Structural Relaxation Enabled by Internal Vacancy Available in a 24-Atom Gold Cluster Reinforces Catalytic Reactivity

Article abstract of DOI:10.1021/jacs.9b07761

Unveiling the mystery of the contribution of nonsurface or noninterface sites in a catalyst to its catalytic performance remains a great challenge because of the difficulty in capturing precisely structural information (surface plus inner) encoded in the catalyst. This work attempts to elucidate the critical role of the internal vacancy in an atomically precise 24-atom gold cluster in regulating the catalytic performance on the hydrogenation reaction of CO2. The experiment results show that the Au24 cluster with internal vacancy can mitigate sintering and exhibit high catalytic activity under relatively harsh reaction conditions, in contrast to the structurally similar Au25 cluster without internal vacancy. Our computational study suggests that the internal vacancy in Au24 provides the cluster with much more structural flexibility, which may be crucial to resisting the aggregation of the cluster and further postponing the deactivation. The hydrogenation and coupling stages of the reaction intermediates are proposed to explain the potential reaction pathway of CO2 with H2 on the Au24 catalyst with internal vacancy.

Full text of DOI:10.1021/jacs.9b07761

Subscriber access provided by University of Texas Libraries
Article
Structural Relaxation Enabled by Internal Vacancy Available
in a 24-atom Gold Cluster Reinforces Catalytic Reactivity
Xiao Cai, Weigang Hu, Shun Xu, Dan Yang, Mingyang
Chen, Miao Shu, Rui Si, Weiping Ding, and Yan Zhu
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.9b07761 • Publication Date (Web): 12 Feb 2020
Downloaded from pubs.acs.org on February 14, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Structural Relaxation Enabled by Internal Vacancy
Available in a 24-atom Gold Cluster Reinforces Catalytic
Reactivity
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Xiao Cai,† Weigang Hu,† Shun Xu,† Dan Yang,† Mingyang Chen,*,‡,∥ Miao Shu,§ Rui
Si,*,§ Weiping Ding,† and Yan Zhu*,†
†School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210093, China
‡Center for Green Innovation, School of Materials Science and Engineering, University of Science
and Technology Beijing, Beijing 100083, China
§Shanghai Synchrotron Radiation Facility, Zhangjiang Laboratory, Shanghai 201204, China
∥Beijing Computational Science Research Center, Beijing 100193, China
ABSTRACT
Unveiling the mystery of the contribution of non-surface or non-interface sites in a
catalyst to its catalytic performance remains a great challenge because of the difficulty
in capturing precisely structural information (surface plus inner) encoded in the
catalyst. This work attempts to elucidate the critical role of the internal vacancy in an
atomically precise 24-atom gold cluster in regulating the catalytic performance on the
hydrogenation reaction of CO2. The experiment results show that the Au24 cluster
with internal vacancy can mitigate sintering and exhibit high catalytic activity under
relatively harsh reaction conditions, in contrast to the structurally similar Au25 cluster
without internal vacancy. Our computational study suggests that the internal vacancy
in Au24 provides the cluster with much more structural flexibility, which may be
crucial to resisting the aggregation of the cluster and further postponing the
deactivation. The hydrogenation and coupling stages of the reaction intermediates are
proposed to explain the potential reaction pathway of CO2 with H2 on the Au24
catalyst with internal vacancy.
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 40
1
2
3
4
5
6
1. INTRODUCTION
7
8
9
The investigations into catalysis of the surface and interface sites of catalysts are
always in the research spotlight, considering that the chemical reactions usually occur
on the surfaces or interfaces of the catalysts and even a tiny alteration in the
surface/interface sites of catalysts can significantly change the catalytic properties.1-10
An interesting question is whether the inner sites of the heterogeneous catalysts can
also have a substantial influence on the catalytic performances. Can one directly
observe such influence in a catalytic reaction? If so, how can the inner and surface
sites cooperatively regulate the catalytic process? In recognizing the difficulty in
correlating the conventional characterization information with inner sites of a
heterogeneous catalyst, a new generation of well-defined catalysts with precise
structures is highly desirable. By solving total structures (both the surface and the core
ones) of such catalysts, one may precisely correlate the catalytic properties with the
exact structure of the catalyst and learn what on earth controls the activity, selectivity,
and stability of the catalyst.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The emergence of atomically precise metal clusters recently has brought precise
synthesis of well-defined catalysts and successful determination of their crystal
structures into bloom.11-18 Based upon the precise structures of these metal clusters,
very fundamental understanding of metal catalysis can be attained at a deep level.19-22
Among the studies on the catalytic properties of metal clusters, site-specific catalysis
of the clusters has constituted one of the most exciting fields, e.g. it has been found
2
ACS Paragon Plus Environment

Page 3 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
that the central atom of the cluster catalysts can apparently impact the catalytic
activity.23-27 These studies show, in the subnanometer size regime, that the influences
of the internal atoms of the cluster catalysts on overall performances should never be
ignored. We question naturally whether the internal vacancies as one type of inner
sites can possibly contribute to the catalytic performances.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To address this fundamental issue, herein we utilize the atomically precise
24-atom gold cluster with internal vacancy and the 25-atom gold cluster without
internal vacancy to investigate how the internal vacancy surrounded by the surface
affects the catalytic reactivity of CO2 hydrogenation, considering that transformation
of CO2 with renewable hydrogenation to yield chemicals and fuels is of significant
importance for mitigating global warming and energy supply crises.28-35 The only
structural difference of [Au24(PPh3)10(SC2H4Ph)5Cl2]+ (hereafter Au24) and
[Au25(PPh3)10(SC2H4Ph)5Cl2]2+ (hereafter Au25) is that Au24 lacks a core gold atom.36
Au24 can be viewed as two vertex-missing icosahedral Au12 units joined by sharing
the vacancy, fused together by five thiolate linkages (Figure 1a), whereas Au25 can be
viewed as two icosahedral Au13 units joined by sharing one vertex gold atom (Figure
1b). The spectroscopic fingerprints of the two clusters allow for an easy identification
of the reversible interconversion of the two clusters via one-core-atom removal and
addition (Figure S1). Our studies suggest that the internal vacancy in the Au24 cluster
provides the system with more structural flexibility, and the resultant
vacancy-protected structural relaxation possibly plays a pivotal role in inhibiting
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 40
1
2
3
4
5
6
7
aggregation of the active cluster into inactive nanocrystal as found for the Au25 cluster
without the internal vacancy.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. X-ray crystal structures of (a) [Au24(PPh3)10(SC2H4Ph)5Cl2]+ and (b)
[Au25(PPh3)10(SC2H4Ph)5Cl2]2+ clusters. Color codes: orange = Au, pink = P, yellow = S, green =
Cl. H atoms are omitted for clarity.
2. RESULTS AND DISCUSSION
As shown in Figure 2, the Au24 and Au25 clusters supported on SiO2 gave rise to
similar selectivity toward dimethyl ether (DME) as the major product. Notably, SiO2
showed no activity in the hydrogenation of CO2 under identical conditions (Figure S2).
In addition, neither charge transfer nor surface ligand exchange occurred between
gold clusters and SiO2 (more details shown in Figure S3 and S4). Either CO2 or H2
was singly added to the reaction system and no product was detected (Figure S2),
which ruled out the participation of the protective ligands of the clusters as carbon
sources in forming the organic products. Importantly, a significant difference of the
two catalysts was reflected in the catalytic activity. The Au24 catalyst can efficiently
convert CO2 to the organic products (Figure 2a), while the Au25 catalyst exhibited
quite a low conversion of CO2 under identical reaction conditions (Figure 2b). The
4
ACS Paragon Plus Environment

Page 5 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
additional experiments at different temperatures and under various reaction gas
pressures further showed that the Au24 catalyst with internal vacancy exhibited
consistently higher catalytic activity than the Au25 catalyst without internal vacancy
for the hydrogenation reaction of CO2 (Figure 2c and 2d). The distinct activities of the
Au24 and Au25 catalysts in the hydrogenation of CO2 seemed to stem from whether the
vacancy is present or absent at the internal sites of the two catalysts.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conv. on Au25
DME
others
Conv. on Au24
DME
others
a
b
100
75
50
25
0
100
75
50
25
0
1200
800
400
0
1200
800
400
0
2
4
6
8
10
2
4
6
8
10
Reaction time (h)
Reaction time (h)
1600
1200
800
400
0
d 1600
c
Au24
Au25
Au24
Au25
1200
800
400
0
120
130
140
1
2
3
Temperature (oC)
Pressure (MPa)
Conv.
DME
MF
EtOH
MeOH
others
800
e
1600
1200
800
400
0
100
80
60
40
20
f
Au24
Au25
600
400
200
0
1
2
3
4
5
6
7
8
Au
*
Au38
Au37
25
Number of cycles
Figure 2. Catalytic performances of CO2 hydrogenation over (a) Au24/SiO2 and (b) Au25/SiO2
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 40
1
2
3
4
catalysts with reaction time. Reaction conditions: 100 mg Au24/25/SiO2 (0.5% Au), 2 MPa reaction
5
6
o
gas (CO2:H2 = 1:3) with 5% N2 as internal standard, 15 mL H2O, 130 C. Note that beside DME
7
8
9
other products formed in small amounts were not further studied. (c) Effect of reaction
temperature on CO2 hydrogenation over Au24/SiO2 and Au25/SiO2 catalysts with 2 MPa reaction
gas for 10 hr. (d) Reaction gas pressure-dependent catalytic performances in CO2 hydrogenation
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
o
over Au24/SiO2 and Au25/SiO2 catalysts at 130 C for 10 hr. (e) Catalytic properties of the
supported Au25*, Au37 and Au38 catalysts without internal vacancies for CO2 hydrogenation with 2
MPa reaction gas at 130 oC for 10 hr. Au25*: Au25(SC2H4Ph)18; Au37:
[Au37(PPh3)10(SC2H4Ph)10X2]+; Au38: Au38(SC2H4Ph)24. (f) Recyclability of the Au24/SiO2 and
Au25/SiO2 catalysts with each 4-hour reaction under 2 MPa reaction gas (CO2:H2 = 1:3) at 130 oC.
To further reveal the contribution of the internal vacancy in Au24 to the catalytic
performance for CO2 hydrogenation, the three reference clusters with icosahedral
Au13 kernels, [Au37(PPh3)10(SC2H4Ph)10X2]+, Au38(SC2H4Ph)24, and Au25(SC2H4Ph)18,
were performed (Figure 2e). It is worth noting that the three clusters have no internal
vacancies: [Au37(PPh3)10(SC2H4Ph)10X2]+ resembles a rod-like structure assembled
from three icosahedral Au13 aligned along the rod axis (Figure S5a and S5b);37
Au38(SC2H4Ph)24 can be viewed as a biicosahedral Au23 core capped by the remaining
fifteen gold atoms (Figure S5c and S5d);38 Au25(SC2H4Ph)18 contains a centered
icosahedral Au13 kernel encapsulated by the remaining twelve gold atoms (Figure S5e
and S5f).23 Clearly, as shown in Figure 2e, the three reference clusters without
internal vacancies exhibited a very low conversion of CO2. This result again
suggested that the Au24 catalyst with internal vacancy was more effective than those
without internal vacancies for the catalytic hydrogenation of CO2.
6
ACS Paragon Plus Environment

Page 7 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
a
Au-Au
Data
Fit
b
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Au25 1st run
Au-P/S
Au foil
Au25 fresh
Au25 1st run
Au1.62+
Au25 fresh
5 6
11900
11920
11940
11960
1
2
3
4
Distance (Å)
Energy (eV)
c
Data
Fit
d
Au-Au
Au-P/S
Au24 5th run
Au24 3rd run
Au foil
Au24 fresh
Au24 1st run
Au24 3rd run
Au24 5th run
Au24 1st run
Au1.62+
Au24 fresh
11900
11920
11940
11960
1
2
3
4
5
6
Energy (eV)
Distance (Å)
Figure 3. (a) XANES profiles and (b) EXAFS spectra with fits of the Au25 catalysts before
reaction and after the first run reaction. (c) XANES profiles and (d) EXAFS spectra with fits of
the fresh Au24 catalyst and the spent Au24 catalysts with reaction cycles. The XANES profile of
Au1.62+ as oxidic Au standard.39
More interestingly, the internal vacancy of Au24 could be conducive, to an
extent, to mitigating the cluster’s aggregation and retaining the cluster’s feature. As
displayed in Figure 2f, the Au25 catalyst lost its catalytic activity after only the first
run reaction, due to the severe aggregation of clusters throughout the reaction. X-ray
absorption near edges structure (XANES) and extended X-ray absorption fine
structure (EXAFS) data showed that Au25 completely degraded into metallic gold
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 40
1
2
3
4
5
6
7
8
9
after the first run (Figure 3a and 3b, Table 1). Transmission electron microscopy
(TEM) experiments further showed that the Au25 clusters after the first run aggregated
to form large-sized nanoparticles (nanocrystal) (Figure S6a-S6c). Note that large Au
nanoparticles gave a very low activity under identical reaction conditions. In contrast,
deduced from the catalytic recyclability of Au24, the catalytic ability of Au24 gradually
decreased during the initial 6 cycles and remained essentially unchanged during the
subsequent cycles (Figure 2f). The gradual deactivation of the Au24 catalyst was
attributed to the fact that the Au24 cluster did not substantially grow into the
nanocrystal over the reaction cycles. From XANES data (Figure 3c and Table 1), the
charge state of the Au24 cluster appeared to converge towards an intermediate between
the positive valence state of the fresh Au24 and the metallic state of Au foil. The
presence of the oxidic Au species in the fresh cluster is mainly due to the electron
donation from the surface Au atoms to the ligands.27 From EXAFS analysis (Figure
3d and Table 1), the peaks for the Au-Au bonds intensified and the peaks for the
Au-P/Au-S bonds abated over the reaction cycles, implying that a part of Au24 clusters
might have been converted into metallic gold clusters during each cycle. Of note,
Au-Au bonds in fresh Au24 were not observed in EXAFS spectra, due to the thermal
fluctuation of Au-Au bonds at room temperature.27,40,41 From Table 1, the
coordination number (CN) of Au-Au in Au24 increased gradually with cycles (the CN
value was 3.6±0.4 for Au24 after the fifth reaction), but the coordination number of
Au-Au in Au25 rose dramatically (the CN value went up to 10.2±0.5 for Au25 after the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
first run, which was close to the CN value of gold foil). In addition, the evolution of
8
ACS Paragon Plus Environment

Page 9 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
the XANES and EXAFS data (Figure S7 and Table S1) of the Au24 catalyst with
reaction time further confirmed that aggregation of the Au24 clusters was much more
reluctant under reaction condition than that of Au25.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Valence of gold (δ) and EXAFS fitting results (R: distance; CN: coordination number; σ2:
Debye-Waller factora; △E0: inner potential correctionb) of Au24 and Au25 with reaction cycles.
Au-P/S
Au-Au
Δ
Sample
R (Å)
CN
R (Å)
CN
12

Au foil
0


2.8580.002
2.32 0.01
3.020.05
2.20.2
0.60.3
Au24 fresh
+0.59

2.310.01
2.620.02
2.10.2
0.90.2
Au24 1st run
Au24 3rd run
Au24 5th run
+0.35
+0.35
+0.28
2.870.02
2.880.02
2.860.01
1.20.4
2.10.4
3.60.4
2.310.01
2.620.02
2.00.2
0.80.2
2.320.01
2.620.05
1.70.2
0.40.3
2.330.01
2.600.02
2.50.3
1.00.4
Au25 fresh
+0.31
+0.01
2.750.06
2.860.01
0.40.1
Au25 1st run


10.20.5
aσ2 = 0.005 and 0.008 Å2 for all the analyzed Au-P/S and Au-Au shells, respectively. b△E0 varies
in the range of 49 eV.
Moreover, X-ray diffraction (XRD) pattern of fresh Au24/SiO2 was similar to that
of spent Au24/SiO2, in which the crystal plane of gold was hardly observed in the fresh
and spent Au24/SiO2 (Figure S8a). However, the XRD pattern of fresh Au25/SiO2 was
different from that of spent Au25/SiO2. As shown in Figure S8b, for the spent
Au25/SiO2 sample, the diffraction peaks at the 2θ of 38.2, 44.5, 64.6 and 77.5o, which
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 40
1
2
3
4
5
6
7
8
9
were typically indexed as the (111), (200), (220), and (311) diffractions, matching the
diffraction pattern of the face-centered cubic (fcc) gold nanocrystal well (JCPDS,
PDF No.03-065-2870). This further indicated that large-sized gold nanocrystal had
been formed from the Au25/SiO2 reaction system. TEM results also proved that the
Au24 cluster grew slightly throughout the cycles (Figure S6d-S6h). The chemical
adsorption capability of the Au24 and Au25 catalysts was probed by in situ
Fourier-transformed infrared (FT-IR) spectroscopy. As shown in Figure S9, the
intrinsic chemical adsorption capability of Au24 dropped with reaction time of CO2
hydrogenation, while the chemical adsorption capability of Au25 had no longer
changes after 2-hour reaction, indicating a slow fall in the activity of Au24 but a sharp
fall in the activity of Au25. These results explicitly revealed the superior stability of
the Au24 catalyst to the Au25 catalyst.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Since the small H2 molecule was one of reagents, a fanciful question is whether
H2 could enter the internal vacancy of Au24 to protect the structure of Au24 during the
reaction. To address this, the two clusters were calcinated at 150 oC under H2/N2 flow
for 1 hour, after which the Au24 cluster remained robust but the Au25 cluster had fully
degraded (Figure S10 and Table S2). The result seemed to suggest that the existence
of H2 might maintain the structure of Au24, but the thermal treatment experiment in a
vacuum for the cluster ruled out the possibility (Figure S10 and Table S2), as the
absence of H2 did not change the outcomes of the treatment.
Furthermore, the reaction conditions such as the reduction atmosphere and high
temperature may not only influence the atom-packing structure of gold atoms but also
10
ACS Paragon Plus Environment

Page 11 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
strongly alter the surface organic ligands. The changes of ligands for the Au24 and
Au25 clusters treated under different conditions were measured by in situ FT-IR
spectroscopy. Figure 4a and 4b showed the IR spectra of the two clusters treated at
different temperatures under the H2/N2 flow. The band at 3054 cm-1 is assigned to the
CH stretching vibration of benzene rings in PPh3 or thiolate ligands.42 The bands at
2914 and 2851 cm-1 are assigned to the asymmetric and symmetric stretching
vibrations of CH2 groups in the thiolate, respectively.43 The band around 1474 cm-1 is
related to the deformation vibration of CH groups in benzene rings.44 The band at
1428 cm-1 is associated with the stretching vibration of P-Ph in PPh3.45 As the
temperature was increased from 50 to 130 oC, for the Au24 cluster, the vibration bands
attributed to the CH, CH2, and P-Ph groups in the ligands only declined slightly
(Figure 4a). In contrast, for the Au25 cluster, the vibration bands associated with the
CH, CH2 and P-Ph groups were dramatically weakened with the elevated temperature
(Figure 4b). This indicated that a greater amount of ligands including PPh3 and
thiolate had been removed from Au25, compared to Au24. Figure 4c and 4d showed the
change of ligands over time for the two clusters under the treatment of H2/N2 and the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
o
constant temperature of 130 C. In particular, the vibration bands associated with the
organic ligands in the Au25 cluster fully disappeared after a 5-hour treatment, whereas
those in the Au24 cluster remained after a 10-hour treatment. The results hinted that
the removal of the ligands from the Au24 catalyst was comparatively slow, but the
exfoliation of ligands from the Au25 catalyst was rapid and thorough.
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 40
1
2
3
4
5
130 oC
130 oC
a
b
6
7
8
9
50 oC
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
50 oC
s(CH)
s(CH)
δas(CH)
(P-Ph)
δas(CH)
s(CH2)
s(CH2)
as(CH2)
as(CH2)
Au24
Au25
(P-Ph)
4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
c
10 h
10 h
d
0 h
0 h
Au25
Au24
4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
Figure 4. In situ FT-IR spectra of the Au24 and Au25 clusters under H2/N2 (5% H2) flow: (a, b) at
different temperature; (c, d) at 130 oC for different time.
The observation that Au25 is prone to aggregation and Au24 exhibits good
aggregation-resistance indicates that Au24 is much more catalytically stable than Au25
during CO2 hydrogenation. This is opposite from the previous observation on the
stability of Au25 and Au24 in a homogenous catalytic reaction.27 The effective physical
forms of Au24 and Au25 in the heterogeneous reaction of CO2 hydrogenation must be
totally different from the forms of Au24 and Au25 in the homogenous reaction. It is
evident from the EXAFS and infrared spectroscopy results that the ligands of Au24
and Au25 are extensively removed during the CO2 reaction due to comparatively high
12
ACS Paragon Plus Environment

Page 13 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
temperature and hydrogen pressure, suggesting that the Au25 and Au24 clusters are no
longer well passivated. The influences of the depassivation on Au25 and Au24 were
examined using ab initio molecular dynamics (AIMD) simulations with NVT
ensembles under the experiment temperature of 400 K for 8 ps (Figure S11 and Figure
5). The structural evolution of the Au24 and Au25 clusters is found to be dependent on
the degree of depassivation. The fully passivated Au24 and Au25 remain biicosahedral
without notable structural reconstruction (Figure 5a). When two thiolates and one
phosphide protective ligands are removed from the clusters, the resultant
[Au25(PR3)9(SR′)3Cl2]2+ undergoes small geometry changes with the rod shape
essentially retained, and the resultant [Au24(PR3)9(SR′)3Cl2]+ exhibits significant
geometry changes towards a cage shape with a central vacancy. When four thiolates,
three phosphides and a chloride ligands are removed from the clusters, the resultant
[Au25(PR3)7(SR′)Cl]2+ and [Au24(PR3)7(SR′)Cl]+ undergo reconstructions into a
close-packed and a cage-like amorphous clusters, respectively. Such trends in
structural evolution extend as the ligands are further removed from the clusters. The
fully depassivated [Au25]2+ reconstructs into a close-packed cluster and [Au24]+
becomes a cage with an interior vacancy. Our AIMD simulation results suggest that
maintaining the biicosahedral shapes of Au24 and Au25 requires essentially full
passivation. Depassivated Au25 reconstructs into the close-packed cluster without any
interior vacancy which, despite being amorphous, is structurally akin to the fcc
crystal. Depassivated Au24 reconstructs into the cage-shaped cluster. The vacancy
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
originating from the biicosahedral [Au24(PR3)10(SR′)5Cl2]+ is preserved during the
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 40
1
2
3
4
5
6
7
8
9
reconstructions for all degrees of depassivation (Figure 5b). Interestingly, the
vacancies of the depassivated Au24 clusters are animate during the structural
evolution, in terms of both structure and location, as indicated by the drifting of the
vacancy in the reconstructing cluster (Figure 5c). In other words, the vacancy is not
annihilated despite the depassivated Au24 cluster undergoes dynamic structure
reconstruction.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We have shown that whether there exists a vacancy in the Au cluster leads to
different reconstruction behaviors following the depassivation. Is the depassivation
thermally viable, will the reconstructing depassivated clusters aggregate, and what
role can the vacancy potentially play during the aggregation? To answer these
questions,
the
thermodynamics
for
the
depassivation
of
[Au25(PPh3)10(SC2H4Ph)5Cl2]2+ and [Au24(PPh3)10(SC2H4Ph)5Cl2]+ and the initial
aggregation of the depassivated clusters were predicted at the density functional
theory (DFT) level. The dissociation of the PPh3 phosphine is calculated to be
endothermic by ~180 kJ/mol and the removals of the SC2H4Ph thiolate and chloride
ligands are predicted to be less than 90 kJ/mol for both clusters (Figure S12),
indicating facile removal of ligands at high temperature in the presence of H2. Thus,
the Au25 and Au24 clusters during the CO2 hydrogenation might be fully or partially
depassivated.
Here
we
consider
the
extreme
situation
where
[Au25(PPh3)10(SC2H4Ph)5Cl2]2+ and [Au24(PPh3)10(SC2H4Ph)5Cl2]+ are fully
depassivated to form [Au25]2+ and [Au24]+. Direct DFT optimization calculations and
AIMD simulations lead to different structures for the fully depassivated clusters.
14
ACS Paragon Plus Environment

Page 15 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
From direct DFT optimization calculations, the fully depassivated [Au25]2+ remains
the biicosahedron, whereas [Au24]+ undergoes a spontaneous relaxation to form a
double-cage structure that contains a small vacancy and a large vacancy (Figure S13).
The AIMD simulations find [Au25]2+ as a close-packed amorphous cluster and [Au24]+
as a single-walled cage, which are 284 kJ/mol and 14 kJ/mol more stable than the
biicosahedron [Au25]2+ and the double-cage [Au24]+, respectively (Figure S13). Note
that the double-cage [Au24]+ not only is nearly as stable as the single-walled cage
[Au24]+, but also structurally resembles the partially depassivated Au24 clusters from
AIMD ([Au24(PR3)7(SR′)Cl]+ and [Au24(PR3)9(SR′)3Cl2]+) to some extent. The
double-cage [Au24]+ and close-packed [Au25]2+ are considered to represent the
depassivated Au24 and Au25 clusters during the CO2 hydrogenation, respectively, and
are used in further computations. As shown in Figure S14, the total energy for the full
depassivation of the Au25 catalyst that leads to the close-packed [Au25]2+ is calculated
to be 1805 kJ/mol with 7/2 equivalents of H2 involved, and the total depassivation
energy for the double-cage [Au24]+ is 1521 kJ/mol. The much lower depassivation
energy of [Au24]+ suggests that [Au24]+ has much lower surface energy than [Au25]2+.
The initial aggregation of the [Au25]2+ and [Au24]+ clusters can be achieved by
atomic gold migration, i.e. gold atoms being transferred from one cluster to another to
form larger clusters, analogous to Ostwald ripening. According to DFT results (Figure
S14), both [Au25]2+ and [Au24]+ exhibit Au dissociation energy comparable to Au
affinity, and the 2[Au25]2+ → [Au24]2+ + [Au26]2+ and 2[Au24]+ → [Au23]+ + [Au25]+
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
migration reactions are both mildly exothermic (∆H = –10 and –17 kJ/mol
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 40
1
2
3
4
5
6
7
8
9
respectively). Therefore, the initial aggregation is spontaneous for both fully
depassivated clusters. Both the optimized [Au23]+ and [Au25]+ contain an interior
vacancy (Figure S14). It is therefore hypothesized that the vacancy of [Au24]+ may be
preserved during the initial aggregation, and is likely to be persistent throughout the
aggregation. We hypothesize that the different aggregation behaviors of the
depassivated Au24 and Au25 clusters are governed by the presence/absence of the
interior vacancy. Previous computational and experimental studies suggested an
intrinsic size-structure relationship for Aun (where n is gold atom number) clusters,
where amorphous and cage structures46-52 dominate small cluster sizes n < 30 while
the fcc-crystal structure motif 53,54 prevails for n = 20 and n > 35. The sizes of the
Au24 and Au25 clusters are only slightly below the critical size for amorphous-to-fcc
transition. Hence, the close-packed amorphous [Au25]2+ is expected to grow into the
fcc cluster facilely. For growth of [Au24]+ beyond the critical size, the persistent
vacancies might divide the particle into regions containing de facto small amorphous
Au clusters that locally follows the intrinsic size-structure relationship of gold
particles, and the whole cluster might remain amorphous. This could be why the Au25
catalysts spontaneous grow into large fcc nanoparticles under experimental conditions
while the Au24 catalysts exhibit much lower aggregation propensity and are not fully
grown into fcc-nanoparticles, as found by the experiment.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
16
ACS Paragon Plus Environment

Page 17 of 40
Journal of the American Chemical Society
1
2
3
4
5
a
b
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c
Figure 5. AIMD simulations for the fully passivated, partially depassivated, and fully
depassivated Au24 and Au25 catalysts with an NVT-ensemble under a temperature of 400 K. (a)
Relaxed cluster structures at 8 ps for Au24 and Au25 clusters with different degrees of
depassivation. (b) Evolution of the vacancy radius in the NVT-AIMD simulations of the fully
passivated, partially depassivated, and fully depassivated Au24 clusters. (c) The distance between
the geometric center of Au atoms and the center of the vacancy as a function of time in the
NVT-AIMD simulations of the fully passivated, partially depassivated, and fully depassivated
Au24 clusters. The atomic vacancies are indicated by light-green spheres. Color codes for
elements: orange = Au, pink = P, yellow = S, green = Cl, and black = C. H atoms are omitted. PR3
(R = H), SR′ (R′ = CH3).
We next turned our attention to the potential reaction mechanism of CO2
hydrogenation on the Au24 catalyst. In situ monitoring of the reaction intermediates
formed on the two catalysts was performed by time-resolved in situ FT-IR
17
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 40
1
2
3
4
5
6
7
8
9
spectroscopy during the interaction of the catalysts with 2 bar reaction gas (CO2:H2 =
1:3) at 130 oC. For the Au24 reaction system, as shown in Figure 6, the bands at 2954,
2824 and 1404 cm-1 are assigned to the asymmetric and symmetric C-H stretches of
methoxy groups, and C-H bending vibrations, respectively.55 The bands at 2892,
1696, 1457 and 1065 cm-1 are associated with formate species, in which the first three
bands arise from the C-H, C=O, and C-O stretching modes of formate groups,56,57 and
the last band is attributed to the C-H scissoring vibration of formate groups.58 The
stretching vibrations of C-O-C of DME were found at 1249 and 1180 cm-1, and the
asymmetric C-H stretching vibration of DME appeared at 2992 cm-1.59-61 The other
three IR bands observed at 3031, 1280 and 1140 cm-1 are respectively assigned to the
CH3 rocking, C-O stretching, and C-H stretching vibrations of methyl formate (MF).62
In addition, the absorption bands close to 3108 and 1642 cm-1 correspond to the O-H
stretching and in-plane deformation vibrations of interfacial water, respectively.63 The
IR band at 2075 cm-1 is assigned to CO adsorbed on Au sites.64 The characteristic
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
δ−
peaks of the carboxylate (CO2 ) species were located near 1515 and 1357 cm-1.57,58
For the Au25 reaction system, the IR signals of the formate, methoxy, and DME
species were also detected (Figure S15). Noticeably, the intensities of these IR bands
for the Au25 catalyst were much lower than those for the Au24 catalyst, implying the
higher reactivity of Au24 than Au25 in CO2 activation and hydrogenation.
18
ACS Paragon Plus Environment

Page 19 of 40
Journal of the American Chemical Society
1
2
3
4
5
a
b
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c
d
Figure 6. Time-resolved in situ FT-IR results of surface species formed on the Au24 catalyst with
2 bar reaction gas (CO2:H2 = 1:3): (a) In situ FT-IR spectra from 3500 to 2450 cm-1; (b) In situ
FT-IR spectra from 2200 to 1000 cm-1; (c) Contour map from 3500 to 2450 cm-1; (d) Contour map
from 2200 to 1000 cm-1.
Going further, the calculated potential energy surface (PES) for the CO2
hydrogenation leading to the DME major product over the fully depassivated
double-cage Au24 derived from [Au24(PPh3)10(SC2H4Ph)5Cl2]+ is shown in Figure 7.
+
A two-stage reaction mechanism is proposed, comprising a one-carbon (C1)
hydrogenation stage and a two-carbon (C2) coupling stage. During the first stage, the
one-carbon (C1) hydrogenation occurs first to form Au(CHO) at an exposed Au
following the reverse water-gas shift (RWGS) reaction path (Figure 7a), similar to the
CO2 hydrogenation catalyzed by Cu/ZnO catalysts.65 The reaction path is briefly
19
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 40
1
2
3
4
5
described as CO2 → Au(COOH) → Au(CO) → Au(CHO). In Au(CHO), O forms an
6
7
additional bond with the Au adjacent to the C-Au. It is noted that the vacancies in
8
9
+
Au24 are well preserved during the C1 hydrogenation stage. The second stage occurs
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
when a second CO2 attacks an Au site that is adjacent to the existing Au(CHO) site,
which results in a (CHO)Au(CHO) intermediate with two C1 ligands (Figure 7b). In
(CHO)Au(CHO), the adsorption of the two CHO ligands is asymmetric, and Au is
bonded to C and O atoms from the two ligands separately. It should be pointed out
+
that the atomic vacancies remain in Au24 upon the formation of (CHO)Au(CHO).
The two CHO ligands of the (CHO)Au(CHO) species can undergo C–O coupling
reaction with ½ equivalent of H2, and spontaneously form a C2 η2 HO-CH-O-CH
ligand bridging the two adjacent exposed Au atoms (∆H = –25 kJ/mol) due to the
+
asymmetric dual-ligand adsorption. Notably, the Au24 double-cage is squeezed upon
+
the formation of HO-CH-O-CH and a vacancy in Au24 disappears consequently.
+
Such relaxation of the Au24 double-cage should release the stress between the two
Au sites caused by the strong η2 adsorption and stabilize the HO-CH-O-CH
intermediate, and hence favors the coupling reaction. Further hydrogenation of
HO-CH-O-CH leads to HC-O-CH, and H2C-O-CH η2 C2 intermediates, with the
+
vacancy remaining disappeared from Au24 . Upon the formation of the subsequent
hydrogenation intermediates and product, i.e. H2C-O-CH2, H3C-O-CH2 and DME, the
missing vacancy grows back (Figure 7b) as the Au24+ catalyst is blown up to return to
the unreacted state. This suggests that the deformation of the double-cage is reversible
and the vacancy can be regenerated within the catalytic cycle when the bridge
20
ACS Paragon Plus Environment

Page 21 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
adsorption is weakened (as for H2C-O-CH2) or destroyed (as for H3C-O-CH2). The
formation mechanism of DME on the 24-atom gold catalyst provides a new insight
that is different from the conventional two-step mechanism comprising methanol
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
synthesis via hydrogenation of CO2 and methanol dehydration to form DME.66
a
b
Figure 7. Calculated potential energy surfaces for CO2 hydrogenation to form DME at the DFT
level with PBE/Aug-cc-pVDZ(-pp): (a) RWGS to form one-site Au-CHO intermediate species
and (b) two-site coupling reaction to form DME. Color codes for elements: orange = Au, pink = P,
yellow = S, green = Cl, black = C, red = O, and white = H. CAT = catalyst. Blue and green
spheres indicate the small and large vacancies in the catalyst. Reaction enthalpies at 0 K are given
in kJ/mol.
+
Note that in the DFT calculations, the fully depassivated Au24 is chosen to
understand the aggregation and catalytic mechanism during the CO2 hydrogenation.
21
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 40
1
2
3
4
5
6
7
8
9
From the experimental results, we are aware that the depassivation of
[Au24(PPh3)10(SC2H4Ph)5Cl2]+ is slower than that of [Au25(PPh3)10(SC2H4Ph)5Cl2]2+,
and some protective ligands remain on the Au24 clusters during the reaction. The fully
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
+
depassivated Au24 is chosen because: (1) the infrared and EXAFS data give average
+
characteristics of the reaction system, which does not rule out the existence of Au24 ;
(2) the partially depassivated Au24 clusters as intermediate states are expected to
contain vacancies since the two extreme states ([Au24(PPh3)10(SC2H4Ph)5Cl2]+ and
+
Au24 ) contain vacancies; (3) the realistic Au24 clusters, which could be partially
depassivated and slightly aggregated, are likely to contain local structures represented
+
by Au24 , which is supported by the AIMD results. Overall, on the basis of DFT
calculations and theory, it is suggested that the depassivated and aggregated
[Au24(PPh3)10(SC2H4Ph)5Cl2]+ could contain non-fcc cage-like subsurface local
structures, which may exhibit breath-like motions to promote the coupling of reaction
intermediates with decent catalytic recyclability.
3. CONCLUSIONS
To summarize, we report the potential contribution of the internal vacancy/cage
in the Au24 catalyst to the catalytic performance on the CO2 hydrogenation. The
experimental studies show that the Au24 cluster with internal vacancy is much more
reluctant to aggregate into large-size nanocrystal and further exhibits enhanced
activity and stability, compared to the Au25 cluster without internal vacancy. DFT
computations and theory suggest that, not only the vacancy of Au24 is possibly
22
ACS Paragon Plus Environment

Page 23 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
preserved during the reaction at different surface ligand coverages, but also
vacancy-protected structural relaxation may mitigate the sintering and deactivation of
the Au24 catalyst. In addition, we propose a molecular-level two-stage reaction
pathway of CO2 hydrogenation on the Au24 catalyst towards DME product. We
anticipate that the exploration of potential influences of internal vacancies on catalytic
properties can bring a new scientific vision in understanding the contributions of
non-surface/non-interface sites on the overall catalytic performances.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4. EXPERIMENTAL AND COMPUTATIONAL METHODS
4.1
Catalysts
synthesis.
[Au24(PPh3)10(SC2H4Ph)5Cl2]+
and
[Au25(PPh3)10(SC2H4Ph)5Cl2]2+ clusters were synthesized using the previous
36
procedures. 0.2 mmol HAuCl4·4H2O in 5 mL pure H2O was added in 10 mL toluene.
After 20 min, the aqueous phase was removed by a glass tube. 0.6 mmol PPh3 was
added to the toluene phase after 0.6 mmol NaBH4 dissolved in 6 mL ethanol was
poured. After 2 h, the solution was evaporated to obtain the black crude product, and
15 mL dichloromethane was applied to extract the precursor. 200 μL PhC2H4SH was
added to the above dichloromethane solution of black product. The solution was kept
at 40 °C for 12 h. The Au25 was purified with hexane and extracted with ethanol.
Then 20 mg purified Au25 clusters and 1.2 g PPh3 were mixed in 10 mL
dichloromethane. The reaction continued for 24 h at 40 °C. The Au24 clusters were
obtained after washed with hexane and extracted with toluene.
[Au37(PPh3)10(SC2H4Ph)10X2]+ was synthesized as follows.37 0.2 mmol HAuCl4·4H2O
23
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 40
1
2
3
4
5
6
7
8
9
in 5 mL pure H2O was mixed with 0.233 mmol tetra-n-octylammonium bromide
(TOAB) in 10 mL toluene. After 20 min, the aqueous phase was removed by a glass
tube. 0.6 mmol PPh3 was added to the toluene solution, after 0.6 mmol NaBH4
dissolved in 6 mL ice-cold water was poured. After 2 h, the solution was evaporated
to obtain the crude product, 15 mL dichloromethane was applied to extract precursor.
200 μL PhC2H4SH was added to the above solution. The solution was kept at room
temperature for 36 h. The product was purified with methanol and extracted with
acetonitrile/DCM (3:1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Au38(SC2H4Ph)24 was synthesized based on the reported method.38 0.5 mmol
HAuCl4·4H2O and 2.0 mmol glutathione were dissolved in 20 mL acetone under
vigorous stirring for 30 min. The mixture was then cooled to 0 °C. After 30 min, a
solution of NaBH4 (5 mmol, dissolved in 6 mL cold water) was rapidly added. After
30 min, 6 mL water was added to dissolve the precipitations. Then the solution was
mixed with 0.3 mL ethanol, 2 mL toluene, and 2 mL PhC2H4SH. The solution was
kept at 80 °C for 40 h. The organic phase was washed with ethanol to remove excess
thiol. The product was extracted with toluene.
Au25(SC2H4Ph)18 was synthesized based on the reported method.23 0.203 mmol
HAuCl4·3H2O and 0.235 mmol TOAB were added in 15 mL THF. After stirred for 30
min, 0.140 mL phenylethanethiol was added to the above mixture. After 1 h, NaBH4
(76.8 mg, dissolved in 5 mL of cold water) was added to the mixture at once. The
reaction was stopped after 3 h. The crude product solution was dried by a rotary
evaporator and washed with methanol and then collected by centrifugation.
24
ACS Paragon Plus Environment

Page 25 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Preparation of Aun/SiO2: The supported Aun catalysts were obtained by stirring a
solution of Aun clusters in dichloromethane with a calculated amount of SiO2 (fumed,
140 m2/g) for 12 h and the solution was removed in flowing nitrogen. The loading
weight of gold was detected by ICP (inductively coupled plasma–atomic emission
spectroscopy).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4.2. CO2 hydrogenation. 100 mg catalyst (～0.5% Au clusters) was added into a
stainless-steel autoclave equipped with an automatic temperature control system, and
the magnetically driven impeller was used to disperse catalyst in 15 mL pure water.
o
The temperature was heated to 130 C and the vessel was then pressurized by 2 MPa
mixed gas of CO2 and H2 with 5% N2 as internal standard under a stirring of 800 rpm.
The ratio of CO2 and H2 is 1:3. After a certain amount of time, the reaction was
stopped. The catalytic products analyzed by GC9860 equipped with thermal
conductivity detector with TDX-1 column and flame ionization detector with FFAP
and PLOT-Q capillary column.
The cycle experiments: The mixture after reaction was centrifuged to isolate the
catalyst, and then the catalyst was washed with pure water several times. After that,
the catalyst was re-introduced into the reactor to start a new run of reaction. The data
presented in Figure 2f are the new performance of the reused catalyst in every 4-hour
run.
4.3. Characterization. UV-vis absorption spectra were recorded using SHIMADZU
UV-1800 with a medium sampling scan rate. Transmission electron microscopy
(TEM) was recorded with a JEM-200CX (JEOL, Japan) at a 100 kV acceleration
25
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 40
1
2
3
4
5
6
7
8
9
voltage. The loading weight of Au was determined via the inductively coupled
plasma–atomic emission spectroscopy (ICP-AES, Optima 5300DV). The binding
energies of the surface species on the catalysts were determined by X-ray
photoelectron spectroscopy (XPS) (Thermo Scientific K-Alpha) using Al K (hν =
1486.6 eV) as the excitation source. Powder X-ray diffraction (XRD) patterns of the
samples were recorded on a Bruker D8 ADVANCE X.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The X-ray absorption fine structure (XAFS) spectra at Au L3 (E0 = 11919.0 eV)
edge were performed at BL14W1 beamline67 of Shanghai Synchrotron Radiation
Facility (SSRF) operated at 3.5 GeV under “top-up” mode with a constant current of
240 mA. The XAFS data were recorded under fluorescence mode with a Lytle ion
chamber at room temperature. The energy was calibrated accordingly to the
absorption edge of pure Au foil. Athena and Artemis codes were used to extract the
data and fit the profiles. The XAFS data of the fresh Au24/Au25 samples were from the
previous report.27 For the X-ray absorption near edge structure (XANES) part, the
experimental absorption coefficients as function of energies μ(E) were processed by
background subtraction and normalization procedures, and reported as “normalized
absorption” with E0 = 11919.0 eV for all the samples and Au foil standard. Based on
the normalized XANES profiles, the average valence of gold can be determined by
the white-line intensity with the help of the above standard references. For the
extended X-ray absorption fine structure (EXAFS) part, the Fourier transformed (FT)
data in R space were analyzed by applying first shell approximation and metallic Au
models for Au-P/Au-S and Au-Au contributions, respectively. The passive electron
26
ACS Paragon Plus Environment

Page 27 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
factors were determined by fitting the experimental data on Au foil and fixing the
coordination number (CN) of Au-Au to be 12, and then fixed for further analysis of
the measured samples. The parameters describing the electronic properties (e.g.,
correction to the photoelectron energy origin, E0) and local structure environment
including CN, bond distance (R) and Debye-Waller factor around the absorbing atoms
(2) were allowed to vary during the fit process. The fitted ranges for k and R spaces
were selected to be k = 313 Å−1 with R = 1.03.0 Å (k3 weighted).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Time-resolved in situ FT-IR experiments were carried out in an elevated pressure
cell with a TENSOR 27 (Sample Compartment LN-MCT Mid). To understand the
reaction mechanism, the temperature was raised from room temperature to 130 °C at
4 °C/min while purging with Ar flow, and then the background spectrum of the
catalyst was acquired. Then the elevated pressure cell was pressurized by 2 bar mixed
gas (CO2:H2 = 1:3) for different periods of time and then measured by in situ FT-IR.
In situ IR spectra of CO adsorption were collected using TENSOR 27 (Sample
Compartment RT-DLaTGS). IR spectra were recorded continuously to study the
surface changes. In order to follow surface changes over reaction time, the elevated
o
pressure cell was pressurized by 2 bar mixed gas (CO2:H2 = 1:3) at 130 C for
different periods before spectra collection. Then after flowing 1 bar Ar for 0.5 h at
150 °C, the sample was then cooled to room temperature and the background
spectrum was collected in Ar flow. CO adsorption was conducted at room
temperature for 30 min, and Ar flow was switched back to purge gaseous CO. IR
spectra of CO adsorbed onto the samples were recorded.
27
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 40
1
2
3
4
5
6
7
8
9
In situ IR spectra of the Au24 and Au25 clusters under H2/N2 flow at different
temperatures or at 130 °C for different time were collected using TENSOR 27
(Sample Compartment RT-DLaTGS). KBr under H2/N2 flow was scanned as a
background spectrum, and Au24 and Au25 dichloromethane solution was respectively
dropped on KBr, then heated under infrared lamp to remove the solvent. The
temperature was raised to 130 °C at 4 °C/min under H2/N2 flow and IR spectra were
recorded at specific temperature. Or the temperature was kept at 130 °C and IR
spectra were collected at specific time. In situ IR spectra of the Au24/SiO2, Au25/SiO2
and SiO2 samples under H2/N2 (5% H2) or Ar flow at 130 °C were collected using
TENSOR 27. KBr was heated under H2/N2 or Ar flow at 130 °C and scanned as a
background spectrum. Each sample was mixed with KBr and then placed into the cell
after compress. The temperature was raised to 130 °C at 4 °C/min under H2/N2 or Ar
flow and then the IR spectra were collected at an interval of 2 min.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4.4. Computational methods. The CO2 hydrogenation reaction catalyzed by the
heterogeneous Au24 and Au25 catalysts was modelled using density functional theory68
(DFT) level with the PBE functional.69 Local geometry optimization and vibrational
frequency calculations were carried out for the important catalytic species during the
hydrogenation catalysis using the Gaussian 09 software package.70
Ab initio molecular dynamics (AIMD)71 simulations were used to model the
reconstruction of Au24 and Au25 catalysts with different degrees of depassivation (i.e.
removal of protective ligands). Four sets of AIMD simulations were performed: (1)
intact [Au24(PH3)10(SCH3)5Cl2]+ and [Au25(PH3)10(SCH3)5Cl2]2+ catalysts, (2) catalysts
with 2 SCH3 and 1 PH3 removed, (3) catalysts with 4 SCH3, 2 PH3, and 1 Cl removed,
28
ACS Paragon Plus Environment

Page 29 of 40
Journal of the American Chemical Society
1
2
3
4
5
6
and (4) fully-depassivated metallic catalysts. The AIMD simulations were performed
with an NVT ensemble (attached to a Nosé-Hoover thermostat72,73) at 400 K (or 130
7
8
9
̊
implemented in the CP2K software package75 at the DFT level of theory with the PBE
exchange-correlation functional and a DZVP-MOLOPT-SR-GTH76 basis set and
pseudopotential.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The fully passivated (fresh) Au24 and Au25 clusters were modelled with
simplified [Au24(PH3)10(SCH3)5Cl2]+ and [Au25(PH3)10(SCH3)5Cl2]2+ models at the
PBE level with the Aug-cc-pVDZ(-pp) basis set and pseudopotential,77-79 where the
PPh3 ligands of the catalysts were replaced by PH3 and the SCH2CH2Ph ligands
replaced by SCH3. The zero-point energy corrections were considered for all of the
relative species and were included in the reaction enthalpies.
The
depassivation
of
the
[Au24(PH3)10(SCH3)5Cl2]+
and
[Au25(PH3)10(SCH3)5Cl2]2+ was tested for each ligand, as given by equations (1) to
(6):
[Au24(PH3)10(SCH3)5Cl2]+ → [Au24(PH3)9(SCH3)5Cl2]+ + PH3
(1)
(2)
(3)
(4)
(5)
(6)
[Au24(PH3)10(SCH3)5Cl2]+ + H2 → [Au24(PH3)10(SCH3)4HCl2]+ + HSCH3
[Au24(PH3)10(SCH3)5Cl2]+ + H2 → [Au24(PH3)10(SCH3)5HCl]+ + HCl
[Au25(PH3)10(SCH3)5Cl2]2+ → [Au25(PH3)9(SCH3)5Cl2]2+ + PH3
[Au25(PH3)10(SCH3)5Cl2]2+ + H2 → [Au25(PH3)10(SCH3)4HCl2]2+ + HSCH3
[Au25(PH3)10(SCH3)5Cl2]2+ + H2 → [Au25(PH3)10(SCH3)5HCl]2+ + HCl
The reaction enthalpies for the full depassivation reactions (equations (7) and (8))
were also predicted.
29
ACS Paragon Plus Environment