Comparison of the photocatalytic efficiency of TiO2, iron oxides and mixed Ti(IV)Fe(III) oxides: photodegradation of oligocarboxylic acids

183

.I Photochem. Photobiol. A: Chem., 84 (1994) 183-193

Comparison of the photocatalytic efficiency of Ti02, iron oxides and mixed Ti(IV)-Fe(II1) oxides: photodegradation of oligocarboxylic acids Marta

I. Litter+

Depatiamento

Quimica de Reactores, Comision National de Enegi

At6mica, Av. de1 Liberiador 8250, 1429 Buenos Aires (Argentina)

A. Navio Imtitutode Ckrzcia de Jos6

Materiafes de Sevilla, Centro Mixto CSICVniversidad Facultad de Qulmica, Universidad de &villa, 41012 Se&a (Spain) (Received

November

2, 1993; accepted

de &villa, and Depanamento

de Quimica Inorgdinica,

March 24, 1994)

Abstract Comparative photo-oxidations of three oligocarboxylic acids, i.e. oxalic acid, EDTA and malonic acid, performed with different catalysts, i.e. TiOa (Degussa P-25), a-FeO,, y-Fe20s, FeIO., and two samples of Fe(IlI)-Ti(IV) oxides containing 0.5 wt.% and 5 wt.% Fe respectively. Different irradiation wavelengths used. It was found that the photocatalytic efficiency depends mainly on the physicochemical properties catalyst, the irradiation wavelength, and the redox potential and adsorption ability of the substrate. The of surface complexes formed between the oxide and the organic compound was analysed. TiOs was found the best catalyst. The relative utility and stability of iron oxides and mixed Fe-Ti oxides is discussed.

1. Introduction

Extensive laboratory research into heterogeneous photocatalysis with powders or small particles of semiconductors has been carried out during the last 15 years, and has quickly become the basis of technological applications in energy production, organic synthesis or detoxification of polluted waters. Several reviews on these subjects were published recently (see refs. l-8 and references therein). TiO, is the most commonly used photocatalyst, owing to its chemical stability and low cost. Iron oxides (less extensively studied) have been considered promising for applications which make use of solar light, because iron oxides have a bandgap smaller than that of TiOz Q&-2.2 eV and 3.0 eV respectively) [A. However, most of the studies of these oxides refer to their corrosion in the presence of reducing agents, and photodissolution mechanisms have been proposed by several authors [9-231. The role of iron oxides as photocatalysts is still under discussion, although several examples have appeared in the literature [24-281. The pos‘Author

to whom correspondence

lOlO-6030/94/$07.00 0 1994 Elsevier SSl>I 1010-6030(94)03858-R

should be addressed.

Science S.A. All rights reserved

were mixed were of the effect to be

sibility of expanding the range of irradiation to the visible range, to promote chemical changes is restricted to a few cases in which surface complexes which absorb visible light are probably involved [lo, 17, 20, 281. In recent years, attention has focused on mixed Ti(IV)-Fe(II1) oxides, because they have been found to be effective for the photo-assisted dinitrogen reduction to ammonia, TiOz not being successful for this transformation (see refs. 29-32 and references therein). Other photocatalytic reactions with these mixed oxides have been reported [24, 33-351, including toluene, dichloroacetic acid and phenol oxidation. Oxalic acid and EDTA are common water pollutants which arise from industrial processes (metallurgy, decontamination of nuclear plants and boilers, and textile industries) or domestic use. They have generally been used as sacrificial agents in photoelectrochemical studies [36-40], and only a few examples in the literature are related to their photocatalytic oxidation [9, 11, 12, 15, 26, 27, 41-451. In this paper, we present the results of the degradation of EDTA, oxalic acid and malonic acid (which is being studied at present as a sub-

MI. Litter, J.A. Navio J Photo-oxidation of oiigocarbo*ylic acids

184

stitute for oxalic acid for nuclear plant decontamination) by short- and near-UV irradiation with selected samples of TiOZ, iron oxides and mixed Ti-Fe oxides. The three substrates were compared with equal conditions of concentration, oxygen atmosphere, pH and temperature, to evaluate the efficiency of the different catalysts and the effect of the irradiation wavelength. This study attempts to help elucidate the complex processes involved in photocatalysis, and to explore how the physical and chemical properties of the oxides may affect their reactivities.

2. Experimental

details

Commercially available samples of TiO, (Degussa P-25) and maghemite (ly-Fe,03) (Hercules HHO) were used as provided. Magnetite (Fe304) and hematite (a-Fe,O,) sample 1 were prepared as described previously [46, 471. Hematite sample 2 was synthesized by calcination of Fe(NO& 49H,O in air at 773 K for 24 h. Iron-doped TiO, powders with nominal iron concentrations of 0.5 wt.% and 5 wt.% were prepared by an incipient wetness impregnation method [48] as follows. Powdered TiOZ (P-25) was added with stirring to a solution of Fe(NO,),-9HZ0 containing the corresponding quantity of iron in the minimum amount of water; after standing for 48 h at room temperature, the liquid phase was evaporated at 383 K for 24 h and the dried solids were fired in air at 773 K for 24 h. The iron content was checked by atomic absorption. The oxides were characterized by chemical analyses, X-ray diffractometry, scanning electron microscopy (SEM) and Brunauer-Emmett-Teller (BET) surface area measurements. Table 1 shows the properties of all the oxide samples. TABLE

1. Properties

Type of oxide

of the oxide samples Specific surface aIca

Particle (pm)

Cm’ g-7 Ti02 (P-25) Fe304 ~-Fe&h rr-Fe,O, 1 a-Fe,Oa 2 5%Fe-Ti OJ%Fe-Ti

67.1 9.7 26.6 67.5 19.5 29.2 29.6

“Provided by the manufacturer. bPure TiOz. CAggrcgates containing iron.

0.03” 0.26 0.50 0.05 0.06 5”+ 1w sb+5ff

size

Na,EDTA (Schuchardt) and oxalic acid (RiedelDe Haen) were of quality grade and were used as provided. Malonic acid (Mallinckrodt 99%) was first dried in vacw until no traces of acetic acid were detected by ionic chromatography. Water was bidistilled in a quartz apparatus. All the other reagents were of analytical grade and used without further purification. Dilute H,SO, or NaOH was used for the pH adjustments. Irradiations in the short-UV range (254 nm) were performed using a low-pressure mercury arc germicidal lamp (Nice, model GlSTS, 15 W). For irradiations at other wavelengths, a high-pressure xenon arc lamp (Osram XBO, 450 W) was used with a water filter of length 50 mm to minimize IR irradiation. For near-UV irradiations, a bandpass filter was used (Schott Catalog No. BG12; thickness of 2 mm; 310 nm < h < 520 nm; maximum transmission at 400 nm). The use of the bandpass filter guaranteed that no short-UV light entered the system, and also gave a photon flux comparable with that from the germicidal lamp. Visible irradiations were performed using a cut-off filter set at 435 or 530 nm (Schott Catalog no. GG435, thickness of 4 mm; no. OG530, thickness of 1 mm respectively). For actinometry in the short-UV region, the ferrioxalate method [49] was used. A photon flux of 9.2X10-” einstein s-l dme3 was calculated. For the remaining spectral regions, Reinecke’s salt actinometry [50] was used. Approximate photon flux values were obtained taking average actinometer quantum yields. The calculated photon fluxes were 3.7 X lo-’ (310-520 nm), 1.24 X 10m4 (A > 435 nm)and7.6x10-5(h>530nm)einsteins-1dm~3. Experiments show that, except in the maghemite-oxalate system, irradiation with tight of h > 435 nm has no effect. Therefore, the near-UV photon flux was calculated by subtracting the irradiation at A> 435 nm, obtained by superimposition of the bandpass filter and the cut-off filter at 435 nm. After correcting for cut-off filter transmissions, a photon flux of 1.2~10~~ einstein s-’ dmb3 was obtained for the range 310-435 nm. More accurate calculations taking into account the light intensity, filter transmissions and actinometer quantum yield variations with wavelength gave a photon flux comparable with the approximate calculation. Photodegradations were carried out as follows. Each oxide sample was suspended (0.5 g dmm3) in a fresh aqueous solution of the corresponding organic acid (5X 10m3 mol dmm3), previously adjusted to pH 3; the oxide concentration guaranteed total absorption of light for all the oxide samples

M.I. Littm, J.A. Navio / Photo-oxidation

in the short-UV and near-UV ranges. The suspension was ultrasonicated for 1 ruin, and a sample of 2 ml was irradiated at 298 K at the desired wavelength in a thermostatted quartz cell (pathlength of 10 mm) for 2 h with magnetic stirring. A water-saturated oxygen stream was bubbled in the suspension throughout the experiment. After irradiation, the suspension was filtered through a Millipore membrane. The photodegradation of the organic acid was evaluated by determining its concentration before and after the irradiation, by comparison with a blank in the dark. Irradiations of the organic acids in the absence of catalyst were performed for all the wavelengths under the same conditions. A Dionex DX-100 ion chromatograph equipped with a 4400 Dionex Integrator was used to determine the oxalic and malonic acid concentrations. The following conditions were used: HPIC-AS4A column; 1.8 Mm CO,‘/l.7 Mm HCOJp eluent; AMMS-II anion micromembrane suppressor; 50 mN H,SO, eluent [Sl]. Fresh standards were used daily for calibration. EDTA determination was carried out by spectrophotometric analysis of the Co” complex [.52]. For the iron oxides and Ti-Fe oxides, the total amount of iron in the solution was determined by the thioglycolate method [53]. In all the spectrophotometric techniques, calibration curves were obtained first. Absorption spectra were obtained using a Shimadzu 210A spectrophotometer, with an integrating sphere for reflectance spectra. A PerkinElmer model 2380 spectrophotometer was used for atomic absorption determinations. The BET areas were determined with a Micromeritics AccuSorb model 2100E physical adsorption analyser. The X-ray diffraction pattern for the present TiOZ P-25 sample was obtained with a Phillips PW1050 diffractometer. The K radiation of copper was employed.

3. Results

In Table 1 the main differential characteristics of all the oxide samples are listed. Other physical properties of the iron oxide samples used in this work can be found in previous papers [P, 46, 471. It can be seen that the P-25 together with the hematite samples showed smaller particle sizes. Bulk and surface characterizations of the mixed Ti-Fe samples can be found elsewhere [54].

of oligocarboxy’ic acids

185

For the TiO, sample, the anatase-to-rutile weight ratio (0.57), as determined by X-ray diffractometry, differs from the ratio reported by the manufacturers (0.80); this value depends on the history of the sample. As reported previously [54], Ti-Fe oxides present lower anatase/rutile ratios, and the sample with 5 wt.% Fe shows X-ray peaks assigned to pseudo-brookite (Fe,TiO,). The mixed oxides have a lower amount of acid and basic hydroxyl groups than does the TiO, sample, as determined previously by a spectrophotometric method [55]. These data are shown in Table 2. SEM and energy-dispersive X-ray (EDX) studies [48, 541 showed that our Fe-Ti oxide samples consist of fine particles and larger aggregates, as a result of the impregnation method used in the preparation. Very large distributions of shapes and dimensions of the particles were observed in the aggregates, with some enrichment in iron at the surface. It was concluded that the aggregates (50 km in size for OSwt.%Fe-TiO, and 180 Frn in size for Swt.%Fe-TiO,) contain non-uniformly distributed iron, whereas the smaller (about 5 pm in size) grains are mixtures of pure anatase and r-utile. In the sample with 0.5 wt.% Fe, Fe3+ forms a solid solution in the TiO,, occupying titanium sites as a substitutional dopant; however, some particles have an amount of iron above the solubility limit, with small amounts of precipitated Fe,O, (probably hematite). In addition, the sample with 5 wt.% Fe contains Fe,TiO, or Fe,O, as separate phases. 3.2. DiJkw reflectance and absorption spectra The spectral characteristics of the iron oxide samples used in this work were reported earlier [12] and were in agreement with earlier data (see refs. 43-47 in ref. 12). Iron oxides present absorption bands in the UV region, in the ranges 260-320 nm and 370-400 nm, corresponding to charge transfer transitions. The points of the onset of the absorption are located approximately at 570 nm (2.2 eV) for hematite, 530 nm (2.3 eV) for maghemite and beyond 800 nm for magnetite.

TABLE

2. Differences in OH group content and anatasehutile (A/R) ratio behveen TiOz and mixed Fe-Ti oxides

Type of oxide

Acidity (crm g-9

Basic@ (Pm g-7

A/R ratio (WW

Ti02 (P-2.5)

5.8 3.3 3.4

13.8 6.5 4.3

0.57 a.45 0.44

5%Fe-Ti O.S%Fe-Ti

M.I. Lifter, I.4

186

Navio 1 Photo-oxidation

3.3. Photocatalytic degradation of oligocarbarylic acids as a function of wavelength In the absence of a catalyst, no degradation of the carboxylic acids was found at any wavelength, with the only exception of oxalic acid under irradiation at 254 nm, which decomposed 19% after illumination for 2 h. Because of the uncertainties in the experimental evaluation of the substrate concentrations, a degradation yield lower than 10% was considered negligible. Preliminary photocatalytic experiments on oxalic acid and EDTA in the presence of different oxide samples were performed under near-UV irradiation (310 530 nm. Therefore, with only this exception, active wavelengths were restricted in all cases to A < 435 nm. Figures 2-4 depict the corresponding results. In these figures, results for oxalic acid and EDTA under near-UV irradiation are included for a better comparison. Broken lines in the figure represent the experimental limit of degradation. It can be seen that the photodegradations were more efficient with short-UV irradiation than with near-UV light (taking into account that similar photon fluxes were used in both series of exper-

As is known [56], TiOz (P-25) presents an absorption edge located at 408 nm (3.04 eV) and a charge transfer band with a maximum at 325 nm. The mixed Ti-Fe oxide spectra resemble that of TiOz with the onset shifted towards the visible end of the range (450 nm for the O.Swt.%Fe-TiO, sample and 500 nm for the Swt.%Fe-TiO, sample, corresponding to 2.75 and 2.5 eV respectively), in agreement with reported data [24, 34, 571. Dispersed iron has little effect on the absorption properties of TiO,. Thus, in the UV region, the 0.5 wt.% Fe sample spectrum is similar to that of TiO,; some differences in the 5 wt.% Fe sample indicate the presence of small amounts of Fe,O, or FeZTiO,.

%

of digocarbmylic

20

0 n

0 TiO, Near-UV

Fe,O,

light

Fig.1.Preljminary results of nm Fe”’

In contrast, an oxidizable substrate attacked directly by holes, i.e. h+ + Sads -

Sad;+ - + final products

(S) can be (3)

or can react with OH’ formedon the semiconductor surface [2, 24, 591, i.e. reaction (1) or h+ + > HO(HZOads) -

HO,,;(

+ H+)

(4)

In the case of carboxylic acids a photo-Kolbe reaction is generaily proposed [l, 21, i.e.

189

M.I. Litter, LA. Navio / Photo-oxidation of oligocarboxylic acids

cb cb ...........................................-.............~... E

(VI ~ E,

E, = 3.2 eV

n

2.2 ev

V S. N H E

CY-Fe& Ti02 Fig. 5. Energy

RCO,-

5

level diagram

RCO;

of the conduction

and valence

band edges for TiOl (anatase)

CO, + other products

(5)

Alternatively, attack by OK radicals can produce hydrogen abstraction in compounds such as EDTA or malonic acid [24, 591. The direct photo-Kolbe product for malonic acid is acetic acid, which can be degraded analogously to CO, and methane or ethane. In our case, we only detected this product by high performance ionic chromatography (HPIC), other products not having been tested for. 4.2. Eficiency of the catalysts: effect ofthe irradiation wavelength As is known, photocatalysis with TiOz is only possible with UV light (A ~400 nm). The same energy restriction holds for mixed Ti-Fe catalysts, because photogeneration of e --h+ pairs from TiOz is not (or only marginally) affected by Fe+3 [60]. In the case of iron oxides, although the bandgap is shifted to the visible range, only charge transfer transitions occurring in the UV region are effective, with spin-flip or d-d transitions, closer to the bandgap, being inactive [9, 10, 15, 201. In the present work, we indeed obtained degradation of oxalic acid with y-Fe,O, at h > 435 nm; however, in this case, we suggest that it arises from the photolysis of surface complexes (see Section 4.3).

and a-Fe203

at pH 3.

As experiments with near-UV light show, oxalic acid is the most reactive substrate for all the oxide samples; the other two acids are fairly unreactive on oxides other than TiOz. The reactivity can be correlated with the redox potential of the compounds and the valence band edge of the catalyst. Because of the irreversibility of the decomposition reactions, the measurement of one-electron redox potentials of carboxylic acids is very difficult, although some data are reported for EDTA and oxalic acid on ft-TiO, catalysts (1.7 V and 2.2 V vs. NHE at pH 3 respectively) [44]. According to our present and previous results [ll, 121, we think that oxalic acid is more oxidizable than EDTA; malonic acid has probably an even more positive redox potential. Consequently, the efficiency of the photocatalytic processes correlates with the redox potentials in the following order: oxalic acid > EDTA > malonic

acid

Differential adsorption of the ligands onto the oxide and surface complex formation also can be invoked to explain different degrees of degradation. Adsorption parameters derived from dissolution experiments of magnetite show that the ability of malonic acid to adsorb is lower than those of EDTA and oxalic acid [61, 621, and similar results can be expected onto maghemite and TiO,. This fact also accounts for the reduced photodegradability of malonic acid.

190

M.I. Littq

J.A. Navio f Photo-oxidatian of oligocarimylic

For the different semiconductors, in Ti02, the hole is generated by excitation of an 02- to Ti4+ charge transfer band, lying at a redox potential of +3.19 V at pH 3 (see Fig. 5). In iron oxides, it is generally assumed that holes are created in the iron band and are located in a deep FeI” trap with less oxidizing power ( +2.44 V at pH 3) [25, 34, 63-651. Also, the production of OH radicals by reactions (1) or (4) (E=2.2 V ‘us. NHE at pH 3 in homogeneous solutions [66]) is less favored in iron oxides. These oxides appear to react only with strong reducing ligands, other electron donors being inactive. The reactivity found for different types of pure iron oxide with oxalic acid (Fig. 1) can be explained by the electronic and structural properties of the corundum and spinel-type forms. Maghemite and magnetite, being spin&, are more reactive than hematite, which presents a low mobility of charge carriers and fast surface recombination [63, 67, 681. The same behavior was found in previous studies [12, 261. In the case of mixed Ti-Fe oxides, it was found that the lifetimes of the electrons and holes are enhanced in comparison with pure TiO, or Fe,O, (from some nanoseconds to several hours), as a result of electron or hole trapping at Fe”’ centers [31,69]. This effect was found in the case of doped samples, where iron replaces titanium centers in the lattice. Consequently, redox processes would be favored by this situation, and the activity of mixed oxides should be similar to or higlier than that of TiOZ [32, 34, 60, 701. However, in the samples used in the present work, iron is distributed non-homogeneously, forming a second phase (iron oxide and/or pseudo-brook&) that acts as a hole trap. This model is that of “coupled semiconductors” [5]: as is shown in Fig. 6, in addition to the direct photoproduction of e--h+ pairs in the Fe,O, aggregates, the transfer of holes from TiO, to Fe,O, is thermodynamically possible. This effect will place holes at the same level as in pure iron oxides, and explains the lower photoactivities found in our case. Also, as a result of the rather low mobility of photoexcited electrons in those phases [60,67,71], it is likely that the e--h+ recombination rate is higher than the trapping rate by other species. Consequently, considering the inherent photophysical properties of the different semiconductors, all the substrates can be easily oxidized on Ti02; the process is less favored on iron and mixed oxides, as is found experimentally. Differences in the photoactivity under near-UV light also can be attributed to other characteristics

acids

of the samples. For example, TiOz P-25 must be more reactive because of its smaller particle size and higher surface area. In maghemite, surface defects also can produce a higher recombination rate [12]. In the Ti-Fe oxide samples, the low amount of OH surface groups (see Table 2), resulting from the calcination process or some interaction of iron with surface hydroxyls of the TiO, precursor, can cause lower adsorption of the substrate, as proposed for mixed Cr(III)-Ti(IV) oxides [72]. Another consequence of the calcination of the samples is the lower anatase/rutile ratio: a lower photoactivity is expected, because rutile shows a much lower capacity to absorbing oxygen [l, 21. However, these differences are very small and seem here to be of less importance. Some inactivation of the samples by photocorrosion (see Section 4.4) also might be considered. Several effects may explain the higher efficiency under short-UV irradiation with all the oxide samples. Owing to higher absorption coefficients, the penetration distance of the photons into the particle is shorter and the e--h’ pairs are formed closer to the surface, making them more available for the substrate [64, 731. This charge separation favors the degradation process, even for the less oxidizable substrates (EDTA and malonic acid) on maghemite and mixed oxides. In addition, direct photolysis at short wavelengths cannot be ruled out, in principle, because oxalic acid and EDTA present some absorption in this range (Earn= 48 M-’ cm-’ and 8 M-’ cm-’ respectively). However, photolysis experiments in the absence of a catalyst resulted in small amounts of degradation, which were only significant in the case of oxalic acid (19%). Moreover, this contribution can be considered negligible in the heterogeneous systems, because the low absorptivity of the substrates means that practically all the light will be absorbed by the oxide. The process H202 5

2OH’

(6)

may be important at shorter wavelengths [3] and can contribute to oxidative processes. Although it can explain the greater reactivity of TiOs, it does not take place in pure iron oxides [25], in mixed oxides, because electrons can be trapped either by titanium or by iron centers, process (6) could contribute, at least partially. 4.3. Role of surface complexes Surface complexes may lead to an enhancement of the overall oxidation rate through

M.I. Litter, J.A. Navb

I Photo-oxidation

of oligocarbmylic

]

acids

191

AE 8 0.25 V

I AE = 0.75 V

TiO, Fig. 6. Diagram

> Ti”‘(Fe”‘)

of “coupled

- Sz

semiconductors”

> Ti”‘(Fe”)

for TiOz and wFe203.

+ S,

(7)

In fact, the formation of surface complexes between strong ligands and surface Fe”’ centers was proposed in reductive dissolution mechanisms of iron oxides [62, 741 and in photochemical reactions [9-13, 15-18, 20-22,28,34], some of them occurring with visible light. In the case of TiO,, although there is no conclusive proof of the existence of such surface complexes, some IR and UV spectroscopic features give evidence of chemical interactions between TiO, and adsorbates [5]. The detection of surface complexes is very difficult, especially in the UV region, because of their low concentration and the strong absorption of the oxides. However, it is likely that they exhibit charge transfer absorption bands at energies in the UV and visible ranges similar to those of the corresponding homogeneous complexes, and analogous photolytic behavior. In Table 4 are shown the reported ranges of the absorption and photolysis quantum yields (elecTABLE 4. Reported absorption and photolysis Ti’” complexes in homogeneous solutions Ligand

Absorption range

data for Fe”’ and

Photolysis quantum yield

Ref.

1.25 (254 nm) 0.35 (254 nm) =10-j (366 nm) Not reported Not reported

49, 78, 13, 82, 78,

(nm) Fe(III)dxalate Fe(III)-EDTA Fe(III)-malonate Ti(IV)-oxalate Ti(IV)-EDTA

24(1-475 22C‘loo MI-360 23&3.50 220-300

76, 71 79 80, 81 83 84

tron transfer reactions to Fez+ or Ti3’ and oxidation products of the ligand) of the complexes of the three carboxylic acids in solution. In the case of Fe”‘, complexes with high stability constants are well known [75] and their photolyses have been studied [76-81]. In the case of Ti’” complexes, information is less scarce 178, 82-841. According to these data, and assuming similar processes in heterogeneous systems, the degradation of each substrate on maghemite under short-UV light irradiation suggests an important contribution of the photolysis of surface complexes, which is probably more important for oxalic acid and EDTA than for malonic acid. At longer wavelengths, this contribution should be important only for oxalic acid and explains the occurrence of degradation under visible light irradiation. On TiOZ (and probably mixed oxides), excitation of Ti(IV)-oxalate complexes by short- or long-UV light might contribute to the reaction [82]. This process is probably of minor importance for EDTA at higher UV wavelengths (h > 300 nm). In the case of malonic acid, irrespective of the effectiveness of the photolysis, the reduced adsorption should produce a lower concentration of surface complexes and, consequently, a lesser contribution of this pathway to degradation processes. 4.4. Photocon-osion In iron oxides in the absence of oxygen, photocorrosion can occur by the attack by conduction band electrons on Fe”‘ surface sites. Although the reduction of oxygen by conduction band electrons is not a thermodynamically competitive process

192

M.I. Litter, J.A. NoGo / Photo-oxidation of aligocarbozylic acids

on iron oxides, the presence of oxygen is important to prevent photocorrosion by reoxidation of > Fe” to >Fe’II at a high rate before it can be detached from the surface (reaction .(Z)) [15, 26, 281. In fact, as shown in Table 3, the photodissolution of iron oxides under an oxygen atmosphere is very low, in agreement with previous results [12, 151. In Ti-Fe oxides, although the fate of conduction band electrons is not clear, some iron dissolution takes place. This can cause inactivation of the catalyst, which can acquire poorer properties than those of P-25 (larger particle size, smaller surface area and smaller amount of active OH groups). Strong inactivation of iron-doped Ti02 samples made by impregnation methods was found after several hours of irradiation [32]. The 5 wt.% Fe sample seems to be more stable, probably because of the presence of pseudo-brookite, which is less easily photocorroded.

4.5. Further remarks about mked Ti-Fe oxides The reported good efficiency of Ti-Fe oxides for nitrogen reduction to ammonia [2!9-321 compared with the efficiency of TiOz was assigned to enhanced charge separation and better adsorption of nitrogen. However, some photo-oxidations carried out with these oxides showed slightly better results than with some samples of TiOZ, but none of them was Degussa P-25 (or they were modified (e.g. calcinated) P-25) [24, 33-351. According to the present results, mixed Ti-Fe oxides do not seem to be better photocatalysts than P-25 for oxidative purposes, unless their properties could be improved. Mixed oxides with a low amount of iron (below the solubility limit, to avoid the formation of separate phases of hematite or pseudo-brookite), obtained by a mild coprecipitation technique at lower temperatures, would give a homogeneous dispersion of iron on Ti02, larger surface areas and smaller particle sizes, and could be better catalysts.

Acknowledgments

Drs. E. San RomPn from INQUIMAE-UBA (Argentina), and D. Meissner and D. Bahnemann from ISFH (Hannover) are thanked for helpful discussions. Further thanks go to Divisibn Fisica de1 Sdlido (CNEA) for X-ray measurements. M.I.L. is a member from CONICET (Argentina).

References 1 N. Serpone and E. Pelizzetti (eds.), Photocutulysis, Fundumenfak and Applicotians, Wiley, New York, 1989. 2 M.A. Fox and M.T. Dulay, Chem. Rev., 93 (1993) 341. 3 0. Legrini, E. Oliveros and A.M. Braun, Chem. Rev., 93 (1993) 671. 4 P. Mulvaney, F. Grieser and D. Meisel, in M. Grltzel and K. Kalyanasundaram (eds.), Kin&x and Catalysis in Microhetemsneous Svstems, Dekker. New York. 1991. D. 303. Chek Rev:, 93 (1963) 267. _ 5 P.V. I&&t, 6 D.W. Bahnemann. in E. Peliuetti and M. Schiavello (eds.), Photochemicnl Conversion and Storage of Solar Ezzew, Kluwer, Amsterdam, 1991, p. 251. 7 T.D. Waite, Rev. Mineral., 23 (1990) 559. 8 D. Bahnemann, D. Bockehnann and R. Goslich, SolarEnergy Mutt-r., 24 (1991) 564. 9 M.I. Litter and M.A. Blesa, J. CoUoid lmerface Sci., 125 (1988) 679. 10 M.I. Litter and M.A. Bless, Can. J. Chem., 68 (1990) 728, 11 M.I. Litter, E.C. Baumgartner, G.A. Urrutia and M.A. Blesa, Environ. Sci. Technol, 2.5 (1991) 1907. 12 M.I. Litter and M.A. Bless, Con. J. Chem., 70 (1992) 2.502. 13 M.I. Litter, M. Villegas and M.A. Blesa, Can. J. Chem., in press. 14 R.M. Cornell and P.W. Schindler, Clays Clay Miner., M (1987) 347. 15 C. Siffert and B. Sulzberger, Langmuir, 7 (1991) 1627. 16 T.D. Waite and F.M.M. Morel, Environ. Sci. TechnoL, 28 (1984) 860. 17 T.D. Waite and F.M.M. Morel, 1. ColIoid Interface Sci., 102 (1984) 121. 18 K.M. Cunningham, M.C. Goldberg and E.R. Weiner, Pho&hem. Photobiol., 41 (1985) 409. 19 M. Griitzel, .I. Kiwi and CL. Morrison,J. Chem. Sot., Faraday Trans. I, 81 (1985) 1883. 20 B.C. Faust and M.R. Hoffmann, Environ. Sci. Technol., 20 (1986) 943. 21 T.D. Waite, A. Torikov and J.D. Smith, J. Calloid Interface Sci., 112 (1986) 412. 22 K.M. Cunningham, M.C. Goldberg and E.R. Weiner, Environ. Sci. TechnoL, 22 (1988) 1090. 23 R.D. Stramel and J.K. Thomas, J. Colloid In&$&e Sci., 110 (1986) 121. 24 D.W. Bahnemann, D. Bockelmann, R. Goslicb and M. Hilgendorl?, in D.G. Crosby, G.R. Helz and R.G. Zepp (eds.), Envimmnental Aspects of Surface and Aquatic Photochemishy, ACS Symp. Ser., in press. 25 C Kormann, D.W. Bahnemann and M.R. Hoffmann, L Photochem. Pholobiol. A: Chem., 48 (1989) 161. 26 J.K Leland and All. Bard, J. Phys. Chem, 91 (1987) 5076. M.-N. Mozzanega and P. Pichat, J. Pho27 J.-M. Herrmann, tochem., 22 (1983) 333. 28 B.C. Faust, M.R. Hoffmann and D.W. Bahnemann, J. Phys. Chem., 93 (1989) 6371. Pho29 M. Schiavello and A. Sclafani, in M. SchiaveUo,(ed.), Phororeactors, toelectrochemirhy, Photocata&sis and NATO-ASI, Ser. C, 146 (1985) 503. 30 R.I. Bickley and J.A. Navio-Santos, in G. Grassi and D.O. Hall (eds.), Photocaralytic Production of Etaergvich Compounak, Elsevier, London, 1985, p. 105. and L. Palmisano, in M. Schiavello (ed.), 31 V. Augugliaro Photocata&sir and Envimnmenf, Kluwer, 1988, p. 425. 32 J. Soria, J.C. Conesa. V. Augugliaro, L. Palmisano, M. Schiavello and A. Sclafani, J. Phys. Chem., 95 (1991) 274.

MI, Litter, J.A. Navio / Photo-oxidation of oligocarboxylic acids 33 J.A. Natio,

34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

52 53 54 5.5

56 57

M. Garcia GBmez, M.A. Pradera Adrian and J. Fuentes Mota, in M. G&net, J. Barraulr, C. Bouchoule, D. Duprez, G. Pirot, R. Maurel and C. Montassier (eds.), Heterogeneous CaraZysis and Fine Chemicals II, Elsevier, Amsterdam, 1991, p_ 445. D.W. Bahnemann, Zsr. J. Chem., 33 (1993) 115. L. Pahnisano, V. Augugliaro, A. Sclafani and M. Schiavello, 1. Phys. Chem., 92 (1988) 6710. T. Freund and W.P. Gomes, Catal. Rev., 3 (1969) 1. J.R. Harbour and M.L. Hair, 1. Phys. Chem., 83 (1979) 652. J.R. Darwent, J. Chem. Sot., Faraday Tmns. 2, 77 (1981) 1703. A. Mills and G. Porter, J. C&m. Sot., Faraday Trans. 1, 78 (1982) 3659. T. Uchihara, M. Matsumura, J. Ono and H. Tsubomura, J. Phys. Chem., 94 (1990) 415. J. Domenech and J.M. Costa,Phorochem. PhotobioL, 44 (1986) 675. J.H. Kennedy and D. Diinwald, J. Elecnochem. Sot., 130 (1983) 2013. J.D. Spikes, Photochem. Pholobioi., 34 (1981) 549. D.N. Furlong, D. Wells and W.H.F. Sasse, A&. J. Chem., 39 (1986) 757. F. Sabin, T. Tiirk and A. Vogler, .I. Photochem Photobiol. A: Chem., 63 (1992) 99. A.E. Regazzoni, G.A. Urrutia, M.A. B1esaandA.J.G. Maroto, I. Inorg. Nud Chem., 43 (1981) 1489, method B. A.E. Regazzoni, M.A. Bless and A.J.G. Maroto, J. Colloid Inte*ce Sci., 122 (1988) 315. R.I. Bickley, J.S. Lees, R.J.D. Tilley, L. Palmisano and M. Schiavello, 3. Chem. Sot., Famday Trans., 88 (1992) 377. C.G. Hatchard and CA. Parker, Proc. R. Sot. London, Ser. A, 235 (1956) 518. E.E. Wagner and A.W. Adamson, J. Am. Chem. Sot., 88 (1966) 394. J. Weiss, Einfiihrung in die Ionenchromatographie: Grundlagen, Instrumentation und Anwendung, Teil 1, Chem. Labor Bet&b, 34 (1983) 293. K.L.E. Kaiser, UIafer Res., 7 (1973) 1465. D.L. Leussing and I.M. Kolthoff, 1. Am. Chem. Sot., 75 (1953) 390. J.A. N&o, M. M&as, M. Gonzilez-Cat&n and A. Juste, J. Mater. Sci., 27 (1992) 3036. J.A. Natio, M. Macias, M. GonzBlez-Cat&n, J.M. Marinas and J.M. Campelo, Proc. 13th Iberoametican Symp. on Cafarusis, Madrid, July 1992. R.I. Bickley, T. Gonzllez-Carrefio, J.S. Lees, L. Palmisano and R.J.D. Tilley, J. Solid State Ctiem., 92 (1991) 178. M. Schiavello, L. Rizzuti, R.I. Bickley, J.A. Navio and P.L. Yue, Froc. 8fh Int. Congress on Cata&& Berlin, June-Juty __^ lY84.

193

$8 E. Pelizzetti and C. Minero, Electrochim. Acta, 38 (1993) 47. 59 Y. Mao, C. Schiineich and K.-D. Asmus, J. Phys. Chem., 95 (1991) 10080. 60 D. Cordishi, N. Burriesci, F. D’Alba, M. Petrera, G. Polizzotti and M. Schiavello, J. Solid State Cbem., 56 (1985) 182. 61 E.B. Borghi, P.J. Morando and M.A. Blesa,Pmc. 8thArgentine Meet. olt Physical Chemistty, Mar de1 Platu, Aptil 1993. 62 M.A. Blesa, P.J. Morando and A.E. Regazzoni, Chemical DiEsolution of Metal Oxides, CRC Press, Boca Raton, FL, 1994. 63 J.H. Kennedy and K.W. Frese,l. E&cfrochem. Sot., I25 (1978) 709. 64 L.G.J. de Haart and G. Blasse, 1. Electrochem. Sot., 132 (1985) 2933. 65 M.P. Dare-Edwards, J.B. Goodenough, A. Hammett and P.R. Trevellick, I. &em. Sot., Faraday Trans. 1, 79 (1983) 2027. 66 J.P. Hoare, in A.J. Bard, R. Parsons and J. Jordan (eds.), Stamford Potential in Aqueous Solutions, Dekker, New York, 1985, p. 49. 67 J. Moser and M. GrHtzel, Helv. Chim. Acta, 65 (1982) 1436. 68 H. Nakanishi, C. Sgnchez, M. Hendewerk, G.A. Somojai, Mater. Res. BuiZ., 21 (1986) 337. 69 J. Moser, M. Grltzel and R. Gallay, Helv. China. Actu, 70 (1987) 1596. 70 J.A. Navio, F.J. Marchena, M. Rome1 and M.A. De la Rosa, J. Photochem. Photobiol. A: Chem., 55 (1991) 319. 71 M. Schiavello, L. Rizzuti, A. Sclafani, I. Majo, V. Augugliaro and P.L. Yue, Adv. Hydrogen Energy, 3 (1982) 821. 72 C. Martin, I. Martin, V. Rives, L. Palmisano and M. Schiavello, J. Catal., 134 (1992) 434. 73 R.W. Matthews and S.R. McEvoy, J. Photochem. Photobiol. A: Chem., 66 (1992) 355. 74 W. Stumm, Chemisby of the Solid-Water Interface, Wiley, New York, 1992, p. 309. 75 A.E. Martell and R.M. Smith, Critical Stability Consfants, Vol. 5, First Supplement, Plenum, New York, 1982. 76 C.A. Parker, Trans. Faraday Sot., 50 (1954) 1213. 77 C.A. Parker, Proc. R. Sot. London, Ser. A, 220 (1953) 104. 78 Y. Uzumasa, M. Nishimura and T. Seo, Bull. Chem. Sot. Jpn., 30 (1957) 438. 79 P. Natarajan and J.F. Endicott, L Phys. Chem., 77 (1973) 2049. 80 R. Van Eldik and J.A. Van den Berg, J. South Afr. Chem. Inst., 23 (1970) 96. 81 J.A. Van den Berg, 1 South AJ?. Chem. Inst., 22 (1969) 12. 82 J. Shiokawa, A. Matsumoto, Asahi Garasu Kogvo Gijutsu Shore&i, Kenkyu Hokoku, 25 (1974) 2Q7. 83 A.K. Babko and L.I. Dubovenko, Zhur. Neorg. Khim., 4 (1959) 372. 84 B. Karadakov and P. Nenova, God. K&h. Khimikotekhnol. Insr. Sofia, 15 (1972) 287.