Formation and Ordering of Gold Nanoparticles at the Toluene-Water Interface

Milan K. Sanyal,* ,†,‡ Ved V. Agrawal,‡ Mrinal K. Bera, †,§ K. P. Kalyanikutty, ‡ Jean Daillant,|Christian Blot, | S. Kubowicz,| Oleg Konovalov,⊥ and C. N. R. Rao‡

Surface Physics DiVision, Saha Insititute of Nuclear Physics, 1/AF Bidhannagar, Kolkata-700064, India,Chemistry and Physics of Materials Unit, Jawaharlal Nehru Centre for AdVanced Scientific Research, JakkurP.O., Bangalore-560064, India, S. N. Bose National Centre for Basic Sciences, JD-Block, Sector-3, Salt Lake,Kolkata-700098, India, LIONS, CEA Saclay, F-91191 Gif-sur-YVette Cedex, France, and European SynchrotronRadiation Facility, Beamline ID10B, P.O. Box 220, 38043 Grenoble, France

ReceiVed: NoVember 6, 2007; In Final Form: January 5, 2008

Microscopic measurements that provide direct information in nanometer length scales are essential to obtaina proper understanding of the interfacial reactions that form nanostructured materials. We present here theresults of a synchrotron X-ray scattering study of the formation and ordering of gold nanoparticles at thetoluene-water interface through a reduction reaction. The observed X-ray reflectivity and diffuse scatteringdata show the formation of a monolayer of “magic clusters” at the water-toluene interface. Each clusterconsists of 13 nanoparticles with about 12 Å diameter, similar to Au-55 nanoparticles, with about an 11 Åorganic layer and an in-plane cluster-cluster separation of 180 Å. The electron density profile of the monolayerof these clusters exhibits three layers of nanoparticles as a function of depth that evolves with time.

Understanding chemical reactivity at the liquid-liquid in-terface is of fundamental importance in several research areasincluding drug delivery and diffusion through biological mem-branes because the effect of the inhomogeneous environmentcan alter the behavior of the reacting molecules.1-6 Thehydrogen bonds in the aqueous surface and the transfer of ions/charge from one liquid into the other can get altered due to thepresence of microscopic roughness, which is related to macro-scopic interfacial tension through capillary wave theory.7-9 Ithas been demonstrated that a toluene-water interface can beexploited to form two-dimensional aggregates of nanoparticlesof metals, metal sulfides, and other materials at the interfaceby clever choice of interfacial chemistry.6 The film of nano-particles produced at the liquid-liquid interfaces is expectedto be very thin because it is known3 that water “fingers” andorganic “fingers” protrude into one another only around 10 Ålasting for tens of picoseconds to facilitate interfacial chemicalreactions. An in situ study of nanoparticle formation andordering should provide us with a unique opportunity to probethe inhomogeneous chemical reactivity and associated ion/charge transfer processes across interfaces.

We report here an in situ study of formation and ordering ofgold nanoparticles at the water-toluene interface using high-energy synchrotron X-ray scattering techniques.7 The primarymotivation of this study is to measure the size of the goldnanoparticles generated through an interfacial chemical reactionand to determine the nature of the ordering of these nanoparticles

at the interface. The other aim of this work is to study thetoluene-water interfacial tension as the nanoparticles form atthe interface by measuring the X-ray diffuse scattering intensity8

as a function of the in-plane wave vector transfer (qy). Althoughliquid surfaces have been studied by diffuse scattering measure-ments of capillary wave fluctuations,8-11 it is only recently thatthe buried liquid interfaces are receiving attention.7,11-13

We have performed high-energy X-ray scattering measure-ments at the ID10B beamline of the European SynchrotronRadiation Facility (ESRF) to investigate the formation ofultrathin films of gold nanoparticles at the toluene-waterinterface. We have used a large trough (7 cm wide and 30 cmlong) for this reaction to keep the in-plane pressure low and anantivibration table to stabilize the interface for carrying out theX-ray measurements. The energy of the monochromatic beamwas set to 21.9 keV to allow the X-ray beam to pass throughthe upper liquid (toluene here), and reflectivity and diffusescattering data were collected from the toluene-water interface.Two thin silicon wafers of equal heights were used near theentry and exit X-ray windows of the Langmuir trough to anchorthe toluene-water interface. In this setup, the X-ray intensityreduces by a factor of 0.142 and 0.0195 as the 21.9 keV beampass through 7 cm of toluene and water, respectively.

We also collected scattering data in the same geometry fromthe toluene bulk by moving down the interface 0.2 mm tosubtract the background arising from bulk scattering. Weneglected the contribution of bulk water scattering in the analysisbecause the incident angle of the beam was low and theabsorption coefficient of water is much higher than that oftoluene. The scattered intensity profiles measured as functionof horizontal and vertical components of the wave-vectortransfersqz and qy, being related to the out-of-plane and in-plane anglesθf and φ,11,14 provide information regarding the

* Corresponding author. E-mail: milank.sanyal@saha.ac.in.† Surface Physics Division.‡ Chemistry and Physics of Materials Unit.§ S. N. Bose National Centre for Basic Sciences.| LIONS.⊥ European Synchrotron Radiation Facility.

1739

2008,112,1739-1743

Published on Web 01/19/2008

10.1021/jp710635e CCC: $40.75 © 2008 American Chemical Society

out-of-plane and in-plane ordering of nanoparticles formed atthe liquid-liquid interface. The position of the interface wasadjusted to minimize the meniscus with controlled addition andremoval of water and by monitoring the sharpness of thereflected beam at smallθi. Reflectivity data of the toluene-water interface were collected using a point detector by changingthe incident and reflected angles and keeping these two anglesequal (θi ) θf). For diffuse scattering measurements, a position-sensitive detector (PSD) was used to collect the data as afunction of the in-plane angle (φ), keeping the incident grazingangle (θi) fixed at 80 millidegrees. All of the measurementswere performed at room temperature (23°C) with the incidentbeam size of 0.017× 1.0 mm2 (V × H) defined usingconventional slits. Details of the experimental arrangement areavailable in the literature.9,10,12

We carried out the reaction at the interface between a metal-organic compound triphenylphosphine gold chloride, Au(PPh3)-Cl (P≡ phosphorus, Ph≡ phenyl), in the toluene layer and thereducing agent, tetrakishydroxymethylphosphonium chloride(THPC), in the aqueous layer.6 The process was repeated severaltimes to confirm reproducibility of the scattering data. It wasshown earlier5 using X-ray diffraction data of the transferredfilms on solid substrates that the reduction reaction with THPCmay produce Au-55 particles with a magic number of atomshaving a diameter of around 12 Å, though transmission electronmicroscopy (TEM) indicated much larger sizes due to electronbeam damage. A TEM study of gold nanoparticles transferredon to a grid from the toluene-water interface6 also indicated aparticle size of about 80 Å diameter though the reaction zoneat liquid-liquid interface is expected to be around 10 Å due toprotrusion of liquid “fingers”. The results of synchrotron X-rayscattering and atomic force microscopy (AFM) presented here,however, indicate formation of an Au-55 “magic particle”having a diameter of around 12 Å. The use of an antivibrationtable that minimizes macroscopic interdiffusion across thetoluene-water interface is of crucial importance to obtain theresults presented here.

In Figure 1a, we have shown six reflectivity curves collectedafter the initiation of the reaction and indicated the time of datacollection. We have also presented in the insets of Figure 1aand b, the reflectivity data and electron density profile of thebare toluene-water interface, respectively. We required around30 min time to collect a reflectivity profile and have notpresented the data collected within 194 min of initiation ofreaction here because the profiles changed during scans. Theoscillations in the measured reflectivity curves indicate thepresence of a thin film at the toluene-water interface.

These curves were fitted using an iterative inversion techniquebased on the Born approximation.15 The inversion scheme isbased on recursive Fourier transform (FT) and inverse Fouriertransform (FT-1) of the equation relating the model electrondensity profile (Fm(z)) with actual electron density profile(Fe(z)) as

Here Re is the experimental reflectivity curve andRm is themodel reflectivity curve calculated from the model profileFm(z) by the standard slicing technique. In each iteration,Fm(z)is replaced byFe(z) obtained from eq 1 and the process iscontinued untilRe andRm become indistinguishable. We usedthe water-toluene profile (with a small increase in toluenedensity due to presence of Au(PPh3)Cl to match the critical

angle) shown in inset of Figure 1b as initialFm(z) and confirmedthe stability of the solutions by setting various total filmthicknesses. The extracted final electron density profiles, forthese six reflectivity data, which always take the value of waterdensity toward the end of the film, are shown in Figure 1b.

We also verified the solutions by simple models having afew slices; one such simple model having six slices for 194min data is also shown with and without roughness convolution,and the corresponding calculated reflectivity data is shown inFigure 1a. The dip in reflectivity data around 0.123 Å-1 andthe subsequent modulation require a strong peak in electrondensity around 70 Å above the water surface. The width of thispeak and the simple model shown in Figure 1b suggest the sizeof the uppermost gold particle to be around 12 Å. There is abroad composite structure around the middle of the film (markedas 40 Å) and a small hump just before reaching the waterdensity. Interestingly, these three layers have a vertical separa-tion of about 30 Å from each other. The electron densitybetween these layers takes the value (∼0.32 electrons/Å3) of atypical organic material. The electron density of the uppermostlayers reduces, and the peak electron densities of all of the peakstend to become equal as time progresses, but the separation ofthe layers remain nearly constant.

The profiles obtained in Figure 1b with continuous reductionof electron density of the uppermost layer can be explainedassuming coexistence of a monolayer of individual nanoparticleshaving around a 12 Å gold core with an 11 Å organic shell and

Figure 1. (a) Variation of reflectivity and fits as function ofqz afterinitiation of the reaction (pink, 194 min; light blue, 224 min; grey,253 min; green, 283 min; red, 312 min; orange, 364 min) and the fits(solid lines). The reflectivity (dashed line) generated from the simplebox model is also shown. Inset: The reflectivity data (<) from thetoluene-water interface and its fit (solid line). (b) The electron densityprofiles (EDPs) (colors same as those in a) as function of depth obtainedfrom fitting. Positions of the two upper peaks as measured from thewater interface (dashed line) 70 Å and 40 Å are marked. A simplemodel without (dashed line) and with roughness convolution (solid line,blue) are also shown.

Fe(z) ) FT-1[xRe

Rm∫F′m(z) exp(iqzz) dz] (1)

1740 J. Phys. Chem. C, Vol. 112, No. 6, 2008 Letters

a layer of a 13-member “magic cluster” of these nanoparticles.With the progress of reaction, the number of individualnanoparticles reduces and only the monolayer of the “magiccluster” remains at the toluene-water interface. To verify thepresence of the monolayer of individual nanoparticles, wemeasured the thickness of the film after transferring it on to asilicon substrate using an atomic force microscope (employingthe tapping mode with Nanoscope IV). We transferred the filmsafter carrying out the reaction for 15 min on both ordinary andantivibration tables (see Figure 2a and b). Although much thickerfilms (120 Å) were formed on an ordinary table, the filmthickness obtained on an antivibration table was around 23 Å.A monolayer film of individual nanoparticles would have givena film thickness of 34 Å. We feel that the reduced thicknessobserved here is due to the partial removal of the organic layerduring transfer from the toluene-water interface. The thicknessof a film on the antivibration table grows slowly with time,and the rate of growth depends on the in-plane pressure. Adetailed analysis of AFM results of the films transferred atvarious time intervals will be published elsewhere.

In Figure 3A, we have shown the measured diffuse scatteringintensity profiles from the interface and the bulk (obtained bylowering the interface by 0.2 mm) as a function ofqy byintegrating the PSD data overqz. The bulk scattering data wassubtracted from the interfacial data (shown as “Interface-Bulk”in Figure 3a), and a small peak around 0.29 Å-1 is observed.This peak corresponds to the separation (d2) between two

Figure 2. AFM images of the films formed on the toluene-waterinterface inside a beaker placed on (a) a simple table and (b) anantivibration table and deposited on a silicon substrate are shown. Theenlarged views of the regions marked by yellow squares are shown asinsets. The height distributions obtained from the enlarged views andheight profiles corresponding to the marked blue regions of the filmsare shown beside the respective images.

Figure 3. (a) Grazing incidence diffraction (GID) data integrated overqz are shown as a function ofqy from the interface (g), bulk (solidline) and interface- bulk (0). The GID peaks are also marked by 2π/d1 and 2π/d2. (b) The variation of scattered intensity from interface(symbols) at two differentqz values and the corresponding calculatedintensity from capillary wave theory with a fixed interfacial tension.(c) GID data (0) normalized to unity and the fits obtained from ourmodel with core sizes 80 Å (solid line) and 12 Å (dashed line). Inset:The GID peak and the fits are shown in linear scale after subtractingthe capillary scattering contribution. (d) Three-dimensional schematicsof our model that involve a 13-member cluster of organic capped goldnanoparticles at the toluene-water interface are shown along with thetwo-dimensional schematics of an individual nanoparticle and a cluster.(e) Two-dimensional schematics for the same model is also shown alongwith the three-layered electron density profile as a function of depthobtained from the fitting of specular reflectivity data.

Letters J. Phys. Chem. C, Vol. 112, No. 6, 20081741

individual nanoparticles and is much weaker than the peakobserved at 0.04 Å-1 that corresponds to the separation (d1) ofclusters discussed below. Because the measurement of diffusescattering data takes a long time after initiation of reaction, thenumber of individual nanoparticles reduced and the clustersdominated the toluene-water interface. We could not detectany rod or arc scattering in the (qy-qz) plane that represent10

two-dimensional and three-dimensional aggregates of theseparticles and clusters due to strong bulk scattering.

The diffuse scattering data was analyzed8-10 using thecapillary wave model of liquid interfaces for determining theeffective interfacial tension of the toluene-water interface. Thelarge diffuse scattering intensity in the lowqy region as well asthe decay of the intensity asqy

-(2-η) (where theη is equal tokBTqz

2/(2πγ) and γ is the interfacial tension) from a liquidsurface can be explained using a logarithmic height-heightcorrelation induced by capillary fluctuations without any fittingparameters.8 In Figure 3b, we have shown diffuse scatteringdata aroundqz values of 0.09 and 0.13 Å-1 after subtractingthe corresponding bulk scattering data. This data could be fittedwith the two straight lines shown in the log-log plots. Theslopes (η - 2) gave8 the value of the interfacial tension as 1.7mN/m; this value is much lower than the expected value (27.8mN/m) of the toluene-water interface. The small value of theinterfacial tension indicates the enhancement of interfacialroughness probably due to the presence of the organic layerdiscussed below at the interface. It will be interesting to studythe effect of water and toluene “fingers” and the associatedhydration shells involved in the solvation processes on capillaryscattering.

In Figure 3c, we have shown measured data near the grazingincidence diffraction (GID) peak and a fit to a calculated profilethat includes scattering from a two-dimensional radial distribu-tion function of gold nanoclusters and capillary scatteringcontributions; details of this formalism has been publishedelsewhere.10 The position of this peak indicates an in-planeseparation of 180 Å, but the calculated profile with a form factorof a 12 Å spherical nanoparticle is found to be broad and couldnot represent the measured data (dashed lines Figure 3c). Aspherical cluster size of about 80 Å was required to fit the GIDpeak. The solid line shows the best fit that includes Gaussianbroadening of 0.015 Å-1 that represents coherent scatteringsurface area of islands of these clusters. In the lower inset, theGID peak and the fits are shown in linear scale after subtractingthe capillary scattering contribution. It is to be noted thatcapillary scattering contribution is a sloping line in log-logscale.

We present a simple model in Figure 3d that can explain theelectron density profile extracted from the reflectivity data andin-plane scattering around the GID peak. This model assumesa “magic cluster” of 13 Au nanoparticles of 12 Å diameter, 1at the center and 12 surrounding it in a compact spherical shell,each having an organic capping of 11 Å. The central cross-section of this cluster is shown in the lower-right corner ofFigure 3d with a particle-particle separationd2 ) 34 Å. Forthe GID fit, we approximated these clusters with a core-shellmodel having an 80 Å central core and an organic capping of11 Å. If we assume that the location of the water-tolueneinterface is around the middle of the cluster, then the lowerportion of the hydrophobic organic capping cannot stay in waterand floats around the upper half of the spherical cluster to forma ring in toluene just above the interface, making the cappinglook like a “semi-spherical cap” on the metal cluster. Theextension of the resultant capping is shown as black circles in

Figure 3d. Simple volume calculations indicate that a 5-Å-thickring generates an average diameter of these circles or cluster-cluster separationd1 ) 180 Å, as observed in GID data. Earlierresults5,6 of similar nanoparticles formed at room temperatureshowed rather monodispersed larger particles (70 to 80 Ådiameter) in TEM, but X-ray diffraction data from thesetransferred films did not show peaks expected from such largeparticles These observations are also consistent with our presentmodel because the 12 Å nanoparticles in the cluster cannot givedistinct X-ray diffraction peaks but 13 of them will inter-diffusewithin the cluster to form bigger particles due to electronirradiation in TEM.5

In conclusion, we have shown by an X-ray scattering studythat a monolayer of clusters having 13 gold nanoparticles of12 Å diameter with large (180 Å) in-plane cluster-clusterseparation forms at the toluene-water interface. The electrondensity of such aligned clusters is expected to exhibit three layersalong the depth with the central electron density values of 0.33,0.37, and 0.33 electrons/Å.3 Reflectivity measurements confirmthe presence of three layers with the lower two layers showingslightly less than the calculated values probably due to partialcoverage. The presence of low electron densities in betweenthese peak values confirms the monodispersity of these alignedclusters. The higher electron density of the top peak indicatesthe presence of individual gold nanoparticles with these clusters(refer to Figure 3e). The presence of individual nanoparticlesin the initial phase of the reaction is evident from AFMmeasurements of early films. As the reaction progresses, thenumber of individual Au nanoparticles reduces and only magicclusters at the toluene-water interface remain. The presenceof an organic layer (shown as black circles in Figure 3d) at thewater-toluene interface lowers the interfacial tension as seenin diffuse scattering measurements. It seems that the presenceof the clusters and the associated organic layer hinders theprogress of the reaction unless the interface is disturbed bysurface pressure and/or vibrations.

Acknowledgment. M.K.S. and V.V.A. acknowledge thesupport of Department of Science and Technology (India) andService pour la Science & la Technologie (France) for carryingout experiment at ESRF synchrotron. M.K.B. acknowledgesC.S.I.R. (India) for funding and support.

References and Notes

(1) Lin, Y.; Skaff, H.; Emrick, T.; Dinsmore, A. D.; Russell, T. P.Science2003, 299, 226-229.

(2) Nikolaides, M. G.; Bausch, A. R.; Hsu, M. F.; Dinsmore, A. D.;Brenner, M. P.; Gay, C.; Weitz, D. A.Nature2002, 420, 299-301.

(3) Benjamin, I.Chem. ReV. 1996, 96, 1449-1476.

(4) Dryfe, R. A. W.Phys. Chem. Chem. Phys.2006, 8, 1869-1883.

(5) Duff, D. G.; Baiker, A.; Edwards, P. P.Langmuir1993, 9, 2301-2309. Duff; D. G.; Baiker, A.; Gameson, I.; Edwards, P. P.Langmuir1993,9, 2310-2317.

(6) (a) Rao, C. N. R.; Kulkarni, G. U.; Thomas, P. J; Agrawal, V. V.;Saravanan, P.J. Phys. Chem. B2003, 107, 7391-7395. (b) Rao, C. N. R.;Kulkarni, G. U.; Agrawal, V. V.; Gautam, U. K.; Ghosh, M.; Tumkurkar,U. J. Colloid Interface Sci.2005, 289, 305-318.

(7) Luo, G.; Malkova, S.; Yoon, J.; Schultz, D. G.; Lin, B.; Meron,M.; Benjamin, I.; Vanysek, P.; Schlossman, M. L.Science2006, 311, 216-218.

(8) Sanyal, M. K.; Sinha, S. K.; Huang, K. G.; Ocko, B. M.Phys.ReV. Lett. 1991, 66, 628-631.

(9) Fradin, C.; Braslau, A.; Luzet, D.; Smilgies, D.; Alba, M.; Boudet,N.; Mecke, K.; Daillant, J.Nature2000, 403, 871-874.

(10) Bera, M. K.; Sanyal, M. K.; Pal, S.; Daillant, J.; Datta, A.; Kulkarni,G. U.; Luzet, D.; Konovalov, O.Europhys. Lett.2007, 78, 56003-56009.

1742 J. Phys. Chem. C, Vol. 112, No. 6, 2008 Letters

(11) Fukuto, M.; Gang, O.; Alvine, K. J.; Pershan, P. S.Phys. ReV. E2006, 74, 031607-1-19.

(12) (a) Fradin, C.; Luzet, D.; Braslau, A.; Alba, M.; Muller, F.; Daillant,J.; Petit J. M.; Rieutord, FLangmuir1998, 14, 7327-7330. (b) Mitrinovic,D. M.; Zhang, Z.; Williams, S. M.; Huang, Z.; Schlossman, M. L.J. Phys.Chem. B1999, 103, 1779-1782. (c) Martin, E. S.; Konovalov, O.; Daillant,J. Thin Solid Films2007, 515, 5687-5690.

(13) Mezger, M.; Reichert, H.; Scho¨der, S.; Okasinski, J.; Schro¨der, H.;Dosch, H.; Palms, D.; Ralston, J.; Honkima¨ki, V. Proc. Natl. Acad. Sci.U.S.A.2006, 103, 18401-18404.

(14) Basu, J. K.; Sanyal, M. K.Phys. Rep.2002, 363, 1-84.

(15) Sanyal, M. K.; Hazra, S.; Basu, J. K.; Datta, A.Phys ReV. B 1998,58, R4258-R4261.

Letters J. Phys. Chem. C, Vol. 112, No. 6, 20081743