Yanan
Wang
and
Thomas
Bürgi
*
Department of Physical Chemistry, University of Geneva, 30 Quai Ernest-Ansermet, 1211 Geneva 4, Switzerland. E-mail: thomas.buergi@unige
First published on 24th January 2022
Ligand exchange reaction (LER) is an important post-synthesis strategy and has been studied widely. The mechanism of this dynamic process for gold nanoclusters proved to be associative (SN2). Many factors affect the LER of clusters, including stability, solubility, chirality, electronic properties and so on. Some of these factors are not well understood and need further exploration. Here, we use a chiral fluoro-substituted ligand (R)-5,5′,6,6′,7,7′,8,8′-octafluoro-[1,1′-binaphthalene]-2,2′-dithiol (8F-R-BINAS) to investigate the stereoelectronic and stereospecific effects during LER on achiral Au25 cluster. It is demonstrated that the fluorine-substituted BINAS significantly decreases the LER reactivity both at the molecule and the related cluster level. The stereoelectronic effect is global and can be transmitted to the cluster surface. In contrast, the stereospecific effect is marginal.
Ligand substitution reactions, in which one or more ligands in a complex are replaced by a different ligand, is a conventional process in inorganic and organic chemistry.16–18 This reaction is also called ligand exchange reaction (LERs) in some fields. During the past decades, the mechanism of this dynamic process has been explored widely, and two possible pathways, dissociative (SN1) and associative (SN2), have been demonstrated.19–21 This process has been revealed for traditional transition metal-complexes or organic frameworks, but after that it also has been shown for thiolate-protected gold nanoclusters, as first studied by Murray and coworkers.22–27 Until now LERs developed into an important post-synthesis method, which has been used for extending the properties and functions of nanoclusters.27–29 Much research has also focused on the reaction sites of LERs and the related effects.30–36 In addition, the mechanism of LERs on nanoclusters is thought to follow an associative (SN2) pathway, according related experimental and computational studies.27,37,38 Understanding the mechanism and the factors that affect this reaction will help the design of atomically precise metal clusters with mixed ligand shells.
LERs on gold nanoclusters are strongly influenced by the chemical properties of the involved ligands and the flexibility of the gold–sulfur interface.39,40 Other factors involved in this dynamic process which affect the reaction rate are the stability of the samples,41 length of surface ligand,26 metal doping,42 diastereoselective interactions,43 and electronic properties of the ligand,24,44 and nanoclusters.23 In 2002, Murray and coworkers demonstrated that the rate of LERs increased with increasing positive electronic charge on the Au core.23 Later they also revealed that ligands with electron-withdrawing substituents reacted faster at shorter reaction time, and ligands with electron-donating substituents were more efficient at longer reaction times.24 However, other work revealed that the ligands with electron-donating groups reacted faster with the cluster.45 After two decades of research the influence of electronic structure on LER seems not completely understood, which calls for more effort.44 Furthermore, in contrast to the diastereoselective LERs with intrinsically chiral nanoclusters,43 the stereospecific effects in LERs of chiral ligands on achiral nanoclusters have not been quantified yet.
Here, using a chiral fluoro-substituted ligand (R)-5,5′,6,6′,7,7′,8,8′-octafluoro-[1,1′-binaphthalene]-2,2′-dithiol (named 8F-R-BINAS in the following), the electronic and stereospecific effects of the ligand during exchange the reaction on achiral Au25 will be systematically investigated. Compared with parent ligand 1,1′-binaphthalene-2,2′-dithiol (BINAS), the fluoro-substituted molecule exhibits higher electron-withdrawing ability without large size change. In addition, the rigid structures of the two chiral ligands also represent good models for the investigation of stereospecific effects. The distance between the two sulfur atoms in the BINAS molecule matches very well the distance between two sulfur atoms in the cluster. Also dithiol ligand prohibit the intercluster ligand exchange at room temperature. Therefore, BINAS is an excellent candidate for the LERs study. We show that the electronic property of ligand induces significant discrimination to the LERs, however, the absolute configuration of the ligand did not significantly affect the rate of exchange, at least at the early stage of exchange where only few chiral ligands are adsorbed on the achiral cluster.
To investigate the electronic effect of the ligand during the exchange process, Au25(2-PET)18 was first mixed with R-BINAS and 8F-R-BINAS separately. Toluene was used as solvent for the reaction, and the molar ratio of cluster to ligand was equal to 1:
20. A small amount of sample was taken from the reaction mixture at different reaction time and the corresponding MALDI-TOF mass spectra were recorded (Fig. 1). The calculated mass of different species after Au25(2-PET)18 exchange with BINAS (Fig. 1A) and 8F-R-BINAS (Fig. 1B) are listed in Table S1.† As confirmed before, BINAS acts as a bidentate ligand thus substituting two 2-PET ligands on the cluster.48,49 Ligand exchange numbers (x = number of BINAS/8F-R-BINAS in the ligand shell) are marked in Fig. 1. The clusters show fragmentation, notably by losing Au4(2-PET)4. The corresponding fragments are also visible in Fig. 1 and the related exchange numbers (x′) were also labelled. Since the fragments are formed during the MALDI measurement, the signals of the fragments were also considered for the quantitative analysis by adding them to the ones of the corresponding intact cluster.
In order to study the kinetics of ligand exchange reactions, the intensity of mass peaks in Fig. 1 were quantified (taking into consideration also fragment peaks as mentioned above) and the results are given in Table S2† (Au25(2-PET)18 + R-BINAS) and Table S3† (Au25(2-PET)18 + 8F-R-BINAS). When reacted with R-BINAS, clusters with up to six R-BINAS ligands were observed after 72 h reaction time, whereas for the ligand exchange reaction with 8F-R-BINAS up to four 8F-R-BINAS ligands could be detected on the cluster. The fractions of the different cluster species were calculated, and their evolution as function of time are shown in Fig. 2. LER can be considered as a consecutive reaction, where a first ligand is exchanged followed by a second one etc. The equations for the related LERs are shown in the ESI (ESI Note 1†). The time-dependent concentration of the species with x = 1 (one BINAS/8F-R-BINAS in the ligand shell of the cluster), red data points in Fig. 2, depends on the two rate constants k1, which describes the first ligand exchange, and k2, which describes the second ligand exchange. The time dependence of the one ligand-substituted species (x = 1) was quantitatively different in the two cases. The maximum fractions of this species were 0.44 for the experiment with R-BINAS and 0.74 in the case of 8F-R-BINAS, which shows that the ratio of the rate constants k1/k2 is different for the two ligands. In order to extract these ratios, the abundance of the parent cluster and the cluster with one exchanged ligand in its shell were fit to a kinetics of a consecutive reaction (pseudo first order) using MATLAB (the code is shown in the ESI, Note 2†). The raw data extracted from the MALDI experiments are given in Tables S2 and S3,† and the fitting curves of these two LERs process are shown in Fig. S4.† From the fit it emerges that the related ratio of rate constants k1/k2 is 2.4 for LERs between Au25(2-PET)18 and R-BINAS, and changes to 4.7 when 8F-R-BINAS is used as the incoming ligand. This means that the second ligand exchange performed slower with 8F-R-BINAS compared to R-BINAS. The significant difference for these two ligands can also be discovered at the initial stage (before 5 h) of the exchange, where the species with two exchanged BINAS ligands (x = 2, green curve) can already be observed before 5 h (Fig. 2A), whereas for the experiment with 8F-R-BINAS the corresponding species arises after 5 h (Fig. 2B). In addition, also the species with three exchanged ligands were delayed for 8F-R-BINAS compared to BINAS. All these observations are consistent with an increasing slowing down of the further ligand exchange reaction once 8F-R-BINAS is incorporated in the ligand shell of the cluster.
The distinct difference in the LER with the two ligands, described above, is ascribed to the electronic effect of fluorine. The 8 fluorine atoms in 8F-R-BINAS have strong electron-withdrawing ability and change the electron density of the aromatic ring. After incorporation of 8F-R-BINAS into the ligand shell, the electronic effect may also extend to the whole cluster (surface). As mentioned before, the mechanism for ligand exchange reaction follows an associative (SN2) pathway.27 The first step for the ligand exchange is nucleophilic attack by the incoming thiol, creating a bimolecular intermediate. Consequently, the 8F-substituted BINAS molecule has lower electron density at the sulfur atom compared with BINAS, which decreases its ability to act as nucleophile. The data shown above indicates that the fluorinated ligand may also affect the reactivity of the cluster, decreasing its ability for subsequent ligand exchange. However, the electronic effect of fluorinated ligand on the clusters may be more complex.
To further study the different properties of BINAS and 8F-R-BINAS in LER and to better distinguish the electronic effects of the fluorinated ligand and the cluster containing the fluorinated ligand, mixtures of R-BINAS and 8F-R-BINAS were used. Here, the molar ratio of Au25(2-PET)18 clusters and free ligand was 1:
15, and the ratio between R-BINAS and 8F-R-BINAS was chosen as 1
:
2 and 1
:
4, respectively in two separate experiments. The calculated mass values of the different cluster species after ligand exchange of Au25(2-PET)18 with R-BINAS and 8F-R-BINAS are listed in Table S4.† The MALDI-TOF spectra as a function of time for the experiment with 1
:
4 R-BINAS
:
8F-R-BINAS ratio are shown in Fig. 3A. The peaks have been labelled using numbers, and the corresponding compositions are given at the right side of the spectra. The mass peak intensities were determined and the percentage of different species were quantified and listed in Table S5.† For the calculation of the percentage, the fragmentation peaks (Au21 species) were also taken into account. The relative abundance of the different species as a function of time are illustrated in Fig. 3B. As expected the fraction of the parent cluster (marked as 0,0) decreased with time and the fractions related to ligand-exchanged species raised. More interesting is the comparison between the cluster species containing one R-BINAS (marked as 1R,0, red trace) and one 8F-R-BINAS molecule (marked as 0,1R(F), green trace) in their ligand shell. Whereas both species increase at about the same rate initially, the abundance of the species containing one R-BINAS ligand decreased again after about 50 h while the abundance of the cluster containing one 8F-R-BINAS ligand continued to increase. At the same time the cluster with both one R-BINAS and one 8F-R-BINAS (light blue line in Fig. 3B) increased strongly. The different behavior shows that the cluster containing one 8F-R-BINAS is less reactive compared to the cluster containing one R-BINAS ligand, which is ascribed to the effect of 8F-R-BINAS on the electronic properties of the cluster.
Importantly, the average numbers of exchanged R-BINAS (black curve) and 8F-R-BINAS (red curve) in the cluster (Fig. 3C) are comparable in the course of the time, showing that the behavior described above is not due to the deactivation of the free 8F-R-BINAS ligand.
At R-BINAS:
8F-R-BINAS ratio of 1
:
2 the behavior was found to be qualitatively similar (Fig. S5†). The MALDI-TOF spectra at species are listed in Table S6.† From the evolution plots of the different times are shown in Fig. S5A† and the derived values of different mass peaks and calculated fractions of the different species in these two experiments (Fig. 3B and Fig. S5B†), some conclusions can be drawn. By increasing the R-BINAS fraction from 20% (ratio 1
:
4) to 33.3% (ratio 1
:
2), the species with one R-BINAS ligand (1R,0) becomes predominant. Furthermore, the species with two R-BINAS ligands (2R,0) exceeds the abundance of the hybrid species (1R,1R(F)), which is opposite at 1
:
4 ratio. At lower ratio, considerably more R-BINAS was incorporated in the ligand shell compared to 8F-R-BINAS, which is reflected in the average number of exchanged R-BINAS and 8F-R-BINAS, which is lower for the fluorinated ligand (Fig. S5C†). The observations described above are also in agreement with the associative pathway for the ligand exchange process, the rate constant being dependent on the concentration of incoming ligands.
The ligand exchange leads to some changes in the optical properties. The absorbance spectra of Au25 after reaction with R-BINAS or R-BINAS/8F-R-BINAS mixture have similar features as shown in Fig. S6A & B,† however, the characteristic features of Au25 become less distinct as the LER proceeds. The spectral changes seem more important for clusters containing 8F-R-BINAS (even for lower exchange numbers). In addition, the circular dichroism spectra of (Au25(2-PET)18−2x(R/S-BINAS)x and Au25(2-PET)18−2x−2y(R-BINAS)x(8F-R-BINAS)y) are provided in Fig. S6C.† The chiral ligands induce optical activity in the clusters. The bands at 350 nm and above are due to the cluster (and not due to transitions within the ligand alone), as comparison with the CD spectra of the ligands shows (see Fig. S2†). The incorporation of 8F-R-BINAS into the ligand shell does not lead to drastic changes of the CD spectrum. However, the band around 350 nm is shifted to higher wavelengths by about 10 nm for the cluster containing 8F-R-BINAS and the feature at around 475 nm is less pronounced. We should mentioned that, the average number of exchanged ligand for Au25(2-PET)18−2x−2y(R-BINAS)x(8F-R-BINAS)y is higher than for Au25(2-PET)18−2x(R/S-BINAS)x here. But the spectral changes also show that 8F-R-BINAS has some influence on the electronic structure of the cluster.
Since the two ligands show distinct behaviour in separate LERs, we may anticipate that using a mixture of the two ligands will not simply lead to a statistical distribution of the two ligands on the cluster. In other words, some combinations of ligands may be more (less) abundant than expected based on the average composition. In order to verify this hypothesis, we calculated the statistical distribution (multinomial distribution). For this we first determined the average composition (average x and average y) for a specific sample by considering all the detected species. Having these numbers from the experiment, one can determine the probabilities required for the calculation of the multinomial distribution. (The program code and related values of probabilities are shown in ESI Note 3.†) This can be done for every sample collected as a function of time during the experiment. We took the data extracted from the experiment shown in Fig. 3 (average number of exchanged R-BINAS and 8F-R-BINAS, Fig. 3C) to calculate the statistical distribution of the cluster species. It should be noted that this calculation requires knowledge of the total number of available sites. It has been shown that only up to seven BINAS ligands can adsorb on Au25 (Ag25) nanoclusters,50 due to steric hindrance, and therefore we chose seven as the number of available sites for the dithiols. The calculated statistical distributions of the different species are illustrated in Fig. S7.† The comparison between the experimental evolution of species and the evolution based on the statistical distribution are shown in Fig. 4.
![]() | ||
Fig. 4 Evolution of different species during Au25(2-PET)18 ligand exchange with R-BINAS and 8F-R-BINAS mixture as function of time. The solid curves are extracted from Fig. 3B (experimental (E) values, 1![]() ![]() |
As demonstrated by Fig. 4, there are clear differences between experiment (solid curves, E) and simulation (dash curves, S), demonstrating that the distribution is not statistical. For example, the species containing one 8F-R-BINAS ligand (0,1R(F)) shows significantly higher abundance than expected based on the statistical distribution. The ligand with one R-BINAS in its ligand shell is initially more abundant than expected based on statistical distribution and then becomes less abundant than expected. The above shows that the LERs between Au25(2-PET)18 (0,0) and a mixture of R-BINAS and 8F-R-BINAS ligands do not lead to statistical distributions of cluster species. There seems to be some specificity, either due to steric of electronic effects. To shed some more light on this issue we analyzed the kinetics of the reaction in more detail.
The reaction network of ligand exchange reactions between Au25(2-PET)18 (0,0) and mixtures of R-BINAS and 8F-R-BINAS ligands is illustrated in Scheme 1. The different involved species of general formula Au25(2-PET)18−2x−2y(R-BINAS)x(8F-R-BINAS)y were labelled as (x, y) in the scheme, and only the initial part of the process with maximum total ligand exchange number equal to 3 (maximum x + y = 3) is considered. The reaction network in Scheme 1 is a simplification, as it does not consider isomers of clusters, which differ in the relative position of the adsorbed R-BINAS and 8F-R-BINAS ligands on the cluster. The reaction network was modelled assuming each ligand exchange reaction as a pseudo first order reaction, which seems reasonable, taking into account the excess of the ligands. The experimental data of the LERs were fit to the model outlined in Scheme 1 using MATLAB (the code is shown in the ESI Note 4†). For example, Fig. 5 shows the fit of the experiments depicted in Fig. 3B (raw data extracted from Table S5;† ratio between R-BINAS and 8F-R-BINAS as 1:
4). The model fits well the experimental data. Similar fitting was done for the experiment with 1
:
2 ratio of R-BINAS and 8F-R-BINAS (raw data taken from Table S6 and Fig. S9A†). The obtained rate constants are given in Fig. S9D.†
![]() | ||
Fig. 5 MATLAB fitting curves of Au25(2-PET)18 ligand exchange with R-BINAS and 8F-R-BINAS mixture with ratio as 1![]() ![]() |
The ratio of the rate constants describing the consecutive exchange of R-BINAS (k1/k3) and 8F-R-BINAS (k2/k6) also are in good agreement with the corresponding ratio extracted in the experiments with the individual ligands (Fig. 2), which underlines the reliability of the approach. The electronic effect of the 8F-R-BINAS modified cluster on the exchange rate can be appreciated by comparing k4 and k6, as k4 describes the reaction of Au25(2-PET)16(R-BINAS)1 with 8F-R-BINAS and k6 the reaction of Au25(2-PET)16(8F-R-BINAS)1 with 8F-R-BINAS. The ratio k4/k6 is about 2.0 for both experiments (1:
4 and 1
:
2 ligand ratios). The effect is also illustrated by the ratio of k3/k5. In this case, however, the fit gave very low values for k5. The kinetic constants indicate the ligand as well as the (ligand-exchanged) Au25 cluster exert an electronic effect on the rate of LERs: the 8F-R-BINAS ligand and the related substituted cluster, Au25(2-PET)16(8F-R-BINAS)1, show lower reactivity compared to non-fluorinated counterparts (R-BINAS and Au25(2-PET)16(R-BINAS)1).
R-BINAS and 8F-R-BINAS have the same configuration, and the discrimination during LERs is mainly due to electronic effects. By replacing R-BINAS with S-BINAS in the ligand mixture, the stereospecific effect can be investigated as well. Here the conditions for LERs follow the previous experiments, with molar ratio between cluster and ligand set as 1:
15, and molar ratio between S-BINAS to 8F-R-BINAS set as 1
:
2 (Fig. S8†) and 1
:
4 (Fig. 6), respectively. The MALDI-TOF results for 1
:
4 ratio are shown in Fig. 6A and the evolution of different species are represented in Fig. 6B (data given in Table S7†). Comparison with the experiment shown in Fig. 3B with the other enantiomer of BINAS (mixture of R-BINAS and 8F-R-BINAS) did not reveal significant differences. Furthermore, the average number of S-BINAS and 8F-R-BINAS on the cluster evolved similarly with time (Fig. 6C). Also for the experiment with 1
:
2 ratio of S-BINAS
:
8F-R-BINAS, the related evolution of the different species (Fig. S8 and Table S8†) was very similar to the one with the other BINAS enantiomer (Fig. S5†). Both experiments were fit using the MATLAB program (Fig. S9B & C†). The related rate constants as shown in Fig. S9† D also revealed the stereoelectronic effect of 8F-R-BINAS ligand as described above. From these experiments we can conclude that diastereospecific interactions are negligible under these conditions, which is not surprising, considering the fact that in our experiments only very few chiral ligands are adsorbed on the achiral cluster, leaving many “achiral sites” free for further incoming ligands. The stereospecific effect originating from the adsorbed chiral ligand is a local effect, unlike the stereoelectronic effect which seems to extend over the whole cluster.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1nr07602g |
This journal is © The Royal Society of Chemistry 2022 |