The use of active carbon pretreated at 2173 K as a support for palladium catalysts for hydrodechlorination reactions

Article abstract of DOI:10.1016/j.cattod.2010.12.039

A commercial active carbon was heat-treated at 2173 K in argon and then subjected to steam gasification, yielding a series of very pure, turbostratic carbon materials, characterized by different specific surface areas and pore volume. These materials served as supports for palladium catalysts. Their pore structure, apparent absence of oxygen containing functional groups and hydrophobic character have a great effect on the dispersion of palladium introduced by impregnation. The use of an acetone solution of palladium acetate rather than an aqueous solution of palladium chloride (a typical Pd precursor) for the impregnation gives better results for preparing more metal dispersed Pd/C catalysts, especially for carbons with smaller micropore volumes. All preheated carbon-supported palladium catalysts showed very good activity and selectivity to CH₂F₂ in CCl₂F₂ (CFC-12) hydrodechlorination, up to 90% at the highest reaction temperature. In contrast, untreated or only HCl-washed carbons showed inferior catalytic properties. Residual phosphorus in the active carbons which were present in the active carbons which have not been subjected to thermal treatment, appears to be responsible for deterioration of catalytic properties of Pd/C. After reaction the presence of interstitial carbon (originating from the CFC-12 molecule) in the Pd lattice was found in the catalysts characterized by lower and medium metal dispersions.

Full text of DOI:10.1016/j.cattod.2010.12.039

Catalysis Today 169 (2011) 223–231
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
The use of active carbon pretreated at 2173 K as a support for palladium catalysts
for hydrodechlorination reactions
a
b
a,c,∗
Magdalena Bonarowska , Wioletta Raróg-Pilecka , Zbigniew Karpi n´ ski
a
Institute of Physical Chemistry, Polish Academy of Sciences, ul. Kasprzaka 44/52, PL-01224 Warszawa, Poland
Warsaw University of Technology, Faculty of Chemistry, ul. Noakowskiego 3, PL-00664 Warszawa, Poland
Faculty of Mathematics and Natural Sciences-School of Science, Cardinal Stefan Wyszy n´ ski University, ul. Wóycickiego 1/3, PL-01938, Warszawa, Poland
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 August 2010
Received in revised form
A commercial active carbon was heat-treated at 2173 K in argon and then subjected to steam gasifi-
cation, yielding a series of very pure, turbostratic carbon materials, characterized by different specific
surface areas and pore volume. These materials served as supports for palladium catalysts. Their pore
structure, apparent absence of oxygen containing functional groups and hydrophobic character have a
great effect on the dispersion of palladium introduced by impregnation. The use of an acetone solution of
palladium acetate rather than an aqueous solution of palladium chloride (a typical Pd precursor) for the
impregnation gives better results for preparing more metal dispersed Pd/C catalysts, especially for car-
bons with smaller micropore volumes. All preheated carbon-supported palladium catalysts showed very
good activity and selectivity to CH2F2 in CCl2F2 (CFC-12) hydrodechlorination, up to 90% at the highest
reaction temperature. In contrast, untreated or only HCl-washed carbons showed inferior catalytic prop-
erties. Residual phosphorus in the active carbons which were present in the active carbons which have
not been subjected to thermal treatment, appears to be responsible for deterioration of catalytic proper-
ties of Pd/C. After reaction the presence of interstitial carbon (originating from the CFC-12 molecule) in
the Pd lattice was found in the catalysts characterized by lower and medium metal dispersions.
1
0 December 2010
Accepted 13 December 2010
Available online 14 January 2011
Keywords:
Turbostratic carbons
Carbon gasification
Pore structure
Carbon-supported Pd catalysts
CF2Cl2
Hydrodechlorination
©
2010 Elsevier B.V. All rights reserved.
1
. Introduction
nation catalyst [1]. Apart from an exhaustive washing with acids
and alkalis, a serious heat treatment of active carbons, at temper-
atures above 1773 K in inert gas, should also serve for effective
purification of industrial active carbons [2,3]. Such pretreatment
removes nearly all more or less volatile contaminants, especially
typical catalytic poisons like S, As and P. This type of active carbon
has already found application as a support of modified ruthenium
catalyst used in ammonia synthesis [4]. Its relatively low spe-
cific surface area and less developed pore structure (compared
to thermally untreated carbons) may be improved by subsequent
gasification using steam or carbon dioxide [2,5,6], making the sup-
port very useful for various applications. It was our hope that
the highly pretreated active carbon should also serve as a use-
ful support for metal-containing catalysts in hydrodechlorination
Besides of their primary application in adsorption processes,
active carbons are commonly used as supports for catalytically
active phases, such as metals or oxides. Catalytic hydrodechlo-
rination (HdCl) of organic wastes is often realized with the use
of carbon-supported Pt, Pd or Ni catalysts. Since this process is
accompanied by massive release of HCl, active carbons, compared
to typical oxidic carriers, are better suited to withstand highly cor-
rosive reaction conditions, making them the support of industrial
choice. Research with the use of such supports as silica or alumina,
serves mainly for short-term model studies, when the functioning
of various metallic (active) phases is tested. One important vari-
able in the preparation of carbon supported hydrodechlorination
catalysts is the selection and pretreatment of the support. In this
respect, van de Sandt et al. found that purification of commercial
active carbons by washing with aqueous NaOH, aqueous HCl and
water resulted in an optimal performance of Pd/C hydrodechlori-
of various harmful compounds, such as CCl . Deprived of a vast
4
majority of functional (e.g. oxygen-carrying) groups, a strongly
hydrophobic nature of highly pretreated carbons would help in
more effective adsorption of CCl4 from wastewaters. Subsequent
hydrodechlorination on metal (usually Pt or Pd) sites located in
the porous structure of carbon should efficiently transform this
harmful molecule into more benign products.
This paper describes the results on preparation and character-
ization of palladium catalysts supported on differently pretreated
samples of Norit GF40 carbon, the support which has been recently
∗ Corresponding author at: Institute of Physical Chemistry, Polish Academy of Sci-
ences, Department of Catalysis on Metals, ul. Kasprzaka 44/52, PL-01224 Warszawa,
Poland. Tel.: +48 22 3433356; fax: +48 22 3433333.
E-mail address: zkarpinski@ichf.edu.pl (Z. Karpi n´ ski).
0
920-5861/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2010.12.039

2
24
M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
Table 1
Characteristics of differently pretreated samples of Norit GF40 carbon.
Two series of palladium/carbon catalysts were prepared by the
incipient wetness impregnation of crushed and sieved carbons
(grain fraction 0.6–1.25 mm), using an aqueous solution of palla-
dium chloride (analytical purity, POCh Gliwice, Poland), slightly
acidified with HCl (analar), or acetone (analytical purity) solu-
tion of palladium acetate (spectral purity from Ventron, Karlsrue,
Germany). After impregnation the materials were dried overnight
Carbon
designation
Pretreatment codea
Mass loss after
gasification, wt%
Ash,
wt%
Density,
g/cm
3
A
B
C
D
E
F
G
H
GF40/H2O
GF40/HCl
GF40/HCl/2173
GF40/HCl/2173/1033
GF40/HCl/2173/1103
GF40/HCl/2173/1120
GF40/HCl/2173/1129
GF40/HCl/2173/1158
–
–
–
1.1
11.24
20.37
28.93
51.3
0.52
0.26
0.16
0.17
0.21
0.30
0.36
0.55
0.29
0.29
0.4
0.39
0.34
0.3
at 373 K. Then they were reduced in a fluidized reactor in an H /Ar
2
flow. During reduction the temperature was increased at a rate
8 K/min from RT to 673 K and kept at 673 K for 3 h. After reduc-
tion the catalysts were flushed with argon and transferred in air to
glass-stoppered bottles, which were stored in a dessicator.
The respective catalyst code includes both the type of carbon
support (as in Table 1) and the kind of Pd precursor, for example,
Pd(Cl)/C denotes the catalyst prepared by deposition of PdCl2 on
carbon C, whereas Pd(Ac)/G stands for the catalyst prepared from
Pd acetate on carbon G. The palladium loading was 2 wt.%.
0.26
0.24
a
For details see Section 2. H2O and HCl stand for washing media, “2173” means the
temperature (in K) of GF40 pretreatment in argon, and the last figure (for example:
1
120) is the temperature of steam gasification.
employed in catalytic studies [7]. In addition, the hydrodechlo-
rination behavior of these catalysts was tested in HdCl of
dichlorodifluoromethane, the reaction selected as a relatively
simple one, frequently investigated in the past decade on differ-
ently supported palladium catalysts, also on carbon-supported Pd
The Pd/C catalysts were characterized by CO chemisorption (at
3
08 K, using a backsorption method with an ASAP 2020 Chem
instrument from Micromeritics), temperature programmed (pal-
ladium) hydride decomposition (TPHD, for details, see [13]), and
X-ray diffraction (XRD, Siemens D5000, Ni-filtered Cu K␣ radia-
tion).
[
1,8–12]. Therefore, it was convenient to compare the catalytic
performance of Pd supported on thermally treated carbon with
previously investigated Pd/support catalysts.
Temperature-programmed desorption (TPD) of the decomposi-
tion products of surface oxygen groups was performed for a few
representative samples of GF40: B, C, G and one catalyst, Pd(Ac)/G.
The TPD experiments were carried out in a flow reactor coupled
to a Dycor Ametek MA200 quadrupole mass spectrometer, in the
manner previously described [14]. A tested sample (mass ∼0.45 g)
was placed in the reactor and heated to 1100 K at a 10 K/min ramp
2
. Experimental
A series of different active carbons were prepared on the basis
of Norit GF 40, the starting material obtained from the Norit B.V.
Company. The first sample, called A, was obtained by washing the
parent carbon with large portions of redistilled water (to remove
the dusty fraction). Another sample (B) was obtained after washing
with hydrochloric acid (analar). The rest of GF 40 was subjected to
a high-temperature heating at 2173 K for 2 h, in argon atmosphere,
yielding a larger batch of sample C. Because such a procedure led to
a considerable carbon graphitization and drastic decrease of spe-
cific surface area (vide infra), subsequent gasification of sample C
with a steam:Ar = 1:1 mixture at temperatures between 1033 and
3
in a flow of helium (25 cm /min), while selected mass signals were
observed at 15 s intervals. The obtained mass spectra allowed to
represent evolution of CO and CO, the main products of decompo-
2
sition of the-surface oxygen-containing groups.
Prior to reaction (and also prior to CO chemisorption and TPD),
all catalysts (all samples ∼0.2 g) were reduced in flowing 10%
3
H /Ar (25 cm /min), raising the temperature from room tempera-
2
ture to 623 K (at 8 K/min), and kept at 623 K for 2 h. The reaction of
dichlorodifluoromethane (CFC-12 from Galco S.A., Belgium; purity
1
158 K for 5 h, led to preparation of next active carbon samples
9
9.9%) with hydrogen (purified over MnO/SiO ) was conducted
characterized by a gradually higher degree of burn-off (samples
D, E, F, G and H). Table 1 collects the sample pretreatment and a
preliminary characteristics of obtained carbon materials.
2
in a glass flow system under atmospheric pressure at 433, 443
and 453 K, in the manner previously described [8,15]. Briefly, feed
−
2
partial pressures were 15 and 150 Torr (1 Torr = 133.3 N m ) of
dichlorodifluoromethane and hydrogen, respectively, in an argon
Surface areas and porosities were measured with an ASAP
2
020 instrument from Micromeritics, employing the BET
carrier and the overall flow rate of the reactant gas mixture was
(
Brunauer–Emmett–Teller), t-Plot, BJH (Barrett–Joyner–Halenda)
3
1
00 cm /min. The reaction mixture leaving the reactor was ana-
and HK (Horvath–Kawazoe) methods and using nitrogen as
adsorbate. Before measuring the adsorption isotherm at 77 K, the
activated carbon was kept at 473 K for 5 h in vacuum to clean its
surface. Possible presence of selected elements in starting, washed
and preheated GF40 carbon was done by the X-ray fluorescence
method. XRF analyses were done with the use of a Philips PW
lyzed by GC. In order to adequately establish changes in the catalytic
behavior, a typical reaction run lasted ∼24 h. The first stage of the
reaction involved a 20-h period at 453 K, when the catalyst perfor-
mance was stabilized. Next, the reaction temperature was lowered,
in 10 K steps, and the next experimental points were collected.
After catalyst screening at the lowest reaction temperature (433 K)
the catalyst performance was tested again at 453 K, giving, in most
cases, a good return to the initial behavior at this temperature. To
avoid secondary reactions, the overall conversion was usually kept
2
400 spectrometer at the Institute of Geology in Warsaw, Poland.
Presence of sulfur in selected active carbon samples was excluded
by the Schöniger method of analysis at the Institute of Organic
Chemistry of PAS in Warsaw. Table 2 shows the analytical results.
Table 2
Results of analysis of activated carbons subjected to different pretreatments.
Carbon sample
pretreatment code )
P, wt%
Cl, wt%
Mg, wt%
0.008
Fe, wt%
Ca, wt%
S, wt%
–b
a
(
Untreated GF40
A (GF40/H2O)
B (GF40/HCl)
0.186
0.137
0.103
0.011
0.018
0.006
0.005
0.004
0.002
0.026
0.015
0.008
0.016
b
b
b
–
–
–
–
b
b
0.147
–
b
b
b
C (GF40/HCl/2173)
–
–
–
a
For sample pretreatment code see Table 1 and text.
Not found.
b

M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
225
Fig. 2. XRD profiles of differently pretreated Norit GF40 samples.
Carbon C (after annealing at 2173 K) shows very small specific sur-
2
face area (26 m /g). A gradually increased degree of gasification
with steam results in a vast recovery of the specific surface area
and porous structure (Table 3).
Upon 51.3% mass burn-off, the specific surface area of carbon
2
2
H (1794 m /g) matches that exhibited by carbons A (1885 m /g)
2
and B (1676 m /g). However, the shape of N adsorption isotherms
2
Fig. 1. TEM images of Norit GF40: (a) starting GF40, (b) GF40 pretreated at 2173 K
carbon C).
of “gasified” carbons loses a “langmuirian” shape, with a regular
increase of hysteresis loop, indicating the presence of mesopores.
Fig. 4 shows the evolution of porous structure with the degree of
gasification of the thermally pretreated GF40. Clearly, both micro-
and mesopores develop upon carbon burn-off.
(
low, i.e. <10%, at the highest reaction temperature. Post-reaction
catalyst samples were also investigated by XRD.
3
. Results and discussion
4
3
3
2
2
0
5
0
5
0
3.1. Turbostratic carbons
H
A
Fig. 1 shows two TEM images showing the morphology of the
starting GF40 carbon and that of the preheated carbon at 2173 K.
The morphology change caused by the preheating is well rec-
ognized: in a completely amorphous structure starting carbon
(
Fig. 1a) it is possible to recognize the presence of parallel graphene
G
layers (Fig. 1b).
F
The X-ray diffraction patterns for the differently pretreated
Norit GF40 carbon samples are shown in Fig. 2. The diffractograms
of thermally untreated GF40 samples (A and B) indicate their amor-
phous character. The only characteristic feature is a diffuse peak
E
◦
corresponding to the (0 0 2) reflection at ∼22.2 . On the other hand,
all thermally treated carbons (irrespective of their further gasifica-
tion) showed turbostratic two dimensional ordering. First, a much
15
10
D
◦
better developed (0 0 2) peak was shifted upwards to ∼24.2 . Next,
the (1 0 0) and (1 0 1) reflections merge into a single (1 0) peak, at
◦
◦
∼
43.5 . Similarly, the (1 1) peak is manifested at ∼79 . It is seen that
even very pronounced gasification (>50%) of highly heated carbon
sample H) which results in an almost complete recovery of specific
(
surface area (Table 3), does not lead to loss of turbostratic charac-
1.0
ter of gasified samples. In addition, a small but very sharp peaks at
C
◦
0.5
∼
26.4 (characteristic for graphite) are well seen on the right side of
the (0 0 2) reflections of turbostratic carbons. So, in line with TEM,
XRD confirms a partial ordering of the carbon.
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Fig. 3 shows nitrogen adsorption–desorption isotherms on dif-
ferently pretreated Norit GF40 active carbons. Carbons A and B (B
is not shown in Fig. 3) exhibit a type I isotherm of nitrogen at 77 K
P/Po
Fig. 3. Adsorption–desorption isotherms of N2 (at 77 K) on differently pretreated
(
according to Brunauer), indicating a large number of micropores.
Norit GF40 active carbons.

2
26
M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
Table 3
Characteristics of untreated and pretreated Norit GF40 active carbon samples.
Pretreatment codea
SBET, m /g
2
t-Plot micropore
area, m /g
t-Plot external
surface area,
m /g
Pore
DFT pore volume, cm /g
3
Carbon
designation
2
b
volume,
2
3
cm /g
Total
Pores <0.5 nm
A
B
C
D
E
F
G
H
GF40/H2O
GF40/HCl
GF40/HCl/2173
GF40/HCl/2173/1033
GF40/HCl/2173/1103
GF40/HCl/2173/1120
GF40/HCl/2173/1129
GF40/HCl/2173/1158
1885
1676
26.5
–c
1583
10.4
–c
93
16.1
34
1.03
0.889
0.814
0.0024
0.408
0.622
0.725
0.861
1.141
0.430
0.074
0.953
0.0036
0.481
0.710
0.825
0.970
1.278
0
0
0
0
0
0
848
815
1256
1423
1526
1794
1192
1315
1319
1333
64
107
207
461
a
For sample pretreatment code see Table 1.
HK (Horvath–Kawazoe) method.
Not measured.
b
c
eration. However, samples C and G show only negligible evolution
of carbon oxides, indicating a very small presence of oxygen func-
tionalities on these materials. A similar situation is with a reduced
Pd(Ac)/G catalyst, where a small peak in CO2 profile, at 700–900 K,
would result from a residual presence of acetate precursor.
In conclusion, the gradually steam-gasified GF40 samples regain
the considerable surface area and porous structure lost in the high
temperature pretreatment of this Norit active carbon. Interestingly
enough, they also acquire a substantial degree of mesoporosity.
Total
Micro (t-Plot)
Meso (BJH)
1
1
0
0
0
0
0
.2
.0
.8
.6
.4
.2
.0
3.2. Carbon-supported palladium catalysts
Deposition of palladium precursors on variously pretreated
GF40/HCl
0
1.1
11.24
20.37
28.93
51.3
Norit GF40 carbon produced different results, was found strongly
dependent on the type impregnating medium and degree of carbon
burn-off produced by steam gasification. Assessment of homogene-
ity of Pd particle size comes from the XRD study of freshly reduced
Carbon burn-off, %
Fig. 4. Development of pore structure by steam gasification of highly pretreated
Norit GF40 active carbon.
(and post-reaction) Pd/C samples (Fig. 6), and indirectly from TPHD
Temperature-programmed desorption from representative
samples of GF40 and one Pd/C catalyst is shown in Fig. 5a (CO2
evolution) and Fig. 5b (CO evolution). It is seen that sample B of
GF40 (without preheating at 2173 K, washed in HCl and pretreated
(temperature programmed hydride decomposition, Fig. 7) investi-
gation, the method which was applied earlier for such diagnostic
purposes [8].
XRD examination established that some catalysts were com-
posed of larger Pd crystallites (∼20 nm in size) and a part of a
in Ar at 773 K) exhibits very pronounced profiles of CO and CO lib-
2
.0x10-8
(b) 1.5x10-7
(
a) 2
Mass signal, a.u.
Mass signal, a.u.
-
8
8
9
1
.5x10
-
7
1
.0x10
-
1
5
.0x10
-
8
5
.0x10
-
.0x10
0
.0
3
0.0
300 400 500 600 700 800 900 1000 1100
00 400 500 600 700 800 900 1000 1100
Temperature, K
Temperature, K
Fig. 5. Temperature-programmed desorption profiles of selected samples of GF40 and Pd(Ac)/G catalyst. (a) CO2 evolution, (b) CO evolution. Carbon B—thick line, carbon
C—dotted line, carbon G—dashed line, Pd(Ac)/C—thin solid line.

M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
227
(
a)
111
(b)
PdCx
111
Pd
PdCx
200
2
00
Pd
Pd
220
PdCx
311
PdCx
2
22
Pd
Pd
PdCx
Pd
PdCx
Pd
PdCx
30
35
40
theta (CuKa)
45
50
40
50
60
70
80
90
2
2
theta (CuKα)
(c)
(d)
PdCx Pd
Pd
111
PdCx Pd
111
Pd
2
00
PdCx
200
PdCx
30
35
40
theta (CuKa)
45
50
35
40
45
50
2
2 theta (CuKa)
Fig. 6. XRD of selected Pd/C catalysts: (a) Pd(Ac)/C, (b) Pd(Cl)/D, (c) Pd(Cl)/F and (d) Pd(Cl)/F). Thin lines—reduced samples, thick lines—post-reaction samples. The presence
of basic reflections from fcc phase of Pd and Pd–C solid solution (PdCx) are marked.
more dispersed Pd material, as concluded from an overlapping of a
very sharp (1 1 1) diffraction peak and a diffuse (1 1 1) Pd reflection
so it is difficult to confirm the presence of a ␤-PdH phase in highly
dispersed Pd material (like for Pd(Cl)/H in Fig. 7). In other cases
the TPHD profiles presented in Fig. 6 exhibit various situations
described above. Results of a more careful analysis of TPHD and
XRD profiles are shown in Table 4, yielding important information
on the degree of homogeneity of Pd materials.
(
Fig. 6c). In other cases (exemplified in Fig. 6a, b and d), the obtained
reflections were found practically not structured, suggesting the
presence of palladium characterized by a narrower distribution of
metal crystallites.
In the case of the TPHD method, which uses the phenomenon of
a facile ␤-PdH formation at a relatively low hydrogen pressure, a
narrow, single TPHD peak implies the ␤-PdH decomposition from
similar in size palladium particles, whereas double/multiple and
structured in shape TPHD peaks suggest the presence of palladium
phase characterized by a broader distribution of particle size [8].
In addition, a shift of peak minimum towards lower temperatures
indicates the presence of smaller Pd particles [8]. Such low tem-
perature TPHD minima are also less intense, because the H/Pd ratio
in bulk Pd hydride decreases when Pd particle size decreases [10],
Table 4 collects the experimental data from CO chemisorption,
TPHD and XRD. It is seen that the use of acetone solution of Pd
acetate gives more homogeneous Pd catalysts than when aque-
ous solution of PdCl2 was used for preparation. In the case of the
latter precursor, for carbons D, E, F and G, the palladium material
is present as small particles as well as larger crystallites. Only in
the case of the carbon H which was the most intensively gasified,
the kind of Pd precursor does not appear to have effect on metal
dispersion.
Table 4 also shows that for Pd/A and Pd/B catalysts there is a
serious conflict between the Pd dispersion data assessed from CO
chemisorption and XRD. A considerable metal dispersion (FE) con-
cluded from the absence of characteristic palladium reflections,
being a sign of a considerable metal high dispersion (FE), is not
reflected in the CO uptake. This is presumably because the surface
of palladium supported on not preheated carbons A and B contains
considerable amounts of phosphorus (Table 2), which as a typical
catalytic poison hampered the CO uptake. This fact is also reflected
in the catalytic behavior of these catalysts (vide infra). For the rest
of Pd/C/catalysts, the agreement between XRD and CO chemisorp-
tion data is fair or good (Table 4), especially if one accepts known
problems with a partial blocking of palladium surface by carbon
species originating from the support [17].
Pd(C)/H
Pd(Cl)/F
Pd(Ac)/F
Pd(Cl)/D
Pd(Ac)/C
Similar results were obtained in preparation of Pd catalysts sup-
ported on turbostratic carbons, pretreated at 1223 K [17]. Fairly
narrow Pd particle size distributions were obtained for the catalysts
made by impregnation of a THF solution of palladium acetylacet-
onate. In contrast, broader distributions were obtained when an
aqueous solution of PdCl2 was used: large Pd particles up to 45 nm
were observed. Although those authors [17] related this fact to
2
50
300
350
400
Temperature, K
Fig. 7. Temperature-programmed hydride decomposition profiles of selected Pd/C
catalysts.

2
28
M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
Table 4
Characteristics of Pd/C catalysts.
Catalyst
Fraction
exposed
Pd particle size, nm
Results of TPHD
designationa
(FE = CO/Pd)
From CO/Pdb
Crystallite size
from XRDc after
reduction
Crystallite size from XRDc
after reaction
H/Pd from
hydride
decomposition
Temperature of
minimum in TPHD,
K
d
d
Pd(Cl)/A
Pd(Ac)/A
Pd(Cl)/B
Pd(Ac)/B
Pd(Cl)/C
Pd(Ac)/C
Pd(Cl)/D
0.0077
0.0187
0.0285
0.0714
0.0173
0.0149
0.137
146
60
<2
<2
<2
<2
<2
No PdH decomposition
d
d
<2
No PdH decomposition
No PdH decomposition
No PdH decomposition
361.5
d
d
39.3
15.7
64.6
75.1
8.2
<2
d
d
<2
32.6
20.2
10 + 22.5
22.2 (PdCx) + 24.9 (Pd)
19.2 (PdCx)
21.1 (PdCx) + 22.3 (Pd)
0.50
0.46
0.35
357.1
339.8, 358.5
(
0.22 + 0.13)e
Pd(Ac)/D
Pd(Cl)/E
0.054
0.124
20.7
9.0
22.3
5.1 + 23.0
21 (PdCx)
3.9
0.36
0.27
(0.20 + 0.07)
0.16
358.1
333.8, 357.7
e
(
PdCx) + 23.4(PdCx) + 23.5(Pd)
Pd(Ac)/E
Pd(Cl)/F
0.199
0.138
5.6
8.1
4.2
4.7 + 20.5
3.6 (PdCx)
4.8 (PdCx) + 19.2 (Pd)
328.5
337.7, 360.7
0.24
(
0.22 + 0.02)e
Pd(Ac)/F
Pd(Cl)/G
Pd(Ac)/G
Pd(Cl)/H
Pd(Ac)/H
0.149
0.185
0.228
0.236
0.225
7.5
6.1
4.9
4.7
5.0
3.9
5.0
3.9
4.1
4.4
4.2 (PdCx)
4.2 (PdCx)
4.6 (PdCx)
6.3 (PdCx)
3.6 (PdCx)
0.21
0.21(0.18 + 0.03)
0.18
0.04
0.16
336.4
330, 353.4
332.7
327.2
335.6
e
a
For further details see Section 2 and Table 3.
Calculated as d = 1.12/FE, according to [16].
From the broadening of (1 1 1) reflections (using the Scherrer equation). Crystallite sizes from the most abundant phases are underlined (PdCx—carbon solution in
b
c
palladium).
d
Extremely diffuse, practically invisible XRD reflections from Pd.
For two minima in TPHD profiles (last column), a H/Pd contribution from each hydrogen liberation is accordingly marked in the brackets.
e
problems associated with catalyst passivation (coalescence of non-
passivated Pd particles upon exposure to air), we believe that in
our case the use of an organic medium (less polar than water)
is beneficial for distributing the Pd precursor in a pore structure
of a hydrophobic carbon material, mostly deprived of oxygen-
containing functional groups. Such a situation takes place when the
pore structure of original GF40 Norit carbon is not fully recovered
after carbon pretreatment (2173 K + gasification), i.e. for samples D,
E, F and G. Carbon H, characterized by much larger surface area and
pore volume (Table 2), is able to accommodate larger amounts of
smaller clusters of Pd precursors more uniformly, irrespective of
their chemical nature and impregnating medium. Therefore, possi-
ble problems associated with a wettability of hydrophobic support
results with an aqueous solution of anionic Pd precursor (PdCl2
in HCl is supposed to turn H PdCl ) and acetone solution of neu-
2
4
tral Pd(CH COO) do not confirm the suggestion mentioned above.
3
2
As long as the porous structure is well developed (i.e. for car-
bon samples G and H) the differences in metal dispersion are not
large. Similar results were obtained by Gurrath et al. [20] who used
different Pd precursors (anionic, neutral and cationic) introduced
to carbons characterized by considerable surface areas and pore
volumes. In this respect, one refers to an earlier report of Rodriguez-
Reinoso et al. [21] on characterization of Fe/C catalysts prepared
from aqueous solutions of iron nitrate and n-pentane solutions of
iron pentacarbonyl. The carboxylic groups of the activated carbon
supports were crucial to facilitate the access of the aqueous solu-
tion to the small pores, whereas this was not the case for nonpolar
pentane solvent.
[
18] are important only when its porous structure is not sufficiently
developed.
Similarly, Ehrburger et al. [19] found a direct correlation
between the degree of dispersion of platinum supported on a
graphitized carbon black and the extent of prior gasification (in
air) of the carbon support. Interestingly, the degree of dispersion
of platinum for the catalyst impregnated with aqueous solu-
tion of H PtCl was found to be 24% compared to 55% obtained
3.3. Catalytic behavior of Pd/C in dichlorodifluoromethane
hydrodechlorination
Fig. 8a exemplifies the procedure of catalyst testing described in
Section 2. It is seen that after catalyst’s “passivation” at 453 K, fur-
ther changes in the reaction temperature (down to 443 and 433 K)
produce stable conversions (and product selectivities) in very short
periods of time and a final “return” to the initial reaction temper-
ature (453 K) brings back the previous catalytic behavior. Catalytic
screening of all Pd/C samples prepared with use of highly pre-
treated active carbons showed that stable conversions are achieved
in a relatively short “passivation” time (<2 h). This is demonstrated
in Fig. 8a and b, where also product selectivities are shown to be
quickly established at a constant level. Methane, difluoromethane
(HFC-32), and, to a much smaller extent, chlorodifluoromethane
(HCFC-22) were found to be the predominant products. For Pd
samples supported on highly pretreated carbons the first two
compounds make up usually more than 97% of all products. The
remaining products (mainly CHClF2 and C H , CClF , CHF , and
2
6
for the same catalyst loading when impregnation was carried
out in a benzene–ethanol mixture. The authors [19] assumed
that a more uniform distribution of chloroplatinic acid (from a
benzene–ethanol solution) on the substrate should give a higher
degree of metal dispersion. Then, they consider that the dielectric
constant of the benzene–ethanol solution mixture is much less than
that of water, so chloroplatinic acid is present essentially as undis-
sociated molecules in the former medium, whereas in the aqueous
solution it is strongly ionized, platinum being present in the anionic
2
−
form, [PtCl6] . In conclusion, the mentioned authors believe that
chloroplatinic acid should be adsorbed nonspecifically, molecularly
and uniformly from a benzene–ethanol mixture. Carbons which are
poor adsorbents for anions, are able to adsorb chloroplatinic acid
from an aqueous solution as agglomerates in the form of islands
or clusters giving rise to a broad distribution of particle sizes. Our
2
6
3
3
CH Cl) make up <4% of all products at 453 K.
3

M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
229
Fig. 8. Time-on-stream behavior in CCl2F2 hydrodechlorination on selected samples of Pd/C catalysts at 453 K: (a) Pd(Cl)/F, where also the screening procedure is shown
catalyst stabilization at 453 K, than the decrease of reaction temperature to 443 K and 433 K, and return to 453 K, see Section 2), (b) Pd(Ac)/H, (c) Pd(Ac)/A, and (d) Pd(Ac)/B.
(
Noticeable exceptions are Pd/C samples based on non-
The selectivity level of ∼90% obtained in this study is characteristic
for the best previously investigated Pd/C catalysts [1,8–12,15]. A
similar conclusion concerns the overall activity (TOF) of the most
active catalysts tested in this study [9,15,22,23].
Fig. 9 shows the correlation between the turnover frequency
and palladium dispersion for thermally treated carbon Pd catalysts.
Contaminated carbons (A and B) are excluded from this relation. It
is seen that among 12 tested Pd/C catalysts, 11 of them show a
preheated activated carbons (A and B), for which larger changes
in overall conversions and in product selectivities are expanded
over longer time intervals (Fig. 8c and d). As it was mentioned in
Section 2 (Table 2), carbon B (washed with HCl), and especially car-
bon A (washed with distilled water) contain nonnegligible amounts
of phosphorus (GF40 is chemically activated using H PO ). Partial
3
4
removal of phosphorus by HCl wash slightly improves the cat-
alytic behavior: both the conversion level as well as the selectivity
towards CH2F2 are increased for Pd(Ac)/B. Further improvement
of the catalyst behavior of Pd(Ac)/A and Pd(Ac)/B observed with
time-on-stream must also result from additional decontamination
during hydrodechlorination (due to HCl and HF evolution) at 453 K.
Because our main attention was focused on the behavior of ther-
mally treated carbon-supported catalysts, further work with the
catalysts supported on carbons A and B was suspended.
The results of catalytic tests at steady state are presented in
Table 5. The turnover frequency, as a measure of catalytic activ-
ity, was calculated by using metal dispersion data assessed from
XRD of used catalysts (FE in the first column of Table 5). With very
few exceptions, the selectivities to CH2F2 (desired reaction prod-
uct) were measured at overall conversions <10%, so they should
result from initial product distributions. It is seen that, apart from
catalysts supported on carbons A and B, the selectivities to CH2F2
are very high, usually between 80 and 90%, for the highest reac-
tion temperature. The trend in the activation energy (higher E ’s
A
for carbons A and B, last column in Table 5) also reflects the need
of removing contaminants from active sites (here phosphorus). In
Fig. 9. TOF as a function of metal dispersion in CCl2F2 hydrodechlorination on Pd/C
catalysts. Black square represents the behavior of Pd(Ac)/C catalysts (average from
three runs).
general, the level of E ’s (52–62 kJ/mol) is similar to that obtained
A
for carbon-supported palladium catalysts in earlier reports [9,10].

2
30
M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
Table 5
Kinetic data for 1.5 wt% Pd/C: turnover frequencies and selectivities towards CH2F2 at three reaction temperatures (433, 443 and 453 K), and apparent activation energies
EA’s).
(
Catalysta (FEb)
TOF, s (SCH2
−1
, %)
EA, kJ/mol
F
2
433 K
443 K
453 K
c
Pd(Cl)/A (0.75 )
0.000769 (53.8%)
0.000787 (43.1%)
0.00105 (69.5%)
0.000852 (69.3%)
0.0152 (85.1%)
0.00434 (83.2%)
0.0151 (85.6%)
0.0112 (90.7%)
0.0141 (87.9%)
0.0186 (88.8%)
0.0227 (91.4%)
0.0134 (90.7%)
0.0132 (90.8%)
0.0168 (92.2%)
0.00808 (75.0%)
0.013 (87.3%)
0.00113 (52.1%)
0.00124 (40.3%)
0.00168 (68.0%)
0.0012 (66.0%)
0.0227 (83.2%)
0.00662 (81.3%)
0.0202 (84.0%)
0.0157 (89.3%)
0.0209 (86.5%)
0.0259 (87.3%)
0.0313 (90.5%)
0.0187 (89.7%)
0.0188 (89.7%)
0.0246 (91.3%)
0.0114 (72.3%)
0.0188 (85.4%)
0.00157e−3 (51.0%)
0.0017 (38.0%)
0.00245 (63.0%)
0.00178 (65.6%)
0.0324 (81.7%)
0.00837 (80.8%)
0.0291 (88.1%)
0.0215 (82.4%)
0.0272 (85.2%)
0.0356 (86.0%)
0.0432 (89.5%)
0.0258 (88.5%)
0.0263 (88.4%)
0.0335 (90.4%)
0.0158 (70.5%)
0.025 (84.0%)
61.9 ± 1.1
c
Pd(Ac)/A (0.75 )
62.8 ± 1.0
58.4 ± 3.4
62.5 ± 1.7
57.9 ± 0.9
55.7 ± 1.0
55.1 ± 2.0
53.3 ± 3.0
55.5 ± 1.4
52.5 ± 0.1
53.7 ± 0.5
53.0 ± 0.3
54.5 ± 0.8
53.8 ± 0.6
56.1 ± 0.9
53.6 ± 1.7
c
Pd(Cl)/B (0.75 )
c
Pd(Ac)/B (0.75 )
Pd(Cl)/C (0.048)
Pd(Ac)/C (0.059)
Pd(Cl)/D (0.054)
Pd(Ac)/D (0.052)
Pd(Cl)/E (0.196)
Pd(Ac)/E (0.316)
Pd(Cl)/F (0.147)
Pd(Ac)/F (0.260)
Pd(Cl)/G (0.270)
Pd(Ac)/G (0.246)
Pd(Cl)/H (0.179)
Pd(Ac)/H (0.316)
a
b
c
As in Table 4.
Calculated as FE = 1.12/d, according to [16], where d is the metal crystallite measured by XRD after reaction (see Table 4).
Approximate FE values assuming that Pd particle size is 1.5 nm in (very diffuse XRD lines).
similar level of catalytic activity (open squares in Fig. 9). Such a
result seems to be in apparent disagreement with our earlier data
obtained for Pd/Al O catalysts, where a mild (i.e. with TOF changes
Pd bulk. Slowing down this process creates a better possibility for
hydrogenating these species to methane.
2
3
within an order of magnitude over the complete FE range) correla-
tion between TOF and metal dispersion was found [15]. Although
a more clear-cut explanation of this difference is not offered here,
we should admit that our present results encompass only a limited
range of palladium dispersion (FE between 0.05 and 0.30). On the
other hand, the black square in Fig. 8, which deviates from the sug-
gested correlation (thick line), shows the average behavior assessed
from three independent runs with the Pd(Ac)/C catalyst.
The fact that >90% of tested catalysts showed a direct correlation
in confrontation with only a single “unfitting” catalyst suggested
to leave this problem behind. Our lack of enthusiasm in this mat-
ter was also associated with the fact that carbon C is the sample
which was thermally treated, but it was not subjected to gasifi-
cation. So, its surface area was very low and pore structure very
limited, making this sample not very attractive as a catalyst sup-
port. We would also suspect that such a drastic thermal treatment
would yield very inhomogeneous carbon material and, in effect,
non-uniform distribution of deposited palladium material.
Fig. 6 showing the XRD profiles of Pd/C catalysts demonstrates
the presence of Pd–C solution in the samples subjected to CCl2F2
hydrodechlorination. Carbiding of palladium-containing catalysts
in CCl2F2 hydrodechlorination is a well-known phenomenon. It
is manifested in a substantial retention of carbon (originating
from CCl2F2 molecule) in the palladium lattice, as the presence of
palladium–carbon solutions (up to a PdC0.13) has been ascertained
using XRD [8,11,15,24–29]. By analogy with the formation of ␤-
palladium hydride, a more effective carbon incorporation to Pd is
expected for larger Pd particles [30]. In line with such an interpre-
tation, McCaulley [30] and Morato et al. [27] found less carbon in
Pd catalysts characterized by smaller crystallite sizes. Fig. 6c and,
especially, d shows that bulk carbiding of smaller Pd crystallites
4. Conclusions
Highly preheated (at 2173 K) commercial active carbon, sub-
jected to steam gasification at different temperatures, yielded a
series of very pure, turbostratic carbon materials, characterized by
different specific surface areas and pore volumes. Gasification of
highly sintered active carbons with steam gradually restores the
surface area and pore structure of the starting carbon. These mate-
rials served as supports for palladium catalysts tested in CCl2F2
hydrodechlorination. The pore structure and hydrophobic charac-
ter of highly preheated carbons have a great effect on dispersion
of palladium precursor introduced by impregnation. Impregnation
with acetone solution of palladium acetate brings better results
than using an aqueous solution of palladium dichloride, especially
for carbons with smaller micropore volumes. It seems that a limited
wettability with water of carbons deprived of functional groups (in
effect of high temperature treatment) results in deposition of larger
clusters of Pd precursors and, in effect, leads to preparation of low
metal dispersed catalysts, and inhomogeneous particle size distri-
bution. All highly preheated carbon-supported palladium catalysts
showed very good activity and selectivity in hydrodechlorina-
tion of dichlorodifluoromethane (selectivity to CH2F2 up to 90%).
In contrast, untreated and or only HCl-washed carbons showed
much poorer catalytic properties. Residual phosphorus, present
in the active carbons which have not been subjected to thermal
treatment, appears to be responsible for deterioration of cat-
alytic properties of Pd/C. Catalytic performance of these catalysts
is markedly improved during reaction, suggesting that the phos-
phorus content is decreasing due to interaction with HCl and HF,
products liberated during hydrodehalogenation.
After reaction the presence of interstitial carbon (originating
from the CFC-12 molecule) in the Pd lattice was found in the cata-
lysts characterized by lower and medium metal dispersions.
(
catalysts Pd(Cl)/F and Pd(Ac)/F with the Pd crystallite size less
than 5 nm) proceeds during reaction. It appears that during CCl2F2
hydrodechlorination, massive amounts of bare C species (after the
1
removal of all halogens) become very effective carbiding agents.
Fig. 8a and b shows that the selectivity to methane slightly increases
at the initial stage of reaction. This effect could be due to a gradu-
ally less effective removal of C1 species via their incorporation to
Acknowledgements
This work was supported by the Polish Ministry of Science and
Higher Education within Research Project No. N N204 161636. We

M. Bonarowska et al. / Catalysis Today 169 (2011) 223–231
231
are also grateful to Norit B.V. Company for providing us with com-
mercially manufactured activated carbon materials.
[14] G.S. Szyma n´ ski, Z. Karpi n´ ski, S. Biniak, A. S´ wi a˛ tkowski, Carbon 40 (2002)
627–2639.
2
[
[
15] W. Juszczyk, A. Malinowski, Z. Karpi n´ ski, Appl. Catal. A 166 (1998) 311–319.
16] S. Ichikawa, H. Poppa, M. Boudart, J. Catal. 91 (1985) 1–10.
References
[17] N. Krishnankutty, M.A. Vannice, J. Catal. 155 (1995) 312–326.
[18] L. Daza, S. Mendioroz, J.A. Pajares, Carbon 24 (1986) 33–41.
[19] P. Ehrburger, O.P. Mahajan, P.L. Walker Jr., J. Catal. 43 (1976) 61–67.
[20] M. Gurrath, T. Kuretzky, H.P. Boehm, L.B. Okhlopkov, A.S. Lisitsyn, V.A.
Likholobov, Carbon 38 (2000) 1241–1255.
[1] E.J.A.X. van de Sandt, A. Wiersma, M. Makkee, H. van Bekkum, J.A. Moulijn, Appl.
Catal. A 173 (1998) 161–173.
[
[
[
[
2] L. Forni, D. Molinari, I. Rossetti, N. Pernicone, Appl. Catal. A 185 (1999) 269–275.
3] X. Zheng, S. Zhang, J. Xu, K. Wei, Carbon 40 (2002) 2597–2603.
4] R.B. Strait, Nitrogen Methanol 238 (1999) 37–43.
[
[
21] F. Rodríguez-Reinoso, C. Salina-Martínez de Lecea, A. Sepúlveda-Escribano, J.D.
López-González, Catal. Today 7 (1990) 287–298.
22] A.L. Dantas Ramos, P. da Silva Alves, D.A.G. Aranda, M. Schmal, Appl. Catal. A
5] Z. Kowalczyk, S. Jodzis, W. Raróg, J. Zieli n´ ski, J. Pielaszek, A. Presz, Appl. Catal.
277 (2004) 71–81.
A 184 (1999) 95–102.
[
[
23] Y.C. Cao, Y. Li, Appl. Catal. A 294 (2005) 298–305.
24] E.J.A.X. van de Sandt, A. Wiersma, M. Makkee, H. van Bekkum, J.A. Moulijn, Catal.
Today 35 (1997) 163–170.
[
[
[
[
6] Z. Kowalczyk, J. Sentek, S. Jodzis, R. Diduszko, A. Presz, A. Tetrzyk, Z. Kucharski,
J. Suwalski, Carbon 34 (1996) 403–409.
7] R.F. Bueres, E. Asedegbega-Nieto, E. Díaz, S. Ordó n˜ ez, F.V. Díez, Catal. Today 150
[
25] E.J.A.X. van de Sandt, A. Wiersma, M. Makkee, H. van Bekkum, J.A. Moulijn, Appl.
Catal. A 155 (1997) 59–73.
(
2010) 16–21.
8] M. Bonarowska, J. Pielaszek, V.A. Semikolenov, Z. Karpi n´ ski, J. Catal. 209 (2002)
28–538.
9] B. Coq, F. Figuéras, S. Hub, D. Tournigant, J. Phys. Chem. 99 (1995) 11159–
1166.
[
[
26] M. Öcal, M. Maciejewski, A. Baiker, Appl. Catal. B 21 (1999) 279–289.
27] A. Morato, C. Alonso, F. Medina, Y. Cesteros, P. Salagre, J.E. Sueiras, D. Tichit, B.
Coq, Appl. Catal. B 32 (2001) 167–179.
5
1
[
[
[
28] M. Bonarowska, A. Malinowski, W. Juszczyk, Z. Karpi n´ ski, Appl. Catal. B 30
[
[
10] B. Coq, J.M. Cognion, F. Figuéras, D. Tournigant, J. Catal. 141 (1993) 21–33.
11] A. Malinowski, W. Juszczyk, J. Pielaszek, M. Bonarowska, M. Wojciechowska, Z.
Karpi n´ ski, Stud. Surf. Sci. Catal. 130 (2000) 1991–1996.
(
2001) 187–193.
29] A. Malinowski, W. Juszczyk, M. Bonarowska, J. Pielaszek, Z. Karpi n´ ski, J. Catal.
77 (1998) 153–163.
30] J.A. McCaulley, J. Phys. Chem. 97 (1993) 10372–10379.
1
[
[
12] M. Bonarowska, B. Burda, W. Juszczyk, J. Pielaszek, Z. Kowalczyk, Z. Karpi n´ ski,
Appl. Catal. B 35 (2001) 13–21.
13] M. Bonarowska, Z. Karpi n´ ski, Polish J. Chem. 82 (2008) 1973–1979.