, C.; Jia, J.; Liao, K.; et. al. Spatial Separation of Charge Carriers in In2O3-x(OH)y Nanocrystal Superstructures for Enhanced Gas-Phase Photocatalytic Activity. ACS Nano. 2016, 10, 5578–5586.

(2)

Berengue, O. M.; Rodrigues, A. D.; Dalmaschio, C. J.; Lanfredi, A. J. C. ; Leite, E. R. ; Chiquito, A. J. Structural characterization of indium oxide nanostructures: a Raman analysis, J. Physics D: Applied Physics. 2010, 43, 045401.

(3)

Ghuman, K.K; Hoch, L. B.; Szymanski, P.; Loh, J. Y. Y.; Kherani, N. P.; El-Sayed, M. A.; Ozin, G. A.; Singh, C. V. Photoexcited Surface Frustrated Lewis Pairs for Heterogeneous Photocatalytic CO2 Reduction, J. Am. Chem. Soc. 2016, 138, 1206-1214.

(4)

Ghuman, K. K.; Wood, T. E. ; Hoch, L. B. ; Mims, C. A. ; Ozin, G. A. ; Singh, C. V. Illuminating CO2 reduction on frustrated Lewis pair surfaces: Investigating the role of surface hydroxides and oxygen vacancies on nanocrystalline In2O3-x(OH)y . Phys. Chem. Chem. Phys., 2015., 17, 14623-14635.

(5)

Peil, N. J.; Hoffmann, M. R. Development and Optimization of a TiO2-coated Fiber-Optic Cable Reactor: Photocatalytic Degradation of 4-Chlorophenol. Environ Sci. Technol., 1995, 29, 2974–2981.

(6)

Sazio, P. J. A.; Correa, A. A.; Finlayson, C. E.; Hayes, J. R. ; Scheidemantel, T. J. ; Baril, N. F. ; Jackson, B. R.; Won, D. J. ; Zhang, F. ; Margine, E. R.; et al. Microstructured Optical Fibers as High-Pressure Microfluidic Reactors. Science, 2006, 311, 1583-1586.

(7)

Lasa, H. d.; Serrano, B.; Salaices, M. Photocatalytic Reaction Engineering. Springer, 2005, ISBN 978-0-38727591-8

(8)

Aymen, A. R.; Assad, A.; Nasrallah, N.; Bouzaza, A. ; Maachi, R. ; Wolbert, D. Photocatalytic Treatment of Petroleum Industry Wastewater using Recirculating Annular Reactor: Comparison of Experimental and Modeling. Environ Sci Pollut Res, 2018, 1, 1-12.

(9)

Lu, W.; Ghoussoub, M.; Wang, H.; Shao, Y.; Sun, W.; Tountas, A. A.; Wood, T. E.; Li, H.; Loh, J. Y. Y.; Dong, Y.C.; et al. Photocatalytic Hydrogenation of Carbon Dioxide with High Selectivity to Methanol at Atmospheric Pressure. Joule. 2018, 2, 1369-1381.

(10)

Chen, H. Y. T.; Giordano, L.; Pacchioni, G. From Heterolytic to Homolytic H2 Dissociation on Nanostructured MgO (001) Films as a Function of the Metal Support, J. Phys. Chem. C. 2013, 117, 20.

ACS Paragon Plus Environment

Page 14 of 22

Page 15 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

(11)

Yang, Y.; Zhang, P. Dissociation of H2 Molecule on the β-Ga2O3 (100) Surface: The Critical Role of Oxygen Vacancy, Solid State Phys. 2010., 374, 10623–10629.

(12)

Brinzari, V. Mechanism of Band Gap Persistent Photoconductivity (PPC) in SnO2 Nanocrystalline Films: Nature of Local States, Simulation of PPC and Comparison with Experiment, Applied Surface Science. 2017, 411, 437-448

(13)

Guerra, J. A.; Zela, F. D. On the Origin of the Urbach Rule and the Urbach Focus, MRS Proceedings, 2013, 1536, 139-145.

(14)

Rai, R. C. Analysis of the Urbach tails in Absorption Spectra of Undoped ZnO Thin Films. J. Appl. Physics. 2013, 113, 153508

(15)

Singh, S.; Li, C.; Panzer, F.; Narasimhan, K. L. ; Graeser, A. ; Gujar, T. P. ; Köhler, A. ; Thelakkat, S. Huettner, M. ; Kabra, D. Effect of Thermal and Structural Disorder on the Electronic Structure of Hybrid Perovskite Semiconductor CH3NH3PbI3, J. Phys. Chem. Lett. 2016, 7, 3014–3021.

(16)

Weiher, R. L.; Ley, R. P. Optical Properties of Indium Oxide, J. Appl. Phys. 1966, 37, 299.

(17)

Albrecht, M.; Schewski, R.; Irmscher, K. ; Galazka, Z. ; Markurt, T. ; Naumann, M. ; Schulz, T. ; Uecker, R. ; Fornari, R. ; Meuret, S.; Kociak, M. Coloration and Oxygen Vacancies in Wide Band Gap Oxide Semiconductors: Absorption at Metallic Nanoparticles induced by Vacancy Clustering - A Case Study on Indium Oxide, J. Appl. Phys. 2014, 115, 053504.

(18)

Arooj, S.; Xu, T.T.; Hou, X.; Wang, Y.; Tong, J.; Chu R.; Liu, B. Green emission of Indium Oxide via Hydrogen Treatment, RSC Adv. 2018, 8, 11828–11833.

(19)

Kumar, M; Singh, V. N.; Singh, F.; Lakshmi, K. V.; Mehta, B. R.; Singh, J. P. On the Origin of Photoluminescence in Indium Oxide Octahedron Structures, Appl. Phys. Lett. 2008, 92, 171907

(20)

Guha, P.; Kar, S; Chaudhuri, S. Direct Synthesis of Single Crystalline In2O3 Nanopyramids and Nanocolumns and their Photoluminescence Properties, Appl. Phys. Lett. 2004, 85, 171907

(21)

Borisenko, V.E; Gaponenko, S.V; Gurin, V.S; Physics, Chemistry and Applications of Nanostructures: Proceedings of the International Conference Nano Meeting, World Scientific Publishing Co. Pte. Ltd, 2011, ISBN-13 978-981-4343-89-3

(22)

Schwartz, R. N.; Wechsler, B. A.; McFarlane, R. A. Photo-EPR study of Light-Sensitive Impurity and Defect Centers in Photorefractive BaTiO3, Phys. Rev. B. 1992, 46, 3263-3271

(23)

Ye, J.; Liu, C.; Ge, Q. DFT study of CO2 Adsorption and Hydrogenation on the In2O3 surface, J. Phys. Chem. C. 2012, 116, 7817–7825.

(24)

Wang, B.; Wang, X.; Lu, L.; Zhou, C.; Xin, Z.; Wang, J. J.; Ke, X. K.; Sheng, G. D.; Yan, S.C.; Zou, Z. G. OxygenVacancy-Activated CO2 splitting over Amorphous Oxide Semiconductor Photocatalyst. ACS Catal. 2018, 8, 516–525.

(25)

Boon, A. Q. M.; Huisman, H. M.; Geus, J. W. Influence of Surface Oxygen Vacancies on the Catalytic Activity of Copper Oxide: Part 2. Oxidation of Methane, J. Mol. Catal. 1992, 75, 293-303.

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

This is the image for the abstract. 85x49mm (150 x 150 DPI)

ACS Paragon Plus Environment

Page 16 of 22

Page 17 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 1. (a) The material parameters (length of the nanorods and XPS peak ratio) and CO production rate from the reaction of H2+CO2 on catalytic In2O3-xOHy nanorods as a function of the annealing time of In(OH)3 nanorods (for 3, 8 and 12 hours) at 250°C. Data reproduced from reference (1); additional details are provided in supplementary information. (b) SEM images of In2O3-xOHy nanorods with average lengths of 1700±200nm. 177x71mm (150 x 150 DPI)

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. (a-b) show the photocurrent rise and decay profiles of S1 and S2 in vacuum and in H2+CO2 atmosphere, respectively. (c-d) show the photocurrent rise (solid) along with the calculated photocurrent (dashed) increase as an exponential function with time constants of 20 / 30 minutes and 18 / 33 minutes for S1 and S2 in vacuum and H2+CO2, respectively. (e-f) show the photocurrent decay along with the calculated decay profile of a stretched exponential function with a decay time constant of 115 / 1500 minutes and 40 / 538 minutes for S1 and S2 in vacuum and H2+CO2, respectively. Macroscopic distributions for the grain surface potential barrier for S1 and S2 in vacuum (g), and CO2+H2 (h). The potential barriers are asymmetrically distributed, giving rise to a sub-linear decay profile for the photocurrent. Associated error bars for β and τD uncertainties of ±0.2 and ±10 mins are also shown. 309x355mm (150 x 150 DPI)

ACS Paragon Plus Environment

Page 18 of 22

Page 19 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 3. (a-b) Urbach analysis of S1 and S2 optical absorption parameters with increasing temperature. The optical adsorption edge can be described by a modified exponential function, the steepness parameter σ with temperature. The Urbach energy Eu values (b-i and b-ii) determined by kT/σ can be fitted with disorder parameters P of 0.35 and 0.40 for S1 and S2, respectively, in vacuum, and with P of 0.45 and 0.60 for S1 and S2, respectively, in CO2+H2. 300x182mm (150 x 150 DPI)

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. EQE spectra of S1 and S2 in vacuum (a-b), H2 (c-d), CO2 (e-f), H2+CO2 (g-h) atmospheres. The EQE tails were deconvolved with Gaussian peaks with linewidths of 0.4-0.5eV and the subpeak energy values were used to determine the average defect energies within the band gap. 163x191mm (150 x 150 DPI)

ACS Paragon Plus Environment

Page 20 of 22

Page 21 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Figure 5. EPR spectra of S1 (a) and S2 (b) in air before and during photoillumination by a 405nm wavelength laser after 3 minutes. The blue laser illumination was aligned with the sample cavity during the EPR measurement. A secondary peak (inset figure) develops especially in the S2 spectra with a corresponding decrease in peak signal intensities. (c) Simulated spectra with the assumption that the isotropic g-tensor of a spin centre changes to an axial tensor under light (grey dashed line), and spin concentration ratio of 0.21 : 1.12 for two distinct centres of isotropic g-tensor each (red dashed line) - for best fit of experimental spectrum. Isotropic g = [ 2.171 ] for a one-spin system fits the simulated spectra for S2 in dark (inset figure). 277x175mm (150 x 150 DPI)

ACS Paragon Plus Environment

The Journal of Physical Chemistry Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6. A sketch of the grain surface band diagram using the parameters determined for S2 in vacuum (a) and CO2+H2 (b) atmosphere. The surface potential barriers have a distribution of energies over the macroscopic nanorod grain system, approximating 0.3 - 0.39 eV, due to the presence of a range of defects. There are at least 5 defects comprising the sub-band gap defect system as determined by EQE. They were assigned to typical metal oxide defects of undercoordinated Indium defects, ionized and neutral oxygen vacancies, under coordinated oxygen defects, and in the case of the In2O3-xOHy system an additional set of OH defects is introduced. The orange lines represent the photoexcitation rate (N(t)) of negative charge carriers from the valence band and neutral Indium defects into the conduction band. After turning off illumination, negative charge carriers then relax into the valence band through the set of defects as sketched by the red line that results in a net photocurrent decay (ΔGph(t)). Under CO2+H2, an increased set of defects due to heterolytic dissociation of H2 and carbonate formation via oxygen vacancy filling and coordination with the oxide surface. The diagrams are not drawn to scale. 237x125mm (150 x 150 DPI)

ACS Paragon Plus Environment

Page 22 of 22