l-Penicillamine-Capped Silver

Chem. Mater. 2007, 19, 2831-2841

2831

Synthesis and Chiroptical Study of D/L-Penicillamine-Capped Silver Nanoclusters Naoki Nishida,† Hiroshi Yao,*,† Tomoyasu Ueda,‡ Akito Sasaki,‡ and Keisaku Kimura† Graduate School of Material Science, UniVersity of Hyogo, 3-2-1 Koto, Kamigori-cho, Ako-gun, Hyogo 678-1297, Japan, and X-ray Research Laboratory, Rigaku Corporation, 3-9-12 Matsubara-cho, Akishima, Tokyo 169-8666, Japan ReceiVed January 4, 2007. ReVised Manuscript ReceiVed March 28, 2007

We report the synthesis and chiroptical properties of silver nanocluster enantiomers. The surface of the silver nanoclusters is covered with L/D-penicillamine or their racemate. The clusters are separated by gel electrophoresis according to their charge and size into well-defined compounds. The mean core diameter of each fractioned component varies from 1 to 3 nm. The separated compounds of silver nanoclusters covered with L- or D-form thiol show strong optical activity or circular dichroism with opposite sign (mirror image relationship) in the metal-based electronic transitions (obtained anisotropy factor is on the order of 1 × 10-3 to 1 × 10-5), suggesting that the nanoclusters have well-defined stereostructures as common chiral molecules do. With a decrease in the mean cluster diameter, the anisotropy factor gradually increases at first, but a steep rise is observed when the diameter becomes smaller than ∼1.5 nm. The chiroptical response of the silver nanoclusters is consequently several-fold larger than that of the gold nanoclusters having the same ligand. The possibilities of a dissymmetric field model as well as an inherently chiral core model are discussed as the origin of the intense optical activity.

Introduction Nanoclusters are a new class of materials made up of tens to hundreds of atoms and/or molecules, and thus they can be seen in intermediates between single atoms/molecules and bulk materials. These nanoclusters are of considerable interest because of their potential application for biosensing, catalysis, and nanoelectronics.1 In particular, metal nanoclusters with the diameter reduced to the order of metal’s Fermi wavelength (e.g., ∼0.5 nm for gold and silver) have attracted strong attention because they give discrete electronic transitions among quantized levels,2 and the advanced charge transport phenomena such as the Coulomb staircase or blockade are expected.3 In small size clusters such as semimetallic carbon fullerenes4 and nanotubes5 on the other hand, low-symmetry structures are established to observe chirality. In the case of metal nanoclusters, some theoretical studies show that the most stable (lowest-energy) isomers of bare gold clusters (e.g., Au28 and Au55) are able to have * Corresponding author. Tel: 81-791-58-0160. Fax: 81-791-58-0161. E-mail: [email protected]. † University of Hyogo. ‡ Rigaku Corporation.

(1) (a) Chen, S.; Ingram, R. S.; Hostetler, M. J.; Pietron, J. J.; Murray, R. W.; Schaaff, T. G.; Khoury, J. T.; Alvarez, M. M.; Whetten, R. L. Science 1998, 280, 2098. (b) Schwerdtfeger, P. Angew. Chem., Int. Ed. 2003, 42, 1892. (c) Whetten, R. L.; Shafigullin, M. N.; Khoury, J. T.; Schaaff, T. G.; Vezmar, I.; Alvarez, M. M.; Wilkinson, A. Acc. Chem. Res. 1999, 32, 397. (d) Huang, T.; Murray, R. W. J. Phys. Chem. B 2001, 105, 12498. (2) (a) Kubo, R. J. Phys. Soc. Jpn. 1962, 17, 975. (a) Sanchez, A.; Abbet, S.; Heiz, U.; Schneider, W.-D.; Ha¨kkinen, H.; Barnett, R. N.; Landman, U. J. Phys. Chem. A 1999, 103, 9573. (3) Davidovic, D.; Tinkham, M. Appl. Phys. Lett. 1998, 73, 3959. (4) Ettl, R.; Chao, I.; Diederich, F.; Whetten, R. L. Nature 1991, 353, 149. (5) IIjima, S. Nature 1991, 354, 56.

a disordered low-symmetry geometry and a chiroptical effect.6 Passivation of small gold nanoclusters with low-symmetry (or chiral) molecules also results in unique electronic and chiroptical responses that are unlike those of the component parts. For example, Whetten and co-workers observed large optical activity in gold nanoclusters protected by a monolayer of glutathione molecules.7 Their gold nanoclusters with the core diameters smaller than 1-2 nm (∼10-40 kDa in mass) were prepared and isolated as a discrete family of massselected fractions. These nanoclusters were small enough to have discrete electronic states (quantum size effect) not found in larger particles, and displayed circular dichroism (CD) from the ultraviolet to the near-infrared. In glutathione passivation, because larger clusters showed a higher anisotropy factor (ratio of the CD signal to the absorption), they attributed that the phenomenon would be the result of a helical or chiral structure at the metal core of the clusters.7b This chiral core model has been supported by theoretical studies by Garzo´n and co-workers: strong Au-S interaction distorts the cluster geometry and induces chiroptical activity regardless of the chirality of the thiolate ligands.6 Note that glutathione is an optically active tripeptide of L-L configuration (γ-L-glu-cys-gly). Recently, we reported large optical activity of gold nanoclusters of 0.6-1.8 nm protected by a (6) (a) Garzo´n, I. L.; Reyes-Nava, J. A.; Rodrı´guez-Herna´ndez, J. I.; Sigal, I.; Beltra´n, M. R.; Michaelian, K. Phys. ReV. B 2002, 66, 073403. (b) Garzo´n, I. L.; Beltra´n, M. R.; Gonza´lez, G.; Gutı´errez-Gonza´lez, I.; Michaelian, K.; Reyes-Nava, J. A.; Rodrı´guez-Herna´ndez, J. I. Eur. Phys. J. D 2003, 24, 105. (c) Roma´n-Vela´zquez, C. E.; Noguez, C.; Garzo´n, I. L. J. Phys. Chem. B 2003, 107, 12035. (7) (a) Schaaff, T. G.; Knight, G.; Shafigullin, M. N.; Borkman, R. F.; Whetten, R. L. J. Phys. Chem. B 1998, 102, 10643. (b) Schaaff, T. G.; Whetten, R. L. J. Phys. Chem. B 2000, 104, 2630.

10.1021/cm0700192 CCC: $37.00 © 2007 American Chemical Society Published on Web 05/02/2007

2832 Chem. Mater., Vol. 19, No. 11, 2007

pair of enantiomers of chiral penicillamine.8 Three fractions of the clusters (containing about 6, 50, and 150 gold atoms, respectively) were separated in gel electrophoresis. These optically active clusters exhibited a clear mirror-image relationship, with the anisotropy factors decreasing with the cluster size increase. On the basis of kinetic and structural considerations, we have attributed the observed optical activity to a dissymmetric field induced by the chiral penicillamine shell. A recent theoretical study based on a particle-in-a-box model supports the dissymmetric field model; the symmetric metal cores can be optically active when perturbed by a dissymmetric field originating from the adsorbates.9 Furthermore, very recent studies by Tsukuda and co-workers10 and Bu¨rgi and co-workers,11 in which the former represents the study of optically active undecagold cluster compounds capped by BINAP (BINAP denotes a phosphine ligand of (R)-/(S)-2,2′-bis(diphenylphosphino)1,1′-binaphthyl) and the latter that of optically active N-isobutyryl-D-/L-cysteine-protected gold nanoclusters, discuss the origin of optical activity in the context of the chiral metal surface deformation. Although the induction of chirality on a cluster surface and the mechanism clarification are definitely important, the origin of observed optical activity in metal-based electronic transitions is fragmentary because they are still few examples of well-defined optically active clusters and a lack of geometrical information. We report here the synthesis and chiroptical properties of chiral D-/L-penicillamine-capped silver nanoclusters with the core size ranged in 1-3 nm. Silver nanostructures have many important applications in catalysts, transparent conducting coating, surface-enhanced Raman spectroscopy (SERS), and antimicrobial agents and, thus, a great amount of research has been completed.12 With regard to the chiroptical properties of silver nanoparticles/clusters, few observations have been reported so far. To the best of our knowledge, the CD measurement of silver nanoparticles has been reported only once before; the work of Park and co-workers.13 In their study, chiroptical response was observed after coating of large (23.5 nm) silver nanoparticles with chiral cysteine or penicillamine, and it was concluded that association between surface ligands through hydrogen bonding would be responsible for the optical activity.13 However, the particles were not size-separated, and thus, it cannot be excluded that the observed optical activity is due to a fraction of very small particles present and/or formed during the addition of the thiols into the silver nanoparticle suspensions. Despite the studies available on gold nanoclusters, the knowledge of chiroptical properties of monolayer-protected silver nanoclusters is much more unsatisfactory, so that the investiga(8) Yao, H.; Miki, K.; Nishida, N.; Sasaki, A.; Kimura, K. J. Am. Chem. Soc. 2005, 127, 15536. (9) Goldsmith, M.-R.; George, C. B.; Zuber, G.; Naaman, R.; Waldeck, D. H.; Wipf, P.; Beratan, D. N. Phys. Chem. Chem. Phys. 2006, 8, 63. (10) Yanagimoto, Y.; Negishi, Y.; Fujihara, H.; Tsukuda, T. J. Phys. Chem. B 2006, 110, 11611. (11) Gautier, C.; Bu¨rgi, T. J. Am. Chem. Soc. 2006, 128, 11079. (12) (a) Stuart, D. A.; Haes, A. J.; Yonzon, C. R.; Hicks, E. M.; Van Duyne, R. P. IEE Proc. Nanobiotechnol. 2005, 152, 13. (b) Philip, R.; Kumar, G. R.; Sandhyarani, N.; Pradeep, T. Phys. ReV. B 2000, 62, 13160. (13) Li, T.; Park, H. G.; Lee, H.-S.; Choi, S.-H. Nanotechnology 2004, 15, S660-S663.

Nishida et al.

tions on the synthesis, size separation, and chiroptical activity of the silver nanoclusters are indispensable. In the present work, we found that chiral penicillamine-capped silver nanoclusters smaller than ∼1.9 nm in core diameter showed measurable optical activity, and had much larger response than the analogous gold nanoclusters with comparable core sizes. The chiroptical properties of the silver nanoclusters are strongly dependent on stereochemistry (D/L-form) of the surface ligands and show clear mirror-image behaviors with each other. Origins of optical activity of silver nanoclusters are discussed from a viewpoint of both a dissymmetric field model and a chiral core model. Experimental Section Materials. Silver nitrate (AgNO3, 99.5%), sodium borohydride (NaBH4, >90%), methanol (GR grade), ethanol (GR grade), and toluene (GR grade) were received from Wako Pure Chemicals and used as received. Electrophoresis-grade acrylamide, N,N′-methylenebisacrylaminde (Bis), tris(hydroxymethylaminomethane) (Tris), glycine, ammonium peroxodisulfate (APS), and N,N,N′N′-tetramethyethylenediamine (TEMED) were received from Nacalai Tesque and used without further purification. Enantiopure L- or D-penicillamine (abbreviated as L-Pen (99%) or D-Pen (99%), respectively) and racemic DL-penicillamine (abbreviated as racPen (97%)) were received from Aldrich and used as received. The chemical structures of L-Pen and D-Pen are shown in Figure 1a. Pure water was obtained by an Advantec GS-200 automatic waterdistillation supplier. Synthesis of Penicillamine-Capped Silver Nanoclusters. Penicillamine-capped silver nanoclusters were prepared using a similar method to that reported previously.8 Briefly, 0.5 mmol of AgNO3 dissolved in water (0.121 M) and 0.5 mmol of penicillamine (L-, D-, or rac-Pen) were at first mixed in methanol (100 mL), followed by the addition of a freshly prepared 0.2 M aqueous NaBH4 solution (25 mL) under vigorous stirring. After further stirring for 1.5 h, addition of ethanol (300 mL) into the solution gave a brown crude precipitate. Note that prolonged storage of the solution made the nanoclusters decompose. The precipitate was then thoroughly washed with ethanol. Finally, a nanocluster powder was obtained by a vacuum-drying procedure. The silver nanocluster sample prepared by using L-, D-, or rac-Pen is called Ag-L-Pen, Ag-DPen, or Ag-rac-Pen, respectively. Polyacrylamide Gel Electrophoresis (PAGE). PAGE was conducted using a slab unit with a gel of 2 mm thickness (ATTO, AE-6200). For the stacking and the separating gels, the total contents of the acrylamide monomers were 3% (acrylamide/Bis ) 93/7) and 28% (acrylamide/Bis ) 93/7), respectively. The stacking and the separating gels were buffered at pH 6.8 and 8.7 with Tris-HCl solution, respectively. The running electrode buffer consisted of glycine (192 mM) and Tris (25 mM) solution. The as-prepared product of penicillamine-capped silver nanoclusters was dissolved in a buffer solution (pH 6.8) at a concentration of 10 mg/mL. The sample solution was loaded onto a stacking gel top and eluted for 5 h at a constant voltage mode (150 V) controlled using a power supply (ATTO, AE-8150) to achieve separation. To extract silver nanoclusters in aqueous solution, we cut out and homogenized a part of the gel containing each fraction, followed by the addition of distilled water. The gel lumps were removed by centrifugation. Instrumentation. Field emission scanning transmission electron microscopy (FE-STEM) was conducted with a Hitachi S-4800 electron microscope operated at 30 kV. FT-IR spectra were measured with a Horiba FT-720 infrared spectrophotometer. For

D/L-Penicillamine-Capped

SilVer Nanocluster

Chem. Mater., Vol. 19, No. 11, 2007 2833

Figure 1. (a) Chemical structure of L- or D-penicillamine (L-Pen or D-Pen). An asymmetric carbon is denoted by “*” in each formula. (b) FT-IR spectra of penicillamine-capped silver nanoclusters along with pure L-penicillamine. In the figure, “L”, “D”, “rac”, and “pure” denote the samples of Ag-L-Pen, AgD-Pen, Ag-rac-Pen, and pure L-Pen, respectively. (c) Typical STEM image of the as-prepared Ag-L-Pen.

the measurements, a KBr disk pellet containing the silver nanocluster powder was prepared. Absorption spectra were recorded with a Hitachi U-4100 spectrophotometer. CD spectra were recorded with a JASCO J-720 or J-820 spectropolarimeter. In the CD measurement, absorption spectrum was simultaneously recorded. Rectangular 1 cm cuvettes made of quartz were used for the measurements. Determination of the Size Distribution of Silver Nanoclusters. Size distributions of silver nanoclusters were determined by a smallangle X-ray scattering (SAXS) technique in solution, where the details were described elsewhere.14,8 Briefly, the SAXS profiles of silver nanoclusters dispersed in solution were first measured, followed by analyzing the profiles on the basis of the assumption that the size distribution of spherical clusters is approximated by the Γ-distribution function. In a solution-phase SAXS measurement, the X-ray scattering is dominated by the metal cores having a large electron density, so that we can determine the core size distribution (14) (a) Sasaki, A. Rigaku J. 2005, 22, 31. (b) Nagao, O.; Harada, G.; Sugawara, T.; Sasaki, A.; Ito, Y. Jpn. J. Appl. Phys. 2004, 43, 7742. (c) SAXS measurements were conducted with a Rigaku ATX diffractometer with Cu Ka irradiation using a rotating anode X-ray generator. (d) The SAXS analyses are based on the assumption that spherical nanoclusters are distributed with the simple Γ-distribution function, so that an obtainable size distribution might come from a deviation from the ideal spherical shape of small clusters. The relatively broad size dispersion of compound 7 comes from a diffusive cluster distribution in the separating gel. However, note that it may be possible to have a small size distribution in the fractioned gel band; for example, it is reported that Au10(SG)10, Au11(SG)11, and Au12(SG)12 appeared in a single gel band, where SG denotes the glutathione ligand (Negishi, Y.; Nobusada, K.; Tsukuda, T. J. Am. Chem. Soc. 2005, 127, 5261).

of silver nanoclusters by this method. Note that the Γ-distribution function (P(D)) of the cluster with a diameter D is given by P(D) )

( ) (

1 M Γ(M) D0

Γ(M) )

M

∫

∞

0

)

D exp - M DM-1 D0

(1)

xM-1 e-xdx

(2)

where D0 and M are the mean diameter of the cluster core and the shape parameter that relates to the dispersion of the estimated diameter, respectively. Γ(M) is the gamma function. Hence, D0 and M are determined on the basis of the SAXS profile analysis.

Results and Discussion Surface Properties of Penicillamine-Capped Silver Nanoclusters. As-prepared sample analyses elucidate the chemical or surface properties of the penicillamine-capped silver nanoclusters. On the basis of the IR spectra (Figure 1b), we have revealed that L-, D-, or rac-Pen chemisorbed on silver surfaces as thiolates by forming an Ag-S bond (as revealed by the disappearance of the S-H stretch mode of penicillamine at ∼2570 cm-1).8 The protecting amino acid on nanocluster surfaces is present in the carboxylate form (as revealed by the appearance of two characteristic peaks of νs(COO-) and νas(COO-) at ∼1390 and ∼1585 cm-1, respectively).8 The appearance of the N-H bending mode (∼1670 cm-1) as well as the absence of the N+-H asymmetric vibration (∼2610 cm-1) indicate the existence

2834 Chem. Mater., Vol. 19, No. 11, 2007

Figure 2. Photographs of the PAGE separation. Seven bands are labeled in the order of mobility of nanoclusters (1 being most mobile): (a-c) results for Ag-L-Pen, Ag-D-Pen, and Ag-rac-Pen samples, respectively; (d) separation of L-Pen-capped gold nanoclusters. The sizes of the separated compounds i, ii, and iii have already been estimated to be 0.57, 1.18, and 1.75 nm, respectively. See ref 8.

of primary NH2 groups.15 According to a typical STEM image of the as-prepared Ag-L-Pen sample (Figure 1c), particles with diameters of 2-4 nm are discernible in the image. Similar images could be taken for the as-prepared Ag-D-Pen and Ag-rac-Pen samples. These results indicate a direct evidence for nanoparticle or nanocluster formation. However, we could not determine the cluster size distributions from the STEM observations, because under the influence of the STEM electron beam irradiation, they were observed to grow and coalesce into larger particles. The particle growth is probably attributed to an electric-fieldenhanced diffusion of small clusters.16 Size Separation and Determination. The nanoclusters can be negatively charged and separated according to their size and charge by PAGE. Photographs of a typical PAGE separation for Ag-L-Pen, Ag-D-Pen and Ag-rac-Pen samples, along with L-Pen-capped gold nanoclusters, are shown in Figure 2. Note that the L-Pen-capped gold nanoclusters were prepared in a way similar to that described in the experimental section and reported previously.8 In the silver nanoclusters (Figures 2a-2c), many bands are observable in the gel under normal illumination for each sample, but the appearance of discrete bands suggests the presence of magic number compounds. The three higher-mobility bands in each sample showed the strictly identical positions with each other, whereas the positions of other bands were somewhat scattered. We then selected (or separated) seven isolated bands located at the same positions with each other. This means that the silver nanoclusters to be fractioned are composed of similar size and chemical components.7,8,11 On the basis of the electrophoretic mobility of the silver nanocluster compounds, the separated compounds are re(15) (a) Silverstein, R. M.; Bassler, G. C.; Morril, T. C. Spectrometric Identification of Organic Compounds, 6th ed.; John Wiley & Sons: New York, 1997. (b) Yanagimoto, H.; Akamatsu, K.; Gotoh, K.; Deki, S. J. Mater. Chem. 2001, 11, 2387. (16) Ito, Y.; Jain, H.; Williams, D. B. Appl. Phys. Lett. 1999, 75, 3793.

Nishida et al.

ferred to as compounds 1-7, with 1 being the most mobile yellowish species. In the following, most attention is devoted to the five representative cluster compounds 1, 2, 3, 5, and 7.17 When distinguishing these compounds in regard to the stereochemistry of the surface ligands, a suffix L, D, or rac is added at the end the compound number; for example, 1L, 2L, ... for the Ag-L-Pen nanocluster sample. Size distributions of the fractioned silver nanoclusters were determined by a solution SAXS measurement. For compounds 1-7, Ag-L-Pen, Ag-D-Pen, and Ag-rac-Pen samples exhibited band positions identical to each other in the separating gel, so we conducted the size analysis representatively for the Ag-L-Pen sample. Figure 3a shows the experimental scattering profiles of the components 2L, 3L, 5L, and 7L along with the simulated curves. The simulated curves well-reproduce the profiles using the Γ-distribution function. Note that the scattering intensity for compound 1 was not enough to be satisfactorily analyzed.18 In compounds 2-7, the obtained parameters are as follows: for 2L, D0 ) 1.30 nm (M ) 14); for 3L, D0 ) 1.47 nm (M ) 21); for 4L, D0 ) 1.60 nm (M ) 22), for 5L, D0 ) 1.86 nm (M ) 40); for 6L, D0 ) 2.20 nm (M ) 18); for and 7L, D0 ) 2.95 nm (M ) 23). The core size distributions are depicted in Figure 3b. As expected, compound 2L is the smallest. Hence, these analyses indicate that the compounds were separated by differing core size (with the smallest having the highest mobility). In PAGE separation of organic macromolecules such as proteins using SDS (sodium dodecyl sulfate), the electrophoretic mobility in a gel is correlated to the macromolecule’s size (molecular weight), shape, and charge.19 The SDS minimizes the shape and charge components by binding to proteins. Consequently, the shape and charge density become invariant, making migration a function solely of size (or molecular weight) through the sieve effect, that is, species with diameters that are too large to pass through a gel pore are effectively sieved. Under these conditions, a linear relationship between the logarithm of size and the relative mobility (Rf value; the ratio between the distance that a compound has moved from its starting to its final point after a fixed period of time and that the elluent has moved) can be obtained.19 Similarly, the spherical metal cluster migration during the gel electrophoresis depends both on the size and charge of the clusters, the latter being dependent on the surface composition.7 If the surface charge density for each separated nanocluster is invariant, a similar analysis can be applied. Figure 4 shows the relationship between the logarithm of the mean core diameter and the compound mobility expressed by Rf. The excellent linear relationship between log(D0) and Rf suggests that surface charge densities of fractioned silver nanoclusters are almost identical to each other. Assuming here that the surface charge density of the (17) In compounds 4-6, featureless absorption spectra were obtained. To distinctly show the trend in spectroscopic properties as a function of size, we have selected compound 5 as representative of the three. Ordinary absorption and CD responses (anisotropy factor) of 4 and 6 are shown in the Supporting Information. (18) We made an effort to analyze a weak SAXS profile of compound 1L, but the obtained mean cluster diameter fluctuated from 0.4 to 1.5 nm. (19) Dube, S.; Flynn, E. Focus 1998, 20, 24.

D/L-Penicillamine-Capped

SilVer Nanocluster

Chem. Mater., Vol. 19, No. 11, 2007 2835

Figure 3. (a) X-ray scattering intensity profile of each numbered silver nanocluster compound. The experimental and the simulated profiles are shown by dots and curves, respectively. (b) Obtained cluster size distributions of each numbered compound.

compound 1 is also the same as the other nanocluster compounds, we may estimate the size of 1L to be 1.05 nm by extrapolation in Figure 4.18 Figure 2d shows the result of PAGE separation for the analogously L-Pen-capped “gold” nanoclusters under the same separation condition. We have already established that the mean core diameter of gold nanoclusters extracted from band i, ii, or iii is determined to be 0.57, 1.18, or 1.75 nm, respectively.8 Comparing the electrophoretic mobility between similar-sized penicillamine-capped silver and gold nanoclusters (for example, compare bands 3 and ii located at a similar position; 3 is 1.47 nm and ii is 1.18 nm in diameter), the silver clusters have higher mobility than gold, indicating that the surface charge density of silver nanoclusters is higher than that of gold nanoclusters.20a In the selfassembled monolayer of alkanethiol on the (111) gold and

silver surfaces, a dense commensurate structure is observed for the thiolate head groups on silver (a lattice constant of ∼0.467-0.477 nm) and commensurate lattice with a larger intersite spacing of ∼0.50 nm on gold.21 Hence, the difference in the surface charge density primarily comes from that in the surface number density of thiols on silver and gold.20b (20) (a) Negishi, Y.; Takasugi, Y.; Sato, S.; Yao, H.; Kimura, K.; Tsukuda, T. J. Phys. Chem. B 2006, 110, 12218. (b) Precisely, a surface charge density of penicillamine-capped metal nanoclusters depends not only on the number of ligands but also on an average charge per ligand. The latter can be changed through the interaction of the charged groups in the molecule with the metal surface. Therefore, if such interactions are different between gold and silver, the difference in the average charge per ligand may contribute to that in the surface charge density. (21) (a) Ulman, A. Chem. ReV. 1996, 95, 1533. (b) Sellers, H.; Ulman, A.; Shnidman, Y.; Eilers, J. E. J. Am. Chem. Soc. 1993, 115, 9389. (c) Zharnikov, M.; Frey, S.; Rong, H.; Yang, Y.-J.; Heister, K.; Buck, M.; Grunze, M. Phys. Chem. Chem. Phys. 2000, 2, 3359.

2836 Chem. Mater., Vol. 19, No. 11, 2007

Figure 4. Plot of the log of cluster diameter versus the mobility represented by the Rf value. An excellent linear fit could be obtained (The line shows the best fit using least-squares method.)

Absorption Spectroscopy. In silver nanoparticles smaller than ∼5 nm, there is some controversy regarding the effect of size on the UV-vis absorption spectrum. For example, a spectral red shift and broadening of the surface plasmon band has been observed with decreasing size of particles (from 3.7 to 1.5 nm).22a,b In tiopronin-protected silver nanoparticles, a blue shift of the surface plasmon band has been detected when the size decreases (from 4.8 to 1.6 nm),22c and the phenomenon is discussed in terms of the Ag-S bond’s electron donation to the nanoparticle.22d In triethylamine/ dodecanethiol-capped silver nanoparticles, disappearance of the common surface plasmon band and new appearance of the quadrupole resonance band (∼350 nm) have been observed when the average size decreases from 4.5 to 2.5 nm.22e,f Furthermore, it is reported that silver nanoparticles of 2.5 nm embedded in silica exhibited no plasmon band.22g These observations suggest the complex sensitivity of small silver nanoparticles to their environments caused by the interface- or surface-related interactions.23 Indeed, the metalsurroundings (including surface ligands) interaction introduces a shift in the energy of the metal valence orbitals and thereby changes the electron density of the metal nanoparticles, which would modify the electronic properties of the nanoparticles and affect their optical absorption.24 Hence the discussion on the absorption spectra of our silver nanoclusters will offer further insight in the size effect on their electronic transitions. (22) (a) Wilcoxon, J. P.; Martin, J. E.; Provrncio, P. J. Chem. Phys. 2001, 115, 998. (b) Pivin, J. C.; Garcı´a, M. A.; Hofmeister, H.; Martucci, A.; Vassileva, M. S.; Nikolaeva, M.; Kaitasov, O.; Llopis, J. Eur. Phys. J. D 2002, 20, 251. (c) Huang, T.; Murray, R. W. J. Phys. Chem. B 2003, 107, 7434. (d) Malinsky, M. D.; Kelly, K. L.; Schatz, G. C.; Van Duyne, R. P. J. Am. Chem. Soc. 2001, 123, 1471. (e) Chaki, N. K.; Sudrik, S. G.; Sonawane, H. R.; Vijayamohanan, K. Chem. Commun. 2002, 76. (f) Jin, R. C.; Cao, Y. W.; Mirkin, C. A.; Kelley, K. L.; Schatz, G. C.; Zheng, J. G. Science 2001, 294, 1901. (g) Weiping, C.; Lide, Z. J. Phys. Cond. Mater. 1997, 9, 7257. (23) Haynes, C. L.; Van Duyne, R. P. J. Phys. Chem. B 2001, 105, 5599. (24) (a) Yang, L, Li, G. H.; Zhang, L. D. Appl. Phys. Lett. 2000, 76, 1537. (b) Ho¨vel. H.; Fritz, S.; Hilger, A.; Kreibig, U.: Vollmer, M. Phys. ReV. B 1993, 48, 18178.

Nishida et al.

Figure 5 shows the UV-vis absorption spectra of the separated compounds 1, 2, 3, 5, and 7 in aqueous solution.17 The first derivatives of these spectra are calculated as shown in Figure 6 to identify the spectral fine structures more clearly. In L-Pen- and D-Pen-capped nanoclusters (panels a and b of Figure 5), the spectral shapes of the separated compounds are almost identical within the same compound numbers, supporting that each numbered compound has similar core size, geometry, and chemical compositions.25,26 The fact is supported by the identical patterns of the first derivative spectra within the same compound numbers (Figure 6). Note that the hump observed at ∼480 nm for compound 2L or 2D tends to disappear within 24 h, and thus, the intensity of the hump was fluctuated from measurement to measurement. Time dependence of the absorption spectrum of compound 2L is shown in the Supporting Information, proving that the compound 2 is relatively unstable. The spectra of compounds 1-3 (1L-3L or 1D-3D) are rather structured, whereas that of compound 5 is less pronounced. Compound 7 solely shows a broad surface plasmon band at around 400 nm. In addition, clear absorption onsets appeared in the vis-near IR region for compounds 1 and 2: ∼600 nm for 1 and ∼710 nm for 2.27 These results indicate that the electronic structures of compounds 1 and 2 are in a quantum confinement regime and better viewed as nonmetallic clusters rather than metallic particles covered by thiols.28 Therefore, it is expected that metal to nonmetal transition in thiolate-capped silver nanoclusters occurs in the 1.3-1.5 nm range. In rac-Pen-capped silver nanoclusters, although the PAGE band positions of the fractioned compounds were identical with those of the l-Pen- or D-Pen-capped samples, the spectral shapes involving peak or shoulder positions of the compounds 1rac-3rac were slightly different from those of 1L-3L (or 1D-3D), respectively. These reproducible features can be confirmed by the patterns of the first derivative spectra of the fractioned nanocluster compounds 1-3; that is, the elliptically marked regions in Figure 6 exhibit slight but distinct differences in the optical response among the respective nanocluster compounds. For compounds 5 (5L, 5D, and 5rac) and 7 (7L, 7D, and 7rac), on the other hand, their spectral shapes were quite similar to each other. It is important to note that such behaviors are substantially different from those for the penicillamine-capped gold nanoclusters we have investigated.8 Hence, rac-Pen-capped silver nanoclusters are not the equimolar mixture of two homochiral (L-Pen-capped and D-Pen-capped) silver nanoclusters, and the core geometry in compounds 1rac-3rac, which influences the optical properties of metal nanoclusters,26 may be different from that in the corresponding (25) It has been demonstrated that monolayer-protected gold nanoclusters differing by one ligand can be separated using PAGE; in this case, absorption spectra of the two separated compounds showed the same absorption onset (Negishi, Y.; Nobusada, K.; Tsukuda, T. J. Am. Chem. Soc. 2005, 127, 5261). Using similar arguments, it is quite reasonable that compound 1 has a different (and smaller) core size fthan those of the other numbered compounds. (26) (a) Sosa, I. O.; Noguez, C.; Barrera, R. G. J. Phys. Chem. B 2003, 107, 6269. (b) Noguez, C. Opt. Mater. 2005, 27, 1204. (27) In compound 2, the position of the absorption onset remains unchanged with the passage of time. See the Supporting Information. (28) Nobusada, K. J. Phys. Chem. B 2004, 108, 11904.

D/L-Penicillamine-Capped

SilVer Nanocluster

Chem. Mater., Vol. 19, No. 11, 2007 2837

Figure 5. (a-c) Absorption spectra of the separated nanocluster compounds for Ag-L-Pen, Ag-D-Pen, and Ag-rac-Pen samples, respectively. For clarity, the spectra of 2, 3, 5, and 7 are off-set by adding a constant.

Figure 6. First derivatives for the absorption spectra of fractioned L-Pen-, D-Pen-, and rac-Pen-capped silver nanoclusters. The spectra of the L-Pen- and rac-Pen-capped clusters were off-set by adding a constant.

homochiral silver nanoclusters. This would be the result of statistical (or random) ligation of the thiols onto a silver cluster surface (that is, the adsorption probability of D-form thiols onto the surfaces is equal to that of L-form isomers),

as has been proved for chiral cysteine adsorption on a gold surface.29 Chiroptical Properties. To obtain information on chiroptical properties of the present silver nanoclusters, we

2838 Chem. Mater., Vol. 19, No. 11, 2007

Nishida et al.

Figure 7. (a) CD spectra of each numbered silver nanocluster compound. Mirror image relationship in the CD signals between each numbered L-Pencapped (green curve) and D-Pen-capped (red curve) silver nanoclusters can be seen. No CD signals were obtained for any of the rac-Pen-capped silver nanoclusters (black curve).

measured the CD spectrum of each numbered compound (Figure 7). For homochiral penicillamine-capped silver nanoclusters, ligand chirality had a dramatic effect on the CD spectra of all the clusters. Note that optically active penicillamine contributes to the CD signal only in the UV region in the wavelength shorter than ∼270 nm. (The ordinary absorption and CD spectra of penicillamine enantiomers are shown the Supporting Information.) The pairs of CD spectra of 1L/1D, 2L/2D, 3L/3D, and 5L/5D show measurable Cotton effects and an almost perfect mirror-image relationship in the range of metal-based electronic transitions, showing that enantiomeric ligands can produce the corresponding enantiomeric silver nanoclusters. This finding indicates that the separated silver nanoclusters have welldefined stereostructures as common molecules do. Without (29) Ku¨hnle, A.; Linderoth, T. R.; Hammer, B.; Besenbacher, F. Nature 2002, 415, 891.

the structural information on the silver clusters, it would be hard to assign the observed CD spectral bands; however, it is conceivable that the Cotton effect in CD spectra is induced through the different (quantized) transitions and their interactions in the silver clusters.8 In the largest clusters of 7L/7D (D0 ) 2.95 nm), on the other hand, almost no CD signals were detected. The result disagrees with that previously reported, showing that very large silver nanoparticles (∼23.5 nm) had measurable CD responses by the addition of chiral penicillamine.13 However, these large particles were not sizeseparated and it cannot be excluded that the observed optical activity might come from a fraction of small clusters formed during the addition of penicillamine to the silver colloids and/or minor species already present in the samples. Indeed, an addition of thiols to metal cluster solution causes the optical changes in both Ag and Au clusters due to a size etching effect, giving rise to a reduced average size.22a

D/L-Penicillamine-Capped

SilVer Nanocluster

Chem. Mater., Vol. 19, No. 11, 2007 2839

Figure 8. Anisotropy factors (g-factors) for each numbered silver nanocluster compound. The solution pH was measured to be ∼9.5. The green and red curves represent the L-Pen-capped and D-Pen-capped silver nanoclusters, respectively. Note that the maximum scale in the ordinate is 0.002 for compounds 2 and 3, whereas that for the other compounds is 0.001.

For rac-Pen-capped silver nanoclusters, no CD signals could be detected in any fractioned sample. An equimolar mixture of two homochiral nanoclusters can cancel out the net optical activity; however, statistical adsorption of L- and D-penicillamine onto the silver surface makes the situation implausible.29 Hence, the surface ligand rac-Pen would cover the silver nanocluster surface statistically, resulting in the disappearance of circular dichroism. To sharply elucidate the size effect on the optical activity of chiral penicillamine-capped silver nanoclusters, we calculated the anisotropy factor (or g-factor), defined as g ) ∆/ ()intensity of the molar dichroic absorption ∆ normalized to the extinction coefficient ), for the separated compounds 1-3, 5, and 7. The results are shown in Figure 8. Typical anisotropy factors varied from ∼1 × 10-3 (in compound 1) to ∼1 × 10-5 (in compound 5) as the nanocluster core size increased from ∼1.05 to ∼1.86 nm, respectively. Finally, the chiroptical response in the metalbased transitions disappeared at the core size of ∼2.95 nm (in compound 7).

Chiroptical Origins: Comparison between Penicillamine-Capped Silver and Gold Nanoclusters. Perhaps the most striking feature of the present study is the observation of very intense optical activity of penicillamine-capped silver nanoclusters compared to that of the corresponding gold nanoclusters. Comparison of maximum anisotropy factors between the homochiral penicillamine-capped silver and gold nanoclusters are summarized in Figure 9. For the gold nanoclusters, the data were collected from our previous study.8 The magnitude of the anisotropy factors is similar to each other for the silver and gold nanoclusters with the core size larger than ∼1.5 nm. Interestingly, it steeply increases at the size smaller than ∼1.5 nm for silver nanoclusters, whereas that of gold nanoclusters increases gradually. Consequently, the anisotropy factors of chiral penicillamine-capped silver nanoclusters were about several-fold (one order magnitude in maximum) larger than those of the corresponding gold nanoclusters with comparable core sizes (5.31,33 Hence the gauche form of penicillamine rotamer is expected to be the most favorable one, but other rotational conformers (anti forms) also coexist.31 In our CD measurements, the data were collected at pH ∼9.5, so that the COO- and/or NH2 groups can be involved in the coordination to the silver surface.34 This remains the pos(30) (a) Stewart, S.; Fredericks, P. M. Spectrochim. Acta, Part A 1999, 55, 1641. (b) Watanabe, T.; Maeda, H. J. Phys. Chem. 1989, 93, 3258. (c) Lee, H.; Kim, M. S.; Suh, S. W. J. Raman Spectrosc. 1991, 22, 91. (31) Lo´pez-Ramı´rez, M. R.; Arenas, J. F.; Otero, J. C.; Castro, J. L. J. Raman Spectrosc. 2004, 35, 390. (32) (a) Craff, M.; Bukowska, J. J. Phys. Chem. B 2005, 109, 9567. (b) Xu, Q.-M.; Wan, L.-J.; Wang, C.; Bai, C.-L.; Wang, Z.-Y.; Nozawa, T. Langmuir 2001, 17, 6203. (33) So´va´go´, I. Handbook of Metal-Ligand Interactions in Biological Fluids; Berthon, G., Ed.; Marcel Dekker: New York, 1995; p 627. The acid dissociation constants (pKa) of D-penicillamine are 10.69 (for NH3+) and 1.90 (for COOH). (34) It has been elucidated that conformation of the carboxylate group is readily changeable through CR-Cβ bond rotation at highly acidic conditions (Brolo, A. G.; Germain, P.; Hager, G. J. Phys. Chem. B 2002, 106, 5982). We then tried to measure chiroptical responses of the silver nanoclusters at a highly acidic condition (pH ∼2) to check the effect of surface penicillamine rotamers. Unfortunately, the extracted silver nanoclusters were very unstable and formed agglomerates, so we could not measure any optical properties at that pH.

D/L-Penicillamine-Capped

SilVer Nanocluster

sibility of chiral surface atom distortion caused by the interaction of Ag-carboxylate or Ag-amine group. We next discuss the observed chiroptical response on the basis of the dissymmetric field model. In scheme (c), formation of a chiral arrangement or adsorption pattern is necessary, and the pattern of the adsorbates does not influence the absorption spectrum but the CD responses.9 Such a pattern has the non-superimposability of an entire assembled structure with its mirror image. On the basis of the absorption spectroscopy, we could suppose that rac-Pencapped silver nanoclusters may have different core geometry from L-Pen- and D-Pen-capped nanoclusters (especially in a size region smaller than ∼1.5 nm). It would probably be due to an unfavorable adsorption pattern caused by the statistical/kinetic ligation of the racemic penicillamine.20 Under the conditions, homochiral ligand-capped silver nanoclusters are able to have specific and sterically stable adsorption patterns regardless of the fact that the silver cores are chiral or achiral. A chiral pattern of the adsorbate is then also possible. In scheme (d), when the ligand contains asymmetric atoms, optical activity of the species attached to the ligand is universally induced from the chiral field transmission through space and by way of the chemical bonds linking the asymmetric center to the chromophore.35a The order of magnitude of the vicinal effect depends on the number of atoms between the asymmetric center and the target species, and on the efficiency of the substituents in transmitting dissymmetry.35b We then think that deduction from the discussion on chirality for transition metal complexes would be helpful. In transition metal complexes, it is well-known that the anisotropy factor originated from the conformational effect (that is, effect of the chelate-ring puckering that can be considered as one of the chiral adsorption patterns) is about one or two order(s) of magnitude larger than that from the vicinal effect: For example, a ∆ or Λ isomer of [Co(en)3]3+ (en: ethylenediamine as a bidentate ligand) exhibit a very large g-factor of ∼2 × 10-2 in d-d electronic transitions.36 The ∆ or Λ isomer of [Co(L-ala)3]3+ (L-ala: basic L-alanine as a bidentate ligand) has a maximum g-factor similar to that of [Co(en)3]3+,37 suggesting that the conformational effect has a dramatic effect on the optical activity of transition metal complexes with chelate rings.38 On the other hand, the maximum anisotropy factors of metal complexes [Co(NH3)4(L-ala)]3+ and [Co(NH3)5(L-alaH)]3+ (L-alaH: zwitterionic L-alanine as a unidentate ligand) are ∼4 × 10-3 and ∼2 × 10-4, respectively, meaning that bidentate ligation improves the optical activity. This trend is explained within the context of the vicinal effect, and the increase is due to suppression of conformational degrees of freedom of the amino acid ligands.37 In the penicillamine-capped silver nanoclusters that behave like a sterically regulated molecular complex, their anisotropy factors show magnitudes similar (35) (a) Pessoa, J. C.; Gajda, T.; Gillard, R. D.; Kiss, T.; Luz, S. M.; Moura, J. J. G.; Tomaz, I.; Telo, J. P.; To¨ro¨k, I. J. Chem. Soc., Dalton Trans. 1998, 3587. (b) Sigel, H.; Martin, R. B. Chem. ReV. 1982, 82, 385. (36) McCaffery, A. J.; Mason, S. F. Mol. Phys. 1963, 6, 359. (37) Yasui, T.; Hidaka, J.; Shimura, Y. Bull. Chem. Soc. Jpn. 1966, 39, 2417. (38) Denning, R. G.; Piper, T. S. Inorg. Chem. 1966, 5, 1056.

Chem. Mater., Vol. 19, No. 11, 2007 2841

to those in the transition metal complexes such as [Co(NH3)4(L-ala)]3+ or [Co(NH3)5(L-alaH)]3+, so the vicinal effect is expected to significantly contributes to the origin of optical activity for such silver nanoclusters. However, unlike the analogous gold nanoclusters, the silver clusters smaller than ∼1.5 nm exhibited a steep enhancement in the anisotropy factor or optical activity (Figure 9), suggesting that other mechanisms mentioned above could be involved. Further investigations related to the surface ligand reactions are required and underway in our laboratory. In any case, as a material design principle in cluster science, small silver nanoclusters covered with chiral adsorbates are a significant candidate for the promising optically active materials. Conclusion In conclusion, we have prepared a pair of optical isomers of silver nanoclusters protected by L/D-penicillamine. The clusters were separated into seven components by polyacrylamide gel electrophoresis according to their charge and size into well-defined compounds. Each fractioned component has the mean diameter of 1.05, 1.30, 1.47, 1.60, 1.86, 2.20, or 2.95 nm. The separated compounds of silver nanoclusters show intense optical activity with opposite sign when covered with L- or D-enantiomer of penicillamine (mirror-image relationship), indicating that the nanoclusters have a welldefined stereostructure as common chiral molecules do. Especially, the chiroptical response of the silver clusters is several-fold larger than that of the analogous gold nanoclusters with comparable core size. No CD signals could be obtained for the racemate-capped silver nanoclusters. With a decrease in the mean cluster diameter, the optical activity or anisotropy factor increased. In addition, a steep increase in the optical activity was observed when the diameter became smaller than ∼1.5 nm. The vicinal effect from asymmetric carbons in the ligands should be the universal origin for the optical activity of the silver nanoclusters, but its enhancement observed in a small size region might be attributed to the additional contributions from a chiral silver core, chiral adsorption pattern, and/or surface chiral distortion. The field of optically active nanoclusters is a very broad one ranging from cluster science to surface science, so that we believe that our findings will give a new development of this fascinating subject in nanoscience. Acknowledgment. Prof. T. Okuyama, Prof. T. Sugimura, and Dr. M. Fujita (University of Hyogo) are acknowledged for allowing us to use the CD spectrophotometer. The present work was financially supported by a Grant-in-Aid for Scientific Research from MEXT (16101003), Scientific Research in Priority Areas: Application of Molecular Spins (15087210), and a grant from the Kawanishi Memorial Shinmeiwa Education Foundation. Supporting Information Available: Absorption spectra of compound 2L as a function of time, ordinary absorption, CD spectra of pure chiral penicillamine in water, and absorption and CD response (anisotropy factor) of the separated compounds 4 and 6 (PDF). This material is available free of charge via the Internet at http://pubs.acs.org. CM0700192