Nanostructured Pd/C catalysts prepared by grafting of model carboxylate complexes onto functionalized carbon

Article abstract of DOI:10.1016/j.jcat.2006.07.019

Pd(5 wt%)/C catalysts were prepared by grafting carboxylate precursors onto functionalized carbon. The functionalization of the support was carried out with HNO₃ or H₂O₂, and the number and nature of oxygenated functions introduced were determined by a combination of simplified Boehm's titration, XPS, BET, and DRIFTS. These functions were then used as anchors for model Pd precursors through covalent bonding. The underlying ligand-exchange mechanism was elucidated through detailed XPS studies. The thermal activation of the grafted materials was followed by in situ mass spectrometry, which demonstrated that it consisted of ligand losses. The activated samples were characterized by SEM, XRD, CO chemisorption, and XPS and used in the reduction of 2-methyl-2-nitropropane (MNP) into t-butylamine (TBA). The catalytic activity was shown to be correlated with the initial carbon acidity and the Pd dispersion.

Full text of DOI:10.1016/j.jcat.2006.07.019

Journal of Catalysis 243 (2006) 239–251
www.elsevier.com/locate/jcat
Nanostructured Pd/C catalysts prepared by grafting of model carboxylate
complexes onto functionalized carbon
S. Hermans a, C. Diverchy a, O. Demoulin b, V. Dubois c, E.M. Gaigneaux b, M. Devillers a,∗
a
Université catholique de Louvain, Unité de chimie des matériaux inorganiques et organiques, Place Louis Pasteur 1/3, B-1348 Louvain-la-Neuve, Belgium
b
Université catholique de Louvain, Unité de catalyse et chimie des matériaux divisés, Croix du Sud 2/17, B-1348 Louvain-la-Neuve, Belgium
c
Meurice Institute, Avenue Emile Gryzon 1, B-1070 Brussels, Belgium
Received 22 May 2006; revised 13 July 2006; accepted 14 July 2006
Available online 6 September 2006
Abstract
Pd(5 wt%)/C catalysts were prepared by grafting carboxylate precursors onto functionalized carbon. The functionalization of the support was
carried out with HNO or H O , and the number and nature of oxygenated functions introduced were determined by a combination of simplified
3
2 2
Boehm’s titration, XPS, BET, and DRIFTS. These functions were then used as anchors for model Pd precursors through covalent bonding. The
underlying ligand-exchange mechanism was elucidated through detailed XPS studies. The thermal activation of the grafted materials was followed
by in situ mass spectrometry, which demonstrated that it consisted of ligand losses. The activated samples were characterized by SEM, XRD, CO
chemisorption, and XPS and used in the reduction of 2-methyl-2-nitropropane (MNP) into t-butylamine (TBA). The catalytic activity was shown
to be correlated with the initial carbon acidity and the Pd dispersion.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Palladium; Carbon; Functionalization; Oxygenated functions; Heterogeneous catalysis; Hydrogenation; Grafting
1. Introduction
fruit stones, coconut shells) for producing active carbon and
the various methods of carbonization used, its physicochemical
characteristics vary greatly from one batch to the other [2–5].
Second, due to its high absorbing power (black body), carbon
is not easily characterized spectroscopically by IR or Raman.
Third, its inertness, which can be advantageous in some cir-
cumstances, becomes a disadvantage when trying to chemically
modify its surface. Hence, the use of carbonaceous materials as
catalyst supports rely mainly on preparation methods developed
empirically.
The transformation of the raw material into a usable car-
bon support can be carried out in two ways: a physical or a
chemical “activation” method [2,4,5]. In the latter case, the fi-
nal high-surface area support is formed in one step by using
a reactant for chemical activation such as zinc chloride, phos-
phoric acid, or potassium hydroxide. In the former case, the raw
material is carbonized at high temperature (700–800 ◦C), fol-
lowed by oxidative “activation” to increase the specific surface
area by developing the pore system. In the final carbon mater-
ial, hetero-atoms, such as O, N, or S, remaining from the source
material are responsible for the surface acidic/basic character.
Carbonaceous materials are ideal supports for heterogeneous
catalysts, because they are cheap, inert, and robust [1,2]. More-
over, the so-called “activated” (or “active”) carbons present
very high specific surface areas that allow high dispersion of the
supported catalytically active phase [2]. The other advantages
of carbon supports compared with inorganic oxides such as sil-
ica and alumina is that their surface favors the metallic reduced
state and interface effects are minimized. In contrast, oxide sup-
ports present surface hydroxyl groups that stabilize oxidized
phases and also are sensitive to harsh conditions, especially
moisture. Consequently, many industrial processes make use of
carbon-supported catalysts, especially for reactions carried out
in the liquid phase [2]. However, the use of carbon materials
as catalyst supports carries several major drawbacks. First, due
to the variety of possible raw materials (e.g., coal, wood, peat,
*
Corresponding author. Fax: +32 10 47 23 30.
E-mail address: devillers@chim.ucl.ac.be (M. Devillers).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.07.019

240
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
Moreover, functional groups are also introduced at the surface
in the process of creating the porosity [2,5]. These functions
are present in modest amounts and can be removed by thermal
treatment at high temperature [3,6–10]. In contrast, to increase
the surface reactivity, the number of oxygenated functions can
be increased by further oxidation in the gas or liquid phase [11].
The gas-phase treatments usually involve oxidation in a flow
of air [12–15], O2 (in N2 or Ar) [8,9,16–22], N2O (in N2) [8]
oxygen plasma [23–25], or ozone [25,26], whereas liquid-phase
functionalization implies stirring a given amount of solid car-
bon in a solution of HNO3 (see below), H2O2 (see below),
(NH4)2S2O8 [16,23,24,27–30], NaOCl [13,16], Ca(ClO)2 [31],
or KMnO4 [16,32], for example. The nature of the possible
oxygen-containing groups present at the surface of an active
carbon (before or after oxidative functionalization) has been
determined to be mainly of the carboxylic, lactonic, pheno-
lic, and carbonyl types [3,11,33]. Consequently, one can speak
of a “surface acidity” for active carbons [11]. Boehm and co-
workers have developed a method to quantify these functions by
titration [3,11,16,33,34]. Oxygenated surface functions can also
be characterized by FTIR or DRIFTS, TPD, XPS, TGA/MS,
and other methods [1,6,8,14,22,31,34–39]. The point of zero
charge (PZC) also has been shown to give a good indication of
the acidic/basic properties of carbons and can be determined by,
for example, mass titration [40].
The liquid-phase functionalization of active carbons with
HNO3 in particular has already been the object of many studies.
Although gas-phase oxidation increases mainly the concentra-
tion of surface hydroxyls and carbonyls (as well as lactones and
anhydrides), nitric acid treatment leads to a marked increase in
carboxylic acid groups, as well as lactones to a lesser extent
[3,7–9,13,14,18,20,29–31,41–44]. It was found that the num-
ber of oxygen functions introduced is dependent on the HNO3
concentration used [45] (although one study reported an oppo-
site effect [32]), or the temperature at which the treatment is
carried out [6,46] and its duration [9]. Usually, the harsh ox-
idizing conditions associated with a HNO3 treatment tend to
modify the textural characteristics of the carbon used [27,39,43,
47–52], but it has also been shown that O-groups can be intro-
duced by HNO3 under mild conditions (low concentration, tem-
perature, and/or duration) without significantly modifying the
porous structure [17,21,40,44,49,52,53]. Similarly, many au-
thors have already carried out carbon oxidation with H2O2 and
studied its particular effects, but less extensively than with nitric
acid. The amount of surface O-groups introduced by H2O2 is
generally lower than that introduced by HNO3 [10,17,27,29,44,
50,51,54,55]. It is possible to slightly augment the amount in-
troduced by increasing the concentration of the H2O2 solution
used [56]. The functions introduced are also acidic in nature,
but are usually of lower strength (mainly lactones rather than
carboxylic acids) [13,23,24,26,30,32], but, surprisingly, H2O2
was also found to be able to create new surface basic sites in
addition to the acidic ones [30]. One paper reported the effect
of pH control on H2O2 functionalization, leading to higher ox-
idative power [57]. In addition, it was found that H2O2 does not
usually modify the textural characteristics of the treated carbon
[25], although it has been shown to have an effect in some cases
[23,24].
The present paper aims to show that it is possible to prepare
carbon-supported catalysts in a controlled manner. Our strat-
egy relies on the idea of introducing acidic functions in known
amounts at the support’s surface and then using them as an-
chors for grafting the catalytically active phase. In particular,
using carboxylate complexes allows a ligand-exchange process
for surface carboxylic groups, thus producing grafted molec-
ular fragments of controlled structure. To go even further, we
have used “probe” molecules containing chelate ligands and/or
fluorine-containing fragments to model the chemical interaction
of the metallic precursor with the prefunctionalized support.
The presence of chelating ligands allows us to direct ligand
substitution toward selective exchanges, whereas the introduc-
tion of F atoms provides a quantitative probe for XPS char-
acterization. The present study aims to control precisely each
step of the catalyst preparation, and via “molecular modeling”
with “probe” molecules, to gain a precise picture of the frag-
ments grafted onto the carbon supports. Studies dealing with
real grafting of metallic species onto carbonaceous materials
remain very scarce, and the methods used usually were not op-
timal [58]. One paper reported the use of a carbon-supported
Pd complex that is simply adsorbed on the surface and remains
intact and active for catalytic hydrogenation [59]. Ryndin et al.
used oxidized diamond to anchor the organometallic complex
[(C3H5)Pd(C5H5)] and found that the resulting Pd particles
after reduction are extremely small (<10 Å) and resistant to
sintering [60]. Immobilization of metal complexes on a car-
bon electrode via their ligands was also carried out through
electrochemical cation radical formation [61]. Recent reports
by Silva et al. presented successful carbon-grafted complexes,
even chiral ones in some cases, such as manganese(III) salen
species, for applications in (enantioselective) epoxidation, for
example [62–65]. Another recent work by another group [66]
used HNO3-functionalized carbon to covalently attach acety-
lacetonate complexes for application in liquid-phase oxidation.
In both these recent cases, the stability toward leaching of the
complexes into the catalytic reaction medium was very good.
Our strategy was used to prepare Pd(5 wt%)/C catalysts,
the performance of which was evaluated (after thermal activa-
tion) in the selective hydrogenation of 2-methyl-2-nitropropane
(MNP) into t-butylamine (TBA) (Scheme 1). The hydrogena-
tion of nitro compounds to the corresponding amines is an in-
teresting reaction from an industrial standpoint, because these
chemicals are used as basic raw materials for urethanes, rubber
chemicals, dyes, and pharmaceuticals [2]. Although the cat-
alytic hydrogenation of aromatic nitro compounds has been
studied extensively, far fewer studies have been devoted to
the transformation of aliphatic nitro compounds into the cor-
responding amines [67–70]. Previous reports on nitropropane
or nitrobutane reduction have shown that Pd/C catalysts display
interesting performance [68–70] and that the surface state of the
support might have an influence on the catalytic activity [70].
The literature reports that in general (and for other reactions),
the augmentation of oxygenated surface groups in the carbon
support could have a double impact in catalysis: (i) via a sim-

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
241
Scheme 1. Reduction of 2-methyl-2-nitropropane (MNP) into t-butylamine (TBA).
ple dispersion effect, when support functionalization leads to
higher metallic dispersions and hence higher activity [20,26,45,
53,71], or (ii) by a more complicated mechanism in which the
O-functions play an active role in the catalytic reaction (e.g., as
adsorption sites) [9,70,72]. In addition to the fact that the in-
troduction of O-functions provides anchoring sites for catalyst
precursor adsorption and thus the potential for high dispersion,
it also changes the surface charge in some pH windows (hence
the precursor–surface interactions) [34,73] and the surface hy-
drophobic nature [13]. Moreover, the final dispersion also might
be influenced by the thermal stability of the O-functions intro-
duced [34,73].
[Pd(O2CCF3)(py)]+, 222.9 [Pd(O2CCF3)]+, 157.2 [bipy+
H+]+. Diacetato bis(diethylamine) palladium (II) [Pd(OAc)2-
(Et2NH)2] resulted from direct reaction of palladium acetate
[Rocc, Pd 47.27%] with diethylamine [Acros, 99+%] [74–
76] (yield: 66%). IR (KBr disk): ν(NH) 3120, νas(COO)
1586, νs(COO) 1375, ν(CN) 1150, ν(CN) 869, δ(OCO) 686,
ω(COO) 611, ρ(COO) 543, ν(Pd-ligand) 441 cm−1. Elemen-
tal analysis: anal. (calcd.): C: 38.89 (38.87), H: 7.69 (7.61),
N: 7.52 (7.56). MS (EI+): m/z 249.6 [Pd(Et2NH)22+–H+]+,
176.6 [Pd(Et2NH)2+–H+]+, 72.0 [Et2NH–H]+, 57.9 [Et2NH–
CH3]+, 44.0 [Et2NH–CH3CH2]+.
The synthesized complexes were then grafted onto the var-
ious samples of treated carbon supports under the following
conditions. The complex was dissolved in water and stirred with
a fixed amount of carbon at room temperature (unless other-
wise stated) for 24 h. The two bipyridine complexes were used
in equimolar ratios with respect to the amount of acid functions
on the carbon sample used, as determined by Boehm’s titra-
tion (see Section 2.2.1), starting from 1 g of active carbon in
50 mL of distilled water. This would correspond to 3–15 wt%
Pd loading in a final activated carbon-supported material. The
amounts for grafting of [Pd(OAc)2(Et2NH)2] were calculated
to get Pd(5 wt%)/C catalysts (0.1742 g of complex and 0.95 g
of active carbon in 50 mL of distilled water). The grafted mate-
rials were then separated by filtration, abundantly washed with
water, and dried under vacuum in an oven at 70 ◦C for 4 h. The
filtrate was kept for further analysis by atomic absorption spec-
trometry using a Perkin–Elmer 3110 spectrometer fitted with a
flame atomizer (after realization of a calibration curve from 1
to 10 mg/L with standard solutions obtained by dilution of a
1 g/L palladium solution from Acros).
2. Experimental
2.1. Materials
The activated carbon used, called SX+ (acid-washed steam
activated carbon, ash content: 6%; SBET = 922 m2/g; SBJH
=
245 m2/g; pore volume [BJH] = 0.41 cm3/g), was supplied
by NORIT and sieved to keep only the fraction of granulome-
try at 50–100 µm. It was submitted to oxidation treatments in
the liquid phase, with HNO3 [1 mol/L; Aldrich, conc. Acros,
p.a.] or H2O2 [Acros, 35 wt% solution] (each time with 50 mL
of reagent for 2 g of SX+), under variable conditions of re-
actant concentration, duration, and temperature, as specified in
the Results section. After treatment, the functionalized carbon
was filtered out, washed with distilled water on a Soxhlet appa-
ratus for 24 h, and dried under vacuum in an oven at 70 ◦C for
4 h.
Carbon-supported palladium catalysts were prepared us-
ing carboxylate palladium complexes as precursors. The se-
lected complexes were synthesized as reported in the litera-
ture and described below. Diacetato(2,2ꢀ-bipyridine)palladium
(II) [Pd(OAc)2(bipy)] was synthesized from palladium ac-
etate [Rocc, Pd 47.27%] and 2,2ꢀ-bipyridine [Acros, 99+%]
in acetone [Fisher Chemicals, 99.99%] as described previ-
ously [47,74] (yield: 90%). IR (KBr disk): νas(COO) 1625,
ν(CN) 1578, νs(COO) 1369 cm−1. Elemental analysis: anal.
(calcd.): C: 40.82 (44.17), H: 3.98 (3.71), N: 6.46 (7.36). MS
(ESI+): m/z 321.0 [Pd(OAc)(bipy)]+, 293.8 [Pd(OAc)(bipy-
HCN)]+, 262 [PdI(bipy)]+, 157.2 [bipy+H+]+. A second
complex, (2,2ꢀ-bipyridine)bis(trifluoroacetato)palladium (II)
[Pd(CF3COO)2(bipy)], was obtained by ligand exchange [47],
using the synthesized [Pd(OAc)2(bipy)] and trifluoroacetic
acid [Acros, 99%] as reactants (yield: 71%). IR (KBr disk):
νas(COO) 1685, νs(COO) 1425 cm−1. Elemental analysis:
anal. (calcd.): C: 34.24 (34.41), H: 1.44 (1.65), N: 5.00
(5.73). MS (ESI+): m/z 375.0 [Pd(O2CCF3)(bipy)]+, 292.9
The grafted materials thus obtained were submitted to ther-
mal activation in a Carbolite tubular oven under nitrogen and
eventually hydrogen flow. The conditions of temperature and
duration of activation were varied to optimize the metallic dis-
persion and are detailed in Section 3.
2.2. Characterization methods
2.2.1. Boehm’s titration
The titration method developed by Boehm et al. [3] was
implemented to assess the amount of acidic functions on the
carbon surface of the treated support samples. This method in-
volves neutralization of the surface acid functions with excess
amounts of selected bases, followed by back titration with hy-
drochloric acid [3]. In this study, only sodium hydroxide was
used as a base, which allows the neutralization (hence quantifi-
cation) of the sum of all phenol, lactone, and carboxylic acid
groups, leading to a value hereinafter termed “total acidity”

242
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
(in mmol/100 g of carbon). In a typical experiment, 50 mL
of NaOH 0.05 mol/L [FIXANAL, Riedel-de-Haën] is brought
into contact with 0.7 g of carbon, and the mixture is stirred for
24 h at room temperature. The carbon is then filtered out, and
the filtrate is titrated with hydrochloric acid, using phenolph-
thalein as an indicator. All of these steps are carried out under
nitrogen flushing, using freshly distilled decarbonated water.
CO (i.e., the sum of nondetected CO in each pulse), following
VNTP
M
D =
× 100,
VMmCP
where D is the metallic dispersion (%), VNTP is the adsorbed
volume under normal temperature and pressure conditions,
M is the atomic weight of the metal (here Pd), VM is the molar
volume (22.414 L/mol) of the active gas at normal conditions
(0 ◦C, 1 atm), m is the dry sample mass, C is the stoichiometric
coefficient for CO adsorption on Pd (a 1:1 stoichiometry was
used in the present study), and P is the wt% of metal in the
sample (here 5).
Topographic SEM images were obtained using a FEG Dig-
ital Scanning (DMS 982 Gemini LEO) electron microscope
fitted with an EDAX analyzer (Phoenix CDU LEAP). Samples
were stuck using double-face adhesive tape on 5-mm-diameter
aluminum specimen stubs from Agar Scientific.
XRD powder analyses were conducted using a Siemens
D5000 diffractometer equipped with a copper anticathode
(λKα = 154.18 pm). The samples were supported on Si mono-
crystals and analyzed at 15–80◦ with increments of 0.03◦ per
4 s. The detected phases were identified by reference to the
JCPDS database. The metal loading in the activated samples
was determined via ICP-AES (Medac, Egham, UK).
2.2.2. Characterization of carbonaceous materials
X-ray photoelectron spectroscopy (XPS) was carried out on
a SSI-X-probe (SSX-100/206) Fisons spectrometer. The pow-
der samples were pressed on small troughs using double-face
adhesive tape and then placed on an insulating homemade ce-
ramic sample holder (Macor, Switzerland). A nickel grid was
fixed above the samples, and an electron flood gun was set at
8 eV to overcompensate for the positive charging of the sam-
ples during the analyses. The analyzed area was approximately
1.4 mm2, and the pass energy was set at 150 eV. Under these
conditions, the resolution determined by the full width at half
maximum (FWHM) of the Au4f7/2 peak of a clean gold ref-
erence sample was around 1.6 eV. Calibration of the energy
scale was done with reference to the Au4f7/2 peak at 84 eV,
and the binding energies were set by fixing the C1s peak [C–
(C,H) component] at 284.8 eV. Three photopeaks (C1s, O1s,
and N1s) were systematically analyzed for the carbonaceous
materials. The F1s peak was taken into account if necessary,
and the Pd3d peak was added for the catalysts. The XPS results
were decomposed with the CasaXPS software using a sum of
Gaussian/Lorentzian (85/15) after subtraction of a Shirley-type
baseline. The constraints used for decomposition of the Pd3d
peak were as follows: imposing an area ratio Pd3d5/2/Pd3d3/2
2.2.3. Characterization of the Pd complexes
The IR spectra were obtained using a FTIR spectrometer
(Bio-Rad FTS 135) at a resolution of 4 cm−1 with KBr pel-
lets containing 1 wt% of the complexes. The three synthesized
complexes were characterized for C–H–N elemental content by
the microanalysis service of University College, London.
Thermograms were recorded on a Mettler Toledo 851e TGA
system under a 100 mL/min nitrogen flow at a heating rate of
10 ◦C/min, between 20 and 500 ◦C. The sample (∼3 mg) was
placed in an alumina container (70 µL).
of 1.5, a difference in the binding energies (Pd3d3/2–Pd3d5/2
)
of 5.26 eV, and a FWHM ratio (for the Pd3d5/2/Pd3d3/2 peaks)
of 1.
DRIFTS studies were conducted with a Bruker Equinox 55
spectrometer accumulating 5000 scans per spectrum with a res-
olution of 4 cm−1. The detector was of the MCT type. The cell
was coupled with a Balzers QMS 200 mass spectrometer. The
samples were first degassed under helium flow (30 mL/min)
for 1 h at room temperature, in a commercial cell (Spectra-
Tech 0030-102) built in the spectrometer. DRIFTS spectra of
functionalized carbon materials were taken at 120 ◦C to avoid
obstruction of the spectra by H2O. Mass spectra of the grafted
samples were recorded using a heating ramp of 2 or 3 ◦C/min.
The specific surface areas were measured on a Micromerit-
ics ASAP 2000 instrument by nitrogen adsorption at −196 ◦C
(77 K). Before analysis, the sample (0.1 g) was degassed for
several hours at 150 ◦C under 4 mm Hg pressure.
Metallic dispersion measurements were conducted by CO
chemisorption with a Pulse Chemisorb 2700 apparatus from
Micromeritics. The sample (about 30 mg) was first reduced un-
der hydrogen flow (30 mL/min) at 150 ◦C for 1 h, then swept
by a helium flow (30 mL/min) at 200 ◦C for 2 h. The sample
was finally brought back to room temperature, at which point it
was subjected to a sequence of CO pulses. The metallic disper-
sion was calculated from the cumulative amount of adsorbed
A LCQ FINNIGAN MAT or a TSQ 7000 FINNIGAN MAT
mass spectrometer was used to analyze the complexes using
electrospray (ESI) and electronic impact (EI) ionization meth-
ods, respectively.
2.3. Catalytic tests
The reaction used for the catalytic tests is the reduction
of 2-methyl-2-nitropropane (MNP) into t-butylamine (TBA)
(Scheme 1). The tests were carried out in a steel Parr 4521
reactor fitted with a temperature control device and a me-
chanical stirrer fixed at 2500 rpm. In these conditions, extra-
granular diffusion limitations are avoided and a kinetic regime
is guaranteed [70]. First, 375 mL of methanol (Fluka, p.a.
>99.8%), a known quantity (about 0.3 g) of catalyst, and 3 mL
of 1-propanol (chromatographic internal standard; Acros, p.a.
>99%) were poured into the reactor and degassed for 15 min
under nitrogen. The mixture was then submitted to pretreat-
ment at 80 ◦C for 1 h, under 17 bar of H2. The reactant (MNP;
Alfa Aesar, 99%) was dissolved in 25 mL of methanol and de-
gassed (under N2) before being added into the reactor. During
catalytic reaction, the reactor was kept under a pressure of about

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
243
12 bar H2 and a temperature of 60 ◦C. Sampling of the reaction
mixture took place at regular time intervals, and the product
distribution was obtained by gas chromatography; further de-
tails have been provided previously [70].
Table 1
Functionalization of SX+ carbon with HNO or H O
3
2 2
b
Treatment
XPS
O/C
BET
Total acidity
2
a
(mmol/100 g C)
32.0
(m /g)
Temperature
Duration (h)
The chromatograph was an Agilent Technologies 6890N fit-
ted with a WCOT fused silica Chrompack CP-WAX 52 CB cap-
illary column (length, 50 m; internal diameter, 0.53 mm; film
thickness, 2 µm). The vector gas was He (50 kPa), and the tem-
perature of the injector and detector was maintained at 250 ◦C.
The samples were automatically injected via a split–splitless in-
jector with a split flow of 20 mL/min. The temperature program
for analysis was as follows: 120 ◦C for 5 min, heating to 230 ◦C
(30 ◦C/min), isotherm at 230 ◦C. The signal received by a FID
detector was directly interpreted by a computer using the HP
Chemstation software. The surface area of each peak was con-
verted into concentration by applying the response factor with
respect to the internal standard.
The specific activity measured can be expressed by either
the rate of MNP conversion or the rate of TBA formation. In
this study, the catalysts are compared on the basis of the for-
mer, because an apparent zero-order is always observed; hence
the slope is proportional to the catalyst amount. The presented
results are normalized with respect to the mass of catalyst used.
Unmodified SX+ carbon
1
–
–
0.041
922
916
HNO 0.25 mol/L
3
2
3
4
5
6
RT
RT
RT
RT
RT
1
56.9
67.4
60.2
58.7
56.9
55.9
51.3
71.1
80.1
138.5
135.8
119.1
131.0
0.054
0.050
0.054
0.052
0.052
0.052
0.049
0.065
0.063
0.12
24
24
48
72
1/2
1
24
1
24
24
24
24
◦
7
50 C
◦
8
50 C
◦
9
50 C
878
913
10
11
12
13
14
Reflux
Reflux
Reflux
Reflux
Reflux
0.099
0.094
0.085
HNO 1 mol/L
3
15
16
17
18
RT
RT
Reflux
Reflux
24
24
24
24
57.9
66.6
254.7
231.0
0.057
0.066
0.16
905
793
0.13
HNO 2.5 mol/L
3
3. Results and discussion
19
20
RT
Reflux
24
24
75.0
325.6
0.073
0.19
897
756
3.1. Functionalization of SX+ carbon
H O 0.2 mol/L
2
2
21
22
23
RT
RT
RT
1
8
24
28.6
32.5
32.5
0.047
0.048
0.047
The functionalization of SX+ carbon was carried out under
various conditions. The concentration (0.25–2.5 mol/L), tem-
perature (room temperature to reflux), and duration (1–72 h)
were varied for the oxidative treatments with HNO3, whereas
the concentration (0.2 or 11.68 mol/L) and duration (1–24 h)
were varied for treatments with H2O2. The support was charac-
terized by several methods to evaluate the extent of functional-
ization. Boehm’s titrations and XPS were used to quantify the
obtained acidity, whereas DRIFTS was applied to determine the
nature of the introduced oxygenated functions. In addition, the
textural state of the modified supports was explored through
BET.
The results of Boehm’s titrations presented in Table 1 show
that the unmodified SX+ carbon (line 1 in Table 1) has some
acid functions on its surface. Any treatment with HNO3 or con-
centrated H2O2 provokes an increase in the number of surface
oxygenated groups. The formation and relative quantities of
these groups were influenced by the nature and concentration
of the reactant used, as well as by the temperature and dura-
tion in cases of treatment with HNO3. It appears that treatment
with HNO3 (lines 2–20) has a greater effect on total acidity
than a treatment with H2O2 (lines 21–26), as mentioned in the
literature [10,17,27,29,44,50,51,54,55]. Moreover, with HNO3,
temperature and concentration seem to have more effect than
duration on the total acidity. More precisely, increasing the tem-
perature to 50 ◦C (lines 7–9) did not have a major influence,
but heating under reflux (lines 10–14, 17, 18, and 20) consid-
erably enhanced the surface acidity. The effects of HNO3 con-
centration [45] and treatment temperature [6,46] on the number
927
904
H O 11.68 mol/L
2
2
24
25
26
RT
RT
RT
1
8
24
48.7
52.9
48.4
0.058
0.059
0.055
a
RT = room temperature.
b
Determined by Boehm’s titration.
of oxygen functions introduced have also been reported previ-
ously.
These titration results were confirmed by determination of
O/C surface ratios by XPS (Table 1). As expected, an increase
in acidity (as determined by Boehm’s titrations), correspond-
ing to an increase in surface oxygenated functions, was related
to higher O/C ratios (measured by XPS). More specifically, a
linear correlation (Fig. 1) was found between the total acidity
measured by Boehm’s titration and the XPS results (O/C sur-
face ratios). For most of the samples, the N1s zone contained
no signal, indicating that no residue of nitric acid remained on
the carbon. Reproducibility and repeatability were verified for
both Boehm’s titration and XPS.
The unmodified SX+ carbon has a BET surface area of
about 922 m2/g. When treated with HNO3, its specific sur-
face tends to decrease slightly, especially at high temperatures
and concentrations (Table 1). This finding is in agreement with
the literature, because it has already been shown that liquid-
phase HNO3 oxidation of carbon does not significantly modify
its textural properties [17,21,40,44,49,52,53], except for severe

244
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
Table 2
IR (main bands) of functionalized carbons
Wavenumber
Functions
−1
(cm
)
1850–1700
1650–1550
ν(C=O): anhydride, lactone, ketone, carboxylic acid
ν(C=O): quinone, amide, carboxylate
ν(C=C): aromatics
1450–1150
950–850
ν(C–O): ether, carboxylic acid, ester, lactone, phenol
ν(C–O): cyclic ether
ν(CO–O–CO): aromatic cyclic anhydride (5-membered)
δ(OH): carboxylic acid
ν(O–O): peroxide
Fig. 1. Linear correlation found between XPS results (O/C surface ratios) and
usually creates weak acid sites rather than carboxylic acids on
the surface [13,23,24,26,30,32].
total acidity (measured by Boehm’s titration) for carbons modified with HNO
3
(F) or H O (2) and unmodified SX+ carbon (Q).
2
2
To summarize, the treatment of SX+ carbon with H2O2
brought about a smaller increase in the oxygenated surface
groups than with HNO3. Moreover, the nature of introduced
groups was variable depending on the reactant. The most highly
functionalized material (with HNO3 2.5 mol/L, for 24 h un-
der reflux) displayed a total acidity of 325.6 mmol/100 g and
only a slightly smaller surface of ∼700 m2/g, corresponding to
2.8 functions/nm2. This compares well with previous studies
in which values up to 1–3 acidic groups/nm2 were found for
highly oxidized carbon nanofibers [77].
3.2. Grafting of Pd complexes onto functionalized carbon
The oxygenated functions introduced on the support were
used as anchors for the grafting of the synthesized metal-
lic precursors. The grafting mechanism, consisting of ligand
exchange, was first studied using [Pd(OAc)2(bipy)] and [Pd-
(O2CCF3)2(bipy)]. Then Pd(5 wt%)/C catalysts were prepared
using [Pd(OAc)2(Et2NH)2].
Fig. 2. DRIFTS spectra for (a) unmodified SX+ carbon, (b) C treated with
HNO 0.25 mol/L under reflux, (c) C treated with HNO 1 mol/L under reflux,
3
3
(d) C treated with HNO 2.5 mol/L under reflux, and (e) C treated with H O
3
2 2
11.68 mol/L. (Note: to allow comparisons, the duration of all treatments here
was taken as 24 h.)
treatments [27,39,43,47–52]. Table 1 shows also that modifi-
cation with H2O2 brought no change in specific surface area,
again in agreement with the literature [25].
3.2.1. Synthesis and characterization of the complexes
The compounds [Pd(OAc)2(bipy)], [Pd(O2CCF3)2(bipy)],
and [Pd(OAc)2(Et2NH)2] were prepared as described in the lit-
erature [47,74,75], and characterized by IR, elemental analysis,
TGA, and mass spectrometry. In their IR spectra, the νs(COO)
and νas(COO) bands were particularly interesting, because the
separation between them (ꢀν(as-s)) varied with the coordina-
tion mode of carboxylate ligands [78]: ꢀν(as-s) ≈ 150 cm−1 for
a chelating and ꢀν(as-s) > 200 cm−1 for an unidentate coordi-
nation mode. These considerations allowed us to determine that
in the starting complex [Pd(OAc)2], both acetate ligands are co-
ordinated in a chelating mode, whereas they become unidentate
in the three synthesized complexes [Pd(O2CR)2(L)] (O2CR =
carboxylate, L = bipy or (Et2NH)2), in agreement with the lit-
erature [47,74–76,78].
Results from TGA analyses of the synthesized complexes,
carried out under nitrogen, are given in Table 3. The compari-
son between calculated and experimental mass losses indicates
that on heating, the complex [Pd(OAc)2(Et2NH)2] lost all of its
ligands. In contrast, the ligands loss for bipyridine complexes
was not complete, probably due to the polyaromatic nature of
Five DRIFTS spectra of carbons with radically different
acidities were recorded. These spectra display (as shown in
Fig. 2) large bands attributable to the oxygenated functions.
Although there is no general consensus in the literature, the
assignment of the main IR bands associated with O-groups
in functionalized carbons is summarized in Table 2 [1,6,8,23,
38,52]. The spectrum for the unmodified carbon showed weak
bands at around 1600, 1200, and 900 cm−1, corresponding to
ν(C=O), ν(C–O), and ν(C–O) or ν(CO–O–CO), based on the
assignments shown in Table 2. An increase in the oxygen con-
tent was clearly visible when the carbon was modified with
HNO3; the bands grew and lengthened, and a new band ap-
peared at 1800 cm−1 [ν(C=O)]. This was expected, because
nitric acid treatment is known to increase mainly the amount of
carboxylic acid groups on carbon surfaces [3,7–9,13,14,18,20,
29–31,41–44]. The spectrum of the carbon modified with H2O2
was identical to that of SX+ apart from the amplification of the
band at 900 cm−1—not surprisingly, because H2O2 oxidation

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
245
Table 3
TGA analyses of the synthesized Pd complexes (under nitrogen)
Complex
Calculated mass loss (%)
Based on a PdO residue
Experimental
mass loss (%)
Final decomposition
temperature ( C)
◦
Based on a Pd residue
[Pd(OAc) (bipy)]
67.8
74.9
67.0
72.0
78.2
71.3
59.7
46.1
73.3
240
270
170
2
[Pd(O CCF ) (bipy)]
2
3 2
[Pd(OAc) (Et NH) ]
2
2
2
the bipy ligand. These observations justified our choice to use
[Pd(OAc)2(Et2NH)2] only as a catalyst precursor.
The mass spectra recorded for each Pd compound show
peaks corresponding to representative fragments arising from
losses of ligands or fragments of them, as well as proton addi-
tion, and high mass entities corresponding to dimers, trimers,
and lightly bound adducts.
3.2.2. Grafting of [Pd(OAc)2(bipy)] or [Pd(O2CCF3)2(bipy)]
onto functionalized carbon
Scheme 2. Most probable grafting model for [Pd(O CR) (bipy)] complexes
onto functionalized carbon.
2
2
The advantages of these two complexes were linked to their
solubility in water and the ease of interpretation because they
presented nitrogen (and fluorine) atoms. But these complexes
also had a disadvantage—results from TGA analysis showed
that they could not be used for catalysis if thermal activation
was used, because the bipyridine ligands could not be com-
pletely removed on heating. For these reasons, these complexes
were used only to investigate the grafting mechanism of car-
boxylate complexes onto functionalized carbon.
The procedure for grafting [Pd(O2CR)2(bipy)] compounds
on carbon samples containing surface oxygenated functions
is believed to proceed via substitution of the carboxylate lig-
ands by surface –COOH groups according to a ligand-exchange
mechanism. Among the possible resulting structures, one was
thought to be the most probable (Scheme 2). Indeed, the pres-
ence of a N-donor chelating ligand limits the number of pos-
sible reactions by favoring substitution of the carboxylate lig-
ands. This hypothesis was confirmed by XPS (see below).
A series of samples was prepared by reacting various batches
of functionalized carbon with equimolar quantities of complex
with respect to the amount of acid functions determined by titra-
tion. The influence of the initial carbon acidity, as well as the
effect of the temperature during the grafting procedure, were
studied. Table 4 shows that the higher the initial carbon acid-
ity, the more the experimental Pd/C ratio differed from the
calculated Pd/C value. This means that at high loading, all of
the palladium introduced was not quantitatively grafted, due
to a saturation effect of the acid functions, perhaps related to
their proximity. These results were confirmed by atomic ab-
sorption analysis of the filtrates (last column of Table 4), which
showed that some residual Pd remained in solution at high
loading. When raising the grafting temperature (Table 4, com-
pare lines 5, 6, and 7), the experimental XPS Pd/C ratios in-
creased slightly, but the percentage of introduced Pd (measured
by atomic absorption analysis of the filtrates) diminished. Thus,
it seems that increasing the temperature allowed better disper-
sion (higher surface Pd concentration) but was detrimental to
the total amount adsorbed. The Pd/N ratios were close to 0.5 in
Table 4
Grafting of [Pd(OAc) (bipy)] (1 to 7) or [Pd(O CCF ) (bipy)] (8 and 9) on SX+ carbon and carbon functionalized with HNO
2
2
3 2
3
Initial carbon acidity
(mmol/100 g C)
Grafting
Pd/N (XPS)
Pd/C × 100 (XPS)
Relative amount of Pd
d
b
c
adsorbed on C (%)
Temperature
Duration (h)
Calculated
Experimental
1
2
3
4
32.0
60.2
138.5
138.5
RT
RT
RT
RT
24
24
24
24
0.52
0.65
0.65
0.65
0.38
0.72
1.66
1.66
0.41
0.74
1.29
1.19
99.99
99.28
70.39
–
a
5
6
7
135.8
135.8
135.8
RT
50 C
100 C
24
24
24
0.62
0.60
0.74
1.63
1.63
1.63
1.02
1.14
1.38
84.03
73.32
66.95
◦
◦
8
9
138.5
138.5
RT
RT
24
24
0.65
0.61
1.66
1.66
1.26
1.05
62.04
–
a
a
b
c
d
Samples washed thoroughly with distilled water in a Soxhlet apparatus after the grafting procedure.
RT = room temperature.
Values calculated by considering 100% grafting of the Pd complex.
The amount of nonadsorbed Pd was measured in the synthesis filtrates by atomic absorption, which allowed to calculate the amount of adsorbed Pd by difference
(here expressed in terms of % of amount introduced initially).

246
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
most cases, confirming the hypothesis that the acetate ligands
were exchanged for surface groups during the grafting proce-
dure, while the bipyridine ligand remained linked to the palla-
dium, as shown in the model displayed in Scheme 2. The Pd/N
ratio became significantly higher for sample 7 (grafted under
reflux), suggesting that the bipyridine ligand was more easily
exchanged (or even destroyed) at higher temperature. Hence, in
the remainder of this study, the grafting temperature was fixed
at room temperature only.
Table 5
Grafting of [Pd(OAc) (Et NH) ] onto SX+ and functionalized carbon samples
2
2
2
Initial carbon acidity XPS
(mmol/100 g C)
Relative amount of Pd
d
adsorbed on C (%)
Pd/N Pd/C (×100)
a
b
b
b
b
c
1
2
3
4
5
6
32.0
72.0
1.01
1.42
0.98
0.65
0.47
1.04
0.68
0.62
0.82
0.90
0.81
0.71
99.39
99.47
99.45
99.71
100.00
99.83
119.1
254.7
325.6
52.9
Comparing the results obtained for samples 3 and 4 (Ta-
ble 4) indicated that the palladium species was really bound
chemically on the carbon surface, because the XPS values af-
ter thorough washing in a Soxhlet apparatus varied very little.
In the same way, grafting of [Pd(O2CCF3)2(bipy)] was carried
out to confirm that chemical links between the palladium com-
plex and the carbon support could be achieved. Samples 8 and 9
gave rise to F/Pd ratios of 0.57 and 0.19, respectively. Thus, the
F/Pd ratio decreased with abundant washing, but the Pd/C and
Pd/N ratios remained very close in these two samples (Table 4),
indicating that the “Pd(bipy)” fragment was actually grafted,
whereas the exchanged trifluoroacetate ligands remained ad-
sorbed on the surface in sample 8 and could be washed out in
sample 9.
a
b
c
d
Nonfunctionalized carbon.
C sample functionalized with HNO .
C sample functionalized with H O .
The amount of nonadsorbed Pd was measured in the synthesis filtrates by
3
2
2
atomic absorption, which allowed to calculate the amount of adsorbed Pd by
difference (here expressed in terms of % of amount introduced initially). In this
case, 100% of palladium introduced corresponds to a final activated catalyst
with a 5 wt% Pd loading.
ble 5. Low Pd/C ratios were obtained. To explore a possible
evolution of the Pd dispersion with the carbon acidity, CO
chemisorption measurements were implemented after activa-
tion (see below). The Pd/N ratios measured on the surface af-
ter grafting of [Pd(OAc)2(Et2NH)2] were variable, indicating
that the ligand-exchange reactions were not as selective as for
[Pd(OAc)2(bipy)] or [Pd(O2CCF3)2(bipy)]. Here several ex-
change mechanisms are plausible, leading to different surface
structures, as presented in Scheme 3. Based on the measured
Pd/N ratios, hypothesis (a) was believed to be the predominant
model. Other schemes are also possible—for example, taking
into account an exchange mechanism involving two surface
acidic functions or the occurrence of a cis/trans rearrangement
in solution, leading to grafted fragments such as those shown in
Scheme 3b or 3c.
3.2.3. Grafting of [Pd(OAc)2(Et2NH)2]
The complex [Pd(OAc)2(Et2NH)2], according to its crystal
structure [79], is obtained predominantly in the trans form. This
could preclude a clean carboxylate-exchange grafting mecha-
nism, as the N-donor ligands (being nonchelating) are also po-
tentially substituted. Due to its neat TGA and elemental analy-
sis, the complex was nevertheless used to prepare Pd(5 wt%)/C
catalysts using functionalized carbon samples of various acidi-
ties. The results obtained after grafting are summarized in Ta-
Scheme 3. Possible models for grafting of [Pd(OAc) (Et NH) ] onto functionalized carbon: (a) exchange of one acetate and one diethylamine ligands for a surface
2
2
2
–COOH group, (b) exchange of two acetate ligands, (c) exchange of two diethylamine ligands, (d) exchange of one diethylamine only, (e) exchange of one acetate
only, (f) exchange of all four ligands for two –COOH groups inside a pore. Models (b) and (c) imply the occurrence of a cis/trans rearrangement of the complex.

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
247
The Pd loading obtained by atomic absorption analysis of
the filtrates, shown in the last column of Table 5, shows that
all of the introduced palladium (corresponding to 5 wt% in the
final catalyst) had been incorporated on the carbon support in
each case. Thus, the resulting samples might be used in catal-
ysis (after activation) and be compared directly. It should be
noted that for the treated carbon samples, the amount of sur-
face functions present on the surface is in all cases sufficient to
graft the totality of the introduced Pd (corresponding to 5 wt%
in the final catalyst). It is only in the case of the untreated sup-
port (SX+) that the amount of surface functions is slightly too
low to guarantee 100% grafting of the Pd; hence some amount
of metal might be only adsorbed. However, we believe that
we have achieved real grafting of molecular metallic fragments
onto the surface of the functionalized carbon supports by the
O-groups acting as anchors for the complex precursors. This
can be related to recent studies dealing with covalent grafting
of coordination compounds on carbon supports [62–66], where
the grafted moieties are then used as “intact” fragments for
catalysis, which could also be an interesting field of applica-
tion for the present study. However, we have chosen to use the
grafted molecules as precursors for the formation, by thermal
activation, of supported metallic particles of controlled size and
dispersion, in parallel to another study dealing with the anchor-
ing of [(C3H5)Pd(C5H5)] onto oxidized diamond to produce
small Pd particles via reduction [60].
(a)
3.3. Activation of materials
To optimize the activation step, the results from TGA analy-
sis of the starting complex [Pd(OAc)2(Et2NH)2], which showed
that the total loss of ligands occurred at 200 ◦C (after about
20 min), were taken into account. In addition, DRIFTS/MS
(Fig. 3) was implemented as an in situ method to follow the
departure of ligands during thermal activation. Indeed, heating
of functionalized carbon samples grafted with Pd complexes
led to desorption of ligands as typical fragments [80], which
were observed by mass spectrometry (MS). These MS results
confirmed the previously established models: [Pd(OAc)2(bipy)]
grafted on carbon gave rise to fragments only from bipyridine,
whereas [Pd(OAc)2(Et2NH)2] gave rise to fragments from both
ligands. This in situ DRIFTS/MS characterization of the grafted
samples allowed us to follow the desorption of ligands from the
surface, starting at 150 ◦C, confirming the TGA results on the
“pure” unsupported complexes. Taking these facts into account,
two activation temperatures were selected for comparison: 200
and 500 ◦C.
First, activation of the grafted samples [Pd(OAc)x(Et2-
NH)y]/C was carried out in a tubular oven under nitrogen flow
at 500 ◦C for 18 h (activation I). The activated catalysts thus
obtained were characterized by CO chemisorption (Table 6;
Fig. 4); the Pd dispersion values thus obtained increased with
the initial carbon acidity in a nonlinear way, reaching a plateau
for highly functionalized carbons, whereas the sample modified
with H2O2 displayed a surprisingly high Pd dispersion value
relative to its carbon acidity count. However, the Pd dispersion
values were rather low, and the XPS Pd/C ratios before activa-
(b)
Fig. 3. In situ mass spectrometric characterization of (a) [Pd(OAc) (bipy)] and
2
(b) [Pd(OAc) (Et NH) ] grafted onto functionalized carbon.
2
2
2
tion were higher than those after activation. Thus, it seemed that
the activation step at 500 ◦C was responsible for agglomeration
of palladium atoms on the carbon surface. Indeed, sintering is
favored by high temperatures [81]. Moreover, it is known that
the oxygenated surface functions of oxidized carbons can be re-
moved by thermal treatment at high temperature, and that the
least resistant functions are the carboxylic acid groups [3,6–10],
which in this study have been chosen as anchors for the metallic
precursor molecules.
Second, the lower temperature activation was implemented
as follows. Activation II was carried out at 200 ◦C for 4 h, in
some cases followed by treatment under hydrogen at 200 ◦C for
2 h (activation III). This hydrogen treatment was added to en-
sure complete reduction of Pd, as the catalysts are to be used
in a hydrogenation reaction. Characterization of the samples
submitted to activations II and III by SEM, XRD (Fig. 5), and
CO chemisorption (Table 6) showed that the dispersion was im-
proved compared with that of the samples activated at 500 ◦C.
In particular, the formation of Pd particles of a nonnegligible
size (∼25 nm), which were visible by SEM (bright dots in
Fig. 5a) and gave rise to sharp X-ray diffraction peaks of Pd
(Fig. 5c) for samples activated at 500 ◦C, was hindered by a
lower-temperature activation (less bright dots observed by SEM

248
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
Table 6
Pd dispersion measured by CO chemisorption after activation and average particle size in nm (for detail of calculation, see text)
Initial carbon acidity
After activation I
After activation II
After activation III
(mmol/100 g C)
Pd dispersion
(%)
Average particle size
(nm)
Pd dispersion
(%)
Average particle size
(nm)
Pd dispersion
(%)
Average particle size
(nm)
32.0
2.2
3.2
4.2
5.3
3.8
63
43
32
26
36
–
–
–
13.6
10.8
–
–
–
10
13
–
–
–
17.0
15.3
–
–
–
8
9
a
72.0
119.1
254.7
a
a
b
52.9
a
C sample functionalized with HNO .
3
b
C sample functionalized with H O .
2
2
Table 7
Catalytic results in terms of conversion rate of MNP
Initial carbon acidity
(mmol/100 g)
Activation
Pd dispersion
(%)
Activity
(mol/(L min g))
32.0
72.0
I
I
I
III
I
2.29
3.20
5.32
17.02
3.76
0.00114
0.00122
0.00154
0.00279
0.00154
254.7
231.0
a
52.9
a
C sample functionalized with H O .
2
2
Scheme 1). This involved determining the conversion rate
of MNP (Table 7). Two intermediates are produced in this
reaction—2-methyl-2-nitrosopropane and t-butylhydroxyla-
mine—but the nitroso compound is very reactive and thus accu-
mulates very little [70]. The values of activity obtained here are
in the range 1×10−3–3×10−3 mol/(Lming), which compares
well with the results obtained under similar testing conditions
with a commercial Pd(5%)/C catalyst from Johnson Matthey
(JM 487), which gave a value of 1.51 × 10−3 (Pd dispersion:
12.3%) [82]. However, much better results were obtained with a
Pd(5%)/C catalyst from Degussa (E10N), which gave a value of
8.68 × 10−3 mol/(ming) (Pd dispersion, 11.2%), again under
similar experimental conditions. The catalytic results for vari-
ous samples are also shown in Figs. 6 and 7, as functions of the
initial carbon acidity and Pd dispersion, respectively. The spe-
cific activity increased linearly with the initial carbon acidity
when activation I and HNO3-modified carbon were used. Ac-
tivation III gave better performance, as well as H2O2-modified
carbon. These observations seemed related mainly to an effect
of Pd dispersion on the activity (see Fig. 7); however, the H2O2-
modified carbon still led to a slightly more active catalyst. This
must be due not to the number of oxygenated functions intro-
duced initially on the surface, but to their nature, because it has
already been shown that they are not innocent in this kind of
reaction [70,72]. The O-groups introduced on the carbon sup-
port in this study thus played a double role: serving as anchors
for the grafting of Pd precursors, thus increasing the disper-
sion, and playing an active role during the catalytic process that
remains to be investigated in greater detail.
Fig. 4. Evolution of Pd dispersion (after activation I) with initial carbon acidity:
(2) carbon modified with H O , (F) carbon modified with HNO .
2
2
3
but Pd signal by EDXS (see Fig. 5b) and broad Pd XRD peaks
(see Fig. 5d)). Moreover, the samples activated under hydrogen
(activation III) displayed sharper XPS Pd photopeaks, corre-
sponding, as expected, to a more reduced surface state of the
metal. The values of dispersion measured by CO chemisorp-
tion (Table 6) allow us to estimate an average particle size,
which is 26–63 nm for the samples activated at 500 ◦C and 8–
13 nm for the samples activated at 200 ◦C, in agreement with
SEM characterization. These values of particle sizes were cal-
culated by taking a cubic model for the Pd particles, which
gives rise to the following expression: dispersion = 5/a, where
a = particle size = length of cube edge expressed in num-
ber of Pd atoms (we have taken the radius of a Pd atom as
0.137 nm) [82]. The fact that the highest dispersion obtained
in this work is only ∼17% can again be ascribed to the ther-
mal fragility of the carboxylic acid functions used as anchors,
even at temperatures as low as 200 ◦C. Indeed, if these functions
were destroyed thermally, then metal agglomeration would be
favored. Finally, the 5 wt% Pd loading in these samples derived
from [Pd(OAc)x(Et2NH)y]/C was verified after activation by
ICP-AES.
3.4. Catalytic results
To quantify the catalytic performance of the Pd(5 wt%)/C
materials obtained from the activated [Pd(OAc)x(Et2NH)y]/C,
the specific activity in the hydrogenation of 2-methyl-2-nitro-
propane (MNP) into t-butylamine (TBA) was measured (see
4. Conclusion
In this work, we have shown that it is possible to use a con-
tinuous approach for carbon-supported catalyst preparation by

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
249
◦
◦
Fig. 5. SEM and XRD characterization of samples activated (a, c) at 500 C for 18 h and (b, d) at 200 C for 1 h, and prepared from carbon modified with HNO
3
1 mol/L under reflux for 24 h. Magnification: (a) ×50,000; (b) ×100,000. The vertical lines in the XRD diffractograms correspond to the expected lines for
metallic Pd.
Fig. 7. Evolution of specific activity with Pd dispersion: (F) activation I and
Fig. 6. Evolution of specific activity with initial carbon acidity: (F) activation I
carbon modified with HNO , (2) activation III and carbon modified with
3
and carbon modified with HNO , (2) activation III and carbon modified with
HNO , (Q) activation I and carbon modified with H O .
3
3
2 2
HNO , (Q) activation I and carbon modified with H O .
3
2 2
but rather to provide materials that could be characterized at
each step of their genesis, thus providing mechanistic insight
and room for optimization.
Surface functionalization of SX+ carbon was achieved us-
ing HNO3 or H2O2. The amount of oxygenated surface func-
tions introduced, as well as their nature, were determined
using Boehm’s titration, XPS, and DRIFTS characterization
developing a controlled methodology for the synthesis of Pd/C
materials. We have exploited the possibility offered by carbona-
ceous materials of modifying the number of surface acid sites,
then using these for grafting coordination compounds selected
as suitable metallic precursors. The described synthetic proce-
dure was aimed not to develop a commercially viable process,

250
S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
methods. The nature of the introduced functions was shown
to be different for each reactant, and their amount was in-
creased by temperature and concentration. The surface func-
tions were then used as anchors for selected Pd carboxy-
late complexes of general formula [Pd(O2CR)2(N-ligand)x].
The ligand-exchange mechanisms were studied by XPS, and
the prevailing models for the grafted fragments were estab-
lished to be “Pd(bipy)” or “Pd(OAc)(Et2NH)” when starting
from [Pd(O2CCR3)2(bipy)] or [Pd(OAc)2(Et2NH)2], respec-
tively. Monometallic Pd(5 wt%)/C catalysts were prepared by
grafting [Pd(OAc)2(Et2NH)2] onto various samples of func-
tionalized carbon. The activation step was optimized (tempera-
ture and duration) on the basis of TGA and MS analyses to give
well-dispersed samples, which were characterized by XRD,
SEM, and CO chemisorption. The catalytic performance in the
hydrogenation of MNP for HNO3-modified samples was re-
lated to the initial carbon acidity (for selected activation condi-
tions) and to Pd dispersion (whatever the activation conditions),
whereas they were unexpectedly improved for H2O2-modified
samples, probably due to the different types of oxygenated sur-
face functions introduced (or variable ratios of the same types),
which are thought to play an active role in the catalytic process
besides serving as anchors during catalyst preparation.
[16] H.-P. Boehm, E. Diehl, W. Heck, R. Sappok, Angew. Chem. Int. Ed. 3
(1964) 669.
[17] R.C. Sosa, R.F. Parton, P.E. Neys, O. Lardinois, P.A. Jacobs, P.G. Rouxhet,
J. Mol. Catal. A Chem. 110 (1996) 141.
[18] G. Farkas, L. Hegedus, A. Tungler, T. Mathe, J.L. Figueiredo, M. Freitas,
J. Mol. Catal. A Chem. 153 (2000) 215.
[19] M. Gurrath, T. Kuretzky, H.P. Boehm, L.B. Okhlopkova, A.S. Lisitsyn,
V.A. Likholobov, Carbon 38 (2000) 1241.
[20] A.E. Aksoylu, M. Madalena, A. Freitas, M.F.R. Pereira, J.L. Figueiredo,
Carbon 39 (2001) 175.
[21] M.A. Fraga, E. Jordao, M.J. Mendes, M.M.A. Freitas, J.L. Faria, J.L.
Figueiredo, J. Catal. 209 (2002) 355.
[22] P.E. Fanning, M.A. Vannice, Carbon 31 (1993) 721.
[23] M. Domingo-Garcia, F.J. Lopez-Garzon, M. Perez-Mendoza, J. Colloid
Interface Sci. 222 (2000) 233.
[24] M. Perez-Mendoza, M. Domingo-Garcia, F.J. Lopez-Garzon, Appl. Catal.
A 224 (2002) 239.
[25] F.J. Lopez-Garzon, M. Domingo-Garcia, M. Perez-Mendoza, P.M. Al-
varez, V. Gomez-Serrano, Langmuir 19 (2003) 2838.
[26] G.C. Torres, E.L. Jablonski, G.T. Baronetti, A.A. Castro, S.R. de Miguel,
O.A. Scelza, M.D. Blanco, M.A.P. Jimenez, J.L.G. Fierro, Appl. Catal.
A 161 (1997) 213.
[27] C. Moreno-Castilla, M.A. Ferro-Garcia, J.P. Joly, I. Bautista-Toledo,
F. Carrasco-Marin, J. Rivera-Utrilla, Langmuir 11 (1995) 4386.
[28] C. Moreno-Castilla, F. Carrasco-Marin, A. Mueden, Carbon 35 (1997)
1619.
[29] B.K. Pradhan, N.K. Sandle, Carbon 37 (1999) 1323.
[30] M. Santiago, F. Stuber, A. Fortuny, A. Fabregat, J. Font, Carbon 43 (2005)
2134.
[31] P. Vinke, M. Vandereijk, M. Verbree, A.F. Voskamp, H. Vanbekkum, Car-
bon 32 (1994) 675.
Acknowledgments
The authors thank R. Rozenberg and J.-L. Habib Jiwan for
mass spectrometric characterization of the carboxylate com-
plexes and useful discussions, A. Iserentant for help with metal
loading determinations, J.-F. Statsyns for technical support, and
D. Desmecht and P. Verhasselt for assistance during the cat-
alytic tests. The NORIT firm is also acknowledged for supply-
ing the SX+ carbon support. Financial support was provided by
the Belgian National Fund for Scientific Research (FNRS) and
the Communauté Française de Belgique (through an “ARC”
concerted research program). Moreover, O.D. acknowledges fi-
nancial support from the “Concorde” package (EC-funded) and
the Belgian FRIA.
[32] M. Acedo-Ramos, V. Gomez-Serrano, C. Valenzuela-Calahorro, A.J. Lo-
pez-Peinado, Spectrosc. Lett. 26 (1993) 1117.
[33] H.-P. Boehm, Angew. Chem. Int. Ed. 5 (1966) 533.
[34] C.A.L.Y. Leon, L.R. Radovic, Chem. Phys. Carb. 24 (1994) 213.
[35] J.M. O’Reilly, R.A. Mosher, Carbon 21 (1983) 47.
[36] I.I. Salame, T.J. Bandosz, J. Colloid Interface Sci. 240 (2001) 252.
[37] K. Laszlo, K. Josepovits, E. Tombacz, Anal. Sci. 17 (2001) il741.
[38] M. Domingo-Garcia, F.J.L. Garzon, M.J. Perez-Mendoza, J. Colloid Inter-
face Sci. 248 (2002) 116.
[39] T. Cordero, J. Rodriguez-Mirasol, N. Tancredi, J. Piriz, G. Vivo, J.J. Ro-
driguez, Ind. Eng. Chem. Res. 41 (2002) 6042.
[40] J.S. Noh, J.A. Schwarz, Carbon 28 (1990) 675.
[41] J.M. Solar, C.A.L. Leon, K. Osseoasare, L.R. Radovic, Carbon 28 (1990)
369.
[42] V. Strelko, D.J. Malik, J. Colloid Interface Sci. 250 (2002) 213.
[43] A.N.A. El-Hendawy, Carbon 41 (2003) 713.
References
[44] G. Soto-Garrido, C. Aguilar, R. Garcia, R. Arriagada, J. Chil. Chem.
Soc. 48 (2003) 65.
[45] J.Y. Li, L. Ma, X.N. Li, C.S. Lu, H.Z. Liu, Ind. Eng. Chem. Res. 44 (2005)
5478.
[46] S.S. Barton, M.J.B. Evans, E. Halliop, J.A.F. MacDonald, Carbon 35
(1997) 1361.
[47] B. Milani, E. Alessio, G. Mestroni, A. Sommazzi, F. Garbassi, E. Zan-
grando, N. Brescianipahor, L. Randaccio, J. Chem. Soc. Dalton Trans.
(1994) 1903.
[1] R. Schlögl, Handbook of Heterogeneous Catalysis, VCH, 1997, Ch. 2.
[2] E. Auer, A. Freund, J. Pietsch, T. Tacke, Appl. Catal. A 173 (1998) 259.
[3] A. Polania, E. Papirer, J.B. Donnet, G. Dagois, Carbon 31 (1993) 473.
[4] A. Ahmadpour, D.D. Do, Carbon 34 (1996) 471.
[5] C. Moreno-Castilla, J. Rivera-Utrilla, MRS Bull. 26 (2001) 890.
[6] G. de la Puente, J.J. Pis, J.A. Menendez, P. Grange, J. Anal. Appl. Pyrol. 43
(1997) 125.
[7] C. Moreno-Castilla, F. Carrasco-Marin, F.J. Maldonado-Hodar, J. Rivera-
Utrilla, Carbon 36 (1998) 145.
[48] N. Krishnankutty, M.A. Vannice, Chem. Mater. 7 (1995) 754.
[49] H. Tamon, M. Okazaki, Carbon 34 (1996) 741.
[50] A. Gil, G. de la Puente, P. Grange, Micropor. Mater. 12 (1997) 51.
[51] J. Choma, W. Burakiewicz-Mortka, M. Jaroniec, Z. Li, J. Klinik, J. Colloid
Interface Sci. 214 (1999) 438.
[52] V. Dubois, Y. Dal, G. Jannes, Stud. Surf. Sci. Catal. 143 (2002) 993.
[53] L. Calvo, M.A. Gilarranz, J.A. Casas, A.F. Mohedano, J.J. Rodriguez, Ind.
Eng. Chem. Res. 44 (2005) 6661.
[8] J.L. Figueiredo, M.F.R. Pereira, M.M.A. Freitas, J.J.M. Orfao, Carbon 37
(1999) 1379.
[9] M.A. Fraga, M.J. Mendes, E. Jordao, J. Mol. Catal. A Chem. 179 (2002)
243.
[10] P.C.C. Faria, J.J.M. Orfao, M.F.R. Pereira, Water Res. 38 (2004) 2043.
[11] H.P. Boehm, Carbon 32 (1994) 759.
[12] P. Ehrburger, O.P. Mahajan, P.L. Walker, J. Catal. 43 (1976) 61.
[13] D.J. Suh, T.J. Park, S.K. Ihm, Carbon 31 (1993) 427.
[14] Y. Otake, R.G. Jenkins, Carbon 31 (1993) 109.
[15] E. Papirer, R. Lacroix, J.-B. Donnet, Carbon 34 (1996) 1521.
[54] E. Ahumada, H. Lizama, F. Orellana, C. Suarez, A. Huidobro, A. Sepul-
veda-Escribano, F. Rodriguez-Reinoso, Carbon 40 (2002) 2827.

S. Hermans et al. / Journal of Catalysis 243 (2006) 239–251
251
[55] M.F.R. Pereira, S.F. Soares, J.J.M. Orfao, J.L. Figueiredo, Carbon 41
(2003) 811.
[69] G. Carturan, G. Facchin, G. Cocco, G. Navazio, G. Gubitosa, J. Catal. 82
(1983) 56.
[56] C. Prado-Burguete, A. Linares-Solano, F. Rodriguez-Reinoso, C.S. Dele-
cea, J. Catal. 115 (1989) 98.
[70] V. Dubois, G. Jannes, P. Verhasselt, Stud. Surf. Sci. Catal. 108 (1997)
263.
[57] V. Gomez-Serrano, M. Acedo-Ramos, A.J. Lopez-Peinado, C. Valenzuela-
Calahorro, Fuel 73 (1994) 387.
[71] F.J. Derbyshire, V.H.J. de Beer, G.M.K. Abotsi, A.W. Scaroni, J.M. Solar,
D.J. Skrovanek, Appl. Catal. 27 (1986) 117.
[58] N. Krishnankutty, M.A. Vannice, J. Catal. 155 (1995) 312.
[59] P.C. L’Argentiere, E.A. Cagnola, D.A. Liprandi, M.C. Roman-Martnez,
C.S.M. de Lecea, Appl. Catal. A 172 (1998) 41.
[60] Y.A. Ryndin, O.S. Alekseev, P.A. Simonoc, V.A. Likholobov, J. Mol.
Catal. 55 (1989) 109.
[61] A. Aramata, S. Takahashi, G.P. Yin, Y.Z. Gao, Y. Inose, H. Mihara, A. Tad-
jeddine, W.Q. Zheng, O. Pluchery, A. Bittner, A. Yamagishi, Thin Solid
Films 424 (2003) 239.
[72] N. Bouchenafa-Saib, P. Grange, P. Verhasselt, F. Addoun, V. Dubois, Appl.
Catal. A 286 (2005) 167.
[73] J.M. Solar, F.J. Derbyshire, V.H.J. Debeer, L.R. Radovic, J. Catal. 129
(1991) 330.
[74] S. Hermans, M. Wenkin, M. Devillers, J. Mol. Catal. A Chem. 136 (1998)
59.
[75] T.A. Stephenson, S.M. Morehouse, A.R. Powell, J.P. Heffer, G. Wilkinson,
J. Chem. Soc. (1965) 3632.
[62] A.R. Silva, M. Martins, M.M.A. Freitas, J.L. Figueiredo, C. Freire, B. de
Castro, Eur. J. Inorg. Chem. (2004) 2027.
[63] A.R. Silva, J.L. Figueiredo, C. Freire, B. de Castro, Catal. Today 102
(2005) 154.
[64] B. Jarrais, A.R. Silva, C. Freire, Eur. J. Inorg. Chem. (2005) 4582.
[65] A.R. Silva, V. Budarin, J.H. Clark, B. de Castro, C. Freire, Carbon 43
(2005) 2096.
[66] A. Valente, A.M.B. do Rego, M.J. Reis, I.F. Silva, A.M. Ramos, J. Vital,
Appl. Catal. A 207 (2001) 221.
[76] L. Soptrajanova, B. Soptrajanov, Spectrosc. Lett. 25 (1992) 1141.
[77] M.L. Toebes, E.M.P. van Heeswijk, J.H. Bitter, A.J. van Dillen, K.P. de
Jong, Carbon 42 (2004) 307.
[78] G.B. Deacon, R.J. Phillips, Coord. Chem. Rev. 33 (1980) 227.
[79] S.V. Kravtsova, I.P. Romm, A.I. Stash, V.K. Belsky, Acta Crystallogr. Sect.
C 52 (1996) 2201.
[80] SDBS Web database, National Institute of Advanced Industrial Science
and Technology, Japan, 2006; http://www.aist.go.jp/RIODB/SDBS/.
[81] J.J. Chen, E. Ruckenstein, J. Catal. 69 (1981) 254.
[82] V. Dubois, Ph.D. thesis, Institut Meurice (Brussels) and Université
catholique de Louvain (Louvain-la-Neuve), Belgium, 2000.
[67] D.C. Iffland, F.A. Cassis, J. Am. Chem. Soc. 74 (1952) 6284.
[68] GB Patent 925,458 to Engelhard Industries (1963).