Unraveling structures of protection ligands on gold nanoparticle Au68(SH)32.

New low-energy atomic structures of the thiolate-protected gold nanoparticle Au68(SH)32 are uncovered, where the atomic positions of the Au atoms are taken from the recent single-particle transmission electron microscopy measurement by Kornberg and co-workers, whereas the pattern of thiolate ligands on the gold core is attained on the basis of the generic formulation (or rule) of the "divide and protect" concept. Four distinct low-energy isomers, Iso1 to Iso4, whose structures all satisfy the generic formulation, are predicted. Density-functional theory optimization indicates that the four isomers are all lower in energy by 3 to 4 eV than the state-of-the-art low-energy isomer reported. Further analysis of the optimized structures of Au68(SH)32 shows that the structure of gold core in Iso1 to Iso4 is consistent with the experiment, whereas the positions of a few Au atoms at the surface of gold core are different. The computed optical absorption spectra of the four isomers are consistent with the measured spectrum. Computation of catalytic properties of Au68(SH)32 toward CO oxidation suggests that the magic number cluster can be a stand-alone nanoscale catalyst for future catalytic applications.

INTRODUCTION

Highly stable <span class="Chemical">thiolatepan>-protected gold nanoparticles (RS-AuNPs) can be viewed as “super atoms” for making cluster-assembled materials whose novel properties may be exploited for applications in catalysis, nanotechnology, or chemical biology. Determination of the structures of RS-AuNPs in the range of 1 to 2 nm has attracted considerable research interests over the past two decades (–). In the laboratory, the most common approach to determine the structures of the RS-AuNPs is x-ray crystallography. However, a critical prerequisite for using the x-ray crystallography technology is to achieve a sizable single crystal of the RS-AuNPs, which generally is a very challenging task. To date, only the structures of the following RS-AuNPs have been fully resolved by x-ray crystallography: Au102(p-MBA)44 (p-MBA: p-mercaptobenzoic acid, SC7<span class="Chemical">O2H5) (), <span class="Chemical">Au25(SCH2CH2Ph)18− (–), Au38(SCH2CH2Ph)24 (), Au36(SPh-tBu)24 (), Au28(SPh-tBu)20 (), Au20(SPh-tBu)16 (), Au24±1(SAdm)16 (), and Au18(SC6H11)14 (). The lack of atomic structures of many RS-AuNPs has largely hindered a comprehensive understanding of the structure-property relationship of RS-AuNPs. Hence, a new experimental technique that can detect the atomic structures of AuNPs without the need for crystallization of RS-AuNPs is greatly desired.

Recently, Azubel et al. () have reported the atomic structure of an RS-pan class="Chemical">AuNP containing 68 Au atoms determined by the powerful single-particle transmission electron microscopy (SP-TEM) combined with density-functional theory (DFT) computation and absorption spectroscopy. This SP-TEM–centered approach is transformative in the structure determination of AuNPs because it no longer requires producing a single crystal of RS-AuNPs. Nevertheless, thus far, SP-TEM can only yield the positions of heavy (that is, gold) atoms. Thus, structure determination of the protection ligands requires theoretical input from DFT computation and absorption spectroscopy measurement. According to the atomic positions of Au determined from SP-TEM, the structure of Au68(SH)32 has been suggested () where some gold-thiolate motifs, such as bridging thiolates and ring-like structures, are predicted. In this communication, we present a series of new low-energy isomers in which the pattern of thiolate ligands on the gold core is determined according to the generic formulation (or rule) of the “divide and protect” (D&P) concept (, ). DFT optimization indicates that all the new isomers are uniformly lower in energy by 3 to 4 eV than the state-of-the-a.rt low-energy isomer reported in ().

We note that the D&P approach combined with the DFT computation has been proven as a viable theoretical approach to predict low-energy or even global minimum structures of RS-AuNPs (–). The D&P approach elucidates that highly stable AuNPs tend to have a symmetric gold core covered by various levels of interfacial -RS-Au-RS- staple motifs (). The lengths (or levels) of staples can vary, but overall they must obey two stoichiometry constraints for a given number of Au atoms in the inner core and in the ligand-core interface (, ). The validation of the D&P concept has been confirmed by the conceptual breakthrough (for example, the staple motifs for the ligand structures) being made in the total structure determination of the Au102(p-MBA)44 by Kornberg and co-workers via x-ray crystallography (), as well as the total structure determination of relatively small-sized clusters, such as <span class="Chemical">Au25(SCH2CH2Ph)18− (–).

RESULTS AND DISCUSSION

First, on the basis of the precise positions of all Au atoms in the inner gold core and in the interface region of the Au68NP determined from the SP-TEM experiment (), we construct a number of different ligand structures with various levels of S-Au-S staple motifs, following the generic rule of the D&P approach (). The generic rule states that an RS-AuNP can be divided into several groups as illustrated by [Au][Au(SR)2] [Au2(SR)3], where a, a′, b, and c are integers. Here, [Au] represents the gold core, which satisfies the constraint condition that the number of “surface” Au atoms (a′) in the gold core equals the sum of end points of the exterior motifs (2b + 2c); that is, each surface Au atom of the gold core is protected by one end point of the staple motif. Hence, the parameters a, a′, b, and c for Au68(SH)32 must satisfy a + a′ + b + 2c = 68, 2b + 3c = 32, and a′ = 2b + 2c. Four initial core structures [Au](a + a′ = 48 to 51) are examined, and each was covered by a certain number of -HS-Au-HS- and -HS- Au- HS-Au-HS- staple motifs: (i) [Au]24+24[Au(SR)2]4[Au2(SR)3]8, (ii) [Au]23+26[Au(SR)2]7[Au2(SR)3]6, (iii) [Au]22+28[Au(SR)2]10[Au2(SR)3]4, (iv) [Au]21+30[Au(SR)2]13[Au2(SR)3]2. The four newly constructed isomer structures together with the previously reported isomer based on the SP-TEM experiment () were optimized using the DFT method implemented in Materials Studio Dmol3 7.0 (, ). The generalized gradient approximation with the Perdew-Burke-Ernzerhof (PBE) () functional and the double numeric polarized (DNP) basis set were adopted. Lastly, the linear and quadratic synchronous transit (LST/QST) method (, ) was used to locate the transition state of CO oxidation on Au68(SH)32. The theoretical powder x-ray diffraction (XRD) curve is calculated using the Debye formula:where s is the diffraction vector length and θ is the scattering angle, satisfying s = 2sinθ/λ. λ and α are determined by the experimental setup and are set to be 0.1051967 nm and 1.01, respectively. B is the damping factor, which reflects thermal vibrations, and is set to be 0.03 nm2. The corresponding atomic numbers are used for the scattering factor f. d is the distance between atoms i and j. The atomic distance d used in the calculation is taken from the optimized structure of clusters.

The optimized structures of the four isomers, <pan class="Chemical">span class="Gene">Iso1 to Iso4, classified by group divisions (i) to (iv), are shown in Fig. 1. The computed total energies of <sppan>an class="Gene">Iso1 to Iso4 are 3.95, 3.28, 3.31, and 3.06 eV, respectively, lower than the state-of-the-art low-energy isomer reported (). The isomer Iso1 appears to be the most stable isomer. The large energy difference between Iso1 and the previously reported isomer indicates that the ligand patterns that can meet the generic formulation of the D&P approach are energetically much more favorable. In Iso1 to Iso4, the 68 gold atoms can be grouped into an Au15 core with an Au atom in the center. Figure 2 displays two orthogonal views of the Au15 core in Iso1 to Iso4 and the Au15 core taken from the previously reported experimental isomer. In addition, the root mean square deviation (RMSD) values as a measure of deviation of the Au15 core in Iso1 to Iso4 from the experimental structure are presented in Table 1. The small RMSD values (about 0.3 Å) indicate that the overall structures of Au15 inner cores in Iso1 to Iso4 are consistent with the experimental one. Moreover, the structural distortion of the Au15 core in all five isomers can be recognized in Fig. 2. The slight distortion is probably induced by the different interfacial -RS-Au-RS- staple structures. As such, the core structure of Au68NP is slightly distorted rather than being highly symmetric.

Fig. 1

Optimized structures of the four isomers Iso1 to Iso4 of Au68(SH)32.

The Au and S atoms are in gold and red, respectively. The H atoms are not shown.

Fig. 2

Two orthogonal views of the 15 Au atoms in the core of the reported isomer from the SP-TEM experiment of Au68 and in the cores of Iso1 to Iso4.

The Au atom in the center is in red.

Table 1

RMSD values as a measure of the deviation of the Au15 core and Au68 in Iso1 to Iso4 from the experimental structure.

The unit is in angstrom.

	Iso1	Iso2	Iso3	Iso4
Au15	0.28	0.28	0.33	0.32
Au68	0.70	0.71	0.69	0.69

The Au anpan class="Chemical">d S atoms are in gold and red, respectively. The H atoms are not shown.

The Au atom in the center is in repan class="Chemical">d.

The fcc-like gold frameworks in <pan class="Chemical">span class="Gene">Iso1 to Iso4 were singled out in Fig. 3. Here, to illustrate the fcc-like structure of gold frameworks more clearly, a common neighbor analysis (CNA) () was undertaken. Detailed information of CNA is given in table S1. The distinct CNA signatures demonstrate that the inner gold core structures of <sppan>an class="Gene">Iso1 to Iso4 exhibit the fcc (100) and (111) surfaces, consistent with the SP-TEM experiment. The analysis of the Au15 core and the fcc-like framework suggests that the interior structures of Iso1 to Iso4 are consistent with the SP-TEM experiment. The RMSD values (about 0.7 Å) for the Au68 structures of Iso1 to Iso4, however, become larger after including the surface Au atoms, as shown in Table 1, suggesting that the positions of the surface Au atoms that bind with the end point of the exterior motifs differ from those in the reported isomer. We note that in the SP-TEM experiment, a minimal electron dose was used (). Could the TEM electron beam alter the positions of several surface Au atoms during the measurement? This question can be addressed through further experimental investigation using known RS-AuNP structures resolved from x-ray crystallography as a benchmark.

Fig. 3

The fcc-like frameworks of Au68 in Iso1 to Iso4.

The 15 Au atoms in the core are in gold, and the other Au atoms are in red.

The 15 Au atoms in the pan class="Chemical">core are in gold, and the other Au atoms are in red.

Computed optical absorption spectra of Iso1 to Iso4, using the time-dependent DFT (TD-DFT) method, are shown in Fig. 4. As a comparison, both experimental and theoretical spectra of the previously reported isomer () are also included in Fig. 4. The locations of four prominent absorption peaks (a, b, c, and d) show that the computed optical absorption spectra of <span class="Gene">Iso1 to Iso4 are largely consistent with the experimental spectrum. Additionally, the simulated XRD curves of Iso1 to Iso4 and the experimental cluster () are shown in Fig. 5. Overall, Iso1 results in the best agreement with the experimental curve. For Iso2 to Iso4, their first, second, and third peaks are slightly shifted compared to the corresponding experimental peaks. Further analysis of the molecular orbital (MO) levels and the corresponding atomic orbital components in each MO of Iso1 to Iso4 are presented in figs. S1 to S4. We find that the strong absorption peaks b to d are mainly contributed by the core [Au] (a + a′ = 48 to 51). Overall, the positions of the major absorption peaks of these five isomers are more or less similar to one another because they have similar core structures. However, as shown in Fig. 4, the shape of the computed absorption curves of Iso1 to Iso4 are different, suggesting that the overall shape of the optical absorption spectrum of Au68(SH)32 is sensitive to the isomer structure.

Fig. 4

Optical absorption spectra of Au68(SH)32.

Top row: The experimental (wine and olive) curves and the theoretical (black) curve plotted by taking the data from Ref. 20. Rows 2 to 5: Computed optical absorption spectra of Iso1 to Iso4. The red curve denotes the vibrational frequency analysis based on the individual vibrational intensities (red vertical lines). The blue curve denotes spectra from TD-DFT computation of the individual optical transitions. In all five rows, the locations of four prominent absorption peaks (a, b, c, and d), from either experiment or theory, are displayed.

Fig. 5

Simulated XRD curves of Iso1 to Iso4 and the experimental Au68(SH)32 cluster.

Top row: The experimental (wine and <pan class="Chemical">span class="Species">olive) curves and the theoretical (black) curve plotted by taking the data from Ref. 20. Rows 2 to 5: Computed optical absorption sppan>ectra of <span class="Gene">Iso1 to Iso4. The red curve denotes the vibrational frequency analysis based on the individual vibrational intensities (red vertical lines). The blue curve denotes spectra from TD-DFT computation of the individual optical transitions. In all five rows, the locations of four prominent absorption peaks (a, b, c, and d), from either experiment or theory, are displayed.

Finally, we examine catalytic properties of <span class="Chemical">pan class="Chemical">Au68(SH)32 by using the CO oxidation as a probe. The computed catalytic reaction pathway for the CO oxidation on the <sppan>an class="Gene">Iso1 cluster is shown in Fig. 6. Because all gold atoms in the clusters are protected by the thiolate groups, several surface staple motifs of Iso1 are removed to make the catalytic reaction, as indicated in previous studies (, ). As shown in Fig. 6, the CO and O2 molecules can be favorably coadsorbed on two neighboring low-coordinated Au atoms, with the coadsorption energy of CO2 and O2 being about −1.24 eV. Upon the coadsorption of CO and O2, the two molecules can move closer while the O-O bond length is elongated. In the first step, a relatively low-energy barrier of 0.18 eV (TS1 in Fig. 6) can be overcome while the two molecular species arrive at a bridge-like metastable intermediate state characterized by the O-C-O-O species, with the O–O bond length being 1.43 Å. In the second step, the O–O bond length is further elongated to 1.85 Å while CO fully grasps an O atom of O2 to form a CO2 molecule. Lastly, CO2 desorbs with cleaving the O–O bond, leaving the other O atom adsorbed on the gold cluster. The second step (TS2 in Fig. 6) is the rate-determining step that requires overcoming a reaction barrier of 0.68 eV. These reaction barriers are comparable to those of typical nanogold catalysts (–), indicating that the Au68 cluster can be a stand-alone nanoscale catalyst for future applications.

Fig. 6

Computed reaction pathway for the CO oxidation on Iso1 cluster.

Here, the symbol * denotes adsorption of the corresponding molecules on Au atoms. Au, S, C, and O atoms are in gold, red, gray, and blue, respectively. H atoms are not shown.

Here, the symbol * denotes apan class="Chemical">dsorption of the corresponding molecules on Au atoms. Au, S, C, and O atoms are in gold, red, gray, and blue, respectively. H atoms are not shown.

CONCLUSION

We have presented a series of new low-energy isomer structures of <pan class="Chemical">span class="Chemical">Au68(SH)32 determined from the generic formulation of the D&P approach. The consistency in the interior structure of Au68 and the peak locations of the optical absorption spectra between the new isomers and the isomer reported from the SP-TEM experiment suggests that these isomers, Iso1 in particular, are promising candidates for the most stable atomic structure of Au68(SH)32. Further computation of catalytic properties of Au68(SH)32 toward CO oxidation suggests that this magic number cluster can be a stand-alone nanoscale catalyst for future applications. Confirmation of the predicted staple motif–based atomic structure of Au68(SH)32 must await future benchmark experiments, for example, using known staple motif–based RS-AuNPs, such as Au102(p-MBA)44, as a benchmark in future SP-TEM experiment.