J. P. Viricelle et al/ Composites Science and Technology 61(2001)607-614 carbon interphase is protected, as confirmed by a polishing and this explains the observation of holes in decrease of the specific area(Table 1) the B4c(2)layer. These micrographs also show that the At 900C, a global weight gain is observed. It is lower level of oxidation of matrix I(SiBC)remains very small. lised after 20 h( Fig. 10). The study with matrix 2 sam- between the layers B4 C(2)and SiC(3)(Fig. 13). Such, e than that measured at 800C, but this gain is not stabi Another point is the observation of a decohesic ples has revealed a change of the oxidation regime of phe enomenon ha as also been observed for samples oxi- B4C around 900C. The boron oxide film is no longer died at lower temperatures(600C)and is probably the protective, which leads to a markedly higher oxidation result of a thermal-expansion mismatch between the rate and to a strong weight gain compared to the constituents during thermal cyclins of nearly.07% is experiment performed at 800C. In the case of the At 1200.C, an initial weight loss composite, the phenomenon is the same. However, as observed during the first 50 min of the isotherm. The the boron oxide film is no longer protective, the inter- behaviour is the same as the previous one at 1000C. A phase can be partially oxidised, which explains the weight gain(0. 2%)is measured between 50 min and 7 lower weight gain at 900C compared to 800C for the h and is followed by a stabilisation and a weak loss at composite. This interpretation is also supported by the the end of the isotherm(2-0.03%). The variation of strong increase in specific surface area(AS=+0. 27 m/ the specific area(Table 1)indicates that the porosity is g, Table 1). Moreover, matrix I oxidation begins and sealed by silica. This behaviour at 1200C is in agree lso generates a weight gain. Thus, as for intermediate ment with the increasing-temperature experiment temperatures, 650 and 700oC, the weight variation is ig. 9). and with the behaviour observed for ID SiC/C/ given by the sum of a weight loss and a weight gain, but SiC composites [4]. The initial weight loss before stabi the second term(gain)is predominant at 900C. lisation is observed for temperatures higher than At 1000C, the initial weight gain during the first 7 h 1050C. Above 900oC, the boron oxide film is no longer is close to that observed at 900 C. But, for longer oxi- protective and the carbon interphase oxidation occurs dation times(7-20 h), a stabilisation occurs and a very However, at 900 and 1000oC, the global weight change mall loss(<0.025%)can be detected. The variation of remains positive (gain) because the formation rate of he specific area is negligible which allows to conclude B2O3(matrix 2 oxidation) and of borosilicate(matrix I that the porosity created by carbon combustion is oxidation) are predominant compared to B2O3 volatili- sealed. At this temperature, the porosity closure is sation rate and the interphase is protected. Above mainly explained by the oxidation of the fibers and of 1050C, the volatilisation term increases very strongly the first silicon carbide layer SiC(1) according to the and the result is an initial weight loss until the oxidation mechanisms previously described for SiC/C/SiC com- product of silicon carbide and Nicalon fibers seals the sites [4]. The small weight loss is the consequence of pores. Then, the weight change is stabilised. For longer B2O3 removal by volatilisation and reaction with water oxidation times, as the oxidation of Sic is limited by vapour which induces composition change of the bor- diffusion, its rate decreases. Thus, the volatilisation term silicate layer. Optical micrograph(Fig. 12)obtained becomes significant and explains the weak final weight from samples oxidised at 1000 C clearly confirms that loss at 1000 and 1200 C. The role of matrix 1 (SiBC)is the main oxidised phase in the matrix is boron carbide. quite limited for the samples which have been cut from The resulting boron oxide glass is removed during the plates. Its main influence is an additional weight gain g ple oxidised at Fig. 13. Optical micrograph(x90) of a sample oxidised at 1000C showing the preferential oxidation of B.C(2). showing a decohesion between B, C(2)and Sic()       !  5          5   -  ( : ),,
!  0  / 0 0      E  /         C,,
!   0         ',  J0 (, -   /  '          0     0  B   ),,
 -   5   0    ! /     6  0       0 / 0 0            C,,
 E      !         %/  !     5   0    !        ! /   /  / 0 0  ),,
    C,,
     -           0     5     

OK,'7 '  0! -  ( =  !  (  0   0     / 0 0 - !          ! +.,  7,,
!  / 0   0        / 0    / 0 0!       0     ),,
 : (,,,
!   / 0 0  0  5 7           ),,
  !  0      74', !            
,,'.1      -     5     00  / /                   :      !              5     5      ( 0                ?B@ -  / 0     F   '
2        / /    /     0      
 0 J0 ('          (,,,
   5               -   0   0      0  0             B'   -  0  /          (       :             /     B'  2 J0 (2                     /       +,,
                /      0   0 : (',,
!   / 0     ,,71       0  5 .,      -             (,,,
 : / 0 0 ,'1      /  .,   7    /        / 6         ,,21 -     5   -  (            -    (',,
   0    /   0       J0 )  /         ($     ?B@ -  / 0                 0   (,.,
 :  ),,
!    5   0             %/  !  ),,  (,,,
!  0  / 0 0    0       '
2  '      (        '
2            :  (,.,
!          0         / 0            A 5       - !  / 0 0     J 0    !          9 !       - !       05    / 6 5 / 0   (,,,  (',,
 -    (   F        /        E  >     / 0 0 J0 ('
 0 ),       (,,,
 /0        B' J0 (2
 0 ),       (,,,
 /0     /  B'  2 +(' , -%   . /
%%  %'  0""1 "!23