e used asball millingn smallerfiberoptic

t particleeases up toreflectione particles,

Journal of Colloid and Interface Science 276 (2004) 354–363www.elsevier.com/locate/jcis

Physical properties of organic particulate UV absorbers usedin sunscreens

II. UV-attenuating efficiency as function of particle size

Bernd Herzoga,∗, Katja Quassa, Erika Schmidtb, Stefan Müllera, Helmut Luthera

a Ciba Specialty Chemicals Inc., Grenzach-Wyhlen, Germanyb Solvias AG, Basel, Switzerland

Received 2 March 2004; accepted 8 April 2004

Available online 12 May 2004

Abstract

In this study the UV-attenuating properties of microparticles consisting of a benzotriazole derivative were investigated, which arabsorbers for UV radiation in cosmetic sunscreens. The particles were micronized in presence of a dispersing agent by means of aprocess. According to the energy input different particle sizes were produced in the range of 0.16 to 4 µm. In order to study eveparticles, the sample with particle size 0.16 µm was fractionated further by centrifugation. Particle sizes were measured usingquasi-elastic light scattering (FOQELS) and laser diffractometry. The UV-attenuating properties of the dispersions with differensizes were assessed using UV spectroscopy. With decreasing particle size the efficiency of the UV extinction of the dispersion incra particle size of 80 nm. For particles smaller than 80 nm the UV extinction decreases again indicating an optimum at 80 nm. Fromspectroscopic measurements it was found that scattering makes about 10%, and absorption 90%, of the UV-attenuating effect of thwhich are obtained at the end of the milling process. 2004 Elsevier Inc. All rights reserved.

Keywords:Sunscreens; Particulate UV absorbers; Particle size; Absorption; Scattering; Reflectance

ainstain

on-reen. Inngaypar-thethe

s-nds

dze de-

d thema-gy of

esows

howtter-inceuchrue

UVasis, are

1. Introduction

Sunscreens used for the protection of human skin agthe harmful effects of solar radiation must contain ceramounts of UV-absorbing substances[1]. Most of the UVfilters used are oil-soluble or even oil-miscible and csequently are incorporated into the oil phase of sunscemulsions. However, solubility sometimes is a problemorder to circumvent this difficulty, UV absorbers showivery poor solubility in oils and being insoluble in water mbe micronized leading to aqueous dispersions of smallticles of the UV-absorbing material. In a previous paperdetermination of the particle size of such particles duringcourse of micronization using wet milling technology hasbeen reported[2]. The aim of the present work is to invetigate how the attenuation efficacy of UV radiation depeon the size of the UV-absorber particles.

* Corresponding author. Fax: +49-7624-122888.E-mail address:bernd.herzog@cibasc.com (B. Herzog).

0021-9797/$ – see front matter 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2004.04.009

t

Inorganic oxide particles like TiO2, which are also useas active substances in cosmetic sunscreens, show a sipendence of the extinction spectrum[3]. The UV attenuationresults as the sum of the semi-conductor absorption anstrong scattering due to the high refractive index of suchterials. As particle size decreases, the band-gap enerTiO2 increases causing a shift ofthe extinction maximumtoward lower wavelengths[4].

For organic colored pigments the absorption increaswith a decrease of particle size while the scattering sha maximum at a certain particle size[5]. The extinction re-sults as a superposition of both effects. Thus, it may san optimum at a certain particle size, even though scaing is not as strong as in the case of titanium dioxide sthe refractive index for most organic materials is lower. Sfeatures valid for colored pigments should similarly be tfor pigment-like organic UV-absorbing particles.

The UV-attenuating properties of particulate organicabsorbers as function of particle size, with special emphon the differentiation between absorption and scatteringsubject of the present investigation.

B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363 355

yl-illre

is

ofom

ll-egum

d toheg af

ecylcol

on

i-c-entn ofithTus-htment

t at-

en-o-

de-tra-

ofs

herer-erth

theis

sionring

(w/v)in

lowo 2%

singndory2).-

calpic

e

-

ermck-

sin-u-ing

INri-

diipub-

Fig. 1. Structure of MBBT.

2. Materials and methods

2.1. Chemicals

The UV absorber used for this study is 2,2′-methy-lene-bis-(6-(2H -benzotriazole-2-yl))-4-(1,1,3,3-tetramethbutyl)-phenol from Ciba Specialty Chemicals Inc. and wbe abbreviated in the following as MBBT. The structuof this substance is shown inFig. 1. Its molar mass is659.00 g/mol. Because of its high melting point (195◦C)and low solubility (<10−8 g/l in water), the substancewell suited for micronization.

Highly diluted dispersions were stabilized by additionTween 20 (polyoxyethylene (20) sorbitan monolaurate, frUniquema).

2.2. Micronization by milling

The micronization of MBBT was performed with a bamilling process[2] in deionized water using decyl glucosid(Plantacare 2000, Cognis) as dispersant and xanthan(Rhodia) as thickener. First, the material was exposea premilling step with a corundum disc mill (Fryma). Tmain milling procedure was performed afterward usinball mill LMZ 60 (Netzsch). After micronization, 100 g othe aqueous dispersion contained 50 g MBBT, 7.5 g dglucoside, 0.2 g xanthan gum, and 0.4 g propylene gly(Fluka).

2.3. Separation of smaller particles by centrifugation

After the ball-milling process a particle size distributiwith a d(0.5) of 150 to 170 nm andd(0.9) of about 300 nmwas obtained[2]. In order to study the behavior of partcles with smaller sizes this distribution was further frationated by centrifugation, since larger particles sedimfaster than smaller ones. For that purpose after completiothe milling process the suspension was diluted 10-fold wbidistilled water resulting in a concentration of 5% MBB(w/v). Afterward the diluted suspension was centrifugeding a Sorvall SS-3 centrifuge with rotor SS-34 in which eigvessels of 50 cm3 volume can take place. The maximuvolume of the suspension for one centrifugation experimwas 225 cm3. The speed of the rotor was held constan15,000 rpm corresponding to 17,600g, and the centrifuga

tion time was varied between two and 60 min. After a ctrifugation run the supernatant was analyzed with respect t(i) MBBT concentration, (ii) particle size, and (iii) UV extinction of the particles.

(i) The MBBT concentration in the supernatant wastermined in order to be able for adjusting the concention afterward for further analysis. After extraction of theactive material with dioxane, in which the solubilityMBBT is 1% (w/v) at 25◦C, 0.05 g of the supernatant waweighed into volumetric flasks of 10 cm3 volume and diox-ane was added up to the calibration marking. After furt10 fold dilution with dioxane the concentration was detmined photometrically (UV/vis spectrometer Perkin ElmLambda 16) with quartz cells of 1 cm optical pathlengusing Lambert–Beer’s law (at a wavelength of 348 nmmolar decadic extinction coefficient of MBBT in dioxane32,000 l mol−1 cm−1).

(ii) The size of the particles in the supernatant suspenwas measured using fiberoptic quasi-elastic light scatte(FOQELS) as described in detail elsewhere[2]. For particlesize measurement the suspension was adjusted to 2%MBBT. However, at centrifugation times longer than 15 mthe concentration of MBBT in the supernatant was be2%. In those cases the suspension was concentrated tusing a rotary evaporator.

(iii) The UV extinction of the particle dispersions wadetermined also at MBBT concentrations of 2% usquartz cells with an optical pathlength of 0.0008 cm aa UV/vis spectrometer with integration sphere access(Perkin Elmer Lambda 16 with Labsphere B009-401From concentration and extinction the decadic molar extinction coefficient can be evaluated or theE(1,1) value,which is the extinction of 1% active material at 1 cm optipathlength. A detailed description of the UV spectroscomeasurements will be given below.

2.4. Fiberoptic quasi-elastic light scattering (FOQELS)

FOQELS [6] is a dynamic light scattering techniquoperating at a scattering angle of 180◦. This experimen-tal setup enables measurementsat high particle concentrations (>1%) without multiple scattering problems since onlyback-scattered light of a thin particle layer is detected[7].The light (λ = 632.8 nm) of a 15 mW He/Ne-laser (Spindl& Hoyer) is guided through a single mode optic fiber of 4 µdiameter which is immersed in the particle dispersion. Bascattered light from the particles (θ = 180◦) then enters thesame fiber and is guided to the detector (ALV SO-SIPDgle photon detector, ALV Laser Vertriebsgesellschaft). Atocorrelation functions of scattered light were recorded usthe correlator of an ALV 5000 instrument. The CONTalgorithm[8] was employed in order to resolve the distbution of diffusional modesG(Γ ), whereΓ is the inverse ofthe time at which the corresponding mode occurs[9]. Fromthis the volume weighted distribution of hydrodynamic raF V(R) can be obtained as has been shown in a previous

356 B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363

ers

ticleun-

distheua-

m-dif-

rn)dif-

inta.

the

uilt-it ise re-esspartwasper-

erure

wasefer

of

samac-ndref-ir

resite-as

am-len-ereav-

ereinc-n in

nterf thertztheere.he-

ionly,ncem-

andention

tion

f the

en-teentstion

lication [2]. It is possible to determine the particle diametd(0.5) andd(0.9) from F V(R). The d(0.5) values are theparticle diameters up to which the area under the parsize distribution curve is equal to 50% of the whole areader the distribution. Correspondingly, thed(0.9) values arethe diameters up to which the area under the particle sizetribution curve is equal to 90% of the whole area underdistribution. This can be expressed via the following eqtions:

(1)

∫ R(0.5)

0 FV(R) dR∫ ∞0 FV(R) dR

= 0.5 and

∫ R(0.9)

0 FV(R) dR∫ ∞0 FV(R) dR

= 0.9,

where 2R(0.5) = d(0.5) and 2R(0.9) = d(0.9). Thed(0.5)

andd(0.9) values obtained in that way can be directly copared to corresponding data of other methods like laserfractometry or disc centrifugation[2].

2.5. Laser diffractometry

For laser diffractometry a Mastersizer S device (Malvewas used. With this technique the angular distribution offracted or scattered light is measured[10]. For large particles(d � λ) the Fraunhofer approximation may be appliedorder to extract the particle size distribution from the daHowever, for small particles with sizes in the order ofwavelength of the incident light or below, Mie theory[11]must be employed. For this purpose the corresponding bin software of the device was used. In order to do so,necessary to know the complex and the real parts of thfractive index of the particles. In the case of the colorlUV-absorbing particles investigated here, the complex(at λ = 632.8 nm) can be set to zero, and the real partdetermined to 1.67 by index-matching light scattering eximents[2].

2.6. UV/vis spectroscopy

UV/vis spectra were measured with a Perkin ElmLambda 16 UV/vis double-beam spectrometer. For measment of clearly dissolved UV absorbers the instrumentused in its standard setup. The sample solution and the rence (pure solvent) were filled into quartz cells ofd = 1 cmoptical pathlength and placed into the respective beamsthe device.

For measurements of the particulate absorbers theinstrument was employed, but with an integration spherecessory in order to collect the direct transmitted light aalso the light scattered in forward direction. Sample anderence cells were placed in the respective light beams at theentrance into the integration sphere, while BaSO4 pressings(barium sulfate for white standard DIN 5033, Merck) wemounted into the sample and reference windows oppothereof. Measurements were performed with a spectral resolution of 2 nm. In order tokeep the samples as thinpossible, sandwich-type quartz cells withd = 0.0008 cm

-

-

-

e

Fig. 2. Lambert–Beer functionality of particulate MBBT ofd(0.5) =160 nm, measured atd = 0.0008 cm with integration sphere accessory.

optical thickness (Hellma, Germany) were used for sples and reference. Since thereproducibility of experimentadata with these cells was limited, for each individual suspsion 10 measurements always with freshly filled cells waveraged. The standard deviation was typically 7% of theerage value. For the determination of extinction coefficientssuspensions containing 1, 2, 2.5, and 3% of MBBT wmeasured in that way. A linear relationship between exttion and concentration was proofed. An example is showFig. 2.

2.7. Reflection spectroscopy

Reflectance measurements were performed with a PerkiElmer Lambda 9 UV/vis/NIR double beam spectromewith integration sphere accessory. Diluted suspensions oparticulate UV absorber (1 to 3%) were filled into quacells with 0.0008 cm optical thickness and mounted insample holder at the corresponding window of the sphThe samples were not “infinitely” thick. The reflection of tsample on white background,Rw, was measured by placing a BaSO4 pressing behind the sample cell. Reflectwith black background,Rb, was obtained correspondingbut with a black cone behind the sample cell. As refereserved a BaSO4 pressing. Before measuring a series of saples, for calibration of the instrument reflectance 0%100%,R0% andR100%, were determined. For measuremof R100%, sample and reference window of the integratsphere were covered with BaSO4 pressings, and forR0% ablack cone was mounted in the sample holder. All reflecdata were corrected according to

(2)R = Rsample− R0%

R100%− R0%· RBaSO4,

whereRBaSO4 was the reflectance of BaSO4 as certified bythe supplier of the powder. As reference served a cell osame optical thickness filled with distilled water.

The diffuse transmission of the samplesT diff was alsomeasured, placing the sample in the light beam at itstrance into the sphere with BaSO4 pressings in the opposisample holders. The experimental setup for measuremof diffuse transmission and reflectance using the integrasphere accessory is schematically drawn inFig. 3.

B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363 357

Fig. 3. (a) Measurement of diffuse transmission. (b) Measurement of reflectanceon white or black background.

thewas

ayerringnly

the-

ue

mis-ardthe

i-outeithn be

of aly ton-

ght

n

er-c

plesugh

tionhe

e

nd

-

dgbe

3. Theoretical aspects

3.1. UV transmission of particulate UV absorbers

The transmission of scattering particles depends onexperimental setup of the measurement. Transmissionmeasured with a directed light beam perpendicular to a lof suspended particulate UV absorbers. When measuthe transmission without an integration sphere, usually olight is detected that is transmitted in the direction ofincident radiation,T direct. In that case the extinction consists of a portion due to absorption,EA, and a portion due toscatteringES, the latter being the sum of the extinction dto forward- and back-scattering,EFS andEBS, respectively,expressed by the equation

(3)lg1

Tdirect= EA + EFS+ EBS.

When an integration sphere is used, the diffuse transsionT diff is measured, where the light scattered in forwdirection also is registered by the detector. In that caseextinction is given by the equation

(4)lg1

Tdiff= EA + EBS.

The difference ofEqs. (3) and (4)is the extinction dueto forward scatteringEFS. This may be obtained expermentally by measuring the spectrum of a sample withand with integrating sphere and calculating the differencof both. The relative contribution of forward scattering wrespect to the total attenuation of the radiation can thedetermined via the ratioEFS/(EA + ES).

It is to be emphasized, that the experimental setupspectrometer with integrating sphere corresponds closethe situation of human skin covered with a thin film of a suscreen: The skin also “detects” the directly transmitted lias well as the light scattered in forward direction (Fig. 4).

3.2. Scattering and absorption coefficients from reflectiospectroscopy[12]

The following equations allow the evaluation of scatting and absorption coefficientsfrom reflection spectroscopi

Fig. 4. Analogy between UV-transmission measurements of thin samusing an integration sphere and collection of light by human skin throthe film of a sunscreening agent.

data. The diffuse transmission measured with integrasphere accessoryT diff was corrected for the decrease of tincident intensityI 0 due to back-scattering (Rb). The cor-rected diffuse transmissionT then becomes

(5)T = Tdiff/(1− Rb).

The scattering coefficientS is calculated according to thequation

(6)S = 1

b

(Ar sinh

(b

T

)− Ar sinh(b)

)1

d,

whered is the optical pathlength of the respective layer, ab is defined via equation

(7)b =√(

(a − Rb)2 − T 2),

where

(8)a =(

1+ R2b − T 2

2Rb

).

The Kubelka–Munk functionK/S (the ratio of absorption coefficientK and scattering coefficientS) is related tothe functiona, which is defined viaEq. (8), in the followingway:

(9)K

S= a − 1.

SinceS is known fromEq. (6), K can also be determinemaking use ofEq. (9). The relative contribution of scatterinwith respect to the total attenuation of the radiation cancalculated from the ratioS/(K + S).

358 B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363

cer-of

naseranangtry,ll be

ized.ilute

per-

herti-itessizeedk

.7the

or-

a-

cle

onl)

-

u-cles.ts

heheldedab-

themesandandat-

lesstillaretion

nd

ionow-tion

e

e

4. Results and discussion

4.1. Particle size distributions after centrifugation ofmicronized samples

The measurement of particle size distributions aftertain amounts of energy consumption during millingMBBT has been reported in detail in[2]. It has been showthat for particle sizes in the range larger than of 0.2 µm ldiffractometry is a suitable method whereas FOQELS cbe used most advantageous for smaller particles in the rof 10 to 500 nm. As is already usual for laser diffractomeparticle sizes obtained from FOQELS measurements wialso given asd(0.5) values as pointed out above.

After energy consumption of 740 kWh/1000 kg themilling process of MBBT is completed and a particle sdistribution with ad(0.5) value of about 160 nm is obtaineAs described above, a corresponding suspension was dto 5% MBBT and centrifuged at 17,600g. By increasing thecentrifugation time the average particle size in the sunatant became smaller. This is shown inFig. 5 in terms ofcumulative particle volume distributions.

With X-ray powder diffraction (XRD) measurements, tline width of diffraction peaks becomes sensitive to pacle size (inverse relationship), if the particles are crystalland smaller than 100 nm. The decrease of the particlein the supernatant with centrifugation time was confirmby XRD, where the full width of half maximum of the peaoccurring at 11.9◦ 2θ (and others) was by a factor of 1broader after the maximum centrifugation time than incase before centrifugation[13]. It was also deduced fromXRD results, that there is no significant amount of amphous material present in the particles.

4.2. UV spectra of MBBT as function of particle size

All extinction spectra referred to in this section were mesured with integration sphere accessory.Fig. 6 shows theE(1,1) spectra of MBBT suspensions with different partisizes obtained during milling, indicated by thed(0.5) values.The E(1,1) value refers to the extinction of a dispersicontaining 1% (w/v) of the UVabsorber at a (theoretica

Fig. 5. Cumulative volume weighted distributions of MBBT dispersions after centrifugation (from below: 0, 2, 5, 15, 60 min).

e

d

optical pathlength of 1 cm. The efficiency of the UV attenation was observed to strongly increase for smaller partiThis is in line with the effects known from organic pigmenof comparable size[5]. The effects can be explained in tfollowing way: With bigger particles certain amounts of tabsorbing material are not reached by light but are shieand cannot contribute to its attenuation. For that reasonsorption will be higher when particles are smaller. Onother hand, in this size range scattering efficiency becosmaller with decreasing particle size. Thus, absorptionscattering efficiency are inversely related in this respect,there might be an optimum of particle size for maximumtenuation of light.

Fig. 7 shows spectra obtained for centrifuged sampwith the amount of bigger particles reduced. There isan increase ofE(1,1) to be measured when the particlesbecoming smaller, but at a certain particle size the extincdecreases again.

The E(1,1) values at 360 nm obtained for milled acentrifuged samples are plotted inFig. 8as function of par-ticle size. There is an optimal effect in terms ofE(1,1)360at ad(0.5) value of 80 nm. The decrease of the extinctat smaller particle sizes can be interpreted in the folling way: At particle sizes smaller than 80 nm the absorp

Fig. 6. UV-extinction spectra of MBBTdispersions with different particlsizes obtained after milling;E(1,1) = extinction of a dispersion with 1%(w/v) active substance atd = 1 cm.

Fig. 7. UV-extinction spectra of MBBTdispersions with different particlsizes obtained after centrifugation;E(1,1) = extinction of a dispersion with1% (w/v) active substance atd = 1 cm.

B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363 359

s-er to

ll ex

av-ndder

ting

-

r-m.µm

ed bgeop-e of

t aof

ateare

ar-

-

at

sedmtthewithsize

lly

ralthe

m,

ofrentthe

Fig. 8.E(1,1) at 360 nm as function of particle size, given asd(0.5).

Fig. 9.E(1,1) at 360 nm (open and closed squares) and〈E(1,1)〉 averagedin the spectral range of 290 to 400 nm (open and closed circles); regresion lines are drawn for visualization of the trends; closed symbols refmilled, open ones to centrifuged samples.

stays constant, but scattering decreases. Thus, the overatinction will be lower.

Alternatively, instead of taking theE(1,1) value at theextinction maximum at 360 nm, one may also use theerage of theE(1,1) values in the range between 290 a400 nm, a measure which is proportional to the area unthe extinction curve in this spectral range and thus indicathe respective band-strength:

(10)⟨E(1,1)

⟩ =400∫

290

E(1,1)λ dλ

/ 400∫290

dλ.

In Fig. 9 the average〈E(1,1)〉 is compared to the maximumE(1,1)360 in the range of particle sizes withd(0.5) be-tween 0.05 and 0.25 µm. Whereas theE(1,1)360 decreaseswith increasing particle size,there is a plateau of the aveage〈E(1,1)〉 for d(0.5) values between 0.05 and 0.18 µThus, decrease of the particle size from 0.18 to 0.05leads to changes of the shape of the spectrum expresshigher values ofE(1,1)360. On the other hand, the averaextinction stays constant in this range, indicating that thetimum band-strength is already achieved at a particle sizd(0.5) = 0.18 µm.

Fig. 10 compares the spectra of particulate MBBT aparticle size of 169 and 38 nm together with the spectrumMBBT dissolved in dioxane. The spectra of the particuland dissolved MBBT preparations are similar, but there

-

y

Fig. 10. Comparison of the UV spectra of MBBT in dispersion at two pticle sizes and that of MBBT dissolved in dioxane.

Fig. 11. Determination of particle sized(0.5) from UV spectroscopic parameters.

some distinct differences: The long-wavelength maximum350 nm of the dioxane solution is shifted toλ = 357 nm forsmall particle suspensions, and this shift is further increafor bigger particles to 360 nm. Additionally, the spectruat all particle sizes (seeFigs. 6 and 7) shows a shoulder a380 nm. It can be noticed that the (negative) slope ofspectral curve between 360 and 380 nm becomes higherdecreasing particle size. Thus it may be used for particlecharacterization. In the range ofd(0.5) of 35 to 350 nm,a linear relationship of the following kind was empiricafound (Fig. 11):

(11)E360− E380

E360− E400= 0.9026− 0.3074 lg

(d(0.5)/nm

),

whereE360−E380 is proportional to the slope of the spectcurve between 360 and 380 nm, which is normalized bydifference of the extinction maximum atE360 and the effectof scattering, approximated with the extinction at 400 nE400 (there is no absorption at this wavelength).

In Table 1theE(1,1) data and the normalized slopesthe spectral curve between 360 and 380 nm for diffeparticle sizes of MBBT are summarized. At the end of

360 B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363

tha

tionds tople.ay

tion

ionroxi-s iseAsre-

ion

ithion

ithion

ctedgithemstion

he

rr

r-hered byfor-fls.

Table 1Spectral properties of MBBT as function of particle size

d(0.5)

(µm)〈E(1,1)〉(290–400 nm)

E(1,1) at360 nm

(E360− E380)/(E360− E400)

4.37 100±7 119±9 0.1140.780 184±13 226±16 0.1420.340 329±23 408±29 0.1400.360 311±22 393±28 0.1440.260 334±23 427±30 0.1630.230 392±27 508±36 0.1790.220 389±27 505±36 0.1790.220 400±28 522±37 0.1790.220 386±27 506±36 0.1890.177 419±29 556±39 0.2160.177 440±31 586±41 0.2160.174 422±29 565±40 0.2190.169 396±28 525±37 0.2110.161 435±30 579±41 0.2130.160 412±29 555±39 0.2130.157 400±28 532±38 0.2120.147 427±30 579±41 0.2410.142 445±31 600±42 0.2300.131 418±29 564±40 0.2340.103 441±31 612±43 0.2790.100 444±31 616±43 0.2810.098 431±30 597±42 0.2790.066 430±30 622±44 0.3810.065 409±29 581±41 0.3330.062 404±28 578±41 0.3450.061 405±28 576±41 0.3350.052 446±31 642±45 0.3670.050 412±29 599±42 0.3700.047 306±21 443±30 0.4210.042 351±25 510±36 0.4210.038 334±23 484±34 0.4270.032 346±24 515±36 0.463

milling process typically a particle size ofd(0.5) = 160 nmis obtained with anE(1,1) of 555± 39 atλ = 360 nm. Thecorresponding decadic molar extinction coefficient at thiswavelength is 36600± 2600 l mol−1 cm−1.

4.3. Extinction due to scattering and absorption fromUV-transmission studies

Fig. 12 shows the spectrum of MBBT particles wid(0.5) = 160 nm at 0.0008 cm optical pathlength andconcentration of 2% measured with and without integrasphere. The difference of those two spectra corresponthe extinction caused by forward scattering of the samThe spectrum of forward scattering obtained in this wshows maxima at the points of inflection of the absorpspectrum.

The structure of the first derivative of the absorptspectrum (measurement with integration sphere) is appmately congruent to the spectrum of forward scattering adepicted inFig. 12. This is consistent with the fact that thscattering intensity is proportional to the refractive index.is known from dispersion theory, in absorption bands thefractive index goes with the first derivative of the absorpt

Fig. 12. Determination of forward scattering of MBBT particles wd(0.5) = 0.16 µm from UV spectra measured with and without integratsphere.

Fig. 13. Determination of forward scattering of MBBT particles wd(0.5) = 1.02 µm from UV spectra measured with and without integratsphere.

spectrum. The spectrum of forward scattering is expeto be also influenced by the 1/λ4-dependence of scatterinintensity and the variation of the refractive increment wwavelength. But the dispersion of the refractive index seto have a major influence on scattering in the absorpband.

Figs. 13 and 14show similar measurements, with tsame experimental parameters as described forFig. 12, butfor different particle sizes of MBBT. Principally, for largeparticles ofd(0.5) = 1.02 µm forward scattering is highe(Fig. 13), with smaller particles ofd(0.5) = 32 nm it isstrongly diminished (Fig. 14). In the latter case the diffeence of measurements without and with integration spis very small and negative values may have been causeexperimental uncertainties. The effect of particle size onward scattering is visualized inFig. 15, where the ratio oextinction caused by forward scatteringEFS and the overalextinction (EA + ES) is depicted for the three particle sizeIn Table 2the data are listed for some wavelengths.

B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363 361

Table 2Extinctions of MBBT particles of different sizes from measurements with and without integration sphere at selected wavelengths

λ (nm) d(0.5) = 0.032 µm d(0.5) = 0.16 µm d(0.5) = 1.02 µm

EFS EA + ES EFS/(EA + ES) EFS EA + ES EFS/(EA + ES) EFS EA + ES EFS/(EA + ES)

280 0 0.372 0 0.029 0.358 0.08 0.110 0.352 0.31300 0.007 0.779 0.01 0.045 0.603 0.07 0.123 0.512 0.24320 0.032 0.647 0.05 0.057 0.576 0.10 0.129 0.491 0.26340 0 0.749 0 0.017 0.594 0.03 0.099 0.501 0.20360 0.043 0.923 0.05 0.059 0.789 0.07 0.132 0.619 0.21380 0.075 0.534 0.14 0.123 0.731 0.17 0.177 0.575 0.31400 0.014 0.033 0.42 0.157 0.339 0.46 0.232 0.369 0.63

Table 3Reflection spectroscopic parameters of MBBT particles withd(0.5) = 160 nm atc = 2.01% andd = 0.0008 cm at selected wavelengths

λ (nm) RBaSO4 T d Rw Rb a b Sd S (cm−1) K (cm−1) S/(K + S)

280 0.8670 0.3689 0.1357 0.0441 9.809 9.758 0.1020 127 1123 0.102300 0.9010 0.1682 0.0620 0.0421 11.557 11.514 0.1547 193 2041 0.087320 0.9350 0.1923 0.0696 0.0449 10.738 10.691 0.1540 192 1874 0.093340 0.9573 0.1693 0.0638 0.0427 11.401 11.357 0.1562 195 2031 0.088360 0.9645 0.0977 0.0517 0.0438 11.318 11.274 0.2062 258 2659 0.088380 0.9698 0.1516 0.0701 0.0519 9.441 9.388 0.2006 251 2117 0.106390 0.9725 0.3932 0.2079 0.0877 4.863 4.759 0.1942 243 938 0.206400 0.9744 0.7296 0.5504 0.1387 1.756 1.443 0.1877 235 177 0.570420 0.9770 0.8462 0.6978 0.1242 1.205 0.673 0.1465 183 38 0.830440 0.9790 0.8690 0.7205 0.1098 1.170 0.607 0.1262 158 27 0.855

n

ter-d

ithor-

r at-

urs

Fig. 14. Determination of forward scattering of MBBT particles withd(0.5) = 0.032 µm from UV spectra measured with and without integra-tion sphere.

Fig. 15. Fraction of forward scattering with respect to the total UV-atten-uating effect for three particle sizes of MBBT.

Fig. 16. Spectra of scattering coefficientS and absorption coefficientK ,and (K + S) from reflection spectroscopic data of MBBT withd(0.5) =0.16 µm.

4.4. Scattering and absorption coefficients from reflectiospectroscopy

As pointed out above, numerical values for the scating coefficients and absorption coefficients can be evaluatefrom reflection spectroscopic measurements.

For this purpose a preparation of MBBT particles wd(0.5) = 160 nm was used at a concentration of 2.01% (cresponding to a molar concentration ofc = 0.0305 mol/l).In Table 3the reflection spectroscopic data are given fonumber of wavelengths.Fig. 16shows the spectra of scatering coefficientS, absorption coefficientK and the sumof both. The maximum of the absorption coefficient occ

362 B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363

l

e

ic

b-ticle00

ivee-

rp-fect.

wasge.unt

ver-hen

ringge o

ere(ex-

ollerara-

tnc-ncesent.fhusa-

te-t ofin-theat-andentsdrp-um

nt ofith

on,or-

nc-.

bay-

vel-ew

oid

ho-

89–

41–

Fig. 17. Fraction of scattering coefficientS with respect to the totaUV-attenuation represented by (S + K) of MBBT with d(0.5) = 0.16 µm.

at λ = 360 nm withK = 2660 cm−1. From K the molardecadic extinction coefficient can be calculated using thequation

(12)ε = K

2.3026c,

wherec is the molar concentration[14]. With c = 0.0305mol/l andK = 2660 cm−1 the result for the molar decadextinction coefficient isε = 37800 l mol−1 cm−1. This iswithin the range of experimental error of the value otained from extinction measurements with the same parpreparation using an integration sphere attachment (366±2600 l mol−1 cm−1).

When investigating UV-absorbing particles the relatcontribution of scattering to the overall extinction is of spcial interest. This is shown inFig. 17 in terms of the ratioS/(K + S) versus wavelength. In the region of the absotion band, scattering makes about 10% of the overall efWith the respective sample ofd(0.5) = 0.16 µm the portionof forward scattering reported in the previous sectionaround 7% of the overall extinction in this spectral ranThe difference of 3% is roughly consistent with the amoof back scattering given by the values ofRb listed in Ta-ble 3.

With the inorganic UV absorber TiO2 at a particle sizeof 100 nm, scattering contributes to about 50% to the oall extinction in the range of the extinction maximum of tspectrum[15]. The reason for this difference to MBBT cabe seen in the much higher refractive index of TiO2 which isabout 2.6 compared to 1.67 of MBBT. However, with TiO2particles as small as 20 nm (a usual size of TiO2 primary par-ticles for application in cosmetic sunscreens), the scattecontribution also decreases to less than 10% in the ranthe maximum of the extinction spectrum.

5. Summary

The spectroscopic properties of particulate MBBT winvestigated as function of particle size. Particle sizes

f

pressed in terms ofd(0.5) values) in the range from 4 t0.16 µm were generated by wet milling technology. Smaparticle sizes down to 32 nm were achieved by size seption using a centrifugation technique.

The UV-attenuating efficiency in terms ofE(1,1)360 in-creased with decreasing particle size up to a maximum aa particle size of 80 nm. With smaller particles the extition started again to decrease.The shape of the extinctiospectrum of particulate MBBT shows significant differencompared to that of the substance in an organic solvThe spectrum changes also characteristically as function oparticle size in the size range of 30 to 300 nm, and toffers an additional possibility for particle size determintion.

Measurements of the extinction with and without ingration sphere allowed the determination of the amounforward scattering of the MBBT particles. This amountcreases with particles size. With a particle size of 0.16 µmportion of forward scattering in comparison to the overalltenuation effect in the spectral range of the absorption bmakes in average about 7%. From reflection measuremthe scattering and the absorption coefficient were determinefor one particle size (0.16 µm). In the region of the absotion band the scattering coefficient is about 10% of the sof absorption and scattering coefficient. Thus, the amouback-scattering can be estimated to 3%, what is in line wthe reflectance measurements.

In conclusion it becomes clear from this investigatithat a certain fineness of MBBT particles is necessary inder to achieve a satisfactory efficacy in terms of UV extition, so that thed(0.5) value should be well below 300 nm

Acknowledgments

Helpful discussions with Dr. Wolfgang Schlenker (CiSpecialty Chemicals), Dr. Werner Sieber (Ciba SpecialtChemicals) and Dr. Beat Freiermuth (Solvias AG) are gratefully acknowledged.

References

[1] N.A. Shaath, in: N.J. Lowe, N.A. Shaath (Eds.), Sunscreens—Deopment, Evaluation, and Regulatory Aspects, Marcel Dekker, NYork, 1997, pp. 3–33.

[2] B. Herzog, A. Katzenstein, K. Quass, A. Stehlin, H. Luther, J. CollInterface Sci. 271 (2004) 136–144.

[3] V.P.S. Judin, V.T. Salonen, SÖFW J. 119 (1993) 491–496.[4] R. Van Grieken, J. Aguado, M.J. Lopez-Munoz, J. Marugan, J. P

tochem. Photobiol. A 148 (2002) 315–322.[5] D.H. Clewell Jr., J. Opt. Soc. Am. 31 (1941) 521–527.[6] H. Wiese, D. Horn, J. Chem. Phys. 94 (1991) 6429–6443.[7] H. Wiese, D. Horn, Ber. Bunsenges. Phys. Chem. 97 (1993) 15

1597.[8] S.W. Provencher, Makromol. Chem. 180 (1979) 201.[9] S. Zhou, S. Fan, S.C.F. Au-yeung, C. Wu, Polymer 36 (1995) 13

1346.

B. Herzog et al. / Journal of Colloid and Interface Science 276 (2004) 354–363 363

ew

–

ein-

94)

[10] P. Bowen, J. Dispers. Sci. Technol. 23 (2002) 631–662.[11] H.C. Van de Hulst, Light Scattering by Small Particles, Wiley, N

York, 1957.[12] G. Kortüm, Reflexionsspektroskopie, Springer, Berlin, 1969, pp. 122

123.

[13] B. Freiermuth, Solvias AG, Basel, unpublished results.[14] W. Schmidt, Optische Spektroskopie, second ed., Wiley–VCH, W

heim, 2000, p. 302.[15] J.L. Robb, L.A. Simpson, D.F. Tunstall, Drug Cosmet. Ind. 154 (19

32–39.