Control of the Metal−Support Interface of NiO-Loaded Photocatalysts

2334

Langmuir 2006, 22, 2334-2339

Control of the Metal-Support Interface of NiO-Loaded Photocatalysts via Cold Plasma Treatment Ji-Jun Zou,† Chang-Jun Liu,*,† and Yue-Ping Zhang‡ Key Laboratory for Green Chemical Technology, School of Chemical Engineering and Technology and Department of Chemistry, Tianjin UniVersity, Tianjin 300072, People’s Republic of China ReceiVed August 4, 2005. In Final Form: December 18, 2005 NiO-loaded semiconductors have been extensively used as the photocatalysts for water splitting. The metal-support interface is an important factor affecting the efficiency. In the present work, the pretreatment methods were studied to produce a more desirable metal-support interface using Ta2O5 and ZrO2 as the support. The traditional method includes a thermal decomposition, reduction at 773 K, and oxidation at 473 K (R773-O473). The thermal decomposition of Ni(NO3)2 makes the Ni atoms migrate into the bulk of the supports, resulting in a diffused interfacial region. Alternatively, a cold plasma treatment was used to replace the thermal decomposition. Metal salts are quickly decomposed by glow discharge plasma treatment at room temperature, avoiding the thermal diffusion of Ni atoms. With the sequent R773-O473 treatment, a clean metal-support interface is produced. Moreover, the metal particles have optimal shapes with a larger surface. In photocatalysis, the clean metal-support interface is more favorable for the charge separation and transfer, and the increased metal surface provides more active sites. NiO/Ta2O5 and NiO/ZrO2 prepared with the plasma treatment exhibit higher activity for photocatalytic hydrogen generation from pure water and methanol solution, respectively. This work shows the potential of cold plasma treatment in the preparation of metal-loaded catalysts and nanostructured materials.

Introduction Photocatalytic water splitting has attracted much attention because it provides an ideal way to utilize solar energy for hydrogen generation.1-7 Water splitting is an uphill reaction with a large positive change in the Gibbs free energy. The function of a semiconductor is to convert photoenergy into chemical energy to drive this reaction. The basic requirement for the semiconductors is that their conduction band should be more negative than the reduction potential of H2O and their valence band should be more positive than the oxidation potential of H2O. For example, Ta2O5 and ZrO2 are such oxides.1,2 Naked semiconductors generally exhibit low efficiency because their surface is not active enough to generate H2. The loading of metals can introduce more active H2-generating sites. Photocatalytic reactions require the cooperation of the supports (semiconductors) and the loaded metals. The efficiency can be improved either by synthesizing better supports or by modifying the properties of the metals. Many investigations have been conducted to explore more efficient supports. The semiconductors with perovskite, layered, and tunnel structure are promising photocatalysts.3-7 Their activity is greatly improved when NiO is loaded as the cocatalyst. Compared with other metals or oxides, NiO is the most effective in most cases, indicating that NiO is significantly important for water splitting.1-6 * To whom correspondence should be addressed. [email protected]. † School of Chemical Engineering and Technology. ‡ Department of Chemistry.

Water splitting over NiO-loaded photocatalysts can be explained as follows: (i) the semiconductors absorb photons and generate electron-hole pairs; (ii) the photoinduced electronhole pairs are separated with electrons in the conduction band and holes in the valence band; (iii) then the electrons are transferred to NiO particles; (iv) finally, H2 and O2 are produced over NiO and semiconductor surfaces, respectively. Step iii is difficult for many semiconductors because their conduction band is more positive or close to that of NiO. A metallic interlayer between the semiconductor and NiO makes the electron transfer easier. Therefore, a direct metal-support interface is desirable for the NiO-loaded photocatalysts. A thermal treatment, including thermal calcinations, reduction at 773 K, and oxidation at 473 K, has been frequently conducted with the aim to produce the favored metal-support interface.1-5 The metal-support interface is very important in regard to the photocatalytic water splitting. To our knowledge, however, few studies have been conducted to optimize the interface structure. Although the traditional treatment is extensively used, the present work will show that it may not work very well to produce the desired metal-support interface. Alternatively, a more desirable interface is produced with a cold plasma treatment. Cold plasmas have been extensively used to convert small stable molecules, to modify the surface of polymers, and to treat catalysts, but the detailed mechanism of the plasma treatment is still unclear.8-16 In the present work, the effect of the plasma treatment will be

E-mail:

(1) Takahara, Y.; Kondo, J. N.; Takata, T.; Lu, D.; Domen, K. Chem. Mater. 2001, 13, 1194. (2) Sayama, K.; Arakawa, H. J. Photochem. Photobiol., A 1994, 77, 243. (3) Domen, K.; Kudo, A.; Onishi, T.; Kosugi, N.; Kuroda, H. J. Phys. Chem. 1986, 90, 292. (4) Zou, Z.; Ye, J.; Sayama, K.; Arakawa, H. Nature 2001, 414, 625. (5) Ishihara, T.; Nishiguchi, H.; Fukamachi, K.; Takita, Y. J. Phys. Chem. B 1999, 103, 1. (6) Kato, H.; Asakura, K.; Kudo, A. J. Am. Chem. Soc. 2003, 125, 3082. (7) Ogura, S.; Inoue, Y. Phys. Chem. Chem. Phys. 2000, 2, 2449.

(8) Li, M.; Xu, G.; Tian, Y.; Chen, L.; Fu, H. J. Phys. Chem. A 2004, 108, 1687. (9) Zou, J.-J.; Zhang, Y.-P.; Liu, C.-J.; Li, Y.; Eliasson, B. Plasma Chem. Plasma Process. 2003, 23, 69. (10) Me¨dard, N.; Soutif, J.-C.; Poncin-Epaillard, F. Langmuir 2002, 18, 2246. (11) Weng, C.-C.; Liao, J.-D.; Wu, Y.-T.; Wang, M.-C.; Klauser, R.; Grunze, M.; Zharnikov, M. Langmuir 2004, 20, 10093. (12) Liu, C.-J.; Yu, K. L.; Zhu, X. L.; Zhang, Y.-P.; He, F.; Eliasson, B. Appl. Catal., B 2004, 47, 95. (13) Dittmar, A.; Kosslick, H.; Herein, D. Catal. Today 2004, 89, 177. (14) Zou, J.-J.; Liu, C.-J.; Yu, K.-L.; Cheng, D.-G.; Zhang, P.-Y.; He, F.; Du, H.-Y.; Cui, L. Chem. Phys. Lett. 2004, 400, 520. (15) He, J.; Ichinose, I.; Kunitake, T.; Nakao, A. Langmuir 2002, 18, 10005.

10.1021/la052135u CCC: $33.50 © 2006 American Chemical Society Published on Web 02/01/2006

NiO-Loaded Photocatalyst Metal-Support Interface

Langmuir, Vol. 22, No. 5, 2006 2335

discussed to show its advantages in the preparation of metalloaded catalysts and nanostructured materials. Experimental Section Preparation. Simple but typical semiconductors, Ta2O5 and ZrO2 powders (99.9%, Tianjin Guangfu Institute of Fine Chemicals), were chosen as the supports. The grain size was about 100 nm according to analyses using X-ray diffraction (XRD) and transmission electron microscopy (TEM) measurements. These powders were used as received without any further treatments. They were impregnated with Ni(NO3)2 (99.0%, Tianjin Bodi Chemical Co.) using wet impregnation. The amount of NiO loaded was 1.0 wt % if not specifically indicated. After being dried in air for 24 h, they were divided into two parts. One was treated with the traditional method, including calcination at 600 K for 1 h, reduction with H2 at 773 K for 2 h, and oxidation with O2 at 473 K for 1 h (R773-O473 treatment), as described by many other investigators.1-5 The other was treated with a plasma method. To do so, the sample was first treated with glow discharge plasma for 30 min. The glow discharge plasma apparatus applied has been previously reported.12,14 The glow discharge plasma was initiated by a high-dc-voltage generator (Tianda Cutting and Welding Setup Inc. Ltd., China) with argon as the plasmaforming gas. The gas temperature of the plasma was measured using an infrared camera (Ircon, 100PHT). The plasma-treated sample was subsequently treated with the R773-O473 treatment. In the following sections, the samples treated with the traditional method are designated as “C”, and those treated with the plasma method are designated as “P”. Characterizations. XRD patterns were obtained using a Rigaku D/Max-2500 V/PC diffractometer with Cu KR radiation. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Perkin-Elmer PHI-1600 spectrometer with Al KR (1486.6 eV) radiation. The binding energy was calibrated using the C1s peak (284.6 eV) of surface adventitious carbon. TEM observations were conducted using a Philips Tecnai G2 F20 electron microscope. Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were conducted using a Shimadzu TGA 50 instrument with a rate of temperature increase of 10 K/min. Optical absorption spectra were recorded using a Shimadzu UV-2100S spectrometer with BaSO4 as the reference. Activity Evaluation. The photocatalytic reaction was conducted in a gas-closed system. A 0.5 g sample of catalyst was dispersed in 350 cm3 of water with a magnetic stirrer in a reactor made of quartz. The light source was a 300 W high-pressure Hg lamp (Shanghai Yaming, GGZ-300). Prior to irradiation, the system was deaerated and Ar was introduced with a pressure of 20 kPa. The amounts of H2 and O2 produced were determined by a mass spectrometer (AVIGmbH, Omnistar). Since the water splitting involves two reactions, H2 generation and O2 generation, the activities for H2 generation from methanol solution (42.8 vol %) and O2 generation from AgNO3 solution (0.1 N) were also evaluated in the same system.

Results and Discussion Core-Shell Structure of Nickel Particles. Figure 1 shows the XRD patterns of NiO/Ta2O5 prepared with both methods. The peaks at 2θ ) 44.7°, 37.8°, and 43.0° are assigned to the Ni(011), NiO(111), and NiO(200) phases, respectively. This suggests that the nickel particle is a combination of metal phase and oxide phase. The peak of Ni is more intense than that of NiO, indicating that the metallic Ni is the dominant phase. Ni2p XPS spectra were analyzed to determine the status of the nickel particles. The sample was pressed into a flat slice, and the spectra were collected with the photoelectrons taking off from the surface at takeoff angles of 30°, 60°, and 90°. The probing (16) Boyen, H.-G.; Ka¨stle, G.; Weigl, F.; Koslowski, B.; Dietrich, C.; Ziemann, P.; Spatz, J. P.; Riethmu¨ller, S.; Hartmann, C.; Mo¨ller, M.; Schmid, G.; Garnier, M. G.; Oelhafen, P. Science 2002, 297, 1533.

Figure 1. XRD patterns of (a) NiO/Ta2O5-C and (b) NiO/Ta2O5P. The unmarked peaks belong to Ta2O5.

Figure 2. Angle-resolved Ni2p XPS spectra of NiO/Ta2O5-C.

depth should be increased with an increase of the takeoff angle.17 The spectrum at 30° gives information at the uppermost layer, while the spectrum at 90° reflects the chemical status in the bulk. As shown in Figure 2, the Ni2p spectrum acquired at 30° contains two peaks centered at 854.1 and 855.8 eV that have been assigned to Ni2+ of NiO and Ni(OH)2, respectively.3 When the takeoff angle is increased to 60°, a new peak assigned to metallic Ni0 appears at 852.9 eV. This peak is more prominent at 90°. With an increase of the takeoff angle, the percentage of Ni2+ decreases whereas that of metallic Ni0 gradually increases. So the uppermost surface of the nickel particle is oxidized, and the bulk is metallic. The R773-O473 treatment produces a metallic core with an external oxidized shell. Structure of the Metal-Support Interface. Figure 3a shows the typical TEM image of NiO/Ta2O5-C. The nickel particle is partly embedded in the Ta2O5 crystal, indicating that nickel atoms at the interface have migrated into the bulk of the support probably because of thermal diffusion. The depth of this region is about 10 nm. The composition of this region (point A in Figure 3a) was analyzed with energy-dispersive X-ray (EDX) as shown in Figure 3b. Ta, O, and Ni lines are observed, and the atomic ratio (17) Tunc, I.; Suzer, S.; Correa-Duarte, M. A.; Liz-Marza´n, L. M. J. Phys. Chem. B 2005, 109, 7597.

2336 Langmuir, Vol. 22, No. 5, 2006

Zou et al.

Figure 3. (a) TEM image of NiO/Ta2O5-C, (b) EDX analysis at point A, and (c) TEM image of NiO/ZrO2-C.

of Ta to Ni is about 2/1. Meanwhile, only Ni and O lines were observed in the bulk of the nickel particle and only Ta and O lines in that of the support. This confirms that the interfacial region of the support contains homogeneously diffused Ni atoms. So a diffused interfacial region, not a metal-support interface, is produced. The typical TEM image of NiO/ZrO2-C is shown in Figure 3c, and a similar diffused interfacial region is observed. Ni atoms in this region share oxygen atoms with Ta (or Zr) atoms and are in oxidized states, which produces an additional NiO layer between the support and metallic Ni. The traditional treatment has been extensively used and was expected to produce a favored metal-support interface. However, a diffused interfacial region is produced that is different from the expected one. Figure 4 shows the TEM images of NiO/Ta2O5-P and NiO/ ZrO2-P. The interface structure is dramatically different from that of NiO/Ta2O5-C and NiO/ZrO2-C. A very thin junction between the support and the loaded metal is observed, indicating that nickel atoms do not diffuse into the bulk of the supports. The nickel particles are loaded onto the surface of the supports with a clean metal-support interface. Evidently, the plasma method produces a favored clean metal-support interface that is important for water splitting, as addressed below. TEM images mentioned above also show different morphologies of nickel particles when they are treated with different methods. The nickel particles treated with the traditional method have a spherical shape with a diameter of 30-40 nm (Figure 3). For the samples treated with the plasma method (Figure 4), the nickel particles are half-ellipsoidal. The length along the metalsupport interface orientation is 20-30 nm, 2 times longer than that along the perpendicular direction (10-15 nm). This provides a higher metal coverage on the support with a larger metal surface. The Ni2p XPS profiles of NiO/Ta2O5-P and NiO/ZrO2-P are more intense than those of NiO/Ta2O5-C and NiO/ZrO2-C, respectively. The corresponding surface atomic compositions are shown in Table 1. The plasma-prepared samples show higher nickel surface concentrations and larger atomic ratios, which are in good agreement with larger metal coverage. It can be concluded that different methods produce different interfacial structures and different particle morphologies. The

Figure 4. TEM images of (a) NiO/Ta2O5-P and (b) NiO/ZrO2-P.

traditional method does not work very well to produce the desired interface, although it is extensively used. Meanwhile, the plasma method can control the metal-support interface and optimize the shape of the metal particle. The two types of interfacial structures are shown in Figure 5. Control of the Metal-Support Interface. It is the difference between the treatment methods that results in different interfacial structures. The traditional method includes thermal calcination and an R773-O473 treatment. The thermal calcination is to decompose Ni(NO3)2 into NiO and the R773-O473 treatment is to produce the core-shell structure. The plasma method also involves the R773-O473 treatment but uses the plasma treatment

NiO-Loaded Photocatalyst Metal-Support Interface

Langmuir, Vol. 22, No. 5, 2006 2337

Table 1. Surface Composition and Atomic Ratio Determined Using XPS surface atomic compositiona (%)

surface atomic ratio

sample

Ni

O

Zr (Ta)

Ni/Zr (Ta)

Ni/O

NiO/ZrO2-P NiO/ZrO2-C NiO/Ta2O5-P NiO/Ta2O5-C

2.5 1.1 1.6 1.2

37.9 33.8 33.0 29.0

15.0 13.1 10.4 9.9

0.17 0.08 0.15 0.12

0.07 0.03 0.05 0.04

a

The composition of surface adventitious carbon is not shown.

Figure 7. Photographs of untreated (left) and plasma-treated (right) Ni(NO3)2/ZrO2 (5 wt % NiO).

Figure 5. Schematic interfacial structures produced by the (a) traditional method and (b) plasma method.

Figure 8. XPS spectra of untreated and plasma-treated Ni(NO3)2/ ZrO2 (5 wt % NiO): (a) survey spectra, (b) Ni2p spectra.

Figure 6. TEM images of calcined (a) Ni(NO3)2/Ta2O5 and (b) Ni(NO3)2/ZrO2 and plasma-treated and calcined (c) Ni(NO3)2/Ta2O5 and (d) Ni(NO3)2/ZrO2.

instead of thermal calcination. In comparison, the formation of the diffused interfacial region is obviously due to the thermal calcination. The following analysis confirms that the thermal calcination generates the diffused regions. To do so, the supports were impregnated with Ni(NO3)2, calcined at 600 K for 1 h, and then characterized using TEM. Figure 6a,b shows that the diffused interfacial regions have been produced. This confirms that it is the thermal decomposition that leads to the thermal diffusion of Ni atoms. When the samples were treated with plasma instead of thermal calcination, no diffusion of Ni atoms was observed. Even with additional thermal calcination at 600 K for 1 h, the plasma-treated samples still show a clean and direct interface without a diffused interfacial region (Figure 6c,d), indicating

that the plasma treatment can effectively avoid the thermal diffusion of Ni atoms. Effect of Plasma Treatment. The supports impregnated with Ni(NO3)2 (5.0 wt % NiO) are green. They become gray after the plasma treatment as shown in Figure 7, suggesting that nickel salts are decomposed into oxides. XPS analyses were conducted to determine the change in chemical state as shown in Figure 8. A N1s peak at 406.8 eV belonging to NO3- appears before the plasma treatment but disappears after that. The Ni2p peak of Ni(NO3)2 is at 856.6 eV. After the plasma treatment, this peak shifts to 855.2 eV assigned to the combined peak of NiO and Ni(OH)2. This demonstrates that Ni(NO3)2 is decomposed into oxides with the plasma treatment. TGA and DTA were carried out to study the change in thermal state as shown in Figure 9. Ni(NO3)2/ZrO2 exhibits a large loss in weight due to the thermal decomposition of Ni(NO3)2 when heated from 300 to 1000 K. The total loss in weight is 7.2%, in good agreement with the loss caused by the thermal decomposition of Ni(NO3)2 (5.0 wt % NiO). Correspondingly, an intense endothermic peak at 600 K is observed. However, the plasma-treated sample shows neither

2338 Langmuir, Vol. 22, No. 5, 2006

Figure 9. (a) TGA and (b) DTA of untreated and plasma-treated Ni(NO3)2/ZrO2 (5 wt % NiO).

Zou et al.

Figure 11. Photocatalytic activity for water splitting: solid lines, H2; dotted lines, O2.

Figure 10. IR thermal image of glow discharge plasma.

an obvious loss in weight nor an endothermic peak, indicating that no decomposition occurs when it is heated. The plasmatreated sample no longer changes when heated in N2, reconfirming that Ni(NO3)2 has been previously decomposed by the plasma treatment. The IR thermal image in Figure 10 shows that the gas temperature of glow discharge plasma is below 320 K, much lower than the temperature required for the thermal decomposition of Ni(NO3)2. The plasma treatment decomposes Ni(NO3)2 at room temperature, thus avoiding the thermal diffusion of Ni atoms. This also suggests that the glow discharge plasma treatment can be used to decompose other metal salts or organic polymers. Especially, many nanostructured materials are thermally sensitive and thermal calcinations may ruin their original structure or surface states. In these cases, the plasma treatment is a better alternative because it can avoid these side effects. The optimization of the metal particle shape caused by the plasma treatment is also very useful for some shape-sensitive catalysts.18,19 Activity. Figure 11 shows the activity of photocatalytic water splitting. The catalysts with the plasma treatment exhibit higher activity than those with the traditional treatment. The reaction rate over NiO/Ta2O5-P and NiO/ZrO2-P is 1.7 and 1.5 times higher than that over NiO/Ta2O5-C and NiO/ZrO2-C, respectively. (18) Haruta, M. CATTECH 2002, 6, 102. (19) Bowker, M.; James, D.; Stone, P.; Bennett, R.; Perkins, N.; Millard, L.; Greaves, J.; Dickinson, A. J. Catal. 2003, 217, 427.

Figure 12. Optical absorbance of photocatalysts prepared by different methods.

Figure 12 shows the optical spectra of the catalysts prepared with both methods. The maximum wavelength (λ) that can excite the photocatalysts is calculated using the following equation:20

λ (nm) ) 1239.9/[E (eV)] E is the band gap of the semiconductors (4.0 eV for Ta2O5 and 5.0 eV for ZrO2).2 The calculated λ is 310 and 245 nm for the Ta2O5- and ZrO2-based photocatalysts, respectively. These catalysts have similar absorbance in the active region despite the preparation method, indicating that they generate the same amounts of charges under irradiation. The high activity of the plasma-treated samples is due to more efficient charge separation and transfer and a more active surface. First, the diffused interfacial region is not as effective as the clean metal-support interface for the charge separation. The diffused Ni atoms incorporate into the crystal of semiconductors in a substitutional or in an interstitial manner, making the semiconductors partly doped. It has been reported that TiO2 doped with Ni2+ exhibits decreased efficiency.21 It is therefore concluded that the doping of semiconductors with lower valence cations (M2+ and M3+) anodically shifts the flat-band potential of semiconductors, resulting in a (20) Nguyen, T.-V.; Yang, O.-B. Catal. Today 2000, 87, 69. (21) Chio, W.; Termin, A.; Hoffmann, M. R. J. Phys. Chem. 1994, 98, 13669.

NiO-Loaded Photocatalyst Metal-Support Interface

Langmuir, Vol. 22, No. 5, 2006 2339

Table 2. Rate of Hydrogen and Oxygen Generation rate of gas evolution (µmol/h) sample

H2 from CH3OH solutiona

O2 from AgNO3 solutionb

NiO/Ta2O5-P NiO/Ta2O5-C NiO/ZrO2-P NiO/ZrO2-C

299 222 172 138

24 23 19 18

a The rate is an average of a 10 h reaction. b The rate is an average of a 6 h reaction.

thicker depletion layer at the proximity of the semiconductorwater interface and a lower in-built electric field.22 This weakens the capability of separating the photoinduced electron-hole charges and thus reduces the reaction efficiency. So the diffused interfacial region is undesirable for the charge separation, while the clean metal-support interface has no such negative effect. Second, the clean interface structure is ideal for electron transfer from the semiconductor to the metal particles. Finally, the smaller size and flatter shape of the metal particles produce a larger metal surface and thus more active sites for H2 generation. The rates of H2 evolution from methanol solution and O2 generation from AgNO3 solution are shown in Table 2. The samples with the plasma treatment show a higher rate for hydrogen evolution, but the rate for oxygen evolution is identical for the two types of catalysts. IR spectroscopic studies have demonstrated that the photoinduced holes can be rapidly trapped by surface adsorbates but the electrons have a relatively long lifetime.23 When an electron acceptor such as Ag+ appears in the solution, the photoinduced electrons are quickly trapped and holes remained for water oxidation. For the H2 generation from methanol solution, (22) Karakitsou, K. E.; Verykios, X. E. J. Phys. Chem. 1997, 97, 11184. (23) Yamakata, A.; Ishibashi, T.; Kato, H.; Kudo, A.; Onishi, H. J. Phys. Chem. B 2003, 107, 14383.

the electron transfer is the rate-determining step. The catalysts with the clean metal-support interface show higher activity, confirming that this interface is more favorable.

Conclusions A plasma method has been used to control the metal-support interface of NiO-loaded photocatalysts. With this plasma method, Ni(NO3)2 is decomposed at room temperature by glow discharge plasma treatment instead of thermal calcination, avoiding the thermal diffusion of nickel atoms. A clean and direct metalsupport interface, different from the diffused interfacial region caused by thermal calcination, is ultimately obtained. Moreover, the surface of the metal particles is increased with a smaller particle size and with a half-ellipsoidal shape. In photocatalytic reactions, this interface is more efficient for the charge separation and transfer. The increased metal surface provides more active sites. NiO/Ta2O5 and NiO/ZrO2 prepared with the plasma treatment show 1.7 and 1.5 times higher activity for photocatalytic water splitting than those obtained with traditional treatment, respectively. The plasma treatment quickly decomposes metal salts or modifies the surface of materials at low temperature, and thus is a promising alternative to thermal calcination to avoid the negative thermal effect. The present work has shown promising perspectives in optimizing the metal-support interface and particle shape of metal-loaded catalysts and also in modifying thermosensitive nanostructured materials. Acknowledgment. Part of the instruments and equipment were donated by ABB Switzerland Ltd., which is greatly appreciated. This work was supported by the National Natural Science Foundation of China (Grant 20225618) and the Research Foundation for Doctorial Program of the Ministry of Education of China (Grant 20030056033). LA052135U