J . Chern. SOC., Faraday Trans. I , 1989, 85(1), 71-78 Chromia/Silica-Titania Cogel Catalysts for Ethene Polymerisation Polymerisation Kinetics Steven J. Conway, John W. Falconer and Colin H. Rochester* Department of Chemistry, The University, Dundee DDl 4HN, Scotland The kinetics of ethene polymerisation over a series of chromia/silica-titania cogels have been studied. The catalysts were activated by reduction in CO at 623 K. Rates of polymerisation were first order in ethene pressure below 373 K. Above 373 K the rates tended towards second-order kinetics. Arrhenius plots were non-linear and exhibited a maximum, the nature of which was dependent upon the reduction time and the catalyst composition. For polymerisation at 273 K the activity of the catalysts passed through a maximum with increasing reduction time.Catalysts which were inactive at 273 K due to extensive reduction were active at 343 K, and in some cases gave bimodal activity curves. Active catalysts were deactivated by evacuation or reduction in CO at 873 K. Rates of polymerisation at 343 K for catalysts deactivated in uacuo at 873 K showed an initial decrease in activity which was followed by a small increase in activity before further deactivation. The results are consistent with a Langmuir-Hinshelwood polymerisation mechanism. Three types of active site have been identified. Heterogeneous catalysts are often prepared by depositing the active component on a solid support. The support can either act as a dispersive medium where both support and active component exist as two separate phases, or the support forms surface complexes with the active component. The Phillips chromium(vr)/silica ethene-polymerisation catalyst is an extreme case of the latter, as unsupported chromia (CrO,) cannot polymerise ethene.' Bulk CrO, decomposes above 473 K, but it is possible to stabilise CrO, up to 1173 K on a silica support by calcination in dry air.This stabilisation is due to the formation of surface chromate and/or dichromate species in which the chromium is linked to the support via Si-0-Cr bonds.2 The incorporation of titania either in or on the silica support results in a chromia catalyst capable of polymerising ethene to yield a polymer with lower average molecular weight than that for polymer generated over unmodified This is thought to be due to a change in the electronic environment of the chromium caused by the formation of Ti-0-Cr as well as Si-0-Cr bonds.The dependence of the polymerisation process on the nature of the chromium and the manner in which the titanium influences this process have here been examined by comparison of the polymerisation kinetics for a series of chromia catalysts supported on silica-titania cogels with titanium contents in the range W.2 wt%. Experimental Catalyst Preparation Silica-titania cogel supports were prepared by a coprecipitation m e t h ~ d . ~ The initial stage involved the coprecipitation of an aqueous sodium silicate solution and a titanium tetrachloride-sulphuric acid mixture. After the gel had been aged for 4 h at 373 K it was 7172 Modijied Phillips Catalyst Kinetics Table 1.Summary of porosity measurements for cogels (after calcination at 873 K) used in the preparation of catalysts pore volume/cm3 g-' titanium surface average pode (wt %) area/m2 g-Ia < 500 A" totalb diameter/A" 0 272 f 5 1.02 2.95 190 1.3 508 f 10 2.34 2.72 212 2.1 273 f 5 1.04 2.34 185 4.2 279 6 1.06 2.18 187 a Determined by N, adsorption. Data refer to a pore diameter of up to 500 A. liquid Determined by washed with a 5% ammonium nitrate solution and several times with distilled water to remove sodium and sulphate ions. Water was extracted from the gel by azeotropic distillation in ethyl acetate and the remaining solvent was removed by drying in air at 373 K for ca. 3 h. A summary of support characteristics are given in table I .The gel was subsequently impregnated with 1 wt% chromium by forming a slurry with a solution of chromium acetylacetonate in ethyl acetate. The resulting slurry was dried on a rotary evaporator. Kinetic Studies Kinetic studies were performed under static conditions using a conventional high- vacuum system linked to a reactor vessel consisting of a cylindrical quartz bulb with diameter 4 cm and volume 148 cm3. Reactant-gas pressures were monitored using a pressure transducer. Samples of catalyst (120 mg) which had been precalcined at 873 K for 16 h in a flow of dry air (50 cm3 min-') were placed in the reactor. The temperature of the catalyst in vacuum was raised to 873 K over a 30 min period and the catalyst was then contacted with oxygen (101 kN m-') for a further 30 min at 873 K.Subsequent evacuation at 873 K for 60 min was followed by a reduction in temperature to 623 K over a 30 min period. The catalyst was then activated by reduction at 623 K in CO (5.33 kN m-2) for various times, with subsequent evacuation for 30min prior to being cooled to the polymerisation temperature. Ethene (ca. 2.2 kN m-') was then admitted to the reactor vessel and the fall in ethene pressure was monitored as a function of time. Reaction temperatures in the range 250-4023 K were maintained with liquid thermostat baths. Two methods of deactivation of reduced catalysts were used; the first involved heating in vacuo at 873 K and the second involved heating in CO at 873 K for various times. Ethene (B.O.C. research grade, > 99.92%) with quoted impurity levels of 1 and 2 ppm of 0, and water, respectively, was subjected to a series of freeze-pumpthaw cycles before use.Carbon monoxide (B.O.C., > 97.5%) was passed through a cold trap (77 K) to remove condensable impurities. Results Calcined supports alone and supported chromia were found to be inactive for low- pressure ( < 3.0 kN m-,) gas-phase ethene polymerisation. However, supported chromia catalysts which had been reduced in CO at 623 K were active. Below 373 K polymerisation rates exhibited a first-order dependence on ethene pressure. Fig. I shows a typical first-order plot. First-order rate constants (k,) wereS. J . Conway, J . W. Falconer and C. H . Rochester 3.2- -- 3.0- ‘E 2 v % 2 2.8- 2 2.6- 0 - 2.4- 73 I I 1 1 - 0 60 120 180 240 300 tls Fig.1. Typical first-order plot for the derivation of rate constants. Example for ethene polymerisation at 343 K over a Cr/silica-titania (2.1 % Ti) cogel reduced in CO (623 K) for 15 min. 3.3 n - 3.2 ‘E 3 v !i 3.1 v) 2 3.0 2.9 4 1.4 - 1.2 b 1.0 2 “E m 2 2 0.8 \ d v 0.6 0 5 I 0 15 20 2 5 30 tlmin Fig. 2. First- and second-order plots for ethene polymerisation at 423 K over a Cr/silica-titania (1.3% Ti) cogel reduced in CO (623 K) for 15 min. deduced per gram of catalyst per second. Above 373 K the pressure dependence of the reaction rate began to deviate from first-order towards second-order behaviour (fig. 2). Polymerisation activities are here expressed as first-order rate constants (kJ. A non- linear dependence of log (polymerisation activity) on reciprocal temperature and the appearance of a maximum polymerisation activity were observed in Arrhenius plots (fig.3). The characteristics of the curves varied with reduction time and support composition. Maximum polymerisation activities were observed at 313 and 273 K for Cr/silica gel and Cr/silica-titania (4.2 O/O Ti) cogel, respectively, reduced in CO for 15 min. Maximum activity was observed at 343 K for both Cr/silica gel reduced in CO for 120 min and Cr/74 Mod$ed Phillips Catalyst Kinetics -1 .o- -Ij - 1 . 8 2 2 I (b) L I I I I 2.5 3.0 3.5 4.0 103 KIT -1 .o- - ly) - 1 . 4 - c I \ M -Y - gJ - 1 . 8 - - - 2 . 2 - Fig. 3. (a) Arrhenius plots for Cr/silica gel reduced in CO (623 K) for I5 min (@) and 120 min (H). (b) Arrhenius plots for Cr/silica-titania (4.2 YO Ti) cogel reduced in CO (623 K) for 15 min (@) and 60 min (m).0 60 120 180 reduction time/min 6 5 " 4 'm - I 53 -Y 2 1 reduction time/min 1 I I I 0 60 120 1 8 0 2 1 0 reduction time/min I 0 s b - reduction time/min Fig. 4. Polymerisation rate at 273 (H), 343 (a) and 373 K (0) over supported chromia catalysts as a function of reduction time in CO at 623 K: (a) 0, (b) 1.3, (c) 2.1, (d) 4.2 wt YO Ti. 0, Reduced 5 h, polymerisation at 273 K. 0, Reduced 12 h, polymerisation at 343 K.75 0 60 120 180 240 0 60 120 180 240 treatment time/min treatment time/min Fig. 5. Polymerisation rate at 273 K (m, 0) and 343 K (0, 0) for two catalysts reduced in CO (15 min at 623 K) followed by heating in U ~ C U O (0, W) or in CO (0, 0) at 873 K. (a) 0, (b) 4.2 wt % Ti.silica-titania (4.2 YO Ti) cogel reduced in CO for 60 min. A marked loss of activity was observed at low polymerisation temperatures (< 283 K). Polymerisation activities for four catalysts at various polymerisation temperatures are shown in fig. 4 as a function of reduction time. At 273 K a maximum activity was attained after a short reduction time (usually 15 min) at 623 K. Further reduction resulted in a loss of polymerisation activity at 273 K, with the loss being promoted by increased titanium content. Upon increasing the polymerisation temperature to 343 K a second maximum at ca. 60-120 min reduction time was observed, after which the activity began to decrease slowly. The appearance of bimodal activity curves with an increase in temperature are shown most clearly for 0 and 2.1 YO titanium.An overall reduction in activity was observed on increasing the polymerisation temperature to 373 K for 0 and 4.2% titanium. High-temperature evacuation or reduction in CO at 873 K of a standard reduced catalyst (standard reduced being defined as reduction in CO for 15 min at 623 K) was found to reduce the polymerisation activity at 273 and 343 K (fig. 5), the activity at 343 K passing through a maximum after an initial loss of activity when treated in uacuo. High-temperature treatment in CO caused a greater decrease in polymerisation activity than heat treatment in uacuo. Also a greater loss of activity at 273 K was observed for a catalyst with a higher titanium content. Discussion The first-order behaviour of the polymerisation rate on ethene pressure (at temperatures < 373 K) has been explained by a Rideal-Eley mechanism whereby polymerisation occurs by reaction of gas-phase ethene with the adsorbed polymer chain (or monomer) or by the Langmuir-Hinshelwood mechanism in which polymerisation proceeds by the reaction of an adsorbed ethene molecule with an adjacently adsorbed polymer chain or monomer. A comparison of theoretical predictions for both mechanisms with experimental observations indicates the Langrnuir-Hinshelwood mechanism to be the more probable mechanism6 for polymer growth.For such a mechanism the propagation reaction is normally envisaged as occurring in two steps,’ uiz. (1) reversible ethene76 ModiJied Phillips Catalyst : Kinetics adsorption at a vacant coordination site and (2) subsequent ethene insertion into the active Cr-C bond: transition state Above 373 K the reaction order showed a tendency towards second-order kinetics (fig.2). At increasing temperatures desorption of coordinated ethene molecules from the active site and a destabilisation of the transition-state complex is expected. In the case of destabilisation of the transition-state complex, insertion of the weakly complexed ethene into the growing polymer chain may be facilitated by the coordination of a second ethene molecule : Such behaviour would result in a second-order dependence of polymerisation rate on ethene pressure.8 The existence of a maximum in the dependence of polymerisation rate on temperature (fig. 3) has been reported before.Groeneveld et al.’ believe this to be a consequence of depolymerisation, although Clark’’ points out this is unlikely at such low temperatures (353 K). Zakharov et a1.l’ proposed that the maximum is a result of the presence of feedstock poisons which are adsorbed on the support at low temperatures, become mobile at higher temperatures and irreversibly deactivate the catalyst. It has been demonstrated that this poisoning mechanism is improbable.’ The dependence of catalytic activity on polymerisation temperature is complex in nature. Some of the features in the Arrhenius plots can be rationalised in terms of a Langmuir-Hinshelwood mechanism. At low temperatures a high surface coverage will be obtained and the rate-determining step will be the insertion of the coordinated ethene.With an increase in temperature the rate constant of the insertion reaction increases. However, when the temperature is high enough, desorption of coordinated ethene predominates and ethene coordination becomes rate-determining. This results in falling polymerisation rates, and thus a maximum is observed. In the temperature regions (up to 373 K) where these features are noted, the reaction is strictly first-order and no deviation towards second-order kinetics occurs. Thus the maximum polymerisation rates and positive activation energies are apparently due to changes in the kinetic parameters and not to a change in the polymerisation mechanism. At low temperatures a marked fall-off in polymerisation activity is observed (fig. 3). This is thought to indicate a possible change in the initiation step of the polymerisation mechanism.12 Low- temperature stabilisation of the Cr-C bonds in the possible initial coordination complexes2 may reduce the ability for rearrangement into a propagation centre (P), this rearrangement being a prerequisite to propagation.S.J . Conway, J . W. Falconer and C . H. Rochester 77 For polymerisation at 273 K the activity of the catalysts passed through a maximum with increasing reduction time (fig. 4). Upon increasing the polymerisation temperature at 343 K, catalysts which were previously inactive began to display increased activity. At 343 and 373 K the activity curves were bimodal in nature. This is consistent with the presence of two active sites for which there are two possible alternative explanations.Case 1 Two oxidation states of active Cr are present. The passing of polymerisation activity at 298 K through a maximum with increasing reduction time in CO at 623 K has been reported. Beck and Lun~ford'~ correlated the polymerisation activity at 298 K with CrI'I concentration. Later Myers and L~nsford'~ found the catalytic activity at 298 K to be inversely proportional to Cr" concentration as indicated by the chemiluminescent intensity associated with the reaction of oxygen with supported Cr". Upon re-evaluation of their catalyst, Myers and Lunsford12 found that the Cr" catalyst which was inactive at 298 K began to exhibit polymerisation activity as the temperature was increased (> 303 K). They concluded that extensive reduction of the Phillips catalyst results in reduction of Cr"' to Cr", and hence causes a change in the nature of the active polymerisation site.Wittgen et al.15 pointed out that CO reduction of chromate surface species can result in the formation of chromium with even valencies, IV and 11. In contrast, dichromate reduction can be expected to result in the formation of valencies from v to 11. For a mechanism involving the reduction of a dichromate species the concentration of CrI'I is expected to go through a maximum when Crrv species are totally reduced, while the concentration of Cr" should steadily increase. Although CO is normally assumed to be a two-electron reducing agent, without a more detailed knowledge of the surface electrochemistry and surface species, it is not possible to assign either chromate or dichromate species as the precursor of the active site.Case 2 Two types of active Cr" are present. Several types of Cr" on CO-reduced catalysts have been reported by different groups, although not all are believed to be active. Three types of Cr", each varying in their coordination number, have been identified using spectrosopic and calorimetric techniques. 16. l7 Other analytical techniques have identified at least two main types of Cr11.'E-20 McDaniel and Welch2' proposed two types of Cr", one producing low-molecular-weight polymer, the other producing polymer of high molecular weight. Merryfield et aZ.22 reported that Cr" becomes deactivated by reduction at high temperatures (> 673 K). This deactivation is not due to a change in oxidation state since the average state remains constant at 2.0.CO chemisorption experiments showed that the amount of chemisorbed CO decreases with increasing reduction temperature (> 623 K). This loss of coordinative unsaturation is thought to be due to a structural rearrangement of the chromium. Such a phenomenon may occur on prolonged reduction at 623 K (fig. 4). It is feasible that a mechanism based on two types of active Cr", each varying in their rate of formation and deactivation, may operate. The incorporation of titania appears to promote either the deactivation or reduction78 Modijied Phillips Catalyst Kinetics of the low-temperature active site (fig. 4). At high polymerisation temperatures high titania (4.2 % Ti) concentrations reduce the activity of the low-temperature active site.Plausible causes of these effects include a change in surface geometry (table 1) or the generation of Lewis-acid sites by titania, which may affect the physical and/or chemical properties of the active chromium. The possibility of Ti-0-Cr bonds being responsible for these effects cannot be discounted. Heating active catalysts at high temperature (873 K) in uacuo or in CO deactivates the catalyst towards ethene polymerisation (fig. 5). Such conditions have previously been observed to result in catalyst deactivation, which was attributed to loss of coordinative unsaturation by aggregation. Deactivation in CO is particularly severe as compared to deactivation in uacuo. This may be due to the formation of thermally stable Cr-carbonyl complexes,23 which allow the Cr greater surface mobility, enabling the Cr to rearrange itself on the surface in an inactive form.Titania also appears to influence the stability of the low-temperature active site. Catalysts treated in vacuo pass through a maximum at 343 K. This is likely to be due to the formation of a new surface complex, possibly involving a change in surface coordination. l2 The existence of two types of active site, each possessing different kinetic parameters, together with the complexity of mechanism will contribute to the non-linearity of the Arrhenius plots. This would also influence the position of the maximum, depending on the relative concentrations of each active site. We thank the S.E.R.C. for a CASE Studentship, BP Chemicals for financial support, and Dr G.W. Downs for helpful discussions. References 1 D. R. Witt, in Reactivity, Mechanism and Structure in Polymer Chemistry, ed. A. D. Jenkins and 2 M. P. McDaniel, Adv. Catal., 1985, 33, 47. 3 T. J. Pullukat, R. E. Hoff and M. Shida, J. Polym. Sci., Polym. Chem. Ed., 1980, 18, 2857. 4 M. P. McDaniel, M. B. Welch and M. J. Dreiling, J . Catal., 1983, 82, 118. 5 R. E. Dietz, U.S. Patent 3887494, 6/1975. 6 A. Clark, Ind. Ing. Chem., 1967, 59, 29. 7 J. P. Hogan, J. Polym. Sci., 1970, 8, 2637. 8 W. Cooper, in Comprehensive Chemical Kinetics, ed. C. H. Bamford and C. F. H. Tipper (Elsevier, 9 C. Groeneveld, P. P. M. M. Wittgen, J. P. M. Lavrijsen and G. C. A. Schuit, J . Catal., 1983, 82, 77. 10 A. Clark, in Polymerisation and Polycondensation Processes Symp., Ind. Eng. Div., 155th National 11 V. A. Zakhorov, Y. I. Ermakov, L. P. Ivanov and V. B. Skomorokhov, Kinet. Catal., 1968, 9, 499. 12 D. L. Myers and J. H. Lunsford, J . Catal., 1986, 99, 140. 13 D. D. Beck and J. H. Lunsford, J. Catal., 1981, 68, 121. 14 D. L. Myers and J. H. Lunsford, J . Catal., 1985, 92, 260. 15 P. P. M. M. Wittgen, 16 B. Fubini, G. Ghiotti, L. Stradella, E. Garrone and C. Morterra, J . Catal., 1980, 66, 200. 17 G. Ghotti, E. Garrone, G . Della Gatta, B. Fubini and E. Giamello, J . Catal., 1983, 80, 249. 18 H. L. Krauss, B. Rebenstorf, U. Westphal and D. Schneeweiss, in Preparation of Catalysts, ed. B. Delman, P. A. Jacobs and G. Poncelet (Elsevier, Amsterdam, 1976), p. 489. 19 P. Morys, U. Gorges and H. L. Krauss, Z . Naturforsch., Teil B, 1984, 39, 458. 20 H. L. Krauss and R. Hopfl, Proc. Eur. Symp. Thermal Anal., 198 1, 2, 175. 21 M. P. McDaniel and M. B. Welch, J . Catal., 1983, 82, 98. 22 R. Merryfield, M. P. McDaniel and G. Parks, J . Catal., 1982, 77, 348. 23 J. N. Finch, J . Catal., 1976, 43, 111. 24 M. P. McDaniel and T. D. Hottovy, J . Colloid Interface Sci., 1980, 78, 31. A. Ledwich (Wiley, Chichester, 1974), p. 43 1. Amsterdam, 1976), p. 155. Meeting A.C.S. (1968). C. Groeneveld, P. J. C. J. M. Zwaans, H. J. B. Morgenstern, A. H. von Heughten, C. J. M. von Heumen and G. C. A. Schmit, J. Catal., 1982, 77, 360. Paper 8/00722E; Received 23rd February, 1988