Journal of the American Ceramic Sociery--Morscher Vol. 80. No 8 Nicalon), three different interphases(one carbon and two was performed on single-fiber microcomposites that is de types of BN), and two different chemical-vapor-infiltrated- scribed in detail elsewhere. 3 The fibers were mounted with matrix(CVI-matrix) vendors were used in this study, as de- epoxy onto cardboard tabs. The epoxy was gripped, and acous- cribed in Table 1. Examples of minicomposite polished cross tic emission(AE) sensors were attached to the epoxy so that sections are shown in Figs. 1((aHe)) for each minicomposite cracking could be monitored. The ultimate failure load had been determined for each minicomposite type (Table I)from The general minicomposite processing approach is identical monotonic tensile tests, and the load for significant micro to common CVI-SiC composites. For cracking was fairly well characterized by the AE activity. A minicomposites(C-Nic), the tows wer precracking load was chosen based on the AE data. The crack ardon. an pacing was then determined from polished longitudinal sec- onsecutive steps by the same vendor tions of each minicomposite(Fig. 2). The average crack spac- shown in Table I. This is not a saturation crack spacing, b a was very uniform, and the interphase thickness was -0.4 um ing for each minicomposite for the applied precracking load is (Fg.1(a) For the bn interphase minicomposites(3MBN-Nic, PBN cause saturation had not occurred at or below the precracking HN, 3MBN-HN), the tows were first coated by a vendor and oad for any of the minicomposites that were tested in this then the coated tows were mounted on graphite racks and com- study. Twenty five to fifty unload/reload cycles were per ited with Sic by a different vendor. The BN interphase formed during this precracking step were produced at different temperatures and differ in com The stress-rupture test was either performed in the univers sition and microstructure. The BN for the 3MBN-Nic(Fig esting machine or in simple dead-weight load stress-rupture (b)and 3MBN-HN (Fig. l(e))minicomposites was deposited rigs. The distance from the top to the bottom of the outside using triethylborane at 1050 C. o This bn has been well char- walls of the furnace was 35 mm. The hot zone was -12 mm acterized on tows and in composites by Brennan. It is uniform The temperatures that were chosen for stress-rupture testing and has >5 at. %oxygen and -15 at carbon. It can be de- were in the range of 700-1200.C. For all the stress-rupture scribed as being composed of small turbostratic crystals that tests, a constant load was applied at room temperature, the temperature was increased(100%C/min) to the desired value (PBN-HN) was processed with boron trichloride(BCl,)and and the time was monitored for-100 h or until failure. If no ammonia(NH,OH)at 1400@C. This BN is more pure with failure occurred after-100 h, the temperature was decreased to ittle oxygen or carbon. The BN is turbostratic with the c-axis room temperature, so that the minicomposite surface could be well aligned normal to the fiber axis. 2 There also is significant observed. After examination, the minicomposite was then re- oating nonuniformity across the tow. The outer fibers of the mounted and reloaded to the original stress-rupture load for ow can have a bn thickness of up to 3 um, whereas the another 100+ h, after which the same procedure was followed interior fibers of the tow are relatively uniformly coated with nless failure had occurred. This type of very-low-cycle ther- BN.5 um thick(Figs. I(c)and(d)) mal fatigue was followed until failure or for times at tempera he minicomposites also differed in physical appearance. ture of up to 1000 h. For some experiments, if the samples did The PBN-HN, 3MBN-Nic, and C-Nic minicomposites were not fail after several hundred hours, the temperature was in- very compact( before and after SiC infiltration), with the fibers creased or the retained room-temperature tensile strength was undled tightly together. The overall diameter of the minicom- determined as noted below osites was <l mm(Fig. 1). The appearance of the 3MBN-HN minicomposites were very fibrous(before and after Sic infil. (3) Scanning Electron Microscopy tration), with the fibers loosely bundled. This specific tow must Minicomposite fracture surfaces and some longitudinal have been spread during the Bn coating step. The result was a tions were examined using a conventional minicomposite that effectively had many single-fiber com microscopy(SEM) microscope(Model JEOL 840A, tes that were loosely connected to each other. The overall Tokyo, Japan)and a field-emission SEM(FESEM) mici diameter of these minicomposites was 2-3 mm. Although the (Model $4500, Hitachi, Tokyo, Japan). The FESEM was used minicomposites were nonuniform in many respects, the test at 5 kv, which enabled energy-dispersive spectroscopy (EDS) nethodology that is described below should have minimized spectra to be obtained for relatively small areas. Under these the importance of these nonuniformities conditions, boron was easily detected for BN (2) Mechanical Testing IlL. Results Tensile testing consisted of, first, precracking the minicom- posites and, second, subjecting them to a constant-load stress (1 Minicomposite Tensile Stress-Rupture Results rupture test. The precracking was performed using a universal The load, time, temperature, and ae data from a typical testing machine(Model 4502, Instron, Canton, MA). The test precrack and stress-rupture test for a PBN-HN minicomposite set-up is almost identical to one that was used in a study that tested at 950"C are shown in Fig. 3. The precracking step consisted of 50 unload/reload cycles with loads of 11-55 kg Both fibers are produced by Nippon Carbon; in this paper, the fibers have been The AE energy per event is shown as a function of time, the majority of which occurred on the first loading cycle Table I. Properties of Minicomposites Tested diameter(um) per tow Interphase ad(kg) (cracks/mm) 50 Carbon19±2[2.311.3[14 Nicalon 202 14.4 10±1[24]647.7[6 0.59 PBN-HN Hi-Nicalon 13.3 PBN14.5±1[2.]11.3[16 3MBN-HN Hi-Nicalon 13.9 4905 BN14.5±1[20]11.3[5y were cach sup plied by B: E: Goodrich aerospace. nIcation 202 a. d Hi-Nicalon were ma ufrcturecd by Nippon carbon tol Nic. PBN-HN. and 3MBN-HN the mber stress, thrich is the smes on th t fibers m the bers were s ll loaded she he oss d divided by the cross-sectionitle area of the fiber. "i he numbe brous, an accurate fiber therefore, the failure and precrack stress are only approximations. It was too difficult to obtain a crack density. Longitudinal polished sections were obtained which showed cracks; however, this minicomposite type was too fibrous to quantify the average crack spacinNicalon),§ three different interphases (one carbon and two types of BN), and two different chemical-vapor-infiltrated￾matrix (CVI-matrix) vendors were used in this study, as de￾scribed in Table I. Examples of minicomposite polished cross sections are shown in Figs. 1((a)–(e)) for each minicomposite type. The general minicomposite processing approach is identical to common CVI-SiC composites. For the carbon interphase minicomposites (C-Nic), the tows were mounted on graphite racks, coated with carbon, and then composited with SiC in consecutive steps by the same vendor. The carbon interphase was very uniform, and the interphase thickness was ∼0.4 mm (Fig. 1(a)). For the BN interphase minicomposites (3MBN-Nic, PBN￾HN, 3MBN-HN), the tows were first coated by a vendor and then the coated tows were mounted on graphite racks and com￾posited with SiC by a different vendor. The BN interphases were produced at different temperatures and differ in compo￾sition and microstructure. The BN for the 3MBN-Nic (Fig. 1(b)) and 3MBN-HN (Fig. 1(e)) minicomposites was deposited using triethylborane at 1050°C.10 This BN has been well char￾acterized on tows and in composites by Brennan.2 It is uniform and has >5 at.% oxygen and ∼15 at.% carbon. It can be de￾scribed as being composed of small turbostratic crystals that are not well aligned with the fiber axis. The PBN material (PBN-HN) was processed with boron trichloride (BCl3) and ammonia (NH4OH) at 1400°C.11 This BN is more pure with little oxygen or carbon. The BN is turbostratic with the c-axis well aligned normal to the fiber axis.12 There also is significant coating nonuniformity across the tow. The outer fibers of the tow can have a BN thickness of up to 3 mm, whereas the interior fibers of the tow are relatively uniformly coated with BN ∼0.5 mm thick (Figs. 1(c) and (d)). The minicomposites also differed in physical appearance. The PBN-HN, 3MBN-Nic, and C-Nic minicomposites were very compact (before and after SiC infiltration), with the fibers bundled tightly together. The overall diameter of the minicom￾posites was #1 mm (Fig. 1). The appearance of the 3MBN-HN minicomposites were very fibrous (before and after SiC infil￾tration), with the fibers loosely bundled. This specific tow must have been spread during the BN coating step. The result was a minicomposite that effectively had many single-fiber compos￾ites that were loosely connected to each other. The overall diameter of these minicomposites was 2–3 mm. Although the minicomposites were nonuniform in many respects, the test methodology that is described below should have minimized the importance of these nonuniformities. (2) Mechanical Testing Tensile testing consisted of, first, precracking the minicom￾posites and, second, subjecting them to a constant-load stress￾rupture test. The precracking was performed using a universal testing machine (Model 4502, Instron, Canton, MA). The test set-up is almost identical to one that was used in a study that was performed on single-fiber microcomposites that is de￾scribed in detail elsewhere.13 The fibers were mounted with epoxy onto cardboard tabs. The epoxy was gripped, and acous￾tic emission (AE) sensors were attached to the epoxy so that cracking could be monitored. The ultimate failure load had been determined for each minicomposite type (Table I) from monotonic tensile tests, and the load for significant micro￾cracking was fairly well characterized by the AE activity. A precracking load was chosen based on the AE data. The crack spacing was then determined from polished longitudinal sec￾tions of each minicomposite (Fig. 2). The average crack spac￾ing for each minicomposite for the applied precracking load is shown in Table I. This is not a saturation crack spacing, be￾cause saturation had not occurred at or below the precracking load for any of the minicomposites that were tested in this study. Twenty five to fifty unload/reload cycles were per￾formed during this precracking step. The stress-rupture test was either performed in the universal testing machine or in simple dead-weight load stress-rupture rigs. The distance from the top to the bottom of the outside walls of the furnace was 35 mm. The hot zone was ∼12 mm. The temperatures that were chosen for stress-rupture testing were in the range of 700°–1200°C. For all the stress-rupture tests, a constant load was applied at room temperature, the temperature was increased (100°C/min) to the desired value, and the time was monitored for ∼100 h or until failure. If no failure occurred after ∼100 h, the temperature was decreased to room temperature, so that the minicomposite surface could be observed. After examination, the minicomposite was then re￾mounted and reloaded to the original stress-rupture load for another 100+ h, after which the same procedure was followed, unless failure had occurred. This type of very-low-cycle ther￾mal fatigue was followed until failure or for times at tempera￾ture of up to 1000 h. For some experiments, if the samples did not fail after several hundred hours, the temperature was in￾creased or the retained room-temperature tensile strength was determined as noted below. (3) Scanning Electron Microscopy Minicomposite fracture surfaces and some longitudinal sec￾tions were examined using a conventional scanning electron microscopy (SEM) microscope (Model JEOL 840A, JEOL, Tokyo, Japan) and a field-emission SEM (FESEM) microscope (Model S4500, Hitachi, Tokyo, Japan). The FESEM was used at 5 kV, which enabled energy-dispersive spectroscopy (EDS) spectra to be obtained for relatively small areas. Under these conditions, boron was easily detected for BN. III. Results (1) Minicomposite Tensile Stress-Rupture Results The load, time, temperature, and AE data from a typical precrack and stress-rupture test for a PBN-HN minicomposite tested at 950°C are shown in Fig. 3. The precracking step consisted of 50 unload/reload cycles with loads of 11–55 kg. The AE energy per event is shown as a function of time, the majority of which occurred on the first loading cycle. § Both fibers are produced by Nippon Carbon; in this paper, the fibers have been designated ‘‘Nic’’ and ‘‘HN,’’ respectively. Table I. Properties of Minicomposites Tested Name of minicomposite† Fiber type‡ Average fiber diameter§ (mm) Number of fibers per tow Interphase Failure load¶ (kg) Precracking load¶ (kg) Average crack density along length of minicomposite (cracks/mm) C-Nic Nicalon 202 15.3 450 Carbon 19 ± 2 [2.3] 11.3 [1.4] 0.44 BN-Nic Nicalon 202 14.4 250 BN 10 ± 1 [2.4] 6.4–7.7 [1.6] 0.59 PBN-HN Hi-Nicalon 13.3 490 PBN 14.5 ± 1 [2.1] 11.3 [1.6] 1.1 3MBN-HN Hi-Nicalon 13.9 490§ BN 14.5 ± 1 [2.0]†† 11.3 [1.5]†† ?‡‡ † C-Nic was supplied by Hyper-Therm Corp., Huntington Beach, CA, or B. F. Goodrich Aerospace, Brecksville, OH; BN-Nic, PBN-HN, and 3MBN-HN were each supplied by B. F. Goodrich Aerospace. ‡ Nicalon 202 and Hi-Nicalon were manufactured by Nippon Carbon, Tokyo, Japan. § The average fiber diameter was determined by measuring the diameters of 50 fibers from a polished cross section for each minicomposite type. ¶ The value in brackets is the fiber stress, which is the stress on the fibers if the fibers were fully loaded, i.e., the load divided by the cross-sectional area of the fiber. ††The number of fibers per tow is assumed to be the same as that for PBN-HN. Because the minicomposite was so fibrous, an accurate fiber count was impossible; therefore, the failure and precrack stress are only approximations. ‡‡It was too difficult to obtain a crack density. Longitudinal polished sections were obtained which showed cracks; however, this minicomposite type was too fibrous to quantify the average crack spacing. 2030 Journal of the American Ceramic Society—Morscher Vol. 80, No. 8