Gold Nanoelectrodes of Varied Size: Transition to Molecule-Like Charging

See allHide authors and affiliations

Science  26 Jun 1998:
Vol. 280, Issue 5372, pp. 2098-2101
DOI: 10.1126/science.280.5372.2098


A transition from metal-like double-layer capacitive charging to redox-like charging was observed in electrochemical ensemble Coulomb staircase experiments on solutions of gold nanoparticles of varied core size. The monodisperse gold nanoparticles are stabilized by short-chain alkanethiolate monolayers and have 8 to 38 kilodaltons core mass (1.1 to 1.9 nanometers in diameter). Larger cores display Coulomb staircase responses consistent with double-layer charging of metal-electrolyte interfaces, whereas smaller core nanoparticles exhibit redox chemical character, including a large central gap. The change in behavior is consistent with new near-infrared spectroscopic data showing an emerging gap between the highest occupied and lowest unoccupied orbitals of 0.4 to 0.9 electron volt.

Nanoparticles of metals and semiconductors have sparked intense interest (1) in anticipation that this unexplored range of materials dimensions will yield size-dependent optical, electronic, and chemical properties suitable for applications in optoelectronic nanodevices, catalysts, and chemical sensors (2–4). Among known preparations of nanoparticles (5–8), recent attention has focused on alkanethiolate monolayer-protected metal clusters (MPCs). Gold MPCs in particular are quite stable and can be prepared with average core diameters of 1.1 to 5 nm. Electrochemical studies have demonstrated that Au MPCs are equivalent to diffusing, nanometer-sized electrodes (9) and can provide electrocatalytic advantages (10). Further, room-temperature solutions of MPCs with monodisperse cores display an electrochemical “ensemble Coulomb staircase” (11), a behavior anticipated and explained based on the sub-attofarad double-layer capacitances (C CLU) of diffusing, nanometer-sized 28-kD metallic Au particles coated with a monolayer (hexanethiolate, C6) dielectric. Analogous staircase phenomena have been reported, using nanometer-sized electrodes (12).

We aim to further understand electrochemical ensemble Coulomb staircases by varying the monodisperse core mass in Au MPC solutions from 8 to 38 kD (core diameters of 1.1 to 1.9 nm). The double-layer capacitive charging seen for larger core sizes changes for smaller MPC core sizes to a molecular redox-like behavior. That is, over a certain range of core sizes, electron orbital-shell effects or pairing effects, or both, begin to dominate, changing the cluster capacitance from one determined by electrostatic processes to one more dominated by bonding interactions.

Coulomb staircases for nanoparticles are usually observed as tunneling currents through a single nanoparticle addressed by a tip probe (Fig. 1A), that undergo stepwise increments with increasing tip-substrate bias (V) (13,14). A model accounting for junction capacitances in a double tunnel-junction circuit (Fig. 1A) predicts that current increments occur at critical voltage biases (V C)Embedded Image(1)where Z is integral nanoparticle charge,e the electron charge, C capacitance of the more resistive junction, and Q O a fraction associated with tip-substrate work function differences. Coulomb staircase charging is normally observed at low temperatures because of the requirement that the stepwise charging energy (E C = e 2/C) greatly exceeds thermal energy,k B T, where k Bis Boltzmann's constant and T is temperature. Equation 1 predicts that if C is constant, consecutive charging steps should occur at a regular spacing ΔV C =e/C.

Figure 1

(A). Schematic STM double tunnel-junction model (13, 14). (B). Schematic electrochemical ensemble Coulomb staircase model.R ct is charge-transfer resistance andZ W is diffusional (Warburg) impedance for MPC transport through the solution. Differential pulse voltammograms for (C) butanethiolate (C4) and (D) hexanethiolate (C6) Au MPCs as a function of uniform core size, in 0.05 M Hex4NClO4/toluene/acetonitrile (2/1 v:v), at 9.5 × 10−3 cm2 Pt electrode; DC potential scan 10 mV/s, pulse amplitude 50 mV. Concentrations are: (C) 14 kD, 0.086 mM; 22 kD, 0.032 mM; 28 kD, 0.10 mM; (D) 8 kD, 0.30 mM; 22 kD, 0.10 mM; 28 kD, 0.10 mM; 38 kD, 0.10 mM. Arrows at lower left indicate ΔE potential steps used in Fig. 4.

Figure 1, C and D, presents electrochemical ensemble Coulomb staircase behavior for MPCs of varied core mass, in the form of differential pulse voltammograms (DPVs) at a Pt electrode. The interfacial double-layer chargings of the uniform electronic charge and core-size MPCs with C4 and C6 coatings (15) produce a series of DPV current peaks (in both positive- and negative-going scans ofE) that occur at the ΔVC =e/CCLU spacings summarized in Fig. 2. The MPCs reach and depart the electrode/solution interface by diffusion, so that the sizes of the current peaks are determined by a series of interfacial electron-transfer and diffusional (“Warburg”) impedances (Fig. 1B). A very large number of MPCs become charged during each DPV peak, being those forming the diffusion layer around the electrode, hence our use (11) of the term “ensemble staircase.”

Figure 2

DPV peak potentials for MPCs inFig. 1, C and D. CCLU values are shown for the ΔV C nearest E = 0 V (solid line brackets), those for adjacent ΔV C spacings are shown by dotted brackets. For the large-core MPCs, theC CLU values correspond to a monolayer dielectric constant of about 5.

The DPV peak spacings for the larger MPC core sizes (22, 28, and 38 kD) differ from those of the smaller ones (8 and 14 kD). Figure 2shows that the central ΔV C spacings [solid brackets, peaks spanning E = 0 V where Z is nearest zero (16)] are similar (0.32 to 0.40 V) for the larger core sizes and correspond toC CLU = 0.4 to 0.5 aF/cluster, or 3 to 4 μF/cm2 when normalized for estimated Au core surface areas (A CLU). These values are similar to capacitances (C DL) of macroscopic alkanethiolate monolayer-coated Au(111) surfaces (17) and to average C DL's of polydisperse MPC solutions (9). Also, the ΔV C spacings for the larger core MPCs (Fig. 2) decrease slightly at more positive and negative E, as expected for double-layer capacitances. Thus, the DPV results for the larger MPC core sizes in Fig. 1, C and D, andFig. 2 can be confidently interpreted as charging of metal cores coated with a dielectric film and in an electrolyte solution.

For the smaller core MPCs (8 and 14 kD), however, the central ΔV C spacings [1.2 and 0.7 V, Fig. 2, solid brackets (18)] are much larger than the central ΔV C spacings of the larger core MPCs. They are also much larger than the ΔV C spacings seen at more positive and negative E, for both small- and large-core MPCs (Fig. 2, dotted brackets). Further, only minor changes inC CLU are expected from changes inA CLU, based on consideration of the concentric sphere capacitance model (9) and given the similarities of core radii to monolayer dielectric thicknesses. The qualitatively different voltammetric behavior of the 8- and 14-kD MPCs is highly reminiscent of redox transformations known for Pt-carbonyl clusters [Ptn(CO)m, n = 24, 26, and 38] (19). Thus, the results in Fig. 1, C and D, and Fig. 2show, within a series of MPCs of nominally constant architecture and composition, a transition from classical metal-like double-layer charging to that resembling electrochemical charging of an electroactive molecule.

The core-size dependent change in Fig. 1, C and D, and Fig. 2 suggest the emergence of a significantly quantized electron-level structure, including, for example, the appearance of a substantial gap between the highest occupied and lowest unoccupied orbitals (HOMO-LUMO) in the neutral (or as obtained) clusters. Such a gap should also be manifest in their optical spectroscopic responses, which are shown in Fig. 3 for three of the above MPCs, along with (for comparison) a larger MPC (66 kD, ∼2.2 nm) and bulk (colloidal) Au. The response function ɛ2 (imaginary part of the complex dielectric function) reflects the density and strength of optical transitions at each transition energy and shows several strong trends with decreasing MPC core size: (i) enhanced strength in the near-infrared region (<1.8 eV), (ii) the appearance of discrete or band-like spectral features, and (iii) an increasing “gap” energy, reflecting primarily the HOMO-LUMO gap, below which (in contrast to the ω–3 divergence of bulk Au) the optical response vanishes. For the 28-kD MPC, the gap is no greater than 0.4 eV (the smallest energy for which the functions are reliable), but the 14- and 8-kD MPCs exhibit larger gaps, estimated as 0.6 and nearly 0.9 eV, respectively. These gaps are reasonable in terms of the electronic structure of Au clusters, which have all of the gross characteristics of the bulk bands for clusters with more than 20 atoms (20), assuming that clusters with closed electron shells are preferentially formed and isolated. The differences between the spectroscopic gaps and electrochemical ΔV Cvalues (0.74 and 1.2 eV, respectively) for the 14- and 8-kD cores are explicable considering that the spectroscopic values are estimated as band edges, whereas the ΔV C results are taken at the DPV peak maxima, not their edges. A similar electrochemical/orbital-energy correlation has been found for Pt-carbonyl clusters (19) and also for C76fullerene (21) and La@C82 metallofullerene (22), but in those cases, no higher homologs were available to observe the transition to the double-layer capacitive charging that occurs for the larger Au MPCs.

Figure 3

Optical response (ɛ2) of monodisperse Au MPCs with various core sizes. The HOMO-LUMO gaps vary with core size: (a) bulk Au, 0 eV; (b) 66 kD, ≪0.4 eV; (c) 28 kD, ≪0.4 eV; (d) 14 kD, ∼0.6 eV; (e) 8 kD, ∼0.9 eV. These curves were derived from a dispersion analysis (Kramers-Kronig relation) of the optical absorbance spectra, across the 0.2 to 6.4 eV range, of dilute solutions at room temperature. Absorbances at lower energies were obtained by extrapolating measured spectra and those in the far UV by assuming convergence to bulk Au values. Dielectric functions (ɛ1and ɛ2) of the (assumed spherical) nanoparticles were computed from absorption coefficients and calculated index of refraction using a model of the effective dielectric constants for the nanoparticle-medium composite (Mie theory), as was done in (8).

That the DPV peaks of Fig. 1, C and D, all correspond to 1e transfers is supported by double potential step chronocoulometry of a C4, 28-kD MPC solution. Figure 4 shows charge-time results from steppingE from its rest value (approximately –0.1 V versus a Ag/AgCl electrode) by ΔE to the valleys between successive DPV peaks (Fig. 1, C and D, lower left, see arrows), and back. The equation for diffusion-controlled charging of MPCs in the forward potential step is (23)Embedded Image Embedded Image(2)The back potential step equation has an analogous time dependence (23) and the same slope terms. CalculatingC CLU from the slope S F of the smallest forward ΔE step (–0.103 V ↔ 0.200 V) chronocoulometric plot in Fig. 4 gives a value (0.55 aF) close to that obtained from the central ΔV C spacing (0.50 aF) in DPV, as expected for both being 1e transfers. That the other DPV peaks are also 1e transfers is shown byFig. 4, in that the slopes of forward and back plots change by integral multiples of those (S F, andS R, respectively) of the smallest ΔE step (–0.103 V ↔ 0.200 V, assuming constantC CLU ) as ΔE is incremented across successive charging peaks.

Figure 4

Plots for forward (A; t < τ) and reverse (B; t > τ) steps in double potential step chronocoulometry for C4, 28 kD MPC solution (steps shown by arrows in Fig. 1, C and D, lower left). Step reversal time 0.25 s. (a) Slopes (C/ms1/2) for smallest (–0.103 V ↔ 0.200 V) forward and reverse potential step are, respectively,S F = 5.1 × 10–9 andS R = –5.68 × 10–9. Using this S F in Eq. 1 with D = 1.54 × 10–6 cm2/s (from microelectrode voltammetry) gives C CLU = 0.55 aF, which is close to that obtained from ΔV C =e/C CLU (0.50 aF, Fig. 2) where 1e transfer is assumed. Normalizing slopes for the other forward and reverse potential steps to S F andS R gives (b) –0.106 V ↔ 0.400 V, 2.0S F, 2.0 S R; (c) –0.100 V ↔ 0.600 V, 2.9 S F, 3.0S R; (d) –0.092 V ↔ 0.800 V, 4.4S F, 4.3 S R; (e) –0.090 V ↔ –0.420 V, –1.3 S F, –0.8S R; (f) –0.090 V ↔ –0.800 V, –4.6S F, –1.9 S R. The forward slope for the –0.090 V ↔ –0.800 V step is enlarged by reduction of solution O2 impurity.

Finally, although the analogy between electrochemical ensemble and classical Coulomb staircase charging is strong, there are differences worth noting. (i) Their equivalent circuits differ as shown in Fig. 1, A and B, the principal reason being that the electrochemical currents are controlled by MPC diffusion. (The equivalent circuits would be more alike were the MPCs attached as a monolayer to the electrode.) (ii) The capacitance (C CLU) that determines ΔV C spacings in the electrochemical case is that of the entire MPC surface, whereas capacitance in the classical experiment (Fig. 1A) is determined by the two junction contacts. (iii) The electrochemical case involves an ensemble of MPCs (as opposed to a single nanoparticle in Fig. 1A), so that macroscopic transport relations can be used to describe their voltammetry, as illustrated in Fig. 4. (iv) It is difficult to conceive of a fractionally charged MPC in solution, so there is no analogy in electrochemical ensemble charging to Q O in Eq. 1.

Solution-phase electrochemical ensemble Coulomb staircase charging has also been observed in experiments performed on arylthiolated Au MPCs, so our results may be forerunners of a general phenomenon. In addition, because the staircase behavior is closely related to MPC core electronic energy structure, it may aid understanding of other nanophase properties, such as the metal-insulator transition of Ag nanoparticles upon compression (24). Finally, although differences in fundamental properties reside in the metal-like and molecule-like charging behaviors, we anticipate that their electrochemical, thermodynamic, and kinetic properties will, upon further study, prove to fit within a common formal representation.

  • * To whom correspondence should be addressed. E-mail: rwm{at} (R.W.M.) or robert.whetten{at} (R.L.W.)


View Abstract

Navigate This Article