Physical Vapor Deposition of One-Dimensional Nanoparticle Arrays

Aug 23, 2007 - One-dimensional (1D) ensembles of 2−15 nm diameter gold nanoparticles were prepared using physical vapor deposition (PVD) on highly ...
0 downloads 0 Views 1MB Size
10372

Langmuir 2007, 23, 10372-10379

Physical Vapor Deposition of One-Dimensional Nanoparticle Arrays on Graphite: Seeding the Electrodeposition of Gold Nanowires C. E. Cross, J. C. Hemminger,* and R. M. Penner* Department of Chemistry and Institute For Surface and Interface Science (ISIS), UniVersity of California, IrVine, California 92697-2025 ReceiVed June 1, 2007. In Final Form: July 5, 2007

One-dimensional (1D) ensembles of 2-15 nm diameter gold nanoparticles were prepared using physical vapor deposition (PVD) on highly oriented pyrolytic graphite (HOPG) basal plane surfaces. These 1D Au nanoparticle ensembles (NPEs) were prepared by depositing gold (0.2-0.6 nm/s) at an equivalent thickness of 3-4 nm onto HOPG surfaces at 670-690 K. Under these conditions, vapor-deposited gold nucleated selectively at the linear step edge defects present on these HOPG surfaces with virtually no nucleation of gold particles on terraces. The number density of 2-15 nm diameter gold particles at step edges was 30-40 µm-1. These 1D NPEs were up to a millimeter in length and organized into parallel arrays on the HOPG surface, following the organization of step edges. Surprisingly, the deposition of more gold by PVD did not lead to the formation of continuous gold nanowires at step edges under the range of sample temperature or deposition flux we have investigated. Instead, these 1D Au NPEs were used as nucleation templates for the preparation by electrodeposition of gold nanowires. The electrodeposition of gold occurred selectively on PVD gold nanoparticles over the potential range from 700-640 mV vs SCE, and after optimization of the electrodeposition parameters continuous gold nanowires as small as 80-90 nm in diameter and several micrometers in length were obtained.

I. Introduction Increasingly, chemists are investigating metal nanowires because of their adjustable photonic properties, e.g. refs 1-6, because they can serve as transducers in chemical sensors, e.g. refs 7-10, and because they can be employed to electrically contact single molecules, e.g. refs 11,12. Metal nanowires with diameters larger than 20 nm can be produced using electron beam lithography (EBL),13 and a variety of novel nanofabrication methods have been demonstrated which permit the preparation of long metal nanowires in a predetermined location on a surface. One successful “top down” strategy is to subdivide a metal film with macroscopic two-dimensional (2D) structures using an ultramicrotome.5 Another strategy is to use a corrugated surface to partition a uniform flux of metal vapor into nanoscopic widths.14-17 In this paper, we describe a new “bottom-up” strategy * Address correspondence to these authors. Email: [email protected] or [email protected]. (1) Garcia, N.; Ponizowskaya, E. V.; Zhu, H.; Xiao, J. Q.; Pons, A. Appl. Phys. Lett. 2003, 82, 3147. (2) Hu, X. H.; Chan, C. T. Appl. Phys. Lett. 2004, 85, 1520. (3) Krenn, J. R.; Lamprecht, B.; Ditlbacher, H.; Schider, G.; Salerno, M.; Leitner, A.; Aussenegg, F. R. Europhys. Lett. 2002, 60, 663. (4) Xu, L. B.; Tung, L. D.; Spinu, L.; Zakhidov, A. A.; Baughman, R. H.; Wiley, J. B. AdV. Mater. 2003, 15, 1562. (5) Xu, Q. B.; Bao, J. M.; Capasso, F.; Whitesides, G. M. Angew. Chem., Int. Ed. 2006, 45, 3631. (6) Zhang, X. P.; Sun, B. Q.; Friend, R. H.; Guo, H. C.; Nau, D.; Giessen, H. Nano Lett. 2006, 6, 651. (7) Bogozi, A.; Lam, O.; He, H. X.; Li, C. Z.; Tao, N. J.; Nagahara, L. A.; Amlani, I.; Tsui, R. J. Am. Chem. Soc. 2001, 123, 4585. (8) Li, C. Z.; Sha, H.; Tao, N. J. Phys ReV B 1998, 58, 6775. (9) Murray, B. J.; Newberg, J. T.; Walter, E. C.; Li, Q.; Hemminger, J. C.; Penner, R. M. Anal. Chem. 2005, 77, 5205. (10) Murray, B. J.; Walter, E. C.; Penner, R. M. Nano Lett. 2004, 4, 665. (11) Xu, B. Q.; Xiao, X. Y.; Tao, N. J. J. Am. Chem. Soc. 2003, 125, 16164. (12) Xu, B. Q.; Zhang, P. M.; Li, X. L.; Tao, N. J. Nano Lett. 2004, 4, 1105. (13) Vieu, C.; Carcenac, F.; Pepin, A.; Chen, Y.; Mejias, M.; Lebib, A.; ManinFerlazzo, L.; Couraud, L.; Launois, H. Appl. Surf. Sci. 2000, 164, 111. (14) Jorritsma, J.; Gijs, M. A. M.; Kerkhof, J. M.; Stienen, J. G. H. Nanotechnology 1996, 7, 263. (15) Melosh, N. A.; Boukai, A.; Diana, F.; Gerardot, B.; Badolato, A.; Petroff, P. M.; Heath, J. R. Science 2003, 300, 112.

for the synthesis of two types of gold nanostructures: (1) Onedimensional (1D) ensembles of 2-15 nm diameter gold nanoparticles with ensemble lengths of up to 1 mm and (2) gold nanowires in the 70-90 nm diameter range with lengths of several micrometers. In both cases, we begin by preparing 1D ensembles of gold nanoparticles in the 2-15 nm diameter range using physical vapor deposition (PVD) on graphite surfaces. In the latter case, we then use these 1D nanoparticle ensembles as “nucleation templates” to catalyze the electrodeposition of gold nanowires. These experiments build upon prior work in which we have used the step edges present on the HOPG basal plane as horizontal templates for the electrodeposition of nanowires composed of a variety of materials including metals. Gold, platinum, and palladium nanowires with minimum diameters in the 80-100 nm range18-20 have been obtained using this “horizontal templating” approach.21 The formation of these nanowires occurs in two steps: First, gold nuclei are formed on step edges and terraces of the HOPG surface, and second, these nanoparticles are grown until they coalesce into electrically continuous nanowires. This mechanism produces polycrystalline nanowires with a distinctive morphology in which grains of the deposited material, each derived from a nanoscopic nucleus, are arrayed in a line along the axis of each nanowire. An important implication (16) Natelson, D.; Willett, R. L.; West, K. W.; Pfeiffer, L. N. Appl. Phys. Lett. 2000, 77, 1991. (17) Natelson, D.; Willett, R. L.; West, K. W.; Pfeiffer, L. N. Phys. ReV. Lett. 2001, 86, 1821. (18) Favier, F.; Walter, E. C.; Zach, M. P.; Benter, T.; Penner, R. M. Science 2001, 293, 2227. (19) Walter, E. C.; Murray, B. J.; Favier, F.; Kaltenpoth, G.; Grunze, M.; Penner, R. M. J. Phys. Chem. B 2002, 106, 11407. (20) Walter, E. C.; Zach, M. P.; Favier, F.; Murray, B. J.; Inazu, K.; Hemminger, J. C.; Penner, R. M. ChemPhysChem 2003, 4, 131. (21) However, this shift is not the nucleation overpotential, which is defined as the extra potential required to induce nucleation at a predefined rate referenced to the reversible potential for the metal or compound of interest. Because gold cannot be stripped from the HOPG surface, this reversible potential is not accurately known.

10.1021/la7016209 CCC: $37.00 © 2007 American Chemical Society Published on Web 08/23/2007

Seeding the Electrodeposition of Gold Nanowires

Langmuir, Vol. 23, No. 20, 2007 10373

Figure 1. Schematic diagram illustrating the relationship between nucleation density, δ, and the minimum nanowires diameter, diamin. As shown here, increasing the nucleation density by a factor of 2 reduces diamin by one-half.

of this mechanism is that the minimum nanowire diameter, diamin, is inversely proportional to the density of nuclei present on the step edges, δ:

diamin )

1 δ

(1)

Equation 1 predicts that an increase in δ by a factor of 2 produces a reduction in the minimum nanowire diameter by one-half, as seen schematically in Figure 1. The highest values for δ seen in purely electrochemical gold nanowire growth experiments is in the 10-13 µm-1 range, corresponding to the 80-100 nm minimum diameter obtained in these experiments.19 In this paper we ask the question: Can higher values of δ and therefore smaller nanowires be obtained by preparing the gold nuclei using PVD instead of electrodeposition? To answer this question, we have investigated the PVD of gold on HOPG surfaces as a function of the substrate temperature and the evaporation rate. Separately, we are interested in understanding the degree to which the particle size and interparticle spacing of the initial 1D ensembles prepared by PVD can be controlled and adjusted using electrodeposition. These experiments have culminated in the identification of PVD conditions that favor the formation of a high density (δ > 20 µm-1) of gold nanoparticles at step edges coupled with virtually no deposition on terraces. Finally, we have investigated the properties of these 1D gold nanoparticle ensembles for nucleating electrodeposited gold to form small nanowires. Early studies of the deposition by PVD of gold onto HOPG, dating back to 1967, have elucidated fundamental aspects of the nucleation and growth of gold on this surface.22,23 In 1975, Wayman and Darby24 were the first to study the distribution of PVD gold on HOPG as a function of the substrate temperature up to 450 °C using SEM. Heyraud and co-workers investigated the equilibrium shape of gold particles deposited on graphite25,26 and the self-diffusion of gold between islands on these (22) Arthur, J. R.; Cho, A. Y. Surf. Sci. 1973, 36, 641. (23) Evans, E. L.; Bahl, O. P.; Thomas, J. M. Carbon 1967, 5, 587. (24) Wayman, C. M.; Darby, T. P. J. Cryst. Growth 1975, 28, 53.

surfaces.27 Ganz et al.28,29 used STM to probe the structure of PVD submonolayers of gold and other metals on HOPG, concluding that 2D gold islands were not closest packed. Wynblatt et al.30 used TEM to study PVD gold islands prepared under conditions favoring the formation of dendritic islands and interpreted these data in terms of a diffusion-limited aggregation (DLA) model. Nishitani et al.31 studied the morphology of gold PVD deposits on HOPG between 293 and 383 K using STM. Compact islands were prepared at the lower end of this temperature range, whereas dendritic islands were seen at the high end of this range. Gladfelter and co-workers32 also observed, by STM, dendritic gold deposits for PVD gold on graphite at room temperature. Anton et al.33,34 detailed the use of in situ TEM to study dendritic growth of gold particles on HOPG. These experiments achieved “real time” recording of the deposition, nucleation, and growth process. Beebe and co-workers35 were the first to focus attention on the decoration of defects by PVD gold. In that work, defects were intentionally formed by bombarding HOPG surfaces with energetic Cs+ and Ga+ ion beams and then annealing these surfaces in air. Deposition of gold at 703 K with 2 h of annealing postdeposition produced gold rings, disks, and mesas, depending on the amount of gold deposited. Finally, Palmer and co-workers36,37 have investigated the PVD deposition of silver on HOPG. In that work, 1D ensembles of silver nanoparticles were obtained on HOPG step edges during PVD deposition at ambient and higher sample temperatures. Electrochemical gold deposition on HOPG has also been investigated previously. Arvia and co-workers38,39 investigated the morphology of electrodeposited gold particles as a function of deposition potential, finding that as the potential is made more negative the shape changes from facetted “Euclidean” shapes to dendritic fractal ones. This growth mode change was attributed to the influence of Cl- anions adsorbed on gold islands. White and co-workers40 studied step edge-deposited gold using ex situ AFM and SEM, concluding that gold deposits on the upper plane of an HOPG step edge defect, and that some electrodeposition of gold is chemically irreversible. In 1996, Zoval et al.41-43 showed that highly dispersed silver and platinum nanoparticles could be created on HOPG surfaces by potentiostatic electrodeposition. (25) Heyraud, J. C.; Metois, J. J. Acta Metall. Mater. 1980, 28, 1789. (26) Heyraud, J. C.; Metois, J. J. J. Cryst. Growth 1980, 50, 571. (27) Drechsler, M.; Metois, J. J.; Heyraud, J. C. Surf. Sci. 1981, 108, 549. (28) Ganz, E.; Sattler, K.; Clarke, J. J. Vac. Sci. Technol., A 1988, 6, 419. (29) Ganz, E.; Sattler, K.; Clarke, J. Phys. ReV. Lett. 1988, 60, 1856. (30) Wynblatt, P.; Metois, J. J.; Heyraud, J. C. J. Cryst. Growth 1990, 102, 618. (31) Nishitani, R.; Kasuya, A.; Kubota, S.; Nishina, Y. J. Vac. Sci. Technol., B 1991, 9, 806. (32) Strong, L.; Evans, D. F.; Gladfelter, W. L. Langmuir 1991, 7, 442. (33) Anton, R.; Kreutzer, P. Phys. ReV. B 2000, 61, 16077. (34) Anton, R.; Schneidereit, I. Phys. ReV. B 1998, 58, 13874. (35) Zhu, Y. J.; Schnieders, A.; Alexander, J. D.; Beebe, T. P. Langmuir 2002, 18, 5728. (36) Francis, G. M.; Goldby, I. M.; Kuipers, L.; vonIssendorff, B.; Palmer, R. E. J. Chem. Soc., Dalton Trans. 1996, 665. (37) Goldby, I. M.; Kuipers, L.; vonIssendorff, B.; Palmer, R. E. Appl. Phys. Lett. 1996, 69, 2819. (38) Martin, H.; Carro, P.; Creus, A. H.; Gonzalez, S.; Andreasen, G.; Salvarezza, R. C.; Arvia, A. J. Langmuir 2000, 16, 2915. (39) Martin, H.; Carro, P.; Creus, A. H.; Gonzalez, S.; Salvarezza, R. C.; Arvia, A. J. Langmuir 1997, 13, 100. (40) Boxley, C. J.; White, H. S.; Lister, T. E.; Pinhero, P. J. J. Phys. Chem. B 2003, 107, 451. (41) Zoval, J. V.; Lee, J.; Gorer, S.; Penner, R. M. J. Phys. Chem. B 1998, 102, 1166. (42) Zoval, J. V.; Biernacki, P. R.; Penner, R. M. Anal. Chem. 1996, 68, 1585. (43) Zoval, J. V.; Stiger, R. M.; Biernacki, P. R.; Penner, R. M. J. Phys. Chem. 1996, 100, 837.

10374 Langmuir, Vol. 23, No. 20, 2007

Cross et al.

Figure 2. Scanning electron microscope images of HOPG surfaces after the deposition of gold (rate ≈ 0.3 nm/s) at various temperatures: (a) T ) 310 K, ΓAu ) 1.3 nm equiv thickness; (b) T ) 373 K, ΓAu ) 0.6 nm; (c) T ) 562 K, ΓAu ) 0.6 nm, (d) T ) 686 K, ΓAu ) 3.3 nm; (e,f) T ) 686 K, ΓAu ) 4 nm.

II. Experimental Methods and Materials II.A. Graphite. All PVD experiments were carried out using 12 mm × 12 mm × 2 mm pieces of highly oriented pyrolytic graphite (HOPG) of ZYB grade or better, with a mosaic spread of 0.8° ( 0.2° (G.E. Advanced Ceramics). The HOPG was cleaved using Scotch tape in air immediately prior to being placed in the vacuum evaporator. II.B. Physical Vapor Deposition. All vapor depositions were performed in an Edwards 306A Coating System with a diffusion pump producing a base pressure of 10-7 Torr. HOPG samples were held ∼11.5 cm above the horizontal crucible at an angle of ∼57 ° from vertical. With samples in place and the vacuum pressure at 5 × 10-6 Torr or below, sample heating was performed by running current through the holders. Each holder could be heated independently. The power supply was controlled via an Omega CNi32 series temperature controller; by reading the temperature from the thermocouple these controllers could adjust, ramp, and hold the temperature to within a tenth of a degree. For each HOPG sample, prior to gold deposition, annealing at 673 K was carried out for at least 3 h to drive off adsorbates and contamination. Samples were then brought to the desired temperature for deposition. Depositions were performed by resistively heating a ceramic crucible with gold (ESPI Metals, 99.999%) for a variety of times. A shutter blocked the samples from the heating gold crucible; before HOPG sample exposure the crucible was allowed to equilibrate for the set current. During crucible heating the pressure rose slightly and was allowed to rebound to 10-7 Torr, before depositions. Immediately following depositions the shutter was replaced over the crucible, and crucible heating was turned off. The total gold deposition was measured with a precision of 0.1 nm using a quartz crystal microbalance (Edwards model FTM5) in the evaporator chamber. Immediately after gold deposition, the sample heating was

turned off, and samples were permitted to cool to room temperature in high vacuum and then were transferred to a desiccator. II.C. Electrodeposition. Electrochemical depositions were carried out using a one-compartment cell. Within this cell, the HOPG sample was held within a Teflon sheath that provided for the exposure of a 0.26 cm2 area of the HOPG basal plane to the gold plating solution using a Teflon O-ring. In addition to this “working electrode”, a platinum counter electrode and a saturated calomel reference electrode (SCE) were also employed. Electrochemical depositions and measurements were carried out using a computer-controlled EG&G Princeton Applied Research model 273A potentiostat. Gold was electroplated from aqueous 0.2 mM AuCl3 (Aldrich, 99.99+%), 0.1 M sodium chloride (Fisher, >99%). II.D. Scanning Electron Microscopy. Scanning electron microscopy (SEM) was carried out on uncoated or gold-coated samples using a Philips FEG-30XL microscope equipped with EnergyDispersive X-ray Analysis (EDX) elemental analysis capabilities. An accelerating voltage of 10 kV was used for SEM imaging, whereas for EDX analysis, 20 kV was used.

III. Results and Discussion III.A. Preparation of High Number Density 1D Gold Nanoparticle Ensembles Using PVD. The first objective of this study was to identify conditions of substrate temperature and deposition rate that provided for the selective decoration of gold at step edges on the HOPG basal plane. Using SEM, the distribution of gold on the HOPG surfaces was examined as a function of temperature over the range from 310 to 686 K (Figure 2). In these images, gold islands that nucleated at the quasilinear step edges were readily distinguished from islands that nucleated randomly on the terraces separating these step edges.

Seeding the Electrodeposition of Gold Nanowires

Langmuir, Vol. 23, No. 20, 2007 10375

Figure 3. High magnification SEM images of 1D Au NPEs prepared by depositing 4 nm of gold at 673 K. The particle diameter is in the 2-15 nm range.

Over this temperature range, the distribution of deposited gold changed dramatically in two ways: First, the number density of gold islands present on terraces decreased with increasing temperature from ∼3 × 109 cm-2 at 310 K (Figure 2a) to 1 × 108 at 562 K (Figure 2c) to ∼0 cm2 at 686 K (Figure 2d). Second, as the nucleation density of gold decreased on terraces, an increase in gold nucleation at step edges was observed until, at 686 K, gold islands were exclusively associated with linear step edge defects. The nucleation density we see at 310 K (e.g., Figure 2a) is 2 orders of magnitude lower than reported by Darby and Wayman44 for gold PVD at air-cleaved HOPG. These investigators attributed the elevated nucleation densities they observed (4 × 1011 cm-2 versus ∼108 for vacuum-cleaved samples) to nucleation at adsorbed gas molecules, but our dramatically lower nucleation densities for similar conditions of sample temperature and deposition flux suggest instead that the HOPG crystal quality and the associated areal density of atomic-scale point defects on terraces play an important role. The decrease in gold island density with increasing sample temperature documented in Figure 2 is qualitative as seen by Wayman and Darby.24 As seen in Figure 2d and Figure 3, the selective decoration of step edges with gold is achieved at a sample temperature of 686 K. The SEM image of Figure 2d shows “bunches” of 2-5 step edges crossing from right top to bottom left. An enlargement of one group of steps (Figure 2d, inset) shows four parallel step edges. The 2-5 nm diameter gold nanoparticles located at the exterior edges of this bunch are 3-5 times larger in diameter than those that have nucleated at the interior step edges, suggesting that gold particles that nucleate at steps within the interior of a step bunch experience a lower flux of gold adatoms and achieve a smaller limiting diameter than particles that nucleate outside the confines of a step bunch. This is consistent with the view that under our PVD conditions the vast majority of incident gold atoms land on the graphite terraces (the step defects make up only a small percentage of the surface). At elevated surface temperatures, the gold atoms on the terraces diffuse rapidly until they encounter a step defect. Since the residence time at the step is longer, there is then a higher probability of nucleation at the step followed by particle growth. In Figure 3, the particle diameter is seen to be 2-15 nm and the number density of gold islands is 30-40 µm-1sa factor of 3 higher than is achieVable for the electrochemical nucleation of gold using potentiostatic pulses.20,45 (44) Darby, T. P.; Wayman, C. M. J. Cryst. Growth 1975, 28, 41. (45) Penner, R. M. J. Phys. Chem. B 2002, 106, 3339.

Surfaces like this, on which the step edges are selectively decorated with one-dimensional gold nanoparticle ensembles (henceforth “1D Au NPEs”) are the focus of the rest of this paper. With the continued deposition of more gold, one might expect the gaps between nanoparticles to be filled with gold and continuous nanowires to be formed, but this does not occur at any temperature and flux condition that we investigated in this study. Addition of more gold to the surface does lead to the formation of elongated particles along the steps, and these particles are also wider in the direction perpendicular to the step. However, we have not found conditions that routinely lead to continuous wires. At the same time addition of more gold to the surface results in the initiation of island growth on the terraces. One straightforward explanation of this behavior is that as the particles at the steps elongate and grow, desorption of gold atoms from the particle back onto the terrace becomes important. Under these conditions, the particles at the steps are no longer only a sink for gold atoms. The result is that even under constant flux conditions the average concentration of gold atoms on the terraces increases. This can then lead to nucleation of particles and islands either on the flat terraces or on isolated defects away from the steps. Under our experimental conditions this appears to compete with the final stages of filling in the gaps between elongated particles at the steps. This behavior is apparent in the SEM images of Figure 2e,f showing a surface on which 4 nm of gold were deposited at 686 K. Can 1D ensembles of gold particles be obtained by redistributing gold from islands on terraces to step edges by postdeposition thermal annealing? Figure 4 shows the results of an experiment in which gold was first deposited at 302 K to produce a high density of dendritic islands on terraces (Figure 4a), and then thermally annealed at 673 K for 5 h (Figure 4b). The predominant effect of annealing is island growth coupled with the smoothing of dendritic island perimeters. However we also found that annealing reduced the area of the HOPG surface that was covered by gold to 60% of its preannealing value. This reduction in area could result either from the loss of gold from the surface by evaporation or by the increase in height of gold islands during annealing. Based on the fact that the annealing temperature employed here is more than 500-600 K below the melting temperature of bulk gold (1337 K) and nanometer-scale gold clusters (>1200 K),46 3D island growth is the most likely (46) Sambles, J. R. Proc. R. Soc. London, Ser. A 1971, 324, 339.

10376 Langmuir, Vol. 23, No. 20, 2007

Cross et al.

Figure 4. SEM image showing the effects of postdeposition annealing. (a) Deposition at 302 K. (b) The same sample after ∼5 h of vacuum heating at 673 K.

Figure 5. (a) Cyclic voltammograms (3 scans) in aqueous 0.2 mM Au(III)Cl3, 0.1 M NaCl at 20 mV s-1 for two electrodes: Freshly cleaved HOPG (top) HOPG on which 3.3 nm of gold was deposited at 673 K (bottom). (Inset) SEM image of gold PVD decorating the step edge of HOPG. In each case, the first voltametric scan is shown with solid lines, and the dashed lines indicate the second and third. (b) Current versus time transients at five deposition potentials, as indicated, for HOPG on which 3.3 nm of gold was deposited at 673 K.

cause of the observed decrease in covered area. As is apparent in Figure 4, gold was not redistributed from terraces to step edges during annealing. The only route to the 1D nanoparticle ensembles seen in Figures 2d and 3 is gold PVD at a sample temperature above 670 K. In the next section of this paper, we explore the use of such vapor-deposited 1D gold nanoparticle ensembles as nucleation templates for the electrodeposition of gold nanowires. The additional complexity of electrodeposition was necessary since, as already indicated, we were unable to obtain gold nanowires at step edges by PVD alone. III.B. Electrodeposition of Gold onto 1D Particle Ensembles. Electrochemical step-edge decoration (ESED) is a purely electrochemical method for preparing nanowires of an electrodeposited material on HOPG surfaces.19,45,47 In the first step of the ESED process, step edges on the HOPG surface are electrochemically oxidized at +0.80 V vs SCE, and then (step 2) ensembles of sub-10-nm diameter nanoparticles are prepared at step edges and terraces by applying a negative voltage pulse (47) Zach, M. P.; Inazu, K.; Ng, K. H.; Hemminger, J. C.; Penner, R. M. Chem. Mater. 2002, 14, 3206.

lasting 10-100 ms, typically. The amplitude of this voltage pulse is 1.0 V or more negative of the reversible potential, Erev, for the deposition of the material of interest. These nanoparticles are then grown very slowly, at 10-100 mV below Erev, until they coalesce into a continuous, polycrystalline nanowire. As already indicated above (e.g., Figure 1), the minimum diameter of the nanowire that forms under these conditions is determined by the number density and diameter of the nanoparticles deposited at step edges, and the highest nucleation densities that have been achieved are in the 10-13 µm-1 range. We substituted 1D Au NPEs prepared using PVD for those generated electrochemically in the ESED procedure. To do this, we omitted the first two steps of the ESED procedure and relied on the PVD gold NPEs to facilitate nucleation at step edges following the application of a small amplitude “growth” pulse. The effect of the 1D Au NPEs on the nucleation of electrodeposited gold can be assessed using cyclic voltammetry. As shown in Figure 5, at a freshly cleaved HOPG surface on which no gold has been deposited, the gold deposition peak on the first voltammetric scan is shifted negative by 300 mV from the peak potential of scans 2, 3, etc. This peak shift, relative to subsequent

Seeding the Electrodeposition of Gold Nanowires

Langmuir, Vol. 23, No. 20, 2007 10377

Figure 6. Low magnification (a) and higher magnification (b) SEM images acquired at the location of the o-ring used to isolate electrodeposition on the HOPG surface. At left in these two images are 1D Au NPEs before electrodeposition and at right, after electrodeposition. In (b) an intermediate zone of reduced gold deposition is present, probably due to shielding of this region by the curved edge of the o-ring.

voltammetric scans, is caused by the overpotential associated with the formation of gold nuclei on the HOPG surface. The magnitude of the peak shift from scan 1 to scans 2, 3, etc. therefore provides a direct measure of the facility with which nucleation is initially occurring on the electrode surface.21 The 1D Au NPEs are expected to facilitate gold electrodeposition in much the same way as electrodeposited gold nuclei, and the cyclic voltammetry data for gold deposition at this surface are consistent with this expectation: The peak shift seen here between scan 1 and scans 2, 3, etc. (Figure 5a, bottom) is just 33 mVsone-ninth of the value measured at a pristine HOPG surface. We investigated the electrodeposition of gold onto 1D Au NPEs at potentials in the range from 640-700 mV vs SCE. This potential range is well positive of the onset for gold deposition at pristine HOPG electrodes (e.g., Figure 5a, top) ensuring that gold deposition occurred selectively on the PVD gold nanoparticles and therefore is confined to the step edges. The distribution of electrodeposited gold was investigated for 150 s potentiostatic depositions at four different potentials: 640, 660, 680, and 700 mV vs SCE. Current versus time transients for these depositions are seen in Figure 5b, showing that, as expected, significantly more gold was deposited as the potential was made more negative. For example, 3.3 mC of gold was deposited at 640 mV, whereas just 0.28 mC was deposited at 700 mV. The selectivity of the gold electrodeposition within this potential range is documented by the images shown in Figure 6 which show an area of the HOPG surface on either side of the rubber o-ring that was used to isolate electrodeposition on the HOPG surface. These images afford a view of the PVD gold seeds without deposited gold (left) and after the deposition of gold (right). From these images, it is apparent that gold deposition occurs only at the location of PVD gold nanoparticles, just as predicted from the CV. However, while gold deposition occurs only on the preexisting PVD “seeds” at potentials within the range from 700-640 mV, not all PVD gold particles undergo growth over this potential range. As shown in the SEM image of Figure 7a, at 640 mV gold was electrodeposited onto all of the PVD gold “seeds” present on the surface (a uniform increase in size is seen for all), but this is not the case for more positive growth potentials. As the growth potential was progressively increased, gold electrodeposition was concentrated at an increasingly smaller fraction of the available PVD gold seeds until, at 700 mV (Figure 7d), gold electrodeposition appears to occur on fewer than half of the available PVD gold nanoparticles. The number of PVD gold particles that become “activated” for electrodeposition varied with the deposition time as well as the deposition potentialsthat is, we found that gold “renucleation” on the available PVD gold seeds was progressive,

not instantaneous. Consequently, when the quantity of electrodeposited gold is normalized, the effect seen in Figure 7 is weaker, but still readily apparent. As an example, Figure 8 shows samples prepared with 1.5-1.6 mC of gold at these same four deposition potentials. To equalize the quantity of deposited gold, the deposition time was varied from 125 s (640 mV) to 350 s (700 mV). In Figures 7 and 8, PVD gold particles on which no electrodeposition occurred can be recognized because they have retained not only the diminutive size but also the faceted shape characteristic of the PVD gold particles prepared at 673 K (see Figures 2d and 3). The formation of continuous nanowires can be monitored by measuring the continuous particle length along the HOPG step edges as a function of the deposition time. Before gold is electrodeposited, this length is of the order of the diameter of the 2-15 nm single gold nanoparticles since these particles are not connected to one another. We analyzed SEM images for multiple samples using ImageJ, an image processing software published by the National Institutes of Health.48 ImageJ provided length versus pixel intensity data from the SEM images of step edge decoration (PVD and ECD), and a computer program was written to analyze the XY data. Histograms of the contiguous particle length after ECD for various deposition durations at two potentialss640 and 700 mVsare shown in Figure 9. At a growth potential of 700 mV, even after a deposition duration of 500 s most wire segments remained 50 nm or less in length. This is because just one-half of all PVD gold particles undergo any growth at this potential, as shown in Figures 7d and 8d. At 640 mV, in contrast, all PVD particles undergo growth, and deposition of approximately the same total amount of gold (in 100-200 s) produces a longer median wire length of 100 nm with 5-20% of wire segments more than 400 nm in length. In principle, based upon eq 1, the nucleation density we achieve for 1D Au NPEs should translate into gold nanowires with minimum diameters in the 25-35 nm range, but the smallest nanowires that are continuous over lengths of more than 1.0 µm were approximately a factor of 3 larger than this minimum value: in the 80-100 nm range. Examples of such nanowires are shown in the SEM images of Figure 10. What accounts for the difficulty in obtaining nanowires in the 25-35 nm range in these experiments? The model represented by Figure 1 and eq 1 assumes a hemispherical particle shape, no particle size dispersion, and a fixed and uniform nucleation density along the step edge. Instead, as we have seen above, gold particles prepared by PVD are irregularly shaped, they possess polydispersity in (48) Abramoff, M. D.; Magelhaes, P. J.; Ram, S. J. Biophotonics Int. 2004, 11, 36.

10378 Langmuir, Vol. 23, No. 20, 2007

Cross et al.

Figure 7. SEM images of 1D Au NPEs after the potentiostatic electrodeposition of gold for 150 s from aqueous 0.2 mM Au(III)Cl3, 0.1 M NaCl at the following potentials (vs SCE): (a) 640 mV, QAu ) 3.1 mC; (b) 660 mV, QAu ) 1.3 mC; (c) 680 mV, QAu ) 0.47 mC; and (d) 700 mV, QAu ) 0.28 mC.

Figure 8. SEM images of 1D Au NPEs after the potentiostatic electrodeposition of 1.5-1.6 mC of gold: (a) 640 mV, tdep ) 125 s; (b) 660 mV, tdep ) 150 s; (c) 680 mV, tdep ) 200 s; and (d) 700 mV, tdep ) 350 s.

terms of their dimensions and polydispersity with respect to their nearest neighbor distances, that is, the “local” nucleation

density. Fluctuations in these three parameters translate into a requirement that the mean nanowire diameter be larger than

Seeding the Electrodeposition of Gold Nanowires

Langmuir, Vol. 23, No. 20, 2007 10379

Figure 9. Statistical comparison of wire growth resulting from ECD onto PVD gold seeds prepared by the deposition of 4 nm equiv thickness of gold at 673 K at two potentials: 700 mV (a) and 640 mV(b). Shown are histograms of lengths of contiguous metal segments measured along the axis of step edges as a function of the deposition time for gold deposition from a solution of aqueous 0.2 mM Au(III)Cl3, 0.1 M NaCl.

Figure 10. SEM images of gold nanowires prepared in two steps: (1) PVD deposition of 3.3 nm of gold at 673 K, (2) electrodeposition of gold at 612 mV vs SCE for 15 s.

predicted by eq 1 in order to achieve intimate grain-to-grain contact along the axis of the incipient nanowire.

IV. Summary On HOPG surfaces, physical vapor deposition provides a means for preparing one-dimensional ensembles of gold nanoparticles that are 2-15 nm in diameter and spaced by