«Kinetic Investigations of Thiolate Protected Gold Nanoparticles: Protein Interactions, Electron Transfer, and Precursor Formation By Brian N. Turner ...»
Improved Synthesis of Alkanethiol Nanoparticles Reduces Current Drift Cleaner preparations of gold nanoparticles were anticipated to solve the problem of negative current drift, presumably due to self-assembly of excess ligand on the UME surface. The improved single phase synthesis discussed below allowed for cleaner nanoparticles without the use of excessive washes with a battery of solvents.
Tetraoctylammonium bromide has been known to stick to the alkanethiolate monolayer in these preparations, as stated by David Schiffrin, one of the authors of the original Brust synthesis paper.155 TOABr, in turn, presumably holds on to excess thiols present in solution, making purification difficult. To get around this problem, Schiffrin used a lengthy soxhlet extraction which reduced the overall yield.155 A better way to attain cleaner alkanethiolate nanoparticle preparations that was found was to eliminate the phase transfer reagent completely. Elimination of the phase transfer reagent was accomplished by substituting lithium borohydride for sodium borohydride, allowing for as solvent system of purely anhydrous tetrahydrofuran, instead of a two phase system. Consequently, nanoparticles which were easier to clean post-place exchange were produced. Upon further review of the literature, a very similar method to the one employed had already been published by Rowe and coworkers. 7 To illustrate how much easier to purify these preparations are, 1H NMRs of two samples of AuDodecanethiol nanoparticles (from the two different preparatory methods) prepared with the same stoichiometric amounts of reactants and purified with identical methods are displayed in Figure 41.
Figure 41: Comparison of 1H NMR spectra of dodecanethiol gold nanoparticles synthesized by the Brust method (blue), and by the method of Rowe and coworkers (red) after identical purification methods. The red spectrum is visibly cleaner than the blue spectrum judging by the peak width and smoothness.
Both nanoparticle preparations were purified by the same steps of centrifuging out of ethanol once, and then sonicating on glass three times with acetonitrile. Broader and more well-defined peak shapes for the nanoparticles prepared by the method of Rowe and coworkers versus those of the Brust method indicate lower remaining excess thiol, and the lack of tetraoctylammonium. The place exchanged products of these preparations were remarkably easier to clean, very amenable to centrifugation, and if additional dry sonication steps were required, they were generally fewer.
SECM Quantitative Observation of AuOctanethiol Electron Transfer Rates Using AuOctanethiol nanoparticles produced with the improved method discussed above, it was possible to generate positive feedback approach curves that could be fit to equation 4-5 with some success for both wire functionalized and unfunctionalized samples. Three samples were studied in the same manner as described above (generating approach curves, switching solutions, generating approach curves, switching back to the original solution, repeat ad infinitum) at three different voltages: 200, 400, and 600 mV. The three different samples were Au228Octanethiol92 (unfunctionalized), Au228Octanethiol90PEPEPS2 (from exchange with disulfide, PEPS2), and Au228Octanethiol89PEPEPS2.6 (from exchange with free thiol, PEPSH). 400 and 600 mV experiments gave comparable results, but the 200 mV experiments were markedly different. The data for the Au228Octanethiol92 was baseline corrected, and a numerically identical correction was applied to the other samples in order that equation 4-5 could be fit. The correction eliminated negative current drift due to moisture, impurities, or electrode irregularities; therefore, the values for k f are relative. The best corrected approach curves generated for wired and unwired gold octanethiol nanoparticles, all at a tip bias of 400mV, are displayed in Figure 42.
Figure 42: SECM approach curve through Au228Octanethiol90PEPEPS2, Au228Octanethiol90PEPEPS2.6, and Au228Octanethiol92. Conditions: 25 µm Pt tip at 0.4 V, 2 mm Pt substrate electrode at 0 V (vs. Ag/Ag+), 1 mL aliquot of a solution of 5 mg nanoparticle in 20 mL dichloromethane with 0.1M TBAPF6. The black curve represents the expected result of mass-transfer-only limited system, in accord with equation 4-6, while the purple curve is the kinetic and mass transfer limited theory with k f = 0.0040 cm/s. Bot wire functionalized particles (red and green) have faster electron transfer kinetics than the unfunctionalized particles (blue).
It would be expected that the place exchange percentage from the disulfide would be lower than that of the thiol, but that given a long reaction time, the values of disulfide place exchange would approach that of the thiol place exchange.
Using equation 4-5, it was possible to determine values for kf, assuming a diffusion coefficient calculated using the Stokes-Einstein equation, DSE = 2.52 x 10-6 cm2/s. The value of DSE is not dissimilar to values obtained for the diffusion coefficients of other gold-alkanethiolate nanoparticles of a similar core diameter by the Taylor dispersion method.156 The values for kf calculated from the curve are gathered in Table 9.
Table 9: Approximate electron transfer kinetic constants determined for Au228Octanethiol92-nPEPEPSn.
0 1.2 2 3.2 2.6 4.0 In all of these experiments, it should be considered that the mixed kinetic and mass transfer limited model described by equation 4-5 might not entirely describe the system, due to the applied baseline corrections. Additionally, it should be noted that the difference between kf values obtained for AuOctanethiolate nanoparticles in these experiments (1.2 x 10-3 cm/s) and AuOctanethiol nanoparticles from the Peterson study (2.4 x 10-2 cm/s). A number of possibilities were considered. Initially, experiments were hampered by a constant drift, and that was, correctly attributed to a dirty nanoparticle preparation. Having identified and dealt with this problem, further deviations and their possible causes were considered. Absorption of water from air by the dichloromethane is one possible source of deviation. It is thought that water can act as an electron sink, stealing electrons from reduced material as it diffuses back to the tip, causing a drop in feedback current. Another possibility is that the shape of the positive feedback curve is slightly different than expected due to a hemispherical electrode surface or a recessed electrode surface. The equations used in calculations of electron transfer rate and current are derived assuming a flat disk geometry. Differences in nanoparticle structure are also possible, but difficult to probe or comment on. Possible factors resulting from this are ligand packing, organization, and nanoparticle geometries with more or less edge and vertex sites resulting from slightly different synthetic conditions. Structural differences may have caused the electrons to find slightly different paths to the nanoparticle from a change in distance between the nanoparticle core and the electrode, from a change in the presentation of the ligand orbitals to the electrode, and/or a change in the nature of the double layer.
In conclusion, the data supports the hypothesis that molecular wire molecules behave as conductors on gold nanoparticles in soltuion, consequently increasing electron transfer kinetics. These functionalized materials can be obtained by straightforward place exchange reactions with the free thiol or disulfide of PEPEPSAc. It seems that a small increase in surface concentration of the wire molecule resulted in a small increase in electron transfer. Much work remains on this topic to characterize the effects of surface concentration, wire size, and wire functional group chemistry. Additionally, MALDIIM-MS should be more useful than 1H NMR for quantifying relative amounts of surface ligands due to less interference.
The wired nanoparticle project is currently ongoing. Dave Crisostomo is currently investigating the effects of alkanethiol ligand length, nanoparticle size and dispersity, and PEPEPS surface concentration. Additionally, other electrochemical properties of wired versus unwired alkanethiol nanoparticles are being probed. From there, it would be valuable to study the variation of molecular wire length by comparison with PEPEPSH to the molecule PEPSH (shorter by one phenylethynyl unit). Due to a lack of success with the standard method to produce PEPSH,157 which was complicated by the lack of commercial availability of the starting material, alternative synthetic methods for PEPSH are included in Appendix E.
I would like to acknowledge Jeremy Wilburn for training me on the use of the SECM and in the fabrication and maintenance of UMEs. I would like to thank Gongping Chen and Fred Hijazi for their assistance with some of the earlier gold dodecanethiol data collected in this study. I would like to further thank Gongping Chen for her assistance with electrode polishing and analysis. Finally, I would especially like to thank Dave Crisostomo for continuing with this work.
Much interest as has been paid to the variation of synthetic parameters and the consequences on nanoparticle formation;63, 158-165 however, the approaches have not adequately addressed fundamental issues about the formation of intermediate species on the way to the final nanomaterial. The experiments that follow attempted to gain insights into the make-up of the precursor material and how it is formed. Short of proposing a full mechanistic understanding of all aspects of gold nanoparticle precursor formation, this work addressed a number of important questions along this path. This chapter focuses primarily on the precursor formation during the synthesis of tiopronin protected gold nanoparticles as first synthesized by Templeton and coworkers122, but slightly modified to better interrogate the system.
The modified Brust method employed by Templeton and coworkers had long been
theorized to proceed by Scheme 5-1:
Scheme 5-1: General reaction scheme for the Brust method of nanoparticle synthesis.
The reactions are unbalanced and Cl- and H+ are not shown as products. Aup(RS)q represents the final nanoparticle with a core of p Au atoms and a shell of q RS molecules.
On the basis of a mass spectrometry study by Geis and coworkers27 and a later powder diffraction study by Simpson and coworkers (43c),166 the tiopronin system is thought to go through the stable Au n(RS)m intermediate in Scheme 5-1, a tetrameric ring structure (Figure 43c). Later, in a landmark x-ray crystallographic study, Kornberg and coworkers (Figure 43a)43 discovered Au-SR-Au staple motifs45 on the surface of gold nanoparticles protected by p-mercaptobenzoic acid (PMBA) (Figure 43b). Further x-ray crystal structures of other monodisperse gold nanoparticles prepared by the Brust method or some variation on it displayed similar capping structures on the surface. Gold thiolate capping structures had been predicted earlier, computationally, by Häkkinen and coworkers.42 Figure 43: Nanoparticle and ring complex structures (a) PMBA nanoparticle crystal structure proposed by Jadzinsky, et. al., 43 from Walter, et. al.167 Copyright 2008, National Academy of Sciences, reused with permission. (b) PMBA staple motifs, from Whetten and Price,45 Copyright 2007, American Association for the Advancement of Science, reused with permission, and (c) Au4Tiopronin4 cyclic structure (R group of tiopronin now shown), Amsterdam density functional theory prediction from Simpson, et. al.166 Copyright 2010, American Chemical Society, reused with permission.
Notably, the precursor material formed before the final borohydride reduction, whether a cyclic or chain polymer structure, bears structural similarity to the capping structures in the reported crystal structures. Harkness and coworkers demonstrated that different ligands prefer different cyclic stoichiometries (some with equal but greater or lesser numbers of gold and thiol per cycle), but that tiopronin prefers the cyclic tetramer (four gold atoms and four thiolate molecules) completely.168 Furthermore, these precursor structures can be liberated from gold nanoparticles and measured using MALDI-MS. It is clear that gold nanoparticles are not always, and possibly never can be, a solid gold core surrounded by a ligand shell. Instead, it appears that there is a solid grand core and a loose ligand-gold shell which structurally resembles the precursor material. Better understanding of the precursor material and its formation, therefore, is important to the continued progress in the synthesis of Brust nanoparticles.
The ligand exchange and reduction of Au(III) complexes has been the subject of a small number of prior studies; consequently, discussion of some of the more prominent results is prudent. Elding has studied the exchange and reduction of Au(III) halides. Notably, as a possible analogue to this study, they fully characterized the ligand exchange kinetics and subsequent reduction of haloamminegold(III) with thiocyanate, 169 with dimethylsulfide,170 and with thione containing ribonucleic acids. 171 The general scheme that they deduced for the reaction of haloamminoaurate with thiocyanate, displayed in Scheme 5-2, seemed to proceed in a similar fashion to what was visually observed for the reaction of gold and tiopronin.
Scheme 5-2: Reaction pathways for the exchange and reduction of haloamminegold(III) by thiocyanate, as proposed by Elmroth and coworkers. 169 Copyright 1989, American Chemical Society, reused with permission.
None of the literature reviewed172-174 reports kinetics and mechanism for the reaction of a terminal organic thiol or thiolate with Au(III)Cl4 -, but the following Scheme 5-3 for this reaction was generated with the cited studies for other sulfur containing molecules in mind.
Scheme 5-3: Plausible reaction mechanism for the reaction of Au(III)Cl4 - with tiopronin featuring competitive ligand exchange and reduction.