Photocatalytic reaction was followed in the same way as described in our previous papers [14] . We used a 15 W germicide lamp (type GCL 307T5L/CELL, Lighttech Ltd., Hungary), which emits predominantly in the wavelength range of 250 - 440 nm, its maximum intensity is at 254 nm. For the visible photocatalytic experiments another type of lamp was used (Lighttech GCL 307T5L/GOLD) with 400 - 640 nm wavelength range and two maximum intensities at 453 and 545 nm. Note that this lamp also emits below 400 nm. The approximate light intensity at the catalyst films is 3.9 mW/cm² for the germicide lamp and 2.1 mW/cm² for the other lamp. The photoreactor (volume: 670 ml) consists of two concentric quartz glass tubes fitted one into the other and a centrally positioned lamp. It is connected to a gas-mixing unit serving for the adjustment of the composition of the gas or vapor mixtures to be photolyzed in situ. The carrier gas was Ar, which was mixed with DME (~1.5%, 330 μmol). The DME/water-containing Ar flow entered the reactor through an externally heated tube to avoid condensation. The gas-mixture was circulated by a diaphragm pump. The reaction products were analyzed with a HP 5890 gas chromatograph equipped with PORAPAK Q and PORAPAK S packed columns. The sampling loop of the GC was 500 μl. The amount of all products was related to this loop. The conversion of DME was mainly calculated taking into account the amount of DME consumed.

For FTIR studies a mobile IR cell housed in a metal chamber was used [14] . Samples were illuminated by the full arc of a Hg lamp (LPS-220, PTI) outside the IR sample compartment. The filtered light passed through a high-purity CaF₂ window into the cell. Infrared spectra were recorded with a Biorad (Digilab. Div. FTS 155) instrument. All the spectra presented in this study are difference spectra. The surface area of the catalysts was determined by BET method with N₂ adsorption at ~100 K. The dispersion of metals was determined by the adsorption of H₂ at room temperature.

TiO₂ of different origins was used: Hombikat, UV 100 (pure anatase, 300 m²/g), TiO₂ nanowire (60 m²/g) and nanotube (40 m²/g). The synthesis of the last two compounds is described elsewhere [30] . For the preparation of N-doped samples TiO₂ was reacted with NH₃ [31] [32] . Titanium tetrachloride was used as a precursor. After several steps the NH₃-treated TiO₂ slurry was vacuum dried at 353 K for 12 hr, followed by calcination at 723 K in flowing air for 3 hr. This TiO₂ is noted with “SX”. The surface area of TiO₂ prepared in this way is 265 m²/g and that of N-doped oxide is 79 m²/g. The nitrogen content of this sample is 2.9%. The bandgaps of these TiO₂ samples have been evaluated in our previous work [14] . We obtained 3.02 eV for pure TiO₂ and 1.98 eV for N-doped TiO₂. Metal-promoted TiO₂ samples were prepared by impregnating pure or doped TiO₂ with the solution of metal compounds to yield a nominal 2 wt% metal. The following salts of Pt metals were used: Pd(NO₃)₂, H₂IrCl₆, RhCl₃∙3H₂O, H₂PtCl₆∙6H₂O and RuCl₃∙3H₂O. For IR studies the samples were pressed in self-support- ing wafers (30 × 10 mm ~10 mg/cm²). For photocatalytic measurements the sample (70 - 80 mg) was sprayed onto the outer side of the inner tube from aqueous suspension. The surface of the catalyst film was 168 cm². The catalysts were oxidized and reduced at 573 K in the IR cell or in the catalytic reactor for 1 hr. DME was the product of Gerling Holz Co. with purity of 99.9%.

The primary aim of IR study is to ascertain the development of adsorbed complexes formed on the effect of illumination on TiO₂, and to establish the influence of metal deposition on these features. Exposing pure TiO₂ to DME at 300 K resulted in a development of intense absorption bands in the C-H stretching region at 2950, 2921, 2878, 2842 and ~2830 cm⁻¹ (Figure 1). In the low frequency range strong bands appeared at 1459, 1253, 1159 and ~1063 cm⁻¹. Weaker bands or shoulder were also traced at ~1592 and ~1384 cm⁻¹. Illumination of the adsorbed DME caused only very slight changes in the high frequency range, but led to the slow attenuation of the bands at 1459 and 1253 cm⁻¹. At the same time the absorption peaks at ~1585 and 1366 cm⁻¹ increased in intensity. These spectral features were also observed on the IR spectra of metal/TiO₂ samples (Figure 2). In the high frequency range a pair of strong absorption bands at 2948 - 2952 and 2833 - 2838 cm⁻¹ of almost same intensity became the dominant spectral features for all M/TiO₂ catalysts. In the low frequency range the absorption features at ~1458, 1253, 1153 and 1047 - 1056 cm⁻¹ were found. We obtained similar spectral features for Ir/TiO₂ (not presented). The effect of the illumination on the IR spectra was almost the same as observed for pure TiO₂. The difference was the appearance of a shoulder at 2936 cm⁻¹ and the development of an intense CO band between 2001 - 2078 cm⁻¹, which grew slightly with the progress of illumination. In the low frequency range new strong absorption features appeared at ~1570 - 1574 and ~1357 cm⁻¹. IR bands observed on different samples and their assignments are presented in Table 1.

The photocataytic decomposition of DME has been investigated on different TiO₂ samples. Whereas DME does not decompose at 300 K on pure TiO₂, illumination induced the occurrence of the reaction to give H₂ and CO₂.

(g) gaseous; (a) adsorbed.

However even on the most effective TiO₂ (Hombikat) the extent of the decomposition was very low, about ~2% - 3%, in 210 min. The photocatalytic effect of TiO₂ nanowire and nanotubes was also tested: we obtained a similar low photoactivity.

The deposition of Pt metals on TiO₂ (Hombikat) markedly enhanced its photoactivity. In Figure 3, we displayed the conversion of DME and the amount of products formed on various catalysts as a function of illumination time. On the most active Rh/TiO₂, the conversion of the decomposition of DME attained ~22% in 210 min. On the less active Ru/TiO₂ it was only ~6.5%. The main products were H₂ and CO₂. A small amount of CO and methyl formate was also formed. A trace of formaldehyde was also detected. The CO/H₂ ratio varied between ~0.023 - 0.044. The formation of methyl formate deserves a special attention. Its amount increased with the illumination time (Figure 4(a)). The largest quantity was measured for Rh/TiO₂ and the lowest one for Pt/TiO₂. When its amount was related to that of H₂ produced, we obtained the highest value for Ru/TiO₂. The ratio of methyl formate/H₂ slowly decreased with the duration of illumination (Figure 4(b)). Some important data for the photocatalytic decomposition of DME are presented in Table 2. Based on the conversion data, the activity order of metals was as follows: Rh, Pd, Ir, Pt and Ru. When the rate of H₂ production is related to the dispersity of the metals, we obtained slightly different order: Pt, Rh, Pd, Ir and Ru.

In order to judge the contribution of thermal effect for the photoreaction, we also examined the thermal reaction on selected catalysts. A measurable reaction on Pt/TiO₂ and Rh/TiO₂ was observed only at 523 K. Attaching a thin thermocouple in the catalyst layer indicated only a temperature rise of only a few degrees during illumination. The results of these control experiments lead us to exclude the contribution of thermal effects to the photodecomposition of DME induced by illumination.

Table 2Table 2

. Some characteristic data for the photolysis of DME on metal-promoted TiO₂

Samples	Dispersion (%)	Conversion (%, 210 min)		Methyl formate (nmol, 210 min)	CO/H2 (210 min)	MF/H2 (210 min)
2% Rh/TiO2	16	22.5	0.063	5.9	0.023	0.052
2% Pd/TiO2	26	21.0	0.042	4.2	0.039	0.042
2% Ir/TiO2	54	12.5	0.024	1.6	0.044	0.028
2% Pt/TiO2	13	12.0	0.096	0.4	0.039	0.007
2% Ru/TiO2	6	6.5	0.023	2.0	0	0.142

= the amount of H₂ formed in 210 min related to the number of metal atoms.

The effect of illumination on the reforming of DME was also investigated. Water exerted a positive influence on the conversion of DME and it appreciably increased the amount of H₂ formed. As the hydrolysis of DME to CH₃OH occurs more easily on the acidic centers of Al₂O₃, some experiments have been performed in the presence of Al₂O_3. Adding Al₂O₃ to Pd/TiO₂ catalyst greatly enhanced the conversion of DME and the formation of H₂. Table 3 contains some characteristic data.

Some experiments have been performed in visible light. These measurements were carried out with TiO₂ (SX), which possessed better performance compared to other N-doped TiO₂ [14] . As the surface area of TiO₂ is markedly lowered by doping with N, the data presented in Figure 5 are related to unit surface area. The results clearly show that whereas pure TiO₂ exhibits very little activity in the visible light, the photoactivity of N-doped sample (SX) is appreciably higher. Similar features were experienced for metal-promoted TiO₂. Figure 6 depicts the photocatalytic effects of three selected metals deposited on pure and N-doped TiO₂ (SX). A comparison immediately reveals that the photoactivity of the metals on N-doped sample is markedly higher than that of M/TiO₂ free of nitrogen. This is reflected in the conversion of DME and in the amounts of the products formed in the photo-induced decomposition.

Adsorption of DME on TiO₂ at 300 K produced several intense absorption bands in the IR spectra. Taking into account the results of previous studies their possible assignment is presented in Table 1. In the high frequency range a pair of strong absorption bands at ~2952 and 2838 cm⁻¹ of almost same intensity became the dominant spectral features for pure and metal-promoted TiO₂ catalysts. These absorption bands are the characteristic vibration of molecularly bonded DME. The dissociation of DME to CH₃O species is indicated by the appearance of another pair of bands at 2936 and 2879 cm⁻¹. The intensities of all these bands underwent a slight attenuation as a result of illumination. A more striking effect of photolysis of adsorbed DME is the appearance of asymmetric stretch of formate bands at ~1570 cm⁻¹, and the development of CO band between 2000 - 2101 cm⁻¹ on TiO₂-supported metals. The fact that we observed the same spectral features for TiO₂ and for M/TiO₂ samples indicates that both the adsorbed DME and the CH₃O species are located on the TiO₂ surface. The appearance of CO band in the IR spectra of metal/TiO₂ samples, however, suggests that the metals can initiate the decomposition of these compounds very likely resided at the metal/oxide interface. An interesting feature of the IR spectra is the absence of dicarbonyl species (M⁺(CO)₂). This possible reason is that hydrogen formed in the photoreactions prevents the oxidative disruption of metal particles leading to the formation of M⁺(CO)₂ surface complex [36] .

The formation of methoxy species suggests the breakage of one of the C-O bonds in the adsorbed DME

As there it is no indication of the IR bands of adsorbed CH₃ radical [37] , it is very likely that the CH₃ has been attached to the oxygen atom of TiO₂ also yielding a Ti-OCH₃ surface compound. Accordingly, instead of step (1), we can count with the reaction of DME with the OH groups of TiO₂

Illumination of adsorbed layer resulted in a slow attenuation of methoxy bands and the appearance of absorption features due to formate species. Its formation is described by the following elementary steps

DME proved to be very resistant towards illumination on TiO₂. Deposition of Pt metals on the TiO₂, however, enhanced its photoactivity, but the low reactivity of DME appeared on these catalysts, too. The effect of illumination can be explained by the donation of photoelectrons formed in the photo-excitation process

to the CH₃O species:

producing a more reactive negatively charged species, which is converted into adsorbed CH₂O and HCOO (Equations (3) and (4)). However, even the photo-induced reaction occurred to only a very limited extent on pure TiO₂, a finding which can be attributed to the fast recombination of the electrons and holes formed in the photo-excitation process (Equation (5)). The formation of H₂, CO₂ and CO suggests the occurrence of the reactions

An interesting and somewhat surprising result of the photocatalytic decomposition of DME is the formation of methyl formate. This compound has been considered as a precursor in the preparation of several materials [38] . Methyl formate is mainly synthesized by dehydrogenation of methanol over Cu-based catalyst at higher temperatures. However, recent works showed that it is also formed in the photocatalytic oxidation [39] and decomposition of methanol on polycrystalline TiO₂ at room temperature [40] . Its production was markedly increased when Pt metals were deposited on TiO₂ [40] . The highest yield of methyl formate was measured for Pt/TiO₂ (62.2) and the lowest one for Ru/TiO₂ (26.0). Recent studies performed under UHV conditions on preoxidized TiO₂(110) disclosed that methyl formate is produced from the photo-oxidation of methanol even at ~200 K [41] . The finding that methyl formate is produced in the photocatalytic decomposition of DME further supports the idea that CH₃O species is involved in its photoreaction The formation of methyl formate can be ascribed to the recombination of CH₂O formed in the dissociation of CH₃O (Equation (3)):

or by the reaction of CH₂O with a further CH₃O species:

An appreciable increase in the extent of photolysis of DME was observed in the presence of H₂O, which can be attributed to the occurrence of the hydrolysis of DME,

e.g. to the formation of more reactive CH₃OH. The addition of H₂O to DME also lowered the extent of CO formation, very likely due to the occurrence of the water-gas shift reaction promoted by illumination. This was confirmed by a separate experiment. Mixing Pd/TiO₂ with Al₂O₃ further enhanced the formation of H₂, which can be also ascribed to the promotion of the hydrolysis of DME to methanol.

The deposition of metals onto TiO₂ greatly improved the photocatalytic effect of the TiO₂. We assume that the CH₃O species formed at the metal/TiO₂ interface is much more reactive than that located on TiO₂. The promoting effect of Pt metals deposited on TiO₂ is generally explained by the better charge carrier separation induced by illumination [12] [13] [42] . In addition we assume that the occurrence of an electronic interaction between n-type TiO₂ and Pt metals is also important. The role of the electronic interaction between metals and TiO₂ has been first demonstrated in the catalytic decomposition of formic acid on Ni deposited on pure and doped TiO₂ [43] [44] . As far as we are aware, TiO₂ was first used as a support in this case [43] . As the work function of TiO₂ (~4.6 eV) is less than that of Pt metals (4.7 - 5.7 eV), electron transfer is expected to occur from TiO₂ to the deposited metals, which increases the activation of adsorbed molecules. We assume that illumination enhances the extent of this electron transfer at the interface of the two solids, leading to increased decomposition.

An important finding of this work is that the incorporation of N into TiO₂ support enhanced the photoactivity of TiO₂ (Figure 5), and particularly that of M/TiO₂ catalysts (Figure 6) and led to the decomposition of DME in the visible light, too. This can be attributed to the lowering of the bandgap of TiO₂.

As we studied the photocatalytic decomposition of several organic compounds on the same catalysts under exactly identical experimental conditions, this gives us a possibility to make a comparison. Some data are presented in Table 4. It shows that HCOOH is the most reactive compound both on TiO₂ and Rh/TiO₂. H₂ was also formed with highest selectivity and yield in the photocatalytic decomposition of HCOOH. Relatively high yield for H₂ formation was obtained in the decomposition of C₂H₅OH on Rh/TiO₂.