Nanostructured Cu/TiO2 Photocatalysts for H2 ... - Wiley Online Library

DOI: 10.1002/cctc.201000289

Nanostructured Cu/TiO2 Photocatalysts for H2 Production from Ethanol and Glycerol Aqueous Solutions. Tiziano Montini,[a] Valentina Gombac,[a] Laura Sordelli,[b] Juan Jos Delgado,[c] Xiaowei Chen,[c] Gianpiero Adami,[a] and Paolo Fornasiero*[a] The sustainable development of the human society requires an increasing use of renewable raw materials and energy sources. In this context, photocatalysis represents a promising and necessary way to produce solar fuels and chemicals.[1–4] The photocatalytic hydrogen production of renewable oxygenated compounds from aqueous solutions could represent an important alternative[5, 6] to the more complex water splitting,[7–9] or to conventional thermal reforming processes.[10] Noble and base metals, including Pt, Au, Pd, Ni, Cu, and Ag, have been reported to be very efficient at increasing the production of H2 in TiO2 photocatalysis.[11] Increasing attention has been devoted to Cu2O and CuO as photocatalysts.[12, 13] Both copper oxides are abundant natural p-type semiconductors and are attractive owing to their virtual non-toxicity. Their bandgaps (2.1 eV for Cu2O and 1.2 eV for CuO) are suitable for photosplitting water to produce hydrogen by using visible light.[14] However, these materials have been shown to be unstable in electrolytic solutions owing to their facile photooxidation.[12] Siripala et al.[15] prepared a Cu2O/TiO2 heterojunction by electrodeposition of copper oxide; the device was shown to perform photoelectrolysis of water, the TiO2 layer providing protection against photocorrosion. Herein, a cheap and active photocatalysts based on Cu nanoparticles dispersed on TiO2 supports, which are capable of operating under solar radiation, are investigated for the production of hydrogen from ethanol and glycerol. Second generation ethanol and sugars, extracted from lignocellulosic parts of vegetables, and glycerol, produced as a by-product of bio-diesel, are attractive and largely available sacrificial agents.[3, 16] Two different TiO2 supports were prepared, the former from titanium isopropoxide (TiO2-SG)[17] by a sol–gel method, the latter from titanyl sulphate (TiO2-PS) by precipitation,[18] followed in both cases by calcination at 450 8C for 6 h. TiO2-SG (surface area 69 m2g 1) was composed of a mixture of polymorphs (Figure S1 in the Supporting Information). The analysis [a] Dr. T. Montini, Dr. V. Gombac, Prof. Dr. G. Adami, Prof. Dr. P. Fornasiero Department of Chemical and Pharmaceutical Sciences and ICCOM CNR Trieste Research Unit University of Trieste, via L. Giorgieri 1, I-34127, Trieste (Italy) Fax: (+ 39) 405583903 E-mail: [email protected] [b] Dr. L. Sordelli CNR-ISTM and IDECAT CNR Unit via Camillo Golgi 19, 20133 Milano (Italy) [c] Dr. J. J. Delgado, Dr. X. Chen Departamento de Ciencia de los Materiales e Ingeniera Metalﬄrgica y Qumica Inorgnica Universidad de Cdiz, E-11510 Puerto Real (Cdiz) (Spain) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cctc.201000289.

574

of its XRD pattern, following the work of Zhang et al.,[19] evidenced the presence of anatase (64 wt %), rutile (8 wt %) and brookite (28 wt %). Mean crystallite sizes of 11, 32, and 11 nm were calculated for anatase, rutile, and brookite, respectively. TiO2-PS possessed a slightly higher surface area (104 m2 g 1). Its powder XRD pattern showed the presence of a pure anatase phase (Figure S1 in the Supporting Information), with a mean crystallite size of 8 nm. The photodeposition of Cu on the surface of both TiO2 supports was performed by using UV/Vis irradiation (2 h) in the presence of copper nitrate and CH3OH as a hole scavenger (Figure S2). XRD analysis of the Cu/TiO2 nanocomposites did not allow the identification of Cu-related phases probably because of their low amount and/or high dispersion. Therefore, the Cu phases photodeposited on TiO2 supports were characterized by using X-ray absorption near-edge structure (XANES) or extended X-ray absorption fine structure (EXAFS) spectroscopy. Immediately after photodeposition, XANES spectra at the Cu K edge was in good agreement with that of Cu foil used as reference standard (Figure 1), thus indicating that copper is deposited in the form of zero-valent copper. The results of the analysis of the EXAFS spectra are summarized in Table 1.

Figure 1. Normalized XANES spectra at the Cu K-edge of the photodeposited samples (after photodeposition by irradiation in CH3OH/H2O solution for 2 h): a) Cu(1.0 %)/TiO2-SG, b) Cu(2.5 %)/TiO2-SG, c) Cu(1.0 %)/TiO2-PS, d) Cu(2.5 %)/TiO2-SG, e) Cu foil, f) Cu2O, and g) CuO.

The Cu Cu bond lengths are in good accordance with the value observed for metal Cu (0.2556 nm). Cu O contributions were also observed in the EXAFS signals of the samples with 1.0 wt % Cu loading. These contributions are likely to have originated from the interaction of the Cu atoms with the TiO2 support, as the Cu O bond lengths do not match those of standard materials, such as Cu2O (0.184 and 0.302 nm) and

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemCatChem 2011, 3, 574 – 577

Table 1. Results from EXAFS analysis at the Cu K-edge of photodeposited samples. Sample

Shell

N[a]

R [nm][b]

DM [nm][c]

Cu(1.0 %)/TiO2-SG

Cu Cu Cu Cu Cu Cu Cu Cu

2.1 5.1 – 7.0 3.1 3.8 – 10.5

0.2070 0.2576 – 0.2548 0.1984 0.2582 – 0.2852

1.2

Cu(2.5 %)/TiO2-SG Cu(1.0 %)/TiO2-PS Cu(2.5 %)/TiO2-PS

O Cu O Cu O Cu O Cu

1.5 1.0 4.0

[a] Average number of neighbors; [b] distance from the scattering atom; [c] mean diameter of the Cu nanoparticles, calculated assuming a cubooctahedral geometry.

which can entirely preserve the metal particle structure, with only a negligible reduction in size. To determinate the Cu particle size distribution and corroborate the EXASF results, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) was used in combination with X-ray energy dispersive spectrometry (XEDS) chemical analysis. Shown in Figure 2 are typical HAADF images of the Cu(2.5 %)/TiO2-SG and Cu(2.5 %)/TiO2-PS samples after exposure to the air. The average particle sizes were 1.8 and 4.3, respectively. To obtain reliable particle-size distributions of the reduced copper samples, that is, simulating the active state of the copper during photoreaction, the materials were also mildly pre-reduced (pure hydrogen at 70 8C) in a quasi in situ equipment and subsequently transferred to the TEM under anaerobic conditions. This approach prevented particle

CuO (0.195, 0.276, 0.290, and 0.309 nm). Assuming a cubooctahedral geometry, the average dimensions of the Cu nanoparticles were calculated,[20] evidencing a very high dispersion of the metal on the surface of TiO2. Cu nanoparticles of about 1 nm were observed for the samples with low metal loading. An increase in the amount of Cu, lead Figure 2. Representative HAADF images and copper particle size distribution of the a) Cu(2.5 %)/TiO -SG and 2 to a moderate increase in the b) Cu(2.5 %)/TiO2-PS. About 60 copper particles were measured for each sample. size of metal nanoparticles for the sample supported on TiO2re-oxidation under atmospheric pressure. The particle size disSG. Conversely, a significant increase in the particle size was tribution and representative STEM images are included in the observed for the samples prepared from TiO2-PS (Table 1). AsSupporting Information (Figure S4). In this case, the mean parsuming that the nucleation of Cu occurs by reduction of metal ticle size of the Cu(2.5 %)/TiO2 -SG and Cu(2.5 %)/TiO2 -PS samions by photogenerated electrons and that the subsequent growth of metal nanoparticles takes place around defects on ples were 1.6 and 4.2 nm, respectively. The TEM data were in the surface of the support,[21] the mean dimensions of Cu good agreement with the EXAFS results. The small differences nanoparticles would suggest that in TiO2-SG a higher number between the fresh and reduced samples can be attributed to of growth centers are actually present with respect to TiO2-PS. the reduction of the CuOx nanoparticles. Indeed, multi-phasic TiO2 materials show a lower rate for elecShown in Figure 3 is the H2 evolution from aqueous solution tron-hole recombination, as electrons and holes can be of ethanol and glycerol, both of which used a UV/Vis light irratrapped in different and, therefore, separated phases.[22–24] In diation or a Solar Simulator. Other by-products were detected this way, the lifetime of surface defects is increased, favoring a in the gas stream: CH4, CO2, ethylene, and acetaldehyde in the case of ethanol, CO2 in the case of glycerol (see the Supporting better dispersion of Cu nanoparticles on the surface of TiO2SG. Information, Figure S5–S10). After exposure to air, all the samples showed a rapid oxidaSamples prepared from TiO2-SG demonstrated better perfortion of Cu nanoparticles, as revealed by the change in color mance with respect to those prepared from TiO2-PS. This could from black to green and by an analysis of the XANES spectra be ascribed to the longer lifetime of the electron/hole pair of of the samples. After exposure to the air for 24 h (Figure S3) multiphasic TiO2 materials.[22–24] Comparing samples with the only Cu(2.5 %)/TiO2-PS still maintains a residual fraction of zerosame support, higher H2 productions were observed with a Cu valent Cu. Furthermore, EXAFS analysis revealed, for the samloading of 1.0 wt % under UV/Vis irradiation and in the presples supported on TiO2-SG, CuO-like species without crystalloence of ethanol, whereas for the glycerol solutions, a Cu loading of 2.5 wt % produced better results. graphic order, as supported by the absence of next nearest Cu Using irradiation produced by the Solar Simulator produced neighbor contributions. On the other hand, Cu(1.0 %)/TiO2-PS a lower photocatalytic activity than if the UV/Vis source was formed small CuO aggregates, whereas Cu(2.5 %)/TiO2-PS used. This might be ascribed to the lower contribution of the showed the same shell structure as that of a freshly irradiated TiO2 support, which adsorbs only UV photons, scarcely present sample. This latter behavior would suggest that around the metal core only an amorphous layer of oxide was formed, ( 4 %) in the Solar Simulator source. Therefore, these materiChemCatChem 2011, 3, 574 – 577


www.chemcatchem.org

575

Figure 3. H2 production rates on photodeposited Cu/TiO2 nanocomposites from ethanol or glycerol aqueous solutions by irradiation in the UV/Vis region (left side) or by Solar Simulator (right side): a) Cu(1.0 %)/TiO2-SG; b) Cu(2.5 %)/TiO2-SG; c) Cu(1.0 %)/TiO2-PS; d) Cu(2.5 %)/TiO2-PS. Please note the different time scales.

als possess an appreciable activity even using visible light. After an initial activation period, the H2 production is rather stable. Ethanol dehydrogenation resulted in the production of large quantities of acetaldehyde, the concentration of which was continuously increased in the gas stream (Figure S5), owing to the equilibration with the products in the liquid phase. Although no acetic acid or its esters are detected by GC/MS analysis (Figure S6), its intermediates could be suggested by the presence of equimolar amounts of CO2 and CH4 (Figure S5). The mechanism by which significant amounts of acetic acid was adsorbed on TiO2 polymorphs was recently discussed in details.[25] Alternatively, CH3CHO could decompose to CH4 and CO, followed by water-gas shift reaction (WGSR) promoted by light.[26] As CH3CHO accumulates in solution, it is evident that its transformation is slower than its formation. As a consequence, owing to the presence of both ethanol and surface acidic sites on the TiO2 support, the corresponding acetaldehyde diethoxy acetal was also formed. The same acidic sites might be considered responsible for the formation of ethylene by dehydration of ethanol (Figure S7). In the case of glycerol, only CO2 was detected in the gas stream (Figure S8), whereas in the liquid phase the main byproducts detected were 1,3-dihydroxypropanone and hydroxyacetaldehyde. Minor amounts of 2,3-dihydroxypropanal, 1,2ethanediol, hydroxyacetone, as well as two isomers of dimethyl-1,4-dioxane were also detected (Figure S9). Notably, no traces of either carboxylic acids or their ethyl esters were ever detected. The minor compounds derive from partial transformations occurring on the main products (Figure S10). For instance, conversion of 1,3-dihydroxypropanone into 1-hydroxy2-propanone was very limited, which lead to the eventual accumulation of 1,3-dihydroxypropanone in solution. This makes possible the sustainable production of 1,3-dihydroxypropanone, an interesting compound used for preparation of skincare products and also as a building block for the synthesis of

576

www.chemcatchem.org

biodegradable polymers.[27] On the other hand, formation of 2,3-dihydroxypropanal opens the way to the full degradation of glycerol to H2 and CO2 through a sequence of intermediates that include alcohols, aldehydes, and carboxylic acids. After the photocatalytic experiments, the resulting liquid phases were analyzed by using inductively coupled plasma atomic emission spectroscopy (ICP-AES) to check a possible leach of Cu during irradiation. If the UV/Vis irradiation source was used, very small amounts of Cu (2–3 % of total) were detected in the liquid phase in both cases. This result is consistent with previous investigations,[28] which indicated that some of the Cu could be oxidized and leached from the catalyst during the reaction. However, at the same time it can be rephotodeposited, owing to the constant presence of the UV photons of the TiO2 semiconductor and the sacrificial agent. Conversely, if Solar Simulator irradiation was used (Figure 3), the amount of Cu detected in the liquid phases was much higher (about 16 % and 9 % for the samples supported on TiO2-SG and TiO2-PS, respectively). These values are independent of both the sacrificial agent (ethanol or glycerol) and the experiment time (24 or 80 h), which suggests that an equilibrium between the Cu ions in solution and the Cu photodeposited on TiO2 was established. In this case, the fraction of UV photons is quite low ( 4 %), not sufficient to re-photodeposit Cu efficiently. Despite this, only minor deactivation was observed for all the experiments performed using the Solar Simulator (Figure 3). Preliminary experiments indicate that the amount of Cu present in solution can be minimized by pulsed UV radiation. To summarize, active and cheap nanocomposites can be produced by photodeposition of Cu on the surface of TiO2 supports. The activity of TiO2 is strongly influenced by its structural characteristics: if a multi-phasic support is used, the high number of defects (electron/hole pairs) produced results in a very high dispersion of the metal and in a higher activity for H2 production, by an increase in the oxidation rate of the sacrificial agents. Minor leaching is observed under UV irradiation, whereas solar-like irradiation causes a considerable fraction of the Cu to be oxidized and subjected to photocorrosion. After a few hours, the amount of Cu ions becomes constant, which suggests that an equilibrium between the Cu photodeposition process is established, induced by the small fraction of UV ( 4 %), and the leaching phenomena. Remarkably, the negative effect on the hydrogen production is not dramatic. A regeneration process, based on the use of short UV irradiation steps, can overcome progressive and serious deactivation. In addition to the production of hydrogen, this photocatalysis could open the way for the synthesis of building blocks of interest for the chemical industry, as the oxidation of carbonyl compounds is the limiting step of the degradation reaction, which occurs on the sacrificial agent.

Acknowledgements Dr. L. Olivi (Elettra Synchrotron, Trieste, Italy) is gratefully acknowledged for the assistance during EXAFS measurements. Pro-


ChemCatChem 2011, 3, 574 – 577

fessors M. Graziani and G. Pitacco (University of Trieste, Italy) are acknowledged for the helpful discussion. University of Trieste, ICCOM-CNR, PRIN2007 “Sustainable processes of 2nd generation for H2 production from renewable sources”, Fondo Trieste, Regione Friuli-Venezia Giulia, Fondazione CRTrieste and MICIINSpain/FEDER-EU (Project MAT2008-00889-NAN) are acknowledged for financial support. X. C. thanks the Spanish MCYT for funding through the “Ramn y Cajal” Program. Keywords: copper · hydrogen production · photocatalysis · renewable feedstock · titania [1] P. T. Anastas, M. M. Kirchhoff, Acc. Chem. Res. 2002, 35, 686 – 694. [2] C. Adams, Top. Catal. 2009, 52, 924 – 934. [3] S. Fernando, S. Adhikari, C. Chandrapal, N. Murali, Energy Fuels 2006, 20, 1727 – 1737. [4] C. Ampelli, G. Centi, R. Passalacqua, S. Perathoner, Energy Environ. Sci. 2010, 3, 292 – 301. [5] D. I. Kondarides, A. Patsoura, X. E. Verykios, J. Adv. Oxid. Technol. 2010, 13, 116 – 123. [6] D. I. Kondarides, V. M. Daskalaki, A. Patsoura, X. E. Verykios, Catal. Lett. 2008, 122, 26 – 32. [7] A. Fujishima, K. Honda, Nature 1972, 238, 37 – 38. [8] Z. Zou, J. Ye, K. Sayama, H. Arakawa, Nature 2001, 414, 625 – 627. [9] S. U. M. Khan, M. Al Shahry, W. B. Ingler, Science 2002, 297, 2243 – 2245. [10] L. De Rogatis, P. Fornasiero in Catalysis for Sustainable Energy Production, 1st ed. (Eds. P. Barbaro, C. Bianchini), Wiley, New York, 2009, pp. 173 – 223. [11] A. V. Korzhak, N. I. Ermokhina, A. L. Stroyuk, V. K. Bukhtiyarov, A. E. Raevskaya, V. I. Litvin, S. Ya. Kuchmiy, V. G. Ilyin, P. A. Manorik, J. Photochem. Photobiol. A 2008, 198, 126 – 134.

ChemCatChem 2011, 3, 574 – 577

[12] P. E. De Jongh, D. Vanmaekelbergh, J. J. Kelly, Chem. Commun. 1999, 1069 – 1070. [13] D. Barreca, P. Fornasiero, A. Gasparotto, V. Gombac, C. Maccato, T. Montini, E. Tondello, ChemSusChem 2009, 2, 230 – 233. [14] Y. S. Chaudhary, A. Agrawal, R. Shrivastav, V. R. Satsangi, S. Dass, Int. J. Hydrogen Energy 2004, 29, 131 – 134. [15] W. Siripala, A. Ivanovskaya, T. F. Jaramillo, S. H. Baeck, E. W. McFarland, Sol. Energy Mater. Sol. Cells 2003, 77, 229 – 237. [16] M. Kleinert, T. Barth, Energy Fuels 2008, 22, 1371 – 1379. [17] V. Gombac, L. De Rogatis, A. Gasparotto, G. Vicario, T. Montini, D. Barreca, G. Balducci, P. Fornasiero, E. Tondello, M. Graziani, Chem. Phys. 2007, 339, 111 – 123. [18] S. Sakthivel, M. C. Hidalgo, D. W. Bahnemann, S. U. Geissen, V. Murugesan, A. Vogelpohl, Appl. Catal. B 2006, 63, 31 – 40. [19] H. Zhang, J. F. Banfield, J. Phys. Chem. B 2000, 104, 3481 – 3487. [20] A. Jentys, Phys. Chem. Chem. Phys. 1999, 1, 4059 – 4063. [21] R. Kydd, J. Scott, W. Y. Teoh, K. Chiang, R. Amal, Langmuir 2010, 26, 2099 – 2106. [22] B. Sun, A. V. Vorontsov, P. G. Smirniotis, Langmuir 2003, 19, 3151 – 3156. [23] C. A. Emilio, M. I. Litter, M. Kunst, M. Bouchard, C. Colbeau-Justin, Langmuir 2006, 22, 3606 – 3613. [24] M. C. Hidalgo, M. Maicu, J. A. Navio, G. Colon, Catal. Today 2007, 129, 43 – 49. [25] A. Mattsson, L. Osterlund, J. Phys. Chem. C 2010 DOI: 10.1021/ jp103263n. [26] L. Millard, M. Bowker, J. Photochem. Photobiol. A 2002, 148, 91 – 95. [27] C. H. Zhou, J. N. Beltramini, Y. X. Fan, G. Q. Lu, Chem. Soc. Rev. 2008, 37, 527 – 549. [28] V. Gombac, L. Sordelli, T. Montini, J. J. Delgado, A. Adamski, G. Adami, M. Cargnello, S. Bernal, P. Fornasiero, J. Phys. Chem. A 2010, 114, 3916 – 3925. Received: August 18, 2010 Published online on December 23, 2010


www.chemcatchem.org

577