Thomas
Bürgi
Department of Physical Chemistry, University of Geneva, Quai Ernest-Ansermet 30, 1211 Geneva 4, Switzerland. E-mail: thomas.buergi@unige.ch
First published on 7th September 2015
The gold–sulphur interface of self-assembled monolayers (SAMs) was extensively studied some time ago. More recently tremendous progress has been made in the preparation and characterization of thiolate-protected gold clusters. In this feature article we address different properties of the two systems such as their structure, the mobility of the thiolates on the surface and other dynamical aspects, the chirality of the structures and characteristics related to it and their vibrational properties. SAMs and clusters are in the focus of different communities that typically use different experimental approaches to study the respective systems. However, it seems that the nature of the Au–S interfaces in the two cases is quite similar. Recent single crystal X-ray structures of thiolate-protected gold clusters reveal staple motifs characterized by gold ad-atoms sandwiched between two sulphur atoms. This finding contradicts older work on SAMs. However, newer studies on SAMs also reveal ad-atoms. Whether this finding can be generalized remains to be shown. In any case, more and more studies highlight the dynamic nature of the Au–S interface, both on flat surfaces and in clusters. At temperatures slightly above ambient thiolates migrate on the gold surface and on clusters. Evidence for desorption of thiolates at room temperature, at least under certain conditions, has been demonstrated for both systems. The adsorbed thiolate can lead to chirality at different lengths scales, which has been shown both on surfaces and for clusters. Chirality emerges from the organization of the thiolates as well as locally at the molecular level. Chirality can also be transferred from a chiral surface to an adsorbate, as evidenced by vibrational spectroscopy.
Fig. 1 Top: model of the c(4 × 2) phase of a SAM in which each molecule is schematically drawn as circle. The two types of shading represent molecules that are distinct within the c(4 × 2) unit mesh (rectangle). Molecules 1 and 2 are equivalent but distinct from molecules 3 and 4. Reproduced from ref. 31 with permission from Elsevier. Bottom: cross section of a SAM formed from decanethiol. Reprinted with permission from ref. 24 Copyright (2005) American Chemical Society. |
Fig. 2 Top: schematic drawing of a monomeric Au–SR–Au staple motif. The organic substituent R of the thiolate can be oriented on the same side with respect to the staple plane (cis) or on opposite sides (trans). Bottom: stereochemistry of monomeric and dimeric staple motifs. Reprinted with permission from ref. 34 Copyright (2014) American Chemical Society. |
The structure of the staple motifs resembles the one of Au(I)–thiolate polymers that are formed when a Au(III) salt is mixed with a thiol during the preparation of thiolate-protected gold clusters according to the Brust synthesis.7 Such polymers are used as therapeutic agents against rheumatoid arthritis (myochrysine) for more than halve a century. X-ray structure determination of gold(I)–thiomalate single crystals revealed a gold–sulphur double helical structure.35 Each helix shows approximate four-fold symmetry. Left- and right-handed helices are found in the unit cell. The left-handed helix contains exclusively (S)-thiomalate, whereas the right-handed helix is built from (R)-thiomalate. In this structure the S–Au–S angle is 178.9° and 169.4° for the two distinguishable gold atoms in the unit cell, whereas the Au–S distances are 2.289 Å and 2.285 Å, respectively. It was furthermore suggested, based on NMR and wide angle X-ray scattering (WAXS)/differential anomalous scattering (DAS), that oligomers of similar structure exist in solution. The structural parameters given above are close to the ones determined for typical staple motifs of thiolate-protected gold clusters. For example for the Au38(PET)24 (PET = 2-phenylethylthiolate) cluster the average S–Au–S bond angle is 168.6° and the average Au–S bond length in the staple is 2.3 Å.11
The staple motifs described above are the most abundant structure elements found up to now in thiolate-protected gold clusters. More recent structure determinations indicate the presence of other binding motifs such as trimeric motifs SR(–Au–SR)3,13 bridging thiolates,12,16 cyclic –Au–SR– structures13 or even naked sulphur atoms.14 However, sometimes it is not straightforward to clearly distinguish between these binding motifs on the highly curved surface of a cluster.
Fig. 3 STM image of self-assembled structures of methylthiolate at low coverage on Au(111), produced by heating the gold crystal predosed with CH3SSCH3 above 200 K for ∼10 min (a). Highlighted are two trans-(CH3S)2Au complexes and adjacent cis- and trans-adatom complexes. Their schematic models are shown in (b) and (c), respectively. Reprinted with permission from ref. 21 Copyright (2009) American Chemical Society. |
The creation of gold adatoms is consistent with the observation of etch pits (vacancy islands) on gold surfaces exposed to thiols. For SAMs of long-chain alkanethiols the gold–sulphur interface is buried and precludes direct observation by scanning probe techniques. Structure determination remains therefore a real challenge and the same is true for calculations, since different structural models have similar energy22 and a key role may be played by intermolecular interactions between the alkyl chains. Such interactions represent a real challenge for density functional theory (DFT) calculations. Using grazing incidence X-ray diffraction (GIXRD) the structure of a hexanethiolate SAM prepared in UHV on Au(111) at high coverage was reinvestigated.23 Best-fit conditions were found for a model with adatoms and vacancies in the topmost Au layer. In this model both vacancies and adatoms are delocalized (partially occupied atomic sites), which reflects the dynamic character indicated by density functional theory based molecular dynamics simulations. The dynamic nature of the interface stems from the interconversion between two different thiol adsorption geometries as well as vacancy migration. The structure derived from the fit to the experiment is therefore an average structure consisting of one-dimensional zigzag chains –S–Au–S–Au–S– as well as thiolates adsorbed in a bridged adsorption mode. The chains resemble the staple motifs found on several thiolate-protected gold clusters10 and the structure of gold–thiolate polymers.35
The X-ray diffraction studies mentioned above provide averaged and static information and do not directly probe the dynamic aspect of the gold–thiolate interface. However, the appearance of partially occupied sites in the best fit model at least indicates some disorder in the Au–S interface and points towards the dynamic nature of the structure.
The diffusion of thiolates on gold was also studied by cyclic voltammetry.45 A phase-separated SAM of two different thiolates shows a different cyclic voltammogram compared to a homogeneously mixed SAM of the same thiolates. Starting from artificially separated domains of 1-undecanethiol (UDT) and 11-mercaptoundecanoic acid it took over 300 hours at 60 °C to convert into a homogeneously mixed SAM. The authors attributed the change in mixing state to the lateral diffusion of the thiolates on gold, since desorption of thiolates into water was considered unlikely. Similarly, a change in mixing state was also observed when annealing the SAM at 60 °C in air. From these experiments a diffusion coefficient D of 10−18 cm2 s−1 was roughly estimated at 60 °C. This is slow, on the order of 1 nm per hour lateral diffusion. Using STM other researchers estimated diffusion coefficients that are roughly one order of magnitude larger for the coalescence of holes in a dodecanethiol SAM at 90 °C or for the diffusion of a single pit in a dodecanethiol SAM.46,47 It is however a priori not clear if the diffusion processes mentioned above in the STM and cyclic voltammetry studies refer to the same physical processes, i.e. diffusion of thiolates, adsorbate complexes or surface gold atoms, which might explain the order of magnitude difference in the estimated diffusion constants for the different experiments.
To sum up, there is evidence for the mobility of thiolates on flat gold surfaces. Several cases have to be distinguished, including (i) the diffusion of thiolates on otherwise free gold surfaces, (ii) the diffusion of thiolates within a fully covered surface and (iii) the diffusion of gold atoms. Process (i) is important during the formation of SAMs (low coverage) and also for microcontact printing. Process (ii) has implications for phase separation processes of mixed SAMs driven by thermodynamic factors. Process (iii) is evident from the movement of etch pits on SAMs and can be used to heal such defects. The mechanism of these processes is largely unclear and especially for processes (ii) and (iii) it is unclear whether a gold–thiolate complex is moving and, if yes, what the stoichiometry of such a complex could be.
The diffusion of thiolates on gold nanoparticles was proposed by Murray and coworkers to rationalize the kinetics of thiolate-for-thiolate exchange reactions.55 However, the rate of diffusion was proposed to be very low, resulting in typical reaction times of weeks at room temperature. Similarly, the extent of photochemical reactions, which depends on the positions of the reacting groups on the gold nanoparticle surface, were observed to be more pronounced for aged samples (after weeks), which was explained by slow lateral diffusion of ligands.56 In contrast, other work indirectly indicated fast diffusion of thiolates on gold nanoparticles.57,58 For example, the results of fluorescence titration experiments using pyrenyl chromophores to monitor the adsorption of ligands on gold nanoparticles were rationalized by equilibration of the ligands on the nanoparticle surface on the time scale of 1–2 hours.58
There is some experimental evidence of phase segregation of mixed thiolates on the surface of gold nanoparticles, driven by thermodynamic factors, which implies some mobility of the thiolates.59 Indication of phase segregation stems from MALDI (matrix-assisted laser desorption/ionization) mass spectrometry,51 scanning tunnelling microscopy (STM),59 and spectroscopy such as electron paramagnetic resonance (EPR),60 infrared (IR)61 and nuclear magnetic resonance (NMR).62 The interpretation of STM is challenging for this purpose but MALDI is an elegant way to study segregation processes on gold nanoparticles (see Fig. 4).51 In MALDI experiments of thiolate-protected gold nanoparticles fragments of composition Au4(SR)4 are often observed. These fragments are thought to be formed by the rearrangement of the staple motifs discussed above during the MALDI process. The ligands in these fragments likely stem from the same location on the nanoparticle. By analysing the distribution of fragments of mixed SR/SR′ thiolate-protected gold particles one can gain information on the distribution of the thiolates on the original cluster, which can in principle vary from completely phase separated (Janus type) to homogeneously (statistically) mixed. For phase segregated particles one would expect large abundance of Au4(SR)4 and Au4(SR′)4 fragments and low abundance of mixed fragments Au4(SR)x(SR′)4−x (x = 1…3). Using this method phase segregation was found for certain ligand combinations on 2–4 nm (core diameter) nanoparticles.51 However, the most pronounced phase separation effects were found for samples that were prepared via ligand exchange rather than direct synthesis. The lack of phase separation in nanoparticles that were prepared by the direct route, even after their heating to 55 °C for one hour indicated that mobility of the thiolates on the gold nanoparticle surface is limited at room temperature or slightly above. The pronounced phase separation via ligand exchange is in line with work on SAMs on Au(111) that showed domainwise exchange of hexadecanethiol with 12-mercaptododecanoic acid in ethanol.63
Fig. 4 Comparison of experimental (open circles) and theoretical (filled circles) ligand distributions for free gold–thiolate complexes and three mixed-ligand AuNPs obtained by ligand-exchange reactions of tiopronin AuNPs with free glutathione, 11-mercaptoundecanoic acid (MUA), or mercaptoundecyltetra-ethylene glycol (MUTEG). Deviation from the theoretical model (indicated by arrows) reveals phase-segregated gold–thiolate monolayers on AuNPs. Various ligand mixtures yield different degrees of nanophase separation. Reprinted with permission from ref. 51 Copyright (2011) John Wiley and Sons. |
Electron paramagnetic resonance (EPR) spectroscopy was also used to study lateral diffusion of thiolates on gold nanoparticles.65,66 Chechik and coworkers designed a disulphide ligand with two spin labels connected by a cleavable bridge.65 After adsorption on gold nanoparticles the bridge was broken and the initially close spin labels could diffuse away from each other, which was probed by EPR spectroscopy. The results suggested that at room temperature there is virtually no lateral mobility, whereas at 90 °C redistribution was observed in the course of several hours. The authors suggested that this redistribution is at least partially due to desorption of ligands and re-adsorption on different nanoparticles.
The studies mentioned above were performed with small nanoparticles (2–4 nm core diameter). Such samples are typically not monodisperse and may contain particles and clusters that are not infinitely stable particularly at elevated temperatures, which may interfere with the interpretation of the obtained results. A way to circumvent this drawback is the preparation of well-defined clusters, which can be obtained by synthesis and subsequent separation for example by size exclusion chromatography.67 In such samples decomposition and size evolution can readily be monitored. Direct evidence for the movement of thiolates on the surface of a gold cluster was obtained from racemization studies of a chiral Au38(PET)24 cluster. The structure of this cluster is known (see Fig. 5)11 and its enantiomers could be separated using chiral chromatography (HPLC, Fig. 5).64
Fig. 5 Top: HPLC chromatogram of the enantioseparation of racemic Au38(PET)24. The structure of the left- and right-handed cluster enantiomers is also shown viewed along the long axis of the cluster. Au core atoms: bold orange, Au atoms of the staples: orange, sulphur in the staples: green. The PET ligand is omitted for clarity. Bottom: circular dichroism (CD) spectra of the two enantiomers. Reproduced from ref. 64. |
Racemization of the cluster was evidenced both by circular dichroism (CD) spectroscopy and HPLC and the kinetics of this process at different temperatures was followed by CD spectroscopy.68 Racemization takes about half an hour at 80 °C, whereas it is very slow at room temperature. Furthermore, it is not affected by the addition of free thiol. The racemization proves the mobility of the staple motifs on the cluster surface. In the racemization process all six dimeric staple motifs (SR–Au–SR–Au–SR) of the cluster are involved. These are arranged in two chiral three-blade fans of the same handedness at the poles of the cluster (Fig. 5). During racemization both fans invert their handedness. No direct information on the mechanism of this reaction is available. Since there is some evidence for thiolate desorption from gold particles into the solution at elevated temperature (see below) a mechanism involving desorption of thiolates has to be considered. There are two arguments against such a mechanism in the discussed case: (i) addition of thiolates to the solution does not affect the kinetics, and (ii) the activation energy for the process was determined to be less than 30 kcal mol−1,68 whereas a Au–S bond is worth 40 kcal mol−1.1,69 Two possible mechanisms without complete Au–S bond breaking have been proposed, one including the sliding of thiolates of the staple motif over the cluster core surface and the other involving a series of inter-staple SN2-type reactions, where old bonds are broken and new bonds are formed simultaneously. Note that in the first mechanism (sliding) the staple motifs stay intact, whereas in the SN2 mechanism staple motifs are broken and new ones are formed, whereby thiolates exchange between different staples. A similar “crossover” mechanism, with concerted Ag–S bond breaking and formation, was proposed to explain the transformation of polymeric –Ag–SR–Ag–SR– chains in the solid of (3-methylpentane-3-thiolato)silver into (AgSR)8 molecules in solution.70 Similarly, for ligand exchange reactions calculations indicate that concerted Au–S bond formation and bond breaking could play a role.71
Racemization was also studied for Au40(PET)24.72 The enantiomers of this chiral cluster could be separated using chiral HPLC,73 but its structure has not been resolved yet by X-ray crystallography. A structural model has been proposed, which is supported by the comparison of calculated and experimental circular dichroism (CD) spectra.74 Racemization of Au40(PET)24 proceeds at about 50 °C higher temperatures compared to Au38(PET)24, although the activation energies are quite similar for the two cases. For the Au38(PET)24 cluster it was furthermore demonstrated that the incorporation of one rigid dithiol, 1,1′-binaphthyl-2,2′-dithiol (BINAS), into the ligand shell significantly affects the racemization.75 Species with composition Au38(PET)22(BINAS)1 were obtained by ligand exchange, followed by HPLC purification.76 The BINAS ligand is thought to bridge two dimeric staple units.77,78 The effect on racemization is quite drastic. Compared to the parent Au38(PET)24 cluster the incorporation of only one dithiolate increases the temperature of racemization by 40–50 °C. The effect of doping on racemization was also studied. Doping of the Au38(PET)24 cluster with two Pd atoms leads to a drastic lowering of the racemization temperature.79 The doped Pd2Au36(PET)24 cluster is thought to have a similar structure as the parent Au38(PET)24 cluster. The effect of doping on the racemization may therefore indicate the importance of electronic effects.
Fig. 6 summarizes the racemization experiments described above. The activation entropies are distinctly different for the investigated clusters. Note that the two enantiomers of the Au38 cluster, exchanged with enantiopure R-BINAS, show slightly different behaviour. This indicates that diastereomeric interactions play a role in the activation step.
Fig. 6 Comparative Eyring plot of different clusters.72,79 The C-Au38(PET)22R-BINAS (A-Au38(PET)22R-BINAS) is the clockwise (anticlockwise) Au38 cluster containing 22 2-PET and one R-BINAS ligand. |
Some more information on the mobility of thiolates on the cluster surface was obtained from ligand exchange experiments with a monothiol ([2.2]paracyclophane-4-thiol).80Fig. 7 shows that there are only four different symmetry-unique thiolates on the Au38(PET)24 cluster. Put in other words, ligand exchange can lead, in principle, to only four isomers (regioisomers) of composition Au38(PET)23(SR′)1, where SR′ is the incoming thiolate. Note that the number of isomers increases drastically with the number of exchanged ligands. When ligand exchange was stopped at short times, a mixture of un-exchanged and singly exchanged species was obtained, as verified by MALDI mass spectrometry. The same sample showed at least three additional weak peaks in the HPLC chromatogram, which were assigned to regioisomers of Au38(PET)23(SR′)1. A single regioisomer was then isolated and HPLC experiments showed that such a species is stable at room temperature and slightly elevated temperatures. This proves that the incoming ligand does not move to different symmetry-unique sites. However, at 80 °C migration of thiolates to different symmetry-unique sites was clearly observed (Fig. 7).80
Fig. 7 Top: model of Au38(PET)24 cluster. Green: Au atoms in the core, yellow: surface Au atoms, orange: sulphur atoms. The PET ligand is omitted for clarity. The letters a–d indicate the four symmetry unique sites of the cluster. The black HPLC chromatogram was obtained by performing ligand exchange with [2.2]paracyclophane-4-thiol (PCP4). The dominant peak is due to unexchanged Au38(PET)24 (enantiomer 2). The smaller peaks are due regioisomers of singly exchanged cluster Au38(PET)23(PCP4)1. Peak 2 was collected and re-injected (red trace). Bottom: HPLC chromatogram (top trace) of the collected peak 2 corresponding to a specific regioisomer of Au38(PET)23(PCP4)1 after heating to 80° C for 90 min. For comparison the HPLC chromatogram of a ligand exchange reaction of racemic Au38(PET)24 with PCP4 is shown. Reprinted with permission from ref. 80. Copyright (2013) American Chemical Society. |
The study mentioned above furthermore shows that ligand exchange at the four symmetry-unique sites is not equally likely because the bands in the HPLC associated with the formed regioisomers have different intensity (see Fig. 7, top).80 NMR experiments showed this also for Au25(PET)18 clusters.81 In the latter cluster only two different positions are available for exchange, namely in the centre of a dimeric staple (called outer ligand by the authors) or at the end of the dimeric staple (called inner ligand by the authors). The experiments showed that exchange is favoured for the outer ligand (middle position, SR–Au–SR–Au–SR) and that the exchange rate at the outer ligand is up to 3.5 times higher than for the inner ligand, depending on the nature of the incoming ligand. Interestingly, a single crystal X-ray study revealed ligand exchange only at the most solvent-exposed site, at the end of the dimeric staple,82 which seems to be in contradiction with the aforementioned study. Possibly crystallization leads to a selection of a subset of ligand-exchange products.
The stability of thiolate-protected gold clusters and nanoparticles indicates that desorption of thiolates or of gold–thiolate complexes from the gold cluster/nanoparticle surface is not an important process, despite the fact that the thiolates are mobile to a certain extent on the gold cluster surface. However, some experimental observations show that thiolates can be exchanged between different clusters. Negishi and coworkers prepared Au25(SC12H25)18 and Au25(SC10H21)18 clusters.83 Mixing of the two clusters in CH2Cl2 solution for 10 minutes at room temperature lead to a mixed thiolate layer as evidence by mass spectrometry, which showed that up to four thiolates exchanged between the particles, as can be seen in Fig. 8. The effect was less pronounced for a Pd doped cluster.
Fig. 8 Negative-ion MALDI mass spectra of a mixture of [Au25(SC12H25)18]− and [Au25(SC10H21)18]− after 10 min in CH2Cl2 (a) and of [PdAu24(SC12H25)18]0 and [PdAu24(SC10H21)18]0 after 10 min in CH2Cl2 (b). The dominant peak corresponds to the parent clusters, whereas the small peaks are due to exchange clusters with mixed ligand shell. Reproduced from ref. 83 with permission from The Royal Society of Chemistry. |
EPR was also used to study exchange of thiolates between different particles.84 Two types of particles, initially covered by different thiolates each, were mixed. Interparticle exchange of ligands then leads to a change in interaction between adjacent ligands, which is sensitively monitored by EPR. Changes were very slow at room temperature, but at 70 °C alterations in EPR line-shapes were observed within minutes, indicating the interparticle exchange of ligands. Interestingly, even after prolonged heating the presence of free ligands in solution was minimal, showing that the steady state concentration of free ligand is low. Based on the kinetics of this reaction the authors proposed a mechanism where the rate-determining step is the desorption of a thiolate species (of unknown nature and stoichiometry), which creates a vacancy for the readsorption of (another) thiolate.
These clusters studies seem to be in agreement with studies performed 20 years ago on SAMs on gold of octadecanethiol containing radiolabeled head groups (35S).85 These studies showed that thiols can desorb into solution on a timescale of a day. The desorption process was not complete, followed pseudo first order kinetics and depended on the solvent. The finding is consistent with earlier studies, which indicated the transfer of thiolates (4-carboxythiophenol) between gold and silver colloids.86
In summary, the gold–thiolate interface is not rigid, both in clusters and in SAMs. Thiolates can move between different sites on clusters. Several studies indicate that this is a slow process at room temperature both for clusters and nanoparticles. At slightly elevated temperatures (70–90° C) migration is evidenced within tens of minutes. Also, entire staple motifs can move on the cluster surface, as is shown by racemization studies. Again, this process is observed above room temperature. In addition, there is evidence both for clusters and for flat gold surfaces that thiolates can desorb into solution even at room temperature within tens of minutes. The stoichiometry of the desorbing species is unknown. However, work by Murray and coworkers indicated that even metal atoms can be exchanged between gold and silver clusters protected by thiolate monolayers87 indicating that even metal atoms/ions can desorb. Despite the large body of experiments described above there is not yet a clear picture emerging, for example as concerns the relative importance of ligand desorption/re-adsorption and direct migration on the surface. Some findings remain somehow contradictory. Possibly there is a particle size effect of these properties, as the work of Stellacci and Murry seems to indicate by showing that the formation of striped (phase segregated) nanoparticles is size dependent.88 Open questions to be addressed in the future concern the stoichiometry of the species desorbing from or migrating on the cluster surface, the role of Au(I) in the staples for these processes and the migration/desorption of dithiolates.
The structure of SAMs of BINAS on Au(111) was also studied by STM, which revealed a 2D chiral arrangement.103 These SAMs were prepared from solution and investigated in air. A honeycomb structure was found, consisting of screw-like entities with 3-fold rotational symmetry, each of which is composed of three BINAS molecules. The formation of orderd structures implies some degree of mobility of the thiolates on the surface. Such a SAM of BINAS on gold was shown later by quartz crystal microbalance (QCM) to discriminate between the enantiomers of thalidomide.104R-thalidomide adsorbed on R-BINAS SAMs but not or hardly on S-BINAS SAMs.
An interesting STM study revealed the assembly of an achiral adamantane tripod (trithiol) on Au(111) into a chiral framework.105 The sample was prepared and examined in UHV. The rigid molecule forms three Au–S bonds with the surface. The adsorbed molecules develop chiral trimers, which act as building block for even larger hexagonal chiral structures. This shows that symmetry can be broken by achiral molecules leading to locally chiral structures on a globally racemic surface. The formation of such highly ordered hierarchical structures again indicates some degree of mobility of the individual molecules on the surface after their adsorption. In these experiments the molecules were evaporated onto the surface at around 330 K. Similarly, supramolecular chiral structures were found for SAMs of a dithiocarbamate on Au(111).106
Waldeck and coworkers prepared SAMs on gold from thiol-terminated chiral scaffold molecules containing a porphyrine chromophore at its end.107 The SAMs were prepared from solution. Illumination by visible light within an electrochemical cell generated a cathodic photocurrent. Interestingly, the authors observed a strong difference of this photocurrent depending on the circularity of the light used (left- or right-circular polarized). This asymmetry in the photocurrent was explained with a symmetry constraint on the electronic coupling between the porphyrine moiety and the chiral scaffold.
Nakanishi and coworkers prepared SAMs containing leucine molecules and studied the crystallization of leucine on top of these SAMs.108 It was found that the crystallization was enantioselective. When immersing a D-leucine SAM in a D-leucine solution, crystals of D-leucine were grown. However, when the D-leucine SAM was immersed in a L-leucine solution, no crystals were formed (and vice versa). Similarly, Amabilino and coworkers prepared SAMs from a chiral resolving agent containing a cyclic disubstituted phosphate group.109 These SAMs were used as templates for the growth of crystals of organic molecules with a similar structure as the ones used for SAM formation.
Using UHV STM Besenbacher and coworkers showed that cysteine on Au(110) forms dimers and that the molecular pairs formed from a racemic mixture are exclusively homochiral, which is evidence for strong chiral recognition.97 Chiral SAMs prepared from homocysteine on Au(111) were used as electrodes to study the redox behaviour of catechins using cyclic voltammetry.110 The formation of the SAM blocks, to some extent, the redox reaction. However, the extent of blocking was different for the two enantiomers of the catechins. The preference for one enantiomer was inverted in acidic solution. These experiments showed that the homocysteine SAM can distinguish enantiomers of catechin. By comparison between catechin and epicatechin the authors concluded that the chiral SAM is able to discriminate the absolute configuration at one of the two chiral centres. Homocysteine SAMs were also integrated into the gate of a field effect transistor, which allowed the discrimination between the enantiomers of alanine.111
SAMs of chiral N-acetyl-L-cysteine112 and glutathione113 on gold were also prepared directly on internal reflection elements, which enabled in situ studies by attenuated total reflection infrared spectroscopy (ATR-IR). The ability of these two SAMs to discriminate between proline enantiomers was then studied using modulation excitation spectroscopy combined with a phase-sensitive detection.114–117 For this, solutions of the two enantiomers of the probe molecule (proline) were flown alternatingly over the SAM and the response of the system was recorded in situ. For the glutathione SAM it was shown that the adsorption and desorption kinetics (but particularly the desorption) are different for the two enantiomers of proline and therefore the chiral SAM can discriminate between them. All these examples show that chiral SAMs have the ability to discriminate between enantiomers of molecules that interact with the SAMs.
The first experimental manifestation of chirality in thiolate-protected clusters stems from circular dichroism (CD) measurements on gold clusters protected by (chiral) glutathione.119 Strong CD signals were observed in transitions that are mainly located in the metal core. Similar observations were made for other gold clusters protected by chiral thiolates.120–126 Until direct structural information was available the origin of the optical activity of was not clear. Particularly it was not clear if the structure of the cluster is intrinsically chiral127 or if the optical activity is “just” induced by the chiral ligands around a symmetric particle core.118,128 The first crystal structure of a thiolate-protected Au102(pMBA)44 cluster then revealed a symmetric core protected by gold–thiolate staples (see Fig. 2 and 3). The staples are arranged in a chiral fashion even though the thiol itself is not chiral. Furthermore, each sulphur atom represents a stereogenic centre. The unit cell of the crystal contains both enantiomers of the cluster and therefore such a sample is racemic and does not show optical activity. A similar situation was later found for the chiral Au38(PET)24 cluster (see Fig. 7).11 The enantiomers of the cluster could be separated into its enantiomers using chiral chromatography.64 This allowed one to measure the CD spectrum of the cluster, which was quite pronounced despite the fact that the cluster is protected by achiral ligands. The enantiomers of other chiral gold clusters were later also separated.16,73,79,129
An interesting case represents the Au25(SR)18 cluster.9 Its Au–S framework is achiral consisting of an Au13 icosahedral core protected by six long staple motifs (three orthogonal pairs). The cluster shows optical activity when covered by a chiral thiolate.130 However, the anisotropy factors are much weaker compared to the Au38 case, which indicates that the chiral arrangement of thiolates gives rise to stronger optical activity compared to the optical activity induced by chiral thiolates.
Chiral Au38(PET)24 clusters were shown to be able to discriminate between enantiomers of a chiral thiols in ligand exchange reactions (Fig. 9). Specifically, a racemic mixture of the cluster was exposed to R-BINAS and the ligand exchange was monitored in situ by chiral HPLC.76 Analysis of the kinetics revealed that the left-handed cluster reacted about four times faster with R-BINAS than the right-handed cluster. Note that the thiolates initially adsorbed on the cluster are not chiral.
Fig. 9 The ligand exchange between Au38(PET)24 and BINAS is diastereomeric. Top: HPLC traces recorded in situ during the ligand exchange between racemic Au38(PET)24 and R-BINAS. The first two peaks belong to the two enantiomer of Au38(PET)24 (A-Au38(PET)24 and C-Au38(PET)24) The next two peaks belong to the corresponding cluster with one R-BINAS incorporated (A-Au38(PET)22(R-BINAS)1 and C-Au38(PET)22(R-BINAS)1). The broad peak at higher retention times corresponds to higher exchange products. All chromatograms were scaled to the intensity of the first peak. Bottom: kinetic analysis of the ligand exchange. Dots are measurements (HPLC) and solid lines are derived from a fit to a kinetic model. Black: A-Au38(PET)24, gray: C-Au38(PET)24, red: A-Au38(PET)22(R-BINAS)1, blue: C-Au38(PET)22(R-BINAS)1. Reprinted with permission from ref. 76 Copyright (2012) American Chemical Society. |
Gellman and coworkers also demonstrated that gold nanoparticles of about 5 nm in size and covered by enantiomers of cysteine show enantioselectivity in adsorption of propylene oxide on their surface.131 The authors used optical rotation to demonstrate enantiospecific adsorption and used the fact that the specific optical rotation is enhanced for molecules interacting with the gold nanoparticles. The analysis can be made in a quantitative manner in order to extract enantiospecific adsorption equilibrium constants.132 In summary, not much has been done to demonstrate and apply the enantiospecific properties of gold nanoparticles and clusters although their potential has been shown. With the recent progress in the preparation of chiral clusters and particles more work will certainly be done in this field.
Vibrational spectra of adsorbed thiolates on gold surfaces provide useful information on interactions between molecules in the SAM, on the orientation of molecules with respect to the surface112,133 and the order within the layer. Information on the order within an alkanethiol SAM is provided by the frequency and width of the C–H stretching bands.134 Vibrational spectra can be obtained by several techniques including infrared and Raman spectroscopy, high-resolution electron energy loss spectroscopy (HREELS) and non-linear optical techniques such as sum-frequency generation. Of particular interest here are the vibrational properties of the Au–S interface, which are quite low in frequency. Although vibrational spectroscopy was frequently used to characterize SAMs of thiolates not much attention was paid to the low frequency region of the spectrum, a fact related with experimental challenges. Raman signals of monolayers on flat surfaces are very small and far infrared (FIR) spectroscopy requires special equipment. Typical Au–S stretching vibrations give rise to bands around 220–240 cm−1 in the spectrum, as evidenced by Raman and HREELS measurements.135–137 A systematic study of the HREELS spectra of SAMs formed by alkanthiolates of different lengths on Au(111) revealed multiple Au–S stretching bands, which were assigned to multiple adsorption sites of the thiolate on the gold surface.138 The authors also observed a dependence on the chain lengths. At the time of the study the staple structure was not yet discovered. It is clear that such a staple structure should give rise to multiple Au–S stretching modes at different frequencies due to the different nature of the Au–S bonds involved (Au–S within the staple and Au–S involving a core gold atom).
In principle a vibrational spectrum contains information about the structure (conformation) of a molecule. In many cases it is however not trivial to extract such information from the spectra.
Creutz and co-workers reported far-IR spectra of alkanethiol capped gold nanoparticles (about 2 nm).143 The spectra are characterized by quite broad bands that change with the length of the alkanethiol. Bands at 260–270 cm−1 and around 180 cm−1 were assigned to Au–S vibrations. Murray and coworkers reported the Raman spectrum of Au25(PET)18. A band at around 290 cm−1 was assigned to Au–S vibrations.144
Halas and coworkers used surface enhanced Raman spectroscopy to study the low-energy region of SAMs of alkanethiols of different lengths (C10, C11…C16) on plasmonic gold particles.145 The low-energy vibrational spectrum (below 400 cm−1), where the Au–S vibrations are expected, showed several bands and the spectra were quite different from the Raman spectra of the alkanethiol liquids. In particular the spectra were characterized by a sharp band that shifted drastically with chain lengths from 367 cm−1 (C10) to 278 cm−1 (C16) and a second band at lower frequency that shifted as well.
These observations were explained by a coupling between the Au–S stretching and longitudinal acoustic modes (LAM) of the alky chain. The latter modes were also observed for the alkanethiol liquids although less sharp and with a smaller shift as a function of chain length. It is likely that the coupling and therefore the spectrum is sensitive to the geometry and binding of the thiolate head-group. However, the coupling also complicates a more direct analysis of the low-frequency vibrational spectra.
In an attempt to relate vibrational signatures to structural information of the Au–S interface Raman and far-IR spectra of well-defined clusters were measured.142,146 By comparison with calculations one can assign different types of Au–S vibrations notably Au–S–C bending modes around 180 cm−1, radial Au–S stretching modes around 220–280 cm−1 and tangential Au–S stretching modes around 320 cm−1 and above. (see Fig. 10).
Fig. 10 Calculated (top) and experimental (bottom) Raman spectrum of the Au38(SCH3)24 and Au38(PET)24 cluster, respectively. Radial and tangential Au–S modes of the staples are schematically represented. Radial vibrations of the long staples are responsible for bands with high intensity. Modes associated with the short staples (symmetric and antisymmetric stretching and tangential vibrations) have lower Raman intensity. Reprinted with permission from ref. 142 Copyright (2014) American Chemical Society. |
Radial modes are lower in frequency and involve the stretching between sulphur atoms of the staple and gold core atoms. Tangential modes involve stretching motions between sulphur and gold within staples. As mentioned above good quality HREELS spectra of alkanethiol SAMs evidence multiple peaks in the 200 cm−1–350 cm−1 spectral region,138 which might indeed be attributed to staple-like structures instead of different adsorption sites of the thiolate. Vibrational spectroscopy has therefore the potential to shed more light on the nature of the Au–S interface on flat SAMs but for this more systematic studies are needed with high quality spectra of the low frequency vibrations.
A special technique that is particularly sensitive to conformation is vibrational circular dichroism (VCD), i.e. the differential absorption of left- and right-circular polarized light by a chiral sample.147,148 This technique has been applied to gold clusters in order to extract information on the conformation of the adsorbed chiral thiolates.122,149–151 VCD signals are in general very weak and the technique cannot be applied to surfaces. However, VCD on clusters is a powerful technique to study the structure of adsorbed molecules.
Achiral molecules and racemic mixtures do not show VCD signals. However, it has been demonstrated that the achiral 2-phenylethylthiol gives rise to strong VCD signals when adsorbed on the chiral Au38 cluster (Fig. 11).152 For the VCD measurements the enantiomers of the cluster were separated. The reason for this is a preferred chiral conformation of the molecule adsorbed on the cluster. More precisely the molecule can exist in (achiral) anti and (transiently chiral) gauche conformation, referring to the relative orientation of the phenyl group and the sulphur atom. The infrared absorption spectra indicated a pronounced abundance of gauche molecules. For the gauche conformation two enantiomeric forms exist and in solution the two forms are equally abundant. Due to the chirality of the cluster one gauche enantiomer becomes more stable, which gives rise to the observed VCD signals. This shows that the chiral cluster transfers its chirality to the achiral molecule. The study furthermore indicates that the conformation of the thiolates on the clusters is different in solution and in the solid state (crystal).
Fig. 11 VCD spectra of the two enantiomers of Au38(PET)24. The bands in the spectra are due to the 2-phenylethylthiolate (PET) adsorbed on the cluster. PET is an achiral molecule, but the chiral cluster stabilizes one enantiomeric form of a transiently chiral gauche conformation, which leads to the VCD signals. The models show the chiral Au–S framework of the cluster and one PET ligand adopting a chiral conformation. Reproduced from ref. 152. |
There is strong evidence for gold ad-atoms or staple-like structures on flat gold surfaces (at least for Au(111)). However, there are also recent studies, which do not give direct evidence for gold ad-atoms. On the other hand, there is now also evidence from single crystal X-ray crystallography for other structural motifs, besides staple motifs, such as bridged adsorption sites on clusters. If a unified model of the Au–S interface will emerge for both the cluster and SAM field remains to be seen. What is clear is that the Au–S interface is more complex than it was thought one decade ago.
The Au–S interface is far from being rigid and static both on flat surfaces and on clusters. Different processes including migration of thiolates and desorption of thiolates (and re-adsorption) contribute to the flexibility of the interface. Migration of thiolates was evidenced for clusters and surfaces at temperatures above ambient whereas there is evidence for desorption even at room temperature both from flat surfaces and from clusters. On flat surfaces the migration of etch pits is also observed and it is not clear if there is an analogous process for clusters. The flexibility of the Au–S interface is an important property for applications of clusters (and SAMs) and it will be important in the future to clarify the composition of the migrating and of the desorbing species.
Chirality of SAMs and of thiolate-protected clusters is an interesting property with possible applications in chiral technology. Chirality in both cases arises at different levels that is at the level of arrangement of the thiolates (organization level) and at the molecular level (local level). In both cases achiral molecules can become chiral upon adsorption which has to do with symmetry breaking imposed by the surface. Chiral recognition and chirality transfer effects have been demonstrated and it remains to be shown if chiral gold clusters may be used for chirality sensing or enantioselective catalysis.
Comparison of properties of the two systems considered here (thiolate SAMs on flat surfaces and thiolate-protected clusters) is sometimes difficult due to the different experimental techniques that have to be applied. In this sense vibrational spectroscopy stands out because it can be equally applied to both systems and the vibrational spectrum is a local property, i.e. whether the surface is flat or curved does not matter much as concerns the vibrational spectrum of the adsorbate. Indeed, spectroscopy of the Au–S vibrations could play an important role to characterize the structure of various Au–S interfaces. Unfortunately this is hampered by the relative complexity of the vibrational signatures. Future theoretical studies are needed here. In addition, there are only few vibrational spectra available of clusters and good quality low-frequency vibrational spectra of thiolated SAMs are surprisingly scarce.
This journal is © The Royal Society of Chemistry 2015 |