W Li, Z.H. Chen /Ceramics International 35(2009)747-753 pressures and becomes horizontal gradually. The highest intrusion increments are quite small, indicating the stability of cumulative Hg volume intruded is only 80 vol. o of the former, cumulative Hg volumes intruded and corresponding to the i.e.,20% pores in the specimens are inaccessible for 2nd horizontal part of the capillary pressure curve in Fi intrusion. As mentioned before, the evacuation before the 2nd According to traditional analysis, this kind of the PSd curve intrusion starting can make the Hg withdraw from the pores free implies the pores in 3D-C SiC possess two main components of shielding as much as possibly, hence the Hg only intrudes one are hundreds of microns-sized large pores, and the other are into the pores that are not occupied by the entrapped Hg after the smaller ones about 20-0. 1 um. while the trend of extrusion the last extrusion [16], and the disparity between the total Hg will be interpreted as the shielding effects of 0.2 um pores volumes of the lst and 2nd intrusion reflects the amount of leading to the surface on the smaller ones from the structural the entrapped Hg. Though not significant, the hysteresis still hysteresis"points of view exists between the 2nd extrusion and intrusion, and the As for the 2nd cycle, the intrusion PSD shows a similar peak extrusion branch is similar and parallel to the former. These style above 20 um with the first. Hence, it is reasonable to bvious hysteresis and Hg entrapment of 3D-C SiC implies believe that the intrusion in the pores of these sizes is reversible, the heterogeneity of inside pores in a wide domain and the the Hg can retreat from these large pores thoroughly, and they complicated connectivity of the system, according to the are free from shielding but leading to the surface directly. By aforementioned theories about miP mechanism comparison with SEM image of morphology of 3D-CdSiC in Fig 3, these large pores are quite matching with A-type pores 3.2. Pore size distribution (PSD) that are chambers between the twisting bundles, both at sizes and locations The pore size distribution of 3D-CSic is defined as the pore While in the range of 20-0. 1 um, the Hg increments of 2nd volume increments per unit logarithm of pore diameter vs intrusion are much lower than the lst ones, and the second peak corresponding pore diameters D, i.e., dV/d(log D)vs D, as has almost disappeared at 1.5 um on the curve. From the plotted in Fig. 2. Obviously, there are several key diameters on analysis of the capillary pressure curves above, it has been e Ist intrusion PSD curve, 1. e, 20 um, 1.5 um and 0. 1 um, known that the content of the entrapped Hg adds up to 20 vol % dividing the curve into three parts. The first is above 20 um, of the total intruded at lst intrusion. Now, it is clear that the especially at 100 um around, whose increasing rate is the entrapment occurs in the pores whose nominal sizes dwell in highest, corresponding to the high slope of intrusion plot in 20-0. 1 um during the lst extrusion. The exact porous structure Fig. l, which implies the great volume contents of the large where the Hg remains can be spotted by focusing on the carbon pores about hundreds of microns. After the first peak at fabrics, for their architecture is one of the main determinants of 100 um, the PSD curve declines drastically with the pressure as-produced 3D-CSiC's porosity. The longitudinal size of the increasing, till 20 um, the dv/(log D) reaches a relative low stitch of 3D-braided fabric adopted in this paper is about value. A convergent structure of pore is proposed to explain this 5.5 cm, to which the length of the specimens sliced is trend, which restricts the Hg when its meniscus is advancing to comparable(4-5 cm), thus the A-type large pores on the the narrow throats, e.g., 20 um-sized, and the increments surfaces of specimens are not individual but repeated inside reduce consequently. Another peak on PSD appears at 1.5 um, One can imagine, if the applied pressure breaks through the one al l height but wider range(20-0. 1 um), than the first threshold of throat, e.g., 20 um-sized, the mercury will rush D um. The last phase is below 0. I um, and the into the large chambers inside massively, but this volume will only be attributed to the 20 um or smaller pores automatically, hat is what has happened on the PSd of lst intrusion. While 0.07 ▲- Ist extrusi 0.06 2nd Intros 005 0.04 0.01 000m 5x10°10° 10 Diameter(A) Fig. 3. Morphology of the cross-section of 3D-CfSic by SEM. Typepressures and becomes horizontal gradually. The highest cumulative Hg volume intruded is only 80 vol.% of the former, i.e., 20% pores in the specimens are inaccessible for 2nd intrusion. As mentioned before, the evacuation before the 2nd intrusion starting can make the Hg withdraw from the pores free of shielding as much as possibly, hence the Hg only intrudes into the pores that are not occupied by the entrapped Hg after the last extrusion [16], and the disparity between the total Hg volumes of the 1st and 2nd intrusion reflects the amount of the entrapped Hg. Though not significant, the hysteresis still exists between the 2nd extrusion and intrusion, and the extrusion branch is similar and parallel to the former. These obvious hysteresis and Hg entrapment of 3D-Cf/SiC implies the heterogeneity of inside pores in a wide domain and the complicated connectivity of the system, according to the aforementioned theories about MIP mechanism. 3.2. Pore size distribution (PSD) The pore size distribution of 3D-Cf/SiC is defined as the pore volume increments per unit logarithm of pore diameter vs corresponding pore diameters D, i.e., dV/d(log D) vs D, as plotted in Fig. 2. Obviously, there are several key diameters on the 1st intrusion PSD curve, i.e., 20 mm, 1.5 mm and 0.1 mm, dividing the curve into three parts. The first is above 20 mm, especially at 100 mm around, whose increasing rate is the highest, corresponding to the high slope of intrusion plot in Fig. 1, which implies the great volume contents of the large pores about hundreds of microns. After the first peak at 100 mm, the PSD curve declines drastically with the pressure increasing, till 20 mm, the dV/d(log D) reaches a relative low value. A convergent structure of pore is proposed to explain this trend, which restricts the Hg when its meniscus is advancing to the narrow throats, e.g., 20 mm-sized, and the increments reduce consequently. Another peak on PSD appears at 1.5 mm, with lower height but wider range (20–0.1 mm), than the first one at 100 mm. The last phase is below 0.1 mm, and the intrusion increments are quite small, indicating the stability of cumulative Hg volumes intruded and corresponding to the horizontal part of the capillary pressure curve in Fig. 1. According to traditional analysis, this kind of the PSD curve implies the pores in 3D-Cf/SiC possess two main components: one are hundreds of microns-sized large pores, and the other are the smaller ones about 20–0.1 mm. While the trend of extrusion will be interpreted as the shielding effects of 0.2 mm pores leading to the surface on the smaller ones, from the ‘‘structural hysteresis’’ points of view. As for the 2nd cycle, the intrusion PSD shows a similar peak style above 20 mm with the first. Hence, it is reasonable to believe that the intrusion in the pores of these sizes is reversible, the Hg can retreat from these large pores thoroughly, and they are free from shielding but leading to the surface directly. By comparison with SEM image of morphology of 3D-Cf/SiC in Fig. 3, these large pores are quite matching with A-type pores that are chambers between the twisting bundles, both at sizes and locations. While in the range of 20–0.1 mm, the Hg increments of 2nd intrusion are much lower than the 1st ones, and the second peak has almost disappeared at 1.5 mm on the curve. From the analysis of the capillary pressure curves above, it has been known that the content of the entrapped Hg adds up to 20 vol. % of the total intruded at 1st intrusion. Now, it is clear that the entrapment occurs in the pores whose nominal sizes dwell in 20–0.1 mm during the 1st extrusion. The exact porous structure where the Hg remains can be spotted by focusing on the carbon fabrics, for their architecture is one of the main determinants of as-produced 3D-Cf/SiC’s porosity. The longitudinal size of the stitch of 3D-braided fabric adopted in this paper is about 5.5 cm, to which the length of the specimens sliced is comparable (4–5 cm), thus the A-type large pores on the surfaces of specimens are not individual but repeated inside. One can imagine, if the applied pressure breaks through the threshold of throat, e.g., 20 mm-sized, the mercury will rush into the large chambers inside massively, but this volume will only be attributed to the 20 mm or smaller pores automatically, that is what has happened on the PSD of 1st intrusion. While Fig. 2. Pore size distribution (PSD) curves of 3D-Cf/SiC from the two MIP cycles. The extrusion data are plotted negatively for convenience of compar￾ison. Fig. 3. Morphology of the cross-section of 3D-Cf/SiC by SEM. Type A represents the inter-bundle large chambers exposed by slicing. Types B and C are both paths between the bundles, and C leads to the sample’s surface, connecting the inside and the outside of the sample. W. Li, Z.H. Chen / Ceramics International 35 (2009) 747–753 749