This is an open access article published under a Creative Commons Attribution (CC-BY)

License, which permits unrestricted use, distribution and reproduction in any medium,
provided the author and source are cited.

Article

Cite This: Ind. Eng. Chem. Res. 2019, 58, 13021−13029 pubs.acs.org/IECR

Mechanisms of Copper-Based Catalyst Deactivation during CO2

Reduction to Methanol
Anže Prašnikar,*,† Andraž Pavlišic,̌ † Francisco Ruiz-Zepeda,∥ Janez Kovac,̌ § and Blaž Likozar*,†
†
Department of Catalysis and Chemical Reaction Engineering, National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana,
Slovenia
∥
Department of Materials Chemistry, National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana, Slovenia
§
Department of Surface Engineering and Optoelectronics, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia
*
S Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via UNIV TOULOUSE III - PAUL SABATIER on January 9, 2023 at 09:10:55 (UTC).

ABSTRACT: Despite the fact that the methanol synthesis process includes industrially some of the most important catalytic
chemical reactions, it is still not clear how different gaseous species impact catalyst component structure. With the goal to
reduce CO2 emissions through hydrogenation to CH3OH, a higher H2O formation rate than in the production from
compressed CO-rich feed should also be considered. It is known that steam accelerates the sintering of metals, several oxide
compounds, and their interfaces. To determine the effect of moisture on the Cu/ZnO/Al2O3 catalysts, a commercial catalytic
material was systematically aged at various gas compositions and analyzed using transient H2 surface adsorption, N2O pulse
efficient chemisorption, X-ray photoelectron spectroscopy, scanning transmission electron microscopy mapping, X-ray powder
diffraction, and N2 physisorption, and the mechanisms of deactivation were observed. A strong consistent relation between the
compacting of Al2O3, the amount of water in the controlled streamflow, and the activity was found. This connected loss of
support resulted in the (re)forming of Cu, ZnO, and Cu/ZnO phases. Copper particle growth was modeled by applying a
physical coalescence model. In the presence of CO and/or CH3OH, zinc oxide material started to cover the Cu granules, while
H2O promoted the development of separate Cu regions.

1. INTRODUCTION materials (Al2O3, SiO2, MgO, etc.), which act as a support for
Methanol is an important chemical for the production of active components.7−9
various products such as dimethyl ether, formaldehyde, and In addition to the thermal sintering, gas atmosphere can
acetic acid. Besides being a feedstock chemical, methanol can accelerate this process by increasing surface diffusion of the
also be mixed with gasoline and used as a mixture.1 It can be catalyst molecules.10−12 The surface diffusion of hydroxylated
produced to capture CO2 emissions from carbon-rich sources ZnO is faster than that for pure ZnO, resulting in a higher
such as steel plants and thermoelectric power stations. sintering rate.12 The rate of bulk diffusion is also dependent on
Conventionally, methanol is synthesized using syn-gas in a the gaseous environment. For example, Gai et al.13 observed
packed bed reactor with a recycle at 240−280 °C and 50−100 increased diffusion of copper through the alumina layer in the
bar with Cu/ZnO/Al2O3 (CZA) catalyst. presence of oxygen, which they correlated with surface energy
The CZA catalysts are usually used for several years with minimization. Recently it was shown, that ZnO grows over Cu
gradual temperature increase in the reactor to compensate for particles under the reaction conditions implemented during
the activity decrease.2 Catalysts can be deactivated by particle
sintering or sulfur and halogenide poisoning.3,4 Poisoning is Received: April 8, 2019
usually eliminated by sulfur/halogenide removal processes;5,6 Revised: June 28, 2019
however, sintering cannot be simply omitted. The rate of Accepted: June 29, 2019
deactivation is usually limited by the addition of refractory Published: July 1, 2019

© 2019 American Chemical Society 13021 DOI: 10.1021/acs.iecr.9b01898

Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

methanol synthesis, resulting in decreased copper surface

area.14,15
The catalysts for the methanol synthesis were optimized for
the conversion of CO-rich syn-gas, however, the deactivation
process could proceed differently in the case of pure CO2
reduction by H2. This would be due to larger H2O and smaller
CO content, since for every molecule of CO2 an additional
molecule of H2O is formed. It is widely known that steam
increases the sintering rate of oxides and metal particles and Figure 1. Experimental work procedure.
therefore decreases active surface area.9,11,12,16−21
Steam impacts the catalyst activity temporarily (reversibly) equipped with CP-Molsieve and PoraPlot U columns.). A pre-
and permanently (irreversibly). Some existing literature has prepared gas mixture of H2/CO2/N2 = 61/29/10 (TPJ, d.o.o.,
been presented on reversible catalyst deactivation by Jesenice) and pure H2 with purities 99.999% (Messer) were
H2O,22−24 which is caused by occupation of the active sites mixed to obtain gas inlet mixtures. In the case of samples R1
with *OH.24 The rate of methanol formation increases in the and R5, distilled water was pumped with a HPLC pump into
presence of CO, which scavenge surface hydroxyls through the the vaporizer where it was mixed with the other gases. The
water gas shift reaction (WGS). The irreversible deactivation aging conditions and inlet compositions are displayed in Table
was also studied for Cu/ZnO/ZrO/Al2O3 catalysts with steam 1.
being added into the stream,9,16 indicating that H2O increased
the rate of deactivation. The rate of deactivation was reduced Table 1. Catalyst Aging Conditions (Temperature, 240 °C;
by the addition of colloid silica, which additionally stabilized Pressure, 50 bar)
the catalyst structure.9
name inlet aging gas composition H2/CO2/H2O/N2 GHSV [h−1]
However, there was no detailed post-mortem catalyst
characterization to determine the deactivation mechanism R1 62/24.8/5.6/7.6 40 000
with H2O or determination of catalyst morphology changes. R2 65/26/0/8 20 000
For this reason, we performed experiments with the CZA R3 65/26/0/8 1 700
catalyst under different aging compositions at 50 bar and 240 R4 65/26/0/8 40 000
°C for 48 h with the purpose of isolating the effect of H2O. R5 93.1/0/6.9/0 40 000
The samples after and before aging were analyzed with various R6 only reduction
techniques to provide a clear picture of structural changes of R7 100/0/0/0 13 000
catalyst samples. The model of H2O effect on Cu particle R8 no aging
growth is presented. The changes of the catalytic activity were
explained by structural transformations. The X-ray photoelectron spectroscopy (XPS) analyses were
carried out on the PHI-TFA XPS spectrometer produced by
2. EXPERIMENTAL SECTION Physical Electronics Inc. Sample powders were brought from
To determine the effect of the H2O on the methanol catalyst, the catalytic reactor under the protective atmosphere of Ar and
approximately 400 mg of pelletized commercial Cu/ZnO/ introduced into an ultra-high-vacuum spectrometer. The
Al2O3 catalyst HiFuel W230 (particle sizes 200−400 μm) was analyzed area was 0.4 mm in diameter and the analyzed
inserted into the parallel packed bed reactor system and aged depth was about 3−5 nm. Sample surfaces were excited by X-
under different conditions. Aging was performed in the same ray radiation from a monochromatic Al source at a photon
reactor as was used for the catalytic tests. The scheme of energy of 1486.6 eV. The high-energy resolution spectra were
parallel packed bed reactor system and filled reactor tube can acquired with a pass energy of 29 eV and an energy resolution
be found in the Supporting Information, section S6. The of 0.6 eV. Because of the favorable electrical conductivity of the
catalyst contains 50.2 wt % of CuO, 30.8 wt % of ZnO, and graphite phase involved in the samples, XPS spectra were not
18.7 wt % of Al2O3 with the addition of graphite as a binder. aligned since no charging effects were observed. The accuracy
Before being aged, the samples were reduced in H2 at 1 bar and of binding energies was about ±0.3 eV. Quantification of
300 °C for 12 h and evaluated in H2/CO2 = 2.5 at 50 bar, surface composition was performed from XPS peak intensities
180−240 °C and GHSV 40 000 h−1 (long form NmLgas taking into account relative sensitivity factors provided by the
mLcat−1 h−1). Catalytic activity was measured again after instrument manufacturer. Two different XPS measurements
aging (shown only for 240 °C). In both catalytic tests, catalysts were performed on each sample and the average composition
were held at 240 °C for 8 h to ensure that the change in was calculated.
catalyst activity is solely due to the irreversible catalyst changes, Samples for the XRD analyses were powdered and
for example, alterations due to irreversible catalyst morphology transferred in the inert atmosphere on the XRD holder and
and not from adsorbed species remaining following aging. The covered with the Kapton film. The analyses were performed on
samples were transferred and prepared in a glovebox (for the the PANalytical X’Pert Pro instrument. The CuKα1 radiation
X-ray photoelectron spectroscopy (XPS) and X-ray diffraction source was used for the scanning from 10° to 90°. Afterward,
(XRD) analyses) or were passivated using N2O (for the Rietveld refinement of the XRD patterns was performed to
scanning transmission electron microscopy (STEM), N 2 extract the crystal size and phase composition using the Fm3m,
physisorption, and chemisorption analyses). Figure 1 shows P63mc, and Fd3m of Cu, ZnO, and Al2ZnO4 spinel,
experimental procedure. respectively. The instrumental resolution function was
Catalytic tests and aging were performed in a parallel reactor determined by refinement of the XRD pattern of a SiO2
system with online gas composition analysis using gas standard powder. For the fitting procedure, X’Pert HighScore
chromatography (Agilent 490 Micro GC, TCD detectors Plus was used with U and W coefficients for description of
13022 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

Figure 2. Particle sizes of Cu (red) and ZnO (blue) determined using XRD by Rietveld refinement. CO2 conversion (green) is measured after
aging at 240 °C, 50 bar, GHSV of 40 000 h−1.

Table 2. Average Volume Gas Compositions of Aging Gas Mixtures

average aging
volume fraction
[%]
name H2 CO2 H2O CO CH3OH N2
R1 H2/CO2 low conv. + 6%H2O 61.9 23.8 6.1 0.3 0.1 7.8
R2 H2/CO2 med. conv. 63.4 23.2 2.4 0.9 1.1 9
R3 H2/CO2 eq. conv. 59.1 20 6 1.6 4.5 8.8
R4 H2/CO2 low conv. 64 24 1.9 0.6 1 8.5
R6 H2 + 6% H2O 93.1 6.9
R7 only reduction
R8 H2 100
R9 no aging

f whm peaks. Additionally, anisotropic broadening, asymmetry, 3. RESULTS AND DISCUSSION

and peak shape was also considered for all samples. The The Cu and ZnO particle sizes and CO2 conversion after
background of patterns was described by the iterative method different treatments are displayed in Figure 2. Average particle
proposed by E. J. Sonneveld and J. W. Visser.25 sizes were determined with XRD using Rietveld refinement,
STEM was performed in a Jeol ARM CF probe corrected while the CO2 conversions were obtained from the second
microscope operated at 200 kV equipped with a SSD Jeol EDX catalytic tests. Since samples were tested in the parallel reactor
spectrometer. Samples were prepared by sonication in ethanol. system, there are minor differences (<6%) between measured
activities at the same conditions. For this reason, results of the
A small part of the solution was then drop casted on Ni grids
second test were normalized using the first catalytic test to
and observed using a microscope. more accurately compare samples. The powder diffractograms,
N2 physisorption was used to obtain BET specific surface weight fractions of crystal phases and particle sizes can be
areas. Measurements were performed on Micromeritics ASAP found in the Supporting Information, section 1. As mentioned,
2020, with the degassing at 200 °C for 17 h with 100 mg of catalyst activity can be affected by the adsorbed species which
sample. N2O pulse chemisorption and H2-TA (H2 transient remain on the catalyst surface after aging. To rule out this
adsorption) were performed on Micromeritics Autochem II possibility, catalyst R5 was inserted back into the reactor and
2920 with 100 mg of catalyst sample. First, to reduce the Cu2O treated for 5 h with H2 at 50 bar and 240 °C. Then the
catalytic test was repeated under the same conditions. The
layer the samples were purged in H2 at 240 °C for 30 min,
catalytic activity of the treated catalyst converged to the same
followed by N2O pulse chemisorption at 50 °C. The N2O gas value as that before the treatment.
(Messer) decomposition was monitored with a daily calibrated We choose to show the effects based on gas composition
Pfeiffer Vacuum Thermostar mass spectrometer (m/z = 28 and with the average aging gas compositions, since it is the most
30). Following the H2-TA, the formed Cu2O layer was reduced representative way to cumulatively describe conditions along
with 5% H2/Ar (Messer) at 40 °C for 1 h. The H2 uptake was the catalyst bed. The average aging gas compositions (Table 2)
monitored with TCD. Since H2 is also used for the adsorption were calculated based on the linear average estimation of
measured inlet and outlet fractions. For the sample aged at
on the reduced Cu, we purged the sample in He (Messer) at
equilibrium conversion (R3), only the outlet molar fraction
220 °C for 1 h to remove the adsorbed H2 and repeated the was used, since from the additional catalytic tests at 2.4-times
H2-TA on the reduced samples. The copper surface area was higher feed flow (GHSV 4000 1/h) and at the same
calculated from the differences between H2 uptake on N2O temperature and pressure, the composition of the outlet gas
treated and reduced samples. was the same within the margin of error. Additionally, due to
13023 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

reactor geometry (cylindrical tube, bed length 11 mm, reaction products could prevent ZnO particle growth, which
diameter 6.3 mm) and low linear velocity (0.6 mm/s), the is discussed below.
axial dispersion becomes significant, resulting in even more From the N2 physisorption we also obtained average pore
uniform gas composition through the bed length. The amount size and pore volume (Table S3). The significant pore volume
of water added for the aging of the samples R1 and R5 is nearly increase of 22% is observed only when the sample aged at
equal to the H2O formed at the equilibrium conversion of CO2 equilibrium conversion (R3) is compared to the sample R6. As
and H2, without water addition. the pore walls are composed mainly from particles, the average
3.1. General Overview of the Aging Impact on the pore size is also related to the particle size as seen in Figure S4.
Catalyst. Aging to a great extent influences particle sizes. The 3.2. Surface Composition Changes. To indicate surface
average particle size in the sample without aging (R8) was 10.0 composition changes, we calculated the Cu surface fraction
nm for Cu and 6.3 nm for ZnO, while CO2 conversion was and exposed Cu crystallite surface fraction (ECSFCu). The
11%. The comparison with the sample after reduction (R6) copper surface area was determined using N2O pulse
demonstrated that catalytic tests do not significantly affect the chemisorption. The method was validated by H2 transient
particle size as Cu and ZnO particle growth were 18% and adsorption (H2-TA) since the N2O chemisorption is also used
16%, respectively. Aging in H2 (sample R7) also shows only a to measure oxygen vacancies.2 Owing to short reduction time
minor effect on the particle size. It means that the thermal (1 h), the low temperature of reduction (240 °C), and low H2
sintering in the reductive atmosphere is not observed in this partial pressure (0.1 bar), the measured N2O uptake is the
time frame. same as the H2 uptake from H2-TA (in SI, Figure S3). To
On the other hand, the difference became distinct when the additionally confirm the above results, we also measured
steam was added to the hydrogen gas (sample R5). The copper surface area of nonreducible support (Cu on Al2O3)
average Cu and ZnO particle size increased by 51% and 100% with both methods. The Cu surface areas measured by N2O
and CO2 conversion dropped by 15%. H2O is responsible for pulse chemisorption and H2-TA can be found in Table S2.
ZnO hydroxylation which increases the surface diffusion and We calculated the Cu surface fraction by the ratio of the
specific Cu surface area, from the N2O pulse chemisorption,
therefore sintering.19,20 The Cu particle growth might also be
with the BET specific surface area (eq 1):
explained by the hydroxylation of the surface as for Ni on
Al2O321 or more likely due to decreased contact between Cu N2O
SCu
and ZnO and Cu and Al2O3 phases. The contact angle Cu surface fraction[%] =
S BET (1)
between Cu and ZnO increases in more oxidizing
atmospheres,26,27 which could result in decreased metal− Figure 3 shows the dependence of the Cu surface fraction on
support interaction and therefore faster sintering. Additional the CO + MeOH aging fraction. Except for the sample aged at
discussion on the sintering mechanism can be found in section
3.4.
Water promotes ZnO and Cu particle growth but effects
could vary during actual reaction conditions. For this reason,
the catalyst was aged in a mixture of H2/CO2 = 2.5 at different
residence times (GHSV of 40 000 h−1 (R4), 20 000 h−1 (R2),
1700 h−1 (R3)), and with the addition of water (GHSV of
40,000 h−1, R1). R4 and R2 samples at high GHSV both show
minor increase in particle size and a low CO2 conversion
decrease, which is in agreement with the work of Sun et al.2
The addition of water at low CO2 conversion (sample R1)
causes the Cu particle size to increase to a similar size as was Figure 3. Cu fraction on the surface decreases during aging at
observed for the sample without CO2 in the aging mixture equilibrium of the conversion, sample R3. The red triangle represents
(R5). However, ZnO particles increase by 225% compared to sample R8 (no aging). The line is to guide the eye.
the amount in R8 which is doubled compared to the value
obtained from aging without the presence of CO2 (R5). One equilibrium conversion (R3), all samples retain almost an
possible explanation of the synergistic effect of H2O and CO2 equal Cu surface fraction compared to the no aging sample R8
on ZnO growth was given by Varela et al.;12 H2O and CO2 (red triangle), including samples with H2O addition. There-
competitively adsorb on ZnO causing formation of Zn(OH)2 fore, at low CO and CH3OH content the copper surface area
and ZnCO3, respectively. Chemisorption of CO2 would not decreases proportionally with BET specific surface area (as
significantly increase surface diffusion, but rather increase seen in Figure 3). Despite large variations regarding (other
mobility of ions from the interior to the catalysts surface. The samples than R3) moisture content, H2O does not have a
CO2 conversion has decreased about the same amount as in significant effect on this Cu surface fraction. The Cu surface
the case with only water and hydrogen in stream (R5). In the fraction was also estimated with XPS using Cu 2p3/2, Zn 2p3/
case of aging at equilibrium conversion (sample R3), the 2, Al 2p, O 1s, and C 1s peaks. The surface composition was
volume fraction of water in the aging mixture was calculated for all analyzed samples and is given in Table S4.
approximately the same as for the samples R5 and R1 The lowest Cu surface fraction is also evident from the ratio of
(∼6%). The copper particle size and CO2 conversion changed surface coverage [Cu]/[Cu + Zn + Al] obtained from XPS
almost the same as for the samples R5 and R1, while (Figure 4). It should be noted that this is a ratio of fractions in
unexpectedly ZnO particles did not increase as for R1, despite atomic percentages.
similar concentrations of CO2 and H2O in the stream. One possible reason of the lowering of the Cu surface
Accordingly, some other mechanism connected with the fraction is the ZnO covering of the Cu particle.14,15,28 To
13024 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

equilibrium conversion (R3) where it increases. Simulta-

neously, sample R3 also exhibits the lowest ECSFCu and the
highest surface Zn/Cu ratio (Figure 9), indicating the covering
of Cu with ZnO and increasing specific activity.
The change of TOF is not correlated to the Cu particle size
since the sizes of R3 and R5 are nearly the same (14.5 and 15.1
nm, respectively), while the TOF differs by 50% (0.06 and
0.089 1/s, respectively). In the work by van den Berg et al.31 it
was shown that TOF is constant for the particles larger than 8
nm, which are smaller particles than observed in this work.
The same result was obtained by Fichtl et al.16 where TOF
for methanol increased during the first 200 h for the catalyst
Figure 4. Cu surface fraction obtained by XPS depending on the aged at equilibrium conversion. This behavior cannot be
aging of the CO + CH3OH fraction. The red triangle represents explained by the change of Cu particle size, since the specific
sample R8 (no aging). activity (TOF) is almost constant for the Cu particles larger
than 10 nm.31 However, by STEM-EDX (Figure 6) and factor
indicate the Cu overlay, we calculated the exposed Cu ECSFCu (Figure 9) it can be observed that the ZnO covers the
crystallite surface fraction: Cu surface and promotes specific catalyst activity. The activity
N2O
SCu 6LCu increase was also reported for Cu(111) covered by ZnO, with
ECSFCu[%] = XRD
XRD
100; SCu = XRD
the maximum activity at 20% Cu(111) covered by ZnO.32,33
SCu ρCu dCu (2) Fichtl et al.16 found actual increase in the activity for the
catalyst aged at equilibrium conditions at 220 °C in the first 50
where SXRD
Cu represents the surface of Cu crystallites, which is
h. Owing to similar conditions, ZnO could similarly overlay the
calculated using copper crystallite size dXRD
Cu , copper loading
surface of Cu and increase the overall catalyst activity.
(LCu = 0.277 gCu/gcat), and copper density (8.92 g/mL),
assuming spherical crystallite shape. The ECSFCu is the lowest However, in our case we did not observe the increase of
(37%) for sample R3, while the values for other samples are in overall catalytic activity (as seen in Figure 2), but it was
the range of 51 ± 6% (Figure 5). Therefore H2O does not observed in the first 10 h of the stability test at 50 bar (Figure
S11). Therefore, too much of the ZnO overlayer can be
formed, and this can decrease the catalyst activity which was
also indicated by Lunkenbein et al.15 The impact on the
methanol selectivity is insignificant with the minimum
selectivity of 60% for the sample R1 and maximum selectivity
of 62% for sample R4. The TOFCO+MeOH in the experiments
with the water addition (R1 and R5) were nearly the same as
in the case of aging at low conversion. In contrast, Fichtl et
al.16 observed a steep decrease in specific activity when 10% of
the steam was added for 12 h at 220 °C and 60 bar. The
difference could come from the fact that the catalyst was
further aged at equilibrium conditions, resulting in increased
sintering due to loss of Al2O3 support for Cu and ZnO.
Figure 5. Exposed Cu surface fraction (ECSFCu) calculated from the Deactivation by coking is dismissed due to the carbon phase
ratio of specific surface area measured with N2O and theoretical surface fraction ranging between 30% and 36% (Table S6) for
specific surface area of Cu crystallites. The red triangle represents all samples and is in the range of the repeatability of analyses.
sample R8 (no aging).
3.4. Modeling of the H2O Influence on Cu Particle
Growth. Sintering of the catalyst can proceed by several ways.
promote Cu overgrowth, in fact the samples with H2O Herein two of the most probable mechanisms will be
addition (R1 and R5) expose the highest ECSFCu (57% and presented. First is the coalescence mechanism which is caused
55%, respectively). Additionally, the STEM-EDX mapping by the increased particle migration by weaker contact between
(Figure 6) clearly shows that for the sample R3, the ZnO phase the metal and support or increased surface diffusion of catalytic
is much more spread over the Cu phase, as compared to material. The other important mechanism is Ostwald ripening
samples R1 and R6 (see also Supporting Information). which is caused by the surface diffusion of catalytic material
3.3. Catalytic Activity Changes. Catalytic activity is and higher thermodynamic stability of larger particles.16,34
frequently reported to be proportionally correlated to Cu Early in the process of sintering, particles would likely sinter by
surface area;29,30 however, in our case we can see only a weak a coalescence mechanism since they are already in contact,
relation of conversion and metal surface area (Figure 7). while in the end phase of sintering, the particles can undergo
The turnover frequency (TOFCO+MeOH) was calculated to the Ostwald ripening mechanism.35 Fichtl et al.16 modeled the
estimate the effect of aging conditions on the specific activity. catalyst deactivation and found out that the particle migration
The activities from the second catalytic test at 240 °C and 50 model describes the particle size distribution the best. For the
bar were used and normalized to SCu from N2O pulse reasons stated above, and because these deactivation studies
chemisorption. As previously mentioned, only the copper were conducted at short aging times, the coalescence model
surface area without ZnOx oxygen vacancies was measured by was chosen.
N2O pulse chemisorption. As shown in Figure 8, aging does The addition of Al2O3 to Cu/ZnO catalyst strongly
not significantly impact the TOF except for the sample aged at enhances the catalyst stability.7 The state of Al2O3 has
13025 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

Figure 6. AC-STEM ADF images with corresponding EDX signal for Cu (green), Zn (red), and 4Al (blue) of samples R1 (low conversion + H2O
(a,d)), R3 (equilibrium conversion (b,e)) and R6 (after reduction (c,f)).

Figure 7. CO2 conversion after aging as a function of Cu surface area. Figure 9. Sum of CO and MeOH turnover frequency and Zn/Cu
Reaction conditions T = 240 °C, p = 50 bar, GHSV = 40 000 h−1. The ratio determined by XPS as a function of exposed Cu surface fraction.
dashed line represents proportional correlation between the Cu Activity measurement conditions: 240 °C, 50 bar and GHSV of
surface area and CO2 conversion. 40 000 h−1. The dashed lines are guides to the eye.

that small Al2O3 particles uniformly cover Cu and ZnO phase.

These small particles could decrease the mobility of the Cu
and ZnO phases. In all studies with aging (in the presence of
H2O and/or CO2), we observe the formation of 7−10% of a
Zn-spinel phase (Al2ZnO4) with a particle size between 7 and
15 nm (Table S1). A portion of Al2O3 (∼25%) is therefore
consumed during the formation of Zn-spinel, while the other
Al2O3 species could be unavailable due to agglomeration. The
peak for the crystalline Al2O3 was not observed in any XRD
diffractogram as in the long term study by Lunkenbein et al.15
To monitor the state of Al2O3, the surface area of Al2O3 was
Figure 8. Sum of CO and CH3OH turnover frequency of the Cu calculated using the atomic fraction of Al from XPS and BET
surface (black circles). Activity measurement conditions: 240 °C, 50 surface area (section 4.1). Since only a minor part of Al2O3 is
bar, and GHSV of 40 000 h−1. The red triangle denotes sample R8 incorporated in a spinel, the entire Al is represented as Al2O3.
(no aging).
To validate the approach of the combination of XPS and BET,
the Cu2O surface area from the XPS is plotted against the Cu
therefore detrimental impact on the sintering behavior of the surface area determined from N2O pulse chemisorption
Cu and ZnO phase and it could change in the presence of (Figure S7). In Figure 10, Cu particle size, ZnO particle size,
water. In the presence of large amounts of steam, the catalyst and Al2O3 surface area are displayed as a function of aging
stability can be additionally increased using SiO2,9 thus doping molar fractions of H2O. It can be observed that all phases
a commercial CZA material. The Al2O3 phase is therefore exhibit a decrease in surface area (or increase of particle size)
sensitive to the moisture content. From the images acquired with increasing amount of steam. The sample R8 shows nearly
with STEM-EDX for the samples R1, R3, and R6, we observe identical features as sample R4, indicating that all catalyst
13026 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

Figure 11. Relation between Al2O3 surface and Cu particle size.

Figure 10. H2O impact on the Cu and ZnO particle size and on the
Al2O3 surface.
is faster than for Ni, which is responsible for increased Ni
growth. Due to the discussed similarity between the systems,
sintering occurred in the first 16 h (at 240 °C) and for this their modified model for Ni particle growth is used (eq 3):
reason is plotted at the same H2O fraction. The S(Al2O3)−
x(H2O) and d(Cu)−x(H2O) relations are very simple despite dCu
0
= (1 + ktx H2Om)1/(n − 1)
different aging conditions. The ZnO particle size is much more dCu (3)
dependent from other parameters as was discussed in the
section 3.2. where dCu is Cu particle size, d0Cu
is the initial crystallite size, k
From the observed experimental data and literature review, is the growth rate constant, t is time, x is H2O molar fraction,
it is now be possible to critically discuss the proposed m represents the nonlinearity of the atmosphere effect on
mechanisms of the H2 O effect on the catalyst. The sintering, and n is related to the growth mechanism.21 For the
hydroxylation of Cu could increase surface diffusion of the particle migration and coalescence, n is 8.21 To isolate sintering
metal particle and therefore sintering as for the case with Ni.21 due to H2O, the Cu particle size of the sample sintered in H2
H2O can adsorb on the Cu surface and dissociate to form (R7) is used as d0Cu.
adsorbed hydroxyl and hydrogen. The presence of surface H2O In Figure 12, Cu particle growth is plotted as a dependence
was proven during methanol synthesis by DRIFT.36,37 Residual of H2O molar fraction. In the figure we show that the model
*OH is also present on the surface, after the ratio between
CO2 and CO decrease.24 Adsorbed H2O/*OH could increase
the surface diffusion of Cu and subsequently cause Cu particle
growth. However, H2O adsorbs much stronger on the Al2O3
and ZnO surface and can also hydroxylate them to form
AlOOH and Zn(OH)2, respectively. As mentioned, Zn(OH)2
has a larger surface diffusion than ZnO and can be therefore
sintered faster. Another mechanism discussed in literature is
that under relatively oxidizing atmospheres, Cu−ZnO
interactions weaken,26,27 which could promote Cu crystallite
growth. However, the loss of the metal−support interaction
due to a much higher increase of ZnO particle size in the case
of R1 (20.5 nm) comparing to R5 (12.6 nm), also does not
have any significant effect on the Cu particle growth (dCu =
13.8 nm (R1), 15.1 nm (R5)). Despite the ZnO overlay over Figure 12. Growth factor for Cu (dCu/d0Cu) depending on the molar
the Cu particles (R3), the Cu particle size still increases by fraction of the water and model results for 48 h. The samples R6 and
approximately the same amount (dCu = 14.5 nm (R3)). R8 (empty circles) were not included in the regression due to a
Since the Al2O3 limits the sintering of other more mobile different aging time.
phases, the most probable catalyst deactivation mechanism
with water is due to a loss of Al2O3 surface and consequently for k is 0.56 s−1 and m is 3. However, with the change of k it is
decrease of Cu and ZnO support. From XPS measurements we also possible to use m between 3 and 5 to obtain sufficient fit.
estimated that Al2O3 is the most abundant phase on the Number m is larger than 1, which is an indication that Al2O3
catalytic surface (50−60%, Table S6). Additionally, TEM sinter faster (and therefore also Cu) when there are
images show small particles of Al2O3 finely dispersed over Cu neighboring Al2O3 influenced by H2O.
and ZnO (Figure 6). The Cu particle growth is closely
correlated with the decrease of the Al2O3 surface (Figure 11). 4. CONCLUSION
In line with our observations and the above-mentioned In the synthesis of methanol from CO2, an equivalent amount
literature, we state that the rate determining step is the of water is also produced. To obtain the impact of water on the
decrease of available Al2O3 for the support of Cu phase. The methanol catalyst, we performed aging experiments at various
decrease of support is correlated to the loss of Al2O3 surface gas compositions and post-mortem analysis of samples using
and therefore the amount of steam in the aging atmosphere. A different methods. We concluded that the most probable cause
model for Ni particle growth in the presence of steam21 is of catalyst deactivation is increasing the rate of sintering of
modified and implemented. The surface diffusivity of Ni(OH)2 Al2O3 with water. The Al2O3 surface decreases with increasing
13027 DOI: 10.1021/acs.iecr.9b01898
Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

molar fraction of H2O, while similar trends also follow the (9) Wu, J.; Saito, M.; Takeuchi, M.; Watanabe, T. The Stability of
growth of the Cu particle size. The Cu particle growth was also Cu/ZnO-Based Catalysts in Methanol Synthesis from a CO2-Rich
fitted to a coalescence model for sintering. While the form of Feed and from a CO-Rich Feed. Appl. Catal., A 2001, 218, 235−240.
Cu and the Al2O3 phase shows simple dependence on the (10) Simonsen, S. B.; Chorkendorff, I.; Dahl, S.; Skoglundh, M.;
amount of H2O in a gas stream, the ZnO particle size is also Sehested, J.; Helveg, S. Direct Observations of Oxygen-Induced
Platinum Nanoparticle Ripening Studied by In Situ TEM. J. Am.
dependent on other reaction products and CO2 in addition to
Chem. Soc. 2010, 132, 7968−7975.
H2O. With various methods we confirm that under aging in the (11) Challa, S. R.; Delariva, A. T.; Hansen, T. W.; Helveg, S.;
presence of large amounts of reaction products (CO, Sehested, J.; Hansen, P. L.; Garzon, F.; Datye, A. K. Relating Rates of
CH3OH), the ZnO phase starts to cover Cu particles. In Catalyst Sintering to the Disappearance of Individual Nanoparticles
contrast, water promotes the growth of each individual phase. during Ostwald Ripening. J. Am. Chem. Soc. 2011, 133, 20672−20675.

■
*
ASSOCIATED CONTENT
S Supporting Information
(12) Varela, J. A.; Whittemore, O. J.; Longo, E. Pore Size Evolution
during Sintering of Ceramic Oxides. Ceram. Int. 1990, 16, 177−189.
(13) Gal, P. L.; Smith, B. C.; Owen, G. Bulk Diffusion of Metal
Particles on Ceramic Substrates. Nature 1990, 348, 430−432.
The Supporting Information is available free of charge on the (14) Schumann, J.; Kröhnert, J.; Frei, E.; Schlögl, R.; Trunschke, A.
ACS Publications website at DOI: 10.1021/acs.iecr.9b01898. IR-Spectroscopic Study on the Interface of Cu-Based Methanol
Rietveld refinement results, XRD diffractograms, vali- Synthesis Catalysts: Evidence for the Formation of a ZnO Overlayer.
Top. Catal. 2017, 60, 1735−1743.
dation of N2O pulse chemisorption method, pore size (15) Lunkenbein, T.; Girgsdies, F.; Kandemir, T.; Thomas, N.;
distribution and pore volume, determination of surface Behrens, M.; Schlögl, R.; Frei, E. Bridging the Time Gap: A Copper/
composition using XPS including estimation of Al2O3 Zinc Oxide/Aluminum Oxide Catalyst for Methanol Synthesis
surface area and verification using N2O pulse chem- Studied under Industrially Relevant Conditions and Time Scales.
isorption, HRTEM images, parallel reactor system Angew. Chem., Int. Ed. 2016, 55, 12708−12712.
scheme, long-term catalytic test (PDF) (16) Fichtl, M. B.; Schlereth, D.; Jacobsen, N.; Kasatkin, I.;

■
Schumann, J.; Behrens, M.; Schlögl, R.; Hinrichsen, O. Kinetics of
Deactivation on Cu/ZnO/Al□O□ Methanol Synthesis Catalysts.
AUTHOR INFORMATION Appl. Catal., A 2015, 502, 262−270.
Corresponding Authors (17) Twigg, M. V.; Spencer, M. S. Deactivation of Copper Metal
*E-mail: blaz.likozar@ki.si. Catalysts for Methanol Decomposition, Methanol Steam Reforming
*E-mail: anze.prasnikar@ki.si. and Methanol Synthesis. Top. Catal. 2003, 22, 191−203.
(18) Hansen, T. W.; Delariva, A. T.; Challa, S. R.; Datye, A. K.
ORCID Sintering of Catalytic Nanoparticles: Particle Migration or Ostwald
Blaž Likozar: 0000-0001-9194-6595 Ripening? Acc. Chem. Res. 2013, 46, 1720−1730.
(19) Borgwardt, R. H. Calcium Oxide Sintering in Atmospheres
Notes
Containing Water and Carbon Dioxide. Ind. Eng. Chem. Res. 1989, 28,
The authors declare no competing financial interest.

■
493−500.
(20) Dargatz, B.; Gonzalez-Julian, J.; Bram, M.; Jakes, P.; Besmehn,
ACKNOWLEDGMENTS A.; Schade, L.; Röder, R.; Ronning, C.; Guillon, O. FAST/SPS
This research was supported by Slovenian Research Agency Sintering of Nanocrystalline Zinc Oxide  Part I: Enhanced
Densification and Formation of Hydrogen-Related Defects in
(Research Core Funding No. P2 0152) and Project FReSMe
Presence of Adsorbed Water. J. Eur. Ceram. Soc. 2016, 36, 1207−
No 727504. The authors are very grateful to Urška Kavčič for 1220.
N2 physisorption measurements, Brett Pomeroy for language (21) Sehested, J.; Gelten, J. A. P.; Helveg, S. Sintering of Nickel
editing, and Matic Grom for the scheme and help with reactor Catalysts: Effects of Time, Atmosphere, Temperature, Nickel-Carrier
operation. Interactions and Dopants. Appl. Catal., A 2006, 309, 237−246.

■ REFERENCES
(1) Olah, G. A. Beyond Oil and Gas: The Methanol Economy.
(22) Sahibzada, M.; Metcalfe, I. S.; Chadwick, D. Methanol
Synthesis from CO/CO□/ H□ over Cu/ZnO/Al□O□ at Differ-
ential and Finite Conversions. J. Catal. 1998, 174, 111−118.
(23) Liu, G.; Willcox, D.; Garland, M.; Kung, H. H. The Role of
Angew. Chem., Int. Ed. 2005, 44, 2636−2639.
(2) Sun, J. T.; Metcalfe, I. S.; Sahibzada, M. Deactivation of Cu/ CO□ in Methanol Synthesis on Cu-Zn Oxide: An Isotope Labeling
ZnO/Al2O3 Methanol Synthesis Catalyst by Sintering. Ind. Eng. Chem. Study. J. Catal. 1985, 96, 251−260.
Res. 1999, 38, 3868−3872. (24) Martin, O.; Pérez-Ramírez, J. New and Revisited Insights into
(3) Forzatti, P.; Lietti, L. Catalyst Deactivation. Catal. Today 1999, the Promotion of Methanol Synthesis Catalysts by CO□. Catal. Sci.
52, 165−181. Technol. 2013, 3, 3343−3352.
(4) Arcidiacono, S.; Bieri, N. R.; Poulikakos, D.; Grigoropoulos, C. (25) Sonneveld, E. J.; Visser, J. W. Automatic Collection of Powder
P. On the Coalescence of Gold Nanoparticles. Int. J. Multiphase Flow Data from Photographs. J. Appl. Crystallogr. 1975, 8, 1−7.
2004, 30, 979−994. (26) Grunwaldt, J.-D.; Molenbroek, A. M.; Topsøe, N.-Y.; Topsøe,
(5) Kung, H. H. Deactivation of Methanol Synthesis Catalyst-A H.; Clausen, B. S. In Situ Investigations of Structural Changes in Cu/
Review. Catal. Today 1992, 11, 443−453. ZnO Catalysts. J. Catal. 2000, 194, 452−460.
(6) Hochgesand, G. Rectisol and Purisol. Ind. Eng. Chem. 1970, 62, (27) Hansen, P. L.; Wagner, J. B.; Helveg, S.; Rostrup-Nielsen, J. R.;
37−43. Clausen, B. S.; Topsøe, H. Atom-Resolved Imaging of Dynamic Shape
(7) Kurtz, M.; Wilmer, H.; Genger, T.; Hinrichsen, O.; Muhler, M. Changes in Supported Copper Nanocrystals. Science 2002, 295,
Deactivation of Supported Copper Catalysts for Methanol Synthesis. 2053−2055.
Catal. Lett. 2003, 86, 77−80. (28) Kuld, S.; Conradsen, C.; Moses, P. G.; Chorkendorff, I.;
(8) Twigg, M. V.; Spencer, M. S. Deactivation of Supported Copper Sehested, J. Quantification of Zinc Atoms in a Surface Alloy on
Metal Catalysts for Hydrogenation Reactions. Appl. Catal., A 2001, Copper in an Industrial-Type Methanol Synthesis Catalyst. Angew.
212, 161−174. Chem., Int. Ed. 2014, 53, 5941−5945.

13028 DOI: 10.1021/acs.iecr.9b01898

Ind. Eng. Chem. Res. 2019, 58, 13021−13029
Industrial & Engineering Chemistry Research Article

(29) Natesakhawat, S.; Lekse, J. W.; Baltrus, J. P.; Ohodnicki, P. R.;

Howard, B. H.; Deng, X.; Matranga, C. Active Sites and Structure-
Activity Relationships of Copper-Based Catalysts for Carbon Dioxide
Hydrogenation to Methanol. ACS Catal. 2012, 2, 1667−1676.
(30) Zhan, H.; Li, F.; Gao, P.; Zhao, N.; Xiao, F.; Wei, W.; Sun, Y.
Influence of Element Doping on La-Mn-Cu-O Based Perovskite
Precursors for Methanol Synthesis from CO2/H2. RSC Adv. 2014, 4,
48888−48896.
(31) van den Berg, R.; Prieto, G.; Korpershoek, G.; van der Wal, L.
I.; van Bunningen, A. J.; Lægsgaard-Jørgensen, S.; de Jongh, P. E.; de
Jong, K. P. Structure Sensitivity of Cu and CuZn Catalysts Relevant to
Industrial Methanol Synthesis. Nat. Commun. 2016, 7, 1−7.
(32) Kattel, S.; Ramirez, P. J.; Chen, J. G.; Rodriguez, J. A.; Liu, P.
Active Sites for CO□ Hydrogenation to Methanol on Cu/ZnO
Catalysts. Science 2017, 355, 1296−1299.
(33) Palomino, R. M.; Ramirez, P. J.; Liu, Z.; Hamlyn, R.; Waluyo,
I.; Mahapatra, M.; Orozco, I.; Hunt, A.; Simonovis, J. P.; Senanayake,
S. D.; Rodriguez, J. A. Hydrogenation of CO□ on ZnO/Cu(100) and
ZnO/Cu(111) Catalysts: Role of Copper Structure and Metal - Oxide
Interface in Methanol Synthesis. J. Phys. Chem. B 2018, 122, 794−
800.
(34) An, B.; Zhang, J.; Cheng, K.; Ji, P.; Wang, C.; Lin, W.
Confinement of Ultrasmall Cu/ZnOx Nanoparticles in Metal-Organic
Frameworks for Selective Methanol Synthesis from Catalytic
Hydrogenation of CO□. J. Am. Chem. Soc. 2017, 139, 3834−3840.
(35) Bartholomew, C. H. Mechanisms of Catalyst Deactivation.
Appl. Catal., A 2001, 212, 17−60.
(36) Yang, R.; Fu, Y.; Zhang, Y.; Tsubaki, N. In Situ DRIFT Study of
Low-Temperature Methanol Synthesis Mechanism on Cu/ZnO
Catalysts from CO□-Containing Syngas Using Ethanol Promoter. J.
Catal. 2004, 228, 23−35.
(37) Fisher, I. A.; Bell, A. T. In-Situ Infrared Study of Methanol
Synthesis from H□/CO□. J. Catal. 1997, 172, 222−237.

13029 DOI: 10.1021/acs.iecr.9b01898

Ind. Eng. Chem. Res. 2019, 58, 13021−13029