J. CHEM. SOC. FARADAY TRANS., 1994, 90(14), 2147-2153 Mechanistic Study of sec-Butyl Alcohol Dehydration on Zeolite H-ZSM-5 and Amorphous Aluminosilicate M. A. Makarova,? C. Williams$ and K. 1. Zamaraev" Institute of Catalysis, Russian Academy of Sciences Siberian Branch ,Prospekt Akademika Lavrentieva , 5,Novosibirsk 630090,Russia J. M. Thomas* Davy Faraday Research Laboratory, The Royal Institution, 21 Albemarle Street, London, UK WIX 4BS The dehydration of sec-butyl alcohol has been studied by in situ FTlR and gas-chromatographic (GC) kinetic methods in the range 60-140°C on zeolite H-ZSM-5 and amorphous aluminosilicate (AAS) samples with a well characterized number and strength of Br~nsted acid sites. Under flow conditions (GC kinetic studies), the reac- tion yields butenes [but-1-ene, (Z)-and (€)-but-e-ene] and water, with an activation energy of 40 1 kcal mol-' determined from steady-state data.Under non-steady-state conditions, the so-called 'stop effect ' is observed : namely, an increase in the rate of butene evolution (as compared with that at steady state) when the flow of alcohol into the reactor is halted. The course of dehydration on H-ZSM-5 in a static IR cell was followed by the appearance and growth of a peak for adsorbed water (water deformation peak at 1640 cm-'). The rate constant determined from the kinetics of water formation in the FTlR experiments (1.1 x s-' at 70°C) is found to be 400 times as high as the rate constant calculated from GC steady-state kinetic data. All these anomalous pheno- mena observed under flow conditions (the low rate of reaction, the high activation energy and the 'stop effect') can be explained by the slowing down of dehydration under these conditions as a result of the reverse reaction, i.e.the hydration of the product butene with product water. When the zeolite pores are free from physically adsorbed reactants (in the FTlR experiments or during the 'stop effect'), the extent of the reverse reaction decreases and the rate of butene formation increases. On AAS, whiFh has acid sites of Fimilar strength, but which has a much more open surface (average pore diameter ca. 50 A compared with 5.5 A for ZSM-5), similar effects are observed, but they are much less pronounced. This probably arises from the lower reactant concen- tration in the AAS at steady state and hence, a lower concentration of water in the vicinity of the active sites.1. Introduction It is generally accepted that the presence of microporous channels is one of the most significant factors influencing reactions in zeolites.' Nevertheless, the mechanism of their influence is still under discussion.2 The present work is part of a series of investigations aimed at elucidating the influence of pore confinement on reactions in zeolites. As a model reac- tion we have chosen the dehydration of butyl alcohols in H-ZSM-5 and results for n-, iso- and tert-butyl alcohol have already been p~blished.~-~ This article is devoted to the fourth isomer, namely sec-butyl alcohol.For a better under- standing of the processes that the alcohol molecules undergo inside the zeolite channels, we have adopted the following approaches, refined by us in earlier studies: (1) comparison of the kinetic parameters of dehydration obtained from studying products remaining adsorbed in the zeolite pores (in situ FTIR studies) and products that are desorbed (GC studies); (2) comparison of the course of reaction under steady-state and non-steady-state kinetic regimes ; (3) comparison of the course of reaction taking place on 'enclosed' microporous ZSM-5 zeolite samples with that on a sample with an 'open' surface, such as amorphous aluminosilicate (with an average pore diameter of ca. 50 A). t Present address : Department of Chemistry, University of Man- Chester Institute of Science and Technology, P.O.Box 88, Manches-ter, UK M60 1QD.1C. Williams was also based at the Davy Faraday Research Laboratory, Royal Institution, London, when the work was carried out. Present address : Koninklijke/Shell-Laboratorium, Badhuisweg 3, 103 1 CM, Amsterdam, The Netherlands. 2. Experimental The samples used in the present study (four samples of ZSM-5 and one sample of amorphous aluminosilicate, AAS) were described in detail in ref. 4-6. Their main characteristics are summarized for convenience in Table 1. GC kinetic studies were carried out using a flow micro- reactor system with on-line GC analysis. Full details were given in ref. 5. Briefly, the reactor was a quartz U-shaped tube loaded with 0.005-0.05 g of a pelletized zeolite catalyst (0.3-0.5 mm fraction).The system allows one to feed into the reactor either pure helium or a helium-alcohol gas mixture (usually 2.0 mol% sec-butyl alcohol, residual water content <10 ppm). The gas flow rate was typically 30-40 cm3min-' at an overall pressure of 1 atm. In order to avoid adsorption of reactants on the tube walls, the sampling system was ther- mostatted at 70°C. Before reaction, the sample was treated Table 1 Characterization of aluminosilicate samples sample crystallite size/pm Si :Al N/1OZ0sites g-' ZSM-5 1 <1 42 : 1 2.3 2 0.5-4 20 : 1 3.3b 4 15-20 35 : 1 2.8 5 4-6' 35 : 1 2.8 AAS loood 1.1 Sample numbers are the same as in ref. 4-6, Scanning electron micrographs for the zeolite samples are given in ref.4 and 6. Detailed characterization of AAS is given in ref. 3(b). a Per g of dehydrated sample. Na :A1 = 0.35 : 1. Aggregates of 0.1 pm crystallites. Average pore diameter ca. 50 A. 2148 for 1 h in a flow of oxygen at 500°C and for 2 h in a helium flow at 450°C. The dehydration reaction was studied in the temperature range 80-140 "C. During reaction, conversion was kept to < 10%. The reaction was zero order with respect to the alcohol. The reaction rate (in terms of butene and water evolution) was defined as : W(C4H8) = F[C4H81/m W(H,O) = F[H,O]/m where F is the helium-alcohol flow rate (cm3 s-'), [C,H,] and [H,O] are the concentrations of products determined by GC (molecules cm-3) and m is the sample mass (g).The rate at which alcohol was incident on the sample or the rate of exit of unreacted alcohol was defined as before:' V = F[C,H,OH]/m where [C,H,OH] is the concentration of butanol determined by GC before and after the reactor. FTIR studies of the dehydration raction were carried out on a Bruker FTIR spectrometer (IFS-l13V), using a ther- mostatted in situ cell. The construction of the FTIR cell was described in more detail in ref. 5. The samples were pressed into self-supporting discs (mass typically 25 mg, p = 6-12 mg an-,). Before adsorption, the discs were calcined for 1 h in air and for 2 h in vacuum (lo-, Torr) at 450°C. The sample was then cooled to the desired temperature and the IR spec- trum of the dehydrated sample was recorded for reference purposes.After injection of the desired amount of alcohol into the cell ([alcohol] : [acid sites] = 0.5 : l), the kinetics of reaction were studied by following changes in the IR spectra with time. In these kinetic studies, spectra (of 10 scans each) could be collected every 25 s or so in the wavenumber inter- val 4000-1200 cm-' (resolution 4 cm-'). By plotting the dif- ference between these spectra and the spectrum of the purely dehydrated zeolite, the spectrum of the adsorbed species was obtained. For non-kinetic work a typical number of scans was 200. 3. Results 3.1 GC Kinetic Studies 3.1.1 Beginning of Reaction and Steady State Dehydration of sec-butyl alcohol on H-ZSM-5 was studied in the temperature interval 80-140°C. The only products of reaction are butene isomers and water; no ether formation was observed.Typical reaction kinetics at 126°C are shown in Fig. 1. At the beginning of reaction the trough in the V(t) curve shows alcohol adsorption. Simultaneously with saturat- ing the sample with alcohol, the rate of product formation increases, reaching a steady-state value when adsorption of alcohol is complete. Note that in the initial non-steady-state phase of reaction, the rates of formation of both products, butene and water, are almost identical. This significantly differs from the case of isobutyl alcohol dehydration described by us in ref. 5, where a much larger amount of water was evolved in comparison with butene before the sample became saturated with alcohol.At lower reaction temperatures, similar kinetics are observed to these at 126 "C. At higher reaction temperatures there is some decrease in the rate of reaction in the region following alcohol adsorption. This most likely arises from oli- gomerization of butene on the catalyst active acid sites (analogous to that observed for the isobutyl alcohol/H-ZSM- 5 and n-butyl alcohol/H-ZSM-5 systems described previously in ref. 3b and 4), resulting in gradual catalyst deactivation. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 i 31 'E mI I A I ,o 0 10 20 " 60 ti me/min Fig. 1 Kinetics of sec-butyl alcohol dehydration on H-ZSM-5. Flow microreactor, sample 2, m = 0.044 g, T = 126 "C, [ROH] = 1.0 mol%, V, = 3.1 x 10l8 molecules of sec-butyl alcohol gc;: s-'.(0) sec-butyl alcohol, (0)butene isomers, (@) H,O. Rate constants for butene formation on zeolite samples of various crystallite sizes are shown in the Arrhenius plot in Fig. 2. These constants were determined using as a reaction rate the value W,, (the reaction rate after saturation of the sample with alcohol) obtained in a series of experiments, k = W,,JN (where N is the number of active sites). At T < 130"C, W,,, is identical to the steady-state rate of reaction. At T > 130°C (the upper point of the plot), when there is no steady state, then Wsatis the rate of butene formation imme- diately on completion of alcohol adsorption, when catalyst deactivation is as yet still minimal (see ref.4 for more details of the W,,, definition). The fact that all the experimental points satisfactorily lie on one and the same line demon- strates the absence of any diffusion limitations for the dehy- dration of sec-butyl alcohol on H-ZSM-5 under these conditions. From this plot a reaction activation energy of 40 f2 kcal mol-' is determined. The kinetics of dehydration of sec-butyl alcohol on the AAS sample at 126°C are the same as those already seen for the ZSM-5 zeolite. The kinetics of butene formation at this temperature are shown in Fig. 3(4. It is very surprising that the rate of dehydration on the AAS is significantly higher than that on the ZSM-5 zeolite. Thus the rate per g is twice that on ZSM-5 [Fig.1 and Fig. 3(a)], while the rate per -3 I -5. n -7-c Iv)1-s = -9. -1 1--131 . . . . . . . . . 2.2 2.4 2.6 2.8 3.0 2 2 103 KIT Fig. 2 Arrhenius plot for the rate constants of butene formation on dehydration of sec-butyl alcohol on different samples of H-ZSM-5 : (0)sample 1, (A)sample 2, (0)sample 4, (0)sample 5 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 c I v) (a-2 2. 10 0 0s; U The Arrhenius plot for the rate constants of sec-butyl alcohol dehydration on AAS (in terms of butene evolution) obtained from experiments such as that described above, is shown in.Fig. 5. These data allow us to estimate two activa- tion energies: E,, ,,, = 18 f1 kcal mol- and E,, ,= 26 f1 kcal mol- '.Therefore, the two types of aluminosilicates, namely the ZSM-5 zeolite and the amorphous sample, behave quite dif- ferently in the dehydration of sec-butyl alcohol. First, for the zeolite samples, the decrease in the reaction rate as compared with Wsat becomes more significant as the reaction tem-perature is increased, and most likely indicates deactivation of the active sites as a result of butene oligomerization on them. For the AAS sample, the drop in the reaction rate as compared with W,,, becomes more prominent on decreasing the reaction temperature and so is apparently not connected with the oligomerization processes. Secondly, the rate con- stant for dehydration on AAS is higher than that on H-ZSM-5 (by a factor of six at 126°C and even more at lower temperatures).Thirdly, the activation energy in the case of H-ZSM-5 (40kcal mol-') is significantly higher than both activation energies determined for AAS by different methods $2$ 0 10 20 30 40 (18 and 26 kcal mol- I). ti me/min Fig. 3 Kinetics of sec-butyl alcohol dehydration on AAS at different temperatures, flow microreactor, rn = 0.0394 g, [ROH] = 2.0 mol%. T/"C: (a) 126, (b) 106, (c) 68. active site (using data from Table 1) is six times higher. This phenomenon was not observed either in the case of n-butyl alcohol dehydration3' where the rate constant (i.e. the rate of reaction per active site) for the H-ZSM-5 zeolite was slightly higher (by a factor of two to three) than that for the AAS, or in the case of isobutyl alcohol and tert-butyl alcohol dehy- dration, where the rate constants for zeolite and AAS were iden tical.4 At lower reaction temperatures the kinetics of butene for- mation on AAS differ from those observed on H-ZSM-5 [Fig.3(b), (c)]. After reaching W,,, (time ca. 3 min), one sees initially a decrease in the rate of butene formation and only after that, a steady-state region. Note that with decreasing reaction tem- perature, the difference between W,,, and W, (which is the steady-state rate) becomes increasingly pronounced. A more detailed picture of the kinetics of dehydration at an interme- diate temperature of 90°C is given in Fig. 4. It is seen that less water than butene is evolved before reaching the steady state.For instance, at the maximum rate of butene evolution (t = 3 min), water is not observed at all. A ..A b I 0 10 20 30 40 50 ti me/min Fig. 4 Kinetics of sec-butyl alcohol dehydration on AAS. Flow microreactor, m = 0.0535 g, T = 9O"C, [ROH] = 2.0 mol%. (0)sec-butyl alcohol, (0)butene isomers, ((3)H,O. 3.1.2 Non-steady-state Conditions :'Stop Efect ' Experiments under non-steady-state conditions were carried out in the following manner. After reaching steady state, the alcohol-helium reaction mixture was replaced by a flow of pure helium. This moment of switching flows was taken as time zero. The 'reply' of the zeolite catalyst to this change in reaction conditions was then followed. Results at 126°C are shown in Fig.6. Thus, after switching from alcohol-helium to pure helium, the rate of butene formation at first increases sharply and then subsequently decreases. Estimating the amount of butene desorbed from the sample during purging gives a value corresponding to less than half of the number of active sites in the catalyst. On returning to the alcohol- helium flow, the new rate of dehydration has dropped to ca. 50% of its original value. This decrease in the reaction rate can obviously be explained by butene oligomerization on the active sites during purging (when the zeolite channels are relatively empty and so the side reaction, oligomerization, is favoured). These results agree with those obtained for iso- butyl alcohol dehydration on H-ZSM-5, where, on switching \ 9-10--12 J4 2.2 2.4 2.6 2.8 3.0 3 2 103 U/T Fig.5 Arrhenius plot for the rate constants of butene formation on dehydration of sec-butyl alcohol on AAS: (0)WJN, (0)W,/N, usually for t = 1-3 min. Each pair of points at the same temperature refer to the same experiment. 2150 20 c Iv) 4J --m10 CJ, 15 h al al w 1s 10-1 al E I$5 'I0 r I Is-u-0 0 5 10 15 time/mi n Fig. 6 Dehydration of sec-butyl alcohol on H-ZSM-5 (sample 2) at 126°C and subsequent desorption of butene when alcohol flow is stopped. The horizontal line on the extreme right denotes the reac- tion rate after returning to the sec-butyl alcohol-helium mixture. back to a flow of reaction mixture, we observed a new reac- tion rate which was reduced by a factor of two as a result of deactivation of the zeolite active sites during the period of helium purge.4 At lower reaction temperatures, butene oligomerization on the active sites is not so prominent. This is shown by the results of a similar experiment to that described above, but at a lower temperature (96"C), Fig.7. In this case, the rate of butene formation increases gradually by one order of magni- tude after replacing the reaction mixture with pure helium, before eventually starting to decrease. On returning to the alcohol-helium stream, the original rate of reaction is fully restored. Moreover, estimation of the area of the butene desorption peak gives a value of one butene molecule desorbed per active site.Thus the picture observed at this temperature is not complicated by subsequent oligomeriza- tion of butene generated inside the zeolite pores. Since zeolite samples of different crystallite sizes behave in a similar manner, we conclude that there is no connection between the observed phenomena and diffusion limitations for butene molecules inside the zeolite pores under steady- state conditions. 201 I II reaction desorption 141 I p 10 0 58 6 4 2 0 0 10 20 30 " time/min Fig. 7 Dehydration of sec-butyl alcohol on different samples of H-ZSM-5 at 96°C and subsequent desorption of butene when alcohol flow is stopped. The horizontal line on the extreme right denotes the reaction rate after returning to the sec-butyl alcohol- helium mixture.(0) sample 4.Sample 1, (A)sample 2, (0) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 16 I 14 reaction - 12 & 10 d 98 I I 26 I I 4 I I 2 I I 0 0 10 20 30 40 50 a ti me/min Fig. 8 Dehydration of sec-butyl alcohol on AAS at 90°C and sub- sequent desorption of butene when alcohol flow is stopped. The hori- zontal line on the extreme right denotes the reaction rate after returning to the sec-butyl alcohol-helium mixture. For dehydration of sec-butyl alcohol on AAS, under non- steady-state conditions we also observed some increase in the rate of butene evolution (Fig. 8). The estimated value of the area of the butene desorption peak is close to the number of active sites in the AAS sample, 1.3 butene molecules desorbed per active acid site.The decay of the kinetics curve of butene evolution is exponential and so it is possible to calculate the rate constant for butene formation from these data. The start- ing point for the treatment of the function is denoted in Fig. 8 by *. This plot gives a rate constant of 1.7 x s-l, which is very close to that determined from W,,, at the beginning of this experiment (1.8 x s-' from Fig. 5) and is almost twice as much as the steady-state value (determined from W,) of 0.9 x s-' (Fig. 5). Therefore, for both H-ZSM-5 and AAS, replacing the reac- tion mixture with pure helium gives rise to the so-called 'stop effect': i.e.an increase in the rate of product (in our case butene) formation as compared with that at steady state. However, the effect is much more significant in the case of the zeolite samples. 3.1.3 Isomeric Distribution of Butenes For reaction on H-ZSM-5, the isomeric distribution of the reaction product, butene, is found to be independent of the zeolite crystallite size (Table 2) or contact time, and almost constant in the studied temperature range. The composition of butenes obtained on AAS also does not vary in the tem- perature range employed, but differs from that observed for the zeolite. For AAS the distribution is shifted more towards but-2-ene, while more but-l-ene is produced by the zeolite samples. In no case was 2-methylpropene (isobutene) observed; this is in good agreement with literature data on the absence of skeletal isomerization at such temperat~re.~ In addition, the results in Table 2 show that the mixture of Table 2 Butene isomeric distribution at 100 "C but-l-ene (E)-but-Zene (Z)-but-2-ene ZSM-5 sample (%) (%) ("/.I 1 27.8 45.5 26.7 2 26.4 43.9 29.7 4 27.3 44.9 27.8 AAS 11.0 45.5 43.5 equilibrium 7.9 63.4 28.7 mixt urea a From ref.7. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0) CC .-.-.e E 8 .-En c 13000 2000 wavenumber/cm -' Fig. 9 FTIR difference spectra for H-ZSM-5 sample 2, showing the course of reaction as a function of time after exposure to sec-butyl alcohol at 70 "C: (a)0.5,(b)5, (c) 17, (6)21, (e) 43 and (f)130 min butenes obtained on both H-ZSM-5 and AAS is far removed from the equilibrium one.3.2 FTIR Kinetic Studies Kinetic studies of reaction using in situ FTIR methods were carried out at 70 "C.Some typical spectra showing changes in the adsorbed species on increasing exposure time to alcohol, are given in Fig. 9. These are difference spectra, subtracting the spectrum of the purely dehydrated zeolite. Let us consider in more detail the first spectrum recorded 0.5 min after adsorption of alcohol. Stretching vibrations of the hydrocarbon skeleton (C-H vibrations) of alcohol mol- ecules lie in the range 2800-3000 cm-': the peaks at 2970 time/mi n Fig. 10 Kinetics of water formation in sec-butyl alcohol dehydra- tion on H-ZSM-5(sample 2) obtained from FTIR studies at 70°C (0)and on treating the data in first-order coordinates (a).Data are taken from Fig.9 (peak at 1640 cm-'). and 2885 cm-' arise from vibrations of CH, groups and the peak at ca. 2940 cm-' from those of CH, groups.8 Two peaks at 1460 and 1380 cm-' correspond to C-H deforma-tion vibrations: that at 1380 cm-' to CH, vibrations and that at 1460 cm-' to vibrations from both CH, and CH, groups.8 Two broad components at ca. 1500 and CQ. 2450 cm-' are associated with the mode of stretching vibrations of zeolite OH groups hydrogen-bonded to alcohol molecules, split because of Fermi re~onance.~,~ A weak peak at 1640 cm-' corresponds to deformation vibrations of water, which reflects the beginning of dehydration of the adsorbed alcohol molecules even at this early stage of reaction.As time increases, one can see a number of characteristic changes in the spectra, indicating the course of dehydration of the adsorbed alcohol molecules. These are: (1) an increase in the intensity of the peak at 1640 cm-', corresponding to an increase in the number of adsorbed water molecules; (2) the disappearance of the broad peak at ca. 1500 an-' and substitution of the broad peak of ca. 2450 cm-' by a rather narrower peak situated slightly to higher wavenumber, corre- sponding to decomposition of adsorbed alcohol molecules and formation of water molecules which are hydrogen-bonded to the zeolite OH groups; (3) an increase in the inten- sity of the peak at CQ.2940 cm-I and some decrease and broadening of the peak at 1380 cm-' corresponding to an increase in the number of CH, groups in the adsorbed reac- tion products at the expense of CH, groups (indicating product oligomerization as previously observed for tert-butyl alcohol6). For kinetic analysis of dehydration under these conditions, the water deformation peak at 1640 cm-' can be used (see ref. 5 and 6). At long reaction time (130 min) it is clear that alcohol decomposition is complete. First, the intensity of the peak at 1640 cm-' does not grow any further. Secondly, the intensity of this peak has reached a value which is four times greater than the intensity of the peak at 1460 cm-'. This is in agreement with the maximum ratio of 4 : 1 observed by us at the end of the reaction in previous studies involving dehydra- tion of other butanol isomer^.^^.^*^ Fig.10 shows the kinetics of water evolution derived from the change in the relative intensity of this peak. (1164(,)m is the intensity at the end of reaction (reaction time 130 min). On the same plot, the data are also given in first-order coordinates. From these data the rate constant for sec-butyl alcohol dehydration at 70 "C is determined to be 1.1 x lo-, s-'. 4. Discussion All the data concerning the rate constants for sec-butyl alcohol dehydration on H-ZSM-5 and AAS samples, esti- mated by the various methods described in Section 3, are summarised for convenience in the Arrhenius plot in Fig.11. 4.1 Reaction on AAS Let us consider the reason for the difference between initial and steady-state reaction rates on the AAS catalyst [Fig. 1 l(b), (c)]. Earlier, studying the dehydration of n-butyl alcohol and isobutyl alcohol, we considered in detail the mechanism of this reaction, shown in Fig. 12 [which has been analysed carefully in ref. 3(b) and 51. After water elimination, R I there are four ways for the active intermediate Al-0-Si to react : (i) desorption of butene (downwards); (ii) interaction with another butene molecule, i.e. oligomerization on active sites (upwards); (iii) interaction with another alcohol mol- ecule, i.e. ether formation (to the right); and (iv) interaction with a water molecule, i.e. back/reverse reaction (to the left).-11. -131 . . . . . -. . . I 2.2 2.4 2.6 2.8 3.0 3.2 103 KIT Fig. 11 Arrhenius plot for the rate constants of sec-butyl alcohol dehydration obtained by different methods on the different alumino- silicate samples. (a) Data are rate constants for H-ZSM-5 from Fig. 2, obtained by GC under steady-state conditions. (b), (c) Data taken from Fig. 5, where for dehydration as AAS, (b) corresponds to the initial rate of reaction (i.e. W,, ,immediately after adsorption) and (c) corresponds to steady-state rates (Wa).(d)Rate constant of dehydra- tion on AAS obtained from non-steady-state experiments (Fig. 8). (e) Rate constant for H-ZSM-5 estimated from FTIR data of Fig. 9 and 10. The last three pathways can, in principle, give rise to the observed slowing down of reaction with time.However, ether formation is unlikely to be of significance for sec-butyl alcohol dehydration. Di-sec-butyl ether is highly reactive and is easily decomposed on acid sites." Furthermore, no traces of ether are observed among the reaction products. As we noted in Section 3.1.1, the decrease in the reaction rate with time observed in the GC kinetic experiments is unlikely to be connected with catalyst deactivation resulting from butene oligomerization on the active sites. In fact, the decrease in rate most likely reflects the slowing down of the reaction by water. The following facts support this suggestion: (1) The decay of the kinetic curve of butene formation after satura- tion of the sample with alcohol is accompanied by retention of water (Fig.4). (2) After the falling phase in the kinetics of butene evolution, a clear steady-state rate is observed. If dehydration on AAS were complicated by oligomerization, such behaviour would not be expected. (3) The difference between Wsa,and W, increases with decreasing reaction tem- -__------.------.,-flu */-~BUR *-..I ..//// ,, ,, t .... ,.I i +Bu I (VI) RH R\?/R'0' I. ' +ROH OH --40 OR --oli I //// Fig. 12 Reaction scheme for butanol dehydration on alumino-silicates (see ref. 5 for more detail). R = C,H, ,Bu = C,H, . J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 perature, which indicates a greater degree of side reaction at lower temperatures.Oligomerization, on the other hand, would be expected to become more prominent as the tem- perature is increa~ed.~ (4) The observed 'stop effect' (Fig. 8) is not accompanied by deactivation of the active sites of AAS. The next question concerns the nature of this suppression of activity by water. Two explanations are possible. The first possibility is that product water molecules compete with alcohol molecules for the active sites and so bring about a decrease in the number of active sites that are interacting with alcohol molecules, and hence, a decrease in the reaction rate. However, the absence of such a reduction in reaction rate in the case of isobutyl alcohol and n-butyl alcohol dehy- dration at rather similar temperat~res~~~' makes this explana- tion unlikely, since one would expect that the adsorption properties of all three butanol isomers on the acid OH groups of AAS should be similar.The second possibility is connected with reverse reaction. Proceeding from the reac- tion scheme given in Fig. 12, it is seen that the active interme- diate *OR, generated on dehydration of a sec-butyl alcohol molecule, can in turn interact back again with product water molecules present in the sample. This suggestion is in agree- ment with all facts (1)-(4) above. It is clear that the more water molecules present on the sample (in the adsorbed layer), the higher is the rate of back reaction and, hence, the lower is the rate of butene evolution [point (l)].When all rates of the various processes: adsorption, desorption, forward and back transformations, become equal, the reac- tion comes to steady state [point (2)]. On decreasing the reaction temperature, the hydration4ehydration equilibrium shifts to the left towards alcohol, thus further slowing the rate of the forward dehydration reaction [point (3)]. Finally, during purging, the catalyst pores release adsorbed reactants (water in particular) and this reduces the possibility of back reaction and hence the rate of butene formation increases [point (4)]. Based on the above considerations, one can conclude that the points on Fig. 1 l(b) and the point at Fig. ll(d), lying on this line, are the true rate constants for sec-butyl alcohol dehydration on AAS; points lying on Fig.ll(c) are the effec- tive rate constants for dehydration in the presence of back reaction. Thus, E,,, = 18 kcal mol-' is the true activation energy of dehydration of sec-butyl alcohol on AAS (i.e. of the forward step of Section I1 in Fig. 12) while E, = 26 kcal mol-'is an effective value. 4.2 Reaction on Zeolite H-ZSM-5 Comparing the results obtained using two independent methods [the rate constants on Fig. 11 (a) obtained by GC studies and the rate constant plotted as Fig. ll(e) from FTIR studies], their striking difference becomes evident. At 70 "C, the rate constant from spectroscopic data is 400 times greater than the value determined from GC steady-state data. Again, as for the case of AAS, we believe that the low steady-state rate of reaction (GC studies) is connected with the back reaction. Based on the scheme in Fig.12, other explanations are hardly likely. Therefore, the decrease in activity cannot be explained by deactivation of the active sites as a result of butene oligomerization on them, because in that case catalytic activity would then depend on the zeolite crystallite size and the 'stop effect' would not be observed because deactivated sites could not produce more butene during purging. The decrease in the rate of butene formation also cannot be a result of a shift of reaction equilibrium to di-sec-butyl ether formation under steady-state conditions because, just as in the case of AAS, even traces of this ether are not detected among the reaction products (although this J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 can be a consequence of diffusion problems for such bulky molecules). Furthermore, the observed kinetics in the initial phase of reaction (period of alcohol adsorption) are com- pletely different to those observed previously for isobutyl alcohol and n-butyl alcoh01~~~~ where ethers are formed. As was mentioned in Section 3, in that case, water evolves ahead of butene, which reflects ether formation inside the zeolite pores. In the case of sec-butyl alcohol, this was not observed: the rates of formation of both products were equal. Analysis of the isomeric distribution of product butenes shows that sec-butyl alcohol is expected to be the only product of the back reaction, since it is known that in liquid- phase catalysis, both but-2-ene and but-l-ene give only sec- butyl alcohol on hydration.' In the FTIR experiments, back reaction does not occur and dehydration proceeds irreversibly.The reaction condi- tions in this case are different from those in the flow reactor and the reaction products are oligomers of butene adsorbed on the active sites (Section 3) which probably cannot be hydrated under these conditions. This means that from FTIR kinetic studies, we determined the true rate constant for sec- butyl alcohol dehydration, while the rate constants and the activation energy from GC kinetic studies are effective values. The idea of back reaction in the sec-butyl alcohol/H-ZSM- 5 system makes it possible to explain the whole complex of observed phenomena (Section 3): (1) the low rate constants under steady-state flow conditions; (2) the high value of E,; (3)the 'stop effect' and its magnitude.Comparison of the results for H-ZSM-5 and AAS shows that the true rate constants for both catalysts [Fig. ll(e) and the point from Fig. ll(b) at 70"C] are close, the difference being a factor of two or so, i.e. the same ratio that was observed by us for isobutyl and n-butyl alcohol dehydration earlier.3'.4 However, in the case of the zeolite system, the sup- pression of dehydration is stronger than for the AAS sample. This most likely arises from the specific adsorption properties of the zeolite pores.The higher the water concentration in the vicinity of the active intermediate, i.e. in the catalyst pores, the higher is the expected rate of back reaction. During the flow reaction at 126 "C, the zeolite micropores are completely filled with reactant molecules, while the macropores of the AAS, on the other hand, are virtually empty (see results for n-butyl alcohol dehydration3'). This, in its turn, is a result of the adsorption properties of the catalyst and hence of the type of pore structure. In the case of a flow reaction on H-ZSM-5 we do not see the initial drop in catalytic activity in contrast to AAS, but rather observe a low steady-state rate from the very begin- ning. This can be explained as follows. The time necessary for a system to reach a steady state is inversely proportional to the sum of the rate constants of the forward and back reac- tions.Therefore, having a higher rate of back reaction (as in the case of H-ZSM-5), the system reaches a steady state more quickly than in the case of AAS, and the initial non-steady- state region of the kinetics can occur during the period of sample saturation with alcohol. In conclusion, note that for the two aluminosilicate samples investigated, namely the AAS and the zeolite, the former gives rise to 'normal' heterogeneous solid-gas cataly- sis. In this case, the dehydration of sec-butyl alcohol under flow conditions proceeds almost irreversibly (or more exactly, with little reversibility). Just after the stage of water elimi- nation the active intermediate formed decomposes along pathway V (Fig.12) with high probability, releasing butene into the gas phase. Reaction on the zeolite catalyst in the temperature range studied is more like the reaction taking place in the liquid phase with a large contribution from the back reaction. In this case, the rate of reaction is lower than 21 53 expected because it is not possible to disperse or separate the reaction products and so they undergo back transformation. When the temperature is increased, the extent of back reac- tion on the zeolite is reduced, presumably owing to a release of adsorbed reactants from the pores. Indeed, extrapolation of Fig. ll(a) and (b)to higher temperatures suggests that at ca. 160°C the rate constant for H-ZSM-5 under flow condi- tions reaches a value twice that for the rate constant of AAS, which is the normal ratio typical for these catalysts (according to the results obtained for the other butyl alcohols and the ratio of the rate constants obtained by different methods for sec-butyl alcohol at 70 "C).5. Conclusions Comparison of sec-butyl alcohol dehydration in zeolite and amorphous aluminosilicate samples has shown the particular influence of pore confinement on reaction. Confinement of reactant molecules to the H-ZSM-5 micropores (in the tem- perature range 60-140 "C) results in an enhancement of the back reaction, namely hydration of the product butene with product water. At these low temperatures the catalyst pores are filled with reactants, in particular water, and this increases the probability of back reaction : all transformations proceed in a pseudo-liquid phase.Kinetically, the influence of pore confinement on this system manifests itself as: (i) anomalously low values of steady-state reaction rates under flow conditions ; (ii) an anomalously high value of the activation energy obtained from steady-state rate constants under flow conditions; (iii) an increase in the reaction rate after interrupting the alcohol- helium feed gas (the 'stop effect'). In the case of the amorphous aluminosilicate sample with larger pores (ca. 50 A diameter) and hence a more open surface, all of the kinetic effects mentioned above are also observed, but to a much lesser degree.The authors thank Dr. E. A. Paukshtis for assistance with the FTIR experiments, Dr. S. V. Dudarev for sample 1 and V. N. Romannikov for providing samples 3,4 and 5. We thank The Royal Society, USSR Academy of Sciences and the SERC for financial support. References 1 (a)N. Y. Chen and W. E. Garwood, Catal. Rev.-Sci. Eng., 1986, 28, 185; (b)M. E. Davis, Ind. Eng. Chem. Rex, 1991,30, 1675. 2 J. Weitkamp, in Catalysis and Adsorption by Zeolites, ed. G. Olhmann, J. C. Vedrine and P. A. Jacobs, Elsevier, Amsterdam, 1991, p. 21. 3 (a) M. A. Makarova, C. Williams, J. M. Thomas and K. I. Zamaraev, Catal. Lett., 1990, 4, 261; (b) M. A. Makarova, E. A. Paukshtis, C. Williams, J. M. Thomas and K. I. Zamaraev, J. Catal., in the press. 4 M. A. Makarova, C. Williams, V. N. Romannikov, K. I. Zamaraev and J. M. Thomas, J. Chem. SOC., Faraday Trans., 1990,86,581. 5 C. Williams, M. A. Makarova, L. V. Malysheva, E. A. Paukshtis, J. M. Thomas and K. I. Zamaraev, J. Chem. SOC., Faraday Trans., 1990,86,3473. 6 C. Williams, M. A. Makarova, L. V. Malysheva, E. A. Paukshtis, E. P. Talsi, J. M. Thomas and K. I. Zamaraev, J. Catal., 1991, 127, 377. 7 Yu. M. Zhorov, lsomerization of OIefins, Khimia, Moscow, 1977. 8 L. M. Sverdlov, M. A. Kovner and E. P. Krainer, Vibration Spectra of Polyatomic Molecules, Nauka, Moscow, 1970. 9 N. D. Sokolov, The Hydrogen Bond, Nauka, Moscow, 1981. 10 The Chemistry of the Ether Linkage, ed. S. Patai, Wiley- Interscience, New York, 1967. 11 I. Dostrovsky and F. S. Klein, J. Chem. SOC.,1955,440. 12 H. Knozinger, in The Chemistry of the Hydroxyl Group, ed. S. Patai, Wiley-Interscience, London, 1971, pp. 642-718. Paper 4/00095A; Received 7th January, 1994