Thermochemical mass-spectrometric investigation under reducing conditions of [Pd(acac)2] adsorbed on magnesium oxide

Article abstract of DOI:10.1016/S0040-6031(98)00378-5

The thermochemical behavior of [Pd(acac)₂] adsorbed onto different magnesium oxides has been studied under reducing conditions by means of the TPRD (temperature-programmed reductive decomposition) technique coupled to mass spectrometry. The thermal profiles of the different decomposition products have been rationalized in terms of chemical interactions with the support, and correlated with the catalytic properties in the dehydrocyclization reaction of n-heptane. The technique is then proposed to be a valid screening methodology for routine testing of catalytic batches without the need of running costly and time-consuming catalytic tests.

Full text of DOI:10.1016/S0040-6031(98)00378-5

Thermochimica Acta 317 (1998) 157±164
Thermochemical mass-spectrometric investigation under reducing
conditions of [Pd(acac)2] adsorbed on magnesium oxide
Carlo Dossia,*, Rinaldo Psarob, Achille Fusia, Sandro Recchiaa,
Vladimiro Dal Santoa, Laura Sordellib
a Dipartimento di Chimica Inorganica, Metallorganica e Analitica, University of Milano, Milan, Italy
b CNR Center `CSSCMTBSO' Via Venezian, 21, 20133 Milan, Italy
Received 2 September 1997; received in revised form 5 May 1998; accepted 11 May 1998
Abstract
The thermochemical behavior of [Pd(acac)2] adsorbed onto different magnesium oxides has been studied under reducing
conditions by means of the TPRD (temperature-programmed reductive decomposition) technique coupled to mass
spectrometry. The thermal pro®les of the different decomposition products have been rationalized in terms of chemical
interactions with the support, and correlated with the catalytic properties in the dehydrocyclization reaction of n-heptane.
The technique is then proposed to be a valid screening methodology for routine testing of catalytic batches without the need
of running costly and time-consuming catalytic tests. # 1998 Elsevier Science B.V.
Keywords: TPRD; Thermoanalytical studies; Palladium catalysts; Mass spectrometry; Surface decomposition
1. Introduction
particularly for catalytic purposes, are generally
dependent on the nature of the chemical interactions
between the organometallic complex and the oxide
support [3±6].
The recent advances in solid-state electronics [1]
and catalysis [2] have been often achieved through the
utilization of organometallic complexes as precursors
of high-purity metal phases with the desired composi-
tion and topology. In this respect, acetylacetonate
complexes have a signi®cant role [3±5], since they
are easily synthesized and/or puri®ed, and their vola-
tility may be ®nely tuned for CVD applications by the
controlled introduction of ¯uorine atoms. Moreover,
the supported metal phase is simply obtained by
controlled thermal removal of acetylacetonate ligands.
However, the ®nal properties of the metal phase,
Thermochemical methods play a primary role in the
characterization of the kinetics of decomposition pro-
cesses of the supported organometallics to yield the
desired metal phase. Many different approaches have
thus been proposed, ranging from conventional DTG
studies [4,5] to more specialized ¯ow techniques for
catalyst characterization [7±10]. In particular, ¯ow
techniques under reducing conditions, primarily the
TPRD technique [8,9], are thought to offer signi®cant
advantages, in that they mimic the actual activation
and catalytic condition of working catalysts. However,
no de®ned correlations between such thermochemical
studies and catalytic activity have been presented so
*Corresponding author. Tel.: +39-26680676; fax: +39-2362748;
e-mail: dossic@csmtbo.mi.cnr.it
0040-6031/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved
P I I S 0 0 4 0 - 6 0 3 1 ( 9 8 ) 0 0 3 7 8 - 5

158
C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
far, preventing a wide-scale utilization of thermoche-
mical techniques for screening purposes. There is, in
fact, an increasing need of simple analytical techni-
ques for routine prescreening of catalytic batches as an
alternative to the costly and time-consuming catalytic
tests.
In this paper, we will present the application of the
TPRD (temperature-programmed reductive decompo-
sition, [8,9]) technique to a series of Pd/MgO catalysts
obtained from [Pd(acac)2] precursors with different
acid/base and metal/support properties, and, conse-
quently, different catalytic behaviour. The information
obtained from the thermal pro®les will be interpreted
as a function of metal/support interactions, with the
®nal purpose of ®ngerprinting (and possibly predict-
ing) their catalytic properties.
night in vacuo (10 2 mbar). For all other supports, the
solution remained lightly yellow-coloured after 24 h
of stirring. The solvent was then removed in vacuo,
and the sample dried overnight in vacuo.
TPRD experiment [8,9] was carried out in a ¯owing
mixture H2(8%)/He in a ¯ow-through glass reactor.
Conventional sample holder for thermogravimetric
analyses could not be used, because of diffusion
limitations of the sample bed. The temperature was
raised from 258 to 5008C at 38C/min using a compu-
ter-controlled temperature programmer (Ascon YM).
Volatile products were monitored by an on-line
quadrupole mass spectrometer (VG Masstorr, 0±130
amu) interfaced by a differentially pumped capillary
inlet.
Data acquisition was carried out in selected ion-
monitoring mode, and presented as a function of the
increasing temperature. A detailed description of the
instrumental setup was reported elsewhere [9].
Catalytic tests were made on samples after the
thermochemical investigation, using a continuous-
¯ow glass microreactor. Quantitative analyses of reac-
tants and products were carried out on an on-line gas
chromatograph (Carlo Erba Instruments HRGC 5160)
®tted with a thermostatted sampling valve and FID,
using a 50 m, silica-fused capillary PONA column.
The test reaction chosen was the catalytic dehydro-
cyclization of n-heptane at 5008C and 1 atm, with a
H2
2. Experimental
[Pd(acac)2] was prepared from K2PdCl4 as reported
in the literature [11].
Reagent-grade MgO (Carlo Erba RPE) was re¯uxed
in doubly distilled water for 2 h and dried in air at
1008C. This type of support is referred to as MgOair.
Highly dehydroxylated magnesium oxide (MgO500
)
was obtained by heating MgOair in air from 25 to
5008C and further evacuation (Pˆ10 5 mbar) at this
temperature overnight. MgOHCl was prepared from
MgO500 by treating with excess 0.9 M HCl and further
drying overnight at 5008C in vacuo (Pˆ10 5 mbar).
Reagent-grade La2O3500 (Strem Chemicals, 99.99%
/C H ratio of 20.
7
16
3. Results and discussion
purity) was treated as MgO500
.
Reagent-grade benzene and dichloromethane
The thermal reduction pro®les in an hydrogen
atmosphere of pure [Pd(acac)2] were monitored at
m/zˆ15, 43 and 85 (Fig. 1). A single sharp peak for
all m/z values was observed at 758C. The signal at m/
Ê
(Fluka) were dried over 5 A molecular sieves.
The oxide supports were impregnated at room
temperature under nitrogen atmosphere with a dilute
solution of [Pd(acac)2]. Typically, 2.0 g of MgO were
impregnated with ca. 115 mg of [Pd(acac)2] dissolved
in 20±30 cm3 of the solvent, in order to have a ®nal Pd
loading of 2 wt%. Benzene was generally used as
solvent, although it may be safely substituted by
‡
zˆ85, due to the [COCH2COCH] ion, was a typical
fragmentation ion in the electron impact (EI) mass
spectrum of acetylacetone; it was, therefore, indicative
of the evolution of acetylacetone (acacH) following a
classical reductive elimination mechanism:
toluene without any difference. Instead, for MgO500
,
‰Pdꢀacac† Š ‡ H2 ! Pd0 ‡ 2 acacH
2
impregnation was conducted either in benzene and in
dichloromethane.
For MgO500, the solution became colourless after
few hours of stirring; the yellow-coloured impreg-
nated material was then ®ltered off, and dried over-
‡
The signals at m/zˆ15 and 43, due to CH3 and
‡
CH3CO ions, respectively, were instead much less
diagnostic. They are related not only to the formation
of acetylacetone, but also to all the possible hydro-

C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
159
Fig. 1. TPRD profiles of pure [Pd(acac)2] in flowing H2 (8%)/He mixture.
genation products: typically, 2,4-pentadiol, acetone,
propanol, and saturated hydrocarbons. However, the
close similarity of the MS traces, and the absence of
any other peaks at higher temperatures, suggest that
only acetylacetone is evolved during the reductive
decomposition of [Pd(acac)2]. Moreover, the intensity
ratios of the three m/z values are those expected on the
basis of the electron-impact mass spectrum of acetyl-
acetone.
product was formed by the further hydrogenation of
acetone deriving from acetylacetone adsorbed onto the
palladium phase.
This evolution of acetone and isopropanol, through
the surface-mediated fragmentation of acetylaceto-
nate ligands, has the additional consequence that some
organic residues are left on the surface. We suggest
that this deposition takes place by two different
mechanisms:
Upon deposition onto a highly dehydroxylated
magnesium oxide surface (MgO500), the resultant
material showed a totally different thermal behaviour
(Fig. 2). In particular, no evolution at m/zˆ85 was
observed, rejecting any formation of acetylacetone
following the previously depicted mechanism.
Instead, a broad and complex evolution of m/zˆ43
and 58, with two peaks at 1058 and 2908C was
observed. The signal at m/zˆ58 is attributed to the
1. on the metal surface, via the polymerization of C2
fragments, or, more generally, of Cx fragments as
it is often observed in heterogeneous catalysis
[12];
2. on the MgO support through the formation of
acetate ions [5] at the metal oxide interface.
‡
The evolution pro®le of [CH3] ion at m/zˆ15
parallels those at m/zˆ43 and 58 up to ca. 3208C,
‡
‡
[CH3COCH3] Áion radical as the parent ion in the
since the [CH3] fragment is present in the electron-
electron impact mass spectrum of acetone. The evolu-
tion of such product was already reported in the
thermal decomposition of [Pd(acac)2]/Al2O3, albeit
under slightly oxidizing conditions [5]. A gas-chro-
matographic analysis of the evolution products under
these temperature ranges con®rmed the formation of
acetone, along with some isopropanol. This latter
impact spectra of acetone and isopropanol. At higher
temperatures, however, an additional evolution peak at
m/zˆ15 is observed, although no signals at m/zˆ43
and 58 can be seen. This observation indicates that, in
this temperature range, the signal at m/zˆ15 can be
due only to the evolution of methane. It is suggested
that this formation of methane derives from the hydro-

160
C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
Fig. 2. TPRD profiles of [Pd(acac)2]/MgO500 impregnated from benzene.
Fig. 3. Evolution profiles of methane in the TPRD of [Pd(acac)2]/MgO500: (A) sample prepared from C6H6; and (B) sample prepared from
CH2Cl2.
genolysis of the organic residues left on the catalyst, as
it is often observed in TPRD pro®les of reforming-
type catalysts [13].
single, broad peak of methane evolution centered ca.
4008C (Fig. 3(a)).
The overall reactivity of MgO-supported [Pd(a-
cac)2] under TPRD conditions may be summarized
as follows:
According to this hypothesis, we subtracted the MS
trace at m/zˆ43 from that at m/zˆ15, resulting in a

C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
161
where CHx(ads) indicates the organic residues left on
the surface.
eventually removed as methane. We found that this
type of interaction is extremely sensitive to the experi-
mental condition and particularly to the nature of the
organic solvent used for the deposition of [Pd(acac)2].
In fact, a [Pd(acac)2]/MgO sample prepared from a
chlorinated solvent such as CH2Cl2 shows a different
TPRD pro®le at m/zˆ58, with three peaks at 1058,
2188 and 2958C instead of the two-peak feature pre-
viously reported for the material impregnated from
benzene (compare Figs. 2 and 4). Secondly, evolution
of methane is signi®cantly increased, starting at tem-
peratures well below 3008C (Fig. 3(b)). We speculate
that the use of a chlorinated solvent modi®es the
surface of MgO support. Potentiometric and infrared
studies show that new, acidic Mg±Cl sites are present
upon thermal activation in hydrogen [14]. These new
surface sites, of Lewis acid character, offer new sites
This profound difference between the thermoche-
mical behaviour of pure (unsupported) and MgO-
supported [Pd(acac)2] resides in the chemical interac-
tion between the organometallic molecule and the
basic sites of the magnesium oxide surface.
In particular, the formation of hydrogenated pro-
ducts along with acetone, and of methane at high
temperatures, may only be rationalized if we assume
that the acetylacetonate ligands remain chemisorbed
on the active palladium atoms. As already pointed out,
the hydrogenolytic rupture of the C±C bond in the 2±3
position then leads to the evolution of acetone and
isopropanol with deposition of organic residues on the
catalyst, possibly at the interface between the growing
palladium particles and the MgO surface, that are
Fig. 4. TPRD profiles of [Pd(acac)2]/MgO500 impregnated from dichloromethane.

162
C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
Table 1
The catalytic activity to n-heptane dehydrocyclization a of some
supported palladium catalysts
of n-heptane (Table 1). Benzene as the impregnation
solvent was signi®cantly superior to dichloromethane,
leading to higher activity materials. The catalytic
activity, expressed as mole percent conversion of n-
heptane to aromatics at ®xed contact time, decreased
from 70 to 45% when dichloromethane was used for
the impregnation of MgO500 support.
In addition, the data in Table 1 clearly show that the
presence of a highly basic surface is the primary
requirement for developing a catalytically active metal
phase. The removal of highly active basic sites of
MgO by exposure to air (MgOair) or to hydrochloric
acid (MgOHCl) leads to a dramatic decrease in cata-
lytic activity. The two palladium samples supported on
MgOair and MgOHCl showed conversions of n-heptane
ca. 10% only.
Support
Impregnation
solvent
Activity b
(mol%)
Aromatics
(mol%)
MgO500
MgO500
MgOair
MgOHCl
La2O5300
C6H6
CH2Cl2
C6H6
C6H6
C6H6
70
45
11
9
84
83
72
58
Ð
<1
a Experimental conditions: Tˆ5008C, Pˆ1 atm, H2/C7H16ˆ20.
b Activity expressed as mole percent conversion of n-heptane to
aromatics.
for the adsorption of the [Pd(acac)2] molecule, and, at
the same time, cause an increased deposition of car-
bonaceous residues, as evidenced by the higher
amount of evolved methane. The presence of surface
acidity on metal catalysts is known to cause an
increased carbon deposition on the surface [15].
The observed differences in the thermal reduction
pro®les of the two [Pd(acac)2]/MgO samples are not
merely instrumental effects, but are strongly related to
chemical activity. The two samples at the end of the
reduction treatment are in fact good catalysts for
hydrocarbon reactions, particularly for the high-tem-
perature dehydrocyclization (aromatization) reaction
The thermal reduction pro®les of [Pd(acac)2]
deposited on MgOair (Fig. 5) and on MgOHCl
(Fig. 6) have also been studied.
In the MgOair sample, the TPRD experiment had to
be stopped at 3508C, when large amounts of water and
CO2 started to be evolved from the decomposition of
magnesium hydroxide and carbonate species, prevent-
ing the MS detector to work under proper conditions.
In both cases, the evolution pro®les in ¯owing H2/
He mixture looked quite different from those of
Fig. 5. TPRD profiles of [Pd(acac)2]/MgOair
.

C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
163
Fig. 6. TPRD profiles of [Pd(acac)2]/MgOHCl
.
[Pd(acac)2] deposited on MgO500. In particular, a low-
temperature peak at m/zˆ85 is observed between 658
and 908C in both thermal pro®les, which is due to the
evolution of free acetylacetone (Figs. 5 and 6). This
closely resembles the behaviour of pure unsupported
[Pd(acac)2] (Fig. 1). At higher temperatures, some
evolution of acetone at m/zˆ58 was then observed
below 3308C, whereas some residual methane was still
formed at 5008C.
These two features, namely the evolution of
acetylacetone at low temperature and the negligible
methane formation from CHx(ads), are necessarily
related to a completely different surface chemistry
of [Pd(acac)2]. The occurrence of a simple physisorp-
tion of the palladium precursor without any signi®cant
chemical interaction with the MgO surface is then
suggested. This fact con®rmed the ability of the TPRD
thermoanalytical technique to differentiate between
chemisorbed and physisorbed state of supported metal
complexes, owing to the advantage of investigating
the sample under reducing conditions, which mimic
the actual working conditions of the catalyst.
This surface reactivity can be explained with the
dramatic decrease in surface basicity caused by sur-
face carbonates or chlorides. The organopalladium
molecule is, thus, no longer able to chemically react
with the surface; as a consequence, the Pd precursor
is not homogeneously distributed over the surface,
and the formation of a highly dispersed, highly
active metal phase is prevented. We con®rmed this
assumption by investigating the thermal behaviour
of a Pd=La2O5300 (Fig. 7) catalyst of very low cata-
lytic activity. Accordingly, the evolution of acetyla-
cetone is a key feature of the thermal pro®le, and
demonstrates how the La2O3 surface is totally
inert towards the adsorption of the acetylacetonate
complex.
4. Conclusions
Thermochemical mass-spectrometric characteriza-
tion under reducing conditions of [Pd(acac)2]/MgO
offers a simple analytical methodology for ®ngerprint-
ing the surface properties of supported palladium
phases. A complex thermal behaviour was demon-
strated to be related to a strong interaction between the
metal complex and the surface, and eventually to a
high catalytic activity.
A simple physisorption, associated with a low
catalytic activity, was instead revealed by few sharp
peaks of ligand evolution.

164
C. Dossi et al. / Thermochimica Acta 317 (1998) 157±164
Fig. 7. TPRD profiles of ‰Pdꢀacac† Š=La2O3500
.
2
Faraday Trans. I 85 (2) (1989) 389; A.R. Siedle, P.M. Sperl,
T.W. Rusch, Appl. Surf. Sci. 6 (1980) 149; J.C. Kenvin, M.G.
White, M.B. Mitchell, Langmuir 7(6) (1991) 1199.
[5] J.R. van Veen, M.S.P.C. DeJong-Versloot, G.M.M. van
Kessel, F.J. Fels, Thermochim. Acta 152 (1989) 359.
[6] C. Dossi, R. Psaro, A. Bartsch, E. Brivio, A. Galasco, P. Losi,
Catal. Today 17 (1993) 527.
This analytical approach could, thus, be proposed as
a valid routine screening methodology of catalytic
batches before running costly and time-consuming
catalytic tests.
[7] C. Dossi, A. Fusi, R. Psaro, G.M. Zanderighi, Appl. Catal. 46
(1989) 145.
Acknowledgements
[8] C.M. Tsang, S.M. Augustine, J.B. Butt, W.M.H. Sachtler,
Appl. Catal. 46 (1989) 45.
This work was supported by the Ministry of Uni-
versity and Scienti®c and Technological Research
(MURST) and the National Research Council (CNR).
[9] C. Dossi, A. Fusi, R. Psaro, Thermochim. Acta 236 (1994)
165.
[10] C. Dossi, A. Fusi, R. Psaro, D. Roberto, Thermochim. Acta
182 (1991) 273.
[11] Gmelin Handbuch der Anorganischer Chemie, Palladium,
Verlag Chimie, Weinheim 1947, p. 302.
References
[12] G. Webb, Catal. Today 7 (1990) 139.
[13] C. Dossi, J. Schaefer, W.M.H. Sachtler, J. Mol. Catal. 52
(1989) 193.
[1] H.D. Kaesz, R.S. William, R.F. Hicks, J.I. Zink, Y. Chen, H.J.
Muller, Z. Xue, D. Xu, D.K. Shuh, Y.K. Kim, New J. Chem.
14 (1990) 527.
[14] M. Kappers, C. Dossi, R. Psaro, S. Recchia, A. Fusi, Catal.
Lett. 39 (1996) 183.
[2] R. Ugo, C. Dossi, R. Psaro, J. Mol. Catal. 107 (1996) 13.
[3] C. Dossi, R. Psaro, A. Bartsch, A. Fusi, L. Sordelli, R. Ugo,
M. Bellatreccia, R. Zanoni, G. Vlaic, J. Catal. 145 (1994) 377.
[4] J.A.R. van Veen, G. Jonkers, W.H. Hesselink, J. Chem. Soc.,
[15] C. Dossi, R. Psaro, L. Sordelli, M. Bellatreccia, R. Zanoni, J.
Catal. 159 (1996) 435; A.T. Bell, Stud. Surf. Sci. Catal. 48
(1989) 91.