Porous boron nitride supports obtained from molecular precursors. Influence of the precursor formulation and of the thermal treatment on the properties of the BN ceramic

Article abstract of DOI:10.1016/S0022-328X(02)01589-9

Boron nitride (BN) porous samples have been prepared in order to be used as noble metal catalyst support from various molecular precursors, using classical thermal methods to expand and preceramise the precursors. Three types of precursors have been teste

Full text of DOI:10.1016/S0022-328X(02)01589-9

Journal of Organometallic Chemistry 657 (2002) 98Á106
/
www.elsevier.com/locate/jorganchem
Porous boron nitride supports obtained from molecular precursors.
Influence of the precursor formulation and of the thermal treatment
on the properties of the BN ceramic
Jose´ Antiono Perdigon-Melon b, Aline Auroux b, David Cornu a, Philippe Miele a,
Be´range`re Toury a, Bernard Bonnetot a,*
a Laboratoire des Multimate´riaux et Interfaces, Universite´ C. Bernard Lyon I, UMR-CNRS 5615, Bat Berthollet, 43 Bd du 11 Novembre 1918,
69622 Villeurbanne Cedex, France
b Institut de Recherches sur la Catalyse, CNRS, 2, Av. A. Einstein, 69626 Villeurbanne Cedex, France
Received 3 September 2001; accepted 21 May 2002
Abstract
Boron nitride (BN) porous samples have been prepared in order to be used as noble metal catalyst support from various
molecular precursors, using classical thermal methods to expand and preceramise the precursors. Three types of precursors have
been tested: ammonia borane, polyborazylene and several aminoborazines and derived polymers. Using trimethylamino borazine
(MAB) and MAB polymers, the surface of the foams was shown to be decreasing when the polymerisation advancement of the
precursor increased meanwhile the crystallisation of the samples decreased. All the samples issued from MAB presented the same
SEM morphology: large pores with a glassy skin surface covering a disorganised inner part, hidden under the thin homogeneous BN
skin, which appeared through the sample breaks. This was related to the melting property of MAB polymers which imbedded the
internal part of the sample and avoided its expansion. Attempts were made, using the bulkier amino group as anilino- and benzyl-
aminoborazine to enhance the porosity of the samples using a precursor with a lower ceramic yield. The change in the specific area
was not important. However, a surface of 30 m2 gꢀ1 was currently obtained. The ability of the precursor to lead to high specific area
seemed to be correlated to the expansion possibilities of the precursor in the solid state prior to fusion. Foams exhibiting a surface of
more than 50 m2 gꢀ1 have been obtained when polyborazylene based precursor was pyrolysed. # 2002 Elsevier Science B.V. All
rights reserved.
Keywords: Boron; Nitride; Catalysis; Support; Powders
1. Introduction
high temperature localised on the metal active catalyst.
The oxide supports were mainly isolators so the high
oxidation energy could not be spread over the support.
Moreover, a certain degree of metal-support interaction
existed in most oxide-supported metal catalysts that
brought a negative influence on the catalytic activity [4]
thus non-oxide supports have been considered [5].
Boron nitride (BN) was one of the more interesting
non oxide ceramic material because of its low density,
excellent resistance to chemical attack, high melting
point, good thermal conductivity and its higher stability
towards oxidation than carbon [1]. The combination of
these specific properties focused the attention on the
possibility to prepare high surface area/high porosity
BN specially useful in the fields of high temperature
composites and ceramic support for catalysis. The high
Catalytic oxidation of gasoline volatile was actually
an important subject with respect to environment
concern. The catalyst supports were usually made
from highly divided oxides providing a large specific
area on which an active noble metal was deposed. In
fact, traditional metal oxides, such as Al2O3, SiO2 and
zeolithe possessed rather low thermal conductivity, thus
metal sintering on hot spots might be severe [2,3]. The
high oxidation energy of hydrocarbon species led to a
* Corresponding author. Tel.: ꢁ
440618
E-mail address: bernard.bonnetot@univ-lyon1.fr (B. Bonnetot).
/
33-4-72-431234; fax: ꢁ33-4-72-
/
0022-328X/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 3 2 8 X ( 0 2 ) 0 1 5 8 9 - 9

J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á
/
106
99
thermal conductivity of BN [6] could be a solution to
disperse the important oxidation energy cleared on the
metal active site and to preserve the catalyst efficiency
for a longer time. Another point that had brought our
interest to BN was its higher stability towards oxidation
than carbon. When a catalyst was used in the destruc-
tion of gasoline volatile, small amounts of carbon were
usually deposed during the oxidation and the efficiency
of the catalyst was lowered. A reactivation of a BN
supported catalyst could be realised by burning the
residual carbon without effect on the catalyst. However,
the classical ways used to get BN led to powders which
were hardly converted into high specific surface sup-
ports [7,8]. The preparation of high surface samples of
boron nitride had to be performed from very different
molecular precursors. Starting from boron oxide, mel-
amine diborate [9] could be converted into BN. Very
thin BN powder had been obtained trough a high
temperature nitruration of boron coming from diborane
pyrolysis [10]. However, the main part of the molecular
precursors used to obtain BN was derived from a
borazine framework [7,8]. High surface BN substrates
had been prepared from aerogels obtained in a THF
solution using a low temperature reaction between TCB
and hexamethyldisilazane following a described method
[11]. The gel suspension had been treated by ammonia
before it was dried using a critical point drier with CO2
as a solvent. The surface area was shown to be very
dependant on the ceramisation temperature of the
sample and a surface of more than 800 m2 gꢀ1 at
1200 8C was described as dropping to 400 m2 g when
BN was stabilised at 1600 8C, the smaller pore of about
0.1 nm being closed by annealing of the aerogel [12,13].
Recently, a first boron nitride supported catalyst had
been prepared from BN powder with a great specific
area but not indication was given on the support
preparation [14]. In the present work, several molecular
BN precursors have been used under different expand-
ing conditions to obtain polymeric foams which have
been ceramised to lead to BN stabilised foams. This
stabilisation had been performed up to 1800 8C in order
to get very stable supports although this high tempera-
ture of treatment was known to lower the surface of the
foams [13]. Very different results have been obtained
which could be related to the precursor but also to the
chemical and thermal treatment used to expand the
precursors before and during the preceramisation. The
main part of the precursors was issued from aminobor-
azines bearing different amino moieties. Trimethylami-
noborazine, MAB, used as BN fibre precursor [15] is
well known, but attempts have been made using
trianilino, tribenzyl aminoborazine, polyborazylene
and ammonia borane. SEM observation of the main
part of the samples had shown large pores leading to a
30 m2 gꢀ1 specific area with a smooth glassy surface,
always obtained using MAB or MAB issued polymers.
Benzylaminoborazine led to a black carbon containing
ceramic due to bad stripping of the amino group of
amino borazine during ceramisation. Using non fusible
precursors such as ammonia borane, polyborazylene
derivative or heavy aminoborazine, thin powders were
obtained exhibiting various specific areas in the range of
20Á
35 m2 gꢀ1 depending of the precursor and the
/
thermal ceramisation profile, but also of the tempera-
ture of ceramisation and of impurities as described for
such compounds [10,13,16].
2. Results and discussion
2.1. Molecular precursors used
Samples of porous boron nitride have been obtained
using different precursors and various thermal and
physical treatments. All the precursors were only con-
stituted from boron, nitrogen and hydrogen. The
simplest compound used (P I) was ammonia borane
NH3BH3. Its pyrolysis had been studied in order to get
borazine (HBNH)3 at moderate temperature [17] and to
get BN coatings [18]. The thermolysis of ammoniaÁ
/
borane under ammonia led to thin white BN powder.
The other precursors were build from a borazinic
framework (BN)3 as the major part of the compounds
used in BN preparation from molecular precursors. Two
different kind of precursors have been studied one
starting from borazine (HBNH)3 and its polyborazylene
issued polymers and the other from aminoborazines
(RNHBNH)3 and their derived polymers. The polymer-
isation of borazine [19Á21] had been widely studied to
/
get spinnable polymers. The main interest of polybor-
azylene was its very high ceramic yield but the chemistry
of this compound was difficult [19]. However, a poly-
borazylene derivative sample had been prepared (P II)
using the polymerisation protocol used for BN matrix
preparation [22,33]. Another kind of precursors was
obtained from amino-borazines. The monomers were
prepared using a reaction between trichloroborazine and
primary amines. Three different amines have been used,
methylamine, aniline and benzylamine to prepare the
corresponding aminoborazines, respectively the tri-
methylamino-borazine or MAB, (CH3NHBNH)3 lead-
ing to the precursor (P III), the trianilinoamino-
borazine, AAB, (C6H5NHBNH)3 leading to (P IV)
and the tribenzylaminoborazine, BAB, (C6H5CH2-
NHBNH)3 used as precursor (P V). The molecular
structure of these aminoborazines were shown on Fig. 1.
The changes in the amino group corresponded to two
criteria: the bulkiness of the group would change the
ability of the aminoborazine to polymerise [23] and their
reactivity toward ammonia used in the preceramisation.
Some amino group, effectively considered as ‘good
leaving’ group in the precursor which means amino

100
J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á
/
106
Fig. 1. Aminoborazines used.
groups easily replaced by ammonia, could lead to an
important change in the behaviour of the preceramic
during the chemical and thermal treatment and induced
differences in the ceramic obtained [24]. Taking into
account the bulkiness of the amino group, the amino-
borazines underwent polymerisation leading to poly-
mers of various structure depending on the
condensation reaction. The classic thermal treatment
yielded the reaction between two amino groups from
two monomer molecules with the evolution of a
molecule of amine and a new bonding between the
two molecules through a bridge. The ability of amino-
borazine to led to polymer was shown to depend upon
the amino group, its bulkiness and its reactivity [25].
Among the aminoborazine prepared, the more reactive
monomer was MAB and the less reactive one was BAB.
MAB had been transformed into a polymer following a
polycondensation reaction described elsewhere [25]. The
physico-chemical properties of the polymer obtained
were described in the experimental part, this polymer
was very similar to the polymer used for polymer fibre
spinning [27]. The other aminoborazines AAB and BAB
had been thermolysed without polymerisation. Their
reactivity had been shown as very different related to the
amino group associated.
their properties, the samples were annealed after the
preceramisation up to 1800 8C in a nitrogen flow. All
the experiments were presented in Table 1 with the
experimental conditions and the main characteristics of
the obtained samples. Out of the specific area, the
crystallinity of porous BN had been measured but only
along the c axis of h BN [28].
The two attempts were performed using the precursor
P I and similar condition without any expansion
pretreatment. After annealing at 1800 8C, the samples
JA 13 and 18 looked like a ‘flour’ powder of BN and
have shown very similar properties. The specific area
was more than 30 m2 gꢀ1 and the samples exhibited a
very high crystallisation characterised by a crystallite
˚
150 A which correspond to
length along the c axis Lcꢂ
/
45 BN layers.
Using the polyborazylene derived precursor P II, two
experiments JA 12 and 15 were made without any
expansion treatment because this polymer decomposed
rapidly when heated [22]. The heating rate and the
ammonia flow were the two modified parameters
(preceramisation protocols A and B, respectively) and
the results were different specially for the crystallisation
˚
of the sample with Lc value dropping from 125 to 20 A
while the surface increased from 20 to 53 m2 gꢀ1
.
The main part had been run on P III because, as it
could be molten, a larger number of thermal treatments
could be done. The expansion of the P III had been done
under low pressure with a very rapid heating and a 1 h
hold at various temperatures for samples JA 01, 02 and
03; a heating in a closed autoclave followed by a rapid
gas release for samples JA 04, 05 and 06, a high
mechanic pressure application while heating followed
by a quick pressure release was applied for samples JA
2.2. Results and discussion
2.2.1. Specific treatments
Several types of thermal treatments had been per-
formed on the different samples in order to get foams
from the fusible precursors. On some samples, a first
prethermolysis with the goal to prepare a more reactive
and more expanded preceramic had been made using
various conditions. Then the pretreated or untreated
samples were preceramised following two protocols as
described in the experimental part. To stabilise and to
crystallise the samples in order to avoid any changes in
07, 08, 16 and 17. A flash thermolyse (100 8C minꢀ1
)
under vacuum had been realised on sample JA 11. The
preceramisation of all this samples had been done using
the protocol A. The precursors P IV and P V have only

J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á
/
106
101
Table 1
Series of samples prepared from molecular precursors
Time (h) Surface (m2 gꢀ1
)
˚
Crystallisation (Lc A)
Sample Precursor
Conditions expansion
Temperature treatment (8C)
JA 0
JA 1
P III
P III
P III
P III
P III
P III
P III
P III
P III
P III
P II
25
150
123
28
Vacuum
Vacuum
Vacuum
250
360
470
225
225
360
250
250
550
1
20
4
JA 2
JA 3
1
1
3.6
18
20
28
1.5
23
7
22
JA 4
JA 5
Gas pressure 1b
Gas pressure 2b
Gas pressure 5b
Press 42 daN
4
140
150
112
30
4
JA 6
JA 7
4
1
JA 8
Press 42 daN
Vacuum
2
0.2
100
86
JA 11
JA 12
JA 13
JA 14
JA 15
JA 16
JA 17
JA 18
JA 19
JA 20
No exp/preceram A
No exp/preceram A
No exp/preceram B
No exp/preceram B
Press 40 daN
Á/
Á/
Á/
Á/
200
250
Á
Á
Á
Á
/
20
34
12
53
28
12
30
21
25
125
163
101
20
P I
/
P III
P II
/
/
P III
P III
P I
1
90
75
Press 40 daN
1
/
No exp/preceram A
No exp/preceram A
No exp/preceram A
/
/
/
140
26
P IV
P V
/
/
16
been treated following the preceramisation procedure A
without pretreatment because they carried bulky amino
groups which render the polymerisation very difficult.
From precursor P III, the thermal treatment under
vacuum on sample JA 01, 02 and 03 lowered drastically
the specific area and the crystallinity of the samples.
However, after pretreatment, a very light foam of
preceramic was recovered and the decomposition and
the polymerisation of the sample JA 03 correspond to
about 50% of weight loss when a ceramic yield was
achieved on the sample. It seemed that the expanded
polymer could not react easily when preceramised and
that the inside part of the polymeric foam was kept into
a solid skin of BN. This morphology will be more
efficiently shown using SEM. An example of the
difference of crystallinity observed on the samples
related to the precursor was given on Fig. 2.
MAB reacted very rapidly with ammonia whereas
BAB was very stable up to 200 8C. Aniline described as
a good ‘living group’ did not react very rapidly if the
temperature was lower than 100 8C. However, the
thermolyse residues for MAB and AAB after TGA
were white, without carbon while the sample issued from
BAB was black consistent with the theoretical weight
loss of 81 for 75% experimental.
The crystallinity of the samples obtained from the
different precursors was very different and always the
ceramics obtained from ammonia borane or polybor-
azylene were more crystallised as reported [29].
2.2.2. Micrographic morphology of the samples
Prior the specific area measurements, the morphology
of the different samples had been examined using SEM.
From precursor (P I) no difference could be noticed
between the two products. The samples were very well
crystallised and the morphology of the BN powder given
in Fig. 4 showed a regularly grained powder.
For samples coming from precursor (P II), important
differences were observed in the crystallisation and
surface. Using a slow preceramisation, the sample
exhibited a glassy out side structure as shown on Fig.
5a, but if the thermolysis was faster with an important
ammonia flow (protocol B), a more porous compound
was obtained as shown in Fig. 5b.
On the sample 15, the specific area related to the pores
repartition had been studied. The pores size was
measured between 25 and 5 nm with about the same
proportion. Only a macro porosity had been shown as
given on Fig. 6 a and b corresponding, respectively to
the sum of the specific area as a function of the pores
size and the ratio of the pores versus their radii.
These three samples JA 17, 19 and 20 have been
ceramised in the same run under exactly the same
conditions but staring from three precursors, P III, IV
and V. If the surface value of the three samples were
comparable, about 20 m2 gꢀ1, the crystallisation was
˚
very different and Lc decreased down to 16 A. More-
over, as shown in Fig. 3, the TGA performed under the
same conditions than the preceramisation protocol A
led to important differences in the ability of the
precursor to react with reductive atmosphere used to
strip off the carbon moieties bound to the amino group
of the aminoborazine. After the preceramisation treat-
ment performed using the TGA apparatus the residue
corresponding to the sample 20 was deeply black
showing the presence of an important amount of carbon
coming from the pyrolysis of the hydrocarbon amino
parts.

102
J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á106
/
Fig. 2. DRX spectra of different samples issued from aminoborazines.
Fig. 4. Sample JA 13 issued from polymer P I.
Fig. 3. TGA of aminoborazines under the preceramisation conditions.
stable glassy structure which had been ceramised with-
out any change. It corresponds to the sample obtained
from very polymerised precursor with a very low surface
and a poor crystallisation as observed with the samples
JA 03, 07 and 17.
From precursors P IV and P V, only powders had
been obtained, looking like the sample obtained from P
I.
The samples prepared from MAB (P III) presented
very similar SEM morphology (Fig. 7a for JA 06, b for
JA 08 and c for JA 11) whatever the thermal or physical
treatment of the polymer had been before ceramisation.
All the samples seemed to be very heterogeneous with
an outside skin very regular and crystallised and an
inside part looking more porous and covered by the
smooth BN skin. This complex structure could be
observed at any scale. For instance the magnification
of the SEM figures on the Fig. 7a and b was 10 times
different but the same morphology was observed with a
skin and an inside porous part.
3. Experimental
3.1. Starting materials
The picture obtained on the sample pretreated using
the flash thermolysis of precursor P III showed clearly
that the molten precursor had been expanded into very
large pores, more than 20 mm, in a very strong and
All experiments were performed under an atmosphere
of pure Ar using standard vacuum-line, Schlenk techni-
ques and an efficient dry box with solvents purified

J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á
/
106
103
Fig. 5. Samples JA 12 (a) and 15 (b) issues from polymer P II.
Fig. 7. SEM pictures of the samples JA 06 and 08 obtained from
precursor P III. (c) Sample JA 11 issued from precursor P III.
NH3BH3 and polyborazylene were used as received. The
trichloroborazine (TCB) was prepared from BCl3 (Al-
phagaz) and NH4Cl (Aldrich, ACS reagent). Ammoni-
um chloride was dried under vacuum at 130 8C for 12 h
before the synthesis. Methylamine (Fluka, 97%), aniline
and C6H5NH2 (Aldrich, ACS reagent) were used as
supplied without further purification.
3.2. Characterisation
1
11B-, H- and 13C-NMR spectra were recorded on a
Bruker DRX 300, respectively at 96.26 MHz (external
reference Et2OBF3 non decoupled), with positive value
down field), 300 MHz (reference Me4Si) and at 75 MHz
(reference Me4Si, total decoupling). The chemical shifts
were explained in ppm and proton and carbon spectra
were obtained from CH2Cl2 solutions. The following
abbreviations are used: d, doublet; s, singlet; t, triplet; q,
quadruplet; m, unresolved multiplet; br, broad. The IR
spectra were recorded on a FTIR Nicolet Magna 550
Fig. 6. (a and b) Sum of the specific area and pores number as a
function of the size of the pores.
using standard methods [32]. The starting materials,
NH4CO3 and KBH4 (Aldrich ACS) for the synthesis of

104
J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á106
/
spectrophotometer in KBr pellets. The ceramic yield was
determined using a TGA B70 apparatus and the study
was performed under ammonia up to 600 8C and then
under nitrogen up to 1000 8C with a 2 8C minꢀ1
heating rate. DCS analysis were ran on a TA 8000
Mettler-Toledo under an Ar atmosphere. SEM images
were obtained from a JEOL 55 CF (CMEABG Lyon).
X-ray powder diffraction (XRD) were obtained with
oxygen and carbon can be related to a partial reaction of
borazine with glyme, leading through BOR bridges to
the incorporation of alkoxide into the precursor.
3.3.3. Synthesis and characterisation of the
triaminoborazine (P III, IV and V)
The aminoborazines were prepared from trichorobor-
azine (TCB) and the corresponding amine. TCB was not
commercially available so it was prepared in the
Laboratory using the standard method using the reac-
tion between BCl3 and NH4Cl in C6H5CH3 [34]. To a
suspension of NH4Cl in refluxing C6H5CH3, BCl3 was
slowly added, the addition rate was regulated by the
reflux of BCl3. After the reaction, TCB was recovered
from C6H5CH3 with a 80% yield from BCl3. IR (cmꢀ1):
n 3450, 3416 (NH); 1438 (BN stretch.); 1031 (NH); 743
CuÁKa radiation using a Philips PW 3710/3020 dif-
/
fractometer equipped with a monochromator. The BET
surface areas were measured using nitrogen at 77 K
using a apparatus build in the Institute of Catalysis. All
the samples were outgassed at 673 K for 4 h prior to the
absorption measurements.
3.3. Preparation of the studied compounds
1
(BCl); 704 (BN bend); H-NMR: (ppm): d 5.29 (t, br,
3.3.1. Synthesis and characterisation of ammoniaÃ
borane, NH3BH3 (P I)
AmmoniaÃborane (NH3BH3) was prepared from
/
(NH); 11B-NMR: (ppm): d 29.7 (s). In a second step,
TCB was reacted with the required amine at low
temperature in a ratio 1/8 leading to the formation of
aminoborazine and amine-chlorhydrate which was se-
parated by filtration. The triaminoborazine was recov-
ered after the evaporation of C6H5CH3 under low
/
NH4CO3 and KBH4 through an exchange reaction
performed in THF [32,33]. IR (cmꢀ1): n 3320 (NH);
2355 (BH); 1380 (BN stretch.); 789 (BN bend); 1H-
1
NMR: (ppm): d 1.49 (q, 3H, JHB
ꢂ
/
93 Hz (BH3); 4.1
pressure at room temperature (r.t.) [26Á35].
/
1
(br t, 3H, JH14Nꢂ
/
35 Hz (NH3); 11B-NMR: (ppm): d
94 Hz). The ceramic yield measured by
Precursor (P III): it was prepared from MAB,
however, the synthesis of pure MAB was difficult
because of its ability to polymerise. Usually a mixture
of 6% molar of dimer is isolated while the remaining
C6H5CH3 is about 15% molar. The precursor looks like
a very viscous oil and the removal of C6H5CH3 is
difficult and leads to an important polymerisation
progression. The MAB is characterised in situ and the
ꢀ
/
22.6 (q, 1JBH
ꢂ
/
TGA is a function of the ceramisation atmosphere. Up
to 1000 8C with a heating rate of 28 minꢀ1 under
nitrogen flow (0.2 l minꢀ1), the ceramic yield was 49%
however, under NH3 (0.2 l minꢀ1) it was only 25%. This
was interpreted by the formation of volatile borazine
favoured by NH3 [31] which was stripped off of the
crucible lowering the ceramic yield.
1
amount of C6H5CH3 can be measured using H-NMR.
IR (cmꢀ1): n 3446, 3250 (NH); 2922, 2830 (CH); 1514
(CH); 1420 (BN stretch.); 1194 (NC); 1029 (NH); 715
(BN bend); 1H-NMR (ppm): d 1.50 (s, br, 1H,
(BNHCH3); 2.5 (s, br, 3H, (NHCH3)), 2.66 (s, br,
3H(NH cycle)); 11B-NMR (ppm): d 25.9 (s, br); 13C-
NMR (ppm): d 27.5 (s, br (NHCH3). This ‘monomer’
will be transformed into polymer using a thermal
polymerisation conducted under an Ar stream. As
shown on Fig. 2, the polymerisation of MAB evolved
methylamine and each methylamine molecule evolved
corresponds to an intercyclic bonding. The reaction can
be monitored by the determination of the quantity of
amine evolved. The polymerisation was carried follow-
ing the standard process described [24]. For the pre-
cursor (P III), the ratio of the amine evolved versus the
monomer quantity was 0.5, which led to a polymer with
3.3.2. Synthesis and characterisation of polyborazylene
(P II)
The borazine was prepared using a two steps classical
method from ammonia borane, NH3BH3, which was
pyrolysed in glyme as described in literature [17].
However, the borazine was not distilled but polymerised
in situ yielding directly a polymer. The solvent was
removed by fitration under an Ar atmosphere and the
remaining borazine and the volatiles were removed as
recommend in the preparation of poly(aminoborane)
[20,21,33]. A white waxy polymer is isolated which is a
mixture of polyborazylene and poly(aminoborane)
whose dissolution is very difficult without decomposi-
tion. IR (cmꢀ1) n: 3400 (NH); 2875 (CH); 2478 (BH);
1
1465 (BN stretch.); 1105 (CN); H-NMR (ppm): d 4.3
(m, br), 3.5 (s, br CH); 11B-NMR (ppm): d 27.6, 21.1
glass transition Tgꢂ
55% at 1000 8C. The elemental analysis (given in %
weight): Bꢂ23.5; Nꢂ47.8; Hꢂ7.2; Cꢂ18.8; oxygen,
/
85.5 8C and a ceramic yield of
(N3B groups), ꢀ
/
17.9 (m, br, N2BH2), ꢀ22.7 (m, br
/
NBH3). The presence of organic moieties in the polymer
/
/
/
/
is confirmed by the elemental analysis (given in %
not measured in presence of boron can be estimated at
less than 2.8%. I.R. n cmꢀ1: 3441 (NH); 2929, 2886
(CH); 1420 (BN stretch.); 1108 (NH); 736 (BN bend);
NMR 1H, d: 1.80 (m, br, BNHCH3); 2.53 (s, br,
weight): Bꢂ
/
23.0; Nꢂ
/
45.8; Hꢂ
/
7.1; Cꢂ18.0; oxygen
/
could not be measured in presence of boron and its
amount can be calculated at about 6%. The presence of

J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á
/
106
105
NHCH3), 3.3Á
/
2.66 (m, br, NH cycle); 11B, dꢂ
27.9 (s, br, NHCH3).
/
25.77 (s,
4. Conclusion
br); 13C, dꢂ
/
The precursor P IV was obtained reacting TCB with
aniline as reported in the literature [26]. The tri anilino
aminoborazine (C6H5NHBNH)3 was prepared with a
45% yield due to its poor solubility in C6H5CH3. The
Porous boron nitride could be prepared from poly-
meric precursors and the physical properties of the
obtained ceramic supports could be adapted to the
required utilisation conditions. Higher specific areas
were obtained from polyborazylene derived precursor.
The preceramisation conditions were shown to be very
independent of the obtained sample and, using the same
precursor, the surface of the porous BN could be two
times higher following different ceramisation protocols.
No microporous structure could be obtained as de-
scribed in the literature. The porous substrates were
composed of broad pores, widely open to the surface of
the sample that could be interesting to depose an active
part of noble metal to prepare a catalyst. The commonly
obtained specific area of about 30 m2 gꢀ1 could be
related to the surface of the powder used in literature
and attempts would be made to prepare noble metal
deposed on the new BN porous supports. The behaviour
of aminoborazine derived precursors showed that the
better results were obtained from the compounds
exhibiting the lower ceramic yield. However, this low-
ering was shown to be difficult to be obtained using very
broad amino group which were known to be hard to
strip from the polymer. A future study would be to
prepare some catalyst samples using first BN foam and
an impregnation technique with Pt or Pd precursor and
other attempts would be made to incorporate the noble
metal precursor into the polymer before the expansion
and the ceramisation of the samples.
1
AAB was characterised using H-NMR. IR (cmꢀ1): n
3421, 3278 (NH); 2922, 3032 (CH); 1501 (CH); 1429
(BN stretch.); 1231 (NC); 1028 (NH); 696 (BN bend);
1H-NMR (ppm): d 3.80 (s, br, 1H, (BNHC6H5); 6.77,
6.94, 7.24 (m, br, 5H, (NHC6H5)), 3.67 (m, br ,3H (NH
cycle)); 11B-NMR (ppm): d 25.2 (s, br). The ceramic
yield was 19.4% up to 1000 8C; it was consistent with
the theoretical weight loss of 21% when AAB was
converted into BN.
The precursor P V was obtained using the same
method [26]. The tribenzylaminoborazine (C6H5CH2-
NHBNH)3 was prepared with a 30% yield. It was
purified by recrystallisation from a chlorobenzene solu-
1
tion. The BAB was characterised using H-NMR. IR
(cmꢀ1): n 3440, 3240 (NH); 2910, 3025 (CH); 1495
(CH); 1443 (BN stretch.); 1028 (NH); 713 (BN bend);
1H-NMR (ppm): d 2.73 (s, br, 1H, (C6H5CH2NHB);
3.12, 3.15 (m, br, 2H, (BNHCH2C6H5); 7.19, 7.27, 7.37
(m, br, 5H, (NHCH2C6H5)), 4.2, 4.32 (m, br, 3H(NH
cycle)); 11B-NMR (ppm): d 30.10 (s, br). The ceramic
yield determined using TGA was 25.6% which was very
high but the residue was black and carbon containing,
so this value was not taken into account and confirmed
the poor ‘living group’ character of the benzyl group.
3.4. Preceramisation and ceramisation conditions
The crude foams after the expanding treatment were
transformed into ceramic by a two steps thermal and
chemical treatment using two different apparatus. First,
a preceramisation was performed from r.t. up to
1000 8C in classic ovens fitted with a silica tube, this
operation was run under reactive ammonia flow (1 L
hꢀ1) up to 600 8C and then under nitrogen flow.
Ammonia was used to strip the organic parts of the
copolymer, as shown in previously [25]. Two heating
rates were used, the standard heating used for ceramic
fibres production Protocol A: 18 minꢀ1 up to 600 8C
and then 58 minꢀ1 up to 1000 8C and a more rapid
preceramisation protocol B: heating rate of 200 8C hꢀ1
under a 6 l hꢀ1 ammonia flow up to 600 8C and then
under nitrogen. Then the fibres were transferred into a
high temperature oven where they were heated up to
1800 8C with a heating rate of 108 minꢀ1 under a
nitrogen atmosphere. The ammonia gas used was from
References
[1] H. List, Gmelin Handbuch der Anorg. Chem, Springer Verlag,
Berlin, Heidelberg, New York, 1974, pp. 1Á82.
/
[2] W.S. Epling, G.B. Hoflund, J. Catal. 182 (1999) 5.
[3] H. Widjaja, K. Sekizawa, K. Eguchi, Chem. Lett. (1998) 481.
[4] J.N. Carstens, S.C. Su, A.T. Bell, J. Catal. 176 (1998) 136.
[5] C. Methivier, B. Be´guin, M. Brun, J. Massardier, J.C. Bertholini,
J. Catal. 173 (1998) 374.
[6] M. Basche, D. Schiff, Mater. Design Eng. 59 (1964) 78.
[7] R.T Paine, C.J. Narula, Chem. Rev. 90 (1990) 73.
[8] R.T. Paine, L.G. Sneddon, Chemtech (1994) 29.
[9] T. Hagio, K. Kobayashi, T. Sato, J. Ceram. Soc. Jpn. 189 (1978)
64.
[10] Y.G. Andreev, T. Lundstrom, J. Alloys Compd. 210 (1994) 311.
¨
[11] C.K. Narula, R. Schaeffer, R.T. Paine, J. Am. Chem. Soc. 109
(1987) 5556.
[12] D.A. Lindquist, T.T. Borek, S.J. Kramer, C.K. Narula, G.
Johnston, R. Scheaffer, D.M. Smith, R.T. Paine, J. Am. Ceram.
Soc. 73 (1990) 757.
[13] D.A. Lindquist, D.M. Smith, A.K. Datye, G.P. Johnston, T.T.
Borek, R. Schaeffer, R.T Paine, Mat. Res. Soc. Symp. Proc. 180
(1990) 73.
Alfagaz Electronic grade N36 (99.96%; H2OB
CnH2n 1 ppm) used without further purification and
nitrogen was from ‘air liquid’ N21 grade (99.999%,
H2OB3 ppm, O2B0.2 ppm).
/100 ppm,
B
/
[14] J.C. Wu, Z. Lin, J.W. Pan, M.H. Rei, Proceedings North
American Catalyst Soc. Meeting (NACS 2001), June 3Á
Toronto, Canada, 2001, p. 294.
/
8, 2001,
/
/

106
J.A. Perdigon-Melon et al. / Journal of Organometallic Chemistry 657 (2002) 98Á106
/
[15] Y. Kimura, Y. Kubo, N. Hayashi, Compos. Sci. Technol. 51
(1994) 173.
[25] F. Guilhon, B. Bonnetot, D. Cornu, H. Mongeot, Polyhedron 15
(1996) 851.
[16] T. Hagio, K. Nonaka, T. Sato, J. Mater. Sci. Lett. 16 (1997) 795.
[17] M.G. Hu, R.A. Geanangel, W.W. Wendlandt, Thermochim. Acta
23 (1978) 249.
[26] R.H. Toeniskoetter, F.R. Hall, Inorg. Chem. 2 (1963) 29.
[27] Aerospatiale Matra, French patent, FR 2 806 422, 2000.
[28] J. Thomas, N.E. Weston, T.E. O’Connor, J. Am. Chem. Soc. 84
(1962) 4619.
[18] P. Goursat, A. Fabre, D. Tetard, A. Lecomte, A. Dauger, P.
Lespade, Silicates Indus 3Á
[19] A.W. Laubengayer, P.C. Moews, R.F. Porter, J. Am. Chem. Soc.
83 (1961) 1337.
/
4 (1993) 45.
[29] B. Toury, C. Duriez, D. Cornu, P. Miele, C. Vincent, M Vaultier,
B. Bonnetot, J. Solid State Chem. 154 (2000) 137.
[30] M.G. Hu, J.M. Van Paasschen, R.A. Geanangel, J. Inorg. Nucl.
Chem. 39 (1977) 2147.
[20] P.J. Fazen, J.S. Beck, A.T. Lynch, E.E. Remsen, L.G. Sneddon,
Chem. Mater. 2 (1990) 1942.
[31] E. Framery, Thesis No. 1616, University of Rennes, 1996, pp. 93Á
/
[21] P.J. Fazen, E.E. Remsey, J.S. Beck, P. Carroll, A.R. McGhie,
L.G. Sneddon, Chem. Mater. 7 (1995) 1942.
99.
[32] D.D. Perrin, W.L. Armarengo, D.R. Perrin, Purification of
Laboratory Chemicals, Pergamon Press, London, 1996.
[33] D.P. Kim, K.T. Moon, J.G. Kho, J. Economy, C. Gervais, F.
Babonneau, Polym. Adv. Technol. 10 (1999) 702.
[22] D.P. Kim, J. Economy, Chem. Mater. 5 (1993) 1216.
[23] K.J.L.Paciorek, W. Krone-Schmidt, D.H. Harris, R.H. Kratzer,
K.J. Wynne, in: M. Zeldin, K.J. Wynne, H.R. Allcock (Eds.),
ACS Symposium Series 360, Chapter 32, 1988, p. 392.
[24] D. Cornu, P. Miele, R. Faure, B. Bonnetot, H. Mongeot, J.
Bouix, J. Mater. Chem. 9 (1999) 757.
[34] H. Noth, H. Sachdev, Z. Naturfosch. Teil. 52 (1997) 1345.
¨
[35] B. Toury, P. Miele, D. Cornu, B. Bonnetot, H. Mongeot, Main
Group Metal Chem. 22 (1999) 231.