Combinatorial discovery of cosolvent systems for production of narrow dispersion thiolate-protected gold nanoparticles.

The effect of aqueous solvent concentration in the synthesis of water-soluble thiolate-protected gold nanoparticles (AuNPs) was investigated for 13 water-miscible solvents and three thiolate ligands (p-mercaptobenzoic acid, thiomalic acid, and glutathione). The results were analyzed by construction of heat maps that rank each reaction result for polydispersity. When solvents were organized in the heat map according to their Dimroth-Reichardt ET parameter (an approximate measure of polarity), two "hot spots" become apparent that are independent of the ligand used. We speculate that one hot spot may arise in part from the metal chelation or coordination ability of solvents that include diglyme, 1,2-dimethoxyethane, 1,4-dioxane, and tetrahydrofuran. The second hot spot arises at concentrations of alcohols including 2-propanol and 1-butanol that appear to selectively precipitate a growing product, presumably stopping its growth at a certain size. We observe some tightly dispersed products that appear novel. Overall, this study expands the number of tightly dispersed water-soluble AuNPs that can be directly synthesized.

Introduction

The landmark Brust–Schiffrin synthesis[1,2] and its derivatives[3−6] preceded a now large body of literature concerning thiolate-ligated gold nanoparticles (AuNPs). This synthesis generally produces polydisperse products which can be purified,[7] etched,[7,8] annealed or “size-focused”[9,10] to give products of atomically precise formulas.[11,12] The 5, 8, 14, 21, and 29 kDa products identified earlier[13−16] are now assigned as Au25(SR)18, Au36(SR)24, Au38(SR)24, Au40(SR)24, Au67(SR)35, Au102(SR)44, and Au144(SR)60.[17−22] The special stabilities of these clusters are explained by electronic or geometric shell filling.[23,24] The widespread adoption of these compounds by chemists,[25] biologists,[26] and physicists[27] is a testament to their robust nature.

Methods based on oxidative etching are now widespread for synthesis of the especially stable clusters.[10,28−31] These methods excel at isolating exceptionally stable clusters with nonpolar ligand shells. Also, the purification of well-defined nanoclusters from similarly sized clusters, such as the purification of Au38(SR)24 from Au40(SR)24, is increasingly well developed.[32,33] For water-soluble clusters (needed for biological applications), application of etching methods is much more limited, and is shown only for two cases so far.[34,35] Synthetic methods are also lagging for clusters that are insufficiently stable to survive the etching process.[21,36]

A less mature method for limiting the polydispersity of a cluster preparation involves selection of reaction conditions so that a single nanocluster product, rather than a product distribution, is favored. This method was developed for water-soluble clusters, notably the p-mercaptobenzoic acid (pMBA) protected Au144(pMBA)60 and Au102(pMBA)44 clusters.[36,37] Reaction conditions that produced these syntheses were found by screening each of the synthetic parameters in the Brust synthesis, including ligand:Au ratio, solvent composition, and reductant molar excess. The direct synthesis differs from etching methods in that ligand is generally not present in excess. Compared to the now relatively mature etching approaches, this direct-synthesis approach may allow production of “less stable” clusters, presently synthetically elusive clusters,[3,5,6] and a larger panel of biologically useful water-soluble clusters than is presently available.

Previous work speculated that solvent composition is among the most important parameters in this direct synthesis of nanoparticles.[7,37] Herein we attempt to gain additional insight into which solvent compositions are associated with narrow dispersion in the Brust-type synthesis. By screening in a combinatorial manner the synthetic effects of aqueous compositions of 13 water-miscible solvents, we observe the emergence of hot spots of overall solvent polarity associated with very narrow product distribution. Overall our results suggest the existence of several so far unidentified discrete water-soluble nanoparticles.

Results

Previous work suggests that two of the most important parameters for direct synthesis are ligand identity and solvent composition. We reasoned that systematic investigation of these two synthetic influences may reveal promising synthetic conditions for novel products. We tested the ligands p-mercaptobenzoic acid (pMBA), thiomalic acid (Tm), and glutathione (GSH, deprotonated: GS) in systematically varied aqueous mixtures of the solvents 1,4-dioxane (dioxane), dimethoxymethane (DMM), tetrahydrofuran (THF), dimethoxyethane (DME), diethylene glycol dimethyl ether (diglyme), N,N-dimethylformamide (DMF), dimethyl sulfoxide (DMSO), acetonitrile (MeCN), 2-propanol (iPrOH), 1-butanol (nBuOH), 1-propanol (nPrOH), ethanol (EtOH), and methanol (MeOH).

The general approach to the combinatorial screening of reaction conditions involved attempting as many as 96 reactions, examining the effect of two synthetic variables at a time, in a matrix as large as 12 × 8. Each reaction was analyzed initially by polyacrylamide gel electrophoresis[7,8] (PAGE). To enable facile comparison of reaction conditions, we scored each lane of analytical PAGE gel corresponding to each tested reaction condition. The scoring system extracts three key pieces of information on a direct synthesis condition: (1) how many products were produced; (2) the apparent dispersity of the products; (3) the relative sizes of the products. An example of how we scored a small set of reactions run in varying concentrations of aqueous ethanol is shown in Figure 1. Relative dispersity is shown as a heat value and relative size and number of products are depicted in the bottom panel. Overall, this scoring system allows compact comparison of the effects of solvent system and ligand choice in gold nanoparticle synthesis. While the PAGE gels give an excellent comparison of synthetic conditions, they may not fully account for minor products present.

Figure 1

Example of a scored PAGE gel, showing heat map that encodes dispersity, and a relative size graph. The darker red in the heat map indicates a more tightly dispersed product and the relative size graph on the bottom encodes the mobility (i.e., relative size) of the products within the same gel.

Solvent Effects on Synthesis of p-Mercaptobenzoic Acid (pMBA) Protected Clusters

The solvent screen with pMBA was carried out with a 1:3.4 Au/pMBA ratio with initial [Au] = 2 mM and 1 equiv of NaBH4 with respect to Au. These initial screening parameters are similar to those of the published Au102(pMBA)44 synthesis which serve as a good starting point for our initial screen.[16,31] The reactions were allowed to proceed for 17 h at 30 °C in a shaking incubator. All reactions were identical except for the solvent composition. We attempted synthesis in aqueous mixtures of each of the 13 solvents, with solvent compositions ranging from 10% to 80% solvent in water. The results of this solvent screen are shown in Figure 2.

Figure 2

Solvent screen results for the synthesis of gold nanoparticles using pMBA as the ligand. Left panel shows the relative size of the clusters as shown by PAGE analysis. The conditions that lack a size indicator indicate no visible product formation. The right panel shows the dispersity of the products with the darkest red being the most monodispersed product. The pale yellow squares indicate no visible product formation.

Figure 2 shows the relative size of the product (with the y-axis defined as 1 – Δx to resemble the bands on the original polyacrylamide gel, Δx is the distance the product traveled from the well). Some screened conditions yield either no product or nonvisible product and those conditions do not have a marker on the relative size graph and these are displayed as pale yellow space in the heat map. The Δx for the polydispersed products is measured in the center of the product streak.

The right panel of Figure 2 shows the dispersity of the product, with tighter dispersion indicated by darker red, and wider dispersion indicated by lighter colors. The x-axis of Figure 2 is organized according to the percentage cosolvent in water. We constructed preliminary heat maps with the solvent identity (y-axis) arranged according to dielectric constant, dipole moment, chelating ability, and density (Supporting Information Figures S1–S12) and observed that when solvents are organized according to polarity (or ionizing power, Dimroth–Reichardt ET parameter[9,10]) of the solvents, hot spots become apparent that are independent of ligand used. We discounted the possibility that the redox potential of the solvents may play a role in product formation because the Au is in redox-stable −Au(I)–SR– complexes before it is exposed to solvents other than water, and we do not see evidence for oxidation or reduction upon solvent addition as judged by stable color of the solutions that form.

Table 1 gives a listing of ET parameters for each solvent used. Figure 2 and other heat maps shown in this paper arrange the solvents from least-polar to most polar. The value of ET parameter of DMM was not available. For the purpose of the heat maps, DMM was placed between dioxane and THF using the dielectric constant of the three solvents.

Table 1

Solvent Used in the Synthetic Screening As Ranked in Dimroth–Reichardt ET parameter[35]

solvent	Dimroth–Reichardt ET parameter (kcal/mol)
1,4-dioxane	36.0
dimethoxymethane (DMM)	N/A
tetrahydrofuran (THF)	37.4
1,2-dimethoxyethane (DME)	38.2
diethylene glycol dimethyl ether (diglyme)	38.6
dimethylformamide (DMF)	43.8
dimethyl sulfoxide (DMSO)	45.1
acetonitrile (MeCN)	45.6
2-propanol (i PrOH)	49.2
1-butanol (n BuOH)	50.2
1-propanol (n PrOH)	50.7
ethanol (EtOH)	51.9
methanol (MeOH)	55.4

In Figure 2 there are two hot spots, one centered at 50% DMM and another centered at 40% nBuOH. On the basis of 1 – Δx values, the products created at these hot spots are not identical. Because of similarities between the hot spots found in the pMBA screen and those found for other ligands, we present a unified, expanded discussion below.

Using Au102(pMBA)44 and Au144(pMBA)60 as standard markers in each PAGE analysis, we found eight solvent conditions that make monodispersed products, as judged by the appearance of a single, very discrete band in PAGE. Some of these solvent conditions yield products that are in between Au102(pMBA)44 and Au144(pMBA)60. These conditions include (50–60% iPrOH, 40% THF, 50–60% dioxane, 30% diglyme, and 40% DME). As judged by relative gel mobility, it is likely that one or more of these products is novel, and not, for instance Au130.[36]

Other conditions are notable for making particles larger than Au144(SR)60, including 20–70% nBuOH, 80% MeCN, and 40–60% DMM. The conditions that makes that most monodispersed products are correlated to the darkest red spots in the heat map.

We further characterized the products of some of the reactions that produced narrow dispersity products by transmission electron microscopy. Figure 3 shows transmission electron micrographs for the 40% THF and 60% nBuOH conditions corresponding to products from each hotspot in Figure 2. TEM reveals narrow dispersity products in each case, with the 40% THF condition corresponding to 1.84 nm ±0.30 nm nanoparticles and the 60% nBuOH condition corresponding to 1.96 ± 0.29 nm. The approximate diameters correspond to molecular formulas of Au228pMBA75 and Au188pMBA66, respectively. Some of the apparent dispersity in this measurement may be attributed to the difficulty in determining the precise edge of sub-5-nm diameter particles in transmission electron micrographs—most of the measured dispersity may be inherent to the TEM technique itself as applied here. Some of the dispersity, especially larger particles, may also be attributed to electron-beam induced sintering of adjacent particles. The particles appear spherical in morphology in each case, and are of sufficiently tight dispersity to form extended 2D hexagonal lattices.

Figure 3

TEM image and histogram of the pMBA-protected gold nanoclusters synthesized using 40% THF (top, 1.84 ± 0.30 nm, Au188pMBA66) and 60% nBuOH (bottom, 1.96 ± 0.29 nm, Au228pMBA75).

After identifying the hotspots, we sought to improve the existing direct synthesis for Au102(pMBA)44. The existing literature synthesis uses 47% aq. MeOH, [Au] = 3 mM, [pMBA] = 12 mM, at a 0.036 mmol Au scale. Four solvent conditions emerged to make clusters that have the same 1 – Δx values on the PAGE analysis (80% nBuOH, 10% DMF, 10% diglyme, and 50% MeOH). A series of refinements, including the adjustment of solvent percentages in finer increments, order of addition, method of addition, temperature, [pMBA], [BH4–], degassed solvents vs nondegassed solvents, equilibration time of the polymer, and the addition of solid NaBH4 vs aqueous NaBH4 were carried out for the synthesis of Au102(pMBA)44. We came to a condition that allows us to make Au102pMBA44 in a 0.5 mmol Au scale in reasonable purity and yield. The reaction is 50 times larger than the original reactions scale and reproducibility may be improved. The synthetic condition changes to the published synthesis are minor, suggesting that the published synthesis was already very well optimized. We found that the published solvent (47% aq. MeOH) made the most monodispersed Au102(pMBA)44. Operational changes that improved the reproducibility and scalability include the rapid mixing of the reagents during polymer synthesis and using solid NaBH4 instead of an aqueous solution. Notably, the approach does not allow elimination of the second product almost always observed in Au102(pMBA)44 syntheses that is generally removed by fractional precipitation.

Solvent Effects on Synthesis of Thiomalate (Tm) Protected Clusters

Similar to the initial solvent screen with pMBA, the solvent screen with Tm as the ligand was carried out with 1:3 Au/Tm ratio with [Au] = 2 mM and 1.5 equiv of NaBH4 with respect to Au. The reactions were allowed to proceed for 17 h at 30 °C in a shaking incubator. The results of this solvent screen are shown in Figure 3.

A hotspot appears centered at 30% THF, with additional areas of tight dispersity observed in isolated conditions of 60 and 80% nBuOH, and 10% and 60–70% EtOH (shown as dark red squares in the heat map). Thus, the optimal solvent for synthesis of narrowly dispersed products using Tm as the ligand can be found in the low percentages of the lower polarity solvents, except in the case of nBuOH which appears to be advantageous to monodispersity at all percentages. The large hot spot at lower concentrations of solvent may share a phenomenological origin with the similar hotspot for pMBA, with a shift toward lower cosolvent concentrations arising because Tm is more charged (thus more polar) compared to pMBA.

Many of the solvents result in two distinguishable tight bands (dioxane, DME, MeCN, nPrOH, and EtOH), as shown in the depiction of two products in the left panel. Since the heat map considers the entire reaction, some of these conditions are scored as polydisperse, even though they may contain notably monodisperse individual products.

The smaller band in each of the two product producing reactions can be further resolved into three products, notable for their distinctive colors (red, green, and yellow; Figure 4, left). These are similar to the colors reported by Tsukuda,[37] for GSH protected clusters smaller than Au25(SG)18. The color order does not match, however, so suggestions of molecular formula cannot be made by analogy, other than these are likely to be clusters smaller than Au25(SR)18.

Figure 4

Solvent screen results for the synthesis of gold nanoclusters using Tm as the ligand. The left panel shows the relative size of the clusters as shown by PAGE analysis. The conditions that lack a size marker indicate no visible product formation. The conditions that have two size markers indicate the formation of two relatively tight bands. The right panel shows the dispersity of the products with the darkest red being the most monodispersed product. The pale yellow squares indicate no visible product formation.

To demonstrate the potential of systematic screening of reaction conditions, we further refined the synthesis that produces this set of three products, so that that only the red product is produced (Figure 4, right). The refinement process included adjustment of temperature, Au:Tm ratio, pH, reaction time, [NaOH], amount of O2 in the solvent, and [BH4–] (see Experimental Procedures for the synthesis of the red product). The parameters that have been adjusted are commonly screened parameters in the refinement of gold nanocluster syntheses in our laboratory. The refinement process is mainly determined by the results of the previous screen until no further optimization can be obtained. Some parameters, such as [NaBH4], need to be revisited after other parameters are set.

Solvent Effects on Synthesis of Glutathione (GSH) Protected Clusters

The solvent screen with GSH as the ligand was carried out with 1:3 Au/GSH with [Au] = 2 mM and 3 equiv of BH4– with respect to Au. The reactions were allowed to proceed for 2 h and 15 min at 4 °C. These conditions were the result of an initial optimization. The result of this solvent screen is shown in Figure 5. As in the case for other ligands, we observe two hot spots in the heat map. One can be found in the low to mid percentages (20–40%) of low polarity solvents (DME and diglyme), the second is found in the low percentages (10–30%) of the alcohols (iPrOH, nBuOH, and nPrOH).

Figure 5

Refinement progress of the “triple product” band to a red band. The PAGE lane on the left is the “triple product” band before the refinement process and the PAGE lane on the right is the resulting red band after the refinement process.

The products observed at the hot-spots when GSH is used as a ligand are less than 1.5 nm in diameter and essentially of insufficient size for routine and accurate analysis by transmission electron microscopy. To characterize the approximate composition and dispersity of GSH protected nanoparticles we used size exclusion chromatography. A Superdex 75 gel filtration column (GE Health Sciences) is used routinely to separate proteins in the 3 to 70 kDa range. By comparison of actual and computed Stokes–Einstein radii, we estimate that the column is capable of resolving spherical clusters with approximate formulas of Au10(SG)11 to Au998(SG)161. We calibrated with protein standards as described in the methods section and shown in Supporting Information Figure S18. We tested experimentally the elution of the product synthesized at 76% aqueous diglyme, which appears to contain several products as observed in the PAGE, although one product appears more abundant (Supporting Information Figure S16). By chromatographic analysis we see that the 76% diglyme product also contains many products and similar to the appearance in the gel, a single product dominates. The dominant product has a Stokes–Einstein radius of 1.11 nm (Figure 7), corresponding to an approximate molecular formula of Au53(GS)26. To arrive at this approximate molecular formula, we assume that GSH contributes 0.35 nm to the total hydrodynamic diameter of 2.44 nm, as calculated previously.[11] The remaining diameter is occupied by gold atoms. The full-width at half-maximum of the main peak is slightly broader than that of the protein standards injected, consistent with a the appearance of the second most abundant product in the gel at lower molecular weight, and the tailing at what corresponds to lower molecular weight also suggests that the PAGE gel is an accurate depiction of the dispersity of the sample.

Figure 7

Superdex 75 Elution Profile for GS protected nanoparticles synthesized in 76% diglyme.

One goal at the outset of this screen was to attempt to find a direct, large scale synthesis of Au25(GS)18.[7,35,38] We found that one of our screened solvent conditions produced a product with the expected orange appearance, and approximately expected gel mobility. MALDI characterization of this product is not consistent with Au25(GS)18 and is instead consistent with a novel nanocluster with properties that are the subject of a future report. It is presently unclear if it is possible to directly synthesize Au25(GS)18 with this direct synthesis approach.

Discussion

Two hot spots of high monodispersity appear in each of the combinatorial solvent screens, independent of the organothiolate ligand that protects the resulting nanoparticles. One of the hot spots occurs in lower polarity solvents such as dioxane, DME, THF, and diglyme. These solvents are interesting in that, with the exception of THF, each is potentially both a solvent and a metal chelating ligand (Figure 6). These coordinating or chelating solvents may interact with either the (Au-SR)n polymer precursors, perhaps to enforce a particular polymer size prior to the NaBH4 reduction step. This is suggested by others as a means to narrow the dispersity of cluster syntheses.[39] These solvents may also interact competitively with organothiolates to occupy the Au0 surfaces of nanoclusters. The multidentate nature of these ligands and their extraordinarily high concentration relative to thiols may make these oxygen-containing molecules capable of such competition. We speculate that either one of these effects may help to stabilize a particular cluster molecular formula, leading to the observed narrow dispersity.

Figure 6

Solvent screen results for the synthesis of gold nanoclusters using GSH as the ligand. Left panel shows the relative size of the clusters as shown by PAGE analysis. The conditions that lack a size indicator indicate no visible product formation. The conditions that have two size markers indicate the formation of two relatively tight bands. The right panel shows the dispersity of the products with the darkest red being the most monodispersed product. The pale yellow squares indicate no visible product formation.

The other hot spot appears for alcohols with longer hydrophobic chains, such as iPrOH, nBuOH, and nPrOH. We speculate that this hot spot arises because the nanoclusters grow until they reach a size at become insoluble in the aqueous alcohol mixture, at which point they precipitate. The precipitated nanoclusters then cease to grow. Under these circumstances, growing and completed clusters are separated in two physical phases, resulting in narrow size distribution.

This work highlights several solvents that do not correlate with monodisperse product preparation. Solvents that do not appear to produce narrow dispersion products include MeOH DMF, DMSO, and MeCN. The absence of monodisperse products in methanol is of special note. Most water-soluble thiolate ligated gold nanoclusters reported to date are synthesized in aqueous methanol.[7,35−38,40−43] We observe here that methanol consistently results in more polydispersed product than its more nonpolar and more bulky counterparts.

Conclusions

Early work on thiolate protected gold nanoclusters identified the approximate masses of especially abundant (magic-sized) clusters. Etching or size-focusing methods now reliably produce these clusters in many laboratories around the world with organic ligand shells. Synthetic methods for making water-soluble clusters and also particles that do not conform to the “magic” sizes are not yet widely available. By combinatorial screening of solvent conditions we observe hot spots of low product dispersity, suggesting routes for direct synthesis of well-defined water-soluble particles. We also observe in some cases the synthesis of well-defined products of sizes apparently in between the magic sizes. For instance we observe five solvent conditions that produce narrow dispersity products that are between Au102(pMBA)44 and Au144(pMBA)60. From the combinatorial solvent screen, there seem to be two modes of solvent effects: chelating or coordinating ability of the solvent and cluster solubility in the solvent. Full development of either one of these modes may greatly expand the availability of water-soluble clusters for applications in biology, for instance as contrast markers,[44] delivery vectors,[4] or RF heating antennae.[45]

Experimental Procedures

Materials

All commercially available reagents were used without further purification. Tetrachloroauric (III) acid (HAuCl4·3H2O 99.99% metal basis, Alfa Aesar), p-mercaptobenzoic acid (>95.0%, TCI America), l-glutathione reduced (≥98%, Sigma-Aldrich), thiomalic acid (≥99.0%, Sigma-Aldrich), NaBH4 (98–99%, MP Biomedicals), 1,4-dioxane (ACS grade, Mallinckrodt Chemicals), tetrahydrofuran (ACS grade, Mallinckrodt Chemicals), dimethoxyethane (99+%, Alfa Aesar), diethylene glycol dimethyl ether (99%, Sigma-Aldrich), dimethylformamide (reagent grade, Amresco), dimethyl sulfoxide (99.9% Fisher Scientific), acetonitrile (ACS grade, Fisher Scientific), 2-propanol (ACS grade, Fisher Scientific), 1-butanol (ACS grade, Fisher Scientific), 1-propanol (99.0%, Mallinckrodt Chemicals), ethanol (200 proof ACS grade, Pharmco-AAPER), methanol (99.9%, Fisher Scientific), and dimethoxymethane (98%, Alfa Aesar). Nanopure water (resistivity 18.2 MΩ-cm) was produced with a Barnstead NANOpure water system.

Solvent Screen Procedures

Solvent Screen with p-Mercaptobenzoic Acid as the Ligand

Synthesis of Polymer with p-Mercaptobenzoic Acid:

p-Mercaptobenzoic acid (3.4 mmol, 0.524 g, 3.4 equiv) was dissolved in H2O (24 mL), and 10 M aq. NaOH (1.6 mL) was added to the suspension. The resulting solution was mixed; the pH was determined to be >13, and the solution was diluted with H2O to a final volume of 50 mL. HAuCl4·3H2O (1 mmol, 0.394 g, 1 equiv) was dissolved in H2O (50 mL) in a separate beaker. The pMBA solution and the gold solution were mixed to yield a clear bright red solution. The bright red solution was stirred at rt overnight, which changes to yellow after about 1 h.

96-Well Plate Solvent Screen:

Fifty microliters of the above polymer solution was distributed in each well using a multichannel pipet, and the appropriate amount of organic solvent was added to each well (e.g., 25 μL for 10%, 200 μL for 80%); then, it was backfilled with the appropriate amount of H2O (e.g., 175 μL for 10%, 0 μL for 80%) to result in a final reaction volume of 250 μL. Four microliters of freshly made 0.125 M aq. NaBH4 was added to each well with a multichannel pipet. The 96-well plate was then placed in an incubating shaker at 30 °C for 17 h MeOH (1 mL) and 2.0 M NH4OAc (25 μL) was added to each well and the 96-well plate was centrifuged in a swinging bucket rotor at 4000 rpm and 4 °C for 10 min. The clear and colorless supernatant was then removed with a multichannel pipet and the precipitate was air-dried. Gel electrophoresis visualization was run on a 15% polyacrylamide gel (19:1, acrylamide/bis(acrylamide)) at 110 V for 1.5 h. The nanoparticle bands were visible by eye and with a UV transilluminator, thus no staining steps were performed for visualization.

Solvent Screen with Thiomalic Acid as the Ligand

Synthesis of Polymer with Thiomalic Acid:

Thiomalic acid (3.0 mmol, 0.450 g, 3.0 equiv) was dissolved in H2O (20 mL), 10 M aq. NaOH (2.0 mL) was added to the suspension. The resulting solution was mixed; the pH was checked to be >13, and the solution was diluted with H2O to a final volume of 50 mL. HAuCl4·3H2O (1 mmol, 0.394 g, 1 equiv) was dissolved in H2O (50 mL) in a separate beaker. The thiomalic acid solution and the gold solution were mixed to yield a clear bright red solution. The bright red solution was stirred at rt overnight, which changes to yellow after about 1 h.

96-Well Plate Solvent Screen:

Fifty microliters of the above polymer solution was distributed in each well using a multichannel pipet, and the appropriate amount of organic solvent was added to each well (e.g., 25 μL for 10%, 200 μL for 80%); then, it was backfilled with the appropriate amount of H2O (e.g., 175 μL for 10%, 0 μL for 80%) to result in a final reaction volume of 250 μL. Three microliters of freshly made 0.25 M aq. NaBH4 was added to each well with a multichannel pipet. The 96-well plate was then placed in an incubating shaker at 30 °C for 17 h. MeOH (1 mL) and 2.0 M NH4OAc (25 μL) was added to each well and the 96-well plate was centrifuged in a swinging bucket rotor at 4000 rpm and 4 °C for 10 min. The clear and colorless supernatant was then removed with a multichannel pipet and the precipitate was air-dried. Gel electrophoresis visualization was run on a 15% polyacrylamide gel (19:1, acrylamide: bis(acrylamide)) at 110 V for 1.5 h. The nanoparticle bands were visible by eye and with a UV transilluminator, thus no staining steps were performed for visualization.

Direct Synthesis of Thiomalic Acid-Protected Red AuNP (Figure 4, Right)

In a 15 mL conical tube, 1 mL of Au–Tm polymer solution (from above, 0.01 mmol Au), 2.5 mL of DME, and 1.5 mL of H2O were added in order. The reaction was allowed to cool in a 4 °C fridge for 10 min; 60 μL of 0.25 M NaBH4 (0.015 mmol, 1.5 equiv. with respect to Au) was added, and the reaction was gently mixed and left to sit at 4 °C (without mixing/shaking) for 3 days. MeOH was then added to the reaction to a total volume of 15 mL; 200 μL of 2.0 M NH4OAc was also added, and the reaction was then shaken to mix and was centrifuged at 4 °C for 10 min. The supernatant was removed, and the resulting pellet (red product) was air-dried. Gel electrophoresis visualization was run on a 22% polyacrylamide gel (19:1, acrylamide: bis(acrylamide)) at 110 V for 1.5 h. The nanoparticle bands were visible by eye and with a UV transilluminator, thus no staining steps were performed for visualization.

Solvent Screen with Glutathione as the Ligand

96-Well Plate Solvent Screen with l-Glutathione as the Ligand:

HAuCl4·3H2O (111.2 mg) was dissolved in H2O (2.824 mL) in a 15 mL conical tube to yield a 100 mM Au solution. l-Glutathione (169.4 mg) was dissolved in 0.3 M aq. NaOH (5.512 mL) in a separate conical to yield a 100 mM glutathione solution (pH = 9.5). Ten microliters of Au solution and then 30 μL of glutathione solution were added to each well of a 96-well plate. The appropriate amount of H2O (e.g., 400 μL for 10% and 50 μL for 80%) and then the appropriate amount of organic solvent (e.g., 50 μL for 10% and 400 μL for 80%) were added. The resulting solution was shaken at 4 °C for 45 min. Ten microliters of freshly made 300 mM aq. NaBH4 was added to each well (500 μL total reaction volume), and the plate was shaken at 4 °C for 2 h and 15 min. MeOH (1 mL) and 2.0 M NH4OAc (25 μL) was added to each well and the 96-well plate was centrifuged in a swinging bucket rotor at 4000 rpm and 4 °C for 10 min. The supernatant was then removed with a glass pipet connected to an aspirator, and the precipitate was dried under vacuum overnight. Gel electrophoresis visualization was run on a 22% polyacrylamide gel (19:1, acrylamide: bis(acrylamide)) at 110 V for 1.5 h. The nanoparticle bands were visible by eye and with a UV transilluminator; thus, no staining steps were performed for visualization.

Transmission Electron Microscopy

Imaging was performed with a JEOL 1400 at an acceleration voltage of 100 kV and images were recordeed on Orius SC1000 (∼4kx3k). The gold nanocluster samples were applied to carbon support film on 400 mesh copper speciment grids. The excess liquid was blotted with a piece of filter paper, and the grids were allowed to dry in air.

Size Exclusion Chromatography

Separation of cluster compounds were conducted on an ÄKTApurifier HPLC system equipped with a Superdex 75 PC 3.2/30 (2.4 mL) size exclusion column. Detection was performed at a wavelength of 254 nm. Samples were injected in Nanopure H2O and eluted at 0.02 mL/min in phosphate buffer (50 mM sodium phosphate, 150 mM NaCl, pH 7.0). Thyroglobulin, γ-globulin, ovalbumin, myoglobin, aprotinin, and vitamin B-12 correesponding to Stokes–Einstein radii of 8.6, 5.1, 2.8. 1.9, 1.35, and 0.85 nm were used to generate a calibration curve for the column.