Ethylene Epoxidation at the Phase Transition of Copper Oxides

Article abstract of DOI:10.1021/jacs.7b05004

Catalytic materials tend to be metastable. When a material becomes metastable close to a thermodynamic phase transition it can exhibit unique catalytic behavior. Using in situ photoemission spectroscopy and online product analysis, we have found that close to the Cu₂O-CuO phase transition there is a boost in activity for a kinetically driven reaction, ethylene epoxidation, giving rise to a 20-fold selectivity enhancement relative to the selectivity observed far from the phase transition. By tuning conditions toward low oxygen chemical potential, this metastable state and the resulting enhanced selectivity can be sustained. Using density functional theory, we find that metastable O precursors to the CuO phase can account for the selectivity enhancements near the phase transition.

Full text of DOI:10.1021/jacs.7b05004

Subscriber access provided by UNIVERSITY OF THE SUNSHINE COAST
Article
Ethylene epoxidation at the phase transition of copper oxides
Mark Greiner, Travis E. Jones, Alexander Yu. Klyushin, Axel Knop-Gericke, and Robert Schloegl
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.7b05004 • Publication Date (Web): 28 Jul 2017
Downloaded from http://pubs.acs.org on July 28, 2017
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 free 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 accessible to all
readers and 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.
Journal of the American Chemical Society 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 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Ethylene epoxidation at the phase transition of copper oxides
*
Mark T. Greiner, Travis E. Jones, Alexander Klyushin, Axel KnopꢀGericke, Robert Schlögl
*
mgreiner@fhiꢀberlin.mpg.de
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
Abstract
Catalytic materials tend to be metaꢀstable. When a material becomes metaꢀstable close to a
thermodynamic phase transition it can exhibit unique catalytic behavior. Using in-situ
photoemission spectroscopy and onꢀline product analysis, we have found that close to the
Cu OꢀCuO phase transition, there is a boost in activity for a kinetically driven reaction–ethylene
2
epoxidation–giving rise to a 20ꢀfold selectivity enhancement relative to the selectivity observed
far from the phase transition. By tuning conditions towards low oxygen chemical potential, this
metaꢀstable state and the resulting enhanced selectivity can be sustained. Using densityꢀ
functional theory we find that the metaꢀstable O precursors to the CuO phase can account for
the selectivity enhancements near the phase transition.
Introduction
Determining the chemical structure of active sites is a key step in understanding how a
heterogeneous catalyst works. It is often found that active sites are highꢀenergy, metaꢀstable
1–3
4
5–7
8
configurations–such as step edges, stacking faults, vacancy defects, latticeꢀstrained sites
9
–11
and surfaces of amorphous materials.
In some cases catalytically active, metaꢀstable
ensembles can only form in-situ by the combined effect of many kinetically and
1
2,13
thermodynamically driven processes.
Unfortunately, the scale and complexity of such
dynamic processes precludes the identities of metaꢀstable sites from being predicted ab-inito
using modern theoretical methods. Contemporary in-situ microscopy and spectroscopy
techniques are used to learn more about such phenomena, identifying metaꢀstable active sites,
understanding how they are formed and how they behave catalytically.
One of the most informative clues about the metaꢀstable nature of active catalysts comes from
observations of catalysts under conditions close to a thermodynamic phase transition. By phase
transition, we are referring to the scenario in which a set of thermodynamic variables (i.e.
temperature or pressure) results in two phases of a material having identical free energies.
When a catalytic material is held in such conditions, small local perturbations in temperature,
chemical potential and concentration can stimulate the surface to flip from one phase to the
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
other with minimal energetic consequence. In a sense, the catalyst becomes thermodynamically
'frustrated'. This bistability gives rise to such catalytic phenomena as spatiotemporal pattern
14
15
16–18
formation, oscillatory reactivity and activity enhancements.
The spatiotemporal pattern formation and reaction oscillations associated with biꢀstable phases
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
were intensively researched in the 1980's and 1990's, where the focus was mainly on reactions
15,18–21
relevant to automotive industry, such as CO oxidation and NO reduction.
In these cases,
the two coexisting phases were surfaces terminated with different adsorbates (e.g. COꢀ and Oꢀ
covered surfaces in the case of CO oxidation). The bistability of the two terminations ensures
that both reactants are present on the surface in high concentrations and in close proximity to
one another.
It remains unclear how influential phase transitions are to general catalysis–e.g. in cases where
oscillations and pattern formation are not observed. If one considers that under reaction
conditions–due to interactions between catalyst and gas phase–generally many surface phases
coꢀexist, and dynamic transitions between them occur continuously. If such phase transitions
are coupled with a step in a catalytic pathway, they can have a large influence on catalytic
performance. Indeed the development of new in-situ characterization methods has provided
22–28
many examples of phase coexistence under reaction conditions,
far fewer studies have so
far as to demonstrate a correlation between phase coexistence or phase transitions and
23,26,29,30
catalytic performance.
Based on the diversity of surface phases with similar free
energies that can be present under reaction conditions and the dynamic nature of catalyst
27
surfaces, it is quite possible that phase transitions play a much larger and more general role in
catalysis than is currently recognized by a large part of the research community. It is important
to point out however, that the notion of 'phase' here must be defined to include not only bulk
phases but also surface phases–such as surface reconstructions and surface terminations.
In the present work, we report on enhanced catalytic activity for a kinetically controlled reaction–
ethylene epoxidation–near the phase transition from Cu O to CuO. In contrast to the rate
2
enhancements near the phase transition in CO oxidation–where the rate enhancement is a
result of a maximum in coꢀadsorption of reactants–we propose that the epoxide selectivity
enhancement near the phase transition on copper oxides is due to the presence of metaꢀstable
adsorbed oxygen species on Cu O that are precursors for CuO formation and form close to the
2
Cu OꢀCuO phase transition. Density functional theory (DFT) calculations indicate that such
2
species offer an alternative reaction pathway to ethylene epoxide that is kinetically favored over
2
ACS Paragon Plus Environment

Page 3 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
the total oxidation pathway. These findings suggest that metaꢀstable surface species formed
near oxide phase transitions could be utilized to enhance catalytic properties.
Results
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 ethylene oxidation reaction several products are possible–namely ethylene epoxide (EO),
acetaldehyde (AcH) and CO , with CO being the most thermodynamically stable product.
2
2
Ethylene epoxide and acetaldehyde, being partial oxidation products, are thermodynamically
less stable than CO , but can be produced preferentially when the path towards partial oxidation
2
products and subsequent desorption is faster than the total oxidation pathway.
It has previously been shown that metallic copper generally oxidizes to CuO when exposed to
31
epoxidation conditions, and exhibits a selectivity to epoxide of only ca. 1%; however, it was
32
also found that during the initial stages of oxidation, a transient selectivity of ca. 28% occurs.
This selectivity enhancement was attributed to the presence of Cu O, forming as an
2
intermediate structure during Cu oxidation, but is not stable under epoxidation conditions.
To begin our investigation, we examined the behavior of Cu O under epoxidation conditions
2
using nearꢀambient pressure (i.e. millibar range) photoemission spectroscopy. Figure 1 shows a
series of in-situ Cu L NEXAFS spectra from Cu O powder measured during a temperature
3
2
ramp (from 50°C to 450°C, 4°C/min) in a 1:1 gas mixture of O :C H at 0.3 mbar total pressure.
2
2
4
One can clearly recognize the transformation from Cu O to CuO by the change in position of the
2
main absorption line seen in Figure 1 (a). At low temperatures, the main line is positioned at
1+
2+
933.56 eV (indicating Cu ). When the temperature reaches 120°C, the first signs of Cu
2+
become apparent, with the formation of a peak at 931.26 eV (indicating Cu ), and by 350°C the
transformation of the surface from Cu O to CuO is complete.
2
In tandem with the in-situ NEXAFS spectra, the reaction products were monitored using gas
chromatography (GC), protonꢀtransferꢀreaction mass spectrometry (PTRꢀMS) and electron
ionization mass spectrometry (EIꢀMS). The selectivity to EO–as measured with GC–is shown in
Figure 1 (b) as well as the corresponding catalyst surface composition–as determined by fitting
the NEXAFS spectra of Figure 1 (a) using reference spectra. From this figure, one can see that
the selectivity to EO is greatly enhanced at the onset of the Cu OꢀCuO phase transition. The
2
maximum in EO selectivity (ca. 21%) is obtained at a temperature of 180°C. At higher
temperatures, the EO selectivity rapidly drops by more than a factor of 10 (to ca. 1.5%). Note
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
selectivity was calculated as [EO]/([EO]+[AcH]+2[CO ]). Epoxide, CO and acetaldehyde were
2
2
the only ethylene oxidation products detected.
The same experiment was performed using CuO as the staring material (shown in Figures S1
and S3). In this case, there was no change to the copper oxidation state during the experiment
and no highꢀselectivity EO production. The only EO production was of lowꢀselectivity (1.5%) at
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
higher temperatures (> 240 °C), similar to what was observed after Cu O was completely
2
oxidized to CuO in the previous results. These findings are consistent with those reported by
32
Jayamurthy et. al.
If conditions are tuned such that Cu O does not transform into CuO at high temperature (i.e. by
2
using lower O partial pressures) then the high selectivity epoxidation reaction is also observed
2
during cooling (see Figure S5). This finding confirms that the high selectivity epoxidation is not
simply due to the burning off of contaminant surface carbon, because such species are
completely absent at high temperature.
Figure 1 - (a) in-situ Cu L
pressure. (b) Ethylene epoxide selectivity measured using gas chromatography during the temperature ramp, with the
corresponding proportions of Cu O and CuO on the surface, as determined from the spectra in (a).
3 2 2 2 4
-edge spectra measured while heating Cu O in a 1:1 mixture of O :C H at 0.3 mbar total
2
While Figure 1 (b) shows that the highꢀselectivity epoxidation reaction is clearly correlated with
the Cu OꢀCuO phase transition, one could argue that this correlation is only coincidental and not
2
causally related to the Cu OꢀCuO phase transition. One could argue, for instance, that Cu O is
2
2
the active phase for epoxidation, and if one could keep the catalyst in a state of Cu O at higher
2
temperatures, the epoxide activity would continue to increase. However, as Figure S5 shows, in
an oxygen poor gas feed of 1:10 O :C H the complete transition of Cu O to CuO can be
2
2
4
2
4
ACS Paragon Plus Environment

Page 5 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
avoided, such that the catalyst remains Cu O at high temperatures. Under these conditions,
2
where the Cu OꢀCuO phase transition is no longer imminent–because oxygen chemical
2
potential decreases with increasing temperature–the catalyst does not exhibit high selectivity
epoxidation, but rather a low selectivity of ca. 1% (similar to CuO). Thus high selectivity on
Cu O is only observed under conditions where the Cu OꢀCuO transition is imminent.
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
2
2
It should be noted that the GC used for these experiments also resolves acetaldehyde.
Interestingly, detectible amounts of acetaldehyde were produced during the high temperature
reaction (with a selectivity for aldehyde of only ca. 1%). During epoxidation on silver–the de
facto standard ethylene epoxidation catalyst–acetaldehyde is never detected because it is
rapidly oxidized to CO before it can desorb from the surface. This observations indicates a
2
distinctly different behavior compared to conventional epoxidation catalysts.
Figure 2 - a) Partial oxidation products and b) selectivity data, measured with PTRMS, during temperature ramps of
2 2 2 4
Cu O (blue lines) and CuO (red lines) in 0.3 mbar of 1:1 O :C H .
In order to investigate the highꢀselectivity epoxidation process further, we repeated the
temperatureꢀprogrammed reaction experiments with and without gasꢀphase oxygen in the
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
reaction feed, using both Cu O and CuO as starting materials. The PTRMS results are
2
presented in Figure 2, representing the production of mass 44 (Note that PTRMS was chosen
due to its extremely high sensitivity to C H O and relative insensitivity to CO due to their very
2
4
2
33
different proton affinities). From Figure 2 (a), it is clear that the boost in EO selectivity near the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
phase transition is due to increased EO production rather than decreased CO production.
2
Secondly, one sees from Figure 2 (b) that when gasꢀphase oxygen is removed from the reaction
feed, Cu O does not produce any detectible EO. This finding suggests that a form of adsorbed
2
oxygen is required for the high selectivity EO reaction to occur. Lastly, comparison of Figures 2
(
c) and (d) shows that CuO produces EO only at high temperatures (> 240°C), with low
selectivity (1ꢀ3 %), and produces EO even in the absence of O . This observation, in addition to
2
the NEXAFS spectra corresponding to the experiment in Figure 2 (d) (see SI)–showing that
CuO is reduced during heating in ethylene–suggests that EO is produced on CuO via a Mars
van Krevelen mechanism.
2 2
Figure 3 - Images of the (a) Cu O(110), (b) Cu O(110) with interstitial oxygen and (c) CuO(110) surfaces.
To understand how adsorbed oxygen species existing under conditions close to the Cu OꢀCuO
2
phase transition might influence ethylene epoxidation activity, we performed a variety of
computational experiments. For these tests, we examined the Cu O(110) surface because it has
2
been predicted from abꢀinitio atomistic thermodynamics to be the thermodynamically most
34
stable Cu O surface under the conditions used in our experiments.
2
A rendering of the Cu O(110) surface is shown in Figure 3 (a). The surface shares a structural
2
similarity with the CuO(110) surface, consisting of rows of Cu ions and a zigzag arrangement of
O ions along the rows of Cu. It is clear from Figure 3 (c) that the CuO(110) surface can be
constructed by filling all the empty zigzag sites (the tetrahedral holes) of the Cu O(110) surface,
2
along with a concomitant distortion to the lattice. To model how the Cu O(110) surface might
2
look prior to the onset of oxidation to CuO, we have placed an additional O ion into a tetrahedral
6
ACS Paragon Plus Environment

Page 7 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
site of the Cu O(110) surface. Such interstitial oxygen ions could potentially form at defect sites
2
such as Oꢀvacancies (as elaborated further in the discussion). The geometry shown in Figure 3
(
b) is the site with the lowest energy Oꢀinterstitial configuration on the Cu O(110) surface. This
2
surface represents Cu O with an excess of oxygen, and recognizing the structural similarity of
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
this interstitial Oꢀion with the stoichiometric CuO(110) surface, one could expect it to represent a
precursor to the nucleation of CuO. It should be noted that the first elementary step in Cu Oꢀ
2
CuO transformation and the fist elementary step in ethylene epoxidation are the same, namely,
O activation. In the geometry of Figure 3 (b), only one interstitial O ion is present (instead of
2
two) because the geometry was formed by relaxing O close to an Oꢀvacancy defect on the
2
Cu O(110) surface.
2
The surfaces shown in Figure 3 (a) and (b)–i.e. Cu O(110) and Cu O(110) with an interstitial
2
2
oxygen defect–were used for subsequent minimumꢀenergy path (MEP) calculations to
determine barriers in the ethylene epoxidation mechanism. The results from these calculations
are summarized in Figure 4. On both surfaces, we calculated the pathway from adsorbed
ethylene to EO and to AcH. For the stoichiometric surface, the O ion used for the reaction was
taken from a normal lattice site, while for the nonꢀstoichiometric surface, the interstitial O ion
was used in the reaction.
2
Figure 4 - (a) calculated reaction pathway for epoxidation on a pristine Cu O(110) surface and (b) a Cu2O(110)
surface containing an O-interstitial ion. The zero energy represents desorbed EO and the respective surface.
Considering first the stoichiometric surface, the energy diagram shows that EO is formed from
ethylene directly (i.e. without passing through the oxametallacycle intermediate typically found
35
on metal surfaces). This result is consistent with previously published calculations that showed
36
a direct pathway to EO on Cu O(100). The barrier to direct epoxidation on the stoichiometric
2
surface was calculated to be ca. 1.7 eV. Due to the strong oxygenꢀcopper bond, breaking the
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
bond and forming EO is endothermic relative to adsorbed ethylene on the stoichiometric
surface. Once formed, EO can either desorb from the surface, revert back to ethylene and
lattice oxygen, or isomerize to AcH. The process of desorption energy is nearly the same as the
barrier of reverting back to ethylene (ca. 0.8 eV). To form AcH it was found that the molecule
must revert back to ethylene, followed by formation of an oxametallacycle, then formation of
AcH. This process has a barrier significantly smaller (by 0.2 eV) than the barrier to formation of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EO from ethylene and lattice O from Cu O. In summary, the stoichiometric surface should
2
preferentially form AcH (or CO ), as opposed to EO, because the barrier to AcH formation is
2
smaller than the barrier to EO formation.
Turning to the Cu O surface containing the Oꢀinterstitial (Figure 4 b), we see that adsorbed EO
2
can be formed by overcoming a barrier of only 0.8 eV. Again, no oxametallacycle intermediate is
involved in the pathway. Interestingly, due the instability of the interstitial O ion, the EO
adsorption energy is only ca. 0.3 eV, indicating that once formed, EO quickly desorbs.
When considering the isomerization of EO to AcH, we find that the reaction passes through an
ethylenedioxy (EDO) intermediate, with a barrier of ca. 1.5 eV, before forming AcH, reminiscent
37
of what has been computed for partially oxidized silver surfaces. Furthermore, no direct path
from ethylene to AcH was found (i.e. the MEP was always EO→EDO→AcH). In summary, these
calculations suggest that an Oꢀrich Cu O surface would have both a higher activity and higher
2
branching ratio to EO through a common intermediate than would the stoichiometric surface.
While it is interesting that epoxidation activity can be enhanced near a phase transition, the
question arises whether a catalyst can be sustained in this metaꢀstable state. To answer this
question, we performed isothermal reactivity and in-situ NEXAFS measurements using various
O :C H ratios. We found that by tuning the oxygenꢀtoꢀethylene ratio in the gas feed it was
2
2
4
possible to hold the catalyst in the state of Cu O indefinitely. Figure 5 shows Cu L spectra and
2
3
GC data from Cu O in a 1:7 mixture of O :C H (0.3 mbar) held at 170°C (i.e. near the highꢀ
2
2
2
4
selectivity reaction maximum) for 10 hours. Under these conditions, the epoxide production and
selectivity reach a steady state, with a selectivity of 10% (note: the selectivity was generally
found to decrease with O partial pressure). The NEXAFS spectra measured during the reaction
2
show that the surface remains primarily Cu O; however, a close examination of the preꢀedge
2
2+
reveals a small concentration of Cu (ca. 0.3 atomic percent) as is evident by the small
2+
shoulder at 931.1 eV in Figure 5 (d). Whether this Cu represents overꢀcoordinated Cu ions in
Cu O or a small amount of CuO on the surface is uncertain.
2
8
ACS Paragon Plus Environment

Page 9 of 19
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
Figure 5 - (a) Cu L
3
2
spectra, (b) epoxide production and (c) epoxide selectivity (measured with GC) of Cu O during a
1
2 2 4
0-hour isothermal reaction in 1:7 O :C H at 170 °C and 0.3 mbar.
It should also be noted that under steadyꢀstate conditions, a small amount of carbonaceous
species form on the surface and contain CꢀC, CꢀO and C=O bonds according to the XPS
spectra (see Figure S6). These species are believed to be adsorbed AcH species that become
strongly bound to the surface, as predicted from the DFT calculations from Figure 4. Note that
the formation of carbon does not increase with time, but reaches a steady state.
Discussion
The results shown in this work demonstrate several important findings: Firstly, Cu O can exhibit
2
lowꢀtemperature highꢀselectivity ethylene epoxidation under conditions in which it is metaꢀ
stable. Complete oxidation to the thermodynamically more stable CuO deactivates this process,
and CuO does not exhibit lowꢀtemperature highꢀselectivity epoxidation activity. The highꢀ
selectivity reaction can be sustained on Cu O if the reaction is carried out under conditions in
2
which the catalyst stays as Cu O but is close enough to the phase transition to enable the
2
formation of the hypothesized adsorbed metaꢀstable species. These conditions are ones where
the oxygen chemical potential in the gas phase is high enough to form the metaꢀstable species
of an Oꢀrich Cu O surface, while the ethylene chemical potential is high enough to prevent
2
complete oxidation of Cu O to CuO. It is possible, based on the observed presence of
2
carbonaceous species at steady state, that strongly bound C H O species slow the Cu O
x
y
z
2
oxidation by blocking sites that would otherwise be susceptible to oxidation.
The inference of an adsorbed oxygen species was made based on the tests with and without O2
in the gas phase. Gas phase O is needed for the lowꢀtemperature highꢀselectivity reaction on
2
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cu O, and in the absence of gas phase O , Cu O is inactive to epoxidation at the corresponding
2
2
2
temperatures (ca. 150ꢀ220°C). Thus we conclude that an adsorbed oxygen species is needed in
the reaction. In contrast, for the high temperature reaction on CuO, no gas phase oxygen is
needed to form epoxide because the source of oxygen is the CuO lattice. This means that
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
without O in the gas phase to reꢀoxidize the CuO it would eventually deactivate via reduction.
2
At high temperatures in the absence of O , Cu O produces CO in a similar manner, where the
2
2
2
oxygen source for forming CO is the lattice oxygen of the Cu O surface.
2
2
Regarding the choice of adsorbed oxygen species to model, we considered several possibilities.
While some oxides can form charged molecular oxygen species like superoxo and peroxo,
38
evidence based on EPR suggests that these intermediate structures are not stable on Cu O.
2
Furthermore, stoichiometric Cu O surfaces have been shown to relatively incapable of O2
2
activation, with activation barriers on the stoichiometric Cu O(111) surface in the range of 1.3
2
39
eV. For O activation to occur, it is believed that defects are required, such as step edges or
2
Oꢀvacancies. The thermodynamic stability of defects in Cu O has been the subject of many
2
40–44
experimental and theoretical studies.
Several native defects are present in significant
concentrations in Cu O. The most dominant defect under most conditions is the neutral Cu
2
vacancy; however, Oꢀvacancies are also relatively stable–second only to Cu vacancies–and
increase in concentration with temperature. At elevated temperatures Oꢀvacancies can even
40
become the dominant defect.
The interstitial structure used here was obtained computationally by relaxing the geometry of O2
in the vicinity of an Oꢀvacancy, indicating a barrierꢀless pathway to formation. A significant
number of Oꢀvacancies will be present on the Cu O surface at elevated temperatures.
2
The proposed interstitial oxygen structure used in this study is both a relatively stable structure,
with an adsorption energy of ꢀ0.5 eV, and structurally analogous to the CuO surface. One would
expect that such species would be in relatively high concentrations under conditions near to the
Cu O→CuO phase transition, and prior to the nucleation of CuO phases. In this structure, the
2
2+
overꢀcoordinated Cu ion would be formally Cu , and one would expect a steadyꢀstate
2+
population of Cu species to be present, whose concentration depends on the balance of their
formation and disappearance rates. There are numerous possible routes to the formation of
such species. As mentioned, O splitting at Oꢀvacancies, or at underꢀcoordinated Cu sites at
2
step edges or kinks are potential routes. Recent atomically resolved in-situ TEM studies have
shown that Cu O grows via terrace growth along step edges, suggesting that O activation may
2
2
10
ACS Paragon Plus Environment

Page 11 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
45
occur at Cu O step edges. An O activation step would be needed for both ethylene
2
2
epoxidation and the Cu OꢀCuO transformation. If activated O species are not removed by
2
ethylene, they will eventually result in the nucleation of CuO. The present data does not enable
us not quantitatively assess the formation and disappearance rate of such species to determine
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
whether the Cu concentration measured at steady state (0.3 atomic percent) is a reasonable
concentration. The NEXAFS spectra at steady state would suggest that 3 out of every 1000 Cu
2+
surface sites are Cu .
Interestingly the proposed activated oxygen species does not behave the same as a terminal
oxygen on the CuO surface, even though they have structural similarities. As shown in Figure
S7, ethylene removes the lattice Oꢀions from CuO, and give rise primarily to the formation of
CO . One can understand this by considering that on a fully oxidized CuO surface, the
2
terminating oxygen species are in a minimum energy configuration, with bond and lattice
distortions at a minimum. In contrast, a single CuOꢀlike site in an otherwise Cu Oꢀlike surface is
2
only metaꢀstable. The interface between such a site and the surrounding lattice results in a
strained bonding environment. This highlyꢀstrained site gives rise to the exothermic epoxidation
step (predicted on the MEP calculations) and accounts for the ease at which epoxide desorbs
from these sites. In stoichiometric CuO, the O ion that gives rise to epoxide is taken from the
surface lattice, as is evident from the reduction of CuO during epoxidation (see Figure S7).
In the recent work by Jayamurthy et. al., where an epoxidation selectivity of 28 % was observed
1+
from supported copper catalysts in a flow reactor, they postulated that the Cu species gives
rise to the highꢀselectivity epoxidation. The high selectivity that was observed was only
transient, and over the course of tens of minutes, it decreased to ca. 1 %. They found that, after
the reaction, the Cu catalyst had oxidized to CuO, and that by reducing the oxidized catalyst
back to Cu they could reproducibly formed the initial structure that generated the higher
selectivity epoxidation. It was concluded that the oxidation state of Cu plays an important role in
the epoxidation reaction, but without in-situ spectroscopy it was uncertain which oxidation state
1+
was correlated to the epoxide activity. It was proposed that a Cu plays an important role in the
epoxidation, and they even postulated that a synergy between oxidation states may play a role.
In the present work, through the use of in-situ spectroscopic methods correlated with inline
product analysis, we can confirm that such a synergy between oxidation states exists, and is
1+
2+
between Cu and Cu . The synergy occurs at the onset of the phase transition between Cu O
2
and CuO. We can confirm that Cu O does not intrinsically exhibit this enhanced epoxide
2
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
selectivity (i.e. when far from the phase transition). This observation is clear from an experiment
where epoxidation was perfromed at high temperature (> 300°C) with an oxygenꢀlean reaction
feed of O :C H = 1:10 (see Figure S5). Under these conditions, the epoxide selectivity peak at
2
2
4
ca. 150ꢀ220 °C, then decreases at higher temperature, while the surface remains Cu O. The
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
selectivity at high temperature is similar to the highꢀtemperature reaction on CuO (ca. 1%). The
explanation for this observation is that, as temperature increases O chemical potential
2
decreases. At elevated temperature and low O2 partial pressure Cu O becomes
2
46
thermodynamically preferred over CuO and thus is not close to a phase transition.
Consequently, the proposed metaꢀstable oxygen species would rapidly recombine and desorb
from the surface. In contrast, near the phase transition, Cu O has the highest oxygen chemical
2
potential it can have without forming CuO. This finding is an indication that the highꢀselectivity
reaction only occurs when Cu O is operated in conditions near to the phase transition, and does
2
not intrinsically exhibit high epoxide selectivity. We propose that the presence of metaꢀstable
oxygen surface species is necessary.
Conclusion
We have found that operating a Cu O catalysts in ethylene oxidation, under conditions at the
2
unset of the Cu OꢀCuO phase transition, gives rise to a large increase in ethylene epoxide
2
activity. Due to this increase in epoxide production, the epoxide selectivity increases by a factor
of twenty (from ca. 1% away from the phase transition to over 20% at the onset of the phase
transition). When fully oxidized to CuO, or when present as Cu O, but far away from the phase
2
transition, catalysts exhibit a ca. 1% epoxide selectivity. In the absence of gas phase O , the
2
epoxidation reaction on Cu O is absent, leading us to conclude that the reaction involves an
2
adsorbed oxygen species that forms on the Cu O surface at the onset of oxidation to CuO. By
2
modeling this surface as a Cu O(110) surface with interstitial O ions (a precursor to the CuO
2
structure), DFTꢀbased minimum energy pathway calculations reveal that such an oxygen
species would give rise in increased epoxidation activity due to its highꢀenergy. These findings
provide new insights into how catalyst metaꢀstability near phase transitions yields valuable
catalytic behavior, and how a kinetically controlled reaction can be enhanced near an oxide
phase transition.
Experimental
All experiments were carried out at the Innovative Station for InꢀSitu Spectroscopy (ISISS)
beamline at the HelmholzꢀZentrum Berlin für Materialien und Energie (HZB) synchrotron light
12
ACS Paragon Plus Environment

Page 13 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
source (BESSY II). The beamline uses a bending magnet and a planeꢀgrating monochromator
10
to provide photons in the energy range of 80ꢀ2000 eV, with a flux of in the range of 6×10
photons/s/100 mA and a spot size at the sample of 100 ꢁm × 80 ꢁm. The end station consists of
a NearꢀAmbient Pressure Xꢀray Photoemission Spectrometer (NAPꢀXPS) produced by SPECS
GmBH. The spectrometer uses a Phoibos 150 analyzer with three differentially pumped stages
separating it from the sample chamber. The pressure in the sample chamber was held constant
using mass flow controllers to flow reactive gases into the chamber and a PIDꢀcontrolled throttle
valve to pump the gases out and maintain constant pressure in the chamber. Samples were
heated from the back side using a focused infrared laser that uniformly illuminates, over an area
of 10 mm × 10 mm, a stainless steel backplate in contact with the sample. Temperature was
monitored using type K thermocouples clamped to the surface of the sample and regulated
using a PIDꢀcontroller connected to the laser power source.
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
XPS spectra were measured with the spectrometer in FixedꢀAnalyzer Transmission (FAT)
mode. Photon energies were always tuned to keep the photoelectron kinetic energies at 150 eV,
making the technique very surface sensitive (λ ~ 6 Å in Cu O). NEXAFS spectra were either
2
measured in TotalꢀElectron Yield (TEY) mode or Auger Electron Yield (AEY) mode. TEY mode
has better signal intensity, but cannot be used for the O Kꢀedge when gasꢀphase oxygen is
present due to the large gasꢀphase absorption signal. In TEY mode, the drain current is
measured by detecting the secondary electrons that hit the nozzle in front of the sample surface
(
i.e. the nozzle that leads to the differentially pumped stages and hemispherical analyzer). In
AEY mode, the analyzer is set to collect electrons of a specified kinetic energy that corresponds
to an Auger emission line (in the case of oxygen, we used a kinetic energy of 493.0 eV). In both
TEY and AEY, the signals are recorded while the monochromator is scanned across the desired
absorption edge. The excitation energy scale was calibrated for the Cu Lꢀedge using the
absorption edge of metallic Cu (932.67 eV at the adsorption edge inflection point) and in the
case of the O Kꢀedge it was calibrated using the π* transition of gasꢀphase O (absorption peak
2
at 530.8 eV).
Copper oxide powders were purchased from Sigma Aldrich and were of 99.999% metals purity.
Powders were pressed into 8 mm diameter pellets using a pressing tool. After loading onto
vacuum, the samples underwent a standard preꢀtreatment procedure. They were first heated in
ꢀ2
10 mbar of O to 400°C to remove adsorbed carbon, while monitoring with C 1s XPS peak.
2
The surface composition was then checked using two survey spectra: one using a photon
energy of 1486.7 eV for a broad overview, and one using a photon energy of 355 eV that is very
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
sensitive to low atomic mass impurities such as S, Cl, and Si. In general the samples were
clean after initial oxidation, but if samples contained impurities they would be exchanged for
new samples. Note that sputter cleaning is not a effective technique for cleaning powders that
are to be tested in catalysis because the surface within the pores of the powder that are not
exposed to the ion beam will remain contaminated. The Cu oxidation state of the starting
material was confirmed using Cu Lꢀedge NEXAFS, O Kꢀedge NEXAFS, Cu 2p XPS and O1s
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
XPS. During initial oxidation, it can happen that CuO begins to form on the Cu O surface. This
2
can be removed either by heating to a high temperature in vacuum (> 550°C) or by heating in a
mild reducing agent like ethylene (> 350°C).
Gas analysis methods included gas chromatography (GC), Quadrupole Mass Spectrometry
(
QMS) and ProtonꢀTransfer Reaction Mass Spectrometry (PTRꢀMS). The three methods were
used in parallel for every experiment. All gas analytical methods were attached, using highꢀ
vacuum connections, to the exhaust line of the differentiallyꢀpumped sample chamber. The GC
was a Varian CPꢀ4900 MicroGC with four columns: two silica molecular sieve columns, one
PPQ column and one alumina column. With this equipment we could detect all the partial and
total oxidation products. The PTRꢀMS was from Ionic Analytik and was used to measure C H O
2
4
(
both EO and AcH) with very high sensitivity and without detection of CO . This ability is
2
afforded by the vastly different proton affinities of C H O and CO . The QMS (Prisma from
2
4
2
Pfeiffer GmbH) was used to detect CO with a higher sensitivity than the GC. All detection
2
equipment was calibrated using a fiveꢀpoint calibration curve made by dosing of test gases
containing the analytes of interest in known amounts.
The temperature ramps (example shown in Figure 1) were performed by first cooling the sample
to 50°C, filling the chamber with the desired composition of O and C H up to 0.3 mbar total
2
2
4
pressure, then initiating a programmed temperature ramp from 50°C to 450°C at a rate of
°C/minute. Photoemission spectra (NEXAFS or XPS) and gas product analysis were measured
during the ramps. For temperature ramps on both Cu O and CuO, gas compositions of 1:1, 1:2,
4
2
1
:5, 1:7 and 1:10 (O :C H ) were tested (see SI). Steady state tests were performed only on
2 2 4
Cu O, using a temperature of 170°C, a total pressure of 0.3 mbar, and gas compositions of 1:1,
2
1
:2, 1:4 and 1:7 (O :C H ). The duration of the tests were either 10 hours or until the activity
2 2 4
became deactivated.
Calculations
14
ACS Paragon Plus Environment

Page 15 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
47
DFT calculations were performed with the Quantum ESPRESSO package using ultrasoft
48
pseudopotentials from the PS Library with a kinetic energy (charge density) cutoff of 30 Ry
49
(
300 Ry) and the PBE exchange and correlation potential. We performed calculations with and
49,50
without the dispersion corrections.
Only the dispersion corrected results are reported. A kꢀ
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
point mesh equivalent to (2x4) for the (110) surface unit cell was used. We employed 4ꢀlayers of
Cu O for all slabs, which were separated by 20 Å of vacuum and the bottom 2 layers were held
2
fixed during ionic relaxation. Cell parameters for the 2x2 slab were 12.02 Å and 8.50 Å. MEPs
were computed with the climbing image nudged elastic band algorithm. The paths were
considered to be converged when the force on the climbing image was less than 0.05 eV/Å and
ꢀ3
the energy change dropped below 10 eV.
Acknowledgements
We would like to thank the MaxꢀPlanck Gesellschaft and the Alexander von Humboldt
Foundation for their generous funding. We would also like to thank the MaxꢀPlanck Gesellschaft
and the HelmholtzꢀZentrum Berlin for use of their infrastructure.
Supporting Information
S1 ꢀ GC data of Cu O and CuO in 1:1 O :C H ; S2 ꢀ NEXAFS spectra of Cu O during
2
2
2
4
2
temperature ramp in 1:1 O :C H ; S3 ꢀ NEXAFS spectra of CuO during temperature ramp in 1:1
2
2
4
O :C H ; S4 ꢀ NEXAFS spectra and PTRMS data of Cu O during several heating cycles in 1:1
2
2
4
2
O :C H ; S5 ꢀ GC data and NEXAFS spectra of heating and cooling ramps of Cu O in diluteꢀO
2
2
2
4
2
epoxidation feed; S6 ꢀ C1s spectra of Cu O during epoxidation reaction; S7 ꢀ NEXAFS spectra
2
measured during heating of CuO in C H .
2
4
References
(1)
2)
(3)
Salmeron, M.; Gale, R. J.; Somorjai, G. A. J. Chem. Phys. 1977, 67 (11), 5324.
Salmerón, M.; Gale, R. J.; Somorjai, G. A. J. Chem. Phys. 1979, 70 (6), 2807.
Zambelli, T.; Wintterlin, J.; Trost, J.; Ertl, G. Science (80). 1996, 273, 1688.
(
(
(
4)
Behrens, M.; Studt, F.; Kasatkin, I.; Kühl, S.; Hävecker, M.; AbildꢀPedersen, F.; Zander,
S.; Girgsdies, F.; Kurr, P.; Kniep, B.ꢀL.; Tovar, M.; Fischer, R. W.; Nørskov, J. K.; Schlögl,
R. Science (80-. ). 2012, 336 (6083), 893.
5)
Polarz, S.; Strunk, J.; Ischenko, V.; Van Den Berg, M. W. E.; Hinrichsen, O.; Muhler, M.;
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Driess, M. Angew. Chemie - Int. Ed. 2006, 45 (18), 2965.
(
6)
Wang, Q.; Biener, J.; Guo, X.ꢀC.; FarfanꢀArribas, E.; Madix, R. J. J. Phys. Chem. B 2003,
107 (110), 11709.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(7)
Lu, G.; Linsebigler, a; Yates, J. T. J. Phys. Chem. 1994, 98 (45), 11733.
(
8)
Amakawa, K.; Sun, L.; Guo, C.; Hävecker, M.; Kube, P.; Wachs, I. E.; Lwin, S.; Frenkel,
A. I.; Patlolla, A.; Hermann, K.; Schlögl, R.; Trunschke, A. Angew. Chemie - Int. Ed. 2013,
52 (51), 13553.
(
(
(
9)
Jia, H.; Stark, J.; Zhou, L. Q.; Ling, C.; Sekito, T.; Markin, Z. RSC Adv. 2012, 2 (29),
0874.
1
10) Reier, T.; Teschner, D.; Lunkenbein, T.; Bergmann, a; Selve, S.; Kraehnert, R.; Schlögl,
R.; Strasser, P. J. Electrochem. Soc. 2014, 161 (9), F876.
11) Pfeifer, V.; Jones, T. E.; Velasco, J. J.; Lez, V.; Massué, C.; Greiner, M. T.; Arrigo, R.;
Teschner, D.; Girgsdies, F.; Scherzer, M.; Allan, J.; Hashagen, M.; Weinberg, G.;
Piccinin, S.; Hä, M.; KnopꢀGericke, A.; Schlö Gl Ac, R. Phys. Chem. Chem. Phys. Phys.
Chem. Chem. Phys 2016, 18 (18), 2292.
(
(
(
12) Schlogl, R. Angew. Chemie - Int. Ed. 2015, 54 (11), 3465.
13) Rocha, T. C. R.; Hävecker, M.; KnopꢀGericke, A.; Schlögl, R. J. Catal. 2014, 312, 12.
14) Nettesheim, S.; von Oertzen, A.; Rotermund, H. H.; Ertl, G. J. Chem. Phys. 1993, 98 (12),
9977.
(
(
(
(
(
(
(
15) Cox, M. P.; Ertl, G.; Imbihl, R.; Rüstig, J.; Rustig, J. Surf. Sci. Lett. 1983, 134 (2), L517.
16) Reuter, K.; Scheffler, M. Phys. Rev. Lett. 2003, 90 (4), 46103.
17) Reuter, K.; Scheffler, M. Phys. Rev. B - Condens. Matter Mater. Phys. 2006, 73 (4), 1.
18) Nibbelke, R. H.; Hoebink, J. H. B. J.; Marin, G. B. Chem. Eng. Sci. 1998, 53 (12), 2195.
19) Imbihl, R. Catal. Today 2005, 105 (2), 206.
20) Imbihl, R. Prog. Surf. Sci. 1993, 44 (3–4), 185.
21) Imbihl, R.; Ertl, G. Chem. Rev. 1995, 95 (3), 697.
16
ACS Paragon Plus Environment

Page 17 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
(
22) Newton, M. A. Chem. Soc. Rev. 2008, 37 (12), 2644.
(23) Gabasch, H.; Hayek, K.; Klotzer, B.; Unterberger, W.; Kleimenov, E.; Teschner, D.;
Zafeiratos, S.; Hävecker, M.; KnopꢀGericke, A.; Schlögl, R.; AszalosꢀKiss, B.; Zemlyanov,
D. J. Phys. Chem. C 2007, 111 (22), 7957.
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
(
24) Rupprechter, G.; Weilach, C. J. Phys. Condens. Matter 2008, 20 (18), 184019.
(25) Piccinin, S.; Zafeiratos, S.; Stampfl, C.; Hansen, T. W.; Hävecker, M.; Teschner, D.;
Bukhtiyarov, V. I.; Girgsdies, F.; KnopꢀGericke, A.; Schlögl, R.; Scheffler, M. Phys. Rev.
Lett. 2010, 104 (3), 35503.
(
26) Newton, M. A.; Jyoti, B.; Dent, A. J.; DiazꢀMoreno, S.; Fiddy, S. G.; Evans, J.
ChemPhysChem 2004, 5 (7), 1056.
(27) Frenken, J.; Groot, I. Research in Heterogeneous Catalysis.
28) Locatelli, A.; Kiskinova, M. Chem. - A Eur. J. 2006, 12 (35), 8890.
(
(29) Kaichev, V. V; Teschner, D.; Saraev, A. A.; Kosolobov, S. S.; Gladky, A. Y.; Prosvirin, I.
P.; Rudina, N. A.; Ayupov, A. B.; Blume, R.; Hävecker, M.; KnopꢀGericke, A.; Schlögl, R.;
Latyshev, A. V; Bukhtiyarov, V. I. J. Catal. 2016, 334, 23.
(
30) Newton, M. A.; Dent, A. J.; DiazꢀMoreno, S.; Fiddy, S. G.; Evans, J. Angew. Chemie - Int.
Ed. 2002, 41, 2587.
(
31) Greiner, M. T.; Jones, T. E.; Johnson, B. E.; Rocha, T. C. R.; Wang, Z. J.; Armbruster,
M.; Willinger, M.; KnopꢀGericke, A.; Schlogl, R. Phys. Chem. Chem. Phys. 2015, 17 (38),
25073.
(
32) Jayamurthy, M.; Hayden, P.; Bhattacharya, A. K. J. Catal. 2014, 309 (0), 309.
(33) Bukhtiyarov, V. I.; Nizovskii, A. I.; Bluhm, H.; Hävecker, M.; Kleimenov, E.; KnopꢀGericke,
A.; Schlögl, R. J. Catal. 2006, 238 (2), 260.
(34) Soon, A.; Todorova, M.; Delley, B.; Stampfl, C. Phys. Rev. B 2007, 75 (12), 125420.
(
35) Kokalj, A.; Gava, P.; de Gironcoli, S.; Baroni, S. J. Catal. 2008, 254 (2), 304.
(36) Ozbek, M. O.; Onal, I.; van Santen, R. A. J. Catal. 2011, 284 (2), 230.
17
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 19
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
(
(
37) Greenley, J.; Mavrikakis, M. J. Phys. Chem. C 2007, 111 (22), 7992.
38) Dyrek, K. Adv. Mol. Relax. Process. 1973, 5 (1–3), 211.
39) Zhang, R.; Liu, H.; Zheng, H.; Ling, L.; Li, Z.; Wang, B. Appl. Surf. Sci. 2011, 257 (11),
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4787.
(
40) Porat, O.; Riess, I. Solid State Ionics 1994, 74 (3–4), 229.
(41) Xue, J.; Diekmann, R. J. Phys. Chem. Solids 1990, 51 (11), 1263.
42) O’Keeffe, M.; Moore, W. J. J. Chem. Phys. 1962, 36 (11), 3009.
(43) Isseroff, L. Y.; Carter, E. A. Chem. Mater. 2013, 25 (3), 253.
44) Raebiger, H.; Lany, S.; Zunger, A. Phys. Rev. B 2007, 76 (4), 45209.
(
(
(45) Zhou, G.; Luo, L.; Li, L.; Ciston, J.; Stach, E. A.; Yang, J. C. Phys. Rev. Lett. 2012, 109
23), 235502.
(
(46) Hallstedt, B.; Risold, D.; Gauckler, L. J. J. Phase Equilibria 1994, 15 (5), 483.
(
47) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.;
Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; Corso, A. D.; Gironcoli, S. d.; Fabris, S.; Fratesi,
G.; Gebauer, R.; Gerstmann, U.; Gougoussis, C.; Kokalj, A.; Lazzeri, M.; MartinꢀSamos,
L.; Marzari, N.; Mauri, F.; Mazzarello, R.; Paolini, S.; Pasquarello, A.; Paulatto, L.;
Sbraccia, C.; Scandolo, S.; Sclauzero, G.; Seitsonen, A. P.; Smogunov, A.; Umari, P.;
Wentzcovitch, R. M.; Paolo, G.; Stefano, B.; Nicola, B.; Matteo, C. C.; Roberto, C.; Carlo,
C.; Davide, C.; Guido, L. C.; Matteo, C. C.; Ismaila, D.; Andrea Dal, C.; Stefano de, G.;
Stefano, F.; Guido, F.; Ralph, G.; Uwe, G.; Christos, G.; Anton, K.; Michele, L.; Layla, M.ꢀ
S.; Nicola, M.; Francesco, M.; Riccardo, M.; Stefano, P.; Alfredo, P.; Lorenzo, P.; Carlo,
S.; Sandro, S.; Gabriele, S.; Ari, P. S.; Alexander, S.; Paolo, U.; Renata, M. W. J. Phys.
Condens. Matter 2009, 21 (39), 395502.
(
(
(
48) Dal Corso, A. Comput. Mater. Sci. 2014, 95, 337.
49) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78 (7), 1396.
50) OteroꢀdeꢀlaꢀRoza, A.; Johnson, E. R. J. Chem. Phys. 2012, 136 (17), 174109.
18
ACS Paragon Plus Environment

Page 19 of 19
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Table of Contents Graphic
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
19
ACS Paragon Plus Environment