Hydrodechlorination of Tetrachloromethane Over Supported Platinum Catalysts. Effects of Hydrogen Partial Pressure and Catalyst’s Screening Protocol on the Catalytic Performance

Two series of silica- and alumina-supported platinum catalysts were investigated in the hydrodechlorination (HdCl) of tetrachloromethane. During an initial period of reaction carried out at a lower H2/CCl4 ratio the catalysts, especially those characterized by high metal dispersion, deactivated with time-on-stream. Two catalyst screening protocols were used. The first one concerned a gradually increased hydrogen partial pressure, whereas during the second one the H2 pressure was decreased. Although, in general, the hydrogen-rich reaction conditions resulted in improved catalyst performance (higher overall activity and selectivity to CHCl3), the second protocol led to even better results. Reasons for such a behaviour are suggested. Because of very high activity of a few tested samples, changes in CCl4 conversion with the hydrogen partial pressure do not reflect real reaction orders in hydrogen. The same reason may lead to falsification of apparent activation energies. In certain cases the relation between conversion and hydrogen pressure showed a maximum, suggesting that HdCl undergoes via a Langmuir–Hinshelwood mechanism, when hydrogen and CCl4 compete for metal surface sites. Both carbon- and chlorine-containing deposits were found in the post-reaction catalyst samples.


Introduction
Recently we have shown that catalytic activity of Pt/Al 2 O 3 in CCl 4 hydrodechlorination (HdCl) exhibits very strong inverse relationship with metal dispersion; highly dispersed Pt samples exhibit very low turnover frequencies [1]. A similar, but a much milder trend, was found for silicasupported Pt catalysts. This significant support effect was attributed to an extensive surface chloriding of small Pt particles interacting with Lewis acid sites of c-alumina. It was proposed that such interactions lead to the formation of electrodeficient Pt sites, which are quickly blocked by produced chloride species. On silica, similar in kind metalsupport interactions do not occur, and, in effect, deactivation of Pt/SiO 2 catalysts is less marked. These results were in line with previously published data by Zhang and Beard [2] for Pt/Al 2 O 3 catalysts with varied metal dispersions catalysts and by Prati and Rossi [3] who showed an exceptional catalytic behavior of Pt/Vycor glass (96 % silica).
The idea that catalyst deactivation is caused by chloride species seems to be supported by the results obtained for CCl 4 HdCl on Pt-Au/Al 2 O 3 catalysts [4]. A higher activation energy for monometallic Pt/Al 2 O 3 (56 vs. *30 kJ/mol for the bimetallics) suggests that chlorine removal from Pt surface would be more difficult for Pt only surface than for bimetallic Pt-Au surface, in line with the magnitude of metal-chlorine bond energy (E Pt-Cl [ E Au-Cl ). However, the ultimate evidence as to the reason of catalyst deactivation (chloride or carbonaceous species) was not provided so far, so it is still a matter of opinion.
In this respect a considerable part of published data seem to be in an apparent conflict with the aforementioned results [1][2][3]. Very good HdCl performance of highly dispersed Pt/ Al 2 O 3 catalysts was reported by various groups [5][6][7]. Choi et al. [5] showed that the kind of metal precursor is more important factor in shaping the HdCl behavior than the size of metal particles. Pt particles of very small size (*1 nm) supported on alumina, prepared from Pt(II) precursors, showed stable, high conversion of CCl 4 above 99 % and selectivity to CHCl 3 above 78 % and a retarded coke formation, compared to that from H 2 PtCl 6 . Analogous results have recently been reported by Bae et al. [8].
Among various factors influencing the HdCl behavior of supported platinum catalysts one should also consider the reaction conditions, such as contact time, reaction temperature, the H 2 /CCl 4 ratio in the reaction mixture and the protocol of catalyst's screening. Shorter contact times allow to keep low conversion levels and to extract useful information about the reaction mechanism from the selectivity pattern, without considering secondary reactions, such as CH 2 Cl 2 formation in HdCl of CCl 4 on Pt/Al 2 O 3 found by Dal Santo et al. [7]. Higher reaction temperatures, i.e. [393 K in the case of CCl 4 HdCl, lead to a complete conversion and to more extensive catalyst deactivation. This detrimental effect is softened by working with hydrogen-richer reaction mixtures (e.g. H 2 /CCl 4 ratio = 9, [5]).
It must be stressed that the HdCl of CCl 4 was investigated by us [1,4] at a relatively low reaction temperature, 343-363 K. Our intention was to avoid operation at very high reactant conversions, i.e. at nearly 100 %, easily reachable for HdCl of CCl 4 at the reaction temperatures C400 K (e.g. [7][8][9]). However, even at such relatively mild reaction conditions (B363 K and the feed H 2 /CCl 4 ratio *7) a considerable catalyst deactivation was observed. It must be recalled that the H 2 /CCl 4 molar ratio *7 was intentionally used by us because a similar reactant ratio was used in the studies we wished to refer to [2,3]. However, it is well known that the H 2 /CCl 4 molar ratios higher than 7 reduce catalyst deactivation [5][6][7]. The aim of this work was the reinvestigation of the same series of alumina-and silica-supported platinum catalysts (as in [1]) in the HdCl of CCl 4 , at different reaction temperatures and H 2 /CCl 4 molar ratios. Less surface deactivation expected at hydrogen-rich conditions should be manifested by higher overall activity and CHCl 3 selectivity (at the expense of dimeric species formation). During the realization of this program we found that the catalyst's screening protocol may also have a considerable effect on the catalytic behavior of Pt catalysts.

Preparation and Characterization of the Catalysts
Preparation and characterization (by CO chemisorption, XRD and TEM) of two series of 1.5 wt% metal-loaded Pt catalysts was reported in [1]. Briefly, they were prepared by incipient wetness of alumina (Sasol Puralox SCCA, 150-200 mesh, 196 m 2 /g) and silica (Davison 62, 120-200 mesh, 268 m 2 /g) with an aqueous solution of H 2 PtCl 6 . After impregnation, Pt/SiO 2 and Pt/Al 2 O 3 precursors were divided into small parts, and by combination of different conditions of calcination, reduction and sintering (in Ar or wet H 2 ) two series of catalysts with platinum dispersion in a relatively wide range were obtained [1]. In this study only selected samples from two series of 1.5 wt% metal-loaded Pt catalysts were subjected to catalyst screening. Table 1 shows the basic characteristics of selected samples, keeping the same catalyst designation as in the previous work [1]. CO chemisorption was used for assessing metal dispersions and Pt particle sizes. For the silica-supported series, Pt particle size was mostly confirmed by XRD diffraction studies. Transmission electron microscopy provided essential information about metal particle growth in platinum catalysts subjected to various regeneration pretreatments [1]. In addition, the previous work showed that a prolonged screening of supported Pt catalysts in HdCl of CCl 4 , at a relatively low reaction temperature (343-363 K), does not lead to detectable metal sintering. Kinetic results obtained in the present study will be discussed in the light of catalyst characterization obtained in that previous work [1].
After kinetic runs the catalysts were investigated by a temperature-programmed hydrogenation (TPH), followed by mass spectrometry (MA200, Dycor-Ametek, Pittsburgh, USA), to detect species which could be removed by hydrogen from the catalysts used. TPH runs were performed by flowing a 20 % H 2 /He mixture (25 cm 3 /min) at a 10°C/min ramp. In TPH runs of the used catalysts (samples *0.085 g), attention was paid to m/z 15 and 16 (methane evolution), and m/z 36 and 38, which are suggestive of HCl liberation from the catalysts used.

Catalytic Tests
Prior to the reaction, the catalyst sample was dried at 393 K for 0.5 h in an argon flow and reduced in flowing 20 % H 2 /Ar (25 cm 3 /min), ramping the temperature from 393 to 673 K (at 8 K/min) and kept at 673 K for 2 h. The reaction of HdCl of tetrachloromethane (analytical reagent from POCh, Gliwice, Poland, purity, [99.6 %), provided from a saturator maintained at 273 K and bubbled in a flow of H 2 ? Ar mixture (29 cm 3 /min, with variable H 2 -to-Ar In a specially designed experiment, one Pt/Al 2 O 3 catalyst characterized by small metal particles (\2 nm, A1) was investigated after pretreating it with HCl ? H 2 mixture at the temperature 363 K for 2 h. To this aim, HCl was produced in situ from CCl 4 and H 2 , in an upstream located reactor, filled with another portion of Pt/Al 2 O 3 , which operated at 573 K. This procedure, taken from [9], generates HCl and methane, according to the reaction: CCl 4 ? 4H 2 ? 4HCl ? CH 4 . Large amounts of produced methane should not block platinum sites at such low temperature as 363 K and at high hydrogen pressure. Previous works on methane/deuterium exchange on platinum [10,11] showed that the reaction goes via a stepwise   mechanism, i.e. it involves the participance of reversibly adsorbed methyl radicals, which should be easily removed from the catalyst surface by flushing it with hydrogen. Table 1 shows the basic characteristics of six supported platinum catalysts selected from a larger representation of platinum catalysts used in [1]. A large variety of conditions of sample pretreatment resulted in preparation of catalysts, characterized by an extensive range of metal dispersion. The catalysts chosen for the present study cover the most interesting (from the catalytic standpoint) range of metal particle sizes, from \2 to *8 nm, therefore their catalytic behavior should also reflect expected surface-sensitivity effects.

Results and Discussion
Our previous studies [1] on HdCl of CCl 4 on supported Pt catalysts were performed using a standard H 2 /CCl 4 ratio *7:1. In the present study the same catalysts were investigated at different hydrogen partial pressures. Figure 1 (left section) shows that at the beginning this Pt/Al 2 O 3 catalyst (characterized by very small Pt particles, \2 nm) exhibits strong deactivation with time-onstream. In addition, the selectivity towards dimeric C 2 H x Cl 6-x products (mainly C 2 Cl 6 and C 2 H 2 Cl 4 ) is high, although somewhat decreasing during the long-term testing. The selectivity for chloroform (desired product) is rather modest, not exceeding 60 %. The overall conversion level at a quasi-steady state is only *0.05. Such a catalytic behavior was attributed to an extensive surface chloriding of small Pt particles interacting with Lewis acid sites of c-alumina [1]. It was also suggested that such interactions lead to the formation of electrodeficient Pt sites, which are quickly blocked by the produced chloride species.
The increase of hydrogen pressure by factor of 2 radically changes the catalytic performance of this highly dispersed alumina-supported catalyst, as it is seen in Fig. 1a. The overall conversion levels, established during the lower H 2 /CCl 4 ratio (6.7; 30.1/4.47 kPa) screening, start to increase gradually and after *20 h of reaction attain a rather high value, *0.5. At the same time, the selectivity towards chloroform reaches *90 % and C 2 -dimers nearly cease from the reaction products. An order of magnitude increase of the overall reaction rate cannot be rationalized when one assumes a slightly positive (fractional) reaction order in hydrogen found in the kinetic studies of HdCl reactions [12,13]. A rough calculation of present data (only two experimental points!) would suggest that the reaction order in H 2 is higher than 3, which does not seem a trustworthy value. In addition, a gradual and rather long-lasting improvement clearly indicates that  On the other hand, if catalyst screening is realized with downward variation of hydrogen pressure, starting from P H 2 = 96.9 kPa (at H 2 /CCl 4 ratio = 21.7), overall changes in the catalytic behavior are different from the previous situation (Fig. 1b). A very high conversion level (*0.9) observed for P H 2 = 96.9 and 60.1 kPa decreases only to *0.7 after the respective hydrogen pressure change to 30.1 kPa (at H 2 /CCl 4 ratio = 6.7). Compared with the upward variation of hydrogen pressure (conversion *0.05 at P H 2 30.1 kPa), this high conversion level is accompanied by a large improvement in CHCl 3 selectivity and an apparent absence of C 2 -dimeric species. Only when the hydrogen pressure was decreased to 21.4 kPa (at H 2 /CCl 4 ratio = 4.8), the level of conversion drops to *0.12, with simultaneous appearance of C 2 -dimeric products. Similar situations were observed for a highly dispersed Pt/SiO 2 catalyst (S1) and the other two Pt/Al 2 O 3 catalysts (A6 and A9), however respective changes in the catalytic behavior were not as large as for the catalyst A1. 1.5 wt% Pt/SiO 2 catalyst characterized by the highest metal dispersion (S1) strongly deactivates during first minutes of reaction (the decrease in conversion from *0.9 to 0.2, results not graphically presented). However, after *1 h of time-onstream its catalytic performance gradually undergoes a significant improvement, reaching *0.8 conversion and a similar level of CHCl 3 selectivity. Such a behavior would be explained by a fast initial deactivation of small Pt particles (as also seen for the highly dispersed Pt/Al 2 O 3 catalysts) caused by an immediate deposition of unreactive species. However, the further progress of reaction, still with H 2 /CCl 4 ratio of 6.7 (P H 2 = 30.1 kPa), allows the restoration of a very high activity of this catalyst. At the same time, the C 2 -dimers cease from reaction products.
The trends associated with different catalyst screening protocol (downward or upward variations in P H 2 ) are collected in Fig. 2. They indicate large changes associated with the catalyst deactivation, accompanied by downward changes of P H 2 or catalyst reactivation, when the hydrogen pressure was increased. We believe that the much better catalytic behavior of the catalyst samples which were first subjected to hydrogen-richer reaction conditions results from a kind of catalyst activation. At very high CCl 4 conversion, when only negligible amounts of C 2 -dimeric species are formed, huge amounts of HCl (inevitable HdCl product) and hydrogen (reactant in excess), must interact with the catalyst surface rendering it more suitable for this reaction. In this respect it should be noted that a considerable induction period in the HdCl behavior of Pt/MgO catalyst disappeared when reaction was conducted after Pt/MgO had been pretreated with 18 % HCl-H 2 gas and reduced with hydrogen at 573 K for 2 h [12]. Kim

A1
A6 A9 S1 S8 S11 silica (S1, S8, S11) supported platinum catalysts. For catalyst notation see Table 1. Apparent activation energies (E A 's) refer to all hydrogen partial pressure conditions HdCl catalysts. We have also tested the effect of catalyst's pretreatment with HCl. To this aim, the experiment with A1 catalyst pretreated with CCl 4 and H 2 (described in subsection 2.2, HCl was produced in situ from CCl 4 ), at 363 K for 2 h resulted in an improvement of catalytic behavior (expressed as a single star in upper section of Fig. 2). It must be stressed that although our conditions of catalyst's chloriding were much milder than those in [12,13], the effect of this pretreatment substantiates the speculations as to the beneficial role of the catalyst's screening protocol starting from hydrogen-rich conditions. It appeared clear from the foregoing that the search for the effect of hydrogen partial pressure on the catalytic behavior should be realized by catalyst screening at a gradually descending H 2 /CCl 4 ratio. Results of such studies are shown in Fig. 3, presented as log (conversion) -log (P H 2 ) relations, just to check if one has a direct insight into the reaction order in hydrogen. Different situations were found for various platinum catalysts. The highly dispersed catalysts (A1 and, especially, S1) showed nearly full conversion at the reaction temperature 363 K, at P H 2 between 30.1 and 96.9 kPa. Such very high conversion levels preclude assessment of real reaction orders. The numbers showed in a few cases (n h 's, from line slopes) in Fig. 3 serve only as an indication of a linear character of certain correlations. Very low apparent activation energies found for hydrogen-rich conditions do not also reflect the real reaction kinetics. In this respect, more realistic is the kinetic data collected for catalyst A1 (and A6) at lower reaction temperatures, i.e. at 343 and 353 K: n h values are close to , i.e. suggestive of the participation of dissociated hydrogen [14,15]. It is also easy to notice an abrupt decrease of the overall conversion when the P H 2 is changed from 30.1 to *21 kPa. But in this case, the n h values higher than 5 (for A1) and 2 (for S1) indicate that under hydrogen deficient conditions, both Pt catalysts with high dispersion are deactivated. On the other hand, the apparent activation energies for A1 and S1 catalysts collected at hydrogen-poor conditions are close to the respective values reported in the literature [1,3,14].
In other cases it is impossible to draw straight line relations (Fig. 3, A9, S8, S11). For less metal dispersed Pt/SiO 2 catalysts (Fig. 3, S8, S11), the respective log-log relations possess maxima. This may suggest that CCl 4 HdCl undergoes via a Langmuir-Hinshelwood mechanism [16], when both reactants (CCl 4 and H 2 , in our case) compete for active surface sites. Similar mechanism was earlier suggested to operate in HdCl of dichlorodifluoromethane over supported Pd catalysts [17]. It should also be recalled that so-called volcano-(or bell-) shaped relations between catalytic activity and hydrogen pressure were frequently reported and interpreted for alkane catalytic hydroconversions (hydrogenolysis, isomerization or dehydrocyclization) on unsupported and supported platinum catalysts [18][19][20]. Paál [21] demonstrated that hydrogen coverage effects in alkane catalytic conversion would lead to a drastic decrease of activation energy, resulting even in inverse Arrhenius plots. Such effects would also operate in catalytic HdCl, although the present results obtained at very highly conversion level render a more detailed discussion difficult. However, it must be emphasized that in the case of S8 and S11 catalysts, which exhibited maxima in the conversion-H 2 pressure relations, the apparent activation energies were found roughly between 27 and 36 kJ/mol, not much dependent on the H 2 pressure.
In order to decide between poisoning effects of deposited chlorine/chloride species and carbonaceous residues, the TPH runs appears helpful in showing how to eliminate chlorine from a catalyst without a considerable removal of carbon. The 1.5 wt% Pt/Al 2 O 3 and Pt/SiO 2 catalysts used in CCl 4 HdCl showed the maximum of chlorine removal rate by hydrogen at the temperature below 600 K (Fig. 4a).
For the alumina-supported catalysts a pronounced tail in HCl trace developed at higher temperatures suggests evolution from the support. Carbon-containing species leave (as methane) all tested catalysts at much higher temperatures. Therefore these carbonaceous deposits should be regarded as most effective active site blockers, in agreement with others [5,6,8,22]. Very low amounts of desorbed HCl and CH 4 from a medium-dispersed Pd/SiO 2 catalyst (S8, fraction exposed [FE] = 0.264, Fig. 4a, b) are in good agreement with our earlier report [1], where Fig. 2 (on p. 252) also shows rather small amounts of liberated HCl and methane from catalyst S5 (FE = 0.377, not tested in the present study). Both the low platinum dispersion and the fact of using silica support appear to rationalize such a behavior, and, in effect, much less marked than in the case of alumina-supported counterparts, catalyst's deactivation.

Conclusions
Two series of silica-and alumina-supported platinum catalysts were investigated in the HdCl of tetrachloromethane at the reaction temperature range 343-363 K, at different H 2 /CCl 4 ratios and very short contact times (B0.26 s).
During an initial period of reaction carried out at a lower H 2 /CCl 4 ratio (6.7 and less), all catalysts deactivated with time-on-stream, the effect was more drastic for highly dispersed Pt catalysts. After a long-term screening and reaching a steady state, the increase of H 2 /CCl 4 ratio to 13.4 brought about significant changes in the catalytic behavior. In particular, this effect was remarkable for the Pt catalysts characterized by the highest metal dispersions, for which the overall conversion rate was increased by an order of magnitude and the selectivity to chloroform raised from *60 to *90 %. Such impressive activity growth cannot be explained by a positive reaction order in hydrogen, the more so as this variation was not immediate but it was progressively developed with time-on-stream. It seems that during the initial reaction period at the lower H 2 /CCl 4 ratio, the catalysts quickly deactivated by unreactive deposits, but at a higher hydrogen pressure these deposits were gradually removed, making the surface of platinum active and selective towards chloroform. A prolonged catalyst screening carried out with gradually decreasing H 2 /CCl 4 ratios (from 21.7 to 6.7, and less) resulted in much better catalyst's performance than it was observed for the reversed order of H 2 /CCl 4 ratio. It appears that the catalysts subjected to hydrogen-rich reaction conditions are better activated, probably because of the action of an intense HCl flux, generated in the reaction. Because of a very high activity of a few tested samples, changes of CCl 4 conversion with the hydrogen partial pressure do not reflect real reaction orders in hydrogen. The same reason may lead to falsification of apparent activation energy. In other cases the relation between conversion and hydrogen pressure shows maxima, suggesting that HdCl undergoes via a Langmuir-Hinshelwood mechanism, when hydrogen and CCl 4 compete for metal surface sites. Both carbonand chlorine-containing deposits were found in the postreaction catalyst samples.