Photodeposition of Copper and Chromia on Gallium Oxide: The Role of Co-Catalysts in Photocatalytic Water Splitting

Authors

  • Dr. G. Wilma Busser,

    1. Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115
    Search for more papers by this author
  • Dr. Bastian Mei,

    1. Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115
    Search for more papers by this author
  • Anna Pougin,

    1. Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115
    Search for more papers by this author
  • Dr. Jennifer Strunk,

    1. Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115
    Search for more papers by this author
  • Ramona Gutkowski,

    1. Analytische Chemie - Elektroanalytik & Sensorik, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany)
    Search for more papers by this author
  • Prof. Wolfgang Schuhmann,

    1. Analytische Chemie - Elektroanalytik & Sensorik, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany)
    Search for more papers by this author
  • Dr. Marc-Georg Willinger,

    1. Anorganische Chemie, Fritz-Haber-Institut, Faradayweg 4-6, 14195 Berlin (Germany)
    Search for more papers by this author
  • Prof. Robert Schlögl,

    1. Anorganische Chemie, Fritz-Haber-Institut, Faradayweg 4-6, 14195 Berlin (Germany)
    Search for more papers by this author
  • Prof. Martin Muhler

    Corresponding author
    1. Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115
    • Laboratory of Industrial Chemistry, Ruhr-Universität Bochum, Universitätsstr. 150, 44780 Bochum (Germany), Fax: (+49) 0234-32-14115===

    Search for more papers by this author

Abstract

original image

Split second: The photocatalytic activity of gallium oxide (β-Ga2O3) depends strongly on the co-catalysts CuOx and chromia, which can be efficiently deposited in a stepwise manner by photoreduction of Cu2+ and CrO42−. The water-splitting activity can be tuned by varying the Cu loading in the range 0.025–1.5 wt %, whereas the Cr loading is not affecting the rate as long as small amounts (such as 0.05 wt %) are present. Chromia is identified as highly efficient co-catalyst in the presence of CuOx: it is essential for the oxidation of water.

Photocatalytic water splitting using semiconductor particles has been studied intensely as a means for producing hydrogen as renewable energy carrier. Although numerous efforts have been made to optimize photocatalytic water splitting systems many challenges still remain, such as the efficient deposition of co-catalysts on the light-absorbing semiconductor particles. Co-catalysts are applied to collect the photogenerated electrons and holes and to provide active sites for the reduction of water, yielding hydrogen (hydrogen evolution reaction, HER), and for the water oxidation, yielding oxygen (oxygen evolution reaction, OER). The deposited co-catalysts are usually present in the form of nanoparticles. NiOx,1 Rh2−yCryO32 and Cr2O3/noble metal (shell/core)3 nanoparticles function as effective HER co-catalysts as well as RuO24 as an effective OER co-catalyst for the overall photocatalytic splitting of water into H2 and O2 (2:1 ratio) in the absence of sacrificial agents. To reduce the cost of the related devices it is highly important to develop noble metal-free co-catalysts based on materials that are readily available.5 It has been shown previously that copper-based co-catalysts can be used as a nontoxic, non-noble-metal co-catalysts on GaN:ZnO6 as well as on TiO27 and ZnO.8 In particular, a mixed oxide based on copper and chromium on GaN:ZnO was introduced as a noble-metal-free co-catalyst for photocatalytic water splitting. It was shown that the activity was strongly dependent on the generation of CuII–CrIII mixed oxide nanoparticles with optimal composition. CuO/GaN:ZnO was reported to be inactive, but in the presence of a K2CrO4 solution, H2 and O2 evolution was observed and it was suggested that an amorphous Cr2O3 shell is deposited through photoreduction, which prevents the back reaction from taking place on the co-catalyst, because it is permeable to protons, but not to O2.6 Furthermore, copper oxides are reported to be active photocatalysts on their own as the bandgaps of the copper oxides Cu2O and CuO are both suitable for the photosplitting of water in the presence of a sacrificial agent and for the overall water splitting reaction using visible light.9, 10 However, copper based co-catalysts are also reported to suffer from deactivation depending on the preparation method and the testing conditions.11

We recently described the stepwise photodeposition of Rh and Cr2O3 on Ga2O3 and on Ga/Zn oxynitride,12 as the Cr2O3/Rh shell/core composite has been described to be the most effective co-catalyst for overall water splitting.2, 3 We demonstrated that the optimum loading of the co-catalysts has to be determined with great care to obtain the highest reactivity with a minimum noble metal loading.12 However, doubts on the validity of the shell/core model remain.6, 12, 13

Here, we demonstrate that photodeposition of Cu from Cu(NO3)2 and Cr2O3 from K2CrO4 on Ga2O3 leads to highly active photocatalysts for overall water splitting, using the stepwise addition of precursors and aqueous methanol as reducing agent. CuOx–Cr2O3/Ga2O3 catalyst was chosen as the simplest highly active photocatalyst allowing us to gain deeper insight.14 Our results clearly indicate that the formation of Cr2O3 (chromia) shells on the CuOx cores is unlikely. Instead, chromia in the presence of CuOx is identified as co-catalyst involved in both the H2 and the O2 evolution reactions.

In the first experiment, a suspension of 1 g of Ga2O3 in 550 mL H2O and 50 mL methanol was exposed to UV light (Figure 1 a). The H2 concentration increased with irradiation time until a stable maximum of about 1.2 % H2 was reached, which can be attributed to the continuous photocatalytic reforming of aqueous methanol over pure Ga2O3:(1)

equation image(1)
Figure 1.

a) Photodeposition of Cu on Ga2O3: Hydrogen evolution as a function of Cu loading. The experiments were performed using 1000 mg of Ga2O3 in a solution containing 50 mL MeOH and 550 mL H2O. b) Overall water splitting with 0.4 wt % Cu/Ga2O3 before and after addition of K2CrO4 using 300 mg of catalyst in 550 mL H2O.

Obviously, pure Ga2O3 has a considerable activity for this reaction upon irradiation with UV light. The CO2 concentration increased simultaneously with the H2 concentration to a value of about 0.002 %. This value is very low, indicating that the high concentration of H2 cannot be accounted for by assuming full oxidation according to the stoichiometry of Equation (1). Therefore, the major part of H2 results from partial oxidation to products such as formaldehyde and formic acid, which remain in the liquid phase.11 Subsequently, the light was turned off and 40 mL of an aqueous solution of Cu(NO3)2 was added to achieve a theoretical loading of 0.4 wt % Cu. The H2 concentration rose to essentially the same maximum value of about 1.2 % (Figure 1 a). The same experiment was performed with different amounts of Cu(NO3)2, sufficient for loadings in the range from 0.025 wt % to 1.5 wt % (Figure 1 a). When applying higher Cu precursor concentrations the H2 evolution rate exceeded the value of bare Ga2O3, whereas for a very low Cu(NO3)2 concentration, corresponding to a theoretical loading of 0.025 wt % Cu, the resulting H2 evolution rate was lower than that of bare Ga2O3 (Figure 1 a). Except for low Cu concentrations, a sudden increase in CO2 concentration was observed during the photodeposition experiments shortly after turning on the light (Supporting Information, Figure SI1), but after a short period also very suddenly the CO2 concentration dropped again to zero and remained zero during continuous irradiation. It is assumed that the first CO2 peak (Figure SI1) originates from the oxidation of methanol during photocatalytic reduction of the Cu(NO3)2 precursor:(2)

equation image(2)

As soon as the CO2 signal disappears, active Cu0 species are present and the H2 concentration [Eq. (1)] increases due to photocatalytic methanol reforming. After having reached a maximum, the H2 evolution gradually decreased as a function of time-on-stream indicating a slow photodegradation of the active Cu0 species, which may be caused by blocking due to strongly adsorbing organic intermediates.11 While it was confirmed that the CO2 concentration remained zero for at least 3 h of irradiation for the samples with Cu loadings in the range of 0.4–1.5 wt %, the CO2 concentration rose after prolonged irradiation time for lower copper concentrations. Elemental analysis confirmed the Cu loadings of 0.03 wt % Cu to 1.6 wt % Cu and, depending on the Cu loading, the samples were bluish–red or pinkish–red in color.

The overall water splitting performance of Cu-modified Ga2O3 was subsequently evaluated (Figure 1 b). In the initial period, the simultaneous formation of H2 and CO2 was observed, which can be attributed to the removal of carbonaceous species from the catalyst surface, possibly remaining on the surface after photodeposition of Cu from aqueous methanol solutions.11 However, no O2 formation was detected, indicating that the 0.4 wt % Cu/Ga2O3 catalyst was not active for water-splitting in the liquid phase. After dosing a small amount of the Cr precursor corresponding to a theoretical loading of 0.05 wt % Cr, water splitting was observed, yielding the stoichiometric ratio of H2/O2=2:1. The activity did not increase further when increasing the amount of photodeposited chromia stepwise to 0.15 wt % Cr. Therefore, small Cr amounts are sufficient to alter the catalyst performance, allowing for overall water-splitting in the liquid phase. It is reasonable to assume that Cu is present as copper oxide under the photocatalytic water-splitting conditions in the absence of methanol as reducing agent.

In order to more closely investigate the effect of the addition of chromia, photodeposition experiments in aqueous methanol solution were performed, in which the Cr precursor corresponding to 0.09 wt % was photodeposited either before, after, or simultaneously with the photodeposition of 0.66 wt % Cu. The rate of H2 evolution is raised significantly in the presence of photodeposited chromia as compared to bare Ga2O3 or the Cr-free Cu-Ga2O3, and it does not seem to matter for the overall H2 production after 3 h, whether the Cr photodeposition is performed before, simultaneously, or after the photodeposition of Cu (Figure 2). Correspondingly, the CO2 evolution is rather similar in the different experiments (Supporting Information, Figure SI2). The CO2 evolution profiles suggest that the Ga2O3 catalysts loaded with Cu and Cu+Cr have a certain storage capacity for CO2 generated from aqueous photocatalytic methanol reforming as a complex function of the Cu loading. Further studies are in progress to elucidate the mechanism of CO2 formation from methanol during the photodeposition. Subsequently, the effect of Cr (0.09 wt %) on the photodeposition of different amounts of Cu was investigated. The H2 evolution rate clearly exceeds those of the bare Ga2O3 or the Cr-free Cu-Ga2O3 catalysts (Supporting Information, Figure SI3). Nevertheless, only a slight dependence of the H2 evolution rate on the Cu loading was observed. The H2 evolution appears to be maximized for a Cu loading of 0.66 wt %.

Figure 2.

Photodeposition of Cu (0.66 wt %) and Cr (0.09 wt %) on Ga2O3: hydrogen evolution as a function of the photodeposition sequence. The experiments were performed using 1000 mg of Ga2O3 in a solution containing 50 mL MeOH and 550 mL H2O.

Transmission electron microscopy (TEM) characterization of samples prepared by photodeposition of either Cu or Cu+Cr showed that the β-Ga2O3 rods are characterized by a porous structure, a rectangular cross-section, and single crystal-like diffraction behavior (Supporting Information, Figure SI4). Photodeposition of Cu led to homogeneously distributed nanoparticles on the surface of the rods. The average particle size is around 2 nm. However, the size distribution is relatively large ranging from below 1 nm up to 10 nm (Figure 3 a, b). Selected-area electron diffraction (SAED, Figure 3 c) and elemental analysis by STEM-EDX (not shown) indicate that the small nanoparticles correspond to Cu2O. Some larger particles of metallic Cu were also detected. In samples with similar Cu loading the presence and distribution of Cu2O and Cu nanoparticles is independent of the presence of Cr. However, due to the low loading and high dispersion, it was not possible to observe Cr in Cr-containing samples by means of TEM. The successful deposition of Cr was demonstrated by X-ray photoelectron spectroscopy (XPS), indicating that Cr is present as CrIII as in Cr2O3 (Supporting Information, Figure SI5). Furthermore, XPS supports the presence of CuI as in Cu2O.15

Figure 3.

a, b) HAADF STEM shows the distribution of small particles on the surface of the β-Ga2O3 rods. c) SAED confirms the assignment of the particles to Cu2O.

The photocatalysts with 0.09 wt % Cr and various Cu loadings were tested for overall water splitting (Figure 4). In all cases, stoichiometric amounts of H2/O2=2:1 were observed after an induction period, in which carbonaceous species were removed from the surface. The O2 production rates as a function of the Cu loading are shown in Figure 4. At low Cu loadings up to about 0.2 wt % Cu, the reaction rate is constant. At higher Cu concentrations, the reaction rate increases, and the optimum Cu loading for water splitting is reached at about 0.66–1 wt % Cu corresponding to the optimum loading for methanol reforming. Furthermore, the catalyst that was prepared by photodeposition of Cu prior to the photodeposition of Cr (first Cu, then Cr) was slightly less active than the sample that was prepared by photodepositing Cr followed by photodeposition of Cu (first Cr, then Cu). Additionally, the results indicate that simultaneous photodeposition of Cu and Cr results in lower photocatalytic activity and that low loadings of Cr (0.09 wt %) are sufficient to drive the overall water splitting reaction.

Figure 4.

Pure water splitting with Cu+Cr-Ga2O3 and Cu+Mo-Ga2O3: activity as a function of the Cu loading. Cr-Ga2O3 was fully inactive.

Selected catalysts were additionally investigated with respect to gas-phase water splitting under well-defined static conditions. Therefore, Cu leaching due to photoxidation and also shadowing effects can be neglected, which may be influenced by the stirring conditions in liquid-phase water splitting. The constant H2 and O2 contents obtained in the gas-phase water-splitting experiments with a constant gas-phase water content of 2.3 % are summarized in Table 1. For all Ga2O3 samples modified with Cu or Cu+Cr gas-phase water splitting was found to occur, whereas for the unmodified bare Ga2O3 sample only a small amount of evolving H2 was detectable. The results summarized in Table 1 suggest that the water-splitting activity is independent of the Cu content in the absence of Cr. However, the rates are very low and highly dependent on the H2O concentration, and the deviation of the H2/O2 ratio from 2:1 points to photodegradation.9 In the presence of chromia higher Cu loadings strongly enhance the overall water-splitting activity of the materials in agreement with the liquid-phase results. For all samples the H2/O2 ratio was almost 2:1, and the highest water-splitting activity was observed for the catalyst 0.09 wt % Cr/0.4 wt % Cu/Ga2O3. Finally, using the gas-phase reactor it was shown for this catalyst that the H2 and O2 evolution rates depend on the gas-phase water content in agreement with the results obtained by Chorkendorff and co-workers16 by varying the water content from 0.3 to 2.3 % (Supporting Information, Figure SI6).

Table 1. Hydrogen and oxygen production rates in gas-phase water splitting with constant water content of 2.3 % at quasi steady state.

Catalyst

H2 production rate [nmol h−1]

O2 production rate [nmol h−1]

0.09 wt % Cr/0.4 wt % Cu

9600

4700

0.4 wt % Cu

1600

670

0.09 wt % Cr/0.025 wt % Cu

4270

1470

0.025 wt % Cu

1600

450

Independent of the reaction conditions, either in the gas or in the liquid phase a high water-splitting rate is only observed when a small amount of chromia is present on Cu-Ga2O3, although Cr2O3–Ga2O3 is not able to split water. In the presence of Cr2O3 the water oxidation rate is catalyzed efficiently, and, as a result, increasing the Cu loading leads to an increase in the amount of water reduction sites and therefore to an increase in the overall water splitting rate. This increase occurs relatively steeply starting at about 0.4 wt % Cu indicating that a certain minimum size of the CuOx nanoparticles is beneficial.

In case of the Cu+Cr system prepared by co-impregnation, Domen and co-workers6 claim that Cu and Cr are present as a CuII-CrIII mixed metal oxide on the photocatalyst. When photodepositing Cr2O3 from K2CrO4 they discuss that an amorphous Cr2O3 shell on the CuOx core might be permeable to protons and H2, but not to O2, thus suppressing the back reaction 2 H2+O2→2 H2O.6 Our results indicate that the formation of Cr2O3 shells on the CuOx cores is unlikely, because depositing Cr first and then Cu resulted in a more active water-splitting catalyst compared to depositing Cu first and then Cr (Figure 4), and the activity does not scale with the Cr/Cu ratio. Catalysts that contain a low ratio of Cr/Cu were found to be as active as catalysts that contain a high Cr/Cu ratio, and it was not possible to increase the activity of the material further by adding Cr. In case of a shell–core structure one certainly would expect to increase the activity by covering the Cu core, which is unlikely as the Cr loading of 0.09 wt % corresponds to just 1.04 atoms per nm2 based on the specific Ga2O3 surface area of 10 m2 g−1 in comparison with about 7–8 Ga atoms per nm2 exposed on the surface of β-Ga2O3(100).18 Furthermore, it can be estimated, assuming an average Cu particle size of 2 nm for the catalyst with 0.4 wt % Cu (Figure 3), that the hypothetical Cr coverage of these Cu particles would amount to only half a monolayer. Instead, the formation of oxidic Cu and Cr nanoparticles in close proximity during photodeposition is considered likely, which is in agreement with the TEM and XPS data.

A significant rate of water oxidation yielding O2 especially at higher concentrations of Cu was only observed after adding chromia, whereas the CuOx–Ga2O3 catalyst was able to generate appreciable amounts of H2 also in the absence of chromia using methanol as sacrificial reagent. These observations imply that chromia is directly involved in the OER, providing active sites for water oxidation on Cu-Ga2O3 similar to RuO2.17 To probe this conjecture, MoOx was photodeposited as OER catalyst using MoO42− prior to Cu photodeposition as MoOx is known to have similar oxidizing properties, and indeed water-splitting activity was observed (Figure 4).

As many photocatalytic studies are based on a combinatorial approach using a standardized deposition of the co-catalysts, the question has to be addressed whether this approach is indeed promising. Our results clearly indicate that the interaction between the n-type semiconductor Ga2O3 and the p-type semiconductor CuOx has to be optimized by stepwise photodeposition. The CuOx particles are assumed to act as electron traps allowing for fast charge separation. Furthermore, additional OER sites are needed in close proximity to the CuOx particles to collect the photogenerated holes needed for simultaneous water oxidation. Therefore, the whole system consisting of the semiconductor with HER and OER sites has to be tuned carefully to obtain highly active water splitting catalysts.

In conclusion, the photocatalytic activity of β-Ga2O3 for overall water splitting is found to depend strongly on the presence of the co-catalysts CuOx and chromia, which were efficiently stepwise photodeposited using Cu2+ and CrO42− as precursors and methanol as sacrificial reagent. The water-splitting activity can be tuned by varying the Cu loading in the range 0.025–1.5 wt %, whereas the Cr loading is not affecting the rate as long as small amounts (e.g., 0.05 wt %) are present. Chromia enhances the water reduction rate over CuOx–Ga2O3 in the presence of methanol, and is essential for the oxidation of water. This insight is considered vital for a further knowledge-based improvement of photocatalysts applied in water splitting.

Experimental Section

Water-splitting and photodeposition experiments were performed in a home-made quartz reactor using UV light as described recently.12 Subsequent to the photodeposition of the co-catalysts, the samples were filtered and freeze-dried. Water-splitting experiments in the gas-phase were performed in a home-made fully metal sealed photoreactor, which is described in detail elsewhere.19 Further experimental details are included in the Supporting Information.

Acknowledgements

This work was co-funded by the German federal state North Rhine-Westphalia (NRW) and the European Union (European Regional Development Fund: Investing In Your Future) and by the Mercator Research Center Ruhr. Fruitful discussions with the H2ECO2 partners and Israel Wachs are gratefully acknowledged.