J. Phys. Chem. C 2007, 111, 18195-18203
18195
Photocatalytic Production of Hydrogen from Water with Visible Light Using Hybrid Catalysts of CdS Attached to Microporous and Mesoporous Silicas Su Young Ryu,† William Balcerski,† T. K. Lee,‡ and Michael R. Hoffmann*,† W. M. Keck Laboratories, California Institute of Technology, Pasadena, California 91125, and Nanopac, 878-6 Hanyang Plaza 5th Floor, Youngtong-dong, Youngtong-ku, Suwon 443-812 Korea ReceiVed: June 21, 2007; In Final Form: August 10, 2007
Microporous and mesoporous silicas are combined with nanoparticulate CdS particles to form hybrid photocatalysts that produce H2 from water/ethanol solutions under visible light irradiation. Catalyst structures are characterized by XRD and SEM. All hybrid materials are active photocatalysts for water splitting, and the order of photoactivity is found to be zeolite-Y > SBA-15 > zeolite-L. Silica cavity size, which determines, in part, CdS particle size, and photocatalytic activity are found to be correlated. Photocatalytic activity is seen to decrease under acidic or basic conditions with associated negative ionic strength effects. In addition, XPS analysis indicates loss of ion-exchanged and Cd2+ ion from the silicate supports occurs during the course of the photochemical reaction in solution with the complete retention of preformed and surface-bound CdS.
Introduction Photocatalytic water splitting into H2 and O2 is an area of great interest due to the potential of hydrogen gas as a cleanenergy fuel source for the future. Since the initial work of Fujishima and Honda,1 many different metal oxide semiconductors and metal sulfides have been reported to be active for water splitting. Although initial efforts involved large band gap (Eg > 3.0 eV) semiconductors that could only utilize UV light, recent efforts have focused on using visible light as the energy source,2-17 as the ultimate goal is to use solar energy to produce hydrogen fuel. Although the overall redox potential of the reaction hν
2H2O 98 2H2 + O2
(1)
is only -1.23 eV (1000 nm) at pH 7, the crucial reaction for hydrogen production is believed to be the initial one-electron transfer to H+ ion
H+ + eaq- h Haq•
(2)
where EH ) -2.5V at pH 7, which falls energetically within the visible region of light. CdS with a band gap energy of 2.4 eV is well-known to have photocatalytic activity for H2 production under visible light irradiation,18-21 although sacrificial electron donors are often used to prevent the photocorrosion CdS in the presence of O2. In an effort to increase photocatalytic yield, CdS is often combined with other semiconductors such as TiO2/CdS,22 ZnO/ CdS, ZnS/CdS,22,23 K4Nb6O17/CdS,24 and K2Ti4O9/CdS composites.18,25 Alternative approaches include coupling CdS with micro- and mesoporous materials to form hybrid or composite photocatalysts. * To whom correspondence should be addressed. Tel: +1 626 395 4391. E-mail:
[email protected]. † California Institute of Technology. ‡ Nanopac.
There are several advantages that micro- and mesoporous materials afford as hosts for semiconductor particles. For example, the pore sizes of the host micro- and mesoporous support material control the resulting particle size of the supported semiconductor prepared in situ. Decreasing particle size often has a known beneficial effect on catalyst efficiency.26-28 Furthermore, encapsulation into a micro- or mesoporous structure should provide some protection against surface-mediated reactions which corrode the catalyst, as exposed surface area will be limited. In addition, a major factor which limits the photocatalytic efficiency of CdS, or any other photocatalyst, is electron-hole recombination. Photogenerated electrons in CdS can move into the attached micro- or mesoporous molecule, while the holes remain behind, providing charge separation and increased photocatalytic efficiency. Since silicates are good candidates to serve as host structures, we have explored the use of zeolite-Y, zeolite-L, and SBA-15 as suitable composite supports. Zeolite-Y is an aluminosilicate with a supercage structure, having a pore diameter of 0.74 nm and a cavity diameter of 1.2 nm, whereas zeolite-L is an aluminosilicate with connected SiO4 and AlO4 tetrahedra that give rise to one-dimensional channels (aperture size 0.71 nm) arranged in a hexagonal structure. Zeolite-L has been shown to have excellent catalytic properties, such as isobutene dehydrogenation and oxidative cyclization.29 It also has been used to make dye composites with optical antenna properties.30 A highly ordered mesoporous silica, SBA-15, which was first reported by Zhao et al.,31 has two-dimensional hexagonal symmetry, extremely high surface areas, and controllable pore size ranging from 3 to 30 nm. Experimental Procedures UV-vis diffuse reflectance spectra were recorded on a Shimadzu UV-2101PC with an integrating sphere attachment (Shimadzu ISR-260), using Ba2SO4 powder as an internal reference. BET surface area measurements were performed on a Micromeritics Gemini 2360 Analyzer. Diffuse infrared Fourier transform (DRIFT) spectra were acquired using a Bio-Rad FTS45 spectrometer with a liquid N2-cooled MCT detector. Spectra
10.1021/jp074860e CCC: $37.00 © 2007 American Chemical Society Published on Web 11/13/2007
18196 J. Phys. Chem. C, Vol. 111, No. 49, 2007 were collected at 8 cm-1 resolution using a Spectra-Tech Collector diffuse-reflectance accessory. The solid samples were held in the sample cup of a Spectra-Tech high-temperature environmental chamber (HTEC) that could be resistively heated to 1000 K ((1 K) and the chamber evacuated to 10 µTorr. Steady-state fluorescence spectra were collected on a Hitachi F-2500 fluorometer with samples excited at 400 nm. SEM images were taken on a LEO 1550VP FESEM, whereas XRD spectra were recorded on a Phillips X’Pert PRO X-ray diffraction system. XPS experiments were performed in an M-probe surface spectrometer (VG Instruments) using monochromatic Al KR X-rays (1486.6 eV). Photolysis experiments were performed by using the collimated output of a high-pressure Hg-Xe arc lamp as a light source. The light beam was passed through an IR filter, a focusing lens, and a 400 nm cutoff filter before reaching the reactor. The reactor was cooled by a fan to keep it at room temperature. Since it is well-known that colloidal CdS suspensionsundergophotocorrosion32-34 andphotocatalyticdissolution35-39 under oxic conditions, the H2 production experiments were carried out under an argon atmosphere. All photolysis reactors were purged with Ar gas for 30 min in order to eliminate O2. Hydrogen gas evolution was measured using gas chromatography (HP 5890 Series II) with thermal conductivity detection (TCD). Due to similar conductivity values for He and H2, nitrogen was used as a carrier gas. The separations were achieved with a molecular sieve column (30 m × 0.32 mm × 12.00 µm). The gas chromatograph oven temperature was 30 °C, as this was found to be the best temperature for spatial resolution between Ar (reactor purge gas) and H2. Calibration curves were linear over a broad range of H2 concentrations. Unless otherwise specified, the catalyst loading was 0.2 g, the reaction solution was 50 mL of 50:50 ethanol/water, and the lamp power was 500 W. Aliquots of the reactor’s headspace gas (total headspace volume ) 100 mL) were syringed out in 50 µL volumes through a rubber septum. Samples were taken every 8-12 h, and several samples were taken at each time point to ensure accuracy. Synthesis of CdS-Zeolite Composites. Cadmium sulfide was synthesized within the pore spaces of aluminosilicate zeolites12,28,40-44 by the direct reaction of asorbed Cd2+ with aqueous sulfide as HS-. Commercially available zeolite-Y (Si/ Al ) 6; Strem Chemicals, Inc.) was used as received. The following procedure was used for all CdS/zeolite samples. A total of 1.0 g of zeolite was added to a 50 mL solution of ethanol containing 0.01 M cadmium acetate. The covered solution was stirred for 48 h. A 50 mL solution of ethanol containing 0.01 M sodium sulfide was prepared in a Vacuum Atmospheres glove box. The sodium sulfide solution was slowly added to the cadmium acetate solution while stirring, resulting in a yellow colored solution. The resulting solution was covered and stirred for 24 h. The pH of this solution was between pH 6.0 to 6.5. The catalyst was then removed from solution by vacuum filtration, washed several times with ethanol, and dried overnight in an oven at 100 °C. The yield from this procedure was always >90%. The relative amounts of all reagents and materials were increased proportionally in order to generate larger amounts of catalyst (i.e., using twice as much zeolite, cadmium acetate, ethanol, etc). Synthesis of Zeolite- L. The synthesis and characterization of zeolite-L has been reported previously.29,45-48 Two different sizes of zeolite-L were synthesized. Large particle size (500600 nm) and a smaller size zeolite-L with a particle size in the range of 30-100 nm.
Ryu et al. Large Zeolite-L. The desired elemental ratio of the synthesis mixture was 3 K2O/1 Al2O3/10.7 SiO2/169.3 H2O. In one flask, 3.175 g of KOH and 1.475 g of Al(OH)3 were added to 6.225 g of H2O, and the resulting solution was refluxed overnight in a mineral oil bath at 100 °C. The solution was initially cloudy but turned clear within 20 min. The second flask had 6.0 g of Aerosil OX50 silica in 22.8 g of H2O. This solution was stirred and placed in an ultrasound bath for 30 min to break up the silica particles but was not refluxed. After the overnight reflux, the Al(OH)3/KOH solution was slowly added to the silica solution, while stirring vigorously, in a Teflon reactor. The resulting solution became viscous and turned white. After 10 min of stirring, the solution was transferred to Teflon-lined tubes which were secured inside metal bomb casings. The metal bombs were put in an oven at 433 K for 5 days. After 5 days, the brownish supernatant was removed, leaving behind a white precipitate, which was scooped out into a centrifuge tube. The precipitate was resuspended in hot water and centrifuged at 6000 rpm for 10 min. The supernatant was decanted, and the product resuspended with vortex shaking. The centrifugation process was repeated 5-7 times until the pH of the supernatant became neutral. The resulting solid was dried at 100 °C in an oven overnight. Small-Scale Zeolite-L. The desired elemental ratio of the synthesis mixture was 10 K2O/1 Al2O3/20 SiO2/400 H2O. Several failed attempts were made before finding a mixing ratio that worked. Initially, a solution containing 12.0 g of silica, 5.6 g of KOH, and 36.0 g of H2O was used, but upon reflux this solution turned into a clear solid gel that stuck to the walls of the glass reaction vessel. Reducing the amount of KOH to 2.8 g led to a similar result. It was determined that more water was needed to fully dissolve the silica. In one flask, 2.8 g of KOH and 1.56 g of Al(OH)3 were added to 18.0 g of H2O. In the second flask, 12.0 g of Aerosil OX50 silica and 8.4 g of KOH were added to 54.0 g of H2O. Both solutions were refluxed overnight and then mixed together and placed in the Teflonlined pressure vessels as described above. The bombs were placed in a rotating oven (rotating at 40 rpm) at 433 K for 24 h. The resulting precipitate was washed as described earlier, and the centrifugation process was repeated 8-11 times until the pH of the supernatant stopped dropping from one consecutive wash to another, giving an end point of pH 9. Unlike the large zeolite-L sample, this sample was hard to resuspend, as the solid was more of a semi-clear gel than a clump of solid powder. Thus, the washing process was likely not as effective, which is why it took more attempts to lower the pH. The resulting solid was dried at 100 °C in an oven overnight. Synthesis of SBA-15. SBA-15 was synthesized following procedures in the literature.49,50 A total of 4.0 g of the surfactant P123 triblock copolymer was added to a solution containing 30 g of H2O and 120 g of 2.0 M HCl. The solution was stirred for 1 h at room temperature, and then 8 g of tetraethyl orthosilicate (TEOS) was added. The solution was stirred for 20 h at 308 K and then aged at 353 K for 5 h. The solution was filtered, and the resulting solid product was washed with ethanol four times and then dried in an oven at 353 K for a few hours. Some of the powder was then calcined at 773 K for 5 h. Experimental Results Before creating the CdS/micro- and mesoporous silica composites, we first confirmed that the starting silicas were synthesized correctly, specifically zeolite-L and SBA-15. Figure 1 shows the XRD spectra of small vs large zeolite-L. As expected, there was increased line broadening in the small
Production of Hydrogen from Water
J. Phys. Chem. C, Vol. 111, No. 49, 2007 18197
Figure 3. SEM images of calcined SBA-15. The scale-bar is 3 µm for (a) the top image and (b) 300 nm for the bottom image.
Figure 1. X-ray diffraction patterns (XRD) of (a) large zeolite-L (top) and (b) small zeolite-L (bottom).
Figure 2. SEM image of small zeolite-L. The scale-bar length is 3 µm.
zeolite-L sample, indicating smaller particle sizes. The XRD of SBA-15 was inconclusive. The characteristic peaks at very low angles were not accessible with the device we employed. Figure 2 shows an SEM image of small zeolite-L, and Figure 3 shows SEM images of synthesized SBA-15 that clearly illustrate the tunnel-like structure of SBA-15.31 The CdS/micro- and mesoporous silica composites were analyzed by various techniques to gain insight into the structure
and chemistry of the materials. Figure 4a shows the UV-vis diffuse reflectance spectra of the starting micro- and mesoporous silicas and the CdS/micro- and mesoporous silica composites, as well as commercial CdS. As expected, incorporation of CdS into the micro- and mesoporous silica shifted the absorption edge into the visible. In order to calculate the band gap energies, the diffuse reflectance spectra were converted into Tauc plots (Figure 4b) by plotting (F(R)hV)n vs hV, where F(R) is the Kubelka-Munk function (a transform of absorbance spectrum) and n ) 1/2 for a direct band gap semiconductor like CdS. For the synthesized compounds, the compound with the smallest band gap energy (CdS/zeolite-Y) is also the one with the smallest cavity size. As the cavity size increases, the band gap energy is seen to also increase, with the widest pore structure (CdS/SBA-15) having the largest band gap energy. This would seem to indicate that the CdS clusters are the smallest in zeolite-Y. Furthermore, zeolite-Y appeared to be the most photoactive of the compounds. Increased photocatalytic activity with decreasing particle size is a well-known phenomenon.27,51 Surface areas were measured by BET and are listed along with the experimental band gap energies in Table 1. In general, addition of CdS decreased measured surface area. In the case of SBA-15, calcination increased surface area by removing the surfactant support molecules. Calcination resulted in a loss of approximately 25% of the sample’s mass, and the surface area was increased by 50%. However, calcined CdS/SBA-15 showed no photocatalytic activity. It is likely that calcination removed the positively charged surfactant tri-block copolymers that served as substitution sites for Cd2+ ion,50 thus preventing the incorporation of CdS particles into structure.
18198 J. Phys. Chem. C, Vol. 111, No. 49, 2007
Ryu et al. TABLE 2: IR Vibrational Frequency of CdS/Zeolite-Y and CdS/ Zeolite-L covalent vibration
zeolite-Y
CdS/zeolite-Y
1639
1639
1449 νasy νasym νasym νsym νsym
Figure 4. (a) UV-vis diffuse reflectance spectra for the various starting micro- and mesoporous silicas, CdS/micro- and mesoporous silica composites, and bulk-phase CdS. (b) Tauc plot of CdS/zeolite-Y, calculated from the UV-vis spectrum in Figure 4a.
TABLE 1: BET Surface Areas and Band Gap Energies Calculated from the Tauc Plots for Various CdS/Micro- and Mesoporous Silica Compositesa material
surface area (m2/g)
band gap (eV)
CdS/zeolite-Y CdS/zeolite-L (small-scale) CdS/zeolite-L (large-scale) CdS/SBA-15** commercial CdS
380 (431) 165 (326) 122 (237) 351 (435, 674*) 50
2.31 2.35 2.38 2.44 2.29
1150 1019 764 591
1583 1388 1186 1150 1020 769 589
zeolite-L
CdS/zeolite-L
1347 1191 1152 1013 776 585
1567 1389 1192 1152 1012 776 585
at 1583 cm-1. The FTIR peak assigned to CdS is much sharper than that of commercial CdS powder. The narrow peak width provides further evidence that the CdS particle sizes are on the order of nanometers compared to the micrometer sized, bulkphase CdS. Bands corresponding to lattice vibrations are observed in the spectral region between 1650 and 400 cm-1 for both zeolite-Y and zeolite-L. For the naked zeolite particles, the IR peaks above 1000 cm-1 are attributed to asymmetric stretching vibrations O-(Al, Si)-O, and the bands below 1000 cm-1 (e.g., 776 cm-1) are attributed to symmetric stretching vibration of O-(Al, Si)-O.52 The band at 1643 cm-1 of zeolite-Y is ascribed to the corresponding covalent bond vibration.53 The IR spectra can be used to examine the interactions of the host zeolite and the intercalated CdS. For example, a new peak at 1567 cm-1 was observed in the case of CdS/zeolite-L, whereas a new peak appeared at 1583 cm-1 in the case of CdS/ zeolite-Y. These new peaks can be attributed to CdS formed in the cavities of the zeolites. Furthermore, the vibrational bands at 1449 and 1348 cm-1 show frequency shifts and intensity enhancements. For example, the band at 1449 cm-1 of zeolite-Y is shifted to a lower frequency (1388 cm-1), whereas the 1348 cm-1 band of zeolite-L is shifted to a higher frequency (1387 cm-1). These frequency shifts and intensity enhancements indicate some structural changes due to strong interactions between the host zeolite and the guest CdS. In this regard, Pichat et al.54 reported that aluminum-deficient zeolite-Y and zeolite-L framework caused a change of frequency of asymmetric stretching vibrations of the region over 1000 cm-1 indicating a structural change. The significant spectral change observed with zeoliteY, which is frequency shifted down from 1449 to 1388 cm-1, is more likely due to weakening of the O-Al(Si) framework, and probably it is caused by coordination of the Cd2+ ion to
a The numbers in parentheses are the BET surface areas of microand mesoporous silica material before combination with CdS. The asterisk (*) denotes the BET surface area of the calcined SBA-15, and the double asterisk (**) denotes the surface area without calcination.
The DRIFT spectra for SBA-15, which was not thermally treated (i.e., not calcined), and the corresponding CdS/SBA-15 composite catalysts are shown in Figure 5. There was not clear evidence of a characteristic peak for CdS, although this may indicate that the CdS in the composite was in an IR inactive form. Furthermore, the hydrocarbon peaks at 2800 cm-1, which are from the surfactant copolymer, were not present in calcined SBA-15. In contrast, the DRIFT spectra of the other CdS/Zeolite composites (Figure 6) have peaks that can be assigned to nanosized CdS. For the zeolite-L composite, the CdS peak was observed at 1567 cm-1, whereas in zeolite-Y, it was observed
Figure 5. DRIFT spectra for naked SBA-15 (not thermally treated; i.e., no calcination) and the CdS/SBA-15 composite.
Production of Hydrogen from Water
J. Phys. Chem. C, Vol. 111, No. 49, 2007 18199
Figure 7. Hydrogen production rates for the various CdS and zeolite or silica composites for irradiation of a 50:50 water-ethanol mixture by volume at λ g. 400 nm.
Figure 6. DRIFT spectra for naked Na-zeolite-Y and the CdS/zeolite-Y composite (top), compared to the small-scale zeolite-L and the CdS/ zeolite-L composite (bottom).
the oxygen of the O-Al(Si) framework. In contrast, asymmetric vibration band at 1348 cm-1 of zeolite-L shifted to upward to 1387 cm -1. The hydrogen production rates (i.e., [H2] versus time profiles) for the various CdS composite materials are shown in Figure 7. CdS/zeolite-Y was found to be the most photoactive (k ) 6.6 µmol H2 hr-1), followed by CdS/SBA-15 (k ) 2.7 µmol H2 h-1), CdS/zeolite-L large (k ) 1.7 µmol H2 h-1), and CdS/ zeolite-L small (k ) 1.3 µm H2 h-1). During the course of the photoreaction, the reaction mixture is observed to turn from yellow to gray (Figure 8). Upon exposure to air, the yellow color is recovered within minutes. The gray color is likely the result of Cd2+ ions being photoreduced to Cd0 metal on the surface of CdS as previously reported. The fluorescence spectra of zeolite-L, which is shown in Figure 9, provide some insight into the lower photoactivity with respect to H2 production. The emission peak of CdS loaded within the zeolite-L channels appears at 790 nm, whereas a weak emission of CdS embedded in the zeolite-Y matrix was observed at 523 nm. The CdS emission is red-shifted by nearly 260 nm when CdS is intercalcated into zeolite-L compared to deposition on the zeolite-Y surface. The zeolite-L material has a onedimensional channel of 3∼6.5 nm with ring opening diameters ranging from 7.1 to 12.6 Å. As the concentration of CdS nanoparticles builds up in the channel axis, the aggregated CdS
particles likely undergo fast radiationless Fo¨rster-type energy migration where the excitation energy is directly transferred from CdS to neighbor CdS without conversion into light.55 Therefore, the observed emission most likely occurs from the edge of CdS particle clusters aggregated along channel resulting emission at longer wavelengths (i.e., lower energies). Several different experiments were performed to probe some of the critical parameters that impact the rates of hydrogen production. The effect of lamp power was explored, and as shown in Figure 10, the hydrogen production rate was linear over the range of lamp output power. The effect of ionic strength on hydrogen production is illustrated in Figure 11. Increasing the ionic strength of the solution with sodium perchlorate decreased the amount of hydrogen produced. The thickness of the electrical double layer around the CdS/zeolite-Y, which is given by the reciprocal of the Debye parameter κ (i.e., κ-1 in units of length), can be determined as follows
κ≡
( ) 2F2µ 0RT
0.5
(3)
where is the dielectric constant of the solvent or mixed solvent system, 0 is the permittivity of free space (8.854 × 10-12 C2 J-1 m-1), µ is the ionic strength of the background electrolyte (mol m-3), R is the gas constant (8.314 J mol-1 K-1), T is temperature in units of K, and F is the Faraday constant (96485 C mol-1). With ) 78.5 (ΧH2O ) 0.5) and ) 24.3 (ΧEtOH ) 0.5), the thickness of the electrical double layer for an ionic strength of 1.0 mol m-3 (10-3 M) at T ) 298 K gives κ-1 ) 7.9 nm. For the concentrations of sodium perchlorate used in the ionic strength experiments (0.01, 0.1, and 0.5 M) and assuming a starting background electrolyte concentration of 10-3 M, the calculated thickness of the double layer is 2.37, 0.78, and 0.35 nm, respectively. The contribution from solvent should be minimal, but it is likely that ions (such as Cd2+) leach out of the catalyst during the course of the experiment. It would appear that the electric double layer can be compressed to at least 2.37 nm with no adverse affect on reaction rate, but compression to ∼1 nm leads to a noticeable drop-off in catalytic efficiency. Figure 12 shows the effect of acid or base addition on hydrogen production. Adding both acid and base decreased photocatalytic activity. In several of the experiments where acid
18200 J. Phys. Chem. C, Vol. 111, No. 49, 2007
Ryu et al.
Figure 8. Color changes observed during irradiation of the CdS/zeolite-Y composite photocatalyst under argon: (a) initial color before illumination; (b) color change to grayish brown during illumination with H2 production; (c) gray color of composite photocatalyst after photolysis; (d) color change back to yellow after a few minutes of exposure to oxygen. Color returns to yellow after several hours of exposure to oxygen. The color change from yellow to gray is due to the photoreduction of Cd2+ on the CdS surface to metallic Cd(0). Colored metal sulfides are well-known to turn black upon prolonged UV-vis irradiation. A yellow color returns upon oxidation by air. Some of the yellow color that reappears may be due to oxidation of S(-II) to orthorhombic elemental sulfur, which is yellow.
Figure 9. Emission spectra of CdS embedded with the CdS/zeolite-Y and CdS/zeolite-L composite catalysts. The excitation was at λex ) 430 nm for CdS/zeolite-L and at λex ) 390 nm for CdS/zeolite-Y at 25 °C.
Figure 10. Hydrogen production rates for CdS/zeolite-Y as a function of increasing light intensity. At 500 W, the emission of the Hg-Xe arc lamp at λ g 400 nm, the He incident light flux per unit area, Io, was determined via potassium ferrioxalate actinometry to be 4.7 × 1015 photons cm-2 s-1. The assumption has been made that Io is proportional to the applied lamp power.
and base were added to the catalyst solution, there was no change in the pH. This is because the catalyst itself has some self-buffering capacity due to the presence of surface >SiOH and >Al-OH groups. At the typical experimental pH (∼9), the silanol groups will be deprotonated and the aluminol groups (i.e., >AlOH2+) will be protonated. Thus, there is capacity to absorb both H+ and OH- from solution. These experiments also increased ionic strength, but that effect alone cannot account
for the decreased photocatalysis rate. In the case of the base experiments, there is a steady drop off in reactivity despite constant pH. It is likely that at increased base concentration there are less available surface protons that can participate in the formation of H2. In the case of acid addition, there is a very sharp drop-off in both pH and reactivity when acid concentration was changed from 1.0 × 10-3 to 2.5 × 10-3 M. Somewhere in this range, it is likely that the capacity for the silanol groups to
Production of Hydrogen from Water
J. Phys. Chem. C, Vol. 111, No. 49, 2007 18201
Figure 11. Hydrogen production rates as a function of the background solution-phase ionic strength for photolysis of CdS/zeolite-Y composite photocatalyst. NaClO4 was used as the background electrolyte. Photolysis of a 50:50 water-ethanol mixture by volume at λ g. 400 nm.
Figure 13. XPS of Cd peaks from CdS/zeolite-Y before (top) and after (bottom) photoreaction.
absorb protons was reached. The excess protons then likely interfered with other surface groups, like S-H, shutting down the reaction. To gain insight into the fate of CdS during the photoreaction, the postreaction catalyst was collected by filtration, washed, and examined by XPS. Figure 13 shows XPS data for the Cd 3d5/2 and 3d3/2 states in CdS/zeolite-Y before and after photolysis. Two distinct forms of Cd are observed, and the ratio of the two changes after the photoreaction. The higher energy peaks in each pairing (414.1 and 407.3 eV) are assigned to CdS, and the lower energy peaks (413.3 and 406.5 eV) are assigned to Cdsubstituted zeolite sites, which would have CdO character. Reference data indicates CdS to have a binding energy between 0.7 and 1.1 eV higher than CdO,56 which is consistent with the 0.8 eV difference in peak positions. Using the Si and Al peaks to normalize the relative peak areas between the two samples (the overall Si and Al content should not change), it was found that there was a 21% loss of Cd that occurred during photolysis but no net loss of sulfur (the XPS sulfur signal was too weak to perform individual peak assignment, only total area could be measured). The loss in intensity is seen as a reduction in the CdO peak, which can be explained as follows. It is likely that not all Cd substitution sites were accessible for formation of CdS and thus remained in the zeolite matrix as Cd2+ counterions. The cadmium at these substitution sites could then be leached out of the zeolite-Y matrix during the 24 h reaction period. Another possibility is that the specific Cd2+ loss was actually from CdS particles that had undergone photocorrosion leaving sulfur atoms attached to the zeolite matrix. The formation of CdS clusters or particles may not occur exclusively within the cavities or pores of the zeolite matrix. A fraction of the CdS may have precipitated on the external surfaces of zeolite-Y, although the XRD patterns of zeolite-Y with embedded CdS were consistent with that of zeolite-Y alone without discernible 2θ peaks for CdS. This result indicates that the inherent framework structure of zeolite-Y was maintained after ion exchange and that CdS was synthesized preferentially within the zeolite-Y cavities. Figure 12. Influence of acid/base addition on hydrogen production for CdS/zeolite-Y. Photolysis of a 50:50 water-ethanol mixture by volume at λ g. 400 nm.
Discussion The observed color change (i.e., yellow to gray) that is observed with CdS/zeolite-Y (Figure 8) during hydrogen
18202 J. Phys. Chem. C, Vol. 111, No. 49, 2007 production is most likely due to the rapid photoreduction of Cd(II) to Cd(I) and then to Cd(0) on the surface of CdS in the absence of oxygen. There is also some probability that S(-II) in the CdS matrix is also oxidized partially by trapped valence band holes to form S(0) and eventually polysulfide ion (S22-). Upon exposure to oxygen, the yellow color is regenerated over several hours with the return of yellow CdS on the nanocomposite CdS/zeolite structures. The photoreduction and color changes of metal sulfides and upon exposure to UV or visible radiation is well documented.55,57-65 In our experiments and those of Gutierrez and Henglein,66 H2 production is preceded by an initial photoreduction of CdS. Based on these observations, a probable set of surface chemical reactions for H2 production in ethanol-water or waterisopropanol mixtures can be proposed. The initial steps in the mechanism involve photoexcitation of CdS followed by surface site electron and hole-trapping at protonated thiol groups, which are preform due to the hydrolysis of the CdS surface in water.
Ryu et al. of the injection of conduction band electrons into the lowerlying bands of the host material. This is consistent with the observed blue shifts, which were noted in the fluorescence spectra that are attributed to “quantum confinement” within zeolite cages.70 We previously noted that absorption band edged of CdS/ zeolite-Y at 524 nm was red-shifted, whereas the fluorescence emission band at λem ) 526 nm was blue-shifted compared to those of the freely suspended CdS colloids in EtOH. A shift in the diffuse reflectance spectrum reflects a relative change in the HOMO and LUMO states, which will depend on the size and structure of the embedded colloids or clusters within the NaY zeolite cavity. The red-shift in the absorption spectrum is most likely due to the formation of CdS clusters in the restricted space of the zeolite cavity (d ∼ 1.3 nm) as opposed to the formation of crystalline cubic or hexagonal CdS. As a consequence, the cavity-bound CdS clusters have very small Stokes shifts, ∆λ (i.e., λex - λem). Based on our observations, it appears that (1) there is an efficient dispersion of CdS clusters into discrete zeolite-Y cages and pores; (2) the quantum confinement of hydrated CdS increases the reduction potential for the bound proton to hydrogen atom electron transfer (i.e., >CdOH2+ + ecb- f >CdOH + H•; and (3) the confined CdS supercluster is a more efficient chromophore in the visible portion of the spectrum. Conclusions Several micro- and mesoporous silicas were combined with nanosize CdS particles to produce catalysts that produce H2 gas from water/ethanol solutions using visible light. All compounds were active photocatalysts for water splitting with the following order of photoreactivity: zeolite-Y > SBA-15 > zeolite-L. Optimization of reaction conditions is critical to achieving a high level of hydrogen production: For example, pH, ionic strength, and water-to-alcohol ratios all have a marked effect on catalyst efficiency. Preventing loss of CdS during the course of the photoreaction is also paramount to creating a stable, longterm use catalyst.
A similar set of photoreactions are likely to occur due to electron and hole trapping at neutral >CdSH and >CdOH sites. Shiragami et al.65 reported that CdS suspensions photolyzed in the presence of CH3OH turn brown due to the internal reduction of lattice Cd(+II) to Cd(0). In addition, they were able to estimate the amount of Cd(0) produced. They also noted that the brown-colored colloidal suspension gave an absorption spectrum that was red-shifted (λabs > 600 nm), and that the yellow color was restored upon the reintroduction of oxygen into the system. Their observations are consistent with those reported herein. Since the early work of Herron and co-workers44,67,68 on CdSzeolite composites, there have been a number of investigations of the photocatalytic properties of CdS encapsulated in zeolite cavities. Most often the enhanced reactivity has been attributed to enhanced electron-hole charge separation and the increased stability44 of the CdS has been explained in terms of stabilization by oxygen atoms in the zeolite cages. Guan et al.69 reported that CdS formed in zeolite-ETS-4 and titanium silicate loaded ETS-4 were effective catalysts for H2 production at λ > 420 nm in the presence of HS- and SO32as sacrificial electron donors at pH 14 (i.e., 1.0 M NaOH) and that the CdS embedded in the ETS-4 cavities was resistant to photocorrosion. Furthermore, nanoparticulate CdS in the ETS4-zeolite showed enhanced activity relative to naked CdS suspensions. The enhanced activity if often explained in terms
Acknowledgment. We are grateful to the Hydrogen Energy Research & Development Center and 21st Century Frontier Research and Development Program of the Ministry of Science and Technology of Korea for research support. References and Notes (1) Fujishima, A.; Honda, K. Nature 1972, 238, 37-41. (2) Kato, H.; Kudo, A. J. Phys. Chem. B 2002, 106, 5029-5034. (3) Ishii, T.; Kato, H.; Kudo, A. J. Photochem. Photobiol. A 2004, 163, 181-186. (4) Tsuji, I.; Kato, H.; Kobayashi, H.; Kudo, A. J. Am. Chem. Soc. 2004, 126, 13406-13413. (5) Tsuji, I.; Kato, H.; Kudo, A. Chem. Mater. 2006, 18, 1969-1975. (6) Yamasita, D.; Takata, T.; Hara, M.; Kondo, J. N.; Domen, K. Solid State Ionics 2004, 172, 591-595. (7) Niishiro, R.; Kato, H.; Kudo, A. Phys. Chem. Chem. Phys. 2005, 7, 2241-2245. (8) Tsuji, I.; Kudo, A. J. Photochem. Photobiol. A 2003, 156, 249252. (9) Diwald, O.; Thompson, T. L.; Goralski, E. G.; Walck, S. D.; Yates, J. T. J. Phys. Chem. B 2004, 108, 52-57. (10) Konta, R.; Ishii, T.; Kato, H.; Kudo, A. J. Phys. Chem. B 2004, 108, 8992-8995. (11) Tsuji, I.; Kato, H.; Kudo, A. Angew. Chem., Int. Ed. 2005, 44, 3565-3568. (12) Thompson, T. L.; Yates, J. T. Chem. ReV. 2006, 106, 4428-4453. (13) Gole, J. L.; Stout, J. D.; Burda, C.; Lou, Y. B.; Chen, X. B. J. Phys. Chem. B 2004, 108, 1230-1240. (14) Di Valentin, C.; Pacchioni, G.; Selloni, A. Phys. ReV. B 2004, 70.
Production of Hydrogen from Water (15) Hitoki, G.; Ishikawa, A.; Takata, T.; Kondo, J. N.; Hara, M.; Domen, K. Chem. Lett. 2002, 736-737. (16) Kasahara, A.; Nukumizu, K.; Hitoki, G.; Takata, T.; Kondo, J. N.; Hara, M.; Kobayashi, H.; Domen, K. J. Phys. Chem. A 2002, 106, 67506753. (17) Kato, H.; Asakura, K.; Kudo, A. J. Am. Chem. Soc. 2003, 125, 3082-3089. (18) Shangguan, W.; Yoshida, A. J. Phys. Chem. B 2002, 106, 1222712230. (19) Koca, A.; Sahin, M. Int. J. Hydrog. Energy 2002, 27, 363-367. (20) Milczarek, G.; Kasuya, A.; Mamykin, S.; Arai, T.; Shinoda, K.; Tohji, K. Int. J. Hydrog. Energy 2003, 28, 919-926. (21) Mrowetz, M.; Balcerski, W.; Colussi, A. J.; Hoffmann, M. R. J. Phys. Chem. B 2004, 108, 17269-17273. (22) Chen, F.; Zhu, Y. P.; Ma, H. L.; Bo, Z. L.; Zhang, J. L. Acta Phys.Chim. Sin. 2004, 20, 1292-1296. (23) Innocenti, M.; Cattarin, S.; Loglio, F.; Cecconi, T.; Seravalli, G.; Foresti, M. L. Electrochim. Acta 2004, 49, 1327-1337. (24) Yoshimura, J.; Tanaka, A.; Kondo, J. N.; Domen, K. Bull. Chem. Soc. Jpn. 1995, 68, 2439-45. (25) Shangguan, W. F.; Yoshida, A. Sol. Energy Mater. Sol. Cells 2001, 69, 189-194. (26) Bahnemann, D. W.; Kormann, C.; Hoffmann, M. R. J. Phys. Chem. 1987, 91, 3789-3798. (27) Choi, W. Y.; Termin, A.; Hoffmann, M. R. Angew. Chem., Int. Ed. 1994, 33, 1091-1092. (28) Chen, X. B.; Burda, C. J. Phys. Chem. B 2004, 108, 15446-15449. (29) Wan, Y.; Williams, C. D.; Duke, C. V. A.; Cox, J. J. Microporous Mesoporous Mater. 2001, 47, 79-84. (30) Ruiz, A. Z.; Li, H. R.; Calzaferri, G. Angew. Chem., Int. Ed. 2006, 45, 5282-5287. (31) Zhao, D. Y.; Feng, J. L.; Huo, Q. S.; Melosh, N.; Fredrickson, G. H.; Chmelka, B. F.; Stucky, G. D. Science 1998, 279, 548-552. (32) Davis, A. P.; Huang, C. P. Water Res. 1991, 25, 1273-1278. (33) Korgel, B. A.; Monbouquette, H. G. J. Phys. Chem. B 1997, 101, 5010-5017. (34) Vucemilovic, M. I.; Vukelic, N.; Rajh, T. Photochem. Photobiol. A 1988, 42, 157-67. (35) Tang, W. Z.; Huang, C. P. Water Res. 1995, 29, 745-756. (36) Davis, A. P.; Hsieh, Y. H.; Huang, C. P. Chemosphere 1994, 28, 663-674. (37) Hsieh, Y. H.; Huang, C. P.; Davis, A. P. Chemosphere 1993, 27, 721-732. (38) Hsieh, Y. H.; Huang, C. P.; Davis, A. P. Chemosphere 1992, 24, 281-290. (39) Hsieh, Y. H.; Huang, C. P. Colloids Surf. 1991, 53, 275-295. (40) Batzill, M.; Morales, E. H.; Diebold, U. Phys. ReV. Lett. 2006, 96, 026103-026104. (41) Peng, H.; Liu, S. M.; Ma, L.; Lin, Z. J.; Wang, S. J. J. Cryst. Growth 2001, 224, 274-279.
J. Phys. Chem. C, Vol. 111, No. 49, 2007 18203 (42) Chen, W.; Wang, Z. G.; Lin, L. Y. J. Lumin. 1997, 71, 151-156. (43) Jentys, A.; Grimes, R. W.; Gale, J. D.; Catlow, C. R. A. J. Phys. Chem. 1993, 97, 13535-13538. (44) Herron, N.; Wang, Y.; Eddy, M. M.; Stucky, G. D.; Cox, D. E.; Moller, K.; Bein, T. J. Am. Chem. Soc. 1989, 111, 530-540. (45) Ko, Y. S.; Ahn, W. S. Bull. Korean Chem. Soc. 1999, 20, 173178. (46) Ko, Y. S.; Ahn, W. S. Powder Technol. 2004, 145, 10-19. (47) Megelski, S.; Calzaferri, G. AdV. Funct. Mater. 2001, 11, 277286. (48) Ruiz, A. Z.; Bruhwiler, D.; Ban, T.; Calzaferri, G. Monatsh. Chem. 2005, 136, 77-89. (49) Xu, W.; Liao, Y. T.; Akins, D. L. J. Phys. Chem. B 2002, 106, 11127-11131. (50) Wang, S.; Choi, D. G.; Yang, S. M. AdV. Mater. 2002, 14, 1311+. (51) Choi, W. Y.; Termin, A.; Hoffmann, M. R. J. Phys. Chem. 1994, 98, 13669-13679. (52) Flanigen, E. M.; Khatami, H. Szymansk. Ha 1971, 201-&. (53) Liu, J.-F.; Li, X.-L.; Li, Y.-D. J. Cryst. Growth 2003, 247, 419424. (54) Pichat, P. J. Phys. Chem. 1975, 79, 2127-2129. (55) Huber, S.; Ruiz, A. Z.; Li, H.; Patrinoiu, G.; Botta, C.; Calzaferri, G. Inorg. Chem. Acta 2007, 360, 869-875. (56) Wagner, C. D.; Naumkin, A. V.; Kraut-Vass, A.; Allison, J. W.; Powell, C. J.; Rumble, Jr. X-Ray Photoelectron Spectroscopy Database; National Standard Reference Database 20, Version 3.4, NIST, Gaithersburg, MD 2006, available on internet at http://srdata.nist.gov/XPS/. (57) Stroyuk, A. L.; Shvalagin, V. V.; Raevskaya, A. E.; Korzhak, A. V.; Kuchmii, S. Y. Theor. Exp. Chem. 2003, 39, 341-346. (58) Albers, C. Z. Phys. Chem. (Leipzig) 1968, 237, 278-80. (59) Niekisch, E. A.; Pflaum, U.; Rompe, R. Z. Phys. Chem. Leipzig 1956, 205, 146-59. (60) Gisolf, J. H. Physica 1939, 6, 84-96. (61) Guntz, A. A. Ann. Chim. Appl. 1926, 5, 363-420. (62) Job, A.; Ernschwiller, G. Compt. Rend. 1923, 177, 313-6. (63) Sviszt, P. Acta Phys. Pol. 1964, 26, 823-7. (64) Shionoya, S.; Watanabe, K. Bull. Chem. Soc. Japan 1957, 30, 11821. (65) Shiragami, T.; Fukami, S.; Wada, Y. J.; Yanagida, S. J. Phys. Chem. 1993, 97, 12882-12887. (66) Gutierrez, M.; Henglein, A. Ber. Bunsen-Ges. 1983, 87, 474-8. (67) Wang, Y.; Herron, N. J. Phys. Chem. 1988, 92, 4988-4994. (68) Wang, Y.; Herron, N. J. Phys. Chem. 1987, 91, 257-260. (69) Guan, G.; Kida, T.; Kusakabe, K.; Kimura, K.; Fang, X.; Ma, T.; Abe, E.; Yoshida, A. Chem. Phys. Lett. 2004, 385, 319-322. (70) Aparisi, A.; Fornes, V.; Marquez, F.; Moreno, R.; Lopez, C.; Meseguer, F. Solid-State Electron. 1996, 40, 641-645.
