Ethane dehydrogenation using a high-temperature catalytic membrane reactor

joumal of

MEE%:E ELSEVIER

Journal of Membrane Science 90 ( 1994) 1 l- 19

Ethane dehydrogenation using a high-temperature membrane reactor

catalytic

E. Gobina, R. Hughes* Department of Chemical Engineering, Universityofsalford, Salford h4.5 4 W, UK

(ReceivedAugust 27, 1993; accepted in revised form December 17, 1993)

Abstract Experiments have been conducted for the dehydrogenation of ethane to ethylene using a high-temperature catalytic membrane reactor under isothermal conditions. Typically, conversions of up to 7 times (cocurrent mode) and 8 times (countercurrent mode) higher than the equilibrium value achievable in conventional fixed-bed reactors have been attained at high sweep flow rates. The membrane used in this study was a thin layer of Pd-23 wt% Ag on porous Vycor glass. The significant improvement in the ethane conversion is attributed to the exclusive and continuous permeation of hydrogen through the membrane. Key words: Catalytic dehydrogenation;

Metal membranes;

Membrane reactors; Gas separations

1. Introduction Membrane reactors are unique in that they are capable of combining chemical reaction and separation in a single-unit operation. The membrane, being selective to one or more of the product or reactant species, can be used to improve the yields of thermodynamically limited reactions. Although the basic concept of membrane reactors has been known for almost three decades [ l-3 1, it is only recently that the application of the concept is being recognised [ 4-6 1. Low-temperature applications ( < 1OOoC) were the first to benefit from this concept and an extensive review has been carried out by Chang and Furusaki [ 7 1. Most of the low-temperature applications utilise organic membranes. The *Correspondingauthor. Tel: 61-745-5081, Fax: 61-745-5999, Email: R. Hughes @ Chemistry. Salford. ac.uk.

availability of high-temperature, porous, ceramic and metallic membranes has opened up new avenues to researchers and engineers for carrying out catalytic processes at elevated temperatures. This is because ceramic and metallic membranes in contrast to their organic counterparts, are robust, can function for prolonged periods at high temperatures with great stability, possess improved pore size controllability and have higher rates of heat transfer. This development has renewed interest in the use of hightemperature catalytic membrane reactors and is the subject of recent reviews [ 8,9]. Table 1 presents a selection of experimental studies, for gasphase reactions, utilising porous ceramic and metallic membranes, attempted by several workers to enhance the conversion of equilibriumlimited reactions. One particular class of reactions of industrial importance is the dehydrogenation of lower al-

0376-7388/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDZO376-7388(94)00011-M

E. Gobina, R. Hughes /Journal of Membrane Science 90 (1994) I I- 19

12 Table 1 Experimental

studies of gas-phase

reactions

using membranes

Membrane material

Catalyst

Chemical studied

Porous A1203

MoSz in pores of membrane

Porous A1203

Cr209 (20 wtl)

Microporous

glass

Al,Os

Pt/A1203

Porous supported multrlayer A1203

Temperature range ( “C)

Enhancement achieved

Ref.

Decomposition of H2S to form H2 and S Dehydrogenation of propane to propene

800

10

Dehydrogenation cyclohexane

215

Two-fold increase on the equilibrium conversion Conversion increased from an equilibrium value of 40.1 to 58.7% Conversion increased from 35% without membrane to 80% with membrane Conversion of ethylbenzene in the presence of the membrane were 15% higher than those in the absence of the membrane Conversion from an equilibrium value of 26 to 56% in the membrane Complete conversion of CO attained in the membrane reactor compared to the equilibrium conversion of 60% Conversion of C6Hr2 was 99.7% in the membrane reactor compared with an equilibrium conversion of only 18.7% The conversion of HI was 4%, a twenty-fold increase from a value of 0.2% in the absence of a membrane

reaction

of

575

Dehydrogenation of ethylbenzene to styrene

600-640

270

Porous Vycor glass

Pt cat. within pores of the membrane (34 wt% Pt )

Dehydrogenation cyclohexane

Microporous glass/thin Pd film

Thin dense film (20 pm) of Pd coating the external surface of the glass

Water gas shift reaction

400

Dense Pd tube

Pt/Al,O,

Dehydrogenation cyclohexane

200

Non-porous tube

Gas-phase

Pd/Ag

(0.5 wt% Pt)

decomposition

Decomposition of hydrogen iodide

kanes. These yield valuable commodity chemicals for the production of alcohols, gasoline blends and polymers. However, these reactions are also limited by their thermodynamic equilibrium and to obtain higher yields, very high temperatures are required resulting in losses in both selectivity and overall catalyst activity. An incentive therefore exists for a process that has the capability of overcoming the thermodynamic equilibrium while maintaining the integrity of the catalyst and improving the product yield. Membrane reactor technology appears to be promising since the selective removal of hydrogen from the reaction products would inevitably tend to shift the equilibrium towards increased product yield. The integration of chemical reaction and separation in a membrane reactor would then result in better rate control because of shorter contact time and reduced temperature of operation. When the membrane has also catalytic proper-

of

of

500

II

12

13

14

15

16

17

ties, even more avenues are opened and, for example, one can couple reactions in a membrane reactor. Experimental investigations that have been attempted for reactions accompanying membrane separation with the objective of shifting the equilibrium towards increased conversion include amongst others (see Table 1) the decomposition of hydrogen sulfide [ 18 1, and decomposition of hydrogen iodide [ 191 using microporous glass membranes. Microporous alumina membranes have been used to investigate the reaction of propane to propylene [ 201, ethane to ethylene [ 2 1 ] and for the dehydrogenation of ethylbenzene to styrene [ 131. In these investigations, the flow of reactant and product species through the membrane followed a Knudsen diffusion mechanism and reactor performance was much improved compared with a conventional catalyst-packed reactor (see Table 1) . However,

E. Gobina, R. Hughes /Journal of Membrane Science 90 (I 994) I l-1 9

computer simulations and experiments [ 141 have established that there is a limitation on the performance of microporous membrane reactors and it is almost impossible to make the reaction proceed beyond a certain level of conversion without the need to recycle the unreacted feed since some ‘slip’ of feed also passes through the micropores of the membranes into the separation side. This limitation can be overcome by using a membrane that is selective only to product(s). This principle has been demonstrated by Itoh [ 161 who succeeded in obtaining almost 100% conversion for the dehydrogenation of cyclohexane when using a palladium membrane instead of microporous glass membrane. A 20-fold increase was obtained for the decomposition of hydrogen iodide by using a non-porous palladium/silver tube [ 171, far higher than that obtained by using a porous membrane [ 19 1. This high increase was attributed to the exclusive separation of hydrogen. Palladium and palladium alloys are the current preferred methods for removing hydrogen exclusively. However, existing commercially available palladium membranes are too thick ( N 53 pm) to provide economic rates of permeation and palladium being an expensive metal, a great thickness also means more cost. Although the selectivity of microporous membranes is limited, they could be employed as substrates on which a thin metal layer of a dense membrane material such as palladium may be deposited to form a coherent layer having exclusive permeation to hydrogen. We have previously reported on the application of a high-temperature membrane reactor for the selective permeation of hydrogen from an N2-Hz gaseous mixture [ 22 ] and the characterisation of Pd-metallised porous Vycor glass produced by the process of magnetron sputtering [ 231. In this paper, as part of an ongoing research into membrane reactor applications, we present results for the catalytic dehydrogenation of ethane in a high temperature catalytic membrane reactor promoted by 0.5 wt% Pd/A1203 catalysts packed in the tube side of the reactor. The significant equilibrium shift observed shows

13

that the concept of overcoming thermodynamic equilibrium limitations with consequent improved product yields is feasible with a careful choice of membrane.

2. Experimental The experimental equipment used in this study is similar to that employed earlier [ 231 and consists of a feed gas delivery system, a high-temperature membrane reactor, a gas chromatograph and an on-line data acquisition system. The membrane is a composite consisting of a porous Vycor glass tubular substrate ( 10 mm o.d., 1.1 mm thick and pore size of 40 A) onto which a thin continuous film of Pd-Ag was deposited by the process of magnetron sputtering. The process of magnetron sputtering can be used to deposit very thin films on almost any porous substrate. It involves ion bombardment of the target, thus ejecting atoms which then adhere onto the porous substrate. By exposing the substrate to a radio frequency (rf ) glow discharge (plasma), the adhesion strength of the film can be enhanced. Film thickness (and hence deposition rates) were estimated using additional samples of porous Vycor glass in the deposition chamber placed adjacent to the actual substrate of interest. Direct measurement of the film thickness on these samples was carried out using scanning electron microscopy (SEM ). The membrane thickness in this investigation was 6 ,um. Film thickness uniformity was measured at various locations along and across the samples by electrical resistivity. The variation along the membrane averaged 6.7%. Film stoichiometry was determined by the electron probe microanalysis technique (EPMA ) . A total of 50 analyses was carried out over a randomly selected area, and an average value obtained. There was very little difference between the Pd: Ag content in the target (77 : 23 wt%), and that on the substrate (23 + 3 wt% Ag ) . There was no light transmission across the membranes indicating a dense structure. Over an extended period of use in the membrane reactor, there was no physical peeling from the substrate.

14

E. Gobina, R. Hughes /Journal of Membrane Science 90 (1994) 1 I-I 9

This is indicative of the good adherence strength of the film. The high-temperature membrane reactor, details of which are schematically shown in Fig. 1, consists of an outer stainless steel shell (with provision for either inflow or outflow of sweep gas) and an inner tube. The membrane tube is centralised in the reactor by a set of moulded graphite rings packed at each end of the reactor. Careful tightening of stainless steel plugs on both ends compresses the rings and completes the seal between the shell and tube. The thickness of the Pd-Ag film was 6 pm and the area available for hydrogen permeation was 47.14 cm2. Three 0.5 mm thermocouples were located along the axis of the catalyst bed to determine the temperature profile inside the reactor. A similar size thermocouple was located at the midpoint of the shell side. Steady-state temperature was attained when the variation in the thermocouple readings were within 22°C of the desired temperature. The catalyst for the dehydrogenation reaction was located inside the tube and consisted of 2.55 g of 0.5 wt% Pd/A1203 cylindrical pellets (3.35 mm o.d.x3.63 mm long) which were carefully and uniformly packed in the bed. The catalysts were characterised by various techniques includ-

Fig. 1. Essential details of the catalytic membrane reactor.

ing BET surface area using nitrogen adsorption, SEM and EPMA. The catalyst was used mainly to promote the dehydrogenation reaction. All experiments were performed at a total feed pressure of 128.7 kPa ( 1.27 bar) while the shell pressure was maintained at 10 1.3 kPa ( 1 bar). The catalytic membrane and the catalyst pellets were reduced in a flow of pure hydrogen at 2.5 cm3 (STP) /min for 30 min at a temperature of 400°C prior to the reaction experiments. The feed gas mixture consisting of a single feed of 50 ~01% ethane in nitrogen was then allowed to flow at a predetermined rate over the catalyst pellets; this continued until temperature stability was achieved. Due to the endothermicity of the dehydrogenation reaction, a small temperature drop in the catalyst bed was observed. This was compensated by increasing the heat input to maintain the desired temperature in the bed. Temperature variation along the reactor was then less than 3 “C. Membrane performance was assessed by comparing the shift observed from the equilibrium conversion of ethane. This equilibrium conversion was determined by closing the shell-side inlet, so that shell-side sweep flow was absent and continuously monitoring the effluent from the catalyst bed via an air-activated automatic sampler. This was continued until constant concentrations were obtained as indicated by identical values of the peak chromatograph areas. The ethane conversion calculated under these circumstances was the equilibrium conversion and was then used for comparison with conversions obtained when the shell side was opened and the sweep gas consisting of pure nitrogen was allowed to flow. Both the permeate and retentate streams were sampled. The shell-side sweep gas flow was then varied and flow directions reversed for either cocurrent or countercurrent operations while maintaining the temperature and feed mole fraction of ethane constant. Reaction experiments were conducted in the reactor using a single feed consisting of a 50 ~01% mixture of ethane in nitrogen, the nitrogen serving as an internal standard for the analysis of gas composition. All experiments were carried out at 660 K under isothermal conditions.

E. Gobina, R. Hughes /Journal of Membrane Science 90 (I 994) 1 I-l 9

3. Results and discussion 3.1. Permeability of membrane under nonreactive conditions The permeability of the membrane to hydrogen was studied using a 60% Ha: 40% Nz gas mixture, varying the driving force and the temperature in order to obtain quantitative information regarding the permeability of the membrane. If the permeate rate of hydrogen gas through the Pd : Ag membrane [ QH ( cm3 ) ] is assumed to obey the half-power pressure law [ 241, then:

QH =Qopt” -pi”)

(A/d,)

(1)

where Q0 is the permeability constant of hydrogen gas through the membrane (cm3 cm/cm2 s atm”‘) ; A is the membrane area available for flow (cm2); and dt is the film thickness of the membrane (cm). A typical plot of Qu versus (P ;I2 -pii ) at various operating temperatures is given in Fig. 2. The permeation rate of hydrogen gas is directly proportional to the difference

in the square roots of the upstream and downstream hydrogen partial pressures. This is consistent with Sieve&s law [ 25 1, and indicates that permeation through the bulk of the metal is the rate-limiting step. A closer inspection of Fig. 2 also reveals that there is a temperature effect on the permeability constant. This was further investigated by maintaining a constant pressure difference while varying the temperature. The results correspond to the Arrhenius law and the permeability constant (Q ) of hydrogen gas through the film can then be expressed as follows: Q,,=7.174x

10m5exp( -6.380/RT)

(2)

where R is the ideal constant (kJ/mol K) and T the absolute temperature (K). The activation energy of 6.38 kJ/mol obtained in this investigation compares well with reported values of 5.73 [26], 6.60 [27], and 5.86 kJ/mol [28]. In these previous investigations, similar Pd-Ag alloy membranes were employed. 3.2. Ethane dehydrogenation experiments The equilibrium conversion of ethane to ethylene was determined twice (before and after experiments in the membrane reactor). An average value of 2.57% was obtained and used for comparison with the membrane reactor conversions. This value is simply the maximum conversion achievable in the absence of hydrogen separation and may be compared with a value of 0.87% estimated from thermodynamic data for the same temperature. A similar three-fold increase in the measured equilibrium constant for this reaction compared to the thermodynamic value was obtained by Champagnie et al. [ 2 I 1. Operation of the catalytic membrane reactor was achieved by varying the flows of the feed and/or sweep gas at constant temperature using a fixed weight of catalyst. The overall conversion of ethane to ethylene, allowing for volume changes was obtained using the equation,

1.6 .

X=( FA;rOFA)x 100% Fig. 2. Effect of pressure and temperature rate of hydrogen.

15

(3)

on the permeation

where X is the overall ethane conversion to eth-

E. Gobina, R. Hughes /Journal ofMembrane Science 90 (1994) 11-I 9

16

ylene; FAOand FA are the respective flow rates of ethane in the feed and product in the tube side. No ethane or ethylene was detected in the shell side exit confirming that the membrane is selective only to hydrogen. 3.3. Conversion vs. time factor The effect of time factor ( W/F,,) defined as the weight of catalyst in the bed, W (g), divided by the molar flow rate of ethane, FAO(mol/s), on the conversion of ethane to ethylene is presented in Fig. 3. The conversion of ethane depends on the mean residence time of ethane in the reactor. This means that conversion increases with the time factor as expected. In the catalytic membrane reactor, at long contact times, the conversion of ethane to ethylene becomes higher than the equilibrium value because product hydrogen is preferentially and exclusively diffused through the membrane. This selective and continuous removal of hydrogen from the reaction zone, results in a shift in the equilibrium towards increased ethane conversion. At very short contact times, equilibrium shift is prevented because the mean residence time of ethane in the reactor is too short. Under conditions of varying feed flow rate at constant catalyst weight, the external mass transfer rate changes with the velocity of the gas

JA

0.5

1.5

2.5

3.5

I

TIblE PKTOH ,!4,FA<>i x

105

vi-at

Z,WOl

Fig. 3. Influence of time factor ( W/F,,) on the conversion of ethane in the membrane reactor. Temperature, 600 K; catalyst weight, 2.55 g; flow rate of sweep gas, 300 cm3 (STP)/ min.

stream. The mass transfer flux, N, to the inner surface of the membrane is given by N= (~1 -pz)Sh

D,,ld

(4) where p1 and p2 are the molar densities of hydrogen on the reaction and permeate side, respective ( mol/cm3), DABis the diffusivity of hydrogen in the reaction mixture (cm2/s), d is the reactor diameter (cm) and Sh is the Sherwood number (k,d/D) . The permeate-side concentration of hydrogen was very small due to the presence of the sweep gas. Under the conditions of the present study, the flow on the reaction (tube) side was laminar and the Sherwood number is given by [ 29 ] :

where Re and SC are the Reynolds and Schmidt numbers, h and h are the fluid viscosities at average bulk temperature and the wall, respectively, and L the length of reactor across which permeation of hydrogen occurred. Using this correlation for Sh and the flux equation (4)) the hydrogen mass transfer flux from the bulk gas to the inner membrane surface was always greater than the permeation flux and therefore mass transfer in the tube side was not limiting. A comparison of the efficiency of the modes of operation of the catalytic membrane reactor is shown in Fig. 4. Both countercurrent and cocurrent modes of operation were examined. In the countercurrent mode, the flow of the sweep gas was in the opposite direction to that of the feed while in the cocurrent mode both were in the same direction. Experimental results show that operation of the membrane reactor in the countercurrent mode is more desirable compared to that in the cocurrent mode. This is because higher conversions can be obtained for the same sweep flow in the countercurrent mode than for the cocurrent mode. This conclusion was also arrived at by computer simulation results, in a membrane reactor hy Itoh [ 16 ] for the dehydrogenation of I-butene and shown experimentally for the dehydrogenation of cyclohexane [ 281. In these studies Pd-Ag membranes similar to that used in the present study were employed.

17

E. Gobina, R. Hughes /Journal of Membrane Science 90 (1994) 11-I 9

Fig. 4. Effect of sweep gas flow rate on ethane conversion. (-- ) Equilibrium; ( 0 ) cocurrent flow; ( A ) countercurrent flow. Time factor ( W/F,,) = 1.144 x 10’ g cat s/g mol; temperature, 660 K, space time, 8.4 s.

tinuously from the reaction zone. Since these PdAg membranes are selective only to hydrogen, an increase in the sweep ratio ensures a large hydrogen partial pressure difference between the shell side and tube side. The partial pressure difference across the membrane is the driving force for hydrogen permeation which results in an equilibrium shift towards higher ethane conversion. At a sweep ratio of 5, for example, the ethane conversion increases from the equilibrium value of 2.5 (without any flow of sweep gas) to 17.66%. This is a 7-fold increase, and is attributed to the selective and continuous removal of hydrogen from the reaction zone. This result compares with a 6-fold increase reported by Champagnie et al. [ 2 I] who employed a Pt-impregnated porous alumina membrane having a pore size of 40 A, for ethane dehydrogenation. By using a dense metallic membrane in this study, we have thus eliminated the problem of back-permeation of reactant from the permeate side to the feed side at high conversions and re-equilibration [ 301 often encountered in porous membrane reactor systems. 3.5. Performance of the membrane reactor

Fig. 5. Membrane reactor conversion as a function of the sweep gas/feed gas flow ratio. Temperature, 660 K; cocurrent flow.

3.4. Effect of sweepjlow/feedjlow Fig. 5 shows the conversion of ethane to ethylene at a temperature of 660 K as a function of the sweep gas ratio; defined as the ratio of the flow rate of sweep gas in the shell side to the feed inlet flow rate. The feed-side inlet consisted of a 50 ~01% ethane in nitrogen mixture. Increasing the sweep ratio increases the ethane conversion. This is due to the fact that in the catalytic membrane reactor, conversion is increased by the selective and exclusive removal of hydrogen con-

Table 2 shows the relationship between membrane performance and rate of reaction at high performance, i.e., at a conversion of N 18%. The rate of Hz permeation is 6.7 times the production rate. This suggests that the membrane has sufficient capacity to permeate the hydrogen and that the flow through the membrane is not rate limiting. Although the hydrogen removal rate is far in excess of the production rate, Fig. 3 suggests that at sufficiently long contact times the conversion attains a virtually constant value. The reason for this is not clear; one possibility is that Table 2 Estimation of catalytic membrane reactor performance Temperature (K)

Ethane conversion W)

H2 production rate (gmof/s)

H2 permeation rate [usingpermeability Eqs. (2) and (3)]

(gmol/s) 660

18

4.0x

10-h

26.7X IO-6

18

E. Gobina. R. Hughes /Journal of Membrane Science 90 (I 994) I I- 19

readsorption of ethylene and hydrogen on the surface of the catalyst pellets occurs under these conditions and this is related to the relative high surface area of the catalyst employed (-200 m2/g). The dehydrogenation of 1 mol of ethane yields equimolar amounts of ethylene and hydrogen. The selectivity of ethylene calculated when no sweep gas was present was 50%. However, when a sweep gas flow was used this selectivity was close lOO%, indicating that almost all of the hydrogen in the gas phase was removed by the membrane.

4. Conclusions It has been shown experimentally that a much higher level of ethane conversion can be attained even in a single-stage catalytic membrane reactor. The significant characteristic advantages are that it is possible to carry out the reaction at a lower reaction temperature and shorter contact time. This results in a shorter reactor length and lower energy requirement for eventual separation of the product(s) and reactant (s ) from each other and from the reaction mixture. Such a reactor could be applied to study other dehydrogenation reactions of industrial importance such as the conversion of ethylbenzene to styrene, butene to butadiene and methanol to formaldehyde. Because the metallic surface is in itself catalytic to a host of hydrogenation reactions, there exists a possibility of coupling the endothermic dehydrogenation with an exothermic hydrogenation reaction. This would assist transport of both hydrogen and heat in the system. Work in this area is continuing.

5. Acknowledgements We thank the SERC for continued financial support in this research. We are also grateful to Dr. Dermot Monaghan for his assistance in carrying out the magnetron depositions in the substrate. Characterisation of the catalyst pellets and palladium film stoichiometry was undertaken by

Johnson Matthey Research Centre, Sonning Common to whom we express our thanks.

6. References [ 1]A.S. Michaels, New separation technique for the CPI, Chem. Eng. Prog., 64( 12) (1968) 31. 2lD.I.C. Wang, A.E. Humphrey and E. Arthur, Enzyme detergents, already a multimillion dollar industry, are but one concrete result from biochemical engineering. Chem. Eng., 86(27) (1969) 108. 3 ] W.C. Pfefferle, Process for dehydrogenation, US Pat. 3,290,406 (1966). [ 41 J.F. Roth, Future catalysis for the production of chemicals, in J.W. Ward (Ed.), Catalysis 1987, Elsevier, Amsterdam, 1988. [ 5 ] V.M. Gryaznov, Hydrogen permeable palladium membrane catalysts, an aid to the efficient production of ultrapure chemicals and pharmaceuticals. Platinum Met. Rev., 30(2) (1986) 68. 6 ] V.M. Gryaznov., VS. Smimov and M.G. Slinko, in J.W. Hightowes (Ed.), Proc. 5th Int. Congr. Catal., Vol. 2, North Holland, Amsterdam, 1973, p. 1139. 17lH.N. Chang and S. Furusaki, Membrane Bioreactors: Present and Prospects, Adv. Biochem. Eng. Biotechnol., 44 (1991) 27. 8]J. Shu, B.P.A. Grandjean, A. Van Neste and S. Kaliaquine, Catalytic palladium-based membrane reactors: a review, Can. J. Chem. Eng., 69 ( 199 1) 1036. [ 9 ] J.N. Armour, Catalysis with permselective inorganic membranes, Appl. Catal., 49 ( 1989) 1. [ lO]T. Kameyama, K. Fukuda, M. Fujishige, H. Yokohawa and M. Dokiya, Production of hydrogen from hydrogen sulfide by means of selective diffusion membrane, Adv. Hydrogen Energy Prog., 2 ( 1981) 569. [ 1llJ.G.A. Bitter, Br. Pat. Appl. 8629135 (1986). [ 1210. Shinji, M. Misono and Y. Yoneda, The dehydrogenation of cyclohexane by the use of porous-glass reactor, Bull. Chem. Sot. Jpn., 55 (1982) 2760. [ 13 ] J.C.S. Wu, T.E. Gerdes, J.L. Pszczolkowski, R.R. Bhave and P.K.T. Liu, Dehydrogenation ethylbenzene to styrene using commercial ceramic membranes as reactors, Sep. Sci. Technol., 25 (1990) 1489. [ 141Y.M. Sun and S.J. Khang, Catalytic membranes for simultaneous chemical reaction and separation applied to dehydrogenation reaction. Ind. Eng. Chem. Res., 27 (1988) 1136. [ 15 ] E. Kikuchi, S. Uemiya, N. Sato, H. Inoue, H. Ando and T. Matsuda, Membrane reactor using microporous glasssupported thin film of palladium. Application to the water gas shift reaction, Chem. Lett., 3 ( 1989) 489. [ 16]N. Itoh, A membrane reactor using palladium. AIChE J., 33 (1987) 1576.

E. Gobina, R. Hughes /Journal oJMembrane Science 90 (1994) 1 l-l 9 [ 171 J. Yeheskel, D. Leger and P. Courvoisier, Thermal decomposition of hydroiodic acid and hydrogen separation, Adv. Hydrogen Energy, 2 (1979) 569. [ 18]T. Kameyama, M. Dokiya, M. Fujishige, H. Yokohawa and K. Fukuda, Possibility for effective production of hydrogen from hydrogen sulfide by means of a porous glass membrane, Ind. Eng. Chem. Fundam., 20 ( 198 1) 97. [ 19]N. Itoh, Y. Shindo, K. Obata, T. Hakuta and H. Yoshitome, Simulation of a reaction accompanied by separation, Int. Chem. Eng., 25 ( 1985) 138. [20]Z.D. Ziaka, R.G. Minet and T.T. Tsotsis, Propane dehydrogenation in a packed-bed membrane reactor, AIChE J. 39 (1993) 526. [Zl]A.M. Champagnie, T.T. Tsostsis, R.G. Minet and I.A. Webster, A high temperature membrane reactor for ethane dehydrogenation, Chem. Eng. Sci., 45 ( 1990) 2423. [22]E. Gobina and R. Hughes, Selective high temperature membranes for hydrogen separation, I993 IChemE Research Event, Institution of Chemical Engineers, Rugby, UK, 1993, p. 522. [ 231 E. Gobina and R. Hughes, High temperature selective membranes for hydrogen separation, Dev. Chem. Eng. Min. Process., accepted for publication.

19

[ 24]G. Bohmholdt and E. Wicke, Diffusion von H und D in Pd und Pd-Legierungen. Z. Physik. Chem. N.F., 56 (1967) 133. [25]A. Sieverts and W. Kumbhaar, Solubility of gases in metals and alloys, Ber. Dtsch. Chem. Ges., 43 ( 1910) 893. [ 26]H. Yoshida, S. Konishi and Y. Naruse, Effects of impurities on hydrogen permeability through palladium alloy, J. Less-Common Met., 89 (1983) 429. [27]F.J. Ackerman and G.J. Koskinas, Permeation of hydrogen and deuterium through palladium-silver alloys, J. Chem. Eng. Data, 17 ( 1972) 5 1. [ 281 J. Chabot, J. Lecomte, C. Grumet and J. Sannier, Fuel clean-up system. Poisoning of palladium-silver membranes by gaseous impurities, Fusion Technol., 14 (1988) 614. [ 29lC.O. Bennett and J. Myers, Momentum, Heat and Mass Transfer, McGraw-Hill, New York, 1982, p. 557. [ 30lK.C. Cannon and J.J. Hacskaylo, Evaluation of palladium-impregnation on the performance of Vycor glass catalytic membrane reactor, J. Membrane Sci., 65 (1992) 259.