Langmuir 2000, 16, 9775-9783
9775
Amphoterization of Colloidal Gold Particles by Capping with Valine Molecules and Their Phase Transfer from Water to Toluene by Electrostatic Coordination with Fatty Amine Molecules Ashavani Kumar,† Priyabrata Mukherjee,‡ Ayon Guha,† S. D. Adyantaya,† A. B. Mandale,† Rajiv Kumar,‡ and Murali Sastry*,† Catalysis and Materials Chemistry Divisions, National Chemical Laboratory, Pune 411 008, India Received June 23, 2000. In Final Form: September 5, 2000
The surface modification of colloidal gold particles with the amino acid valine is demonstrated. Selfassembly of valine on the gold particles is accomplished in the aqueous phase, linkage with the gold particles possibly occurring through covalent interaction of the amine group with the surface gold atoms. Derivatization with the amino acid in this manner imparts amphotericity to the gold particles, the particles being negatively charged at pH values greater than 6 (the isoelectric point of valine) and positively charged below this pH. The charge reversal on the gold particles is demonstrated by electrostatic self-assembly of the colloidal gold particles on glass (which is negatively charged at pH > 3) by immersion of the substrate alternately into the valine-capped gold solutions maintained at pH ) 3.5 and 8.5. The phase transfer of the aqueous valine-capped gold colloidal particles by electrostatic linkage with fatty amine molecules present in toluene is also described. The two-phase method described herein is extremely simple and results in hydrophobic gold particles which may be precipitated out of solution and redissolved in different nonpolar organic solvents without significant variation in the particle size distribution. The electrostatically driven multilayer assembly of the valine-derivatized gold particles on glass substrates, the phase transfer process, and the stability of the hydrophobized gold particles have been investigated with a variety of techniques such as UV-vis, infrared and X-ray photoemission spectroscopies, ellipsometry, and thermogravimetry/differential thermal analyses.
Introduction There has been much recent interest in the surface modification of colloidal particles by self-assembly of suitable surfactants (the so-called three-dimensional selfassembled monolayers, or 3-D SAMs), motivated to a large extent by the impressive advances in the understanding of self-assembly on planar surfaces.1 Colloidal gold, in particular, has been studied in detail, and numerous reports on the surface modification with alkanethiols,2 ω-functionalized alkanethiols,3 aromatic thiols,4 and primary amines5 have appeared in the literature. The ability to modify the surface of colloidal gold with terminally functionalized thiol molecules is important in the use of the colloidal particles as scaffolds for novel chemical reactions,6 in gold-sol-based immunoassays,7 and in the self-assembly of the particles in thin film form.8 * To whom correspondence should be addressed. Phone: +91 20 5893044. Fax: +91 20 5893952/5893044. E-mail: sastry@ ems.ncl.res.in. † Materials Chemistry Division. ‡ Catalysis Division. (1) Ulman, A. An Introduction to Ultrathin Organic Films: From Langmuir-Blodgett to Self-Assembly; Academic Press: New York, 1991. (2) (a) Brust, M.; Walker, M.; Bethell, D.; Schiffrin, D. J.; Whyman, R. J. Chem. Soc., Chem. Commun. 1994, 801. (b) Porter, L. A.; Ji, D.; Westcott, S. L.; Graupe, M.; Czernuszewicz, R. S.; Halas, N. J.; Lee, T. R. Langmuir 1998, 14, 7378. (3) (a) Weisbecker, C. S.; Merritt, M. V.; Whitesides, G. M. Langmuir 1996, 12, 3763. (b) Templeton, A. C.; Hostetler, M. J.; Kraft, C. T.; Murray, R. W. J. Am. Chem. Soc. 1998, 120, 1906. (4) (a) Brust, M.; Fink, J.; Bethell, D.; Schiffrin, D. J.; Kiely, C. J. Chem. Soc., Chem. Commun. 1995, 1655. (b) Johnson, S. R.; Evans, S. D.; Mahon, S. W.; Ulman, A. Langmuir 1997, 13, 51. (c) Mayya, K. S.; Patil, V.; Sastry, M. Langmuir 1997, 13, 3944. (5) Leff, D. V.; Brandt, L.; Heath, J. R. Langmuir 1996, 12, 4723.
Although ionizable terminal groups such as carboxylic acid groups3a,4c,8a,b have been anchored to colloidal gold particles in order to electrostatically stabilize the colloidal particles in an aqueous phase, to the best of our knowledge there have been no reports on the use of amphoteric capping agents for colloidal gold. We demonstrate herein that the amino acid valine may be used to cap gold colloidal particles and render them amphoteric. The gold surface coordination of valine molecules is readily effected in the aqueous colloidal phase, the binding of valine to gold occurring possibly through a weak covalent linkage of the amine groups with the particle surface as suggested by Leff, Brandt, and Heath for primary amines.5 The pI of valine is ca. 6; therefore, above this pH the gold colloidal particles acquire a net negative charge whereas they are positively charged at pH values lower than 6. In the first part of this paper, the amphoteric nature of the valinecapped gold particles is demonstrated by sequential electrostatic assembly of gold particles at solution pH ) 3.5 and pH ) 8.5 onto negatively charged glass substrates as illustrated in Scheme 1. We have recently demonstrated such a process for the growth of multilayers of positively charged gold and negatively charged silver particles on (6) Templeton, A. C.; Hostetler, M. J.; Warmouth, E. K.; Chen, S.; Hartshorn, C. M.; Krishnamurthy, V. M.; Forbes, M. D. E.; Murray, R. W. J. Am. Chem. Soc. 1998, 120, 4845 and references therein. (7) (a) Storhoff, J.; Elghanian, R.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L. J. Am. Chem. Soc. 1998, 120, 1959. (b) Sastry, M.; Lala, N.; Patil, V.; Chavan, S. P.; Chittiboyina, A. G. Langmuir 1998, 14, 4138. (8) (a) Patil, V.; Malvankar, R. B.; Sastry, M. Langmuir 1999, 15, 8197. (b) Sastry, M.; Mayya, K. S.; Patil, V.; Paranjape, D. V.; Hegde, S. G. J. Phys. Chem. B 1997, 101, 4954. (c) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221.
10.1021/la000886k CCC: $19.00 © 2000 American Chemical Society Published on Web 11/11/2000
9776
Langmuir, Vol. 16, No. 25, 2000
Scheme 1. Diagram Showing the Various Stages in the Sequential Electrostatic Assembly of Valine-Capped Gold Particles on Glassa
a Step 1: electrostatic assembly of positively charged gold particles on a negatively charged glass substrate during immersion in the gold solution at pH ) 3.5. Step 2: electrostatic assembly of negatively charged gold particles on positively charged gold particle-covered glass during immersion in the gold solution at pH ) 8.5. The process may be repeated to grow multilayer assemblies.
glass, the charging of the particles achieved by ionization of surface-bound 4-aminothiophenol and 4-carboxythiophenol molecules for gold and silver particles, respectively.9 For completeness, the essential details are repeated in this paper. In the first step, negatively charged glass substrates are immersed in the valine-capped gold solution (pH ) 3.5) and a gold particle monolayer is electrostatically assembled on the glass surface. This leads to charge reversal of the glass surface and the possibility of electrostatically assembling a succeeding layer of negatively charged valine-capped gold particles (pH ) 8.5) as shown in step 2 of Scheme 1. Thereafter, the process may be repeated to form multilayer structures of the Au particles. The stability of the valine-capped colloidal gold solutions at different pH values has also been studied, the details of which are given in this part of the paper. One of the emerging challenges in the synthesis of nanoparticles in colloidal form is directing the particles into different solvents providing varying physicochemical environments. Colloidal particles stabilized with suitable surfactants in volatile organic solvents are particularly attractive because they are known to self-assemble into close-packed structures on evaporation of the solvent.10 Most reports on the growth of colloidal metal particles such as gold follow the two-phase approach developed by Brust and co-workers wherein metal ions are transferred to the organic layer using a phase transfer reagent and thereafter are reduced in the organic medium in the presence of suitable capping agents such as alkanethiols2a or alkylamines.5 In a markedly different approach, Sarathy et al. have demonstrated that colloidal gold, platinum, and silver particles first synthesized in aqueous media may be transferred into a hydrocarbon environment such as toluene by acid-facilitated coordination of the particles with alkanethiols present in toluene.11 This method has also been applied to the acid-facilitated transfer of oleatestabilized colloidal silver particles by Wang, Efrima, and Regev.12 This work does not use a surfactant in the organic (9) Kumar, A.; Mandale, A. B.; Sastry, M. Langmuir 2000, 16, 6921. (10) Wang, Z. L. Adv. Mater. 1998, 10, 13 and references therein. (11) Vijaya Sarathy, K.; Kulkarni, G. U.; Rao, C. N. R. J. Chem. Soc., Chem. Commun. 1997, 537. (12) Wang, W.; Efrima, S.; Regev, O. Langmuir 1998, 14, 602.
Kumar et al.
phase; the stabilization of the silver particles in the organic medium is effected by a change in the orientation of surface-bound oleic acid molecules.12 In the second part of this paper, we advance our investigation on valine-capped colloidal gold particles to show that the gold particles synthesized in water can be quantitatively transferred into toluene by coordination to octadecylamine (ODA) molecules present in the organic phase. Under appropriate pH conditions of the aqueous colloidal phase, phase transfer of the valine-capped gold particles occurs because of electrostatic coordination of the carboxylate ions in the gold-surface-bound valine molecules and ODA in toluene, this linkage rendering the gold particles hydrophobic. The resulting ODA-stabilized gold particle powder may be dissolved in different organic solvents and cast in thin film form by simple solvent evaporation. The method demonstrated herein combines the simplicity of the two-phase transfer method without the use of acid to facilitate the phase transfer of the colloidal particles. Presented below are details of the investigation. Experimental Details Gold colloidal particles were synthesized by borohydride reduction of HAuCl4 (1.5 × 10-3 g of HAuCl4 in 100 mL of water) as described elsewhere.8a This procedure yields a clear red solution at pH ) 9 with gold particles of size 35 ( 7 Å.8a The gold particles were thereafter capped with the amino acid L-valine (Sigma Chemicals) by mixing a carefully weighed quantity of the amino acid to yield an overall valine concentration of 10-4 M in the colloidal solution.13 (Please note that the capping of the gold particles with valine was carried out at pH ) 9 where the amine groups in the amino acid are unprotonated. This has important implications for the covalent binding of valine to the gold surface, as will be discussed subsequently.) The pH of the valine-capped gold solution was adjusted to 8.5 and 3.5 by addition of dilute HCl. The pI of valine is ca. 6, and therefore the colloidal gold particles are expected to acquire net negative and positive charges above and below pH ) 6, respectively. The stability of the colloidal solutions at these pH values is a good indicator of the surface binding of the valine molecules and was studied by UV-vis spectroscopy measurements carried out on a HewlettPackard 8542A diode array spectrophotometer operated at a resolution of 2 nm. After synthesis, capping, and stabilization of the gold colloidal particles with valine, multilayers of the colloidal particles were formed on borosilicate glass substrates by sequential immersion of the substrate in the pH ) 3.5 and pH ) 8.5 gold colloidal solutions for 24 h as illustrated in Scheme 1. The sequential assembly of the valine-capped gold colloidal particles on glass was studied by UV-vis spectroscopy and ellipsometry. Care was taken to thoroughly wash the substrates with deionized water prior to measurement of the optical properties. Ellipsometry measurements of the growth of individual layers of valine-capped Au particles on glass were performed using a manually operated Gaertner L 119 null ellipsometer operated in the polarizercompensator-sample-analyzer (PCSA) mode at an angle of incidence of 60°. One side of the glass substrate was made optically rough to prevent specular reflection of the laser beam from that surface, and the other, optically flat surface was exposed sequentially to the different colloidal solutions. The compensator was a quarter-wave plate set with the optical axis at 45° to the plane of incidence. The light source was a He-Ne laser (5 mW), the wavelength being 6328 Å. The ellipsometric angles Ψ and ∆ were determined from a measurement of the polarizer and analyzer angles in four zones to correct for any instrument (13) Assuming all the chloroaurate ions in solution (∼4.4 × 10-6 mol in 100 mL of solution) are reduced and the gold particles have a diameter of 35 Å, the amount of valine required to fully cap the gold particle surface (assuming an area per valine molecule on the gold particle surface of 25 Å2) can be shown to be ∼5.4 × 10-7 mol per 100 mL of water. The actual concentration of valine in the aqueous colloidal solution was 1 × 10-6 mol per 100 mL and was thus taken in excess of that required to fully cap the gold particle surface.
Amphoterization of Colloidal Gold Particles
Langmuir, Vol. 16, No. 25, 2000 9777
Figure 2. UV-vis spectra of aqueous colloidal gold particles capped with valine monolayers: (A) at pH ) 8.5 immediately after capping with valine (solid line) and after 48 h of capping (dashed line) and (B) at pH ) 3.5 immediately after capping with valine (solid line) and after 48 h of capping (dashed line).
Figure 1. (A) Picture showing the two-phase valine-capped gold hydrosol-ODA-containing toluene layers before (test tube on the left) and after (test tube on the right) phase transfer of the gold particles into toluene. (B) UV-vis spectra recorded from the ODA-stabilized valine-capped gold particles in toluene (curve 3), chloroform (curve 2), and n-hexane (curve 1). The curves have been displaced for clarity. misalignment. A photomultiplier was used to determine the extinction (null) condition precisely. The model used in the ellipsometry analysis is discussed subsequently. A film consisting of four layers of valine-capped gold colloidal particles grown on glass as well as drop-cast films of the colloidal solution at pH ) 3.5 and pH ) 8.5 on Si(111) wafers were subjected to chemical analysis using X-ray photoemission spectroscopy (XPS). The measurements were carried out on a VG MicroTech ESCA 3000 instrument at a pressure >1 × 10-9 Torr. The general scan and Au 4f, C 1s, N 1s, and O 1s core-level spectra were recorded with unmonochromatized Mg KR radiation (photon energy ) 1253.6 eV) at a pass energy of 50 eV and an electron takeoff angle (angle between electron emission direction and surface plane) of 60°. The overall resolution was ∼1 eV for the XPS measurements. The core-level spectra were background corrected using the Shirley algorithm,14 and the chemically distinct species were resolved using a nonlinear least-squares procedure. The core-level binding energies (BE) were aligned taking the adventitious carbon binding energy as 285 eV. A four-layer valine-capped gold particle film was formed on oxidized Si(111) wafers in a manner similar to that adopted for the glass substrates, and the film thus formed was analyzed by Fourier transform infrared (FTIR) spectroscopy. FTIR measurements of the gold particle film were made on a Shimadzu PC8201 PC instrument in the diffuse reflectance mode at a resolution of 4 cm-1. In a typical phase-transfer experiment, 100 mL of the valinecapped gold colloidal solution at pH ) 8.5 was added to 100 mL of a 2 × 10-4 M solution of octadecylamine (ODA, Sigma) in toluene to yield immiscible layers of the colorless organic solution on top of the red-colored gold hydrosol (Figure 1A, test tube on the left). Vigorous shaking of the test tube resulted in extremely rapid transfer (within 30 s) of the gold colloidal particles into the (14) Shirley, D. A. Phys. Rev. B 1972, 5, 4709. An application written in Mathcad (a commercial mathematical software package available from Mathsoft Inc.) by one of us (M.S.) was used for the background analysis and is available from www.mathsoft.com.
organic phase, and this was observed by the red coloration of the organic layer (and a corresponding loss of color from the aqueous layer) when the two phases separated (Figure 1A, test tube on the right). This occurs spontaneously and did not require the addition of acid which is an essential ingredient in the alkanethiolcoupled phase-transfer protocol.11 The toluene solution of the gold particles was rotary evacuated and yielded a brownish solid which was repeatedly washed with ethanol and filtered to remove uncoordinated ODA molecules in the powder. The resulting powder could be readily dissolved in different organic solvents such as chloroform, carbon tetrachloride, benzene, and so forth without any apparent change in the particle size distribution (PSD), and these solutions were characterized using UV-vis spectroscopy measurements carried out on a Shimadzu dualbeam spectrophotometer operated at a resolution of 0.5 nm. A carefully weighed quantity of the gold powder was subjected to thermogravimetry (TGA) and differential thermal analysis (DTA) on a Seiko Instruments model TG/DTA 32 at a heating rate of 10 °C/min. Some of the gold powder was dissolved in chloroform, and a film of the gold particles was formed by solvent evaporation on a holey transmission electron microscope (TEM) grid for microscopy measurements and on a Si(111) substrate for FTIR spectroscopy measurements. The TEM measurements were carried out on a JEOL model 1200EX instrument operated at an accelerating voltage of 120 kV.
Results and Discussion 1. Amphoterization and Stability of Valine-Capped Gold Particles. There are few reports in the literature on the surface modification of colloidal gold particles with amino acids. An important concern, therefore, would be the stability of the valine-capped gold colloidal particles in solution at the pH values at which the electrostatic assembly on glass is accomplished (pH ) 3.5 and pH ) 8.5). Parts A and B of Figure 2 show the spectra recorded from the valine-capped gold solutions at pH ) 8.5 and pH ) 3.5, respectively, immediately after capping with valine (solid lines in both figures) and after 48 h of capping with the amino acid (dashed lines in both figures). The figure shows that there is little change in the spectra of the colloidal gold solution even after aging for 48 h at pH ) 3.5 (Figure 2B) whereas there is a small broadening and red shift of the gold surface plasmon resonance for the solution at pH ) 8.5 (Figure 2A). The change in the optical properties of the solution is consistent with some degree of aggregation of the gold particles at the higher pH value.4c However, the solution was still clear and it was stable for many weeks, and therefore this aspect is not expected to influence the main conclusions of this work. This is an important result and indicates that during the time of immersion of the glass/Si(111) substrates in the colloidal solutions at pH ) 3.5 and 8.5 (24 h), the gold particles are
9778
Langmuir, Vol. 16, No. 25, 2000
stable. Therefore, UV-vis features observed in the gold particle films on glass may be safely attributed to factors associated with adsorption on the glass surface. From the UV-vis results presented in Figure 2, it is inferred that valine is bound to the surface of the gold particles and, furthermore, that ionization of the carboxylic acid and amine functional groups at solution pH values of 8.5 and 3.5, respectively, is responsible for the particle stability. Leff, Brandt, and Heath5 have studied the stabilization of gold colloidal particles with alkylamine molecules and have concluded that a weak covalent interaction occurs between the amine group and the surface gold atoms. By comparison with earlier results on the self-assembly of alkylamines on planar gold surfaces where the monolayers were found to be unstable,15 they have shown that finite size effects present during adsorption on colloidal particle surfaces are responsible for the stability of the 3-D SAMs of primary amines.5 We believe that a similar covalent interaction between the amine group in the valine molecule and the gold surface leads to the capping postulated from the UV-vis measurements mentioned above. The formation of a -NH-Au0 bond at high pH appears probable given that the colloidal particles are stable at low pH values as well. The stabilization at low pH could occur through protonation of the amine groups via formation of a -NH2+-Au0 complex, thus providing the necessary electrostatic stabilization. The strength of the valine-gold particle interaction will be dealt with in more detail in the latter half of this paper during the discussion of the TGA/DTA measurements of the phase-transferred gold particles. The surface modification of colloidal gold particles with primary amines in the study of Leff, Brandt, and Heath was carried out in an organic medium (toluene)5 whereas the capping with valine in this study is done in the aqueous phase, and this is an important point of deviation. The sign of the charge on the valine-capped colloidal gold particles at pH ) 8.5 and 3.5 may be conveniently studied by electrostatic assembly of the particles on an oppositely charged surface. Such a layer-by-layer electrostatically controlled self-assembly process has been used with success in forming superlattices of polyelectrolytes, charged biopolymers such as DNA and proteins, and inorganic colloids.16 The demonstration of this process with the valine-capped gold particles is done by immersion of borosilicate glass substrates in the colloidal solution as briefly mentioned in the Introduction and illustrated in Scheme 1. It is known that the negative charge at the glass-water interface is due to dissociation of surface silanol groups,17 and therefore immersion of the glass substrates in the valine-capped gold colloidal solution at pH ) 3.5 should lead to electrostatic self-assembly of the gold particles on the glass surface (step 1, Scheme 1). Thereafter, the charge reversal on the glass surface resulting from the adsorption of positively charged gold particles may be used to electrostatically assemble negatively charged valine-capped gold particles by immersion in the colloidal solution at pH ) 8.5 (step 2, Scheme 1) and can be repeated to yield multilayer structures (Scheme 1). The UV-vis spectra recorded from a glass substrate after the various immersion cycles mentioned above are shown in Figure 3. The spectrum (15) Xu, C.; Sun, L.; Kepley, L. J.; Crooks, R. M. Anal. Chem. 1993, 65, 2102. (16) (a) Decher, G. Science 1997, 277, 1232 and references therein. (b) Lvov, Y.; Ariga, K.; Onda, M.; Ichinose, I.; Kunitake, T. Langmuir 1997, 13, 6195. (17) (a) Poptoshev, E.; Rutland, M. W.; Claesson, P. M. Langmuir 1999, 15, 7789. (b) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554.
Kumar et al.
Figure 3. UV-vis spectra recorded from a glass substrate after immersion in the valine-capped colloidal gold solutions at different pH levels for 24 h (see text for details). The different immersion cycles are identified in the legend shown in the figure.
recorded after the first immersion of the glass substrate in the pH ) 3.5 gold solution for 24 h shows the presence of gold particles on the glass surface with a well-defined resonance at ca. 590 nm (Figure 3, squares). The substrate was carefully washed with deionized water prior to measurement of the UV-vis spectrum. A blue coloration of the glass substrate was clearly visible to the naked eye at this stage. Comparison of the spectrum of the gold particle film on glass with the UV-vis spectrum recorded from the valine-capped gold solution at pH ) 3.5 (Figure 2B) shows a large red shift in the surface plasmon resonance from the solution value of ca. 520 nm. The red shift in the surface plasmon resonance occurs because of overlap of the dipole oscillation modes from neighboring particles and thus indicates fairly close packing of the gold particles on the glass surface.4c As a further test of the sign of the charge on the gold particles, a bare glass substrate was immersed in the valine-capped gold solution at pH ) 8.5 for 24 h, and the spectrum recorded is shown in Figure 3 (solid line). At this pH, both the glass and gold particle surfaces are expected to be negatively charged. The presence of gold particles on the glass surface is clearly below the detection limits of the UV-vis spectrophotometer for this control experiment, and this provides further support to the valine-induced amphotericity of the gold colloidal particles. The formation of the second layer after immersion for 24 h of the now positively charged gold particle-glass surface in the pH ) 8.5 gold solution can clearly be seen from the UV-vis spectrum (Figure 3, circles). The resonance at 590 nm has shifted to larger wavelengths indicating further close-packing of the gold particles, and a new, well-defined plasmon resonance is observed at ca. 525 nm. The presence of a less densely packed layer of gold particles is indicated during this cycle of immersion. Further immersion in pH ) 3.5 (triangles, Figure 3) and pH ) 8.5 (Figure 3, diamonds) gold solutions clearly shows the growth of the resonance at ca. 530 nm as well as a continued increase in the longer wavelength resonance intensity. Thus, the electrostatic self-assembly process studied through UV-vis measurements of the gold particle films on glass clearly supports the contention that the valine-capped gold particles are amphoteric.
Amphoterization of Colloidal Gold Particles
Langmuir, Vol. 16, No. 25, 2000 9779
An interesting but as yet not too well understood aspect of the layer-by-layer electrostatic assembly of valinecapped gold particles as described above is the following. Deposition of the first layer of gold particles on negatively charged glass substrates at pH ) 3.5 is easily understood in terms of attractive electrostatic interactions. However, during subsequent immersion in the valine-capped gold colloidal solution at pH ) 8.5 it may be expected that both the gold particles in solution and those bound to the glass surface would acquire a negative charge by ionization of the carboxylic acid groups in the surface-bound valine molecules. The UV-vis measurements presented in Figure 3 (and the ellipsometry measurements to be discussed below) clearly show, however, that additional layers continue to be deposited as the colloidal solution pH is alternated between 8.5 and 3.5. A possible explanation for this observation may be a change in the ionization constant of the carboxylic acid and amine functional groups in the surface-bound valine molecules arising from assembly of the gold particles on the glass surface. The pKA of terminal carboxylic acid groups in mixed self-assembled monolayers (SAMs) increases as the concentration of the groups is progressively reduced in the SAM.18 Variation in the ionization constant of carboxylic acid groups bound to colloidal particle surfaces of varying curvature has also been observed.19 After immobilization of the colloidal particles on the glass surface, interactions between the particles would be enhanced (relative to that existing in solution), and this could conceivably result in a variation in the ionization constant of both the amine and carboxylic acid groups in the gold-surface-bound valine molecules on the glass surface. Hydrogen bonding between valine molecules in successive layers of the gold particles may also play a role in realizing the multilayer structures formed. Further work is required before this aspect of the work is understood completely. Although the UV-vis spectroscopy measurements shown in Figure 3 indicate that the valine-capped gold particles do adsorb sequentially on glass substrates at pH ) 3.5 and 8.5, information on the packing density of the particles in the individual layers may be more accurately obtained from ellipsometry. As in the case of the UV-vis measurements, the ellipsometric angles were recorded after immersion of the substrate in the respective colloidal solutions for 24 h and thoroughly washing and drying the substrates prior to measurement. Ellipsometry measurements of the Au layers sequentially adsorbed on a glass substrate as described earlier were analyzed in terms of the Bruggeman effective medium model.20 This effective medium formalism was chosen because it is applicable even for large filling fractions of the colloidal particles in the individual layers.20 In the analysis, the sequentially adsorbed colloidal particles are considered to occupy layers of thickness equal to the size of the gold particles (35 Å). The individual layers thus constitute a two-component mixture of the colloidal particles of dielectric function C (filling fraction f) and air (M ) 1). For such a twocomponent system, the governing equation relating the effective medium (air plus colloidal particles) dielectric function BR to the dielectric functions of the components and the filling fraction is20
(1)
The dielectric function for the gold particles was taken from a compilation by Johnson and Christy21 and is (1.315-1.524i)2 at 6328 Å (laser wavelength in the ellipsometry measurements). After formation of the first gold particle layer on the glass surface, the dielectric function of the layer of thickness 35 Å (the size of the gold particles) was determined from the ellipsometric angles Ψ and ∆, and thereafter the filling fraction was calculated using eq 1. The substrate refractive index in this case was 1.5 that of glass. Measurements were made on at least five different points of the film surface and averaged. For the first gold layer, an average filling fraction f ) 37.4% was calculated. This translates into a particle density of 5.83 × 1012 particles/cm2 of the glass surface. Assuming an area per valine molecule on the gold particle surface of 25 Å2, the number of positive charges on each gold particle may be easily shown to be ca. 154 e. Assuming a typical value for the surface charge density on borosilicate glass (25 Å2 per charge, 0.64 C/m2),17b it is seen that for this concentration of gold particles on the surface, the positive charge density due to gold ((5.83 × 1012)154 ≈ 8.98 × 1014 cm-2) is larger than the negative charge density on the glass surface ((25 × 10-16)-1 ) 4 × 1014 cm-2). Thus, there is clearly overcompensation of the negative charge on the glass surface because of adsorption of the positively charged valine-capped gold particles, which clearly supports the subsequent electrostatically controlled deposition of the negatively charged gold particles at pH ) 8.5 (step 2, Scheme 1). In a similar manner, the filling fraction for the next layer of gold particles was calculated from the ellipsometric angles and eq 1, taking the glass plus the layer of gold particles as the new substrate. The average filling factor for the second layer of gold particles was calculated to be 28.6%, and it is easy to show that charge overcompensation occurs in this case as well. The filling fractions for the next two gold layers were also calculated and were 21.6% and 22.5%, respectively. Thus, there is clearly a reduction in the particle density as the valinecapped gold multilayer film is built up by electrostatic assembly. The ellipsometry calculations are in agreement with the UV-vis measurements of the valine-capped gold multilayer films shown in Figure 3 and discussed earlier. Lvov et al.16b have pointed out that electrostatic assembly of rigid inorganic particles does not lead to efficient packing of the particles in the individual layers and a flexible polyelectrolyte “glue” was observed to lead to better layerby-layer growth. This may be one of the reasons for the reduction in particle density observed in this study. We point out that the model used for the ellipsometry analysis presented above is highly idealized and it is assumed that each successive layer of colloidal particles grows directly on top of the previous layer even though large voids (resulting from less than optimum packing of the particles in the layer below) exist in the substrate layer. From a purely electrostatic viewpoint, this growth mode does not appear to be unrealistic given that unfilled regions of the substrate would expose charges of the same sign as that of the incoming colloidal particles. We are currently attempting to develop a model along the lines proposed by Yang et al.22 taking into account electrostatic interactions between the particles to understand better the growth mode of such colloidal particle films. FTIR spectra recorded from a four-layer valine-capped gold film deposited on an oxidized Si(111) substrate in
(18) (a) Bain, C. D.; Whitesides, G. M. Langmuir 1989, 5, 1370. (b) Lee, T. R.; Carey, H. A.; Biebuyck, H. A.; Whitesides, G. M. Langmuir 1994, 10, 741.
(19) Gole, A. M.; Sathivel, C.; Lachke, A.; Sastry, M. J. Chromatogr. A 1999, 848, 485. (20) Granqvist, C. G.; Hunderi, O. Phys. Rev. B 1977, 16, 3513. (21) Johnson, P. B.; Christy, R. W. Phys. Rev. B 1972, 6, 4370. (22) Yang, S.; Viot, P.; Van Tassel, P. R. Phys. Rev. E 1998, 58, 3324.
f
C - BR M - BR + (1 - f) )0 C + 2BR M + 2BR
9780
Langmuir, Vol. 16, No. 25, 2000
Figure 4. FTIR spectra recorded from a four-layer valinecapped gold particle film on an oxidized Si(111) substrate: (A) in the range of 1400-1800 cm-1 and (B) in the range of 25003550 cm-1. The main resonances are identified in the figure and discussed in the text.
the spectral ranges of 1400-1800 cm-1 and 2500-3550 cm-1 are shown in parts A and B of Figure 4, respectively. Figure 4A shows two prominent resonances at 1642 and 1544 cm-1. The resonance at 1642 cm-1 arises because of excitation of the NH3+ deformation vibration mode (the “amino acid I” band),23 whereas the peak at 1544 cm-1 is assigned to absorption by the ionized carboxyl groups in the multilayer film.23 The NH3+ resonance is also known to have a component in this frequency range (in the range of 1485-1550 cm-1, the “amino acid II” band).23 In addition to these bands, a relatively weaker resonance is observed at ca. 1740 cm-1. This feature is attributed to absorption by un-ionized carboxyl groups in the film23 and clearly confirms the amphoteric nature of the valine molecules in the multilayer gold film. Figure 4B shows the spectrum recorded from the four-layer valine-capped gold film in the region of the N-H and C-H stretching vibrational modes. A broad and intense resonance centered at 3355 cm-1 is clearly seen in Figure 4B and is assigned to the N-H stretching vibration.24,25 The fact that this peak is broad indicates that there may be some hydrogen bonding between amine and carboxylic acid groups of valine in successive layers in the multilayer valine-capped gold film. The methylene antisymmetric and symmetric vibration modes are observed at 2920 and 2850 cm-1. The frequency of these resonances indicates that the valine molecules on the gold particle surface are in a close-packed state.25 The features shown in Figure 4 are in agreement with reported values for amino acids and confirm the presence of an appreciable concentration of valine in the multilayer gold films. Furthermore, the valine molecules on the gold particle surface are in ionized states consistent with the electrostatic deposition protocol illustrated in Scheme 1. X-ray photoemission spectroscopy (XPS) measurements were carried out on a four-layer gold particle film on glass as well as drop-coated films of the valine-capped gold particles on Si(111) substrates from the pH ) 3.5 and 8.5 colloidal solutions. Figure 5A shows the N 1s core-level spectra recorded from the pH ) 3.5 and 8.5 gold solution films, the spectra having been displaced vertically for clarity. The pH values of the colloidal solutions are given in the figure. It is clear that the N 1s core level consists of two chemically distinct species in both of the films, which have been stripped into the individual components (23) Bellamy, L. J. The Infra-red Spectra of Complex Molecules; John Wiley: New York, 1960; Chapter 13, p 234. (24) Templeton, A. C.; Chen, S.; Gross, S. M.; Murray, R. W. Langmuir 1999, 15, 66. (25) Hostetler, M. J.; Stokes, J. J.; Murray, R. W. Langmuir 1996, 12, 3605.
Kumar et al.
Figure 5. (A) N 1s core-level spectra recorded from drop-coated films of valine-capped gold on Si(111) substrates at colloidal solutions of pH ) 3.5 and 8.5. The spectra have been displaced vertically for clarity. The chemically distinct components are shown for the pH ) 8.5 film (see text for details). (B) N 1s core-level spectrum recorded from a four-layer valine-capped gold film deposited on a glass substrate. The chemically distinct species are shown in the figure and discussed in the text. Table 1. Parameters Obtained from a Nonlinear Least-Squares Curve Resolution Analysis of the Core Levels from Valine-Capped Gold Particle Films valine-capped Au film pH ) 8.5 (N 1s) pH ) 3.5 (N 1s) four-layer gold film (N 1s) pH ) 8.5 (C 1s) pH ) 3.5 (C 1s) four-layer gold film (C 1s) four-layer gold film (Au 4f) four-layer gold film (O 1s)
BE(1) BE(2) BE(3) (eV) (eV) (eV) A(2)/A(1)a A(3)/A(2)a 398.9 403.5 398.5 403.7 399.0 403.3
0.74 2.05 2.08
285.0 287.1 290 285.0 287.3 290.1 285.0 287.5 290.5
4.88 1.10 4.04
84.0
87.6
0.73
529.0 532.1
9.32
0.11 0.19 0.17
a A(1), A(2), and A(3) are the peak areas of components 1, 2, and 3, respectively, in Figures 5 and 6.
by a nonlinear least-squares procedure. The two components are shown in Figure 5A for the pH ) 8.5 gold film, and the parameters obtained from the fits are listed in Table 1 (for clarity, the components in the pH ) 3.5 gold film have not been shown; the fitting parameters may be obtained from Table 1). The two components at ca. 399.0 and 403.5 eV BE (peaks 1 and 2, respectively, in Figure 5A and Table 1) are assigned to electron emission from the un-ionized and ionized amine groups in the goldsurface-bound valine molecules, respectively. Table 1 shows that the peak area ratio of the high BE component to the low BE component increases for the pH ) 3.5 valinecapped gold particle film in comparison with the pH ) 8.5 film (Figure 5A). The gold particle layers formed at pH ) 8.5 would contain amine groups from valine in the predominantly un-ionized form, whereas the amine groups in the layers self-assembled at pH ) 3.5 would be, to a large extent, protonated. The N 1s core-level XPS results of Figure 5A thus clearly provide chemical evidence of the amphotericity of the gold particles in the self-assembled multilayer film. Figure 5B shows the N 1s spectrum recorded from the four-layer valine-capped gold particle film on glass along with the two chemically distinct species discussed above. The parameters obtained from a fit to the N 1s spectrum are listed in Table 1. It is clear that this spectrum shows the characteristics of the individual layers at pH ) 3.5 and pH ) 8.5 (Figure 5A). The C 1s spectra recorded from the pH ) 3.5 and pH ) 8.5 valine-capped gold particle films on Si(111) are
Amphoterization of Colloidal Gold Particles
Figure 6. (A) C 1s core-level spectra recorded from drop-coated films of valine-capped gold on Si(111) substrates at colloidal solutions of pH ) 3.5 and 8.5. The spectra have been displaced vertically for clarity. The chemically distinct components are shown for the pH ) 3.5 film (see text for details). (B) C 1s core-level spectrum recorded from a four-layer valine-capped gold film deposited on a glass substrate. The chemically distinct species are shown in the figure and discussed in the text.
shown in Figure 6A, and the chemically distinct components are shown for the pH ) 3.5 film (Table 1 lists the parameters obtained from the fits to the C 1s core level for both films). The pH at which the gold particle film was deposited is indicated next to the spectra, which have been displaced vertically for clarity. The C 1s spectra could be satisfactorily stripped into three components at ca. 285.0, 287.3, and 290.3 eV BE. The 285 eV component is assigned to electron emission from the methyl groups in valine and adventitious carbon on the film, and the components at 287.3 and 290.3 eV BE are assigned to the carbon coordinated to the amine group and the carbon in the carboxylic acid group, respectively.26,27 From stoichiometry considerations, the two higher BE components would be expected to have roughly the same intensities. We believe the 287.3 eV component may also contain contributions from carbons in the ionized carboxylic acid groups. That this is likely is indicated by the increase in intensity of component 2 (287.3 eV BE, Figure 6A) relative to component 3 (290.3 eV BE) at pH ) 8.5 in comparison with the C 1s spectrum recorded at pH ) 3.5 (see Table 1 for ratios of the areas of the different components). We are not aware of any reports on core-level BE measurements of ionized carboxylic acid functional groups and therefore hesitate in making an unequivocal statement on the quantitative aspects of the C 1s core-level spectra. The C 1s core-level spectrum recorded from a four-layer valine-capped gold particle film on glass is shown in Figure 6B with the resolved components. That this spectrum is a superposition of the C 1s spectra from the individual layers (Figure 6A) is borne out by the coincidence in BEs of the individual components of the four-layer gold film with the low- and high-pH valine-capped gold films as well as the peak area intensity ratio values which lie within the range determined by the individual layers (Table 1). The Au 4f spectrum (Supporting Information, Figure 1) recorded from the four-layer valine-capped gold film on glass showed the 4f7/2 component at 83.8 eV with a spinorbit splitting of 3.7 eV. There was no evidence of additional components. This is in agreement with the XPS results of Leff, Brandt, and Heath for alkylamines on gold5 and clearly shows that higher oxidation state gold atoms do not exist in the clusters after coordination with valine. (26) Davies, M. C.; Lynn, R. A. P.; Davis, S. S.; Hearn, J.; Watts, J. F.; Vickerman, J. C.; Johnson, D. Langmuir 1994, 10, 1399. (27) Sastry, M.; Ganguly, P. J. Phys. Chem. A 1998, 102, 697.
Langmuir, Vol. 16, No. 25, 2000 9781
The O 1s spectrum (Supporting Information, Figure 2) from the multilayer gold film on glass could be decomposed into two components at 528.6 and 532.3 eV, which are assigned to electron emission from oxygen in the ionized and un-ionized carboxylic acid groups, respectively. The higher BE component may also contain contributions from the glass substrate if the surface gold particle layers are sufficiently patchy. 2. Phase Transfer of Aqueous Valine-Capped Gold Particles into Toluene Containing ODA. Figure 1A shows a picture of the test tubes before (left) and after (right) phase transfer of valine-capped gold particles into toluene (layer on top) by electrostatic coordination with ODA molecules present in the organic phase. Powders of the gold particles stabilized by ODA in toluene could be readily obtained by rotary evaporation of the solvent. After the powder was washed with ethanol and further rotary evaporation was performed, the hydrophobized gold powder was dissolved in n-hexane (refractive index n ) 1.374), chloroform (n ) 1.442), and toluene (n ) 1.49). The UV-vis spectra recorded from the gold colloidal particles in the different solvents are shown in Figure 1B (curve 3, toluene; curve 2, chloroform; curve 1, n-hexane). The surface plasmon resonance of colloidal gold is seen prominently in all the spectra with little evidence of aggregation of the particles. There is a shift in the surface plasmon resonance wavelength with refractive index of the solvent (528 nm for toluene, 524 nm for chloroform, and 520 nm for n-hexane). The increase in the resonance wavelength with increasing refractive index of the medium in which the gold particles are dispersed is in agreement with the predictions of the Mie theory of the optical properties of colloidal solutions and has been observed for gold colloids earlier.28 The phase transfer of the valine-capped gold particles into toluene in the manner described above may be explained along these lines. The pI of valine is ca. 6, and consequently the gold particles would be negatively charged at pH ) 9, the as-prepared colloidal solution pH. During mixing of the organic/aqueous layers, it is expected that the amine groups from ODA in the toluene phase would become protonated (pKB of ODA ) 10.8) while in contact with water and would thereafter electrostatically bind to the negatively charged gold particles in the aqueous phase. This would hydrophobize the colloidal gold particles and render them soluble in the toluene phase, as observed. This, to the best of our knowledge, is the first report on electrostatically controlled phase transfer of colloidal gold synthesized in an aqueous medium. Phase transfer of the valine-capped gold particles as described above did not occur when the pH of the colloidal phase was adjusted to 4 (using dilute HCl). At pH ) 4, the carboxylic acid groups in the amino acid would not be ionized and, as a consequence, the attractive electrostatic interaction between the gold particles and ODA in toluene would be negligible. This result clearly highlights the important role played by electrostatic interactions in the phasetransfer protocol described herein. Another important point to note is that a critical concentration of the ODA molecules in toluene was required (ca. 10-4 M) above which the phase transfer of the gold particles occurred (using equal volumes of the aqueous and organic phases). Below this critical concentration, the gold particles were observed to be immobilized at the water-organic solvent interface and transfer into the bulk of the organic phase did not occur. This indicates that a certain minimum hydropho(28) Underwood, S.; Mulvaney, P. Langmuir 1994, 10, 3427.
9782
Langmuir, Vol. 16, No. 25, 2000
Figure 7. Transmission electron micrograph of a film of ODAstabilized valine-capped gold particles formed from a chloroform solution on a holey grid. The scale bar corresponds to 1000 Å.
Figure 8. Thermogravimetric (left axis) and differential thermal analysis data (right axis) obtained from the ODAstabilized valine-capped gold colloidal particles, with the stabilization with ODA effected by transfer into toluene (see text for details).
bicity provided by the ODA coordinated to the gold-bound valine molecules is required to drive the phase transfer. A thin film of the ODA-stabilized valine-capped gold particles formed on a holey TEM grid by solvent evaporation was analyzed by TEM, and the micrograph obtained is shown in Figure 7. The scale bar in the figure corresponds to 1000 Å. It can be seen that the gold particles are fairly well dispersed and there are regions in the superlattice film where a close-packed structure of the gold particles exists. Nearly 150 particles in the TEM micrograph shown in Figure 7 were size analyzed, and a mean particle size of 41 Å and a standard deviation of 7 Å were determined from the analysis. This value is in good agreement with the PSD obtained for the uncapped as-prepared gold colloidal particles using the synthesis protocol described in ref 8a (35 ( 7 Å). This result also indicates that there is no significant size selectivity associated with phase transfer of the colloidal gold particles by coordination with ODA. Figure 8 shows the TGA (left axis)/DTA (right axis) data obtained for ODA-stabilized valine-capped gold powder heated in air. A number of features can be seen in the figure, and we tentatively assign them to the following physical processes. An endothermic process is
Kumar et al.
Figure 9. (A) FTIR spectra in the N-H stretch range obtained from a film of ODA-stabilized valine-capped gold particles on a Si(111) substrate after different heat treatments (see text for details). (B) FTIR spectra in the N-H bend and methylene scissoring range obtained from a film of ODA-stabilized valinecapped gold particles on a Si(111) substrate after different heat treatments (see text for details). The temperatures are indicated next to the respective curves.
observed at ca. 53 °C, and no weight loss is seen at this temperature. This feature is attributed to melting of the ordered regions in the hydrocarbon chains, possibly the interdigitated segments between neighboring particles.29,30 Another endothermic process is seen at 87 °C and is accompanied by a 9% weight loss. This process is attributed to desorption of uncoordinated ODA molecules in the solid. An exothermic process occurs at around 200 °C and is accompanied by a 40% weight loss. We attribute this weight loss to desorption of the valine-ODA complex from the solid. If we assume an area per valine molecule on the gold surface to be 25 Å2 and a linkage of one ODA molecule to each valine molecule, the percentage contribution of the valine-ODA component to the solid can be shown to be ca. 20%. Thus, the weight loss recorded at 200 °C is nearly double the calculated value, indicating that a small percentage of the gold particles may be desorbing at this temperature as well. Two more exothermic processes are observed at 320 and 475 °C and are accompanied by almost complete weight loss in the sample analyzed. It appears that the gold colloids almost completely desorb when the valine-ODA stabilizing layer is removed. To understand better the various temperature-dependent processes occurring in the ODA-stabilized valinecapped gold powder (Figure 8), a parallel FTIR study of a gold film on a Si(111) substrate was performed wherein the spectra were recorded after heating the film at 100 and 200 °C for 1 h. These temperatures correspond to features observed in the TGA/DTA data (Figure 8) discussed above. The FTIR data obtained after different heat treatments (the temperatures are indicated next to the respective curves) in different spectral regions are shown in Figure 9A,B. Figure 9A shows the FTIR spectra from the gold film in the N-H stretching region. Two features at 3330 and 3160 cm-1 were clearly observed in the as-prepared (room-temperature) and 100 °C films and correspond to the N-H stretch vibrations from ODA molecules that are uncoordinated or electrostatically bound to the valine molecules on the gold surface and from the surface-bound valine molecules, respectively.5,8a Such a shift in the N-H stretch vibration mode on (29) Terrill, R. H.; Postelwaithe, T. A.; Chen, C.; Poon, C.; Terzis, A.; Chen, A.; Hutchinson, J. E.; Clark, M. R.; Wignall, G.; Londono, J. D.; Superfine, R.; Falvo, M.; Johnson, C. S., Jr.; Samulski, E. T.; Murray, R. W. J. Am. Chem. Soc. 1995, 117, 12537. (30) Patil, V.; Mayya, K. S.; Pradhan, S. D.; Sastry, M. J. Am. Chem. Soc. 1997, 119, 9281.
Amphoterization of Colloidal Gold Particles
formation of salts with anions such as PtCl6- has been observed in Langmuir-Blodgett films of ODA.31 In earlier FTIR studies on electrostatic complexation of carboxylic acid-derivatized colloidal gold particles with ODA films, we have observed that the N-H stretch frequency is not altered on coordination and this was explained as being a consequence of the weak nature of the electrostatic interaction.8a It is for this reason that we attribute the feature at 3330 cm-1 to both the uncoordinated and valinebound ODA molecules. After heating the film at 100 °C for 1 h, it is observed that the N-H stretch feature from ODA molecules in the film (3330 cm-1) is reduced in intensity whereas the N-H stretch mode from the valine molecules bound to gold is unaltered. This result indicates that at 100 °C there is loss of ODA from the film and that this is likely to be due to uncoordinated ODA molecules in the solid. This clearly lends additional support to the TGA/DTA weight loss/exothermic process observed at 87 °C (Figure 8). After the film is heated at 200 °C for 1 h, the features corresponding to both ODA molecules and surface-bound valine are significantly reduced in intensity. This indicates desorption of both valine and ODA from the film and agrees with the interpretation of the TGA/ DTA feature at 200 °C. Figure 9B shows the FTIR spectra recorded from the gold films after different heat treatments in the range of 1300-1750 cm-1. Two prominent bands are observed at 1468 and 1561 cm-1 in the room-temperature and 100 °C films and correspond to the methylene scissoring8a,25 and N-H bend5 vibrations, respectively. A marginal decrease in intensity of these two modes of vibration is observed on heating to 100 °C and is consistent with the TGA/DTA feature at 100 °C which was interpreted as arising from desorption of uncoordinated ODA molecules. After the film was heated to 200 °C, the intensity of these two bands (31) Bardosova, M.; Tregold, R. H.; Ali-Adib, Z. Langmuir 1995, 11, 1273.
Langmuir, Vol. 16, No. 25, 2000 9783
is reduced considerably, suggesting almost complete loss of ODA molecules from the gold film. Weak resonances at 1470 and 1560 cm-1 remain after this heating cycle and possibly arise from the methyl groups and amine groups in the gold-surface-bound valine molecules, respectively. The FTIR results presented above to a large extent corroborate the physical interpretation of the features observed in the TGA/DTA data shown in Figure 8. To conclude, the amphoterization of colloidal gold particles because of surface coordination of the amino acid valine has been demonstrated. The valine-capped colloidal gold solution is extremely stable at both pH ) 8.5 and 3.5, indicating electrostatic stabilization via ionization of the carboxylic acid and amine functional groups, respectively, at these pH values. The gold particle surface charge reversal at two pH values on either side of the valine isoelectric point has been indirectly validated by electrostatically controlled sequential self-assembly of the gold particles on negatively charged glass substrates. Chemical analysis of the multilayer gold films on glass by XPS provides additional support to the valine-induced amphotericity of the gold particles. The aqueous valine-capped gold particles could be transferred into toluene by electrostatic coordination with octadecylamine molecules. The resulting hydrophobic gold particle powder could be readily dissolved in different organic solvents. Acknowledgment. P.M. and A.K. thank the Council for Scientific and Industrial Research (CSIR), Government of India, for research fellowships. Supporting Information Available: XPS spectra of the Au 4f (Figure 1) and O 1s (Figure 2) core-level spectra from a four-layer valine-capped gold film on a glass substrate. This material is available free of charge via the Internet at http://pubs.acs.org. LA000886K
