Photocatalytic Purification of Water

The ‘enhancement factor’, E-factor, for a photoprocess is defined as:

(iii)

There are many reports of enhanced rates for the photocatalytic purification of water (E-factors ≥ 1) for Pt/TiO₂ and Pd/TiO₂ semiconductors compared to their non-metallised counterparts. Table III contains details from a selection of these reports in terms of the titania source, method and amount of pgm deposited, organic pollutant under test and reported E-factor (9–15).

In most of the systems listed in Table III the amount of Pt (or Pd) deposited on the titania particles is invariably ca. 1% w/w. This amount is a consequence of the vast majority of previous work on hydrogen evolving photocatalytic systems: too much pgm will result in a greatly reduced photocatalytic efficiency due to enhanced electron-hole combination and UV-shielding of the TiO₂ particles by the pgm deposits, whereas too little pgm will result in low photocatalytic activity due to the paucity of low-overpotential water reduction sites. The optimum pgm value is usually ca. 1 wt.%.

From Table III (and later Tables) the most popular method of depositing Pt/Pd onto titania is by photodeposition (16, 17). The E-factors in Table III are generally low, typically from 2 to 4. By contrast, for the reduction of water to H₂ by sacrificial electron donors, such as methanol, photocatalysed by a Pt/TiO₂ or Pd/TiO₂ dispersion, the E-factor is > 1000 if not infinitely large. But, as noted earlier, if pgm is not present little or no hydrogen is usually evolved (16, 17). In many cases, the major difference between these two types of photocatalytic systems: i.e. those focused on organic mineralisation (Equation (i)) and those centred on water reduction (Equation (ii)) is the presence or absence of oxygen. Thus, one type of photosystem can be converted to function as the other simply by purging the system with nitrogen or air.

Finally, it is clear from Table III that for apparently identical photocatalytic systems, that is the same pgm, semiconductor and organic pollutant, there are significant variations in the E-factors (for instance, for methanol or trichloroethylene (TCE)) reported by different research groups (8, 10–12).

Table IV contains details of photocatalytic systems for water purification where E-factors ≤ 1 have been reported (8, 11, 14, 18–20). It is clear that Pt/TiO₂ and Pd/TiO₂ photocatalysts are not necessarily always more effective photocatalysts for water purification compared to the original unmetallised titania forms.

Initially the variation in the E-factor appears to be due to the variation in the nature of the pollutant rather than a variation in the photocatalyst. For example, from Table III the photodestruction of dichloroacetic acid (DCA) always appears to be more active (E-factor ≥ 1) for a Pt/TiO₂ or Pd/TiO₂ photocatalyst than for the original TiO₂ (E-factor - 1) (13, 14) whereas, from Table (IV), the photodestruction of 4-chlorophenol (4-CP) sensitised by the same UV 100 titania photocatalyst is often found to be unaffected or inhibited (E-factor ≤ 1) by deposits of Pt or Pd (13, 14). These results and others indicate that the E-factor varies from pollutant to pollutant. This variation could be attributed to the effect of the pgm on one or more of the key reaction parameters which indude: the adsorption isotherms of the reactants, the kinetics of the electron-transfer processes and the desorption isotherms of the products. However, Tables III and IV also show examples where, for apparently the same or very similar photosystems, E-factors > 1 and < 1 have both been reported for the same pollutant, for example TCE, (8, 10–12).

Significant differences in the E-factor for apparently identical photocatalytic systems may occur because the systems are not as alike as they first appear. In fact, one problem in semiconductor photocatalysis is that the systems have many critical variables. These variables indude: source and history of the semiconductor, method of depositing the pgm, method of agitation, method of aeration, photoreactor design and construction (material of which it is made), illumination source (the wavelength(s) and intensity of excitation), pollutant concentration, pH, ionic strength, and temperature. Various combination of these factors could easily lead to an overall E-factor > 1 or < 1.

In the case of the photomineralisation of TCE (11), using Pt deposited by a chemical reduction method using zinc, it appears that the variation in E-factors in Tables III and IV is due to the source of titania, with Aldrich anatase (E-factor = 3) appearing more active than Degussa P25 (E-factor < 1). The reason for this is presendy unclear.

In other cases, the cause for the E-factor variation may be hidden in the experimental detail. Thus, the work of Chen et al. gives the E-factor for the photomineralisation of methanol by Pt/TiO₂ as 7.8 at pH 5.1, but as 2.4 at pH 10.9 (8). Thus, it is even possible for different groups studying the same photocatalytic system to report different E- factors and the difference is due simply to a difference(s) in one or more of the key experimental parameters (pH or ionic strength, for instance).

In fact, the rate of photomineralisation of methanol sensitised by unplatinised titania is ten times faster at pH 10.9 than at pH 5.1, so in this system, increasing the pH improves the overall rate of photocatalysis much more for TiO₂ than for Pt/TiO₂ (8). The reasons for the apparent variation in E-factor and rates with pH remain as yet unclear.

One of the classic works in semiconductor photocatalysis, by Wang et al., concerns the efficiency of Pd as an oxygen reduction catalyst in the photocatalytic destruction of sodium 2,2-dichloro- propionate (DCP) by titania (9). This work is often cited by researchers as an example of a pgm (Pd) that significantly enhances the overall rate of photominetalisation by catalysing the, presumed slow, oxygen reduction step by photogenerated electrons on TiO₂. In the destruction of DCP, dissolved HCl is produced, along with CO₂. Wang used the HCl production, via the associated change in pH of the reaction solution, to monitor the progress of the reaction (9). [H⁺] versus irradiation time profiles for this system, using titania impregnated with 0, 0.01 or 2 wt.% Pd as the photocatalyst, are illustrated in Figure 2(a) and show that the overall rate of the photocatalytic oxidation of DCP is increased 3-fold with 0.01 wt.% Pd, but 7fold for 2 wt.% Pd (9).

Recent work has shown that the presence of a pgm does not always enhance a photocatalytic system even when the organic pollutant, pgm and semiconductor are the same. For instance, in a study of the role of Pt and Pd in semiconductor photocatalysis, Chen and coworkers (8) investigated the same DCP/(Pd/TiO₂) system as used by Wang (9). Figure 2(b) illustrates some results of this work which show that 1 wt.% Pd significantly depresses (E-factor = 0.5), rather than enhances, the overall kinetics of photomineralisation (8)! The only obvious difference between the two studies is the source of the titania: Hombikat UV 100 (Wang (9)) and Degussa P25 (Chen (8)). The type of the titania used thus appears to be a critical factor in determining whether a pgm will have a positive or negative effect on the overall kinetics of photomineralisation. In fact, despite more than two decades of research, it is still not clear why some forms of titania are better than others at destroying particular organic pollutants. For instance, why is unplatinised Hombikat UV 100 usually found to be excellent at destroying DCA but less effective for 4-CP? Why is this situation reversed when unplatinised Degussa P25 is used as a photocatalyst? Only more detailed work will reveal the elusive answers to these apparently simple questions.

Another example of the unpredictable and contradictory nature of semiconductor photocatalysis has been provided by Tanaka and coworkers in a study of the effect of the crystallinity of titania on its photocatalytic activity (21). Twelve different commercial samples of titania, with and without Pt deposits, were tested as photocatalysts for the destruction of phenol. The results, in the form of plots of initial rate as a function of the % anatase content of the photocatalyst, are illustrated in Figure 3 (21). They indicate that whereas platinisation of rutile TiO₂ generally enhances the rate of destruction of phenol, it depresses it for anatase TiO₂. Just as progress seems to be made towards a general rule, a glance at the data in Figure 3 reveals that Aldrich anatase is an exception, that is, platinisation of a 100% sample of anatase titania significantly enhances its rate of destruction of phenol! It was also noted that platinisation has only a small effect if the organic pollutant is TCE or chloroacetic acid (21)! There appears to be no obvious explanation for this exceptional behaviour by Aldrich anatase, or why the ‘rule’ that: platinised rutile is better than anatase (if it exists) is pollutant specific.

The most likely explanation for the results in Tables III and IV and the examples discussed is that Pt and Pd may enhance or depress the kinetics of the semiconductor photocataiytic destruction of organic pollutants in aqueous solution, depending upon a combination of negative or positive effects arising from the selected reaction conditions. The variation in E-factor is due to the complicated nature of the overall process, the rate of -which depends upon factors such as: adsorption of pollutant, oxidation of pollutant and the concomitant reduction of oxygen and desorption of products. All these processes may be positively, negatively or unaffected by the presence of a pgm, and are likely to depend upon:

From Tables III and IV it can be seen that if the E-factor is positive it is never very large, typically 2–4, and may often be negligible or < 1. This seems to cast doubt on the future role of pgms in semiconductor photocatalysis for purifying water.

It is worth commenting on recent work by Kamat and colleagues on Au/TiO₂ photocatalysts, from which it appears that nanoparticle deposits of Au are photooxidised to Au⁺ ions by photogenerated holes and/or hydroxyl radicals on the surface of the titania particles (22). The authors suggest that adsorbed or intercalated Au⁺ ions may then act as electron-hole recombination centres and so cause a reduction in the efficiency of such photocatalytic systems during long-term irradiation. It is not clear how much of a problem this oxidative dissolution mechanism represents in the photocataiytic destruction of organic pollutants in water or air. Certainly the photooxidative dissolution of Pt or Pd deposits on titania has not been recognised as a major complicating process despite the many photocataiytic studies carried out using Pt/TiO₂ and Pd/TiO₂ photocatalysis. However, pgm oxidative dissolution is worth bearing in mind, especially as it may be partly or fully responsible for any observed slow loss of photocataiytic activity in such systems on prolonged irradiation.

Photocatalytic Purification of Air

More recently the photocatalytic purification of air has become a focus of attention and air conditioning devices and air purifying tiles and paving stones are now available, based on the process summarised by Equation (i) and Figure 1. The organic to be treated in this case is a volatile organic carbon (VOC), usually at a low level (typically < 100 ppm). While not obvious from Equation (i) or Figure 1, water vapour plays an essential role in the overall process even in the gas phase. The nature of the role is uncertain but it is generally agreed that if water vapour were not present the overall photomineralisation process would rapidly grind to a halt. This is because in most photocataiytic systems for air purification, water vapour is the major source of adsorbed hydroxyl groups. However, it is often also a major product, thus, in most instances Equation (i) can proceed quite efficiently under very dry conditions. In some examples of Equation (i) for air purification, a high partial pressure for water vapour appears partly to inhibit the photomineralisation process, possibly due to competitive adsorption between the water and VOC molecules on the semiconductor surface (23). In other examples, increasing the partial pressure of the water vapour appears to have either little effect or to promote the overall photocatalytic process (23).

A possible concern about the real usefulness of Equation (i) as a method of air purification, comes from Peral and Ollis in their study of the deactivation of titania photocatalysts by heteroatomcontaining VOCs, such as decamethyl tetrasiloxane (DMTS), indole and pyrrole (24). These workers found that VOCs containing Si or N heteroatoms appear to generate products that cause the irreversible and complete deactivation of titania photocatalysts. The nature of the poisons is unknown. These findings and the concerns they raise about the usefulness of Equation (i) for air purification appear to have been largely ignored by other researchers in the field. Obviously, more work on heteroatom VOCs and their possible poisoning effects needs to be conducted.

There have been many attempts to improve the performance of photocatalytic air purification systems by incorporating a pgm, such as Pt or Pd, on the semiconductor surface. But, as with the water purification work, there has been mixed success. Table V contains data from reports on the enhancement, or otherwise, by Pt and Pd in Equation (i) for air purification (25—31). The E- factors are not large and may be ≤ 1. Again they appear to vary considerably, depending on the source and nature of the titania, method of pgm deposition and nature of the pollutant (amongst other things).

One major difference between the photocatalytic systems for air and water purification is the relative ease of carrying out the process at elevated temperatures (T > 100°C). This ease has been exploited recently by Kennedy and Datye in a study of the photomineralisation of ethanol by platinised and unplatinised titania (26). The study shows that while the rate of photocatalytic mineralisation of ethanol by TiO₂ is largely unaffected when the reaction temperature is increased from 50 to 170°C, with Pt/TiO₂ the rate is markedly increased at temperatures > 140°C. This is illustrated in Figure 4(a) using plots of the rates of CO₂ generation as a function of reaction temperature for UV-illuminated TiO₂, Pt/TiO₂ and non-illuminated Pt/TiO₂ photocatalysts (26). The results show that the enhanced rates of ethanol destruction recorded for Pt/TiO₂ at T > 140°C are not due simply to the catalysed thermal oxidation of ethanol by the Pt deposits, a measure of which is given by the plot of the dark rate of CO₂ generation as a function of temperature in Figure 4(a). Instead, there seems to be a synergistic effect between the semiconductor photocatalytic and the Pt-catalysed thermal oxidation modes of ethanol mineralisation at temperatures over 140°C, that is, region B in Figure 4(a). It is suggested that this synergistic effect arises because the TiO₂ photocatalyst generates acetaldehyde as an intermediate and this species is then rapidly oxidised on the Pt islands via a thermal process at the elevated temperatures associated with region B (26).

The findings of Kennedy and Datye (26) are very encouraging with regard to Pt and Pd usage in Equation (i): they indicate that air purification by semiconductor photocatalysis may be achieved more efficiently if the process is carried out at elevated temperatures and in the presence of a pgm, such as Pt or Pd.

However, generalising from one result is not possible. This is shown in recent work by Falconer et al. (30) on the photomineralisation of gaseous acetaldehyde by Pt/TiO₂ at elevated temperatures. The titania they used was the same as used by Kennedy and Datye, Degussa P25 (26). Some of Falconer’s results (30) are illustrated in Figure 4(b) as plots of (aldehyde) % conversion to CO₂ versus reaction temperature for their UV-irradiated TiO₂ and Pt/TiO₂ and non-irradiated Pt/TiO₂ photocatalysts. The titania photocatalyst becomes poisoned at T > 100°C and while platinisation depresses the rate of mineralisation of acetaldehyde at ambient temperatures, the observed enhanced rates at elevated temperatures appear to be associated with additional thermal oxidation of acetaldehyde on the Pt islands, rather than a synergistic effect. As a consequence, the variation in % conversion of acetaldehyde versus reaction temperature profile for Pt/TiO₂ in Figure 4(b) appears to be the sum of the average % conversion of acetaldehyde due to photomineralisation at ambient temperature on TiO₂ plus the variation with temperature in the % conversion due to thermal oxidation of acetaldehyde by oxygen, catalysed by Pt. The presence of Pt on the titania prevents, or slows down, the formation of poisoning agents associated with this system, especially at elevated temperatures (30).

These two studies show that pgms may have an important role to play in the semiconductor photocatalytic purification of air at elevated temperatures (26, 30). This role may be as a synergistic catalyst that enhances the rate of photomineralisation of the VOC (as with the Pt/TiO₂-ethanol system of Kennedy and Datye (26)), or as a simple catalyst to thermally oxidise the VOC under test and/or any poisoning intermediates that may be generated (as in the Pt/TiO₂-acetaldehyde systems of Falconer (30)). Thus, the exact nature of the role played by a pgm deposit in semiconductor photocatalysis for air purification appears to depend, amongst other things, on the pollutant.

Looking to the near future, it is probable that, tempted by the promise of synergy and greatly enhanced purification rates, many more studies of air (and possibly water) purification by semiconductor photocatalysis will be conducted at elevated temperatures, T > 100°C, and this work will be dominated by semiconductor photocatalysts with pgm deposits.

Visible Light Studies

The review so far has focused on the role of Pt and Pd as catalytic materials to purify water and air by semiconductor photocatalysis, dominated by the semiconductor titania. However, titania as a photocatalyst has one major drawback: it absorbs only UV light and so can only utilise 2–3% of the solar spectrum. One major objective of semiconductor photocatalysis research is therefore to develop a stable, active photocatalyst that utilises a wider part of the solar spectrum.

Despite much effort little progress had been made until recently, when Kisch and coworkers looked at the visible light sensitisation of the photodegradation of 4-CP by TiO₂ (amorphous or crystalline) impregnated with pgm chlorides, such as chlorides of Pt(IV), Pd(II) and Rh(III) (32). Their most active sample appeared to be a TiO₂- chloroplatinate(IV) photocatalytic material, prepared by stirring the TiO₂ powder into a H₂PtCl₆ solution and calcining the yellow-brown filtered product at 200°C for 2 hours. Rutile TiO₂ (from Aldrich) did not appear to chemisorb the H₂PtCl₆ and so did not work as a photocatalyst using visible light. The activity of the P25 TiO₂- chloroplatinate(IV) product appeared low due to presence of Pt in both the (II) or (0) oxidation state, as well as in the desired (IV) oxidation state.

Their most active and effective TiO₂-Pt(IV) chloride samples generated to date for visible light photocatalysis use 100% anatase TiO₂ (from Kerr- McGee) and amorphous TiO₂ (prepared by a sol-gel process) (32). The suggested role of the Pt(IV) chloride in such TiO₂-Pt(IV) chloride photocatalytic systems for water purification is illustrated in Figure 5 (32) for the destruction of the test pollutant, 4–CP.

In this scheme the electronically-excited Pt complex, created by the absorption of UV or visible light, undergoes homolytic cleavage of a Pt—Cl bond to generate intermediates of Pt(III) and an adsorbed chlorine atom. The former species injects an electron into the conduction band of the semiconductor which subsequently reduces oxygen to superoxide and, eventually, leads to the generation of a hydroxyl radical. Hydroxyl radicals and the adsorbed chlorine atoms are assumed to oxidise the 4-CP, eventually converting it to CO₂ and HCl. The yellow-brown TiO₂-Pt(IV) powders appear quite stable under visible light irradiation and can be used for several days as photosensitisers for water purification (32).

However, this system inevitably degrades, especially under UV light illumination, as the Pt(IV) chloride is slowly photoreduced by the organic pollutant, in this case 4-CP. Other, less effective TiO₂-noble metal chloride, visible light photosensitisers for water purification have been prepared by Kisch and coworkers, using gold and rhodium chlorides (33). It appears likely that these other sensitisers operate via a mechanism similar to that indicated in Figure 5, and thus also suffer from the same problem of a slow but, ultimately limiting, photoinstability. Therefore, although such systems have attracted a great deal of attention, their future as a route to sensitising Equation (i) with visible light, appears to be very limited.

Conclusions

Semiconductor photocatalysis can effect the complete mineralisation of a wide range of organic pollutants, invariably using a UV-absorbing titania semiconductor, generally either Aldrich anatase, Hombikat UV 100 or Degussa P25. The efficacy of Pt and Pd deposits on titania as oxygen reduction catalysts has been extensively studied, aimed at improving the overall purification rate. Enhanced and depressed rates of photomineralisation with platinisation (or palladisation) have been reported for both water and air purification systems. Any positive E-factor is not usually large, typically 2—4, and can vary considerably. The large variation in the E-factor is attributed to the complicated nature of the overall process and depends on many factors including: adsorption of pollutant, oxidation of pollutant and concomitant reduction of oxygen and desorption of products, all of which may be positively or negatively affected by the presence of a pgm. Important rate-determining reaction conditions include: (a) the nature of the pollutant, (b) the source and history of the semiconductor, (c) the method (and amount) of pgm deposited, and (d) the reaction solution pH and temperature.

Although semiconductor photocatalysis can be used to purify air, VOCs containing a heteroatom, such as N and Si, appear able to deactivate the photocatalyst. Air purification can be conducted easily at temperatures > 100°C and the presence of Pt deposits markedly improves the overall rate of VOC photomineralisation. For some VOCs, such as ethanol, the effect is a synergistic combination of semiconductor photocatalysis and thermal oxidation of the pollutant, catalysed by Pt. With other VOCs, such as acetaldehyde, this effect appears to be the sum of the two process, although the Pt deposits may protect the semiconductor from poisoning at very elevated temperatures. Chloro-Pt(IV)-titania and other chloro-pgm-titania complexes have recently attracted attention as possible visible light sensitisers for water or air purification. However, at present they lack long term photostability.