Chemical states of metal-loaded titania in the photoreduction of CO2

(1)

Chemical states of metal-loaded titania in the

photoreduction of CO

2 I-Hsiang Tseng, Jeffrey C.-S. Wu

*

Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan 10617, ROC Received 18 November 2003; received in revised form 10 March 2004; accepted 20 March 2004

Available online 31 July 2004

Abstract

Various sol–gel procedures and post-treatments were applied to modify the distribution of Cu on the surface of Cu/TiO2catalysts in order

to increase the production of methanol in the photoreduction of CO2. The chemical states of Cu in 2 wt.% Cu/TiO2were characterized in detail

as a follow-up the high Cu dispersion found in previous studies. XRD, XPS and XAS analysis reveal that the active Cu state for the photoreaction of CO2is suggested to be highly dispersed Cu(I). The photoactivity decreases when Cu(I) changes to Cu(0) or aggregates after

reduction with H2. An optimal distribution of Cu exists between the surface and bulk of TiO2particles. The photocatalytic activity attains

maximum when near 25% of the total Cu loading is located on the outermost surface of a TiO2particle. Cu/TiO2is a more active catalyst than

Ag/TiO2because Cu particles act as electron trapping sites while still maintain the mobility of photoelectrons.

Keywords: Cu/TiO2; Ag/TiO2; CO2; Photoreduction; XAS; Fluorescence

1. Introduction

The use of transition-metal-loaded titania as a photo catalyst in photoreactions has been extensively studied. Well-controlled metal-loaded titania can efficiently inhibit the recombination of photo-generated hole-electron pairs

[1–5]; rapidly transfer electrons/holes to adsorbed reactants

[6,7], and even modify the bandgap structure with a con-comitant red-shift of the intrinsic absorption edge[8–10].

Greenhouse gases such as CO2, CH4and CFCs are the

primary causes of global warming. CO2can be transformed

into hydrocarbons in a photocatalytic reaction. The advan-tage of photo reduction of CO2is to use inexhaustible solar

energy. Our previous work on the photocatalytic reduction of CO2showed that loading with Cu promoted CO2reduction

activity and improved the selectivity of the product toward methanol[11]. However, TiO2-supported Ag and Pt did not

show these effects.

This study focuses on how copper loading promotes photoreduction, and which chemical state of Cu contributes

mainly to enhance the activity of TiO2. Various preparation

procedures and precursors were applied in the synthesis of Cu/TiO2. A series of Cu/TiO2catalysts were studied and Ag/

TiO2was also prepared and analyzed for comparison.

2. Experimental

The Cu/TiO2catalysts were prepared using a modified

sol–gel method, illustrated inFig. 1. A specified amount of copper precursor, CuCl2 or Cu(CH3COO)2, was mixed

during the hydrolysis/polycondensation period which lasted 0–8 h.Table 1lists the steps in the catalysts preparation. A number was assigned to each notation; for example, CuCl2

-3 h indicates that copper chloride was added on the third hour during hydrolysis/polycondensation period. The final stable transparent green sol was dried and calcined at 500 8C in flowing air. After the catalysts had been crushed into a powder in a mortar, they were ready for use. Some catalysts were further reduced in H2 at 300 8C for 3 h. A reduced

catalyst was denoted by an ‘‘r’’ in front of the notation, as in r-CuCl2-3 h. The same procedure was implemented to

pre-pare Ag/TiO2. In this case, the silver precursor AgNO3was

www.elsevier.com/locate/cattod

* Corresponding author. Tel.: +886 2 2363 1994; fax: +886 2 2362 3040. E-mail address: cswu@ntu.edu.tw (Jeffrey C.-S. Wu).

(2)

mixed with the TiO2 sol on the 8 h during hydrolysis/

polycondensation period. A commercial pure TiO2powder,

Degussa P25, was used for comparison.

The photoreduction of CO2was carried out in a

cylind-rical quartz reactor with a capacity of 300 ml. The catalyst of 0.3 g powder was suspended in 0.2 N NaOH solution. Ultra-pure CO2from Air Products and Chemicals Co., USA was

bubbled through the reactor for at least 4 h to ensure that all dissolved oxygen was eliminated. The illumination system included a mercury lamp (Ultra-Violet Products Inc., USA, 11SC-1) with a wavelength of 254 nm in the center of the reactor. A detailed description of the photoreactor system was described in previous work[11].

The UV–vis spectrum was obtained using a diffuse reflectance UV spectrophotometer (VARIAN, Cary100). The crystalline phase was identified by X-ray diffractometry (XRD) on a MAC M03XHF instrument (Material Analysis and Characterization, Japan). X-ray photoelectron spectro-scopy (XPS) was performed using a VG Microtech MT500 with an Mg-Ka X-ray source. All binding energies were

referenced to oxygen (1s) at 530.7 eV or carbon (1s) at 285.6 eV. A scanning electron microscope (SEM), equipped with an energy-dispersive spectrometer (EDS) LEO 1530 FEG SEM EDS, was used to observe the morphology of

the catalyst particles and to measure their elemental com-position. The element distribution was analyzed by measur-ing the X-ray radiation emitted after an excitmeasur-ing electron beam (15 keV) had scanned a small area of a compressed catalyst pellet. A powder sample cell (JASCO FP-1061) was filled with a fixed amount of catalyst to measure the emitted fluorescence spectra under an exciting wavelength of 290 nm, using a spectrofluorometer (JASCO FP-777) at room temperature.

The X-ray absorption spectra (XAS) of the Cu K-edge for all catalysts were measured at the Wiggler 17C station of the Taiwan Synchrotron Radiation Center in the Hsinchu Science-based Industrial Park. The powder sample was pressed in a sample holder orientated at 458 to the incident X-ray beam in a sample box. A fluorescence mode and an Ni-filter were employed to improve the resolution. For Cu, the X-ray photon energy was varied across and beyond the absorption edge of the measured atom from 200 eV below the copper absorption edge at 8979 eV, to 800 eV above it. The spectra of pure Cu2O, CuO powder and Cu foil were

measured as standard references. The fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) of all species were derived using WinXAS 2.33 software [12].

3. Results and discussion

Different preparation procedures were employed in order to obtain various copper distributions in TiO2particles. The

XRD patterns of all Cu/TiO2catalysts showed anatase as the

only phase of TiO2and no characteristic peaks associated

with Cu/CuOx crystalline could be detected. The copper

particles were thus considered to be highly dispersed throughout the TiO2 structure [13]. Fig. 2(a) shows the

UV–vis spectra of the Cu/TiO2and the pure TiO2catalysts.

The absorption edge of TiO2shifts toward the visible region

upon the addition of either copper or silver (Fig. 2(b)). The tailing absorption peaks can be regarded as the extra tail states in the bandgap because the Cu or Ag atoms were added to the TiO2 matrix [14,15]. The extension of the

absorption edge to longer wavelengths for Cu/TiO2indicates

a good contact between TiO2and Cu grains[16]. Similar

results are obtained for Ag/TiO2 catalysts with various

I.-H. Tseng, J.C.-S. Wu / Catalysis Today 97 (2004) 113–119 114

Fig. 1. Sol–gel procedures for preparation of the catalysts.

Table 1

Various steps in catalysts preparation

Catalysts Notation Precursor adding during hydrolysis H2reduction at 300 8C

0.6 7 wt.% Cu/TiO2 – 8 h No

2 wt% Cu/TiO2 CuCl2-0 h Beginning No

CuCl2-1 8 h 1 8 h No

r-CuCl2-1 8 h 1 8 h Yes

CuAc2-8 ha 8 h No

r-CuAc2-8 h 8 h Yes

1 9 wt.% Ag/TiO2 – 8 h No

All catalysts were calcined at 500 8C.

a

(3)

loadings, as shown inFig. 2(b). The baseline in the visible light region is clearly raised, the reason for which has been suggested to be the presence of Ag clusters[17–19].

Fig. 3presents the photocatalytic activities of the cata-lysts. After 30 h of irradiation with UV (254 nm), the activity of the catalysts followed the order CuCl2-1 h

CuCl2-3 h > CuCl2-0 h > CuAc2-8 h > CuCl2-8 h, with a

maximum methanol yield nearly 1000 mmol/gcatal. The

optimal silver loading was also found to be 2 wt.% Ag, but its performance was below 300 mmol/gcatalafter 30 h of

irradiation. The reduction of Cu/TiO2 in H2decreases the

yield of methanol, so did the loading of TiO2with silver.

Fig. 4shows the average rates of methanol production using

catalysts with various metal loadings during the initial 6 hs of UV irradiation. Most rates of methanol production cat-alyzed by 2 wt.% Ag/TiO2did not exceed those obtained by

2 wt.% Cu/TiO2. Only in the region of high metal loadings,

the catalysts with Ag loadings were more active than that with high Cu loading.

Valence electrons of TiO2can be promoted to the

con-duction band by photons with excess bandgap energy. Once photoelectron/hole pairs have been generated in the TiO2

particles, they may either recombine or react with the species adsorbed on the TiO2surface. Whether a reaction

between the electron/hole pairs and the adsorbates can

Fig. 2. UV–vis spectra of (a) 2 wt.% Cu/TiO2 and TiO2; (b) Ag/TiO2

catalysts.

Fig. 3. Methanol yields of 2 wt.% Cu/TiO2under UV irradiation (254 nm).

Fig. 4. Effects of Cu or Ag loadings on the initial 6 h production rate of methanol.

(4)

proceed depends on the energy positions of the conduction and valence bands[20,21]. In the photoreduction of CO2, the

energy position of the conduction band (ECB) can be

regarded as the reduction capability of TiO2. The ECBlevel

generally depends on the pH of the solution in which TiO2is

suspended. A higher pH value corresponds to a more negative ECB level indicating that the higher reduction

capability of TiO2. In this study, methanol was the main

liquid product of the photoreduction of CO2. TiO2particles

were suspended in a CO2-saturated solution, whose pH value

was near 7. The redox potential of CO2/methanol is

0.62 eV. The ECB level of anatase TiO2, is 0.75 eV at

pH 7, and so is sufficiently negative to reduce CO2 to

methanol[21–23]. Consequently, the production of metha-nol was observed under our reaction conditions.

Metal-loaded TiO2showed a higher photocatalytic

activ-ity than pure TiO2, and Cu/TiO2yielded more methanol than

did Ag/TiO2, in this study. The photoelectrons can be

trapped at the surface Cu sites due to the redox potentials (E8) of Cu2+/Cu+(0.1 eV) and Cu+/Cu0(0.5 eV), which are lower than the ECBlevel of TiO2[24,25]. Thus Cu particles

can serve as electron acceptors to restrain electron/hole pairs from recombination. However, E8 of Ag+/Ag0is near 0.8 eV. The photoelectrons are strongly trapped by the Ag particles and are therefore hard to transfer to the adsorbates on the surface of TiO2. The XPS spectra of Ag/TiO2catalysts with

various Ag loadings in Fig. 5 indicate that Ag0 is the dominant chemical state. An increased Ag loading shifts the chemical state to Ag+. The interaction between TiO2and

silver is significant in that Ag+can trap electrons efficiently to become Ag0, as suggested by Epifani et al. [17] and Vamathevan et al. [26]. This fact may explain why the photoreduction of CO2 on Ag/TiO2 is less efficient than

on Cu/TiO2.

The fluorescence emitted by the TiO2catalysts was also

observed in this study.Fig. 6shows the photoluminescence of Degussa P25, TiO2, Cu- and Ag-loaded TiO2. The

wavelength of fluorescence was in the range 400–500 nm under UV excitation at around 290 nm. The emitted fluor-escence of metal-loaded TiO2was in the same region as that

of P25 and TiO2, but at much lower intensity. One of the

origins of photoluminescence can be due to the recombina-tion of excited electron/hole pairs[27–29]. Accordingly, the decline in the intensity of fluorescence reveals the decrease in the rate of recombination of electron/hole pairs on metal-loaded TiO2catalysts.

The Cu/Ti molar ratios, determined using XPS and EDS analysis which can respectively represent the surface and bulk values, indicate the portion of Cu distribution in the catalyst particles. In order to quantify the comparison, the molar percentage of the Cu located at the surface was estimated with the following two assumptions: (1) particles are spherical with a mean radius of 25 nm according to the particle size distribution and the TEM micrograph of the catalysts; (2) the collection depth of XPS measurements is close to 1 nm from the surface. Then, the quantitative data from XPS yields a molar ratio of 11 vol.% on the outer shell, and the results of EDS refer to the bulk. Fig. 7 plots the portion of Cu or Ag located in the outer shell versus the total loading. For Cu/TiO2, the portion of surface Cu decreases

monotonously with the amount of Cu loaded. In contrast, the portion of Ag on the surface increases with the Ag loading. The findings that higher Cu loading gave lower dispersion of Cu on the surface of Cu/TiO2catalysts also consisted with

our previous TPR study[11].

Fig. 8 shows the yield of methanol is influenced by the portion of Cu located on the surface for 2 wt.% Cu/TiO2

catalysts which were prepared by various adding time of I.-H. Tseng, J.C.-S. Wu / Catalysis Today 97 (2004) 113–119

116

Fig. 5. Ag(3d) XPS spectra of Ag/TiO2.

(5)

CuCl2. The yield of methanol declined as the portion of Cu

on the surface increased. An optimal amount of copper had to be dispersed on the outer surface to promote the produc-tion of methanol. The optimal Cu proporproduc-tion on the surface (catalyst CuCl2-1 h) was about 25%. A large amount of

copper dispersed inside the TiO2 particles would lead to

increase the photoelectron/hole recombination rate in the bulk thus reducing the activity[2–4]. Zhang et al. suggested that a shallow charge-trapping site was favorable for

suppression of surface recombination but that an optimal amount of metal inside the TiO2particles was also required

to transfer charges and maintain their mobility. However, when CuAc2was used as a precursor, the surface Cu ratios of

both 2 wt.% Cu/TiO2catalysts, CuAc2-0 h and CuAc2-8 h,

were also close to 25% (not shown), but their methanol yields, 450 600 mmol/gcatal, after 30 h of irradiation were

lower than those of 2 wt.% Cu/TiO2catalysts, when CuCl2

was used (Fig. 3). Additional factor, that is, Cl ion, might also influence the photo activity.

Fig. 9displays the XPS of Cu indicating that the copper on the surface of TiO2 exhibits in multiple-oxidation

states. A comparison with the spectrum of Cu2O indicates

that Cu(I) is the primary species on Cu/TiO2, according to

the position and the shape of the Cu(2p) XPS peaks. Similar results were also obtained regardless of the various preparation procedures.Figs. 10–12show the FT-EXAFS spectra of 2 wt.% Cu/TiO2catalysts. The first-shell

neigh-boring peaks are characteristic of the isolated Cu(I) and Cu(0) particles at 1.5 and 2.2 A˚ , respectively. The peak at 2.6 A˚ indicates notable aggregation of copper particles in the catalysts[29,30]. Most of the copper in the catalysts was stable as Cu(I) after calcination as shown in Fig. 10. That is, although copper particles on TiO2 existed in

multiple-oxidation states, the isolated Cu(I) dominated which consisted with our early study from XPS analysis

[31]. The unique first-shell neighboring atoms in the FT-EXAFS spectra prove that Cu on TiO2is well-dispersed.

This fact also explains the absence of Cu diffraction peaks in the X-ray diffractograms.

Fig. 11shows that a large amount of Cu(I) was reduced to Cu(0) in the r-CuCl2-1 h and r-CuCl2-3 h catalysts after

reduction in H2. The aggregation of copper particles were

Fig. 7. Effects of the metal loadings on the metal portion in the outermost shell of the catalyst particles.

Fig. 8. Effects of surface Cu portion of 2 wt.% Cu/TiO2catalysts on the

methanol yields after 30 h irradiation.

(6)

also found in r-CuAc2-8 h catalyst as shown inFig. 12. The

copper particles on the catalysts CuCl2-1 h, CuCl2-3 h and

CuAc2-8 h are easily aggregated and reduced to Cu(0). From

the above characterization and photo activity results, the H2

-reduced Cu/TiO2 catalysts perform poorly because either

Cu(0) states are formed or Cu(I) particles are aggregated. However, the activities of these three catalysts are still better than those of other catalysts. Besides the major Cu(I) state, the strong capability of electron reception also contributes to the methanol yield.

4. Conclusion

The chemical states and the location of Cu on TiO2play

important roles in the photoreduction of CO2. Isolated Cu(I)

is regarded as the primary active site for photoreduction. The distribution of Cu on/in the TiO2 particles is critical to

maximize the yield of methanol. A maximum methanol yield of1000 mmol/gcatal. was obtained with the 25 molar

% Cu of total loading located on the surface of TiO2

particles. The mobility of photoelectrons is also an impor-tant factor in enhancing photocatalysis. The photo activity of Ag/TiO2was lower than those of Cu/TiO2due to the strong

affinity between Ag clusters and photoelectrons.

Acknowledgements

The authors thank the National Science Council of Taiwan, the Republic of China, for financially supporting this research under Contract NSC92-ET-7-002-001-ET. The authors also thank Dr. Jyh-Fu Lee of the Wiggler 17C station of the Taiwan Synchrotron Radiation Center and Ms. Chaol-ing Lai of the Surface Analysis Lab at the National Taiwan University for their effort in instrumental analysis.

References

[1] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69.

[2] N. Serpone, D. Lawless, R. Khairutdinov, J. Phys. Chem. 99 (1995) 16646.

[3] N. Serpone, D. Lawless, R. Khairutdinov, J. Phys. Chem. 99 (1995) 16655.

I.-H. Tseng, J.C.-S. Wu / Catalysis Today 97 (2004) 113–119 118

Fig. 10. FT-EXAFS (Cu Ka) spectra of CuCl2-derived 2 wt.% Cu/TiO2.

Fig. 11. FT-EXAFS (Cu Ka) spectra of CuCl2-derived 2 wt.% Cu/TiO2

after H2reduced.

Fig. 12. FT-EXAFS (Cu-Ka) spectra of Cu(CH3COO)2derived-2 wt.% Cu/

(7)

[4] Z. Zhang, C.-C. Wang, R. Zakaria, J.Y. Ying, J. Phys. Chem. B 102 (1998) 10871.

[5] S. Ikeda, N. Sugiyama, B. Pal, G. Marci, L. Palmisano, H. Noguchi, K. Uosaki, B. Ohtani, Phys. Chem. Chem. Phys. 3 (2001) 267. [6] K.Y. Song, Y.T. Kwon, G.J. Choi, W.I. Lee, Bull. Korean Chem. Soc.

20 (1999) 957.

[7] A. Di Paola, E. Garcia-Lopez, S. Ikeda, G. Marci, B. Ohtani, L. Palmisano, Catal. Today 75 (2002) 87.

[8] Y. Sakata, T. Yamamoto, T. Okazaki, H. Imamura, S. Tsuchiya, Chem. Lett. 1998 (1998) 1253.

[9] M. Iwasaki, M. Hara, H. Kawada, H. Tada, S. Ito, J. Colloid Interface Sci. 224 (2000) 202.

[10] R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, Y. Taga, Science 293 (2001) 269.

[11] I.-H. Tseng, W.C. Chang, J.C.S. Wu, Appl. Catal. B: Environ. 37 (2002) 37.

[12] T. Ressler, J. Synchroton Radiat. 5 (1998) 118.

[13] X. Bokhimi, O. Novaro, R.D. Gonzalez, T. Lo´pez, O. Chimal, A. Asomoza, R. Go´mez, J. Solid State Chem. 144 (1999) 349. [14] P.A. Forsh, A.G. Kazanskii, H. Mell, E.I. Terukov, Thin Solid Films

383 (2001) 251.

[15] K.K. Chattopadhyay, J. Dutta, S. Chaudhuri, A.K. Pal, Diamond Relat. Mater. 4 (1995) 122.

[16] X. Bokhimi, A. Morales, O. Novaro, T. Lo´pez, O. Chimal, M. Asomoza, R. Go´mez, Chem. Mater. 9 (1997) 2616.

[17] M. Epifani, C. Giannini, L. Tapfer, L. Vasanelli, J. Am. Ceram. Soc. 83 (2000) 2385.

[18] M. Vollmer, U. Kreibig, Optical Properties of Metal Clusters, Springer-Verlag, Berlin, Germany, 1995.

[19] L. Kundakovic, M. Flytzani-Stephanopoulos, Appl. Catal. A: Gen. 183 (1999) 35.

[20] E. Pelizzetti, M. Visca, in: M. Gratzel (Ed.), Energy Resources through Photochemistry and Catalysis, Academic Press, New York, 1983, p. 263, (Chapter 8).

[21] M.A.A. Schoonen, Y. Xu, D.R. Strongin, J. Geochem. Expl. 62 (1998) 201.

[22] H. Yoneyama, Catal. Today 39 (1997) 169.

[23] A. Fujishima, T.N. Rao, D.A. Tryk, J. Photochem. Photobiol. C. 1 (2000) 1.

[24] S. Ikeda, N. Sugiyama, B. Pal, G. Marcu, L. Palmisano, H. Noguchi, K. Uosaki, B. Ohtani, Phys. Chem. Chem. Phys. 3 (2001) 267. [25] B. Douglas, D. McDaniel, J. Alexander, Concepts and Models of

Inorganic Chemistry, third ed., John Wiley and Sons, New York, 1994, Appendix E.

[26] V. Vamathevan, H. Tse, R. Amal, G. Low, S. McEvoy, Catal. Today 68 (2001) 201.

[27] H. Yoneyama, S. Haga, J. Phys. Chem. 93 (1989) 4833.

[28] R. Janes, M. Edge, J. Rigby, D. Mourelatou, N.S. Allen, Dyes Pigments 48 (2001) 29.

[29] M. Anpo, M. Matsuoka, K. Hanou, H. Mishima, H. Yamashita, H.H. Patterson, Coord. Chem. Rev. 171 (1998) 175.

[30] L.-S. Kau, K.O. Hodgson, E.I. Solomon, J. Am. Chem. Soc. 111 (1989) 7103.