Nanostructured Indium Oxide Coated Silicon Nanowire Arrays: A

Sep 6, 2016 - The field of solar fuels seeks to harness abundant solar energy by driving useful molecular transformations. Of particular interest is t...
2 downloads 19 Views 4MB Size
Download PDF
Nanostructured Indium Oxide Coated Silicon Nanowire Arrays: A Hybrid Photothermal/ Photochemical Approach to Solar Fuels Laura B. Hoch,† Paul G. O’Brien,‡ Abdinoor Jelle,§ Amit Sandhel,† Douglas D. Perovic,§ Charles A. Mims,∥ and Geoffrey A. Ozin*,† †

Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario M5S 3H6, Canada Department of Mechanical Engineering, Lassonde School of Engineering, York University, Toronto, Ontario M3J 1P3, Canada § Department of Materials Science and Engineering, University of Toronto, 184 College Street, Toronto, Ontario M5S 3E4, Canada ∥ Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, Ontario M5S 3E5, Canada ‡

S Supporting Information *

ABSTRACT: The field of solar fuels seeks to harness abundant solar energy by driving useful molecular transformations. Of particular interest is the photodriven conversion of greenhouse gas CO2 into carbon-based fuels and chemical feedstocks, with the ultimate goal of providing a sustainable alternative to traditional fossil fuels. Nonstoichiometric, hydroxylated indium oxide nanoparticles, denoted In2O3−x(OH)y, have been shown to function as active photocatalysts for CO2 reduction to CO via the reverse water gas shift reaction under simulated solar irradiation. However, the relatively wide band gap (2.9 eV) of indium oxide restricts the portion of the solar irradiance that can be utilized to ∼9%, and the elevated reaction temperatures required (150−190 °C) reduce the overall energy efficiency of the process. Herein we report a hybrid catalyst consisting of a vertically aligned silicon nanowire (SiNW) support evenly coated by In2O3−x(OH)y nanoparticles that utilizes the vast majority of the solar irradiance to simultaneously produce both the photogenerated charge carriers and heat required to reduce CO2 to CO at a rate of 22.0 μmol·gcat−1·h−1. Further, improved light harvesting efficiency of the In2O3−x(OH)y/SiNW films due to minimized reflection losses and enhanced light trapping within the SiNW support results in a ∼6-fold increase in photocatalytic conversion rates over identical In2O3−x(OH)y films prepared on roughened glass substrates. The ability of this In2O3−x(OH)y/SiNW hybrid catalyst to perform the dual function of utilizing both light and heat energy provided by the broad-band solar irradiance to drive CO2 reduction reactions represents a general advance that is applicable to a wide range of catalysts in the field of solar fuels. KEYWORDS: solar fuels, photocatalysis, photothermal catalysis, broadband solar irradiance, gas phase, silicon nanowires, indium oxide he emerging field of solar fuels seeks to efficiently store radiant solar energy in the form of chemical bonds, which can then be released on demand and act as a drop-in replacement for traditional fossil fuels.1−7 Researchers in this field are striving to develop an “artificial leaf”, essentially a material, or combination of materials, that can convert light energy in the form of solar photons into chemical energy, using water and/or CO2 as a feedstock, to generate useful chemical species. Enabling this technology would allow the greenhouse gas CO2, emitted from energy production and manufacturing exhaust streams, to be converted into valuable products such as fuels or platform molecules, thereby creating significant economic and environmental benefits.8−11 Generally, solar

T

© 2016 American Chemical Society

energy can be harnessed and utilized via both photocatalytic and photothermal processes. In photocatalytic processes, semiconductors are used to absorb sunlight, generating e−/h+ pairs that drive the conversion of CO2 to various products.12−14 In photothermal processes sunlight energy is used to generate heat, which is responsible for driving the reaction in a method more analogous to traditional thermal catalysis.15−19 However, despite many advances in the field of solar fuels, realizing Received: August 11, 2016 Accepted: September 6, 2016 Published: September 6, 2016 9017

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

www.acsnano.org

Article

ACS Nano

under the same conditions. Furthermore, by testing the photocatalytic rate of the vertically aligned SiNWs evenly coated by In2O3−x(OH)y nanoparticles under filtered light, we demonstrate that photogenerated electrons transferred from the SiNW support to the In2O3−x(OH)y nanoparticles are not the primary factor driving the RWGS reaction and are not responsible for the substantial enhancement in activity. Instead we propose that the higher CO production rates over the evenly coated In2O3−x(OH)y/SiNW hybrid structures are a result of improved light harvesting efficiency, which is supported by both reflectance and emissivity measurements. These results demonstrate that the evenly coated In2O3−x(OH)y/SiNW hybrid structures utilize incident solar radiation more efficiently by simultaneously increasing reaction rates and removing the need for external heating, thus improving the overall energy efficiency of the process.

materials that absorb and convert light energy from the entire solar spectrum is particularly challenging because the sun provides a broadband light source; over 98% of the solar irradiance received at the earth’s surface comprises photons ranging in wavelength from 300 nm in the UV region to 3 μm in the NIR region. In this context, to achieve optimal efficiency, it is important to develop photocatalysts that utilize light from the entire solar spectrum.20,21 Previously, our group has demonstrated that nonstoichiometric, hydroxylated In2O3−x(OH)y can function as an efficient gas-phase photocatalyst to reduce CO2 to CO via the reverse water gas shift (RWGS) reaction using both UV and visible light.22,23 However, indium oxide has a relatively wide band gap (2.9 eV) and absorbs only the blue and UV portions of the solar spectrum.24 Incident light with energy less than the band gap is either transmitted or reflected and does not contribute to the RWGS reaction. Further, optimal temperatures for this reaction are in the range of 150−190 °C, which is necessary to facilitate desorption of H2O to regenerate the catalytic site.22,23 This means that an external heating source is required to elevate the In2O3−x(OH)y catalysts to these temperatures, which decreases the overall energy efficiency of the process. Recently, our group has also demonstrated that vertically aligned silicon nanowires (SiNWs) can be used as a photoactive support for Ru-based catalysts, which are traditionally activated using thermal energy, to produce CH4 from CO2 via the Sabatier reaction.25 The vertically aligned nanowires etched into a Si wafer enhance the light harvesting capabilities of these nanostructured supports, forming so-called “black silicon”, which exhibits reflectance values of less than ∼3% on average throughout most of the solar spectrum.25−27 Furthermore, since ∼81% of the solar irradiance is composed of photons with an energy greater than the band gap of Si (1.1 eV), the SiNWs absorb >75% of the solar irradiance after accounting for reflection losses. Due to its small band gap, light energy absorbed by the SiNWs is primarily converted into heat via thermalization and nonradiative recombination of photoexcited electron−hole pairs. By using concentrated light at an intensity of ∼20 kW/m2 (∼20 suns) temperatures of 150 °C can easily be reached without the need for external heating.25 Such concentrated solar energy can be generated by simple, inexpensive parabolic trough solar concentrators.28 Using a thin Ru film deposited on these photoactive SiNW supports, our group was able to demonstrate efficient conversion of CO2 to CH4 with rates on the order of 1 mmol·gcat−1·h−1 without the use of external heating.25 Further, these experimental results suggest that, in addition to supporting the reaction photothermally by providing heat energy, a small fraction of photons absorbed by the SiNWs were able to generate electron−hole pairs that photochemically accelerate the reaction by facilitating the formation of active hydrogen atoms on the Ru surface that participate in the overall photomethanation reaction.25 On the basis of these promising results, we sought to determine if SiNW supports could enhance the reaction rates of our In2O3−x(OH)y nanoparticle photocatalysts, improve the utilization of the solar spectrum, and remove the need for external heating. To this end, we have prepared vertically aligned SiNWs evenly coated by In2O3−x(OH)y nanoparticles, which can reduce CO2 to CO at a rate of 22.0 μmol·gcat·h−1 without using external heating. This stands in stark contrast to both In2O3−x(OH)y-coated glass substrates and In2O3−x(OH)y nanoparticles deposited on SiNW supports in a bilayer configuration, which exhibit substantially lower reaction rates

RESULTS AND DISCUSSION Fabrication and Characterization of In2O3−x(OH)y/ SiNW Hybrid Nanostructures. In2O3−x(OH)y/SiNW hybrid films were prepared by drop-casting a ligand-free, colloidally stable indium hydroxide nanoparticle suspension onto freshly prepared 1 cm2 SiNW arrays. The resulting films were dried and then calcined at 250 °C for 3 h to convert the indium hydroxide nanoparticles into nonstoichiometric, hydroxylated indium oxide (In2O3−x(OH)y) via controlled thermal dehydration. This produced an even coating of ∼1.5 mg of In2O3−x(OH)y over the SiNW array, as shown in the scanning electron microscope (SEM) images in Figure 1a. The higher magnification inset in Figure 1d indicates that the In2O3−x(OH)y nanoparticle sizes are relatively small, in most cases much less than 100 nm in diameter. Two control samples were also prepared. First, in order to investigate the effects of the In2O3−x(OH)y nanoparticle loading and distribution over the surface area of the SiNW support, the colloidal indium hydroxide nanoparticles were dried and precalcined to produce porous In2O3−x(OH)y nanoparticle aggregates, 1.5 mg of which was deposited onto the SiNW support to form the bilayer structure shown in Figure 1b. The higher magnification inset in Figure 1e illustrates the nanoparticle aggregates are still porous and the particle size is similar to the evenly coated In2O3−x(OH)y/SiNW hybrid films. As an additional control to examine the effect of the substrate, the same quantity of indium hydroxide nanoparticle suspension was drop-cast onto a roughened glass substrate and calcined under the same conditions, producing a porous In2O3−x(OH)y nanoparticle film (Figure 1c). The higher magnification inset in Figure 1f indicates that porosity and particle size are comparable to the other two films. The energy dispersive X-ray (EDX) mapping data in Figure 1d−f and Figure S1a−c confirm that the observed particles coating the SiNW and glass substrates are indium-containing. In particular, on the evenly coated In2O3−x(OH)y/SiNW hybrid films (Figures 1d and S1a), it is important to note that the distribution of indium is very even throughout the nanowire structure, indicating that the nanoparticles are able to penetrate all the way to the base of the nanowires, even with the simple drop-casting technique used. One possible reason for this is the SiNWs are highly hydrophilic, and, upon addition, the aqueous suspension of indium hydroxide nanoparticle suspension is visibly drawn into the internanowire spaces by capillary forces. On the other hand, in the case of the In2O3−x(OH)y/SiNW bilayer structure (Figures 1e and S1b), very little indium is 9018

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

ACS Nano

reactor window. For comparison, similar reaction rate tests were carried out at the same temperature, but in the dark. A summary of the GC/MS data is presented in Figures S3−S6 in the Supporting Information. 13C-Labeled CO2 was used to distinguish carbon-containing products produced by the direct reduction of 13CO2 and carbon-containing products resulting from reactions with adventitious sources of carbon, which are composed of the naturally abundant 12C isotope. Figure 2

Figure 1. Cross-sectional scanning electron microscope (SEM) images of (a) evenly coated In2O3−x(OH)y/SiNW hybrid films, (b) In2O3−x(OH)y/SiNW bilayer films, and (c) In2O3−x(OH)y on roughened glass. Energy dispersive X-ray (EDX) mapping of (d) evenly coated In 2 O 3 − x (OH) y /SiNW hybrid films, (e) In2O3−x(OH)y/SiNW bilayer films, and (f) In2O3−x(OH)y on roughened glass, showing the distribution of silicon (blue), indium (red), and oxygen (green) within the three samples. All scale bars in the EDX mapping data correspond to 2 μm. The higher resolution SEM inset on the right of parts d−f illustrates the porosity and nanoparticle distribution of the respective sample.

Figure 2. 13CO production rates of evenly coated In2O3−x(OH)y/ SiNW, bilayer In2O3−x(OH)y/SiNW, and In2O3−x(OH)y/glass in the dark and under illumination, with or without external heating.

shows a comparison of the 13CO production rates for the three samples. From these data, it is clear that no 13CO was produced under dark conditions for any of the films; however under ∼20 kW/m2 (∼20 suns) of illumination a substantial difference in the 13CO production is observed. The highest activity is observed for the evenly coated In2O3−x(OH)y/SiNW film, which produces on average 22.0 μmol·gcat−1·h−1, while under identical conditions the bilayer In2O3−x(OH)y/SiNW film produces 13CO at a rate of 14.4 μmol·gcat−1·h−1, a decrease of approximately 35%. As illustrated in the reactor temperature profiles in Figure S7 in the Supporting Information, under the high light intensity from the Xe lamp, both the bilayer and evenly coated In2O3−x(OH)y/SiNW films easily reached a temperature of 150 °C, which as demonstrated in previous studies is required to facilitate catalytic turnover of the active sites in the In2O3−x(OH)y nanoparticle photocatalyst by releasing the water byproduct of the RWGS reaction.22,23 These results clearly demonstrate that the SiNW substrate is capable of facilitating the photocatalytic reduction of 13CO2 to 13 CO by In2O3−x(OH)y via the RWGS reaction without the use of an external heat source. On the other hand, the In2O3−x(OH)y/glass control sample shows negligible 13CO production in both the dark at 150 °C and under ∼20 kW/m2 (∼20 suns) of illumination. Under concentrated light from the Xe lamp and without external heating, the In2O3−x(OH)y/glass control sample was only able to reach a maximum of 110 °C (Figure S7), which is too low to facilitate the RWGS reaction on the In2O3−x(OH)y surface.22,23 In order to provide a fairer comparison between the In2O3−x(OH)y/glass and In2O3−x(OH)y/SiNW films, a third test was performed wherein external heating was used to elevate and maintain the temperature of the In2O3−x(OH)y/glass sample at 150 °C while the sample was illuminated with the Xe

observed in the SiNW layer, indicating that a true bilayer structure was formed. The powder X-ray diffraction (PXRD) patterns shown in Figure S1d of the Supporting Information confirm that after calcination the indium hydroxide nanoparticles deposited on both the SiNWs and on glass produce pure crystalline cubic, bixbyite indium oxide. The diffraction patterns are almost identical, with the only difference being the two very sharp peaks at 33° and 62° 2θ (marked with an asterisk) in the evenly coated and bilayer In2O3−x(OH)y/SiNW patterns. These reflections originate from the [100] face of the Si wafer, which was etched to make the SiNW substrate. The fact that only those two peaks appear, and not the rest of the peaks in the Si diffraction pattern, indicates that the preferred orientation is maintained and the SiNWs remain singlecrystalline. Photocatalytic Activity of the In2O3−x(OH)y/SiNW Hybrid Nanostructures. We performed a series of reaction rate tests to evaluate and compare the photocatalytic activity of t h e e v e n l y c o a t e d I n 2 O 3 − x (O H) y /S iNW, b ilayer In2O3−x(OH)y/SiNW, and In2O3−x(OH)y/glass films toward the RWGS reaction (CO2 + H2 → CO + H2O). The three films were sealed inside a custom-designed stainless steel reactor with a quartz window, which was charged with a 1:1 mixture of H2 and 13CO2 gas (stoichiometric for the RWGS reaction) to a total pressure of 2 atm. During the photocatalytic reaction rate tests, the sample was illuminated with a 300 W Xe lamp, focused to an intensity of ∼20 kW/m2 (∼20 suns) through the 9019

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

ACS Nano lamp. In this case, under illumination at 150 °C, the RWGS reaction proceeded at a rate of 3.8 μmol·gcat−1·h−1 over the In2O3−x(OH)y/glass sample, which is roughly 6 times lower than the RWGS rates observed over the evenly coated In2O3−x(OH)y/SiNW sample under similar temperature and light conditions. The external quantum efficiency (EQE) and internal quantum efficiency (IQE) for the evenly coated and bilayer In 2 O 3 − x (OH) y /SiNW films as well as the In2O3−x(OH)y/glass film for the case in which external heat was provided are plotted in Figure S9. In calculating the QE values reported in Figure S9 only incident photons with energy greater than the band gap of In2O3−x(OH)y (λ < ∼450 nm), which are capable of driving the RWGS, were considered. The trends of the IQE and EQE values of the In2O3−x(OH)y/SiNW and In2O3−x(OH)y/glass films in Figure S9 are consistent with the 13CO production rates reported in Figure 2. Investigation of the Substrate’s Role in Enhancing the CO2 Reduction Rates. The reaction rate results presented in Figure 2 along with the temperature profiles shown in Figure S7 clearly demonstrate that the SiNW support heats up under high-intensity irradiation from the Xe lamp and transfers thermal energy to the In2O3−x(OH)y nanoparticle catalysts to drive the RWGS reaction. However, it is not clear whether photoexcited charge carriers generated within the SiNW support provide any photochemical contribution to the reaction by either directly reducing CO2 to CO or transferring photogenerated electrons to the In2O3−x(OH)y nanoparticle catalysts in a manner that enhances the RWGS reaction rate. We see no evidence of 13CO production on bare SiNWs in the absence of In2O3−x(OH)y, indicating that the SiNW support alone does not function as a photochemical or thermochemical catalyst. However, as illustrated in the band offset diagram for Si and In2O3−x(OH)y shown in Figure 3a, the conduction band

band gap of In2O3−x(OH)y is around 425 nm (2.9 eV); so by blocking all photons with an energy greater than 495 nm, we can selectively photoexcite electrons in the SiNW support without photoexciting charge carriers across the band gap of In2O3−x(OH)y. As shown in Figure 3b, the RWGS reaction rate when the 495 nm cutoff filter is used is significantly lower than the rates observed when the evenly coated In2O3−x(OH)y/ SiNW film is tested without using a filter. However, a small amount of 13CO is observed, even with the 495 nm cutoff filter, corresponding to a 13CO production rate of 0.8 μmol·gcat−1·h−1 (Figure 3b). This indicates that while photogenerated electron transfer from the SiNW support to the In2 O 3−x (OH) y nanoparticle catalysts may occur to some extent, it is not the dominant cause for the significant enhancement of the RWGS reaction rates over the In2O3−x(OH)y/SiNW films compared to those rates observed over the In2O3−x(OH)y/glass films. Another possible explanation for the observed enhancement in 13CO production rates for the In2O3−x(OH)y/SiNW catalyst film compared to the rates observed over the In2O3−x(OH)y/ glass film is the dramatically different optical properties of the two substrates. Figure 4a shows the reflectance spectra of the e v e n l y c o a t e d I n 2 O 3 − x (OH ) y / S i N W , t h e bi l a y e r In2O3−x(OH)y/SiNW, and In2O3−x(OH)y/glass films as well as the bare SiNW film. The reflectance spectrum of the evenly coated In2O3−x(OH)y/SiNW sample overlaps almost perfectly with the reflectance spectrum of the bare SiNW film, averaging a reflectance of just 3.4% over the entire UV to NIR spectral range. By contrast, both the bilayer In2O3−x(OH)y/SiNW and the In2O3−x(OH)y/glass films show substantially higher reflectance across the entire spectral range. The bilayer In2O3−x(OH)y/SiNW film averages 8.6% reflectance over the entire UV to NIR spectral rangemore than double that of the evenly coated In2O3−x(OH)y/SiNW filmindicating that although the amount of In2O3−x(OH)y nanoparticles deposited on both films is identical, the distribution of these nanoparticles plays a key role in the optical properties of the resulting film. The In2O3−x(OH)y/glass film has an average reflectance of 12.5% over the entire spectral range. However, it should be noted that while both the In2O3−x(OH)y/SiNW samples are opaque due to the opacity of the SiNW support, the In2O3−x(OH)y/glass sample is translucent, as shown in the photograph in Figure 4c. In the case of the In2O3−x(OH)y/ SiNW samples, where transmittance is negligible (see Figure S10), low reflectance implies enhanced light trapping and absorption in the SiNW support. On the other hand, for the In2O3−x(OH)y/glass sample, where transmittance is not negligible, while the reflectance averages around 12.5%, the amount of light actually absorbed by the sample is significantly lower. In order to estimate the light absorption by the In2O3−x(OH)y/glass film, we measured a second reflectance spectrum in which the transmission was blocked by placing a standard Spectralon reflector at the rear side of the glass substrate such that transmitted light was reflected back through the front side of the In2O3−x(OH)y/glass film (red dashed curve in Figure 4a). As expected, the In2O3−x(OH)y/glass film absorbs very little light throughout the visible and near-IR region of the spectrum and only begins to absorb strongly at wavelengths shorter than 500 nm, which corresponds to the onset of absorption of In2O3−x(OH)y. The effect of this difference in optical properties between the In2O3−x(OH)y/ SiNW films and In2O3−x(OH)y/glass film can be clearly seen in Figure 4d−f. These photographs show the three samples in the reactor under identical illumination of ∼20 kW/m2 (∼20 suns).

Figure 3. (a) Band diagram indicating the respective energy level offsets for the conduction band (CB) and valence band (VB) of Si and In2O3−x(OH)y. (b) 13CO production rates for the evenly coated In2O3−x(OH)y/SiNW under ∼20 kW/m2 (∼20 suns) of illumination, both with and without a 495 nm high-pass cutoff filter.

(CB) minimum of Si lies above the CB of In2O3−x(OH)y, indicating that photogenerated electron transfer from Si to In2O3−x(OH)y may be possible. In order to determine if the transfer of photogenerated electrons from the SiNW support to In2O3−x(OH)y enhances the RWGS reaction rate, we measured the 13CO production rates of the evenly coated In2O3−x(OH)y/ SiNW film both with and without a 495 nm high-pass cutoff filter. By using a filter, a portion of the solar spectrum is removed; so in order to ensure a fair comparison between the two tests, we increased the intensity of the Xe lamp to ensure the temperature of the SiNW support reached 150 °C. The 9020

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

ACS Nano

Figure 4. (a) UV−vis−NIR diffuse reflectance spectra of the sample films. Due to the transparency of the In2O3−x(OH)y/glass film, a second reflectance spectrum is also shown in which the transmission is blocked by covering the back of the sample film with a 100% reflectance standard. For comparison, the reflectance spectrum of bare SiNWs is also shown. (b) Hemispherical infrared reflectance for the four catalyst films measured from 3 to 25 μm, with the emissivity values for each film provided in the inset. (c) Photographs of the evenly coated In2O3−x(OH)y/SiNW film (left), bilayer In2O3−x(OH)y/SiNW film (middle), and In2O3−x(OH)y/glass film (right). (d−f) Photographs of the sample films inside the reactor under the same light intensity of ∼20 kW/m2 (∼20 suns): (d) evenly coated In2O3−x(OH)y/SiNW film, (e) bilayer In2O3−x(OH)y/SiNW film, and (f) In2O3−x(OH)y/glass film.

Figure 5. Comparison between the absorption spectrum of In2O3−x(OH)y nanoparticles and the photon utilization of the solar irradiance for (a) In2O3−x(OH)y/SiNW hybrid materials and (b) In2O3−x(OH)y/glass films. The yellow and red shading illustrate the different light harvesting mechanisms between photons with energy greater than the band gap of the In2O3−x(OH)y, which have the ability to create an electron/hole pair within the nanoparticles (yellow), and photons with energy less than the band gap of In2O3−x(OH)y, which are absorbed and converted into heat energy within the SiNW support (red). (c) Schematic illustration of the effect of nanostructuring on the film’s interaction with incident light. The blue arrows represent incident solar irradiation, and the red shading for the evenly coated and bilayer In2O3−x(OH)y/SiNW films illustrates photothermal heat generation.

The evenly coated In2O3−x(OH)y/SiNW film (Figure 4d) exhibits some reflection; however, the dark color of the

substrate is still visible despite the intense illumination. In the case of the bilayer sample (Figure 4e), the reflection is visibly 9021

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

ACS Nano higher than the evenly coated In2O3−x(OH)y/SiNW film. However, the reflection from the In2O3−x(OH)y/glass film (Figure 4f) is so bright that it is saturating the camera’s detector, indicating that a substantial portion of the incident light is reflected away from the sample and is not utilized to drive the photocatalytic reduction of CO2. Light harvesting ability is an important property for any photodriven process, and as these results demonstrate, this type of reflective loss can greatly impact the photocatalytic performance. The amount of light reflected increases with increasing refractive index (RI) mismatch across the material interface, which in this case is determined by the RI of the reactant gases and the surface of the catalyst film. Here it can be noted that the RI of silicon is ∼3.4−4, depending on which spectral region is under consideration, and the RI of the reactant gases can be assumed to be equal to 1, similar to that of air.29 Porous materials with subwavelength features, such as the SiNWs used in this study, exhibit effective RI values between that of the mediums they are composed of. Thus, the SiNW support will have an effective RI value between 1 and 4, depending on the volume fraction of its gas and solid phases.26,30 Consequently, there exists an effective RI gradient at the surface of the SiNW support, starting with a value of 1 just outside the surface and gradually increasing to ∼4 at the base of the Si nanowires. This RI gradient reduces the abrupt RI contrast at the SiNW surface, significantly lowering the amount of light reflected. Moreover, once incident light has been transmitted into the SiNW region, light trapping can be further enhanced by scattering and internal reflections within the vertically aligned >5 μm long nanowires,26,30 as illustrated in the schematic diagram in Figure 5c. The fact that reflection losses are reduced in the evenly coated In2O3−x(OH)y/SiNW film is evident in the diffuse reflectance spectra plotted in Figure 4a. For example, at 450 nm, which corresponds to solar photons with energy just below the band edge of In2O3−x(OH)y, the percent reflectance for the evenly coated In2O3−x(OH)y/SiNW film is only 3.6%, whereas that of the In2O3−x(OH)y/SiNW bilayer film and In2O3−x(OH)y/glass film are 14.5% and 18.7%, respectively. The significant difference in reflection between the evenly coated In2O3−x(OH)y/SiNW film and the other two films can be attributed to changes in optical properties caused by the agglomerated In2O3−x(OH)y nanoparticle film, which forms the top layer of the bilayer In 2 O 3−x (OH) y /SiNW and In 2 O 3 − x (OH) y /glass films. While, individually, the In2O3−x(OH)y nanoparticles, which are approximately 20 nm in size, do not reflect any significant amount of light, when the In2O3−x(OH)y nanoparticles agglomerate into larger particles with a diameter greater than ∼100 nm, the combined reflection from these larger nanoparticle aggregates can be significant. The high reflective losses for the In2O3−x(OH)y/glass film are not surprising, as this substrate was not designed to harvest light efficiently; however, the substantial difference in reflected light between the evenly coated and bilayer In2O3−x(OH)y/ SiNW films clearly demonstrates the importance of the distribution of In2O3−x(OH)y nanoparticle photocatalysts on the support. This is illustrated in the schematic diagram in Figure 5c, which shows a greater amount of reflection from the In2O3−x(OH)y/glass and In2O3−x(OH)y/SiNW bilayer film, as compared to the evenly coated In2O3−x(OH)y/SiNW hybrid film. Another important parameter to consider when designing a photothermal support for solar-driven catalysis is its emissivity,

which describes the ability of an object to emit energy in the form of thermal radiation. Emissivity ranges from 0 to 1, where a value of 1 indicates that the object emits the maximum amount of radiation, equivalent to that of an ideal blackbody. The emissivity of an object is the fraction of thermal energy it radiates compared to that of an ideal blackbody at a given temperature, which can be calculated using the Stefan− Boltzmann law. In order to maximize the temperature of a photothermal support for solar-driven catalysts, it is desirable to have high absorption values over the solar spectral region, from ∼300 to ∼3000 nm, but to have as low of an emissivity value as possible over the mid-infrared region to minimize thermal radiation losses. To evaluate the emissivity of the photocatalysts studied in this work, we measured the hemispherical reflectance of the three catalyst films and the bare SiNW support over the midinfrared spectral region, from 3 to 25 μm, plotted in Figure 4b. These reflectance measurements were performed with and without a gold-plated mirror placed on the rear side of each catalyst film, and the resulting spectra were used to estimate the emissivity (for more details on this calculation, see Figure S12 in the Supporting Information). As shown in the inset in Figure 4b, the emissivity of the In2O3−x(OH)y/glass, bare SiNW, bilayer In 2 O 3 − x (OH) y /SiNW, and evenly coat ed In2O3−x(OH)y/SiNW hybrid films were estimated to be 0.97, 0.44, 0.56, and 0.59, respectively. The emissivity value of 0.97 for the In2O3−x(OH)y/glass film is comparable to values generally reported for glass, while the emissivity value of 0.44 for the bare SiNW wafer is comparable to that reported in the literature for Si.31 According to Kirchoff’s law, at a given wavelength and at thermal equilibrium, the emissivity of an object equals its absorption. For the NIR spectral region, considering 1.1 μm < λ < 6 μm, the photon energy is too low to be absorbed via band-to-band transitions in Si (with its band gap of 1.1 eV); therefore the emissivity is very low in this region ( ∼6 μm, phonon absorption due to lattice vibrations becomes more pronounced and the emissivity increases.31 With the addition of the In2O3−x(OH)y nanoparticles, the emissivity of the SiNW increases from 0.44 for the bare SiNW film to 0.56 and 0.59 for the bilayer and evenly coated In2O3−x(OH)y/SiNW films, respectively. This increase is most likely attributed to absorption of mid-infrared photons by free carriers occurring in the conduction band of the In2O3−x(OH)y nanoparticles.32 Furthermore, the fact that the emissivity of the evenly coated In2O3−x(OH)y/SiNW hybrid film is slightly larger than that of the In2O3−x(OH)y/SiNW bilayer film suggests that the hybrid film absorbs more incident radiant energy in the mid-infrared spectral region. In looking at the cross-sectional SEM images shown in Figure 1a and b it is likely that the clumps of agglomerated In2O3−x(OH)y nanoparticles on the SiNW surface for the In2O3−x(OH)y/SiNW bilayer film scatter and reflect some portion of the mid-infrared spectral region, thereby lowering its absorption, or equivalently its emissivity, compared with the evenly coated In2O3−x(OH)y/SiNW hybrid film. However, despite this slight increase in emissivity upon addition of the In2O3−x(OH)y nanoparticles, the emissivity of the In2O3−x(OH)y/SiNW films is much less than that of the In2O3−x(OH)y/glass films. Thus, the high absorption over the solar spectral region coupled with relatively low emissivity values in the mid-infrared region explain why the In2O3−x(OH)y/SiNW films are able to heat to a much greater temperature (150 °C) compared to that of the In2O3−x(OH)y/ 9022

DOI: 10.1021/acsnano.6b05416 ACS Nano 2016, 10, 9017−9025

Article

ACS Nano glass films (110 °C) under the same intensity of solar simulated radiation. Designing a Catalyst Support for Improved Photocatalytic Performance. As this study illustrates, the substrate can play an active role in enhancing the photocatalytic performance of an already active nanomaterial. In photocatalysis, utilizing light from the entire solar spectrum is important to achieve optimal efficiency. Vertically aligned SiNW substrates are an attractive option that can harvest light energy from the UV to near-IR spectral regions. As demonstrated herein, light with energy above the band gap of In2O3 is utilized to create e−/h+ pairs that drive the photochemical reduction of CO2 to CO, while light with energy below the band gap of In2O3−x(OH)y is absorbed by the SiNWs and converted into heat energy to facilitate water desorption and catalytic turnover of the active site on In2O3−x(OH)y (illustrated schematically in Figure 5a). This stands in stark contrast to the solar light harvesting ability of the In2O3−x(OH)y/glass film (Figure 5b), which is unable to utilize solar photons with energy less than that of the band gap of In2O3−x(OH)y, highlighting the distinct advantage of using the SiNW support. Further, as evidenced in the SEM images shown in Figure 1, the In2O3−x(OH)y nanoparticles, on average, have a greater degree of contact with the SiNW support for the evenly coated hybrid In2O3−x(OH)y/SiNW film as compared to the bilayer In2O3−x(OH)y/SiNW film. We expect that heat generated in the SiNW support upon thermalization of absorbed photons may be more readily transferred to the In 2 O 3−x (OH) y nanoparticles in the evenly coated film compared to the bilayer film. In the bilayer film only the bottom of the In2O3−x(OH)y nanoparticle layer is in contact with the SiNWs, and there is likely a decreasing temperature gradient going from the In 2 O 3−x (OH) y nanoparticles at the bottom of the In2O3−x(OH)y layer, which are adjacent to the SiNWs, to the In2O3−x(OH)y nanoparticles situated at the upper surface of the layer, which are not in direct contact with the SiNW support. Therefore, we expect that heat transfer from the SiNW support to the In2O3−x(OH)y nanoparticles is less pronounced in the bilayer structure, resulting in lower photocatalytic activity. Further, as the SEM images in Figure 1 illustrate, the In 2 O 3− x (OH) y nanoparticles on the evenly coated In2O3−x(OH)y/SiNW substrates are more exposed to the surrounding gases compared to the In2O3−x(OH)y/SiNW bilayer structure, which is composed of In 2 O 3−x (OH) y nanoparticle agglomerates. We expect this increased accessibility to the surface area of the In2O3−x(OH)y nanoparticles in the evenly coated SiNW film will improve the gas−nanoparticle contact and reduce mass transfer limitations that may result from restricted reactant or product gas diffusion. The strategies described herein are not limited to CO2 reduction applications and could foreseeably be applied to enhance the photocatalytic activity of many different materials for a variety of gas-phase chemical reactions. Further, as demonstrated in this study, it is not necessary for photogenerated electron transfer to occur in order to observe a substantial increase in reaction rates. This type of hybrid support structure, which combines both photothermal effects and improved light harvesting ability, would be suitable for any gas-phase reaction that requires both light energy and heat energy.

CONCLUSIONS In this study, we examined the impact of using vertically aligned SiNWs as an active support for In2O3−x(OH)y-based gas-phase photocatalysts for the reduction of CO2 to CO via the RWGS reaction. These hybrid nanostructured materials utilize the solar spectrum more efficiently than In2O3−x(OH)y alone. Solar photons with energy greater than the band gap of In2O3−x(OH)y (>2.9 eV) contribute to the photocatalytic reduction of CO2, while sub-band-gap photons (