Hydrodechlorination of 1,1-dichlorotetrafluoroethane and dichlorodifluoromethane catalyzed by Pd on fluorinated aluminas: The role of support material

Article abstract of DOI:10.1006/jcat.1998.2342

The gas phase hydrodechlorination of CF₃CFCl₂ to CF₃CFH₂ and CF₂Cl₂ to CF₂H₂ catalyzed by Pd supported on Al₂O₃, a series of fluorinated Al₂O₃, and AlF₃ was investigated. A combination of reaction kinetics investigations and characterization by in situ FTIR spectroscopy has been performed. It has been found that for reactions involving CF₃CFCl₂, all catalysts exhibit a rapid and significant decrease in activity; however, little change in activity with time on stream occurs with CF₂Cl₂. FTIR investigations suggest the occurrence of direct reaction between the CFC and the support material, which results in the consumption of hydroxyl groups during the early stages of reaction. The effect of fluorination of support on catalytic behavior of Pd is discussed.

Full text of DOI:10.1006/jcat.1998.2342

Journal of Catalysis 182, 219–227 (1999)
Article ID jcat.1998.2342, available online at http://www.idealibrary.com on
Hydrodechlorination of 1,1-Dichlorotetrafluoroethane and
Dichlorodifluoromethane Catalyzed by Pd on Fluorinated
Aluminas: The Role of Support Material
Kintu Early, Vladimir I. Kovalchuk, Ferenc Lonyi, Subodh Deshmukh, and Julie L. d’Itri
Department of Chemical Engineering, University of Pittsburgh, Pittsburgh, Pennsylvania 15261
Received July 17, 1998; revised November 16, 1998; accepted November 16, 1998
Al2O3 suffered a 20–50% loss in activity during the same
10 h time period (9).
The gas phase hydrodechlorination of CF3CFCl2 to CF3CFH2
and CF2Cl2 to CF2H2 catalyzed by Pd supported on Al2O3, a se-
ries of fluorinated Al2O3, and AlF3 was investigated. A combination
of reaction kinetics investigations and characterization by in situ
FTIR spectroscopy has been performed. It has been found that for
reactions involving CF3CFCl2, all catalysts exhibit a rapid and sig-
nificant decrease in activity; however, little change in activity with
time on stream occurs with CF2Cl2. FTIR investigations suggest the
occurrence of direct reaction between the CFC and the support ma-
terial, which results in the consumption of hydroxyl groups during
the early stages of reaction. The effect of fluorination of support on
c
Clearly, understanding the molecular phenomena related
to interactions of the reactant molecule with the support
material is critical to improving performance. The work
described here considers the role of the support in the
hydrodechlorination reaction. A combination of reaction
studies and FTIR investigations have been conducted to
elucidate the extent to which the support participates in the
CF3CFCl2 and CF2Cl2 hydrodechlorination reactions. Com-
parisons of alumina, fluorinated aluminas, and aluminum
fluoride as supports have been conducted.
catalytic behavior of Pd is discussed.
1999 Academic Press
EXPERIMENTAL
INTRODUCTION
It is well established that many transition metals ex-
Catalyst Preparation
The initial step in preparing the catalysts involved the
hibit catalytic activity for the hydrogenolysis of C–Cl bonds synthesis of a series of fluorided aluminas (2, 5, 10, and
(1–3). Noble metals are particularly attractive because of 30% F by weight). This was accomplished by adding 20 g
their high activities. The hydrogenolysis of carbon–chlorine
-Al2O3 (Vista B) to a precalculated amount of 30% HF
bonds in chlorofluorocarbons is carried out primarily using solution (Mallinckrodt AR) diluted in 800 cm3 of distilled
a palladium catalyst because the selectivity toward the fully H2O. The slurry was mixed at room temperature for 4 h and
dechlorinated hydrofluorocarbon is high (4, 5). However, left overnight. The solid was removed by filtration and the
previous work has shown that the catalytic behavior of Pd liquid was analyzed for F content using a fluorine selective
depends not only on the metal, but also on the type of sup- electrode. No fluorine was detected in excess of the detec-
5
port (4, 6–8). For example, in the hydrodechlorination of tion limit (<10 M), indicating that HF reacted stoichio-
CF2Cl2 selectivities toward CF2H2 in excess of 80% were metrically with the oxide. Afterward, the fluorided supports
obtained using a Pd supported on AlF3 or Al2O3 at conver- were dried in air at 393 K. Then they were calcined in 10%
sions below 10% , whereas Pd supported on ZrO2 exhibits O2/He (30 ml/min) while the temperature was increased at
a much lower selectivity ( 31% ) at the same conversion 1 K/min from room temperature to 773 K, and then held
(4, 9).
at this temperature for 3 h. Subsequently, the material was
Catalytic activity is also influenced by the type of sup- cooled to room temperature and impregnated with Pd as
port material. A fivefold difference in steady-state activity described below. It is worth noting that despite the differ-
(307 h 1 vs 61 h 1) was reported for Pd supported on AlF3 ence in density of alumina and aluminum fluoride, no sig-
in comparison to Al2O3 (4); the metal dispersion was ap- nificant change in alumina volume was observed during flu-
proximately the same. Catalyst stability is also affected by orination. This is consistent with the previous research (10),
the type of support materials used for such dechlorination and it was rationalized therein in terms of fluoride ions oc-
reactions. Palladium supported on AlF3 did not deactivate cupying interparticle spaces in the expanding AlF3 lattice,
during CF2Cl2 hydrodechlorination, while Pd supported on resulting in the reducing of pore volume and surface area.
219
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

220
EARLY ET AL.
Catalysts containing 1 wt% Pd were prepared by impreg- Catalytic Experiments
nating the fluorinated Al2O3 samples, -Al2O3 (Vista-B),
The hydrodechlorination of CF3CFCl2 (courtesy of
and AlF3 xH2O (Aldrich) with a Pd(II)–acetylacetonate
(>99% , Strem Chemicals) solution. For each sample, 0.3 g
of Pd(II)–acetylacetonate was dissolved in 250 ml toluene
(>99.99% , J.T. Baker) and the resulting solution was added
to 10 g of desired support. The slurry was stirred at room
temperature for 1 h in a rotary evaporator and then the
solvent was removed at 338 K under continuous evacua-
tion (125 Torr) in approximately 4 h. The resulting material
was dried overnight in air at 383 K. Finally, the sample was
transferred to a 18 mm i.d. Pyrex tube, heated to 773 K
(1 K/min) in 10% O2/He flow (30 ml/min), and held under
these conditions for 8 h.
DuPont, >99% ) was conducted in an atmospheric flow-
through microreactor containing a quartz U-tube reactor
(i.d. = 10 mm; height of the catalyst bed 3 mm). The re-
action products were analyzed by gas chromatography us-
ing a 5% Fluorocol 143 HMW/60/80 Carbopack B column
(3 m, from Supelco) maintained at a constant temperature
of 353 K. When necessary, products were identified by a
GC/MS equipped with the same column as the GC.
To remove water and contaminants that may have accu-
mulated during storage of the catalysts, prior to reaction the
catalyst sample (50 mg) was heated from room temperature
to 773 K at 4 K/min in 10% O2/He flow (20 ml/min) and held
at 773 K for 30 min. The sample was cooled to room temper-
ature and then purged with 30 ml/min N2 (UHP, 99.999% ,
Liquid Carbonic) for 10 min. Then it was treated in H2
(30 ml/min) as the temperature was increased at 8 K/min to
573 K and held at this temperature for 1 h. The catalyst was
subsequently purged with flowing N2 (30 ml/min) at 573 K
for 10 min before the system was cooled to the reaction tem-
perature (473 K). The reaction was started by switching the
gas flow from N2 to the premixed reactant mixture con-
sisting of 6 ml/min H2, 2 ml/min CF3CFCl2, and 92 ml/min
He (UHP, 99.999% , Liquid Carbonic) as balance (total flow
Catalyst Characterization
Metal dispersions were determined by hydrogen chemi-
sorption at room temperature using the double isotherm
method (11). The catalyst sample (500 mg) was placed into
a quartz U-tube reactor (i.d. = 16 mm; height of the cata-
lyst bed 6 mm), heated to 773 K (4 K/min), and calcined
in a 10% O2/He flow (20 ml/min) for 30 min. The sample
5
was evacuated at 773 K to less than 10 Torr for 15 min
and cooled to room temperature under the same vacuum.
The sample was exposed to H2 (30 ml/min) at room temper-
ature. While maintaining the flow of H2, the temperature
was increased at 8 K/min to 573 K and held at this tem-
perature for 1 h. The sample was then evacuated at 573 K
for 1 h and cooled to room temperature at <10 5 Torr. The
mean particle size of Pd was calculated using the empirical
equation developed by Benson et al., d = 0.9/D, where D
is the dispersion of palladium and d is given in nanometers
(11). Catalyst surface areas were determined by the BET
method using adsorption isotherms of nitrogen at 78 K. Pre-
treatment consisted of an overnight evacuation at 373 K,
and did not include the oxidation and reduction steps. The
physical and chemical properties of the catalysts are sum-
marized in Table 1.
rate = 100 ml/min; H2/CF3CFCl2 molar ratio = 3.0). This re-
1
sulted in a space velocity of 100,000 h
.
For the reaction studies involving difluorodichlorome-
thane (CF2Cl2, courtesy of DuPont, >99% ), a lower space
velocity (10,000 h 1) was used in order to maintain a con-
version level comparable to the CF3CFCl2 reactions. A total
flow rate of 30–50 ml/min was used for the CF2Cl2 reaction
studies. Thus an increased quantity of catalyst compared
to that of CF3CFCl2 dechlorination reactions was charged
into the reactor to provide the necessary space velocity. All
other parameters were the same as for the CF3CFCl2 inves-
tigations.
Rates per gram of catalyst were calculated by applying
the differential reactor approximation in the form
Rate (mmol/g s) = F y/W,
TABLE 1
where F is the total molar flow rate, y is the mole fraction of
reactant converted, and W is the catalyst weight. Turnover
frequencies (TOF) were then calculated using the expres-
sion
Characteristics of the Supported 1 wt% Pd Catalysts
Support
surface
Pd
dispersion
(H/Pd)
Mean particle
size of Pd
(nm)a
Fluorine
Support
Al2O3
2% F–Al2O3
5% F–Al2O3
10% F–Al2O3
30% F–Al2O3
AlF3
content (% ) area (m2/g)
TOF (s 1) = R MWPd/M d,
0
2
5
10
30
n.a.
245
207
195
183
47
0.12
0.14
0.12
0.06
0.07
0.06
7.5
6.4
7.5
15.0
12.9
15.0
where R is the rate per gram of catalyst, MWPd is the molec-
ular weight of Pd, M is the metal loading percent, and d is
the particle dispersion expressed as a percentage.
In all experiments the conversion of reactants was main-
tained below 15% . It was demonstrated in a earlier work
(12) that there is a linear relationship between conversion
33
a d = 0.9/D (11), where D is Pd dispersion.

CFC DECHLORINATION CATALYZED BY SUPPORTED Pd
221
H2 into the cell while the catalyst pellet temperature was
maintained at 473 K. The reaction was run for the desired
length of time. To regenerate the catalyst the cell was evac-
uated to a pressure of <10 5 Torr at 473 K for 10 min and
the pellet was heated to 673 K (8 K/min) while maintain-
ing the same vacuum. Then, the used catalyst pellet was
reduced in H2 (30 ml/min) at 573 K following the same pro-
cedure as for the fresh catalyst. The irreversible changes in
the support and the metal after reaction were probed by
CO chemisorption by monitoring changes in intensity and
position of the characteristic CO vibrations.
TABLE 2
Initial and Steady-State Selectivities of the Supported 1 wt% Pd
Catalysts Studied in the Reactions of CF3CFCl2 Dechlorination
Selectivitya
Conver-
CF3CH3
CF3CFH2
CF3CFHCl
sion (% )
Support
Al2O3
2% F–Al2O3
5% F–Al2O3
10% F–Al2O3
30% F–Al2O3
AlF3
1 h
15 h
1 h 15 h
1 h
15 h
1 h
15 h
45
47
52
80
86
76
76
77
73
71
69
53
43
40
38
15
7
18
18
19
21
24
37
3
4
4
3
6
8
6
5
7
8
6
9
12.0
7.0
5.0
7.0
30.5
27.0
6.0
4.5
3.5
4.5
4.5
3.0
RESULTS
16
a The remaining products were CH4, C2H6, and C2H4.
Table 1 summarizes the characterization of the six cata-
lysts used for this investigation. While it must be pointed
out that these results are dependent upon the preparation
procedure, there were two trends that were apparent from
these analyses. First, as fluorine content increased during
the conversion of alumina (245 m2/g) to aluminum fluoride
(33 m2/g), the surface area decreased by approximately an
order of magnitude. Second, there were two distinct dis-
persion regimes. For catalysts containing up to 5% fluorine
the Pd dispersion was 12–14% , whereas for the high fluo-
rine content catalysts the dispersion was only 6–7% . This
resulted in corresponding regimes of Pd particle size (6.4–
7.5 nm for low F, and 12.9–15 nm for high F content cata-
lysts).
With respect to the reaction kinetics the hydrodechlori-
nation of 1,1-dichlorotetrafluoroethane (CF3CFCl2) cata-
lyzed by Pd supported on -Al2O3 containing 0 to 5% flu-
orine exhibited an initial deactivation period followed by a
gradual approach to steady state (Fig. 1). Catalyst deacti-
vation was most rapid during the first 5 h on stream, during
which up to 30–40% of the initial activity was lost. The rapid
and space velocity for CF3CFCl2 dechlorination over
Pd/Al2O3 at conversions between 0 and 20% , suggesting
the absence of transport limitation. For several catalysts,
however, the conversion during first few hours on stream
was around 30% (Table 2), decreasing eventually due to
deactivation. Of course, the error bar for TOFs is larger for
higher conversions.
Infrared Studies
Investigations were conducted using a Mattson Research
Series II FTIR spectrometer equipped with a MCT detec-
tor with a resolution of 4 cm 1. A self-supporting pellet
of 5–10 mg/cm2 “thickness” was pressed from the catalyst
sample and placed into a quartz IR cell (volume = 140 cm3).
The pellet temperature was measured at the outer wall of
the quartz cell, approximately 6 mm from pellet surface.
The catalyst pretreatment consisted of heating to 773 K
(4 K/min) in 10% O2/He flow (25 ml/min) and immediately
cooling to room temperature. Oxygen traces were removed
5
by evacuating the cell for 15 min (<10 Torr) before
exposing the catalyst to pure H2. The reduction was then
performed as described earlier. The reduction step was fol-
lowed by evacuation (<10 5 Torr) at 573 K for 1 h in order
to remove chemisorbed H2 from the sample.
For experiments involving CO as a probe molecule, after
the previously described pretreatment the pellet was ex-
posed to 10 Torr of CO (99.99% , Liquid Carbonic) at room
temperature for 5 min. The gas phase and weakly adsorbed
CO were removed by evacuation for 10 min (<10 5 Torr).
Spectra of the catalyst pellet were taken before and after
CO chemisorption. Thus, the intensity of the bands asso-
ciated with the chemisorbed CO was determined by sub-
tracting the spectrum of the pellet prior to exposure to CO
from that acquired after removing the gaseous and weakly
interacting CO.
FTIR was used to monitor changes in the catalyst sur-
face under in situ reaction conditions. The reaction was
conducted by introducing 10 Torr CF3CFCl2 and 10 Torr
FIG. 1. Activity vs time on stream in dechlorination of CF3CFCl2 over
1 wt% Pd supported on Al2O3 (1), 2% F–Al2O3 (2), and 5% F–Al2O3 (3)
at 473 K.

222
EARLY ET AL.
initiallymore active (more than three timeshigher than with
alumina). Even though the activity decreased markedly, the
long-term activity of the high fluorine content catalysts was
still 40% higher than that obtained with the 2–5% F con-
taining samples. It is worth noting that the higher fluorine
content catalystshave similar, but lower, Pd dispersion com-
pared to low fluorine containing samples (Table 1). As was
shown previously, an increase in particle size of alumina-
supported Pd from 11 to 53 nm results in an increase in the
1
hydrodechlorination TOF of CF3CFCl2 from 2.3 to 5.0 s
(12). A similar result, a twofold increase in TOF when metal
particle size in Pd/Al2O3 catalysts was increased from 1.5 to
8 nm, was observed for of CF2Cl2 dechlorination (9). How-
ever, there is a relatively small change in TOF when Pd
dispersion decreased from 12–14% to 6–7% (Figs. 1 and 2).
Such a small change does not allow us to address this issue
in the discussion below.
FIG. 2. Activity vs time on stream in dechlorination of CF3CFCl2
over 1 wt% Pd supported on 10% F–Al2O3 (1), 30% F–Al2O3 (2), and AlF3
(3) at 473 K.
Surprisingly different behavior was observed when
CF2Cl2 was the reactant, as shown in Fig. 3 for 1% Pd
supported on Al2O3, 30% F–Al2O3, and AlF3. More sta-
ble activity as a function of time on stream was observed
when CF2Cl2 was the reactant in comparison to CF3CFCl2
(Figs. 1 and 2). The catalyst deactivated less than 30%
during the first several hours on stream with CF2Cl2, (inde-
pendent of the support used) compared to 50% deactiva-
tion when using CF3CFCl2. It was also evident that under
similar conditions, the one carbon CFC (CF2Cl2) had an
initial activity of 30 to 80 times less than that of its two
carbon analog (CF3CFCl2). As mentioned in the Experi-
mental section, the space velocity was a factor of 10 lower
for the CF2Cl2 reaction, and thus the contact time was a fac-
tor of 10 greater. Considering the results on a TOF basis,
the alumina-supported catalyst had lower activity than the
30% F or AlF3. This trend was the same for both reactants.
Selectivity was also a function of time on stream.
Table 2 summarizes the initial versus steady state selectivity
deactivation was followed by a slower rate of deactivation,
approximately 1% per h for the duration of the experi-
ment. On a “per exposed metal atom” basis (Fig. 1), the
initial activity of the alumina-supported catalyst was nearly
1
double that of the 2 and 5% fluorine samples (720 h vs
360 h 1). At the end of each 20 h experiment, the activity
of the alumina-supported catalyst was still higher than that
of the other two catalysts.
The qualitative time-on-stream behavior for catalysts
with high fluorine contents (10% F–Al2O3, 30% F–Al2O3,
AlF3) was similar to that of the samples containing up to
5% F, namely an initial deactivation period followed by a
slow approach to steady state (Fig. 2). However, a number
of features were markedly different. The magnitude of de-
activation was higher for the 30% F and AlF3 catalysts, each
losing 80% of its initial activity. These catalysts were also
FIG. 3. Activity (A) and selectivity (B) vs time on stream in dechlorination of CF2Cl2 over 1% Pd supported on Al2O3 (1), 30% F–Al2O3 (2), and
AlF3 (3) at 473 K.

CFC DECHLORINATION CATALYZED BY SUPPORTED Pd
223
behavior for the hydrodechlorination of CF3CFCl2. Three
TABLE 3
major products were formed—CF3CFH2, CF3CH3, and
CF3CFHCl—comprising >90% of the products. The re-
maining products were methane, ethane, and ethylene. For
the low F content materials (Al2O3, Al2O3–F(2% ), and
Al2O3–F(5% )) the selectivity to CF3CFH2 increased from
45 to 75% in 15 h time on stream. In the same time
period the selectivity to CF3CH3 decreased from 40 to
20% . Surprisingly, the selectivity to CF3CFHCl remained
nearly constant at 5% .
Initial and Steady-State Selectivities of the Supported 1 wt% Pd
Catalysts Studied in the Reactions of CF2Cl2 Dechlorination
Selectivitya
Conver-
CF2H2
CH4
CFHCl
sion (% )
Support
Al2O3
30% F–Al2O3
AlF3
1 h
15 h
1 h
15 h
1 h 15 h
1 h
15 h
53
72
66
61
73
67
36
20
18
31
19
18
1
2
2
1
2
2
6.0
4.0
4.5
6.0
3.5
3.5
The high F content catalysts exhibited different behav-
ior (Table 2). For the Al2O3–F(10% ), the selectivity to
CF3CFH2 decreased from 80 to 71% , while the selectiv-
ity to CF3CH3 increased from 15 to 21% . There was also a
slight increase from 3 to 8% in CF3CFHCl formation. Sim-
ilar behavior was observed for the 30% fluorinated sample
and the aluminum fluoride; the selectivity to CF3CFH2 de-
creased from 86–69% and 76–53% , respectively. There was
also a corresponding increase in the production of CF3CH3
(7–24% for the Al2O3–F(30% ) and 16–37% for the alu-
minum fluoride). The selectivity to CF3CFHCl remained
constant for each catalyst, at 6 and 8% , respectively.
Figure 3B shows the selectivity behavior of 1% Pd sup-
ported on Al2O3, Al2O3–F(30% ), and AlF3 catalysts for
the CF2Cl2 dechlorination reaction as a function of time on
stream. The major products of this reaction were CF2H2
and CH4, accounting for >85% of the total products. Other
products observed were CF2HCl, C2H6, CH2FCl, CH3Cl,
CF3Cl, and CH3F, none of which comprised more than 3%
of the reaction products. Several important observations
are made from Fig. 3B. With the alumina support, the selec-
tivity to the desired product (CF2H2) increased 20% in the
first 5 h time on stream, which corresponded to a decrease in
the selectivity to the other products (CH4, CF2HCl, C2H6,
CH2FCl, CH3Cl, CF3Cl, and CH3F). However, there was
no significant change (<2% ) in the time-on-stream selec-
tivity with the 30% F–Al2O3 and aluminum fluoride. Also,
the selectivity toward the desired product was lower for the
alumina support ( 60% ) than for the fluorinated supports
(65–73% ). A complete summary of the time on stream se-
lectivity for the C1 and C2 reactants is given in Tables 2
and 3.
a Other products observed were CF2HCl, C2H6, CH2FCl, CH3Cl, CF3Cl,
and CH3F, none of which comprised more than 3% of the reaction prod-
ucts.
However, more detailed investigations are needed to an-
swer whether the electron state of Pd depends on the de-
gree of support fluorination, as the position of CO absorp-
tion band is determined by both electron state of adsorption
center and CO coverage.
The intensity of the bands associated with CO adsorbed
on the catalyst varied as a function of fluorination. The
single peak at 2100 cm 1, characteristic of CO linearly
bonded with Pd, decreased in intensity as a function of flu-
1
orine content. The peak and shoulder just below 2000 cm
have been previously assigned to CO bridge-bonded to
two Pd atoms (13). The most dramatic change in inten-
sity, however, was associated with CO adsorption onto the
support. Spectrum 1 in Fig. 4 shows the adsorption of CO
on Al2O3 in the absence of Pd, demonstrating that the
1
peaks below 1800 cm are associated with the alumina
support; most belong to vibrations of surface carbonates
(15). The peaks below 1800 cm 1 were most intense for the
Pd on Al2O3 (spectrum 2). The reason for their increased
Characterization of each catalyst was done by FTIR.
Figure 4 shows the FTIR spectra associated with CO ad-
sorbed on Pd supported on Al2O3, 2% F–Al2O3, 10% F–
Al2O3, and AlF3. The bands at wave numbers between 2200
1
and 1800 cm are characteristic of CO adsorbed on Pd
(13), while the bands in the 1200 to 1800 cm 1 range have
been previously assigned to CO interacting with the Al2O3
support (14). The position of each band did not shift as a
function of the degree of fluorination. This is a particularly
important observation with respect to the vibrations asso-
ciated with CO adsorbed on the metal (1800–2200 cm 1),
as previous investigations have suggested that a shift in the
FIG. 4. IR spectra of CO adsorbed on pure alumina (1) and 1% Pd
position is indicative of an electronic change of the Pd (15). supported on Al2O3 (2), 2% F–Al2O3 (3), 10% F–Al2O3 (4), and AlF3 (5).

224
EARLY ET AL.
of the Al2O3 spectrum from the fluorinated samples
(4 and 5) revealed that several vibrations were absent for
the fluorinated samples. In addition, the difference spectra
showed that the fluorination of the support resulted in the
1
formation of a new peak at 3710 cm
.
Carbon monoxide adsorption was also used to moni-
tor the changes in the catalyst surface after reaction with
CF3CFCl2 in the presence and absence of hydrogen. Sim-
ilar to the effect of prefluorination, the bands in the low-
frequency region (1800–1300 cm 1) were absent after in situ
reaction of CF3CFCl2 alone and in the presence of H2 at
200 C (Fig. 6A). For spectra 2 and 3 the sample pellet was
pretreated, as described in the Experimental section, prior
to CO adsorption. The Pd was also altered during the hy-
drodechlorination reaction. There was a decrease in the
1
intensity of the 2080 cm peak and a slight increase in
FIG. 5. IR spectra of the surface OH groups of 1% Pd supported on
the peak at 1920 cm 1, in both cases in comparison to the
fresh sample.
Al2O3 (1), 2% F–Al2O3 (2), and 10% F–Al2O3 (3). Spectra 4 and 5 repre-
sent the difference spectra obtained by subtraction of spectrum 1 from
spectra 2 and 3, respectively.
Exposure of the catalyst to CF3CFCl2 alone and in the
presence of H2 affected the concentration of the surface
hydroxyl groups (Fig. 6B). The difference spectra presented
in Fig. 6B show the loss of OH groups during reaction with
CF3CFCl2 alone and in the presence of H2. The behavior
was similar to the effect of prefluorinating the support.
intensity compared to pure support is not clear and needs
further investigation. However, the intensity of these peaks
dramatically decreased for the 2% F–Al2O3 catalyst as a re-
sult of the decrease in the surface concentration of oxygen
resulting from fluorination. At higher fluorine content the
peaks were absent (spectra 4–5).
DISCUSSION
Fluorination also influenced the hydroxyl region of the
FTIR spectra. Figure 5 shows the FTIR spectra of the OH
region for the Al2O3, 2% F–Al2O3 and 10% F–Al2O3 cata-
lysts. The characteristic peaks associated with hydroxyl
The corrosive and highly reactive nature of the halo-
genated species which are formed on the surface of the
catalyst or as products of the hydrodehalogenation reac-
tion contributes to the highly dynamic behavior observed
in these systems. The results of the present investigation
provide a framework in which to understand the interac-
tion of these halogenated species with the supported metal
1
groups on the catalyst surface are in the 3000 to 4000 cm
range (16). Several broad bands between 3680 and
3745 cm 1 were present. These bands have previously been
assigned to OH groups of varying acidity (16). Subtraction
FIG. 6. IR spectra of CO adsorbed on (A) and surface OH groups of (B) fresh 1% Pd/Al2O3 catalyst (1), regenerated 1% Pd/Al2O3 catalyst
following the 90 min exposure to CF3CFCl2 + H2 at 473 K, and the fresh 1% Pd/Al2O3 catalyst after 90 min exposure to CF3CFCl2 at 473 K. Curves 4
and 5 represent the difference spectra obtained by subtraction of spectrum 1 from spectra 2 and 3, respectively.

CFC DECHLORINATION CATALYZED BY SUPPORTED Pd
225
catalyst. Two primary facets of the chemistry are worth Support Acidity
considering: the influence of halogenated species on the
supported metal catalysts and the effect of these changes
on the overall catalytic performance.
An understanding of the factors which affect support
acidity is always important because of its impact on side
reactions. Several authors (21–26) have concluded that it is
possible to create, by fluorination, both strong Brønsted
acid sites and strong Lewis acid sites on the surface of
-alumina. The FTIR results highlight the dynamic be-
Fluorination Mechanism
It is well established that alumina is fluorinated by a vari- havior of surface hydroxyl groups during fluorination—one
ety of reactants. Typical reactants include gaseous or aque- group forms while the other is consumed (Figs. 4 and 5). It
ous HF; ammonium bifluoride in either the gas or the aque- is generally accepted that, after fluorination, the less acidic
ousphase;and fluorocarbonssuch asHCF3, CF4, and C6H5F hydroxyl groups are removed while hydroxyls of strong
(10, 17). The reactions of alumina with the HF and NH4HF2 acidity are created (25). The possible production of strong
are generally carried out at temperatures near 600 C (to Brønsted acid sites is most plausible for the AlF3 used in
produce anhydrous AlF3 in the case of gas phase reactions). this investigation, given the conditions of pretreatment. As
The reactivity of organic fluorine containing compounds mentioned above, calcination of hydrated AlF3 at tempera-
varies significantly. A temperature range of 400–600 C is tures above 300 C favors hydrolysis of Al–F bonds. Under
usually used for aluminum oxide fluorination with organic such conditions the resulting concentration of Al2O3 is in
fluorocarbons. However, as was shown by IR spectroscopy, excess of 8% (20).
CHF3 interacts with the surface of - and -Al2O3, result-
The effect of fluorine atoms in the vicinity of a hydroxyl
ing in CO formation even at room temperature and pres- group will increase the acidity of the support. Fluorination
sures of only several Torr (18). Thus, it appears reasonable of -Al2O3 usually results in an increase in the total num-
to conclude that in the present system the CF3CFCl2 re- ber of acid sites and the acid strengths for both Brønsted
actant modifies the alumina, as supported by the results and Lewis sites (27). Infrared studies of CO2 and CO ad-
presented in Fig. 6B. In the presence of hydrogen this pro- sorption show that fluoride strengthens certain Lewis acid
cess is even accelerated because both HF and HCl formed sites ( -sites) of alumina (17). Due to the inductive effect
during dechlorination reaction readily react with alumina. of fluoride, the partial replacement of O and OH by F re-
The support of the Pd/Al2O3 catalyst under conditions of sults in an increase in the acidity of fluorinated aluminas.
CF2Cl2 hydrodechlorination (200 C) isreported to turn into However, even pure aluminum fluoride itself, without any
a mixture of aluminum fluoride and aluminum oxifluoride trace alumina, can possess strong Lewis acidity (28).
(4, 9).
The results of support fluorination via exposure to HF
bring up several points that are worthy of discussion. When
Catalytic Performance
aqueous HF is used, the surface areas of the fluorinated
One trend that has been established in the present inves-
samples are relatively low (Table 1), as much as five times tigation is that both catalytic activity and selectivity change
lower than the parent Al2O3 in the case of30% F–Al2O3 (the as the degree of halogenation of the support changes. The
surface area of aluminas fluorinated with CHF3 decrease activity for all of the catalysts which were tested (1% Pd
by less than a factor of two for samples with a compara- deposited on Al2O3; AlF3; and aluminas containing 2, 5, 10,
ble content of fluorine) (10). The reason for this difference and 30 wt% of fluorine) decreased during the first 5–10 h
becomes evident from consideration of the alumina fluo- time on stream (Figs. 1 and 2). There are several possible
rination conditions. An acid will catalyze the hydration of reasons for this deactivation: deposition of carbonaceous
aluminum oxide, resulting in its partial dissolution followed species, deposition of halogens, consumption of support
by reprecipitation. Consequently, this results in a dramatic oxygen atoms which results in a decreased rate of reaction
change of the initial Al2O3 structure and a significant de- with the support, and loss in Pd dispersion during runs. The
crease in surface area, even if there is a low degree of flu- last reason is unlikely, because before catalytic test all cata-
orination. Indeed, aqueous HF-prepared AlF3 samples do lysts were presintered at 500 C for 8 h in flowing oxygen.
not always exhibit a strong correlation between AlF3 con-
The FTIR data clearly show direct reaction with the sup-
tent and surface area (10). The AlF3 used in the present port. However, taking into account that fluorocarbons show
investigation had a surface area of 33 m2/g. This is substan- significant rate of interaction with bulk alumina at tem-
tially higher than previously reported for pure rhombohe- peratures above 400 C and typical reaction temperatures
dral AlF3 (3.3 m2/g) (10). This suggests that residual Al2O3 are 100–200 C lower (10), it is reasonable to suggest for
is present, which leads to the higher surface area (19). It is the present investigation that dichlorotetrafluoroethane re-
reasonable to conclude that the aluminum oxide in AlF3 is acts with the alumina surface much faster than with the
formed by the water of crystallization, which is capable of bulk. Moreover, given that the types of products which are
hydrolyzing Al–F bonds during calcination (20).
formed during the initial deactivation period are the same

226
EARLY ET AL.
as those formed during the steady-state period, it is reason- increasing the acidity of the support can be more clearly
able to conclude that direct reaction with the support does understood. Carbonaceous deposits form more rapidly on
not strongly contribute to the overall performance of the the more acidic supports. If AlF3 and 30% F/Al2O3 in-
system.
deed possess the highest acidity, one can visualize a rapid
The inhibiting effect which chlorine has on the hydro- deposition of coke onto the support. This has the direct
dechlorination reactions has been established in a couple of consequence of making the oxide site unavailable to re-
investigations. It was shown for chlorobenzene dechlorina- act with halogen species spilling over from the metal on
tion over alumina-supported palladium that the reaction ki- which the reaction occurs. Carbonaceous deposits can mi-
netics can be interpreted in terms of a steady state between grate onto the surface of metal particles, making it inac-
the chlorination of the metallic surface by chlorobenzene cessible for reactant adsorption. The relatively low deac-
and hydrogen chloride and the reduction of the chlorine- tivation for the three samples with a low concentration
covered metallic surface (29). The reaction order in HCl of fluorine fits well with this interpretation in that their
during CF3CFCl2 dechlorination over polycrystalline Pd lower acidity would result in a slower rate of carbona-
foil is reported to be negative (30); thus as Cl surface cover- ceous deposit formation. Hence, more sites would be avail-
age increases, reaction rate decreases. In the present work, able for halogen uptake. It is noteworthy that the more
the Pd surface is free of chlorine at the start of the reaction. carbon atoms in a hydrocarbon molecule, the easier the
Deactivation of the catalyst occurs as the system ap- molecule undergoes coking on acid sites of a catalyst (31).
proaches a steady-state coverage of chlorine (halogen).
This might be another reason for the lower rate of catalyst
In the context of influencing the coverage of chlorine deactivation for CF2Cl2 dechlorination compared to that of
(or fluorine) on the Pd surface, the support material can CF3CFCl2.
be considered a “sink” during the initial transient period.
The impact which the change in the catalytic material
Molecules of HCl or chlorine atoms formed via the dissoci- has on selectivity is most pronounced when the ratio of
ation of carbon–chlorine bonds (as well as H atoms formed defluorinated product to dechlorinated product (CF3CH3/
from dissociation of adsorbed hydrogen) should be suffi- CF3CFH2) is compared. For bulk Pd, the selectivity of
ciently mobile to “spillover” onto the support. Once asso- CF3CFCl2 conversion toward CF3CFH2 is 78–82% for the
ciated with the support, they can react with the support Pd(111) plane (32) and reaches 85% for polycrystalline Pd
according to the scheme as shown below:
foil (30). All the catalysts studied have similar, close to bulk
Pd, selectivity; however, the catalysts supported on Al2O3
or low fluorinated supports (2% F/Al2O3, 5% F/Al2O3) ex-
hibit this selectivity after 15 h on stream and the catalysts
supported on 10% F/Al2O3, 30% F/Al2O3, and AlF3 exhibit
it after 1 h on stream. Thus, the results can be divided into
two categories: selectivity changes associated with low flu-
orinated supports and selectivity changes associated with
highly fluorinated supports.
The selectivity changes with time on stream can be un-
derstood in terms of the influence of adsorbed/absorbed
fluorine on dehalogenation ability of Pd. Coq et al. (9) sug-
gested that fluorine atoms occluded into Pd were responsi-
ble for the decreased selectivity toward CH2F2 in the reac-
tion of CF2Cl2 dechlorination. The same might be true for
CF3CFCl2 dechlorination, i.e., fluorine incorporated into Pd
favors defluorination. In case of low fluorinated supports
the availability of oxygen atoms in the vicinity of metal
particles creates the driving force for defluorination of the
substrate followed by the further fluorination of support.
In contrast to the increasing selectivity with time on stream
for the low fluorine containing supports, the decrease in
selectivity for the highly fluorinated supports suggests the
accumulation of halogen (namely fluorine) with time on
stream on the metal surface and/or in bulk since Pd itself
has some intrinsic defluorination activity. The reactivity of
CF2Cl2 is much lower compared to that of CF3CFCl2, and
changes in selectivity with time are less pronounced.
The activity results for CF2Cl2 are consistent with the
idea that chlorine accumulation on the metal surface con-
tributes to the decrease in activity at early time on stream.
The CF2Cl2 molecule is a factor of 2–5 (depending on the
support) less reactive than CF3CFCl2 (Fig. 3A). This lower
reactivity would result in a lower rate of Cl atoms de-
position (if the rate determining step is cleavage of the
first C–Cl bond). Nevertheless, “spillover” of halogen from
the metal to the support can still occur, resulting in a
change of support acidity as described for the CF3CFCl2
system.
In light of the role of the support as a sink for halogen
during the transient period of the reaction, the impact of

CFC DECHLORINATION CATALYZED BY SUPPORTED Pd
227
CONCLUSIONS
6. Bickle, G. M., Suzuki, T., and Mitarai, Y., Appl. Catal. B 4, 141 (1994).
7. Kellner, C. S., and Rao, V. N. M., U.S. Patent 4,873,381 (1989).
8. Coq, B., Hub, S., Figueras, F., and Tournigant, D., Appl. Catal. A 101,
41 (1993).
9. Coq, B., Cognion, J.-M., Figueras, F., and Tournigant, D., J. Catal. 141,
21 (1993).
The catalytic performance of Pd supported on Al2O3,
AlF3, and partially fluorinated aluminas was studied for
the hydrodechlorination of CF3CFCl2 and CF2Cl2. For re-
actions involving CF3CFCl2, all catalysts exhibit a rapid and
10. McVicker, G. B., Kim, C. J., and Eggert, J. J., J. Catal. 80, 315 (1983).
significant decrease in activity; however, little change in ac- 11. Benson, J. E., Hwang, H. S., and Boudart, M., J. Catal. 30, 146 (1973).
12. Karpinski, Z., Early, K., and d’Itri, J. L., J. Catal. 164, 378 (1996).
13. Eischens, R. P., Pliskin, W. A., and Francis, S. A., J. Phys. Chem. 22,
tivity with time on stream occurs with CF2Cl2. The catalyst
deactivation during transient period is rationalized in terms
of chlorine (fluorine) poisoning and carbonaceous species
1786 (1954).
14. Binet, C., Jadi, A., and Lavalley, J.-C., J. Chim. Phys. 86, 451 (1989).
deposition on the Pd surface. It was also established that
the CFC interacts with the support, resulting in halogena-
tion of the surface. Alumina is converted to an aluminum
fluoride-type species followed by the consumption of sur-
face hydroxyl groups. The relative rate of support fluorina-
tion by the reactant is proportional to the inverse initial F
concentration.
15. Little, L. H., “Infrared Spectra of Adsorbed Species.” Academic Press,
London/New York, 1966.
16. Knozinger, H., and Ratnasamy, P., Catal. Rev. Sci. Eng. 17, 31 (1978).
17. Peri, J. B., J. Phys. Chem. 72, 2917 (1968).
18. Borovkov, V. Yu., private communication, Pittsburgh, 1998.
19. Moerkeren, A., Behr, B., Noordeloos-Maas, M. A., and Boelhouver,
C., J. Catal. 24, 177 (1972).
20. Stilmar, F. B., U.S. Atomic Energy Commission, U.S. Patent 2,533,217
(1951).
21. Webb, A. N., Ind. Eng. Chem. 49, 261 (1957).
ACKNOWLEDGMENTS
22. Strand, P., and Kraus, M., Collect. Czech. Chem. Commun. 30, 1136
(1965).
The authors thank the National Science Foundation (CTS-9502090 and
CTS-9509580) for providing financial support. The DuPont Corporation,
especially Dr. Leo Manzer and Dr. C. Steve Kellner, is gratefully acknowl-
edged. Special thanks are also extended to Dr. Gary McVicker of Exxon
Corporation for his technical assistance and intellectual input.
23. Matsuura, K., Watanabe, T., Suzuki, A., and Itoh, M., J. Catal. 26, 127
(1972).
24. Yushechenki, V. V., and Antipina, T. V., Kinet. Katal. 11, 134 (1970).
25. Scokart, P. O., Selim, S. A., Damon, J. P., and Rouxhet, P. G., J. Colloid
Interface Chem. 70, 209 (1979).
26. Kerkhof, F. P. J. M., Oudejans, J. C., Moulijn, J. A., and Matulewicz,
E. R. A., J. Colloid Interface Chem. 77, 120 (1980).
27. Corma, A., Forne´s, V., and Ortega, E., J. Catal. 92, 284 (1985).
28. Hess, A., and Kemnitz, E., J. Catal. 149, 449 (1994).
29. Coq, B., Ferrat, G., and Figueras, F., J. Catal. 101, 434 (1986).
30. Ribeiro, F. H., Gerken, C. A., Somorjai, G. A., Kellner, C. S., Coulston,
G. W., Manzer, L. E., and Abrams, L., Catal. Lett. 45, 149 (1997).
31. Gates, B. C., Katzer, J. R., and Schuit, G. C. A., “Chemistry of Catalytic
Processes,” Chap. 1. McGraw–Hill, New York, 1979.
32. Rupprechter, G., and Somorjai, G., Catal. Lett. 48, 17 (1997).
REFERENCES
1. Fung, S. C., and Sinfelt, J. H., J. Catal. 103, 220 (1987).
2. Anderson, J. R., and McConkey, B. H., J. Catal. 11, 54 (1968).
3. Campbell, J. S., and Kemball, C., Trans. Faraday Soc. 57, 809 (1961).
4. Coq, B., Figueras, F., Hub, S., and Tournigant, D., J. Phys. Chem. 99,
11159 (1995).
5. Manzer, L. E., and Rao, V. N. M., Adv. Catal. 39, 329 (1993).