Hydrodechlorination of CHClF2 (HCFC-22) over Pd–Pt Catalysts Supported on Thermally Modified Activated Carbon

Commercial activated carbon, pretreated in helium at 1600 °C and largely free of micropores, was used as a support for two series of 2 wt.% Pd–Pt catalysts, prepared by impregnating the support with metal acetylacetonates or metal chlorides. The catalysts were characterized by temperature-programmed methods, H2 chemisorption, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and scanning transmission electron microscopy (STEM) with energy dispersive spectroscopy (EDS). Overall, the results confirmed the existence of well-dispersed Pd–Pt nanoparticles in the bimetallic catalysts, ranging in size from 2 to 3 nm. The catalysts were investigated in the gas phase hydrodechlorination of chlorodifluoromethane (HCFC-22). In this environmentally relevant reaction, both the ex-chloride and ex-acetylacetonate Pd–Pt/C catalysts exhibited better hydrodechlorination activity than the monometallic catalysts, which is consistent with the previous results of hydrodechlorination for other chlorine-containing compounds. This synergistic effect can be attributed to the electron charge transfer from platinum to palladium. In general, product selectivity changes regularly with Pd–Pt alloy composition, from high in CH2F2 for Pd/C (70–80%) to the selective formation of CH4 for Pt/C (60–70%).


3
In this work, we decided to reinvestigate the Pd-Pt/C catalysts more thoroughly. This time, we decided to pretreat our samples at higher than 300°C pre-oxidation temperatures, prior to catalyst reduction. This pretreatment will be referred to as the 'pre-calcination' along the text.
Because the pretreatment resulted in a considerable improvement of catalytic performance (especially for Pd/C catalysts), we also performed different tests of catalyst characterization, evaluating for changes in metal dispersion, crystallite size of the metal phase, and changes in the pore structure produced by pre-oxidation. Temperature Programmed Oxidation (TPO) of carbon-supported metal catalysts was intended to provide information on the forms of carbon removed at different stages of catalyst pre-calcination [21].

Results
Catalytic screening of both series (ex-acac and ex-chloride) of Pd-Pt/Norit1600 catalysts in CHClF2 hydrodechlorination showed that stable conversions, always <3%, were achieved after a relatively long time-on-stream (~16 h), as reported in our previous paper [17].
Detailed data showing the catalytic properties of Pd-Pt/Norit1600 catalysts pre-calcined at different temperatures are in Tables S1-S4, placed in the Supplementary Materials. The effect of pre-calcination temperature of the catalysts is demonstrated in Figures 1 and 2. As in our previous work [17], for the Pd-rich samples, CH4 and CH2F2 (HFC-32) were found to be the predominant products, making up more than 90% of all products (Tables S1-S4 and Figure 1).
For the monometallic Pt samples, CH3F and CHF3 made up ~30% of all products. During stabilization, changes in product selectivity were relatively small, and often CH2F2 formation was increased at the expense of methane.
The product selectivity presented in Figure 1 (A, B, C) shows the best selectivity to CH2F2 for both Pd100 catalysts (approaching 80%, for reaction temperature ~251°C, Tables S1-S4) and rather smooth changes in the selectivity with addition of platinum. Apart from the parallel, smooth increase in CH4 selectivity, the catalysts richer in platinum demonstrated the formation of CH3F and even CHF3. Regular variations of product selectivity with Pd-Pt alloy composition would imply no synergy in the behavior of the bimetallic system, but rather a linear cumulative effect in the catalytic action of the alloys. These relations resemble very much the previous selectivity patterns obtained for the same catalysts pre-calcined at 300°C for 1 and 2 h [17].  Still more interesting results relate to the evolution of catalytic activity of Pd-Pt/Norit1600 systems produced by different pre-calcination conditions. Figure 2 shows that the increase in pre-calcination temperature produces an increase in the activity of all catalysts, however, the Pd-rich catalysts (Pd100 and Pd80Pt20) gain the most. The turnover frequency of Pd100(acac)/Norit1600 experiences a spectacular increase, by nearly an order of magnitude, from ~0.00041 s -1 to 0.00263 s -1 . catalysts pre-calcined at 300 o C and for Pd-Pt(Cl) non-calcined samples were taken from [17].
Our earlier study [17] showed that the ex-acac Pd-Pt/Norit1600 catalysts were less active than the ex-chloride ones. This is recalled in Figure 2. The reasons for this difference could not be seen rationalized in terms of possible contamination with carbon-containing species because the ex-acac catalysts were subjected to intensive oxidation at 300°C, whereas the exchloride samples were not pre-oxidized. Nevertheless, pre-calcination of the ex-chloride samples at 400°C also increases their reactivity ( Figure 2). However, a strong maximum for Pd60Pd40 for non-calcined ex-Cl samples is flattened upon pre-calcination. The red triangle in Figure 2 represents the catalytic behavior of 3 wt.% Pd/Norit1800, the catalyst prepared using the Norit carbon preheated in helium at 1800°C [23]. Its basic characteristics is in the Supplementary Materials (Table S4 and SET S1). Its good catalytic performance in CHClF2 6 hydrodechlorination is achieved without calcination in oxygen. On the other hand, additional calcination practically does not change its catalytic activity (result not shown).
The TOF values have been calculated taking into account the metal dispersions determined by hydrogen chemisorption, H/(Pd+Pt). Metal dispersion did not change significantly with the change in pre-calcination conditions (data in Tables S1-S4 vs. relevant dispersion values reported in [17]). In line with the chemisorption studies, XRD examination of reduced Pd/Norit1600 catalysts confirmed the presence of only minor changes in the metal crystallite size at different pre-calcination temperatures ( Figure 3). Thus, if pre-calcination of Pd/Norit1600 does not lead to a significant change in metal dispersion, the large increase in the activity of this catalyst could result from additional cleaning of its surface from carbon. The problem of decorating the surface of palladium with carbon from the carrier has been known for years [20][21][22]. The relative consistency of the size of the palladium particles resulting from the metal dispersion measurement (dPd = 7 1.12/(H/Pd), [24]) and the size of the crystallites Pd ( Figure 3) suggests that metal surface decoration does not take place here. It should be added that the pre-treatment of all ex-acac catalysts included their initial pre-calcination at temperatures ≥300°C, a step suggested for carbon elimination from the palladium surface [20][21][22]. Nevertheless, we decided to carry out TPO measurements of Norit-supported palladium catalysts that could shed light on the forms of carbon eliminated in the course of oxidation.
The results of TGA-TPO studies of the Pd100(acac)/C, Pt100(acac)/C and Norit1600 carbon are shown in Figure 4. As carbon oxidation (seen as evolution of CO2) is catalyzed by Pd and Pt, a considerable decrease in the burn-off temperature of carbon, by ~200 o C, compared to the oxidation of bare Norit1600 carbon can be observed. However, the shape of TPO profiles for Pd/C and Pt/C does not contain any additional characteristic peaks, which were observed in TPO of Pd/C catalysts by Tengco et al. [21] and could be ascribed to the burn-off of carbon species released from the surface and subsurface layer of palladium.
However, our observations are consistent with the results of the cited work because the preoxidation of Pd/C at 300°C used in our preliminary catalyst pretreatment appears to be sufficient in order to remove superficial carbon from palladium. Therefore, massive carbon removal (TG) accompanied by vast CO2 evolution (m/z 44) at temperatures above 300°C (especially at 400°C, see dotted line in Figure 4) should be attributed to the oxidation of support carbon in close contact with metal, called 'proximate' carbon in [21].
There are some differences in the course of TPO profiles for both metals. The onset of this increase is for the Pt/C catalyst delayed by ~30°C compared to the behavior of Pd/C (Fig. 4).
This 'delay' would be due to the fact that there is less 'proximate' carbon near platinum, which, in turn, may be due to the lower adherence of platinum to carbon. It is known that carbon nanotubes are better wetted by palladium than by platinum [25]. However, after exceeding the temperature of ~400°C, platinum shows a higher rate of carbon oxidation. In conclusion, we suggest that the massive firing of pore walls catalyzed by the presence of entrapped metal particles removes steric obstacles for free access of gas reactants to the entire metal surface, making it more active.   Table 1. Figure 5 shows N2 adsorption isotherms for 2 wt.% Pd/Norit1600, 2 wt.% from the blue one (for Norit) is larger for Pd/C than for Pt/C. The results in Table 1 show that the impregnation of Norit1600 with platinum only slightly decreases the BET surface area and pore volume of the system. In the case of palladium, the corresponding changes are larger. This may result from the fact that, on average, the Pt100(acac) catalyst contained smaller metal particles than the Pd100(acac) (~3 vs. ~7 nm, Table 1 in 17). However, it should be noted that the volume of introduced platinum was ~1.5 times smaller than the volume of introduced palladium (with the same 2 wt. % metal loading in all catalysts).
After oxygen treatment at 400°C, the N2 adsorption isotherm for Pt/C is more markedly shifted toward higher N2 uptake than the isotherm for Pd/C. This may result from the higher catalytic activity of platinum (than palladium) in carbon oxidation. Platinum appeared to be a uniquely good metal in soot oxidation [26]. Overall, interesting conclusions follow from the evolution of micropore volume (tplot). Impregnation of Pd leads to a loss of micropore volume by 0.00389 cm 3 /g (= 0.00577-0.00188, Table 1). Although the total volume of introduced palladium ~0.0017 cm 3 /g would be accommodated in the micropores, it is certain that only a part of Pd nanoparticles blocks the micropores, probably at their outlets. The analysis of the distribution of metal particle size ( Table 1 in [17]) indicated the presence of a large variety of differently sized metal particles, some of them would not fit the micropores. Pd/C precalcination at 400°C for 15 min recovers a substantial part of micropores, suggesting that very small Pd nanoparticles would now be more accessible to the gas reactants.  Pt/Norit1600 (bottom). The isotherm for Norit1600is included in the graphs. Figure 6 presents variations in the pore size distribution in impregnated and calcined Norit1600, Pd/C and Pt/C catalysts. Again, the pore structure of the Pt100/C catalyst does not differ much from the structure of Norit1600 (black vs. white bars in Figure 6). For Pd100, the differences are larger: all black bars (with exception for very large pores) are smaller than the  From the results of the physical characterization of Norit1600 supported Pd and Pt catalysts, one can conclude: 1) Oxidation of Pd-on-highly preheated Norit catalyst, which leads to nearly an order of magnitude increase in catalytic activity cannot be the result of marked changes in metal dispersion changes because such changes have not been found.
2) Possible decontamination of palladium surface from carbon by oxidation at 350-400°C is also rejected as a basic reason for the activity increase. Preliminary pre-calcination of Pd/Norit1600 samples at 300°C for 1-2 h should remove the carbon from the metal surface [21,22]. TPO profiles of such catalysts do not contain any signs of presence of such carbon species.
3) TGA-MS studies show the beginning of a massive removal of a "proximate" carbon at the temperature 350°C. Metal-catalyzed burning of carbon support changes the pore structure of Norit1600. In particular, the micropore volume is vastly increased along with catalyst oxidation. Metal nanoparticles, wetting small pores of the support, presumably lose contact with the pore walls as a result of oxidation. Such a catalyst represents enhanced reactivity, proving the accessibility of the active sites to reactants. show the sharp maximum at 40 at.% Pt, but after pre-calcination at ≥350°C, this maximum disappears. This means that the synergistic effect in catalytic action of Pd and Pt reported in our previous paper [17] ceases to exist. The previously discussed product selectivity data are also in opposition to the synergy. Figure 7 summarizes the results of our previous and present study. The course of the relation between the TOF and Pd-Pt alloy concentration after burning carbon from pore walls does not contain any maxima but tends to reflect the inferred Pd content in the surface layer of Pd-Pt alloys (inset in Figure 7). This suggests that the course of  Good catalytic performance of non-calcined 3 wt.% Pd/Norit1800 catalyst used in our earlier work [23], shown in Figure 2 (red triangle), should be commented on. A relatively low BET surface area and pore size distribution in this sample (SET S1 in Supplementary Materials) appear to favor good access of reactants to active sites. The very small micropore volume in this impregnated catalyst (0.0010 cm 3 /g, SET S1), in combination with the lack of micropores in carbons heated at 1800°C [29], confirms the absence of Pd particles in micropores.
This result calls for further studies with use of very highly pretreated carbons.

Catalyst Preparation
The preparation and basic pretreatment were described in our previous paper [17]. Briefly, As mentioned in the previous report [17], the removal of the organic part of acetylacetonates achieved by calcination in flowing 1% O2/Ar (50 cm 3 /min, at STP) at 300°C for 1 h. In our previous work, the pre-calcined catalysts were subjected to reduction in the flow of the H2/Ar mixture (20 cm 3 /min, STP) at 400°C before the reaction. In the present work, the stored catalysts were subjected to more severe secondary calcinations in O2/Ar at 320°C, 350° and 400°C.
Some kinetic runs were performed with a few Pd, Pt and Pd-Pt catalysts prepared from the chloride precursors. Their preparation was also described in [17].

Catalytic Tests
The hydrodechlorination of chlorodifluoromethane was conducted in a glass flow system equipped with a gradientless reactor, as described previously [17]. Prior to each reaction run, the catalyst (0.20 g sample) was subjected to pre-calcination step in flowing 1% O2/Ar (20 cm 3 /min, STP) at 320°C and 350°C for 1 h, or at 400 o C for 15 min. The pre-calcination was followed by a short purge with Ar at 400°C, and the reduction in 10% H2/Ar (20 cm 3 /min, STP) at 400°C for 1 h. After reduction, the catalysts were cooled in H2/Ar flow to the desired initial reaction temperature, i.e., 270°C. For a typical reaction run, the total flow of the reactant mixture was 48 cm 3 /min and consisted of CHClF2 (1 cm 3 /min), H2 (8 cm 3 /min) and Ar (39 cm 3 /min), fixing the GHSV at 5760 h -1 . This high value allowed the maintenance of low conversions, usually <3%, and minimized secondary reactions. The reaction was carried out until a steady state was achieved at 270°C (~16 h). Then, the reaction temperature was gradually decreased to 260°C and 250°C and new experimental points were collected. A typical run lasted ~20 h. The post-reaction gas was analyzed by gas chromatography. Some kinetic runs were performed with a few pre-calcined Pd-Pt/C catalysts prepared from metal chlorides. In those cases, the reduction in H2/Ar at 400°C lasted 3 h. Finally, the 3 wt.% Pd supported on Norit carbon preheated at 1800°C, used in our previous studies [23], was also tested.

Catalyst Characterization by H2 Chemisorption, XRD, Temperature Programmed
Oxidation, and Physical Adsorption (BET, pore structure) As in our previous work [17], metal (Pd, Pt) dispersion was determined using hydrogen chemisorption, following the procedure recommended by Bartholomew for Fe, Co and Ni catalysts [30]. A reduced and outgassed catalyst was saturated with hydrogen at 70°C, and the amount of adsorbed hydrogen was measured through temperature programmed desorption in an argon stream, increasing the temperature 20°C/min.

Conclusions
Two series of Pd-Pt catalysts (ex-metal acetylacetonates and ex-metal chlorides) supported on Norit activated carbon preheated at 1600°C have been reinvestigated in the hydrodechlorination of chlorodifluoromethane. An additionally adopted catalyst oxidation pretreatment, at 320, 350 and 400°C, resulted in an order of magnitude increase in catalytic activity of the monometallic Pd/C. Neither possible changes in metal dispersion nor metal decontamination from carbon were found responsible for this effect. TGA-MS studies show the beginning of massive removal of a "proximate" carbon at the temperature 350°C, i.e., at the onset of the increase in catalytic activity. Metal-catalyzed burning of carbon support modifies the pore structure of Norit1600. The micropore volume in particular is vastly increased as an effect of catalyst oxidation. Metal nanoparticles, wetting the small pores of the support, presumably lose contact with the pore walls as a result of oxidation. Such a catalyst demonstrates enhanced reactivity, hence proving the accessibility of the active sites to reactants. Therefore, the reactivity tuning of Pd/C catalysts using oxidation is not due to changes in metal dispersion but from unlocking the active metal surface, originally inaccessible in Pd particles tightly packed in the pores of carbon. In agreement with our speculations, a non-calcined Pd/C catalyst supported on the carbon preheated at 1800°C showed good catalytic behavior. Separate runs with 3 wt.% Pd/carbon under somewhat intensified reaction conditions presented very good catalyst stability and excellent selectivity to CH2F2 (>90%).
In our discussion we did not consider the possible inhibiting role of residual chloride originating from the metal precursor for two reasons. First, the ex-chloride catalysts were generally more active than the ex-acac ones. Second, during the reaction of CHClF2 hydrodechlorination, the metal surface, regardless of its origin, is covered by a variety of active Cl/F and HCl/HF species, so the possible role of residual chloride from the precursor would be difficult to distinguish.
Turnover frequencies, product selectivities and activation energies,