Magnesium- and iron-doped chromium fluoride/ hydroxyfluoride: synthesis, characterization and catalytic activity B. Adamczyk, A. Hess and E. Kemnitz" Humboldt- Universitat zu Berlin, Institut fur Chemie, Hessische Sty. 1-2, D-10115 Berlin, Germany The calcination of a-CrF, -3H20 results in the formation of a chromium hydroxyfluoride with the pyrochlore structure. The stepwise replacement of chromium by iron and magnesium leads to considerable alterations in the structure and the surface properties of the calcination products, accompanied by significant changes in the catalytic activity. The dismutation of dichlorodifluoromethane and the dehydrochlorination of l,l,l-trichloroethane act as a probe reactions for Lewis acid sites. The syntheses of the catalysts were carried out by coprecipitation of mixed metal fluoride trihydrates and subsequent calcination procedures.The stepwise replacement with iron leads to a rebuilding of the lattice from the cubic pyrochlore structure of CrF3 -,(OH), into the pseudo-hexagonal tungsten bronze (HTB) structure of P-FeF,. The maximum catalytic activity towards CH,CCl, dehydrochlorination was obtained for the 65% iron sample, which is accompanied by a maximum BET surface area and a maximal number of Lewis acid sites. CrF, -,(OH), exhibits a dramatic loss of catalytic activity as well as BET surface area. Possible explanations are given by comparing the pseudo-HTB structure with the cubic pyrochlore structure with regard to the accessibility of the Lewis acid metal cations. Upon substitution of chromium by magnesium we obtained a maximum Lewis acidity for samples with 65-92% magnesium leading to a corresponding maximum in catalytic activity.The Br~nsted acidity of both systems is predominantly weak. Bulk and surface hydroxy groups are distinguished. Various fluorinated aliphatics are produced by heterogeneously catalysed fluorination of the appropriate chloro compounds with HF [eqn. (l)]. Oxides, oxyfluorides and fluorides of Al, Fe and Cr have been reported as suitable fluorination catalysts. They also catalyse isomerization reactions [eqn. (2)] and dismutation reactions [eqn. (3)]. R-C1+ HF-+R-F +HCl (1) CHF, -CHF, +CF3 -CH2F 2 CC12F2 +CCl,F +CClF, (3) Previous studies have shown that catalyst activation is neces- sary when starting from an oxide During the activation period a chemical reaction between the solid surface and the haloalkane and/or HF takes place. In the case of y-Al,03, phases are formed which are similar to P-AlF, offering a particular catalytic For this reason the catalytic behaviour of p-metal fluorides was examined.First, it was found that active P-AlF, and P-CrF, cannot be synthesized via the same route. A further important factor is that the irreversible phase transformations into the stable a-phases must be avoided since the a-fluorides are catalytically less active or inactive. As is widely reported for oxides, partial substitution of the metal cations affects the catalytic In the case of partially replaced fluorides (Al/Mg and Al/Cr) the following results were obtained.A fluoride with a Al/Cr molar ratio of 1 gave the highest conversion with regard to CCl,F, dismu-tation, which can obviously be attributed to the significant enlargement of the specific surface area and the improved accessibility of the Lewis acid sites. Alterations in the strength of the sites were not observed. In contrast, the Mg-doped aluminium fluorides offered a decrease in Lewis acid strength with increasing Mg content, although the specific surface area had increased. Consequently, the catalytic activity was diminished with increasing Mg con- tent. MgF, was catalytically inactive. There is no uniform route to the synthesis of the p-metal fluorides. While P-AlF, can be obtained by thermal degradation of a-AlF3.3H,O under a self-produced atmosphere,8 P-CrF, is not formed via this route.Instead, a chromium hydroxyfluoride having a pyrochlore structure was formed under these conditions. This product is catalytically inactive, whereas P-CrF, synthesized by thermal decomposition of (NH4)3CrF6 exhibits catalytic activity with regard to CCl,F, dismutation. The different catalytic behaviour was recently discussed on the basis of chemical and structural differences between the hexagonal tungsten bronze (HTB) p-metal fluorides and the pyrochlore hydroxyfluorides.' The present work deals with the systems Fe/Cr and Mg/Cr. Experimenta1 Preparation of catalysts A defined mass of Fe(NO,), 9H20 [or Mg( NO,), * 6H20] and/or Cr(N03), 9H20 was dissolved in ethanol.The mixture was added dropwise to a stirred 40 mass% hydrofluoric acid solution. The precipitate (a-M,M,F, 3H20) was separated, washed with small amounts of water and ethanol and dried in air. Then, the sample powder was covered by an aluminium foil to allow a self-produced atmosphere and was subsequently heated under an argon flow (2 K min-', up to 300 "C for mixed Fe/Cr and 410°C for Mg/Cr compounds). The appro- priate final temperature was maintained for 2 h. All samples were characterized by X-ray powder diffraction (XRD 7 Seiffert-FPM, Cu-Ka radiation). The determination of the fluoride contents was carried out according to ref. 10. The standard deviation was <3%. For analysis of the metal contents, 10-20 mg of each sample was melted with an excess of KNO, (500 mg) in a Pt beaker and dissolved in concentrated H3PO4 (Fe samples) or 50 mass% H,S04 (Mg samples), respectively. The concentration was determined by ICP AES (50 mg l-l, Unicam 701).The relative experimental error did not exceed 2%. The determination of the specific surface area was performed using an ASAP 2000 (Micrometrics) instrument based on the nitrogen BET method (max. exptl. error & l?h). FTIR photoacoustic spectroscopy of chemisorbed pyridine This method was used for determining the nature of the acid sites (Lewis/Br~nsted) by adsorption of pyridine on the catalyst J. Muter. Chem., 1996, 6(lo), 1731-1735 1731 surface and subsequent detection of the characteristic IR bands.This method is sensitive to both Lewis and Brarnsted acid sites, although the sensitivity decreases rapidly with decreasing surface areas. The whole procedure was carried out under the same conditions as described previously." Determination of catalytic activity Dismutation. The dismutation of CC12F2 [eqn. (3)] proceeds on Lewis acid site^."^,^ The products undergo further dismu- tation reactions. In detail, a constant mass of 0.6 g of the catalyst was used in a flow reactor (nickel tube). A residence time of 2 s was set up by adjusting the appropriate gas flow. First, the sample was calcined under a nitrogen flow at 400 "C for 1 h and then subsequently treated with a CC12Fz flow at 390°C. The composition of the gas phase at the exit of the reactor was determined by gas chromatography (column: Poraplot u; i.d.0.53 mm; length 25 m). The conversion of the starting substance CCl,F, is given with an absolute experimen- tal error which did not exceed +0.5% conversion. Dehydrochlorinationof l,l,l-trichloroethane. Since P-FeF, is unstable at the dismutation temperature (phase transformation into the a-modification) it was necessary to introduce an additional probe reaction that is sensitive to Lewis acid sites. The dehydrochlorination of l,l,l-trichloroethane [eqn. (4)] is reported to depend on Lewis acid centresl29l3 and has the advantage of running at lower temperatures. C1,C-CH3 +C12C=CH2 +HCl (4) Ballinger and co-w~rkers~~'~~ employed this method in order to investigate the dehydrochlorination activity of y-alumina above 130°C.Of all the samples studied only P-CrF, as a standard material [formed by thermal decomposition of (NH,),CrF,] exhibits remarkable activity for dehydrochlori- nation reactions below 130 "C. Therefore, for better compari- sons all measurements were carried out above 130°C for all samples. In detail, 200-300mg of the catalyst (pore diameter 0.2-0.5 mm) were fed into a flow reactor and calcined at 200 "C under dry nitrogen for 2 h. The residence time (t, =1.7 s) was adjusted by controlling the flow rate using a mass flow controller (MKS) according to the height of catalyst packing. Then, a constant stream of nitrogen was allowed to flow through a thermostatted bottle with l,l,l-trichloroethane.All tubings were also thermostatted to avoid any condensation. The reactor temperature was 130°C. The conversion was followed by on-line GC (Shimadzu GC14a; column: Porapack Q, 2m) equipped with a thermostatted gas sampling valve. The experimental error did not exceed +5%. Double-bond isomerisation of but-l-ene. This probe reaction depends on the presence of Brarnsted acid Under the selected conditions (low temperature) (E/Z)-but-2-ene is the only product. The apparatus is similar as described above. But-l-ene (1 ml min-l) was mixed with the adjusted nitrogen flow. The reaction was carried out at 100 "C. After 4 min on- stream a gaseous sample was injected into the GC (MGC 4000; column: Poraplot U; 30m).Results Fe/Cr samples X-Ray diffraction and wet analysis. With pure a-FeF, 3H20, the HTB P-FeF, is obtained after calcination at 300°C. Temperatures >400 "C lead to the irreversible formation of a-FeF,. In contrast, if a-CrF, 3H20is calcined under compar- able conditions, a pyrochlore chromium hydroxyfluoride phase is formed that is stable even above 400°C. Owing to the similarity of the ionic radii [Fe 0.65 A;Cr 0.62 A (ref. 17)] the formation of solid solutions should be expected. In fact, depending on the Fe concentration there are two ranges (see Fig. 1): a concentration range from 100 to 65% Fe with the HTB structure and a second range from 41 to 0% Fe with the pyrochlore structure. As can be concluded from Table 1 the HTB range consists of metal trifluorides, whereas the pyrochlore range comprises hydroxyfluorides.It can be seen from Fig. 1 that Fe and Cr replace each other within the appropriate structures although the shift of the lattice constant is of the same order of magnitude as the experimental error. Nevertheless, particularly in the case of the pyrochlore phases a shift towards higher angles (lower d values, lower lattice constants) is observed with increasing Cr content. Besides, there is a certain amount of X-ray amorphous species present, particularly at ca. 50% Fe. The higher the Cr content, the higher the tendency to hydrolysis as described by eqn. (5). a-CrF, ~3H20-,Cr(OH),F,-x+xHF+(3-x)H20 (5) With increasing insertion of Fe into the chromium hydroxyfluo- ride lattice, the amount of bulk OH groups diminishes.Surface acidity. Fig. 2 shows the FTIR photoacoustic spectra of pyridine chemisorbed on the solid surfaces. The band at approximately 1450 cm-' can be assigned to coordinatively bonded pyridine (Lewis acid sites), whereas the band at about 1493 cm-' represents both Lewis and Brarnsted acid sites." If Fig. 1 XRD patterns (offset) of a-CrF, .3H20 calcination products with stepwise replacement by Fe. Note the change from the pyrochlore structure of CrF,-,(OH), into the HTB J3-FeF, structure with increasing Fe content. Table 1 Atomic compositions and BET specific surface areas specific surface area/ sample compositiona m2 g-' Cr 100/1 CrFO 39(OH)261 9 Fe 16 CrO 84Fe0 16F2 5 (OH)O 5 11 CrO 77Fe0 23F1 84(OH)116 13.5 CrO 59Fe0 41F2 95(OH)005 17 2 CrO 35Fe0 6SF3 38 9 Fe 90 Cro 1Feo 9F3 34.5 Fe 100 FeF, 24 Cr lOO/II CrFO 41(OH)259 9.5 Mg 26 CrO 74Mg026F1 42(OH)132 15 2 Mg 46 CrO 54Mg0 46F1 86(OH)068 17 5 Mg 56 CrO 44Mg0 56F2 ,,(OH), 32 31.1 Mg 65 CrO 35Mg065F2 06(OH)0 27 49.4 Mg 86 CrO 1dMg08GF1 95(OH)0 19 77.8 Mg 92 CrO OsMgO 9zF1 97(OH)0 11 78 Mg 100 MgF2 23 2 ~~ ~ ~ ~ a Atomic composition determined by ICP AES and by See1 lo 1732 J.Muter. Chern., 1996, 6(lo), 1731-1735 Cr l00fl Fc I6 Fe23 1493 1452 wavenumber/cm -' Fig. 2 FTIR photoacoustic spectra (offset) of pyridine chemisorbed on Fe-replaced a-CrF, *3H,O calcination products (background correction, 80 mg catalyst mass, 30 p1 pyridine adsorption at 150 "C, flow system) Brsnsted acid sites are absent then the intensity of the 1493 cm-' vibration is about one third of that of the 1450 cm-' band.Changes in the strength of the sites would cause a wavenumber shift (compare Mg/Cr system, sample Mg100). Here, this is not observed. As can be seen from Fig. 2, the intensity of the 1450 cm-I band reaches its maximum at the Fe65 sample and it decreases with further increases of the Cr content. This graduation is also observed for the specific surface areas (Table 1). Furthermore, one can state that the comparatively low BET areas correspond to the pyrochlore samples (small number of Lewis acid sites), whereas the HTB structure enables improved access to the Lewis acid sites.An enhanced number of Lewis acid sites with increasing Cr content have been observed for the HTB phases. This can be explained by a steady loss of crystallinity, accompanied by an increase of BET surface area, which improves the accessibility of the Lewis acid sites (metal cations). The strength of the sites is unaffected. The Brsnsted acidity in the Fe/Cr system is altogether very weak; a further graduation shows that the acidities of the samples Fe41 and Fe65 slightly exceed those of the others, but they remain weak. Hence, it can be concluded that the content of bulk OH groups does not correlate with the obtained Brsnsted acidity. Therefore, we assume that the large number of OH functions in the pyrochlore chromium hydroxyfluoride CrlOO is a fixed part of the bulk and does not act as a Brsnsted acid.Catalytic activity. If P-FeF, is present in the system then a maximum temperature of about 350 "C must not be exceeded (formation of the cr-phase). Therefore, the dismutation could not be applied as a probe reaction for Lewis acid sites. Hence, the dehydrochlorination of l,l,l-trichloroethane was used, which operates at lower temperature (see Experimental). Fig. 3 illustrates the obtained absolute and specific conversions (related to the surface area). There are considerable differences in the conversion depending on the crystal structure. The graduation corresponds with the appropriate number of Lewis acid sites indicated by the intensity of the 1450cm-' band in Fig.2. HTB phases offer a higher activity for Lewis acid-catalysed 2 absoluteconversion Y BET nonnalircd conversion 80 Pyrochlon phases Fe (mol%) FeF, Fig. 3 Conversion of l,l,l-trichloroethane us. Fe content fc)r various Fe-replaced a-CrF, *3H20 calcination products (catalyst mass = 300 mg, residence time =1.7s, 130"C) reactions than the pyrochlore phases. According to Fig. 2 the Lewis acidity of the pyrochlore samples is weak. Thr, double- bond isomerization of but-1-ene reveals very low coilversions (<3%). This is in good agreement with the results obtained from pyridine adsorption results (Fig. 2) which found very weak Brsnsted acidities only. Mg/Cr samples X-Ray diffraction and wet analysis.The thermal decompo- sition of the Mg/Cr precipitates results predominantly in the formation of amorphous products and MgF,, whereas the intensities of the latter diminish with increasing Cr content (Fig. 4). Samples Mg26 and Cr100/II exhibit very small reflec- tions which can be assigned to the pyrochlore chromium hydroxyfluoride. The higher calcination temperature possibly leads to the reduced crystallinity compared to thr, sample Cr100/I in the Fe/Cr system. The difference in the ionic radii [Mg 0.72 A; C c 0.62 A (ref. 17)] would be large enough to cause a detectablt shift of the lattice constant. As can be seen in Fig. 4 there is no shift of reflections with increasing dopant concentration f>x either MgF, or chromium hydroxyfluoride.A mutual substitution within the lattices does not take place. Instead, the appropriate dopant causes decreases in crystallinity of the host lattices. According to Table 1, one can state that the OH'F ratio increases with rising Cr content. -_ \ 0 .-3-$ a P A A - /c Mg 86 Mg 65 Mg 56 VIC Cg!-.-d Mg46 -Mg26 /I.,.).,.,.,.,,,.,. 5 10 IS 20 15 30 35 40 45 {O . 5'5 Cr ioom.-$0 28ldegrees Fig. 4 XRD patterns (offset) of rx-CrF, *3H,O calcination products with stepwise replacement by Mg. Note the change from the pyrochlore structure of CrF,-,(OH), (weak reflections) into the MgF, structure with increasing Mg content. J. Mater. Chem., 1996, 6(lo), 1731-1735 1733 Surface acidity. Fig. 5 shows the FTIR photoacoustic spectra of pyridine chemisorbed on the solid surfaces.Starting from Cr100/II, the number of Lewis acid sites increases with increas- ing Mg content, reaches its maximum at sample Mg86 and decreases to a comparatively low value at Mg100. This is accompanied by the formation of a distinct shoulder at lower wavenumbers (Mg92) at the position where MglOO exhibits the corresponding band. This can be explained by an attenu- ation of Lewis acid strength due to the lower positive charge of the Mg2+cations. The maximum number of Lewis acid sites is in agreement with the maximum of BET surface area (Table 1). It is remark- able that only 8% Cr in an MgF, environment (Mg92) leads to a dramatic change in the acidic properties (Fig.5). The Brernsted acidity behaves as follows. Samples Mg92, Mg86 and Mg65 exhibit a weak band at approximately 1540 cm-' that can be assigned to hydrogen-bonded pyridine (weak Brernsted sites). As can be concluded from the low intensity (Fig. 5), a catalytic activity based on Brmsted acid sites is not expected. Catalytic activity. Fig. 6 presents the CCl,F, conversion (Lewis acid probe reaction) as a function of the Mg content. The graduation corresponds to the number of Lewis acid sites derived from Fig. 5 with the exception of MgF,. In that case, the conversion is almost zero although there are a certain number of Lewis acid sites. The reason is that MgF, possesses significantly weaker Lewis acid sites, illustrated by the shift towards lower wavenumbers in Fig, 6.Obviously, their strength is no longer sufficient to catalyse the dismutation. As can be concluded from the surface-normalized conver- sions (Fig. 6) the Mg dopant causes an enlargement of the surface area whereas the specific conversion is not increased with increasing Mg content. Regarding the Brernsted probe reaction, the following results were obtained. The conversion of but-1-ene was generally very low, but samples Mg92, Mg86 and Mg65 were slightly more active (2-6%). This can be confirmed by the results of pyridine complexes, where very weak Brernsted acidity was found at these samples. Obviously, the bulk OH content (Table 1) does not correlate with the Brernsted acidity, otherwise it would increase with increasing Cr content.LPy+BPy LPy Mg 100 Mg 92 Mg86 Mg 65 Mg 56 Mg 46 Mg 26 Cr 100/11 *1.1.1.1.1.1.1 1540 1520 1500 1480 1460 1440 1420 Fig. 5 FTIR photoacoustic spectra (offset) of pyridine chemisorbed on Mg-replaced a-CrF, 3H20 calcination products (background correction, 80 mg catalyst mass, 30 pl pyridine adsorption at 150"C, flow system) 1734 J. Muter. Chem., 1996, 6(lo), 1731-1735 VI 70 1 absolute conversion=BETnormalizedconversion 4 9 60 50 h8 v C040.-r9 30 C8 20 10 0 Mg (mol%) Fig. 6 Conversion of CC12F2 us. Mg content for various Mg-replaced a-CrF, -3H20 calcination products (catalyst mass = 600 mg, residence time =2 s, 390 "C) Conclusions In the case of the Fe/Cr system significant differences in the catalytic activity can be attributed to the following factors.First, two different structures were obtained and secondly, different compositions were found. Metal trifluorides were formed by the HTB phases, whereas metal hydroxyfluorides predominated in the case of the pyrochlore structures. The latter exhibit a weak Lewis acidity, the HTB phases offer improved Lewis acidities which correspond to enhanced cata- lytic activities with regard to Lewis probe reactions. The catalytically less-active pyrochlore phases consist of M( F,OH), octahedra. Their special linkage results in the formation of hexagonal channels which pass along all six plane diagonals of the cubic unit cell.9 A cleavage of this low-density structure should cause a large number of Lewis sites to be attainable.According to our results this is not the case. We believe that the hydroxy groups which are present in all pyrochlore phases hinder the free access to the metal cations. The presence of hydroxy groups within the pyrochlore lattice enables the formation of hydrogen bridge bonds on the surface which is accompanied by a partial shielding of the Lewis acid sites. The OH groups do not act as Brmsted acid sites. HTB metal trifluorides which also contain hexagonal channels obvi- ously enable an enhanced access to the metal cations due to the absence of OH. Generally, the availability of the sites is improved if the crystallinity is lowered, e.g. when a certain amount of a dopant is introduced.This can also be monitored by increasing BET surface areas. Concerning the Mg/Cr system the catalytic activity was influenced mainly by the strength of the Lewis acid sites. This can be illustrated, e.g., by the samples MglOO (MgF,) and Mg92. MgF, possesses Lewis acid sites that are not strong enough to catalyse the Lewis probe reaction. In contrast, the introduction of 8 mol% Cr results in a significant enhancement of the Lewis acid strength, and the catalyst is now active. With further increases in the Cr content the specific conversion also rises until the pyrochlore lattice has been formed. The latter is again inactive for the reasons described earlier. Brernsted acidity, unsurprisingly, does not play an important role in systems containing Mg.We are grateful to the Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie for financial support. References 10 F. Seel, Angew. Chem., 1964,76,532. 11 A. Hess and E. Kemnitz, J. Catal., 1994,149,449. A. Hess and E. Kemnitz, J. Catal., 1994,148,270. 12 J. Thomson, G. Webb and J. M. Winfield, J. Chem. SOC., Chem. E. Kemnitz and A. Hess, J.Prakt. Chem., 1992,334, 591. Commun., 1991,323. L. Kolditz and G. Kauschka, 2.Anorg. Allg. Chem., 1977,434,41. 13 J. Thomson, G. Webb and J. M. Winfield, J. Mol. Catal., 1991, L. Kolditz, U. Calov, G. Kauschka and W. Schmidt, 2. Anorg. 68,347. 5 6 7 Allg. Chem., 1977,434, 55. L. E. Manzer, Eur. Pat., 331991,1989,to E. I. Du Pont de Nemours and Co. L. E. Manzer, US Pat., 4766259,1988,to E. I. Du Pont de Nemours and Co. S. Hirayama, PCT Int. Appl., WO 8910341, 1989, to Showa 14 15 16 17 T. H. Ballinger and J. T. Yates, Jr., J. Phys. Chem., 1992,%, 1417. T. H. Ballinger, R. S. Smith, S. D. Colson and J. T. Yates, Jr., Langmuir, 1992,8,2473. S. E. Tung and E. McIninch, J. Catal., 1964,3,229. A. F. Wells, Structural Inorganic Chemistry, Clarendon Press, Oxford, 1993. Denko K. K. 8 9 D. H. Menz and U. Bentrup, Z. Anorg. Allg. Chem., 1989,576,186. E. Kemnitz, A. Hess, G. Rother and S. Troyanov, J. Catal., 1996, Paper 6/03296F; Received 13th May, 1996 159, 332. J. Mater. Chern., 1996,6(lo), 1731-1735 1735