Comparison of two soft chemistry routes for the synthesis of silicon carbide
X. Deschanels, G. Arrachart, D. Herault1, M El Ghazzal, F. Hodaj2, C. Delchet, C. Rey, J.
Ravaux, T. Zemb
ICSM-UMR 5257 - CEA/CNRS/UM2/ENSCM Marcoule, BP 17171, 30207 Bagnols-surCèze, France.
1
SM2 UMR 7313 - Centre Universitaire de St Jérôme. EGIM Sud. Case A 62 Av. Escadrille
Normandie-Niemen, 13397 Marseille Cedex 20.
2
SIMaP – UMR CNRS 5266, Grenoble INP UJF, BP 75 38402 St Martin d'Hères Cedex.
Abstract
We compare the influence of molecular and colloidal precursors on the synthesis of a hybrid
material containing SiC and porous carbon. The temperature of production of crystalline SiC
is remarkably independent of the synthesis route chosen. The excess of carbon in the initial
mixture is the source of the excess porous carbon binding the crystalline SiC domains in the
final products. Increasing initial surface of contact between carbon and silicon translates into
mesoporosity. Simultaneous control of the division state and relative amounts of Si and C into
the precursors allow the control of textural and chemical characteristics of the final powder
obtained. A simple mechanism is proposed to explain the evolution of the surface area as a
function of the volume fraction of residual carbon in the powder synthesized.
Introduction
Silicon carbide is a compound with excellent mechanical strength, thermal stability, and good
thermal conductivity properties such as high strength, high thermal conductivity1. Because of
these properties, SiC is used in advanced ceramics applications such as: electronics (diodes,
semi conductor devices), technical ceramics (cutting tools, abrasive, brake discs, heating
elements), catalysis (support at extreme process conditions) as well as nuclear industry
(expansion layers for the nuclear fuel elements of high temperature reactors). Industrially, the
Acheson2 process is generally used for the production of silicon carbides. This process
includes a carbothermal reduction of silica sand by petrol coke. The main drawback of this
process is the use of high temperature reaction (1800-2200°C), which leads to coarse-grained
powders with low surface area (0.1-15 m2/g) 3.
One possible way to lower the temperature of this synthesis product is to reduce the size of
the initial products from micrometric to nanometric scale4. The ultimate reduction in size is
the inclusion of C and Si in the same molecule. A recent review shows examples of sizedependent effects in nanomaterials 5, for example the melting temperature of 2.5 nm gold
particles was reported to be about 930K, much lower value than its bulk value of 1336K6.
This is related to the fact that a larger fraction of atoms is located at the surface of the
nanomaterial.
Several studies suggest that the temperature of the carbothermal reduction of SiO2 by carbon
decreases when colloidal or nanometric precursors are involved in the reaction. H.P. Martin7
uses nanosized precursors (colloidal silica oxide and sugar) to synthesize -SiC at
temperature as lower as 1300°C. Kevorkijian8 synthesizes -SiC powders by carbothermic
reduction of carbon-black and colloidal silica at temperature ranging from 1330°C to 1800°C.
Lin 9 studies the effects of the starting products: quartz or colloidal silica, and carbon sources
(acetylene carbon black and phenolic resin) on the carbothermal synthesis of the SiC. Finally,
Ishihara
10
synthesizes silicon carbide powders by application of sol-gel processing starting
with fumed silica powder as silicon source, and phenolic resin as carbon source.
Another attempt to decrease the temperature of the synthesis of SiC is to include the silicon
and carbon atoms in the same molecule. For example, after the pyrolysis under inert
atmosphere of the poly(silylene)-diacetylene (1400°C), Corriu
11
describes the crystallization
of -SiC. Depending on the composition of the polymer, the ratio C/Si can be adjusted to
produce a composite SiC with an excess of carbon. The formation of sp2-carbon matrix
surrounding SiC particles leads to efficient reducing conditions and can be used for the
carbothermal reduction of other oxides. By this way, mixed-metal carbides ceramics
(SiC/TiC, SiC/ZrC…) were prepared at lower temperature, i.e. few hundred degrees,
compared to that using the carboreduction of the parent oxides with carbon black 4. Others
authors12 describe the usage of carbonaceous silicon gels for the synthesis of SiC. Moreover,
the synthesis route and the processing conditions (nature and size of the precursors) have a
great influence onto the physical characteristics of the final products, particularly on the
surface area (Table1).
Silicon carbide is an interesting compound for catalysis application
13,14
. To be useful as
catalyst support, SiC must be prepared with high surface area, typically higher than 20m2/g. In
addition to the effect on temperature synthesis of SiC, the different methods presented above
also impacts the final characteristics of silicon carbide obtained. Overall, these methods can
be classified into two categories. Direct methods consist of the reaction of the oxide or the
metal oxide with solid carbon at high temperature and lead to powders with low surface area.
In contrast, other synthesis routes involving reaction between a liquid, a gaseous phase or the
pyrolysis of a gel, the decomposition of an organic precursor leads to compounds with higher
surface area. The results of these two routes are compared in table 1.
In order to quantify the relation between initial Si/C surface of contact, and final porosity, we
have chosen to study a simple system in which the characteristics of precursors can be
modified at different scales: molecular, nanometric, and finally micrometric. This system
consists of a mixture of carbohydrate in the form of simple “sugar” and silica. In the present
study, two different routes are compared for the production of silicon carbide powders.
The first, called “colloidal route” is similar to the one used by H-P Martin7b and consists in
the carbo-reduction of a colloidal silica (nanometric or micrometric powders) by sugar. The
second route, called “molecular route”, uses a molecular chemistry approach based on the
immobilisation of polycarbosilane in sugar. Both synthesis routes have in common the use of
an identical carbon source to achieve the carboreduction. The physical characteristics of the
final products (surface area, grains size…) are discussed in terms of the nature of the oxide
precursors: colloidal silica or fumed silica, the composition of the sugar: mannitol, sorbitol,
sucrose, and sugar/oxide ratio.
Experimental
Synthesis of the precursors
Colloidal precursors
The overall carbothermal reduction of a silica oxide can be described as depicted in equation
(1) :
SiO2(s) + 3C(s) → SiC(s) + 2 CO(g)
(1)
Several carbohydrates (sucrose, sorbitol, mannitol) were used as carbon sources, all these
products were purchased from Aldrich. One considers the pyrolysis of the carbohydrate
according to the following reactions:
C12H22O11
∆
12C + 11H2O
(2)
∆
C6H14O6
∆
C42H70O35
6C + 6H2O + H2
(3)
42C + 35H2O
(4)
Others gaseous species are produced during the thermal decomposition of such
carbohydrates15 (CH4, CnH2n, CO2, CO) inducing a loss of the available carbon usable for the
carbothermal reaction (5), (6). However these reactions are used to adjust the amount of
carbon in the precursors. Silica oxides (fumed and micrometric silica) were used to elaborate
the silicon carbide powders. Fumed silica is amorphous silica with a chain-like particle
morphology composed of submicron-sized spheres, which are highly branched into
aggregates from 10 to 30 spheres in length sizing 0.1 to 0.2 microns. The surface area is close
to 360m2/g which correspond to an effective average particle size of 7 nanometers.
Micrometric silica powders were purchased from Aldrich, which is composed of coarse grains
with a grain in the order of 1-10 micrometers and its surface area is close to 6m2/g. These
oxides were mixed together with a quantity of carbohydrates (sucrose, manitol, sorbitol, cyclodextrine) dissolved into an aqueous solution corresponding to the final composition
targeted. The suspension obtained guarantees a homogeneous dispersive mixing in a simple
way. The amount of carbohydrate is adjusted in a proportion ranging between 1 and 12 times
the quantity needed by the reaction (1) in order to obtain a complete conversion of the parent
compounds. The ratio C/Si quantifies the deviation from stoichiometry, i.e. C/Si=3 means the
reagents are mixed together in the proportion of the reaction (1). This ratio was calculated
considering the decomposition of sucrose or cyclodextrine according to reactions (2), (3) and
(4). After a complete dissolution of sugar in the silica sol the suspension was frozen and
freeze-dried to obtain a powder. The precursors (P1, P2, P3, P4, P5) described in the table 2
were all obtained according to this process. The last stage, corresponding to the thermal
conversion, is described in the next section of this article.
Molecular precursor
The molecular precursors C1 (Figure 1) based on sucrose carbohydrate were obtained
according to the synthesis described below.
Reagent grade 3-(triethoxysilyl)propylisocyanate, were commercially available and used
without purification. All reactions were carried out under argon using a vacuum line and
Schlenk techniques. Solvents were dried and distilled just before use. The synthesis of this
precursor
is
composed
of
two
steps.
Firstly
the
compound
ter-[3-
(triethoxysilyl)propylcarbamate]-saccharose
was
obtained
by
the
reaction
of
3-
(triethoxysilyl)propylisocyanate with saccharose and triethylamine (precursor M1). Secondly
the ter-[3-(triethoxysilyl)propylcarbamate]-saccharose reacts with ethanol to form a xerogel
(precursor X1). The synthesis of these compounds is described below.
- Precursor C1
To a solution of 2 g (5.84 mmol) of saccharose in dry DMF (20 ml) at 90°C were added under
argon 1.77 g (17.5 mmol) of freshly distilled triethylamine and 4.33 g (17.5 mmol) of 3(triethoxysilyl)propylisocyanate. The mixture was stirred at 90 °C for 7 days. The solvent was
then removed under vacuum for 32 hours at room temperature. The resulting oil was
coevaporated with dry toluene several times to obtain a brown solid. This solid was washed
with heptane then dried to obtain 5.5 g (87%) of desired product with a small amount of
polycondensed product.
Elem. Anal. Found: C, 41;97; H, 6.82; N, 3.93; O, 35.3; Si, 8.90. Calc. for C42H85N3O23Si3: C,
46.52; H, 7.90; N, 3.88; O, 33.97; Si, 7.77.FTIR max/cm-1 3334, 2973, 2927, 2886, 1699,
1531, 1444, 1246, 1056, 949; 13C CP MAS NMR 75 MHz (, ppm) 156.9, 103.2, 90.2, 71.7,
58.1, 43.5, 23.2, 18.2, 7.7. 29Si CP MAS NMR 60 MHz (, ppm) : -45.1 (T0), -52.8 (T1)
- Xerogel M1
A suspension of 1 g (0.92 mmol) of ter-[3-(triethoxysilyl)propylcarbamate]-saccharose in
technical ethanol (15 ml) was stirred at 110°C for 24 hours then without stirring for 48 hours.
The reaction mixture was then cooled at room temperature and concentrated for 7 days. The
resulting xerogel was filtered and washed with ethanol, followed by Soxhlet extractions with
ethanol (15 hours) then acetone (15 hours). The final solid was dried under vacuum to give
0.551 g (79%) of brown powder.
Elem. Anal. Found: C, 35.43; H, 5.43; N, 4.48; O, 37.7; Si, 10.01. Calc. for C48H80N6O37Si6:
C, 38.39; H, 5.37; N, 5.60; O, 39.4; Si, 11.22. FTIR max/cm-1 3330, 2934, 2360, 2335,
1693, 1528, 1001, 13C CP MAS NMR 75 MHz (, ppm) 158.7, 104.9, 92.1, 73.5, 63.6, 44.4,
24.6, 11.2. 29Si CP MAS NMR 60 MHz (, ppm) : -57.4 (T2), -64.4 (T3). Level of
condensation > 95%. BET surface area < 10 m2·g-1.
Pyrolysis of the precursors
After the different treatments presented above, the precursors were obtained as white or
brown powders depending on the routes chosen. In a first step, these powders can be
compacted to obtain a better confinement of the silica oxide by the surrounding carbohydrate.
In a second step, the compacts were pre-heated at 800°C during 3 hours in flowing argon to
have a complete decomposition of the carbohydrate into carbon. Then the formation of SiC
was carried out at higher temperature (1100-1550°C) during 4 hours under flowing argon. It
has been suggested that under reducing conditions the silica leads to the formation of the
gaseous SiO species 7a. Thus, the carbothermal reduction of silica can be described as follow:
SiO2 (s) + C(s) → SiO(g) + CO(g)
(5)
SiO(g) + 2C(s) → SiC(s) + CO(g)
(6)
Pressure and temperature have been used by several authors to control the reaction kinetics.
Dynamic vacuum was employed by Zeng12a to decrease the temperature of the carbothermal
reaction at 1100°C. This parameter has not been investigated in this work. Most of the
syntheses are carried out at temperature above 1400°C under flowing argon. In these
conditions reactions (5) and (6) become predominant8b. Experimentally the condensation of
silica indicating the presence of SiO gas was observed in the furnace after the pyrolysis of the
samples.
The conversion of these precursors into silicon carbides has been made in a tubular furnace
and in the TGA apparatus. A flow of argon over 30 l/h was used in the tubular furnace to
evacuate the gaseous species produced during the thermal treatment. For the samples treated
into the TGA apparatus the argon flux was lower 20 ml/min leading to an incomplete
conversion. The products obtained have a colour ranging from gray-blue to black depending
on the excess of added carbohydrate.
Characterization
Infrared spectra were recorded on a FT-IR Perkin-Elmer 1600 neat, in KBr pellets or in a
solution of CCl4. The liquid 1H and
13
C were recorded on a Bruker Avance 300.
29
Si NMR
was recorded with a Bruker Avance 200 at a frequency of 40 MHz. The 29Si CP MAS NMR
spectra were recorded on a Bruker FTAM 300 as were
13
C CP MAS NMR spectra, in both
cases, the repetition time was 10 and 5 s with contact times of 5 and 3 ms. The Tn and Qn
notations are given respectively for ((SiO)n(R)SiO3-n) and ((SiO)nSiO4-n) environments.
Thermal analyses were carried out with a Netzsch STA 409 thermo gravimetric analyzer. The
surface areas, the pore size distribution were obtained by ASAP 2020 using nitrogen gas as
adsorptive. Before these measurements, the samples were degassed at 350°C for 4h under a
vacuum of 10-2 mbar. Elemental analyses of carbon and oxygen were performed respectively
by LECO CS230 and TCH600 analyzers. The oxygen content in the powder was determined
by combustion analysis under helium and IR detection of the CO2-vibration. Combustion
method was also used to determine the carbon content. The SiC powder was heated in an
oxygen flow. Added Fe powder was used to assist the combustion. The CO2 formed was used
again to detect the carbon content. Transmission Electron Microscopy (TEM) observations
were carried out at 100 kV (JEOL 1200 EXII). Samples for TEM measurements were
prepared by milling of the samples in an agate mortar, followed the dispersion of such powder
in a ethanol under ultrasonification and deposition on copper grids. Powder X-ray diffraction
patterns were measured on a Bruker D8 advance diffractometer using CuK radiation in
Bragg-Brentano geometry. SEM analyses were obtained with a FEI QUANTA FEG 200
ESEM environmental scanning electron microscope, coupled with an EDX Bruker XFlash
4010 SDD micro analyzer.
Results and discussion
Various carbohydrates (sucrose, manitol, sorbitol, -cyclodextrin) and silica oxide of different
grain-size (micrometric and nanometric) were used for the synthesis of the carbide precursors
(see Table 2). The ratio C/Si is referred to the number of silicon and carbon atoms in the
molecular precursor or in the blend silica oxide+carbohydrate.
Silicon carbide formation
A ratio C/Si higher than 3 is necessary to obtain single-phase SiC after the heat treatment at
1550°C of the P1-type precursors (sucrose + fumed silica). It is surprising that according to
reaction 1, a ratio C/Si = 3 is enough to synthesize silicon carbide, because a part of the silica
is converted into gaseous SiO and also a part of the carbon is lost due to the breakdown of
sucrose into gaseous species (CO, CO2, CH4 and alkenes). These two antagonistic phenomena
compensate each other. Our results are consistent with those of Martin 7, and Cerovic16. The
main phase detected by XRD in these conditions is the cubic phase (-SiC) (Figure 2),
however a shoulder is observed on the XRD pattern at the position 33.5°<2<34.5°. This
indicates the presence of low amount of -SiC or stacking faults in the sample. Depending on
the thickness of the powder bed analyzed by XRD, two peaks were observed for 43°<2<45°
and 50°<2<52° on the diffraction pattern (grey area). This artefact is related to the signal of
the sample holder used to analyse the powders. Below the ratio C/Si=3, cristobalite was
detected, by XRD, in parallel the oxygen content is very high (16.6 wt.%), indicating the
presence of silica into the sample (Table 3). For a value of C/Si higher than 9, a broad peak
was detected at position 20°<2<28° as the consequence of the presence of amorphous carbon
in the sample (Figure 2). Elemental analysis confirms this point and jointly the amount of
oxygen on the final products decrease down to 0.4wt.% for ratio C/Si=36 (Table 3). The
elemental analysis is more sensitive than XRD analysis to detect the free carbon.
Figure 3 shows the evolution of the X-rays patterns as a function of the temperature. The
formation of silicon carbide occurs for temperature close to 1300°C. After reaction at 1250°C
the sample is almost amorphous: no crystalline phase is detectable by X-rays diffraction, the
sample probably consists of silica and carbon. Otherwise the reaction is complete at 1450°C.
More precisely, TGA analysis shows that the reaction starts when the temperature reach
1300°C and ends at 1550°C (Figure 4). This result is consistent with bibliographic data (Table
1). The evolution of the oxygen content in the samples versus the temperature of the heattreatment is another way to follow the progress of the carbothermal reaction. This parameter
decrease down to a value of 1.1wt.% when the temperature increase up to 1550°C (Figure 5).
Conversely the pyrolysis in the same conditions of the precursors P2 (fumed SiO2 + sorbitol)
and P3 (fumed SiO2 + mannitol) leads to the formation of cristobalite. No silicon carbide was
observed. This result17 can be attributed to the evaporation of these carbohydrates when they
reach their boiling point (296°C and 290°C for sorbitol and mannitol respectively). Unlike
mannitol or sorbitol, sucrose and cyclodextrine are decomposed into carbon and gaseous
species15 at low temperatures (<800°C), and the remaining carbon is efficient for the
carboreduction of the silica at high temperature (>1300°C). TGA (Figure 4) confirms this
point. The conversion of the precursor type-P5 into silicon carbide is comparable to that of the
precursor P1. However, for the same value of the ratio C/Si the oxygen content is higher in
powders made from micrometric silica presumably due to their lower reactivity during the
carboreduction.
The doping of sol–gel matrices with organic molecules with simple mixture of organic and
inorganic matrices is one of the simplest ways to entrap organic molecules in an inorganic
network18. Adjustment of the ratio C/Si is simple however weak interactions between the
organic and inorganic matrices are related to the dispersion of the organic fragment within the
structure, but control of the distribution is a major drawback as segregation can easily occur.
The phase segregation may have an impact during the preparation of silicon carbide; the
covalent approach in the “molecular route” can be employed to overcome this drawback.
Molecular precursor C1 can be used directly in the preparation of SiC with a theorical C/Si
ratio value equal to 14 (C/Si=11.6 from elemental analysis). The hydrolysis-condensation of
bis or tris(trialkoxysilyl) organic molecules led to hybrid silica with covalent linkages
between the organic and the silicate network. Bridged organo-polysilsesquioxanes can be
obtained directly after hydrolysis-condensation reaction of C1 leading to a material M1 which
consist of a highly condensed siloxane networks with organic fragments covalently bonded to
the silica network (level of condensation >95% estimated by
29
Si solid state NMR from the
T2 and T3 silicate centres). This approach leads to a final material M1 with a theoritical ratio
C/Si equal to 8. In order to increase the amount of silicon, Tetraethyl orthosilicate (TEOS) can
be added during the sol–gel process. The C/Si ratio can be modified dealing with the rate of
polycondensation or co-hydrolysis with TEOS.
In our study the preparation of silicon carbide by the “molecular route” was investigated
starting from the molecular precursor C1 and M1 with a theoretical ratio C/Si respectively
equal to14 and 8. The pyrolysis at 1550°C of these two molecular precursors C1 and M1 lead
to the formation of -SiC (Figure 6).
Let us now consider more precisely the transformation of the molecular precursor C1
(molecular route) into carbide. After the pyrolysis of the precursor at 1100°C, XRD analysis
shows no evidence of crystalline silica or other crystalline compounds, but Si, O and carbon
were principally detected by EDX microanalysis. Elemental analyses confirm the presence of
oxygen, residual nitrogen and carbon arriving from the decomposition of the starting
products. These observations demonstrate the presence of amorphous silica and amorphous
carbon in such sample, and one can conclude that the mechanism of the formation of silicon
carbide from these molecular precursors is similar to that observed with colloidal precursors,
in a first step up to 1000°C the product is decomposed into C and SiO2, and in a second step
for temperature higher than 1300°C the carbothermal reduction occurs. TGA analysis of C1molecular route precursor presented on figure 4 is consistent with the mechanism of the
formation of SiC described above.
Effect of the synthesis route on the temperature of the carboreduction
One of the objectives of this study was to evaluate the influence of the route used for the
synthesis of silicon carbide onto the temperature of the carboreduction. Whatever the
synthesis route, two significant mass losses were observed (Figure 4) during the pyrolysis of
the precursors P1 (fumed SiO2 + sucrose) and C1 (molecular route). The first loss observed at
temperatures below 800°C corresponds to the decomposition of the organic precursors. The
second loss is observed for temperature ranging from 1300 to 1500°C can be attributed to the
formation of the carbides. In the molecular route, the silica resulting from the decomposition
of the precursor is associated in a highly cross-linked carbon network which can be suitable
for its carboreduction, while through colloidal route the reactants are mixed together at a
nanometric (P1) or micrometric (P5) scale. In all cases, no evidence of the temperature
change was observed with the nature of the precursors. The carboreduction still starts for a
temperature close to 1300°C. The temperature of the carboreduction for the precursor P5 was
not identified by TGA analysis, but furnace pyrolysis done at 1550°C clearly showed a
complete conversion of the precursor into -SiC.
In term of temperature conversion, our results are similar to those obtained by Corriu 19 in the
case of the pyrolysis of the Poly(dimethylsilylene)diacetylene. In this work, the formation of
-SiC is observed after the pyrolysis of the organosilicon polymer at temperature of 1400°C.
However, the mechanism leading to the conversion of this polymer into a ceramic is different.
In the range of 400 to 800°C, the thermal decomposition of the organosilicon polymer leads to
the formation of an amorphous SiC embedded into a cross-linked carbon network resulting
from the decomposition of the diacetylene groups. Finally, in the work of Corriu the
formation of -SiC was observed after a heat treatment a 1400°C. The main difference is due
to presence of oxygen in our compounds which lead to the formation of silica during the
decomposition of the precursors and leads to a classical carbothermal reduction in the same
way if we had used a precursor oxide.
Results obtained in this study confirm that there is no effect on the size or nature of the
precursors on the temperature of the cabothermal reduction. This result may seem surprising
since chemical Si-C link are already formed into the molecular precursor. In fact, during the
heat treatment, these precursors are first transformed into oxides before undergoing the
carboreduction. However, the use of these two synthesis routes produces a relative decrease in
the carboreduction temperature compared to conventional carboreduction Acheson process,
i.e. which presents a conversion temperature higher than 2000°C. This difference can be
attributed to the molecular intimacy of de SiO2/C mixture. In this case, the reaction is not
limited by diffusion process of species in the solid state as is the case in the Acheson process,
this peculiarity has also been invoked by Cerovic16.
Finally, the synthesis of carbide by the colloidal route offers more flexibility than the
molecular route to convert the precursors indeed it can be easily to adjust the ratio C/Si and
thus have a better control of the composition of the sample.
Characterization of the samples
SEM was employed to observe the morphology of SiC particles synthesized with various C/Si
ratios. SEM observations reveal a similar crystallite size (≈80-120nm) of the powders
elaborated from the colloidal route with the precursors P1 (fumed SiO2 + sucrose) and P4
(fumed SiO2 + -cyclodextrin) for a ratio C/Si close to 4 (Figures 7a, 7b). In contrast the
particles obtained from the micrometric silica, precursor P5, are agglomerated with a
characteristic size higher than 1 micrometer. This is particular clear for samples elaborated for
a ratio C/Si=24. However, these particles are constituted with smaller crystallites (≈200300nm) (Figure 7e).
When the ratio C/Si increases the amount of the carbon gangue increases and the
microstructure of the powders becomes more heterogeneous for all the samples. For the P1type samples (fumed silica + sucrose), the size of the agglomerates is about 700-1000nm and
the size of the constitutive crystallites is about 200nm (Figure 7d). SEM observations of the
powders elaborated from the precursor C1 (molecular route) reveal particles with lower
crystallites size, typically 100nm (Figure 7c). The difference in the crystallite size is a
consequence of the better homogeneity of such materials.
The specific surface areas (SBET) of the various samples pyrolysed at 1550°C are plotted
versus the ratio C/Si on Figure 8. This parameter is strongly dependant on the nature of the
precursor. The evolution of the surface area of the materials P1 (fumed silica + sucrose)
versus C/Si is characterized by 4 steps. In the first step, for ratio C/Si lower than 3 the final
product is a mixture of cristobalite+SiC, the surface area decreases as a consequence of the
sintering of the fumed silica grains. Uchino20 measures a decrease of the surface area of
fumed silica heat-treated at 1000°C, 1100°C and 1200°C, respectively 188-115 and lower
than 1m2/g. In a second step, for values of C/Si ranging from 3,6 to 6 the transformation of
silica into single phase SiC is complete. The surface area varies moderately; it is broadly
representative of the external surface of SiC crystallites formed. The SiC crystallites observed
by SEM have a characteristic size of about 100nm (Figure 7). The value of the diameter of the
particle, d=125nm, estimates from the surface area measurement, if we assume that the SiC
particles are spherical, is in good agreement with this observation, i.e. d=6000/(S) where  is
the density of SiC (3.2g/cm3) and S the surface area 15m2/g.
A sharp increase of the surface area was observed for samples with ratio C/Si between 6 and
12. SEM observations reveal the presence of a much more important matrix of porous carbon
coating the SiC grains (Figure 7) as a consequence of the high amount of residual carbon. No
significant change was observed in the grain size of the SiC crystallites (100nm), confirming
that the porosity is located in the carbon matrix surrounding the silicon carbide grains. The
maximum surface area developed by this material (200m2/g) is one order of magnitude greater
than the outer surface that would be developed (10m2/g) by powders with a grain size of 100
nm which supports the hypothesis of the location of the porosity in the matrix of residual
carbon.
For values of C/Si ranging from 12 to 36, the specific surface area of the powders decreases to
tend to the surface area of sucrose (1m2/g) pyrolyzed under the same conditions. In contrast,
materials made from micrometric silica + sucrose (P5) has low specific surface area even
when they have high residual carbon content (Table 2). This difference will be discussed in
the next section dealing with the mechanism of formation of porosity into these powders.
Finally, it is clear that the molecular route does not offer much flexibility in the variation of
the ratio C/Si in the sample. Therefore, the only points of comparison between the molecular
and the colloidal route were obtained for ratio C/Si equal to 8 and 14. The specific surface of
such materials are the highest measured i.e. 722m2/g for C1, 567 m2/g for M1, and the
morphology of its particles doesn’t present intra-particle porosity. The surface area of the
samples C1 and M1 is nearly the same which is consistent with the final amount of carbon
measured on these samples after heat treatment at 1550°C (see Table 3). This result can be
interpreted by the fact that although the C/Si is higher in the precursor C1, the alkoxy groups
associated with the silicon is removed at low temperatures and are not operative for
carbothermal reaction. The grain size of the SiC particles is close to 70-100nm, so we can
conclude as the precursors of type P1 (fumed silica + sucrose), the porosity is located in the
residual carbon phase.
Formation of the mesoporous structure
The evolution of the specific surface area of all the materials from the various precursors (C1,
P1, P5) are reported as a function of volume fraction of carbon on figure 9. The volume
fraction of carbon is inferred from the results of the chemical analysis of this element, by
considering that the final material consists of only 2 phases SiC and carbon. As shown on this
figure, the nature of the precursor has a significant impact on the specific surface area. The
highest surface was obtained for C1 type sample elaborated from the molecular precursor
(722m2/g), the samples P1 (fumed silica + sucrose) have an intermediate surface (200m2/g)
and finally the samples P5 (micrometric silica + sucrose) present the lowest values. The
specific surface area of P1 type materials (fumed silica + sucrose) increase up to 200m2/g for
a volume fraction of carbon close to 50% and then decrease to reach the surface area of
sucrose pyrolyzed at 1550°C. The adsorption-desorption isotherm of the sample P1 (C/Si=12)
(Figure 10) undergoes an abrupt changes when the relative pressure is in the medium range
from 0.6 to 0.9, and this is salient feature of mesoporous materials. The shapes of the
isotherms and the hysteresis loop suggest that the sample possesses a mesoporous structure
located in the matrix of residual carbon.
The evolution of the specific surface area versus volume fraction of carbon can be interpreted
as consequence of the release of CO gas during the carbothermal process at higher
temperature. Further investigations were performed on the evolution of the surface area versus
the pyrolysis temperature for the samples P1 (fumed silica + sucrose) at fixed ratio C/Si=12
(see table 1). After pyrolysis at 800°C, the powders possess a high surface area close to
250m2/g largely related to the surface area of the fumed silica. Progressively as the pyrolysis
temperature increases up to 1250°C, the surface area decreases down to 4m2/g due to the
sintering of the surrounding carbon matrix. Above 1300°C, the gaseous species produced by
the carboreduction between the silica oxide and carbon causes the formation of mesoporosity
which greatly increases the surface area of the powders. A same effect was observed during
for molecular precursor, i.e. after pyrolysis at 1150°C and 1550°C the surface area are
respectively 170 and 722 m2/g. Under these conditions, for low volume fraction of residual
carbon (≈0), SiC grains are bonded together by a thin layer of amorphous carbon and gases
produced during carboreduction can percolate by the interstices between these grains. As the
volume fraction of carbon increases, the gases produced by the carbothermal reaction escape
from the samples by creating narrow channels through the carbon matrix which increase the
surface area of the material. For high value of carbon content, there is not enough gas
produced by the carbothermal reaction and the surface area and the porosity volume decrease.
These processes are shown schematically in Figure 11, which resumes the core of the
mechanisms for porosity control as proposed in this paper. This porosity of excess porous
carbon is not a drawback of the routes proposed here, but can be used to synthesize
nanocomposite carbides with a second network.
The determination of pore size distribution calculated with the BJH model from adsorptiondesorption isotherms of nitrogen indicate that the characteristic diameter of the mesoporosity
is smaller for the sample derived from the pyrolyis of the molecular precursor compared to
that obtained after the pyrolysis of the precursor colloidal (Figure 12). The pore volume of
this material is also higher, which explains the large difference in surface area occupied by
these materials (Table 3). The size of the pore seems to be related to the size of silica
particles; in these conditions the material prepared from the molecular precursors has the
greater characteristics. In contrast the P5 type samples (micrometric silica + sucrose) exhibit
the lowest specific surface area. For these samples, the silica particles are agglomerated in
clusters whose characteristic size is micrometer. These silica particles are much less dispersed
in the sample volume, and therefore the channels created by the escape of the gaseous species
during the carbothermal reaction are larger, leading to a material with lower specific surface.
The evolution of the final surface area developed by the materials is reported on figure 13
according to the initial contact surface between the oxide and sugar. For colloidal type
precursors, the initial contact surface is calculated from the oxides surface area and the
volume fraction of sugar in the precursors. For the molecular type precursor, the initial
contact surface was calculated by dividing the atomic volume of silicon and carbon by the
length of the Si-C bond. This representation shows that the final surface area of the products
obtained is strongly dependent on the initial surface of contact; the molecular route is a good
way to increase this parameter.
Conclusion
Two different synthesis routes have been used for the production of nanocrystalline -SiC
powders. The colloidal route consists in the carboreduction of colloidal silica by sugar
(sucrose, mannitol, sorbitol, cyclodextrin) at high temperature. The molecular route uses a
molecular chemistry approach based on the immobilisation of polycarbosilane in sugar via a
sol gel process which leads to the formation of -SiC after high temperature conversion. In
both case, -SiC powders with a high yield of submicrometric particles were successfully
synthesized for temperature range between 1300 and 1550°C. The powders obtained by the
two routes present a mesoporous structure. These materials have a structure composed of
grains of carbides embedded in a matrix of carbon. Their characteristics depend on the
amount of residual carbon and the mesoporosity is located into the carbon matrix surrounding
the carbide grains. The highest surface area, 722m2/g, was measured for a sample elaborated
from the molecular route. A mechanism related to the release of the CO gas produced during
the carbothermal reduction is proposed to explain the evolution of the surface area versus the
volume fraction of residual carbon. The molecular route seems to be the best way to increase
the surface area of the material, and in future studies we will try to evaluate the effect of
polymerization of the silica network on this parameter.
Acknowledgements
The autors thank the GNR Matinex for its financial support, Pr. Corriu and Pr. Cerveau for
there fruitful discussions and M. Georges for his work on the molecular route. They also
thank the MATINEX French Research Group for its financial support.
1.
(a) Okada, K.; Kato, H.; Nakajima, K., Preparation of silicon-carbide fiber from
actived carbon and gaseous silicon monoxide. Journal of the American Ceramic Society 1994,
77 (6), 1691-1693; (b) Pham-Huu, C.; Bouchy, C.; Dintzer, T.; Ehret, G.; Estournes, C.;
Ledoux, M. J., High surface area silicon carbide doped with zirconium for use as catalyst
support. Preparation, characterization and catalytic application. Applied Catalysis a-General
1999, 180 (1-2), 385-397; (c) Bao, X.; Nangrejo, M. R.; Edirisinghe, M. J., Preparation of
silicon carbide foams using polymeric precursor solutions. Journal of Materials Science 2000,
35 (17), 4365-4372.
2.
Acheson E.G. U.S. Patent 492 Production of artificial crystalline carbonaceous
material. 1893.
3.
Gupta, P.; Wang, W.; Fan, L. S., Synthesis of high-surface-area SiC through a
modified Sol-Gel route: Control of the pore structure. Industrial & Engineering Chemistry
Research 2004, 43 (16), 4732-4739.
4.
Corriu, R. J. P., Ceramics and nanostructures from molecular precursors. Angewandte
Chemie-International Edition 2000, 39 (8), 1376-1398.
5.
Roduner, E., Size matters: why nanomaterials are different. Chemical Society Reviews
2006, 35 (7), 583-592.
6.
Koga, K.; Ikeshoji, T.; Sugawara, K., Size- and temperature-dependent structural
transitions in gold nanoparticles. Physical Review Letters 2004, 92 (11).
7.
(a) Martin, H. P.; Ecke, R.; Muller, E., Synthesis of nanocrystalline silicon carbide
powder by carbothermal reduction. Journal of the European Ceramic Society 1998, 18 (12),
1737-1742; (b) Martin, H. P.; Muller, E.; Knoll, Y.; Strienitz, R.; Schuster, G., Silicon carbide derived from silica sol and sugar. Journal of Materials Science Letters 1995, 14 (9),
620-622.
8.
(a) Kevorkijan, V. M.; Komac, M.; Kolar, D., Low-temperature synthesis of sinterable
SiC powders by carbothermic reduction of colloidal SiO2. Journal of Materials Science 1992,
27 (10), 2705-2712; (b) Krstic, V. D., Production of fine high-purity -silicon-carbide
powders. Journal of the American Ceramic Society 1992, 75 (1), 170-174.
9.
Lin, Y. J.; Tsang, C. P., The effects of starting precursors on the carbothermal
synthesis of SiC powders. Ceramics International 2003, 29 (1), 69-75.
10.
Ishihara, S.; Tanaka, H.; Nishimura, T., Synthesis of silicon carbide powders from
fumed silica powder and phenolic resin. Journal of Materials Research 2006, 21 (5), 11671174.
11.
Corriu, R.; Gerbier, P.; Guerin, C.; Henner, B., The thermal-conversion of
poly(sylene) -diacetylene metal-oxide composites - A new approach to -SiC-MC ceramics.
Angewandte Chemie-International Edition in English 1992, 31 (9), 1195-1197.
12.
(a) Zheng, Y.; Wang, R.; Wei, K. W., Synthesis and characterization of high surface
area silicon carbide by dynamic vacuum carbothermal reduction. Journal of Materials Science
2008, 43 (15), 5331-5335; (b) Jin, G. Q.; Guo, X. Y., Synthesis and characterization of
mesoporous silicon carbide. Microporous and Mesoporous Materials 2003, 60 (1-3), 207212.
13.
Krawiec, P.; Kockrick, E.; Borchardt, L.; Geiger, D.; Corma, A.; Kaskel, S., Ordered
Mesoporous Carbide Derived Carbons: Novel Materials for Catalysis and Adsorption. Journal
of Physical Chemistry C 2009, 113 (18), 7755-7761.
14.
Ledoux, M. J.; Pham-Huu, C., Silicon carbide - a novel catalyst support for
heterogeneous catalysis. Cattech 2001, 5 (4), 226-246.
15.
Tomasik, P.; Palasinski, M.; Wiejak, S., The thermal decomposition of carbohydrates.
1. The decomposition of monosaccharides disaccharides and oligosaccharides. Advances in
Carbohydrate Chemistry and Biochemistry 1989, 47, 203-278.
16.
Cerovic, L.; Milonjic, S. K.; Zec, S. P., A comparison of sol-gel derived siliconcarbide powders from saccharose and actived carbon. Ceramics International 1995, 21 (4),
271-276.
17.
X. Deschanels, M. El. Ghazzal., C. Delchet, D. Herault, V. Magnin, R. Podor, G.
Cerveau, T. Zemb, R. Corriu, Synthesis of carbide compounds derived from colloidal oxide
and carbohydrate. Progr. Colloid Polymer. Sci. 2010, 137, 47-52.
18.
D. Avnir, L. C. K., D. Levy, U. Schubert, A. B. Wojcik In The Chemistry of Organic
Silicon Compounds Z. Rappoport, Y. A., Ed. Wiley: New York, 1998,; Vol. Vol. 2, p 2317.
19.
Corriu, R. J. P.; Gerbier, P.; Guerin, C.; Henner, B. J. L.; Jean, A.; Mutin, P. H.,
Organosilicon polymers - Pyrolysis chemistry of poly (dimethylsilylene)diacetylene.
Organometallics 1992, 11 (7), 2507-2513.
20.
Uchino, T.; Aboshi, A.; Kohara, S.; Ohishi, Y.; Sakashita, M.; Aoki, K., Microscopic
structure of nanometer-sized silica particles. Physical Review B 2004, 69 (15).
Colloidal route
SiO2 grains embedded in
carbohydrate matrix
∑ini.~1-5m2/g
SiO2
Carbohydrate
~1µm
T?
Molecular route
Colloidal SiO2 in
carbohydrate matrix
∑ini~20-100m2/g
~10nm
« Hybrid precursor »
∑ini~500m2/g
O1.5Si
T?
Carbide grain
Carbon and porosity
Highlight : Representation of the various routes used in this study
Carbohydrate
unit
T?
SiO1.5
Figure 1 : Molecular precursor C1.
* -SiC
a Amorphous carbon
+ Cristobalite
+ a
*
*
*
*
a. u.
C/Si=36
C/Si=24
C/Si=18
C/Si=15
C/Si=12
C/Si=9
C/Si=6
C/Si=3,6
C/Si=3
C/Si=2,4
10
20
30
40
50
60
70
80
2(°)
Figure 2 : X-ray diffraction pattern of heat-treated precursor P1, fumed silica + sucrose,
(1550°C-4h) versus the ratio C/Si.
* -SiC
*
SiO2
amorphous
*
a.u.
*
T=1550°C
T=1450°C
T=1250°C
Fumed SiO2
15
25
35
45
55
65
75
2(°)
Figure 3 : X-rays patterns of heat-treated precursor P1, fumed silica + sucrose C/Si=12, versus
the pyrolysis temperature for a dwell time of 4 hours.
120
1600
1400
mf/mi (%)
P1
80
P3
C1
60
T°C
1200
1000
800
T(°C)
100
600
40
400
20
200
0
0
0
100
200
300
400
500
600
Time (min)
Figure 4 : TGA curves (in argon) of precursors P1 (fumed silica + sucrose C/Si=3.6), P3
(fumed silica + mannitol C/Si=3.6), C1 (molecular route, C/Si=8).
Oxygen content (wt.%)
30
25
20
15
10
5
0
700
900
1100
1300
1500
Pyrolysis temperature (°C)
Figure 5 : Oxygen content of samples P1 (fumed silica + sucrose C/Si = 12) versus the
temperature for a dwell time of 4h of the heat-treatment.
+
+ Cristobalite
* -SiC
P3
C1
a. u.
*
*
*
+
0
20
+
*
40
60
80
100
2(°)
Figure 6: X-ray diffraction pattern of heat-treated precursor C1 (molecular route) and P3
(fumed silica + mannitol) at 1550°C-4h.
a)P4-C/Si=4.5
b)P1 – C/Si=3.6
c) C1-C/Si=8
d) P1 - C/Si=24
e) P5-C/Si=24
Figure 7 : SEM micrograph of a)P fumed silica + CD C/Si=4.5, b) P1 fumed silica + sucrose
C/Si=3.6, c)C1 molecular precursor C/Si=8, d)P1 fumed silica + sucrose C/Si=24, e) P5
micrometric silica + sucrose -C/Si=24 after pyrolyses at 1550°C-4h under argon.
SBET (m2/g)
1000
100
P1
P4
P5
C1
M1
10
1
0
10
20
30
40
C/Si
Figure 8 : Specific surface area (BET) of carbides elaborated from various precursors versus
the ratio C/Si (processing conditions 1550°C-4h under Ar).(P1:fumed silica+sucrose, P4:
fumed silica+-cyclodextrin, P5: micrometric silica+sucrose, C1: Molecular precursor, M1:
Xerogel).
2
Specific area (m /g)
1000
100
P5
Sucrose
10
C1
M1
1
0
20
40
60
80
100
%vol Cres.
Figure 9 : Specific surface area versus residual carbon volume fraction after the pyrolysis at
1550°C under Ar of samples P1, P5, C1, M1 and sucrose.
500
P1
P5
400
Vads cm3/g
C1
300
200
100
0
0
0,2
0,4
0,6
0,8
1
p/p0
Figure 10 : :Adsorption-desorption isotherm of P1 fumed silica+ sucrose (C/Si=12), P5
micrometric silica + sucrose (C/Si=12) and C1 molecular precursor.
0% Vol. C
50% Vol. C
80% Vol. C
Figure 11 : Schematic representation of the porosity formation due to the carbothermic
reduction of silica for samples obtained after the pyrolysis of precursors P1.
0,8
C1
P1
P5
dV/dlog(d) (cm3/g.A)
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0
10
20
30
40
50
Pore diameter (nm)
Figure 12 : Pore size distribution (BJH) of P1 fumed silica+ sucrose (C/Si=12), P5
micrometric silica + sucrose (C/Si=9) and C1 molecular precursor.
800
Final area (m2/g)
700
600
500
400
300
0% Vol. C
50% Vol. C
80% V
200
100
0
0
100
200
300
400
500
600
Initial contact area (m2/g)
Figure 13 : Final surface area of the products after pyrolysis at 1550°C under Ar versus initial
contact area of silica and carbohydrate in the precursor.
0,35
CO (Mol/g Cfree)
0,3
Micrometric SiO2
sucrose
0,25
+
Molecular route
0,2
0,15
Colloidal SiO2 +
sucrose
0,1
0,05
0
0
500
1000
1500
2000
2500
Final surface area (m2/g Cfree)
Figure 14 : Gas release versus final surface area of the products.
3000
3500
Method
Solid
reaction
Reactions and processing conditions
state C + silica (1700-1800°C)
Si+C (1350°C)
Si+C
SiO2+carbon black (1450-1550°C)
Surface
area
(m2/g)
6.7
<30
6
9-23
reference
Wei1
Larpkiattaworn 2
Yamada 3
Krstic4
Carbo reduction
SiO2 + sugar (1500-1800°C)
SiO2 + sugar (1500-1800°C)
6-30
115
Martin5
Cerovic 6
Gas solid reaction
C+SiCl4+H2 (1000°C)
30
Moene 7
145
112
1793
410-830
167
450-620
Lednor Ruiter
Jin 8
Li 9
Krawice 10
Zheng 11
Gupta 12
Pyrolysis of
organosilcon
polymer
an [(C6H6)SiO1.5]n (N2-1000°C)
TEOS + phenolic resin
Polymer+silica hybrid (1000°C-Ar/H2)
CVI dimethyl dichlorosilane+ SBA15
TEOS+sacc 1100°C-vaccuum
Phenyltrimethythosilane
Table 1 : Effect of the influence of the processing parameters onto the surface area of SiC.
Carbon source
Precursor composition
Name
C/Si
C1
14
Molecular route
Sucrose
Sucrose
8
Colloidal route
Sucrose
Sorbitol
Mannitol
Cyclodextrin
Sucrose
Fumed silica + sucrose
Fumed silica + sorbitol
Fumed silica + mannitol
Fumed silica + cyclodextrin
Micrometric silica + sucrose
Table 2 : Precursors used for silicon carbide synthesis.
P1
P2
P3
P4
P5
2 – 36
3,6
3,6
4,5
2–4
Sample
C/Si
BET
XRD
C wt(%)
P1 – 800°C
4
256
SiO2a
49.6
26.3
P1 – 1000°C
4
207
SiO2a +C a
51.3
24.8
P1 – 1250°C
4
11.3
SiO2a +C a
48.8
21.9
P1 – 1450°C
4
260
SiC
66.9
1.1
P1 – 1550°C
4
190
SiC
51.2
P1 – 1550°C
2.4
61
SiC+cristo
21,2
16.6
P1 – 1550°C
3
18
SiC
28,3
2.1
P1 – 1550°C
3.6
16
SiC
28,8
1.8
P1 – 1550°C
4,5
11
SiC
28.5
1.9
P1 – 1550°C
5.4
7
SiC
27.8
P1 – 1550°C
6
16
SiC
28,2
P1 – 1550°C
7
168
SiC
51.8
P1 – 1550°C
9
195
SiC
58,2
1.2
50,7
P1 – 1550°C
9
194
SiC
50.2
1.5
38.3
P1 – 1550°C
12
194
SiC + C
54.9
1.1
45.7
P1 – 1550°C
12
190
SiC
51,2
P1 – 1550°C
15
173
SiC+C
73,6
1.1
71,6
P1 – 1550°C
18
136
SiC+C
78.1
0.7
77.1
P1 – 1550°C
24
69
SiC+C
80,1
0.4
79,4
P1 – 1550°C
36
140
SiC+C+SiO2
0.4
P2 - 1550°C
3.6
P3 - 1550°C
3.6
nd
SiC+SiO2
6.1
P4 - 1550°C
4.5
15
SiC
27.8
O wt(%) C vol. (%)
Porosity
(cm3/g)
3,6
1
1.3
0.34
39.9
3.1
5
10
0.018
Sample
C/Si
BET
XRD
C wt(%)
O wt(%)
P5 – 1550°C
4.5
15
SiC
28.8
3.1
P5 – 1550°C
9
53
SiC+cristo
51.6
5
P5 – 1550°C
12
29
SiC+C
58
1
C amorphe
94.5
0.67
2.2
Sucrose - 1550°C
C1 - 1550°C
14
722
SiC + C
45.5
M1 – 1550°C
8
567
SiC + C
45.2
Table 3 : Characteristics of the samples
C vol.
(%)
Porosity
(cm3/g)
0.049
30.6
30.5
0.55