Hydrodechlorination of Tetrachloromethane over Palladium Catalysts Supported on Mixed MgF2-MgO Carriers

Pd/MgO, Pd/MgF2 and Pd/MgO-MgF2 catalysts were investigated in the reaction of CCl4 hydrodechlorination. All the catalysts deactivated in time on stream, but the degree of deactivation varied from catalyst to catalyst. The MgF2-supported palladium with relatively large metal particles appeared the best catalyst, characterized by good activity and selectivity to C2-C5 hydrocarbons. Investigation of post-reaction catalyst samples allowed to find several details associated with the working state of hydrodechlorination catalysts. The role of support acidity was quite complex. On the one hand, a definite, although not very high Lewis acidity of MgF2 is beneficial for shaping high activity of palladium catalysts. The MgO-MgF2 support characterized by stronger Lewis acidity than MgF2 contributes to very good catalytic activity for a relatively long reaction period (~5 h) but subsequent neutralization of stronger acid centers (by coking) eliminates them from the catalyst. On the other hand, the role of acidity evolution, which takes place when basic supports (like MgO) are chlorided during HdCl reactions, is difficult to assess because different events associated with distribution of chlorided support species, leading to partial or even full blocking of the surface of palladium, which plays the role of active component in HdCl reactions.


Introduction
Catalytic hydrodehalogenation (HdCl) of harmful organic compounds, such chloromethanes, CFCs or chlorobenzenes on supported metal catalysts, appear a useful method for transforming those detrimental compounds into valuable chemicals [1,2]. The active phase of these catalysts are metals, among which the best performance show platinum, palladium and nickel [2][3][4][5]. Recently published review article by Colombo et al. [6] highlights the suitability of iron catalysts in removing chlorocarbons from the water. In addition, the performance of the iron in water decontamination can be greatly improved by doping with other metals such as palladium, copper, silver and nickel. Besides, the role of support should not only be considered as a metal redispersing medium. Carbon is the support of industrial choice for a number of reasons, including high specific surface area and stability against corrosion by halogen-containing products of hydrodehalogenation (HCl, HF). Research with the use of other supports, such as silica, alumina, zeolites or magnesia, served mainly for short-term model studies, when the functioning of various metallic phases is tested. The role of acido-basic properties of a support is not well understood, although several suggestions were made in this respect. Studied in the reaction of hydrodechlorination of 1,1-dichlorotetrafluoroethane, Pd catalysts supported on AlF 3 and MgF 2 (characterized by very dissimilar acidities) did not show large differences in their catalytic behavior [7]. It would speak against any special effect of support acidity on the hydrodechlorination performance. However, it is also known that the acidic or basic character of the support would modify metal particles, making them electrodeficient or electron-rich. The electron deficiency was found to depend on the concentration and the location of Brønsted acid sites [8]. Such effect would lead to a weaker interaction of the metal with reaction intermediates contributing to higher activities and product selectivities to, frequently desired, partly dehalogenated compounds [5,9]. Similarly, optimal combination of Pd σ+ and Pd • species has recently been invoked for the best hydrodechlorination performance [10,11]. Electrodeficient palladium, Pd 2+ , has been reported to improve the activity of Pd catalysts in HdCl reactions [12][13][14][15][16][17].
On the other hand, the presence of support acid centers would also lead to accumulation of polymeric species and catalyst's fouling by coking [18,19]. Less acidic supports would act as chlorine "sink", keeping the metal surface clean for a considerable time of reaction. This does not seem to function for highly acidic supports which, in addition, produce polymeric species, which in turn would be spilt over onto the metal surface [18].
In this respect, advantages of MgO as appropriate support of the hydrodechlorination catalyst are presented by several authors [20][21][22]. The high activity of the Pt/MgO catalyst was ascribed to basic properties of MgO. The basicity of this oxide seems to be responsible for retarding the coke formation and suppression of C 2 oligomers (C 2 Cl 4 and C 2 Cl 6 mainly) which are the principal reasons for the catalyst deactivation. Clarke et al. [23] studied MgO-supported Pt catalysts in the hydrogenolysis of C 5 -C 6 hydrocarbons and came to conclusion that large changes in the catalytic behavior of platinum result from negative charge from the magnesia O 2− ions to the metal.
Suitability of MgF 2 -supported metal catalysts for HdCl reactions was proved in numerous studies [7,14,20,21,[24][25][26][27][28]. Because of a variety of investigated reactants, metal precursors (metal salts vs. phosphine complexes of metals, [28]), metal dispersion and structural forms of MgF 2 (high surface area form [26] vs. well-crystalline [25]) it difficult to describe shortly this issue in much detail. However, a general overview appears to appreciate very good overall activity, selectivity to desired products and stability of MgF 2 -supported metals, usually exceeding the performance of their counterparts supported on oxidic carriers. The most often explanation takes into account the beneficial effect of MgF 2 on metal centers making them electrodeficient. MgO → MgF 2 transformation during HdCl of CFCs over MgO-supported metals was also considered as important in shaping the active form of metal catalysts [29].
We have decided to study the hydrodechlorination of CCl 4 carried out over palladium catalysts supported on MgO-MgF 2 carriers, in strong belief that the acido-basic properties of these mixed materials are well correlated with their composition. By increasing the fluorine content of the magnesium oxide fluoride, the Lewis acidity increases whereas the basicity decreases [30]. Therefore, these sites can be tuned over a wide range of MgO-MgF 2 composition thus giving access to optimized catalytic activity and selectivity of these phases as was found for Michael additions [31]. However, during any hydrochlorination reactions which produce HCl the problem becomes more complicated. The reaction leads to a rapid transformation of MgO into MgCl 2 [20,21], which has to be considered in description of a working state of the Pd/MgO-MgF 2 catalyst. Full or even partial replacement of MgO by MgCl 2 should also result in an increase in the strength of Lewis acid sites on the surface of the catalyst.

Catalyst Characterization
Characterization of prepared MgO-MgF 2 carriers was reported in earlier publications, the "sol-gel" series in [32] and the "carbonate" series in [33] and are recalled in Table 1. For the "sol-gel" series, the content of magnesium oxide had a considerable effect on the surface area of the MgO-MgF 2 systems, i.e., the surface area of MgO-MgF 2 was seven times larger than that of pure magnesium fluoride and still larger than that of pure MgO. All the "sol-gel" samples were mesoporous with the mean pore diameters ranging from 8 to 18 nm. The size of the pores strongly depended on the composition of the samples and declined with increasing MgO content in MgO-MgF 2 samples. For the "carbonate" series the surface area of the mixed MgO-MgF 2 supports was 2-3 times greater than that of pure MgF 2 . In general, the surface area of mixed supports increased with increasing amount of MgO introduced. The BET surface area of MgO was almost five times greater than that of MgF 2 . All the "carbonate" series samples were mesoporous with the mean pore diameters ranging from 8 to 24 nm. The XRD studies showed the presence of separate crystalline MgO and MgF 2 phases [33]. Table 1. Support characterization (adapted from [32,33] The present sample characterization concerns only the Pd-loaded MgO-MgF 2 catalysts, with attention focused on the metal component which is commonly regarded as the active phase in hydrodechlorination reactions. Table 2 shows metal dispersion data based on H 2 and CO chemisorptions. Considering the methodological (pulse vs. static) differences, the results obtained are internally consistent. In addition, the TPHD studies (Table 2) show larger H/Pd ratios for poorly dispersed palladium catalysts, in agreement with known correlations [34][35][36]. In the absence of XPS studies, these results would suggest that our palladium material reduced at 380 • C has not been significantly chemically modified by possible interactions with MgO-MgF 2 supports.
However, it is found that introduction of a larger amount of palladium (3%) onto a support characterized by a smaller surface area and lower pore volume (here MgF 2 ) results in the formation of relatively large Pd particles, i.e.,~8 nm (as assessed from CO chemisorption) compared to very small (~2 nm or less) metal particles deposited on MgO and MgO-MgF 2 carriers. It appears that the support of less developed pore structure is not able accommodate larger amounts of dissolved metal precursor during impregation whereas a highly developed pore structure leads to evenly distributed small portions of the precursor. More important is that such large difference in metal dispersion makes a direct comparison of catalytic pattern in these series difficult. Fortunately, smaller differences in surface areas in the "carbonate" series of MgO-MgF 2 supports and lower palladium loadings employed (1% instead of 3%) allowed to prepare the catalysts characterized by comparable metal dispersions ( Table 2). The results of XRD studies of reduced 3 wt % Pd/MgO-MgF 2 catalysts are presented in Figure 1. A variety of large diffraction peaks from crystalline phases of MgO and MgF 2 , overlapping reflections from highly dispersed palladium, prevent identification of the metal. Only in the case of 3 wt % Pd/MgF catalysts, the (200) reflection from the Pd (fcc phase) of low intensity would be identified. Rough estimation of Pd crystallite size from that reflection (using the Scherrer formulae) gives 12-15 nm, i.e., the value in a limited agreement with the Pd particle sizes assessed from chemisorption (Table 1). Accordingly, the diffractograms of the less metal loaded (1 wt %) Pd/MgO-MgF 2 catalysts resemble very much the profiles of unloaded supports [33], although a very small (200) reflection from palladium was also detected in the profile of the 1 wt % Pd/MgF 2 (XRD data not shown). The results of XRD studies of reduced 3 wt % Pd/MgO-MgF2 catalysts are presented in Figure 1. A variety of large diffraction peaks from crystalline phases of MgO and MgF2, overlapping reflections from highly dispersed palladium, prevent identification of the metal. Only in the case of 3 wt % Pd/MgF catalysts, the (200) reflection from the Pd (fcc phase) of low intensity would be identified. Rough estimation of Pd crystallite size from that reflection (using the Scherrer formulae) gives 12-15 nm, i.e., the value in a limited agreement with the Pd particle sizes assessed from chemisorption (Table 1). Accordingly, the diffractograms of the less metal loaded (1 wt %) Pd/MgO-MgF2 catalysts resemble very much the profiles of unloaded supports [33], although a very small (200) reflection from palladium was also detected in the profile of the 1 wt % Pd/MgF2 (XRD data not shown).

Catalytic Activity
All tested Pd/MgO-MgF2 catalysts deactivated during the hydrodechlorination of CCl4 at 90 °C, but the character of deactivation varied between different samples.
Reasons for deactivation of HdCl metal catalysts were analyzed in previous publications. They are considered as metal surface chloriding by a liberated HCl [38][39][40][41], massive deposition of carbonaceous deposits on metal surface [42][43][44][45] and metal sintering caused by long-term operation [46]. It was rather well established that smaller Pd particles (~2 nm or less) deactivate faster than larger Pd crystalllites (~10 nm in size) [36]. Accordingly, our magnesia-containing Pd samples in the sol-gel series which are characterized by very small metal particles ( Table 2) showed very pronounced deactivation during kinetic runs and low final activity ( Figure 2). On the other hand, the superior catalytic performance, with a small decrease of conversion (from ~100% to ~93%) showed the 3 wt. % Pd/MgF2 catalysts (sol-gel series), characterized by poor metal dispersion. Its counterpart from the carbonate series also exhibited high conversion, but a marked deactivation was observed in this case (from ~100% to ~56%, Figure 2, inset). Such somewhat different "deactivation" behavior is understood when one considers that the former catalyst was screened at near full conversion level, i.e. the amount of catalyst sample was too large (always ~0.4 g). However, the turnover frequencies for both catalysts at final conversions (after 15-16 h of reaction) are similar (Table 3), so the apparent difference in the behavior of two 3 wt % Pd/MgO catalysts seen in Figure 2 is easily understood in terms of different metal particle sizes (7.9 vs. 15.6 nm), which differ by factor of ~2. For this metal particle range (~10 nm) one should not expect any special structure-sensitivity, which was usually observed for metal Figure 1. XRD profiles of reduced 3 wt % Pd/MgO-MgF 2 catalysts ("sol-gel" series). Reflections for MgF 2 and MgO are presented by bars proportional to their relative intensities. Inset: diffractograms of 3 wt % Pd catalysts of the "carbonate" series. To simplify the notation the composition of MgO-MgF 2 mixture the % symbol is usually dropped in further text.

Catalytic Activity
All tested Pd/MgO-MgF 2 catalysts deactivated during the hydrodechlorination of CCl 4 at 90 • C, but the character of deactivation varied between different samples.
Reasons for deactivation of HdCl metal catalysts were analyzed in previous publications. They are considered as metal surface chloriding by a liberated HCl [38][39][40][41], massive deposition of carbonaceous deposits on metal surface [42][43][44][45] and metal sintering caused by long-term operation [46]. It was rather well established that smaller Pd particles (~2 nm or less) deactivate faster than larger Pd crystalllites (~10 nm in size) [36]. Accordingly, our magnesia-containing Pd samples in the sol-gel series which are characterized by very small metal particles ( Table 2) showed very pronounced deactivation during kinetic runs and low final activity ( Figure 2). On the other hand, the superior catalytic performance, with a small decrease of conversion (from~100% to~93%) showed the 3 wt. % Pd/MgF 2 catalysts (sol-gel series), characterized by poor metal dispersion. Its counterpart from the carbonate series also exhibited high conversion, but a marked deactivation was observed in this case (from~100% to~56%, Figure 2, inset). Such somewhat different "deactivation" behavior is understood when one considers that the former catalyst was screened at near full conversion level, i.e. the amount of catalyst sample was too large (always~0.4 g). However, the turnover frequencies for both catalysts at final conversions (after 15-16 h of reaction) are similar (Table 3), so the apparent difference in the behavior of two 3 wt % Pd/MgO catalysts seen in Figure 2 is easily understood in terms of different metal particle sizes (7.9 vs. 15.6 nm), which differ by factor of~2. For this metal particle range (~10 nm) one should not expect any special structure-sensitivity, which was usually observed for metal dispersions between 0.2 and 1 [47]. It must be emphasized that the final activity of MgF 2 -supported Pd catalysts was higher than that previously obtained for Pd/Al 2 O 3 [48] and active carbon-supported palladium [49,50] catalysts. This effect would speak for the beneficial effect of support acidity in electronic modification of palladium surface (creation of Pd σ+ species).
Molecules 2016, 21, 1620 6 of 16 dispersions between 0.2 and 1 [47]. It must be emphasized that the final activity of MgF2-supported Pd catalysts was higher than that previously obtained for Pd/Al2O3 [48] and active carbon-supported palladium [49,50] catalysts. This effect would speak for the beneficial effect of support acidity in electronic modification of palladium surface (creation of Pd σ+ species). On the other hand, very different palladium dispersions in the sol-gel series prevent any speculation about a possible role of support in shaping the catalytic behavior. Therefore, it is concluded that palladium particle size of palladium, which is commonly considered as the catalytically active phase, must be a dominant factor. Thus, to investigate a possible role of support one needs to test palladium catalysts characterized by a similar metal dispersion. As mentioned earlier preparation of the carbonate series of 1 wt % Pd/MgO-MgF2 catalysts, gave the catalysts characterized by similar metal particle sizes (~4 nm) ( Table 2). Now the MgF2-supported catalyst showed a poorer performance than its counterparts characterized by 3 wt % metal loading ( Figure 3). Again, such effect would be ascribed to the metal particle size effect. Unexpectedly, the activity time on stream behavior of 1 wt % Pd/MgO was similar to that shown by 1 wt % Pd/MgF2, although larger differences in product selectivity were found (next subsection). On the other hand, the MgO-MgF2 supported catalyst showed an interesting behavior: For a longer time on stream the overall activity was higher and more stable than that for Pd/MgO and Pd/MgF2. This effect will be discussed in the last part of discussion.  On the other hand, very different palladium dispersions in the sol-gel series prevent any speculation about a possible role of support in shaping the catalytic behavior. Therefore, it is concluded that palladium particle size of palladium, which is commonly considered as the catalytically active phase, must be a dominant factor. Thus, to investigate a possible role of support one needs to test palladium catalysts characterized by a similar metal dispersion. As mentioned earlier preparation of the carbonate series of 1 wt % Pd/MgO-MgF 2 catalysts, gave the catalysts characterized by similar metal particle sizes (~4 nm) ( Table 2). Now the MgF 2 -supported catalyst showed a poorer performance than its counterparts characterized by 3 wt % metal loading ( Figure 3). Again, such effect would be ascribed to the metal particle size effect. Unexpectedly, the activity time on stream behavior of 1 wt % Pd/MgO was similar to that shown by 1 wt % Pd/MgF 2 , although larger differences in product selectivity were found (next subsection). On the other hand, the MgO-MgF 2 supported catalyst showed an interesting behavior: For a longer time on stream the overall activity was higher and more stable than that for Pd/MgO and Pd/MgF 2 . This effect will be discussed in the last part of discussion.
Molecules 2016, 21, 1620 6 of 16 dispersions between 0.2 and 1 [47]. It must be emphasized that the final activity of MgF2-supported Pd catalysts was higher than that previously obtained for Pd/Al2O3 [48] and active carbon-supported palladium [49,50] catalysts. This effect would speak for the beneficial effect of support acidity in electronic modification of palladium surface (creation of Pd σ+ species). On the other hand, very different palladium dispersions in the sol-gel series prevent any speculation about a possible role of support in shaping the catalytic behavior. Therefore, it is concluded that palladium particle size of palladium, which is commonly considered as the catalytically active phase, must be a dominant factor. Thus, to investigate a possible role of support one needs to test palladium catalysts characterized by a similar metal dispersion. As mentioned earlier preparation of the carbonate series of 1 wt % Pd/MgO-MgF2 catalysts, gave the catalysts characterized by similar metal particle sizes (~4 nm) ( Table 2). Now the MgF2-supported catalyst showed a poorer performance than its counterparts characterized by 3 wt % metal loading ( Figure 3). Again, such effect would be ascribed to the metal particle size effect. Unexpectedly, the activity time on stream behavior of 1 wt % Pd/MgO was similar to that shown by 1 wt % Pd/MgF2, although larger differences in product selectivity were found (next subsection). On the other hand, the MgO-MgF2 supported catalyst showed an interesting behavior: For a longer time on stream the overall activity was higher and more stable than that for Pd/MgO and Pd/MgF2. This effect will be discussed in the last part of discussion.

Product Selectivities and Activation Energies
We consider that chlorine-free compounds (i.e., hydrocarbons) are the desired products of CCl 4 hydrodechlorination because chloroform, a previously desired product, is classified as an extremely hazardous substance. In this respect all tested here MgF 2 -supported palladium catalysts showed the best results, as indicated in Table 3, where one column aggregates C 2 -C 5 hydrocarbon products. A higher activation energy for MgO-supported palladium would suggest that the removal of surface chlorine (as HCl) is more difficult than for the MgF 2 -supported metal. This comparison clearly confirms the suitability of MgF 2 -supported metal catalysts for HdCl reactions [14,20,21,[24][25][26].

Working State of MgO-MgF 2 Supported Palladium Catalyst
The used samples of catalysts were investigated by XRD, temperature programmed hydrogenation of post-reaction deposits and IR spectroscopy of adsorbed pyridine (on selected samples). The results shown in Figures 4-8 and Table 4 allow to make cautious speculations about the working state of HdCl catalysts. X-ray results (Figures 4 and 5) indicate that palladium, when it is detectable in the diffractograms, is largely transformed into a PdC x phase. Such transformation is often observed for HdCl reactions carried out over Pd catalysts [49,51] and here confirmed for MgF 2 -supported Pd catalyst used in the CCl 4 reaction. This phenomenon may decide about the time scale behavior of product selectivity pattern because at an initial period of reaction palladium acts as carbon sink decreasing the selectivity to methane [47,51]. Figures 4 and 5 showed big changes for the magnesium-containing support component. Direct transformation of MgO into MgCl 2 is also not unexpected, as a similar effect was observed for Pt/MgO catalyst by Choi et al. [20]. However, a closer inspection of diffractograms allows estimation of the extent of transformation and, hopefully, the state of crystallinity of formed MgCl 2 . Figure 4 shows that magnesia in 3 wt % Pd/MgO catalyst is only partly transformed into MgCl 2 . In MgO-MgF 2 supported 3% Pd loaded catalysts the extent of transformation appears more marked. Figure 5 shows analogous data for the carbonate series of Pd catalysts. Here the degree of MgO → MgCl 2 transformation is still more pronounced. For the 1 wt % loaded catalysts the presence of MgCl 2 and MgO phases is only slightly marked (for 1 wt % Pd/MgO) or even completely invisible for 1 wt % Pd/37MgO-63MgF 2 catalyst. These observations along with much broader reflections from the MgCl 2 in the 3 wt % Pd/MgO catalyst that the formed MgCl 2 phase has a tendency to spread out over the catalyst surface.
Molecules 2016, 21, 1620 9 of 16 catalyst. These observations along with much broader reflections from the MgCl2 in the 3 wt % Pd/MgO catalyst that the formed MgCl2 phase has a tendency to spread out over the catalyst surface.  Temperature programed hydrogenation of post-reaction deposits presented in Figures 6 and 7 shows that all tested catalysts possess different amounts of carbon and chlorine. MgF2-supported catalysts are relatively deprived of chlorine: only some small amount of HCl is liberated at low temperature (≤200 °C), suggesting reduction of chloride associated with the palladium. A nonappearance of high temperature HCl evolution (observed for MgO-containing catalysts) indicates lack of chlorine on the surface of MgF2 (Figure 7). This, in turn, suggests no changes in the acidity of this catalyst which would be associated with chlorine. However, CH4 evolution from this catalyst was quite pronounced ( Figure 6) but still not as high as for other catalysts, which we believe is a primary reason of catalyst deactivation in HdCl reactions. Interestingly, Figure 7 shows no chlorine liberation for Pd/MgO at a low temperature region. We believe that it is not due to a lack of chlorine on palladium surface but rather because the surface of palladium is effectively blocked by an impermeable coating by MgCl2, [52] or simply closed in a porous structure by MgCl2 "stoppers". As it was suggested earlier, a decreased crystallinity of MgCl2 in this catalyst allows assumption that MgCl2 is either spread out over the support or a topochemical reaction MgO-MgCl2 [52] would lead to dramatic changes in the support texture, with possible burial of Pd particles. The MgO-MgF2 supported catalysts showed   Temperature programed hydrogenation of post-reaction deposits presented in Figures 6 and 7 shows that all tested catalysts possess different amounts of carbon and chlorine. MgF2-supported catalysts are relatively deprived of chlorine: only some small amount of HCl is liberated at low temperature (≤200 °C), suggesting reduction of chloride associated with the palladium. A nonappearance of high temperature HCl evolution (observed for MgO-containing catalysts) indicates lack of chlorine on the surface of MgF2 (Figure 7). This, in turn, suggests no changes in the acidity of this catalyst which would be associated with chlorine. However, CH4 evolution from this catalyst was quite pronounced ( Figure 6) but still not as high as for other catalysts, which we believe is a primary reason of catalyst deactivation in HdCl reactions. Interestingly, Figure 7 shows no chlorine liberation for Pd/MgO at a low temperature region. We believe that it is not due to a lack of chlorine on palladium surface but rather because the surface of palladium is effectively blocked by an impermeable coating by MgCl2, [52] or simply closed in a porous structure by MgCl2 "stoppers". As it was suggested earlier, a decreased crystallinity of MgCl2 in this catalyst allows assumption that MgCl2 is either spread out over the support or a topochemical reaction MgO-MgCl2 [52] would lead to dramatic changes in the support texture, with possible burial of Pd particles. The MgO-MgF2 supported catalysts showed Temperature programed hydrogenation of post-reaction deposits presented in Figures 6 and 7 shows that all tested catalysts possess different amounts of carbon and chlorine. MgF 2 -supported catalysts are relatively deprived of chlorine: only some small amount of HCl is liberated at low temperature (≤200 • C), suggesting reduction of chloride associated with the palladium. A nonappearance of high temperature HCl evolution (observed for MgO-containing catalysts) indicates lack of chlorine on the surface of MgF 2 (Figure 7). This, in turn, suggests no changes in the acidity of this catalyst which would be associated with chlorine. However, CH 4 evolution from this catalyst was quite pronounced (Figure 6) but still not as high as for other catalysts, which we believe is a primary reason of catalyst deactivation in HdCl reactions. Interestingly, Figure 7 shows no chlorine liberation for Pd/MgO at a low temperature region. We believe that it is not due to a lack of chlorine on palladium surface but rather because the surface of palladium is effectively blocked by an impermeable coating by MgCl 2 , [52] or simply closed in a porous structure by MgCl 2 "stoppers". As it was suggested earlier, a decreased crystallinity of MgCl 2 in this catalyst allows assumption that MgCl 2 is either spread out over the support or a topochemical reaction MgO-MgCl 2 [52] would lead to dramatic changes in the support texture, with possible burial of Pd particles. The MgO-MgF 2 supported catalysts showed both the low and high temperature HCl evolution, evidencing that both MgCl 2 spreading as well as textural changes would be less dramatic for mixed MgO-MgF 2 systems, where a part of palladium species is deposited on MgF 2 . It looks like the very huge HCl evolution from Pd/MgO occurred at very high temperature largely results from reduction of MgCl 2 .
both the low and high temperature HCl evolution, evidencing that both MgCl2 spreading as well as textural changes would be less dramatic for mixed MgO-MgF2 systems, where a part of palladium species is deposited on MgF2. It looks like the very huge HCl evolution from Pd/MgO occurred at very high temperature largely results from reduction of MgCl2.  The results of IR spectra of adsorbed pyridine (Figure 8a and b and Table 4) allow to confirm the presence of Lewis acid centers on the surface of all tested catalysts. The analysis of population and strength distribution of these centers was carried out by considering the absorbance of 1446 cm −1 band, associated with pyridine coordinated to the Lewis acid center. The largest amounts of pyridine left both the low and high temperature HCl evolution, evidencing that both MgCl2 spreading as well as textural changes would be less dramatic for mixed MgO-MgF2 systems, where a part of palladium species is deposited on MgF2. It looks like the very huge HCl evolution from Pd/MgO occurred at very high temperature largely results from reduction of MgCl2.  The results of IR spectra of adsorbed pyridine (Figure 8a and b and Table 4) allow to confirm the presence of Lewis acid centers on the surface of all tested catalysts. The analysis of population and strength distribution of these centers was carried out by considering the absorbance of 1446 cm −1 band, associated with pyridine coordinated to the Lewis acid center. The largest amounts of pyridine left The results of IR spectra of adsorbed pyridine (Figure 8a,b and Table 4) allow to confirm the presence of Lewis acid centers on the surface of all tested catalysts. The analysis of population and strength distribution of these centers was carried out by considering the absorbance of 1446 cm −1 band, associated with pyridine coordinated to the Lewis acid center. The largest amounts of pyridine left on the surface after desorption at 150 • C were found for used 1 wt % Pd/37MgO-67MgF 2 catalyst. The number of such centers was nearly three times higher than the respective value for the reduced catalyst. However, these centers are rather weak because after desorption at 200 • C only 28.5% of them still bind pyridine. On the other hand, at analogous conditions, as far as 66% of pyridine is left on the reduced catalyst. A quantitative parameter characterizing the strength of acid centers is the ratio of the 1446 cm −1 absorbance after pyridine desorption at 300 and 200 • C. Table 4 displays that the highest value of this parameter (0.64) showed the reduced 1 wt % Pd/37MgO-67MgF 2 catalyst, whereas after HdCl it decreased to 0.26. A similar value was found for 1 wt % Pd/MgO after reaction (0.25). Therefore, it appears that during the HdCl reaction the strongest acid centers are neutralized with simultaneous formation of weak acid centers. It would be worthwile to investigate the acid strength evolution in the course of reaction and combine it with the catalytic performance (activity and selectivity). on the surface after desorption at 150 °C were found for used 1 wt % Pd/37MgO-67MgF2 catalyst.
The number of such centers was nearly three times higher than the respective value for the reduced catalyst. However, these centers are rather weak because after desorption at 200 °C only 28.5% of them still bind pyridine. On the other hand, at analogous conditions, as far as 66% of pyridine is left on the reduced catalyst. A quantitative parameter characterizing the strength of acid centers is the ratio of the 1446 cm -1 absorbance after pyridine desorption at 300 and 200 °C. Table 4 displays that the highest value of this parameter (0.64) showed the reduced 1 wt % Pd/37MgO-67MgF2 catalyst, whereas after HdCl it decreased to 0.26. A similar value was found for 1 wt % Pd/MgO after reaction (0.25). Therefore, it appears that during the HdCl reaction the strongest acid centers are neutralized with simultaneous formation of weak acid centers. It would be worthwile to investigate the acid strength evolution in the course of reaction and combine it with the catalytic performance (activity and selectivity). In summary, the role of support acidity is quite complex. On the one hand, a definite, although not very high Lewis acidity of MgF2 is beneficial for shaping high activity of palladium catalysts. The MgO-MgF2 support characterized by stronger Lewis acidity than MgF2 contributes to very good catalytic activity for a relatively long reaction period (~5 h) but subsequent neutralization of stronger acid centers (by coking) eliminates them from the catalytic amphitheater. On the other hand, the role of acidity evolution, which takes place when basic supports (like MgO) are chlorided during HdCl reactions, is difficult to assess because different events associated with distribution of chlorided support species, leading to partial or even full blocking of the surface of palladium, which plays the role of active phase in HdCl reactions. Therefore, a unique disposal of straightforward tuneability of acido-basic properties of MgO-MgF2 of varied composition is weakened by the interaction with chloride species. It must recalled that these sites, tuned over a wide range of MgO-MgF2 composition, gave an unproblematic access to optimized catalytic activity and selectivity of these phases for the Michael addition of 2-methylcyclohexane-1,3-dione to methyl vinyl ketone [31]. By increasing the In summary, the role of support acidity is quite complex. On the one hand, a definite, although not very high Lewis acidity of MgF 2 is beneficial for shaping high activity of palladium catalysts. The MgO-MgF 2 support characterized by stronger Lewis acidity than MgF 2 contributes to very good catalytic activity for a relatively long reaction period (~5 h) but subsequent neutralization of stronger acid centers (by coking) eliminates them from the catalytic amphitheater. On the other hand, the role of acidity evolution, which takes place when basic supports (like MgO) are chlorided during HdCl reactions, is difficult to assess because different events associated with distribution of chlorided support species, leading to partial or even full blocking of the surface of palladium, which plays the role of active phase in HdCl reactions. Therefore, a unique disposal of straightforward tuneability of acido-basic properties of MgO-MgF 2 of varied composition is weakened by the interaction with chloride species. It must recalled that these sites, tuned over a wide range of MgO-MgF 2 composition, gave an unproblematic access to optimized catalytic activity and selectivity of these phases for the Michael addition of 2-methylcyclohexane-1,3-dione to methyl vinyl ketone [31]. By increasing the fluorine content of the magnesium oxide fluoride, the Lewis acidity increases whereas the basicity decreases [30]. This correlation is also compatible with recent results on toluene hydrogenation over Ir/MgO-MgF 2 catalysts [53], where the best iridium catalysts were supported on MgF 2 -rich carriers characterized by a relatively low surface basicity.

Preparation of MgO-MgF 2 Supports
Two series of MgO-MgF 2 supports were prepared. The first one was synthesized by the sol-gel method starting from magnesium methoxide treated with an aqueous solution of hydrofluoric acid, as described in [32]. The amount of HF solution was adjusted to obtain 40 and 70 mol % MgF 2 in the samples. The resulting dense gels of MgO-MgF 2 were subjected to ageing for 40 h at RT, and then to drying at 80 • C for 3 h. The dried samples of this "sol-gel" series were calcined for 4 h at 400 • C and sieved (selected fraction 0.25-0.5 mm). The MgF 2 support was obtained by the sol-gel method from Mg(OCH 3 ) 2 and anhydrous HF (48.8% HF in methanol, Sigma-Aldrich, Polska, Poznań, Poland) in a way analogous to the above described synthesis for MgO-MgF 2 , but under rigorously anhydrous conditions. MgO was obtained by the sol-gel method by hydrolysis of magnesium methoxide (120 cm 3 of 0.5 M solution) in water and treated similarly to MgO-MgF 2 . Another series of MgO-MgF 2 supports of different MgF 2 contents (0, 60% and 100%) were obtained in the reaction of basic magnesium carbonate (4MgCO 3 ·Mg(OH) 2 ·5H 2 O) powder with controlled amounts of 40 wt % aqueous solution of HF, as described in [33]. The resulting dense gels of this "carbonate" series were aged for 40 h at room temperature under stirring, followed by drying at 80 • C for 24 h and calcining under air flow at 500 • C for 4 h. The MgO support was obtained by decomposition of 4MgCO 3 ·Mg(OH) 2 ·5H 2 O at 500 • C for 4 h.

Preparation of MgO-MgF 2 Supported Palladium Catalysts
The 3 wt % Pd/MgO-MgF 2 catalysts were prepared by impregnation of both series of MgO-MgF 2 supports with an acetone solution of palladium acetate using the incipient wetness technique. The 1 wt % Pd-loaded catalysts were prepared by wet impregnation of the "carbonate" supports with a methanol solution of palladium acetate. After impregnation and drying (at 80 • C for 20 h), the resulting solids were transferred to glass-stoppered bottles and kept in a desiccator. Prior to chemisorption and reaction studies, all catalysts were reduced in flowing 20% H 2 /Ar (50 cm 3 /min), ramping the temperature from 20 to 380 • C (at 8 • C/min), and kept at 380 • C for 2 h.

Characterization of Catalysts
The catalysts were characterized by H 2 and CO chemisorption, XRD, temperature-programmed methods and IR spectroscopy of adsorbed pyridine. Irreversible uptake of hydrogen was measured in a pulse method system at 70 • C, to avoid the formation of β-PdH. CO adsorption was measured at 35 • C in a static system, using a double isotherm method (ASAP 2020 from Micromeritics). H ad /Pd and CO ad /Pd ratios were taken as measures of metal dispersion. Prior to characterization, the catalysts were pretreated in the same way as before kinetic experiments (next subsection), i.e., with the final reduction at 380 • C for 2 h. Reduced and post-reaction catalysts were investigated by X-ray diffractometry (PANalytical Empyrean, Almelo, The Netherlands, with Ni-filtered CuK α radiation and sample spinner).
After H 2 chemisorption, the samples were cooled to room temperature and subjected to a temperature programmed study in 10% H 2 /Ar flow, ramping the temperature from 20 to 150 • C, at 8 • C/min. Since the samples had already been reduced, the aim of such experiments was to monitor hydrogen evolution during decomposition of Pd hydride, in the temperature programmed hydride decomposition (TPHD). For details, see [34]. Post-reaction catalyst samples were investigated by temperature programmed hydrogenation (TPH-MS). TPH-MS runs were carried out in a flowing