close

Вход

Забыли?

вход по аккаунту

?

Chemical Vapor Deposition of Silicon Carbide and Silicon NitrideЧChemistry's Contribution to Modern Silicon Ceramics.

код для вставкиСкачать
[I251 J . L. Locarelli, G. Ries, Angw. Makromol. Chem. 35, 57 (1974).
[I261 J . L. Lorarelli, G. Ries, J. Polym. Sci. Polym. Lett. Ed. 11, 257 (1973).
11271 D. J. Srein. G. fahrbach. H . J . Adlrr. Angew. Makromol. Chem. 38.
67 (1 974).
[l28] M. Mursuo. S. Saguye in G. E. Molau: Colloidal and Morphological
Behaviour of Block and Graft Copolymers. Polym. Colloids Proc.
Symp. 1970. Plenum Press, New York 1971, p. I.
[129] G. KBmpf. M. Hoflmunn. H. Krcjmer, Ber. Bunsengs. 74. 851, 859
( I 970).
[I301 7: Hushimoro, K . Nagaroshi, A. Todo, H . Hasegnwn. H . Kawei, Macromolecules 7, 364 (1974).
11313 B. Gallar. Pure Appl. Chem. 38. 1 (1974).
11321 8. B. Bradjord in [l28], p. 21.
[I331 B. Prus:ynski. J . 72risse. A. Skoulios. Makromol. Chem. 176, 3483
(1975).
11341 D. J. Meier in J. J. Burke, K W e b : Block and Graft Copolymers.
Syracuse University Press, New York 1973. pp. 105ff.
[I351 D.J. Meier, Polym. Prepr. Am. Chem. SOC.Div. Polym. Chem. f5.171
(1974).
[I361 D. J. Meier. Appl. Polym. Symp. 24.67 (1974).
[I373 D. J . Meier. Vortragstagung der Fachgruppe Makromolekulare Chemie der GDCh, Bad Nauheim 1975.
[I383 E. Hel/aud in [S].
I1391 E. Heljund. Am. Chem. Rcs. 8,295 (1975).
[140] S. L. Aggarwal, Polymer 17. 938 (1976).
[I411 R. W Seymour, J . R. Oaerron, L. S. Corley, Macromolecules 8. 331
I1975).
[I421 M. Kerker. J. Opt. SOC.Am. 65,376 (1975).
1143) H . Breuer in G. H. Frisrhot: The Physies of Non-Crystalline Solids.
4th International Conference. Clausthal-Zellerfeld 1977. Trans. Techn.
Publ., Rockport. p. 88.
[I443 C. B. Burknull, D. Clayron, Nature Phys. Sci. 231. 107 (1971).
[I451 J. P. Berry. J. Polym. Sci. 50, 107.315 (1961).
[I461
11473
[I483
[I491
[I501
[l5l]
[lS2]
[153]
[I541
[l55]
[I561
[157]
[I581
I1591
[160]
11611
[I621
[163]
[I643
[I651
[I661
[I671
[168]
[I691
[I703
J. P. Berry. J. Polym. Sci. A I, 993 (1963).
J. P. Berry. J. Polym. Sci. A3, 2027 (1965).
R P. Kambour. J. Polym. Sci. A2.4159.4165 0964).
R. f . Kambour, J. Polym. Sci. A-2, 4. 17, 349 (1966).
R. f . Kambour, IUPAC-Symposium Leiden 1970. Preprints 11. 977.
R. P. Kambour, J. Polym. Sci. Macromol. Rev. 7, 1 (1973).
J. J. Benhow, Proc. Phys. SOC. London 78,970 (1961).
L. C. Cessna. S. S. Srernrrein, Polym. Prepr. Am. Chem. SOC. Div.
Polym. Chem. 8. I516 (1967).
S. S. Srernsrein, L. Ongchin, A. Siluermun, J. Appl. Polym. Symp. 7, 175
(1968).
S. S. Srentsrein, L. Ongchin. Polym. Prepr. Am. Chem. SOC.Div. Polym.
Chem. 10, 11 17 (1969).
M. Marsuo, 7: 7: Wong. 7: K . Kwei, J. Polym. Sci. A-2. 10, 1085
(I 972).
J. N. Goodier, J. Appl. Mech. 55, A 39 (1933).
G. Rehage, G. Goldbach, Angew. Makromol. Chem. I . 125 (1967).
F. Huaj, H . Breuer. J. stab en on^. Angew. Makromol. Chem. 58/59,
95 (1977).
H. Breuer, F. Huaj, J . Scabenow. J. Macromol. Sci. Phys. B14, 387
(1977).
H. Breuer. Vortrag auf "Conference Toughening of Plastics': The
Plastics and Rubber Institute, London, July 1978.
L. Bohn. Angew. Makromol. Chem. 20, 129 (1971).
J . A. Schmirr, 1UPAC-Symposium Toronto 1968. Preprints II. A9.
M. Marsuo, A. Ueda, X Kondo, Polym. Eng. Sci. 10,253 (1970).
S. S. Srernrrein, IUPAC-Symposium Leidcn 1970. Preprints II. 973.
D. Kranz, L. Morbirrer, K. H . Orr, R. Casper. Angew. Makromol.
Chem. 58/59.213 (1977).
R. P. Kamhour. R. R. Russel. Polymer I?. 237 (1971).
G. Kiimpj, Angew. Makromol. Chcm. fi0ffil. 297 (1977).
H . Blimel. Kautsch. Gummi, Kunrisi. 24. 517 (1971).
H. A. J . Burreaerd, J. Polym. Sci. Polym. Symp. 49. 149 (1975).
Chemical Vapor Deposition of Silicon Carbide and Silicon NitrideChemistry's Contribution to Modern Silicon Ceramics
By Erich Fitzer and Dieter Hegenf.1
Dedicated to Professor Horst Pommer on the occasion of his 60th birthday
Pure silicon carbide and silicon nitride have valuable properties in bulk pore-free form;
however, their industrial exploitation has hardly been possible so far. Neither compound
can be melted or sintered in pure form; hot pressing or sintering at normal pressure requires
the presence of additives; and the reaction-sintering process in which only Si and C or Si
and N are employed as additives affords porous materials.-The novel process of chemical
vapor deposition has partly overcome the drawbacks of the previous methods. In the new
process S i c is produced, e.9.. by pyrolysis of CH3SiC13, and SioN4 by reaction of Sic14
with NH3. This technique can also be used for pore filling in objects made of S i c and
SkN4 (gas phase impregnation) and for producing extremely fine Sic and Si3N4 powder
and S i c monofilaments suitable as components for S i c composites. Moreover, gas phase
impregnation can also give fiber composites.
1. Introduction
The tempestuous advances in the field of physical metallurgy, solid state physics, and fracture mechanics have rele-
1'3
Prof. Dr. E. Fitzer. Dr. D. Ktgen
lnstitut Nr Chemische Technik der Univenitiit
Kaisentr. 12, D-7500 Karlsruhe I (Germany)
gated the chemical basis of materia ...developments to a background role in modem materials science. It has become apparent, however, that for stringent technical demands greater
attention should be focused on chemical aspects in the solution
of problems. In the case of the silicon ceramics dealt with
in this paper (Sic, Si3N4,silicon nitride solid solutions), which
are considered for use in the manufacture of components
for gas turbines and other topical high-temperature applica-
Anycu'. Chrm. Inr. Ed. Engl. 18.295-304 r 1979)
0 k l a g Cltenrie. GmbH, 6940 Weinheim. 1979
295
05 70-0833/79~0404-0295S 02.50/0
tions"], chemical vapor deposition could provide new impetus
because these materials cannot be produced by fusion methods
(see Fig. 1l 2 -41); moreover, standard techniques of powder
Si + Sic
Si
Si C
C
2. Scope and Limitations of Silicon Ceramics
Silicon ceramics do not occur in nature; they are man-made
inorganic materials, extending from metallic conductors, ~iz.
silicides of transition metals of Group 4 and 5, cia the stoichiometric covalent compounds S i c and Si3N4 having semiconductor properties, to the silicon nitride solid solutions (sial o n ~ ) I ~in. ~which
l
the nitrogen and silicon have been partly
substituted by oxygen and aluminum, respectively (see Fig.
1). The structures of the metallic silicides, of which molybdenum disilicide (MoSi2) is the most significant technologically['I, show some resemblance to the interstitial structures
of carbides and borides. However, the greater size of the
silicon atom leads to lattice widening while still permitting
direct Si-Si bonding which will modify the bonding character
of the metallic host.
Silicon carbide is strictly stoichiometric and crystallizes,
like elemental silicon, in a diamond type structure showing
some range of variation of the stacking sequence. The strong
covalent Si-C bond leads to a refractory nature, extreme
hardness, and a very high theoretical Young's modulus which
is indeed realized in industrial S i c monofilaments (up to
450000 MN/m2). However, dislocation movement is blocked
in these compounds, causing high temperature strength but
also brittleness. This undesirable brittleness is reduced in silicon nitride and sialons, but at the expense of elasticity and
high temperature strength (Table 1).
Table 1. Properties of S i c and Si,N4 (s=coefficient of thermal expansion)
and strength (0)of technical products made of these materials. The lower
strengths are for reaction-bonded materials and the higher values for hotpressed materials.
SIC
SijNl
Bonding
strength
[kJ/mol]
3! x
lo-'
[I<-']
TdtC
(TRT
~ ~ z o O - L
rC]
[MN/m2]
[MN/m*]
437
773
58
35
2700
2200
300-800
150 -700
300- 500
150 -600
The outstanding property of all of these refractory silicon
compounds is their chemical inertness in an oxidizing environ-
150
B
Fig. 1. Phase diagrams of the binary systems siliconicarbon [Z] and siliconpitrogen [3]; equilibrium diagram for the ternary system silicon/oxygen/nitrogen [4]. s=solid, l=liquid, v=vapor. A, B, C=equilibrium points of
the phases at lS00K.
metallurgy or ceramics lead not only to production problems
but also to serious restrictions on quality.
296
100
20
10
2000 2500
0.3
p [g/cm3 1
+
t [hl-
Fig. 2. Flexural strength oRI of a) reaction-bonded commercial
and
b) reaction-bonded commercial Si3N4 [ 9 ] as functlon of porosity (P) and
density ( p ) ; weight increase (m/rno) with time on oxidation of Sic at 1500°C
(the numbers next to the curves refer to the density p [gicm'] 181); d)
change in resistance Q/Qo of SIC on cycling between 20°C and 1500°C
(I without glass layer; I 1 with glass layer).
Aiigrw. Chem. Int. E d . Engl. 18.295-304( 1979)
ment which is due to the high affinity of silicon for oxygen.
In an aqueous environment, passivity results from reaction
between silicon and oxygen. O n thermal oxidation, for instance
in air, glassy SiOz surface layers are formed which strongly
suppress further oxidation. Oxidation was found to show
a parabolic time dependence (Fig. 2)[8,91.Mass transport in
such SiOz layers is effected solely by non-ionized oxygen,
and is additionally hindered by counterdiffusion of volatile
by-products of oxidation, such as CO and N2 respectively.
Comparison of the temperature dependence of the overall
oxidation rate of silicon, silicon nitride, silicon carbide, and
MoSi2 as well, reveals similar values for the activation energy,
Fig. 3. Temperature dependence of oxidation rate constant (Ox.) of various
a) SiJN4 with 30% La2O3, air; b) Si3N4 with
silicon ceramics [10-12].
5 % MgO, air; c) sialon with 5-30 % Al (air); d) sialon with 50 % Al
(air); el MoSi, (air); 0 S i c powder (02);
g) S i c powder (air); h) Si3N4
powder (air).
and shows that the mechanism of the rate-controlling step
is always the same (Fig. 3)14*51.
The activation energies are
also comparable with that of oxygen permeation through
S i 0 2 glass films.
The blocking efficiency of the S i 0 2 surface layers is greatly
reduced by small quantities of foreign oxides. For instance,
the oxidation rate of MoSiz is one order of magnitude higher
than that of pure silicon carbide, because this compound
forms a SiOz glass layer with small amounts of Mo02["],
and does not form counterdiffusing by-products (Fig. 3)"'- I 'I.
Nevertheless, the air oxidation of MoSiz at 1500°C is one
order of magnitude slower than that of 20/80 Cr/Ni alloys
at only 1000"C''31.One can envisage the oxidation resistance
of pure silicon carbide as being ten times better at 1400°C
than that of today's best "superalloys" at their maximum
temperature of application, which is still 1200°C. In such
considerations, the superior resistance of silicon ceramics to
"hot corrosion" resulting from attack by combustion products
has not been taken into account['4!
The problem remains of how to utilize this chemical stability
by provision of bulk materials without open porosity. In
silicon ceramics having open pores, intercrystalline oxidation
will occur, forming S i 0 2 glass lamellae between the grains.
This causes a deterioration of the material on cooling from
the oxidation temperature because of the extremely low thermal expansion coefficient of SiOz glass (a=0.5 x
K-')
as compared with that of silicon carbide ( a = 5 x
K-').
Not only the mechanical properties, mainly impact strength
and creep resistance, but also the thermal and electrical conductivity will be considerably impaired. The last mentioned
effect is well known to the laboratory chemist from the increase
in resistance of commercially available SIC heating elements
in intermittently operated high-temperature combustion
furnaces.
How can we produce pore-free SIC and Si3N4 bulk materials? Pure SIC and Si3N4 powders do not sinter in the conventional sense. In contrast, the silicides of the transition metals
exhibit similarly high values of surface energy as are known
for metals, and such silicides are therefore readily sinterable.
Thus wire-wound or tubular MoSiz heating elements for working temperatures up to 1600°C can be produced by powder
. <
Table 2. Comparison of manufacturing processes for silicon ceramics (RB = reaction bonded, H P = hot pressed, CP = cold pressed).
Process
Reaction sintering
Sic
Reaction sintering
Si3N4
Hot pressing
SIC, Si3N4,Sialon
Pressureless sintering
Sic
Starting
materials
S i c powder
Si3N4 powder
Si ceramic powder
Si ceramic powder
pitch binder
Si powder
A ~ z O SMgO,
,
YzOa
B, B4C, C
Process steps
Molding or extrusion at RT
Carbonization at T= 1000°C
Siliciding at T=2000"C
Molding at RT
Nitriding at T= 1500°C
Molding at T> 1S00"C
Molding at RT
Sintering at T= 2000°C
Product
RB-SiC
Rb-Si3N4
HP-Sic, HP-Si3N4
CP-Sic
Properties
High porosity
Low strength
High porosity
Low strength
Low porosity but
heterogeneous material
High RT strength
Low porosity but
heterogeneous material
High HT strength
Remarks
Process well controlled
Also suitable for complicated
shaped sections
Inexpensive raw materials
Process controlled
Reaction usually incomplete
Suitable only for thinwalled sections
Inexpensive raw materials
Process difficult to control
Suitable only for simple
sections
Expensive raw materials
Process not yet controlled
Suitably only for simple
sections
Expensive raw materials
+
Angew. Chem. lnt. Ed. Engl. 18.295-30411979)
+
+
+
291
metallurgical techniques” ’1. In the case of S i c and Si3N4
powders, even the finest grain sizes (< 0.2 pm diam.) cannot
be sintered in the absence of high melting, nonvolatile additives
such as B4C, MgO, A1203, or Y203116-181.Mechanical pressure is usually also applied during sintering (hot pressing
at ca. 1600”C), imposing severe restrictions on the shapes
and sizes of the products.
The most serious disadvantage introduced by such additives
is their influence on oxygen transport within the SiOz surface
layers (see Fig. 3), and furthermore on the creep behavior
of the polygranular bulk material. Intercrystalline oxides will
act as plastifiers at high temperatures. The sialons, which
partly oxidize to A1203, exhibit a similar deterioration of
their high temperature properties compared with pure silicon
carbide and silicon nitride“ 1‘.
Silicon ceramics free from additives can be prepared by
the reaction bonding (reaction sintering) process. Table 2 compares the process steps with those of other metallurgical processes. Reaction sintering starts with the pure carbide or nitride
powder. As in ceramics the material is shaped prior to sintering
with the aid of a temporary binder, which consists of one
of the components of the silicon material (carbon in the case
of Sic and silicon in case of Si3N4).This binder is converted
into the final silicon compound, which should act as bridge
or matrix connecting the primary powdered silicon compounds, at high temperatures (7,
1.50O0C)by chemical reaction (siliciding with molten or vaporized silicon in the case
of Sic; nitriding with volatile nitrogen compounds (Nz or
NH3) in the case of Si3N4). This transformation amounts
to an impregnation process; the compact must have open
pores and a porous final structure is obtained. The pores
can subsequently be filled by impregnation with an excess
of molten silicon; however, a heterogeneous structure with
impaired high temperature properties will result. Furthermore,
grain growth is promoted at high temperatures by dissolutionreprecipitation cycles.
The novel process of chemical vapor deposition could at
least partly eliminate or compensate the drawbacks of the
reaction bonding technique:
1) Subsequent “gas impregnation” treatment densifies
porous Sic and Si3N4 bulk materials and additionally seals
the pore entrances on the surface[’g1.
2) Direct preparation of finest Sic and Si3N4 powder
obviates the need for grinding processes for size reduction.
Such grinding procedures always lead to surface passivation
by oxygen adsorption and thus also inhibit sintering processes.
3) Sic monofilaments exhibiting extremely high Young’s
modulus and high tensile strength become available for reinforcement purposes.
4) Gas phase impregnation permits fabrication of pore-free,
shaped objects, mainly fiber-reinforced composites.
Successfulapplication of chemical vapor deposition requires
an accurate knowledge of the kinetics of deposition and of
the influence of the various process parameters.
3. Principles of Chemical Vapor Deposition of Sic
and Si3N4
Chemical vapor deposition (CVD) can be considered as
“precipitation” of a solid reaction product from a supersatur298
ated fluid matrix. In contrast to the precipitation processes
occurring from solid or liquid solutions, vapor deposition
does not occur during cooling but during heating. The thermally activated process steps are mainly pyrolysis reactions
having very complicated mechanisms. As in other heterogeneous chemical gas-solids reactions, transport steps can
play a very important role during gas phase
With regard to the energy supply, we distinguish between
two extreme alternatives: the “hot wall arrangement” in which
bl
Substrate
isothermal
Temperature
gradient
Gas
I
Diffusion
vl
I
Crystallization
v2
I
Chemical reaction
“3
Fig. 4. a) Hot-wall and b) cold wall arrangements for chemical vapor deposition: 1, resistance heating elements; 2, induction coil. c) Rate-determining
steps of chemical vapor deposition.
heating of the substrate, where deposition should take place,
is performed by radiation from a surrounding heater, and
the so-called “cold wall arrangement” with a heat source
inside the substrate (Fig. 4a, b).
The mechanism of vapor deposition even on smooth planar
surfaces is very complicated. Figure 4 shows a simplified Arrhenius plot from which it is seen that, especially at high
temperatures (left-hand side of the diagram), the overall
deposition rate will be controlled by mass transport in the
outer gaseous surface layer (slowest consecutive step).
These limitations become more critical in case of chemical
vapor deposition on inner surfaces because of the long diffusion
Angew. Chem. f n t .
E d . Engl. 18,295-304(1979)
I
paths. Only by means of complicated experimental set-ups
(temperature or pressure gradient)["] is it possible to support
diffusion by convective transport (cf. Fig. 4c). In order to
achieve good impregnation, the overall deposition rate within
the pores must never be controlled by the transport rate
but only by the chemical reaction rate. Consequently, depositionmustbeperformedatlow temperatures, and low concentrations should be used. Furthermore, gas phase nucleation must
be avoided. If finely powdered products are desired, however,
high supersaturation is necessary to ensure spontaneous gas
phase nucleation and very fast cooling to avoid crystal growth.
The driving force of chemical vapor deposition is the thermodynamic instability of the gaseous compounds used as precur-
on use of CH3SiCI3, i.e. the compound with a 1 : 1 ratio
of carbon and silicon[' 6 s 261.
-1
I
r C"C1
-
Fig. 6. Yield of SIC (-) and free carbon (---) during pyrolysis of various
and (CH,),SiCI ( 0 )
chloro(methy1)silanes CH,SiCI, ( o ) , (CH3)2SiCII (O),
~41.
Fig. 5 . G i b h Cree energy of formation of various compounds arising during
vapor phase deposition of S i c and Si3N4 [ 2 2 ] ,
sors. Figure 5 shows data on the free energy of formation
of some compounds involved in the deposition of S i c and
Si3N4[221.Chloro(methy1)silanes are mostly used for S i c
deposition because easy control of the Si/C ratio in the feed
can thus be achieved. However, cyclic silicon-carbon compounds ("carbosilanes") may be formed as intermediate~['~]
and deposit as tarry by-products in the cooler areas of the
apparatus; on further heating they can initiate gas phase
nucleation.
CH,SiC13
(CH,),SiCl,
(CH,),SiCl
--
1' 1 -
+ Sic
\ /
,si'I ,\'sicI
:c,,,,c:
"
+
(Si)
S i c + (C)
--+ S i c
+
C
+ HC1
The question whether a stoichiometric deposit or a deposit
with an excess of silicon or carbon is obtained depends on
the type of precursor. As S i c has practically no solubility
for free silicon or carbon (Fig. I), an excess of silicon or
carbon can impart a heterogeneous structure to the deposit
(see Fig. 7)["]. However, the conditions during deposition
will also influence the chemical composition of the deposit.
Figure 6[241shows the yields of S i c and excess carbon
deposited on a cold wall on decomposition of chloromethylsilanes. As can be seen, deposition of free carbon does not
occur below 1000°C whereas free silicon is often found even
Angrw. Chrm. Itit. E d . Enql. 18.295-304(1979)
Fig. 7. Alternating deposition of S i c and carbon caused by oscillating reaction
[ 2 5 ] . The layers are about 0.5-1 pm thick.
In general, SiC14and NH3 are used as precursor compounds
for the deposition of Si3N4 because they are inexpensive and
easy to handle. As shown in Figure 5, however, NH4Cl may
be formed from the feed and the by-product HCI if backmixing
of the reaction gases in not avoided.
Low deposition temperatures, which are necessary for reaction rate control of the overall kinetics, have the additional
advantage of ensuring fine-grained and even amorphous structure of the deposits[20].Scanning micrographs (Fig. 8) show
the influence of deposition temperature on the morphology
of chemical vapor-deposited S i c and Si3N4 for some typical
examples['*].
The key to successful chemical vapor impregnation lies
in a thorough comprehension of the deposition kinetics under
various conditions. Figure 9a shows overall deposition rates
of silicon carbide from trichloro(methy1)silane and of silicon
nitride from tri- or tetra-chlorosilane and ammonia[28,291 as
a function of deposition temperature and gas concentration.
These data were obtained with a hot wall arrangement. This
is also the reason why we have found a maximum in the
case of silicon nitride deposition; the fall in the deposition
rate dith further temperature increase is caused by gas phase
nucleation which consumes the precursor before the gas flow
299
Fig. 8. Morphology of SIC and SiaNl layers (secondary electron image) [28].
reaches the surface of the substrate. The flattening of the
rate of growth of S i c layers at high temperatures indicates
mass transport control of the overall deposition rate, even
on plane surfaces. This transition from reaction control to
diffusion control is better recognized in the Arrhenius plot
from the break in the curve. For cold wall silicon carbide
deposition on plane substrate surfaces, this diffusion control
starts at deposition temperatures above 1050°C.The situation
in a fluidized bed, i. e. a hot wall arrangement, is also shown
for sake of comparison; diffusion control is then observed
only above 1250°C.
185510
The specific reaction rate for Si3N4 deposition is much
higher than that for Sic. This is why very low concentrations
are necessary in order to avoid gas phase nucleation.
4. Gas-Phase Impregnation of Porous SIC and Si3N4
Objects
If information is avallable about mass transport and reaction
rates at various temperatures, the maximum impregnation
depth can be predicted. In reaction kinetics, dimensionless
numbers like thesecond Damkohlernumber Dalland the Thiele
number are used for calculations on diffusion in porous
catalysts. Figure 10["~ shows the ratio 11 of the deposition
rate in the pores to the deposition rate on the surface. In
the interest of maximum possible impregnation. q should
+
5
1.0
0.8
lot
0.6
p.
I
1
1100
-
1300
T [TI
1500
209\
02
1
05
06
07
l i T [K-'I
08
1
09
Fig. 9. Growth rates of S i c and S I N , layers deposited from the gas phase
[28, 291. a) The nunibers next lo the curves refer to the concentralions
or the precursor compounds in the gas phase (in rnol/l); h) the numbers
next to the curves refer to the activation energiesfor decomposition (in kJ/mol).
300
0.1
Fig. 10. Ratio '1 of the rate of deposition in the pores and on the surfnce
as a function of the second Damkohler group Da,, [30]. Dall= k I'iD; k =rate
constant, D =diffusion coeikienr, I =depth of pores to be impregnated.
.4ngew. C h e i . l i i t . Ed. Engl. 18.295-304 f 1 9 7 9 )
lie near the maximum value. 1 stands for the depth of the
pores, which are assumed to be cylindrical.
We have precalculated such theoretical impregnation depths
using the deposition rate as a measure of the reaction rate
in the reaction rate-control regime (Fig. 9). The diffusion
coefficients were calculated with data shown in Table 3.
Table 3 . Diffuslon coefficients i n hydrogen of some gaseous species formed
on chemical vapor deposition of SIC and Si3N4.
900°C
D [cmZ6] a t T=
Ioo0"C
1100°C
0.50
049
0 63
0 95
0 93
I.32
0.51
0.50
0.64
0.96
0.94
I33
Cpd
CH3SiC'I3
(CH3)*SiCI2
SICI,
NH3
CHI
HCl
1200°C
0.54
0.52
0.65
0.97
0.96
I .34
0.56
0.54
0.67
0.98
0.97
1.36
These predicted values were compared with experimentally
observed impregnation depths in Figures 12b and 12c. The
results confirm the applicability of such kinetic data for a
precalculation of impregnation depths. The deviation of the
experimental curves in the direction of lower impregnation
depths is caused by a rapid narrowing of the pore entrances
at the higher temperatures. However. there is an ever-present
danger of merely converting open porosity into closed porosity
by preferred deposition at the pore entrances. Thus, pore
filling of originally open porosity can only be achieved in
part.
We used five different types of reaction-bonded S i c and
Si3N4 in our experiments on densification by S i c chemical
vapor impregnation (CVI) (Fig. 13)[32.33!
They differed considerably in grain size and pore characteristics.
____
,
Attempts weremade to test the results by model experiments
on graphite samples with drilled holes as large model pores
(Fig. 1
and with polygranular graphite having a defined
pore spectrum["] (Fig. 12a) as substrate. At 1500°C closing
of the pore entrances is observed, even at very large diameters,
owing to preferential surface deposition. In practical work.
successful impregnation of substrates with smaller pore
entrances requires much lower impregnation temperatures,
i. e. about 800--900"C.
c
/
30.
25.
-2t
2015-
L
10-
5-
Fig. I I . Deposition of Sic in cylindrical model pores in graphite [28]. From
left to right: d=400, 600, 800, 6 0 0 p .
1000
800
600
400
1200
r
LOCI
-
UOO
16w
Fig 12. a ) Poresize distributron of porous graphite (V=pore volume, r = p o r e
radius); b) calculated (-1 and experimental (---) depth of impregnation I
in porous graphite (see Fig. 12a). and c) in model pores. The pore radius
is given next to the curves. Impregnation was with CH3SiC13.
Angel\. Chum 111r.Ed. Engi. 18. 295-304 ( 1 9 7 9 )
lo'
r [nml-
m'
800
900
T
[TI
-
a00
Fig. 14. Results of gas-phase impregnation of SIC with CH3SiC13at 800°C.
a1 Pore size distrlbution of sample no. 2 (see Fig. 13) after impregnation
for various periods; b) maximum degree of pore filling accomplished by
impregnation (samples 1-5, see Fig. 13).
Table 4 shows that the strength of all samples, except no.
5 , can be increased by densification. However, even in this
last case, the surface properties are improved to the same
301
extent as with the other samples. The resistance to oxidation
became the same as that of smooth S i c single crystal surfaces
in all cases.
Table 4. Results of gas phase impregnation of five reaction-bonded S i c
materials of various porosities. Samples 1-5, see Fig. 13. a is before, b
after, impregnation. The values in the columns headed "Si" and "C"refer to
wt.- Y, of free Si and C, respectively.
C
Sample Si
3.0
I
2
3
4
-
5
-
07
2.7
0.8
0.3
6.5
0.5
Bulk
density
[g/cm3]
Flexural
strength
[MN/m*]
Oxidation
constant
[cm/h'/2]
a
h
a
b
a
b
2.10
222
2.47
2.44
2.33
2.45
2.47
2.67
2.64
2.34
80
87
113
143
100
4.9x10-'
44x10-'
2x10-'
3xlO-I
9x10-l
~
8
~
~
~
47
40
90
82
100
x
~
X
X
I
l
I
I
x I
O
O
O
O
O
- ~
- ~
- ~
- ~
- ~
It was interesting to test how densification affects the high
temperature strength. Figure 15 shows the improvement of
the short time flexural strength and its variation with
increasing testing temperature. It is not influenced up to
1400"C, but drops sharply with further increase of the testing
temperature. Most striking is the effect of chemical vapor
impregnation of RB-Sic on creep resistance (see Fig. 16).
f-f
.o-
E
Q+\
0
I
0
:
Fig. 15. High temperature flexural strength (short-time HT strength) of reaction-bonded SIC a ) before and b) after impregnation with CH3SiCI3(samples
1-5; see Fig. 13).
30
64
90
120
t CminlFig. 16. Creep behavior ofreaction-bonded SIC at 1350°C: a, before impregnatlon (load 30 MN/m*); b and c, after impregnation (load 50 and 30 MN/m*,
respectively). Sample no. 1, see Fig. 13.
The CVD of Si3N4 in general and the densification of
porous RB-Si3N4 samples by CVI in particular are much
more difficult. Not only does the reaction gas consist of two
components but the reaction rate is also much higher. As
302
5. Preparation of Sic and Si3N4Powders
bl
al
N
I
the overall kinetics for CVI must be controlled by the reaction
step, practical execution of in-pore deposition becomes very
difficult. Furthermore, in the case of silicon nitride there is
a high tendency for gas phase nucleation. The conditions
for the gas-phase impregnation therefore have to be shifted
not only to low temperatures (about 900"C-1000"C), but
also to low gas concentrations. However, the latter condition
increases the risk of diffusion control.
We selected a cold wall arrangement mainly in order to
avoid gas phase reactions. We further used indirect heating
with the aim of achieving a thermal gradient (see Fig. 4b)[331.
Finally, both reactive gases were introduced through separate
tubes, and the flow of the gases was directed to the surface
site from where impregnation should start['*]. In spite of all
these preventive measures, we were unable to avoid preferential
densificationof thesurfacelayersdue to the very high deposition
rate even at 1OOO"C.The maximum degree of impregnation was
20 % on complete transformation of closed pores into sealed
There is no increase in strength on Si3N4 gas-phase
impregnation but the surface properties are decisively
Bocker and HausnerC35l recently described the preparation
of sinter-active S i c powders with grain sizes in the submicron
range by pyrolysis ofCH3SiH3in argon carrier gas at temperatures between 1000 and 1800°C but did not report the results
of sinter experiments.
Previously described processes for the production of S i c
or silicon monoxpowder by reaction of silicon
ideC3'l with hydrocarbons do not give well-defined products.
The transformation of amorphous into crystalline S i c by
subsequent heat treatment (18 ~ - 2 0 0 0 ° C ) of the product
obtained from pyrolysis of methylsilanes or chlorosilanes at
600-1100°C is described by Enk and N i ~ k l [ ~ * ] .
Apart from the preparation of S i c powder by chemical
vapor deposition, the plasma process can be used for the
same purpose. N e u e n s c k w ~ n d e r [described
~~~
the reaction
between SiCI4 and hydrocarbon compounds in a direct current
torch. K ~ k n ' ~ used
' ] reactive anodes consisting of SiOz and
carbon. F y ~ n s ' ~ reported
'~
the reaction of solid SiOz with
hydrocarbons. Only S ~ / i n g e r [used
~ ~ ] chloro(methy1)silanes
as reactive compounds for cracking in a high frequency plasma.
The S i c powder used by P r o ~ k a z k ain[ ~his
~ ~first sintering
experiments without applying mechanical pressure was apparently a powder prepared by the plasma process. The excess
of carbon needed in these sintering experiments was probably
necessary to reduce the oxides introduced by the plasma process.
Several methods of producing Si3N4 powder have already
been described. M ~ z d i y a s n i [ ~reported
~l
the pyrolysis of
Si(NH)2.
studied the reaction of SiCI4 with NH3.
Inclusions of HCI were found to be unavoidable. Prochazka
prepared Si3N4 powder from SiH4 and NH3 in
et
a tubular reactor at temperatures above 1450°C. Kijima["]
introduced hydrogen into the same reaction gas. The resulting
powder is finer than 0.1 pm.
Angew. Cliern. Int. E d . Engl. 18.295-304( 1 9 7 Y )
Most interesting with respect to the sinteling behavior of
CVD Si3N4 powder is the work of Thummler et a/.r481.
The
idea was to deposit P-Si3N4[281andto utilize the P % mtransformation occurring during sintering heat treatment to initiate
and assist the sintering process. It is hoped to achieve products
of high density without using high melting additives.
epitactically and thus increases the diameter of the core fiber
(Fig. 19, left). Alternatively, a carbon fiber yarn may be
impregnated from the gas phase with S i c (Fig. 19, right).
6. SIC Fibers and Fiber-ReinforcedComposites
The first experiments on the preparation of SIC monofilaments were performed with W fibers as substrate using a
technique already described for B
50* 511. Diefend o ~ g f ~reported
~]
tensile strengths up to 3500 MN/m2 for
fibers deposited from CH3SiCI3.
Our own studies were performed with carbon monofilaments as substrate^'^'.^^. 541. The strength and flexibility of
Fig. 19. Fracture of unidirectionally reinforced SiCjSiC (left) and C/SiC
the monofilaments depend strongly on the deposition temperacomposites (right).
ture. Coarse-grained products must be avoided. The tensile
strength varies between 2500 and 3500 MN/m2. A Young's
As shown in Figure 19, such unidirectionally reinforced
modulus of up to 400 x f03 MN/mZ can be r e a ~ h e d ' ~ ~ . ~ ~ l .
composites are stronger than all other Sic materials, with
flexural strengths up to 1000 MN/m2. However, this strength
is limited to the direction parallel to the fibers. Nevertheless,
these composites are interesting and show promise for special
future applications.
7. Outlook
Fig. 17 SIC monofilament with carbon core.
Sic monofilaments with carbon cores (see Fig. 17) have
superior high temperature properties to those with W cores
because tungsten reacts with silicon carbide above 600°C.
Similar results are described by De Boltr5" and M ~ H e n r y [ ~ ~ ] .
No literature is available on the formation of Si3N4monofilaments. However, preliminary experiments have shown that
it should be possible to coat carbon fiber yam with Si3N4
by CVD[281.
Production of monolithic silicon ceramic objects from the
gas phase is precluded by frozen-in stresses. However, CVD
can be used for preparation of Sic-fiber and C-fiber reinforced
composites with a Sic matrix. The steps involved are shown
schematically in Figure 18. Such composites have a final closed
porosity of 15-20 %. The Sic matrix preferentially grows
Chemical vapor deposition promises improvements in the
production and quality of silicon ceramics. For instance, comparison of the high temperature strength properties shows
that chemical vapor impregnation of reaction-bonded silicon
carbide materials (Fig. 20, e) with pure silicon carbide (Fig.
20, d) gives product having strengths formerly only achievable
by liquid silicon impregnation (Fig. 20, c) which leads to
silicon excess and attendant low creep resistance. In contrast,
oxidation and creep resistance of vapor impregnated silicon
carbide is enormously improved.
C
1
2
0
200
LOO
600
800
1000
1200
1LW
r mi 4
Fig. 20. High-temperature flexural strength of unidirectionally reinforced
SIC/SiC and C/SiC composites (a) and gas-phase impregnated reactionbonded SIC (d) compared with commercially available hot-pressed S i c (b)
and reaction-bonded S i c with 5-1 5 % excess of Si (c) and untreated reactionbonded SIC (e).
Fig. 18. Stepwise production of unidirectionally reinforced S i c composites
(fiber-reinforced composites) by gas-phase impregnation.
Arzgnr. Chem. Int. Ed. Engl. 18. 295-304 ( I 979)
Fiber-reinforced Sic composites (Fig. 20, a) exhibit high
temperature strength data at 1400"C, superior to those of
all today's leading hot pressed materials (Fig. 20, b).
303
The production of ultrafine powders by chemical vapor
deposition could lead to a breakthrough of cold pressing
techniques without mechanical pressure for high quality SIC
and for Si3N4.
Received: January 18, 1979 [A 268 IE]
German version: Angew. Chem. 91. 316 (1979)
W Bunk. M . BBhmrr: Keramische Komponenten fur Fahrzeug-Gasturbinen. Springer, Berlin 1978.
[2] C. E . Loxrll The Silicon-Carbon Phase Diagram. National Research
Laboratories, TM-436, 1959.
[3] M. Noivolnj, E . Partme. R. Kieflrr. F . Benesou.skj. Monatsh. Chem.
85. 255 (1954).
[4] K . Blrgrn in P. P u p p ~ r .Special Ceramics 6 Brit. Ceram. Res. Assoc..
Penkhull 1973.
[5J L. J . Guuckler, H. L. Lukus. G. Pcrzobv. J. Am. Ceram. Sac. 58. 46
(1975)
[6] K . H . Jack. J . Mater. Sci. 1976, 1135.
[?I J . S<hiichriitg, High Temp. High Pressures 1978, 3.
[XI 0.Ruhicth. R. Schmidtr, Ber. Dtsch. Keram. Ges. 43. 173 (1966).
[9] J . W Ediiigtort. D. J . Row/I/Te. J L. Hritshal/, Powder Metall. lnt.
7, 8 2 (1975).
[lo] J . Sclilirhriiig, L. J . Gouckier. Powder Metall. Int. 9, 36 (1977).
[I I] E. Fitzer. K. Rriiimurh in F Benrsosskr: Hochtemperaturwerkstoffe.
6th Plansee Seminar, Reutte 1968. Springer, Vienna 1969.
1121 E. Firrer. R. Ehr in R. C. .Marsha//, J. W funsr. C . E. R w r r . Silicon
Carbide. University of South Carolina Press. Columbia 1973.
[I31 E . F i t z c v , H . Herhrr, J . Srhiichririg, Werksl. Korros. 24, 274 (1973).
[14] J . Srlilichri!ig in P. Popper: Special Ceramics 6. Brit. Ceram. Res. Assoc.,
Penkhull 1973.
[IS] E. Firzrr, 0. Ruhisch, Elektrowirine 5. 163 (1958).
[I61 R. 4. Allieyro. L. B. Coffin, J . R. Piriklrpough, J. Am. Ceram SOC.
3Y, 386 (1956).
[I71 S. Prochuika, R . J . Churles. Am. Ceram. SOC. Bull. 5 2 , 885 (1973).
[IS] R. F . Coe. R. J . Lumbj, J . F. Puwson in P. Popper. Special Ceramics
5 Brit. Ceram. Res. Assoc., Penkhull 1972. p. 361.
1191 E. Firrer: International Symposium of Factors in Densification and
Sintering of Oxides and Non-Oxide Ceramics. Hakone (Japan), Oct.
1978.
[20] E. Fitzer, W Frirz, D. Krlir, VDI-Z. 114. 1221 (1972)
[ Z l ] W V Koriensky in P. L. Walker: Chemistry and Physics of Carbon.
Vol. 9. Marcel Dekker, New York 1973, pp. 17311.
[22] JANAF, Thermochemical Tables of the Elements. Washington 1965
[23] G. Fritz, J . Grobr, 2. Anorg. All&. Chem. 229. 302 (1959).
1241 M . Bonnke. Dissertation, Universitat Karlsruhe 1965.
[I]
[25] M . B m n k r , E. Firrer, Ber Dtsch. Keram. Ges. 41. 180 (1966).
1261 J . E. Spruirli' Fundamentals of Chemical Vapour Depositlon. 2nd
Internatioiial Conference. The Electrochemical Society. New York 1970,
p. 279
[27] E. Frtzer, Chem.-1ng.-Tech. 41, 331 (1969).
[ 2 8 ] H. J Biihler. Disbertation, Universltit Karlsruhe 1976.
[29] L. Agqor, W Fritz, Cheni.-lng.-Tech. 43. 472 (1971).
[30] E. F i r m . D. Krhr, M . Srrhebkor, Chem.-1ng.-Tech. 4 5 , 1244 (1973)
131 ] M. Suhrhkur, Dissertation, UnibersitPt Karlsruhe 1973.
[32] D. Heyeii. Dissertation, Univcrsitat Karlsruhe, projected for 1979.
[33] H . Srrohineier. projected dissertation. Universitit Karlsruhe.
[34] F . 7%iintm/errr ul.. Sci. Ceram. 8. I33 ( I 975).
1351 W Blickrr, H . Huusuer, Ber. Dtsch. Keram. Ges. 55. 233 (1978).
[36] British Titan Products, Brit. Pat. I174859 (1967).
[37] Degussa. Brit. Pat. 960537 (1962).
1381 E. Enk, J . N i c k / . DBP 1047 180 (1958)
1391 E . S e ~ i ~ i i ~ ( . / i ~ i . i r iDBP
i B e ~ ,I 250796 (1964).
[40] U:E. Kuhn, Can. Pat. 695 133 (1962).
[41] 4 . W. F ~ U J ICan.
. S . Pat. 812157 (1966).
[42] R. ,U.Suirriger. Ind. En&. Chem. Prod. Res. Dev. I / , 230 (1972).
1431 S . P r o r h u ~ k i i n J . J . Burkc,, A . E . Gurum, R. N. K u r i . Ceramics for
High Performance Applications. Brook Hill Publ. Co.. Chestnut Hill.
Mass 1974, p. 239.
1441 K . S. Muxh.awi. c'. M . CnrtXe. J. Am. Ceram. Soc. 56. 628 (1973).
1451 P . E . D. Morgon, Final Reports NTlS-AD-778-373, Franklin Institute
Res. Lab., Philadelphia 1973.
1461 S . Prochiirku. c' D. Greskocich, General Electric Company. Technical
Information Series No. 77 C R D 104 11977).
1471 K . Kijinta. International Symposium of Factors in Densiiication and
Smteiing, Ilakone (Japan), Oct. 1978.
[4X] F Thiimmler, E. Por;, G. G r a r w h i , personal communication.
[49] L G. D[ILIC,.\.J . C . Withers, D. F. Buzarre, Final Report, Contract
NOW 64-0176 C, Bureau of Naval Weapons, Washington 1965.
[SO] J . C. Wirhrw, C L. ,Md'amd/~.ss, R Schwarz in: Advanced Fibrous
Reinforced Composites. Society of Aerospace and Process Engineers.
New York D 33 (1966).
[Sl] J A. Herzog, Metall 1967, 525.
[52] R. J . DiefiiidorfL personal communication.
[53] H. D. Eiihhorii. Diplomarbeit, Universitat Karlsruhe 1976.
[54] J . Horn, Diplomarbeit, Universitit Karlsruhe 1977.
[ 5 5 ] E. Firztdr, D. Kehr, &
Sohebkar
!l. in R. C. Marshall, J . W Faust. C
E. R J o n Silicon Carbide. University of South Carolina Press, Columbia
1973, p. 33.
[56] E. Fitre,, D.Kehr. D. Moriti. ,W Sahehkar, Chemical Vapour Deposition
Conference. London 1975, p. 589.
[57] H . Drbolt. :5 Krukonis, Arco Systems Division, Lowell, Mass., AFMLTR-73-140 (1972).
[58] K . D. M c H e i i r j . . R. E. Pesslrr. J . Compos. Mater. Y. 73 (1975).
C O M M U N JCATJONS
A New Entry to Naphthalene O x i d e s [ * * ]
By Richard R. Schmidt and Roy Arrgerbuuer1"l
Dedicated to Professor Horst Pommer on the occasion of his
60th birthday
Arene oxides and dihydroarenediols are of considerable
interest as metabolites of aromatic hydrocarbons"]. The discovery of the pronounced carcinogenic nature of epoxytetrahydroarenediols12] has prompted several recent synthesed3'.
In our work on the de no00 synthesis of sugars uiu cycloaddit i ~ n ' ~ we
' , have now found a facile and productive method
[*I
Prof. Dr. R. R. Schmidt, Dipl.-Chem R. Angerbauer
Fachbereich Chemie der Universitat
Postfach 77 33, D-7750 Konsranz (Germany)
[**I This work was supported by the Deutsche Forschungsgemeinschaft
and the Fonds der themischen Industrie. We are grateful to BASF AG
for a gift of cyclooctatetraene.
304
Aiiyew. Chrm. Inr. Ed. Engl. 1 R ( 1 9 7 9 ) N o . 4
0 Verlay Chemir, GmbH, 6940 Weinheim, IY7Y
0570-0833~7Y;0404-0304 5 02.5010
Документ
Категория
Без категории
Просмотров
2
Размер файла
1 072 Кб
Теги
nitrideчchemistry, chemical, deposition, vapor, silicon, modern, contributions, ceramic, carbide
1/--страниц
Пожаловаться на содержимое документа