M. Qamar*a,
M. O. Fawakhiryb,
A.-M. Azadc,
M. I. Ahmeda,
A. Khana and
T. A. Salehb
aCenter of Excellence in Nanotechnology (CENT), King Fahd University of Petroleum and Minerals, Dhahran 31261, Kingdom of Saudi Arabia. E-mail: qamar@kfupm.edu.sa
bDepartment of Chemistry, King Fahd University of Petroleum and Minerals, Dhahran 31261, Kingdom of Saudi Arabia
cAcense LLC, 8941 Dutton Drive, Twinsburg, OH 44087, USA
First published on 20th July 2016
Achieving selective transformation of organic functional groups in an energy efficient and environmentally benign way is an important yet challenging endeavour in the field of chemical science. Utilization of heterogeneous photocatalysis for selective conversion holds great potential from cost, energy, and environment viewpoints. In this work, we report the fabrication of nanocomposites comprising hypostoichiometric tungsten oxide (WO3−x) modified with platinum nanoparticles and/or reduced graphene oxide (RGO) and their deployment in highly selective (>99%) and efficient (>80%) conversion of aromatic alcohols into corresponding aldehydes under simulated sunlight and ambient conditions. Efficacy of the nanocomposites was investigated by studying the oxidation of benzyl alcohol (BA), 4-methoxybenzyl alcohol (4-MBA) and cinnamyl alcohol (CA) into corresponding aldehyde, in terms of alcohol concentration, RGO and/or Pt loading, and the amount of photocatalyst employed. Systematic investigations on the chemical stability, recyclability, photocatalytic and photoelectrocatalytic aspects helped to shed light on the role of excitons towards the observed selectivity. A plausible mechanism to explain the observed attributes of the photocatalyst is proposed that provides impetus to design future photocatalysts with anticipated attributes for selective oxidation reactions.
In this context, photocatalytic processes which utilize abundant sunlight, relatively inexpensive and stable photocatalysts, and water as a solvent, offer promising alternative. Though earlier studies exploiting photocatalysis have focused more on environmental cleanup, H2 production, and CO2 reduction etc.,4–8 its deployment for the synthesis of fine chemicals has also been explored.9–11 Thus, transformation from amine to imine,12,13 nitro to azo,14 aniline to azobenzene,15 hexane to hexanone and hexanol,16 and, alcohols to corresponding aldehydes have been well documented.17–21 However, semiconductor-mediated photocatalysis for selective oxidation is still in its infancy; this is true particularly in the case of alcohol oxidation in a green way. Some of the photocatalysts include CdS/graphene and CdS/graphene/TiO2 composites.22,23 Although selective and complete oxidation of benzyl alcohol to benzaldehyde was achieved in aqueous suspensions of Au/CeO2, the reaction rate of benzaldehyde formation was rather low (3.0 μmol h−1).24,25 Pristine titania in its various modifications (rutile,19,26 anatase,21 and brookite27) as well as that surface-modified with Nb2O5,28 Pt,29 and transition metals30 has been extensively investigated as a model photocatalyst. It was found that rutile titania demonstrated the highest selectivity for alcohol oxidation.17 Recently, chemoselective oxidation of alcohols using monolayers of niobic acid (HNb3O8) in the form of 2-D nanosheets under visible light irradiation, was achieved.31 As stated above, however, the selectivity and conversion still remain low and the use of organic solvents was required to achieve higher selectivity.
We have recently reported rather highly selective and efficient conversion of aromatic alcohols into corresponding aldehydes in aqueous suspensions of Pt/Bi2WO6 (ref. 32) and Ag3PO4.33 This was achieved possibly by more decisive roles played by the modification of the conduction band potential of these photocatalysts and more active reduction sites in water. It is believed that a facile oxidation of alcohols tends to diminish the generation of ˙OH radicals. As these radicals are short-lived and hence highly reactive, they can attack the organic moiety leading the formation of various by-products. The restricted formation of such radicals and their limited concentration presumably contributed towards high selectivity of alcohol to aldehyde conversion. One of the intrinsic prerequisites of the photocatalysts for high selectivity appeared to be a combination of high oxidation potential of valence band holes and low reduction potential of conduction band electrons. In the present work, we employed WO3 as the photocatalyst; it has a rather small band gap (∼2.8 eV), a deeper valence band (+3.1 eV) and positive conduction band (+0.3 eV).34,35 Moreover, effective adsorption of solar spectrum, favorable physico-chemical properties and resilience to photocorrosion make it a promising candidate for visible-light-driven photocatalysis. It is demonstrated that hypostoichiometric as well as Pt-modified WO3 nanoparticles could be utilized as efficient and recyclable photocatalyst for selective conversion of aromatic alcohols, such as benzyl alcohol (BA), 4-methoxybenzyl alcohol (4-MBA) and cinnamyl alcohol (CA), into corresponding aldehydes. Furthermore, conversion was carried out in water under simulated sunlight at ambient conditions. Effect of various parameters such as alcohol concentration, RGO, Pt and catalyst loading was investigated. Systematic investigations on the chemical stability, recyclability and, photocatalytic and photoelectrocatalytic aspects helped to shed light on the role of excitons towards the observed selectivity. A plausible mechanism in terms of inhibition of formation of radicals (˙OH, O2−˙) is proposed.
The PEC activity of the materials was evaluated in a 3-electrode cell assembly connected to a potentiostat using sodium sulphate solution (0.1 M) as the electrolyte, saturated calomel electrode as the reference and coiled platinum wire as the counter electrode. The working electrode was prepared from a homogeneous suspension of Nafion® and WO3-based catalyst in ethanol made by sonication and deposited on an ITO substrate. A xenon lamp with a cut-off filter was employed to obtain radiations of λ > 420 nm.
The morphological and microstructural features of the tungsten oxide made in this work are shown in Fig. 2. As can be discerned, the compound consists of particles of spherical, rod-like and platelet shapes with size in the range of 40 to 70 nm. The HR-TEM image (2C) depicts assembly of large number of individual particles stacked together, while the SAED pattern (2D) is indicative of high degree of crystallinity. The interplanar distance was computed to be 0.36 nm, which is in good agreement with the 〈200〉 reflection in the monoclinic phase of WO3.
![]() | ||
Fig. 2 The FE-SEM (A) TEM (B) and (C) HR-TEM images of as-synthesized WO3 compound; the SAED pattern is shown in (D). |
It is well known that color of its oxide differ, depending on the oxidation state of tungsten in them.36 The color changes are a result of loss of oxygen which generates additional valence states in the WO3 parent structure, either W5+, W4+ or a mixture of W6+ and W5+ in various ratio depending on the extent and manner of oxygen loss.37,38 Consequently, the cation-to-cation charge transfer between the parent hexavalent tungsten (W6+) and its reduced (W(6−x)+) state, together with concomitant loss of oxygen is responsible for the color change.37 For example, intermediate compounds such as WO2.90 (dark blue) or WO2.72 (violet) have been reported to form upon reduction.36 In the present case, however, darkening of color could be associated with the possibility of formation of WO3 that is slightly hypostoichiometric with respect to oxygen. Thus, it could be represented by the general formula WOx (3 < x < 2). Ample evidence of this speculation was provided by TG, Raman and XPS investigations, as would be discussed later.
In order to elucidate the existence of oxygen nonstoichiometry in the tungsten oxide prepared in this work, a programmed thermogravimetric run was conducted simultaneously in a Netzsch STA 449F3 TG/DSC instrument (Fig. 3).
![]() | ||
Fig. 3 Cyclic thermogravimetry profile of the as-prepared tungsten oxide in air (glow rate: 25 Sccm min−1; heating and cooling rate: 10° min−1). |
The compound was heated from room temperature (25 °C) to 900°, cooled to 100° and heated again to 900 °C in air (flow rate: 10 Sccm min−1) at a heating/cooling rate of 10° min−1. The color of the sample before (left) and after (right) the run is also shown. Substantial weight loss up to 400 °C in the first heating cycle could be ascribed to a combined loss of surface-adsorbed water as well as structural water (from tungstic acid, WO3·H2O). The weight change during the remainder of the heating and cooling profile is insignificantly small and could be attributed to buoyancy error.
Had the presumed oxygen uptake by the nonstoichiometric oxide occurred, the thermogram would have shown concomitant weight increase while heating in oxygen rich environment. Moreover, the sample color would have changed to yellow, had the compound been fully oxidized to WO3. On the other hand, as seen from Fig. 3, the post oxygen-annealed sample appears highly agglomerated and grey in color. It can, therefore, be concluded that the compound synthesized in this work by hydrothermal process was a stable tungsten blue oxide (WO2.9) of stable stoichiometry.
The X-ray photoelectron spectra for W and O in the as-prepared WO3 are compared with those of commercial power in Fig. 4. Distinct differences in the two cases confirm corresponding changes in oxidation states of tungsten signifying the stated departure from the stoichiometric chemistry for WO3. For example, for pure WO3 (Fig. 4A), the binding energy peaks at 35.78 and 37.91 eV belong to the 4f7/2 and 4f5/2 spectral lines, respectively, are in good agreement with those for tungsten in 6+ oxidation state.39,40 In the case of blue WO3, the 4f7/2 and 4f5/2 spectra are slightly shifted (Fig. 4C); they appear at 35.92 and 38.26 eV, together with an additional peak at 37.7 eV. A similar trend is observed in the case of spectra for oxygen, where XPS O 1s peak consists of two components. The slightly different binding energy in the two cases indicates the presence of oxygen in somewhat different environments, as a result of the difference in the oxidation state of tungsten. One at a binding energy of 530.8 eV is assigned to the oxygen in WO3 and the second rather small peak at 532.2 eV (Fig. 4B) could be assigned to water molecules adsorbed on the sample surface. In the case of blue WO3, the intensity of smaller peak is substantially attenuated with a concomitant shift in the peak position to lower binding energy (531.5 eV; Fig. 4D). When taken in combination, both tungsten and oxygen spectra reinforce the speculation that the blue tungsten oxide is indeed an oxygen-deficient compound. This is further corroborated by the details of systematic XPS investigation correlating different oxidation states of tungsten with binding energy and oxygen vacancy documented elsewhere.36,41,42 The existence of WO bonding with tungsten in 5+ oxidation state was also further confirmed by comparing the Raman spectra of commercial WO3 and tungsten blue oxide in Fig. S2;† they are very similar except in terms of the signal size and, the appearance of an additional peak at ∼950 cm−1, arising from lattice distortion leading to the formation of W (with 5+ oxidation state)
O bond owing to oxygen vanacy.43
![]() | ||
Fig. 4 Comparative XPS spectra of W and O in commercial WO3 powder (A and B) and as-prepared nanoscale WO3 powder (C and D). |
The results of UV-visible diffuse reflectance spectroscopic study on the absorption characteristics of WO3 are shown in Fig. 5. The absorption threshold for both samples extends well into visible region indicating they are capable of absorbing visible light. However, the spectra shown in Fig. 5A and C, reveal stronger absorption in the visible regime by WO3 made in this work as compared to those of commercial WO3. The corresponding band gap derived from the extrapolation of Kubelka–Munk unit vs. photon energy plots in Fig. 5B and D were found to be approximately 2.8 and 2.9 eV, respectively.
![]() | ||
Fig. 5 Diffuse reflectance spectroscopy (DRS) signatures of commercial WO3 (A and B) and as-prepared WO3 (C and D). |
For the investigation on the photocatalytic propensity of the tungsten blue oxide towards selective photo transformations, aromatic alcohols were chosen as representative moieties. All the reactions were carried out in water at room temperature and ambient pressure under simulated sunlight. The photoreactor were properly operated so as to preclude the escape of reaction product(s) during reaction. For quantification, calibration curves with known concentrations of alcohol as well as the aldehyde were generated with a series of their aqueous solutions (0.125, 0.25, 0.5, 1.0 and 1.5 mmol L−1) by injecting them in a GC-MS unit and validating them by analyzing representative samples through ultra-performance liquid chromatography (UPLC). The selective photocatalytic behavior and activity of tungsten oxide samples were investigated by studying the oxidation of benzyl alcohol into benzaldehyde over a period of up to 12 h under specified experimental conditions. Approximately 2% conversion was observed with commercial WO3, while the conversion was about 6% with tungsten blue oxide; the selectivity toward benzaldehyde formation, however, was very high (>99%) in both the cases and remained unchanged. With all other parameters remaining identical, the difference in the photocatalytic activity could be attributed to (1) blue WO3 is presumably endowed with some additional Lewis centers which could induce secondary dark-stage reactions and/or higher adsorption of alcohols thereby increasing the overall photocatalytic activity, and (2) the higher light absorption by the tungsten blue oxide, caused by additional sub-bandgap introduced by non-stoichiometricity, in the visible regime. As a result, its photocatalytic efficiency was higher than that of the commercial WO3 powder. Since hypostoichiometric WO3 showed better alcohol oxidation, it was employed in subsequent photocatalytic experiments. However, even though it showed better alcohol oxidation characteristics, the conversion was still modest (∼6%) and not attractive for commercial adaptation. Such low photocatalytic efficiency is usually rationalized in terms of undesired electron–hole pair recombination. Theoretically, this recombination process is more prominent in those systems whose reduction potential is more positive than that of molecular oxygen. This makes the conduction band electrons more amenable to recombination with the holes. One of the most accepted and successful strategy to mitigate the recombination tendency and perk up the process efficiency, is to create subenergy bands in the vicinity of Fermi level by introducing Pt nanoparticles and/or RGO as electron mediators. Exploiting this methodology, Pt nanoparticles were photodeposited onto the surface of hypostoichiometric WO3. As seen from Fig. S3A and S3B,† the Pt nanoparticles are spherical and ranged between 2 and 5 nm in diameter with rather homogeneous distribution. The FT-IR and Raman spectra substantiate the conversion of graphite into RGO. The comparative IR spectra (Fig. S3C†) and Raman (Fig. S3D†) indicate graphene oxide is quantitatively transformed to RGO. The peak at 1600 cm−1 attributed to CC stretching is the main characteristic peak in both RGO and graphene oxide, while the symmetric stretching vibration at 3420 cm−1 is due to –OH group in adsorbed water on them. The C–O stretching vibrations (1300–1000 cm−1) are more prominent in graphene oxide than in RGO indicating the reduction of the former into RGO.44 From Raman spectrum, the number of layers can be determined using the intensity ratio of the 2D and G peaks (I2D/IG) which appear at 2690 cm−1 and 1580 cm−1. The I2D/IG ∼ 2, 2 > I2D/IG > 1, and I2D/IG < 1, corresponds to monolayer, bilayer, and multilayers RGO, respectively.44 In the present case, the ratio of G to 2D peaks was computed to be 2.05, indicating the formation of monolayer RGO.
A representative time-dependent variation in concentration of benzyl alcohol and benzaldehyde in the presence of Pt/WO3 under radiation is shown in Fig. 6A. After ∼12 h of exposure to simulated visible light, the yield of benzaldehyde on the basis of mole-to-mole conversion was approximately 80% with >99% selectivity. Formation of other products beside benzaldehyde, if any, was below the quantification threshold. Greater than 13 fold improvement in BA oxidation was observed with Pt/WO3 while the improvement was less than 2 fold in the case of RGO/WO3. The performance and selectivity of these photocatalysts in the case of 4-methoxybenzyl alcohol (4-MBA) and cinnamyl alcohol (CA) was also investigated and the results are compared in Fig. 6B. As can be seen, each formulation (viz., Pt/WO3, RGO/WO3 and Pt/RGO/WO3) was effective in the oxidation of higher alcohols as well with selectivity (on mole-to-mole conversion basis) exceeding 99%. The formulation Pt/RGO/WO3 showed the highest conversion efficiency for all the alcohols with greater than 75% oxidation efficiency (∼90% in the case of cinnamyl alcohol) and >99% selectivity. These results are summarized in Table S1† and compared with those reported for other photocatalysts.
One of the shortfalls of selective oxidation by photocatalysis is the overoxidation of reactants/products as the reaction progresses. This is attributed to the presence of highly reactive but radicals, such as ˙OH, O2−˙ or HO2˙, and their subsequent complexation either with the substrate (photocatalyst) itself or with the reactants or products. With the platinized tungsten blue oxide, however, very selective oxidation of benzyl alcohol into benzaldehyde was observed even after 12 h of irradiation. Recently,45 simultaneous oxidation of benzyl alcohol (reactant) and benzaldehyde (product) on titania-based photocatalyst has been attributed to the preferential complexation of benzaldehyde with the TiO2 surface. On the other hand, modification of TiO2 surface with WO3 selectively precluded aldehyde overoxidation. Thus, in order to investigate if a similar surface complexation phenomenon is operative in the case of pristine and platinized–tungsten oxide surfaces as well, adsorption studies were conducted by stirring the aqueous suspensions of each photocatalyst with benzyl alcohol, benzaldehyde and their 1:
1 (v/v) mixture, for 24 h in dark. The catalyst was isolated from solution and FTIR and Raman spectroscopy was carried out to detect any surface bound alcohol or aldehyde on the catalyst while their quantification was done by GC-MS. None of these measurements hinted the presence of trace of alcohol or aldehyde on the catalyst surface. This led to conclude that WO3 surface is passive towards surface complexation of benzyl alcohol or benzaldehyde. Controlled experiments under identical experimental conditions without catalyst or radiation did not show any variation in the alcohol concentration, indicating that both light and catalyst were required to initiate and sustain the desired oxidation.
The observed selectivity is rationalized in terms of radical generation during irradiation step and their subsequent utilization in the photocatalytic step. As stated above, the non-selectivity of a photocatalytic process stems from the formation and increase in the concentration of radicals such as ˙OH, O2−˙, HO2˙, etc. The O2−˙ radicals are formed by the facile reduction of molecular oxygen by excited electrons in the conduction band while the holes in the valence band can oxidize water and generate the ˙OH radicals. These radicals, though short lived, are extremely reactive, and readily interfere in the regular scheme of a chemical process, making the photocatalytic process somewhat non-selective. We have previously shown that water oxidation does not occur in the presence of alcohol; alcohol is preferentially valence band hole-mediated oxidized thereby prohibiting the formation of ˙OH radicals.32,33 On the other hand, the presence of O2 may affect the overall photocatalytic process in one of the two ways: as an oxidant (electron acceptor) or through direct incorporation.
To examine the first possibility (formation of O2−˙ via electron capture from conduction band) oxidation of alcohol was carried out on Pt/WO3, with and without molecular O2; later was achieved by bubbling N2 through the photoreactor. As readily seen from Fig. 7A, the conversion was not affected by the nature of ambient environment, indicating that the role of O2 was passive in the oxidation process. It may, therefore, be concluded that: (a) O2−˙ radicals are not formed, and (b) alcohol oxidation on the Pt/WO3 surface is essentially mediated by the valence band holes, rather than through direct interaction with molecular oxygen. Furthermore, the inability of tungsten blue oxide to reduce O2 to O2−˙ could be due to the lower (more negative) redox potential of O2/O2−˙ reaction (−0.16 eV, vs. NHE at pH = 7)46,47 than that of WO3 (+0.3 eV vs. NHE at pH = 7).34,35 The results summarized in Fig. 7 further substantiate this observation. The anodic polarization profiles for pure WO3 and Pt/WO3 in the presence of molecular oxygen with and without light (Fig. 7B) show a nearly 2-fold improvement in the photoelectrocurrent density in the case of platinized WO3 over its pure counterpart. This could be attributed to the interfacial (metal–semiconductor interface) charge-transfer process, facilitated by Pt nanoparticles through The inhibition/suppression of electron–hole pair recombination. Such improvement in the photocatalytic activity of metalized semiconductors has been discussed in detail elsewhere.4,40,41 As is also evident from Fig. 7B, no photocurrent generated in the dark.
Since the photoelectrodes may show a certain degree of current drift over a time scales of 5–10 min and hence creates ambiguity between photocurrents under illumination and currents under dark. Fig. 7C and D show the variation of photocurrent in the case of pure WO3 and Pt/WO3, respectively, in the dark and under illumination by switching the light off and on for ∼10 min at a constant applied load of 1.0 V, in the presence and absence of oxygen in the reaction mixture. The results are self-explanatory. For example, the anodic photoelectrocurrent is comparable in both the cases, and unaffected by the presence or absence of molecular O2, corroborating the observation that WO3 (pure or platinized) is incapable of reducing O2 to produce O2−˙. Also, the absence of substantially higher photocurrent in the case of Pt/WO3 in the presence of O2 eliminates the possibility of its surface reduction. Thus it seems a rational postulation that the increased photocurrent in the case of Pt/WO3 is caused by proton reduction, leading to H2 evolution at the cathode. In the light of these observations, a plausible mechanism involving electron and proton transfer reactions in the conduction band and holes in the valence band that governs the alcohol to aldehyde oxidation via photochemical reaction of the type discussed in this paper. At energy equal to or greater than the band gap of a semiconductor photocatalyst, electrons are excited from the valence band (vb) to the conduction band (cb), creating a hole (h+) in the valence band. Under conditions of charge separation, the exciton (electrically neutral electron–hole pair) would move to the catalyst surface and participate in redox reactions with the absorbed species. Usually, in the case of allowed state of reduction potential, the conduction band electrons reduce the molecular O2 in the ambient into O2−˙ radicals. However, as shown above, direct reduction of O2 by the mobile valence band electrons (ecb−) on tungsten blue oxide or Pt/WO3 surface was not favored. Instead, the (ecb−) species presumably shuttle between the semiconducting oxide (WO3) and the metal (Pt) surface. This would, in principle cause electron accumulation and create an electron sink, thereby prompting H+ reduction and H2 evolution at the cathode. When both alcohol and water coexist, the (hvb+) species may either oxidize the: (a) surface bound H2O and produce OH˙ radicals, and/or (b) alcohol to aldehyde. Our studies indicate that option (b) is the preferred pathway.
In order to correlate extent of alcohol oxidation on its concentration, the catalyst loading, and the amount of Pt and/or RGO, its dependence on these factors was studied and the results are shown in Fig. 8. The oxidation shows monotonic increase with the concentration of benzyl alcohol up to 1.5 mM beyond which it decreases (Fig. 8A). It would appear that the number of active sites on the photocatalyst surface is sufficiently large to oxidize benzyl alcohol up to 1.5 mM, beyond this conversion decreases. The decrease in conversion at higher concentration may presumably be due to the increased amount of alcohol or formed aldehyde molecules occupying and/or blocking all the active catalytic sites present on the surface of the photocatalyst and lead to decrease in the conversion.48 Since the presence of Pt nanoparticles on the surface of WO3 improved the photocatalytic activity significantly, its amount would play a significant role in the optimization of oxidation. Dependence of alcohol conversion on the Pt loading on WO3 is presented in Fig. 8B. As could be seen, the highest conversion was obtained with 1.5 wt% Pt. Platinization of tungsten oxide could have led to the creation of a Schottky barrier at the interface of metal and the semiconducting oxide allowing seamless shuttling of electrons between the two.4,41,42 Decrease in the oxidation efficiency at higher Pt loading could be an artifact of substantial coverage of photocatalyst surface with Pt, thereby creating a shield and preventing the incident photons from impinging on the semiconducting oxide and triggering the photoelectrochemical reaction.36 The parabolic conversion curve seen in Fig. 8C is characteristic of a typical heterogeneous catalytic process. Fig. 8D shows the influence of RGO loading on the conversion; in this case WO3 modified with 0.5 wt% RGO caused optimal oxidation.
The stability and the recyclability of Pt/WO3 were examined under the identical experimental conditions as applied for the study of photocatalytic oxidation of BA. After the completion of each run, the catalyst was collected, washed, dried (at 300 °C under vacuum) and utilized for the next runs. Respective conversion of benzyl alcohol and formation of benzaldehyde as a function of different reaction cycles are shown in Fig. 9. As could be seen, the efficacy of the photocatalyst was maintained after several runs suggesting that Pt/WO3 is stable and recyclable photocatalyst for the selective oxidation of BA. Furthermore, the XRD patterns recorded after several runs did not indicate any change in crystal structure of Pt/WO3 indicating the chemical stability of the photocatalyst under applied experimental conditions.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11841k |
This journal is © The Royal Society of Chemistry 2016 |