DOI: 10.1002/cssc.201500383

Full Papers

Time- and Energy-Efficient Solution Combustion Synthesis of Binary Metal Tungstate Nanoparticles with Enhanced Photocatalytic Activity Abegayl Thomas,[a] Csaba Jan‚ky,*[b, c] Gergely F. Samu,[b, c] Muhammad N. Huda,[d] Pranab Sarker,[d] J. Ping Liu,[d] Vuong van Nguyen,[d] Evelyn H. Wang,[e] Kevin A. Schug,[e] and Krishnan Rajeshwar*[a, f] In the search for stable and efficient photocatalysts beyond TiO2, the tungsten-based oxide semiconductors silver tungstate (Ag2WO4), copper tungstate (CuWO4), and zinc tungstate (ZnWO4) were prepared using solution combustion synthesis (SCS). The tungsten precursor’s influence on the product was of particular relevance to this study, and the most significant effects are highlighted. Each sample’s photocatalytic activity towards methyl orange degradation was studied and benchmarked against their respective commercial oxide sample ob-

tained by solid-state ceramic synthesis. Based on the results herein, we conclude that SCS is a time- and energy-efficient method to synthesize crystalline binary tungstate nanomaterials even without additional excessive heat treatment. As many of these photocatalysts possess excellent photocatalytic activity, the discussed synthetic strategy may open sustainable materials chemistry avenues to solar energy conversion and environmental remediation.

Introduction The global need to focus on renewable and earth-abundant resources has resparked much attention on inorganic oxide semiconductor materials for solar energy conversion[1] and photo[a] A. Thomas, Prof. Dr. K. Rajeshwar Department of Chemistry & Biochemistry University of Texas at Arlington Arlington, TX 76019 (USA) E-mail: [email protected] [b] Prof. Dr. C. Jan‚ky, G. F. Samu Department of Physical Chemistry and Materials Science University of Szeged Szeged, Rerrich Sq. 1, 6720 (Hungary) E-mail: [email protected] [c] Prof. Dr. C. Jan‚ky, G. F. Samu MTA-SZTE “Lendìlet” Photoelectrochemistry Research Group Szeged, Rerrich Sq. 1, 6720 (Hungary) [d] Prof. Dr. M. N. Huda, P. Sarker, Prof. Dr. J. P. Liu, Dr. V. van Nguyen Department of Physics University of Texas at Arlington Arlington, TX 76019 (USA) [e] E. H. Wang, Prof. K. A. Schug Department of Chemistry & Biochemistry University of Texas at Arlington Arlington, TX 76019 (USA) [f] Prof. Dr. K. Rajeshwar Center for Renewable Energy Science & Technology University of Texas at Arlington Arlington, TX 76019 (USA) Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cssc.201500383. This publication is part of a Special Issue on “Green Chemistry and the Environment”. To view the complete issue, visit: http://onlinelibrary.wiley.com/doi/10.1002/cssc.v8.10/issuetoc

ChemSusChem 2015, 8, 1652 – 1663

catalytic environmental remediation.[2, 3] Titanium dioxide (TiO2) has been the most extensively studied oxide due to its excellent photoelectrochemical and photocatalytic behavior.[4] However, TiO2 has a relatively wide band gap (3.0–3.2 eV), thus restricting light absorption to the UV region of the solar spectrum that comprises only ~ 4 % of the solar spectrum. On the other hand, this material exhibits little or no absorption in the visible region (comprising ~ 50 % of the solar spectrum). Aside from TiO2, tungsten trioxide (WO3) has been widely studied, serving as an attractive material for both water oxidation and environmental remediation.[5] Tungsten trioxide shares characteristics similar to TiO2 in terms of chemical inertness and photo- and chemical stability in aqueous solutions in a relatively broad pH range. Importantly, WO3 has a band gap energy (Eg) of 2.7 eV (vs. 3.2 eV for anatase TiO2), with its absorption edge located just at the cusp of the visible light spectrum. The high valence band edge energy (VB) of + 3.1 eV makes WO3 a suitable photocatalyst for water photo-oxidation; however, its conduction band (CB) is more positive than that necessary for spontaneous solar hydrogen evolution.[6] Driven by the above considerations and a recognition of the critical need to discover families of new materials, the search for oxide semiconductors besides TiO2 and WO3 has gained considerable momentum in recent years. Appreciable efforts have been devoted to the shifting of the VB and CB edge positions of different compound semiconductors (“bandgap engineering”) to tailor their interfacial energetics to targeted photooxidation or photoreduction processes, respectively.[1, 6] This can be done by doping/alloying or synthesizing composites, in many cases using TiO2 or WO3 as one component (Scheme 1).

1652

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers

Scheme 1. Approximate CB and VB edge positions of the synthesized binary tungstates. The interfacial energetics for TiO2 and WO3 are also shown for comparison.

ic.[21] See Figure 1 A and B for model structure representations of each of the discussed binary tungstates. On the other hand, introducing a monovalent cation such as Ag favors a complicated network structure of silver tungstate (Ag2WO4), which can exhibit three different structural phases; a-, b-, and g-Ag2WO4.[22] Among those polymorphs a-Ag2WO4 is thermodynamically most stable[22] and belongs to orthorhombic symmetry Pn2n.[22, 23] All tungsten atoms are six-coordinate and form WO6 octahedra. These WO6, W2O6, and W3O6 octahedra are connected by sharing edges and grouped together at a particular position (see Figure 1 C). On the other hand, the number of different sites occupied by the silver atoms in Ag2WO4 is six, which form different polyhedra.[23] Literature survey indicates that synthesis of ZnWO4, CuWO4, and Ag2WO4 has been dominated by conventional methods such as hydrothermal synthesis or solid-state reaction.[13, 18, 24–30] These methods suffer from similar drawbacks, most importantly the long reaction time (usually several hours) and the need

A shift of the band edge positions can alter the energy band gap of these materials, which in turn may enhance photocatalytic activity. Such a shift may also shrink the band gap, thus increasing solar light absorption without changing the reduction/ oxidation potential of the photogenerated charge carriers. There Figure 1. Ball-and-stick model of the crystal structure for ZnWO (A), CuWO (B), and Ag WO (C). 4 4 2 4 are many examples where the photocatalytic or photoelectrochemical performance of binary for high energy input to maintain the elevated temperature (or even ternary) tungstates outperformed WO3,[6] notably, during the synthesis. We note here that minimization of both Bi2WO6,[7–9] SnWO4,[10, 11] ZnWO4,[12–14] AgBiW2O8,[15] and these factors is crucially important from a sustainability perAgInW2O8.[16, 17] spective, especially if we plan to utilize the materials in appliWe present below the strategy of incorporating Zn, Cu, and cations such as solar fuel generation or environmental remeAg cations into WO3 to form ZnWO4, CuWO4, and Ag2WO4, rediation. spectively, thereby opening up an avenue for tuning interfacial The process known as solution combustion synthesis (SCS) energetics. Introduction of a heteroatom in the WO3 structure addresses the above challenges admirably well; however, can result in the formation of one of two major crystal strucsomewhat surprisingly, its use has been quite limited in the tures: wolframite (smaller divalent cations), scheelite (larger diphotocatalysis community. In this method, reaction times are valent cations), or other possible crystal structures (monovalent short, no special equipment is needed, and the process is cost cations). Accordingly, small divalent cations (ionic radius effective.[31–33] The above-listed benefits are rooted in the exo< 0.77 æ) such as Zn and Cu tend to form wolframite structures [18] thermicity of the reaction, along with the expulsion of noxious Zinc tungstate (ZnWO4) crystallizes to (WO6 octahedra). gases, which together result in small-sized solid particles of the a monoclinic wolframite structure (P2/c), with both Zn and W synthesized products. SCS has been deployed to prepare oxide forming ZnO6 and WO6 octahedra connected by edge sharing. semiconductors, such as TiO2,[34, 35] ZnO,[36] Fe2O3,[37] and WO3.[5] Copper tungstate (CuWO4) crystallizes in a triclinic structure ¯ with symmetry P1. In the structure of CuWO4, both Cu and W Furthermore, the role of fuel was carefully studied, even as a synthetic strategy tool to obtain doped oxide as a result of are surrounded by six oxygen atoms to form CuO6 and WO6 the SCS process.[33, 37] On the other hand, much less is known octahedra.[19] These two different octahedra connected by about the role of precursors and their effect on the resultant edge sharing oxygen atoms, form infinite zigzag chains. Almaterials. Finally, to the best of our knowledge, there is no though CuWO4 is triclinic, its structure is topologically related precedence in the literature on the SCS of binary tungstates. to that of monoclinic wolframite (P2/c).[20] The CuO6 octahedra In this study, SCS was employed to prepare three members demonstrate Jahn–Teller distortion to remove the degeneracy of the binary tungstate family, namely ZnWO4, CuWO4, and of Cu2 + 3d orbitals. This distortion elongates the octahedron, Ag2WO4. The precursor influence on the morphological and causing reduction in the symmetry from monoclinic to triclinstructural attributes of the resultant materials was examined ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

1653

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers systematically by using two different tungsten precursors in the synthesis process. Moreover, the photocatalytic activity of each sample was evaluated for its ability to degrade an organic pollutant, namely, methyl orange dye. Apparent rate constants were compared and contrasted with those obtained for their commercial counterparts prepared by solid-state reaction. The major innovation of this study was the fact that the SCS-synthesized new families of binary tungstates, obtained within five minutes with very little energy input, outperformed their commercial counterparts (synthesized using time- and energyinefficient methods) in photocatalytic tests.

Results and Discussion Thermogravimetric analysis The use of thermal analysis can provide a better understanding of the SCS process. Hence, for the simulation of the SCS procedure, thermogravimetric analysis/differential scanning calorimetry (TGA/DSC) measurements were performed on each of the precursor mixtures. As an example, TGA/DSC profiles for the formation of CuWO4 from two different precursor mixtures are given in Figure 2 A and B. The tungstate samples prepared using both precursors showed multiple mass loss regimes. The initial mass loss between 100–150 8C corresponds to the loss of water from the precursor mixture in both cases. Using ammonium tungstate as precursor, all subsequent mass loss occurs in one single step above 300 8C, which originates from the combinations of fuel and metal-salt-precursor decomposition (and combustion). Note, however, that the above-mentioned temperature is only the temperature where combustion initiates; the actual temperatures after the combustion can be much higher than those indicated on the x-axis (given by the programmed-temperature ramp, see the Experimental Section). On the other hand, samples prepared using the sodium tungstate precursor had an expanded temperature range of 120– 210 8C, which was assigned to the decomposition of the fuel and the metal salt precursor. At temperatures over 480 8C an-

other mass loss regime was observed, which corresponds to the loss of residual carbon in the matrix. Quantitative assessment of the TGA curves was also performed. As the first step, the initial mass loss (related to water evaporation) was subtracted. Subsequently, the relative mass losses were determined for both mixtures and these values were compared with the theoretically calculated values. The good agreement between the actual and stoichiometric mass losses for both the sodium (34 % vs. 31 %) and the ammonium precursor (39 % vs. 41 %) confirmed that the proposed combustion reaction occurred in both cases, albeit with different intensity. DSC analysis yielded both endothermic and exothermic peaks. A fairly large and broad endothermic peak was witnessed at the early stage of analysis, which is assignable to the initial loss of moisture within the precursor material and correlates with the first mass loss on the corresponding TGA profile. More intense and sharp exothermic peaks were observed for the precursor mixtures containing (NH4)2WO4, compared to those with Na2WO4 (Figure 2 B), perhaps not surprisingly in light of its combustible nitrogen content. This exothermic peak suggests that a high combustion temperature is reached during the reaction, which also coincides with the oxidative decomposition of the fuel. Crystal structure and morphology Powder X-ray diffraction (XRD) was employed to characterize the crystalline phases of the synthesized and respective commercial oxide samples. The as-synthesized materials were annealed at various temperatures up to 500 8C (except for CuWO4, where the maximum temperature was 400 8C because of the limited thermal stability of this particular oxide) to enhance the degree of crystallinity and to remove any carbon residue and impurities that may be present. Figure 3 A–C illustrate the XRD patterns mapping the changes in the sample crystallinity from as-synthesized to the sample annealed at the highest temperature. The most vital message of these data is that most SCS samples [except ZnWO4 (Na) and Ag2WO4 (NH4)]

Figure 2. TGA (A) and DSC (B) profiles of the SCS precursor mixtures for CuWO4.

ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

1654

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers precursors show distinctly different purity and degree of crystallinity. Therefore, careful Rietveld refinement analysis was performed on the experimental data obtained at the highest annealing temperature [see the fitted XRD patterns (Figure S1) and more details in the Supporting Information]. Briefly, Rietveld analysis of the samples synthesized with the sodium tungstate precursor yielded single-phase structures in agreement with their respective JCPDS file. The materials prepared using the ammonium tungstate precursor, however, showed the presence of an additional minority WOx phase (Figure S1 A–F). Additionally, what is indeed surprising, the commercial samples (except silver tungstate) were not of a single phase and exhibited low crystallinity. The presence of additional peaks in the XRD pattern for the ammonium-derived samples may be a result of the use of a low pH value caused by the addition of HNO3. Recent literature supports this notion; in the synthesis of ZnWO4 using the hydrothermal process, WO3 formation was noted in the XRD pattern when the solution mixture was at a low pH range (pH < 4).[13, 29] Additionally, the more exothermic environment (and consequently a faster reaction) in the case of the ammonium precursor can be another rationale for the appearance of the minority WO3 phase. The average crystallite sizes were calculated for the synthesized samples using the most intense peaks by using the Scherrer equation (Table 1; more details in the Supporting in-

Table 1. Calculated average crystallite sizes of the various tungstate samples as a function of annealing and tungstate precursor. SCS sample

Crystallite size [nm] as-synthesized

annealed

ZnWO4 (Na) CuWO4 (Na) Ag2WO4 (Na)

1œ7 16 œ 1 22 œ 1 19 œ 1 10 œ 1 21 œ1 11 œ 2 2œ6 8œ2

32 œ 1 22 œ 1 33 œ 2 23 œ 1 16 œ 2 20 œ 1 21 œ10 14 œ 11 15 œ 3

ZnWO4 (NH4) CuWO4 (NH4) Ag2WO4 (NH4)

Figure 3. XRD patterns for A) ZnWO4, B) CuWO4, and C) Ag2WO4 samples. * represents the presence of WO3 in the synthesized sample.

are at least partially crystalline in nature, even without any additional heat treatment. However, the various oxides and more interestingly even the same oxide synthesized using different ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

ZnWO4 WO3¢x CuWO4 WO3¢x Ag2WO4 WO3¢x

formation). The first clear trend is that the synthesized particles are nanocrystalline in nature, and the results show an increase in crystallite size with increase in annealing temperature, as expected. Interestingly, the calculated crystallite size for the assynthesized ZnWO4 (Na) sample was as low as 1 nm. After annealing, all tungstate samples synthesized using the ammonium precursor showed a smaller particle size compared to their counterparts using the sodium-containing precursor. The smaller particle size appears to be associated with less single-phase material as a result of the higher reaction temperature and the larger amount of gas released during the synthesis. Note, however, that the Scherrer equation can only be applied to wellcrystallized samples; otherwise XRD peak broadening can reflect low crystallinity instead of nanosized entities. Therefore, the obtained crystallite sizes are considered as only estimates in these cases.

1655

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers TEM images of the tungstate samples were additionally obtained to compare any contrasting morphologies that may arise from the use of different precursors (Figure 4). In general, the particle sizes obtained from TEM were larger than the calculated crystallite sizes from XRD data (or in other words the size of the crystalline domains). As for CuWO4 (similarly for both precursors), the as-synthesized samples yielded particle sizes ranging from 15–50 nm with the size increasing with increasing annealing temperature. Figure 4 A shows samples with rounded edges signaling that they are at least partly in an amorphous phase. Closer inspection, however, revealed crystalline domains, where the lattice fringes were visible (corresponding to the [011] lattice plane),[38] in agreement with the previously presented XRD data (Figure 4 I). In contrast, the as-synthesized ZnWO4 (Na) was completely amorphous with agglomerated small-sized nanoparticles. For the annealed ZnWO4 samples, however, the observable sharp edges of much larger particles were seen, indicating improved crystallinity consistent with the XRD patterns. The most interesting morphological feature was noted for the ZnWO4 (NH4) samples (Figure 4 E and F), where the sample showed partial crystallinity and well-defined hexagonal nanorod morphology even without heat treatment, again corroborating the previously shown XRD patterns. Note, that this type of morphology was observed for ZnWO4 before but as a result of a lengthy hydrothermal synthetic procedure.[13, 14, 26] High-resolution (HR)TEM images confirmed that ZnWO4 nanorods preferentially grow along the [100] direction. Interplanar spacings were determined to be 0.575 and 0.468 nm, corresponding to the [010] and [100] lattice ChemSusChem 2015, 8, 1652 – 1663

Figure 4. TEM images of A, B) CuWO4 (Na) as-synthesized and annealed at 400 8C, respectively; C, D) ZnWO4 (Na) as-synthesized and annealed at 500 8C, respectively; E, F) ZnWO4 (NH4) as-synthesized and annealed at 500 8C, respectively; G, H) Ag2WO4 (NH4) as-synthesized and annealed at 500 8C. I, J) HRTEM images of the nanoparticles in panels A and E.

www.chemsuschem.org

1656

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers planes, respectively, which is entirely consistent with earlier reports.[13, 39] Besides the rod-shaped crystals, some sphericalshaped nanoparticles can be also seen in the images for the as-synthesized samples, whereas strictly rod-shaped crystals (lengths 100–150 nm) are present after annealing. Ag2WO4 (Na) appeared to form much smaller particles compared to the other cases with sizes in the 4–10 nm range, with many of the particles covered by a thin carbon film (as residue from the SCS precursor). Notably, such carbon coating can be indeed useful for certain applications, such as in Li-ion batteries, an alternative application area of these binary tungstates.[39, 40] Upon subsequent heat treatment, this carbon matrix is removed and large aggregates of the small nanoparticles can be observed. Overall, we can conclude that a variety of morphologies can be achieved upon using different precursors and subsequent heat treatment, tailoring the morphology of these oxides towards a targeted application. Further structural analysis of the synthesized tungstates was performed using Raman spectroscopy. Using group theory, different studies have reported the existence of as many as 18 Raman-active vibrational modes for ZnWO4 and CuWO4.[41–43] Experimental results show that 10–11 of the Raman active vibrations were observed for both oxides within the frequency range of the experiment (see Table S1). Of the ten Ramanactive bands observed, 5–6 internal stretching modes were present due to the W¢O bonds in the WO6 octahedra of the metal tungstate structure. Importantly, assignment of the Raman bands (see Figure 5 and Figures S2 and 3, and Table S1) confirmed the formation of the respective compound structures for all samples. In addition, trends very similar to the sample crystallinity are seen here and similar to those concluded from XRD (see the Supporting Information). The positions of the most intense band in each sample corresponding to the W¢O stretching mode were 907, 899, and 883 cm¢1 for ZnWO4, CuWO4, and Ag2WO4, respectively. The 805 cm¢1 vibrational mode characteristic of WO3 appeared in all samples prepared using ammonium tungstate, thus confirming the multiphase structure.[44] Optical properties

Figure 5. Raman spectra of ZnWO4 (Na), CuWO4 (Na), and Ag2WO4 (Na) samples.

Figure 6. Tauc plots for the different ZnWO4 samples.

optical measurements yielded direct bandgap values in the range 3.24–3.89 eV (depending on the synthesis and heat treatment), all were within the range reported in previous studies (see Table 2). CuWO4 and Ag2WO4 samples showed indirect bandgap values in the 1.80–2.04 and 2.44–3.10 eV ranges, again in reasonable accord with the literature-reported values. It is worth noting the very sizeable spread in literature

UV/Vis diffuse reflectance spectroscopy was employed to estimate the band gap values of the synthesized oxides. This was achieved by generating Tauc plots, namely, a plot of the Kubelka–Munk function versus photon energy (ahn0.5 vs. hn).[45] Figure 6 contains an example for ZnWO4. Table 2. Optical properties of the synthesized samples prepared using different tungstate precursors. Table 2 displays the experimental Sample Band gap [eV] Calculated band gap Reported values band gap values for the syntheas-synthesized annealed at highest temperature values [eV] [eV] sized powders along with the direct band gap calculated optical properties (see 3.89 œ 0.02 3.24 œ 0.03 ZnWO4 (Na) below). The resulting tungstate 2.92 3.02–5.85[25, 26, 46] ZnWO4 (NH4) 3.76 œ 0.02 3.45 œ 0.03 products were observed to be indirect band gap white, yellow-brown, and gray in 2.04 œ 0.01 2.03 œ 0.01 CuWO4 (Na) 2.10 1.78–2.79[21, 46–48] CuWO4 (NH4) 1.91 œ 0.01 1.80 œ 0.01 color for ZnWO4, CuWO4, and 3.10 œ 0.01 – Ag2WO4 (Na) Ag2WO4, respectively (see Fig1.22 3.06–3.55[23, 27, 28] Ag2WO4 (NH4) 2.44 œ 0.01 2.83 œ 0.02 ure S4). For ZnWO4, experimental ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

1657

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers

Figure 7. Electronic band structure for Ag2WO4 (A), CuWO4 (B), and ZnWO4 (C).

values, presumably reflecting the extreme sensitivity of the optical characteristics to synthesis and post-synthesis treatment details.

Electronic band structure calculations Computational studies were performed to rationalize the above optical properties, and to correlate them with the photocatalytic activity of these materials (discussed below). The electronic band structure for each of the tungstate materials were calculated along the special symmetry points in the Brillouin zone. For Ag2WO4, an indirect band gap value of 1.17 eV was found between X and G (Figure 7 A). Along the Y–S–X region, the mid-gap bands are degenerate. Moreover, the less dispersive nature of the bands in both VB and CB along this region suggests higher effective masses and, hence, lower conductivity for both electrons and holes. The computed band gap corresponds to optical transition due to electron transfer from O 2p at the top of the VB to W 5d/Ag 4d at the bottom of the mid-gap CB or from Ag 4d at the top of the VB to O 2p at the bottom of the mid CB. Both these values are smaller than the experimentally measured band gaps for as-synthesized Ag2WO4 (Na) and Ag2WO4 (NH4) samples. In Figure 7 B, an indirect band gap with a minimal value of 0.707 eV occurs along the Y!G region for CuWO4. The VB along the G!X regions is very flat, suggesting higher effective mass of holes. The CB is divided into two regions in which the first region contains only two bands that are mid-gap levels 0.707 eV higher than the VB and dispersive throughout all symmetry points. The bottom of the CB of the second region is about 3.087 eV higher from the VB maximum. They are flat along G!X and R!T but dispersive along the other symmetry points. The minimal optical band gap was calculated to be 2.1 eV. By comparing optical absorption and band structure, it can be concluded that the electron transition does not occur between the top or just below the top of the VB and bottom or just above the bottom of the CB, which are dominated by Cu 3d and W 5d, respectively, since the d–d transition is forbidden. Hence, the origin of the optical gap is attributed to elecChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

tron transfer from occupied O 2p states within the VB to the unoccupied Cu 3d mid bands (first part of CB). The DFT + U electronic band gap structure for ZnWO4 represents a direct band gap of 2.94 eV (very close to the calculated optical absorption) occurring at the Z point. In Figure 7 C, the VB is dispersive throughout all regions, indicating lower effective masses of holes. Similar dispersive features are also found for the CB along the regions except for D to B, which suggests higher effective masses of electrons, while other regions indicate lower effective masses of electrons. In addition, the presence of these lower effective masses of electron regions suggests an ease of transfer of electrons from the VB to the CB. The calculated optical band gap (2.92 eV) corresponds to the electronic band gap. Since the d–d electron transition is forbidden and the top of VB is largely contributed by O 2p, the first peak at 3.43 eV corresponds to electron transfer from O 2p at the top of VB to W 5d at the bottom of CB at the Z point. Photocatalytic activity Figure 8 A–C compares the photocatalytic activity profiles for methyl orange (MO) dye degradation for the blank case (no photocatalyst present), the commercial sample (obtained by high temperature ceramic synthesis), and the as-prepared nanoparticles (for both precursors). Clearly, the freshly synthesized nanoparticles showed good photocatalytic activity compared to the commercial samples under UV/Vis light irradiation. In fact, ZnWO4 (Na) exhibited the best photocatalytic activity, most likely owing to its pure, monoclinic single-phase structure and relatively small particle size as confirmed by Rietveld analysis of the XRD data and TEM images. After 20 min irradiation, both SCS ZnWO4 samples had degraded ~ 90 % of the MO dye whereas the commercial sample required 50 min. Additionally, the results show that both samples of CuWO4 and Ag2WO4 outperformed their respective commercial counterparts, with a striking difference for CuWO4. To confirm the reliability of UV/Vis spectrophotometry for monitoring the MO degradation, LC–MS was also employed to follow the photocatalytic reaction. As shown in the example of ZnWO4 (Na) in Figure 9, the ion peak area for MO (m/z = 304.07

1658

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers

Figure 8. MO photodegradation using the as-synthesized samples under UV/Vis light irradiation.

Figure 10. Comparison of the apparent rate constant values obtained from linear regression of MO photodegradation plots (Figure 8).

Figure 9. LC–MS monitoring of methyl orange photodegradation using the as-synthesized ZnWO4 (Na) sample under UV/Vis light irradiation

in negative ion mode) decreased significantly with increasing reaction time. More importantly, the degradation pattern is similar to that derived from UV/Vis spectrophotometry: the MO concentration reached zero within 40 min. Detailed mechanistic analysis of the degradation process is beyond the scope of this study. Nonetheless, major intermediates were detected using MS2, a tandem mass spectrometry technique for structural determination. An oxidative degradation pathway was identified similar to those presented for experiments using TiO2 earlier, involving demethylation and hydroxylation as intermediate steps.[49, 50] Linear regression was performed for the first 40 min of each photodegradation profile data (Table S2). The apparent rate constants compared in Figure 10 suggest that the SCS samples exhibited faster kinetics for MO dye degradation than their respective commercial samples, with the kapp values ordered thus: ZnWO4 > CuWO4 ~ Ag2WO4. The enhanced degradation rate (especially for ZnWO4) can be rooted in several factors such as optical, electronic, structural/morphological, and surface chemical properties. The impact of the above factors, however, is complex and convoluted. In the following sections an attempt to deconvolute these factors is presented by elaborating on the previously shown XRD and TEM data, as well as on photoelectrochemical and surface area measurements. ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

Any discussion of the physicochemical attributes of a given photocatalyst must start with its surface area [see Brunauer– Emmett–Teller (BET) surface area data in Table S3] as this parameter manifests most directly in photocatalysis by dictating the amount of substrate species that can be initially bound by adsorption. In evaluating surface area contributions, however, surface chemical properties also have to be taken into consideration because these factors strongly influence the adsorption process. As the first step, the kinetics of the adsorption was studied, and we found that the equilibrium was reached approximately within 40–50 min (see Figure S5). Therefore, a 60 min incubation period was employed before each photocatalysis (and adsorption) experiment. The amount of the adsorbed dye was determined for all tungstate samples (Figure S6). Considering the absolute values of the adsorbed dye amounts, most values fall in the same range (15–25 mmol g¢1), except for CuWO4 (commercial) and Ag2WO4 (NH4) for which much higher values were obtained (~ 65 mmol g¢1). The most important message from the adsorption studies is that dye adsorption is not the rate-limiting step in our experiments. If that were the case, the CuWO4 (commercial) and Ag2WO4 (NH4) samples should have outperformed their counterparts because of the vastly larger adsorbed dye amounts. But this was clearly not the case; in fact, CuWO4 (commercial) showed the worst photocatalytic activity among all studied samples (Figure 8).

1659

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers

Figure 11. A) Photovoltammogram of a CuWO4 (NH4) sample recorded between 0.1 and 1.1 V in 0.1 m Na2SO4, at a sweep rate of 2 mV s¢1 using a 300 W Hg– Xe arc lamp. B) Comparison of the onset potential for the three tungstate samples in 0.1 m Na2SO3.

These observations suggest that various factors other than surface area (and dye adsorption) play a role in the increased photocatalytic activity of the SCS-synthesized tungstate nanoparticles. To reveal possibly existing differences in the surface chemistry, surface area-normalized values of the adsorbed dye amounts are also given in Figure S6. By using this approach we can filter out the effect of higher surface area, and the adsorbed amounts reflect the affinity of MO to the studied tungstate surfaces only. Importantly, this representation reveals that the two anomalously high dye loadings in the case of CuWO4 (commercial) and Ag2WO4 (NH4) stemmed solely from a surface-area effect because the normalized values are in good agreement with those obtained for their counterparts. It is also important that the normalized values are similar for each oxide sample, independently from the precursor. If we compare the different tungstates, a similar trend can be seen to that shown in Figure 10 (ZnWO4 > CuWO4 ~ Ag2WO4), indicating that surface chemistry may play a role in the observed PC performance. In solid-state science, crystallinity is well accepted to improve charge transport dynamics and, therefore, this parameter should be of importance in dictating the photocatalytic activity. However, this factor only seems to be a minor contributor in our case. Note, that samples with vastly different crystallinity [e.g., ZnWO4 (Na) and ZnWO4 (NH4)] show very similar performance. We may assume that the previously mentioned give-and-take relationship between phase purity versus crystallinity plays a key role even here. Similarly, the multiphasic composition (together with differences in the surface chemistry, see Figure S6) of the commercial sample may well be the reason for the inferior performance of these materials. Differences among the oxidative photocatalytic activity of the different metal tungstates can be understood by comparing their electronic properties, specifically their VB edge posiChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

tions. In this vein, linear sweep photovoltammograms were recorded for the various tungstate samples. This voltammetry technique has been described elsewhere;[44] briefly, it employs a slow potential scan while irradiation of the electrode is periodically interrupted. In this manner, both the dark and the light-induced photoresponse of the samples can be assessed in a single experiment. Interestingly, the best photoelectrochemical performance (i.e., highest photocurrents) was observed for CuWO4 (Figure 11 A). ZnWO4 generated stable but small photocurrents whereas Ag2WO4 was unstable under these conditions. Photovoltammograms, employing a narrower potential window spanning the open-circuit potential under illumination (Figure 11 B), were also recorded in sulfite ion-containing electrolyte in which sulfite acted as the hole scavenger (electron donor). The onset potential of these curves can be associated with the Fermi level of the various semiconductor samples, and thus their relative position can be correlated to the position of the CB edge.[51] The obtained values increase in the Zn, Ag, Cu-series (see the arrows in Figure 11 pointing to ¢0.2, ¢0.1, and + 0.05 V respectively). Using the above data and taking into account the determined band gap energies (also depicted in Scheme 1), we may assume that the high photocatalytic activity for ZnWO4 (compared to its Ag- and Cu-containing counterparts) can be attributed to the more positive VB edge position and thus the higher oxidation power of the photogenerated holes. In contrast, despite the excellent photoelectrochemical properties (Figure 11 A) of CuWO4, it exhibits rather inferior photocatalytic behavior for our test reaction because of the less positive VB edge position.

Conclusions In this study, binary tungstate nanoparticles of ZnWO4, CuWO4, and Ag2WO4 were synthesized by solution combustion synthe-

1660

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers sis (SCS) using two different precursors. Structural analysis showed a pure monophasic structure forming for the samples synthesized using the Na2WO4 precursor whereas a biphasic structure was observed for samples when using (NH4)2WO4. Most as-synthesized oxide samples were crystalline in nature; however, the degree of crystallinity improved upon subsequent heat treatment. TEM images confirmed the formation of various nanocrystalline structures for which the observed morphology strongly depended on the employed precursor molecule. The optical band gap energies (as determined by diffuse reflectance UV/Vis spectroscopy) were in good agreement with previous literature reports and our own computational data. The as-prepared tungstate materials showed good photocatalytic activity (relatively high apparent rate constants) for the photodegradation of methyl orange (MO) dye. Taking into account the specific surface area as well as the adsorbed dye amount, it was revealed that factors other than surface area [e.g., surface chemical properties, phase purity, and valence band (VB) edge position] contributed to the increased photocatalytic activity of the materials. These conclusions were further supported by DFT (+ U) calculations and photoelectrochemical measurements. This latter technique (together with the energy band gap values) allowed comparison of both the conduction band (CB) and VB band edge positions of the tungstate samples. Overall, we found that surface chemical properties and the VB edge position were the two dominant contributors to superior photocatalytic activity of SCS ZnWO4 compared to the other studied materials. This study, in a broader vein, also illustrates the utility of inserting different transition metal ions into various structural frameworks (e.g., WO3) and thus tuning their optical and optoelectronic attributes for targeted applications.[6] As demonstrated in this study, SCS is an extremely simple and versatile approach for this purpose. In this direction, further studies on the synthesis of various narrow band gap binary oxide semiconductors (e.g., delafossites) for solar energy conversion and environmental remediation applications may be profitable; such studies are in progress.

Experimental Section Materials Zinc nitrate hexahydrate—Zn(NO3)2·6 H2O (Alfa Aesar), copper(II) nitrate hemihydrate - Cu(NO3)2·2.5 H2O (Alfa Aesar), and silver nitrate—AgNO3 (Fisher) were used as the zinc, copper, and silver precursors, respectively. Sodium tungsten oxide dihydrate— Na2WO4·2 H2O (Alfa Aesar) and ammonium tungsten oxide— (NH4)2WO4 (Alfa Aesar) were used as the tungsten precursors and urea as the fuel. All chemicals were used without further purification. Double-distilled water (Corning Megapure) was used to prepare all solutions. Commercial samples of ZnWO4, CuWO4, and Ag2WO4 (all Alfa Aesar) were used as reference materials for benchmarking the characteristics of the combustion-synthesized nanopowders. The synthesized samples are summarized in Table 3. ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

Table 3. Synthesized samples prepared from different tungsten precursors. Sample

Metal precursor

ZnWO4 (Na) CuWO4 (Na) Ag2WO4 (Na) ZnWO4 (NH4) CuWO4 (NH4) Ag2WO4 (NH4)

Zn(NO3)2·6 H2O Cu(NO3)2·2.5 H2O AgNO3 Zn(NO3)2·6 H2O Cu(NO3)2·2.5 H2O AgNO3

Tungsten precursor

Fuel

Na2WO4·2 H2O urea (NH4)2WO4

Synthesis of tungstates Stoichiometric amounts of the respective metal, the sodium tungstate precursor, and fuel were placed in a crucible and homogenized in double distilled water. The solution mixture was then transferred to a preheated furnace at 350 8C and left for 5 min. This allows for dehydration of the precursor mixture and the promotion of spontaneous combustion. As for ammonium tungstate, it was first homogenized in 1.5 m HNO3 medium and then placed in a crucible with the metal precursor and fuel solution. The solution was placed in a preheated furnace at 350 8C and left for 5 min. Once the SCS process was completed, the samples were removed, finely grinded in a mortar and pestle, and thermally annealed at temperatures 400 8C, 450 8C, and 500 8C for 30 min. Thereafter, each sample (both the as-synthesized and the heat treated) was washed with double-distilled water to remove any soluble residue from the precursor species, filtered, and dried in an oven at 100 8C.

Physical characterization To model the SCS procedure, thermogravimetric analysis/differential scanning calorimetry (TGA/DSC) analyses on the precursor mixtures were carried out using a TA Instruments model Q600 instrument. The precursor mixtures were placed in an alumina crucible in air atmosphere with a flow rate of 100 mL min¢1 and at a heating rate of 10 8C min¢1 up to 1000 8C. Powder X-ray diffraction (XRD) measurements were performed within the angle range 2q = 108– 808 using a Rigaku Ultima IV instrument with a CuKa radiation source (l = 1.5406). Rietveld refinements were carried out on MDI Jade 8 software, using initial model structures from reported literature. Refinements were also performed on each of the samples with reduced scale and background parameters and individual FWHM (full width half maximum) curves to account for all the peaks in the XRD pattern. The peak shapes were fitted using the Gaussian profile function with a displacement selection used allowing for any deviations from the model structure. UV-vis diffuse reflectance spectra were collected on a PerkinElmer Lambda 35 spectrophotometer equipped with an integrating sphere. High-resolution transmission electron microscopy (HR–TEM) was performed on a Hitachi H-9500 instrument at various magnifications. Raman spectra were obtained from a Horiba Jobin Yvon Labram Aramis spectrometer at an excitation wavelength of 633 nm, using a He– Ne laser by averaging 64 spectra. For BET surface area, a Quantachrome NOVA 3000 adsorption test instrument was utilized for the measurements.

Computational details The present calculations were performed within the framework of the standard frozen-core projector augmented-wave (PAW) method[52, 53] using DFT as implemented in Vienna ab initio simula-

1661

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers tion package (VASP 4.6) code.[54, 55] In the PAW method, a nonlinear core-correction is not necessary because it is an all-electron-like method. Exchange and correlation potentials were treated in the generalized gradient approximation (GGA) as parameterized by Perdew-Burke-Ernzerhof (PBE).[56, 57] The PBE functional does not contain any empirically optimized parameters, and hence works better on a wide range of elements. It is well known that underestimation of electron localization is a major failure of standard DFT calculations, in particular, for systems with localized d and f electrons.[58–60] This failure manifests in the general trend of DFT to underestimate the band gap and to produce incorrect solutions for some 3d based metal oxides. In order to correct this shortcoming, in our calculations we employed an on-site Coulomb correlation through the Hubbard-based U correction parameter.[60–62] In the present work, we used U = 7 eV externally providing Coulomb correlation to only Cu 3d and Zn 3d orbitals. Hence, all results presented here are based on DFT + U calculations for both ZnWO4 and CuWO4 and DFT calculation for Ag2WO4. The basis sets were expanded with plane-waves with a kinetic energy cut-off of 400 eV, and the BZ integrations were performed using the second-order Methfessel-Paxton method.[63] The optimization of ZnWO4, CuWO4, and Ag2WO4, was done using 11 Õ 9 Õ 11, 5 Õ 9 Õ 9, and 5 Õ 5 Õ 9 Monkhorst-Pack k-point sampling.[64] However, more refined 9 Õ 13 Õ 13, 15 Õ 13 Õ 15, and 7 Õ 7 Õ 11, k-point samplings were used for ZnWO4, CuWO4, and Ag2WO4 respectively for optical property calculations. For visualization of the crystal structures, VESTA (Visualization for Electronic and Structural Analysis) was used.[65, 66] Calculations were performed at the High Performance Computing (HPC) Center at the University of Texas at Arlington.

Photocatalysis experiments The oxidative photocatalytic activity of the tungstate nanoparticles was evaluated by their ability to degrade methyl orange (MO) as a probe dye in aqueous solution. A photocatalytic reactor consisting of an outer cylindrical glass vessel (400–700 nm transmittance) and an interchangeable quartz/glass inner vessel was used for the photocatalysis process. The light source, a 400 W medium pressure Hg arc (incident photon flux estimated by potassium ferrioxalate actinometry was 5.21 Õ 10¢4 Einstein per minute), was placed in the inner vessel, which was also equipped for water circulation aimed at maintaining a constant reactor temperature, as well as filtering infrared radiation. The outer reaction vessel was equipped with a sample collection port, in addition to a gas purge connection and a stirrer to ensure dispersion of the mixture and to prevent the catalyst from sedimentation. Testing of the photocatalytic activity was carried out using a 250 mL MO (50 mm) reaction solution along with the photocatalyst (dose: 2 g L¢1) while being purged with O2 gas. The reaction suspension was initially stirred in the dark for 60 min to attain adsorption equilibrium of MO on the catalyst surface. After 60 min, the solution was irradiated with UV-visible light. Sample aliquots of 5 mL were periodically removed from the outer vessel every 10 min with the suspension being centrifuged and filtered (0.45 mm PVDF membrane filter) to remove the photocatalyst powder.

Detection of degradation The extent of MO degradation was assessed by measuring the absorbance of the solution using an Agilent 8453 UV-visible spectrophotometer. Blank runs were performed under identical experimental conditions without the photocatalyst. ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

The photocatalytic process was also monitored by LC-MS analyses, performed on a Shimadzu LCMS-ITTOF (ion trap—time-of-flight) mass spectrometer (Shimadzu Scientific Instruments, Inc., Columbia MD).[67] Samples were taken from the reaction vessel at different times (t = 0, 10, 20, 30, 40, 50, 60, 70, and 80 min) during the reaction and were analyzed to determine the degradation of methyl orange and the formation of intermediates and products. 30 mL of sample was injected onto a reverse-phase column (Raptor C18, 2.1 Õ 100 mm; 2.7 mm, Restek Corporation, Bellefonte, PA, USA) and binary pumps carrying mobile phase of 10 mm ammonium acetate or 0.1 % formic acid and acetonitrile with a flow rate of 0.3 mL min¢1 were used to obtain separation for scouting purposes. A 10 min gradient was developed with 5–99 % acetonitrile followed by the washing and equilibrating steps. The eluent entered an electrospray ionization (ESI) interface, which was operated in either positive or negative ionization mode. Interface voltages for positive or negative ionization modes were + 4.50 kV or ¢4.00 kV, respectively. Mass range was set to record from 100–500 m/z. The curved desolvation transfer line and heat block temperatures were both set at 200 8C. Nebulizing gas (nitrogen) was flowed at 1.5 L min¢1, and the detector voltage was set at 1.64 kV. MS2 was performed to identify the major intermediates formed during the photocatalytic reaction. Precursor ion isolation width was set at 3.0000 Da with 10 ms ion accumulation time. Collision energy was at 27 % and collision gas was at 50 % with a frequency of 45.0 kHz.

Photoelectrochemical measurements All photoelectrochemical measurements were performed on an Autolab PGSTAT302 instrument, in a classical one-compartment, three-electrode electrochemical cell. Various tungstate nanoparticles were spray coated on ITO glass electrodes (ca. 0.1 mg cm¢2) and were used as working electrodes. A large Pt foil counter-electrode and an Ag/AgCl/3 m KCl reference electrode completed the cell setup. The light source was a 300 W Hg-Xe arc lamp (Hamamatsu L8251). The radiation source was placed 2 cm away from the working electrode surface. Photovoltammetry profiles were recorded in both 0.1 m Na2SO3 and 0.1 m Na2SO4 electrolyte, using a slow potential sweep (2 mV s¢1) in conjunction with interrupted irradiation (0.1 Hz) on the semiconductor coated electrodes. All experiments were performed at the laboratory ambient temperature (20 œ 2 8C).

Acknowledgements The authors thank Dr. Jiechao Jiang, facility manager of the UTA Characterization Center for Materials and Biology, for his technical assistance with the HR-TEM images. KR thanks the National Science Foundation (CHE-1303803) for partial funding support. CJ thanks the Hungarian Academy of Sciences for financial support through the Momentum Excellence Program. The work of MNH was supported partially by National Science Foundation (CBET1133672). E.H.W. and K.A.S. thank Restek Corporation for research support. Finally, we thank the two anonymous reviewers for constructive criticisms of an earlier manuscript version. Keywords: bandgap engineering · combustion synthesis · photocatalysis · semiconductors · solar energy

1662

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers [1] R. M. Navarro Yerga, M. C. Alvarez Galv‚n, F. del Valle, J. a Villoria de La Mano, J. L. G. Fierro, ChemSusChem 2009, 2, 471 – 485. [2] K. Rajeshwar, J. Phys. Chem. Lett. 2011, 2, 1301 – 1309. [3] K. Rajeshwar, A. Thomas, C. Jan‚ky, J. Phys. Chem. Lett. 2015, 6, 139 – 147. [4] A. Linsebigler, G. Lu, J. Yates, Chem. Rev. 1995, 95, 735 – 758. [5] W. Morales, M. Cason, O. Aina, N. R. de Tacconi, K. Rajeshwar, J. Am. Chem. Soc. 2008, 130, 6318 – 6319. [6] C. Jan‚ky, K. Rajeshwar, N. R. de Tacconi, W. Chanmanee, M. N. Huda, Catal. Today 2013, 199, 53 – 64. [7] J. Tian, Y. Sang, G. Yu, H. Jiang, X. Mu, H. Liu, Adv. Mater. 2013, 25, 5075 – 5080. [8] S. Sun, W. Wang, L. Zhang, E. Gao, D. Jiang, Y. Sun, Y. Xie, ChemSusChem 2013, 6, 1873 – 1877. [9] L. Zhang, D. Bahnemann, ChemSusChem 2013, 6, 283 – 290. [10] I.-S. Cho, C. H. Kwak, D. W. Kim, S. Lee, K. S. Hong, J. Phys. Chem. C 2009, 113, 10647 – 10653. [11] J. Ungelenk, C. Feldmann, Chem. Commun. 2012, 48, 7838 – 7840. [12] H. Fu, J. Lin, L. Zhang, Y. Zhu, Appl. Catal. A 2006, 306, 58 – 67. [13] R. Shi, Y. Wang, D. Li, J. Xu, Y. Zhu, Appl. Catal. B 2010, 100, 173 – 178. [14] C. Zhang, H. Zhang, K. Zhang, X. Li, Q. Leng, C. Hu, ACS Appl. Mater. Interfaces 2014, 6, 14423 – 14432. [15] N. R. de Tacconi, H. K. Timmaji, W. Chanmanee, M. N. Huda, P. Sarker, C. Jan‚ky, K. Rajeshwar, ChemPhysChem 2012, 13, 2945 – 2955. [16] J. Tang, Z. Zou, J. Ye, J. Phys. Chem. B 2003, 107, 14265 – 14269. [17] S. Song, Y. Zhang, Y. Xing, C. Wang, J. Feng, W. Shi, G. Zheng, H. Zhang, Adv. Funct. Mater. 2008, 18, 2328 – 2334. [18] D. W. Kim, I.-S. Cho, S. S. Shin, S. Lee, T. H. Noh, D. H. Kim, H. S. Jung, K. S. Hong, J. Solid State Chem. 2011, 184, 2103 – 2107. [19] L. Kihlborg, E. Gebert, Acta Crystallogr. Sect. B 1970, 26, 1020 – 1026. [20] J. Forsyth, C. Wilkinson, A. Zvyagin, J. Phys. Condens. Matter 1991, 3, 8433 – 8440. [21] J. E. Yourey, B. M. Bartlett, J. Mater. Chem. 2011, 21, 7651 – 7660. [22] A. Van den Berg, C. Juffermans, J. Appl. Crystallogr. 1982, 15, 114 – 116. [23] L. Cavalcante, M. Almeida, W. Avansi, R. Tranquilin, E. Longo, N. Batista, V. Mastelaro, S. Li, Inorg. Chem. 2012, 51, 10675 – 10687. [24] G. B. Kumar, K. Sivaiah, S. Buddhudu, Ceram. Int. 2010, 36, 199 – 202. [25] G. Huang, C. Zhang, Y. Zhu, J. Alloys Compd. 2007, 432, 269 – 276. [26] P. Siriwong, T. Thongtem, A. Phuruangrat, S. Thongtem, CrystEngComm 2011, 13, 1564 – 1569. [27] J. Tang, J. Ye, J. Mater. Chem. 2005, 15, 4246 – 4251. [28] E. Longo, D. Volanti, V. Longo, L. Garcia, I. Nogueira, M. A. P. Almeida, A. Pinheiro, M. Ferrer, L. S. Cavalcante, J. Andres, J. Phys. Chem. C 2014, 118, 1229 – 1239. [29] M. Hojamberdiev, G. Zhu, Y. Xu, Mater. Res. Bull. 2010, 45, 1934 – 1940. [30] R. Zhang, H. Cui, X. Yang, H. Liu, H. Tang, Y. Li, Micro Nano Lett. 2012, 7, 1285 – 1288. [31] A. G. Merzhanov, J. Mater. Chem. 2004, 14, 1779 – 1786. [32] K. C. Patil, M. S. Hegde, T. Rattan, S. T. Aruna in Chemistry of Nanocrystalline Oxide Materials: Combustion Synthesis Properties and Applications, World Scientific Publishing, Singapore, 2008. [33] K. Rajeshwar, N. R. de Tacconi, Chem. Soc. Rev. 2009, 38, 1984 – 1998. [34] K. Nagaveni, M. S. Hegde, N. Ravishankar, G. N. Subbanna, G. Madras, Langmuir 2004, 20, 2900 – 2907.

ChemSusChem 2015, 8, 1652 – 1663

www.chemsuschem.org

[35] G. Sivalingam, K. Nagaveni, M. S. Hegde, G. Madras, Appl. Catal. B 2003, 45, 23 – 38. [36] S. Kikkawa, S. Hosokawa, H. Ogawa, J. Am. Ceram. Soc. 2005, 88, 308 – 311. [37] K. Deshpande, A. Mukasyan, A. Varma, Chem. Mater. 2004, 16, 4896 – 4904. [38] C.-L. Li, Z.-W. Fu, Electrochim. Acta 2008, 53, 4293 – 4301. [39] H. Shim, I. Cho, K. Hong, A.-H. Lim, D.-W. Kim, J. Phys. Chem. C 2011, 115, 16228 – 16233. [40] H.-W. Shim, A.-H. Lim, G.-H. Lee, H.-C. Jung, D.-W. Kim, Nanoscale Res. Lett. 2012, 7, 9 – 16. [41] Y. Liu, H. Wang, G. Chen, Y. D. Zhou, B. Y. Gu, B. Q. Hu, J. Appl. Phys. 1988, 64, 4651 – 4653. [42] E. Ross-Medgaarden, I. Wachs, J. Phys. Chem. C 2007, 111, 15089 – 15099. [43] T. T. Basiev, A. Y. Karasik, A. A. Sobol, D. S. Chunaev, V. E. Shukshin, Quantum Electron. 2011, 41, 370 – 372. [44] C. Jan‚ky, N. R. de Tacconi, W. Chanmanee, K. Rajeshwar, J. Phys. Chem. C 2012, 116, 4234 – 4242. [45] A. B. Murphy, Sol. Energy Mater. Sol. Cells 2007, 91, 1326 – 1337. [46] T. Montini, V. Gombac, A. Hameed, L. Felisari, G. Adami, P. Fornsaiero, Chem. Phys. Lett. 2010, 498, 113 – 119. [47] J. Ruiz-Fuertes, D. Errandonea, A. Segura, F. J. Manjon, Z. Zhu, C. Y. Tu, High Pressure Res. 2008, 28, 565 – 570. [48] S. J. Naik, A. V. Salker, Solid State Sci. 2010, 12, 2065 – 2072. [49] C. Baiocchi, M. C. Brussino, E. Pramauro, A. B. Prevot, L. Palmisano, G. Marc†, Int. J. Mass Spectrom. 2002, 214, 247 – 256. [50] K. Dai, H. Chen, T. Peng, D. Ke, H. Yi, Chemosphere 2007, 69, 1361 – 1367. [51] K. Rajeshwar in Electron Transf. Chem. (Ed.: V. Balzani), Wiley- VCH, Weinheim, 2001. [52] G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 11 – 19. [53] P. Blçchl, Phys. Rev. B 1994, 50, 17953 – 17979. [54] G. Kresse, J. Furthmìller, Phys. Rev. B 1996, 54, 11169 – 11186. [55] G. Kresse, J. Furthmìller, Comput. Mater. Sci. 1996, 6, 15 – 50. [56] J. Perdew, J. Chevary, S. Vosko, Phys. Rev. B 1992, 46, 6671 – 6687. [57] J. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865 – 3868. [58] F. Zhou, C. Marianetti, M. Cococcioni, D. Morgan, G. Ceder, Phys. Rev. B 2004, 69, 201101. [59] M. V. Ganduglia-Pirovano, A. Hofmann, J. Sauer, Surf. Sci. Rep. 2007, 62, 219 – 270. [60] V. Anisimov, J. Zaanen, O. Andersen, Phys. Rev. B 1991, 44, 943 – 954. [61] V. Anisimov, I. Solovyev, M. Korotin, M. Czyzyk, G. Sawatzky, Phys. Rev. B 1993, 48, 16929 – 16934. [62] I. Solovyev, P. Dederichs, V. Anisimov, Phys. Rev. B 1994, 50, 16861 – 16871. [63] M. Methfessel, A. T. Paxton, Phys. Rev. B 1989, 40, 3616 – 3621. [64] H. Monkhorst, J. Pack, Phys. Rev. B 1976, 13, 5188 – 5192. [65] K. Momma, F. Izumi, J. Appl. Crystallogr. 2008, 41, 653 – 658. [66] F. Izumi, K. Momma, Solid State Phenom. 2007, 130, 15 – 20. [67] J. S. Barnes, K. A. Schug, J. Agric. Food Chem. 2014, 62, 4322 – 4331.

Received: March 17, 2015

1663

Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim