Open Archive Toulouse Archive Ouverte (OATAO)

OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.

This is an author-deposited version published in: http://oatao.univ-toulouse.fr/

Eprints ID : 2434

To link to this article :

URL : :http://dx.doi.org/10.1016/j.apcata.2006.10.039

To cite this version : Cordier, Anne and Rossignol, Fabrice and Laurent,
Christophe and Chartier, Thierry and Peigney, Alain ( 2007) A new fast method
for ceramic foam impregnation: Application to the CCVD synthesis of carbon
nanotubes. Applied Catalysis A General, vol. 319 . pp. 7-13. ISSN 0926-860X

Any correspondence concerning this service should be sent to the repository

administrator: staff-oatao@inp-toulouse.fr
 A new fast method for ceramic foam impregnation: Application
to the CCVD synthesis of carbon nanotubes
Anne Cordier a, Fabrice Rossignol b, Christophe Laurent a, Thierry Chartier b, Alain Peigney a,*
a
CIRIMAT, UMR CNRS-UPS-INP 5085, Centre Interuniversitaire de Recherche et d’Ingénierie des Matériaux,
Université Paul-Sabatier, 31062 Toulouse cedex 9, France
b
SPCTS, UMR CNRS 6638, Science des Procédés Céramiques et de Traitements de Surface, ENSCI, 87065 Limoges cedex, France

Abstract
A new process that allows preparing, in a single step, good washcoats of catalytic materials for the catalytic chemical vapour deposition
(CCVD) synthesis of carbon nanotubes (CNTs) in reticulated ceramic foams is reported. It is shown that the washcoats, obtained by impregnation
using viscous slurries made of finely divided powders dispersed in different media, cover the total surface of foams with good adhesions. The
catalytic activity with regards to the CNT synthesis is finally verified, showing that our new fast impregnation process makes possible to get
materials with final architectures suitable for heterogeneous catalysis applications.

Keywords: Ceramic foam; Impregnation process; Washcoat; Synthesis of carbon nanotubes

1. Introduction by blowing air through the foam pores and (iii) drying the
impregnated foam. These three steps may be repeated as much
Several studies have shown the interest to use consolidated as it is necessary before calcination [9]. The thickness and the
ceramic foams as catalyst supports instead of packed powders homogeneity of the washcoat depend on the foam character-
[1,2]. Indeed, the low-pressure drop and the high geometrical istics (material, walls porosity and thickness, pores size,
surface area of ceramic foams lead to a better gas turbulence porosity architecture) and on the intrinsic characteristics of the
and thus, to a higher catalytic efficiency [3,4]. However, slurry (powder grain size and shape, powder loading, nature of
ceramic foams usually exhibit a low specific surface area dispersion medium) [10]. Thus, an optimum slurry viscosity
(1 m2 g1) after consolidation by sintering at high tempera- has to be found in order to get a homogeneous coating of the
ture. In order to overcome this problem, they can be foam using a minimum number of impregnations, while
impregnated by a slurry made of finely divided powders that preventing from pores obstruction. In most of previously
are characterised by a high specific surface area like g-Al2O3 reported studies, three successive impregnations were needed
powders [5,6]. This so-called washcoat is then decorated by the to obtain a good deposit when a reticulated foam was used
catalytic nanoparticles. [11,12], which means a rather long process due to all the
In this method, the main difficulty is to deposit the washcoat necessary drying steps.
on the whole foam surface with a homogeneous thickness and a In this study, we propose an original process allowing the
good adhesion and without closing the foam porosity. The deposition of homogeneous layers of catalytic materials on
coating process currently used consists in: (i) dipping the reticulated ceramic foams in only one step using high viscosity
ceramic foam into a slurry containing the powders, water and slurries. Four different powder dispersion media are tested to
additives [7,8], (ii) removing the excess of slurry by draining or get the slurries. After characterisation, one among the
impregnated foams is chosen and a catalytic chemical vapour
* Corresponding author. Tel.: +33 5 61 55 61 75; fax: +33 5 61 55 61 63. deposition (CCVD) treatment is conducted in order to in situ
E-mail address: peigney@chimie.ups-tlse.fr (A. Peigney). synthesise carbon nanotubes (CNT) [13].

doi:10.1016/j.apcata.2006.10.039
2. Materials and experiments process proceeds in less than 10 min. The combustion product
is first calcined during 1 h at 600 8C to remove remaining
2.1. Commercial ceramic foams carbon residues. Then, an a-alumina-type phase is obtained
after calcination at 1100 8C (900 8C h1, 30 min).
The commercial ceramic foams (Aluminium Martigny,
France) are mainly composed of a-alumina (more than 80%) 2.3. Attrition-milling of the powder
and in lower proportions of mullite and cristobalite. Their shape
is cylindrical (diameter = 35 mm; height = 22 mm) (Fig. 1(a)), The a-Al1.8Fe0.2O3 solid solution powder is milled by attrition
with an opened porosity of 50 pores per inch (ppi), which at 2000 rpm for various durations using a vessel and a rotor made
corresponds to pore diameters between 0.5 and 1.5 mm of Nylon. High purity a-alumina balls of a diameter in the range
(Fig. 1(b)). The wall surface is rough and porous, with a pore from 200 to 300 mm are used as milling agent. The operation is
size of less than 1 mm (Fig. 1(c)). The ceramic foams are performed in ethanol in which 1 mg of dispersant (BEYCOSTAT
impregnated using a slurry made of the chosen catalytic C213, CECA, France) per meter square of powder surface is
material powder and they are further dried in air. Then, the initially introduced. The ratio between the powder volume and
impregnated foams are calcined under flowing air in order to the milling ball volume is fixed at 0.5. After attrition-milling, the
remove the dispersant and any other organic compounds. a-alumina balls and the powder are separated by rinsing in
ethanol. Finally, after a filtering step, the powder is dried in air.
2.2. Preparation of catalytic material powder
2.4. Preparation of slurries of catalytic powder and
The combustion route [14,15], which is used to prepare the impregnation of ceramic foams
catalytic material composed of an oxide solid solution
(Al1.8Fe0.2O3), has already been described in details in a The different slurries are prepared by simply mixing with the
previous paper [16]. Al(NO3)3, 9H2O and Fe(NO3)3, 9H2O are help of ultra-sounds the attrited oxide powder in the dispersion
dissolved in deionised water together with the required amounts medium in the presence of the above-mentioned C213
of citric acid and urea. The mixture of citric acid and urea (25% dispersant. The different tested dispersions media are:
citric acid and 75% urea) is used as the fuel, in a quantity equal
to the double of the so-called stoichiometric ratio. A Pyrex dish - DEG: pure diethylene glycol (Prolabo-VWR),
containing the solution is placed in a furnace preheated at - PEG: mixture of 50 wt.% of aqueous solution of polyethylene
550 8C. The solution immediately starts to boil and undergoes glycol 6000 (diluted at 50 wt.% Prolabo-VWR) and 50 wt.%
dehydration. The resulting paste froths and then blazes. No diethylene glycol,
flame occurs and a rather light material is produced which - PVA 4/125: aqueous solution of RHODOVIOL 4/125 (diluted
swells to the capacity of the Pyrex dish. The total combustion at 10 wt.% Prolabo-VWR),

Fig. 1. (a) Macroscopic view of the commercial ceramic foam; (b) macroscopic view of the commercial ceramic foam porosity; (c) SEM image of the pore wall.
 Fig. 2. Scheme of the process used for the impregnation.

- PVA 30/5: aqueous solution of RHODOVIOL 30/5 (diluted at 2.6. Characterisations

4 wt.% Prolabo-VWR).
The specific surface area of the a-Al1.8Fe0.2O3 powder
The slurries are kept under constant magnetic stirring during before and after attrition was measured by the BET method
1 h to reach a good homogeneity. The flowchart of the process is using N2 adsorption at liquid N2 temperature (Micromeritics
shown in Fig. 2. At first, the glass containing the foam is Flow Sorb II 2300). The particle size distributions of powders
submitted to vacuum using a water jet filter pump, so that the are determined using a laser particle size analyser (Malvern
slurry is allowed to flow down into the box. However, since the mastersizer 2000). The viscosity of the slurry is measured at
viscosity is too high to penetrate satisfactorily within all the 20 8C for different shear rates (from 65 to 775 s1) using a
pores of the ceramic foam, a complete impregnation is viscometer (TVe-05 Lamy) and following the DIN 53 019 and
performed upon the admission of air until the atmosphere NF EN ISO 3219 standards. The starting commercial ceramic
pressure is reached. This process is performed alternatively for foams and the impregnated foams (F1–F4) are observed by
each side of the foam without any intermediate drying steps. SEM (JEOL JMS 35CF) in order to evaluate the washcoat
Then, the impregnated foam is dried at room temperature and quality. The washcoat amounts are measured by weighting the
calcined in air at 600 8C (150 8C h1, 60 min). The four foams before and after the impregnation and calcination steps.
impregnated foam will be designated in the following as F1, F2, Then, after the CCVD treatment, the resulting nanocomposite
F3 and F4, prepared using DEG, PEG, PVA 4/125 and PVA 30/ foams are observed by FEG-SEM (JEOL-6700F) in order to
5, respectively. The best one will be selected for use as the evaluate both the quantity of CNT and the overall quality of the
catalytic material for the CCVD formation of CNTs. deposited carbon (we consider that a high carbon quality is
reflected by a high proportion of CNT compared to other carbon
2.5. CCVD synthesis of carbon nanotubes forms). The carbon content is measured by flash combustion.

The selected impregnated foams are transformed into a 3. Results and discussion
nanocomposite foam by a CCVD treatment, which consists in a
heating under a H2–CH4 gas mixture (20 mol.% CH4, heating 3.1. Attrition-milling
and cooling rates at 5 8C min1, maximum temperature
1025 8C, no dwell) [17]. By this process, a CNT-Fe/Fe3C– As shown in a previous paper, the product obtained by
Al2O3 composite foam is obtained, with a very homogeneous combustion and further calcinations is monophased and
dispersion of the CNTs around the oxide grains. composed of a a-Al1.8Fe0.2O3 solid solution [16]. This powder

Fig. 3. (a) The particle size distribution for different attrition-milling times. (b) The mean particle size vs. the attrition-milling time.
 Table 1
Values of the flow consistency index K and the flow behaviour index n for the
different slurries
Dispersion medium K (Pa sn) n
DEG 0.472 0.7387
PEG 1.229 0.7286
PVA 4/125 0.206 0.8480
PVA 30/5 0.445 0.7011

index and n is the flow behaviour index, which traduces the

deviation from the behaviour of a Newtonian fluid.
The values of K and n are reported in Table 1. Whatever the
slurry, the flow behaviour indexes n are less than unity, which
Fig. 4. The apparent viscosity of the different slurries vs. the shear rate. indicates a shear-thinning behaviour. The values of n are almost
equivalent, excepted for the slurry prepared with PVA 4/125,
consists of a few tens of micrometers veil-like agglomerates with showing a less pronounced deviation from Newtonian fluid
a specific surface area of 2.7  0.3 m2 g1. Since Agrafiotis et al. (Table 1). Such a shear-thinning behaviour is common for
[5] have shown that the adhesion of an alumina deposit on a solutions made of large polymeric molecules in a solvent.
honeycomb ceramic is greatly improved by a slurry composed of Indeed, on increasing shear rate, long molecular chains
fine particles, we intend to decrease the grain size while gradually align themselves and produce less resistance.
increasing at the same time their specific surface area by However, the slurries differ by the values of their flow
implementing an attrition-milling treatment. The evolution of the consistency index K (Table 1). K corresponds to the value of the
grain size during durations of attrition-milling up to 270 min is viscosity for a shear rate equal to 1 s1. Thus, the higher this
reported in Fig. 3(a). Up to t = 60 min, a spectacular decrease of value, the more viscous the slurry is at low shear rates. The K
the particles size is observed. For longer attrition-milling times, value for the slurry prepared with PEG is three times higher
the decrease is much lower and progressively the mean grain than for those prepared with DEG or PVA 30/5 and six times
diameter tends to a minimum value around 250 nm at higher than for the one prepared with PVA 4/125 (Table 1).
t = 270 min (Fig. 3(b)). There are about four orders of magnitude Shear rates, imposed to the slurry during impregnation in the
between this mean particle size and the pore size of the foam commercial ceramic preform pores with diameters around a
walls, which allow a good adhesion and prevent from obstruction millimeter, are expected to be low. Thus, the most efficient
of the porosity. The 270 min attrition-milling treatment was impregnation should be obtained for slurries prepared with PVA
selected as the most appropriate for this work. After drying the 4/125.
milled powder, the specific surface area is 25.1  0.7 m2 g1,
that is to say ten times larger than that of the non-milled powder. 3.3. Influence of the dispersion medium on the oxide
deposits
3.2. Viscosity of the different slurries
The commercial ceramic foams are impregnated by the
The apparent viscosities of the different slurries made of slurries as described in Section 2. Then, they are dried and
30 wt.% of attrited powder in the conditions described in calcined at 600 8C for 1 h, with a heating rate of 150 8C h1.
Section 2.4, are reported in Fig. 4. The observed apparent The as-obtained F1, F2, F3 and F4 samples are weighted,
viscosity variations as a function of the shear rate can be fitted observed by SEM, and the characteristics of the oxide deposits
by the Ostwald–De Waele model: are compared (Table 2). The relative weights (wt.%) of the
deposits, calculated for 100 g of dry impregnated foam, are
h ¼ Kg n-1
reported in Table 2. The thickness of the deposits, their
where, h is the apparent viscosity (Pa s) and g is the shear rate homogeneity and the possible presence of cracks, which are
(s1). The constant K (Pa sn) represents the flow consistency evaluated upon several SEM images (see examples in Fig. 5),

Table 2
Macroscopic and microscopic characteristics of the impregnated ceramic foams
Sample Dispersion medium Deposit (wt.%) Deposit thickness (mm) Cracks Microscopic homogeneity
Minimum Average Maximum
F1 DEG 36.6  3.7 0 29 52 Many Poor
F2 PEG 24.1  2.4 5 22 44 Few Acceptable
F3 PVA 4/125 24.5  2.5 15 30 47 Very few Good
F4 PVA 30/5 29.4  2.9 0 35 52 Few Poor
Fig. 5. SEM images of the deposits obtained using different dispersion media: (a) and (b) DEG, (c) and (d) PEG, (e) and (f) PVA 4/125, (g) and (h) PVA 30/5.

are also given in Table 2. It is to be noticed that the average, the F1 sample, obtained with a slurry prepared with DEG, is hence
minimum and the maximal values of the thicknesses are given of poor quality in term of homogeneity. On the contrary, the
to give an idea of the regularity of the oxide layer. deposit of the F2 foam covers all the wall surfaces of the foam
For the F1 foam, the relative weight of the deposit (36.6%) is (thickness between 5 and 44 mm), although it still contains both
slightly higher than that of the other foams (between 24.1 and large and small cracks (Fig. 5(c) and (d)). The deposit in the F2
29.4%), but its thickness appears to be very irregular (between 0 sample, corresponding to a slurry prepared with PEG, has
and 52 mm). The zero value corresponds to non-covered areas globally some acceptable characteristics. However, all the
of the foam walls (Fig. 5(b)). Moreover, many large and small characteristics of the deposit in the F3 sample are clearly better
cracks can be observed on this deposit (Fig. 5(a) and (b)). The than those of the other samples. Indeed, the deposit thickness is
more regular (thickness between 15 and 47 mm) and all the there is no reason to suspect that the deposition process used in
surface of the foam substrate appears to be covered by the oxide the present work in order to obtain this material as a washcoat
layer with only very few small cracks detected (Fig. 5(e) and produced any change in the distribution of the Fe3+ ions.
(f)). The microscopic homogeneity is also better than that of the The CCVD treatment, a selective reduction of a-
other foams. Instead, for the F4 foam, the deposit appears to be Al1.8Fe0.2O3 in H2–CH4 atmosphere, leads to the formation
rather irregular (thickness between 0 and 52 mm) and many of nanometric Fe particles at a relatively high temperature
areas of the foam substrate are not covered by the oxide layer [13,17,18] and those located at the oxide grain surface
(Fig. 5(h)), although this deposit contains few cracks immediately catalyse the decomposition of CH4 and the
(Fig. 5(g)). Thus, in spite of the presence of less numerous subsequent nucleation and growth of CNT. Each active particle
cracks and good microscopic homogeneity, the deposit in the leads to one CNT, no more than a few nanometers in diameter.
F4 sample, corresponding to a PVA 30/5 slurry, is considered to Post-reduction Mössbauer spectrometry studies [17,21] have
be globally of poor quality. Thus, we conclude that the revealed that the reduced species present in the specimen are
increasing order of quality of the deposits in the impregnated a-Fe, Fe3C and a g-Fe–C alloy and a correlation of the
foams is the following F1, F4, F2, F3. proportion of theses species with the proportion of CNT
The poor homogeneity of F1 and F4 deposits cannot be allowed to show that the nanoparticles responsible for the
attributed to a too high viscosity of the slurries at low shear rate, formation of the CNT are in g-Fe–C form at high temperature
as a better deposit is obtained with the more viscous slurry but are found as Fe3C by post-reduction Mössbauer analysis. It
prepared with PEG (Table 1). The new one step impregnation has been shown that the majority of the CNTs are double-
process proposed in this paper is thus, suitable for the foam walled, which is discussed elsewhere [18,22] along with the
impregnation by slurries of consistency index up to 1.2 Pa sn. CNT formation mechanisms. The fact that the solid solution is
So, we can infer that most of the defects come from a powder in the form of a washcoat should not basically change this,
re-agglomeration during the process, likely because of the use although one may expect that the reduction may be easier than
of the phosphate ester dispersant that may not be suitable for when it is in the form of a powder bed. This could cause the
DEG and PVA 30/5 dispersion media. Agglomeration can lead formation of metal surface particles at slightly lower
to the blocking of a part of the pores of ceramic foams, and thus, temperatures.
to a non-complete deposition during impregnation. The foams The CCVD treatment, performed on the impregnated foam
impregnated by the slurries prepared with PEG or PVA 4/125 F3, leads to a nanocomposite foam. The global carbon content
have a homogeneous deposit. The foams F2 and F3 could in measured is 1.9  0.1 wt.%. When this value is reported to the
principle be used for the CCVD synthesis of CNT. However, as proportion of catalytic material (24.5  2.5 wt.%, Table 2), the
the deposit in the F3 sample, corresponding to a slurry prepared carbon content becomes 7.6  1.2 wt.%. This value can be
with PVA 4/125, exhibits the best characteristics among the compared to those obtained using the same solid solution (a-
four samples, we have retained this impregnated foam to Al1.8Fe0.2O3) as catalytic material in the form of a powder bed.
perform the synthesis of CNT. It is similar to that obtained with a non-milled powder [18] and
slightly higher than that obtained with an attrited-milled
3.4. Application to the CCVD synthesis of CNT powder (7.6 versus 5.8 wt.%) [19]. This latter result probably
reflects a better CH4 supply to the catalytic material when
A Mössbauer spectrometry study [16], both at room deposited as a washcoat.
temperature and 80 K, of the starting oxide (a-Al1.8Fe0.2O3) The quality of the carbon was investigated by SEM. Some
after calcination at 1100 8C showed that the spectra could be examples of the obtained images are given in Fig. 6. The low
adequately fitted with a ferric quadrupole doublet representing magnification image of the nanocomposite foam surface
Fe3+ ions substituting for Al3+ ions in the corundum lattice. (Fig. 6(a)) shows a large amount of filaments covering that
Another study in the attrited powder [23] revealed no change, and surface. The observation at high magnification (Fig. 6(b))

Fig. 6. SEM images of CNT grown on the nanocomposite foam made with PVA 4/125 dispersion medium: (a) low magnification image, (b) FEG-SEM high
magnification image.
shows that these filaments have the usual morphological prepared using different dispersion media based on DEG, PEG,
characteristics of bundles of CNT [18], between 10 and 20 nm PVA 4/125 and PVA 30/5. Only impregnation by slurries based
in diameter, which interconnect the oxide grains. The small on the use of PVA 4/125 and PEG allows getting homogeneous
species appearing at the surface of the oxide grains (white dots deposits, covering the total surface of the foam. Other slurries
Fig. 6(b)) are metallic nanoparticles, which have not been lead to poor quality deposits, mostly due to the use of a non-
activated for the CNT synthesis and which are usually covered appropriate dispersant in these systems rather than to a viscosity
by carbon capsules [18]. The carbon synthesised during the that would be too high. The catalytic activity of the foam
CCVD treatment of the impregnated foam is, however, mostly impregnated by the slurry based on PVA 4/125 is finally tested
in the form of CNT. No nanofibers are observed. The large with regards to the CCVD synthesis of CNT. A large amount of
amount of CNT and the good quality of carbon prove that, CNT bundles, without nanofibers, is obtained, in relation with
during the CCVD treatment, the CH4 and H2 supply to the oxide the high reactivity of the catalytic material and the efficient
grains was efficient. supply in H2 and CH4. Thus, this new fast process is suitable for
The new process proposed in this paper allows to impregnate the impregnation of reticulated foams for heterogeneous
a reticulated ceramic foam in only one step without decreasing catalysis application.
the catalytic material activity. These results are in good
agreement with those obtained using a Mg0.9(Co0.75Mo0.25)0.1O
oxide as the washcoat [11], in spite of the fact that the CNT References
content remains much smaller in the present work. However,
[1] H. Ogawa, H. Horie, In Jpn Kokai Tokkyo Koho, (Nissan Motor Co., Ltd.,
this large difference is similar to what is already observed when Japan), Jp, 1991, , 4 pp..
using these materials as powder beds [18–20]. In fact, the main [2] M. Haruta, Y. Souma, H. Sano, Int. J. Hydrogen Energy 7 (1982) 729.
advantage of using a-Al1.8Fe0.2O3 lies more in its better [3] J.T. Richardson, Y. Peng, D. Remue, Appl. Catal. A 204 (2000) 19.
compatibility with the a-alumina foam and its subsequent [4] J.T. Richardson, D. Remue, J.K. Hung, Appl. Catal. A 250 (2003) 319.
better adhesion than in a gain in the CNT quantity. Although a [5] C. Agrafiotis, A. Tsetsekou, I. Leon, J. Am. Ceram. Soc. 83 (2000) 1033.
[6] A. Sirijaruphan Jr., J.G.G. Rice, R.W. Wei, D. Butcher, K.R. Roberts, G.W.
detailed study of the CNT number of walls distribution was not Spivey, J. Appl. Catal. A 281 (2005) 11.
attempted here, it is worth noting that it appears [23] that using [7] R.A. Clyde, In U.S, (USA), Us, 1975, , 8 pp..
an oxide catalyst as a foam instead of using it as a powder bed [8] Y. Peng, J.T. Richardson, Appl. Catal. A 266 (2004) 235.
favours the formation of single-walled nanotubes over double- [9] A. Worner, C. Friedrich, R. Tamme, Appl. Catal. A 245 (2003) 1.
[10] P. Avila, M. Montes, E.E. Miro, Chem. Eng. J. 109 (2005) 11.
walled. This phenomenon is under study and will be reported
[11] A. Cordier, E. Flahaut, C. Viazzi, Ch. Laurent, A. Peigney, J. Mater. Chem.
elsewhere. 15 (2005) 4041.
[12] P. Jiang, G. Lu, Y. Guo, Y. Guo, S. Zhang, X. Wang, Surf. Coat. Technol.
4. Conclusion 190 (2005) 314.
[13] A. Peigney, Ch. Laurent, F. Dobigeon, A. Rousset, J. Mater. Res. 12 (1997)
In this paper, we introduce a new process to prepare, in only 613.
[14] K.C. Patil, Bull. Mater. Sci. 16 (1993) 533.
one step, a good washcoat of catalytic oxide material in a [15] J.J. Kingsley, L.R. Pederson, Mater. Res. Soc. Symp. Proc. 296 (1993)
reticulated ceramic foam to be further used for the synthesis of 361.
CNT by CCVD. In a first step, the attrition-milling of the [16] A. Cordier, A. Peigney, E.D. Grave, E. Flahaut, Ch. Laurent, J. Eur.
catalytic material allows to obtain a finely divided powder Ceram. Soc. 26 (2005) 3099.
(mean particle size of 250 nm, specific surface area of [17] Ch. Laurent, A. Peigney, A. Rousset, J. Mater. Chem. 8 (1998) 1263.
[18] A. Peigney, P. Coquay, E. Flahaut, R.E. Vandenberghe, E. De Grave, Ch.
25 m2 g1), resulting in an increased catalytic activity and Laurent, J. Phys. Chem. 105 (2001) 9699.
allowing a good adhesion of the oxide to the wall surface of the [19] Ch. Laurent, A. Peigney, E. Flahaut, A. Rousset, Mater. Res. Bull. 35
foam. In a second step, the impregnation of reticulated ceramic (2000) 661.
foams by slurries are performed by a process including [20] E. Flahaut, A. Peigney, W.S. Bacsa, R.R. Bacsa, Ch. Laurent, J. Mater.
Chem. 14 (2004) 646.
successively the setting of the foam under vacuum, pouring of
[21] A. Cordier, V. Gonzaga, A. Peigney, E.D. Grave, Ch. Laurent, in press.
the slurry and then penetration within the foam caused by air [22] E. Flahaut, A. Peigney, Ch. Laurent, J. Nanosci. Nanotechnol. 3 (2003)
admission. Four slurries, with a shear-thinning behaviour and 151.
flow consistence indexes ranging from 0.2 to 1.2 Pa sn are [23] S. Rul, A. Peigney, W.S. Bacsa, Ch. Laurent, in press.