Phenylacetylene hydrogenation coupled with benzyl alcohol dehydrogenation over Cu/CeO2: A consideration of Cu oxidation state

Article abstract of DOI:10.1016/j.jcat.2020.11.002

We have examined the effect of copper oxidation state in the continuous gas phase coupled phenylacetylene hydrogenation (to styrene) with benzyl alcohol dehydrogenation (to benzaldehyde) over Cu/CeO₂. Analysis by H₂-TPR, XPS, XRD and

Full text of DOI:10.1016/j.jcat.2020.11.002

Journal Pre-proofs
Phenylacetylene Hydrogenation Coupled with Benzyl Alcohol Dehydrogena-
tion over Cu/CeO2: A Consideration of Cu Oxidation State
Chiara Pischetola, Stephen M. Francis, Federico Grillo, Christopher J.
Baddeley, Fernando Cárdenas-Lizana
PII:
S0021-9517(20)30447-4
https://doi.org/10.1016/j.jcat.2020.11.002
YJCAT 13961
DOI:
Reference:
To appear in:
Journal of Catalysis
Received Date:
Revised Date:
Accepted Date:
25 May 2020
20 October 2020
2 November 2020
Please cite this article as: C. Pischetola, S.M. Francis, F. Grillo, C.J. Baddeley, F. Cárdenas-Lizana,
Phenylacetylene Hydrogenation Coupled with Benzyl Alcohol Dehydrogenation over Cu/CeO2: A Consideration
of Cu Oxidation State, Journal of Catalysis (2020), doi: https://doi.org/10.1016/j.jcat.2020.11.002
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Inc.

Phenylacetylene Hydrogenation Coupled with
Benzyl Alcohol Dehydrogenation over Cu/CeO2:
A Consideration of Cu Oxidation State
Chiara Pischetolaa, Stephen M. Francisb, Federico Grillob,
Christopher J. Baddeleyb and Fernando Cárdenas-Lizanaa,*
aChemical Engineering, School of Engineering and Physical Sciences,
Heriot-Watt University, Edinburgh EH14 4AS, Scotland
bEaStCHEM School of Chemistry, University of St. Andrews,
St. Andrews, Fife KY16 9ST, Scotland
*corresponding author:
tel.: +44(0)131 451 4115, e-mail: F.CardenasLizana@hw.ac.uk
1

Abstract
We have examined the effect of copper oxidation state in the continuous gas phase coupled
phenylacetylene hydrogenation (to styrene) with benzyl alcohol dehydrogenation (to
benzaldehyde) over Cu/CeO2. Analysis by H2-TPR, XPS, XRD and STEM-EDX analyses
demonstrates the generation of a range of Cu oxidation states (Cu0 (13-77%), Cu+ (13-74%),
Cu2+ (0-55%)). The stepwise formation of styrene and ethylbenzene was promoted in the stand-
alone phenylacetylene hydrogenation. An increase in Cu0/Cu+ (from H2-TPR and XPS)
enhanced H2 chemisorption and styrene TOF, but with low hydrogen utilisation efficiency. The
formation of benzaldehyde and toluene was promoted in the stand-alone dehydrogenation of
benzyl alcohol, where benzaldehyde selectivity and TOF correlate with the concentration of
Cu0. Full hydrogen utilisation, exclusive benzaldehyde/styrene formation and a (3-fold) greater
styrene TOF (to attain 100% yield) was achieved in the coupled process, where
hydrogenation/dehydrogenation activity correlates with Cu+/Cu0. This opens new directions
for sustainable "hydrogen free" hydrogenations over non-noble Cu catalysts.
Key
words:
Coupling
process,
phenylacetylene
hydrogenation,
benzyl alcohol dehydrogenation, copper oxidation sate, Cu/CeO2.
2

1 Introduction
Hydrogenation products are an important class of chemicals often produced industrially in
batch systems using pressurised (5-100 bar) non-renewable H2 [1]. Coupling catalytic
dehydrogenation with hydrogenation involves in situ hydrogen generation, transfer and
utilisation [2]. It can be considered as a promising alternative that circumvents the requirement
for external H2 with the possible generation of two high value products. However, catalytic
coupling processes have only been demonstrated for a limited number of molecular
combinations to date, most notably the coupled reduction of aldehydes/ketones (e.g. furfural
[3], acetophenone [4]) with dehydrogenation of aliphatic alcohols (e.g. 2-butanol, 2-propanol
[5]). We are not aware of any report in the literature that involves -C≡C- bond transformation
in tandem dehydrogenation-hydrogenation, the focus of this work. Two active sites are required
for dehydrogenation and hydrogenation in the coupled system. Several catalysts have now been
employed, primarily noble multi-metallic systems including bimetallic (e.g. Co-Ru [5] and Au-Cu
[6]) and alloy catalysts (e.g. NiRuPtAu [7] and PdCu [3]). The use of a non-noble monometallic
catalyst is an attractive consideration in the development of an effective catalyst, given the associated
straightforward synthesis and low cost. However, there is a requirement for a bifunctional behaviour
where the same metal promotes both steps. The metal oxidation state can control catalytic response
[8]. Indeed, while metallic Au and Ag promote hydrogenation (e.g. nitroarenes [9,10]), the electron
deficient counterparts (Auδ+ and Ag+) have been deemed active in the oxidative dehydrogenation of
alcohols (e.g. benzyl alcohols, ethanol [11,12]). The use of monometallic catalysts in coupled
dehydrogenation-hydrogenation is quite limited. Nevertheless, Jae et al. [13] examined the liquid
phase coupled 2-propanol/5-(hydroxymethyl)furfural reaction over Ru/C and identified Ru0 and
Ru4+ as the active species for dehydrogenation and hydrogenation, respectively. In a recent study,
Reddy and co-workers [14] investigated the tandem 2-propanol/hydrocinnamaldehyde process
using a series of oxide (SiO2, Al2O3, MgO and ZrO2) supported Cu catalysts and attributed the
3

differences in catalytic response to variations in copper dispersion and acid-base properties of the
support. Copper promotes -C≡C- bond hydrogenation [15] and alcohol dehydrogenation [16].
The active form of copper in alkyne hydrogenation still remains unknown and there is no general
consensus regarding the copper oxidation state responsible for catalytic activity in alcohol
dehydrogenation, i.e. Cu+ [17] and/or Cu0 [18] have been identified as the active species. The
ultimate copper oxidation state is determined by the Cu precursor [19], synthesis method [19],
activation conditions [20] and nature of the carrier [19]. It is difficult to establish an explicit link
between dehydrogenation/hydrogenation response and Cu oxidation state from the available
literature.
Herein, we have considered the continuous production of styrene, a precursor of
polystyrene with an annual worldwide production of 1 million tonnes per year [21], via
hydrogenation of phenylacetylene coupled with benzyl alcohol dehydrogenation to
benzaldehyde, an important intermediate in the pharmaceutical and cosmetic industries [22]
(Figure 1). The target styrene and benzaldehyde products can be readily separated by liquid-
liquid extraction [23]. The industrial manufacturing of styrene by catalytic dehydrogenation of
ethylbenzene over Fe-K-Cr oxide-based catalysts is inefficient due to (i) the multi-step nature
of the process, (ii) catalyst deactivation and (iii) formation of propylene oxide as a by-product
[24]. Gas phase catalytic hydrogenation of phenylacetylene has emerged as a viable alternative
with a lower associated cost but over-hydrogenation (to ethylbenzene) is difficult to fully
circumvent, notably at conversions >80% [25]. Commercial benzaldehyde production can
involve hydrolysis of benzyl chloride or oxidation of benzyl alcohol using (toxic) chromium
and manganese salts with production of hazardous waste [22]. Gas phase dehydrogenation of
benzyl alcohol in N2 at P = 1 atm offers clear economic and environmental advantages [22],
but an exhaustive search through the literature unearthed only two studies using MgO [22] and
hydrotalcite [26] supported copper catalysts, where toluene was formed as a by-product.
4

Figure 1: Schematic reaction network for coupled process (framed) of benzyl alcohol
dehydrogenation to target (I) benzaldehyde (with undesired C-O hydrogenolysis to (II) toluene)
and phenylacetylene semi-hydrogenation to target (III) styrene (with undesired over-
hydrogenation to (IV) ethylbenzene). Note: Solid arrows denote target reactions, dashed arrows
indicate routes to undesired products while dotted arrows illustrate utilisation of hydrogen
released from benzyl alcohol dehydrogenation in the hydrogenation of phenylacetylene and
carbon-oxygen hydrogenolysis to toluene.
We provide in this work the first direct correlation between dehydrogenation-
hydrogenation response and Cu oxidation state in the gas phase benzyl alcohol
dehydrogenation coupled with phenylacetylene hydrogenation over a series of Cu/CeO2
catalysts. We demonstrate that product distribution can be tuned by changing the Cu oxidation
state and switching from the stand-alone (independent dehydrogenation and hydrogenation) to
the coupled process.
2 Materials and Methods
2.1 Catalyst Preparation and Activation
The CeO2 support and bulk CuO (Sigma-Aldrich, 99.99%) were used as received. Five
CeO2 supported (1 and 15% wt.) Cu catalysts were synthesised by deposition-precipitation or
5

standard impregnation and activated in flowing hydrogen (60 cm3 min-1, 2 K min-1 to 573-773
K; BOC, >99.99%). The metal content, preparation method and final temperature during
hydrogen thermal treatment were selected in order to generate a series of catalysts with
different Cu+/Cu0 molar ratio. The samples synthesised by deposition-precipitation or
impregnation are denoted in this paper by the prefix "DP" and "IM", respectively, followed by
the Cu content (1 or 15% wt.) with the associated activation temperature (573 or 773 K)
specified in parentheses. DP1(573), DP15(573) and DP15(773) were synthesised (at 353 K) by
drop wise addition of a sodium hydroxide solution (2 M, Fisher Scientific, ≥97%) to a
suspension containing the CeO2 carrier (3 g) and the metal precursor (Cu(NO3)2∙3H2O, 40-600
cm3, 1 × 10-2 M, Sigma-Aldrich, 99%), kept under constant (650 rpm) agitation until pH = 10.
The system was aged (353 K, 650 rpm, 4 h) to guarantee homogeneous deposition of Cu2+
species. The precipitate was extracted by filtration, washed (distilled water until pH = 7) and
dried (45 cm3 min-1 N2, 393 K, ca. 12 h, BOC, >99.99%). IM1(773) and IM15(773) synthesis
was based on standard impregnation with aqueous CuCl2 (35-525 cm3, 2 × 10-2 M, 4 g of CeO2,
Sigma-Aldrich, 99.99 %), where the slurry was heated (353 K, 650 rpm) until total evaporation;
drying step as above. The catalyst precursors were sieved (75 μm, ATM fine test sieve),
activated in H2 (step as above) and passivated at room temperature (60 cm3 min-1 1% v/v O2/N2)
for ex situ characterisation.
2.2 Catalyst Characterisation
The copper content was determined by atomic absorption spectroscopy (AAS, Shimadzu
AA-6650 spectrometer, air-acetylene flame) from the extract in 25% v/v HNO3/HCl.
Temperature programmed reduction (H2-TPR)/ desorption (H2-TPD) and hydrogen
chemisorption analysis (at 498 K) were carried out in the commercial CHEM-BET 3000 unit
(Quantachrome Instrument, detector = TCD, software = TPR WinTM). H2-TPR (17 cm3 min-
1, 5% v/v H2/N2, 2 K min-1 to 573-773 K) was employed to determine the extent of Cu and
6

CeO2 reduction, and the amount of surface chlorine [19], i.e. profile curve fitting (software =
CasaXPS 2.3.17) serves to quantify the amount of hydrogen consumed during each transition
step (α = Cu2+ → Cu+, β = Cu+ →Cu0,  = Ce4+→Ce3+ and δ = 2Cl*→2HCl; Cl* indicates
adsorbed chlorine on the catalyst surface) and extract the concentration of copper species
CuT+PR
Cu0 Ce3+ Cl*
TPR ), Ce ( TPR ) and adsorbed chlorine ( TPR ). Post H2-
2+
TPR
3+
Cu
(denoted
,
and
TPR, physisorbed hydrogen was removed by treatment in flowing N2 (65 cm3 min-1, 1.5 h) at
the final activation temperature. Hydrogen consumption during activation was corrected using
H2-TPD (50 cm3 min-1, N2, 50 K min-1 to 1073 K for 1 h; the profiles for the four Cu/CeO2
catalysts are included in Figure S.1 of the Supplementary Information) recorded in parallel
directly after TPR to explicitly determine hydrogen utilised in the reduction of ceria. The
samples were then cooled to room temperature and underwent hydrogen chemisorption by
pulse (10-20 μl) titration. H2 uptake reproducibility was better than ±5% and the values quoted
in this work are the mean; there was no detectable hydrogen uptake on the ceria support alone.
X-ray photoelectron spectroscopy (XPS) measurements were performed on a Scienta
ESCA 300 spectrometer using a monochromatic Al anode (Kα 1486.6 eV, 10 kV, 20 mA).
Before analysis, the Cu/CeO2 catalysts were activated in situ in a pre-chamber under H2
atmosphere (base pressure = 10-8 mbar, H2 partial pressure = 10-6 mbar; 2 K min-1 to 573-773
K) and subsequently subjected to ultra-high vacuum conditions (ca. 10-8 mbar). Full range
survey (0–1100 eV) and high-resolution spectra (over the Cu 2p, Ce 3d core levels and Cu
LMM Auger region) were collected; analyser pass energy = 150 eV. The Ce 3d uꞌꞌꞌ peak was
calibrated at binding energy (B.E.) = 917 eV and used as an internal standard to compensate
for charging effects [27]. Spectral curve fitting and quantification used the CasaXPS software,
employing relative sensitivity factors provided by Scienta. The distribution of Cu and Ce
oxidation states was also derived from XPS curve fitting by fixing the B.E. for each Cu species
(Cu0 and Cu+ (Cu 2p3/2 B.E. ≤ 933 eV) and Cu2+ (>933 eV)) and Ce peaks (uꞌꞌꞌ peak B.E. = 917
7

eV with the 3d5/2 and 3d3/2 peaks separated by -18.3 ± 0.3 eV) according to reported literature
[27,28], using the Gaussian-Lorentzian curves after subtracting a Shirley background with a
full width at half maximum (fwhm) fixed at 2.6-3.6 eV over Cu 2p3/2 [29] and 2.6-4.9 eV over
Ce 3d [30] regions. The concentration of Cu in different oxidation states derived from XPS
data fit was compared with that determined according to the method described by Jasieniak
and Gerson as reported in [31], i.e.
B 1+ A B

S 
S
2+
XPS
Cu (%) =
×100
(1)
(2)
A + B
A - A B B

S 
S
Cu0XPS + CuX+PS (%) =
×100


A + B
where A and B are the experimental areas of the main and shake-up 2p peak of Cu2+,
respectively, while AS/BS represents the theoretical value for the ratio between the main and
shake-up 2p peaks of pure Cu2+ (i.e. Cu(OH)2 AS/BS = 1.57; CuCl2 = 1.63 [31]). It is important
to note that the values for the distribution of Cu oxidation states extracted from curve fitting of
experimental data and using the method of Jasieniak and Gerson did not exceed 5%, considered
2+
XPS
Cu
Cu+XPS Cu0
and XPS ) are the mean.
acceptable [31]; the values quoted in this study (i.e.
,
XPS data recorded from the Cu LMM region was used to calculate the modified Auger
parameter α', defined as the sum of the binding energy of the Cu 2p3/2 and the kinetic energy
of the Cu LMM peaks [31]. The distribution of copper oxidation states was also examined
using the Wagner plot generated with data extracted from the analysis of the photoelectron and
Auger electron regions. Reference data (i.e. B.E., K.E. and I') extracted from [31,32] for Cu2+,
Cu+ and Cu0 were also included in the plot to facilitate the elucidation of the surface copper
species.
The (powder) X-ray diffractograms (XRD) were collected with a Bruker/Siemens D500
instrument (radiation = Cu Kα, scan rate = 0.02º step-1, range = 5º≤ 2θ ≤ 85). Diffractograms
8

were compared with the JCPDS-ICDD reference standards, i.e. Cu (004-0836), CeO2 (043-
1002) and CeClO (153-4659), for identification. Metal particle morphology was evaluated by
probe corrected scanning transmission electron microscopy (STEM) using a FEI Titan Themis
unit equipped with an energy dispersive X-ray (EDX) detector operated at an accelerating
voltage of 200 kV. The scanned images were collected on a Fischione High Angle Annular
Dark Field or FEI Annular Dark Field detector using the FEI TIA software (version 4.9 SP1)
for data acquisition and analysis. Specimen preparation involved (dry) deposition on a holey
carbon/Au (or Ni) grid (300 Mesh). The surface area-weighted mean copper nanocrystal size
(d) was obtained from a measurement of up to 170 individual metal nanoparticles using the
relationship:
n d3

i
i
i
d 
(3)
n d2

i
i
i
where ni is the number of particles of diameter di. Copper particle size distribution fitting was
performed using Gaussian line shape functions. The identification of elements from EDX
analysis was based on the Gatan DigitalMicrograph software (Bruker database), which
revealed the presence of (i) Cu (signals at 0.93, 8.05 and 8.90 keV) from the supported metal
phase, (ii) Au (2.20, 9.71) or Ni (0.85, 7.48, 8.27) from the grid and (iii) Ce (4.84-5.80) and O
(0.53) from the support.
2.3 Catalyst Testing
Reactions (stand-alone hydrogenation of phenylacetylene in toluene under H2, stand-alone
dehydrogenation of benzyl alcohol under N2 and coupled dehydrogenation/hydrogenation of
benzyl alcohol/phenylacetylene under N2) were carried out in a (continuous) fixed-bed glass
tubular reactor (i.d. = 15 mm, 433-573 K, 1 atm), after (in situ) catalyst activation under
conditions of negligible mass/heat transfer constrains. The organic reactant(s) was(were)
9

vaporised and reached reaction temperature in a preheating zone of (1 mm) borosilicate glass
beads situated above the catalytic bed. The reaction temperature in the catalyst bed was
monitored (continuously) with a thermocouple and kept constant (±1 K) by mixing the
Cu/CeO2 catalyst (1-50 mg) with (75 µm) ground glass. The organic reactant(s) was(were) fed
into the reactor at a fixed calibrated flow rate (2 × 10-2 cm3 min-1) via a glass air-tight syringe
(Hamilton) and Teflon line with a microprocessor-controlled infusion pump (Model 100 kd
Scientific). The stand-alone (uncoupled) reactions were conducted in a co-current flow of N2
with benzyl alcohol (gas hourly space velocity (GHSV) = 3 × 103 h-1, molar copper to reactant
feed rate (n/Fbenzyl alcohol) = 8 × 10-5 – 4 × 10-4 h) and H2 with phenylacetylene (GHSV = 3 × 103
h-1, n/Fphenylacetylene = 2 × 10-4 – 5 × 10-2 h, H2 content 80 times in excess to the stoichiometric
requirement for the formation of styrene). Reactant conversion in the stand-alone reactions was
varied by changing the n/F, altering the amount of catalyst while maintaining the total volume
of the bed constant (884 mm3). The coupled reaction was conducted in N2 (GHSV = 3 × 103
h-1, n/Fphenylacetylene = 9 × 10-5 – 2 × 10-4 h, Alkyne:Alcohol molar ratio = 1-191). The reactor
effluent was condensed in a liquid N2 trap for subsequent analysis, which was made by capillary
gas chromatography (Perkin-Elmer Auto System XL, detector = FID, column = DB-1 J&W
Scientific (length = 50 m, i.d. = 0.33 mm, film thickness = 0.20 μm)). Qualitative analysis for
the presence of HCl was carried out by flowing the gas through a two-necked gas-scrubbing
bottle containing an aqueous NaOH solution (2 × 10-3 M, pH = 8.9) kept under constant
agitation at 300 rpm with independent pH (Corning 440 pH Bench-meter) analysis. Before each
run, the system was flushed with N2. The reactants (benzyl alcohol (Acros organic, 99%) and
phenylacetylene (Sigma-Aldrich, 98%)) and products/solvent (toluene (Sigma-Aldrich, 99.8%),
benzaldehyde (Fischer Scientific, ≥99%), ethylbenzene (Fluka, ≥99%) and styrene (Sigma-
Aldrich, ≥99%)) and the gasses (He, N2, and O2 (BOC, >99.99%)) were used as supplied,
without further purification. Data acquisition and analysis were carried out using the
10

TurboChrom Workstation (version 6.3.2) chromatography data system. Calibration plots (not
presented) were used to calculate reactant/product molar fractions (xi). In a series of
preliminary blank tests, there was negligible conversion by passing (i) each reactant in a stream
of N2 or H2 over the (CeO2) support alone or through the empty reactor or (ii) phenylacetylene
in a stream of N2 over the activated Cu/CeO2 catalysts. Reactant (i) conversion (Xi) is defined
by
Reactant - Reactant




i,out ×100
i,in
(4)
Xi (%) =
Reactant


i,in
and selectivity in terms of product "j" (Sj) is given by
Product


j,out
(5)
Sj (%) =
×100
Reactant - Reactant




i,in
i,out
where the subscripts “in” and “out” indicate "inlet" and "outlet" gas streams, respectively.
Catalytic activity is also expressed as initial production rate (R, molproduct molCu-1 h-1) defined
as:
F × Xi,0  Sj
i
-1
Cu
R (molj mol h-1) =
(6)
n
where Xi,0 represents the initial conversion of reactant “i” determined from time on-stream
measurements as described in detail in the Supplementary Information (Figure S.2)
according to [33]. Turnover frequency (TOF, initial production rate per surface copper, h-1)
was obtained using the metal dispersion (D) derived from microscopy measurements according
to:
R
TOF (h-1) 
(7)
D
Hydrogen utilisation efficiency in the stand-alone hydrogenation vs. coupled process was
assessed by:
11

H2 supplied
H2 utilisation efficiency 
(8)
H2 consumed
where "H2 supplied" is the molar hydrogen provided, either (i) via alcohol dehydrogenation or
(ii) from an external gas cylinder supply, and "H2 consumed" indicates the amount utilised in
phenylacetylene conversion corrected with the surface hydrogen post-TPR from H2-TPD; H2
utilisation values >> 1 indicate low efficiency. Carbon mass balance and data reproducibility
was better than ±10%.
3 Results and Discussion
3.1 Catalyst Characterisation
3.1.1 H2-TPR/XPS/H2 Chemisorption Analyses
Catalytic performance in dehydrogenation/hydrogenation reactions is dependent on the
concentration of surface active sites. In the case of Cu/CeO2, this is linked to the Cu0 and Cu+
concentration at the surface. In addition, the metal hydrogen uptake capacity is of practical
importance in hydrogenation applications [34]. Hydrogen TPR measurements were conducted
to examine the reducibility of the copper phase. The surface composition and
chemical/electronic state of copper in the activated catalysts was determined by XPS analysis.
The hydrogen uptake capacity of the Cu/CeO2 catalysts was measured by hydrogen titration at
reaction temperature (= 498 K).
The H2 TPR profiles for the DP1 (Ia), DP15 (Ib), IM1 (IIa) and IM15 (IIb) catalysts are
shown in Figure 2; hydrogen consumption and distribution of Cu oxidation states derived from
curve fitting the H2-TPR response is presented in Table 1. Each profile exhibits a broad positive
peak (H2 consumption) over the temperature range ca. 400-500 K (DP samples) and ca. 600-
670 K (IM), which are close to the range of values reported (413-505 K (DP) [35,36] and 523-
648 K (IM) [37,38]) for the conversion of Cu2+ into metallic Cu. A stepwise Cu2+ → Cu+ →
Cu0 sequence has been proposed elsewhere [20] based on H2-TPR measurements where a
12

mixture of the three copper species prevailed at T ≥ 580 K [39]. The volume of hydrogen
consumed exceeded (by a factor up to 2) the amount required for the formation of metallic
copper in all the catalysts, but was significantly lower (up to a 96-fold) than that required for
the reduction of CeO2 → Ce2O3. This result is indicative of additional hydrogen consumption
by (i) surface Cl species generated from the (CuCl2) metal precursor [19] and/or (ii) the CeO2
carrier, that can result in partial reduction, notably at the Cu-support interface [40].
Figure 2: H2-Temperature programmed reduction (H2-TPR) profiles for Cu/CeO2 catalysts
prepared by (I) deposition-precipitation (DP) and (II) impregnation (IM) with (a) 1% wt. and
(b) 15% wt. Cu loading. Note: Raw data are shown as open symbols () while curve fitted and
envelope are represented by solid and dashed lines, respectively. Hydrogen consumption peaks
associated with Cu2+→Cu+, Cu+→Cu0, Ce4+→Ce3+ and 2Cl*→2HCl transitions are
represented by α, β, γ and δ, respectively.
13

Table 1. Physico-chemical properties of Cu/CeO2 catalysts.
DP15
DP1(573)
IM1(773)
IM15(773)
(573)
(773)
Metal loading (%wt.)
1
15
1
15
Cu2+
(-)
(50)
(-)
(20)
(
TPR , %)
Cu+
80a / 79b (74)
655a / 1180b (14)
73a / 78b (23)
1049a / 1180b (66)
α ( TPR , %)
Cu0
H2-TPR
21a / 79b (26)
470a / 1180b (36)
1547a / 9877b (15)
56a / 78b (77)
190a / 1180b (14)
139a / 9877b (1)
β ( TPR , %)
H2 consumed
(μmol g-1)
Ce3+
106a/ 11504b (1)
180a / 11504b (2)
γ ( TPR , %)
Cl*
-
-
-
1060a / 2360b (55)
935 (25)
δ ( TPR , %)
2+
XPS
Cu
(%)
(%)
(%)
-
934 (55)
933 (13)
932 (32)
934 (19)
-
Cu+XPS
Cu0XPS
XPS
B.E. (eV)
933 (74)
930 (26)
932 (51)
930 (30)
933 (23)
931 (77)
932 (62)
930 (13)
d (nm)c
D (%)d
2
7
8
4
13
8
50
15
12
25
aexperimental value; btheoretical amount of hydrogen required for transition step; cmean particle size from STEM analysis (see Eqn. 3); dcopper dispersion
from STEM analysis. Note: Hydrogen consumption peaks during TPR associated with Cu2+→Cu+, Cu+→Cu0, Ce4+→Ce3+ and 2Cl*→2HCl transitions are
represented by α, β, γ and δ, respectively, where * denotes adsorbed Cl.
14

The H2-TPR profile for the DP1(573) sample (Figure (Ia)) presents, after deconvolution,
a main peak at Tmax = 446 K, corresponding to the reduction of Cu2+ → Cu+, the hydrogen
uptake associated with this signal matched (to ±1%, Table 1) the amount required for the
α transition. A peak at higher Tmax (= 454 K) and lower intensity can be related to a partial
reduction to metallic copper (i.e. Cu+ → Cu0) [41]. The associated hydrogen consumption was
significantly lower than the amount required for the full  transition (21 vs. 79 mol g-1; Table
1) and is consistent with a mixture of Cu+ (74%) and Cu0 (26%) post H2-TPR. The presence of
Cu+ and Cu0 species after thermal treatment in H2 at 423-673 K was also observed by Dandekar
et al. (for Cu/C) [42] and Kundakovic and Flytzani-Stephanopoulos (for Cu/CeO2) [43]. An
additional positive signal at 461 K can be attributed to partial reduction of the support, i.e.
Ce4+→ Ce3+; , with the occurrence of oxygen vacancies as a result of spillover hydrogen, i.e.
migration of H to the ceria support. The degree of copper reduction appears to be sensitive to
the strength of metal-support interactions [44] where surface defects on the carrier can stabilise
transition metal nanoparticles, rendering them less susceptible to reduction [43]. While the
presence of Cu+ post H2-TPR can be attributed to strong copper interaction with CeO2 oxygen
vacancies, there is also the possibility of a stabilisation of Cu+ species via the redox equilibrium
Ce4+ + Cu0 ↔ Ce3+ + Cu+ [45]. The H2-TPR profile for the DP15(573) catalyst (Figure (Ib))
is very similar to that recorded for DP1(573) (i.e. peaks at Tmax 442 ± 4 K, 451 ± 3 K and 464
± 3 K), albeit a far greater difference between the experimental and theoretical hydrogen
requirements for the (, , ) transition steps, which results in a modified distribution of copper
and Ce3+ species with different oxidation states (Cu2+ = 50%, Cu+ = 14, Cu0 = 36, Ce3+ = 15;
Table 1). The presence of Cu2+ and Cu+ following activation in hydrogen at T ≥ 573 K has
been previously demonstrated by CO chemisorption [42], H2-TPR [43] and XPS measurements
[45] for oxide- (CeO2 and ZrO2) and carbon-supported Cu catalysts. The generation of larger
copper nanoparticles on DP15(573) (vs. DP1(573)) post H2-TPR is likely due to the higher
15

metal loading [46]. The lower metal dispersion in DP15(573) leads to (i) inhibition of hydrogen
diffusion in the Cu cluster and (ii) weak metal-support interactions linked to the small area of
contact at the interface between the metal nanocrystals and the CeO2 carrier which, in turn,
hinders Cu2+ reduction and lowers the concentration of Cu+. The H2-TPR profiles measured for
the IM samples (Figure 2(II)) also exhibited three main peaks at T ≤ 658 K but they were
shifted to a significantly higher temperature (by ~185 K) relative to the DP catalysts. This
suggests variations in metal-support interactions that can be linked to the different (i) metal
precursor (Cu(NO3)2∙3H2O vs. CuCl2) and (ii) preparation method (DP vs. IM). A surface Cl
content > 1 ppb (>1 ppm for 1-4% wt. supported copper catalysts prepared via CuCl2
impregnation [19]) has been demonstrated to facilitate Cu surface mobility with the formation
of larger metal particles [47]. As the IM samples were prepared using CuCl2, it is to be expected
that these catalysts will bear larger metal nanoparticles, interacting weakly with the support
and requiring higher reduction temperatures. Chen et al. [48], studying the hydrogen TPR of
several Cu/CeO2 samples reported a ca. 50 K increase in reduction temperature for samples
pre-treated in different atmospheres (i.e. air, vacuum, hydrogen) and attributed this to a change
in the electronic state of Cu nanoparticles that is dependent on metal dispersion with a shift to
higher reduction temperature for bigger particles. Moreover, a stronger Cu/support interaction
in the DP catalysts [19] improves the stabilisation of the excess electrons involved in the
reduction step [49], lowering the temperature requirements. The H2-TPR profile for the
IM15(773) (Figure 2(IIb)) exhibits, in addition to the peaks at T ≤ 658 K, signals at Tmax = 670
K, 678 K and 690 K, indicative of a multi-modal copper size distribution. The two H2-TPR
signals at 670 and 678 K can be tentatively attributed to α and β transitions, respectively, during
the reduction of larger Cu nanoparticles [50]. The H2-TPR signal at 690 K can be associated
with the generation of HCl from CuCl2 (i.e. δ transition) linked to (i) Cl bonded to Cu and/or
(ii) oxychloride species on the support. A similar rationale has been provided by Bond and co-
16

workers [19] for the TPR hydrogen consumption peak recorded at Tmax = 700 K in their study
of Cu/SiO2 and Cu/Al2O3 prepared with CuCl2 as metal precursor. The results from H2-TPR
analysis demonstrate the generation of supported Cu particles with oxidation states in the range
Cu0 = 14-77%, Cu+ = 14-74% and Cu2+ = 0-50% (Table 1). The 1% wt. Cu/CeO2 samples
exhibit full reduction of Cu2+, whereas this step is hampered in the 15% wt. catalysts, which
we attribute to inhibited hydrogen diffusion. Weak metal support interactions in IM (vs. DP)
catalysts with increasing metal content (15% vs. 1% wt.) lowers Cu+ content. IM15(773)
deviates somehow from this trend, where stabilisation via Cl- complexation [19] results in a
high concentration of copper (I).
The H2-TPR measurements provide critical information on bulk characteristics which can
differ greatly from surface properties [51]. The distribution of Cu oxidation states on the
catalyst surface can be extracted from data fitting in the Cu 2p core level. However, the results
can be compromised by the overlapping of copper species (i.e. Cu+ and Cu0) [52]. It has been
demonstrated [52], that analysis of X-ray induced Auger transitions (i.e. electrons excited as a
consequence of the relaxation process following the ejection of an electron from a core orbital)
can serve as an additional measurement to investigate copper on the catalyst surface, where the
spectra of Cu+ and Cu0 exhibit distinct peak(s) and shape. Surface copper and Ce3+ was
examined by XPS analysis of the Cu 2p3/2 (Figure 3(a)), Cu LMM Auger (Figure 4) and Ce
3d regions (Figure 3(b)); the values derived from curve fitting of the XPS measurements are
shown in Table 1 and Figure 5(I). XPS analysis of the 1% wt. loaded samples (DP1(573)
(Figure 3(Ia) and IM1(773) (IVa)) confirmed a total reduction of Cu2+ to generate a mixture
of Cu+ and Cu0 species. In contrast, the 15% wt. catalysts (DP15(573) Figure 3(IIa), DP15(773)
(IIIa) and IM15(773) (Va)) exhibit an additional XPS signal at 943 ± 1 eV that can be related
to the presence of Cu2+, i.e. shake-up satellite peak that can be linked to the O 2p → Cu 3d
transition [53].
17

Figure 3: XPS spectra for Cu/CeO2 (I) DP1(573), (II) DP15(573), (III) DP15(773), (IV)
IM1(773) and (V) IM15(773) and CeO2 (VI) fresh (untreated) and (VII) thermally treated in
H2 to 773 K over the (a) Cu 2p3/2 and (b) Ce 3d regions (intensity expressed as normalised
counts). Note: Raw data are showed as open symbol (), curve fitted and envelope are
represented as solid and dashed lines, respectively. The markers included in (b) illustrate the
position of the 3d5/2/3d3/2 levels for Ce4+ (v/u, v"/u" and v'''/u'') and Ce3+ (v0/u0 and v'/u').
A greater Cu2+ content was recorded with increasing metal content (15 vs. 1% wt.), lower
activation temperature (573 compared to 773 K) and for samples prepared by IM (vs. DP). The
18

results in the Cu LMM kinetic energy region (K.E.) 900-930 eV for DP15(573) (Ia) and
IM15(773) (Ib), as representatives, are depicted in Figure 4; Auger spectra of standards Cu2+,
Cu+ and Cu0 are included to illustrate the level of similarity. The Auger spectrum for the
DP15(573) presents a main peak at K.E. = 918.1 eV that is characteristic of Cu0 (K.E. = 918.6
eV, [31]), although the profile shape is similar to that of Cu2+. On the other hand, the main
signal for IM15(773) (915.1 eV) can be ascribed to Cu+ (915.2 eV, [31]) although the response
largely converges with the profile for Cu2+. The results from the analysis in the LMM Auger
region demonstrate that major contributions due to Cu+ and Cu0 for DP15(573) and IM15(773)
catalysts, respectively, can be excluded. This is in agreement with the low concentrations of
Cu+ and Cu0 (i.e. 13%, Table 1) extracted from the curve fitting of the XPS spectra. The
Wagner plot can be used as an additional tool for the elucidation of copper oxidation states on
the catalyst surface [32]. In this approach, the photoelectron binding energy (B.E.) is plotted as
a function of the kinetic energy of the Auger electron (K.E.) and represented by lines with
slopes of -1 intercepting α' (see Materials and Methods section) and -3 intercepting I' (i.e.
initial state parameter; values in this work extracted from [32]). The values of K.E. and B.E.
for the DP15(573) and IM15(773) samples in the Wagner plot (Figure 4(II)) were obtained
from the survey spectra (see Figure S.3 in Supplementary Information) and the high-
resolution spectra (Figure 4(I) and 3(a)). The results for DP15(573) suggest the presence of
Cu2+, (i.e. experimental data follows straight line with slope = -1 and αꞌ = 1851.33 eV [31]). In
the case of IM15(773), the points fall in the region characteristic of Cu+ species (i.e. straight
line with slope = -3 and Iꞌ = 3713.7 eV). Moreover, the calculated value of αꞌ = 1852.05 ± 0.05
eV for DP15(573) is close to that characteristic of Cu2+ while that of IM15(773) (αꞌ = 1847.30
,
± 0.10 eV) suggests the presence of Cu+ (1847.51 eV in CuCl [31]).
19

Figure 4: (I) XPS spectra over the Cu Auger LMM region for (a) DP15(573) and (b) IM15(773)
Cu/CeO2 (intensity expressed as normalised counts). (II) Cu 2p3/2-Cu LMM Wagner plot
including data reported in the literature (Cu2+ (✕), Cu+ () Cu0 (), [31]) and experimental
values extracted from survey (open symbols) and high-resolution regions (solid symbols) of
DP1(573) (△), DP15(573) (,●), DP15(773) (◇,◆), IM1(773) (□) and IM15(773) (▽,▼).
Note: Raw data in (I) are presented as open symbols () while spectra for standards Cu2+, Cu+,
Cu0 are denoted as solid, dashed and dotted lines, respectively. Solid and dashed-dotted lines
in (II) represent modified Auger parameter (α') and initial state parameter (I'), respectively.
The predominance of Cu2+ and Cu+ in DP15(573) and IM15(773), respectively, confirms the
reliability of the distribution of copper oxidation states extracted from XPS data fitting in the
20

2p3/2 region. The presence of Cu+ (DP1(573) and DP15(773)) and Cu0 (IM1(773)) species was
also proved for the rest of the samples by data inclusion in the Wagner plot, which further
confirms the results from XPS fitting.
Consistent values for both measurements (H2-TPR and XPS) were obtained in the case of
the low loading DP1(573) (i.e. Cu2+ = 0%; Cu+ = 74; Cu0 = 26, Table 1) and IM1(773) (0; 23;
77). However, there are inconsistencies in those obtained with each method for the high loading
samples. We should flag that there is a dearth of published work recording Cu oxidation states
based on more than one analytical technique with XPS [28] or CO chemisorption [42] typically
used. Nonetheless, there are reports in the literature that have considered a combination of XPS,
XANES and EXAFS [54] or H2-TPR and O2-TPO measurements [55]. The discrepancy
observed between H2-TPR and XPS data can be attributed to the type of information obtained
with both methodologies. Indeed, H2-TPR is a "bulk" analytical technique while XPS is
applicable for surface analysis. A source of disagreement can result from the information depth
of the XPS measurements (ca. 3 nm [56]). This suggests copper nanoparticles >4 nm in the
high loading Cu/CeO2 catalysts, in agreement with previous studies [57,58] that have reported
Cu nanocrystals of 8-18 nm for oxide (SiO2, Al2O3) supported copper catalysts with similar
loading (16-20% wt.) prepared by IM and DP. Although the combination of analytical methods
employed in this work is complementary, catalytic reactions occur on the catalyst surface,
consequently the distribution of Cu oxidation states based on XPS measurements was adopted
in subsequent tests.
The Ce 3d profile for the fresh (untreated) CeO2 (Figure 3(VIb)) and hydrogen treated
((VIIb)) are similar and dominated by three spin-orbit doublets (peaks v/u, v"/u" and v'''/u''' at
883 eV/901 eV, 888/907 and 899/917, respectively) that can be associated with the 3d5/2/3d3/2
levels of Ce4+ [30]. Two additional spin-orbit doublets (peaks v0/u0 and v'/u'; 880 eV/899 eV
and 886/904) are also in evidence and can be attributed to the 3d5/2/3d3/2 levels of Ce3+ [30].
21

The extracted surface concentration of Ce3+ (11 ± 1 %) matches the value quoted by Zhang et
al. for bulk CeO2 [30]. The equivalent XPS response was expected considering the higher
(≥823 K) temperature requirement for the Ce4+ → Ce3+ reduction [59]. A comparison of the
XPS spectra of Cu/CeO2 (Figure 3(Ib-Vb)) with those of CeO2 ((VIb-VIIb)) suggests that
incorporation of Cu served to facilitate CeO2 surface reduction with an increase (by up to 28%)
in the concentration of Ce3+. Enhancements of up to 40% in Ce3+ surface content has been
reported post-Pd incorporation [60]. This can be tentatively attributed to spillover hydrogen on
Cu/CeO2 that can participate in a superficial reduction of the oxide carrier, a result that is
consistent with H2-TPR measurements. The correlation between concentration of Ce3+ and
surface metallic copper (CuXPS0, %) is shown in Figure 5(I), where there is a clear linear
increase in Ce3+ density with increasing amount of Cu0 (0 → 77%).
-1
Figure 5: Dependence of (I) surface Ce3+ content (CeXPS3+, %) and (II) H2 uptake (μmol gCu
)
on surface Cu0 concentration (CuXPS0, %) for () DP1(573), () DP15(573), (◆) DP15(773),
() IM1(773) and (▼) IM15(773). Note: The result for pure CeO2 in (I) is represented as a
symbol (). Dashed lines provide a guide to aid visual assessment.
This tendency suggests H2 adsorption at Cu0 sites, and finds agreement with theoretical work
(density functional theory (DFT)) demonstrating that molecular hydrogen does not chemisorb
on Cu2+ or Cu+, but occurs on Cu0 [61] via dissociative adsorption [62]. Hydrogen uptake
22

measured at reaction temperature (498 K) on all the Cu/CeO2 catalysts (1-147 μmol gCu-1) was
similar (1-158 μmol gCu-1) to values reported previously for (Al2O3 and SiO2) supported Cu
catalysts [63,64], but far lower than the uptake for standard transition metal catalysts used in
industrial hydrogenations (e.g. 2023 μmol gPd-1 for Pd/Al2O3 [65]). This is consistent with the
high energy barrier for H2 dissociation due to the filled d band of copper [66]. The hydrogen
chemisorption vs. CuXPS0 linear relationship shown in Figure 5(II), where H2 uptake capacity
increases with increasing Cu0 surface content (13 → 77%), is further evidence for hydrogen
chemisorption on metallic copper.
3.1.2 XRD/STEM-EDX
The morphology of the supported copper nanoparticles was examined by combined XRD
(Figure 6) and STEM-EDX measurements (Figures 7 and 8). All the catalysts present
diffraction peaks at 2θ = 28.5°, 33.1°, 47.5°, 56.3°, 59.1°, 69.4°, 76.7° and 79.1° corresponding
to the (1 1 1), (2 0 0), (2 2 0), (3 1 1), (2 2 2), (4 0 0), (3 3 1) and (4 2 0) planes of CeO2 with a
cubic fluorite-type structure (JCPDS-ICDD reference 043-1002). The XRD patterns for the
high loading Cu samples (DP15(573) (II), DP15(773) (III) and IM15(773) (V)) present
additional signals at 2θ = 43.4°, 50.5° and 74.2° characteristic of (1 1 1), (2 0 0) and (2 2 0)
planes of metallic Cu. The absence of Cu0 diffraction peaks in the 1% wt. Cu/CeO2 catalysis
(i.e. DP1(573) (I) and IM1(773) (IV)) can be linked to the low metal loading and/or the
presence of small (<5 nm) Cu particles [67]. The intensity (I) and width (W) of the main Cu
XRD peak (2θ = 43.4°) is sensitive to copper nanoparticle size where a higher I/W ratio is
indicative of larger metal nanoparticles [67].
23

Figure 6: XRD patterns for activated (I) DP1(573), (II) DP15(573), (III) DP15(773), (IV)
IM1(773) and (V) IM15(773) Cu/CeO2 catalysts. Note: Peak assignment based on JCPDS-
ICDD reference data: () CeO2 (043-1002), () Cu (004-0836) and () CeClO (153-4659).
Taking the three high loading Cu/CeO2 samples, the variation in I/W suggests the following
sequence of increasing copper nanoparticle size: DP15(573) (I/W = 50 counts degree-1)
~ DP15(773) (60) < IM15(773) (440). The XRD diffractogram pattern for IM15(773) exhibits
additional signals at 2θ = 12.9°, 25.4°, 30.9°, 34.1°, 40.8°, 44.3°, 51.7°, 57.1°, 76.9° and 79.1°
24

due to the (0 0 1), (1 0 1), (1 1 0), (1 0 2), (1 1 2), (2 0 0), (2 1 1), (2 1 2), (1 1 5) and (3 1 2)
planes of CeClO. This result is in line with the report of Soria et al. [68], who demonstrated by
electron paramagnetic resonance (EPR) analysis, using oxygen as a probe molecule, the
formation of a cerium oxychloride layer post hydrogen thermal treatment to 773 K on the
surface of Au/CeO2 and Rh/CeO2 prepared by impregnation using a chlorine containing metal
precursor. The absence of detectable XRD signals for CeClO in IM1(773) (IV) can be
attributed to the lower metal and Cl content.
Figure 7: (I) Representative STEM images with (II) associated particle distribution of (a)
DP15(573) and (b) IM15(773) Cu/CeO2 catalysts. Note: Black arrows in the (Ia) STEM image
indicate isolated Cu nanoparticles.
STEM-EDX analysis (see micrographs for DP15(573) (Ia) and IM15(773) (Ib) in Figure
7, as representative samples) have revealed copper nanoparticles at the nanoscale (1-20 nm)
for all the catalysts in this work. High magnification micrographs of individual nanoparticles
(I) with associated EDX spectrum (II) on the surface of DP1(573) (a) and DP15(573) (b), as
representatives, in Figure 8 demonstrate that the Cu component is present as discrete
25

nanoparticles with a pseudo-spherical morphology. Metal particle size distribution is illustrated
by the histograms presented in Figure 7(II); mean Cu diameters (d) are shown in Table 1.
Figure 8: (I) Representative high magnification STEM images with (II) associated EDX
analysis of isolated Cu nanoparticle (dashed frame) on (a) DP1(573) and (b) DP15(573)
Cu/CeO2 catalysts.
The nature of the metal precursor, metal loading and preparation method are critical in
determining the ultimate size distribution [69]. The mean diameter obtained from STEM
analysis for DP1(573) (2 nm) and IM1(773) (4 nm) is consistent with the absence of XRD
peaks characteristic of copper. Taking the same metal content and final activation temperature
(i.e. 15% wt. and 773 K), the Cu/CeO2 samples prepared by impregnation using CuCl2 (vs.
deposition precipitation with Cu(NO3)2∙3H2O) are characterised by a broader distribution of
larger Cu particles, indicative of weaker copper-CeO2 interactions. As expected, the samples
with the higher metal content treated at increased activation temperature bear the larger Cu
nanoparticles, which can be attributed to greater mobility and agglomeration of Cu species that
results in a wider size distribution [69]. This trend is in agreement with the distribution of
copper oxidation states from H2-TPR and XPS measurements. The degree of heterogeneity for
the high loading samples is illustrated by the results for the IM15(773) catalyst (Figure 7(IIb))
26

with an apparent bimodal distribution of smaller (ca. 9 nm) and larger (ca. 14 nm) Cu
nanocrystals, consistent with the four H2-TPR peaks attributed to α and β transitions in Figure
2(IIb).
3.2 Catalytic Response
The characterisation measurements demonstrate the generation of a series of Cu/CeO2
samples with copper oxidation states in the range Cu0 = 13-77%, Cu+ = 13-74% and Cu2+ = 0-
55%. The impact of variations in the copper oxidation state on the tandem dehydrogenation-
hydrogenation of benzyl alcohol with phenylacetylene has been considered. It has been
demonstrated that Cu2+ is inactive in hydrogenation (e.g. isoprene conversion over 1% wt.
Cu/ThO2 [70]) and dehydrogenation reactions (e.g. ethanol transformation using 20% wt.
Cu/Al2O3 [17]). In order to confirm the inactivity of Cu2+ in the catalytic conversion of benzyl
alcohol and phenylacetylene, a (commercial) CuO was purchased and tested in the two
(independent) reactions. CuO exhibited a lack of initial activity in both (dehydrogenation and
hydrogenation) reactions. Based on these results, Cu0 and Cu+ were considered to establish the
impact of metal oxidation state on catalytic performance. We first examine the catalytic
response in the stand-alone phenyl acetylene hydrogenation, using an external hydrogen supply,
and benzyl alcohol dehydrogenation (in N2) over the five Cu/CeO2 catalysts. A compilation of
the studies in the open literature on hydrogenation of phenylacetylene and dehydrogenation of
benzyl alcohol is presented Table S.1 of the Supplementary Information; where pertinent,
our results are related to the trends emerging from the tabulated studies.
3.2.1 Stand-alone Phenylacetylene Hydrogenation
The stand-alone phenylacetylene hydrogenation over all the catalysts generated styrene
and ethylbenzene as the only two products, with no evidence of ring reduction (to
ethylcyclohexane) which has been observed for reaction over Pt/C [71]. The stepwise (-C≡C-
27

→ -C=C- → -C-C- bond) hydrogenation of phenylacetylene to ethylbenzene through styrene
as reaction intermediate is presented in Figure 1 (steps (III) and (IVb)), which also includes a
direct (parallel -C≡C- → -C-C- reduction) pathway (step (IVa)) [72]. At conversions ≤50%,
reaction over all Cu catalysts generated styrene as the main product with secondary generation
of ethylbenzene, consistent with the work of Merrill [73] working under high H2 pressure (10
atm) using Cu/Al2O3. In contrast, under similar reaction conditions full hydrogenation to
ethylbenzene was promoted over Au/Al2O3 [34]. In each case, we recorded a temporal loss of
activity to attain a pseudo-steady state after ca. 2 h (see variation of phenylacetylene conversion
(Xphenylacetyelene) with time on-stream over DP1(573), as a representative, in Figure 9(I)).
Figure 9: Variation of (I) phenylacetylene conversion/styrene molar flow (Xphenylacetylene, %;
Fstyrene, mmol h-1) and (II) benzyl alcohol conversion/ benzaldehyde molar flow (Xbenzyl alcohol, %;
Fbenzaldehyde, mmol h-1) with time on-stream (t, h) in stand-alone () and coupled process (▲)
over DP1(573). (III) Comparison of selectivity to styrene (Sstyrene, %) (open bars) at specific
Xphenylacetyle (grey bars) over DP1(573), DP15(573), DP15(773), IM1(773) and IM15(773). Note:
Dashed grey lines in (I-II) are the result of time scale fitting to obtain the initial Xphenylacetylene
according to the method described in the Supplementary Information section. Reaction
conditions: P = 1 atm, T = 498 K.
Loss of activity in hydrogenation reactions over copper catalysts has been previously
reported and ascribed to carbon deposition [39] and/or strong reactant/product adsorption on
Cu [74]. In order to gain some insight into the reaction mechanism, the dependence of styrene
selectivity on phenylacetylene conversion was investigated. At similar levels of conversion,
28

variations in the Cu0/Cu+ of the five catalysts in this work did not impact on product selectivity
(Xphenylacetylene = 15 ± 1% → Sstyrene = 93 ± 2%, Figure 9(III)). In contrast, product distribution
was sensitive to phenylacetylene conversion (Figure 10(I)).
Figure 10: (I) Styrene selectivity (Sstyrene, %) as a function of phenylacetylene conversion
(Xphenylacetylene, %,) and (II) variation of benzaldehyde selectivity (Sbenzaldehyde, %) with Cu+/Cu0
molar ratio in independent hydrogenation/dehydrogenation (open symbols) and coupled
process (solid symbols) over (,▲) DP1(573), (,✕) DP15(573), (,) DP15(773), (,|)
IM1(773) and (,▼) IM15(773) Cu/CeO2 catalysts. Note: Solid lines provide a guide to aid
visual assessment. Reaction conditions: P = 1 atm, T = 498 K, Xbenzyl alcohol ~6%.
We recorded formation of ethylbenzene in trace quantities (Sethylbenzene ≤5%) at low
phenylacetylene conversions (≤3%) and a shift to preferential ethylbenzene generation at the
expense of styrene with increasing phenylacetylene conversion (Xphenylacetylene >50%). These
results
are
consistent
with
a
predominantly
(stepwise)
phenylacetylene → styrene → ethylbenzene mechanism following paths (III) and (IVb) in
Figure 1, typical for hydrogenation of alkynes in gas phase operation [75]. It is important to
29