Ind. Eng. Chem. Res. 1995,34, 2358-2363
2358
Hydrogenolysis of Methyl Formate by HdCO Mixtures with CuO/ZnO/Al203 Based Methanol Synthesis Catalysts G. BracaJ A. M. Raspolli Galletti, N. J. Laniyonu, and G. Sbrana* Dipartimento di Chimica e Chimica Industriale, Universita di Pisa, Via Risorgimento 35, 1-56126 Pisa, Italy
E. Micheli, M. Di Girolamo, and M. Marchionna SNAMPROGETTI S.p.A., Research Division, Via Maritano 26, 1-20097 San Donato Milanese, Italy
Ternary copper-based catalysts of type CuO/ZnO/Al203, CuO/ZnO/Crz03, and CuO/ZnO/La203 have been advantageously used in the hydrogenolysis of methyl and ethyl formate to methanol. The catalysts are active under mild conditions (150-185 "C; 5-10 MPa) both in gas-solid and in gas-liquid-solid phases with high selectivities to methanol (80-90%). The decarbonylation of the formic ester is the only side reaction, but the catalysts, different from other copper-based systems, do not suffer from the presence of CO in the reaction gases. Accordingly the hydrogenolysis may be carried out with a hydrogen feed containing CO without any poisoning of the catalyst or decrease of the selectivity to methanol. A preliminary screening of different copper catalysts points out that the insensitivity to CO might be related t o the presence of basic sites on the catalysts.
Introduction Several process alternatives have been proposed for converting CON2 to methanol at milder conditions with respect to the current copper-catalyzed process (Wainwright, 1988) in order to improve the thermodynamic efficiency of the process. Among these alternatives, one of the most notable is the synthesis of methanol via an alkyl formate, generally methyl formate (Wainwright, 1988). CH,OH
+ CO r+ HCOOCH,
(1)
+ 2H, - 2CH30H CO + 2H, - CH,OH
HCOOCH, net:
(2)
(3)
In the first step, methanol is carbonylated to methyl formate, and then its subsequent hydrogenolysis results in the gain of a methanol molecule over the total reaction. While the carbonylation step is commercially available (Reutemann and Kiezcka, 1989), the hydrogenolysis, in order to be economically advantageous, must be carried out using a hydrogen feed containing the CO not converted in previous carbonylation stages; as a consequence, the hydrogenolysis catalyst must be insensitive to CO poisoning. Furthermore the same catalyst should be able to supress the undesired side reactions of decarbonylation and decarboxylation (eqs 4 and 5); the latter reaction may follow an intermediate hydrolysis step occurring when water is present (eq 6): HCOOCH3 HCOOCH3 HCOOCH3
+
H20
-
CHjOH CH4
+
+ CO
(4)
CO2
(5)
* CH30H + HCOOH COP + H2
* To whom correspondence should be addressed. +Deceased.
(6)
In the gas-phase hydrogenolysis described in the literature over copper chromite (Evans et al., 1983; Gormley et al., 1992)or silica-supported copper catalysts (Monti et al., 1985,1986a) at temperatures from 120 t o 190 "C and atmospheric pressure two main effects of carbon monoxide have been observed: (i) a reversible kinetic inhibition attributed to Hz displacement from the surface and (ii) a long-term irreversible and progressive poisoning attributed to the deposition on the surface of the catalyst of a HCHO polymer as a consequence of the reduced hydrogenation rate (Evans et al., 1983; Monti et al., 1985, 1986a). Also in the liquid phase, in spite of some preliminary optimistic conclusions (Sorum and Onsager, 1984) it is reported that carbon monoxide inhibits the hydrogenolysis rate of methyl formate in the presence of either copper chromite catalysts (Monti et al., 198613; Gormley et al., 1992)or the more active copper catalysts (Cu Raney) (Gormley et al., 1992). With the latter catalysts the inhibition by CO was found to be lower at higher temperatures. In any way decarbonylation of methyl formate takes place at 130-170 "C accounting for 10-20% of the converted product (conversion 7080%)in the gas-solid (Evans et al., 1983) and 1-2% at 160 "C in the liquid phase (Gormley et al., 1992). Economic reasons related to the use as hydrogenating stream of a syngas with some residual CO not converted in the carbonylation stage prompted us to search for a catalyst not irreversibly poisoned by CO and sufficiently active in the hydrogenolysis of the ester. The choice was oriented toward ternary (CuO/ZnO/Alz03, CuO/ZnO/ Cr2O3, etc.) copper-based catalysts active in methanol synthesis (low-pressureprocess) which should not likely suffer in the presence of CO. Actually, catalysts of this type have been previously proposed for the hydrogenolysis of higher esters (acetates and esters of fatty or dicarboxylic acids) at temperatures in the range 200-300 "C, with good selectivities toward the alcohols produced (Turek et al., 1994). The present paper deals with the applications of the CuO/ZnO/Ah03 and other copper-zinc-based catalysts to the gas-liquid or gas-phase hydrogenolysis of methyl and ethyl formate to methanol.
0888-5885/95/2634-2358$09.00/00 1995 American Chemical Society
Ind. Eng. Chem. Res., Vol. 34, No. 7, 1995 2359 Table 1. Chemical Composition and Physical ProDerties of the Catalysts ~~
elemental a i l y s i s catalystQ
A B C
D E'
physical properties
cu
Zn
Al
Cr
La
(wt%)
(wt%)
(wt%)
(wt%)
(wt%)
55.0 34.2 27.5 55.0 58.3
18.2 38.2 27.9 18.2 18.5
4.4 5.1 4.6 4.4
8.6
alkali (ppm) 890 290 300 2720
3.4
surface area (m2/g) 101 53 96 101 92
pore volumeb (cm3/g) 0.32 0.19 0.19 0.32 nd
pore radius (A) 61 88 39 61 nd
bulk density (g/cm3) 1.41 1.12 1.40 1.12 nd
A CuO/ZnO/AlzOS; B, CuO/ZnO/Al203; C, CuO/znO/Al203/Cr203;D, CuO/ZnO/Al203/alkaliadded as KzC03; E, CuOJZnOJLa203. Hg porosimetry. ' The sample was amorphous by XRD analysis; no other characterization except for surface area was performed on this catalyst due to the highly hygroscopic character of the lanthanum catalyst (nd = not determined).
Experimental Section Catalyst Preparation. Catalyst compositions have been chosen on the basis of catalysts described in the literature to be active in CO hydrogenation (Bartand Sneeden, 1987). The composition and the main physico-chemical properties of the catalysts are summarized in Table 1. The catalysts were prepared according to procedures described in the literature (Gherardi et al., 1983); nitrates of the elements were dissolved in demineralized water (approximate concentration of overall metals ca. 1 M) in the desired ratios, and coprecipitation was carried out at constant temperature (60 "C) by means of a slight excess of a NaHC03 solution (ca. 1 M). The precipitate was digested for 1h and then washed with hot water up to the disapperance of NO3- ions; the cake was dried a t 100 "C overnight and then calcined in air at 350 "C for 4 h. Catalyst D was obtained from the corresponding unpromoted catalyst (catalyst A) by incipient wetness impregnation with an aqueous solution of K2C03 and subsequent drying at 120 "C for 16 h. It was, finally, calcined in air a t 350 "C for 4 h. Catalyst E was obtained under the same conditions as the other catalysts but with the following major differences: the precipitation was carried out by maintaining a constant molar ratio of 2 between co32-anions (precipitatingagent, a 2 M K2CO3 solution) and the sum of the metals (Cu Zn La). In this case, the precipitate was then washed until the disappearance of Nos- occurred. A Fisons Instruments Sorptomatic model 1900 with N2 adsorption was used to measure the surface area, the total volume, and the distribution of pores; elemental composition of the final catalysts was determined by atomic adsorption analysis. Apparatus and Procedure. 1. Liquid-Phase Experiments. Experiments were carried out in a 1 L stainless-steel autoclave fitted with gas inlet and outlet lines and a liquid sampling line. The catalyst pellets (4 = 2 mm) were loaded in a perforated basket fitted on the arm of the stirrer thus allowing a good gas-liquid-solid contact. The catalysts (10-20 g) were introduced in the container, and after the evacuation of air and moisture, the reduction stage was carried out in the same reactor by treatment with hydrogen (10 MPa) and methanol (10-100 mL) at 200 "C for 4-5 h. After the reduction, methanol was evaporated in vacuo and the reduced catalyst kept in HZatmosphere before use. The liquid reactants (pure methyl or ethyl formate) were loaded by suction into the autoclave which was pressurized, a t room temperature, with hydrogen or H2/ CO mixtures up to the required pressure.
+
+
The autoclave was subsequently heated to the desired reaction temperature without stirring; when the reaction temperature was reached a sample of liquid was taken to determine the actual composition, confirming that no reaction occurs while the temperature is raised. Then, the reaction was started by stirring. A constant pressure was maintained ( f 0 . 3 MPa) by supplying pure hydrogen from a reservoir. The experiments were generally carried out at alkyl formate conversions of nearly 20-40% (time = 6 h). The overall material balance was performed by drawing off the liquid through a siphon and discharging gas and the residual liquids through cooled traps at -70 "C condensing there all evaporated material. Accountabilities, based on methyl formate, higher than 95% were obtained. The liquid reaction mixture was analyzed by GC (Sigma 3B/HWD Perkin Elmer instrument equipped with Porapack PS packed column; peak integration was performed using a Perkin Elmer Sigma 15 integrator) by withdrawing the samples (1.5 mL) through the outlet valve in a cooled trap. GC analyses were performed using a silica gel column for the determination of COZand a molecular sieve 5 A column for the determination of CO, CH4, and C2H6. Further details on the sampling procedure and analysis may be obtained elsewhere (Raspolli Galletti et al., 1985). 2. Gas-Phase Experiments. Experiments were carried out in a tubular fixed-bed laboratory unit equipped with a 10 mm i.d. stainless steel reactor; about 5 g of catalyst (14-20 mesh) was loaded in the reactor in each test, and the space above and below the catalyst bed was packed with inert a-alumina. Reactor heating was provided by an electric oven, and the temperature of the catalytic bed was measured with an internal thermocouple. Hydrogen was supplied by cylinders and measured by a mass flow meter, while methyl formate was fed by means of a Waters HPLC pump; the gas and liquid feeds were mixed at the entrance of the reactor, and the upper part of the inert filling also ensured preheating and vaporization of the reactants. The effluent from the reactor was cooled down and condensed a t -10 "C; the liquid product was then collected in a gas-liquid separator. Samples of the effluent gas stream were periodically analyzed on line by a 5730 Hewlett-Packard gas chromatograph equipped with both a thermal conductivity detector (TCD) and a flame ionization detector (FID) Porapak QS and Carbosieve columns were used in order to separate CO and C02 and light hydrocarbons. Analyses of the liquid products were performed on a 5890 Hewlett-Packard gas chromatograph, equipped
2360 Ind. Eng. Chem. Res., Vol. 34,No. 7, 1995 100
R
tures of the DRIFT cell are reported elsewhere (Basini et al., 1991).
8ol\ 0
Results and Discussion
5
10
20
15 t-lme.
25
h
Figure 1. Ethyl formate hydrogenolysis with CuO/ZnO/Al203 (catalyst A). Reaction conditions: ethyl formate, 320 g; catalyst, 10.1 g; temperature, 185 "C; PH*, 10 MPa; rpm, 200. 0, Ethyl formate; A, ethanol; 0, methanol; 0 , methyl formate.
with a flame ionization detector; products separation was effected by a 50 m capillary column (HewlettPackard, HP5 5% cross-linked phenylsilicone). Before feeding the methyl formate the catalyst was pre-reduced raising the temperature to 300 "Cby 100 "CAI in a Hfl2 stream, the Hz content being progressively increased from 5 to 100%. Before collecting the liquid sample, the reaction was performed for 1 h in order t o stabilize the reaction conditions;afker that the catalytic tests were performed for 1 or 2 h, and the catalyst was left under hydrogen flow overnight. Reaction conditions were selected in order to ensure a complete vaporization of reactants depending on the total pressure; total contact time (GHSV, H2 MF; MF = methyl formate)) has been kept constant. Material Balance. The material balance was always higher than 95%. Data Presentation. Conversion and selectivity are defined by the following equations:
+
conversion % = [(mol of ester initial mol of ester final)/mol of ester initial] x 100 Selectivities to methanol in the case of methyl formate (MF) hydrogenolysis are evaluated as selectivity % = [0.5(mol of MeOH produced)/ (mol of converted MF)] x 100 and in the case of ethyl formate (EF) hydrogenolysis from the relation selectivity % = [(mol of MeOH
+ 2(mol of MF produced))/ (mol of converted EF)] x 100
Separated repeated experiments showed that the experimental errors were generally within 24.5% in the gas phase (for both conversion and selectivity) and f2.5% for conversion and f 2% for selectivity in the liquid phase. Spectroscopic Analysis. IR spectroscopic experiments were performed with a Nicolet 20 SXC spectrometer equipped with a mercury, cadmium telluride (MCT) detector (resolution of 4 cm-l). The experimental fea-
Liquid-PhaseHydrogenolysis. Preliminary information on the performance of a ternary methanol synthesis catalyst CuO/ZnO/Al203 (catalyst A, Table 1) has been obtained by studying the hydrogenolysis of ethyl formate in the liquid phase, which allowed a more immediate indication of the methanol formation by reduction of the formyl moiety. Actually, the hydrogenolysis took place rapidly at a not too high temperature (185 "C) (Figure l), with formation of methanol (eq 7) and methyl formate produced by a transesterification reaction (eq 8) with a selectivity to C1 products of about 70-75%. Decarbonylation of the ester (eq 9) is the only side reaction, whereas no decarboxylationtook place (eq 10):
+ 2H2
-
+ C2H,0H CH,OH + HCOOC,H, zz C2H,0H + HCOOCH, HCOOC,H, CO + C2H,0H HCOOC,H,
HCOOC,H,
-
CH,OH
CO,
+ C2H,
(7) (8) (9) (10)
A rapid decarbonylation initially occurred in the first hour accounting for more than 40% of the converted ester; successivelythis reaction proceeded more slowly so that after 24 h it was reduced to less than 30% at a conversion of the ester of 70%; this seemed to indicate that evolved CO probably poisons the basic sites of the catalyst where the ester is activated for the decarbonylation. It was then surprisingly found that the catalyst was not poisoned by the initial presence in the gases of CO (10%) but, on the contrary, a slight advantage was gained by its presence both in activity and in selectivity of the process (Table 2). The promising results obtained with ethyl formate prompted us to extend the experimentation to methyl formate. Initially the following points were ascertained: (1) methyl formate could be hydrogenolyzed,beginning to display some reactivity a t 125 "C; an increase of the temperature in the range 125-200 "C resulted in a decrease of the selectivity to methanol and an increase of the decarbonylation reaction (Table 3);(2) a chemical kinetic regime was assured by a stirring velocity of 200 rpm (velocities checked, 100, 150, 200, and 300 rpm); (3)the recycled catalyst maintained nearly unchanged the reaction activity and selectivity; the catalyst was actually recycled 7 times without significant loss of activity (Figure 2); and (4)no production of methanol was observed by direct CO-Hz reaction under the hydrogenolysis conditions (when the reaction was carried out using ethyl acetate as the medium a t 185 "C, under 11 MPa of H$CO (lO/l), with 10.1 g of catalyst no methanol was formed after 24 h; under these conditions ethyl acetate did not react). To give a quantitative account of the performances of the ternary Cu0/ZnO/Al~03(catalyst A), some experiments were carried out varying P H ~and pco. As in the case of the copper-basedcatalysts (Monti et al., 1986b; Liu et al., 1988; Gormley et al., 19921, the methyl formate disappearance rate may be considered nearly of first order in (Table 41, even though the
Ind. Eng. Chem. Res., Vol. 34,No. 7, 1995 2361 Table 2. Hydrogenolysis of Ethyl Formate in the Liquid Phase with CuO/ZnO/AlzOS (Catalyst A)"
time(h) 1 4 8 12 24
hydrogenolysis with pure Hz P H ~ , 10 MPa C1 products (mol %) MeOH HCOOMe conv (%) 2.4 7.9 12.4 14.7 14.5
1.4 4.3 6.9 20.0
hydrogenolysis with a HdCO mixture pnz, 9.3 MPa; pco, 0.7 MPa sel to MeOH (%)
8.4 24.8 40.0 50.0 69.0
C1 products (mol %) MeOH HCOOMe
57 70 73 73 71
conv (%)
sel to MeOH (%)
9.3 26.2 43.5 56.2 73.0
71 73 76 74 76
3.3 8.5 12.4 14.3 14.9
2.1 8.1 13.0 25.6
Reaction conditions: ethyl formate, 320 g; catalyst, 10.1 g; temperature, 185 "C; rpm, 200.
a
Table 3. Hydrogenolysis of Methyl Formate: Effect of the Temperaturea temp "C
conv (%I
sel to MeOH (%I
sel to CO (%)
125b 150 175 200
5.8 13.0 23.0 50.0
99.0 85.0 80.0 78.0
trace 12.0 18.2 22.0
a Reaction conditions: methyl formate, 389 g; catalyst (A), 19.8 g; time, 6 h; pressure, 11 MPa; rpm, 200. Conditions as in a except times, 5 h.
Table 6. Hydrogenolysis of Methyl Formate: Effect of Dmna PCO (MPa) initial final
1.3 5.0 7.4
P H ~(MPa) initial final
2.5 2.8 5.9 7.7
7.7 7.9 7.5 7.5
conv (%)
sel to MeOH (%)
23.0 26.0 28.0 27.0
79 82 86 85
5.2 6.4 6.5 7.2
a Reaction conditions: methyl formate, 389 g; catalyst (A), 20 g; temperature, 175 "C; time, 6 h; rpm, 200.
Table 6. Gas-PhaseHydrogenolysis of Methyl Formate: Effect of Temperaturea ~
~~
run
temp "C
conv (%)
MeOH sel (%)
1 2 3
150 170 190
35.8 65.7 80.3
97.8 96.5 92.3
Catalyst, CuO/ZnO/Al203 (catalyst A); LHSV, 1.0 h-l; H m F , 5.5 M; pressure, 2.5 MPa.
Table 7. Gas-Phase Hydrogenolysis of Methyl Formate: Effect of Hz Pressurea ~~
2
0
3 *
