CHIMICA ACTA

ELSEVIER
Analytica Chimica Acta 311 (1995) 115-122

ANALYTICA

Application of TiO, sol for UV radiation measurements

Jiansheng Huang, Huiling Ding, William S. Dodson, Yuzhuo Li
Department of Chemistry, Clarkson Unic~ersity, Potsdam, NY 13699, USA 1994; revised 21 March 1995; accepted 21 March 1995 Received 27 December

Abstract
The use of a chemically stable TiO, sol as an actinometer for UV radiation was investigated. The sol produces a visible color change under UV radiation. It can be regenerated to its original transparency upon exposure to oxygen. The actinometer can be used to compare the strengths of different light sources for a wide range of photoinduced chemical processes involving TiO, particles such as photodegradation of industrial waste or organic synthesis on TiO,, photoinduced deterioration of TiO,-containing materials, and the effectiveness of a TiO,-based sunscreen product. Other potential applications include ly-ray radiation monitoring, oxygen detection, and oxygen permeability measurement for polymer films.
Kryword.cc TiO,; Actinometer; Polymers

1. Introduction
Irradiation of TiO,,

proper wavelength (Eq. 1). TiO,+hv-+e-+h+

a semiconductor, with a causes direct band gap excitation

(1)

The electron in the conduction band (e-> and the electron hole (h+) in the valence band can quickly recombine or interact with an adsorbed electron ac. ceptor and an electron donor, respectively [I]. The two crystalline forms of TiO,, anatase and rutile, exhibit similar physical and chemical properties. For example, they both have low toxicity and high stability in the presence of strong acids and bases. However, their behaviors in the presence of light are very different. For anatase, electron transfer processes

* Corresponding

author.

between excited TiO, particles and adsorbed electron donors/acceptors are found to be much more efficient than for its rutile counterpart. Therefore, anatase has been extensively used for applications involving such electron transfer processes as phototo-electrochemical energy conversion, photoinduced waste treatment, and organic synthesis [2-61. For rutile TiO,, the charge recombination between electron and electron hole is much faster than electron transfer reactions between TiO, particles and adsorbents. This translates to a low photoactivity and an effective conversion of UV light into heat. The low photoactivity and its desirable UV absorption spectrum that covers the entire erythemal curve make rutile TiO, an ideal choice as UV blocker for sunscreen products [7]. Rutile TiO, particles are also widely used as whitener pigments for papers and paints due to its low photoactivity and high reflective index in the visible region. For the forementioned applications, an actinome-

0003-2670/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0003-2670(95)00174-3

116

J. Huang et al. /Analytica

Chimica Acta 311 (1995) 115-122

ter is needed to compare the effectiveness of different light sources. The actinometry is particularly important for a process involving sunlight due to the fluctuation of its strength from time to time and location to location. A conventional actinometer involves either a photolabile compound, such as potassium ferrioxalate, or a photoelectrometer [8,9]. A disadvantage of these conventional methods is that the UV-visible absorption spectra of the actinometers are generally wider than TiO,s absorption profile. The results, therefore, have to be carefully corrected for the actual overlap between the two spectra. The main objective of this work is to develop a new actinometer based on the photochemical properties of TiO, particles. A clear advantage of the new method is that it measures the light with wavelengths which is covered and potentially utilized by TiO, particles. In this paper, the basic principle of the new actinometer is first introduced. The characteristics of the actinometer are then described followed by a discussion of factors that influence the sensitivity of the actinometer.

viscosity measurements were performed with a Cannon-Fenske viscometer. Light scattering experiments were conducted on a Brookhaven Instrument BI200SM Light Scattering Goniometer. All measurements were repeated at least three times either parallel or sequentially to insure the validity. 2.2. Preparation of TiO, sol samples

2. Experimental

2.1. General
Ethylene glycol (99%) was obtained from Alfa Products. Titanium(IV) isopropoxide (97%) glycerol (95%), potassium ferrioxalate, and l,lO-phenanthroline were obtained from Aldrich. Ethyl alcohol was obtained from Pharmaco Products and hydrogen chloride (38%) was Baker analyzed grade. All reagents were used as received without purification. UV-visible spectra were obtained on a Perkin Elmer Lambda 6 UV-Vis spectrophotometer. Irradiations were conducted in a Rayonet photochemical reactor equipped with up to 16 light bulbs (General Electric F8T5 BLB, F15T8 BLB, or F15T8 BL) depending upon the desired radiation strength. For most experiments, radiation strengths were calibrated with a broadband Gentec Joulemeter (ED500). For wavelength dependency experiments, a set of bandpass filters (Esco Products, Oak Ridge, NJ) was used. A literature procedure was followed for the actinometry experiments using potassium ferrioxalate [8]. The

TiO, sols were prepared by hydrolysis of titanium isopropoxide, Ti(OC3H7)4 and followed by a condensation at room temperature [lO,ll]. While stirring, a calculated amount (40-200 ~1) of concentrated (13 M) HCl was added to 12 ml of a solvent. To the solution, Ti(OC,H,), was then added to achieve the volume ratio of Ti(OC3H7)4 versus solvent ranging from 0.01 to 0.4. The solution was stirred for 10 min and then allowed to age in the dark. Polyethylene (PE) films (ca. 40 ,um thickness) were prepared from pellets with a blown film line. The films of polyvinyl chloride (PVC) and polyvinyl alcohol (PVA) were prepared by casting a solution of polymer without any plasticizer and allowed to dry in the air at room temperature. Methylene chloride was the choice of solvent for polyvinyl chloride and hot water for polyvinyl alcohol. The aged TiO, sol was injected into a rectangular polymer film package and purged with nitrogen to remove the dissolved oxygen. The prepared polymer film package (20 X 10 X 1 mm) was then sealed and ready for measurements. For pH dependency and reusability studies, regular 10 mm quartz cuvettes as well as polymer packages were used.

3. Results and discussion

3.1. Trapped electrons in TiO, matrix

In addition to the redox reactions and charge recombination process, the electrons and/or electron holes can be trapped on the semiconductor particle surface. For example, in the presence of a good electron donor such as an alcohol for the electron hole (Eq. 2), TiO,(e-) can be readily trapped [12]. Both electrons and electron holes can be trapped in the presence of an alcohol and acetic acid [13]. RCH,OH + h+ + H+ f RCHO + 1/2H, (2)

J. Huang et al./Analytica

Chimica Acta 311 (1995) 115-122

I

117
I I I I

The trapped electrons give a strong absorption in the visible range. The photophysical and chemical aspects of the electron/electron hole trapping have been carefully investigated by Kamat [3], Gratzel [12], Vinodgopal et al. [13] and Serpone et al. [14]. The electron and/or electron hole trapping mainly occur on anatase and amorphous TiO, particles. Other than its potential use in energy storage, such a controllable charge separation system has not been recognized for any practical applications. To a certain extent, the number of electrons trapped in the matrix is linearly proportional to the number of photons absorbed by TiO, and the number of effective band-gap excitations. The absorption spectrum of the trapped electrons, therefore, can be used for actinometry. Due to the similarity between their UV absorption spectra, an anatase based actinometer should also be useful for a system containing rutile. In this paper, a sol containing mostly anatase and some amorphous TiO, in alcohol is used. 3.2. Spectroscopic properties of the TiO, samples

3.5

__
3.0 --~~
-.

PE PVA

-.-.-

PVC SolA
so,
B .._..

2.5

_..-.._

. .

so, c

: c iii

2.0
/:li
;j!i : j:;; :

-SoID .._. pc

i 9 1.5

;:

;:l!

Ji\!
iti ,...

:!I ;,..
I!!

1.0

0.5

I-0.0

200

300

400

500

600

Wavelength

(nm)

Fig. 1. The UV absorption spectra of polymer films (PE = polyethylene, PVA = polyvinyl alcohol, PVC = polyvinyl chloride, PC = Pyrex cuvette) and TiO, sols with different concentrations (see Table 1 for the notation of sample A-D).

The TiO, sol prepared according to the procedure described in the Experimental section was colorless and stable in the dark for months. The UV absorption spectra of the TiO, sols with different concentrations are shown in Fig. 1. It is known that the UV

absorption spectra of quantum size semiconductor particles are blue shifted compared to those of bulk size particles [ 151. For TiO,, the quantum size effect does not cause any significant spectrum change un-

Table 1 UV-visible

absorption

spectra of TiO, sols enclosed in between PVC films and their sensitivities Cut-off wavelength (nm) 356.6 348.4 338.8 334.7 338.3 337.5 337.1 337.0 348.4 348.4 Sensitivity B (AOD/mJ 4.66 2.95 1.94 1.63 2.64 3.33 3.32 0.57 2.54 2.94 d b cm-)

to BL and BLB lamps

Sample pH and initial a, Ti(OiPrJ,:solvent (v/v) A, 1.36, B, 1.36. C, 1.36, D, 1.36. E, 1.06, F, 0.89, G, 0.76. H, 0.00, B, I .36. B, 1.36. 0.375: I 0.25O:l 0.12S:l 0.063:1 0.12S:l 0.125: 1 0.125: 1 0.125:1 0.250: 1 0.750: 1

SD. 0.07 0.08 0.04 0.05 0.05 0.05 0.04 0.13 0.04 0.04

R 0.998 0.997 0.997 0.996 0.997 0.998 0.999 0.995 0.998 0.998

a TiO, sol in between PVC films. b Sensitivity to BL lamps, obtained from the slope (B) of the linear plot of each set of experimental data (Y = BX). where YtAODJ represents the TiO,(e-) absorption, X is the UV radiation dose in mJ/cm *, SD. is the standard deviation, and R is the correlation coefficient. The Y intercept is essentially zero for all samples. BLB lamps were used. d BLB lamps were used and the sensitivity was corrected with emission spectra difference between BL and BLB lamps.

118

J. Huang et al./Analytica

Chimica Acta 311 (1995) 115-122

less the particles are smaller than 0.1 nm [16]. The particle size distributions for the prepared TiO, sols were measured with a light scattering technique. The results indicated that the particles were 50-80 nm in size with some 300-500 nm aggregates. Although the particle sizes vary somewhat depending upon the type of solvent used and the amount of Ti(OC,H,), added, the particles are all within the category of bulk particles. Therefore, the slight differences in the UV absorption profiles (Fig. 1) for the sols that contain different amounts of TiO, are mainly due to the concentration effect. The results shown in Table 1 summarize the UV cut-off wavelengths, at which the absorbance equals to unity, for different concentrations of TiO,. It is known that the energy gap of semiconductor particles, AV, depends on the pH of the environment (Eq. 3) [17]. TiO, sols with different pH, therefore, should have different absorption onsets, a reflection of the different energy band gaps. AV= V,, - 0.059pH (3)

The TiO, sol is injected into a polymer package or a glass cuvette (Pyrex, Vycor, or Quartz) before use. The UV absorption spectra of three different types of polymer materials (PVA, PVC, and PE) and Pyrex glass were also overlayed in Fig. 1. All containers except Pyrex have practically no absorptions above 220 nm. An extensive UV irradiation of the polymer films may cause photochemical changes, it usually requires much higher doses [18,19]. Therefore, the interference from the polymer films is negligible. Although quartz or Vycor materials allow more UVB light passing through, Pyrex glass is more economical to be used in a solar driven catalytic processes.

3.3. Viscosities

of the samples

It can be estimated from Eq. 3 that a 1 nm blue shift of the onset requires an acidic sol to be 0.15 unit lower in pH. The sols prepared for this study have a high content of amorphous TiO, and particles with a large amount of hydroxyl groups that can act as buffer sites. Therefore, a slight change in the amount of acid in the system does not influence the UV absorption spectrum (Table 1). This property allows TiO, sols to resist the perturbation from a small amount of acid introduced during the electron hole quenching process (Eq. 2).

The viscosity of a sol is important in terms of maintaining its physical shape in a polymer package. Table 2 lists the viscosities of some prepared TiO, sols and their solvents [20]. A TiO, sol prepared from titanium isopropoxide in ethyl alcohol (0.25:1, v/v) is ca. 1.5 times more viscous than its pure solvent, ethyl alcohol. A similar sol in ethylene glycol is three times more viscous than its solvent while a sol in glycerol is 13 times more viscous. The viscosity increase seems to be proportional to the number of hydroxyl groups in a solvent molecule. This may be related to the fact that the increase in viscosity is mainly caused by the network formed among the polymeric TiO, and the solvent. It is also shown in Table 2 that the viscosity increase is

Table 2 Relative viscosity Solvent

of TiO, sols in different solvents and their sensitivities Ratio a 7j of solvent (mm* s-l) 1.35 360.3 15.6 77 of sol (mm s1.95 4761 15.8 17.2 19.0 23.5

to BL lamps
771/%

Sensitivity B

S.D. 0.04 0.03

1
1.44 13.20 1.22 1.51 3.02 5.01

Ethanol Glycerol Ethylene Glycol

0.25:l 0.2S:l 0.06: 1 0.13:1 0.2.5:1 0.38:1

3.16 2.95

1.96

0.04

a Initial volume ratio of Ti(OC,H,), to solvent (v/v). b Sensitivity to BL lamps, obtained from the slope (B) of the linear plot of each set of experimental data (Y=BX), where Y (AOD) represents the TiO,(e-) absorption, X is the UV radiation dose in mJ/cm*, S.D. is the standard deviation, and R is the correlation coefficient (R = 0.998 for all three cases). The Y intercept is essentially zero for all samples.

J. Huang et al./Analytica
0.45

Chimica Acta 311 (1995) 115-122

119

0.40

0 35

0.30 Q) ii 2 025 z In 2 0.20

Radiation time (min)

advantage for a dosimetric system to maintain its absorption profile during the measurement. The trapped electrons can be kept for months in the absence of 0,. However, they quickly react with oxygen to give oxygen radical anions. Therefore, after illumination, the colorlessness of the sol could be easily restored by purging with air, allowing the trapped electrons to be scavenged by oxygen (Eq. 5). O,(g) + em+ O,-(ads) (5)

0.15

0.10

0.05 : ) IOC

I.

I.

I.

-I

400

500

600

700

900

900

Wavelength

(nm) ) in glycerol sol

Fig. 2. UV-visible absorption spectra of TiO,(eenclosed between PVC films.

directly proportional to the Ti(OC3H7)4 concentration. Although a TiO, sol with ethyl alcohol as the solvent is more viscous than its pure solvent, it is not enough to sustain a ridged shape in the package. Therefore, ethylene glycol and glycerol are much better choices. 3.4. W radiation of TiOz samples The trapped TiO,(e-) in the matrix gives a strong absorption in the visible range (Fig. 2). The intensity of this absorption is proportional to the UV dose as expressed in a linear equation: Y = 1.63X, S.D. = 0.05,
R = 0.998 (4)

where Y (AOD) represents the TiO,(e-) absorption of a sample B (Table 1) irradiated with a UV (F15T8 BL) lamp, X is the UV radiation dose in mJ/cm, the Y intercept is essentially zero, S.D. is the standard deviation, and R is the correlation coefficient of a linear regression fit. As described before, the hole-quenching process does not generate enough protons to change the acidity of the sol significantly. Therefore, the UV absorption spectrum of a sol should remain unchanged throughout the irradiation process. It is an

Under the influence of oxygen, the blue color of TiO,(e-) disappears instantaneously. The sol packages can thus be reused. The self-recovery process can be controlled by the selection of different polymer films. The oxygen permeability and the thickness have a direct impact on the regeneration of the transparency. The retention of the blue color (trapped electrons) was found to be a function of the polymer films oxygen permeability [19]. For example, the PVA film retains the blue color for days whereas PE film gives the shortest retention time (ca. 1.2 h). The TiO?/PVC package completely discolored in 40 h in air without UV radiations. The rates of discoloration, therefore, can be controlled by using polymer films with different O2 permeabilities. Since it would require only a trace amount of oxygen to quench the trapped electrons in the matrix, precautions must be taken in the selection of polymeric materials and the control of oxygen content in order to achieve desirable and consistent results. At the same time, the high sensitivity of this process could be used in the detection of oxygen for an enclosed environment and the evaluation of the permeability of a polymeric film. Although the solvent is consumed during the electron hole quenching and recovery processes, a regenerated sol has the same characteristics as a fresh one and can be reused several times for UV radiation dose measurements. Eq. 4 shows the linear relationship between absorbance at 540 nm and the LJV radiation dose under the same strength for several UV radiation dose measurements. Within the experimental error (+3%), the sol package has the same sensitivity after several regenerations. Because the actual number of electrons trapped each time in the matrix is relatively small, up to ca. 70 per particle [12], the portion of solvent that is consumed during the quenching of electrons should also be small

120

J. Huang et al./Analytica

Chimica Acta 311 (1995) 115-122

enough so that the function is not compromised.

of the dosimeter package

3.0r-----l
2.5 -

------

Sample C
Sample E
fz

3.5. Influence of Ti(OC, H7j4 concentration vents

and sol-

....-.... Sample
-

-.-.-.-. Sample G

A linear relationship between the absorbance at approximately 540 nm and the UV radiation dose was found (Table 1) for four different Ti(OC,H,), concentrations in ethylene glycol. The sensitivity of the TiO, sol samples towards UV light increases with the Ti(OC3H714 concentration. It was also found that when the ratio of Ti(OC,H,), to ethylene glycol was larger than 0.35, the sol became an unstable gel at room temperature. The optimum Ti(OC3H7)4 to solvent volume ratio was determined to be 0.25, at which both high sensitivity and stability were achieved. Three different solvents, ethyl alcohol, ethylene glycol, and glycerol were employed to study the influence of the matrices on the sensitivity of the titanium dioxide sol packages for UV radiation measurements. When the initial volume ratio of Ti(OC,H,), to solvent was 0.25, the prepared titanium dioxide sol in ethyl alcohol yields the highest sensitivity. The sensitivity of the sol in ethylene glycol is slightly lower than that in ethyl alcohol and that of the sol in glycerol is much lower (Table 1). The sensitivity seems to be proportional to the electron donating abilities of the solvents that follow the order of ethyl alcohol > ethylene glycol > glycerol. Although ethyl alcohol gives the highest sensitivity, the sample has a less rigid shape due to its lower viscosity. In addition, its high volatility makes it difficult to keep the sample to a constant volume. Therefore, ethylene glycol is the overall optimum choice. Although the UV absorption spectra are not significantly affected by the pH of TiO, sols, the absorption profiles of the trapped electrons are. As shown in Fig. 3, the absorption maxima of the trapped electrons are significantly blue shifted in the sols with higher HCl concentrations. This observation is consistent with the fact that the conduction band energy is significantly decreased in a more acidic medium [12,17]. Since the valence band is also lowered at the same time, the band gap of the semiconductor particles is not significantly affected

Sample H

0.0 400

500

600

700

Wavelength

(nm)

Fig. 3. The absorption maxima of the trapped electrons in TiO, sols in Pyrex cuvettes with different pHs (see Table 1 for notation of samples B and D-H).

(Eq. 3). The UV absorption onsets of the particles, thus, do not have a noticeable change (Table 11. The sensitivity of a TiO, sol to the UV light is a function of the acid concentration as illustrated by the slopes listed in Table 1. The sensitivity of a sol increases as the pH of the sample decreases. This is due to the fact that the conduction band of the TiO, particles is lowered and it is easier or more efficient for alcohol molecules to donate electrons to the conduction band. This effect reaches its maximum at ca. pH 0.9. A further addition of acid causes the absorption maximum of the trapped electrons to submerge into the UV absorption of TiO, particles (Fig. 3). The observed sensitivities for the samples with a pH < 0.9 were based on the shoulder peaks near the TiO, absorptions. Therefore, their apparent sensitivities are lower than the sample with pH 0.9. 3.6. Sensitivities to different excitation sources

Two types of UV lamps were used to examine the sensitivity of the actinometer towards different excitation sources. The light sources are the Black Light phosphor lamps (BLB) with the maximum

.I. Huang et al. /Analytica

Chimica Acta 311 (1995) 115-122

121

radiation peak of 350 nm and a half-height peak width of 40 nm, and the Black Light lamps (BL) with the peak of 350 nm, a larger peak width (ca. 50 nm) and a less cut-off in the blue region compared to the BLB lamps. The BLB lamps are widely used in commercial suntan booths while the BL lamps are the choices for many photochemical reactors in laboratory. The dosimeter is more sensitive (AO.D./min at 540 nm) to the BL lamps than to the BLB lamps under the same condition (Table 1). The difference in responses is due to the lower wavelengths that are blocked in the BLB lamps to protect the suntan-booth goers. The photoexcitation efficiencies corrected with the actual photon flux are about the same for both lamps. For sample B as listed in Table 1, the quantum yield of trapped electrons was estimated to be 0.12 based on the molar absorptivity (1200 M- cm- ) reported by Gratzel [12] and the total photon flux at 337 f 5 nm calibrated by using potassium ferrioxalate [8]. We have also tested the application of the dosimeter for a solar process involving TiO, particles. Specifically, an aromatic nitro compound was reduced to its amino derivative on TiO,. The reaction progress (conversion of the nitro compound) was monitored by liquid chromatography (LC). The solar energy was simultaneously measured by the TiO,based actinometer and a photoelectrometer. It is found that the LC results have a good linear correlation with the light strengths measured by the TiO,based dosimeter. The results, however, do not correlate well with the light strengths measured by a photoelectrometer because the photoelectrometer responds to a wider range of wavelengths and is also sensitive to other factors such as temperature. Therefore, the new dosimeter can offer a simple and reliable measurement for the relative sunlight strengths and the quantum efficiency of a process involving TiOz. To measure the amount of light that has been endured by a material containing anatase and/or rutile TiO, such as paper or paint, one may also use the TiO, sol based dosimeter. However, precautions must be taken for light scattering and competitive absorption by other chromophores such as inorganic and/or organic coatings. For example, it has been reported that vectorial electron injection from excited organic dye into the conduction band of a semicon-

ductor particle leads to charge separation and subsequent electron transfer reactions [ 131. A dosimeter was also made into a small polymer package containing the TiO, sol (Fig. 2). Such a device could be used by a suntan-booth/beach goer who can determine the need of applying a (TiO, based) sunscreen product based on the degree of color change. The dosimeter color change indicates how much protection would be rendered if a rutile TiO,-based sunscreen product had been applied. The charge separation at the semiconductor surface initiated by other energy sources such as y-ray radiation. Preliminary results indicate that a TiO, based dosimeter is more sensitive to y-radiation than other methods including PVC films and glasses [21-231. The detailed results will be the subject of a forthcoming report.

4. Conclusion A new type of UV actinometer was developed. It can be used to measure a wide range of radiation doses with a set of tailored sample conditions. Potential applications in related areas include oxygen demeasurement for tection and oxygen permeability polymeric material.

Acknowledgements The work is supported in part by the Center for Advanced Materials Processing at Clarkson University and the U.S. Environmental Protection Agency under the Cooperative Agreement CR822063-01-0, a program managed by Mr. William Barnard at EPA.

References
[l] H. Al-Ekabi, in V. Ramamurthy (Ed.), Photochemistry in Organizied and Constrain Media, VCH, Weinheim, lY91, p. 495. [2] G.R. Helz, R.G. Zepp and D.G. Crosby, Aquatic and Surface Photochemistry, Lewis Publishers, 1994, p. 261. [3] P.V. Kamat, Chem. Rev., 93 (1993) 267. [4] M.A. Fox and M.T. Duly, Chem. Rev., 1 (1993) 341. [5] C. Joyce-Pruden, J. Pross and Y. Li. J. Org. Chem., 57 (1992) S087.

122

J. Huang et al./Analytica

Chimica Acta 311 (1995) 115-122 [16] J.M. Nedeljkovic, Ph.D. Dissertation, Clarkson University, 1991, p. 23. [17] Y. Matsumoto, T. Yoshikawa and E. Sato, J. Electrochem. Sot., 136 (1989) 1389. [18] E. Reichmanis, C.W. Frank and J.H. ODonnell, in Irradiation of Polymeric Materials, American Chemical Society, Washington, DC, 1993. [19] F. Rodriguez, in Principles of Polymer Systems, Hemisphere Publishing, New York, 1989, p. 454. [20] N.B. Vargaftik, in Handbook of Physical Properties of Liquids and Gases, Hemisphere Publishing, New York, 1983, p. 406. [21] H.M. Khan, G. Ahmad, A. Sattar and S.K. Durrani, J. Radioanal. Nucl. Chem., Articles, 125 (1988) 127. [22] B. Whittaker and C.A. Lowe, J. Applied Radiation and Isotope, (1967) 89. [23] B.L. Qupta and R.H. Bhat, Chemistry Symposium, Madras, 1970, p. 57.

[6] F. Mahdavi, T. Bruton and Y. Li, .I. Org. Chem., 58 (1993) 744. [7] L.H. Kaminester, Arch. Dermatol., 117 (1981) 66. [8] S. Murov, I. Carmichael and G.L. Hug, in Handbook of Photochemistry, Marcel Dekker, New York, 1993, p. 299. [9] B.L. Diffey, in N.J. Lowe and N.A. Shaath (Eds.), Sunscreens, Development, Evaluation, and Regulatory Aspects, Marcel Dekker, New York, 1990, p. 93. [lo] B.E. Yoldas, Appl. Opt., 21 (1982) 2960. [ll] E. MatijeviC, M. Budnik and L. Meites, J. Colloid Interface Sci., 61 (1977) 302. [12] M. Gratzel, in E. Pelizzetti and N. Serpone (Eds.), Homogeneous and Heterogeneous Photocatalysis, Reidel, Dordrecht, 1986, p. 91. [13] K. Vinodgopal, I. Bedja, S. Hotchandani and P.V. Kamat, Langmuir, 10 (1994) 1767. [14] N. Serpone, D. Lawless, J. Disdier and J.-M. Herrman, Langmuir, 10 (1994) 643. [15] P.V. Kamat and N.M. Dimitrijeevic, Solar Energy, 44 (1990) 83.