Kinetic Studies of Catalyst Poisoning during ... - ACS Publications

Kinetic Studies of Catalyst Poisoning during Methanol Synthesis at High Pressures. Bernard J. Wood, William E. Isakson, and Henry Wise. Ind. Eng. Chem...
0 downloads 0 Views 3MB Size
Ind. Eng. Chem. Prod. Res. Dev. 1980, 19, 197-204 Klingsberg, E., "Pyridine and Its Derivatives", Chapter 11, pp 152-163, Interscience, New York, N.Y., 1960. Mekhtiev, S. D., Rizaev, R. G., Novruzova, A. Sh., Akhrnedov, S. M., Ranev, M. S., U.S.S.R Patent 253798 (1972). Newey, H. A,, Erickson, J. G., J . Am. Chem SOC.,72, 5645 (1950). Zellner, R . J.. US. Patent 2796421 (1955).

197

Received for reuieu December 21, 1979 Accepted February 19, 1980

Presented in part at the ACS/CJS Congress in Honolulu, Hawaii, April 1979, Division of Industrial and Engineering Chemistry.

Kinetic Studies of Catalyst Poisoning during Methanol Synthesis at High Pressures Bernard J. Wood,* William E. Isakson, and Henry Wise Solid State Cata/ysis Laboratory, SRI International, Menlo Park, California 94025

Quantitative specific poisoning rates have been obtained for both low-pressure process and high-pressure process methanol synthesis catalysts exposed to H,S, COS,and thiophene. In addition, changes in the surface composition and the topography of the catalyst during poisoning have been observed by Auger electron spectroscopy and by scanning electron microscopy, respectively. The results of these measurements suggest that sulfur selectively adsorbs on the surface of copper species in the low-pressure process catalysts. The rate data may be used in catalyst selection and process design for a wide range of operating conditions with sulfur contaminants in the feedstock.

Introduction Two types of mixed metal-oxide catalysts are being used industrially for the conversion of CO/Hz gas mixtures to methanol. The catalysts employed in the high-pressure process have been reported (Strelzoff, 1970) to have a greater resistance to poisoning by sulfur-containing contaminants in the feed gas than the low-pressure process catalysts, but no quantitative data on the sulfur resistance of these catalysts are available. In order to obtain data on the sulfur-poisoning kinetics we have examined the decay in catalytic activity of several methanol synthesis catalysts on exposure to hydrogen sulfide, carbonyl sulfide, and thiophene in the CO/Hz feed gas. Our measurements were performed under methanol synthesis conditions a t temperatures and pressure typical of industrial usage. For these measurements we used a Berty gradientless reactor (Berty, 1974). We determined methanol yield as a function of temperature, space velocity, total gas pressure, and exposure to feedstocks containing gaseous sulfur compounds a t low concentrations (ppm levels). Experimental Details A schematic diagram of the gradientless reactor system used in our studies is shown in Figure 1. The gas supply section included three separate gas sources. One contained a cylinder of high-pressure nitrogen used to purge the system and to set pressures on dome-loaded regulators. The second source contained two cylinders, one of nitrogen and one of hydrogen, used to supply hydrogen-in-nitrogen mixtures (1to 10% hydrogen) to reduce and activate the catalysts. The third source was the gaseous feedstock composed to 33 vol 90CO, 65 vol % Hz, and 2 vol 70 C 0 2 supplied in special cylinders made of aluminum alloy (84 ft3 (STP) pressurized to 1650 psig). The gas flow and pressure control system was designed to supply: (1) the nitrogen purge gas a t the desired pressure with flow monitored and controlled by the vent rotameter at the outlet of the system, (2) the nitrogenhydrogen mixture for catalyst activation by separate flow regulators and rotameters followed by in-line mixing before admission to the reactor system, and (3) the reaction gas 0196-4321/80/12 19-0197$01.00/0

mixture with pressure control achieved with special dome-loaded regulators manufactured by Grove Regulator Co. (these regulators were chosen because of their wide range of pressure operation and the use of aluminum as a material of construction of those parts that are in contact with HzS). To achieve pressure greater than those available from the gas cylinders (1650 psig), we installed a Model 46-13421 diaphragm compressor manufactured by Aminco (American Instrument Co., Silver Spring, Md.). It has a maximum output pressure of 10000 psi. At 5000 psi this compressor delivers 11 scf/h with a 500 psi suction pressure. The system included a bypass to allow operation without the compressor and a surge vessel to dampen the pressure fluctuations from the compressor. A Grove regulator maintained proper downstream pressure. Constant flow to the reactor was assured by maintaining a constant regulated upstream pressure and a constant compressor stroke. The catalytic studies were carried out with a gradientless reactor designed by Berty (1974) and manufactured by Autoclave Engineers. The reactor contains a stationary draft tube catalyst basket. The gas mixture is stirred with a turbine-blade impeller driven by a variable speed Magnedrive assembly. To permit methanol synthesis studies a t high hourly space velocities (5000-80000 h-l) and modest flow rates, we fabricated a cylindrical insert for the catalyst basket that enabled us to use as little as 10 cm3 of catalyst. The effluent from the Berty reactor flowed through a needle valve to reduce the pressure to 1 atm. A portion of the effluent stream was diverted to the sampling loop of the gas chromatograph. The remaining gas entered a cooling coil, immersed in tap water, and passed through a 1-L stainless steel cylinder where liquid products were separated and collected. The gas continued through a rotameter to the outside atmosphere. The methanol yield was determined by gas chroamtographic analysis of aliquots taken from the effluent stream. Methanol was separated from the other components on a Porapak S column and detected quantitatively in a

0 1980 American

Chemical Society

198

Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 2. 1980

PRCDUCT COOLER CrnIlW

Crxlllnp

w.nr

W.W

--w

ilO"I0

Figure 1. Schematic diagram of catalyst test system.

thermal conductivity cell. Methanol yields are reported in terms of the fraction of carbon monoxide converted. A flame photometric detector (manufactured by Tracor, Inc., Analytical Instruments, Austin, Texas) was used to measure the concentration of hydrogen sulfide in the reactor effluent by allowing an aliquot of gas to pass through a Porapak QS column. The detector response was Calibrated against a wet chemical collection and titration technique (NIOSH Analytical Method for Hydrogen Sulfide, No. S4, validated by SRI). Before each experimental run, the reactor was charged with a weighed mass of catalyst, then purged with dry nitrogen. The catalyst was reduced by slowly raising the reactor temperature to 500 K while a dilute hydrogen stream (2 vol 70 H2 in dry N,) flowed through the bed at a total pressure of 1 atm. The flow of reducing gas was sustained for at least 24 h. To test the gradientless operation of the Berty reactor we made some initial runs with a low-pressure process catalyst. Under constant reaction conditions (temperature, pressure, and space velocity), the methanol yield remained constant as the reactor impeller velocity was varied over the range 500 to 1500 rpm. In all subsequent runs, we used an impeller velocity of 11000 rpm. For our study we examined two low-pressure process industrial catalysts. One carried the designation C79-1 (manufactured by United Catalysts, Inc., Louisville, Ky.); the other, a proprietory formulation, is labeled "B". Also we examined a high-pressure process catalyst, designated Zn-0312 T. (manufactured by Harshaw Chemical Co., Cleveland, Ohio). The catalysts were in the form of ' / 4 x ' / 4 in. pellets. The sulfur compounds used as feedstock contaminants were hydrogen sulfide (H,S), carbonyl sulfide (COS) and thiophene (C4H4S). In all cases, the sulfur-bearing impurity was admixed to the syngas by the gas supplier (Airco Industrial Gases, Santa Clara, Calif.) and stored in aluminum alloy cylinders. Hydrogen sulfide was supplied at a number of concentration levels in the range of 0.06 to 75 ppm; COS, 0.2 to 2 ppm; and C4H4S,3 to 5 ppm. The concentration of sulfur compounds in each cylinder was

determined at frequent intervals during the course of the experimental study. The sulfur impurity level was found to remain constant for a particular cylinder, indicative of good mixing and negligible loss by reaction with the cylinder walls. Experimental Results (A) Low-Pressure Process Catalysts. Under our experimental conditions (475 to 500 psig, 460 to 515 K) and in the absence of sulfur contaminants the freshly reduced catalysts exhibited typical aging characteristics, exemplified by high initial activity followed by a gradual decrease to a constant conversion level. Thus the methanol yield dropped to about one-half of its initial value after about 4.5 h. In some long-duration runs with sulfur-free syngas, we allowed the feed gas to enter the reactor continuously during the daytime, but during the night the feed was replaced with a stream of dry nitrogen a t 1 atm pressure. When the reactor was repressurized with syngas after these overnight intervals, the catalytic methanation activity matched closely the activity observed at the end of the preceding day. Methanol yield varied predictably with space velocity. In Figure 2 we show typical results as a function of the reciprocal space velocity, 7 . The deviation from linearity in these plots for large values of 7 (i.e., low space velocities) cannot be attributed to the approach to equilibrium conversion. At 517 K and 475 psig the calculated equilibrium yield (Natta, 1955) of methanol is about 40 vol '70. More likely these results reflect the appearance of desorption limited rate processes or inhibition by reaction products (Atroshchenko et al., 1971). While mass transport in the gas phase in the Berty reactor is fast relative to chemical reaction rate at sufficiently high space velocities (Figure 2), the relative magnitude of intrapellet (pore) diffusion had to be established separately. Intraparticle mass transport can be characterized empirically by evaluation of the Weisz-Prater criterion CP (Weisz and Prater, 1954; Weisz, 1973), which specifies that diffusional effects are negligible if @ = (dn/dt)(l/C)(R2/U) 5 0.3

Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 2, 1980 Table I. Catalyst Effe'ctiveness Assessmenta mean pellet bulk density, diameter. m m ~cm-, 6.3 X 3.1 (cylinders) 2.9 (powder 6-8 mesh) 0.85 (powder 16-24 mesh)

1.05 1.12 1.22

flow rate, (STPI L.h-'

CH,OH formation rate mol,h-'.(g of cat.).'

125-250 55-160 42-120

6.2 x 10-3

7.1 x 10-3

6.3 x 10-3

C79-1; 1' = 503 K, P = 500 psig.

Catalyst:

"rr10

mass of charge.

199

T

I

I

I

4

5

L

f

0

1

3 RECIPROCAL SPACE V E L O C I T Y

6

I Ihr x 10')

Figure 2. Methanol yield as a function of reciprocal space velocity: feed gas, 33% CO, 65% H,,2 % CO,; P = 475 psig; catalyst, C79-1, 'I4 x l/s-in. pellets.

I 192

1%

I

I

I

I

I

2 00

204

2 08

2 12

2 16

1 I T IK-'

where (dnldt) is the observed reaction rate per unit volume of catalyst, C is the reactant concentration (outside the pellet), R is the pellet radius, and D is the diffusivity of the reactant in the gaseous medium. All these parameters were measured for our system with the exception of D, which was estimat'ed for our reaction conditions from Gilliland's empirical formula (Perry, 1950). For the highest methanol-synthesis rate observed in our experiments with a pelleted catalyst, we calculate a value of = 0.03 indicative of the absence of intraparticle diffusional resistance. Parallel experiments with catalysts of widely different particle sizes supported this conclusion. As shown in Table I, the rate of methanol formation exhibited little variation when the mean particle size of the catalyst was reduced by nearly an order of magnitude. In the well-stirred catalytic reactor a direct measure of the reaction rate can be obtained from the difference in inlet and outlet concentration of reactant (Mahoney, 1974). In the absence of a mole change during the reaction under study at the relatively low conversions encountered in our measurements, a good approximation for the reaction rate P per mass of catalyst W is given by r =F (Xi - X,) W

where Xi and X, are the CO concentrations in the feed stream and the effluent and F is the feed rate of reactant.

)i

lo3)

Figure 3. Effect of temperature on rate of conversion of CO to methanol: feed gas, 33% CO, 65% H,,2% CO,; P = 475 psig; catalyst, C79-1, X '/8-in. pellets.

For the fractional yield of methanol Y = (Xi- X,)/Xi, one obtains P

= YXiF/ W

(2)

In our experiments, Xi = 0.0147 mol/L (STP),and W was in the range 5 to 30 g. Consequently, we calculated values of reaction rate in units of mol.h-'.(g of catalyst)-' from measured values of yield and feed rate. During our experiments with the C79-1 catalyst, we varied the reactor temperature in the range 460 to 515 K. The measured values of reaction rate fit an Arrhenius plot (Figure 31, from which we derive an activation energy E = 107.2 kJ/mol (25.6 kcal/mol) for methanol synthesis in the absence of sulfur compounds in the feed stream. After establishing a steady methanol yield at specified temperature, pressure, and space velocity, we began catalyst poisoning experiments. The sulfur-free syngas was replaced with a CO/H2/C02 gas mixture of the same composition but with a known concentration of a sulfurcontaining impurity. The rate of catalyst poisoning was determined by monitoring the decrease in methanol yield as a function of time. A t inlet concentrations of H2S in the range from 1.6 to 40 ppm, catalyst activity decay followed a logarithmic law (Figure 4).

200

lnd. Eng. Chem. Prod. Res. Dev., Vol. 19, No 2, 1980

Table 11. Deactivation of Low-Pressure Process Catalysts by Hydrogen Sulfide during Methanol Synthesisa

H,S activity concn,b decay ni rate, (mo1.L") kni ( h - ' ) catalyst C79-1'

x

lo6

1.3 2.6 26.0

poisoning rate constant, k (L.mo1-l h-l) x

lo2

x

1.1 2.5 15.3

0.85 0.96 0.59 k = 0.80 0.14

*

Space velocity: l o 4 h - ' ; pressure: 500 psig; feedstock: 32 CO, 66 H,, 2 CO, (vol %); '/, x 1/8-in.catalyst pellets. At temperature and pressure of reactor. T= 503 K. a

An exponential decay in activity caused by catalyst surface poisoning is characteristic of a heterogeneous reaction system in which deactivation occurs by a first-order reaction in a kinetically controlled regime, Le., where diffusional transport is not rate limiting. One model (Khang and Levenspiel, 1973) for such a so-called sideby-side deactivation process is available for gradientless reactor conditions in which the catalyst is bathed in a gaseous environment of uniform composition. According to this model the slope of the exponential decay curve of reaction rate with exposure is the product of the poisoning rate constant k and the concentration ni of the sulfur contaminant in the reactor inlet stream. However, if the decay rate deviates from an exponential curve, then the mass transport processes may have begun to affect the poisoning rate. Such behavior might occur with pelleted catalysts when the active sites on the outer surfaces of the pellet are completely poisoned. A t this point further loss in activity will be governed by the rate at which poison diffuses into the pore structure of the pellet. Factors that determine the time (measured from the onset of poisoning) required for such intraparticle diffusion to become ratecontrolling include the mass of the bed, the pore structure of the catalyst, the size of the pellets, and the concentration of the poison in the feedstock. We observed exponential decay for periods as long as 55 h with 1 / 4 x l/s-in. pellets of C79-1 catalyst exposed to syngas containing 3 ppm H2S or less (Figure 4). At much higher levels of H2S (40 ppm) deviations from exponential decay were observed, undoubtedly due to mass-transfer limitations and intraparticle diffusion. The analysis of our data for the lowpressure process catalysts yields the values of k presented in Tables I1 and I11 for H2S as a poison, and in Table I11 for various sulfur-containing compounds. (B) High Pressure Process Catalyst. To compare the sulfur resistance of high- and low-pressure process catalysts, we made measurements with the Zn-0312 catalyst at 3000 psig and 573 K. 1Jnder these typical industrial conditions, we found it necessary to reduce the mean

TIME I N SYNGAS lhourl

Figure 4. Typical decay in catalytic activity during methanol synthesis with H,S-contaminated feedstock: feed gas, 33% CO, 65% Hz, 2% CO,, 3.2 ppm of H,S; space velocity = lo4 h-l, T = 503 K, P = 500 psig; catalyst, C79-1, 1 / 4 X 1/8-in.pellets.

catalyst particle diameter to 0.85 mm (16 to 24 mesh) and to increase the space velocity to values greater than 40 000 h-' to attain gradientless conditions in the reactor during methanol synthesis (Figure 5 ) . Catalyst activity decayed exponentially with time of exposure to feedstocks containing sulfur-bearing contaminants in the concentration range 0.08 to 1.5 ppm. The deactivation rate under reaction conditions as a function of H2S concentration is shown in Figure 6, and the average values of the poisoning rate constants for H2S and thiophene, normalized to lo4 h-l space velocity for purpose of comparison, are summarized in Table 111. Discussion Significant quantitative differences exist in the sulfur poisoning resistance of the catalysts examined in this study. After aging to constant activity for methanol synthesis in the absence of sulfur contaminants, catalyst "B"

Table 111. Comparison of Poisoning Rate Constants for Methanol Synthesis Catalystsa feedstock contaminant process type

catalyst

low pressure

(279-1

H2S

"B"

thiophene HIS

high pressure

Zn-0312

type

cos

cos

H,S thiophene

concn, ppm

reaction temp, K

1.6-33 0.6-9 3-5.4 0.08-40 0.01-7 0.0 8-1.5 5.2

503 503 503 514 514 573 573

poisoning rate constant k , b (L.mol-'.h-') X 0.80 * 0.14 0 1.02 ?- 0.18 5.5 ?- 1.9 0 0.095' 0.041'

a Evaluated a t space velocity = l o 4 h-'. Average values. ' Evaluated a t space velocity = 8.07 X l o 4 h - ' , normalized to space velocity = l o 4 h - ' o n the basis that the rate of loss of both methanol production and number density of active sites due t o poisoning have the same dependence o n space velocity.

Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 2, 1980 30

1 1 I I

I

7

I

I

201

25

J

0

: 20 y1

a

0

3 y1 c

x u

i

15

z U 3

sz

0

2

10

Y U

-

0

r

05

1

2

0

a 0

01

02

04

05

R E C I P R O C A L SPACE V E L O C I T Y T-hr

05

I , (mole

10

’ )

x

7

8

lo6 5

POT

Figure 6. Deactivation of methanol synthesis catalyst Zn-0312 at u different H2Sconcentrations in feed stream: space velocity, 80650;

03

6

4

3

3

0

3

H2S CONCENTRATION

06

07

08

X lo4

Figure 5. Methanol yield as a function of reciprocal space velocity: feedstock, 33 vol 9’0 CO, 2 vol 90 COP,balance H,; pressure, 3000 psig; temperature, 566 I