lable at ScienceDirect

Environmental Pollution 158 (2010) 3482e3489

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Aging of TiO2 nanocomposites used in sunscreen. Dispersion and fateof the degradation products in aqueous environment

Jérôme Labille a,b,*, Jinghuan Feng a, Céline Botta a, Daniel Borschneck a,b, Magali Sammut a,Martiane Cabie c, Mélanie Auffan a,b, Jérôme Rose a,b, Jean-Yves Bottero a,b

aCEREGE UMR 6635 CNRS/Aix-Marseille Université, Europôle de l'Arbois, 13545 Aix-en-Provence, Franceb International Consortium for the Environmental Implications of Nanotechnology iCEINT, Europole de l'Arbois, 13545 Aix en Provence, FrancecCP2M, Université d'Aix Marseille, Case 221, 13397 Marseille cedex 20, France

Aging in aqueous conditions of TiO2 nanocomposite used in sunscreenstable in suspension.

s induces rapid generation of nanometric byproducts remaining

a r t i c l e i n f o

Article history:Received 2 November 2009Received in revised form10 February 2010Accepted 12 February 2010

Keywords:NanoparticleTitanium dioxideAlterationLifecycleDispersionAggregationNatural organic matter

* Corresponding author at: CEREGE, Europole Méd13545 Aix en Provence cedex 04, France.

E-mail address: labille@cerege.fr (J. Labille).

0269-7491/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.envpol.2010.02.012

a b s t r a c t

Aging in water of a TiO2-based nanocomposite used in sunscreen cosmetics has been studied asa function of light and time. It consisted initially in a TiO2 core, coated with Al(OH)3 and poly-dimethylsiloxane (PDMS) layers. Size measurement, coating alteration, and surface charge were followedby laser diffraction, TEM/EDS, ICP-AES and electrophoretic mobility measurement.

The nanocomposite rapidly underwent progressive dispersion in the aqueous phase, enabled by thedissolution of the PDMS layer. A stable suspension of colloidal byproducts from 50 to 700 nm in size wasformed. Their positively charged Al(OH)3 surface was evidenced with an isoelectric point around 7e8,controlling the dispersion stability. The critical coagulation concentrations measured with NaCl and CaCl2was 2 � 10�2 and 8 � 10�3 M respectively. The presence of natural organic matter affected the colloidalstability according to the NOM/byproduct ratio. A 2 wt% ratio favored bridging flocculation, whereas a 20wt% ratio induced sterical stabilization.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Manufactured nanomaterials have moved well-beyond labora-tory settings and are now found in a variety of commercial prod-ucts. As of August 2009, the inventory on nanotechnology-basedconsumer products has grown by nearly 379 %, from 212 to 1015products, since March 2006 (NanotechProject, 2009). Cosmetics,clothing and personal care products are the predominant types ofnano-products on the current market, representing more than 50%.Current trends all indicate that human exposure to nanomaterialswill continue to increase in the coming years as more products aredeveloped and put on the market. Continued application devel-opment, and increased production of nanomaterials, have bothresulted in elevated anxieties concerning the potential risks thatare associated with nanotechnology (E.C., 2004; EPA, 2005;SCENIHR, 2005). Accidents, normal aging during regular use, orinappropriate storage of the related wastes are possible ways for

iterranéen de l'Arbois, BP80,

All rights reserved.

the dissemination of nanomaterials, or their related byproducts,into the environment.

The more recent reviews on the risks of nanotechnology allagree that the surface properties of nanoparticles play a significantrole in determining the risks that these materials may pose to theenvironment and human health (ICON, 2008). On one hand, thisraises the need for nanoparticles characterization to identify thecausality between observed hazards and specific physical andchemical properties (EPA, 2005). On the other hand, this raises thequestion of the environmental relevance of the studied nano-particles. Most of the nanoparticles involved in industrial processesare indeed surface-modified, functionalized, or coated so as to getnew surface properties that match those of the final commercial-ized nanomaterial. The probability that free and bare nanoparticleswill be released from the nanomaterial over its lifecycle is very low.The related products released will more probably consist of nano-particles embedded in more or less altered matrixes.

The size, surface properties (i.e. bioavailability) and environ-mental impacts of these byproducts depend on the type(s) ofphysical and chemical reactions that the nanomaterial is exposed toduring aging and on their extents of progress. If the first steps of

J. Labille et al. / Environmental Pollution 158 (2010) 3482e3489 3483

aging certainly produce macrometric and weakly bioavailablebyproducts consisting of millions of nanoparticles cementedtogether in a matrix, more advanced aging steps standing further inthe lifecycle may assume higher probability in releasing nano-metric and bioavailable byproducts. These latter byproducts arelikely to display unknown and unstable surface properties. Deter-mining the characteristics of these byproducts is of critical impor-tance to predicting their environmental behavior. To date however,little experimental data exists on the aging reactions that nano-materials may encounter during their lifecycle, or on the fate of theassociated degradation byproducts that are formed from thesereactions.

Among the wide variety of commercialized nanomaterials, allrequesting further investigation thorough environmental assess-ment (Senjen, 2009), the UV blocker nanoparticles used insunscreens appear as the nanomaterial category with the highestpriority for future exposure studies (EWG, 2009; Wijnhoven et al.,2009). Titanium dioxide (TiO2) nanocomposites have been used insunscreens for over 20 years, well before the recent concern withthe nanotechnology risk assessment. The typical size of TiO2nanoparticles in sunscreen ranges from 10 to 100 nm (Nohyneket al., 2007) for esthetic and efficiency reasons. An estimated1000 tons of nanoparticles were used in sunscreens worldwidefrom 2003 to 2004 (Börm et al., 2006). Early studies revealed lowcutaneous penetration (EWG, 2009) and absence of toxicity(SCCNFP, 2000) with these nanocomposites. However, when theyare washed from the skin, their dumping into the aquatic envi-ronment most probably induces their aging and generation of newresidues with new surface properties, for which the fate and impactare unknown.

Indeed, TiO2 nanoparticles are photocatalytic when exposed toUV light (Reddy et al., 2007; Huang et al., 2008; Jones et al., 2008;Tsuang et al., 2008), which can cause oxidative stress and celldamage in the ambient environment. Such UV reactivity is usuallyeliminated in sunscreens by using TiO2 nanoparticles that arecoated with magnesium, silica, alumina or zirconium (Mills and LeHunte, 1997; Wakefield et al., 2004; Pan et al., 2009). However, onecan wonder about the lifetime of this inert coating on thosenanocomposites that are released from the sunscreen during itslifecycle.

This is the topic of the current communication, which is aimedat experimentally studying the aging and fate of a TiO2 e basednanocomposite that is commonly used in sunscreens. It was shownin a preliminary work (Auffan et al., in press) that after contact withwater, the TiO2 nanocomposite becomes hydrophilic and formsaggregates in solution due to the desorption and oxidation of theouter amphiphilic coating layer.

However, no data was yet given on the kinetics of this trans-formation, and on the dispersion and fate of the altered nano-composites in the environment. This has been complemented inthe present investigation, where the alteration reaction was fol-lowed with time, and where the residues formed were character-ized in terms of size, surface charge, and chemistry. The effect ofsalt and natural organic matter concentrations on the colloidalstability of the suspended byproducts was also studied in order todescribe the mechanisms that determine the dispersion andbioavailability of the byproducts.

2. Materials and methods

2.1. Studied nanocomposite

The TiO2 nanocomposite studied here, T-Lite� SF (BASF, Germany), was selectedfor relevance criteria based on its wide use in sunscreen. According to themanufacturer, the nanocomposite consists of a TiO2 core that is comprised ofnanometric lattice short sticks having an average size of 10� 50 nm. The TiO2 core is

coated with a layer of aluminum oxide [Al(OH)3] whose purpose is to cancel anyphotocatalytic effect during UV excitation. The Al(OH)3 also enhances grafting ofan organic and amphiphilic outer coating, which favors dispersion of the nano-composite in the organic cream. A polydimethylsiloxane (PDMS), also calledmethicone/dimethicone copolymer as a trade name, is used for this purpose inT-Lite� SF. The copolymer consists of [Si-O(CH3)2]n repeat units. The globalnanocomposite is supplied as a pulverulent white powder. It was not submitted tofurther treatment prior to the experimental aging procedure described below.

2.2. Physicalechemical characterization of the nanocomposite

The mineralogical characterization of the nanocomposite was achieved by X-raydiffraction. It was performed on powder samples with a Panalytical X'Pert Pro MPDdiffractometer using cobalt Ka radiation (l¼ 1.79�A) at 40 kV and 40 mA (Auffanet al., in press). The total counting time during XRD measurements was 10 h.

The original and altered nanocomposites were analyzed by transmission elec-tronmicroscopy (TEM) to investigate the size and shape of the TiO2 crystallites. TEMobservations have been performed on a JEOL 2010F operating at 200 kV, equippedwith KEVEX EDS analysis system for chemical analysis. Samples were prepared byevaporating a droplet of a TiO2 dispersion on a carbon coated copper grid at roomtemperature.

The chemical composition of the original nanocomposite was determined byICP-AES using a Jobin-Yvon Horiba Ultima C, after alkaline dissolution of the powder.The procedure consisted of melting at 1100 �C a mixture of commercial powder/LiBO2 at a weight ratio of 1:2, cooling till glass formation, and dissolution in HCl(2.4 N) as the carrier solvent for ICP measurement.

2.3. Aging procedure

The aging of the TiO2 nanocomposites was studied by introducing 100 mg ofthe nanocomposite powder into a large mouth 1L glass beaker containing 250 mLof ultrapure water. Aging was simulated by submitting this mixture to magneticstirring at 690 rpm. Natural sunlight was simulated using a sodium discharged400W lamp, located 30 cm from the open reactor, and continuously cooled viaa connection to air extraction. Water loss from the sample due to evaporationwas compensated for by adding appropriate amounts of ultrapure water to thereactor on a daily basis. To study the aging of TiO2 samples in the absence of light,the reactor was covered and capped with aluminum foil. These samples were sub-jected to the same procedure under the same time and temperature conditions.During all aging experiments, the solution pH was not modified and was frequentlychecked. No significant evolution was observed from the initial value around6.3� 0.2.

2.4. Kinetics of aging

In order to study the kinetics of the aging reaction, small aliquots (20 mL) of thenanocomposite mixture were sampled under stirring at different times up to 7 days.Each aliquot was then submitted to the following characterization procedure.

Dispersion kinetics: Laser diffractionwas used to measure the size distribution inthe aqueous mixture. The apparatus used was a Malvern Mastersizer S (MalvernInstruments�, Malvern, UK) displaying a measuring range from 50 nm to 900 mm.

In order to quantify possible changes to the different phases that constitute thenanocomposite, the leaching of dissolved byproductswas studied. Samples were firstfiltrated through a 25 nm membrane filter made of mixed cellulose ester(VSWP02500, MilliporeTm) so as to extract the liquid from the solid matter. UsingICP-AES, Al and Si elements were quantified in the filtrate as respective probes forthe possible dissolution of Al(OH)3 and PDMS coatings.

The nanocomposite's surface charge was measured in terms of electrophoreticmobility in order to better understand the evolution of the nanocomposite surfacechemistry with aging time. The apparatus usedwas aMalvern Zetasizer Nano Z fromMalvern Instruments� (Malvern, UK), working in mixed field mode (Minor, 1997).

2.5. Characterization and fate of the byproducts recovered after 48 h of aging

In a second approach, the time corresponding to 48 h of aging was selected aspresumably sufficient to get a steady state in the alteration reaction of the nano-composite. The aging experiment was stopped at this time, and the byproducts wereallowed to settle in the dark for 48 h in order to separate the colloidal phase fromreadily settleable aggregates. The colloidal suspension was further characterized asfollows.

Its solid content was quantified by gravimetric analysis after drying overnight inan oven at 100 �C.

The z potential of the colloidal byproducts was measured as a function of pH, inorder to determine their isoelectric point. For this purpose, the pH was adjustedfrom 3 to 10 using HCl or NaOH. Two different samples were used to move the pHfrom natural toward more alkaline or more acidic conditions to avoid memoryartifact.

Turbidity measurements were used to determine the limit conditions for thecolloidal stability of the byproduct dispersion in terms of critical coagulation

J. Labille et al. / Environmental Pollution 158 (2010) 3482e34893484

concentration (CCC) of salt. The colloidal byproduct suspension initially recoveredwas first diluted at 400 % in Milli Q water, and then distributed in a series of testtubes. The electrolyte (NaCl or MgCl2) concentrationwas adjusted from 10�3 to 10�1

and 10�4 to 10�2 mol/L, respectively in order to study the effect of ion valence oncoagulation. Samples were left 48 h in the dark to allow settling of the coagulatedones, The supernatant turbidity was measured then using a Hach 2100 ANturbidimeter.

In order to investigate the effect of natural organic matter (NOM) on the stabilityof the byproduct suspension, the same approach was followed as previouslydescribed. In addition to varying the electrolyte concentration, NOMwas introducedin the suspension at a ratio of 2 or 20 wt% with regard to the dry weight of thebyproduct. Three polysaccharides having high molecular weights (MW), from rhi-zospheric bacterial origin were tested: (i) Dextran, (neutral, MW¼ 2�106 Da) waspurchased from Sigma Aldrich and used without further treatment; (ii) Gellan(anionic, MW¼ 3�106 Da) (Bian et al., 2002) was supplied by Kelco (USA) and usedwithout further treatment; (iii) YAS34 (anionic, MW¼ 2�106 Da) (Villain-Simonnetet al., 1999) was obtained from ARD (Pomacle, France) and purified as describedelsewhere (Labille et al., 2005) by ethanol precipitation. Lower MW sources of NOM,humic acid and tannic acid (Sigma Aldrich), were also tested without furthertreatment.

3. Results and discussion

3.1. Original nanocomposite characterization

The XRD pattern of the initial nanocomposite (Fig. 1a) is char-acteristic of the TiO2 rutile form, which also contains some anatase

Fig. 1. X-ray diffraction pattern of the initial TiO2-based nanocomposite studied, referenimposed (a); TEM micrograph of the nanocomposite made of coated TiO2 sticks (b); High re(c); Schematic view of the nanocomposite formulation consisting of Al(OH)3 and PDMS coa

in a 1:10 ratio, as indicated by the shoulder at 29.5�. No other latticestructure was detected in the nanocomposite, implying the amor-phous structure of the Al(OH)3 and PDMS coatings.

The observation of the nanocomposite with TEM reveals thatthe rutile nanoparticles are in the stick form with 5e10 nm crosssection per 50e200 nm length (Fig. 1b). These sticks are arrangedtogether in large clusters having an average size of 200 nm. Onhigher magnification images, the 0.33 nm spaced lattice fringescorresponding to the (110) planes of TiO2 rutile confirmmost of theparticles are crystallized in the rutile structure (Fig.1c). TEM imagesalso show nanoparticles are embedded in an amorphous layer.However it is not possible to distinguish inside this coatingbetween both Al(OH)3 and PDMS amorphous layers. The Al(OH)3coating probably assumes the role of inter crystallites cement,as illustrated in Fig. 1d. The concentrations of TiO2, Al(OH)3, andSi-O(CH3)2 in the nanocomposite are given in Table 1.

3.2. Nanocomposite dispersion

As was anticipated for the amphiphilic PDMS coating, the initialbehavior of the nanocomposite followed expectations for a hydro-phobic material in aqueous media. The powder was initiallyretained at the airewater interface consistent with its hydrophobic

ce peaks (Swanson et al., 1969) characterizing anatase and rutile phases are lowersolution TEM image of crystallized rutile sticks constituting the original nanocompositetings (d).

Table 1weight composition of the initial nanocomposite; a : given by the manufacturer;b : measured in this study by ICP-AES.

wt %

TiO2 79e89a

Al(OH)3 11� 2.5b

Si-O(CH3)2 5� 1b

J. Labille et al. / Environmental Pollution 158 (2010) 3482e3489 3485

nature. Nevertheless, under stirring the solid/liquid contact wasforced, and the behavior of the nanocomposite evolved rapidly.Surprisingly, as soon as 30 min of aging, the solid phase started todisperse in the aqueous phase. The size distribution of the nano-composite recorded at this time (Fig. 2) is mainly constituted oflarge sizes around 200 mm certainly corresponding to hydrophobicagglomerates arranged in lumps to minimize the solid/liquidsurface contact. However, a second and minor class of sizes alsoarises around 300 nm, corresponding to well-dispersed particulatebyproducts, hereafter named colloidal byproducts. Its relativecontribution tended to increase rapidly with time, and reachedvalues of up to 35 vol% after 48 h of aging (Figs. 2 and 4a). In termsof a number concentration, this fraction corresponds to the majorclass of the size distribution. This submicron size enabled the cor-responding colloidal residues to remain stable in suspension whenstirring was stopped.

From the 400 mg/L initial nanocomposite/water mixture thatwas first prepared, a 100 mg/L stable suspension of colloidalbyproducts was finally obtained after 48 h of aging followed by 48 hof settling. From this suspension 25 wt% of the original nano-composite thus gets dispersed as submicron sized particles afteronly 48 h of aging. TEM (Fig. 3a) and HRTEM (Fig. 3b) observationsof these dried colloids revealed that they consist of the sameclusters of rutile lattice sticks as those constituting the initialnanocomposite (Fig. 1b). Such easy dispersion in the aqueous phasestrongly suggests a modification of the initially hydrophobicnanocomposite surface. This is in agreement with earlier observa-tions by Auffan et al. (Auffan et al., in press) who found that theamphiphilic coating agent was altered and desorbed in water.

3.3. Kinetics of the alteration reaction of the nanocomposite

The dispersion kinetics of the colloidal byproducts also dis-played a dependence on the incident light radiation. This isparticularly evident during the first 24 h of aging, where lightexposure enabled faster dispersion than in the absence of light(Fig. 4a), suggesting faster degradation of the amphiphilic coatingdue to light alteration.

Fig. 2. Volumic size distribution of the nanocomposite byproducts formed from agingunder light at 30 min, 2 h and 48 h.

This observation is confirmed by the time resolved quantificationof the dissolved Si that was released from the nanocomposite andinto solution, which shows the desorption and alteration of thePDMS coating (Fig. 4b) in the corresponding time ranges. In the caseof light-induced alteration, the increase in Si dissolved and releasedreached a maximum plateau after 1 day of aging. In the absence oflight, the plateau was obtained after 4 days of aging, confirming theslower dissolution of the PDMS. Surprisingly, at the plateau, the Siconcentration that was released and subsequently dissolved insolutionwas three times higher (6 ppm) in the absence of light thanwas measured in the presence of light (2 ppm). This correspondsrespectively to ca. 90% and 30% of the total Si initially present in thenanocomposite/water mixture. However, this does not contradictthe scenario according to which light radiation favors aging. Indeed,lower release of Si under light was certainly due to the remobiliza-tion of the PDMS residues at the surface of the nanocomposite, asshown by Auffan et al. (Auffan et al., in press). UV radiation is knownto enhance PDMS degradation by substituting hydroxyl groups tocarbonyl groups (Graubner et al., 2004). The resulting moleculemade of [Si-O(OH)x(CH3)2-x] units is more hydrophilic and reactive,and certainly ready to readsorb onto the nanocomposite surface. Thisresulted in a lower concentration of released Si than induced fromalteration in the absence of light where the slower degradation ofPDMS did not follow the same steps.

The time evolution of the byproduct z potential also confirmssuch surface degradation favored under light radiation (Fig. 4c).Negatively charged at the very first measurable step of thedispersion process, the nanocomposite surface chargewas certainlycontrolled by the newly formed silanol groups in the altered PDMSstructure. Silanols display an isoelectric point around pH 2,implying a negative charge at the natural pH (6.5) of the currentexperiment. This initial state was followed by an abrupt surfacecharge reversal, as measured by the positive z potential obtainedafter 3 h and 18 h of aging under light and dark conditions,respectively. Since this evolution was not accompanied by anysignificant variation in pH, the positive charge could not beassumed to arise from the amphoteric silanol groups only. Thisfurther confirms the desorption of the (Si-O)-based PDMS from thenanocomposite surface, and the subsequent exposure of theunderlying Al(OH)3 layer. This latter surface coating is character-ized by an isoelectric point of pH 7e8 (Thomas et al., 2002),implying a positive surface charge at pH 6.5.

The slower kinetics that were characteristic of the reactions inthe absence of light allowed us to identify a good correlationbetween z potential and the release of Si from the nanocomposite(Fig. 4b and c). Furthermore, we were able to characterize thedifferent steps in the alteration reaction. The first 12 h of aging, arecharacterized by a continuous increases in Si release (from 1 to4 ppm) and negative surface charge (from �10 to �70 mV). Thiscertainly corresponds to the progressive fractionation of the PDMScoating, inducing on one hand a continuous release of dissolved Siin solution, and on another hand a decreasing size of the PDMSsubunits remaining sorbed at the nanocomposite surface. Thislatter scenario implies an increasing surface density in the nega-tively charged (Si-O-) ends, and thus a progressively greater nega-tive surface charge. After 18 h of aging, when all of the PDMS wasdesorbed from the nanocomposite, further aging did not induceany significant evolution in the positive z potential controlled bythe Al(OH)3 surface, whereas the concentration in dissolved Si insolution still increased slowly up until 4 days. This likely corre-sponds to the time required for complete fractionation and disso-lution of the PDMS subunits that were previously desorbed andremaining free in suspension.

During exposure to light, the non-monotonous time evolutionfor z potential observed beyond the plateau in Si release is certainly

Fig. 3. TEM (a), HRTEM (b) and EDS spectrum (studied spot diameter¼ 150e200 nm) (c) of the colloidal nanocomposite formed via 48 h aging under light and recovered in thesupernatant after 48 h settling.

J. Labille et al. / Environmental Pollution 158 (2010) 3482e34893486

due to a more complex and unstable surface chemistry. The read-sorption of the hydroxylated PDMS subunits [Si-O(O-)] likelypartially screens the positive charge of the Al(OH)3 surface (Fig. 4b),and/or creates some interparticle bridges. As a consequence, thealtered nanocomposites are more likely to aggregate together,which results in the lower extent of sizes below 700 nm withregard to the “dark” reacted nanocomposite beyond 48 h of aging(Fig. 4a).

Attempts to quantify the release of Al from the nanocompositeduring aging by ICP-AES did not detect any measurable concen-tration in solution. However, Al was clearly detected by TEM/EDSanalysis on the altered nanocomposite, as evidenced by the corre-sponding peak at 1.5 keV (Fig. 3c). These results indicate that the Alremained mainly under a solid form at the nanocomposite surfaceduring aging. However, it may have undergone in situ dissolution/precipitation in a more stable mineral form when exposed to theaqueous phase (Auffan et al., in press).

Fig. 4. Time evolutions of the volume fraction of particles below 700 nm in size (a); ofthe release in dissolved Si (error bars of the order of 1%) (b); and of the z potential ofthe colloidal byproducts (c), induced from aging of the nanocomposite in the dark(black triangles) or under light (white squares).

3.4. Limit conditions to the colloidal stability of the byproductsformed

The colloidal stability of the byproduct suspension is mainlydriven by their positive surface charge, which would result instabilization of the dispersion by repulsive electrostatic interac-tions. Nevertheless, the balance between attractive and repulsiveinterparticle forces logically evolves with the pH and ionic strengthof the dispersing solution. Titration of the z potential values asa function of pH for the stable colloidal suspensions formed underdark and light conditions found that the isoelectric point for thesesuspensions were at pH 8 and 7.3 respectively (Fig. 5a). It is worthnoting however that the difference in the isoelectric points is notsignificantly different given the pH gap between each point. ThisIEP range is characteristic for an aluminum hydroxide surface(Thomas et al., 2002; Wu et al., 2006), in agreement with theremoval or consumption of the silanol groups from the surface ofthe suspended nanocomposites. Adjusting the solution pH close tothis isoelectric point will induce neutralisation of the amphotericaluminol groups, and thus annihilation of repulsive electrostaticinteractions, at the benefit of the short-range and attractive van derWaals forces. This results in aggregation of the nanocomposites andsedimentation of the now formed aggregates.

An increase in salt concentration screens the repulsive electro-static interactions between particles through compression of theelectric double layer that surrounds each charged particle. Thecritical salt concentration above which attractive forces becomepredominant, termed the critical coagulation concentration (CCC),determines the limit for colloidal stability. For the byproductsuspension formed with or without light, CCCs of 2�10�2 M NaCland 8� 10�3 M MgCl2 were determined using turbidity measure-ments (Fig. 5b). Lower salt concentrations of 5�10�3 M NaCl or2�10�3 M MgCl2 were sufficient to initiate slow coagulationmechanism and partial clarification of the supernatent. In a Cl�

concentration consideration, as plotted in Fig. 5b, the slight diver-gences observed between the different curves are not significant.Very similar CCCs are measured whatever the cation valency of thesalt used, This is due to the positive surface charge of the alterednanocomposites, which is screened by anionic counterions insolution (e.g., Cl- ions). Thus, the screening of positively chargedsurfaces is not dictated by cation valency, as initially investigated inthis experiment, but only depends on the Cl� concentration. Whatis more surprising is that the colloidal byproducts formed underlight or in the dark display a similar CCC despite their knowndifferent surface chemistries. A lower CCC was indeed expected forthe former, considering its lower zeta potential. This could be the

Fig. 5. Physical chemical behavior of the colloidal byproduct suspension formed after 48 h aging. z potential evolution according to pH (a); Supernatant relative turbidity after 48 hsettling, according to chloride ions concentration brought by NaCl or MgCl2 addition (b).

J. Labille et al. / Environmental Pollution 158 (2010) 3482e3489 3487

result of a competition between adsorbing Cl� counterions and[Si-O(O�)] units remaining on the particle surface. Total desorptionof the PDMS subunits may be achieved when increasing saltconcentration, resulting in a new surface chemistry very similar tothat obtained in dark condition of aging.

Moreover, the CCC reported in Fig. 4b appears relatively lowwhen compared to values obtained with monovalent salts on othertypes of colloids or nanoparticles. Indeed, CCCs of approximately5�10�2e10�1 M are more commonly measured for nC60 (Fortneret al., 2005), hematite (Zhang and Buffle, 1996), or natural claycolloids (Tombacz and Szekeres, 2004) in the same pH range (6e7).This is certainly due to the proximity of the byproducts isoelectricpoint, which implies a low surface charge and favors interparticleaggregation.

The presence and concentration of NOM in solution affected thestability of the byproduct suspension. At 2 wt% of NOM, the globaleffect was the destabilization of the suspension by flocculation,whereas at 20 wt%, it induced further stabilization (Fig. 6). Thiseffect has been observed elsewhere at the sameweight proportionsof NOM with geogenic colloids (Labille et al., 2003). These obser-vations are in good agreement with theoretical expectations forbridging flocculation and steric stabilization, respectively (LaMer,1964; Fleer and Scheutjens, 1986; Lafuma et al., 1991; Adachi andWada, 2000). The former mechanism requires low amounts of

Fig. 6. Turbidity measurement of the byproduct supernatant according to the presenceof natural organic matter and salt concentration, after 48 h settling. NOM/byproduct¼ 2 wt% (histograms) or 20 wt% (extension bars).

macromolecules to be sorbed to the colloid surface with regard tothe total adsorption capacity of the colloid to initiate interparticlebridging. Conversely, steric stabilization is obtained at highmacromolecule concentrations, where each surface is saturatedwith sorbed macromolecules, resulting in repulsive steric interac-tions and thus stabilization of the colloids. Previous studies haveobserved NOM-induced stabilization of metal oxide nanoparticles(Yang et al., 2009; Zhang et al., 2009), as well as for hydrophobiccarbon nanoparticles (Hyung et al., 2007; Xie et al., 2008; Chappellet al., 2009). However, flocculation or destabilization of nano-particles at lower NOM concentrations has not previously beenreported, since most of these earlier experiments were carried atNOM concentrations higher than 1 mg/L. This finding has impor-tant consequences on the global fate of the nanoparticles inaqueous environments. Typically, in surface water where NOMconcentrations range from 5 to 10 mg/L, the formation of stablenanoparticle dispersions may be favored. Conversely, in ground-waters where NOM concentrations range from 0.1 to 2 mg/L theflocculation of nanoparticles may be more likely.

The different NOM macromolecules that were evaluated hereat concentrations of 2 wt% were not all equal in terms of theirability to flocculate the colloidal byproducts. The YAS34 andhumic acid both demonstrated the greatest ability to flocculatethe byproducts, enabling an 80e90 % decrease in the turbidity inthe absence of any salt. The other NOM sources required that saltbe added to the solution to induce significant destabilization ofthe suspension. Salt addition indeed makes neighboring particlesto come closer to each other, which may be a prerequisitecondition for NOM-induced bridging flocculation, depending onthe macromolecule bridging efficiency. The less efficient macro-molecule, the more salt addition required. Tannic acid and gellanrequiring salt concentration lower than the CCC thus appear moreefficient flocculent than dextran which require salt additionhigher than the CCC. The potential effects of molecular weight andspatial conformation cannot explain these discrepancies, sinceYAS34 and humic acid are totally different in all these aspects,whereas YAS34 and gellan are more similar from this point ofview. Nevertheless, the effects of the NOM anionicity and of thelocalization of this charge in the macromolecule do not appearunreasonable as determining parameters. Indeed, the humic acidmolecule is themost negatively charged, and the repeat unit of theYAS34 macromolecule contains acidic residues on the end of theside chains, well exposed to the outermost (Villain-Simonnetet al., 1999; Labille et al., 2005). These characteristics favor rapidattractive electrostatic interactions with the positively charged

J. Labille et al. / Environmental Pollution 158 (2010) 3482e34893488

byproducts, which do not occur for the other types of NOM thatwere evaluated here. Indeed, the least efficient flocculent, dextran,has a neutral charge. The weak flocculent, gellan, displayed acidicresidues in the backbone of the macromolecule, hidden beside thefolded back side chains, thus few exposed to the outermost (Bianet al., 2002; Labille et al., 2005). Finally, the tannic acid wasa relatively good flocculent as soon as a minor salt addition wasachieved. This condition is certainly due to its lower molecularweight compared to the humic acid, implying shorter interparticlebridges, which require prior approach of the neighboring particlesvia salt-induced partial screening of repulsive electrostaticinteractions.

4. Conclusion

Findings from this work provide an avenue by which to moreaccurately assess exposure routes for TiO2 nanocomposites andtheir degradation byproducts that are commonly used insunscreens. The TiO2 nanocomposite undergoes rapid alterationand desorption of its hydrophobic coating layer in aqueous media.This subsequently promotes the dispersion of these materials inaqueous environments. Of the byproducts formed, about 25 wt%get dispersed as stable colloids and are therefore bioavailable to themicroorganisms and filter-feeders. Depending on solution pH(PIE¼ 7e8), ionic strength (anion concentration> 5 � 10�3 M) andNOM concentration (w2 wt%) the colloids tend to aggregate andsettle out of the water column. A similar fate also characterizes theremaining 75 wt% of the byproducts, which are more readilysettleable due to their larger initial size. However, even whenextracted from the water column, the byproducts should not beclassified as bio-unavailable. There they are indeed incorporatedinto geogenic sediments, and may come into contact with benthicfauna where they may be internalized by the grazer organisms.

From a photocatalytic point of view, the preservation of the Al(OH)3 mineral layer at the nanocomposite surface suggests that thebyproducts remain inert in the aqueous conditions used in thiswork. This finding is in agreement with the negative test forsuperoxide generation obtained by Auffan et al. (Auffan et al., inpress). However, the concern remains when considering theacidic media encountered in the digestive systems of the exposedorganisms, which will favor Al(OH)3 dissolution. This raises theneed for ecotoxicological studies on these organisms exposed notonly to the original nanocomposite, as already achieved (Wienchet al., 2009), but also to its byproducts that are formed fromaging under relevant conditions.

Ultimately, global risk assessment data for this type of nano-composite should be put into perspective with regards to the scaleof relative hazards, at the top of which stands skin cancer due tosunlight exposure in absence of sunscreen use. At a lower level,further consideration about the relevance of using nanoparticles insunscreens should weigh the efficiency and safety of TiO2 and ZnOas UV blockers against alternative organic substituants. Avoidingmineral UV blocker in sunscreens would implicitly requiresunscreen ingredients that are less efficient, less stable, offermarginal UVA protection, or that absorb through the skin (EWG,2009), and which may have stronger negative impact on theenvironment (Danovaro et al., 2008).

Acknowledgements

This research was funded by the French National ProgramsNANOALTER (INSU/EC2CO/CYTRIX) and AGING NANO & TROPH(ANR-08-CESA-001). The authors gratefully thank Dr. Jean-PaulAmbrosi for his help in alkaline dissolution procedure, and Dr.Jonathan Brant for the language correction.

References

Adachi, Y., Wada, T., 2000. Initial stage dynamics of bridging flocculation of poly-styrene latex spheres with polyethylene oxide. Journal of Colloid and InterfaceScience 229, 148e154.

Auffan, M., Pedeutour, M., Rose, J., Masion, A., Ziarelli, F., Borschneck, D., Chaneac, C.,Botta, C., Chaurand, P., Labille, J., Bottero, J.Y., Surface structural degradation ofTiO2-based nanomaterial used in cosmetics. Environmental Science & Tech-nology, in press.

Bian, W., Chandrasekaran, R., Rinaudo, M., 2002. Molecular structure of the rham-san-like exocellular polysaccharide RMDP17 from Sphingomonas paucimobilis.Carbohydrate Research 337, 45e56.

Börm, P., Robbins, D., Haubold, S., Kuhlbusch, T., Fissan, H., Donaldson, K., 2006. Thepotential risks of nanomaterials: a review carried out for ECETOC. Part FIbreToxicol 3, 11.

Chappell, M.A., George, A.J., Dontsova, K.M., Porter, B.E., Price, C.L., Zhou, P.H.,Morikawa, E., Kennedy, A.J., Steevens, J.A., 2009. Surfactive stabilization ofmulti-walled carbon nanotube dispersions with dissolved humic substances.Environmental Pollution 157, 1081e1087.

Danovaro, R., Bongiorni, L., Corinaldesi, C., Giovannelli, D., Damiani, E., Astolfi, P.,Greci, L., Pusceddu, A., 2008. Sunscreens cause coral bleaching by promotingviral infections. Environmental Health Perspectives 116, 441e447.

E.C, 2004. Nanotechnologies, a Preliminary Risk Analysis on the Basis of a Work-shop. European Commission: Health and Consumer Protection DirectorateGeneral, Brussels, pp. 11e29.

EPA, Morris, J., Willis, J., 2005. Nanotechnology white paper. United States Envi-ronmental Protection Agency.

EWG, 2009. Sunscreen Investigation. Section 4 Nanotechnology and Sunscreens.Environmental Working Group. http://www.ewg.org/cosmetics/report/sunscreen09/investigation/Nanotechnology-Sunscreens.

Fleer, G.J., Scheutjens, J.M.H.M., 1986. Interaction between adsorbed layers ofmacromolecules. Journal of Colloid and Interface Science 111, 504e515.

Fortner, J.D., Lyon, D.Y., Sayes, C.M., Boyd, A.M., Falkner, J.C., Hotze, E.M.,Alemany, L.B., Tao, Y.J., Guo, W., Ausman, K.D., Colvin, V.L., Hughes, J.B., 2005. C-60 in water: nanocrystal formation and microbial response. EnvironmentalScience & Technology 39, 4307e4316.

Graubner, V.M., Jordan, R., Nuyken, O., Schnyder, B., Lippert, T., Kotz, R., Wokaun, A.,2004. Photochemical modification of cross-linked poly(dimethylsiloxane) byirradiation at 172 nm. Macromolecules 37, 5936e5943.

Huang, Z., Zheng, X., Yan, D., Yin, G., Liao, X., Kang, Y., Yao, Y., Huang, D., Hao, B., 2008.Toxicological effect of ZnO nanoparticles based on bacteria. Langmuir 24, 4140.

Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic matter stabi-lizes carbon nanotubes in the aqueous phase. Environmental Science & Tech-nology 41, 179e184.

ICON, 2008. Towards Predicting Nano-Biointeractions: An International Assessmentof Nanotechnology Environment, Health and Safety Research Needs. In: ICONreports (Ed.). Number 4, May 1, 2008, International Council on Nanotechnology<http://icon.rice.edu/>.

Jones, N., Ray, B., Ranjit, K., Manna, A., 2008. Antibacterial activity of ZnO nano-particle suspensions on a broad spectrum of microorganisms. FEMS Microbi-ology Letters 279, 71e76.

Labille, J., Thomas, F., Bihannic, I., Santaella, C., 2003. Destabilization of montmo-rillonite suspensions by Ca2þ and succinoglycan. Clay Minerals 38, 173e185.

Labille, J., Thomas, F., Milas, M., Vanhaverbaeke, C., 2005. Flocculation of colloidalclay by bacterial polysaccharides. Effect of macromolecule charge and structure.Journal of Colloid and Interface Science 284, 149e156.

Lafuma, F., Wong, K., Cabane, B., 1991. Bridging of colloidal particles throughadsorbed polymers. Journal of Colloid and Interface Science 143, 9e21.

LaMer, V.K., 1964. Coagulation symposium introduction. Journal of Colloid Science19, 291e293.

Mills, A., Le Hunte, S., 1997. An overview of semiconductor photocatalysis. Journal ofPhotochemistry and Photobiology A 108, 1e35.

Minor, M., vanderLinde, A.J., vanLeeuwen, H.P., Lyklema, J., 1997. Dynamic aspects ofelectrophoresis and electroosmosis: a new fast method for measuring particlemobilities. Journal of Colloid and Interface Science 189, 370e375.

NanotechProject, 2009. The Project on Emerging Nanotechnology. Partnershipbetween the Woodrow Wilson International Center for Scholars and the PewCharitable Trusts. www.nanotechproject.org.

Nohynek, G., Lademann, J., Ribaud, C., Roberts, M., 2007. Grey goo on the skin?Nanotechnology, cosmetic and sunscreen safety. Critical Reviews in Toxicology37, 251e277.

Pan, Z., Lee, W., Slutsky, L., Clark, R.A.F., Pernodet, N., Rafailovich, M.H., 2009.Adverse effects of titanium dioxide nanoparticles on human dermal fibroblastsand how to protect cells. Small 5, 511e520.

Reddy, K.M., Feris, K., Bell, J., Wingett, D.G., Hanley, C., Punnoose, A., 2007. Selectivetoxicity of zinc oxide nanoparticles to prokaryotic and eukaryotic systems.Applied Physics Letters 90.

SCCNFP, 2000. Opinion concerning titanium dioxide. In: E.C. (Ed.), ScientificCommittee on Cosmetic Products and Non-Food Products Intended forConsumers. European Commision.

SCENIHR, 2005. Opinion on the appropriateness of existing methodologies to assesthe potential risks associated with engineered and adventitious products ofnanotechnologies. In: E.C. (Ed.), Scientific Committee on Emerging and NewlyIdentified Health Risks. European Commission.

J. Labille et al. / Environmental Pollution 158 (2010) 3482e3489 3489

Senjen, R., 2009. Nanomaterialse health and environmental concerns. In: Hontelez, J.(Ed.), Nanotechnologies in the 21st Century. European Environmental BureauNanotechnology Policy Officer. http://eeb.org/publication/general.html.

Swanson, H.E., Mcmurdie, H.F., Morris, M.C., Evans, E.H., 1969. Standard X-raydiffraction patterns. National Bureau of Standards Monograpg 25 Section 7, 82.

Thomas, F., Prelot, B., Villieras, F., Cases, J.M., 2002. Electrochemical properties ofsolids at the aqueous-solid interface and heterogeneity of surface. ComptesRendus Geoscience 334, 633e648.

Tombacz, E., Szekeres, M., 2004. Colloidal behavior of aqueous montmorillonitesuspensions: the specific role of pH in the presence of indifferent electrolytes.Applied Clay Science 27, 75e94.

Tsuang, Y., Sun, J., Huang, Y., Lu, C., Chang, W., Wang, C., 2008. Studies of photokillingof bacteria using titanium dioxide nanoparticles. Artificial Organs 32, 167e174.

Villain-Simonnet, A., Milas, M., Rinaudo, M., 1999. Comparison between the phys-icochemical behaviour of two microbial polysaccharides: RMDP17 and rham-san. International Journal of Biological Macromolecules 26, 55e62.

Wakefield, G., Lipscomb, S., Holland, E., Knowland, J., 2004. The effects of manganesedoping on UVA absorption and free radical generation of micronised titaniumdioxide and its consequences for the photostability of UVA absorbing organicsunscreen components. Photochemistry and Photobiological Sciences 3, 648e652.

Wiench, K., Wohlleben, W., Hisgen, V., Radke, K., Salinas, E., Zok, S., Landsiedel, R.,2009. Acute and chronic effects of nano- and non-nano-scale TiO2 and ZnOparticles on mobility and reproduction of the freshwater invertebrate Daphniamagna. Chemosphere 76, 1356e1365.

Wijnhoven, S.W.P., Dekkers, S., Hagens, W.I., de Jong, W.H., 2009. Exposure tonanomaterials in consumer products. Letter Report. In: RIVM (Ed.). NationalInstitute for Public Health and the Environment.

Wu, H.X., Wang, T.J., Jin, Y., 2006. Film-coating process of hydrated alumina on TiO2particles. Industrial & Engineering Chemistry Research 45, 1337e1342.

Xie, B., Xu, Z.H., Guo, W.H., Li, Q.L., 2008. Impact of natural organic matter on thephysicochemical properties of aqueous C-60 nanoparticles. EnvironmentalScience & Technology 42, 2853e2859.

Yang, K., Lin, D.H., Xing, B.S., 2009. Interactions of humic acid with nanosizedinorganic oxides. Langmuir 25, 3571e3576.

Zhang, J.W., Buffle, J., 1996. Multi-method determination of the fractal dimension ofhematite aggregates. Colloids and Surfaces A-Physicochemical and EngineeringAspects 107, 175e187.

Zhang, Y., Chen, Y., Westerhoff, P., Crittenden, J., 2009. Impact of natural organicmatter and divalent cations on the stability of aqueous nanoparticles. WaterResearch 43, 4249e4257.