Hydrogenation of CO2 to Methanol by Pt Nanoparticles Encapsulated in UiO-67: Deciphering the Role of the Metal-Organic Framework

Article abstract of DOI:10.1021/jacs.9b10873

Metal-organic frameworks (MOFs) show great prospect as catalysts and catalyst support materials. Yet, studies that address their dynamic, kinetic, and mechanistic role in target reactions are scarce. In this study, an exceptionally stable MOF catalyst consisting of Pt nanoparticles (NPs) embedded in a Zr-based UiO-67 MOF was subject to steady-state and transient kinetic studies involving H/D and ¹³C/¹²C exchange, coupled with operando infrared spectroscopy and density functional theory (DFT) modeling, targeting methanol formation from CO₂/H₂ feeds at 170 °C and 1-8 bar pressure. The study revealed that methanol is formed at the interface between the Pt NPs and defect Zr nodes via formate species attached to the Zr nodes. Methanol formation is mechanistically separated from the formation of coproducts CO and methane, except for hydrogen activation on the Pt NPs. Careful analysis of transient data revealed that the number of intermediates was higher than the number of open Zr sites in the MOF lattice around each Pt NP. Hence, additional Zr sites must be available for formate formation. DFT modeling revealed that Pt NP growth is sufficiently energetically favored to enable displacement of linkers and creation of open Zr sites during pretreatment. However, linker displacement during formate formation is energetically disfavored, in line with the excellent catalyst stability observed experimentally. Overall, the study provides firm evidence that methanol is formed at the interface of Pt NPs and linker-deficient Zr₆O₈ nodes resting on the Pt NP surface.

Full text of DOI:10.1021/jacs.9b10873

Subscriber access provided by Columbia University Libraries
Article
Hydrogenation of CO2 to Methanol by Pt Nanoparticles
Encapsulated in UiO-67: Deciphering the Role of the MOF
Emil S. Gutterød, Andrea Lazzarini, Torstein Fjermestad, Gurpreet Kaur, Maela Manzoli, Silvia Bordiga,
Stian Svelle, Karl P. Lillerud, Egill Skulason, Sigurd Øien-Ødegaard, Ainara Nova, and Unni Olsbye
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.9b10873 • Publication Date (Web): 03 Dec 2019
Downloaded from pubs.acs.org on December 5, 2019
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 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
Hydrogenation of CO2 to Methanol by Pt Nanoparticles Encapsulated
in UiO-67: Deciphering the Role of the MOF
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
Emil S. Gutterøda, Andrea Lazzarinia, Torstein Fjermestadb, Gurpreet Kaura, Maela Manzolic,
Silvia Bordigaa,d, Stian Svellea, Karl P. Lilleruda, Egill Skúlasone, Sigurd Øien-Ødegaarda,
Ainara Novab, Unni Olsbyea*
aCentre for Materials Science and Nanotechnology, Department of Chemistry, University of Oslo, Sem
Saelandsvei 26, N-0315 Oslo, Norway
bHylleraas Centre for Quantum Molecular Sciences, Department of Chemistry, University of Oslo, P.O.
Box 1033, Blindern, N-0315 Oslo, Norway
cDepartment of Drug Science and Technology and NIS - Centre for Nanostructured Interfaces and
Surfaces, University of Turin, Via P. Giuria 9, Turin 10125, Italy
dDepartment of Chemistry, NIS Interdepartmental Centre and INSRM reference centre, University of
Turin, via Quarello 15A, I-10135 Turin, Italy
eScience Institute and Faculty of Industrial Engineering, Mechanical Engineering and Computer
Science, University of Iceland, VR-III, 107 Reykjavik, Iceland
Abstract
Metal-organic frameworks (MOFs) show great prospect as catalysts and catalyst
support materials. Yet, studies that address their dynamic, kinetic and mechanistic role in
target reactions are scarce. In this study, an exceptionally stable MOF catalyst consisting of Pt
nanoparticles (NPs) embedded in a Zr-based UiO-67 MOF was subject to steady-state and
transient kinetic studies involving H/D and 13C /12C exchange, coupled with operando infrared
Page 1 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 35
1
2
3
4
5
6
7
8
9
spectroscopy and density functional theory (DFT) modeling, targeting methanol formation from
CO2/H2 feeds at 170 C and 1-8 bar pressure. The study revealed that methanol is formed at
the interface between the Pt NPs and defect Zr nodes, via formate species attached to the Zr
nodes. Methanol formation is mechanistically separated from the formation of co-products CO
and methane, except for hydrogen activation on the Pt NPs. Careful analysis of transient data
revealed that the number of intermediates was higher than the number of open Zr sites in the
MOF lattice around each Pt NP. Hence, additional Zr sites must be available to formate
formation. DFT modelling revealed that Pt NP growth is sufficiently energetically favored to
enable displacement of linkers and creation of open Zr sites during pretreatment. However,
linker displacement during formate formation is energetically disfavored, in line with the
excellent catalyst stability observed experimentally. Overall, the study provides firm evidence
that methanol is formed at the interface of Pt NPs and linker-deficient Zr6O8 nodes resting on
the Pt NP surface.
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
Keywords: CO2 hydrogenation, methanol, formate, transient kinetic analysis, kinetic isotope
effect, operando FTIR, DFT, MOF, UiO-67.
Introduction
Atmospheric levels of CO2 have risen at an alarming rate since the first half of the 20th
century following our continuous and increasing use of fossil fuels. Large cuts in CO2 emissions
can be made through utilization of greener alternatives of energy production such as solar and
wind power; however, these energy sources suffer from lack of continuity in energy output and
requires efficient methods for large scale energy storage in order to compete with fossil fuels.1
Page 2 of 35
ACS Paragon Plus Environment

Page 3 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
One of the solutions to this problem is production of easily stored liquid fuels with high
volumetric and gravimetric energy density, such as methanol, from CO2 and green hydrogen.2
This allows for continued use of already existing infrastructure.3 Substantial research efforts
have already been dedicated to the topic of valorizing CO2 through hydrogenation, mainly with
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
focus on the reverse water-gas shift (RWGS) reaction, methanation and methanol production.4-
13
The product distribution of CO2 hydrogenation varies significantly with the nature of the
catalyst and operating conditions.7 Most prior studies show that Pt-based systems are highly
selective for the RWGS reaction, with minor selectivity towards methane.11, 14-19 In that respect,
Kattel et al.18 performed a computational study of CO2 hydrogenation over Pt nanoparticles
supported on SiO2 and TiO2. They concluded that a RWGS and CO hydrogenation path,
forming CO and small amounts of methane, dominate over those materials. This finding is in
agreement with experimental evidence of CO formation mainly via surface carbonates.15, 16
Surface formates are also observed in several studies of the RWGS15, 16, 20, 21 and of
WGS22-24 reactions; however, the significance of such a pathway over Pt-based catalysts is
debated. In this regard, Burch, Goguet and Meunier23 conducted a critical analysis of the
experimental evidence for and against a formate mechanism over highly active Pt and Au WGS
catalysts. They argue that most published results do not provide definite evidence for or against
a formate pathway for the WGS reaction, and in the cases where reliable data are available, it
is at most a minor and slow reaction pathway.
Recent studies show that when supporting Pt on a methane producing Co-oxide
catalyst, methanol selectivity is observable under favorable conditions of low temperature and
elevated pressure25-27. Furthermore, one prior study28 demonstrated selectivity towards
Page 3 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 35
1
2
3
4
5
6
7
8
9
methanol over oxide-supported Pt where the supports alone are inactive in CO2 hydrogenation:
out of a set of catalysts producing mainly CO (> 91 %), Pt/ZrO2 showed the highest CH3OH
selectivity, reaching 6 % at 200 °C and 10 atm pressure.
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
Over the Zr-MOFs UiO-66 and -67, functionalized with Cu and Cu/Zn, high methanol
selectivity is ascribed to an important role of the Zr-MOF interface in the reaction.
Rungtaweevoranit et al.29 found XPS evidence of Zr reduction in the presence of Cu when
studying Cu NPs deposited on UiO-66 as a model system for the encapsulated version of the
catalyst (i.e. Cu in UiO-66). Similar findings from XPS on CuZn@UiO-67 samples with 100%
bipyridine-type linkers treated with reaction gas showed indications of Zr(IV) reduction to Zr(III),
argued as caused by H adsorption.30 In combination with H2- and CO2-TPD results, the authors
suggested participation of the Zr-cluster in the reaction by means of H-spillover from Cu and
CO2 adsorption on unsaturated Zr-sites. When the catalyst was prepared with regular UiO-67
(i.e. biphenyl-type linkers), both the CH3OH selectivity and catalytic stability decreased.
Although the CuZn@UiO-67 material showed substantial activity in methanol formation,
notably, both the crystallinity and specific surface area of the MOF were severely reduced
already by the deposition of copper. Thus, the material did not exhibit the well-defined MOF
structure of UiO-67 during the subsequent experiments.30
In a previous contribution, we reported the CO2 hydrogenation activity and selectivity
of an exceptionally stable Pt Zr-MOF catalyst, UiO-67-Pt, at ambient pressure. This catalyst
maintains its well-defined MOF structure even after long-term operation and is therefore well
suited as a model system for studying the influence of the MOF framework on the reaction. In
the current study, the focus is set on elucidating the role of the UiO-67 framework in CO2
hydrogenation to methanol through a kinetic investigation. In addition to standard steady-state
Page 4 of 35
ACS Paragon Plus Environment

Page 5 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
kinetic measurements, we employ H/D- and 13C/12C SSITKA, and operando FT-IR
measurements, methods scarcely employed in MOF catalysis literature. In combination with
DFT calculations, this work reveals unprecedented insight in the Pt–Zr-MOF interplay that
leads to methanol formation during conversion of CO2/H2 mixtures at 170 °C and 1–8 bar.
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
Experimental Details
Details of material preparation and standard characterization methods (N2 adsorption,
1H-NMR, EDX, TGA, TEM and CO-chemisorption) are provided in the SI.
Operando FT-IR spectroscopy was performed using a Bruker Vertex70 instrument
equipped with a liquid nitrogen-cooled MCT detector. Each spectrum was collected by
averaging 64 acquisitions with a resolution of 2 cm−1. The sample (pressed in a self-sustained
pellet of approx. 4 mg) was mounted inside a low free-volume cell from AABSPEC (model
#CXX), by which pressure, temperature and gas flow are controllable. Due to the low sample
amount, the CO2 conversion was too low to be determined reliably and parallel experiments
under comparable conditions were performed with the focus on gas-phase analysis (vide infra).
The sample was activated at 350 °C (5 °C/min ramp) in 10 % H2/He (10 ml/min) for 4 hours,
and then cooled to 170 °C in 10 ml/min He. The sample was kept under CO2 hydrogenation
(CO2/H2 = 1/6, 10 ml/min) reaction conditions for two hours, before the H2 flow was exchanged
to D2.
Catalytic testing was performed in a fixed-bed flow setup with a straight stainless steel
reactor (7 mm I.D.) operated under 1–8 bar, where effluent species is analyzed with an on-line
Q-MS (Pfeiffer) and GC-TCD-FID (Agilent). The MOF samples were reduced for 4 hours at
350 °C (5 °C/min ramp) in 20 ml/(min∙0.1gcat) flow of 10 % H2/Ar under ambient pressure.
Following the activation procedure, the reactor was cooled to 240 °C in inert flow, then
Page 5 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 35
1
2
3
4
5
6
7
8
9
operated for 4 hours reaching steady-state under a set of reference conditions (1/6/3
CO2/H2/inert, 20 ml/min, 200 mg, τ = 0.01 gcat∙min/ml and ambient pressure) before changing
conditions. The pre-reduced commercial samples Pt/SiO2 (abcr), Pt/Al2O3 (Chimet) and Pt/C
(Chimet), were heated directly to 240 °C under reference conditions, which were maintained
until steady state was reached. Dependent on the experimental aim, 0.05–0.2 g catalyst was
tested for CO2 hydrogenation in range: 5–25 % CO2, 40–90 % H2, T = 170 °C, p = 1–8 bar and
contact time τ = 0.004–0.04 gcat∙min/ml. A given set of operating conditions was fixed until
steady state was reached and for at least two hours. During kinetic studies, changes in reaction
conditions were performed in a random sequence. Each 3–4 set of conditions were the
reference conditions. They showed that the change in catalyst performance was negligible
during the kinetic studies. As reported in ref.17 a minor increase in the catalytic activity and
change in selectivity is observed for UiO-67-Pt during long-term operation. The same
procedure as described above was followed in the H/D exchange experiments but with D2
instead of H2. H/D SSITKA experiments were performed by operating the catalysts at steady
state under reference conditions, then switching the feed (1/6/3 CO2/H2/inert) rapidly to another
feed containing D2 instead of H2 (1/6/3 CO2/D2/inert) using an electronically controlled 4-port
2-way valve. Switches back and forth between the two feeds were performed in intervals of 8
hours. 13CO2/12CO2 SSTIKA experiments were performed in the same manner. The m/z values
traced for each specie in the respective experiments are tabulated in Table S1.
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
Intermediates and catalyst structures were optimized using Density functional theory (DFT)
calculations. The calculations were carried out using the mixed Gaussian and plane wave
method31-33 as implemented in CP2K 6.1.34, 35 The functional was PBE+D3, the atom-centered
basis set was DZVP-MOLOPT-SR-GTH,36 and the plane wave kinetic energy cut-off was 360
Ry. Further details are provided in the SI.
Page 6 of 35
ACS Paragon Plus Environment

Page 7 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
Results and Discussion
9
The UiO-66 series of Zr-MOFs (UiO-66, -67, -68) consists of Zr6O8 nodes connected
by dicarboxylate-terminated linkers. The ideal linker-to-node ratio is 6:1, but prior investigations
showed that factors like synthesis conditions (type of modulator and concentration,
temperature and crystallization time) and activation conditions may strongly affect this ratio,
leading to materials that are commonly referred to as having “missing cluster” or “missing
linker” defects, respectively37-43. Missing linker defects may be capped by modulator, anions of
the MOF precursor salts, solvent or OH-/H2O pairs.39, 42, 44, 45
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 current study was carried out using UiO-67 with 90 % biphenyl-4,4’-dicarboxylic
acid (BPDC) linkers and 10 % 2,2’-bipyridine-5,5’-dicarboxylic acid (BPYDC) linkers, hereafter
called UiO-67-Pt. Prior studies have demonstrated that the bipyridine entity in BPYDC is the
preferred anchoring site for Pt salts in this MOF.17, 46-48 In the current study, 1H NMR analysis
of digested material showed that the as-synthesized MOF contained 11 BPYDC linkers, 13
benzoic acid and 2 formic acid ligands, respectively, per 100 BPDC linkers (Table S3).
Furthermore, Thermo-Gravimetric Analysis (TGA) measurements indicated that the as-
synthesized MOF had a linker-to-node ratio of 5, suggesting that, in addition to benzoic acid
and formic acid, the material contained Cl- or OH-/H2O pairs, adding up to an average of 4 out
of 24 Zr coordination sites per Zr-node that were not connected to a linker molecule (Table
S3).
After wet impregnation with the Pt NP precursor, K2PtCl4, the BPYDC and benzoic acid
contents of UiO-67-Pt decreased slightly, while the formic acid content increased to 4 per 100
BPDC linkers (Table S2). Activation in a reducing atmosphere (10 % H2/Ar flow at 350 °C, 1
Page 7 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 35
1
2
3
4
5
6
7
8
9
bar, 4 h) transformed the Pt precursor salt into the active catalytic entity for CO2 hydrogenation,
Pt NPs.17, 46, 48 In the current case, Transmission Electron Microscopy (TEM) investigations
after activation showed the presence of Pt NPs homogeneously dispersed within the MOF
framework (Figure S5). The Pt NPs supported on UiO-67 have average diameter of 3.6  0.7
nm (Figure S5), i.e. larger than the diameter of the tetrahedral (1.2 nm) and octahedral (2.3
nm) cavities of the UiO-67 structure. Most of the Pt NPs displayed spherical shape, however,
careful inspection revealed the presence of NPs with squared borders and irregular shape,
possibly exposing well defined terraces, after activation and after reaction (Figure S5 and
Figure S6). Such features can arise from strong Pt-support interaction, and indeed, limited
broadening of the Pt NP size distribution was observed after prolonged testing (Figure S6). A
schematic illustration of a 3.6 nm Pt particle embedded in an 8 unit cell-enclosed octahedral
cavity, mimicking a representative Pt NP observed by TEM, is presented in Scheme 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
Scheme 1. Left: TEM micrograph showing the close packed (1 1 1) layers of Zr6O8 clusters in
UiO-67, with a spacing of 15.5 Å, overlaid with a 1600 atom Pt NP in the structure of UiO-67
Page 8 of 35
ACS Paragon Plus Environment

Page 9 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
viewed along (1 1 2) which is perpendicular to (1 1 1). Right: A 1600 atom Pt NP in UiO-67
viewed in the same direction. 6 Zr6O8 clusters have been decoupled from the MOF lattice to
accommodate the NP, and are decorating the NP surface.
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
CO2/H2 conversion and product selectivity obtained over Pt/C, Pt/SiO2, Pt/Al2O3 and
UiO-67-Pt at 170 °C and 1–8 bar is shown in Figure 1. Substantial selectivity differences were
observed at similar conversion levels (0.4–1.5 %): Over UiO-67-Pt, the methanol selectivity
increased from around 3 to 19 %—corresponding to a turn-over-frequency (TOF) of 0.01 s−1—
when the pressure increased from 1 to 8 bar (Figure 1). This is, to the best of our knowledge,
the second report of significant methanol formation from CO2 over a Pt-based catalyst where
the support alone is inactive in the reaction.28 Under the same conditions, there was only a
slight increase in methane selectivity from 1.2 to 1.6 %. Over Pt/Al2O3, both the methane and
methanol selectivity reached 10 % under 8 bar pressure (Figure 1). In contrast to UiO-67-Pt,
methane selectivity increased substantially with increasing pressure. Finally, over Pt/SiO2 and
Pt/C (Figure 1), CO was the only carbon-containing product observed, in accordance with the
theory predictions of Kattel et al. for unsupported Pt NPs.18 The formation of methanol over
UiO-67-Pt points to strong metal-support interactions, as previously reported for Cu NPs
embedded in UiO Zr-MOFs,29, 30 and may suggest that the MOF support plays an active role
during reaction, similarly to Al2O3 in the Water Gas Shift (WGS) and CO2 hydrogenation
reactions.8, 24, 49
Page 9 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 35
1
2
3
4
5
6
7
8
9
100
80
2
1
0
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
20
0
0
2
4
6
8
0
2
4
6
8
0
2
4
6
8
Ptotal [bar]
Figure 1. CO2 conversion (left axes, diamonds) and product selectivity (right axes) during CO2
hydrogenation under 1–8 bar total reaction pressure and T = 170 °C. Left: UiO-67-Pt. Middle:
Pt/Al2O3. Right: Pt/SiO2 (grey, blue) and Pt/C (black). CO, CH4 and CH3OH selectivity is
depicted as squares, circles and triangles, respectively. τ = 0.01 gcat∙min/ml, CO2/H2/inert =
1/6/3. For Pt/Al2O3 τ = 0.02 gcat∙min/ml.
To further assess the role of Pt NPs vs. support in CO2 hydrogenation, steady-state
H2/D2 exchange experiments were performed under CO2 hydrogenation conditions over UiO-
67-Pt, Pt/Al2O3, Pt/SiO2, UiO-67 and SiO2. The transient evolution of the HD molecule following
a feed switch from CO2 + H2 to CO2 + D2 is shown in Figure S9. Importantly, formation of HD
(and other products) was observed only over the Pt containing materials but not over the UiO-
67 and SiO2 supports alone, showing that dissociation of H2/D2 only occurs when Pt is present.
The amount of HD formed over UiO-67-Pt and Pt/Al2O3 was larger than over Pt/SiO2, and in
all cases 1–2 orders of magnitude higher than the amount of exposed Pt atoms in Pt
nanoparticles (Table 1). This observation, in combination with the observed HD tailing, strongly
Page 10 of 35
ACS Paragon Plus Environment

Page 11 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
suggests H/D exchange with hydroxyl groups on the support materials, either directly by
hydrogen spillover to/from the Pt NPs, or by H/D exchange with the water molecules formed
during reaction.
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. Pt amount, NP diameter (dNP), CO-uptake (NA,CO,RT ), estimated exposed Pt surface
(Nsurf,Pt) and the amount of HD formed (NA,HD) after switching from H2+CO2 to D2+CO2 at 170
°C, 1 bar.
c
d
Catalyst
Pt amount
(wt %)
2.7a
dNP
NA,CO,RT
(μmol/gcat)
Nsurf,Pt
NA,HD
(nm)
(μmol/gcat)
(μmol/gcat)
1200
UiO-67-Pt
Pt/Al2O3
Pt/SiO2
Pt/C
3.6 ± 0.7 1.7 ± 0.3
55
5b
5b
5b
1.4b
5 ± 2
2b
36
13
-
200
76
-
1100
540
-
aTheoretical amount of impregnation. A Pt amount of 2.4 ± 0.4 wt % was estimated from
c
EDX analysis (see SI). bObtained from the provider. Pulse-chemisorption at room
temperature. dEstimated from TEM by following the procedure described in the SI.
Insights into the origin of HD tailing and formed amount over the UiO-67-Pt sample
were obtained by a parallel operando FT-IR experiment under comparable conditions (1 bar,
170 °C, CO2/H2 = 1/6). When exchanging H2 for D2, the sharp signal of Zr-μ3-OH at 3669 cm-1
Page 11 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 35
1
2
3
4
5
6
7
8
9
39 decreased to zero with time while another equally sharp and intense peak, corresponding to
the deuterated species Zr-μ3-OD, increased at 2705 cm-1 (Figure S16). Interestingly,
quantification of the molar number of H in HD, NA,HD, yields 1200 mol H per gram catalyst,
which corresponds to about 70 % of the theoretical number of Zr-μ3-OH groups in the sample.
No other indication of H/D exchange (neither on Pt nor in linkers) was found from FT-IR
experiments. More detailed description of the spectra obtained during FT-IR experiments over
UiO-67-Pt, including the hydroxyl group region and the C-H bonds present in the aromatic
linkers (carboxylates region is omitted as out of scale), is given in the supporting section
(Figure S16). It is important to note that the chemical integrity of the catalyst is preserved during
reaction and changes in the spectra are therefore caused by reaction products interacting with
the sample. The most evident change is caused by the progressive increase of CO on the Pt
nanoparticles (Figure 2). The shape and frequency of this signal is compatible with carbonyls
linearly adsorbed on Pt atoms at the surface of NPs, as we already addressed in our previous
study on an analogous material.17 The size of the particles (3.6 nm) and the temperature (170
°C) justify the absence of bridged carbonyl species at lower frequencies. Changes in one of
the smaller peaks in the IR spectra is the most novel observation of this study: the peak arising
at 2745 cm−1 (Figure 2) is due to the appearance of bidentate formate groups39 (which, notably,
are absent after activation, see Figure 2) most likely coming from the progressive process of
CO2 reduction.The frequency of this weak feature is compatible with νs(COO) + δ(CH) vibration
of bidentate formates directly connected to open Zr-sites of the Zr-nodes in the MOF
framework.39, 50, 51 Further bands associated to formate are visible in the top part of Figure S16
in the region 3000–2800 cm−1. The spectrum of the sample collected during CO2 hydrogenation
after the subtraction of the activated one (Figure S17) highlights the formation of four additional
bands at 2910, 2888, 2868 and 2854 cm−1, ascribable to various formate species at the Zr-
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
Page 12 of 35
ACS Paragon Plus Environment

Page 13 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
node.39 This is to the best of our knowledge, the first report of formate formation at the MOF
Zr-node under CO2 hydrogenation reaction conditions.
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
0.03
0.01
2800
2760
2720
2120
2080
2040
2000
1960
1920
Wavenumber [cm-1]
Figure 2. FT-IR spectra of UiO-67-Pt collected during CO2 hydrogenation (CO2/H2 = 1/6, 10
ml/min, 170 °C, 1 bar) at different times (thick black curve for t = 0 min, grey scale from darkest
to brightest for 0 < t < 120 min, thick red curve for t = 120 min). The left figure shows the
magnified spectral region of the ν(C-H) for formate groups, while the right figure depicts the
spectral interval typical for CO linearly adsorbed on metal nanoparticles. Full range spectra
are reported in Figure S16.
The importance of these moieties is even more evident, thanks to the isotopic exchange
experiment between H2 and D2. When exchanging H2 for D2, the signals of the formate groups
shifted to lower wavenumbers; the signals in the 2950–2850 cm−1 region shifted to the region
of the CO2 roto-vibrational profile and are therefore not detectable. Conversely, the band at
2745 cm−1 shifted to 2168 cm−1 39 (Figure S16) and the intensity vastly increased as compared
to the H-analogue (IHCOO/IDCOO ≈ 0.02). The intensity increase indicates an inverse kinetic
Page 13 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 35
1
2
3
4
5
6
7
8
9
isotope effect, although possible differences in the extinction coefficient of the two species
cannot be excluded. Importantly, a corresponding, gradual increase in methanol production
rate was observed upon H2/D2 exchange under steady-state CO2 hydrogenation conditions
(Figure S14). At isotope equilibration, this difference in H- and D-methanol production rates
corresponds to an inverse Kinetic Isotope Effect (KIE), rH/rD = 0.36. Moreover, the temporal
scale of exchange was very similar for the formate species and for methanol in the parallel
H2/D2 exchange experiments performed in the FT-IR transmission cell and the test-setup,
respectively (Figure 3) (See Experimental section for details). Together, these experiments
provide firm evidence that the Zr-formate species is a key intermediate in the methanol
formation path.
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
1.0
0.3
0.2
0.1
0.0
H2
HD
H2O
0.8
0.6
0.4
0.2
0.0
HDO
CH3OH
CH4
CH3OH (31) (inverse)
Zr-DCOO
0
1
2
3
4
1.0
0.8
0.6
0.4
0.2
0.0
13CD3OD
13CD4
13CO
13CO2
Kr
Zr-3-OH
Zr-3-OD
Zr-DCOO
0
50
100
150
0
50
100
150
Relative time [min]
Page 14 of 35
ACS Paragon Plus Environment

Page 15 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Figure 3. Top left: The normalized intensity of H2 (m/z = 2), HD (m/z = 3), H2O (m/z = 18), HDO
(m/z = 19), CH3OH (m/z = 31) and CH4 (m/z =15) when switching from CO2+H2 to CO2+D2 at
t = 0. Top right: Comparison of Zr-DCOO and CH3OH (inverse) during H/D exchange. Bottom
left: Normalized absorbance of Zr-μ3OH (open diamonds), Zr-μ3OD (filled diamonds) and
deuterated formate (triangles) during exchange of H2 to D2 at steady state CO2 hydrogenation.
CO2/H2(D2) = 1/6, 10 ml/min, 170 °C, 1 bar. Bottom right: The normalized intensity of Kr (m/z
= 84), 13CO2 (m/z = 45), 13CO (m/z = 29), 13CD4 (m/z = 21) and 13CD3OD (m/z = 35) products
when switching from 13CO2 + D2 to 12CO2 + D2 at t = 0. T = 170 °C, 1 bar, τ = 0.01 gcat∙min/ml.
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
Furthermore, during a complementary (13CO2+D2)/(12CO2+D2) exchange experiment
(Figure 3), 13CD3OD decreased steadily to zero in about 200 minutes, similarly to what
observed for formate and methanol during the H2/D2 switch. The slope of the normalized
intensity of methanol in a semi logarithmic plot (Figure S15) is inversely proportional to the
mean surface residence time (−1/τres) of intermediates leading to the formation of methanol
(Table 3), and is characteristic of formation from a single pool of intermediates.52
Overall, the transient experiments provide firm evidence that formate species, attached
to the Zr nodes in a bidentate configuration, are formed by H transfer from an adjacent Pt NP.
The inverse KIE observed for methanol formation (rH/rD = 0.36) is a signature of reactions
where the rate limiting step involves hydrogen addition to an sp- or sp2-hybridized carbon,
leading to a hybridization change (sp to sp2, or sp2 to sp3).10, 53 CO2 hydrogenation to methanol
via formate species involves two such steps, CO2 hydrogenation to form formate, and formate
hydrogenation to (probably) form dioxymethylene.54 The observation of abundant formate
species by FT-IR further suggest that hydrogenation of the formate species is the rate limiting
step of methanol formation in UiO-67-Pt. Previously, an inverse KIE (albeit not as strong as in
Page 15 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 35
1
2
3
4
5
6
7
8
9
our case) was reported for CO2 to methanol over Cu/ZnO/Al2O3, Cu/SiO2, Cu/MgO and
Pd/SiO2.10 In that case, DFT calculations predicted an inverse KIE for hydrogenation of the
formate species, in line with our results. A schematic presentation of the postulated reaction
mechanism is shown in Scheme 2.
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
Scheme 2. Schematic presentation of the postulated reaction mechanism of CO2
hydrogenation to the formate intermediate in CH3OH formation at the Pt–Zr-node interface.
Integration of the 13C/12C transient response of methanol showed an intermediate
species concentration of 23 µmol/g (Table 3), i.e. close to half the amount of Pt surface species
estimated for the ensemble of Pt NPs identified by TEM (3.6 nm average diameter, Table 1).
The schematic illustration of such a particle embedded in the MOF structure suggests that the
average Pt NP contains 640 surface Pt sites, and is surrounded by 32 Zr nodes that have a
total of 224 Zr-sites accessible for coordination at the Pt  MOF interface, corresponding to
112 bidentate formate species (Scheme 1, Figure 2). These numbers yield a formate:Pt
surface site ratio of 112:640 (0.2), substantially lower than the estimated 24:55 (0.4) ratio from
transient experiments.
Importantly, the numbers imply that additional Zr-sites are available to formate
formation around each Pt NP. In this respect, we hypothesized that Zr-sites might become
accessible by breaking Zr-linker bonds, either during Pt NP formation, or during the catalytic
Page 16 of 35
ACS Paragon Plus Environment

Page 17 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
reaction. The hypotheses were investigated by periodic Density Functional Theory (DFT)
calculations. (See Experimental Details section for a brief description of the methodology and
SI for further computational details and model construction). A Pt89 NP occupying the
tetrahedral cavity of UiO-67 (Figure S18), Pt89tet, was found to be an adequate model, justified
by the assumption that the interface between the Pt NP and the linker/Zr node is similar when
the NP has a diameter of 3.6 nm or ≈ 1 nm.
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
In the perfect MOF structure, all 24 coordination sites of each Zr-node are occupied by
the carboxylate groups of the linker molecules (L). We first investigated the opening of such
sites and computed the free energy profile of the system when a linker decoordinates from the
node and slides along the Pt NP surface (Figure 4 and Figure S22). Decoordination of the
linker presumably occurs via σ(Csp2-Csp2)-bond rotation at the O2C-C(Aryl) fragment, which
X
generally involves low energy barriers. In Figure 4, LX and O2 indicates the coordination mode
of the linker or formate oxygen to the Zr-node (X= number of Zr-O bonds) and * indicates the
atom adsorbed to the Pt89 NP. These calculations showed that it is unfavorable to open two
Zr-sites (ΔG(L2 →L0) = 16 kJ/mol), while opening one Zr-site is favorable (ΔG(L2 →L1) = -54
kJ/mol)), indicating that the interaction with the NP weakens the node-linker Zr-O bonds
favoring the formation of vacant sites. Considering next the catalytic reaction, the first step of
formate formation is the adsorption of a CO2 molecule which was found to coordinate its C
1
atom to the Pt surface and the O atom to the opened Zr-site (ΔGads(L1+ CO2 → L1-C*O2 ) = -20
kJ/mol) (Figure 4). Continuing on the formate formation pathway from the L1-C*O2
1
intermediate, the two subsequent intermediates, L1-C*O2 +2H*, ΔG(L2 + CO2 +H2 → L1-
1
1
1
C*O2 +2H*) = -144 kJ/mol and L1-HCO2 +H*, ΔG(L2 + CO2 +H2 → L1-HCO2 +H*) = -107 kJ/mol,
correspond to H2 adsorption on the Pt NP and a formate + hydride species, respectively (Figure
4). The H2 adsorption is exergonic by -70 kJ/mol, and the subsequent H transfer to the C atom
Page 17 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 35
1
2
3
4
of CO2 is endergonic by 37 kJ/mol. Interestingly, the transformation from L1-HCO2 +H* to L0-
1
5
6
7
2
2
HCO2 +H*, ΔG(L2 + CO2 +H2→ L0-HCO2 +H*) = -29 kJ/mol, where the formate species is
8
9
1
2
coordinated to two Zr-sites is highly endergonic, ΔGr(L1-HCO2 +H* → L0-HCO2 +H*) = 78
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
kJ/mol. The transformation from L1-HCO2 +H* to L0-HCO2 +H* comprises two structural
changes: i) The formate moiety goes from being coordinating to one Zr-site to being
coordinated to two Zr-sites; ii) the linker decoordinates from one Zr-site, opening two Zr-sites.
1
2
To estimate the contribution to the free energy change, ΔG(L1-HCO2 +H* → L0-HCO2 +H*), of
i) and ii), a computational experiment was performed: while keeping the linker in a “completely
open position” (L0), the formate moiety was oriented to its Zr-monodentate configuration
1
2
1
HCO2 . This resulted in a potential energy increase of 45 kJ/mol.
These results indicate that the coordination of formate by one Zr-O bond
(monodentate) is thermodynamically preferred in nodes initially saturated by linkers (L2), while
the coordination of formate by two Zr-O bonds (bidentate) is thermodynamically preferred in
nodes initially having lost some linkers (L0). Therefore, the observation of the latter coordination
mode of formate by FT-IR suggests that these species have been generated in Zr nodes with
empty coordination sites at the start of the reaction, as represented in Scheme 2.
Page 18 of 35
ACS Paragon Plus Environment

Page 19 of 35
Journal of the American Chemical Society
1
2
3
4
5
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
tet
Pt89
L2, 0
1
L1, -54
L0, 16
L1-C*O2 , -74
1
1
2
L1 –C*O2 +2H*,
L1-HCO2 +H*,
L0-HCO2 +H*,
-144
-107
-29
Figure 4. 3D representations of intermediates of the reaction pathway towards the formation
of the formate species coordinated to two Zr sites. Values are free energies in kJ/mol. LX and
O2X indicates the coordination number of the linker and formate to the Zr-node, and * the atom
adsorbed to Pt. O, C and H atoms from CO2 and H2 are in pink, black and green; and for the
linker, in red, grey and white.
Page 19 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 35
1
2
3
4
5
6
It was then explored whether the growth of the Pt NP during activation could cause the
formation of such Zr nodes with open coordination sites. Insight into this possibility was
7
8
9
tet
obtained by computing the free energy of the Pt particle growth from Pt55 to Pt89tet. These
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
calculations showed that the reaction free energy of the system decreases as the Pt NP grows,
and the decrease in energy will eventually exceed the energy loss of linker detachment (Figure
S19, Figure S20 and Figure S21). Thus, during the nanoparticle growth, linkers will detach
from the Zr nodes, and the Zr nodes will decorate the growing Pt nanoparticle. Returning to
Scheme 1, the number of Zr nodes that needs to be removed from a perfect MOF lattice in
order to create the 3.6 nm model Pt NP is minimum 6. The corresponding number of additional
bidentate formate sites is 54. In combination with the 112 formate sites of the surrounding
framework bound nodes, this yields a total number of formate sites of 166 and a formate-to-Pt
surface atom ratio of 0.3, in reasonable agreement with the experimentally observed numbers.
Support for the computational results were found from on-line Mass Spectrometry (MS)
measurements performed during activation in 10% H2/Ar atmosphere at 350 °C. The MS data
revealed traces of phenyl-containing fragments in the effluent gas, suggesting that modulator
and/or linker molecules desorbed from the material during Pt NP formation (Figure S8). The
crystallographic features of the material were unchanged (Figure S2, Figure S5 and Figure S6)
and the BPYDC/BPDC ratio remained constant during subsequent testing (Table S2).
Furthermore, no linker fragments were observed during a second activation of UiO-67-Pt after
testing, in line with the excellent catalyst stability observed under reaction conditions.
Having established the importance of the Zr-nodes and formate intermediates for
methanol formation over UiO-67-Pt, the next issue is whether methanol formation could be
Page 20 of 35
ACS Paragon Plus Environment

Page 21 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
decoupled from CO and CH4 formation, hence, optionally leading to higher methanol
selectivity. To this end, we first turn to classical kinetic experiments.
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
Contact time variation experiments, under otherwise constant conditions, showed that
methanol selectivity is constant when CO2 conversion increases, suggesting that it is a primary
reaction product (Figure 5). Conversely, methane selectivity increases with increasing CO2
conversion, while CO selectivity decreases, suggesting that CH4 is mainly a secondary
product, formed via CO. The latter result is in accordance with our previous studies of an
analogous catalyst at higher temperature.17
100
98
12
9
96
6
94
3
92
0.3
0.2
0.1
6
4
2
0
0.0
0.5
1.0
1.5
0.00 0.01 0.02 0.03 0.04
Conversion [%]
Contact time [gcatmin/ml]
Figure 5. Contact time variation during CO2 hydrogenation to CO (squares), CH4 (circles) and
CH3OH (triangles) at 170 °C, 1 bar, CO2/H2/He = 1/6/3 and τ = 0.004–0.04 gcat∙min/ml. Left:
Selectivity versus conversion. Right: Rate of product formation versus contact time.
Next, partial pressure variation experiments were performed in order to assess reaction
orders for each product. This assessment was complicated by the decreased formation rate of
all products with increasing contact time (Figure 5), indicating strong adsorption of one or
Page 21 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 35
1
2
3
4
5
6
7
8
9
several reaction products, thereby poisoning the active site(s).10, 19 Indeed, when correlating
the rate decrease with the partial pressures of the majority products, CO and H2O, all products
have a reaction order close to negative 1 in p(CO+H2O), CH4 slightly less negative (Figure S10
and Table 2). Based on FT-IR results reported above, showing that CO adsorbed on Pt (2042
cm-1)17 dominates the Pt surface under the respective reaction conditions, CO was assessed
as the main contributor to the inhibition, likely suppressing the coverage in H8 by competitive
adsorption. When taking into account the variable concentration of CO and H2O, positive
reaction orders in pH2 and pCO2 were observed for all products, but with substantial differences
(Table 2).
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
Considering first the majority product, CO, its formation rate depends strongly on pCO2,
but less on pH2 (Table 2). Furthermore, no Kinetic Isotope Effect (KIE) was observed for CO
during the (CO2+H2)/(CO2+D2) transient experiment (Table S5 and Figure S14), showing that
breaking or making of H-H or H-O bonds (which would otherwise lead to a primary KIE)10, 55 is
not rate-determining for CO formation under the conditions studied here. Finally, the number
of surface intermediates leading to CO formation and their mean residence time were
calculated from the 13C/12C transients (Table 3 and Figure 3). The normalized 13CO signal
rapidly decreased to around 0.05 within the first 15 minutes then slowly reached zero in the
following 150 minutes. It is interesting to note that the number of surface intermediates leading
to CO formation represents half of the Pt surface atoms in Pt NPs, estimated from TEM
measurements (Table 1 vs. Table 3). This observation, in combination with the partial coverage
of the Pt NPs by Zr-nodes (and linkers), the inhibiting effect of CO, the high predicted barrier
of CO desorption from Pt,18 and the observation of a positive correlation between facile CO
desorption and rate of CO formation in our previous study,17 strongly suggest CO desorption
as rate-limiting step in the RWGS reaction over UiO-67-Pt.
Page 22 of 35
ACS Paragon Plus Environment

Page 23 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
Table 2. Reaction orders in total pressure (ptotal), and in pH2, pCO2 and p(CO+H2O) (1 bar) for
the rate of conversion (X) and CO, CH4 and CH3OH formation at 170 °C over UiO-67-Pt.
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
pi
X
CO
CH4
CH3OH
Ptotal
0.1
-1.1
0.2
0.9
0
0.3
-0.7
0.9
0.1
1.1
-1
CO+H2Oa
-1.1
0.2
0.9
b
H2
1.7
0.7
b
CO2
aEstimated from contact time variation experiments (Figure 5 and Figure S10). The reaction
orders represent the average of two experiments. bReaction orders when taking into account
variable pCO and pH2O in the reactor.
Turning next to methane formation rate, it depends strongly on pH2 and weakly on pCO2
(Table 2). Considering the high coverage of CO, as well as the presumed indirect formation of
methane via CO, this result is not surprising. In the (CO2+H2)/(CO2 + D2) transient experiment,
an inverse KIE of 0.6, i.e. intermediate between CO (KIE = 1) and methanol (KIE = 0.36), was
observed for methane (Table S5 and Figure S14). Intriguingly, the inverse KIE was installed
within the 15 minutes resolution of the gas analysis, hence, much more rapid than the transient
behavior of methanol (Figure S14). Thus, the rate-determining step of methane formation
involves bonding with hydrogen,55-58 but the rate-determining step is not the same as for
methanol formation. Indeed, the much more rapid transient behavior of methane compared to
formate, disqualifies formate as a significant intermediate to methane formation. This result
implies that, except for hydrogen activation, methane formation is mechanistically decoupled
from methanol formation. The normalized CH4 formation rate (represented by the m/z = 15
Page 23 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 35
1
2
3
4
5
6
7
8
9
signal of CH3) also decreases rapidly to zero (comparable to H2) during the H/D exchange
(Figure 3). Due to their mass overlap with the much more abundant water fragments, the time
evolution of partially exchanged methane/methyl species could not be followed. The transient
behavior of methane during the 13C/12C switch (Figure 3) was markedly different from that
observed in the H/D transient: The normalized 13CD4 formation rate decreased to 0.5 during
the first 7 minutes and then slowly to zero in the following 160 minutes. This distinct shape of
the isotope transient indicates methane formation from two pools in parallel,52, 59 one rapidly
converted to products and the other more slowly. Integration of the transient curve showed that
the number of surface intermediates leading to methane formation is low (Table 3). Moreover,
about 3% of the methane-forming intermediates react fast and is responsible for about 50 %
of the steady-state methane formation rate, while the other 97 % react slowly (Table 3 and
Figure S15).
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
Returning finally to methanol, its formation depends strongly on both reactant partial
pressures, in line with the observation that it is a primary product, formed independently of CO
(Figure 5). The 13C/12C transient experiment shows that methanol is formed from a similar
number of surface intermediates as CO, but their turn-over rate is much slower, hence leading
to the 9 % methanol selectivity observed under the respective conditions (Table 3).
Table 3. Mean surface residence times τres and the number Nad of surface intermediates
leading to the formation of 13CO, 13CD4 and 13CD3OD at 170 °C (1 bar), calculated from
integration of the curves in Figure 3 and the isotope-independent steady-state reaction rates.
INT(CO)
26 ± 3
INT(CD4)
3 ± 0.3
INT(CD3OD)
23 ± 2
Nads (μmol/gcat)
τres (s)
0.5×103
2.1×103
3.8×103
Page 24 of 35
ACS Paragon Plus Environment

Page 25 of 35
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
Considering finally the non-carbon products HD, H2O and HDO, they exhibited slow or
partially slow transient responses (Figure 3). HD has a sharp initial peak with considerable
tailing over the course of the following hour, indicating formation from parallel pools and/or
exchange with hydrogen-containing surface species.60 H2O showed transient characteristics
suggestive of parallel pools, similar to CO and CH4, and the HDO signal increased rapidly to a
maximum within a few minutes then slowly decreased over the course of 2 hours, closely
following the methanol signal. The long surface lifetime of these products is indicative of a
long-lived source of H participating in their formation, presumably also in the formation of
methanol. Interestingly, quantification of the mol H in HD, H2O, HDO and CH3OH yields 1500
μmol H per gram catalyst, which corresponds to about 80 % of the theoretical amount of mol
Zr μ3-OH groups in the sample (HD accounts for 70 %, as reported above).
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
Conclusion
Mechanistic aspects of CO2 hydrogenation over UiO-67-Pt and the role of the UiO-67
framework have been investigated in detail by employment of steady-state and transient kinetic
studies, coupled with operando infrared spectroscopy and DFT modeling.
It was observed that Pt NPs embedded in the MOF structure are responsible for
hydrogen activation, and that formate species are formed at the Zr nodes by reaction between
adsorbed CO2 and hydrogen spill-over from an adjacent Pt NP. These results demonstrate
that the Pt NPs strongly interact with defect Zr nodes during reaction, and hence, that Zr nodes
decorate the surface of the Pt NPs. Formate species are the most abundant intermediates in
the reaction path to methanol, and transient results suggest that formate hydrogenation is the
rate-limiting step of methanol formation. Importantly, the abundance of formate species is
Page 25 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 35
1
2
3
4
5
6
limited by the number of Zr sites made available by linker detachment due to Pt NP growth
during catalyst activation.
7
8
9
CO and methane formation are mechanistically separated from methanol formation,
except for the hydrogen dissociation step. The main route to methane formation is proposed
as CO hydrogenation. Moreover, the presented data are consistent with CO desorption being
the rate limiting step of the reverse water gas shift reaction over UiO-67-Pt.
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
Acknowledgement
E.S.G, A.L., G.K., S.Ø.-Ø., S.B., S.S., K.P.L. and U.O. acknowledge the Research Council of
Norway for financial support (FRINATEK ToppForsk Grant No. 250795 CONFINE). We
further acknowledge Chimet for providing Pt/Al2O3 and Pt/C catalysts.
T.F. acknowledge the Norwegian Metacenter for Computational Science (NOTUR) for
computational resources (projects number nn4654k and nn4683k), Michele Cacella for useful
advice on the methodology, Sri Harsha Pulumati for fruitful discussions, Jingyun Ye and J.
Karl Johnson for help with reproducing their calculations of reference 53.
T.F., E.S. and A.N. acknowledge support by the ’Nordic Consortium for CO2 Conversion’
(NordForsk project No. 85378, site.uit.no/nordco2).
A. N. acknowledge the support from the Research Council of Norway (FRINATEK Grant No.
250044 and Center of Excellence Grant No. 262695).
Author Information
Corresponding Author
*unni.olsbye@kjemi.uio.no
Page 26 of 35
ACS Paragon Plus Environment

Page 27 of 35
Journal of the American Chemical Society
1
2
3
4
Associated Content
5
6
7
Supporting Information
8
9
Catalyst preparation, characterization, catalytic testing, operando FTIR and computational
details.
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
References
1.
Bellotti, D.; Rivarolo, M.; Magistri, L.; Massardo, A. F., Feasibility study of methanol
production plant from hydrogen and captured carbon dioxide. J. CO2 Util. 2017, 21, 132-138.
2.
Chu, S.; Cui, Y.; Liu, N., The path towards sustainable energy. Nat. Mater. 2016, 16,
Olah, G. A., Beyond Oil and Gas: The Methanol Economy. Angew. Chem. Int. Ed 2005,
16.
3.
44 (18), 2636-2639.
4.
Porosoff, M. D.; Yan, B.; Chen, J. G., Catalytic reduction of CO2 by H2 for synthesis
of CO, methanol and hydrocarbons: challenges and opportunities. Energy Environ. Sci. 2016,
9 (1), 62-73.
5.
carbon dioxide. Chem. Soc. Rev. 2011, 40 (7), 3703-3727.
6. Li, W.; Wang, H.; Jiang, X.; Zhu, J.; Liu, Z.; Guo, X.; Song, C., A short review of
Wang, W.; Wang, S.; Ma, X.; Gong, J., Recent advances in catalytic hydrogenation of
recent advances in CO2 hydrogenation to hydrocarbons over heterogeneous catalysts. RSC
Adv. 2018, 8 (14), 7651-7669.
7.
Kattel, S.; Liu, P.; Chen, J. G., Tuning Selectivity of CO2 Hydrogenation Reactions at
the Metal/Oxide Interface. J. Am. Chem. Soc. 2017, 139 (29), 9739-9754.
Page 27 of 35
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 35
1
2
3
4
8.
Wang, X.; Shi, H.; Kwak, J. H.; Szanyi, J., Mechanism of CO2 Hydrogenation on
5
6
7
Pd/Al2O3 Catalysts: Kinetics and Transient DRIFTS-MS Studies. ACS Catal. 2015, 5 (11),
8
9
6337-6349.
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
9.
Wu, H. C.; Chang, Y. C.; Wu, J. H.; Lin, J. H.; Lin, I. K.; Chen, C. S., Methanation of
CO2 and reverse water gas shift reactions on Ni/SiO2 catalysts: the influence of particle size
on selectivity and reaction pathway. Catal. Sci. Technol. 2015, 5 (8), 4154-4163.
10.
Kunkes, E. L.; Studt, F.; Abild-Pedersen, F.; Schlögl, R.; Behrens, M., Hydrogenation
of CO2 to methanol and CO on Cu/ZnO/Al2O3: Is there a common intermediate or not? J.
Catal. 2015, 328, 43-48.
11.
Chen, X.; Su, X.; Duan, H.; Liang, B.; Huang, Y.; Zhang, T., Catalytic performance
of the Pt/TiO2 catalysts in reverse water gas shift reaction: Controlled product selectivity and
a mechanism study. Catal. Today 2017, 281, 312-318.
12.
Hartadi, Y.; Widmann, D.; Behm, R. J., Methanol formation by CO2 hydrogenation on
Au/ZnO catalysts – Effect of total pressure and influence of CO on the reaction characteristics.
J. Catal. 2016, 333, 238-250.
13.
hydrogenation of carbon dioxide to methanol. J. Catal. 2016, 343, 127-132.
14. Román-Martínez, M. C.; Cazorla-Amorós, D.; Linares-Solano, A.; Salinas-Martínez de
Gaikwad, R.; Bansode, A.; Urakawa, A., High-pressure advantages in stoichiometric
Lecea, C., CO2 hydrogenation under pressure on catalysts PtCa/C. Appl. Catal., A 1996,
134 (1), 159-167.
15.
Kim, S. S.; Park, K. H.; Hong, S. C., A study of the selectivity of the reverse water–
gas-shift reaction over Pt/TiO2 catalysts. Fuel Process. Technol. 2013, 108, 47-54.
Page 28 of 35
ACS Paragon Plus Environment

Page 29 of 35
Journal of the American Chemical Society
1
2
3
4
16.
Goguet, A.; Meunier, F. C.; Tibiletti, D.; Breen, J. P.; Burch, R., Spectrokinetic
5
6
7
8
investigation of reverse water-gas-shift reaction intermediates over a Pt/CeO2 catalyst. J.
Phys. Chem. B 2004, 108 (52), 20240-20246.
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
17.
Gutterød, E. S.; Øien-Ødegaard, S.; Bossers, K.; Nieuwelink, A.-E.; Manzoli, M.;
Braglia, L.; Lazzarini, A.; Borfecchia, E.; Ahmadigoltapeh, S.; Bouchevreau, B.; Lønstad-
Bleken, B. T.; Henry, R.; Lamberti, C.; Bordiga, S.; Weckhuysen, B. M.; Lillerud, K. P.;
Olsbye, U., CO2 Hydrogenation over Pt-Containing UiO-67 Zr-MOFs—The Base Case. Ind.
Eng. Chem. Res. 2017, 56 (45), 13206-13218.
18.
Importance of synergy between Pt and oxide support. J. Catal. 2016, 343, 115-126.
19. Kim, S. S.; Lee, H. H.; Hong, S. C., A study on the effect of support's reducibility on
the reverse water-gas shift reaction over Pt catalysts. Appl. Catal., A 2012, 423, 100-107.
20. Tibiletti, D.; Goguet, A.; Meunier, F. C.; Breen, J. P.; Burch, R., On the importance of
Kattel, S.; Yan, B.; Chen, J. G.; Liu, P., CO2 hydrogenation on Pt, Pt/SiO2 and Pt/TiO2:
steady-state isotopic techniques for the investigation of the mechanism of the reverse water-
gas-shift reaction. Chem. Commun. 2004, (14), 1636-1637.
21.
Jacobs, G.; Davis, B. H., Reverse water-gas shift reaction: steady state isotope
switching study of the reverse water-gas shift reaction using in situ DRIFTS and a Pt/ceria
catalyst. Appl. Catal., A 2005, 284 (1), 31-38.
22.
Kalamaras, C. M.; Panagiotopoulou, P.; Kondarides, D. I.; Efstathiou, A. M., Kinetic
and mechanistic studies of the water–gas shift reaction on Pt/TiO2 catalyst. J. Catal. 2009, 264
(2), 117-129.
23.
Burch, R.; Goguet, A.; Meunier, F. C., A critical analysis of the experimental evidence
for and against a formate mechanism for high activity water-gas shift catalysts. Appl. Catal., A
2011, 409-410, 3-12.
Page 29 of 35
ACS Paragon Plus Environment