Adsorption of Alkyl Polyglucosides on the Solid/WaterInterface: Equilibrium Effects of Alkyl Chain Length and

Head Group Polymerization

Maria K. Matsson,*,†,‡ Bengt Kronberg,† and Per M. Claesson†,‡

YKI, Institute for Surface Chemistry, Box 5607, SE-114 86 Stockholm, Sweden, andDepartment of Chemistry, Surface Chemistry, Royal Institute of Technology,

Drottning Kristinas vag 51, SE-100 44 Stockholm, Sweden

Received October 21, 2003. In Final Form: March 1, 2004

The equilibrium adsorption behavior of two n-alkyl-â-D-glucosides (octyl (C8G1) and decyl (C10G1)) andfour n-alkyl-â-D-maltosides (octyl (C8G2), decyl (C10G2), dodecyl (C12G2), and tetradecyl (C14G2)) from aqueoussolution on a titania surface, as measured by ellipsometry, has been investigated. The main focus has beenon the effect of changes in the alkyl chain length and headgroup polymerization, but a comparison withtheir adsorption on the silica/water and air/water interfaces is also presented. Some comparison with thecorresponding adsorption of ethylene oxide surfactants, in particular C10E6 and C12E6, is given as well.For all alkyl polyglucosides, the maximum adsorbed amount on titania is reached slightly below the criticalmicelle concentration (cmc), where it levels off to a plateau and the amount adsorbed corresponds roughlyto a bilayer. However, there is no evidence that this is the actual conformation of the surfactant assemblieson the surface, but the surfactants could also be arranged in a micellar network. On hydrophilic silica,the adsorbed amount is a magnitude lower than on titania, corresponding roughly to a layer of surfactantslying flat on the surface. A change in the alkyl chain length does not result in any change in the plateaumolar adsorbed amount at equilibrium; however, the isotherm slope for the alkyl maltosides increases withincreasing chain length. Headgroup polymerization on the other hand affects the adsorbed amount. Thealkyl glucosides start adsorbing at lower bulk concentrations than the maltosides and equilibrate at higheradsorbed amounts above the cmc. When compared with the ethylene oxide (EO) surfactants, it is confirmedthat the EO surfactants hardly adsorb on titania, since the measured changes in the ellipsometric anglesare within the noise level. They do, however, adsorb strongly on silica.

IntroductionAlkyl polyglucosides, CnGm, are a class of surfactants

that has become of increasing interest during the pastfew years, for which environmental concern is one reason.The alkyl polyglucosides can be synthesized from renew-able raw materials and are biodegradable and nontoxic.1Furthermore, they show good detergency properties, whilebeing mild to the skin,2 and are currently used in somepersonal care products. The surfactants consist of anumber of glucose units linked to an alkyl chain via anether bond. This study focuses on alkyl glucosides andmaltosides; i.e., the headgroup is built from one or twoglucose units.

Knowledge of the adsorption behavior of a surfactantis crucial for understanding the underlying mechanismsfor applications such as wetting, detergency, and lubrica-tion. The aim of this work was to determine and rationalizetrends in the adsorption behavior of alkyl polyglucosidesdue to structural changes, i.e., head and tail group sizes,and to correlate this to similar relationships for the moreknownethyleneoxide (EO)surfactants.Another importantissue is to explain the differences in adsorption of alkylmaltosides on the metal oxide/water and silica/waterinterfaces respectively, as well as to elucidate the differ-ences in adsorption behavior between alkyl polyglucosidesand ethylene oxide surfactants on these surfaces.

The bulk behavior of alkyl polyglucosides has beeninvestigated in some detail,3-7 as has the adsorptionbehavior at the air/liquid interface.8-14 However, studieson the adsorption process at the solid/liquid interface arestill limited. Investigations of the dispersion of titaniapigments in aqueous solutions using technical gradesurfactants show that alkyl polyglucosides adsorb ontitania and that the adsorbed amount exceeds that of amonolayer.15 It was suggested that the adsorption couldbe explained by hydrogen bonding of the slightly acidichydroxyl groups on the surfactant with the basic OHgroups on the oxide surface. On alumina, this has beenfurther investigated, and again the authors concluded thathydrogen bonding is the main driving force for adsorp-

* To whom correspondence should be addressed. E-mail:maria.matsson@surfchem.kth.se.

† YKI, Institute for Surface Chemistry.‡ Royal Institute of Technology.(1) Garcıa, M. T.; Ribosa, E.; Campos, J.; Sanchez Leal, J.Chemisphere

1997, 35, 545.(2) Schmid, K.; Tesmann, H. In Detergency of Specialty Surfactants;

Friedli, F. E., Ed.; Dekker: New York, 2001; Vol. 98, p 1.

(3) Boyd, B. J.; Drummond, C. J.; Krodkiewska, I.; Grieser, F.Langmuir 2000, 16, 739.

(4) Dorfler, H.-D.; Gopfert, A. J. Dispersion Sci. Technol. 1999, 20,35.

(5) Nilsson, F.; Soderman, O.; Hansson, P.; Johansson, I. Langmuir1998, 14, 4050.

(6) Nilsson, F.; Soderman, O.; Johansson, I. Langmuir 1996, 12, 902.(7) Shinoda, K.; Carlsson, A.; Lindman, B. Adv. Colloid Interface

Sci. 1996, 64, 253.(8) Bocker, T.; Thiem, J. Tenside, Surfactants, Deterg. 1989, 26, 318.(9) Drummond, C. J.; Warr, G. G.; Grieser, F.; Ninham, B. W.; Evans,

D. F. J. Phys. Chem. 1985, 89, 2103.(10) Kahl, H.; Enders, S.; Quitzsch, K. Colloids Surf., A 2001, 183-

185, 661.(11) Kjellin, U. R. M.; Claesson, P. M.; Vulfson, E. N. Langmuir 2001,

17, 1941.(12) Rosen, M., J.; Sulthana, S., B. J. Colloid Interface Sci. 2001,

239, 528.(13) Aveyard, R.; Binks, B. P.; Chen, J.; Esquena, J.; Fletcher, P. D.

I. Langmuir 1998, 14, 4699.(14) Shinoda, K.; Yamaguchi, T.; Hori, R. Bull. Chem. Soc. Jpn. 1961,

34, 337.(15) Smith, G. A.; Zulli, A. L.; Grieser, M. D.; Counts, M. C. Colloids

Surf., A 1994, 88, 67.

4051Langmuir 2004, 20, 4051-4058

10.1021/la035959p CCC: $27.50 © 2004 American Chemical SocietyPublished on Web 04/14/2004

tion.16 Comparisons between adsorption on different solidsurfaces show that the adsorption is high on metal oxideswhile low on silica surfaces.17,18 Ethylene oxide basedsurfactants on the other hand adsorb strongly on silica,but scarcely on metal oxides. This difference is especiallyinteresting since this latter adsorption process also hasbeen explained by invoking hydrogen bonding.19 We notethat the standard explanation put forward to explain theadsorption of nonionic surfactants on hydrophilic surfacesfrom aqueous solutions is hydrogen bonding. The maindifficulty with this hypothesis is that water is able to formvery strong hydrogen bonds with both the surface and thesurfactant headgroup and it is hard to see how replace-mentsof suchhydrogenbondswithsurfactantheadgroup-surface hydrogen bonds will result in an energy gain.Nevertheless, the enthalpy of adsorption for poly(ethyleneoxide) on silica is negative.20 We conclude that the drivingforce for adsorption is not well-understood; however, onemay perceive a situation where water hydrogen bondedboth to the surface and the surfactant headgroup mediatesthe surfactant-surface interactions. On hydrocarbon andat air/water interfaces, both surfactants adsorb to anapproximate monolayer.21-23 So far, the techniques usedto study the adsorption behavior of alkyl polyglucosideshave been solution/depletion and calorimetry, but ellip-sometry, the main technique used here, has not previouslybeen employed.

Experimental Section

Materials. A series of monodisperse alkyl â-D-polyglucosides,CnGm (C8G1, C8G2, C10G1, C10G2, C12G2, C14G2) g 99% pure,purchased from Anatrace, were used without further purification.For comparative measurements two ethylene oxide surfactants,C10EO6 and C12EO6 g 99% pure, purchased from Nikko Chemi-cals, were used. Polished silicon wafers, thermally oxidized toproduce a SiO2 layer thickness of approximately 300 Å and thencut into slides with a width of 12.5 mm, were provided by Dr.Stefan Klintstrom, University of Linkoping, Linkoping, Sweden.Titanium slides, prepared by evaporating an approximately 1500Å thick layer titanium on polished silicon wafers, which thenwere cut into slides with a width of 8 mm, were provided by BoThuner, University of Linkoping. The silica-covered silicon waferswere cleaned for 10 min at 80 °C in a mixture of NH4OH (29%),H2O2 (30%), and H2O at a 1:1:5 ratio, followed by a cleaning for10 min at 80 °C in a mixture of HCl (37%), H2O2 (30%), and H2Oat a 1:1:6 ratio. They were then stored in ethanol until used. Justprior to the measurements the wafers were plasma-cleaned inlow-pressure air for 5 min using a radio frequency glow dischargeapparatus (Harrick PDC-3XG, Harrick Scientific Corp., Ossining,NY). The titanium-covered silica wafers were flushed in ethanoland dried with nitrogen to remove particles and then cleanedusing only the plasma cleaner in the same way as the silicaslides. All glassware used in the experiments, for storing andpreparation of surfactant solutions, were cleaned using surfac-tant-free Deconex and then thoroughly rinsed consecutively withwater, ethanol, and water. The water used in all experimentswas treated by a Milli-Q Plus unit (Millipore, Bedford, MA)including ion exchange, active carbon adsorption, and reverseosmosis before the final 0.22 mm filtration step, yielding ultrapurereagent-grade water of resistivity 18.2 MΩ cm (at 25 °C).

Methods. The main experimental method used to investigatethe adsorption properties of the surfactants was null ellipsometry.The instrument used is a Multiskop (Optrel GdBR, Berlin,Germany), with horizontally aligned measurement arms withan angle of incidence of 68°, equipped with a 532 nm laser andset up according to the PCSA arrangement. The instrument hasbeen extensively described elsewhere.24 In principle the ellip-someter measures the ellipticity of light before and after reflectionat a substrate. From these measurements, the ellipsometricangles Ψ and ∆, which reflect the difference in polarizationbetween the incident and the reflected light, are obtained. Ψ isrelated to the change in the amplitude upon reflection, and ∆describes the phase shift. The ellipsometric angles are relatedto the optical parameters of the substrate as

where λ is the wavelength of the light andφ0 the angle of incidence.The optical parameters are the refractive indices of the substrateand adsorbent and their respective layer thickness as well as therefractive index of the ambient media.25 By studying the baresubstrate in two ambient media, air and water, both the complexrefractive index of the bulk material and the thickness and therefractive index of the upper oxide layer can be calculated.26

From the change in polarization upon reflection during adsorp-tion, the thickness and the refractive index of the surfactantlayer are calculated, and from those the adsorbed amount, usingthe de Feijter’s formula:

where n0 is the refractive index of the surrounding media, n andd are the refractive index and the thickness of the adsorbed layer,and dn/dc is the refractive index increment of the surfactantsolution. In the calculations we have invoked a model of theadsorbed layer as being a homogeneous planar film characterizedby its thickness and refractive index. We note that the errorsobtained for n and d during evaluation are significantly largerthan the errors in the adsorbed amount. This is due to the errorsin n and d being coupled, and they largely cancel out when theadsorbed amount is calculated.27 During the measurements thesubstrate was placed vertically in a 5 mL quartz cuvette equippedwith tubing for inlet and outlet liquid and stirred at a constantrate of 300 rpm. The cuvette was temperature-controlled to within(0.1 °C. The adsorption process was studied as the bulk con-centration was increased stepwise, either by adding surfactantfrom a concentrated stock solution or by pumping a solution atthe desired concentration through the tubing. All measurementshave been performed at pH 5.8 and 25.0 °C unless otherwisestated.

The refractive index increment, dn/dc has been determinedfor each surfactant using an interferometric refractometer(Optilab DSP, Wyatt Technology Corp., Santa Barbara, CA) Theinstrument is based on a wave front shearing technique andmeasures the difference in refractive index between a storedreference liquid and a liquid sample stream. The refractive indexis measured at a fixed wavelength of 450 nm, and the instrumentis equipped with a temperature-control system, here set to 25°C. The difference in refractive index was measured for fivedifferent concentrations of surfactant, with water as the referencesolution.

ESCA28 was performed to determine the chemical compositionof the titanium slide surface. This is partly to verify that theupper surface layer is titania and partly to show that the chosencleaning procedure leaves a clean surface as result, withoutdamaging the original surface. With ESCA, the upper 5 nm of

(16) Zhang, L.; Somasundaran, P.; Mielczarski, J.; Mielczarski, E. J.Colloid Interface Sci. 2002, 256, 16.

(17) Kiraly, Z.; Borner, R. H. K.; Findenegg, G. H. Langmuir 1997,13, 3308.

(18) Zhang, L.; Somasundaran, P.; Maltesh, C. J. Colloid InterfaceSci. 1997, 191, 202.

(19) Partyka, S.; Zaini, S.; Lindeheimer, M.; Brun, B. Colloids Surf.1984, 12, 255.

(20) Trens, P.; Denoyel, R. Langmuir 1993, 9, 519.(21) Somasundaran, P.; Snell, E. D.; Xu, Q. J. Colloid Interface Sci.

1991, 144, 165.(22) Tiberg, F.; Landgren, M. Langmuir 1993, 9, 927.(23) Kiraly, Z.; Findenegg, G. H. Langmuir 2000, 16, 8842.

(24) Harke, M.; Teppner, R.; Schulz, O.; Motschmann, H. Rev. Sci.Instrum. 1997, 68, 3130.

(25) Ellipsometry and polarized light; Azzam, R. M. A., Bashara, N.M., Eds.; North-Holland: Amsterdam, 1977.

(26) Landgren, M.; Jonsson, B. J. Phys. Chem. 1993, 97, 1656.(27) de Feijter, J. A.; Benjamins, J.; Veer, F. A. Biopolymers 1978,

17, 1759.(28) Practical surface analysis: By Auger and X-ray photoelectron

spectroscopy; Briggs, D., Shea, M. P., Eds.; Wiley: Chicester, U.K., 1983.

tan(Ψ)ei∆ ) f(λ,φ0,optical params) (1)

Γ )(n - n0)dn/dc

d (2)

4052 Langmuir, Vol. 20, No. 10, 2004 Matsson et al.

the surface is studied in great detail, investigating the relativeamount of different elements and their chemical environmenton the surface. First a wide spectrum was run to detect elementspresent in the surface layer. This was followed by detailed spectrafor all elements detected in the wide spectrum, from which therelative surface compositions were obtained. The spectra wererecorded using a Kratos AXIS HS X-ray photoelectron spec-trometer (Kratos Analytical, Manchester, U.K.). The sampleswere analyzed using a monochromator (Al X-ray source). Theanalysis area was below about 1 mm2.

AFM29,30 can be used for imaging, chemical mapping and forcespectroscopy. Here, the surface was investigated using thetapping-mode technique, giving a visual description of the sur-face roughness and structure. With AFM, details in the nano-meter range are easily detectable. Some measures of the rough-ness can also be calculated; here, the surface roughness isdescribed by the Ra- and Rq-values and the surface area difference.The Ra-value is the average surface roughness, and the Rq-valueis the root-mean-square average of the deviation from a planefit to the surface; they are calculated according to the followingequations:

where z(x) is the vertical deviation from a plane fit of the surfacein the point x and L is the length of the distance measured. Thesurface area difference is the difference between the real areaof a rough surface, Areal, which the surfactant recognizes, andthe measured projected area, Aproj. The instrument used was amultimode nanoscope IIIa AFM (Digital Instruments, SantaBarbara, CA). The projected surface area imaged was 1 and 0.25µm2, and each image was scanned line by line with 512 scanlines. During the measurement, errors causing height changescan occur. These errors have, before calculating the roughnessparameter, been smoothed out using the flatten function of theAFM.

To determine macroscopic hydrophobicity of the substrate,the contact angle of pure water on the surface was measuredusing a contact-angle goniometer, A 1100 (Rame Hart, Inc., NJ),after plasma cleaning, after adsorption from a solution at aconcentration above the critical micelle concentration (cmc)followed by drying with N2, and after rinsing with pure waterfollowed by drying with N2.

Results

Surface Characterization. From ESCA measure-ments it can be concluded that the upper 5 nm of thetitanium surface consists mainly of Ti and O, althoughsome carbon is still left at the surface, Figure 1. The bondenergy spectrum for Ti shows peaks at values corre-sponding to pure Ti and TiO2, but no peaks correspondingto other titanium oxides. The ratio nTi/nO for the layer isapproximately 0.45. The carbon content in the uppersurface layer has been significantly lowered after treat-ment with the plasma cleaner, compared with a slide thatonly was flushed with ethanol and dried with nitrogen.The titanium content is about the same both before andafter plasma treatment.

The surface characterization gives Ra ) 4.349 nm andRq ) 3.463 nm for the titania surfaces when measuredover 1 mm2, giving Ra/Rq ) 1.26, which correspondsapproximately to a Gaussian distribution of the topologicalheight features. Calculation of the real surface area using

an internal AFM routine gives a surface area ratio,(Areal - Aproj)/Aproj ) 0.092. Figure 2 shows an AFM imageof the titania surface, where it also can be observed thatthe upper layer of the substrate seems to have a granularstructure.

After cleaning with plasma, water completely wets boththe silica and the titania slides (θ ) 0); i.e., the surfaceis clearly hydrophilic.

(29) Binnig, G.; Quate, C. F.; Gerber, C. Phys. Rev. Lett. 1986, 56,930.

(30) Atomic force microscopy/scanning tunneling microscopy; Cohen,S. H., Bray, M. T., Lightbody, M. L., Eds.; Plenum Press: New York,1994.

Figure 1. Overview ESCA spectrum for the Ti substrate,showing peaks from C at 280-290 eV, Ti at 450-470 eV, andO around 530 eV. The insert shows a detail 2p spectrum for theTi peak, where the peaks at 459 and 465 eV correspond totitanium in TiO2, while the small peak at 454 eV correspondsto titanium in Ti metal form.

Figure 2. Image of the titanium surface topography, as seenby AFM including the calculated Ra- and Rq-values. The surfacewas investigated using the tapping-mode technique.

Figure 3. Adsorption isotherm for C10G2, showing the adsorbedamount (open circles) and the corresponding headgroup areaper surfactant (filled circles), assuming that the surfactantsassemble in a bilayer. The solid line is drawn to guide the eye.

Ra ) 1L ∫0

L|z(x)| (3)

Rq ) x1L ∫0

Lz(x)2 dx (4)

Adsorption of Alkyl Polyglucoside Langmuir, Vol. 20, No. 10, 2004 4053

Adsorption at the Solid/Liquid Interface. Theellipsometer has been used to measure equilibrium dataat different concentrations to obtain adsorption isothermsfor the surfactants. The measured refractive index incre-ments, dn/dc, necessary to calculate the adsorbed amountsfrom the optical characteristics of the surfactant filmobtained from the ellipsometer, are listed in Table 1,together with a summary of the plateau adsorbed amountsfor all studied surfactants and other properties of thesurfactants.

Starting with the maltosides, Figure 3 shows theadsorption isotherm for C10G2, both the adsorbed amountand the calculated headgroup area per surfactant. Fromthis it is seen that the adsorption, as registered by theellipsometer, commences at concentrations around halfthe cmc and that maximum adsorption is reached atconcentrations slightly below the cmc of the surfactant.Above cmc no additional adsorption takes place, but theadsorption levels out at a plateau value of about 7.6 µmol/m2, which corresponds to an apparent headgroup area ofapproximately 0.22 nm2. This is clearly a too-small valuefor being a monolayer, which indicates that at theadsorption plateau bilayer structures are formed. Theaverage area per molecule in the bilayer is thus 0.44 nm2

in each layer.The glucoside with a C10 hydrocarbon chain, C10G1,

shows different adsorption behavior compared to C10G2.At 25 °C, the reproducibility between measurements islow, as can be seen from Figure 4a. The isotherm levelsout to a stable plateau for all individual measurement,but at values varying between 9.1 and 13.1 µmol/m2.However, when increasing the temperature to 30 °C, thiswide distribution disappears and repeated measurementsgive an average plateau value of 8.8 µmol/m2, shown inFigure 4b. Combining the isotherms of C10G1 and C10G2,Figure 5, it is easily observed that the adsorption beginsat a lower concentration relative to the cmc for the

glucoside and levels out at around the same concentration,slightly below the cmc, but at a higher adsorbed amount.The plateau value corresponds to an apparent headgrouparea per surfactant of 0.18 nm2, i.e., 0.36 nm2 in eachlayer.

To investigate the effect of the alkyl chain length onadsorption, measurements with four alkyl maltosides andtwo alkyl glucosides with different hydrophobic chainlengths were carried out. As illustrated in Figure 6a,b,the adsorption starts around the same concentrationrelative to the cmc and has approximately the sameequilibrium values at different concentrations for all fourmaltosides. The slope of the ascending section of theisotherm increases slightly with increasing chain length.The isotherms for the two glucosides also show similarities,and they also level out at the same concentration, as canbe seen in Table 1, and by comparing Figure 4b and Figure

Table 1. Presentation of Literature Values and Measured Results for the Different Surfactantsa

surfactantM

(g mol-1)cmc

(mM)dn/dc

(mL g-1)Γcmc* a

(mg m-2)Γcmc*/** a

(mmol m-2)A/molecule*/** a

(nm2)A/molecule at air/liquid

(nm2)

C8G1 292.4 22b 0.146 2.6 8.7/7.9 0.38/0.42 0.38b

C8G2 454.4 19c 0.146 3.4 7.5/6.8 0.44/0.49 0.42c

C10G1 320.4 2.0d 0.153 2.8 8.8/8.0 0.38/0.42 0.40d

C10G2 482.6 2.0d 0.146 3.6 7.6/6.9 0.44/0.48 0.49d

C12G2 510.6 0.15e 0.142 3.8 7.4/6.7 0.45/0.50 0.50e

C14G2 538.6 0.015f 0.142 4.1 7.6/6.9 0.44/0.48a Adsorbed amount and area per molecule, assuming bilayer structure, are presented both for an ideally smooth surface (*) and corrected

with surface roughness (**). b Reference 11. c Reference 3. d Reference 12. e Reference 9. f Reference 8.

Figure 4. Adsorption isotherms for C10G1 at (a, left) 25 and (b, right) 30 °C. Each individual measurement series reaches a plateauvalue around cmc, but the scatter around the average is almost a magnitude larger for the lower temperature. The averages forthe two temperatures are also different; while the adsorption at 30 °C corresponds to that expected for a double layer, the adsorptionat 25 °C significantly exceeds it. The solid lines represent average values, drawn to guide the eye.

Figure 5. Adsorption isotherms for C10G1 at 30 °C (filled circles)and C10G2 at 25 °C (open circles), showing the effect of headgrouppolymerization. The glucoside starts adsorbing at lower sur-factant concentration. The solid lines are drawn to guide theeye.

4054 Langmuir, Vol. 20, No. 10, 2004 Matsson et al.

7. However, at concentrations slightly below the cmc, theC8G1 shows a maximum, presumably due to the presenceof a low amount of impurities with higher affinity to thesurface. This is confirmed by surface tension measure-ments, showing a slight dip in surface tension just beforereaching the cmc, Figure 8a, which is not observed for theother surfactants, Figure 8b. It can be observed, bycomparing Figures 8a and 7 that the minimum in thesurface tension curve is smaller than the maximum inthe adsorption isotherm curve; i.e., ellipsometry is moresensitive to impurities than surface tension measure-ments. The same observation has previously been madefor impure SDS solutions.31 We suggest that the sensitivitydifference is mainly due to ellipsometry data being directlyproportional to the adsorbed amount, while the reductionin surface tension is given by the integral of the adsorbedamount as a function of the chemical potential of thesurfactant. This sensitivity difference is also affected bythe difference in the surface/volume ratio between thetechniques, which for the surface tension measurementsis 0.6 m-1, while being 0.4 m-1 for the ellipsometrymeasurements.

To verify that the adsorbed amount is independent ofthe route of adsorption, additional measurements weremade. For instance, the maximum adsorbed amount hasbeen obtained by adding highly concentrated surfactantsolutions to the cuvette and compared with the above-

described measurements where the bulk concentrationwas increased stepwise. Further, measurements wherethe concentrations below the cmc have been obtained fromdilutionof thebulksolutionhavebeenconducted. InFigure9, a typical adsorption/desorption curve for one of thesurfactants is shown and the maximum adsorbed amountreached here is the same as that obtained when deter-mining the adsorption isotherm by the stepwise additionmethod. Hence, the adsorbed amount is independent ofthe path, as it should be for an equilibrium situation.

The adsorption of C10G2 on silica surfaces and of ethyleneoxide surfactants, C10EO6 and C12EO6 on titania was alsomeasured. No adsorption of C10EO6 and C12EO6 on titaniacould be detected. For C10G2 on silica, Figure 10, a smalladsorption, about an order of magnitude lower than ontitania, could be detected.

Discussion

Surface Characterization. Three main conclusionscan be drawn from the ESCA measurements. First, theonly titanium oxide present in the upper 5-10 nm is TiO2,supported both by the location of the peaks and the Ti/Oratio. Second, in the upper layer, titanium is also presentas pure Ti, indicating either that the TiO2 layer is lessthan 5 nm thick or that there is some Ti close to theimmediate surface. Since Ti oxidizes spontaneously, themost probable reason for the peak corresponding to pureTi is that the oxide layer is less than 5 nm. This is alsoin agreement with ellipsometry measurements on thesesurfaces, which yield an oxide thickness of 3.5-4.5 nm.Third, the cleaning process clearly removes carbon fromthe surfaces.

The slightly rough profile observed in the AFM mea-surements is still smooth enough to allow for successfulellipsometry measurements using the TiO2 substrates.However, the TiO2 surface is not as smooth as the oxidizedhydrophilic silica, which usually has an Ra-value of lessthan 0.5 nm. We note that the roughness is so high thatit requires us to consider that the real surface area issignificantly larger than the projected one. This has to betaken into account when evaluating the adsorbed amountper unit area.

Adsorptionat theSolid/LiquidInterface. Generallyspeaking, the isotherms on titania for all surfactants havea, for nonionic surfactants on hydrophilic surfaces, typicalsigmoidal shape, with a fairly steep slope between thecritical surface aggregation concentration (csac) and thecmc, which indicates cooperative adsorption. Since thereis not a step change in the adsorbed amount, the adsorption

(31) Arnebrant, T.; Backstrom, K.; Jonsson, B.; Nylander, T.J. ColloidInterface Sci. 1989, 128, 303.

Figure 6. Isotherms for four different alkyl maltosides C8G2 (filled diamonds), C10G2 (open circles), C12G2 (filled squares), and C14G2(open triangles) (a, left) as a function of the concentration and (b, right) as a function of the concentration scaled with cmc. In a,the arrows indicate the cmcs of the different maltosides. For C8G2 a linear extrapolation of the slope, whose intercept with the x-axisis used as a definition of csac, is also indicated. The solid lines are drawn to guide the eye.

Figure 7. Typical adsorption isotherm of C8G1 (filled dia-monds), showing the effect of small amounts of impurities inthe surfactant. The arrow indicates the cmc. The solid line isdrawn to guide the eye.

Adsorption of Alkyl Polyglucoside Langmuir, Vol. 20, No. 10, 2004 4055

process cannot be considered as a phase separation, butthe growth of the adsorbed layer is continuous. This meansthat the bilayer thickness increases, or that the surfacemicelle structure changes during the adsorption process.That no significant adsorption for the maltosides isregistered in the low adsorption region does not auto-matically give the conclusion that no adsorption occurs atthese concentrations, but it can also be attributed to thatthe relative error of the ellipsometry measurementsincreases with decreasing registered adsorbed amounts;e.g. a measured value of Γ ) 0.2 µmol/m2 would have arelative error of around 15%.32

For C10G1, the first measurements at 25 °C wereperformed as for the other surfactants, by adding a smallamount of concentrated solution to the cuvette. However,nonreproducible results were obtained. Phase studies haveshown that this particular surfactant has a two-phaseregion between 0.1 and 19 wt% of the surfactant in water(cmc ) 0.07 wt%),5 so the concentrated stock solution wouldthen be in the two-phase region. The first measure takenwas to change the addition process to pumping solutionthrough the cuvette at the desired concentrations. Thisdid nevertheless not change the large distribution betweenmeasurements. Another study of the surfactant bulkproperties shows that as the temperature decreases tobelow room temperature, C10G1 crystallizes in water.3Therefore adsorption measurements at higher tempera-tures were performed, assuming that if the reason for theunreliable resultsobtainedat25°Cwassurfactant crystalsadsorbing at the interface, the reproducibility would besignificantly better at 30 °C. This was, as seen above, alsothe case.

To verify that increasing the temperature would notgive other effects as well, the isotherm for C12G2 was alsomeasured at 30 °C. For this surfactant, no significantdifference in adsorbed amounts between the temperaturesis observed, as can be seen in Figure 11. In particular itshould be noted that, unlike C10G1, C14G2 gives reproduc-ible results, though it also is measured below its Krafftpoint. This is most likely due to the very slow rate ofcrystallization for this surfactant. Since the conversioninto crystals in solution was observed to take roughly 2

(32) Tiberg, F. J. Chem. Soc., Faraday Trans. 1996, 92, 531.

Figure 8. Surface tension measurements of (a, left) C8G1 and (b, right) C10G2. The solid lines are polynomial fits of the secondorder of ln(c) for the surface tension, and the dotted lines represent the equilibrium surface tension above the cmc.

Figure 9. Adsorption/desorption cycle for C14G2 obtained whenadding a concentrated stock solution to the cuvette to obtaina bulk concentration well above the cmc, equilibrating andfinally rinsing with water.

Figure 10. Comparison of the adsorbed amount of C10G2 ontitania (open circles) with that on silica (filled circles), showinga difference of around 1 order of magnitude in adsorption. Thesolid lines are drawn to guide the eye.

Figure 11. Adsorption isotherms for C12G2 at 25 (open squares)and 30 °C (filled diamonds), showing the temperature depen-dence for the maltosides. The solid line is drawn to guide theeye.

4056 Langmuir, Vol. 20, No. 10, 2004 Matsson et al.

days, reproducible results could be obtained from freshlyprepared solutions.

The observation that the difference between the maxi-mum adsorbed amount within the groups of maltosidesand glucosides respectively is minimal leads to theconclusion that the maximal adsorption of alkyl polyglu-cosides is independent of the alkyl chain length. Theheadgroup polymerization on the other hand seems toaffect the adsorption process and in particular the plateauadsorbed amount significantly, indicating that the maxi-mum adsorption is limited by the size of the headgroup.The alkyl chain length does nevertheless affect the slopeof the isotherms, and from the adsorption isotherms, somemeasure of cooperativity can be obtained. The free energydifference between a surfactant in a micellar-like ag-gregate at the surface and a surfactant in a free micellecan be estimated as

To estimate csac, the slope of the ascending section ofthe isotherm is extrapolated to a straight line; see theisotherm for C8G2 in Figure 6a. Then the csac is definedas the concentration at the intercept with the x-axis. Theestimated csacs and calculated ∆µ0 are listed in Table 2.Csac is 0.4-0.7 cmc, and the free energy difference is 0.35-0.93 kT; hence, there is a rather weak interaction betweenthe surfactant molecules and the surface. The magnitudeof ∆µ0, and thereby the extra driving force for self-assemblying at the surface rather than in bulk solution,is larger for the short-chain surfactants. A plausibleexplanation is that the hydrophobic interaction betweenthe tails increases with the surfactant chain length, andthus the extra contribution due to interactions betweenthe headgroup and the surface becomes less noticeable.These measures can be compared with the interactionbetween EO surfactants and silica surfaces, whichamounts to <0.5 kT, within the range of the values formaltosides and glucosides on titania reported here.22

When discussing the adsorbed amounts, the surfaceroughness of the substrate also has to be considered, asexplained above. When recalculating the adsorbed amountwith the real surface area, the headgroup areas obtained,Table 1, are around 0.50 nm2 for the maltosides and 0.42nm2 for the glucosides, if we assume bilayer structure.Thevalue for themaltosides is lower comparedwithresultsfrom a previous study on adsorption of C12G2 on soliddispersed oxide particles, including titania (rutile), wherethe surface area per molecule was found to be 0.30 nm2,i.e., 0.60 nm2, assuming bilayer conformation.18 Thisdifference in surface area might be explained by differencesin the titania surface. A micellar structure on the surfacerather than a complete bilayer formation is however alsopossible for both glucosides and maltosides. For C8G1, ahydrodynamic radius, rH ) 0.25 nm is obtained for anaggregation number n ) 82.33 In a close-packed hexagonalarray this gives an apparent area of 0.52 nm2/molecule.

For C12G2, the hydrodynamic radius, rH ) 0.27 nm isobtained for an aggregation number n ) 120.34 This givesa surface area per molecule of 0.42 nm2. At the air/waterinterface the measured headgroup area per surfactant isapproximately 0.45-0.50 nm2 for alkyl maltosides and0.36 nm2 for alkyl glucosides, as can be seen in Table 1.This area corresponds to a monolayer. When comparingthis with the headgroup area calculated from the adsorbedamount at the titania/water interface, it can be concludedthat the area per molecule assuming bilayer conformationis closer to the area at the air/liquid interface than thearea per molecule assuming micellar structure, thoughboth surface conformations are possible.

Finally, the differences between the adsorption ofethylene oxide surfactants and alkyl polyglucosides needto be addressed. For the ethylene oxide surfactants, theformation of bilayer aggregates on silica has been exten-sively studied.21,35,36 The fewer studies devoted to adsorp-tion at metal oxides, including titania surfaces, havealmost exclusively not observed any adsorption,37,38 exceptfor the studies made by Luciani and Denoyel,39 whereadsorption corresponding to almost a bilayer was observedat pH corresponding to the point of zero charge for thetitania surface. For alkyl polyglucosides the measuredadsorbed amounts at the silica/water interface are ap-proximately a magnitude lower than at the titania/waterinterface,17,18 inagreementwith theresultspresentedhere.

For the ethylene oxide surfactants on silica, the entireethylene oxide chain is suggested to bind to the silicasurface. The reason for this assumption is the observationthat there is a linear dependence between increasingethylene oxide chain and plateau adsorbed amount,corresponding to around 9 Å2/(ethylene oxide unit). Thealkyl chain is then directed vertically into the interior ofthe adsorbed layer, since increasing the alkyl chain lengthdoes not affect the plateau adsorbed amount much.21 Themajor part of the adsorption takes place between csac andcmc and is driven by hydrophobic interaction between thealkyl chains.

For the alkyl polyglucoside, the increase in area permolecule when changing from glucoside to maltoside istoo small to assume that both sugar groups adsorb entirelyto the substrate.

Conclusions

The adsorption of alkyl polyglucosides on titania isstrongly cooperative and reaches a plateau value aroundcmc. This value corresponds roughly to a bilayer for allsurfactants. The plateau adsorption of alkyl maltosidesand glucosides is practically independent of the hydro-phobic chain length, while increasing the headgrouppolymerization decreases the plateau adsorbed amounton titania. The slope of the isotherm increases withincreasing chain length, indicating increasing coopera-tivity.

The measured area per molecule of the surfactantssuggests that the surface conformation could be either inthe form of a bilayer or as micelles. Since the adsorbedamount does not increase stepwise at csac, the composition

(33) Esumi, K.; Arai, T.; Takasugi, K. Colloids Surf., A 1996, 111,231.

(34) Bucci, S.; Fagotti, C. Langmuir 1991, 7, 824.(35) Brinck, J.; Jonsson, B.; Tiberg, F. Langmuir 1998, 14, 5863.(36) Brinck, J.; Tiberg, F. Langmuir 1996, 12, 5042.(37) Fukushima, S.; Kumagai, S. J. Colloid Interface Sci. 1973, 42,

539.(38) Kumagai, S.; Fukushima, S. J. Colloid Interface Sci. 1976, 56,

227.(39) Luciani, L.; Denoyel, R. Langmuir 1997, 13, 7301.

Table 2. Measures of Cooperativity for the AlkylMaltosides (csac Estimated from Adsorption Isotherms)

surfactant cmc (mM) csac (mM) ∆µ0 (kT)

C8G2 19a 7.5 -0.93C10G2 2.0b 0.92 -0.78C12G2 0.15c 0.076 -0.67C14G2 0.015d 0.011 -0.35

a The cmc values are from reference 3. b Reference 12. c Refer-ence 9. d Reference 8.

µ0ads - µ0

mic ) kT ln(csaccmc) (5)

Adsorption of Alkyl Polyglucoside Langmuir, Vol. 20, No. 10, 2004 4057

of the adsorbed layer will change during the adsorptionprocess, meaning for micellar conformation that thesurface micelle structure is altered.

Since alkyl polyglucosides of the same chain length showthe same molar plateau adsorbed amount and surfacearea, the alkyl chains are most likely directed close tovertically into the interior of the adsorbed layer. Thesurface area is increased for maltosides compared withglucosides, suggesting that the headgroup adsorption isat least partly parallel to the surface, though the areaincrease is not large enough to allow for complete surfacecontact of the entire maltoside headgroup.

The phase behavior in bulk is crucial for understandingthe adsorption properties at the solid/liquid interface, ascan be seen from the adsorption measurements of C10G1,where the reproducibility is low around room temperature,while significantly better around 30 °C. Measurements ofC14G2 were however also performed below the Kraffttemperature. For this surfactant, equilibrium in bulk is

reached very slowly (days), thereby allowing for repro-ducible and reliable measurements.

Alkyl polyglucosides hardly adsorb on silica, whileethylene oxide surfactants adsorb strongly on this oxidesurface. In contrast, ethylene oxide surfactants show veryweak adsorption on titania. The difference has yet to beexplained; however, not only surfactant properties butalso surface properties need to be investigated anddocumented to enable this. The different affinity to silicaand metal oxide surfaces for the alkyl polyglucosides andthe ethylene oxide surfactants gives interesting possibili-ties for uses in e.g. separation applications.

Acknowledgment. M.K.M. gratefully acknowledgesthe financial support from the VINNOVA CompetenceCenter for Surfactants Based on Natural Products, SNAP.Marie Ernstsson and Mikael Sundin are acknowledgedfor contributions in the ESCA measurements.

LA035959P

4058 Langmuir, Vol. 20, No. 10, 2004 Matsson et al.