Ligand Engineering of Immobilized Nanoclusters on Surfaces: Ligand Exchange Reactions with Supported Au 11 (PPh 3 ) 7 Br 3

The properties of gold nanoclusters, apart from being size-dependent, are strongly related to the nature of the protecting ligand. Ligand exchange on Au nanoclusters has been proven to be a powerful tool for tuning their properties, but has so far been limited to dissolved clusters in solution. By supporting the clusters previously functionalized in solution, it is uncertain that the functionality is still accessible once the cluster is on the surface. This may be overcome by introducing the desired functionality by ligand exchange after the cluster deposition on the support material. We herein report the first successful ligand exchange on supported (immobilized) Au11 nanoclusters. Dropcast films of Au11(PPh3)7Br3 on planar oxide surfaces were shown to react with thiol ligands, resulting in clusters with a mixed ligand shell, with both phosphines and thiolates being present. Laser ablation inductively coupled plasma mass spectrometry and infrared spectroscopy confirmed that the exchange just takes place on the cluster dropcast. Contrary to systems in solution, the size of the clusters did not increase during ligand exchange. Different structures/compounds were formed depending on the nature of the incoming ligand. The feasibility to extend ligand engineering to supported nanoclusters is proven and it may allow controlled nanocluster functionalization.


Synthesis of the F-GSH ligand.
The synthesis of the Fluorescein-labeled glutathione (F-GSH) was carried out following a procedure reported by Landino and coworkers. 6 Small adjustments were made in the reduction step, when the pH value of the aqueous solution was adjusted with 10 N KOH to 7.4 instead of using a buffer when adding tris(2-carboxyethyl)phosphine hydrochloride (TCEP) (to a concentration of 50 mM). Completion of the reduction to F-GSH was indicated by formation of a yellow-orange precipitate (around 30 minutes to 2 hours when stirred at room temperature). Complete reduction was confirmed by TLC (Rf value 0.34 with acetonitrile/water/glacial acetic acid (80:20:1) as mobile phase). Separation and purification was then performed using C8 or C18 solid phase extraction columns and washing with water until no excess TCEP could be detected anymore (free TCEP gives a yellow solution when mixed with a 10 mM solution of 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) in 0,1 M phosphate buffer at pH 7.4). The final product F-GSH was then eluted with methanol, the solvent removed under reduced pressure and the yellow-orange powder stored under a N2 atmosphere. UV-Vis (MeOH, nm): 370, 425 (shoulder), 455, 480. Photoluminescence (MeOH, nm): 520, 545, 603 (shoulder).
Ligand Exchange with GSH in solution to Au25(SG)18. The ligand exchange was conducted according to literature. 1 An excess of GSH (10 eq., 56 mg, 0.18 mmol) in 10 ml of Milli-Q water was added to a solution of Au11(PPh3)7Br3 (1 eq., 10 mg, 0.0024 mmol) dissolved in 10 ml of CHCl3. The reaction mixture was stirred rapidly for 6 hours at 50 °C under nitrogen atmosphere. Afterwards, phases were separated and the colored aqueous phase washed several times with CHCl3. After evaporation of the solvent, Au25(SG)18 was yielded as a brown solid. Purification of the product was achieved by polyacrylamide gel electrophoresis (PAGE). UV-Vis (CH2Cl2, nm): 410, 455, 505 (shoulder), 670.
Ligand Exchange with GSH on an Al2O3 surface. A solution of Au11(PPh3)7Br3 (1 eq., 2.5 mg, 0.61 µmol) in 1-2 ml of DCM was dropcasted onto a polished Al2O3 surface (on an aluminum plate). Prior to use, the plates were polished with Metadi Diamond Polishing Compound (1 and ¼ micron, Buehler) and rinsed with EtOH. After evaporation of the solvent, the sample was placed in a solution of GSH (160 eq., 30 mg, 0.098 mmol) in 10 ml H2O:MeOH (8:2). After 3 days at room temperature, the plate was taken out and rinsed with 8:2 H2O:MeOH and small amounts of MeOH to remove excess ligands. PM-IRRAS spectra were recorded of Au11(PPh3)7Br3 in the beginning, during the reaction and of the final product, referred to as [Au11:GSH] in the following. For MALDI measurements of [Au11:GSH], a small amount of sample was scratched off the plate and dissolved in MeOH.
Ligand Exchange with F-GSH on an Al2O3 surface. The reaction was carried in the same way as the ligand exchange with GSH on an Al2O3 surface described in the last paragraph. A solution of Au11(PPh3)7Br3 (1 eq., 2.5 mg, 0.61 µmol) in 1-2 ml of DCM was dropcasted onto a polished Al2O3 surface (of an aluminum plate). Prior to use, the plates were polished with Metadi Diamond Polishing Compound (1 and ¼ micron, Buehler) and rinsed with EtOH. After evaporation of the solvent, the sample was placed in a solution of F-GSH (160 eq., 68 mg, 0.098 mmol) in 10 ml H2O:MeOH (8:2). After 1 day at room temperature, the plate was taken out and rinsed with 8:2 H2O:MeOH and small amounts of MeOH to remove excess ligands. The product is referred to as [Au11:F-GSH] in the following.
Ligand Exchange with GSH on a ZnSe crystal. A solution of Au11(PPh3)7Br3 (1 eq., 5 mg, 1.2 µmol) in 1-2 ml of EtOH was dropcasted onto a ZnSe ATR crystal. After evaporation of the solvent, the crystal was positioned in the flow cell. A 0.01 M solution of GSH in 8:2 H2O:MeOH was pumped through the cell at 0.05 ml/min over a period of 30 hours, allowing in-situ ATR-IR monitoring of the reaction. The solution was then removed and the product [Au11:GSH] rinsed with 8:2 H2O:MeOH and small amounts of MeOH to remove excess ligands. Spectra of the dry dropcast films of Au11(PPh3)7Br3, [Au11:GSH] and the exchange ligand GSH were also recorded.
Ligand Exchange with 2-PET on an Al2O3 surface. A solution of Au11(PPh3)7Br3 (1 eq., 3 mg, 0.73 µmol) in 1-2 ml of DCM was dropcasted onto a polished Al2O3 surface (on an aluminum plate). Prior to use, the plates were polished with Metadi Diamond Polishing Compound (1 and ¼ micron, Buehler) and rinsed with EtOH. After evaporation of the solvent, the sample was placed in a solution of 2-PET (200 eq., 20 µl, 0.15 mmol) in 10 ml of toluene. After one day at room temperature, the plate was taken out and rinsed with hexane and small amounts of DCM to remove excess ligands. A new solution of 2-PET (200 eq., 20 µl, 0.15 mmol) in 10 ml of 8:2 toluene:CHCl3 was prepared and the sample placed therein for another day. It was then again rinsed with hexane and DCM and then placed in a solution of 2-PET (200 eq., 20 µl, 0.15 mmol) in 10 ml of 7:3 toluene:CHCl3 for additional 2 days. Afterwards, it was washed with hexane and DCM for a final time. PM-IRRAS spectra were recorded of Au11(PPh3)7Br3 in the beginning, between the changes of the solvents and of the final product, referred to as [Au11:2-PET] in the following. For MALDI measurements of [Au11:2-PET], a small amount of sample was scratched off the plate and dissolved in MeOH.
The same experiment was also carried out using hexane instead of toluene as the main solvent, yielding a product [Au11:2-PET]'.

Characterizations Techniques
UV-Vis spectra of the clusters were recorded on either a PerkinElmer 750 Lambda or a Varian Cary 50 Bio spectrometer at room temperature, using quartz glass cuvettes with 1 cm path length. Depending on the nature of the sample, different solvents (DCM, THF, H2O) were used. Measurements with solid samples were performed using a PerkinElmer 750 Lambda spectrometer with an integrating sphere with an inner diameter of 60 mm. The spectra were recorded in diffuse transmittance mode, using cuvettes with 0.5 or 1 cm path length.
Photoluminescence spectroscopy was conducted employing an Edinburgh FSP920 photoluminescence spectrometer equipped with a XE900 Xenon arc lamp, Czerny-Turner monochromators TMS300 and a S900 photomultiplier R928 detector. A step size of 2 nm and 0.25 s dwell time were used. Both dissolved samples in different solvents (DCM, MeOH, H2O) in quartz glass cuvettes (90° geometry) and supported analytes on the aluminum sample plates were measured. Emission scans with a fixed excitation at 400 nm (solution, solid state) or at 450 nm (solid state) were recorded. When recording spectra of analytes in solution, a 455 nm cut-off filter was inserted after the sample to prevent that second order diffraction radiation was reaching the detector.
MALDI mass spectra were measured using either a prototype Axima MALDI ToF 2 time of flight (TOF) MS (Shimadzu, Kratos Analytical) or a Bruker Autoflex mass spectrometer. The spectra were obtained working at near threshold laser irradiances of the nitrogen laser in positive linear mode. Averaging 300-600 spectra of single laser pulses (λ = 337 nm at 50 Hz) obtained by rastering over the sample gave the spectra displayed within this publication. Depending on the samples, different matrixes (DCTB or THAP), solvents (DCM, CHCl3, MeOH) and analyte:matrix ratios (between 1:10 and 1:100) were used.
Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance-250 spectrometer at 250 MHz ( 1 H) or at 101.20 MHz ( 31 P). The chemical shift is reported in parts per million from tetramethylsilane ( 1 H) or 85% phorphoric acid ( 31 P). CDCl3 was used as solvent.
Polyacrylamide gel electrophoresis (PAGE) was performed on a Bio-Rad Protean II xi cell holding one or two 1.5x16x16 cm slab gels. The separation and extraction procedure was conducted according to published protocols, 12,13 with only slight differences. The gels were prepared native without SDS, with 30 wt% and 3 wt% monomer (acrylamide:bisacrylamide = 19:1) in separating and stacking gel, respectively. The samples were dissolved in 1:1 H2O/glycerol mixture and loaded into the wells of the stacking gel. A 1.92 M glycine and 250 mM tris base eluting buffer was used. The electrophoresis was run at constant voltage (150 V) for 6 hours. Afterwards, the gel pieces containing clusters were crushed and extracted in 4 °C cold water for 1 day. The gel residue was then removed by filtration and 2% acetic acid was added. The solution was then concentrated under reduced pressure and the clusters precipitated by addition of 1.5 ml of methanol and centrifugation. After redissolving in 600 µl of H2O, the precipitation and centrifugation were repeated twice with smaller volumes (1 ml and 0.75 ml) of methanol, to achieve higher purity.
ATR-IR was measured on a Bruker Vertex 70 FTIR spectrometer with a liquid nitrogen cooled MCT detector. A specac flow cell was used for the ligand exchange reaction, through which the solution of GSH in 8:2 H2O/MeOH was flown at low speed (0.05 ml/min). The sample was dropcasted on a ZnSe crystal (52(48) mm x 20 mm x 2 mm) with EtOH or H2O. All spectra were recorded at room temperature.
PM-IRRAS spectra were recorded on a Bruker Tensor 27 FT-IR spectrometer, with the external beam port being connected to a Bruker PMA 50 unit. A photoelastic modulator PEM 90 (Hinds) at 50 Hz was used for modulation; demodulation of the signal was achieved with a Stanford Research SR830 DSP lock-in amplifier. The spectrometer was equipped with a MCT detector, which was cooled with liquid nitrogen. Samples were mounted on an IRRAS sample holder and measurements were done at an angle of incidence of 80°. All spectra were recorded at room temperature, using a high pass filter with a cutoff at 3800 cm -1 . The samples were dropcasted centrally on small planar aluminum plates (see above) and the plates were dried for at least 5 hours before measuring. The same position on the plate was chosen for all measurements.  Table S1. With these ablation parameters it can be ensured that the Au complex is ablated completely, though sample thickness varies along the scan route due to sample preparation method.
For signal quantification a dried droplet approach combined with micro grooves is applied and intensity ratios are used for quantification 14,15 . Therefore liquid single standards of P, S (Specpure®, Alfa Aesar, ThermoFisher, Germany) are pairwise mixed with an Au (BDH Prolabo ®, VWR Chemicals, Belgum) single standard. For each analyte/Au pair, a separate calibration based on mass ratios is prepared. All further quantification is likewise based on the intensity ratio of each analyte respectively and the 197 Au signal. The calibration ranged from 0.01 -0.20 µg analyte/µg Au. Microgrooves were produced on aluminum substrates by using the laser ablation system. For each calibration solution four lines with 5 mm length, 0.4 mm distance in between, 100 µm spot size and 5.5 J/cm 2 were ablated. These microgrooves are filled with the liquid standard mixtures, the solvent evaporates and the remaining residue is ablated with the same instrument parameters as the samples. The transient signal, achieved by ablation of the samples or standards, was divided into several subsections. Of each subsection (region) an intensity ratio was formed. 24 of such ratios for one single analyte (6 regions for one line, 4 lines/sample) were averaged and regression curves could be obtained thereby. Within the calibration range correlation coefficients (COD) of r 2 = 0.98 for P/Au and r 2 = 0.96 for S/Au were calculated.

Ligand exchange reactions of Au11(PPh3)7Br3 with Thiolates in solution Ligand exchange with GSH in solution to Au25(SG)18
The exchange reaction was followed with UV-Vis and MALDI-MS. The decrease of the characteristic bands of Au11(PPh3)7Br3 in the organic layer over the course of the reaction is clearly visible in the UV-Vis spectra in Figure S2. The absorption in the H2O phase intensified during the ligand exchange, with the characteristic bands of Au25(SG)18 appearing at 410, 455 and 670 nm. Purification by PAGE then yielded Au25(SG)18 in high purity. When looking at the MALDI mass spectra of the CHCl3 phase ( Figure S3) recorded over the progress of the reaction, the vanishing of Au11(PPh3)7Br3 is clearly visible and most prominent in the first 4 hours of reaction. Afterwards, only smaller fragments are observed in the organic phase, which might be related to decomposed clusters.
Complimentary MALDI mass spectra of the aqueous phase containing the main product Au25(SG)18 could not be acquired at the instruments available.

Ligand exchange with 2-PET in solution to [Au25(PPh3)10(SC2H4Ph)5Br2] 2+
The exchange reaction was followed with UV-Vis and MALDI-MS. In UV-Vis ( Figure S4), the formation of the most characteristic bands of [Au25(PPh3)10(SC2H4Ph)5Br2] 2+ at ≈445 cm -1 and ≈680 cm -1 can be observed even after only 15 minutes. This indicates quite rapid exchange. After 24 hours, the spectrum clearly resembles that of [Au25(PPh3)10(SC2H4Ph)5Br2] 2+ . The same trend is observed by MALDI-MS in Figure S5. After only 15 minutes, the clusters have already started to grow, resulting in a polydisperse mixture. With progressing reaction, a size-focusing process occurs, which yields very pure [Au25(PPh3)10(SC2H4Ph)5Br2] 2+ after 24 hours.  In addition, as leaching of the clusters was indicated by a slight color change of the solution of the [Au11:2-PET] system, LA-ICP-MS measurements of the solution before and after the reaction were obtained ( Figure S8). An increase of the 197 Au signal was noted, confirming the assumed leaching to the solution. In addition, also an increase in the 31 P signal could be found, which indicates dissolution of ligand protected Au units. However, further assessment would be required to determine their exact composition. Figure S8: Comparison of the LA-ICP-MS intensities of the reaction solution of the [Au11:2-PET] system: before (orange) and after the reaction (green).

(S)TEM images of Au11, [Au11:GSH], [Au11:GSH] and Au25 nanoclusters
To further ensure the stability of the clusters upon exposure to excess thiols, the reacted products [Au11:GSH] and [Au11:2-PET] were investigated with TEM ( Figure S9a and Figure S9b). Similar to MALDI-MS, a small amount of dropcast sample was scratched off the support plate, dissolved in DCM and dropcasted onto a copper grid for microscopy. The distribution was found to be very uniform, with a particle size of ≈1 nm. Similar values were reported for supported Au11 before, 16 and could also be observed for the pure starting cluster Au11(PPh3)7Br3. HAADF-STEM ( Figure S9c) was chosen in this case, due to strong charging effects in TEM mode. To further confirm the different reactivities of Au11 in solution and on a surface, HAADF-STEM images of Au25(SC2H4Ph)18were obtained ( Figure S9d). They showed clearly identifiable isolated clusters of ≈1.2 nm size, which is significantly larger than Au11. This provides further evidence for the stability of the Au11 nanoclusters under the conditions of the ligand exchange, with no Au25 being formed. To exclude random absorption of the GSH ligands on the ZnSe crystal serving as support surface during the ATR experiment, a blank experiment was conducted. In that case, a solution of GSH in 8:2 H2O:MeOH was flown through the ATR cell for a duration of 6h, thereby being in contact with the blank crystal surface. As can be seen in Figure S13, the difference in absorbance of the dry crystal surface before and after the blank experiment is negligible and can be related to changes in the ambient gas atmosphere (small differences in the H2O and CO2 stretching vibration region). It is therefore assumed that binding of the GSH ligands to the ZnSe support does not play a significant role within this experiment.  The ligand exchange was conducted in the same way as described in the experimental section for product [Au11:2-PET], with the only exception being the use of hexane instead of toluene as main solvent. This caused only minimal differences in the PM-IRRAS and MALDI-MS spectra, as well as in the TEM images, as comparison of Figure S15 with the corresponding data of [Au11:2-PET] (see main part of the publication) shows. It is therefore concluded that the emerging products are of similar nature.

Additional photoluminescence spectra
Photoluminescence spectra were recorded for the unreacted Au11(PPh)3Br3 and the exchange products of the ligand exchange with GSH and 2-PET, Au25(SG)18 and [Au25(PPh3)10(2-PET)5Br2] 2+ , using excitation at 400 nm ( Figure S16). Due to instrumental limitations, the emission spectra could only be recorded up to 850 nm. As seen in Figure S16, Au11(PPh)3Br3 did not exhibit any photoluminescence in the recorded energy region, also when varying excitation energies. The emission spectrum of Au25(SG)18 showed a band with an onset at ≈620 nm, which is in agreement with recorded spectra 17,18 .
[Au25(PPh3)10(2-PET)5Br2] 2+ also showed photoluminescence activity, exhibiting an emission band with a maximum >850 nm, again resembling published spectra 19 . Spectral data of the as-synthesized F-GSH were also obtained ( Figure S17). The fluorescein labeled thiol showed prominent fluorescence above 500 nm. Figure S17: Absorption (left) and emission (right) spectrum of F-GSH, both measured as solution in methanol. The emission spectrum was recorded with a fixed excitation at 400 nm (the same as for the other dissolved analytes).
Photoluminescence spectra were also recorded of the products of the ligand exchange with Au11 on an alumina surface, [Au11:GSH] and [Au11:2-PET]. The spectra were obtained as solid-state photoluminescence spectrum. Due to instrumental limitations, the emission spectra could only be recorded up to 700 nm. Excitation at both 400 and 450 nm was tested, however, no major influence on the observed emission spectra was found. Figure S18 shows the emission spectra of the products in comparison to the spectrum of the exchange product with the fluorescein tagged GSH, [Au11:F-GSH].
As can be seen, the exchange with the F-GSH ligand induces fluorescence to the previously inactive Au11 clusters, whereas exchange with GSH and 2-PET does not evoke photoluminescence.