You are on page 1of 8

ARTICLE IN PRESS

G Model
CATTOD-8355; No. of Pages 8

Catalysis Today xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

How oxide carriers control the catalytic functionality of the Cu–ZnO


system in the hydrogenation of CO2 to methanol
F. Arena a,b,∗ , G. Mezzatesta a , G. Zafarana a , G. Trunfio a , F. Frusteri b , L. Spadaro a,b
a
Dipartimento di Chimica Industriale e Ingegneria dei Materiali, Università degli Studi di Messina, Salita Sperone 31 c.p. 29, I-98166 S. Agata, Messina, Italy
b
Istituto CNR-ITAE “Nicola Giordano”, Salita S. Lucia 5, I-98126 S. Lucia, Messina, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The reactivity pattern of Al2 O3 (CuZnAl), CeO2 (CuZnCe) and ZrO2 (CuZnZr) supported Cu–ZnO systems
Received 31 August 2012 in the synthesis of methanol via CO2 hydrogenation in the range of 453–513 K at 3.0–5.0 MPa has been
Received in revised form 5 February 2013 addressed. The CuZnCe system shows superior surface methanol productivity, though textural and chem-
Accepted 12 February 2013
ical effects of zirconia carrier account for the better performance of CuZnZr catalyst. Characterization data
Available online xxx
of “steady-state” catalysts show significant surface coverage by CO2 irrespective of metal surface area
(MSA). Direct relationships among activity, CO2 uptake and oxides surface area (OSA) point out a dual-
Keywords:
site Langmuir–Hinshelwood reaction mechanism, involving hydrogenation and CO2 adsorption sites at
Cu–ZnO catalyst
Oxide carrier
the surface of both metal and oxide phases. The influence of space–velocity on selectivity signals the
CO2 hydrogenation occurrence of a parallel-consecutive path leading to methanol and CO, while higher reaction rate and
Methanol methanol selectivity with lowering contact time signal a negative influence of water formation on the
Surface functionality catalyst performance.
Reaction mechanism © 2013 Published by Elsevier B.V.

1. Introduction to Bell and Fischer who pointed out an active role of zirconia car-
rier on the reactivity of Cu/ZrO2 catalyst [18], the lack of direct
Environmental and economic advantages deriving from exten- relationships between metal exposure (MSA) and activity lead us
sive recycle of carbon dioxide emissions and exploitation of fuels to ascribe the superior functionality of Cu–ZnO/ZrO2 catalysts to
alternative to LPG, gasoline and gasoil pressed in the recent the synergism of active sites at the surface of both metal and
years a big scientific concern onto the CO2 -hydrogenation to oxide phases [5,6]. Further, in the light of the peculiar catalytic
methanol and/or dimethylether [1–4]. In this respect, many stud- functionality of CeO2 -based systems [7,12], we argued that oxide
ies documented a superior performance of Cu–ZnO/ZrO2 catalysts carriers and promoters control the reactivity of the active Cu
[4–8], generally ascribed to a lower water-affinity than commer- phase by determining texture, exposure of active sites and inter-
cial alumina-based ones [3,4,8,9]. However, the lack of probative action pattern with reagents, products and reaction intermediates
evidences on the CO2 -hydrogenation mechanism of oxide pro- [3–7,10].
moted Cu systems represents to date the major drawback for Therefore, this study is aimed at providing a comparative assess-
catalyst development [3,4,7,10–12]. Indeed, though methanol syn- ment of the CO2 -hydrogenation pattern of Al2 O3 , ZrO2 and CeO2
thesis reaction on Cu catalyst relies on a classical “one-site” supported Cu–ZnO systems in the range of 453–513 K at 3.0 and
Langmuir–Hinshelwood (L–H) model [13,14], a mix of structural 5.0 MPa. Characterization of steady-state catalysts indicates a fun-
and/or electronic effects determines the positive influence of many damental role of CO2 adsorption on the catalytic functionality,
oxide promoters on the CO2 -hydrogenation pattern of the Cu while textural and chemical effects of zirconia carrier determine the
catalyst [3,7,10]. In particular, ZnO plays an essential promoting superior performance to Cu–ZnO/ZrO2 catalyst, fixed by methanol
−1
influence on the catalytic functionality of Cu owing to enhanced space–time–yield of 1.2 kgCH3 OH kg−1 cat h at 10% CO2 conversion
metal dispersion and indefinite electronic effects of ZnO patches, per pass.
leading the stabilization of Cu–Zn alloy(s) and electron-deficient
Cuı+ sites at the Cu–ZnO interface [3,5,14–17]. Moreover, according
2. Experimental

∗ Corresponding author at: Dipartimento di Chimica Industriale e Ingegneria dei


2.1. Materials
Materiali, Università degli Studi di Messina, Italy. Tel.: +39 0906765484;
fax: +39 090391518. Ceria (CuZnCe) and zirconia (CuZnZr) supported Cu–ZnO sys-
E-mail address: Francesco.Arena@unime.it (F. Arena). tems with a Cu/Zn atomic ratio of ca. 3 were prepared by the reverse

0920-5861/$ – see front matter © 2013 Published by Elsevier B.V.


http://dx.doi.org/10.1016/j.cattod.2013.02.016

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

2 F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx

Table 1
Physico-chemical properties of the calcined and reduced catalysts.

Catalyst Sample Chemical Composition (wt %) SA (m2 /g) PV (cm3 g−1 ) MSA (mCu 2 g−1 ) D (%) dCu a (nm)

CuO ZnO Al2 O3 ZrO2 CeO2

Calcined 35.0 33.1 31.9 – – 105 0.23 – –


CuZnAl
Reducedb 79 0.29 36 21 5

Calcined 43.2 14.5 – 42.3 – 154 0.34 – –


CuZnZr
Reducedb 132 0.28 73 32 3

Calcined 42.7 13.7 – – 43.6 47 0.24 – –


CuZnCe
Reducedb 34 0.17 10 5 21
a
Average Cu particle size obtained from the formula dCu = 104/D(%) [4].
b
Reduced samples at 573 K (1 h) in H2 flow.

co-precipitation under ultrasounds irradiation route of Cu2+ , Zn2+ , was controlled by a thermocouple in contact with catalyst
Ce3+ and ZrO2+ nitrate precursors, according to the procedure else- bed.
where described [4], keeping constant loading of CuO (≈42 wt%), Before testing, catalysts were reduced in situ at 573 K in H2
ZnO (≈14 wt%) and oxide carrier (≈44 wt%). After precipitation, cat- flow (100 stp mL/min) at atmospheric pressure (1 h). Thereafter,
alysts were filtered, dried at 373 K (16 h) and further calcined in the reactor was cooled down to 453 K, admitting the reaction
air at 623 K (4 h). A commercial Cu–ZnO/Al2 O3 (CuZnAl) methanol mixture and raising the pressure to 3.0–5.0 MPa. Heating–cooling
synthesis catalyst (G66A, Sud Chemie AG) was taken as reference. cycles at any pressure were performed to ascertain the lack of
Powdered catalysts were pressed at 40 MPa and then crushed and significant deactivation phenomena. The reaction stream was ana-
sieved to 40–70 mesh particle size fraction. The list of the cata- lyzed by a GC equipped with a two-column analytical system
lysts, with their relative formulation and main physico-chemical connected to FID (CH3 OH, CH3 OCH3 ) and TCD (CO, N2 , CO2 , H2 )
properties, is reported in Table 1. [4]. Methanol and CO were the only detected reaction prod-
ucts, methane and dimetylether being always at trace levels, if
any.
2.2. Methods

Surface area (SA) and pore volume (PV) values were obtained 3. Results and discussion
by the standard BET and BJH method elaboration respectively
of nitrogen adsorption/desorption isotherms (77 K), taken by 3.1. Physico-chemical properties
an automated gas adsorption device (ASAP 2010, Micromeritics
Instrument). Before the adsorption measurements, samples were The data in Table 1 show a different chemical composition of
outgassed at 423 K under vacuum for 2 h. the studied catalysts, as reference CuZnAl catalyst includes similar
Metal surface area (MSA) and dispersion (D) were determined by loadings of CuO (≈34 wt%), ZnO (≈38 wt%) and Al2 O3 (≈29 wt%),
“single-pulse” (0.65 mL) N2 O titration measurements at 363 K [4–7]. while lab-made CuZnZr and CuZnCe contain comparable higher
Before measurements, catalysts were reduced at 573 K in H2 flow loading (≈43 wt%) of CuO and oxide carrier at the expense of ZnO
(100 stp mL/min) for 1 h and, then “flushed” in a N2 carrier flow at (≈14 wt%). This corresponds to a Zn/Cu atomic ratio of 0.3–0.4
583 K for 15 min and further cooled down to 363 K, for the N2 O pulse (Table 1), previously indicated as an optimum for Cu–ZnO/ZrO2
injection. A chemisorption stoichiometry of N2 O:Cu = 1:2 and a sur- catalyst composition [4].
face concentration of 1.46 × 1019 Cu atom/m2 were assumed for Then, reference CuZnAl system features a SA of 105 m2 /g low-
metal surface area (MSA) and dispersion (D) calculations, respec- ering to 79 m2 /g after the reduction treatment (−SA≈25%), while
tively [4]. the calcined CuZnZr system displays the largest SA (154 m2 /g)
Temperature programmed surface reaction (TPSR) tests in the and the smallest decay (−SA≈14%) further to reduction, con-
range of 293–573 K were performed at heating rate of 6 K/min firming the most effective role of zirconia as textural promoter
using Ar or 75% H2 /Ar mixture as carriers (30 stp mL/min). Before of the Cu–ZnO system [4–7]. Whereas, the lowest SA (47 m2 /g),
measurements catalysts (50 mg) were reduced for 1 h at 573 K decreasing to 34 m2 /g on the reduced sample (−SA≈28%),
in H2 flow (50 stp mL/min), then cooled down to 493 K admit- denotes a poorer efficiency of ceria carrier to promote sur-
ting a stoichiometric CO2 /H2 /Ar (1/3/2) mixture (60 stp mL/min) face texture and resistance to sintering of CuZnCe system
and keeping stationary conditions for 30 min. Thereafter, cata- [7].
lysts were quenched to 298 K still under reaction mixture flow, On the whole, N2 O-titration data (Table 1) indicate the occur-
and then flushed by the carrier until stabilization of signals base- rence of a close relationship between metal sites exposure (MSA)
line (≈20 min). Profiles of reagents and products were obtained and SA, as shown in Fig. 1. The CuZnCe catalyst is characterized
by a Quadrupole Mass Spectrometer (Hiden Analytical, HPR 20), by the lowest metal dispersion (D, 5%) and surface area (MSA,
equipped with a heated (453 K) inlet capillary system, monitoring 10 m2 /g), though a MSA/SA ratio considerably lower (≈0.3) than
the following mass-to-charge signal ratios: 2 (H2 ), 40 (Ar), 18 (H2 O), CuZnZr and CuZnAl systems (Fig. 1) denotes an enhanced ten-
28 (CO), 31 (CH3 O), 44 (CO2 ). dency to sintering, probably induced by an incipient reduction of
Catalyst testing was carried out in the range of 453–513 K at ceria carrier [7]. Higher dispersion (32%) and MSA (73 m2 /g) mir-
3.0–5.0 MPa using a gas–liquid feed automated plant, equipped ror the largest surface development of the CuZnZr system (Fig. 1),
with an Inconel micro-reactor (7–10 mm) jacketed with a stainless- while a MSA/SA ratio of 0.55 signals a very strong promoting
steel round to ensure isothermal reaction conditions. The reactor influence of zirconia carrier also on texture of metal phase [4,7].
was loaded with 0.5 g of catalyst diluted (1/1, wt/wt) with The CuZnAl sample features intermediate dispersion (21%) and
same-sized ␣-Al2 O3 and fed with a stoichiometric CO2 /H2 /N2 MSA (36 m2 /g) values (Fig. 1A), accounting for a MSA/SA ratio of
(23/69/8) mixture at rate varying between 80 and 500 stp mL/min 0.46 (Fig. 1). These data imply that the oxide-surface-area (i.e.,
(GHSV, 8800–55,000 NL kgcat −1 h−1 ), while temperature OSA = SA − MSA) to SA ratio (OSA/SA) depicts an opposite trend

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx 3

80 0.8

MSA/SA (mCu2/mcat2) - OSA/SA (mOx2/mcat2)


OSA/SA

MSA (mCu2/gcat) - OSA (mOx2/gcat)


60 0.6

40 0.4
MSA/SA

20 OSA 0.2

MSA
0 0.0
30 70 110 150
SA (mcat2/g)
Fig. 1. Influence of surface area (SA) on the extent of metal (MSA) and oxide surface area (OSA) and MSA/SA and OSA/SA ratios.

to the MSA/SA one (Fig. 1), resulting larger for CuZnCe system methanol decomposition (MD) is a linear combination of previous
(0.71). reactions [4]:
Determining the texture of the final catalysts [3–7,11], it is
expected that oxide carriers control the CO2 -hydrogenation func- CO2 + 3H2  CH3 OH + H2 O (MS)
tionality of the Cu–ZnO system. CO2 + H2  CO + H2 O, (RWGS)

CH3 OH  CO + 2H2 . (MD)


3.2. Effect of temperature and pressure on activity and surface
functionality As a rule, conversion increases with both pressure and tem-
perature, while selectivity rises with pressure and decreases with
The activity pattern of the studied catalysts (F, 80 stp mL/min; temperature. Moreover, though all catalysts exhibit their per-
GHSV, 8800 NL h−1 kgcat −1 ) at 3.0 and 5.0 MPa is shown in Fig. 2, formance within the limits of thermodynamics, conversion-yield
comparing CO2 conversion (XCO2 ), CH3 OH selectivity (SCH3 OH ) and values higher than 20–30% of corresponding equilibrium values
yield (YCH3 OH ) in the range of 453–513 K with relative thermody- suggest that at 493 K, and mostly 513 K, the reactivity pattern of
namic equilibrium values. These were obtained from the equation CuZnAl and CuZnZr catalysts is controlled by the approach to equi-
system of equilibrium constants KMS and KWGSR (see Appendix librium composition (Fig. 2).
A of Supplementary Information) relative to methanol synthesis Namely, the commercial CuZnAl system (Fig. 2A and A ) shows
(MS) and reverse water gas shift (RWGS) reactions respectively, as conversion values increasing from 3.2 and 4.8% (453 K) to 13.5 and

Fig. 2. Conversion, selectivity and yield data (dotted lines) in the range of 453–513 K at 3.0 (A–C) and 5.0 MPa (A –C ) of CuZnAl (A, A ), CuZnZr (B, B ) and CuZnCe (C, C )
catalysts. Continuous lines refer to equilibrium values (experimental conditions: Fmix , 80 stp mL/min; CO2 /H2 /N2 , 23/69/8; wcat , 0.5 g).

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

4 F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx

0.25 CuZnAl 0.25 CuZnAl


A) A’)

SSY (mmol CH3OH mcat -2 h-1)


CuZnZr CuZnZr

SSA (mmol CO2 mcat -2 h-1)


• 0.20 CuZnCe •
0.20 CuZnCe


• 0.15 0.15

0.10 0.10

0.05 0.05
3.0 MPa 3.0 MPa
0.00 0.00
453 473 493 513 453 473 493 513
Temperature (K) Temperature (K)

0.25 0.25
CuZnAl
B) CuZnAl
B’)

SSY (mmol CH3OH mcat-2 h-1)


CuZnZr CuZnZr
SSA (mmol CO2 mcat -2 h-1)

CuZnCe •
• 0.20 0.20 CuZnCe


• 0.15 0.15

0.10 0.10

0.05 0.05

5.0 MPa 5.0 MPa


0.00 0.00
453 473 493 513 453 473 493 513
Temperature (K) Temperature (K)

−1 −1
Fig. 3. (A, B) Specific surface rate of CO2 conversion (SSA, mmolCO2 m−2   −2
cat h ) and (A , B ) methanol formation (SSY, mmolCH3 OH mcat h ) in the range of 453–513 K at 3.0 (A,
A ) and 5.0 MPa (B, B ) (experimental conditions: Fmix , 80 stp mL/min; CO2 /H2 /N2 , 23/69/8; wcat , 0.5 g).

19.5% (513 K) at 3.0 and 5.0 MPa, respectively. Although selectiv- mmol mCu −2 h−1 ) and CH3 OH production (SMSY, mmol mCu −2 h−1 ),
ity lowers from 90–93% (453 K) to 44–61% (513 K), the yield rises as shown in Fig. 4.
steadily with temperature and pressure from minimum values of Therefore, while the poor process performance of the CuZnCe
2.9–4.5% (453 K) to 5.9–11.9% (513 K) at 3.0 and 5.0 MPa, respec- catalyst relies on the low efficiency of ceria as structural promoter
tively. With comparable or higher methanol selectivity levels and of the Cu–ZnO system, such findings lead to ascribe its enhanced
CO2 conversion values rising from 4.4–5.7% (453 K) to 19.3–22.4% surface functionality to the peculiar reactivity of the ceria carrier
at 3.0 and 5.0 MPa respectively, the CuZnZr catalyst (Fig. 2B and [7,12].
B ) shows the highest yields, increasing from 3.8–5.1% (453 K)
to 9.7–14.3% (513 K). At least, with a conversion increasing from
3.3. Effect of contact time on the activity–selectivity pattern
2.0–2.2% (453 K) to 7.2 (3.0 MPa) and 8.5% (5.0 MPa) at 513 K (Fig. 2C
and C ), the CuZnCe catalyst shows the lowest activity and selectiv-
The potential constraints of thermodynamic equilibrium on
ity levels considerably higher than the previous systems (81–97%).
reaction kinetics and catalyst performance prompted us to inspect
Then, methanol selectivity higher than equilibrium values (Fig. 2C
the influence of CO2 conversion on the reactivity pattern of the
and C ) might depend on the poor activity of the system, result-
CuZnZr system. Conversion–selectivity-yield data, summarized in
ing in the occurrence of kinetic conditions in the whole range of
Table 2, disclose a decreasing trend of conversion with flow rate
temperature.
while methanol selectivity rises steadily. Moreover, conversion
Overall, these data provide the following reactivity scale in
data disclose an increasing reaction rate with space–velocity in the
terms of methanol yield
whole range of temperature (Fig. 5A and B), denoting a general neg-
ative effect of CO2 conversion on the reaction kinetics [3,19,20].
CuZnZr > CuZnAl > CuZnCe. Then, at T > 473 K increasing reaction rate data are consistent with
the aforesaid negative influence of the approach to equilibrium
Despite such data confirm the superior process performance of the composition, while at T ≤ 473 K this effect might depend mostly
CuZnZr system [3–8], yet a definitive assessment of the effects of on kinetic factors. However, the Weisz–Prater criterion signals the
carrier on the catalytic functionality of the Cu–ZnO system must lack of diffusional resistances on reaction kinetics at any tempera-
take into account the total and metal surface development of the ture (see Appendix B of Supplementary Information). Therefore, the
catalysts (Table 1). Then, the specific surface rates of CO2 con- positive influence of flow rate on reaction kinetics must be linked
−1
version (SSA, mmolCO2 m−2 cat h ) and methanol production (SSY, to a negative influence of CO2 conversion and, particularly, to the
−1
mmolCH3 OH m−2 cat h ) in the range of 453–513 K at 3.0 and 5.0 MPa consequent increase of water partial pressure across the catalyst
are compared in Fig. 3. Despite the lower activity, the CuZnCe cat- bed, which would determine an incipient oxidation of metal Cu
alyst features SSA values higher than both CuZnAl and CuZnZr sites (e.g., Cu + H2 O → CuO + H2 ), depressing reaction rate at higher
at 3.0 MPa (Fig. 3A), while at 5.0 MPa CuZnCe and CuZnAl cata- contact times. In fact, a selectivity pattern strongly dependent on
lysts show a comparable SSA in the range of 453–513 K, slightly flow rate at T < 493 K (Table 2) signals a marked influence of conver-
higher than CuZnZr catalyst (Fig. 3B). Consequence of much higher sion on the rate of product formation, likely linked also to the state
selectivity levels, yet, the CuZnCe catalyst ensures SSY significantly of the catalyst surface under steady-state conditions [21]. Namely,
higher than CuZnAl and CuZnZr systems in the whole range of tem- for differential (<10%) extents of CO2 conversion, in the absence
perature (Fig. 3A and B ). Besides, as a consequence of the lowest of any thermodynamic constraints, parallel paths of CH3 OH (MS)
MSA/SA ratio (Fig. 1), the CuZnCe catalyst features an even higher and CO formation (RWGS) should result in a constant CH3 OH-to-CO
reactivity in terms of specific metal rate of CO2 conversion (SMSA, selectivity ratio (SCH3 OH /SCO ), according to the fact that the forward

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx 5

0.8 CuZnAl 0.8 CuZnAl

SMSY (mmol CH3OH m Cu-2h-1)


CuZnZr
A) A’)

SMSA (mmol CO2 mCu-2h-1)


CuZnZr
CuZnCe CuZnCe
0.6 0.6

0.4 0.4

0.2 0.2

3.0 MPa 3.0 MPa


0.0 0.0
453 473 493 513 453 473 493 513
Temperature (K) Temperature (K)

1.0 CuZnAl 0.8 CuZnAl


B) B’)

SMSY (mmol CH3OH m Cu-2h-1)


SMSA (mmol CO2 m Cu-2h-1)

CuZnZr CuZnZr
0.8 CuZnCe CuZnCe
0.6
0.6
0.4
0.4

0.2
0.2

5.0 MPa 5.0 MPa


0.0 0.0
453 473 493 513 453 473 493 513
Temperature (K) Temperature (K)

−1 −1
Fig. 4. (A, B) Specific metal surface rate of CO2 conversion (SMSA, mmolCO2 m−2 Cu
h ) and (A , B ) methanol formation (SMSY, mmolCH3 OH m−2
Cu
h ) in the range of 453–513 K
at 3.0 (A, A ) and 5.0 MPa (B, B ) (experimental conditions: Fmix , 80 stp mL/min; CO2 /H2 /N2 , 23/69/8; wcat , 0.5 g).

rates depend only on the partial pressure of reagents (see Appendix (MD) route at T ≤ 493 K. Moreover, despite SCH3 OH /SCO values of ca.
C of Supplementary Information) 40 suggest that methanol is a primary product of the CO2 hydro-
genation reaction (453 K) [20], the steep decrease of the SCH3 OH /SCO
SCH3 OH rateCH3 OH kMS · pCO2 kMS
= = = , (1) ratio proves that the MD reaction is favoured at higher temperature
SCO rateCO kRWGS · pCO2 kRWGS and conversion. While, the progressive flattening of the SCH3 OH /SCO
where kMS and kRWGS are the pseudo-first order constants of parallel ratio with temperature (Fig. 5A and B ) mirrors the increasing con-
MS and RWGS reactions, including the hydrogen dependence. On tribution of the parallel RWGS path [20]. Then, the network of
the other hand, the SCH3 OH /SCO ratio should result a direct function MS, RWGS and MD reactions catches the main clues of the surface
of conversion for a consecutive path. In particular, assuming that CO2 -hydrogenation pattern of supported Cu–ZnO catalysts, since
at differential (<10%) extents of CO2 conversion the average par- the catalytic functionality is dominated by MS and MD paths at
tial pressure of reagents and products across catalyst bed can be T ≤ 473 K, while the parallel RWGS, characterized by higher activa-
roughly considered equal to the arithmetic average and taking into tion barrier than MS and MD reactions, becomes important at high
account only forward formation rates, far away from thermody- temperature [20].
namic constraints, the SCH3 OH /SCO ratio results a straight function Consequence of the strong influence of process conditions on
of the reciprocal of CO2 conversion (1/XCO2 ), according to the fol- catalyst performance, the CuZnZr features a steady rise of the
lowing equation (see Appendix C of Supplementary Information) methanol space–time-yield with flow rate (Fig. 6), up to the
  remarkable value of 1.2 kg kgcat −1 h−1 at ≈10% CO2 conversion per
SCH3 OH kMS 1 pass (T, 513 K; P, 5.0 MPa).
=2· · − 1, (2)
SCO kMD XCO2

where kMS and kMD are the pseudo-first order constants of 3.4. Catalyst steady-state, active sites and surface functionality
consecutive MS and MD reactions. Then, experimental selectivity-
conversion data of the CuZnZr catalyst at different space–velocity In spite of the extensive literature on the reactivity pattern of
were inspected by Eq. (2), as shown in Fig. 5A and B . A satisfac- oxide-promoted Cu catalyst, the nature of active sites responsi-
tory fit of experimental SCH3 OH /SCO data (r2 , 0.90–0.99) proves the ble of the various surface functionalities is still matter of debate
consecutive path of CO formation via the methanol decomposition [3–7,11,12,15–18,21]. For instance, Klier et al. proposed [22] that

Table 2
Activity data of CuZnZr catalyst in the range of 453–513 K at 3.0–5.0 MPa and different feed flow rate.

Fmix (stp mL/min)  (s) P (MPa) TR , 453 K TR , 473 K TR , 493 K TR , 513 K

XCO2 − SCH3 OH − YCH3 OH (%)

3.0 4.4 87 3.8 8.4 74 6.2 13.9 57 7.9 19.3 50 9.6


80 0.4
5.0 5.7 90 5.1 10.0 78 7.8 17.4 65 11.3 22.4 64 14.3
3.0 3.6 94 3.4 5.0 84 4.2 7.9 64 5.1 13.3 48 6.4
160 0.2
5.0 4.3 94 4.1 6.2 84 5.2 11.0 71 7.8 16.9 60 10.1
3.0 2.5 95 2.4 3.3 88 2.9 5.0 70 3.5 9.5 51 4.8
330 0.1
5.0 3.1 96 3.0 4.7 90 4.2 7.0 74 5.2 11.0 60 6.6
3.0 2.0 97 1.9 2.3 90 2.1 4.3 77 3.3 7.5 55 4.1
500 0.06
5.0 2.7 97 2.6 3.5 91 3.2 5.6 78 4.4 9.7 62 6.0

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

6 F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx

2.0×10 - 5 2.0×10 - 5
3.0 MPa 5.0 MPa

reaction rate (molCO2 gcat-1 s-1)


reaction rate (molCO2 gcat-1 s-1)
• A) • B)
513K
1.5×10 - 5 1.5×10 - 5
• •
513K
1.0×10 - 5 1.0×10 - 5
493K
493K
473K
5.0×10 - 6 5.0×10 - 6
473K 453K
453K
0 0
0 200 400 600 0 200 400 600
flow rate ( stp mL/min) flow rate (stp mL/min)

40 40
3.0 MPa 5.0 MPa
A’) B’)
453K
30 453K 30
SCH3OH/SCO

SCH3OH/SCO
20 20

473K
10 473K 10

513K 493K 493K


513K
0 0
0 20 40 60 0 10 20 30 40
1/XCO2 1/XCO2

Fig. 5. (A, B) Influence of flow rate on the rate of CO2 conversion at 3.0 (A) and 5.0 (B) MPa and different temperature. (A , B ) Fitting of methanol-to-CO selectivity ratio
(SCH3 OH /SCO ) as a function of the reciprocal CO2 conversion at 3.0 (A ) and 5.0 (B ) MPa by Eq. (2) (experimental conditions: CO2 /H2 /N2 = 23/69/8; wcat , 0.5 g).

oxidized Cu is active and reduced Cu is not active in methanol water, at variance of MS and MD reactions proceeding in a strongly
synthesis, arguing that copper incorporated into the Zn lattice reducing environment [21]. This determines a different surface
as Cu+ was the active site for reactant adsorption [15,22]. In state of metal sites and, in turn, significant changes in the reac-
the case of methanol decomposition, both Cu+ and Cu0 seem tivity towards MS–RWGS and MSR–WGS processes. In addition,
to be active; however, because of the reducing environment of dispersion and morphology of metal particles determine affinity
methanol decomposition, Cu0 was thought to be the active site on and interaction strength of the various reaction intermediates with
Cu–ZnO/Al2 O3 catalyst [23]. Moreover, while Fisher and Bell sug- surface Cu sites [4,5]. All these evidences suggest that the activity-
gest that metal Cu serves as a sink for atomic hydrogen during MD selectivity pattern depends on different functionalities driven by
on Cu/ZrO2 /SiO2 catalyst [24], Choi and Stenger come to conclusion different active sites.
that Cu2+ centres drive both MD and methanol steam reforming In this context, according to Choi and Stenger [21], the nega-
(MSR) reactions [21]. Despite the similar physico-chemical charac- tive effect of contact time on rate and methanol selectivity would
teristics of the catalysts driving MS, WGS, MSR and MD reactions, mostly reflect the influence of an increasing water partial pressure
the big uncertainty affecting the nature of active sites depends also across the catalyst bed, leading to an incipient oxidation of Cu sites
on the different composition and redox potential of the reacting (CuO) [21]. This is consistent with the fact that the consecutive MD
atmosphere, deeply influencing the state of surface Cu atoms [21]. path is much more evident for CuZnZr and CuZnAl catalysts, since
Indeed, MSR and WGS reactions imply the occurrence of preva- their high dispersion would enhance the oxidation of Cu sites by
lently oxidative conditions in the presence of huge amounts of water [4,5,20].
Then, to shed lights into the surface “steady-state”, reduced cat-
1.5
alysts (573 K) were kept under “stationary” reaction conditions at
493 K for 30 min in a CO2 /H2 (1/3) mixture flow and, then, quenched
5.0 MPa
to 293 K (<10 s). Temperature programmed analyses in the range of
STY (kg CH3OH kg cat-1 h -1)

293–573 K under Ar disclose abundance and interaction pattern of



adsorbed reagents and products with the catalyst surface, whereas
1.0
the reactivity of adsorbed CO2 species was probed by analyses in
• 3.0 MPa
H2 stream of catalyst samples subjected to the above saturation
treatment. Using proper response factors for the various m/z val-
ues monitoring reagents and products, mass-balance calculations
0.5 (C.M.B.) on the released CO2 , CH3 OH and CO both in Ar and H2
streams result always in a satisfactory agreement (100 ± 5%), as
observable from Table 3.
Under Ar stream all the catalysts display a slight release rate
0.0 of CO2 and CO in the range of 293–413 K, both rising suddenly
0 5 10 15 at higher temperature (Fig. 7A–C; solid lines), along with trace
-1 -1
amounts of methanol in the range of 353–423 K (Table 3), “frozen”
flow rate (stp mLg

cat • s ) on the catalyst surface. Quantitative data in Table 3 indicate that
the extent of released CO2 and CO varies from 125 (CuZnCe) to
Fig. 6. Influence of flow rate on methanol space-time-yield (STY) at 513 K
and 3.0 and 5.0 MPa (experimental conditions: reaction mixture composition,
306 ␮mol gcat −1 (CuZnZr) and from 10 (CuZnAl) to 62 ␮mol gcat −1
CO2 /H2 /N2 = 23/69/8; wcat , 0.5 g). (CuZnZr), respectively. In particular, the CO2 profile of CuZnAl

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx 7

CO A) CO B) CO C)

pressure (a. u.)


pressure (a. u.)

pressure (a. u.)


CO2 CO2 CO2

H2O H2O H2O

CH3OH CH3OH CH3OH

293 333 373 413 453 493 533 573 293 333 373 413 453 493 533 573 293 333 373 413 453 493 533 573
Temperature (K) Temperature (K) Temperature (K)

Fig. 7. Temperature programmed profiles of CO2 , CO, CH3 OH (black lines) and H2 O (grey lines) of “steady-state” CuZnAl (A), CuZnZr (B) and CuZnCe (C) catalysts in Ar (solid
lines) and H2 (dotted lines).

(Fig. 7A) and CuZnZr (Fig. 7B) samples consists of a broad release 30
with maxima at 443–453 K and 503–513 K, while the CuZnCe cat-

Surface COx release (µmol•m-2 )


alyst shows only a main component centred at 513 K (Fig. 7C).
MSA
Further, CuZnZr and CuZnCe systems show a CO profile with a
broad maximum at ca. 513 K, while for the CuZnAl one it is much 20
featureless showing a poorly resolved maximum at ca. 433 K.
Characterized by a poor interaction strength with Cu0 sites [4,5],
adsorbed CO is expected to desorb at much lower temperature
and, then, it should arise from the splitting of CO2 to CO and oxy-
10 OSA
gen atoms, causing an incipient oxidation of metal particles, and
CO strongly adsorbed on Cuı+ sites [4,5]. Although a poor water
release indicates that products do not arise from surface reac- SA
tions between adsorbed CO2 and H2 , a larger water signal seems
0
to account for the higher water-affinity of CuZnAl catalyst, induced
both by alumina support and higher ZnO load [3,4]. Referred to 5 10 15 20 25 30 35
total, metal and oxide surface area values (Table 1) the released
D (%)
amount of COx against dispersion result in the trends shown in
Fig. 8. These document an apparent negative influence of dispersion Fig. 8. Influence of metal dispersion on the specific “steady-state” adsorption capac-
on metal surface COx uptakes, similar to TOF of Cu sites (see infra) ity referred to total (SA), metal (MSA) and oxide surface area (OSA).
[4–6,20]. The SA and OSA uptakes decrease with dispersion accord-
ing to analogous trends (Fig. 8), resulting still larger on CuZnCe
system exposing a larger fraction of oxide phases (Fig. 1). Indeed, no and the corresponding rise of CH3 OH, CO and water ones (Table 3).
contribution of promoter and carrier to surface adsorption would Then, the CO2 profile monitors spectator CO2 moieties, while the
account for a metal surface adsorption capacity decreasing from missing part of the profile recorded in Ar highlights concentra-
≈30 (CuZnCe) to 4–6 ␮mol mCu −2 while, with reference to OSA, the tion and type of reactive species. Spanning the range of 333–573 K
surface uptakes are comprised between 12 and 3–6 ␮mol mOx −2 , and with broad maxima at 373–413 K and 483–503 K, respectively,
respectively (Fig. 8). In any case, these document a strong promot- methanol profiles look similar for all the catalysts. An integral area
ing effect of ceria carrier on the CO2 adsorption capacity of the ca. two orders of magnitude larger than that recorded in Ar (Table 3)
Cu–ZnO system, ascribable to its peculiar ability to form surface substantiates the prevalent MS functionality of the studied sys-
and bulk oxygen vacancies at metal–oxide interface under reducing tems, confirming also that methanol is a primary product of the
atmosphere [25]. CO2 -hydrogenation reaction at very low temperature (<373 K). The
The profiles of reagents and products under H2 stream lack of CO at lower temperature probably depends on the inhibi-
(Fig. 7A–C; dotted lines) are quite different both in qualitative and ting effect of H2 on CO2 splitting, while the steady rise at T > 373 K
quantitative terms, since they mirror the ongoing hydrogenation of to a much larger extent than in Ar mirrors instead the increasing
adsorbed CO2 , probed by the strong decrease (60–80%) of its signal contribution of MD and RWGS reactions [21].

Table 3
Temperature programmed data of “steady-state” catalysts.

Catalyst Ar stream amount desorbed (␮mol/gcat ) H2 stream amount desorbed (␮mol/gcat )

CO2 CO CH3 OH products CO2 CO CH3 OH products

CuZnAl 125 10 2 137 35 45 59 139


CuZnZr 306 62 2 370 104 196 69 369
CuZnCe 259 29 3 291 104 143 51 298

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016
ARTICLE IN PRESS
G Model
CATTOD-8355; No. of Pages 8

8 F. Arena et al. / Catalysis Today xxx (2013) xxx–xxx

4×10-0 1 In particular, the main results of the work can be summarized


Specific surface rate (µmol CO2 m-2 s-1)


A) in the followings:

3×10 -0 1 MSA • Oxide carrier controls texture and adsorption properties of the
Cu–ZnO system;
• Methanol is a primary reaction product at low temperature and
2×10 -0 1
extent of CO2 conversion;
OSA • CO formation occurs via a consecutive-parallel reaction network,
mostly at T ≥ 473 K;
1×10 -0 1
• Steady-state CO2 adsorption is not related to metal surface expo-
SA sure (MSA);
0 • The catalytic activity depends on a synergism of metal and basic
5 10 15 20 25 30 35 oxide sites, pointing to a dual-site reaction path;
• Textural and chemical effects of zirconia carrier confer a supe-
D (%)
rior process performance to ZrO2 -supported system, ensuring
−1
4×10 -0 1 methanol space time yields up to 1.2 kgCH3 OH kg−1 cat h at ≈10%
Specific surface rate (µmolCO2 m-2 s-1)


MSA
B) of CO2 conversion per pass.

3×10 -0 1

Appendix A. Supplementary data


2×10 -0 1

OSA Supplementary data associated with this article can be


found, in the online version, at http://dx.doi.org/10.1016/j.
1×10 -0 1 cattod.2013.02.016.
SA
References
0
5 10 15 20 25 30 35 [1] G.A. Olah, A. Goeppert, G.K. Surya Prakash, Journal of Organic Chemistry 74
(2009) 487.
D (%) [2] C. Song, Catalysis Today 115 (2006) 2.
[3] X.-M. Liu, G.Q. Lu, Z.-F. Yan, J. Beltramini, Industrial and Engineering Chemistry
Fig. 9. Influence of metal dispersion on the specific rate of CO2 conversion (473 K) Research 42 (2003) 6518.
referred to total (SA), metal (MSA) and oxide surface area (OSA) at 3.0 (A) and 5.0 [4] F. Arena, K. Barbera, G. Italiano, G. Bonura, L. Spadaro, F. Frusteri, Journal of
(B) MPa. Catalysis 249 (2007) 185.
[5] F. Arena, G. Italiano, K. Barbera, S. Bordiga, G. Bonura, L. Spadaro, F. Frusteri,
Applied Catalysis A General 350 (2008) 16.
Therefore, the oxide carrier controls the catalytic functionality of [6] F. Arena, G. Italiano, K. Barbera, G. Bonura, L. Spadaro, F. Frusteri, Catalysis Today
143 (2009) 80.
the Cu–ZnO system by influencing its steady-state CO2 adsorption [7] G. Bonura, F. Arena, G. Mezzatesta, C. Cannilla, L. Spadaro, F. Frusteri, Catalysis
pattern, as it is evident by comparing kinetic activity data (473 K) at Today 171 (2011) 251.
3.0 and 5.0 MPa normalized to the SA, MSA and OSA of the studied [8] C. Yang, Z. Ma, N. Zhao, W. Wei, T. Hu, Y. Sun, Catalysis Today 115 (2006)
222.
catalysts (Fig. 9). Once more the MSA specific activity as a func-
[9] S.-I. Fujita, M. Usui, H. Ito, N. Takezawa, Journal of Catalysis 157 (1995) 403.
tion of dispersion results in typical decreasing trends, pointing out [10] W. Wang, S. Wang, X. Ma, J. Gong, Chemical Society Reviews 40 (2011) 3703.
an apparent higher specific activity of larger Cu particles [3–5,20]. [11] J. Słoczyński, R. Grabowski, P. Olszewski, A. Kozłowska, J. Stoch, M. Lachowska,
J. Skrzypek, Applied Catalysis A General 310 (2006) 127.
However, it is consequence of assuming a competitive one-site L–H
[12] K.A. Pokrovski, A.T. Bell, Journal of Catalysis 241 (2006) 276.
mechanism, disregarding any contribution of promoter and car- [13] S. Askgaard, J.K. Nørskov, C.V. Ovesen, P. Stoltze, Journal of Catalysis 156 (1995)
rier on the catalytic functionality of the system. In fact, activity 229.
data normalized to the OSA result in constant specific rate val- [14] I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and
Kinetics, WILEY-VCH GmbH & Co. KGaA, Weinheim, 2005.
ues, irrespective of dispersion and catalyst composition (Fig. 9), [15] J. Nakamura, I. Nakamura, T. Uchijima, Y. Kanai, T. Watanabe, M. Saito, T. Fuji-
stressing the fundamental contribution of the oxide carrier on the tani, Journal of Catalysis 160 (1996) 65.
CO2 -hydrogenation functionality of the Cu–ZnO system. Hence, [16] V.E. Ostrovskii, Catalysis Today 77 (2002) 141.
[17] D. Granjean, V. Pelipenko, E.D. Batyrev, J.C. van de Heuvel, A.A. Khassin, T.M.
chemical effects of the oxide carrier [3,5–7,11,12,25] substantiate Yurieva, B.M. Weckhuysen, Journal of Physical Chemistry C 115 (2011) 20175.
the dual-site nature of the CO2 -to-methanol hydrogenation reac- [18] I.A. Fisher, A.T. Bell, Journal of Catalysis 172 (1997) 222.
tion on supported Cu–ZnO systems. [19] Q. Sun, Y.-L. Zhang, H.-Y. Chen, J.-F. Deng, D. Wu, S.-Y. Chen, Journal of Catalysis
167 (1997) 92.
[20] A. Karelovic, A. Bargibant, C. Fernàndez, P. Ruiz, Catalysis Today 197 (2012)
4. Conclusions 109.
[21] Y. Choi, H.G. Stenger, Applied Catalysis B Environmental 38 (2002) 259.
The effects of temperature, pressure and space–velocity on the [22] K. Klier, V. Chatikavanij, R.G. Herman, G.W. Simmons, Journal of Catalysis 74
(1982) 343.
activity–selectivity pattern of Al2 O3 , ZrO2 and CeO2 supported [23] W.H. Cheng, Applied Catalysis A General 130 (1995) 13.
Cu–ZnO systems in the synthesis of methanol via CO2 hydrogena- [24] I.A. Fisher, A.T. Bell, Journal of Catalysis 184 (1999) 357.
tion have been addressed. [25] J.B. Wang, H.-K. Lee, T.-J. Huang, Catalysis Letters 83 (2002) 79.

Please cite this article in press as: F. Arena, et al., How oxide carriers control the catalytic functionality of the Cu–ZnO system in the hydrogenation
of CO2 to methanol, Catal. Today (2013), http://dx.doi.org/10.1016/j.cattod.2013.02.016

You might also like