5492
Langmuir 2005, 21, 5492-5500
Ligand Heterogeneity on Monolayer-Protected Gold Clusters Yang Song, Amanda S. Harper,† and Royce W. Murray* Kenan Laboratories of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290 Received February 8, 2005. In Final Form: March 30, 2005 This paper describes the effects of oxidative electronic charging of the Au cores of the monolayerprotected clusters (MPCs), Au140(S(CH2)5CH3)53 and Au38(SCH2CH2Ph)24, on nuclear magnetic resonance (NMR) spectra of their monolayer ligand shells. Previously unresolved fine structure in the 13C NMR hexanethiolate methyl and C5 methylene resonances is seen in spectra of solutions of monodisperse Au140(S(CH2)5CH3)53 MPCs, reflecting magnetically inequivalent ligand sites. Incremented increases in positive cluster core charge, effected by electrochemical charging, cause the spectral fine structure of the methyl resonance to coalesce, becoming a single peak at the Au1403+ charge state. The spectral changes are reversible; charging back to the original core charge state regenerates the methyl 13C resonance fine structure. Adding an equimolar quantity of a Au(I) thiolate complex, AuI[SCH2(C6H4)C(CH3)3], to an uncharged Au140(S(CH2)5CH3)53 MPC solution in d2-methylene chloride causes partial spectral coalescence. 13C NMR spectra of Au (SCH CH Ph) MPCs exhibit roughly comparable spectral changes upon positive 38 2 2 24 core charging to the ‘0’, ‘+1’, and ‘+2’ states. The NMR results indicate that exchange between magnetically inequivalent sites occurs at rates of 100 to 400 s-1, a rate believed to be too fast to be accountable by actual exchanges of ligands between different sites on the Au core. We also describe changes in core electronic spectra of Au140(S(CH2)5CH3)53 induced by positive charging, measured using spectroelectrochemistry.
Introduction A special chemical property of gold monolayer-protected clusters1 (MPCs), particularly those that have only a few hundreds of atoms (or less) in the cluster core, is the quantization2 of the charge states of their cores. MPCs of small core size (99%), tetrabutylammonium hexafluorophosphate (Bu4NPF6, Aldrich, >99%), tetra-n-octylammonium bromide (Oct4NBr, 98%), hexanethiol (HS(CH2)5CH3), Aldrich, >99%), 2-phenylethane thiol (PhC2SH, 99%), sodium borohydride (NaBH4, 99%), dichloromethane (CH2Cl2, Fisher, 99.9%), acetonitrile (CH3CN, Fisher, 99.9%), toluene (Fisher), cerium sulfate (Ce(SO4)2, Aldrich, >95%), silver nitrate (AgNO3, Aldrich, 99.9%), d2-methylene chloride, d-chloroform, d8-toluene, and d14-hexane (NMR solvents, Cambridge Isotope Laboratories, Inc.) were all used as received. Hydrogen tetrachloroaurate trihydrate (from 99.999% pure gold) was prepared using a literature procedure17 and stored in a freezer at -20° C. Low-conductivity water was obtained from a Millipore Nanopure water purification system. Synthesis of Au140 Hexanethiolate MPCs. Hexanethiolatestabilized gold clusters (Au140C6 MPCs) were synthesized by a modified Brust procedure.11 Briefly, hexanethiol (∼3.5 mL) and AuCl4- (∼3.1 g) were combined in a 3:1 molar ratio in 200 mL of toluene and a 10-fold molar excess of reductant (∼3.8 g of NaBH4 in water) was added at 0 °C with vigorous stirring. The reduction was allowed to proceed for 30 min, after which the water layer was discarded and the toluene removed to a state of a moist black sludge using rotary evaporation. The product was extracted overnight by adding ca. 200 mL of ethanol to the round-bottom flask; the “ethanol-soluble fraction,” (EtOH-soluble Au140C6 MPCs) was dried and washed with acetonitrile. According to previous studies,3c EtOH-soluble Au140C6 MPCs show improved monodispersity; the main constituent is Au140(S(CH2)5CH3)53 with an average core mass of 29 kDa.18 Synthesis of Au38 Phenylethanethiolate MPCs. Monodisperse Au38(PhC2S)24 MPCs can be synthesized3b by the Brust method and isolated by solvent extraction steps. Briefly, 3.10 g of AuCl4- and 3.43 mL of phenylethanethiol (PhC2SH, Aldrich, >98%) were stirred in toluene until the solution became colorless, whereupon it was cooled to 0 °C in an ice water bath and 3.80 g of NaBH4 dissolved in 60 mL ice cold water added to the Au(I) thiolate solution with vigorous stirring. The immediately darkened solution was vigorously stirred in an ice bath for 24 h, after which the water layer was discarded and the organic layer was washed thrice with 50 mL of Nanopure water and then rotary evaporated to a black sludge. The sludge was extracted with 200 mL of acetonitrile overnight, the extract was rotary evaporated, ethanol was added, and the suspended solid collected on a fine-porosity fritted glass filter to yield, after being washed generously with ethanol, 0.20-0.25 g of a brown powder of composition Au38(PhC2S)24.3b Electrochemistry. Electrochemical measurements were performed with a Bioanalytical Systems (BAS-100B) electrochemical analyzer. Differential pulse voltammetry (DPV) and rest potentials of MPC solutions were measured in 0.1 M Bu4NClO4/CH2Cl2 solutions in a single-compartment cell containing a 1.6-mm-diam Pt disk working electrode, a Pt flag counter electrode, and a Ag/Ag+ (Ag/1 mM AgNO3/0.1 M Bu4NPF6/ CH3CN) reference electrode.
5494
Langmuir, Vol. 21, No. 12, 2005
Song et al. slide walls, side-by-side Pt mesh working and Pt wire counter electrodes, and an Ag wire quasi-reference. One end of the thin cell was fitted with a Teflon valve for solution-filling under Ar and contacts for the three electrodes. By masking, only the solution around the working electrode was optically monitored. The reference spectrum was of the dry, air-filled cell. Preliminary cyclic voltammetry (5 mV/s) was done to identify appropriate electrolysis potentials to produce MPC0, MPC1+, MPC2+, or MPC3+ solutions. A typical electrolysis required 5-10 min, according to the current-time response, whereupon an absorbance spectrum was taken. Experiments were also done by monitoring absorbance changes at selected wavelengths during a slow potential scan. Electrochemical control was with a locally built potentiostat21 interfaced to a PC via a Keithley DAS-HRES 16-Bit A/D board and using locally written software. Spectra were taken using UNICAM UV4 and Shimadzu UV-1601 UV-visible spectrophotometers.
Results and Discussion
Figure 1. (A) Differential pulse voltammetry (DPV) of 0.02 mM EtOH-soluble Au140C6 MPCs and (B) of 0.02 mM Au38PhC2 MPCs, both in degassed CH2Cl2/50 mM Bu4NClO4 and at a 0.4-mm-diam Pt working electrode. The DPV pulse height was 50 mV. Bulk electrolysis of the Au140C6 MPC solution (i.e., to charge the MPC cores) was carried out in a fritted three-component cell with compartments containing Pt mesh working and Ag/Ag+ reference electrodes and 0.1 mM MPC/50 mM Bu4NClO4 in CH2Cl2, 50 mM Bu4NClO4/CH2Cl2 electrolyte solution only, and a Pt mesh auxiliary electrode and 50 mM Bu4NClO4/CH2Cl2 electrolyte solution, respectively. Air was not excluded. Figure 1 shows differential pulse voltammetry of Au140 and Au38 MPCs. The oneelectron current peaks and valleys are labeled according to the charge states of the MPC cores in the solution in equilibrium with the indicator electrode at the corresponding potential, which follows Nernstian principles, as shown previously.2c,4 By electrolyzing cluster solutions to equilibrium potentials (tracked using indicator electrode “rest” potentials) corresponding to each valley, MPC0, MPC1+, MPC2+, and MPC3+ solutions were prepared. (MPCs charged more positively than MPC3+ were insufficiently stable to satisfactorily survive the lengthy procedure.19) After the electrolysis, the CH2Cl2 solvent was removed by rotary evaporation and the supporting electrolyte (Bu4NClO4) removed (confirmed by 1H NMR) by repeated washing with 20 mL portions of acetonitrile. Au38PhC2 MPCs were charged via a chemical process reported earlier.5 Ten milliliters of an aqueous solution (5 mM in Ce(SO4)2 and 0.1 M in NaClO4) was stirred rapidly in a scintillation vial with 6 mL of a CH2Cl2 solution (0.1 mM in Au38PhC2 MPC and 50 mM in Bu4NClO4 electrolyte). After various periods of stirring time, the mixture was allowed to phase-separate and the electrochemical rest potential of the organic layer with MPCs was measured on a clean Pt electrode vs. Ag/Ag+ reference electrode. This process was continued until a rest potential corresponding to clusters of the desired charge state was reached. The supporting electrolyte (Bu4NClO4) was not removed since its resonances do not interfere with the proton-decoupled, aromatic 13C NMR signals of the phenylethanethiolate ligands of charged Au38PhC2 MPCs. NMR Spectra. MPC spectra (both 1H and proton-decoupled 13C) in various deuterated solvent solutions were collected using a Bruker AC500 spectrometer. Each sample of Au140 MPC contained ∼20 mg cluster/mL (equivalent to ca. 0.04 M thiolate ligand). Samples were filtered through a Nalgene (25 mm, 0.45 µm) syringe filter to eliminate any possibility of inhomogeneous line broadening caused by insoluble solids. A 5 s relaxation delay time and 64 scans were taken for each 1H NMR spectrum. Proton-decoupled 13C NMR spectra were collected over a 15-h period. Spectroelectrochemistry. Thin-layer spectroelectrochemistry20 was done in a 1-mm path-length cell with quartz
Electrochemical Properties and Preparation of MPC Charge States. Differential pulse voltammetry (DPV) of Au140C6 MPCs (Figure 1A), in the interval between ca. +0.9 and -0.5V, shows a series of six quantized double-layer charging peaks2 that are approximately evenly spaced about the potential of zero charge (EPZC). The even spacing means that the MPC capacitance, CMPC, remains relatively constant near EPZC. An ∼70% monodispersity of the MPC sample is estimated11 on the basis of relative peak-to-background-valley currents near EPZC. Equilibrium potentials measured at indicator electrodes in MPC solutions reflect their core charge state compositions.4 The valley minima in Figure 1A represent the equilibrium potentials of solutions of the indicated, integral Au140C6 charge states, while the current peaks lie at the “formal potentials” and represent equimolar mixtures of the indicated charge state couples. A solution of as-prepared Au140C6 MPCs (“MPCas-prep”) exhibits a rest potential that is slightly negative of EPZC, reflecting a mixture of MPC0 and MPC1- charge states probably from residual reductive charge from the synthetic preparation.2c MPCas-prep solutions were electrolyzed to potentials (according to the valleys of Figure 1A) of MPC0, MPC1+, MPC2+, and MPC3+ charge states. The charged MPCs in these solutions, when isolated, purified of supporting electrolyte, stored dry, and redissolved for NMR experiments, recover4-6 (to within (10 mV typically) the predried rest potential in each case, demonstrating stability of the charge states. The MPC3+ solutions were similarly electrolytically returned to the original MPCas-prep state. The voltammetric peaks for Au38PhC2 MPCs (Figure 1B) are not evenly spaced3,14 because this nanoparticle displays a molecule-like energy gap (1.69 V) between the first oxidation and the first reduction step. Electron-deficient Au38 MPCs are not as robust as are Au140 MPCs, and care must be taken to minimize decomposition.19 In the present study, the Au38PhC2 MPCs were charged by chemical5 oxidation because it could be accomplished more quickly, and NMR measurements were initiated immediately on the charged solutions, without electrolyte removal (see Experimental Section). Core Charge Effects on NMR Spectra of Au140 MPCs. NMR spectra of the organic monolayers of gold MPCs display, compared to those of unbound ligands, 1H,18,22-25 13C, 26-28 and 31P 28 resonances that generally (24) Kohlmann, O.; Steinmetz, W. E.; Mao, X.; Wuelfing, W. P.; Templeton, A. C.; Murray, R. W.; Johnson, C. S. J. Phys. Chem. B 2001, 105, 8801. (25) . Schaaff, T. G.; Shafigullin, M. N.; Khoury, J. T.; Vezmar, I.; Whetten, R. L. J. Phys. Chem. B. 2001, 105, 8785.
Ligand Heterogeneity on Gold Clusters
Langmuir, Vol. 21, No. 12, 2005 5495
Figure 2. 1H (upper) and proton-decoupled 13C (lower left) NMR spectra of EtOH-soluble, as-prepared Au140C6 MPCs in CD2Cl2 and (lower right) expanded 13C NMR signal of the methyl group (-CH3). 13C peak assignments according to ref 25.
are substantially line-broadened. There has been sustained interest in understanding the origin(s) of the broadening, which for alkanethiolate and related ligands is lowest for terminal nuclei and highest for nuclei closest to the gold core.18,27 Badia et al.26b present evidence from 13C solid-state NMR that a diversity of chemical shifts, presumably reflecting a heterogeneity of Au-S surface binding sites, is an important line-broadening mechanism. This analysis was supported by Schaaff et al.25 in solution 1H and 13C NMR on purified Au 145C6 MPCs and by detailed NMR measurements by Kohlmann et al.24 on water-soluble tiopronin (N-2-mercaptopropionylglycine) MPCs. Mentioned sources of the responsible surface heterogeneities have been different nanocrystalline faces,25 different core shapes and sizes, and even defects.24 Another broadening source is slow rotational diffusion of the bulky MPCs (analogous to effects seen for large proteins), which scales with nanoparticle size.18,27 The NMR results reported here are obtained using a higher-field (500 MHz) instrument than used previously in investigations of alkanethiolate-protected MPCs (although higher-field measurement have been done24 for tiopronin-protected Au MPCs) and, additionally, like Schaaff et al.25 with highly purified nanoparticle samples. The 1H NMR literature for gold alkanethiolate MPCs reports18,22-25 two broadened resonances at ∼0.8-0.9 and 1.2-1.5 ppm, corresponding to the methyl and methylene (>βH) groups, respectively. In the 1H NMR spectrum in Figure 2 (upper) these peaks now show slight fine structure. More useful are 13C NMR observations, shown in Figure 2 (lower). The 13C NMR chemical shifts seen are similar (26) (a) Badia, A.; Cuccia, L.; Morin, D. F.; Lennox, R. B. J. A. Chem. Soc. 1997, 119, 2682. (b) Badia, A.; Demers, L.; Dickinson, L.; Morin, F. G.; Lennox, R. B. J. Am. Chem. Soc. 1997, 119, 11104. (c) Badia, A.; Gao, W.; Singh, S.; Demers, L.; Cuccia, L.; Ewven, L. Langmuir 1996, 12, 1262. (27) Terrill, R. H.; Timothy, A. P.; Chen, C.; Poon C.; Terzis, A.; Chen, A.; Hutchison, J. E.; Clark, M. R.; Wignall, G.; Londono, J. D.; Superfine, R.; Falvo, M., Johnson; C. S.; Samulski, E. T.; Murray, R. W. J. Am. Chem. Soc. 1995, 117, 12537. (28) Petroski, J.; Chou, M. H.; Creutz, C. Inorg. Chem. 2004, 43, 1597.
Figure 3. Proton-decoupled 13C NMR spectra of the methyl group (-CH3) of EtOH-soluble Au140C6 MPCs in four different NMR solvents, CDCl3, CD2Cl2, d8-toluene, and d14-hexane, as indicated. Spectra were accumulated for 15 h to improve the S/N ratio.
to those in the previous25-28 reports on alkanethiolateprotected Au MPCs and, like them, display increasing broadening (relative to free thiols) for carbon positions closer to the S-Au core bond. On an expanded scale, (Figure 2, lower right) the methyl carbon peak displays a previously25 unresolved, complex pattern of peaks. A fine structure similarly appears in the C5 resonance (see Supporting Information). The results in Figure 2 (lower) were repeatable, being observed for several synthesized and purified batches of Au140C6 MPCs, both in details of the fine structure and chemical shifts. The fine structure
5496
Langmuir, Vol. 21, No. 12, 2005
Song et al.
Figure 4. Proton-decoupled 13C NMR spectra (measured in CD2Cl2) of the methyl of EtOH-soluble Au140C6 MPCs at different charge states, as indicated. Spectra were accumulated for 15 h to improve the S/N ratio.
Figure 5. Proton-decoupled 13C NMR signal of the methyl group on EtOH-soluble Au140C6 MPCs before and after adding an equimolar concentration of AuI[SCH2(C6H4)C(CH3)3].
is not due to spin-splitting since the spectra are protondecoupled. The details vary with the NMR solvent (CD2Cl2, CDCl3, d8-toluene, d14-hexane, Figure 3), although the overall appearance is not widely changed. Figure 3 shows that the observed fine structure is not peculiar to a given solvent and also implies that solvation of the monolayer ligands plays a role in the evident magnetic inequivalency of methyl 13C chemical shifts seen in Figures 2 and 3. The improved spectral resolution in Figures 2 and 3 can be attributed to a combination of the higherfield NMR observation and a good sample monodispersity, but further underlying spectral details may remain to be discovered in these, as well as in previously observed, NMR resonances for MPCs. What is the source of the fine structure? (No definitive answer to this question will be offered, although an appealing possibility will be eliminated.) A useful insight, that the fine structure can be coalesced, was provided by 13 C NMR spectra of solutions of MPCs with variously charged cores. Figure 4 shows spectra taken in a serial charging experiment. Figure 4 (upper left) shows the methyl 13C NMR spectrum of as-prepared MPCs. Clockwise in the figure, the subsequently acquired spectra are of the MPC0, MPC1+, MPC2+, and MPC3+ charge states, returning to the original, slightly negative rest potential of the MPCas-prep.
Figure 4 shows that charging the MPCas-prep sample to MPC0 produces a successive narrowing of the methyl fine structure pattern until, at the MPC3+ charge state, the original complex pattern of different chemical shifts for the methyl resonance has coalesced into a single, slightly broadened resonance. Importantly, the changes in Figure 4 are reversible; when the MPC3+ solution is electrolytically charged back to the rest potential of the MPCas-prep sample, the original spectral fine structure reappears. The 13C NMR resonances of the C5 carbon show fine structure similar to that in Figure 4 and similar effects of core charge (see Supporting Information). The C4 and C3 carbon spectra also display fine structuresbut not as distinctlys and narrowing as the MPC core was charged positively (results not shown). Figure 5 shows a further experiment in which an equimolar concentration of the soluble AuI[SCH2(C6H4)C(CH3)3] complex29 was added to an MPCas-prep solution. The partial coalescence of the 13C NMR methyl resonance resembles that caused by +1 core charging. The coalescence of methyl 13C NMR resonances seen in Figure 4 mimics the well-known phenomenon of exchange (29) A Au(I) hexanethiolate (Au(I)-SC6) complex would have been a more ideal choice for this experiment. This material however forms a poorly soluble polymer after synthesis, purification, and drying.
Ligand Heterogeneity on Gold Clusters
Figure 6. 1H and proton-decoupled (lower right).
13C
Langmuir, Vol. 21, No. 12, 2005 5497
NMR spectra of Au38PhC2 MPCs in CD2Cl2 and
between magnetically inequivalent nuclei, which, if sufficiently fast on the NMR time scale, averages the chemical shift values so that a single resonance peak is observed.30,31 The impetus to increasing the apparent exchange rate in Figure 4 is obviously the positive core charging. An orderof-magnitude estimate of the rate constant, kC, of the exchange between magnetically inequivalent methyl carbons can be obtained from a simple, two-state expression30 that relates rate to the chemical shift difference ∆υ (in Hz) between the nonexchanging states: kC ) 2.2 ∆υ. Using, on one hand, the chemical shift differences between the peripheral peaks, and on the other, those between the central pair of peaks in the MPCas-prep spectrum in Figure 4 yields estimates of kC from 100 to 400 s-1. While the above is a very rough rate constant estimate, it is useful to compare it to other processes which conceivably could dynamically change the ligands in an MPC monolayer. If the differences in ligand chemical shifts are due to different binding sites, then exchanges of ligands between those sites on a time scale comparable to the NMR time scale might average out those differences. One attractive way that such exchange-between-sites could occur is through rapid on-off (dissociative) release of thiolate ligands or of Au(I) thiolate complexes. However, existing6,15,16 data on exchange kinetics make this pathway unlikely; Au140C6 MPC exchanges6 its hexanethiolate ligands with HS(CH2)6OH in a second-order process whose rate constant increases with positive core charge (7 × 10-3 M-1 s-1 for MPCas-prep to 14 × 10-3 M-1 s-1 for MPC3+). Au140C2Ph, where C2Ph is phenylethylthiolate, similarly ligand-exchanges15 with p-HS-Ph-X in an associative (30) Friebolin, H.; Transl. by Becconsall, J. Basic One- and twoDimensional NMR Spectroscopy, 3rd ed.; Wiley-VCH: Weinheim, 1998; Chapter 11. (31) It is common in NMR studies of chemical exchange to vary the sample temperature; in principle, lowered temperature may slow the exchange process sufficiently to prevent peak coalescence. Loweredtemperature 13C NMR measurements were carried out on solutions of MPCas-prep and MPC3+. Rather than sharpening the resonances, they were broadened, giving only featureless, similarly broadened peaks. Apparently, viscosity-related, slowed rotational diffusion began to dominate.
13C
NMR signal of the aromatic region
second-order reaction whose rate increases with more electron-withdrawing X (4 × 10-3 M-1 s-1 for X ) OH to 14 × 10-3 M-1 s-1 for X ) NO2). The first-order rate constants for these exchange reactions15 were in the 10-4-10-3 s-1 range, at ligand concentrations similar to those employed here, and are thus 105-106-fold slower than the above coalescence rate constant. The above rate determinations refer to the initial, faster stage of the exchange process, which presumably occurs on core vertexes and/or edge sites. For example, in a typical15 Au140 ligand exchange, 10 min might be required for the first 10% of ligands on the MPC core to be replaced. (A dissociative pathway for exchange of these ligands would be first order in MPC, and indeed, in a few circumstances, the kinetic results suggested a minor, slower, first-order pathway6,15,16 (k ≈ 10-4-10-5 s-1) that was promoted by the addition of a Au(I) complex.) After the initial, faster ligand exchange, the reaction slows down considerably and ca. 50% of the ligands (presumably on terracelike sites) resist exchange over a period of days. The latter result shows definitely that ligand migration over an MPC surface even as small as Au140 is extremely slow. Thus, ligand surface migration can also be ruled out as a source of the NMR coalescence, at least any migration involving terrace sites. Thus, while ligand-exchange kinetics roughly parallel the spectral coalescence in the Figure 4 and 5 NMR results, being facilitated by positive core charge and Au(I) complexes, a quantitative comparison of data on ligand exchanges show that they occur on time scales that are orders-of-magnitude slower than the apparent NMR exchange process. This comparison makes clear that the NMR exchange process cannot involve reactions in which thiolate ligands are exchanged with unbound ones, or between MPCs, or migrate between magnetically inequivalent sites on a given Au core. Chemical-shift heterogeneity that results from the MPC sample being a mixture of slightly different core sizes or shapes would also seem to be ruled out as a source of the methyl exchange processsexcept for the possibility of rapid changes
5498
Langmuir, Vol. 21, No. 12, 2005
Song et al.
Figure 7. DPV voltammetry (top) of Au38PhC2 MPCs in CD2Cl2. The solution was chemically charged to the rest potentials shown by the arrows, corresponding to solutions of MPC0, MPC1+, and MPC2+ (from right to left). At the bottom are the corresponding 13C NMR spectra in the aromatic region.
between different core shapes. As said in the Introduction, the accelerating effects of oxidizing conditions on ligand exchange must be attributed to electron deficient cores, not to the intervention of dissociative ligand exchange pathways. The above arguments eliminate an initially appealing scenario for the exchange of magnetically inequivalent ligands, notably their physical exchange from one site to another. This presents a perplexing issue: how can ligands change their magnetic environment without dissociating from the core? We envision two possibilities, involving (a) chain solvation and/or orientation fluctuations and (b) intracore electron exchanges. Both are hard to evaluate. The first possibility (a) might involve a diversity of ligand folding and/or solvated structures within the MPC’s monolayer. The diversity of differently folded/solvated/ etc. structures might, in an underlying way, reflect different kinds of binding locations on the Au nanocrystalline core. Acceleration of the rates of interconversion of such structures with increasing core charge (possibly with participation by the ClO4- counterionsalthough we have no evidence for association except32 at high electrolyte concentrations) could yield the Figure 4 results and also the effects of solvents in Figure 3. The second possibility (b) speculates that the core electron deficiency in the MPC+n forms is localized on different core surface sites. That is, to make MPC1+, an electron is lost from (for example) a vertex-type core site; the electron deficiency is from that type site, as opposed to being smeared over the entire Au-S interface. MPC defect sites (vertexes and edges) are chemically different from terracelike, both in the population of ligands per surface atom33 and electron densities there;34,35 such surface heterogeneity is a plau(32) Guo, R.; Georganopoulou, D.; Feldberg, S. W.; Donkers, R.; Murray, R. W. Anal. Chem. 2005, 77, 2662.
sible source of the dispersity in 13C NMR chemical shifts, as speculated before.25,26 If such localization of electron loss exists, there could ensue an averaging effect from the rapid exchange of the electron deficiency between similar favored localized sites (i.e., such as the different vertex sites on the core). Intramolecular electron exchange is well-known in chemistry as a site-averaging mechanism; the electron exchange between the three bipyridine rings36 in [Ru(bpy)2(bpy-)]1+ is an example. A problem with (b) is that such electron exchange should plausibly be much faster than the above 100-400 s-1 exchange rate. Having no hard evidence for either of these suggestions, we must in the end say that the source of spectral averaging in Figures 4 and 5 remains unknown, although exchange of ligands has been shown to be an unlikely source. Core Charge Effects on NMR Spectra of Au38 MPCs. The 1H NMR and 13C NMR spectra of Au38(PhC2S)24 MPCs in Figure 6 are similar to previously reported spectra.4b Line broadening for these MPCs is much less severe than with the larger Au140C6 MPCs, and the resonances are nearly as sharp as those of the free ligand. There are noticeable effects4b on chemical shifts, both for proton and 13C resonances, as compared to the free ligand (not shown here). Further, in 13C NMR spectra, the relatively unbroadened signals for the four aromatic carbons of phenylethylthiolate ligands are now doublets, showing that two distinct monolayer ligand environments exist for Au38(PhC2S)24 MPCs. This observation re-enforces (33) The ratio of thiolate ligands to surface Au on MPCs is much greater than the 1:3 value known for flat Au(111), so the ligand population is necessarily larger on edges and vertexes. (34) Ha¨kkinen, H.; Barnett, R. N.; Landman, U. Phys. Rev. Lett. 1999, 82, 3264. (35) Ha¨berlen, O. D.; Chung, S.-C.; Stener, M.; Ro¨sch, N. J. Chem. Phys. 1997, 106, 5189. (36) Gex, J. N.; DeArmond, M. K.; Hanck, K. W. J. Phys. Chem. 1987, 91, 251.
Ligand Heterogeneity on Gold Clusters
Figure 8. Proton-decoupled 13C NMR spectra of a solution of Au38PhC2 MPCs in CD2Cl2 after passing through a silica gel column (used to clean samples of Oct4NBr). The electrochemical rest potential of the solution after the silica gel treatment became more positive, at a value corresponding to a MPC1+ state.
previous ones, discussed above, concerning variability of ligand chemical shifts. The effect of charging the Au38PhC2 MPC cores is shown in the expanded spectra of the aromatic carbon sites in Figure 7. Voltammetry and UV-visible spectra of Au38 PhC2 MPCs measured before and after 13C NMR spectral acquisition (∼15 h) detect the occurrence of a small amount of aggregation, but most of the MPCs remain unchanged. Figure 7 shows that, in the MPC1+ state, the chemical shift doublets seem to become “scattered”. (This is also seen in spectra of Au38 PhC2 MPC solutions (Figure 8) after elution from silica gel columns as an alternative part of the cleanup process; they apparently become oxidized (judged by rest potentials) on the column.) Upon attaining the MPC2+ charge state, however, the peaks in Figure 7 (left) begin to coalesce, similar to what is seen in Figure 4. Applying the approximation kC ) 2.2 ∆ν, as above, on the basis of ∆ν values from Figure 7 (lower right) leads to rough estimates of kC of 300-600 s-1, similar to the Au140C6 estimate. A dynamic averaging of the magnetically inequivalent aromatic carbons of Au38PhC2 MPCs appears to occur when positively charged, like Au140 MPCs. Ligand-exchange reaction rates16 for Au38C2Ph MPCs are very similar to those reported4b for Au140PhC2 MPCs, with no evidence of an important dissociative exchange pathway. Accordingly, any role of ligand dissociation, as discussed above for Au140PhC2 MPCs, is again ruled out as a source of the peak coalescence, leaving the possibilities (a and b) mentioned above. In addition, because of the molecule-like character of Au38PhC2 MPCs,4 the possibility of broadening from changes in the nanoparticle’s magnetic characteristic should be mentioned. Spectroelectrochemistry of Au140C6 MPCs. UVvisible spectroelectrochemistry of Au38PhC2 MPCs in various oxidation states has been reported.13 This MPC exhibits a steplike spectral fine structure, like other smallcore MPCs,37 and low-energy absorbance nearest the band edge in the spectrum is partially bleached upon oxidation of the MPC. The bleaching was ascribed to depletion of the HOMO level, analogous to the absorbance bleaching effect and quenching of narrow band-edge photoluminescence seen for semiconductor nanoparticles (e.g., quantum dots).38 Au140C6 MPCs do not exhibit an observable band edge (
