Catalytic reforming of oxygenated hydrocarbons for hydrogen with low levels of carbon monoxide

Article abstract of DOI:10.1002/anie.200351664

Vaporization of water leads to low levels of CO: Now, it is possible to produce low-CO-containing, H₂-rich streams by simple, aqueous-phase reforming of oxygenates. The schematic of ethylene glycol reforming on a Pt catalyst shows the composition of a gas bubble in the aqueous-phase reactor. Lowering the H₂ and CO₂ pressures, by increasing the water pressure, lowers the CO concentration in the product. WGS = water-gas shift.

Full text of DOI:10.1002/anie.200351664

Communications
mass-derived oxygenated hydrocarbons in liquid water at
temperatures near 500 K. This process has the advantage that
the reforming of oxygenated hydrocarbons and the water–gas
shift (WGS) reaction are both thermodynamically favorable
at the same low temperatures, thus making it possible to
conduct both reactions in one reactor. These low levels of CO,
combined with the fact that our low-temperature process uses
biomass-derived renewable feedstocks, makes aqueous-phase
reforming of oxygenates an attractive process for applications
in portable devices.
Herein, we show results from studies of the aqueous-
phase reforming of ethylene glycol (EG) to produce H2, with
specific emphasis on the levels of CO produced as a function
of the process conditions. We show the importance of the
reaction operating conditions for achieving low levels of CO
in the effluent gas from our single-reactor reforming process,
and we elucidate the factors that control the ultimate levels of
CO that can be attained.
Aqueous-phase reforming of oxygenates to produce H2
takes place on metal catalysts[9] through C C cleavage to
make CO and H2, [Eq. (1)].
ꢀ
C2H6O2 Ð 2 CO þ 3 H2
ð1Þ
ꢀ
(Undesired alkanes are formed through C O cleavage.)
Adsorbed CO undergoes WGS [Eq. (2)], which increases the
Fuel-Cell-Grade H2 from Hydrocarbons
CO þ H2O Ð CO2 þ H2
ð2Þ
Catalytic Reforming of Oxygenated
Hydrocarbons for Hydrogen with Low Levels of
Carbon Monoxide**
amount of H2 produced and removes CO from the catalyst
surface.[10]
The lowest partial pressure of CO that can be achieved
depends on the thermodynamics of the WGS reaction and the
operating conditions, [Eq. (3)].
Rupali R. Davda and James A. Dumesic*
PCO PH
The production of fuel-cell-grade hydrogen, containing very
low levels of CO, is critical for the global transition to a
hydrogen society. In spite of ongoing efforts to make fuel cells
more CO-tolerant,[1] the use of hydrogen fuel-cells in portable
applications (e.g., automobiles, battery replacement devices)
is particularly challenging, as the production of hydrogen by
steam-reforming of hydrocarbons[2] requires a complex com-
bination of multiple processes[3–7] to achieve the required low
CO levels (e.g., 100–500 ppm).[1] An important step toward a
simple process for the production of hydrogen containing low
levels of CO is made possible by the discovery[8] that
hydrogen can be produced by catalytic reforming of bio-
2
2
PCO
¼
ð3Þ
KWGS P
H2
O
KWGS is the equilibrium constant for the vapor-phase
WGS and Pj are partial pressures. Since H2, CO2, and small
amounts of alkanes (primarily CH4) are produced by aque-
ous-phase reforming,[8] gas bubbles are formed within the
liquid-phase flow reactor. The pressure in these bubbles can
be approximated to be equal to the system pressure, [Eq. (4)].
X
Pbubble ꢁ Psystem ¼ PH
þ
Pj
ð4Þ
2 O
products
The partial pressures of the reaction products and water vapor
are dictated by the feed concentrations, system pressure and
temperature, as outlined below.
[*] Prof. Dr. J. A. Dumesic, R. R. Davda
Chemical Engineering Department
Universityof Wisconsin
For dilute product concentrations and system pressures
above the saturation pressure of water, the bubbles contain
water vapor at a pressure equal to its saturation pressure at
the reactor temperature, and the remaining pressure is the
sum of the partial pressures of the product gases. The extent
of vaporization, y, is defined as the percent of water in the
vapor phase relative to the total amount of water flowing into
the reactor. In contrast, for systems operated at pressures that
are near the saturation pressure of water, all the liquid water
Madison, WI 53706 (USA)
Fax: (+1)608-262-5434
E-mail: dumesic@engr.wisc.edu
[**] This work was supported byU.S. Department of Energy(DOE),
Office of Basic EnergySciences, Chemical Sciences Division. We
thank Sibongile Nkosi and Jonathon Tomshine for help in collecting
reaction kinetics data.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
4068 ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200351664
Angew. Chem. Int. Ed. 2003, 42, 4068 –4071

Angewandte
Chemie
may vaporize, and the composition of the bubble is dictated
by the stoichiometry of the feed stream. At this condition, the
partial pressure of water is below its saturation pressure
because the water vapor is diluted by the reforming product
gases. Higher concentrations of EG lead to lower partial
pressures of water because of greater dilution from H2 and
CO2 produced by reforming reactions. As the system pressure
is increased, the partial pressure of water vapor increases until
it reaches the saturation pressure of water, at which point any
further increase in the system pressure leads to partial
condensation of water.
The above arguments indicate that the conditions which
favor the lowest levels of CO from reforming of oxygenates
are those which lead to the lowest partial pressures of H2 and
CO2 in the reforming gas bubbles; and, these conditions are
achieved by operating at system pressures that are near the
saturation pressure of water and at low EG feed concen-
trations. As the system pressure increases and the extent of
vaporization decreases below 100%, the partial pressures of
H2 and CO2 in the bubble increase, thereby leading to higher
equilibrium concentrations of CO. Similarly, as the EG
concentration in the feed increases, higher partial pressures
of H2 and CO2 are developed, even for the case of complete
vaporization, again leading to higher equilibrium CO con-
centrations. This dependence of the CO concentration on the
operating conditions for EG reforming has been corroborated
with detailed experiments, as outlined below.
Table 1 summarizes results for aqueous-phase reforming
of EG at feed concentrations of 2 wt% (runs 1–7), 5 wt%
(runs 8–13) and 10 wt% (run 14). The upflow reactor
(1.27 cm (0.5 inch) stainless-steel tube) containing the Pt/
Al2O3 catalyst was divided into two separately heated
reaction zones. Reforming reactions were carried out in the
lower section (denoted as the reforming zone), maintained at
498 K (see Supporting Information for further details). The
temperature of the top section (denoted as the shift zone),
system pressure, feed concentration, and feed flow rate were
varied in the different runs listed in Table 1. The weight
hourly space velocity (WHSV) of EG, defined as the grams of
EG per gram of catalyst per hour, was maintained such that
complete conversion to gas phase products was achieved for
all runs. The performance of the catalyst was stable for long
periods of time on stream (e.g., 2 weeks). Results from
replicate runs agreed to within ꢂ 10%.
The data in Table 1 show that the dry effluent gas stream
(after condensation of water) for each reaction condition
consists primarily of H2 and CO2. The CO concentration in
the effluent gas and the corresponding equilibrium concen-
tration are reported for each condition. For the 2% and 5%
EG feeds, detailed studies were conducted for system
pressures of 25.8, 32.0 and 36.2 bar, with the shift zone of
the reactor controlled at various temperatures. For the 10%
feed, data were collected at 25.8 bar with the shift zone at
498 K. Higher system pressures lead to lower reaction rates,[11]
resulting in a slightly lower conversion of the 10% EG feed.
As the saturation pressure of water at 498 K is equal to
25.1 bar, liquid water is completely vaporized at a system
pressure of 25.8 bar. For 2% EG at this pressure (run 1), the
H2 pressure in the bubble is calculated to be 0.77 bar, and this
condition leads to a low equilibrium CO concentration of
66 ppm in the reactor effluent. At system pressures of 32 and
36.2 bar (runs 2 and 3, respectively), the H2 pressures are 4.60
and 7.53 bar, respectively, with only 18 and 11% vaporization
of water occurring in each case. These conditions lead to
higher equilibrium CO concentrations of 380 and 617 ppm,
respectively. For the 5% EG feed at a system pressure of
25.8 bar (run 8), complete vaporization leads to a H2 partial
pressure of 1.79 bar, which results in a higher equilibrium CO
concentration (163 ppm) compared to run 1 for the 2% feed.
The effect of increasing system pressure for the 5% feed is
similar to the behavior of 2% EG, as presented in runs 9 and
10 of Table 1.
The effect of H2 pressure on the equilibrium CO concen-
tration is illustrated in Figure 1a and 1b for the 2 and 5%
feeds, respectively. It can be seen that the observed concen-
trations of CO are slightly lower than the calculated
equilibrium values, especially for experiments conducted at
higher system pressures. These lower values of CO concen-
trations are not caused by consumption of CO through
methanation reactions. In particular, we conducted experi-
Table 1: Results for ethylene glycol (EG) reforming over 3% Pt/g-Al2O3.
Run Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
EG Feed Concentration
[wt%]
2
2
2
2
2
2
2
5
5
5
5
5
5
10
WHSV [g/gcat/hr]
Shift Temperature, Ts [K]
System Pressure [bar]
% Vaporization of Water, y 100
H2 Pressure in Vapor Phase 0.77
[bar]
0.0054 0.0054 0.0054 0.0054 0.0054 0.0027 0.0081 0.0135 0.0135 0.0135 0.0135 0.0135 0.0054 0.0270
498
498
32.0
18
498
36.2
11
508
32.0
91
515
36.2
90
498
32.0
16
498
32.0
16
498
25.8
100
1.79
498
32.0
41
498
36.5
22
505
32.0
100
2.20
512
36.2
100
2.33
498
32.0
42
498
25.8
100
3.40
25.8
4.60
7.53
1.05
1.14
4.51
4.66
4.62
7.87
4.63
H2O Pressure in Vapor
Phase [bar]
24.4
25.1
25.1
30.1
34.1
25.1
25.1
22.6
25.1
25.1
28.5
32.4
25.1
20.7
H2 [mol%]
CO2 [mol%]
CH4 [mol%]
C2H6 [mol%]
Experimental CO [ppm]
Equilibrium CO [ppm]
69.8
28.7
1.38
0.16
60
69.6
28.7
1.43
0.25
326
70.0
28.6
1.06
0.32
564
69.1
28.9
1.64
0.28
84
68.9
28.8
2.02
0.36
89
68.3
29.3
1.66
0.32
363
70.4
28.4
0.99
0.23
328
69.9
28.5
1.35
0.20
118
69.7
28.9
1.09
0.26
305
70.9
28.3
0.51
0.27
558
69.3
29.1
1.34
0.27
163
69.0
29.0
1.70
0.35
183
70.0
28.3
1.37
0.28
341
69.6
28.9
1.34
0.14
196
66
380
617
88
95
380
379
163
382
638
185
195
377
341
Angew. Chem. Int. Ed. 2003, 42, 4068 –4071
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 4069

Communications
Figure 2. Effect of WHSV on the% deviation of CO concentration from
equilibrium. The numbers alongside each point correspond to the run
numbers (see Table 1). Triangles: 2% EG (runs 1–3, 6–7); Circles: 5%
EG (runs 8–10, 13); and Squares: 10% EG (run 14).
increase with increasing space velocity, thus suggesting that
these deviations may be caused by reaction kinetics effects
involving the WGS reaction, such as adsorbed CO species
react more rapidly with water to form CO2 and H2, compared
to the slower desorption to form gaseous CO. Aupretre,
et al.[12] have shown that H2 and CO2 were the primary
gaseous products from steam reforming of ethanol, and CO
was produced by the reverse WGS reaction.
Although the lowest levels of CO are obtained when the
reactor is operated near the saturation pressure of water, this
condition is not feasible for larger, nonvolatile feeds such as
sorbitol and glucose, which undergo undesirable decomposi-
tion reactions when vaporized.[13] It thus becomes imperative
to operate the reformer at system pressures above the
saturation pressure of water to maintain liquid phase
conditions. We now propose a new process, which we
denote as ultra shift, to achieve very low levels of CO in the
product gas from aqueous-phase reforming of oxygenated
hydrocarbons, conducted at pressures above the saturation
pressure of water. This process involves the vaporization of
liquid water (in the shift zone) to dilute the H2 and CO2 in the
bubbles, thereby favoring increased conversion of the WGS
reaction and leading to lower CO concentrations.
Experiments were designed to demonstrate the feasibility
of the ultra-shift process, and these data are included in
Table 1. Reforming of a 2% EG feed, conducted at a system
pressure of 32 bar with the shift-zone at 498 K (run 2),
resulted in a CO concentration of 326 ppm. When the
temperature of the shift zone was increased to 508 K
(run 4), 91% of the water vaporized, thereby decreasing the
H2 pressure to 1.05 bar, and lowering the measured CO
concentration to 84 ppm. In the case where the system
pressure was 36.2 bar (runs 3 and 5), increasing the temper-
ature of the shift zone from 498 to 515 K decreased the H2
pressure from 7.53 to 1.14 bar and the measured CO concen-
tration from 564 to 89 ppm. At the higher temperatures, the
effect of diluting the H2 and CO2 pressures with water vapor
supersedes the less favorable thermodynamics, leading to
Figure 1. Effect of H2 pressure on CO concentration for a) 2% EG
(WHSV=0.0054 hꢀ1) and b) 5% EG (WHSV=0.0135 hꢀ1). Each point
corresponds to a shift temperature of 498 K. Open symbols represent
equilibrium CO and filled symbols represent observed CO.
ments in which water was co-fed with a simulated reformer
gas (containing 1000 ppm CO in a 30:70 CO2:H2 mixture)
over the Pt/Al2O3 catalyst at 498 K and a system pressure of
25.8 bar, such that the partial pressures of water and H2 in the
reactor were 21.6 and 2.95 bar, respectively (conditions
similar to run 14). Under these conditions, we did not detect
any measurable amounts of CH4 and the concentration of CO
in this simulated reformer gas was converted to the WGS
equilibrium value of 298 ppm.
[11]
Our previous studies of EG reforming over Pt/Al2O3
have shown that reaction rates are lower at higher system
pressures. Hence, we suggest that lower values of observed
CO concentrations (compared to equilibrium values) at
higher system pressures are caused by lower WGS rates.
Accordingly, we conducted experiments to study the effect of
WHSV on the deviations of the effluent CO concentrations
from the equilibrium values. The WHSV was altered for 2 and
5% feeds at 32 bar (runs 6–7 and run 13) by changing the feed
flow rate through the reactor. Also, for a particular volumet-
ric flow rate, the WHSV is different for the three feed
concentrations. Figure 2 shows the deviation of the observed
CO concentration from the equilibrium value (expressed as a
percentage of the equilibrium value) as a function of WHSV
for runs conducted at 498 K. Higher temperatures (runs 4, 5,
11 and 12) were not included in this plot, because the resulting
higher rates reduced the CO deviations from equilibrium. It is
seen from Figure 2 that the CO deviations from equilibrium
4070 ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 4068 –4071

Angewandte
Chemie
ultra shift. Similar effects were observed for 5% EG, as seen
in runs 9–12.
Figure 3a and 3b shows the effects of ultra shift on the
levels of CO attained from reforming of 2 and 5% EG feeds.
It is seen that when the shift temperature is maintained at the
reforming temperature of 498 K, the effluent concentration of
[1]M. Murthy, M. Esayian, W. Lee, J. W. Van Zee, J. Electrochem.
Soc. 2003, 150, A29.
[2]J. Rostrup-Nielsen, Phys. Chem. Chem. Phys. 2001, 3, 283.
[3]H. Gunardson, Industrial Gases in Petrochemical Processing,
Marcel Dekker, New York, 1998.
[4]G. J. Grashoff, C. E. Pilkington, C. W. Corti, Platinum Met. Rev.
1983, 27, 157.
[5]P. G. Gray, M. I. Petch, Platinum Met. Rev. 2000, 44, 108.
[6]A. N. J. Van Keulen, in US 6,207,307 USA, 2001.
[7]C. S. Gittleman, R. Gupta, in US 2002110503 USA, 2002.
[8]R. D. Cortright, R. R. Davda, J. A. Dumesic, Nature 2002, 418,
964.
[9]R. R. Davda, J. W. Shabaker, G. W. Huber, R. D. Cortright, J. A.
Dumesic, Appl. Catal. B, 2003, 43, 13.
[10]J. Novakova, L. Kubelkova, Appl. Catal. B 1997, 14, 273.
[11]J. W. Shabaker, R. R. Davda, G. W. Huber, R. D. Cortright, J. A.
Dumesic, J. Catal., 2002, 215, 344.
[12]F. Aupretre, C. Descorme, D. Duprez, Catal. Commun. 2002, 3,
263.
[13]B. M. Kabyemela, T. Adschiri, R. M. Malaluan, K. Arai, Ind.
Eng. Chem. Res. 1999, 38, 2888.
Figure 3. Effect of system variables on CO concentration for a) 2% EG
(WHSV=0.0054 hꢀ1) and b) 5% EG (WHSV=0.0135 hꢀ1). Circles: CO
concentration versus system pressure at shift temperature Ts =498 K,
Squares: Ultra shift at higher shift temperatures. Open symbols repre-
sent equilibrium CO and filled symbols represent observed CO.
CO increases as the system pressure is increased. When the
shift temperature is increased such that a majority of the
water vaporizes, then the product CO concentration, even at
the higher pressures, decreases to below 100 ppm for 2% EG
and below 200 ppm for 5% EG. Importantly, when the
effluent from the reactor is subsequently cooled to a lower
temperature to condense water, then the pressure of the
noncondensable gases increases and approaches the system
pressure. This process of ultra shift, involving initial vapor-
ization followed by condensation of water, thus leads to the
desirable production of fuel-cell-grade H2 at high pressures
and containing very low levels of CO.
Received: April 15, 2003 [Z51664]
Keywords: carbon monoxide · fuel cells · heterogeneous
.
catalysis · hydrogen · water–gas shift
Angew. Chem. Int. Ed. 2003, 42, 4068 –4071
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4071