Experimental and Mechanistic Understanding of Aldehyde Hydrogenation Using Au25 Nanoclusters with Lewis Acids: Unique Sites for Catalytic Reactions

Article abstract of DOI:10.1021/jacs.5b07716

The catalytic activity of Au₂₅(SR)₁₈ nanoclusters (R = C₂H₄Ph) for the aldehyde hydrogenation reaction in the presence of a base, e.g., ammonia or pyridine, and transition-metal ions M^z+, such as Cu⁺, Cu²⁺, Ni²⁺ and Co²⁺, as a Lewis acid is studied. The addition of a Lewis acid is found to significantly promote the catalytic activity of Au₂₅(SR)₁₈/CeO₂ in the hydrogenation of benzaldehyde and a number of its derivatives. Matrix-assisted laser desorption ionization (MALDI) and electrospray ionization (ESI) mass spectrometry in conjunction with UV-vis spectroscopy confirm the generation of new species, Au_25-n(SR)_18-n (n = 1-4), in the presence of a Lewis acid. The pathways for the speciation of Au₂₄(SR)₁₇ from its parent Au₂₅(SR)₁₈ nanocluster as well as its structure are investigated via the density functional theory (DFT) method. The adsorption of M^z+ onto a thiolate ligand -SR- of Au₂₅(SR)₁₈, followed by a stepwise detachment of -SR- and a gold atom bonded to -SR- (thus an Au-SR unit) is found to be the most likely mechanism for the Au₂₄(SR)₁₇ generation. This in turn exposes the Au₁₃-core of Au₂₄(SR)₁₇ to reactants, providing an active site for the catalytic hydrogenation. DFT calculations indicate that M^z+ is also capable of adsorbing onto the Au₁₃-core surface, producing a possible active metal site of a different kind to catalyze the aldehyde hydrogenation reaction. This study suggests, for the first time, that species with an open metal site like adducts [nanoparticle-M]^(z-1)+ or fragments Au_25-n(SR)_18-n function as the catalysts rather than the intact Au₂₅(SR)₁₈.

Full text of DOI:10.1021/jacs.5b07716

Article
pubs.acs.org/JACS
Experimental and Mechanistic Understanding of Aldehyde
Hydrogenation Using Au25 Nanoclusters with Lewis Acids: Unique
Sites for Catalytic Reactions
Gao Li,†,§ Hadi Abroshan,‡,§ Yuxiang Chen,‡ Rongchao Jin,*,‡ and Hyung J. Kim*,‡,#
†State Key Laboratory of Catalysis & Gold Catalysis Research Centre, Dalian Institute of Chemical Physics, Chinese Academy of
Sciences, Dalian 116023, China
‡Department of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States
#School of Computational Sciences, Korea Institute for Advanced Studu, Seoul 130-722, Korea
S
* Supporting Information
ABSTRACT: The catalytic activity of Au25(SR)18 nanoclusters (R = C2H4Ph)
for the aldehyde hydrogenation reaction in the presence of a base, e.g.,
ammonia or pyridine, and transition-metal ions Mz+, such as Cu+, Cu2+, Ni2+
and Co2+, as a Lewis acid is studied. The addition of a Lewis acid is found to
significantly promote the catalytic activity of Au25(SR)18/CeO2 in the
hydrogenation of benzaldehyde and a number of its derivatives. Matrix-
assisted laser desorption ionization (MALDI) and electrospray ionization
(ESI) mass spectrometry in conjunction with UV−vis spectroscopy confirm
the generation of new species, Au25‑n(SR)18‑n (n = 1−4), in the presence of a
Lewis acid. The pathways for the speciation of Au24(SR)17 from its parent
Au25(SR)18 nanocluster as well as its structure are investigated via the density
functional theory (DFT) method. The adsorption of Mz+ onto a thiolate ligand
“SR” of Au25(SR)18, followed by a stepwise detachment of “SR” and
a gold atom bonded to “SR” (thus an “Au-SR” unit) is found to be the
most likely mechanism for the Au24(SR)17 generation. This in turn exposes the Au13-core of Au24(SR)17 to reactants, providing an
active site for the catalytic hydrogenation. DFT calculations indicate that Mz+ is also capable of adsorbing onto the Au13-core
surface, producing a possible active metal site of a different kind to catalyze the aldehyde hydrogenation reaction. This study
suggests, for the first time, that species with an open metal site like adducts [nanoparticle-M](z‑1)+ or fragments Au25‑n(SR)18‑n
function as the catalysts rather than the intact Au25(SR)18.
Au3 sites (Figure S1) or the staple motifs of the Au25(SR)18
nanoclusters are usually deemed as catalytically active sites in
the catalytic reactions.12,13 For instance, DFT predicts that both
phenylacetylene and iodobenzene molecules are adsorbed on a
Au3 site of the Au25(SR)18 nanoclusters in the Sonogashira
cross-coupling reaction.11 By contrast, the Au13 icosahedral core
would not be able to bind directly to the reactants in the
catalytic reactions despite its strong electronic influence on the
staple motifs and Au3 sites.10,14,15
Recently, nanogold catalysts have been widely applied in the
hydrogenation reactions.16−21 In their recent study of chemo-
selective hydrogenation of nitrobenzaldehyde to nitrobenzyl
alcohol, Jin and co-workers were able to achieve the catalytic
performance of a series of Aun(SR)m nanoclusters (where n and
m varied from 15 to 99 and from 14 to 42, respectively) with
∼100% selectivity using H2 gas as the hydrogen source and
pyridine as the base in both homogeneous and heterogeneous
catalytic systems.13,22 DFT calculations indicated that both the
INTRODUCTION
■
Over the past decade, atomically precise gold nanoclusters have
emerged as a novel and promising nanomaterial in nanoscience
and nanotechnology.1−3 Due to their unique catalytic proper-
ties for a wide range of different reactions, such as oxidation,
hydrogenation, and carbon−carbon coupling reactions, to name
a few,4−9 they have gained overwhelming research interests.
Compared to conventional gold nanocatalysts (for example,
naked nanoparticles) with unknown surface structure, these
gold nanoclusters are characterized by well-defined frame-
works,3 e.g., [Au25(SR)18]−[TOA]+ (SR = thiolate ligands,
TOA = tetraoctylammonium), so that they provide an excellent
arena to investigate how the atomic structure and their catalytic
properties are related, especially from the reaction mechanisms
point of view. Specifically, [Au25(SR)18]−[TOA]+ (or simply
Au25(SR)18) nanoclusters are composed of a 13-atom
icosahedral core (Au13) and six staple shell motifs Au2(SR)3
(“SRAuSRAuSR”, see Figure S1 in the
Supporting Information). These nanoparticles enable a detailed
analysis via quantum chemistry methods to understand the
catalytic reactions.10,11 On the basis of DFT calculations, the
Received: July 23, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
malononitrile (DCTB) was used as the matrix in MALDI analysis. The
nanoclusters and the matrix were dissolved in dichloromethane and
then spotted onto the steel plate and dried in vacuum prior to MALDI
analysis. 1H NMR was run on a Bruker 300 MHz spectrometer.
Electrospray ionization (ESI) mass spectra were obtained using a
Waters Q-TOF mass spectrometer equipped with a Z-spray source.
The sample was dissolved in toluene (1 mg/mL) and then mixed with
a dry methanol solution of CH3CO2Cs (50 mM) by a 1:1 vol ratio.
Thermogravimetric analysis (TGA) was conducted with a ∼1 mg
Au25(SR)18 sample under N2 atmosphere (flow rate 50 mL/min) on a
TG/DTA 6300 analyzer (Seiko Instruments, Inc.), and the temper-
ature was maintained at 150 °C for 60 min. Fourier-transform infrared
(FT−IR) measurements were recorded on a Brukers/Tensor 27
instrument (resolution, 1 cm−1; scans, 8; range, 1000−4000 cm−1).
Immobilization of the Au25(SR)18 Nanoclusters on CeO2
Supports. The CeO2 oxide is chosen as the support for Au25(SR)18
nanoclusters. According to a previous work by Li et al., CeO2 can lead
to a higher catalytic efficiency of the supported nanoclusters than other
oxide-supports (e.g., TiO2, SiO2) in the catalytic hydrogenation
reaction.22 For immobilization of nanoclusters Au25(SR)18, typically,
10 mg [Au25(SR)18]−[TOA]+ nanoclusters were dissolved in 30 mL
dichloromethane, and 1g CeO2 was added. After stirring for 6h at r.t.,
the Au25(SR)18/CeO2 catalysts were collected by centrifugation and
dried in vacuum. The catalysts were then annealed at 423 K for ∼1 h
in a vacuum oven. It has been proven previously that Au25(SR)18
nanoclusters loading on the oxides after the thermal treatment (423 K)
is intact.8,12
Typical Procedure for the Hydrogenation Reaction. In a
typical hydrogenation reaction, benzaldehyde (0.1 mmol), ammonium
hydroxide (5 μL), Au25(SR)18/CeO2 (100 mg, 1 wt % loading of
Au25(SR)18), Mz+ salt (promoter, 0.14 μmol), and 1 mL water were
added to a reactor (Parr instrument company, 22 mL capacity, series
4700) and gaseous H2 was introduced until 18 bar. The reaction
temperature was kept at 323 K for 10 h. After the catalytic reaction, the
catalyst was separated by centrifugation (∼1000 rpm, 1 min). Then
the supernatant was extracted by ethyl acetate, and the product was
obtained by the solvent removing. The conversion of aldehyde was
determined by 1H NMR analysis. In recycling reaction tests, the
catalyst was collected by centrifugation (4000 rpm) for 2 min after the
reaction. Next, the catalyst was washed with water and acetone to
remove the salts and other complexes, and dried in vacuum. The
catalyst thus prepared was then reused in a fresh reaction medium with
no Mz+ salt added.
CHO and NO2 groups of the reactant interact strongly
with the “SRAuSR” staples on the Aun(SR)m
nanoclusters; the adsorption energy of the nitrobenzaldehyde
−
on the Au25(SCH3)18 nanocluster is ∼ −1.03 eV (here and
hereafter, a negative value means a favorable interaction).13 It
should be noted that there has been a significant effort to
develop transition metal complexes for efficient CO bond
hydrogenation.23−26 Despite their high performance, however,
catalytic active sites were reported to be difficult to generate. In
view of this challenge as well as several disadvantages of
homogeneous catalysis, including the poor catalyst recovery
and the difficulty of removing catalysts from the products, it is
desirable to develop heterogeneous catalysis. The advantages
include excellent recyclability of catalysts, easy separation of
products, and better thermal stability.
Lewis acid, e.g., transition metal ions Mz+, is known to
promote gold nanoparticle-catalyzed reactions.27,28 For exam-
ple, the hydrogenation rate and allylic alcohol chemoselectivity
were found to improve in the presence of Co2+ for the
chemoselective hydrogenation of α,β-unsaturated carbonyl
compounds, catalyzed by PVP-protected Au0 nanoparticles
(PVP = polyvinylpyrrolidone).27 Despite its widespread use as
a promoter in the catalytic reactions, however, the mechanism
for the promotion via Lewis acid is still not known. Thus, it
would be extremely worthwhile to investigate the molecular
details of its role in the hydrogenation reactions.
In this article, we report on the Lewis acid-promoted
hydrogenation of aldehyde to alcohol catalyzed by Au25(SR)18/
CeO2 (hereafter, SR = SC2H4Ph) in the presence of a base
(pyridine or NH3·H2O) under mild conditions (18 bar H2 at
323 K). By monitoring the reaction media using UV−vis
spectroscopy in conjunction with MALDI and ESI mass
spectrometry, we have found that Au25(SR)18 nanoclusters
undergo detachment of SR ligands and Au atoms in the
presence of a Lewis acid, resulting in the formation of
Au25‑n(SR)18‑n (n = 1−4) species with catalytic active sites for
the aldehyde-to-alcohol hydrogenation. Possible configurations
of Au24(SR)17 are examined using the DFT and time-dependent
DFT (TD-DFT) methods. Our calculations indicate that the
catalytic active site located on the Au13-core of Au24(SCH3)17 is
capable of activating the covalent bond of H2 heterolytically,
leaving H− on the core for the ensuing hydrogenation. To our
knowledge, this is the first report with convincing experimental
and theoretical evidence that the Au13-core is responsible for
the catalytic reactions.
Computational Details. DFT optimization was performed using
the B3PW91 hybrid density functional. The 6-31G** basis set was
employed for H, C, N, O, and S,29−32 while the valence aug-cc-pVDZ
basis set together with the appropriate Stuttgart-Dresden effective core
potentials (ECPs) was used for transition metal ions, Cu+, Cu2+, Ni2+,
and Co2+.33−35 For gold atoms, the LANL2DZ basis set was used.36
The B3PW91 functional with the LANL2DZ basis set has been shown
to predict accurately the structure of Au nanoclusters.37−40 The
reaction energy is determined as the energy difference ΔE = ΣEpro
−
EXPERIMENTAL SECTION
ΣErea, where ΣEpro and ΣErea are total energies of the products and
reactants, respectively. The integral equation formalism polarizable
continuum model (IEFPCM) was applied to study the solvation effect
on reaction energetics.41 TD-DFT calculations of optical absorption
spectra were also performed and compared with experimental UV−vis
results. All calculations were carried out with the Gaussian09
package.32
■
Chemicals. Methylene chloride (HPLC grade, 99.9%, Sigma−
Aldrich), ammonium hydroxide (99%, Sigma−Aldrich), ceria (99%,
Sigma−Aldrich), pyridine (99%, Sigma−Aldrich), methanol (absolute,
200 proof, Pharmco), Co(OAc)2 (99.99%, Sigma−Aldrich), Cu-
(OAc)2 (99.99%, Sigma−Aldrich), Cu(OAc)1 (97%, Sigma−Aldrich),
Cu(NO3)2 (purum p.a., 98.0−103% (RT), Sigma−Aldrich), and NiCl2
(98%, Sigma−Aldrich), Cr(NO3)3 (99.0%, Sigma−Aldrich). All
chemicals were used as received. Nanopure water (resistivity 18.2
MΩ·cm) was purified with a Barnstead NANOpure Diwater system.
All glassware was thoroughly cleaned with aqua regia (HCl:HNO3 =
3:1 vol %), rinsed with copious Nanopure water, and then dried in an
oven prior to use.
Characterization. UV−vis spectra were acquired on a Hewlett-
Packard (HP) Agilent 8453 diode array spectrophotometer at room
temperature. MALDI-MS was performed with a PerSeptiveBiosystems
Voyager DE super-STR time-of-flight (TOF) mass spectrometer.
Trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenyldidene]-
RESULTS AND DISCUSSION
■
Characterization of the Au25(SR)18/CeO2 Catalyst.
Previous studies showed that the Au25(SR)18 nanoclusters are
stable under the conditions of 20 bar of hydrogen at 80 °C.22
Since the reaction condition employed in our current study (18
bar at 50 °C) is milder than the earlier cases, the clusters are
believed to be stable in our case as well. Isothermal stability
analysis of the unsupported Au25(SR)18 nanoclusters showed
no discernible loss of ligands during the isothermal process
B
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(Figure S2), suggesting that the thermal treatment (at 150 °C
in a N2 atmosphere) does not induce the ligand loss. Further,
the FT−IR analysis was performed for the Au25(SR)18/CeO2
catalyst before and after the 150 °C treatment in vacuum and in
an O2 atmosphere (1 bar). Prior to the FT−IR measurements,
the samples were washed after the thermal treatment using pure
ethanol three times and dried in vacuum at room temperature.
The FT−IR spectra of the sample before and after the
treatment in vacuum are superimposable (Figure S3), strongly
indicating that the Au25(SR)18 nanoclusters on the surface of
CeO2 indeed remain intact. These results are fully consistent
with previous work by Jin and co-workers.8,12 By contrast, the
sample treated in O2 shows ligand loss during the thermal
treatment (Figure S4), in good agreement with previous
study.42
Catalytic Performance of the Au25(SR)18 Nanocluster
with Lewis Acids in the Chemoselective Hydrogenation
of 4-Nitrobenzaldehyde. The results for the catalytic
efficiency of gold nanoclusters in the absence and presence of
different Lewis acids, viz., transition metal ions, Co2+, Ni2+, Cu+,
Cu2+, and Cr3+, under otherwise the same reaction conditions
are compared in Table 1. Several aspects are noteworthy. First,
time)]) reaches 70.8 h−1 when Co(OAc)2 is added (molar ratio
of Co2+: Au25(SR)18 = 2:1) (Table 1, entry 2). We note that
this level of promotion exceeds the 73.2% conversion obtained
with Au/PVP nanoparticles with the same Co2+ in earlier
studies.22,27 Other transition-metal ions result in the catalytic
efficiency enhancement of Au25(SR)18/CeO2 to a varying
degree. Specifically, moderate to significant conversions (50.0
to 78.2%) are achieved with Ni2+, Cr3+, Cu+, and Cu2+. The
degree of promotion of the catalytic activity of Au25(SR)18/
CeO2 by Mz+ is in the order Co2+ > Ni2+ ≈ Cr3+ > Cu+ > Cu2+.
Third, near 100% chemoselectivity for the alcohol product was
observed in all the catalytic reactions (Table 1). It should be
noted that no conversion was detected in the case of blank
experiments in which Au25(SR)18 nanoclusters are absent and
only Lewis acid and CeO2 oxide are present (Table 1, entry 9
and Table S1). This confirms that catalytic activity arises from
the gold nanoclusters.
To shed additional light on roles of Mz+, we studied the
catalytic activity of the unsupported Au25(SR)18 in the presence
and absence of Co(OAc)2. The reaction conditions were: 0.13
μmol unsupported nanoclusters, 0.26 μmol Co(OAc)2 (if used)
in 1 mL of toluene/ethanol (1:1, v/v), 0.1 mL of H2O, 0.1
mmol 4-nitrobenzaldehyde, 5 μL pyridine, and 18 bar H2 at 323
K for 10 h. The unsupported catalysts yielded, respectively,
17.9% and 1.3% conversion with 100% selectivity toward
alcohol product in the presence and absence of the Lewis acid.
These results strongly support that the Lewis acid is indeed a
promoter. Also notable is that the conversion rate of the
unsupported catalysts is considerably lower than the supported
ones. Two aspects are worthy of note here: First, metal oxides
are commonly used to disperse and stabilize nanoparticles and
to promote the reactions in heterogeneous catalysts.43,44 The
nature of the catalyst’s support can play a crucial role in the
catalytic activity of the nanoparticles. Despite extensive efforts,
however, the details of the interactions of the particles and the
supports still remain elusive due to the complexity of the
systems. Nonetheless, the high activity of gold nanoparticles on
CeO2 is often attributed to the excellent redox properties of
ceria which lead to easy formation and diffusion of oxygen
defects.45 Second, because of the low solubility of the catalyst in
polar solvents, we conducted the reactions with the
unsupported catalyst in a solvent mixture (1 mL of toluene/
ethanol (1:1, v/v) + 0.1 mL of H2O) that is considerably less
polar than water used in the supported case. According to the
hydrogenation mechanisms proposed below, solvation can play
an important role by facilitating (or even inducing) the
elongation of the HH bond of H2. Therefore, lowering the
solvent polarity would result in the deceleration of the
heterolytic bond cleavage of H2 and thus of the aldehyde-to-
alcohol conversion.
Table 1. Influence of Different Lewis Acids (Mz+Xz) on the
Chemoselective Hydrogenation of 4-Nitrobenzaldehyde (1)
to 4-Nitrobenzyl Alcohol (2) Catalyzed By Au25(SR)18/
a
CeO2.
b
selectivity [%]
b
entry
catalyst
Mz+
conv. [%]
2
3
4
1
2
Au25(SR)18
Au25(SR)18
Au25(SR)18
Au25(SR)18
Au25(SR)18
Au25(SR)18
Au25(SR)18
Au25(SR)18
13.4
90.1
75.8
78.2
61.1
50.0
50.8
90.8
n.r.
∼100
∼100
∼100
∼100
∼100
∼100
∼100
∼100
Co(OAc)2
Cr(NO3)3
NiCl2
3
4
5
Cu(OAc)1
Cu(OAc)2
Cu(NO3)2
CoCl2
6
7
8
9
Co(OAc)2
Co(OAc)2
c
10
11
12
Au25(SR)18
82.5
87.3
88.4
>99.5
∼100
∼100
d
[Au25Co]
e
[Au25Co]
a
Reaction conditions: 100 mg Au25(SR)18/CeO2 catalyst (∼1 wt %
Au25(SR)18 loading) in 1 mL water, Mz+/Au25(SR)18 = 2:1 (mol/mol),
0.1 mmol 4-nitrobenzaldehyde, 5 μL pyridine, 18 bar H2 at 323 K, 10
b
h. The conversion (conv.) of 4-nitrobenzaldehyde and selectivity
c
1
were determined by H NMR spectrum. n.r. = no reaction. Reaction
conditions: 4 g Au25(SR)18/CeO2 catalyst (∼40 mg Au25(SR)18
loading) in 30 mL water, 1.05 mmol Co(OAc)2, 5 mmol 4-
Another important aspect of our results in Table 1 is that
anions do not play any significant role in the hydrogenation
reaction. As presented in entries 6 and 7, copper salts with
different anions, Cu(NO3)2 and Cu(OAc)2, yield essentially the
same conversion rate (50.0 vs 50.8%). The cobalt salts,
Co(OAc)2 and CoCl2, show the same behavior (entries 2 and
8). These results strongly suggest that the anions act merely as
a spectator.
To gain insight into the performance of the catalyst in large
scale synthesis, we scaled up the catalytic system and examined
reaction kinetics under the conditions as indicated in Table 1.
The conversion rate and selectivity for the alcohol product for
this scaled-up system were found to be 82.5% and 99.5%,
d
nitrobenzaldehyde, 300 μL pyridine, 18 bar H2 at 323 K, 10 h. 2nd
e
d,e
and 3rd reuses of the catalyst recovered from entry 2. No Mz+ was
added to the reaction system.
in the absence of a Lewis acid, Au25(SR)18/CeO2 catalyst yields
a very low conversion at 323 K (13.4%, Table 1, entry 1). When
the temperature is raised to 353 K, however, the conversion can
reach ∼93% as reported previously.22 As mentioned above, in
this work, the temperature was kept at 323 K to investigate the
effects of the Lewis acids. Second, the conversion improves
markedly to 90.1% and the turnover frequency (TOF =
[reacted mol of aldehyde]/[(mol of Au clusters) × (reaction
C
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
respectively (Table 1, entry 10). These results are close to those
reported as entry 2 of Table 1 for the smaller scale catalysis
(90.1% and 100%, respectively). Therefore, Au25(SR)18/CeO2
in the presence of Co(OAc)2 appears to have good potential as
a catalyst for large scale synthesis.
group (Figure S5). In such a case, one may be able to improve
the reaction conversion by increasing the activity of the
reducing agent. Indeed the conversion of o-nitrobenzaldehyde
improves markedly to 97.1% when ammonia (NH3) is used as
the base instead of pyridine (Table S2).
We investigated hydrogenation of benzaldehyde and its
derivatives to study the influence of substitution. The results
obtained in the presence of Co(OAc)2 and pyridine are
compiled in Table 2. The conversion of unsubstituted
To assess the recyclability of the gold nanocluster catalysts,
we examined the catalytic activity of the reused [Au25(SR)18/
CeO2 + Co2+] catalyst (collected from entry 2 of Table 2) in
the hydrogenation reaction of 4-nitrobenzaldehyde. The
catalyst was recovered after the reaction by centrifugation
(4000 rpm), washed with water and acetone, and then dried in
vacuum. By using the recycled catalyst thus prepared, a fresh
reaction was run with reactants in a fresh solvent under
identical reaction conditions but in the absence of Lewis acids.
The recycled catalyst showed nearly the same activity and
selectivity as the fresh catalyst; it yielded 87.3% conversion for
the second cycle and 88.4% for the third cycle (Table 1, entries
9 and 10). Since no appreciable loss of catalytic activity and
selectivity was observed after 3 cycles (higher cycles were not
tested), we conclude that the [Au25(SR)18/CeO2 + Co2+]
catalyst is characterized by a good recyclability in hydro-
genation reaction.
Table 2. Hydrogenation of a Range of Substrates with Nitro
and Aldehyde Groups over Au25(SR)18/CeO2 in the
a
Presence of Co(OAc)2 Salt
We now proceed to examine catalytic mechanisms for
hydrogenation. The fact that the presence of Mz+ ions improves
catalytic properties of the nanoclusters indicates that Lewis
acids play an important role. We consider two different possible
mechanisms for this promotion: (1) parts of the nanocluster
surface, such as “SR” and “AuSR” units, are
removed so that gold atoms of the nanoclusters become
exposed/reactive to the reactants; (2) the Lewis acid is
adsorbed onto the Au25(SR)18 nanoclusters, thereby providing a
new active site on the nanoclusters for the reaction.
Characterization of the Au25‑n(SR)18‑n (n = 1−4)
species. To gain insight into how Lewis acids promote the
hydrogenation reaction, we studied unsupported Au25(SR)18
nanoclusters in the presence and absence of a Lewis acid in
solution. The unsupported nanoclusters can be easily
characterized by UV−vis, MALDI−MS, and ESI−MS. In
Figure 1A, UV−vis spectra of two different solutions are
displayed: (i) a solution (dichloromethane/methanol 1:1, v/v)
of Au25(SR)18 nanoclusters in the presence of only ammonia at
room temperature, and (ii) the same solution but in the
presence of both ammonia and Co(OAc)2 string for 10 h under
the same conditions. We note that the acid and Au25(SR)18
concentrations, Co2+:Au25(SR)18 = 2:1 (mol/mol), employed
here for UV−vis and MS analysis of unsupported nanoclusters
are the same as those used in the hydrogenation reaction study
with supported nanoclusters above (Table 1). Comparison of
the optical spectra of the two solutions shows that the
nanoclusters undergo structural changes when Co(OAc)2 is
added. For further insight, we performed MADLI−MS analysis.
The results are shown in Figure 1B. Solution i (black line) is
characterized by two peaks, which are assigned to Au25(SR)18
(m/z: 7394.3, theoretical m/z: 7394.2) and its fragment
Au21(SR)14 (m/z: 6058.1, theoretical m/z: 6057.4) caused by
MADLI-MS. In the case of solution ii (red line), the main peak
centered at m/z 7060.3 is assigned to Au24(SR)17 (theoretical
m/z: 7060.0). Since Au24(SR)17 is not formed during the
MADLI−MS analysis as evidenced by the result for solution i, it
is the presence of Lewis acid that leads to the generation of
Au24(SR)17 from its parent Au25(SR)18 nanoclusters. Never-
theless, some Au25(SR)18 are still present as indicated by the
much less intense peak at m/z = 7394.3.
a
Reaction conditions: the same as noted in Table 1.
benzaldehyde was found to be 51.4% (Table 2, entry 1). The
substitution with a methyl or a methylthio group exerts a
considerable influence on the hydrogenation reaction as the
conversion of the aldehydes with these substituents decreases
to 37.7 and 26.8%, respectively (Table 2, entries 3 and 4). An
iodo group has a similar but lesser effect (Table 2, entry 2).
Such a decrease in the conversion efficiency is expected because
these substituents (methyl, methylthio, and iodo groups) at the
para-position are electron donating groups and therefore tend
to lower the electron affinity of the carbonyl group of aldehydes
toward reduction to alcohol. We also examined the effect of the
position of substitution, viz., a nitro group at the ortho-, meta-,
and para-positions (Table 2, entries 5−7). The substitution at
the meta- or para-position increased the reaction conversion to
∼90%. Since the nitro group is an electron withdrawing group,
it increases the electron affinity and thus reactivity of the
carbonyl group toward reduction to alcohol. However, the
substitution at the ortho-position (o-nitrobenzaldehyde) led to
the lowest conversion among the three nitrobenzaldehyde
reactants (Table 2, entries 5−7). Possible reasons include the
intramolecular interactions between two closely located nitro
and carbonyl groups and steric hindrance arising from the nitro
D
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
indicates that ions I−VI and VII-IX carry 1+ and 2+ charges,
respectively. The peak assignments are as follows: peak I at m/z
= 7997.87 is assigned to [Au25(SC2H4Ph)18Co4Py3Cs]+
(theoretical m/z: 7999.70), peak II at 7663.87 to
[Au25(SC2H4Ph)18CoPyCs]+ (theoretical m/z: 7664.66), peak
III at 7328.91 to [Au24(SC2H4Ph)17CoPyCs]+ (theoretical m/z:
7329.80), peak IV at 6994.94 to [Au23(SC2H4Ph)16CoPyCs]+
(theoretical m/z: 6995.79), peak V at 6660.93 to
[Au22(SC2H4Ph)15CoPyCs]+ (theoretical m/z: 6660.78), peak
VI at 6326.86 to [Au21(SC2H4Ph)14CoPyCs]+ (theoretical m/z:
6 3 2 7 . 7 9 ) , p e a k V I I a t 4 0 6 5 . 4 0 t o
[Au25(SC2H4Ph)18Co4Py3Cs2]2+ (theoretical m/z: 4066.30),
peak VIII at 3898.40 to [Au25(SC2H4Ph)18CoPyCs2]2+
(theoretical m/z: 3898.36), and peak IX at 3731.42 to
[Au24(SC2H4Ph)17CoPyCs2]2+ (theoretical m/z: 3731.35).
Their experimental isotopic patterns match very well with the
simulated ones (Figure S6). Of note, the CoPy adduct
appearing in the ESI spectrum was not detected in the
MALDI mass spectra as the latter is a “hard” ionization method.
The MALDI and ESI results suggest that both possibilities
mentioned abovei.e., removal of “AuSR” units of the
staple motif on Au25(SR)18 nanoclusters and generation of the
adduct from Au25−n(SR)18−n in the presence of Lewis acid (e.g.,
Co2+ ion) and basecan provide mechanistic explanations for
catalytic promotion.
We monitored the removal of a “AuSR” unit of
Au25(SR)18 nanoclusters in the presence of a Lewis acid using
UV−vis and MALDI−MS (Figure 2). We first studied the
dichloromethane/methanol mixed solution of Au25(SR)18
nanoclusters and the Co(OAc)2 salt (Co2+:Au25(SR)18 = 2:1,
mol/mol) at room temperature (Figure 2A). We observed that
the distinct peak at 670 nm gradually decreased, indicating that
the Au25(SR)18 nanoclusters are slowly converted to other
nanoclusters (Au25−n(SR)18−n). To understand the influence of
the base, we repeated the same experiment but with ammonia
added to the system (all other experimental conditions
remained the same). Similar to the case of no ammonia
(Figure 2A), the UV−vis signature of Au25(SR)18 decays but
Figure 1. (A) UV−vis and (B) MALDI mass spectra of Au25(SR)18
nanoclusters at room temperature stirring for 10 h in a mixed
dichloromethane/methanol solution: (i) in the presence of only
ammonia (black line), and (ii) in the presence of Co(OAc)2
(Co2+:Au25(SR)18 = 2:1, mol/mol) and ammonia (red line). The
peak marked with an asterisk (*) is the fragment Au21(SR)14 caused by
MALDI, rather than an impurity. (C) Positive mode ESI mass
spectrum of Au25(SR)18 nanoclusters in the presence of Co(OAc)2
(Co2+:Au25(SR)18 = 2:1, mol/mol) and pyridine.
For additional information, a sample of Au25(SR)18 nano-
clusters with Co(OAc)2 (Co2+:Au25(SR)18 = 2:1, mol/mol) and
pyridine was characterized by the ESI−MS method. ESI−MS is
much softer than the MALDI analysis and typically does not
cause fragmentation of nanoclusters. Several peaks were
observed in the ESI mass spectrum of the sample (Figures
1C and S6). The spacing of the isotope patterns of the peaks I
to VI and VII to IX is 1 and 0.5 (Figure S7), respectively. This
Figure 2. UV−vis spectra tracing of a methanol solution of Au25(SR)18 nanoclusters in the presence of (A) only Co(OAc)2, (B) Co(OAc)2 and
ammonia, (C) NiCl2 and ammonia, and (D) Cu(OAc)2 and ammonia. (E) UV−vis spectra of Au25(SR)18 with Mz+Xz at 10 min (and at 5 min in the
case of Co(OAc)2). (F) MALDI mass spectra of the Au25(SR)18 nanoclusters in the presence of Cu(OAc)2 and ammonia. In all cases,
Mz+:Au25(SR)18 = 2:1 (mol/mol). In (F), the peak marked with an asterisk (*) is a fragment caused by MALDI, rather than an impurity.
E
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. (A) Framework of Au25(SR)18 nanoclusters (Au25 NC) based on the results of ref 10 and three different “AuSR” fragments
(denoted as AuS: α,β,γ) for possible removal. Detachment of fragments α and β results in Au24(SR)17 with configurations (B) Au24 NC-1 and (C)
−
−
Au24 NC-2, respectively. Theoretical absorption spectra of Au25(SCH3)18 and Au24(SCH3)17 in configurations Au24 NC-1 and Au24 NC-2 are
presented in (D), (E), and (F), respectively. Color code: green, Au on the staple shell; orange, Au atoms in the Au13 core; yellow, sulfur atoms.
Other atoms are not shown for clarity.
considerably faster with ammonia present in the system (Figure
2B). It is worth mentioning that the UV−vis spectrum in the
presence of Co(OAc)2 and ammonia in Figure 2B almost
becomes a decaying curve after ∼60 min (the 400 to 1100 nm
range). Other Lewis acids, NiCl2, Cu(OAc)2, Cu(OAc)1, and
Cr(NO3)3 (Mz+:Au25(SR)18 = 2:1, mol/mol) with ammonia
yield essentially the same, albeit slower, temporal evolution of
the UV−vis spectrum (Figure 2, parts C and D, and Figure S8).
The spectra at 10 min (and at 5 min for Co2+) in Figure 2E
show an excellent correlation with the trend of the reaction rate
Structure of Au24(SR)17 Species. There are three
nonequivalent Au−SR pairs on the Au25(SR)18 nanoclus-
ters10,11 (denoted as Au25 NC in Figure 3) that can be
removed upon exposure to Lewis acids: two different pairings
of Au of a “SRAuSR” staple motif with SR (i.e., α
and β fragments), and one for Au of the Au13-core (γ pairing).
Since the removal of a γ unit from the Au13-core is likely to
induce strong disruptions of Au25 NC’s structure and
characteristics (e.g., cluster rearrangement and major electronic
property changes), we consider here only the removal of
fragments α and β that results in the formation of new species
Au24 NC-1 (Figure 3B) and Au24 NC-2 (Figure 3C),
in Table 1, i.e., conversion rate decreases in the order Co2+
>
Ni2+ ≈ Cr3+ > Cu+ > Cu2+. By contrast, Au25(SR)18
nanoclusters remain intact when only ammonia is present at
50 °C (Figure S9). These results indicate that ammonia acts as
a promoter for the Lewis acid-induced removal of a “Au
SR” unit from the parent Au25(SR)18 nanoclusters. Addi-
tionally, we studied the removal process by MALDI−MS at
different reaction times. As shown in Figure 2F, the intensity of
the mass peak at m/z 7060 corresponding to the Au24(SR)17
species gradually increased, indicating the Au25(SR)18 nano-
clusters are converted to Au24(SR)17 in the presence of Lewis
acid. This provides further evidence that the Au24(SR)17 species
is newly formed, rather than fragments of Au25(SR)18 like the
Au21(SR)14 species.
−
respectively. DFT predicts that the optimized Au24(SCH3)17
structure of Au24 NC-1 is more stable than that of Au24 NC-2
by 21.2 kcal/mol. This is in good agreement with previous
studies10,11,46 and also with our AuS bond length results
(Table S3).
The calculated optical absorption spectra of Au25 NC, Au24
NC-1, and Au24 NC-2 are shown in Figure 3. We considered
only the two lowest absorption bands. The maximum intensity
lines of these absorption bands of Au25 NC are located at 1.72
eV (720 nm) and 2.74 eV (452 nm), consonant with
measurements (1.80 and 2.75 eV).10 Perhaps the most salient
aspect of our results is that the lowest absorption band nearly
disappears for Au24 NC-1, while it remains as a shoulder in the
case of Au24 NC-2 (Figure 3). The calculated spectrum of Au24
NC-1 agrees well with the featureless experimental UV−vis
results in Figure 2A−C. It is also possible that the absence of
specific band structures in the UV−vis spectra is due to the
presence of different complexes. However, in view of TD-DFT
as well as the DFT results for the relative stability above, we
believe that the structure of the Au24SR17 species generated in
the solution of Lewis acids is mainly characterized by Au24 NC-
1 in Figure 3B.
The mass spectrometry results as well as the data in Table 1
suggest that the catalytic activity is associated with the newly
generated Au25−n(SR)18−n species such as Au24(SR)17. This
means that Lewis acids may be not only directly involved in the
hydrogenation reaction, but also responsible for the generation
of Au24(SR)17 species with active sites for catalysis. As a
consequence, a better adsorption of Lewis acids onto the
nanoclusters and/or a faster generation of Au24(SR)17 would
lead to a higher promotion of the hydrogenation reaction (e.g.,
Co2+ > Cu2+). In what follows, we consider in detail the
structure of Au24(SR)17, its production pathways from the
parent Au25(SR)18 nanoclusters in the presence of a Lewis acid
and a base, and potential mechanisms for ensuing hydro-
genation reactions.
Mechanism of Au24(SR)17 Generation. To rationalize the
removal mechanism of a “AuSR” pair unit from the
−
Au25(SCH3)18 nanoclusters, we investigated possible adsorp-
tion sites of a Lewis acid (Mz+ ion) on Au25(SCH3)18− using the
F
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Proposed mechanisms for speciation of Au24(SR)17 (Mechanism 1) and formation of adduct [nanoparticle-M](z‑1)+ (Mechanism 2) in the
presence of Mz+ salt and ammonia. A gold atom (Au1) from the Au13-core becomes exposed to the solvent in Mechanism 1 (upper pathway), while
the transition metal ion at M1 becomes an active site in Mechanism 2 (lower pathway). The calculated energies for different steps are given in Table
3. Color code: Au, green; S, yellow. The orange section represents the Au13 core and Im2* denotes [M(NH3)3SCH3](z‑1)+
.
a
Table 3. DFT Results for Energies (kcal/mol) for Elementary Steps of Mechanisms 1 and 2 in Figure 4
Cu+
Cu2+
Ni2+
Co2+
−183.6
E1 = ΔE(Re → Im1)
−66.9
−197.6
−188.0
ΔE(Im1 → Im2)
+60.2
+69.9
+55.2
+42.5
ΔE(Im2 → Im3)
ΔE(Im3 → Pr)
−20.8 (∼−14)
+64.1 (∼0)
−6.7 (∼+25)
−27.5 (∼+11)
+36.6 (∼+11)
−59.3
−20.8 (∼−14)
+64.1 (∼0)
−127.7 (∼+15)
−148.5 (∼+1)
−84.4 (∼+1)
−20.8 (∼−14)
+64.1 (∼0)
−132.8 (∼+14)
−153.6 (∼0)
−89.5 (∼0)
−20.8 (∼−14)
+64.1 (∼0)
−141.1 (∼+18)
−161.9 (∼ + 4)
−97.8 (∼+4)
E2
E3
E4
#
E2
#
E3
−65.5
a
Solution-phase energetics estimated using the IEFPCM and Born solvation free energies are given in parentheses. All energy values are relative to
the reactant state Re.
DFT method. We considered Cu+, Cu2+, Ni2+, and Co2+ for
Mz+ in the calculations. In Mechanism 1 (upper pathway in
Figure 4), the sulfur atom of unit α (Figure 3) forms a bond
with Mz+ (Re → Im1). This adsorption process is energetically
exothermic in the gas phase by 66.9, 197.6, 188.0, and 183.6
kcal/mol with M+S bond length 2.29, 2.30, 2.36, and 2.45 Å
for Cu+, Cu2+, Ni2+, and Co2+, respectively (Table 3). In
aqueous solution, the exothermicity is expected to decrease due
to the solvation stabilization of the charged reactants. (For
insight into solvation effects, the IEFPCM results for solution-
phase energetics are given in Table 3 and in Table S4). For
clarity, we mention that the coordination number (CN) was
assumed to be 4 for the isolated metal complex, i.e.,
M(NH3)4z+, and one of the four NH3 moieties was assumed
to be displaced from the coordination shell as the complex
attaches itself to the cluster. As a result, the Lewis acid (Mz+
ion) adsorbed on the cluster surface is coordinated to three
NH3 molecules. Although we do not exclude possibilities of CN
higher or lower than 4 for the isolated metal complex, our
results for CN = 4 clearly show that a considerable amount of
energy is released with the formation of a Mz+S bond. This
excess energy can facilitate the removal of the entire Mz+S
moiety (Im2*), i.e., thiolate group plus metal complex, from
the gold nanocluster and thus the formation of Au25(SCH3)17
(Im1 → Im2). The removal of Mz+S is an endothermic
process with ΔE = +60.2, + 69.9, + 55.2, and +42.5 kcal/mol
for Cu+, Cu2+, Ni2+, and Co2+, respectively. We parenthetically
note that ΔE for Im1 → Im2 in the gas phase decreases in the
order Cu2+ > Cu+ > Ni2+ > Co2+, in concert with the
experimental trend of the conversion rate of hydrogenation
(Table 1). One potential explanation is that the thiolate group
removal is a key step of the hydrogenation reaction.
The resulting nanoclusters Au25(SCH3)17 have a gold atom
that is bonded to only one thiolate group. Two ammonia
(NH3) molecules can coordinate to this gold atom exothermi-
cally with the AuN distance of 2.39 Å and energy release of
20.8 kcal/mol (Im2 → Im3). The coordinating NH3 molecules
eventually detach the gold atom from the nanocluster, resulting
+
−
in the formation of Au(NH3)2 and Au24(SCH3)17 (Im3 →
Pr, Figure 4). While the final step is endothermic by +64.1
kcal/mol in the gas phase, endothermicity becomes negligible
in aqueous solutions because of significant solvation stabiliza-
G
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
tion of the charged products. The overall reaction of the
inorganic complex with the gold nanoclusters based on
Mechanism 1 can be written as follows:
nanoclusters influences their catalytic activity in hydrogenation
reactions.
Hydrogenation Mechanism. Since essentially all surface
−
gold atoms of Au25(SC2H4Ph)18 are protected by thiolate
[Au25(SCH3)18]− + M(NH3)4 + NH3 → [Au24(SCH3)17]−
z+
groups, the Au25 NC nanoclusters are inert toward adsorption
of nonradical molecules on their surface.48,49 Our calculations
indeed show that there are no strong interactions of
(z−1)+
+
+ [M(NH3)3(SCH3)]
+ Au(NH3)2
−
Au25(SCH3)18 with possible adsorbates (e.g., PhCHO, NH3,
with ΔE= +36.6, −84.4, −89.5, and −97.8 kcal/mol (gas-
phase) and ∼11, ∼1, ∼0, and ∼4 kcal/mol (solution-phase) for
Cu+, Cu2+, Ni2+, and Co2+, respectively.
and H2), in accordance with recent theoretical studies.48,49
However, the removal of an Au-SR fragment or the formation
of an adduct exposes open metal atoms (a gold atom of the
core Au13 or on staple motifs, or the Lewis acids) to the solvent
(Figures 4 and S10). DFT predicts that these sites, i.e., Au1,
Au2, and M1 in Figures 4 and S10, are active toward adsorption
of reactants. In view of these results, we propose hydrogenation
mechanisms (Figure 5), in which bond activation of H2 on
these metal sites is the first step.
Mechanism 1 proposed here explains our experimental
finding that the presence of base NH3·H2O accelerates the
disappearance of the absorption band at ∼670 nm (Figure 2,
parts A vs B). NH3 is usually a better lone-pair donor than
CH3OH to coordinate and form a complex with Au. Similarly,
it can also explain another experimental finding above, i.e., NH3
leads to a higher conversion than pyridine (see Hydrogenation
Mechanism below), in that NH3 is a better lone-pair donor
than pyridine to coordinate to gold.
Upon removal of the capping NH3 from the metal site (Au1,
Au2, or M1), the HH bond of a H2 molecule becomes
activated with the aid of a base (Figure 5). Specifically, the base
We briefly pause here for perspective. While our DFT (and
UV−vis) results indicate that Au24 NC-1 (Figure 3B) is the
dominant configuration of Au24SR17 species generated in the
solution of Lewis acids, the effect of the CeO2 support was not
accounted for in the calculations. CeO2 can play a crucial role
to, for example, rearrange the structure of the nanoparticles and
change the active sites. Therefore, even if the direct generation
of Au24 NC-2 from its parent Au25 NC via removal of a β unit
may be suppressed (see Figure 3 above), it can be produced
from Au24 NC-1 via, e.g., ligand exchange47 on the support. The
formation of Au24 NC-2 yields an active gold site (Au2) on a
staple motif rather than on the core (see Figure S10).
(NH3) abstracts a proton from H2 to form NH4 with H− on
+
the active metal site (Re*Au1 → Im1*Au1, Re*M → Im1*M,
Re*Au2 → Im1*Au2). The removal of capping NH3 and
adsorption of a H2---NH3 unit are endothermic and
exothermic, respectively. Overall, this two-step process at the
Au1 and M1 active sites is endothermic by +4.9 and +19.7 kcal/
mol, respectively. For Au2, this step is found to be exothermic
by 3.5 kcal/mol. In a polar solvent like water, however, the
product state would become considerably stabilized compared
to the reactant because of the highly charge-separated nature of
the NH4 ---H− moiety. DFT calculations indicate that the
+
presence of a base, such as NH3, is crucial for the activation of
H2. Without a base, the interaction of H2 with M1 and Au2 is
mainly physical and that with Au1 is rather weak, so that H2
tends to stay away from Au1. Thus, a base is needed for
chemical adsorption of H2 and ensuing HH bond activation
on Au1 and M1. This participation of a base in the H2
activation as well as in the removal of a gold atom discussed
above (Mechanism 1) provides a theoretical explanation for our
experimental findings that no hydrogenation reaction occurred
in the absence of a base and that a stronger base (NH3 vs.
pyridine) yields a higher conversion efficiency (Table S2). Of
note, the lone-pair on nitrogen is in sp3 and sp2 orbitals in the
cases of NH3 and pyridine, respectively. In the case of the Au1
active site, PhCHO becomes physically adsorbed on the surface
of the nanoclusters via interactions of its O and C (CHO
Mechanism of the Adduct [nanoparticle-M](z−1)+
Formation. In Mechanism 2, i.e., the lower pathway in Figure
4, a considerable energy release via Mz+S bond formation
(Re → Im1) can result in the removal of coordinating
ammonia molecules (Im1 → Im2# → Im3#) rather than the
interacting thiolate group. ΔE for Im1 → Im2# and Im2#→
Im3# steps are +7.6 and −6.2 kcal/mol, respectively, for Mz+
=
Cu+. As illustrated in Figure 4, Mz+ becomes adsorbed on the
Au13-core surface by coordinating to three core Au atoms
(Im3#). This in turn can induce a considerable elongation (0.05
Å) of nearby AuS bonds, exposing their thiolate groups even
more to the Lewis acids in the solvent. Therefore, it is also
possible that Im3# follows the pathway of Mechanism 1
discussed above, i.e., Im3# → Im2 → Im3 → Pr, by losing one
#
+
group) with NH4 and H−, respectively, in an exothermic step
of the “AuSR” pair units (ΔE = EPr − EIm3 = +102.1
with ΔE = −5.3 kcal/mol (Im1*Au1 → Im2*Au1, Figure 5A).
kcal/mol).
Finally, H+ (of NH4 ) and H− will transfer to PhCHO and
+
The net charge of Au25 NC in the present study was −1.
Since the charge state of nanoclusters can modulate their
solvation energy and spin multiplicity, it could exert a
considerable influence on reaction kinetics and thermody-
namics of the Au-SR removal (Mechanism 1) and/or adduct
formation (Mechanism 2). For example, we expect that the
solvation effect on overall reaction thermodynamics of the
adduct formation would vary significantly with the charge state
because the reactant and product nanocluster states, i.e., Au25
NC and Im3# of Mechanism 2, are characterized by different
electric charges. By contrast, in the case of Mechanism 1, the
corresponding change of the solvation effect with the charge
state would not be as significant because Au25 NC and Au24
NC-1 have the same charge. It would thus be both interesting
and worthwhile to study in the future how the charge state of
produce PhCH2OH exothermically with energy release of 15.4
kcal/mol (Im2*Au1 → Re*Au1, Figure 5A).
In the case of hydrogenation catalyzed by the M1 and Au2
sites (Figure 5, parts B and C), the ammonium ion formed
(Im1*M and Im1*Au2) may become solvated in the reaction
media and act as a donor for protonation of aldehyde (Im1*M
→ Im2*M and Im1*Au2 → Im2*Au2). For M1 and Au2, this
step is found to be endothermic by +26.0 and +46.8 kcal/mol
in water, respectively, while it is +111.6 and +158.4 kcal/mol in
the gas phase. Since the products, NH4 , Au25SR18CuH−, and
+
Au24SR17H2− are charged, they are strongly stabilized in water,
compared to the gas phase (see Table S4). This means that
solvation can facilitate (or even induce) the elongation of the
HH bond (g in Figure 5B) to produce charged species
H
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the process were characterized by MALDI/ESI mass
spectrometry in conjunction with UV−vis absorption spectros-
copy. DFT predicts that the optimized structure of
−
Au24(SCH3)17 in the Au24 NC-1 configuration is 21.2 kcal/
mol more stable than that in the Au24 NC-2 configuration. The
optical absorption spectrum of the former calculated via TD-
DFT shows a good agreement with the experimental results.
On the basis of these observations, the mechanism of the
removal of a “AuSR” fragment to produce Au24(SR)17
from the Au25(SR)18 parent nanoclusters in the presence of
Lewis acids (e.g., Cu+, Cu2+, Ni2+, and Co2+) and ammonia is
proposed. DFT results further indicate that the surface gold
atoms on the Au13-core or staple shell, or the Lewis acids bound
on the Au13-core can provide catalytic active sites for
hydrogenation. On the basis of experimental and DFT results,
we elucidated the hydrogenation mechanism of reducing
aldehyde to alcohol. The catalytic hydrogenation is also applied
to a range of benzaldehyde derivatives, and the gold
nanoclusters catalyst with Lewis acid shows good catalytic
activity. While our study reported here contributes novel
insights on aldehyde hydrogenation catalyzed by Au nano-
clusters, some aspects were not properly accounted for in our
computational analysis. These include the molecular aspect of
the solvent environment and the role of the CeO2 support. The
latter can modulate the electronic and structural and thus
catalytic properties of the gold nanoclusters significantly.
Therefore, it would be worthwhile in the future to extend the
current study to investigate details of molecular interactions
between the nanoparticles and supports and their influence on
catalytic efficiency and pathways for hydrogenation and other
reactions catalyzed by gold nanoparticles.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.5b07716.
■
S
Thermal stability test of the unsupported Au25(SR)18
nanoclusters and the Au25(SR)18/CeO2 catalyst, blank
reaction results, MADLI mass spectrum of the
Au25(SC2H4Ph)18 nanoclusters, experimental and simu-
lated isotopic distribution patterns of Au25−n(SR)18−n (n:
1−4) nanoclusters, NMR Data, and stabilization energy
of different species in solution with respect to the gas
phase, etc. (PDF)
Figure 5. Proposed mechanism for hydrogenation of benzaldehyde in
the presence of H2 and NH3 on (A) site Au1, (B) site M1, and (C)
site Au2. The values of interatomic distances are a* = 2.26, a = 2.56, b
= 1.82, c = 1.05, d = 1.64, c* = 1.03, d* = 1.63, e = 2.85, f = 1.87, a** =
2.03, g = 0.80, h = 2.04, i = 1.74, i* = 1.53, j = 1.73, k = 1.04, l = 1.61,
and l* = 1.59 Å. (For comparison, the equilibrium bond length of an
isolated H2 molecule is 0.74 Å.) The same color code is used as in
Figure 4.
AUTHOR INFORMATION
■
+
Corresponding Authors
*rongchao@andrew.cmu.edu
*hjkim@cmu.edu
(NH4 and H−). This is followed by the reduction of
protonated benzaldehyde to benzyl alcohol (Im2*M → Re*M
and Im2*Au2 → Re*Au2, Figure 5, parts B and C). This final
reduction step is exothermic with ΔE = −152.1 and −175.6
kcal/mol, respectively, for the M1 and Au2 sites in the gas
phase, while −89.3 and −86.8 kcal/mol in the solution phases.
Author Contributions
§G.L. and H.A. contributed equally to this work.
Notes
The authors declare no competing financial interest.
CONCLUSIONS
■
ACKNOWLEDGMENTS
This work is supported in part by DOE BES Grant DE-FG02-
12ER16354 (RJ) and by NSF Grant No. CHE-1223988 (HK).
The Au25(SR)18/CeO2 catalyst in the presence of a Lewis acid
and a base (ammonia or pyridine) exhibits good catalytic
performance, including catalytic activity and selectivity, in the
aldehyde hydrogenation reactions. The presence of Lewis acids
in solution leads to a significant enhancement in the catalytic
activity of Au25(SR)18/CeO2 nanoclusters. Among all Lewis
acids we studied, Co2+ yields the best catalytic conversion. New
nanocluster species Au25‑n(SR)18‑n (n = 1−4) generated during
■
REFERENCES
■
(1) Yamazoe, S.; Koyasu, K.; Tsukuda, T. Acc. Chem. Res. 2014, 47,
816−824.
(2) Taketoshi, A.; Haruta, M. Chem. Lett. 2014, 43, 380−387.
I
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(3) Qian, H.; Zhu, M.; Wu, Z.; Jin, R. Acc. Chem. Res. 2012, 45,
1470−1479.
(33) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007−1023.
(34) Fuentealba, P.; Preuss, H.; Stoll, H.; Szentpaly, L. V. Chem. Phys.
́
Lett. 1982, 89, 418−422.
(4) Li, G.; Jin, R. Acc. Chem. Res. 2013, 46, 1749−1758.
(5) Boronat, M.; Leyva-per
834−844.
́
ez, A.; Corma, A. Acc. Chem. Res. 2014, 47,
(35) Igel-Mann, G.; Stoll, H.; Preuss, H. Mol. Phys. 1988, 65, 1321−
1328.
(36) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299−310.
(37) Assadollahzadeha, B.; Schwerdtfegera, P. J. Chem. Phys. 2009,
131, 064306.
(6) Liu, J.; Krishna, K. S.; Losovyj, Y. B.; Chattopadhyay, S.; Lozova,
N.; Miller, J. T.; Spivey, J. J.; Kumar, C. S. S. R. Chem. - Eur. J. 2013,
19, 10201−10208.
(38) Taketsugu, T.; Lyalin, A. J. Phys. Chem. Lett. 2010, 1, 1752−
(7) Shivhare, A.; Ambrose, S. J.; Zhang, H.; Purves, R. W.; Scott, R.
W. J. Chem. Commun. 2013, 49, 276−278.
1757.
́
́
(39) Faza, O. N.; Rodríguez, R. A.; Lopez, C. S. Theor. Chem. Acc.
(8) Li, G.; Jin, R. J. Am. Chem. Soc. 2014, 136, 11347−11354.
(9) Li, G.; Liu, C.; Jin, R. Chem. Commun. 2012, 48, 12005−12007.
(10) Zhu, M.; Aikens, C. M.; Hollander, F. J.; Schatz, G. C.; Jin, R. J.
Am. Chem. Soc. 2008, 130, 5883−5885.
2011, 128, 647−661.
(40) Zubarev, D. Y.; Boldyrev, A. I. J. Phys. Chem. A 2009, 113, 866−
868.
(41) Scalmani, G.; Frisch, M. J. J. Chem. Phys. 2010, 132, 114110−
114115.
(42) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet,
M. J.; Delmon, B. J. Catal. 1993, 144, 175−192.
(43) Chng, L. L.; Erathodiyil, N.; Ying, J. Y. Acc. Chem. Res. 2013, 46,
1825−1837.
(44) Wu, Z.; Jiang, D.; Mann, A. K.; Mullins, D. R.; Qiao, Z. A.;
Allard, L. F.; Zeng, C.; Jin, R.; Overbury, S. H. J. Am. Chem. Soc. 2014,
136, 6111−6122.
(45) Li, Y.; Shen, W. Chem. Soc. Rev. 2014, 43, 1543−1574.
(46) Lopez-Acevedo, O.; Hakkinen, H. Eur. Phys. J. D 2011, 63, 311−
314.
(11) Heaven, M. W.; Dass, A.; White, P. S.; Holt, K. M.; Murray, R.
W. J. Am. Chem. Soc. 2008, 130, 3754−3755.
(12) Li, G.; Jiang, D.; Liu, C.; Yu, C.; Jin, R. J. Catal. 2013, 306, 177−
183.
(13) Li, G.; Jiang, D.; Kumar, S.; Chen, Y.; Jin, R. ACS Catal. 2014, 4,
2463−2469.
(14) Tlahuice-Flores, A.; Whetten, R. L.; Jose-Yacaman, M. J. Phys.
Chem. C 2013, 117, 20867−20875.
(15) Aikens, C. M. J. Phys. Chem. A 2009, 113, 10811−10817.
(16) Pan, M.; Brush, A. J.; Pozun, Z. D.; Ham, H. C.; Yu, W.;
Henkelman, G.; Hwang, G. S.; Mullins, C. B. Chem. Soc. Rev. 2013, 42,
5002−5013.
(47) Fernando, A.; Aikens, C. M. J. Phys. Chem. C 2015, 119, 20179−
20187.
(17) Yamane, Y.; Liu, X.; Hamasaki, A.; Ishida, T.; Haruta, M.;
Yokoyama, T.; Tokunaga, M. Org. Lett. 2009, 11, 5162−5165.
(18) Corma, A.; Serna, P. Science 2006, 313, 332−334.
(19) Cano, I.; Chapman, A. M.; Urakawa, A.; van Leeuwen, M. J. Am.
Chem. Soc. 2014, 136, 2520−2528.
(48) Lopez-Acevedo, O.; Kacprzak, K. A.; Akola, J.; Hakkinen, H.
̈
Nat. Chem. 2010, 2, 329−334.
(49) Liu, C.; Lin, S.; Pei, Y.; Zeng, X. C. J. Am. Chem. Soc. 2013, 135,
18067−18079.
(20) Ren, D.; He, L.; Yu, L.; Ding, R.; Liu, Y.; Cao, Y.; He, H.; Fan,
K. J. Am. Chem. Soc. 2012, 134, 17592−17598.
(21) Zhu, Y.; Qian, H.; Drake, B. A.; Jin, R. Angew. Chem., Int. Ed.
2010, 49, 1295−1298.
(22) Li, G.; Zeng, C.; Jin, R. J. Am. Chem. Soc. 2014, 136, 3673−3679.
(23) Zhang, G.; Vasudevan, K. V.; Scott, B. L.; Hanson, S. K. J. Am.
Chem. Soc. 2013, 135, 8668−8681.
(24) Rosler, S.; Obenauf, J.; Kempe, R. J. Am. Chem. Soc. 2015, 137,
̈
7998−8001.
(25) Gartner, D.; Welther, A.; Rad, B. R.; Wolf, R.; von Wangelin, A.
̈
J. Angew. Chem., Int. Ed. 2014, 53, 3722−3726.
(26) Zhang, G.; Scott, B. L.; Hanson, S. K. Angew. Chem., Int. Ed.
2012, 51, 12102−12106.
(27) Mertens, P. G. N.; Wahlen, J.; Ye, X.; Poelman, H.; De Vos, D.
E. Catal. Lett. 2007, 118, 15−21.
(28) Mertens, P. G. N.; Vandezande, P.; Ye, X.; Poelman, H.;
Vankelecom, I. F. J.; De Vos, D. E. Appl. Catal., A 2009, 355, 176−183.
(29) Becke, A. D. Phys. Rev. A: At., Mol., Opt. Phys. 1988, 38, 3098−
100.
(30) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(31) Burant, J. C.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1996,
105, 8969−8972.
(32) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.
P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.;
Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin,
K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.;
Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega,
N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.;
Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.;
Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.;
̈
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.;
Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian, Inc.,
Wallingford CT, 2009.
J
DOI: 10.1021/jacs.5b07716
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX