Glutathione Disulfide as a Reducing, Capping, and Mass-Separating Agent for the Synthesis and Enrichment of Gold Nanoclusters

Water-soluble nanoclusters, which are facilely enrichable without changes in the original properties, are highly demanded in many disciplines. In this contribution, a new class of gold nanoclusters (AuNCs) was synthesized using glutathione disulfide (GSSG) as a reducing and capping agent under intermittent heating mode. The as-prepared GSSG–AuNCs had a higher quantum yield (4.1%) compared to the conventional glutathione-protected AuNCs (1.8%). Moreover, by simply introducing the GSSG–AuNC solution to acetonitrile at a volume ratio of 1:7, a new bottom phase was formed, in which GSSG–AuNCs could be 400-fold enriched without changes in properties, with a percentage recovery higher than 99%. The enrichment approach did not need additional instruments and was potentially suitable for large-scale enrichment of nanoclusters. Further, density functional theory calculations indicated that the hydrogen bonding between GSSG and acetonitrile plays a key role for the bottom phase formation. Our work suggests that the highly emissive GSSG–AuNCs possess great potential not only in fluorescent measurements but also in other scenarios in which high-concentration AuNCs may be needed, such as catalysis, drug delivery, and electronic and optical industries.


Heating Modes
Two heating modes were employed in synthesis, viz., continuous and intermittent heating. For a typical continuous heating process, the reactants were held at 80 • C by keeping the round-bottomed bottle in a water bath. In a typical intermittent heating experiment, the round-bottomed bottle was placed in the water bath at 80 • C and ambient temperature, alternately, each for 4 h. The times needed for the reaction solution in the round-bottomed bottle to reach the desired temperatures were measured, being 14.5 min in the heating process, allowing the temperature of the reaction solution to increase from 24.7 • C to 80.0 • C (average 3.81 • C/min), and 13.4 min in the cooling process, during which the temperature of the reaction solution decreased from 80 • C to 24.7 • C (average 4.13 • C/min). For the cooling process, the room-temperature cooling water in the water bath was refreshed every 3 min until the temperature of the reaction solution reached equilibrium.

Synthesis of GSSG-AuNCs
Under vigorous stirring, freshly prepared aqueous solutions of 8 mL of 20 mM HAuCl 4 and 6 mL of 100 mM glutathione disulfide were mixed with 66 mL of triple-distilled water in a round-bottomed flask at ambient temperature. Two minutes later, a 24 h heating process, either under continuous mode or under intermittent heating mode, was initiated by placing the bottle in the 80 • C water bath under gentle stirring. After reaction, the precipitates were removed by centrifugation at 2000 rpm for 10 min, and the supernatant was stored at 4 • C until use.

Synthesis of GSH-AuNCs
For the continuous heating mode, the procedures in the literature [35] were followed. Briefly, the freshly prepared aqueous solutions of 8 mL of 20 mM HAuCl 4 and 2.4 mL of 100 mM glutathione were mixed first. The mixture was then diluted with triple-distilled water to 80 mL at room temperature, followed by a 24 h continuous heating under gentle stirring. For the intermittent heating mode, the parameters same as those of GSSG-AuNCs were adopted. The orange-emitting GSH-AuNCs obtained were stored at 4 • C until use.

Characterization
UV-VIS absorption spectra were recorded on a TU-1901 UV-VIS spectrophotometer (Beijing Purkinje General Instrument, Beijing, China) using a 1 cm cell. PL spectra and phosphorescence decay experiments were performed on an FLS980 spectrometer (Edinburgh Instruments Ltd., Livingston, UK). The absolute QYs were measured with a C11347-11 Quantaurus-QY absolute PL quantum yield spectrometer (Hamamatsu, Japan). Transmission electron microscopy (TEM) images of the gold nanoclusters were taken on a Talos F200S transmission electron microscope (FEI, Hillsboro, OR, USA) operated at 200 kV. Dynamic light scattering (DLS) measurements were performed for colloidal solutions using a ZetaPlus laser light scattering system (Brookhaven, NY, USA). Matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF MS) was carried out on a MALDI-TOF/TOF 5800 analyzer (ABSciex, Framingham, MA, USA) equipped with a neodymium:yttrium-aluminum-garnet laser source emitting at a wavelength of 349 nm, using α-cyano-4-hydroxycinnamic acid (CHCA) as the matrix. X-ray photoelectron spectroscopy (XPS) was performed using an ESCSLAB 250XI X-ray photoelectron spectrometer (Thermo Fisher Scientific, Leicestershire, UK) equipped with a monochromatized Al-Kα source at a power of 2721 eV. Inductively coupled plasma-atomic emission spectrometry (ICP-AES) measurements were carried out on a SPECTRO ARCOS EOP axial-view inductively coupled plasma spectrometer (SPECTRO Analytical Instruments GmbH, Kleve, Germany). The plasma power was 1.28 kW, and the coolant, auxiliary, and nebulizer flows were set at 13, 0.8, and 0.8 L/min, respectively.

Acetonitrile-Initiated Phase Separation
Typically, 1 mL of GSSG-AuNCs was transferred into a centrifuge tube containing 7 mL of acetonitrile. The mixture was shaken thoroughly for 10 s and then was allowed to stand for 2 h or was centrifuged at 2000 rpm for 10 min, followed by phase separation at ambient temperature. The upper phase was transferred to another clean centrifuge tube.
To investigate the luminescent properties of the AuNCs in the upper and lower phases after the enrichment, the acetonitrile in the solutions was removed by reduced-pressure evaporation at 70 • C. The residues were dissolved, separately, in 1 mL triple-distilled water and stored at 4 • C before analysis.

Density Functional Theory Calculation
Density functional theory (DFT) computations were performed following the protocols of a previous work using the Gaussian 16 package [36] with modifications. The geometric structures of GSSG-acetonitrile and GSH-acetonitrile complexes were optimized using the basis set of 6-31G(d,p) in water, and the basis set def2tzvp was employed to calculate the adsorption energy. The M06-2X method was adopted for all calculations. Based on the theory [37], the adsorption energy E ads was evaluated by the following formula where E (ACN+GSSG/GSH) , E ACN , and E (GSSG/GSH) represent the energies of the optimized combined structure (GSSG-ACN or GSH-ACN), acetonitrile, and GSSG or GSH, respectively.

Synthesis of GSSG-AuNCs
Experimental conditions such as the initial concentration ratio of the starting materials, heating temperature, heating time, and heating mode affected the luminescence properties of GSSG-AuNCs. Note that in the widely adopted protocols for preparing the GSH-protected AuNCs, the initial concentration of HAuCl 4 was 2 mM [35]; the influence of the concentration ratio between GSSG and HAuCl 4 in this experiment was investigated by keeping HAuCl 4 at 2 mM, while varying the concentration of GSSG for rational comparison. Upon increasing the GSSG concentration, the emission intensity of the resultant GSSG-AuNCs was enhanced due to the increasing amount of luminescent nanoclusters generated, reaching a maximum at a GSSG:HAuCl 4 ratio of 3.75:1 (Figure 1a,b). Further increasing the GSSG content resulted in reduced fluorescence of the solution because a large amount of precipitations were found at the bottom of the flask. High reaction temperature favors the formation of nanoclusters; the emission intensity, after reaction for 24 h, increased from 1.66 × 10 4 at 30 • C to 2.31 × 10 5 at 70 • C, reaching a maximum at 80 • C (Figure 1c,d). Regarding the heating time, enhanced luminescence was obtained with elongated reaction time until 24 h; further increasing the reaction time reduced the emission intensity (Figure 1e,f). It was observed that intermittent heating yielded stronger fluorescent AuNCs. For the GSSG-AuNCs obtained under continuous heating mode (trace O2, Figure 2), the emission intensity was 2.73 × 10 5 ; however, the emission intensity was 3.23 × 10 5 for the GSSG-AuNCs synthesized under intermittent heating mode (trace O3).
Disulfide bonds are reductive, and the literature suggests that in the presence of chloroauric acid (HAuCl 4 ), a disulfide compound (denoted RSSR, for example) can be oxidized to RSSO 2 R, RSO 2 H, and RSO 3 H [30]. Similar to the GSH-mediated AuNC synthesis process, the reaction between HAuCl 4 and GSSG involves two critical steps: (i) The chloroauric acid was mixed with excess GSSG at room temperature, and Au(III) was quickly reduced to Au(I) within 2 min (Au(I) was detected by XPS analysis; Figure S1a), forming Au(I)-GSSG complexes. (ii) The solution was then heated in a water bath, and Au(I)-GSSG was further reduced by GSSG, forming Au nanoclusters (denoted as GSSG-AuNCs; Au(0) was detectable by XPS; refer to Figure S1a), with the Au crystal cores being capped by Au(I)-GSSG motifs. The results obtained by altering the reaction temperature, reaction time, and heating mode reveal that the properties of AuNCs were influenced by the reaction kinetics. Two reactions took place in opposite ways, affecting the nanocluster structures during formation of the AuNCs. One is the reduction of the Au(I) ions to Au(0) atoms, and the other is the etching of the nanocluster cores by the thiolates [38]. After the nanoclusters were formed, excessive heating, whether by elevating the heating temperature or by prolonging the heating time, favors the etching reactions. thesis process, the reaction between HAuCl4 and GSSG involves two critical steps: (i) The chloroauric acid was mixed with excess GSSG at room temperature, and Au(III) was quickly reduced to Au(I) within 2 min (Au(I) was detected by XPS analysis; Figure S1a), forming Au(I)-GSSG complexes. (ii) The solution was then heated in a water bath, and Au(I)-GSSG was further reduced by GSSG, forming Au nanoclusters (denoted as GSSG-AuNCs; Au(0) was detectable by XPS; refer to Figure S1a), with the Au crystal cores being capped by Au(I)-GSSG motifs. The results obtained by altering the reaction temperature, reaction time, and heating mode reveal that the properties of AuNCs were influenced by the reaction kinetics. Two reactions took place in opposite ways, affecting the nanocluster structures during formation of the AuNCs. One is the reduction of the Au(I) ions to Au(0) atoms, and the other is the etching of the nanocluster cores by the thiolates [38]. After the nanoclusters were formed, excessive heating, whether by elevating the heating temperature or by prolonging the heating time, favors the etching reactions.  Based on the experiments, the strongest photoluminescent GSSG-AuNCs could be obtained through intermittently heating the mixture of 2 mM HAuCl 4 and 7.5 mM GSSG for 24 h at 80 • C and ambient temperature alternately, each for 4 h. Unless otherwise stated, the GSSG-AuNCs obtained under optimal conditions were used throughout the experiments.  The synthesis conditions for R2 were adopted from the literature [35].
Based on the experiments, the strongest photoluminescent GSSG-AuNCs cou obtained through intermittently heating the mixture of 2 mM HAuCl4 and 7.5 mM for 24 h at 80 °C and ambient temperature alternately, each for 4 h. Unless othe stated, the GSSG-AuNCs obtained under optimal conditions were used througho experiments.

Morphological and Compositional Investigations
To understand the mechanism of higher QYs of the GSSG-AuNCs, TEM, MALDI-TOF MS, and XPS experiments were carried out, and the results were car analyzed.
The TEM image (Figure 3a) showed that the as-prepared GSSG-AuNCs spherical, with an average diameter of 1.97 ± 0.25 nm. As observed by high-reso transmission electron microscopy (HR-TEM), the lattice fringes of the GSSG-A were well resolved (inset, Figure 3a). The interplanar distances of 0.23 nm and 0. can be assigned to the (111) and (200) lattice planes of gold nanocrystals, respec However, although the GSH-AuNCs possess a similar size and morphology as G

Photophysical Properties
Under a UV lamp, the Au(I)-GSH and Au(I)-GSSG complexes formed in the first step did not emit noticeable luminescence (Figure 2, left panel), whereas the GSH-and GSSG-protected AuNCs emitted red and amber fluorescence, respectively. The GSSG-AuNCs possessed maximal excitation and emission wavelengths at 424 nm and 610 nm, respectively. The photoluminescence lifetime measurement of the GSSG-AuNCs showed a microsecond-scale average lifetime (τ1 = 2.28 µs, 46.19%, τ2 = 9.45 µs, 53.81%, Figure S2), indicating that the luminescence is attributed to ligand-to-metal charge transfer (LMCT) or ligand-to-metal-metal charge transfer (LMMCT) that generates radiative relaxation via a metal-centered triplet state [39]. However, the luminescence intensities of the GSHprotected AuNCs obtained under both continuous and intermittent heating modes were weaker (Figure 2, traces R2 and R3). The absolute QY for GSH-AuNCs obtained under optimal conditions was determined to be 1.8%, which was similar to the reported values, i.e., 1% [40], 2% [41,42], and 1.2% [11]. In contrast, the GSSG-AuNCs possessed significantly higher luminescent efficacy, offering an absolute QY of 4.1%.

Morphological and Compositional Investigations
To understand the mechanism of higher QYs of the GSSG-AuNCs, TEM, DLS, MALDI-TOF MS, and XPS experiments were carried out, and the results were carefully analyzed.
The TEM image (Figure 3a) showed that the as-prepared GSSG-AuNCs were spherical, with an average diameter of 1.97 ± 0.25 nm. As observed by high-resolution transmission electron microscopy (HR-TEM), the lattice fringes of the GSSG-AuNCs were well resolved (inset, Figure 3a). The interplanar distances of 0.23 nm and 0.21 nm can be assigned to the (111) and (200) lattice planes of gold nanocrystals, respectively. However, although the GSH-AuNCs possess a similar size and morphology as GSSG-AuNCs, their lattice stripes are hard to recognize ( Figure S3). The TEM results partially interpret the slightly shorter emission wavelength of the GSSG-AuNCs (610 nm) relative to that of the GSH-AuNCs (630 nm, Figure 1). As proposed by other researchers [43], disruption of the crystal packing leads to red shift of the nanocluster photoluminescence.
to determine the size distribution of the AuNCs from the diagram because the scattered light intensity of individual particle increases dramatically with the particle size [44], we can deduce from the results that the GSSG-AuNCs possess a higher aggregation degree in aqueous solution. The dense aggregation of GSSG-AuNCs increases intra-and inter-complex aurophilic Au(I)-Au(I) interactions, while restraining intramolecular vibrations and rotations of the Au(I)-GSSG complexes, and as a consequence, enhanced photoluminescence can be obtained [35]. The spectra indicate that GSSG-AuNCs and GSH-AuNCs have similar ionization and core fragmentation patterns, implying that they have the same core composition. However, the GSSG-AuNCs had relatively stronger fragment intensities in DLS measurements showed that the hydrodynamic diameter of the GSH-AuNCs was approximately 600 nm, while hydrodynamic diameters larger than 1000 nm were detected in a significant portion of the GSSG-AuNCs (Figure 3b). Although it is difficult to determine the size distribution of the AuNCs from the diagram because the scattered light intensity of individual particle increases dramatically with the particle size [44], we can deduce from the results that the GSSG-AuNCs possess a higher aggregation degree in aqueous solution. The dense aggregation of GSSG-AuNCs increases intra-and inter-complex aurophilic Au(I)-Au(I) interactions, while restraining intramolecular vibrations and rotations of the Au(I)-GSSG complexes, and as a consequence, enhanced photoluminescence can be obtained [35].
The molecular compositions of AuNCs were determined by MALDI-TOF MS under negative ion mode. Note that the MALDI analysis of water-soluble nanoclusters is challenging; nevertheless, the spectra of the AuNCs were successfully recorded (Figure 3d). The most abundant peak at m/z 5324.99 can be assigned to [Au 25 S 8 + 4Cl] − , and a series of peaks with m/z spacings of 32 and 197 are the consequences of the loss of [S] and [Au], respectively. The spectra indicate that GSSG-AuNCs and GSH-AuNCs have similar ionization and core fragmentation patterns, implying that they have the same core compo-sition. However, the GSSG-AuNCs had relatively stronger fragment intensities in the m/z range of 2000-4000, while the GSH-AuNCs had a higher abundance for fragments beyond m/z 5000. The results indicate that the water-soluble AuNCs obtained are a mixture of nanoclusters with different molecular weights; this observation agrees with the previous reports on GSH-AuNCs [45]. Moreover, previous research based on GSH-AuNCs revealed that the length of the Au(I)-thiolate motif varies with the nanocluster molecule size and that the QY generally increases with the decreasing nanocluster molecular weight [39]. Our results are in line with the reports. The GSSG-AuNCs, which possess relatively high abundance of smaller fragments, offer higher QYs.
X-ray photoelectron spectroscopy (XPS) studies depicted two core peaks with S 2p doublets (Figure 3c). The peak of lower binding energy was deconvoluted into four peaks centered at 162.7, 163.5, 164.0, and 164.5 eV, which were ascribed to S-Au, S-H, S-C, and S-S bonds, respectively. Using a similar approach, two peaks at higher binding energies of 167.1 and 168.18 eV corresponding to S (IV) and S (VI) could be deduced. The best fit of the data revealed that the relative proportion of S-Au bonds in the GSSG-AuNCs was 2.58%, higher than that in the GSH-AuNCs (0.697%, Figure S4). This result partially contributes to the stronger PL of the GSSG-AuNCs, because it is well known that aggregation of the Au-S motifs is one of the main sources of the Au nanocluster luminescence.
We also recorded Au 4f XPS spectra in GSSG-AuNCs and GSH-AuNCs for comparison ( Figure S1). The Au 4f XPS peaks in GSSG-AuNCs showed lower binding energies. For example, in the 4f 7/2 spectra range, the binding energies of Au(0) and Au(I) were 84.3 and 84.7 eV, respectively, in the GSH-AuNCs; they shifted to 83.6 and 84.5 eV, respectively, in the GSSG-AuNCs. It is well known that Au(0) can interact with electron-donating thiol and disulfide groups. Our experiments suggest that the Au(0)-disulfide interaction is stronger. Similarly, the low energy shift of Au(I)-binding energy in GSSG-AuNCs reveals that the Au(I)-GSSG interaction is stronger than the Au(I)-GSH interaction. The stronger Au(0)-disulfide and Au(I)-GSSG interactions favored the high aggregation degree of the nanoclusters in the aqueous phase that was observed in the DLS measurements.
The unique properties of the synthesized GSSG-AuNCs, -i.e., regularly packed nanocluster crystals that are more heavily aggregated in water, higher distribution of smaller nanoclusters, and more abundant Au(I)-GSSG motifs that cap the nanocluster cores,gave rise to the blue-shifted, intensity-enhanced fluorescent emission of GSSG-AuNCs.

Enrichment and Redispersion of GSSG-AuNCs
Organic solvents are often used to purify and concentrate water-soluble AuNCs through precipitation. We found that when acetonitrile was mixed with the aqueous GSSG-AuNC solution at certain volume ratios (the optimum was 7:1), a new liquid phase, instead of precipitates, was formed at the bottom of the centrifuge tube ( Figure 4). Approximately 2.5 µL of the new phase was obtained from 1 mL of the original aqueous GSSG-AuNC solution, offering an enrichment factor of 400-fold. The new phase had good fluidity. The composite viscosities of the bottom phase between 25 and 50 • C, measured by rheological performance tests, were about 20 times the viscosity of water ( Figure S5).
The fluorescence emission intensities of the AuNCs collected from the upper acetonitrilerich phase and the bottom phase differed dramatically ( Figure S6), with count readings of 2.40 × 10 3 and 2.22 × 10 5 , respectively, suggesting that the majority of the GSSG-AuNCs were extracted to the bottom phase. Moreover, experiments indicated that the GSSG-AuNCs recovered from different phases shared similar fluorescence spectra (Figure 4c), fragment patterns (Figure 4b), and morphologies ( Figure S7), implying that the properties of the GSSG-AuNCs did not change significantly after phase separation and the simultaneous enrichment process. Nanomaterials 2021, 11, x FOR PEER REVIEW 9 of 15  (Table S1 of Supporting Information), confirming the photoluminescence measurements.
The results suggest that our method can provide quick enrichment of nanomaterials with a high enrichment factor and recovery as compared with the reported methods (Table  1). Unlike enrichment approaches that are based on nanofluidic devices [14,46], chromatography [28], or high-speed centrifugation [29], our method needs no sophisticated instruments. Importantly, by simply mixing the aqueous AuNC solution and acetonitrile and allowing the mixture to stand for 2 h, phase separation can be accomplished. Therefore, the method can be facilely applied to large-scale enrichment of GSSG-AuNCs.  (Table S1 of Supporting Information), confirming the photoluminescence measurements.
The results suggest that our method can provide quick enrichment of nanomaterials with a high enrichment factor and recovery as compared with the reported methods (Table 1). Unlike enrichment approaches that are based on nanofluidic devices [14,46], chromatography [28], or high-speed centrifugation [29], our method needs no sophisticated instruments. Importantly, by simply mixing the aqueous AuNC solution and acetonitrile and allowing the mixture to stand for 2 h, phase separation can be accomplished. Therefore, the method can be facilely applied to large-scale enrichment of GSSG-AuNCs. Moreover, the bottom phase was readily redispersed in triple-distilled water. Interestingly, the redispersed GSSG-AuNCs can be re-enriched using acetonitrile following the same procedures. According to the ICP-AES measurements, the overall recovery of Au element in the 100 µL lower phase through three redispersion/re-enrichment cycles was 92.85%.

GSSG as the Mass-Separating Agent
To gain a deep understanding of the mechanism of phase separation, 20 mL of the aqueous GSSG-AuNC suspension was dialyzed (membrane dialyzer, 2 kDa cut off, Solarbio, Beijing, China) against 450 mL (150 mL × 3 times) of triple-distilled water. The nanoclusters trapped in the dialysis bag and the matrices (dialysate) outside were collected separately, evaporated to dryness to remove acetonitrile, and then dissolved in 20 mL of triple-distilled water, -denoted as dialyzed nanocluster solution and dialyzed matrix solution, respectively. At a volume ratio of 1:7, the addition of the dialyzed nanocluster solution to acetonitrile resulted in precipitates; conversely, a new phase was formed upon mixing the dialyzed matrix solution with acetonitrile. The experimental results implied that the residual GSSG in the original GSSG-AuNC solution may play a crucial role in phase separation.
To verify the hypothesis, 10 mL of 7.5 mM GSSG solution was mixed thoroughly with 70 mL of acetonitrile. A colorless new phase was found at the bottom of the bottle, suggesting the feasibility of GSSG as a mass-separating agent. It is widely accepted that hydrogen bonding plays a key role in aqueous two-phase separation. In the GSSG-AuNC solution, residual GSSG and its oxidization products formed hydrogen bond networks with water molecules. GSSG is sparingly soluble in acetonitrile. However, acetonitrile, which is fully miscible with water at room temperature, can form hydrogen bonds with the GSSG species in the aqueous GSSG-AuNC solution. In the presence of largely excessive acetonitrile, acetonitrile molecules might replace the water molecules of the GSSG-water networks in the aqueous nanocluster solution. As a result, some water molecules were "pushed" to the upper acetonitrile-rich phase.
The potential application of GSSG as a mass-separating agent was investigated by enriching the GSH-AuNCs. To this end, 0.2 mL of 30 mM GSSG was mixed with 0.8 mL of GSH-AuNCs and was then introduced to 7 mL acetonitrile and shaken thoroughly, and a yellow droplet of ca. 5 µL was found at the bottom of the tube after centrifugation. The ICP-AES test indicated that 98.32% of the Au species were distributed in the lower phase.

Mechanism Studies
Since GSH is also sparingly soluble in acetonitrile, and can form hydrogen bonds with acetonitrile in aqueous solution, experiments were carried out to test whether adding GSH-AuNC solution to acetonitrile could form a new liquid phase. It was observed that mixing GSH-AuNC solutions of different pH values (4.0-11.0) with acetonitrile at various volume ratios (1:15-1:1) resulted in precipitates. Similar results were obtained upon adding GSH solutions to acetonitrile.
GSSG differs from GSH chemically [48]. The more flexible molecular chain of GSSG affects the balance of the molecular conformational and intermolecular energy [49], which are key factors affecting the onset of nucleation and the subsequent precipitation. In addition, the pKa values of GSH are reported to be 2.12 (-CO 2 H), 3.59 (-CO 2 H), 8.75 (-SH), and 9.65 (-NH 3 + ) [50], considerably different from the corresponding pKa values of GSSG, which are reported to be 1.80, 2.39, 3.29, and 3.92 for the four carboxyl groups and 8.94 and 9.68 for the protonated amine groups [51]. The differences affect the protonation/deprotonation of the amine and carboxyl groups and, consequently, the hydrogenbonding capabilities, which greatly dictate whether the molecule can act as a massseparating agent [52]. For example, GSSG in aqueous solution can form intramolecular hydrogen bonds upon the assistance of two water molecules, while GSH cannot [37]. Further, the chemical differences also influence their interaction with acetonitrile. Our DFT simulations indicate that the -COOH, -NH 2 , and -CONH groups of both GSSG and GSH molecules can form hydrogen bonds with acetonitrile in aqueous solution ( Figures 5 and S11). However, the adsorption energy of acetonitrile to the carboxyl is the strongest, which are calculated to be −6.15 and −5.99 kcal/mol for GSSG and GSH, respectively ( Table 2), implying that the hydrogen bond between GSSG and acetonitrile is more stable. Our simulations and observations suggest that the stronger hydrogen bonding between GSSG and acetonitrile molecules stabilizes the bottom phase, preventing the formation of precipitates.

Conclusions
We synthesized new AuNCs using GSSG as both a reducing and a capping agent. Although the GSSG-AuNCs shared similar molecular compositions and morphologies with GSH-AuNCs, the former had a higher distribution of low-molecular-weight nanoclusters, possessed more surface Au-thiol ligands, were more regularly packed in the solid state, and were more heavily aggregated in aqueous solution. These characteristics rendered the GSSG-AuNCs significantly strongly photoluminescent with a QY considerably higher than that of GSH-AuNCs. Moreover, using GSSG as a mass-separating agent, the as-prepared GSSG-AuNCs in aqueous solution could be facilely enriched by as high as 400-fold, with recoveries higher than 99%, by simply mixing with acetonitrile at proper volume ratios. Furthermore, the enriched GSSG-AuNCs retained their original morphological and chemical properties and could be readily redispersed in water. The unique properties of the GSSG-AuNCs enable them to be potentially applied not only in fluorescent imaging but also in catalysis, drug delivery, and electronic and optical industries. Importantly, GSSG, a new mass-separating agent found by this work, may find its place not only in enrichment of nanoclusters but also in other disciplines such as separation and extraction.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/nano11092258/s1, Table S1. Au (µg) content in the original GSSG-AuNC solution, upper phase and bottom phase; Figure S1. XPS spectra of Au 4f of GSSG-AuNCs (a) and GSH-AuNCs (b); Figure S2. Fluorescence lifetime decay profiles of GSH-AuNCs (a) and GSSG-AuNCs (b); Figure S3. TEM image of the GSH-AuNCs; Figure S4. XPS spectra of S 2p of GSH-AuNCs; Figure S5. Comparison of the composite viscosity between water (filled blue square, refer to the left Y axis) and the bottom phase (filled red circle, refer to the right Y axis); Figure S6. Photoluminescent spectra of the upper acetonitrile-rich phase (black line) and the bottom GSSG-AuNC-enriched phase (red line); Figure S7. TEM images of the GSSG-AuNCs before (a) and after (b) enrichment; Figure S8. Representative gas chromatogram of the bottom phase; Figure S9. Capillary electrophoresis of the standard mixture of GSH (1) and GSSG (2)  Data Availability Statement: All data generated and analyzed during this study are included in this paper and the attached supporting information.