H.E. Curry-Hyde and R.F. Howe (Editors), Nalural Gas Conversion I1 1994 Elsevicr Science B.V. 393

CONVERSION OF METHANOL TO LIGHT OLEFINS OVER SAPO-17 MOLECULAR SIEVE

Shah Nawaz*, Stein Kolboe* and Michael St6cker**

*) Department of Chemistry, University of Oslo, P.O.Box 1033, N-0315, Blindern, Oslo, Norway.

**) Department of Hydrocarbon Process Chemistry, SINTEF-SI, P.O. Box 124, Blindern, N-03 14 Oslo, Norway.

Abstract The zeolite erionite and its isostructural silicoaluminophosphate analogue, SAPO- 17.

were studied with respect to their catalytic behaviour in the MTO process, the emphasis being laid on SAPO-17. It proved superior to erionite both with respect to olefln selectivity and catalyst lifetime. Ethylene constituted almost 70 mol% and propylene almost 20 mol% of the product, compared to a maxtmum of 40 mol% ethylene and 30 mol% propylene with erionite. It was active for 10-20 h, the erionite for less than 2 h, and erionite exhibited ethylene/ propylene selectivity for only one hour. Experiments with SAPO- 17 in range 350-475OC showed optimum behaviour around 425OC. Outside this temperature lifetime and selectivity deteriorated quickly. A deactivated sample of SAPO- 17 could be restored to full activity by calcination in air at 500-550°C.

1. INTRODUCTION

Light olefins are important starting materials for the manufacture of many industrial products such as polymers, detergents, plasticizers etc. At present they are mainly obtained from crude oil based resources. Methanol, which may be obtained from coal or natural gas, may be an interesting alternative raw material.

Protonated ZSM-5, in particular, is an excellent catalyst for converting methanol to hydrocarbons, but tends to produce a hydrocarbon mixture which is more useful for gasoline than as a light oleflns petrochemical feedstock [ 1-31. Work aiming at stopping the reaction at the light olefin stage, e.g. by modifying the catalyst or by choosing optimum reaction conditions, has been reported [4-61.

On the other hand, small pore zeolites like erionite, chabazite, ZK-5 etc, with 8-oxygen ring pore openings are reported to be selective for an MTO process 171, but they deactivate quickly, so frequent regenerations are necessary. Modification by dealumination seems to alleviate the problem [8].

Recently a new class of zeotype molecular sieves based on aluminophosphates (ALP04,s) has been discovered 19-10). They are neutral, but acidity can be introduced by substituting some of the P5+ ions by Si4+ (forming SApOs) [ 1 1 - 121. The acidity is weaker, and the acid site density can be varied. They appear to be promising catalysts for light olefln formation from methanol [ 131. Prominent examples are SAPO-34 (chabazite structure) [ 13-14] and SAPO-17 (erionite structure). In this work SAPO-17 and erionite were selected for further study.

394

2. EXPERIMENTAL

2.1. Catalyst synthesis and characterization SAPO-17 was synthesized according to Lok et al. [ 111 and characterized by

X-ray diffraction, elemental composition determined by X-ray fluorescence, TPD of ammonia, MAS NMR spectroscopy on 29Si and 31P, study of template removal during calcination, and thermogravimetric measurements.

NH4-erionite was provided by VEB Chemie Kombinat Bitterfeld, Germany. It was converted to H-form by calcination at 550 O C . It was characterized using the same instrumentation as was used for SAPO- 17. 2.2. Reactor testing

The catalytic evaluation experiments were performed in a fixed-bed flow reactor using 0.4-1.0 g catalyst, which was placed between two layers of quartz wool. Methanol feed was evaporated and diluted with nitrogen. The measurements were carried out keeping a partial pressure of methanol and nitrogen mostly at 0.4 bar and 0.6 bar, respectively, and a Weight Hourly Space Velocity (WHSV) of 0.5 h-' unless speclfled otherwise. The reaction temperature was varied between 350-475OC. All samples were activated in-situ at 4OO0C in an atmosphere of nitrogen for one hour in the reactor oven before methanol feed was admitted. The products were analyzed on-line by gas chromatography. The reactor equipment and analytical systems have been described earlier [ 141. Deactivated catalysts were in some cases regenerated at 55OOC in a flow of air (75 ml/min) for 4 hours and retested.

CO, CO, and hydrogen were analyzed, but the amounts were negligible. 2.3. Thermogravimetry

Thermogravimetric studies were carried out using a Stanton Redcroft STA 785 instrument. The samples were heated (5'C/min) from 20 to 56OoC in flowing nitrogen (20 ml/min) and kept at that temperature until constant weight. The nitrogen flow was then replaced by air (15 ml/min) in order to study the rate of coke oxidation.

3. RESULTS AND DISCUSSION

3.1. Characterization Elemental composition: The elemental composition of the samples gave for erionite the ratio Si/Al =3.5 while in SAPO-17 the corresponding acidity determining ratio (AltPI/Si is 70, showing that the acid site density in erionite may be 20 times higher than in SAPO-17.

Powder X-rav diffraction: The X-ray diffraction pattern showed our SAPO- 17 to contain a small admixture of, probably, SAPO-35 [11,15]. The erionite may contain an offretite intergrowth, but is otherwise pure.

3.2. Reactor testing Product distribution: The product distributions (including unconverted oxygenates) during the methanol conversion over SAPO- 17 and erionite at 425 OC are shown in Figs. 1 (a) and (b). Over both catalysts the olefln formation increases with time on stream until it reaches a maximum and then declines sharply. The released hydrocarbons are in the C -C, range, with only small

395

100

80

d, 60

0

$

I 40

-

20

0 0 5 10 1 5 2 0 25 0 1 2 3

TOS fhours) TOS (hours)

(a) (b 1

Fig. 1. Product distribution in the reactor effluent and its dependence on t h e at 425OC. (a) over SAPO-17; (b) over erionite. (Note the different time scales for TOS).

amounts of C,, In both cases. But there are also clear differences between the two catalysts. The catalysts lifetimes before deactivation are quite different. As seen from Fig. 1 a), SAPO- 17 was active for 20 hours before the activity declined while the erionite sample, as Is seen from Fig. 1 b was active only a short time, about 2 h.

At peak activity, the amount of C,-C, olefins is about 85 mol% over SAPO- 17, whereas the erionite gave 70 %, and that was only in a short time interval. Furthermore, SAPO-17 was particularly selective for the formation of ethylene. During the active period of methanol conversion into hydrocarbons, the ethylene selectivity increased gradually with time on stream, up to above 60 mol% , while the formation of other hydrocarbons decreased correspondingly. Erionite did not show such a high selectivity.

Saturated Hvdrocarbons: Also regarding formatton of saturated hydrocarbons erionite and SAPO-17 show very different behaviour. Erionite starts by giving propane as the dominating product, and propane, although declining, remains a major product during the whole peridd the erionite is active. Over SAPO-

active. Details are seen from Fig. 2. Methane is also formed over z 30

17 propane is never a dominating product, and it falls almost to zero while ~ ~ ~ 0 - 1 7 is still fully

$

U MelhanelErionite --t PropanelSAPO-17 + Melhane/SAPO-17

both catalysts, initially in a 2 similar quantity (< 5 mo1%). It E 20

falls to zero over erionite as the 5 catalyst deactivates. Over SAPO- 10 17 the methane formation remains stable, and it soon 0 becomes the major saturated 0 1 2 3 4 5

hydrocarbon (-3mol% 1. Experi- ments with no catalyst, showed that the methane was not formed on the reactor walls.

Time on stream (hours)

ng. 2. Methane and propane in the reactor effluent over SAPO- 17 and erionite at 425OC.

396

Effect of Methanol Partial Pressure: A number of catalytic testing experiments were performed on SAPO- 17, where the partial pressure of methanol was varied. They were all performed at 40OoC. The partial pressures of methanol were 0.4, 0.2 and 0.1 bar. IGHSV was kept constant, so WHSV varied correspondingly.) From the plots of conversion versus time on stream (Fig. 31, it is apparent that the catalyst lifetime increased when the methanol pressure was lowered. The

of hydrocarbons before regeneration became necessary was, however, essentially constant. This may suggest that coke formation is mechanistically connected with hydrocarbon formation.

Effect of TemDerature: Experiments carried out at different temperatures showed the deactivation rate to be highly temperature dependent. This is shown in Fig. 4, where the conversion of methanol versus time on stream at 350,400, 425 and 475OC are presented. A rapid deactivation is observed both at the lowest and at the highest reaction temperatures. The longest catalyst lifetime was observed around 425 OC. We believe that two different catalyst deactivation mechanisms are operating at the two extremes of temperatures. Around 35OoC, we believe the deactivation to be due to the formation of long hydrocarbon chains formed by a series of condensation and poly-merization reactions of m e t h a n o l a n d / o r o le f in molecules. Due to the low reaction temperature they do not decompose and hence they stay within the framework. The pores of the catalyst are thus filled with this complex and the catalyst is deactivated. On the other hand, around 45OoC and higher, the deactivation of the catalyst is probably due to the formation of heat resistant, carbon rich "coke", probably of polyaromatic character, which may render the active sites inaccessible, thus causing catalyst deactivation.

In order to characterize the de- activating coke further, two samples of S M O - 17, deactivated at 350, respectively 425 O C , were put through thermogravimetric experiments. In each case, 32.6 mg of the sample was heated from 2O-56O0C at a programmed heat-ing rate of 5'C/min in nitrogen flow (20 ml/min). The samples were then kept isothermally at 56OoC until a constant weight

0 20 4 0 6 0 8 0 1 I

Time on stream (hours)

Fig. 3. Methanol Conversion over SAPO-17 at various feed rates (and partial pressures).

0 5 1 0 1 5

Time on stream (hours)

Fig. 4. Conversion of methanol into hydrocarbons at reactton temperatures of 350. 400, 425 and 475OC. The WHSV Is 1.0 h' at pMeoH=0.4 bar.

Ndragan+- A s was observed (about one 34

hour), the carrier gas was then replaced by flowing air (15 ml/min). The weight- - losses in such experiments E

33- - Temp.'C - F 32- - Deactwated (350%)

31 - are shown in Fig. 5. Al- 30-

though the total amount of ' 29-

28 - coke in the two samples is very similar, the weight loss

391

600

Y 450 2

a - 300 2

Q P

150 E :

0

27 4 I

0 100 200 300 curves are quite diffgrent.

Both samples show a weight loss slightly above 1

Fig. 5. Thermal analysis of the carbonaceous mg about 10 min after start of the experiment when the residues in two samples of SAPO-17 de- temperature has reached activated by working at 350, respectively -100 OC, but this weight 425OC. Note initial use of N, (20 mL/min) and loss is undoubtedly due to its replacement by air (15 mL/min). loss of adsorbed water. The low temperature deactivated sample exhibits a pronounced weight loss starting slightly below 450 OC and tapering off slightly before the maximum temperature of 550 OC is reached. The high temperature deactivated sample shows a hardly noticeable weight loss in the same temperature range. When nitrogen is replaced by air, burning off of the residue remaining at this high temperature is seen to start immediately. The resistance to oxidation is, however, much higher in the sample deactivated at high temperature than in the low temperature deactivated one, which almost immediately attain a stable weight (coke free), whereas the former only slowly reaches a stable weight, which may be slightly lower, but is essentially the same as for the other sample.

This indicates that the low temperature coke is rather different from the high temperature analogue, and this difference persists even after heating the coked catalysts to 550 OC, which is 125 OC higher than the temperature where the high temperature coke was formed.

In a separate experiment a sample deactivated at 350 OC was taken to 400 OC (in N, stream), where a non-negligible weight loss takes place during a couple of hours. It could, however, not be revived to an active state by this procedure.

Time (min)

Regeneration: SAPO- 17 which has converted methanol until deactivation may be regenerated. Samples deactivated by converting methanol to hydrocarbons a t 425OC, pmethanol=0.4 bar, WHSV = 1 h-' were regenerated by heating in air a t 55OoC for 4 hours. A renewed test under identical reaction conditions failed to show any change in catalytic properties or catalyst lifetime. This is in agreement with the result obtained by Kaiser [ 131, who also found that SAPO- 17 (as well as SAPO-34) can be repeatedly regenerated by calcination in air. We did, however, not carry out any long-term deactivation and regeneration experiments.

4. CONCLUSIONS

SAPO- 17 is a promising catalyst for converting methanol to light olefms. It is

398

particularly selective for forming ethylene which is the dominant product. It deactivates only moderately fast. The simultaneous catalytic testing of erionite and SAPO-17 clearly demonstrated that, in spite of being essentially isostructural, the two materials differ strongly in their catalytic properties. SAPO-17 is superior to erionite both with respect to catalyst lifetime and selectivity into light o l e h s when converting methanol.

The resistance of SAPO- 17 to deactivation has a maxlmum around 425 OC and falls sharply at temperatures 50 OC higher or lower. Deactivated SAPO- 17 can be fully regenerated.

6. ACKNOWLEDGEMENTS

S. Nawaz wishes to thank the "SPUNG Committee" of the Royal Norwegian Council for Scientific and Industrial Research and Norsk Hydro for financial support. Thanks are due to Dr. Frank RCSher, Leipdg (Germany) for providing the erionite sample.

6. REFERENCES

[ 11 C. D. Chang and A. J. Silvestri, J. Catal., 47 (1977) 249. 121 C. D. Chang, Catal. Rev.-Sci. Eng. 26 (1983) 1. 131 S. Yurchak, Stud. Surface Sci. Catalysis, 36 ( 1988) 25 1. 141 W. W. Kaeding and S. A. Butter, J. Catal., 61 (1980) 155. 151 C. D. Chang, C. T-W. Chu and R. F. Socha, J. Catal., 86 (1984) 289. 161 C. T-W. Chu and C. D. Chang, J. Catal., 86 (1984) 297. (71 C. D. Chang, W. H. Lang and A. J. Silvestri, U. S. Patent 4,062,905 (1977). 181 S. Cartlidge and R. Patel, Stud. Surface Sci. Catalysis, 49 (1989) 1151. [9] E. M. Flanigen, B. M. Lok, R.L. Patton and S. T. Wilson, Proc. 7th Int. Zeolite

Conference, Tokyo 1986, Kodansha-Elsevier 1986, p. 103. [ 101 J. A. Rabo, Periodica Polytechnica Chemical Engineering, 32 ( 1988) 2 1 1. 1111 B. M Lok, C. A. Messina, R. L. Patton, R. T. Gajek, T. R. Cannan,E.

M.Flanigen, US. Patent. 4,440,871. 1121 B. M. Lok, C. A. Messina, R. L. Patton, R. T. Gajek, T. R. Cannan, E. M.

Flanigen, J. Chem. SOC. 106 (1984) 6092. 1131 S. W. Kaiser, Arab. J. Sci. Eng. 10 (1985) 361. 1141 S. Nawaz, S. Kolboe, S. Kvisle, K. P. Lillerud, M. Stdcker and H. M. Bren,

Stud. Surface Sci. Catalysis, 61 ( 199 1) 42 1. [15] R. Von Ballmoos and J. B. Higgins, Collection of simulated XRD Powder

Patterns for Zeolites, Zeolites 10 (1990) 313-5 14.