Carbon Nanotubes in Novel Ceramic Matrix Nanocomposites: A. Peigney, Ch. Laurent, E. Flahaut, A. Rousset
Carbon Nanotubes in Novel Ceramic Matrix Nanocomposites: A. Peigney, Ch. Laurent, E. Flahaut, A. Rousset
Carbon Nanotubes in Novel Ceramic Matrix Nanocomposites: A. Peigney, Ch. Laurent, E. Flahaut, A. Rousset
Received 14 October 1999; received in revised form 28 October 1999; accepted 8 November 1999
Abstract
Novel carbon nanotubes±metal±ceramic nanocomposite powders and dense materials have been prepared and their micro-
structure and mechanical properties have been investigated. After a brief review on the structure, synthesis and physical properties
of carbon nanotubes, we describe an original catalytic method that produces ceramic±matrix composite powders that contain in situ
grown nanotubes. The synthesis parameters that favour the obtention of very high quantities of nanotubes are discussed. The
quality of the nanotubes is also addressed. The microstructure and mechanical properties of the materials prepared by hot-pressing
of these powders are presented. The in¯uence of carbon nanotubes in such composites is discussed in view of potential applications.
# 2000 Elsevier Science Ltd and Techna S.r.l. All rights reserved.
Keywords: B. Nanocomposites; C. Mechanical properties
The speci®c surface area of the powders, measured by determined that the bundles are always smaller than 100
the BET method (using N2 adsorption at liquid N2 nm in diameter, appear to be very ¯exible, and some
temperature), are labelled So for the oxide solid solu- have been traced for 100 mm. For given Tr and tr, the
tion, Sn for the nanocomposite powder obtained after increase in the CH4 ratio in the H2-±CH4 mixture
reduction and Son for the oxidized nanocomposite powder increases the quantity of nanotubes but is detrimental to
obtained after a reoxidation treatment in air at 900 C their quality. Indeed, more CNTs bundles appear on
which eliminates all carbon. As proposed elsewhere SEM images for 30% CH4 (Fig. 2c) than for 12% CH4
[4,11,12], the dierence S=SnÿSon essentially repre- (Fig. 2b), but numerous large, short ®laments, also
sents the quantity of nanotube bundles in the composite appear in the ®rst sample. Milling the starting solid
powder. The carbon content (Cn) in the reduced powder solution by the attrition method allows the synthesis of
was determined by ¯ash combustion. The ratio S /Cn , powders in which the relative dispersion of CNTs bun-
can be considered as representing the quality of the dles and matrix grains is considerably enhanced (Fig.
nanotubes, a higher ®gure for S/Cn denoting a smaller 2d). Both MWNTs and SWNTs, a few nanometers in
average tube diameter and/or more carbon in tubular form diameter, are found on HREM images of nanocompo-
[4,11,12], which we consider a better quality of carbon. site powders (Fig. 3a,b). The CNTs are free of deposits
of amorphous carbon when the quality parameter is
3.2. Results and discussion high (S/Cn 5 200 m2/g). Most MWNTs appear to
have only 2 or 3 walls. XRD analyses conducted on
Generally, the reduced powder is so densely agglom- CNTs±FeAl2O3 powders always show the presence of a-
erated that it retains the shape of the reduction vessel Fe, but sometimes also of Fe3C or Fe5C2.
when transferred in a storage box. As shown on the Increasing the reduction temperature Tr (Fig. 4a,b)
SEM observations (Fig. 2a), the web of CNTs bundles enhances the carbon content in the nanocomposite
gives a cohesion to the nanocomposite powder. We have powder (Cn) and favours the quantity of nanotubes
Fig. 2. SEM images of some nanocomposite powders: (a) 5 wt% Fe, 30% CH4, Tr=1050 C, tr = 1 h; (b) 10 wt% Fe, 12% CH4, Tr = 1070 C, tr =
6 min; (c) 10 wt% Fe, 30% CH4, Tr = 1070 C, tr = 6 min; (d) attritor-milled solid solution, 10 wt% Fe, 12 % CH4, Tr = 1050 C, tr = 1 h.
680 A. Peigney et al. / Ceramics International 26 (2000) 677±683
Fig. 3. HREM images of nanotubes present in the nanocomposite powders: (a) a 3 walls carbon nanotube; (b) numerous SWNTs having similar
diameters (about 3 nm).
Fig. 4. (a): The carbon content (Cn) and S = SnÿSon versus the reduction temperature. (b): S/Cn versus the reduction temperature [10 wt% Fe,
18% CH4, tr = 6 min. for (a) and (b)].
(S) because of a higher CH4 supersaturation level in For CNTs±M-MgAl2O4 nanocomposite powders (M:
the gas atmosphere, but also provokes a decrease in Fe, Co, Ni and their alloys), prepared in the same con-
carbon quality (S/Cn). The macroscopic parameters of ditions than A3, the quantity and quality parameters
the powders which present either the greater quantity depend on the nature of the catalyst [16,17]. Indeed, 7.7
(A1 and MA1) or the better quality (A2 and MA2), wt% of the Fe/Co alloy and 4 wt% Co give the best
obtained with optimized synthesis conditions, are results with respect to the quantity (S = 19.8 m2/g)
reported in Table 1. For CNTs±Fe±Al2O3 powders, the and quality (S/Cn.= 354 m2/g) respectivelly [16,17]. A
best results are obtained with 10 wt% Fe, Tr=1050 C novel CNTs±Co±MgO powder has been prepared show-
and a very short dwell time (6 min), the greater quantity ing that the synthesis method is eective in the case of
(S=24.6 m2/g) and the better quality (S/Cn = 585 MgO-matrix composites, but the conditions have yet to
m2/g) being obtained with a large (30%) and a very be optimized.
small (3%) proportion of CH4 respectively. We have
calculated that a speci®c surface area of 585 m2/g cor-
responds to MWNTs made of 2 or 3 walls which is in 4. Hot-pressed CNTs±Fe±Al2O3 nanocomposites
good agreement with HREM observations. For rein-
forcement applications, we infer that a good compro- 4.1. Experimental methods
mise is found with A3, prepared using 18% CH4. This
powder contains a huge quantity of CNTs bundles : S The powders were hot-pressed in graphite dies at
represents more than 300 000 km of bundles in a 1475 C in a primary vacuum. Massive specimens, in the
gramme of powder [4]. form of pellets 20 mm in diameter and 2 mm thick, were
A. Peigney et al. / Ceramics International 26 (2000) 677±683 681
polished to a ®nish better than 6 mm with diamond was expected from the addition of very long carbon
grids. The densities were calculated from the mass and nanotubes bundles is not observed. Indeed, if the frac-
dimensions of the so-obtained pellets. Scanning electron ture strength of some composites is higher than that of
microscopy (SEM) observations of the surface and Al2O3, it remains lower than those of the carbon free
fracture of some specimen were carried out. We also Fe±Al2O3 nanocomposites [13] Similar observations are
measured the mechanical characteristics of some com- made for the fracture toughness values. No correlation
posites. The fracture strength (f ) was measured by the was found between the mechanical properties and the
three-point-bending test on parallelipipedic specimens quantity or quality parameters. Moreover, specimen E,
(1.8 1.8 16 mm) machined with a diamond saw. The prepared with a composite powder issued from an
fracture toughness (KIc) was measured by the single- attritor-milled solid solution, has even poorer mechan-
edge-notch-beam (SENB) method, on similar specimens ical properties than the other specimens.
notched using a diamond blade 0.3 mm in width. The SEM images of fractures (Fig. 5) show that the CNTs
calibration factor proposed by Brown and Srawley [15] bundles remain present in the hot-pressed alumina-
was used to calculate the SENB toughness from the matrix composites, but in a smaller quantity than in the
experimental results. Cross-head speed was ®xed at 0.1 starting powder. In specimen E (Fig. 5a,b), the bundles
mm/min. are very homogeneously dispersed between the sub-
micrometric sized alumina grains and the cohesion
4.2. Results and discussion between the CNTs bundles and the matrix appears to be
rather poor. Some CNTs bundles are tight, others are
From the densities of massive specimen (Table 2), relaxed (Fig. 5a) and the CNTs which compose the
taking into account the densities and the proportions of bundles are resolved (Fig. 5b). On the surface fracture
the dierent carbon species, we have calculated that the of C2 (Fig. 5c) some bundles, which seem to have been
relative densities are in the 91±100% range. Previous entrapped in the matrix grains during hot-pressing, are
works [13,14] have shown that the hybridization of cut near the grain surface, suggesting some degree of
microcomposites and nanocomposites could result in a bundle pull-out. On the same image, it can be seen that
further improvement in both the fracture strength and the bundles have been constrained and are bent without
fracture toughness. The mechanical properties obtained being broken. The fracture are intergranular for speci-
on CNTs±Fe±Al2O3 nanocomposites (some being men E and mostly mixed intergranular-transgranular
reported in Table 2) show that the additive eect that for the other specimens. We have also determined that
Table 1
Parameters of the nanocomposites powders which contains the greatest quantity of noanotubes (A1 and MA1), which contains nanotubes of the
better quality (A2 and MA2)and which presents a good compromise between quantity and quality (A3), with Al2O3 (labelled Ai) and MgAl2O4
(labelled MAi) matrices. The parameters of a novel CNTs±Co±MgO nanocomposites powder are also mentiones
Code Wt % metal Matrix Cn (wt %) Sn (m2/g) Son (m2/g) S (m2/g) S/Cn (m2/g)
Table 2
Some characteristics of the powders, density and mechanical properties of the massive (hot-pressed) composites. For E, the oxide solid solution was
attritor-milled before the reduction treatment. Some data on a-Al2O3 and 5 and 10 wt% Fe-Al2O3 nanocomposites (from Ref. 13) are shown for the
sake of comparison
Code wt%Fe Cn (wt%) S (m2/g) S/Cn (m2/g) ds (g/cm3) f (MPa) KIc (MPam1/2)
Fig. 5. Scanning electron micrographs of the fractures of some composites: (a,b) specimen E; (c) specimen C2.
the intergranular Fe particles are smaller in CNTs±Fe± may impair the mechanical resistance and a partial
Al2O3 than in Fe±Al2O3 nanocomposites probably annihilation of the dierent reinforcement mechanisms
because the graphene sheets which wrap up these parti- may operate. Possibly the nanotube volume fraction
cles in the powders inhibit some diusion processes should be greatly increased : we have calculated that the
during hot-pressing. CNTs volume fraction is about 20% in the powder but
In the absence of CNTs [13], the fractures are essen- it probably decreases notably during the hot pressing.
tially transgranular because the intragranular metal We presently work to increase the CNTs volume frac-
nanometric particles promote such a fracture, even for a tion in the composite powders and to optimize the hot
matrix with a micrometric grain size. Thus, in CNTs± pressing treatment with the aim to increase the relative
Fe±Al2O3 composites, the reinforcement mechanisms density and to preserve the CNTs from degradation
involving the intragranular metal particles would be far during this process. Moreover, the hot pressing of
less operative. However, some of the SEM observations CNTs±M±MgAl2O4 and CNTs±M±MgO (M: Fe, Co,
seem to indicate that the nanotubes bundles could dis- Ni and their alloys) nanocomposites is in progress to
sipate some fracture energy, particularly for composites evaluate the in¯uence of the nature of the ceramic
with micrometric matrix grains. The residual porosity matrix and metal particles on the microstructure and
A. Peigney et al. / Ceramics International 26 (2000) 677±683 683
mechanical properties. Possibly, CNTs bundles should show that the reinforcement eect that was expected
be aligned to contribute to the reinforcement. from the addition of very long carbon nanotubes bundles
is not observed. The improvement of the composite
microstructure, the change in the nature of the matrix
5. Conclusions and attemps to align the CNTs are works actually in
progress.
Several experimental works have recently con®rmed
the theorically predicted outstanding mechanical prop-
erties of carbon nanotubes (CNTs). Consequently,
CNTs emerge as potentially attractive materials as References
reinforcing elements in ceramic matrix composites. [1] S. Iijima, Nature 354 (1991) 56.
The ®rst diculty to achieve a CNTs-ceramic com- [2] M.R. Falvo, G.J. Clary, R.M. Taylor II, V. Chi, F.P. Brooks Jr.,
posite from as-prepared CNTs, is to disperse homo- S. Washburn, R. Super®ne, Nature 389 (1997) 582.
geneously the CNTs in a ceramic powder. We have [3] M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund, Science of full-
proposed a novel catalysis method for the in-situ pro- erenes and carbon nanotubes, Academic Press, San Diego.
[4] A. Peigney, Ch. Laurent, A. Dobigeon, A. Rousset, J. Mater.
duction, in a composite powder, of a huge amount of Res. 12 (1997) 613.
CNTs bundles, which cover all the matrix grains look- [5] S.B. Sinnott, C.T. White, D.W. Brenner, Mat. Res. Soc. Symp.
ing like a web. The elaboration conditions of CNTs±Fe± Proc 359 (1995) 241.
Al2O3 and CNTs±M±MgAl2O4 (M: Fe, Co, Ni and [6] M.M.J. Treacy, T.W. Ebbesen, J.M. Gibson, Nature 381 (1996)
678.
their alloys) nanocomposite powders have been opti-
[7] W.G. Wildoer, L.C. Venema, A.G. Rinzler, R.E. Smalley, C.
mized with the aim to obtain both a great quantity of Dekker, Nature 391 (1998) 59.
CNTs (up to 300 000 km of CNTs bundles in a gramme [8] S. Iijima, T. Ichihashi, Nature 363 (1993) 603.
of powder) and a good quality of carbon, i.e. a smaller [9] T. Guo, P. Nikolaev, A. Thess, D.T. Colbert, R.E. Smalley,
average tube diameter and/or more carbon in tubular Chem. Phys. Lett. 243 (1995) 49.
form. Because these two parameters generally vary in [10] K. Hernadi, A. Fonseca, J.B. Nagy, D. Bernaerts, A. Fudala,
A.A. Lucas, Zeolites 17 (1996) 416.
opposite directions, a compromise must be found. For [11] Ch. Laurent, A. Peigney, A. Rousset, J. Mater. Chem. 8 (5)
reinforcement applications, with CNTs±Fe±Al2O3 pow- (1998) 1263.
ders, we infer that a good compromise is obtained with [12] Ch. Laurent, A. Peigney, O. Dumortier, A. Rousset, J. Eur.
18% CH4 in the H2±CH4 gas mixture, 10 wt% Fe and a Ceram. Soc. 18 (1998) 1995.
reducing thermal treatement performed at 1050 C with [13] X. Devaux, Ch. Laurent, M. Brieu, A. Rousset, in: Composites
Materials, eds A. T. Di Benedetto, L. Nicolais, R. Watanabe,
a short dwell time (6 min). Elsevier Science Publishers, Amsterdam, Netherlands, 1992.
Massive CNTs±Fe±Al2O3 composites have been pre- [14] O. Quenard, Ch. Laurent, M. Brieu, A. Rousset, NanoStruct.
pared by hot-pressing. SEM observations show that the Mater. 7 (1996) 497.
CNTs bundles remain present in the composites, but in [15] W.F. Brown, J.E. Srawley, ASTM Spec. Tech. Pub., 410, ASTM,
Philadelphia, PA, USA, 1966.
a smaller quantity than in the starting powder.
[16] A. Govindaraj, E. Flahaut, Ch. Laurent, A. Peigney, A. Rousset,
Although some of the SEM observations seem to indi- C.N.R. Rao, J. Mater. Res. 14 (6) (1999) 2567.
cate that the nanotubes bundles could dissipate some [17] E. Flahaut, A. Govindaraj, A. Peigney, Ch. Laurent, A. Rousset,
fracture energy, the obtained mechanical properties C.N.R. Rao, Chem. Phys. Let. 300 (1999) 236.