2056
Ind. Eng. Chem. Res. 1994,33, 2056-2065
Production of Hydrogen from Methanol. 2. Experimental Studies Raphael 0. Idem and Narendra N. Bakhshi' Catalysis and Chemical Reaction Engineering Laboratory, Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Saskatchewan, Canada S7N OW0
Various copper-aluminum catalysts with copper concentration ranging from 0 to 27.8 wt ?6 copper were prepared by coprecipitation techniques. These catalysts were calcined at temperatures in the range of 300-700 "C and then reduced in a H2 atmosphere a t 300 "C. Methanol decomposition and steam reforming reactions were studied in a microreactor at atmospheric pressure over a temperature range of 170-250 "C and methanol space velocities (WHSV) of 26.4 and 16.7 h-l, respectively. Methanol conversion was found to be a strong function of catalyst reducibility and copper concentration. Also, reaction efficiency depended strongly on the amount of Cu2O formed in the activated catalyst. In addition, reaction temperature, type of feed, and catalyst characteristics had tremendous effects on H2 yield and selectivity in the production of hydrogen from methanol. Maximum H2 production efficiency of 78 mol ?6 was obtained in the steam reforming reaction a t 250 "C with the catalyst containing 27.8 wt % copper and calcined a t 700 "C. Table 1. Methanol Conversion as a Function of Catalyst Characteristics.
Introduction Hydrogen continues to attract a strong interest for use as an energy carrier in such applications as fuel cells and mobile vehicles (Uehara, 1992) despite difficulties in storage and handling. A number of routes for hydrogen production for various applications have been proposed, and the most promising appears to be the use of methanol as a chemical carrier, mainly due to the advantages derived in methanol storage and transportation (de Bokx et al., 1988; Kliman, 1983; Othmer, 1982) and for environmental reasons (Uehara, 1992; Dupont and Degand, 1986). Hydrogen can be made by catalytic decomposition or steam reforming of methanol. In the steam reforming process, it is known that more hydrogen is produced than is available in methanol alone on account of CO shift taking place in the presence of steam. If one wishes to use methanol as H2 storage, then it is of interest to know the maximum HZ production efficiency and the operating conditions under which this would be obtained. Such information is not available in the literature. In the literature, hydrogen production from methanol has been performed mostly using copper-containing commercial catalysts such as Cu-ZnO/A1203 and Cu/ZnO (Amphlett et al., 1991; Su and Rei, 1991). In recent literature, Su and Rei (1991) have studied both chemisorptive and reaction products of the methanol-water system on NiO/A1203 and CuO/ZnO catalysts in order to develop a better understanding of the effects of catalyst constituents on reaction mechanism in the steam reforming reaction. However, it is still not clear how each metal species in each catalyst affects the catalyst activity. In addition, there are conflicting views in the literature regarding the oxidation state of the active metal species for the methanol-steam reforming reaction (Matsukata et al., 1988; Takezawa et al., 1982; Agaras et al., 1988). In this investigation, we have studied both methanol decomposition and steam reforming reactions at 3 reaction temperatures on 15 coprecipitated Cu-A1 catalysts of various copper concentrations. This work involved a detailed characterization of these catalysts (Idem and Bakhshi, 1994). Also, an attempt was made to sort out the oxidation states of copper which are active for the reactions as well as establish the relationship between catalyst characteristics and catalyst performance in the production of hydrogen from methanol. In addition, performance with respect to time of the best catalyst
copper calcn copper copper reaction concn temp catalyst area dispersion temp Xmd X,, (wt %) (OC) identity (m2/a) (%) ("C) (mol%) (mol %) 5.79 300 p-2A 0 NA 47 250 49 13 200 10 170 2 3 500 p-2B 0 NA 250 61 49 200 16 16 3 170 6 700 p-2C 16.0 250 52 39.5 61 23 200 20 170 8 8 15.3 300 p-3A 30.8 11.6 63 250 75 38 200 30 10 8 170 500 p-3B 33.4 20.4 250 70 79 200 38 33 12 170 12 700 p-3C 69.7 29.5 250 72 80 40 200 36 13 170 13 24.1 300 p-4A 36.5 250 70 79 9.9 53 200 34 170 15 11 500 p-4B 40.2 10.7 80 250 82 200 54 36 170 13 15 700 p-4C 79.0 25.2 81 250 88 200 55 40 170 15 23 27.8 300 p-5A 33.0 250 79 81 9.5 56 200 36 12 15 170 500 p-5B 34.5 250 83 10.0 80 200 57 38 16 170 16 88 13.7 250 83 700 p-5C 47.2 200 59 40 170 19 26 "Reaction pressure = 1 atm; methanol WHSV,d = 26.4 h-1; methanol WHSV, = 16.7 h-l; X,methanolconversion; md, methanol decomposition reaction; sr, steam reforming reaction (1:l molar mixture of methanol and steam as feed).
operating at optimum reaction conditions was examined. These results are presented in this paper.
Experimental Section Catalysts. A total of 15 catalysts were used in this study. Their identities are given in Table 1. Details concerningtheir preparation and characterization are given elsewhere (Idem and Bakhshi, 1994).
0888-588519412633-2056$04.50/0 0 1994 American Chemical Society
Ind. Eng. Chem. Res., Vol. 33, No. 9,1994 2067
A
20
Figure 1. Schematic diagram of experimental rig for the production of hydrogen from methanol. 1, N2;2,5% H2 in Nz; 3, pressure gauge; 4, on/off valves; 5, feed reservoir; 6,syringe pump; 7, deoxygenation/dehydrationcatalyst; 8, needle valve; 9, vaporizer; 10,temperature indicator; 11, temperature controller; 12, fixed bed reactor; 13,catalyst bed; 14, heating block; 15, electrically heated furnace; 16, insulator; 17, water cooled condenser; 18, gae/liquid separator; 19, gaseous produds, 20,liquid products; 21, gas chromatograph; 22,fume hood.
Experimental Studies. A. Equipment. Figure 1 shows the schematic diagram of the experimental rig used. Catalyst performance was studied in a stainless steel (SS316),microreactor (10-mmi.d. and 460-mmoverall length) placed in an electrically heated furnace. The furnace temperature was controlled by a Series SR22 microprocessor-based autotuning PID temperature controller (supplied by Shimaden Co. Ltd., Tokyo, Japan) through a K-type thermocouple inserted into a heating block placed concentrically within the furnace. A separate thermocouple was used to monitor the temperature of the catalyst bed. This arrangement was capable of ensuring an accuracy of fl "C for the catalyst bed temperature. B. Operating Conditions. All runs were conducted at atmospheric pressure at reaction temperatures in the range of 170-250 "C. Each run was performed using 2 g of calcined catalyst mixed with 3 g of Pyrex glass all in the size range of -8 to +10 mesh. This mixture was loaded on a stainless steel grid positioned centrally within the reactor. This arrangement produced a catalyst bed height to particle diameter ratio (LID,)greater than 40. This was done to ensure a condition as close as possible to plug flow. The flow rate of the liquid feed was regulated by a syringe pump (Eldex Model Bl00 S). A typical run for both methanol decomposition (methanol WHSV of 26.4 h-l) and methanol-steam reforming (methanol WHSV of 16.7 h-l and feed consisting of a 1:l molar mixture of methanol and water) was performed as follows: the catalyst was loaded in the reactor and reduced in situ at 300 "C with H2 a t a flow rate of 100 mL/min for 2 h. The feed was then pumped at the desired methanol space velocity to the vaporizer maintained at about 250
"C. The vaporized feed from the vaporizer entered the reactor in a stream of N2 gas (99.995% purity; obtained from Linde). Methanol decomposition is an endothermic reaction. Therefore, it was necessary to stabilize the reactor temperature before actual reaction data were taken. For example, in order to make a test run at 250 "C the reactor was initially heated to a higher temperature and then allowed to stabilize at 250"C. This stabilization procedure took 1h. The initial part of this stabilization period was the time necessary for the reduced catalysts to undergo mild reoxidation (induction period). This took about 20 min. All reaction data in this paper were taken after the reactor was stabilized a t the desired reaction temperature. The product mixture during both activation and actual reaction was condensed with chilled water to separate gaseous and liquid products. However, onlyproducts from actual reactions were analyzed. C. Analysis of Products. The gaseous product was analyzed on-line with a gas chromatogragh (Model HP 5880)using a Porapak Q column in series with a Porapak R column, a thermal conductivity detector (TCD), and helium (99.995% purity; obtained from Linde) as the carrier gas. The liquid product was analyzed by manual injection into the same set of columns using the same gas chromatograph and analysis conditions as in the gaseous products. D. Identification of the Oxidation State of Copper during Reaction. The catalysts were characterized in order to identify the oxidation state of copper during both methanol decomposition and steam reforming reactions. The procedure consisted of activating each catalyst by
2058 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994 100
100
md (250 o C )
-ES
80
.-
60
sr (250 oC)
g
2
e,
sr (250
60
OC)
t
t
md (200 oC:
e
8
/-9 A
40
40
I
I
0
0
e
md (170 oC) sr (170 oC)
E
m
c,
md (250 oC)
0
.I e,
L
8
>
80
E^
;.
sr (200 oC)
v)
e
-8
20
m
2o
sr (200 oC)
"a md (200
md U sr
0
0 0
10
20
30
10
40
reduction in Hz at 300 "C for 2 h followed by reoxidation at 250 "C in the vaporized feed stream for 1 h and then allowingit to cool to room temperature in flowingnitrogen. The catalyst was then analyzed by the temperatureprogrammed-reduction (TPR) technique. Details of the TPR procedure are given elsewhere (Idem and Bakhshi, 1994).
Results and Discussion Each of the 15 coprecipitated copper-aluminum catalysts was screened in the microreactor for both methanol decomposition and steam reforming reactions in order to obtain understanding of the relationship between catalyst performance and catalyst characteristics. Catalyst Activity. Catalyst activity was evaluated in terms of methanol conversion (mol %) for both methanol decomposition and steam reforming reactions. Characteristics of coprecipitated Cu-Al catalysts together with the corresponding methanol conversions at three reaction temperatures (170,200, and 250 "C) are given in Table 1. Figure 2 shows a typical behavior of methanol conversion with catalyst copper concentration. The conversions in the figure were obtained using catalysts calcined a t 700 "C. Itisseen thatinitiallyat 250and200"Cbothmethanol decomposition and steam reforming reactions showed an increase in methanol conversion with catalyst copper concentration. Beyond 24.1 wt % copper, the increase in methanol conversion was not significant and a plateau was reached in the methanol decomposition reaction. At 170 "C, both reactions showed a slight increase in methanol conversion with copper concentration in the range of 5.79-15.3 wt 5% copper. However, a more significant increase in methanol conversion with copper concentration was obtained between 15.3 and 27.8 w t 5% copper for methanol decomposition reaction, and between 24.1 and 27.8 w t 5% copper in the steam reforming reaction. Our results are in general agreement with those of Matsukata et al. (1988). These workers observed a maximum in the variation of methanol conversion with copper concentration at about 38 wt 9% copper in the steam reforming reaction conducted a t 250 "C. However, it is not clear from their work whether the copper concentrations were based on dried catalyst precursors or calcined catalysts. They did not study the methanol decomposition reaction by itself.
(170 oC)
I
I
20
30
40
Copper Dispersion, %
Catalyst copper concentration, wt % Figure 2. Variation of methanol conversion with catalyst copper concentration. md, methanol decomposition reaction (WHSV = 26.4 h-1); sr, steam reforming reaction (methanol WHSV = 16.7 h-' and feed consistingof a 1:1 molar mixture ofmethanol and steam);reaction pressure = 1 atm; catalysts calcined at 700 "C.
-
OC)
Figure 3. Variation of methanol conversion with catalyst copper dispersion. md, methanol decomposition reaction (WHSV = 26.4 h-l); sr, steam reforming reaction (WHSV = 16.7 h-1 and feed consisting of a 1:l molar mixture of methanol and steam); reaction pressure = 1 atm. 6
I
sr (200 oC) md (200 o C ) (170 oC)
sr
b "
I
10
I
I
I
20
30
40
Copper dispersion, % Figure 4. Variation of turnover number for methanol conversion with copper dispersion.md, methanol decompositionreaction (WHSV = 26.4 h-l); sr, steam reforming reaction (WHSV = 16.7 h-1 and feed consisting of a 1:l molar mixture of methanol and steam); reaction pressure = 1 atm; ordinate multiplied by 1029.
Dispersion of the catalytically active metal also appeared to affect the reaction products. A typical relationship between methanol conversion and copper dispersion using catalysts containing 15.3 wt % copper and calcined at 300,500, and 700 "C is shown in Figure 3. These results alone indicate that the increase in methanol conversion with copper dispersion was not very significant. Matsukata et al. (1988), who examined the relationship between methanol conversion and calcination temperature of the catalysts, observed a significant increase. According to these authors, the higher the calcination temperature, the greater the dispersion of copper even though they did not determine any copper dispersion. Calcination at a high temperature produces catalysts having highly dispersed copper crystallites and large metallic copper area. Our results indicate that not all the available copper area is utilized effectively as active sites for the reactions. This can be seen clearly in Figure 4, which shows a typical relationship between turnover number (TON) for methanol conversion and copper
Table 2. Turnover Number for Methanol Conversion as a Function of Copper Dispersion. TON
copper concn (wt % ) 5.79 15.3 24.1 27.8
copper dispersion ( % ) 39.5 29.5 25.2 13.7
[(mol %/coppeiitom) x 10-231 md sr 5.98 7.24 1.79 1.61 1.53 1.39 2.13 2.01
%
e-
TON, turnover number for methanol conversion;md, methanol decomposition reaction; sr, steam reforming reaction; WHSV,d = 26.4 h-1; methanol WHSV,, = 16.7 h-l; reaction temperature = 250 'C; reaction pressure = 1 atm; catalysts calcined at 700 OC. (I
dispersion for catalysts containing 15.3 wt % copper and calcined at 300, 500, and 700 "C. Figure 4 shows that TON for methanol conversion decreased as the copper dispersion increased. This may be explained on the basis of our earlier results (Idem and Bakhshi, 1994) that the increase in copper area and dispersion obtained by calcination at high temperatures is attributed to the formation of large amounts of CuA1204 and the metallic copper obtained from its reduction will still be retained in the spinel structure. The TON results show that this metallic copper is less efficient for methanol conversion than the metallic copper obtained from the reduction of other species such as CuO, Cu20, and Cug(OH)2C03. Table 2 shows the variation of TON for methanol conversion with copper dispersion for catalysts containing 5.79, 15.3, 24.1, and 27.8 wt %. These catalysts were calcined at 700 "C. It is seen from the table that TON decreased rapidly as the copper dispersion decreased from 39.5% (i.e., 5.79 wt % Cu) to 29.5% (i.e., 15.3 wt 7% Cu). At lower copper dispersions (i.e., higher copper concentrations), the decrease in TON was not very significant and a minimum was reached at a copper dispersion of 25.2 % (i.e., 24.1 w t % Cu). Also, it is seenin Table 1that methanol conversion increased with available copper area (i.e., copper concentration) up to the catalyst containing 24.1 wt 9% copper which also had the highest copper area. Although a higher copper concentration (27.8 w t % Cu) resulted in a reduction in copper area, methanol conversion obtained from this catalyst was higher. These results show that even though the available copper area and copper dispersion play important roles in the reactions, the controlling factors are the amount of copper in the catalyst and the calcination temperature which control the type of copper species that will be present in the calcined catalyst. It was interesting to observe that catalyst reducibility also affected methanol conversion. The minimum temperature at which the reduction of a catalyst can commence in a reducing atmosphere is usually taken as a measure of the catalyst reducibility (Klisurski, 1970). The lower this temperature, the greater the catalyst reducibility. In this study, reducibility was evaluated in terms of lowest TPR peak temperature. Such typical behavior between methanol conversion and catalyst reducibility is illustrated using Figures 2 and 5. Figure 2 shows the relationship between methanol conversion and catalyst copper concentration for catalysts calcined at 700 OC while Figure 5 shows the variation of catalyst reducibility with copper concentration. It is seen that methanol conversion increased as catalyst reducibility increased. Catalyst reducibility seems to provide a more useful parameter than copper concentration alone because it takes into account the effects of both copper concentration and calcination temperature. Klisurski (1970) has reported a similar relationship between methanol conversion and catalyst reducibility. However, this worker used only single oxide catalysts such as CuO, Mn203, C0304,and TiO2.
lv1 . YI
c
3
k E *
40
-
$
30-
i4
0
:
-wa
100
I
200
300
400
Reduction temperature, OC Figure 6. TPR spectra of activated catalysts. a, 5.79 w t % Cu and calcined at 700 "C;b, 15.3 wt % Cu and calcined a t 300 O C ; c, 24.1 w t % Cu and calcined at 700 O C ; d, 27.8 w t 5% Cu and calcined at 700 O C . 1, Methanol decomposition reaction; 2, steam reforming reaction.
A. Oxidation State of Copper during Reaction. A TPR analysis of all the activated catalysts was performed in order to determine the oxidation state of the active species during methanol decomposition and reforming. Typical results are presented in Figure 6 for the coprecipitated Cu-A1 catalyst system. Only a single characteristic peak of Cu20 was observed at about 210 OC. The amount of copper reoxidized to Cu20 was calculated from H2 consumption in the TPR experiments. On the other hand, the amount of total metallic copper present on the catalyst was determined earlier (Idem and Bakhshi, 1994) by isothermal reduction in hydrogen at 300 OC. The results showed that only a fraction of the reduced copper was reoxidized to Cu20 for all catalysts. The balance was metallic copper. These results show that the activated catalysts contained both Cuo and Cu' species. It is extremely important to know the composition of Cu'Jand CUIin the catalyst because catalyst activity depended on this characteristic. Methanol conversion increased with the amount of Cu2O species in the catalyst. This can be seen in Figure 7, which showsa typical relationship between
2060 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994
I
01
0.02
0.06 Amount of Cu20, g/g
0.10
activated catalyst
Figure 7. Variation of methanol conversion with amount of Cu20 in activated catalysts for the steam reforming reaction.
methanol conversion and the amount of Cup0 in the activated catalyst using catalysts p-2C, p-3A, p-4C, and p-5C for the steam reforming reaction for all three reaction temperatures. This type of information has not been reported previously in the literature. However, similar type of work concerning Cu/ZnO catalysts for methanol synthesis has been reported by Chinchen and Waugh (1986). The role of CUI and CuO species in methanol decomposition is discussed below. Methanol decomposition may occur according to two parallel reactions, i.e., H2 production (methanol dehydrogenation, eq 1) and water production (methanol dehydration, eq 2). Even though it is indicated (Amphlett et al., 1991; Su and Rei, 1991; Isogai et al., 1989) that in the case of copper-based catalysts copper species are responsible for methanol dehydrogenation, no information is available on the distribution of copper oxidation states for this reaction. In this study, an attempt was made to determine these oxidation states. It has been mentioned earlier that an induction period of about 20 min was needed before the catalysts became active. This observation suggests that the CuO species present at the start of the reaction has a low activity. However, during the induction period, a fraction of Cuo was reoxidized to Cu20 as was indicated earlier by the TPR analysis of activated catalysts (Figure 6). The initial reaction of methanol and steam on the catalyst produced CO2 which was responsible for the reoxidation of Cuo to CUI. It would appear therefore that the catalysts became active only when both Cuo and Cul were present. Therefore, the induction period could be taken as the time required to produce an appropriate CUO-CU~O combination that is active for the reaction. Klier et al. (1982), during the methanol synthesis reaction from synthesis gas on CuO/ZnO catalysts, have made a similar observation that an induction period of about 20 min was necessary for Cuo and Cul species to be formed in order for this reaction to take place. Similarly, Petrini et al. (1983) observed an induction period in the CO shift reaction to H2 and C02 using CuO/ZnO catalysts. In order to determine the species responsible for methanol dehydration (eq 2), test runs were carried out with feeds consisting of methanol alone and a 1:l molar mixture of methanol and water over catalysts p-1B (0 wt '3 copper) and p-5B (27.8 wt '3 copper). These runs were conducted at 250 "C with the catalysts calcined at 500 "C.
When methanol alone was passed over catalyst p-1B (i.e., A1203only), only dehydration products dimethyl ether (CH30CH3) and HzO were obtained. This shows that catalyst p 1 B possessed active sites responsible for methanol dehydration. It also indicates that these sites do not promote methanol dehydrogenation. On the other hand, when a 1:lmolar mixture of methanol and water was passed over this catalyst, no measurable methanol conversion was observed indicating that the active sites on A1203which are responsible for methanol dehydration are deactivated due to the presence of water in the feed. This catalyst did not yield any dehydrogenation products. Similar tests were conducted on catalyst p 5 B (27.8 wt '3 copper). In this case, the catalyst contained copper. When methanol alone was passed over this catalyst, dehydrogenation products (Le., H2, CO, etc.) were most predominant (see Table 3a). However, with a feed consisting of a 1:l molar mixture of methanol and water, dehydration products dimethyl ether and water were totally absent (Table 3b). These results show that methanol dehydration reaction was catalyzed by active sites on alumina alone, whereas methanol dehydrogenation reaction was catalyzed by copper species. Also, water in the feed poisoned the active sites on alumina. Thus, no methanol dehydration reaction took place. A similar poisoning effect by water was observed by Figoli et al. (1971),who studied the poisoning actions of acids and bases on the activity of alumina for methanol dehydration. According to these authors, water poisoned the active sites on alumina responsible for methanol dehydration thereby rendering them inactive for reactions in which water was formed as a reaction product. Product Distribution. The products obtained from methanol decompositionreaction using coprecipitated CuA1 catalysts were H2, CO, CH4, C02, dimethyl ether (CH3OCH3),and methyl formate (HCOOCH3). In the case of methanol-steam reforming reaction, only H2, CO, Cop, and methyl formate were obtained. The following system of reactions was formulated to account for these products: 2CH,OH = HCOOCH,
(1)
2CH,OH = CH,OCH,
+ H2 + H,O
(2)
HCOOCH, = CO + CH30H
(3)
+ CH, = CO, + H,
HCOOCH, = CO, CO + H,O
(4)
(5)
This system is in generalagreement with the one proposed by Kobayashi et al. (1976). However, the methanol dehydration reaction (eq 2) was not included by these workers. In the present study, the effects of catalyst characteristics and reaction temperature on product distribution have been evaluated using these reactions. The product distribution results are presented in parts a and b of Table 3 as a function of reaction temperature and copper concentration for methanol decomposition and steam reforming reactions, respectively. Even though catalyst characteristics play an important role in bringing about any reaction, the effects of such characteristics on methanol decomposition and steam reforming reactions have not been discussed in the literature. In this study, catalyst characteristics showed interesting effects on some of the products formed during
Ind. Eng. Chem. Res., Vol. 33, No. 9,1994 2061 Table 3. Product Distribution as a Function of Copper Concentration and Calcination Temperature. (a) Methanol Decomposition Reaction (WHSV = 26.4 h-1) product distribution (mol %) copper concn (wt %) calcn temp (OC) reaction temp ("C) Hz CO COZ CH, CHsOCHs 250 13.0 300 67.3 3.1 0.3 2.8 5.79 67.4 1.0 0.1 0 200 3.5 0 51.3 0 170 3.6 0 250 13.2 3.1 0.7 2.4 500 67.6 0 0 0 68.6 200 4.1 0 0 0 55.4 170 3.0 2.9 2.1 0.8 700 68.7 250 13.7 0 0 0 70.0 200 2.9 0 0 0 65.3 170 0.3 5.3 0.9 64.1 19.0 300 6.2 250 15.3 0.1 0 0.1 69.7 200 2.6 0 0 0 64.4 170 0.2 5.6 65.2 18.4 0.9 4.7 250 500 0 0.1 0.1 70.8 200 2.6 0 0 0 60.5 170 2.0 5.5 1.4 4.1 65.4 23.4 700 250 0.1 0 2.4 0.1 69.7 200 0 0 0 2.3 57.9 170 8.3 0.6 7.7 64.7 250 17.2 300 24.1 0 0.1 0.1 69.2 2.5 200 0 0 0 60.9 2.0 170 8.1 0.6 7.5 65.3 250 17.0 500 0 0 0 2.1 70.0 200 0 0 0 62.2 170 2.0 0.7 7.3 17.3 8.0 65.9 250 700 0 0.1 0.1 70.3 200 6.1 1.1 0 0 0 68.2 170 16.6 8.2 0.6 7.6 63.7 250 300 27.8 0 0 0 2.8 72.9 200 2.1 0 0 0 62.7 170 7.7 15.6 8.2 0.5 66.6 250 500 0 0 0 2.4 72.9 200 0 0 0 0.3 64.0 170 17.0 8.3 0.7 7.6 66.2 250 700 0 0 0 5.5 75.0 200 0 0 0 1.7 73.8 170
comer concn (wt % ) 5.79
15.3
24.1
27.8
0
Reaction pressure = 1 atm.
HCOOCH; 13.5 28.9 45.1 13.0 27.3 41.6 11.8 27.1 34.4 4.5 27.5 35.4 5.2 27.4 37.5 0.2 27.7 39.8 1.5 28.1 37.1 1.5 27.9 35.8 0.8 23.4 30.7 3.3 24.3 35.2 1.4 24.7 35.7 0.2 19.5 24.5
(b) Steam Reforming Reaction (Methanol WHSV = 16.7 h-9 product distribution (mol %) calcn temp (OC) reaction temp ("C) Hz coz co HCOOCHa 74.2 24.8 1.0 0 250 300 24.9 74.6 0.3 0.2 200 63.0 0 16.0 21.0 170 74.1 1.2 0 24.7 250 500 74.9 0.1 0 25.0 200 74.1 24.7 0 1.2 170 74.4 24.8 0 0.8 250 700 0.1 0 25.0 74.9 200 4.0 24.0 0 72.0 170 1.5 24.6 0 73.9 250 300 75.0 25.0 0 0 200 74.2 24.7 1.1 0 170 24.7 1.2 0 74.1 250 500 74.9 24.9 0.2 0 200 25.0 0 0 170 75.0 24.5 1.9 0 250 73.6 700 24.9 0.2 0 74.9 200 75.0 25.0 0 0 170 73.9 24.6 1.5 0 250 300 74.9 24.9 0.2 0 200 25.0 75.0 0 0 170 74.1 24.7 1.2 0 250 500 74.9 24.9 0.2 0 200 75.0 25.0 0 0 170 24.7 1.3 0 250 74.0 700 24.9 0.2 0 74.9 200 25.0 0 0 75.0 170 24.6 1.7 0 73.7 250 300 0 0 75.0 25.0 200 0 0 75.0 25.0 170 1.2 0 74.1 24.7 250 500 0.5 0 24.9 74.6 200 0 0 25.0 170 75.0 1.5 0 24.6 73.9 250 700 0.1 0 74.9 24.9 200 0 0 75.0 25.0 170
2062 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994
220
260
Low "3
300
340
TPD peak temperature, OC
Figure 8. Variation of concentration of dimethyl ether in the overall product with low NHs TPD peak temperature of catalyst in methanol decompositionreaction. Catalysts 1,p-5B; 2, p-4B; 3, p-3B; 4, p-2B; reaction temperature = 250 OC, reaction pressure = 1 atm.
methanol decomposition and steam reforming reactions. These are discussed below. A. Effects of Catalyst Characteristicson Methanol Decomposition. One of the major reaction products, namely, dimethyl ether (CHsOCHs), obtained from eq 2 was strongly influenced by the acidity of the active sites of the catalyst. Two types of ammonia TPD peaks were identified earlier (Idem and Bakhshi, 1994) for each coprecipitated Cu-A1 catalyst: the low-temperature peak representing the weak acid sites and the high-temperature peak representing the strong acid sites. Also, it was shown that the low ammonia desorption peak temperature (weak acid site) decreased with increasing copper concentration. Figure 8 shows the low ammonia desorption peak temperatures (weak acid site) for catalysts p-2B, p-3B, p-4B, and p-5B and the corresponding concentration of dimethyl ether (CH30CH3) in the overall product when these catalysts were used in the methanol decomposition reaction. It is seen that dimethyl ether concentration increased with a decreasing strength of the acid sites. This is typical of the relationship between the concentration of dimethyl ether in the overall product and the low ammonia desorption temperature. A t the relatively low reaction temperatures used in this study, methanol which chemisorbs strongly on strong acid sites cannot desorb to permit the formation of dehydration products. Figoli et al. (1971) has attributed methanol dehydration (eq 2) to weak alumina acid sites. Equation 2 shows that water is formed under the conditions where dimethyl ether is formed. This water then reacts with CO according to reaction 5 to give more Cog. This explains why the relationship between the concentration of COz in the product and catalyst copper concentration followed a trend similar to that exhibited by dimethyl ether product (Table 3a). On the other hand, the concentration of methyl formate in the product generally decreased with increasing catalyst copper concentration and copper dispersion (Table 3a). B. Effect of Catalyst Characteristics on Steam ReformingReaction. Since the objective is to maximize Hz production, it is important to determine what catalyst characteristics will minimize the formation of undesired products. Table 3b showsthat the main undesired product in the steam reforming reaction, methyl formate, was produced only with catalysts p-2A, p-2B, and p-2C containing very low amounts of copper (5.79 wt % ), and
catalyst p-3A containing a relatively high amount of copper (15.3 wt %) but calcined at a low temperature (300 "C). Methyl formate was not produced with catalysts such as p-3B and p3C, which contained the same amount of copper as catalyst p-3A (15.3 wt 5%) but calcined at higher temperatures, and the other catalysts containing higher amounts of copper. As was discussed earlier, increase in catalyst calcination temperature results in increase in copper dispersion. Also, increase in both copper concentration and calcination temperature results in increase in catalyst reducibility (Figure 5). These results show that larger Hz production is favored with catalysts of high copper concentration and calcination temperature (Le., catalyst with high reducibility and copper dispersion). Effects of Reaction Conditions on Product Distribution. A theoretical examination of thermodynamic equilibria involved in the catalytic steam reforming of methanol has been carried out by Amphlett et al. (1981). However, experimental study involving Catalystsat various reaction conditions is necessary in order to determine the actual production of H2. We have determined experimentally (see Table 3) the product distribution obtained over 15coprecipitated Cu-Al catalysts at various operating conditions for methanol decomposition and steam reforming reactions. This is discussed below. A. Effects of Reaction Temperature on Methanol Decomposition Reaction. Table 3a shows that with coprecipitated Cu-Al catalysts dimethyl ether product was obtained only at the highest reaction temperature used (250 "C). Yakerson et al. (1967) studied the TPD of methanol on alumina and found major peaks a t 135,190, and 370 "C, respectively. These workers observed that methanoldesorbed unchanged in the 135 "C peak whereas dehydration products were responsible for the higher temperature peaks. These results indicate that temperature has a strong effect on the formation of dimethyl ether and consequently on HZselectivity. Also, Table 3a shows that CO, COZ, and methane concentrations in the overall product increased with reaction temperature. This behavior is typical of products from endothermic reactions (eqs 3 and 4). This therefore suggests an increasing rate of methyl formate decomposition with reaction temperature and, consequently, a decrease in the concentration of methyl formate in the product with increasing reaction temperature. In addition, it is seen from Table 3a that, in the case of methanol decomposition reaction, the amount of CO formed was greater than the corresponding amount of methane at any reaction temperature, with methane being formed only at 250 "C. These results can be attributed to the presence of copper species in the catalysts and the relatively low reaction temperatures, both of which favored reaction 3 in preference to reaction 4 (de Bokx et al., 1988; Agaras et al., 1988; Isogai et al., 1989; Su and Rei, 1991). On the other hand, the behavior of the concentration of Hz in the overall product with reaction temperature shows that a maximum exists. This may be explained as follows: Although high reaction temperature is favorable for high Hz yield (endothermic reaction), it has a detrimental effect on Hz concentration because of the formation of increased amounts of dimethyl ether, carbon monoxide, carbon dioxide and methane at high reaction temperatures (eqs 2-4). Such typical relationship between Hz concentration in the product and reaction temperature is shown in Figure 9. The net result is that maximum Hz concentration in the product is obtained at the intermediate reaction temperature of 200 "C. B. Effects of Reaction Temperature on the Steam Reforming Reaction. Table 3b shows that COOand HZ concentrations in the product exhibited a maximum with
Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994 2063 Table 4. Hydrogen Production Efficiencies of Coprecipitated Cu-A1 Catalyts. ~
copper calcn concn temp (w %) (OC) 5.79 300 500 700
II
I
15.3
300 500
160
180
200
220
240
260
Reaction temperature, OC Figure 9. Variation of HZconcentration in the overall product with reaction temperature in methanol decomposition reaction.
reaction temperature for the catalysts on which methyl formate was also produced. These were catalysts p-2A, p-2B, and p-2C containing low copper concentration (5.79 w t % ) and catalyst p-3A containing a relatively high copper concentration (15.3 wt %) but calcined at a low temperature as was discussed previously. The formation of methyl formate product at low reaction temperature (170 "C) was responsible for the low C02 and Hzconcentrations whereas the low C02 and H2 concentrations at high temperature could be attributed to the formation of the CO product. For these catalysts, maximum H2 concentration in the overall product is obtained a t the intermediate reaction temperature (200 "C). On the other hand, catalysts p-3B and p-3C containing 15.3 wt 5% copper and calcined at 500 and 700 "C, respectively, and the other catalysts with higher copper concentration showed a decrease in the concentrations of C02 and H2 in the product with increasing reaction temperature. This decrease in C02 and Hz concentrations resulted from the formation of increasing amounts of CO with reaction temperature. At 170 "C, no CO was produced on account of it being consumed in the CO shift reaction (exothermic reaction). No methyl formate was produced a t 250 "C for all catalysts. Compared with the Hz selectivity (5') of 35-84 mol % obtained from these latter catalysts in the methanol decomposition reaction, an extremely high selectivity (S > 90 mol % ) was obtained in the steam reforming reaction at the three reaction temperatures used in this study (see Table 4). These results indicate that, within this temperature range and using these latter catalysts, maximum H2 production is not affected significantly by H2 selectivity in the steam reforming reaction. It is seen from Table 3 that while larger amounts of methyl formate were formed in the methanol decomposition reaction, virtually no methyl formate was formed under corresponding reaction conditions in the steam reforming reaction. Kobayashi et al. (1976) made similar observations. Also, Table 3 shows that methane and dimethyl ether products were obtained only in the methanol decomposition reaction. As was discussed earlier, these results were attributed to a shift in thermodynamic equilibrium of reaction 4 and the poisoning effects ofWriter 6~ alumina active sites, respectively. These results show clearly that the presence of water in the feed is necessary for maximum H2 selectivity. These experimental results on the effects of reaction conditions on product distribution are in general agreement with the
700 24.1
300 500 700
27.8
300 500 700
reaction yield efficient (mol of H reaction (mol % mol ofCH& temp ('C) R b RE, ymd Ym 250 32 47 0.69 1.41 200 0.14 10 7 0.29 170 1 1 0.02 0.02 250 34 0.73 58 1.75 200 7 0.13 16 0.48 2 0.04 170 3 0.08 0.79 250 40 59 1.77 200 12 0.34 23 0.69 170 4 0.08 6 0.19 250 1.05 53 59 1.78 200 18 38 1.14 0.36 0.09 170 5 8 0.22 250 1.15 58 67 2.00 21 0.49 200 38 1.13 170 12 5 0.09 0.36 1.34 250 67 67 2.01 200 22 0.43 40 1.19 170 13 5 0.09 0.39 250 1.15 66 58 1.98 200 0.40 53 20 1.58 4 0.09 170 15 0.45 250 1.23 76 61 2.29 0.43 200 54 22 1.61 170 15 5 0.11 0.45 250 77 69 1.37 2.31 200 55 26 0.52 1.64 170 15 13 0.25 0.45 250 74 55 1.09 2.22 200 56 26 0.51 1.68 170 15 5 0.10 0.45 250 76 65 2.29 1.30 200 56 27 0.53 1.68 170 16 7 0.14 0.48 250 70 1.40 78 2.35 200 59 34 0.67 1.76 170 19 19 0.38 0.57
&
selectivit (mol %
d)
Smd
Sw
68.7 54.7 27.4 70.7 40.6 32.1 75.3 61.3 47.3 70.1 60.3 45.4 72.9 63.7 39.3 84.1 60.1 35.4 72.7 58.7 40.0 74.7 60.5 42.3 78.1 66.2 54.6 67.5 70.9 43.2 78.4 70.4 44.6 79.6 84.3 72.8
96.1 97.3 39.6 95.4 99.6 91.1 96.9 99.6 75.0 94.3 100 91.8 96.4 99.2 100 93.0 99.6 100 94.3 99.2 100.0 95.4 99. 100 95.0 99.2 100 93.5 100 100 95.4 98.0 100 94.3 99.6 100
md, methanol decomposition reaction; sr, steam reforming reaction (feedconsists of a 1:l molar mixture of methanol and steam); reaction pressure = 1 atm.
theoretical work of Amphlett et al. (1981). However, products such as methyl formate and dimethyl ether were not included in their analysis. On the other hand, we did not obtain coke as a product probably due to the relatively low reaction temperatures used in this study, although it was indicated as a possible product at high pressures and temperatures in their analysis. Hydrogen Production Efficiency. If we desire to use methanol as Hz storage, it will be desireable to know the maximum amount of H2 that can be made by a catalyst under various operating conditions. This has been designated as "hydrogen production efficiency". This type of information for both methanol decomposition and steam reforming reactions has not been discussed previously in the literature. Hydrogen production efficiency was evaluated in terms of H2 yield (Y) and selectivity (S)to account for the quantity and quality of H2 production, respectively. In addition, hydrogen production efficiencywas evaluated in terms of reaction efficiency (RE)in order to take catalyst activity into consideration. Ha production efficienciesfor various catalysts are given in Table 4. As can be seen, hydrogen yields, Y, obtained from steam reforming reaction were larger than those from methanol decomposition reaction. Also, all the catalysts showed superior H2 selectivity, S, for the steam reforming reaction than for the methanol decomposition reaction. As was discussed previously for the steam reforming reaction, unwanted reactions such as methanol dehydration and methane formation were suppressed due to acid site poisoning and thermodynamic
2064 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994 Table 5. Comparison of Performance of Catalysts Used by Various Workers
X (mol %) reference present study
catalyst p-5c
Amphlett et al.6 (1991) Mataukata et al. (1988) Su and Rei (1991) Cheng et al. (1989)
C18HC catalyst cu/Al2os CuO/ZnO Cu-Cr-Mn-Si
WIFM~ 115.6" 72.v 146.2" 12P 51.84" 156ad
T ('0
P (atm)
250
1
250 250 280 250
1 1 1 2
X,d 88
S (mol % )
x m r
srad
m sr
83
80
94 lW
806 65 58
91
100 92
g of cat min/mol CHaOH for steam reforming reaction. Calculated from approximate rate equation given in reference assuming 100% H2 selectivity. Assumed. g of cat minlmol CH30H for methanol decomposition reaction.
shift, respectively, whereas desired reactions such as methyl formate decomposition and CO shift reactions were facilitated. Also, Table 4 shows that reaction efficiency, RE, was superior in the steam reforming reaction than in the methanol decomposition reaction. Reaction efficiency, RE, is a more useful parameter than Hz selectivity alone, since it takes into account the effect of catalyst activity also. Superior catalyst activity in the steam reforming process was attributed to the formation of an optimum CUO-CUZOcombination on the catalysts as discussed below. As was shown earlier, activated catalysts contained both CuOand CuzO species suggestingthat in order for methanol dehydrogenation to occur a mixture of Cuo and Cu' was needed. Interestingly, Klier et al. (19821, Chinchen and Waugh (1986), and Ghiotti and Boccuzzi (1987) also observed a similar type of behavior for CuO/ZnO catalysts though the reaction under study was the formation of methanol from CO, H2, and COz. While Klier et al. (1982) observed that a maximum existed in the variation of the rate of methanol synthesis with the COz/CO ratio used in the reaction, Chinchen and Waugh (1986) and Ghiotti and Boccuzzi (1987) showed that the proportion of CuaO formed by reoxidation of metallic copper increased with both CO2/CO and HzO/Hz ratios. In this study, the amount of metallic copper reoxidized to CuzO by activation of catalyst p-5C (27.8 wt 5% copper and calcined at 700 "C) with methanol alone (Figure 6, curve d l ) was 0.035 g/g of activated catalyst while the amount of CuzO obtained from the same catalyst using a 1:l molar mixture of methanol and water (Figure 6, curve d2) was 0.08 g/g of activated catalyst. This increase in the amount of CUZOproduced is responsible for the increase in reaction efficiency (RE) in the steam reforming reaction (see Table 4). It is known that 50 mol % more Hz can be produced for the steam reforming reaction when compared to the methanol decomposition reaction alone. However, the Cu-A1 catalyst must be present in its active form. These results show that water provided the extra Hz and also participated in conditioning the copper species on account of C02 formation into an optimum Cu+-CuO combination appropriate for the desired reactions. In this study, we have provided extensive data concerning the production of Hz from methanol under various conditions. It is therefore possible to determine the optimum conditions for maximum Hz production. Table 4 shows that maximum HZproduction was obtained in the steam reforming reaction at 250 "C with the catalyst containing 27.8 wt % copper and calcined at 700 "C. Catalyst Activity versus Time Studies. Catalyst activity versus time studies were conducted on the best catalyst (Le., 27.8 wt % copper and calcined at 700 "C) operating at optimum reaction conditions (feed consisting of a 1:l molar mixture of methanol and water and a reaction temperature of 200 "C). The catalyst activity remained stable, and no decrease in methanol conversion was observed over an 8-h period.
After these studies, the catalyst was then characterized for BET surface area, pore structure, and copper species present. These characteristics were found to be similar for the fresh and used catalysts. Activity Comparison. It is known in catalysis literature that various factors such as catalyst constituents and preparation variables affect catalyst performance. However, comparison of different catalysts used by different workers for H2 production from methanol is not available in the literature. It was decided to compare the performance for Hz production of the best catalyst from this study (p-5C) with those of other catalysts reported in the literature (Amphlett et al., 1991; Cheng et al., 1989; Matsukataet al., 1988; Su and Rei, 1991). These are given in Table 5. As can be seen, comparison is difficult as different reaction conditions have been used by different workers. However, these results show that our optimum catalyst provides the best performance.
Conclusions 1. Catalyst activation resulted in the formation of CuzO and Cuo species which were active in methanol dehydrogenation. 2. Methanol conversion increased with both catalyst copper concentration and catalyst reducibility. 3. Reaction efficiency was found to be a strong function of the amount of CuzO formed in the activated catalyst, and this amount depended on both copper concentration and type of feed. 4. Reaction temperature and catalyst characteristics had tremendous effects on Ha yield and selectivity in both methanol decomposition and steam reforming reactions. Maximum hydrogen production efficiency was obtained in the steam reforming reaction with the catalyst containing 27.8 wt % copper and calcined at 700 "C.
Acknowledgment The financial support provided by the Canadian Commonwealth Scholarship and Fellowship Plan, and the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged.
Nomenclature F = molar flow rate (mol/min) md = methanol decomposition reaction P = reaction pressure (atm) RE = reaction efficiency (mol %) S = Hz selectivity (mol %) sr = methanol-steam reforming reaction T = reaction temperature ("C) TON = turnover number for methanol conversion (mol % /Cu atom) W = weight of catalyst (g) X = methanol conversion (mol %) Y = Hzyield (mol of Hz produced/mol of methanol fed)
Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994 2065 Subscripts md, 1 = methanol decomposition reaction sr, 2 = methanol-steam reforming reaction Mo = methanol in the feed
Literature Cited Agaras, H.; Carrella, G.; Larborde, M. A.; Copper catalysts for the steam reforming of methanol. Analysis of preparation variables. Appl. Catal. 1988,45,53-60. Amphlett, J. C.; Evans, M. J.; Jones R. A.; Mann, R. F.; Weir, R. D. Hydrogen production by the catalytic S t e a m reformingof methanol. Part 1: The Thermodynamics. Can. J.Chem. Eng. 1981,59,720" . . I
111.
Amphlett, J. C.; Mann, R. F.; Peppley, B. A.; Stokes, D. M. Some design considerations for a catalytic steam reformer for a PEM fuel cell power generating system. Proc. Zntersoc. Energy Convers. Eng. Conf. 1991,26th (3), 642-649. Cheng, W. Carbon dioxide calcination of methanol dissociation catalysts. US Patent Number 4826798, 1989. Chinchen, G. C.; Waugh, K. C. The chemical state of copper during methanol synthesis. J. Catal. 1986, 97, 280-283. de Bokx, P. K.; Balkernende, A. R.; Geus, J. W. The mechanics and kinetics of methane formation by decomposition of methanol on a Ni/SiOz catalyst. J. Catal. 1988,117,467-484. Dupond, R.; Degand, P. R. Make pure hydrogen from methanol. Hydrocarbon Process. 1986,45-46. Figoli, N. S.; Hillar, S. A.; Parera, J. M. Poisoning and nature of alumina surface in the dehydration of methanol. J. Catal. 1971, 20, 230.
Ghiotti, G.; Boccuzzi, F. Chemical and physical properties of copper based catalysts for COshiftreactionandmethanolsynthesis.Catal. Rev.-Sci. Eng. 1987, 29 (2&3), 151-182. Idem, R. 0.;Bakhshi, N. N. Production of Hydrogen from Methanol. 1. Catalyst Characterization Studies. Znd. Eng. Chem. Res. 1994, preceding paper in this issue. Isogai, N.; Takagawa, M.; Watabe, K.; Yoneoka, M.; Yamagishi, K. Catalyst composition for decomposition of methanol. European Patent, Number 0 342 614,1989.
Klier, K.; Chatikavanij, V.; Herman, R. G.; Simmons, G. W. Catalytic synthesis of methanol from CO/Hz iv. The effects of COz. J.Catal. 1982, 74, 343.
Kliman, M. L. Methanol, natural gas, and the development of alternative transportation fuels. Energy 1983,8, 859. Klisurski, D. Relationship between the selectivity and activity of oxide catalysts in the oxidation of methanol and the initial temperature of their reduction. Kinet. katal. 1970, 11 ( l ) , 263264.
Kobayashi, H.;Takezawa, N.; Minochi, C. Methanol reforming reaction over copper containing mixed oxides. Chem. Lett. 1976, 1347-1350.
Matsukata, M.; Uemiya, S.; Kikuchi, E. Copper-alumina spinel catalysts for steam reforming of methanol. Chem. Lett. 1988,761764.
Othmer, D. F. Methanol is the best way to bring Alaska gas to market. Oil Gas J. 1982,80 (44), 84. Petrini, G.; Montino, F.; Bossi, A.; Garbassi, F. In Preparation of catalysts, iii. Poncelet, G., Grange, P., Jacobs, P. A., Eds; Elsevier: Amsterdam, 1983; p 735. Su, T.; Rei, M. Steam reforming reactions over nickel and copper catalysts. J. Chin. Chern. SOC.1991,38,535-541. Takezawa, N.; Kobayashi, H.; Hirose, A.; Shimokawabe, M.; Takahashi, K. Steam reforming of methanol on copper-silica catalysts. effects of copper loading and calcination temperature on the reaction. Appl. Catal. 1982,4, 127. Uehara, I. R&D program on hydrogen energy in Japan. Tenth Anniversary Conference of Hydrogen Industry Council. Kananaskis,Alberta, Canada; HydrogenIndustry Council 700,lthAvenue S.W.: Calgary, Canada, 1992. Yakerson, V. I.; Rozanov, V. V.; Rubenshtein, A. M. Thermodesomtion studies on adsorbate-adsorbent interaction. Surf. Sci. 1968. ii, 22i. Received for review December 27, 1993 Revised manuscript received April 29, 1994 Accepted May 31, 1994. Abstract published in Advance ACSAbstracts, July 15,1994.
