`Coatings on SiC Fibers for Non-Oxide High Temperature
`Resistant CMCs
`
`M. Tsirlin, Y. E. Pronin, E. K. Florina, S. Kh. Mukhametov, M. A. Khatsernov
`SRC ofRF GNIIChTEOS, Moscow
`H.M. Yun
`NASA Glenn Research Center, Cleveland, OH
`R.Riedel, E. Kroke
`Technische Universitiit, Darmstadt
`
`1
`
`Introduction
`
`Interphase coatings are necessary constituents of ceramic matrices/ceramic fibers composites
`(CMC). They provide high fracture toughness and prevent a catastrophic destruction of the
`essentially brittle materials. The interphases for non-oxide high temperature CMC intended
`for a long-term work in oxidizing environments (HTI) have gone a long way of development.
`Starting from monolayered adhesion controlling coating of carbon with working temperature
`up to 350-400°C, they were improved to analogues BN coating fit for temperatures up to 600-
`700°. The work on alloying BN with Si [1] seems gave the possibility to raise the temperature
`of exploitation up to 1000-1100°C and to enhance the oxidation and hydrolysis resistance.
`Nevertheless this level is not enough for the hot units with oxidizing environment in
`aviation engines [2]. High temperature (1300-1400 °C) and long-term work (thousands of
`hours) at cycling thermo-mechanical conditions, which are inevitable in this application, put
`forward especially high demands to HTI. A prolonged stability of the adhesion control and a
`fibers protection against the oxidation, connected with matrix cracking at the loading, should
`be ensured. So much attention is paid to more complicated two-layered structure. It is also
`necessary to take into account, that economic and technological aspects of CMC production
`dictate the first turn application of thin fibers in tows such as Nicalon, Hi-Nicalon, Hi-Nicalon
`S, Tyranno, Sylramic etc., which are very sensitive to oxidation and processing of coatings.
`So the preservation of fiber strength during deposition is a serious problem. The protection of
`uncoated fiber ends in the composite, formed by machining, should be considered as a
`separate problem. All these problems are not overcome until now.
`According to the theoretical model [3] the strength of a brittle fiber with a brittle coating is
`an inverse function of"layer thickness to fiber diameter" and "Joung module coating to fiber"
`ratios. The critical thickness of layer (until the strength of fiber begins to fall down) is about
`1,5 % of fiber diameter at the second ratio -1. Thus a dense layer with a strong adhesion on
`thin fibers with diameter 11-14 gm should have the thickness no more than 0,15-0,25/am and
`so make it very problematic to succeed in putting regular fully dense coating. The
`experimental data for SiC/B/W monofilaments gave rather good support to this theory. But in
`thin fiber processing even layers of low modulus materials (BN, E=90 GPa) on SiC base
`fibers (E=250-400 GPa) decrease the strength sufficiently. The detail experimental data [4]
`
`GE-1018.001
`
`
`
`150
`
`discovered a surface damage of fibers after staying in the gas components of BN CVD at
`about 1000°C, 1-10 kPa, 1 min (due to recrystallization in Ar, corrosion in BF3 and depletion
`of free carbon in NI-I3). The formation of a dense oxide film (E=120 GPa) of about 0,5 lam on
`the fiber surface in preoxidation prevents the gas evolving after the internal pyrolysis [5] but
`reduces considerably the strength after oxidizing because the action of oxide layer, irregular
`oxidizing of free carbon and other surface damages.
`The intensive study during the last years gave evidence that the protection of main
`components of HTI against the oxidation is more complicated problem, than it had been
`postulated [6]. The comparison of many Si-, Ti- or B-based high oxidation resistant nonoxide
`materials is shown in Fig. 1 [7,8,9 et al.]. They all form dense meltable at middle temperatures
`oxide layers (SiO2, TiO2, B203 or mixtures of them) - a barrier for 02 penetration inside the
`bulk material. Si-based compounds demonstrate very close data for the resistance, depending
`mainly on the kind of processing. The CVD SiC has nearly the best properties. The reported
`data for SiBN3C [10] are very interesting but have yet no confirmation. Even the slowest
`oxidation can form on SiC at 1300 -1400 °C a limited SiO2 layer (about 0,5 micron) during
`6-10 hours. So any protective layer can succeed only to decelerate the oxidation of fibers and
`the major protective role goes over to the matrix. So the self-healing of cracks in the matrix
`under the influence of mechanical and thermal stresses becomes a very important technical
`task. Ceramics of Si-B-N-C type [ 11,12] are of great interest in this aspect as well.
`
`I,E-05
`
`1,E-06
`
`1,E-07
`
`.= ~,E-09
`
`5,5
`
`6
`
`6.5
`
`7
`
`7,5
`
`8
`
`8~
`
`T4 (10-4 K-~)
`
`Figure 1: Oxidation kinetics ofnonoxide ceramic materials [7,8,9 at al.]: 1-8: SiC; 9: TiSi2; 10: TiC; 11: MoSi2;
`12: Si3N4; 13: SiBN3C; 14: NiAI; 15: Si; 16: Ti; 17-20: Ti3SiC2. CVD, PLS, HP, 7- via PCS, O-
`monocrystalline, metals (other materials) for comparison
`
`The stability of HTI adhesion control demands also a new solution. Carbon disappears after
`400-500°C in air and opens the way for oxygen inside CMC. BN evaporats at 800-900 and
`melts in the interval 900-1100°C. Other Si, Ti, B-base materials form dense melting oxides at
`the work temperatures too. Being good for oxidation protection oxides harm adhesion
`stabilization and fiber strength at cycling temperatures.
`Above considerations have led the authors to a more complex (three layered) but
`technically more radical principal scheme of interphase for HT-CMC. The layer on the fiber
`surface should be porous with low shear strength to control the adhesion and compensate the
`CTE difference between the fiber and the next layer. The second layer should be a protective
`
`GE-1018.002
`
`
`
`151
`
`one. It ought to be dense and high oxidation resistant. The third layer is analogues to the first
`one to control the adhesion between the protective layer and the matrix. What is especially
`important and obligatory: both the first and the third layers should not change (melt, dissolve
`or evaporate) for long periods of cyclic exploitation.
`The reality of theoretical structure and detailed positive and negative properties of used
`components could be evaluated only after an experimental study. So the authors are carrying
`out systematic experimental work with a scope of perspective components in order to
`combine an appropriate structure for HTI. In this paper experimental results on coating Hi-
`Nicalon fibers with MoSiz and Si-B-N-C as protective layers and Ti3SiC2 as adhesion
`controlling layer are represented. The all three compounds are objects of intent attention in the
`researches of HT- materials field during last ten years.
`
`2
`
`Experimental Procedure
`
`The vertical vacuum hot walls reactor was used for putting interphase coatings by CVD and
`RVD in continuous and stationary conditions in vacuum, Ar, H2 and N2 environments. The
`cylindrical graphite heater formed the constant temperature zone of 200 mm length for the
`temperature up to 1400°C. There are the reel out and reel up units (for continuous work) and
`the fiber and plate specimens holder (for fabricating discrete specimens). The installation was
`provided with the equipment for measuring and transportation of gas mixtures, condensation
`and absorption of the wastegases and the temperature measurement. The concentrations of
`vapors were controlled by the temperature in evaporators. The temperature in the furnace was
`measured by the 1R-pyrometer and was calibrated with a thermocouple before the run.
`The Mo-Si protective layers were fabricated by CVD + RVD two-stage process on Hi-
`Nicalon monofilaments and tows. Mo-foil and SiC/C monofilaments of 140 gm diameter
`were used as model objects. The mixtures of hydrogen with MoC15 were used for
`molybdenum layers and with SIC14- for their following siliconizing by RVD. The layers of
`0,l to 0,7 gm were fabricated on Hi-Nicalon, up to 2-3 gm - on SiC/C filaments and up to 20
`micron - on Mo-foil. The conditions: the temperature of Mo-deposition - 700-800°C, of Mo-
`Si (siliconizing) - 850-1050°C, the pressure - 1- 5 kPa and the vapors concentrations about 1
`tool. % for MoC15 and 2-30 % mol. for SIC14 in hydrogen.
`Ti3SiC2 layers were fabricated by CVD. To create the possibility for element and phase
`local analyses the first experiments were carried out in stationary conditions on graphite plates
`180x 12x2 mm and Nicalon tows fixed together in the working zone of furnace. The coatings
`had the thicloaessof 100-300 micron after two hours run on graphite or on the preliminary SiC
`layer of 5-10 microns. Then the process was realized on Hi-Nicalon fibers at continuous
`conditions (about 3 m per run) providing the coatings of about 0,5-1 micron after two and five
`minutes runs. Me2SiC12, MeHSiClz and HSiC13 were used as the source of Si (the first two of
`them - as the source of C also), CH4 and CC14 - as the source of C, .TIC14 - as the source of Ti
`with the ratio Ti:Si:C = 3:(1 to 3): (0,5 to 2). The ratio of reagents to H~ was about 1 :(10-17).
`The PIP Si-B-N-C coatings were put over Mo-Si coating or directly on the filaments in Na
`box, equipped with horizontal tube furnace. The Hi-Nicalon tows were impregnated with
`polymer solutions in TGF, n-hexane, toluene or xylene (concentration 0,02 to 2 mas. %).
`Then tows were dried, heat treated at 800-1000°C and cooled with the furnace (2 to 5 cycles).
`PCS and two types of PBCSZs with initial ratios of Si/B = 2 and 3 were used [11].
`
`GE-1018.003
`
`
`
`152
`
`Thermostability and oxidation resistance tests of fibers were carried out in the horizontal
`furnace with the graphite tubular heater with the alumina tube enclosed. It was possible to
`work with vacuum, N2, Ar and air environment. Temperatures 1200 to 1400°C and times of
`heating 1.5, 5, 9 and 20 to 60 hours were used. Nicalon and Hi-Nicalon S were also tested at
`times for comparison. The changing of specimens appearance and structure were the main
`criteria of oxidation resistance. The tensile strength before and after the heat-treatment was
`evaluated by Instron machine testing. Specimens were analyzed by optical microscopy, SEM,
`EPMA, ESCA, WDXA, XRD and AES.
`
`3 Results and Discussion
`
`3.1 Mo-Si coating by CVD
`
`The Mo-foil siliconized at 1200°C formed two-layered coating of about 14 ~tm in sum The
`ratio Si/Mo (at. %) = 1,80-1,75, which is close to MoSi2, was detected by SEM-EPMA from
`the coating surface to 8-9 gm deep, then it fell sharply to 0,56 (close to MosSi3) near 14 gm
`deep and then became close to zero. These data correspond well with ones of the RVD of
`porous aluminum preform [13]. The Mo-coating on SiC/C monofilaments was regular up to
`thickness of 1,0-1,5 micron and had only very slight tendency to defoliation under
`mechanical impulses. The WDX spectrum (Camebax) detected only molybdenum. The
`coating siliconized at 1200°C was like the coating on Mo-foil. The thimaer (950°C) layer had
`other characteristics. The data of Camebax analysis for main elements Si, Mo and O (C was
`counted for 100% balance) in Table 1 showed the depletion of the coating in two layers. The
`Mo-layer adjoins the fiber surface, above a Si-O layer is located. The oxygen is absorbed
`most likely by porous Si-layer on the surface. These data gave real evidence that silicon and
`molybdenum up to 950°C do not forna a monolayer of a molybdenum silicide.
`
`Table 1: Local concentrations of the main elements in the points across the Mo-Si coating on
`SiC/C (950°C/CVD) fibers (at. %%).
`
`Location
`Fiber
`First layer
`Outer uneven material
`
`Mo
`0.02
`71.52
`5.45
`
`Si
`45.52
`
`3.47
`54.61
`
`O
`0.45
`2.91
`1.25
`
`C
`54.02
`22.10
`38.69
`
`Sum
`100.0
`100.0
`100.0
`
`The layers of Mo and Mo-Si on monofilaments in the tow looked dense and regular, were
`0,1 to 0,3 micron (Fig.2a,b). There were no places with weak adhesion to native fiber surface,
`cracks and sticking fibers together. Increasing the temperature and time made the layers
`thicker but cruder and creasing and the cracks appeared. AES gave evidence of Mo-layer
`having thickness of 0.2 - 0.5 micron against a background of Si, C and O. Signals of Si and C
`probably arose from underlayer material of fiber, oxygen located as usually close to the
`surface. For the Mo-Si-coatings (Fig.2c) there were a thin (about 20-50 nm) Si-O layer on the
`very surface, then Mo-layer aligning to the native fiber surface (Fig.3a). The heat treatment at
`1400°C in N2 to initiate a reaction between Mo and Si gave no positive results: Mo-layer
`disappeared substituted by SiO~-layer of 200-250 nm thick (Fig. 3b). So these results also
`show, that there are no MoSi2 or other chemical Mo-Si compounds in the coating..
`
`GE-1018.004
`
`
`
`153
`
`3.2 Si-C and Si-B-N-C coating by PIP-technology
`
`One or two impregnations were enough to form coating thickness of 0,2-0,3 micron
`¯ (Fig.2d). The regular structure had no sufficient defects on the large part of the surface but
`nevertheless local bulbs were formed in some places even for the most diluted solutions and
`accordingly for the least thickness. At larger concentrations these bulbs grew, forming pours
`and sticking of monofilaments. Regular dense two-layered coatings of Mo-Si and Si-B-N-C
`were fabricated with very diluted solutions of polymer and high boiling solvents. They
`differed from SiC PIP coatings, which were friable [14].
`
`Figure 2: High temperature interphases on Hi Nicalon after the deposition: a,b: CVD Mo; c: CVD Mo-Si; d: PIP
`Si-B-N-C; e: CVD Ti-Si-C. Oxide layers after 1300°C/20h testing in air on: f: Hi-Nicalon as received; g: with
`the Mo-Si coating; h: with the Si-B-N-C coating; i: with (Mo-Si) + (Si-B-N-C) coating; k: with Ti-Si-C coating.
`
`a) b)
`Figure 3: AES data of Hi-Nicalon with the Mo-Si coating before (a) and after (b) heat-treating at 1400°C/20h in
`N2
`
`Depths, n131
`
`GE-1018.005
`
`
`
`154
`
`Ti
`
`Ti
`
`a
`
`b
`
`Figure 4: a) Ti-Si-C-coating from (CH3)H~SiCI2 and TIC14 on graphite~ b) Ti-Si-C-coating from (CH3)HSiCI2
`and TIC14 on Nicalon with the SiC interlayer; c) Ti-Si-C-coating from CH4, TIC14 and HSiCI3 on graphite with
`the SiC interlayer.
`
`3.3 Ti-Si-C coating by CVD
`
`The coatings of large thickness were analyzed by EPMA. The data were put in the triangle
`thermodynamic diagram [15] (Fig.4). All the layers had rough multilayered structure. For
`runs with MeHSiC12 and Ti:Si:C=3:I:I on graphite substratum the layers were about 400
`micron (Fig.4a) The composition changed from areas 7 to 2 crossing potentially monophase
`Ti3SiC2 area 1 only near the outer surface (pp.9,10). Then it is returning back in area 8(p.11).
`
`GE-1018.006
`
`
`
`155
`
`On the SiC interlayer of 2-5 micron (Fig.4b) the rate of deposition decreased dramatically.
`The layers were about 25-35 micron. Pointsl-5 are located in SiC of fiber and interlayer. The
`area 1 is reached close the interlayer in points 8,9. Then the coating was enriching gradually
`with Ti (plS. 10-15). The mean composition (XRD data, mas.%) of the grinded up Coating was:
`C:TiC:Ti3SiC2 = 18,1:56,6:25,3. In the runs with HSiC13 on the SiC interlayer of 1-2 micron
`the ratio Ti:Si:C=3:3:1(0,5) was chosen to decrease Ti and C concentrations. This correlates
`with tendency from other works [16,17]. The lower deposition rate decreased additionally, the
`layers about 10-15 micron being formed. So we can see (Fig.4c) that after initial graphite
`(pp.l-3) interlayer is not indicated and point 4 is close to area 1. Then the coating was also
`enriching gradually with Ti and C (pp.5-8). The continuous runs fabricated regular thin (about
`0,5 micron) coating on Hi-Nicalon (Fig.2e), which was similar to others above.
`The thick layers looked very porous but were hard and strong during diamond machining.
`A sufficient influence of substrate material, large surplus of C and Ti and so TiC and other
`hard phases are evident in runs both with MeHSiCI2 and HSiC13. Runs with the SiC interlayer
`(Fig.4b,c) showed the possibility to realize the coating of very close to monophase Ti~SiC2
`in a thin layer near the fiber surface. At the same time our experiments showed great
`difficulties in stable fabrication and estimation of Ti-Si-C phase with properties close to ones
`of the real Ti~SiC2 at production scale. Even little deviations in conditions increase the
`content of strong carbides and silicides decreasing Ti~SiC2 plasticity.
`
`3.4 Oxidation resistance of coated fibers
`
`Hi-Nicalon as received after the test at 1300°/54h in air had the evidence of serious.
`degradation (Fig.2f). The surface layer of about 5 micron was formed.having deep cracks and
`creasing surface The ESCA surface analysis showed only Si and O in about stoichiometry
`ratio for SiO2. The degradation of sufficient scale was observed also at 1400 °C during 20 h.
`Hi-Nicalon with coatings tested at 1300°/20h in air (Fig.2g,h,I,k) showed better saving of
`the structure. The surface has changed a little but remained regular without cracks. In the
`surface layer, which has doubled in thickness (from 0.3 to about 0.5-0,6 micron), large
`concentration of oxygen was detected. During 30-60 h this layer grew with time very slowly.
`The strength of the fibers was established in absolute meaning and related to Hi-NicalOn
`fiber as received. In all testing very large decrease of strength was established. After the BN
`coating processing the strength decreased about 21%, after Mo-Si- about 38%, after Si-B-N-C
`and Ti-Si-C - 50%. After heating in Ar at 1400°C/lh the saved strength of Hi-Nicalon was
`65, with BN - 46 and with Mo-Si, Si-B-N-C and Ti-Si-C - about 6% of initial one of Hi-
`Nicalon.
`
`4 Conclusion
`
`The experimental data gave evidence of the perceptibility of the Hi-Nicalon fibers to
`oxidizing and processing conditions. So a dense protective layer is necessary but should not
`be put directly on fiber surface. Intermediate compensatory porous layer should divide them
`to avoid strong adhesion and compensate the difference of CTE’s at definite moderate shear
`stress. This layer should not change in any way during allCMC working period.
`The CVD Mo-Si system have some serious negative points: 1) it demands too high
`temperature (about 1200°C) to form MoSi~ layer; 2) the 950-1000°C CVD Mo-Si-phase
`
`GE-1018.007
`
`
`
`156
`
`transforms in pure SiO2 after high temperature work in air, loses molybdenum and
`dramatically decreases strength of fibers. This phenomenon correlates both with the theory of
`interaction between brittle fibers and brittle layers [3], with the data on damaging fiber surface
`during CVD [4] and with the data on limited protective action of dense SiO2 layers on
`Nicalon and Hi-Nicalon [5]. As the oxidation resistance of MoSi2 is at the level of a middle
`quality SiC (Fig.l) it is rational to substitute SiC layer of good quality for protective Mo-Si
`phase. CVD SiC coating able to transform in dense SiO2 layer could be fabricated at about
`1000°C using simpler metering and blending systems.
`PIP Si-B-N-C system forms regular dense coating, which transforms into dense SiO~ layer
`with a noticeable protective properties. This differs from porous friable coating of PIP SiC
`[14]. But all these three kind of coatings, being used for tows, after the transformation into
`SiOz melt and form sticking of monofilaments. These sticks are probably a major reason for
`reducing tow strength after cooling.
`The CVD Ti-Si-C system can be fabricated as monophase Ti3SiC~ in a form of thin regular
`layer but stable processing ability is doubtful. The silicon and titanium carbides form at any
`deviations from very narrow set of conditions. These carbides enhance sufficiently the shear
`strength of material. The second negative point is a potent influence of under layers (carbon,
`for example) on composition of the main layer. It should also be noted that even at the bbst
`variant this material transforms to the mixture of TiO2 and SiOz, which melts at working
`temperatures and cannot save constantly a required adhesion level.
`During further research some special forms of stable high melting oxides as adhesion
`controlling materials and CVD SiC or PIP Si-B-N-C as protective ones will be considered.
`
`6
`
`References
`
`[1]
`[2]
`
`G. N. Morscher, J. Hurst, D. Brewer, J. Am. Ceram. Soc. 2000, 83, 1441-49.
`CERAMIC FIBERS and COATINGS, Advan. Mater. for the Twenty-First Century, Nat.
`Mater. Adv. Board, Publ. NMAB-494, p.54, Nat. Acad. Press, Washington, D.C. 1998.
`M.Kh. Shorshorov, L.M. Ustinov, A.M. Tsirlin et al., J.Mater. Sci. 1979, 14, 1850-1861.
`[4] F. Rebillat, A. Guette, L. Espitalier, R. Naslain, Key Eng. Mater. 1999, 164-165, 31-34.
`[5] T. Shimoo, F.Toyoda, K. Okamura, J. Am. Ceram. Soc. 2000, 83,1450-56.
`[6] K.L. Luthra, J. Am. Ceram. Soc.1997, 80, 3253-57.
`[7] T. Narushima, T. Goto, T. Hirai, J. Am. Ceram. Soc. 1989, 72, 1386-90.
`[8] T.A. Kircher, E.L. Courtright, Materials Sol.and Eng., 1992, A155, 67-74.
`[9] L.U. Ogbuji, E.J. Opila, J. Electrochem. Soc., 1995, 142,925-30.
`[10] H.-P. Baldus, G. Passing in Advanced Structural Fiber Composites (Ed.: P. Vincenzini)
`TECHNA, Italy, 1995,125-132.
`[11] R. Riedel, J. Bill, A. Kienzle, Appl. Organometallic Chem., 1996, 10,241-256.
`[12] H.-P. Baldus, M. Jansen, Angew. Chem. Int. Ed. Engl. 1997, 36, 328-43.
`[13] N. Patibandla, W.B. Hillig, J. Am. Ceram. Soc. 1993, 76, 1630-34.
`[14] A.M. Tsirlin, V.G. Gerlivanov, E.K. Florina, Y.E. Pronin et al. in High Temperature
`Ceramic Matrix Composites III (Eds. K. Niihara, K. Nikano, at all), Ceram. Soc. Jap.,
`Trans Tech Publications, Switzerland, Germany,1998, pp. 399-402.
`[15] E. Pickering, W.J. Lackey, S. Crain, Ceram. Eng. Sci. Proc. 1998, 19, 541-52.
`[16] C. Racault, F. Langlais, R. Naslain, Y. Kihn, J. Mater. Sci. 1994,29, 3941-48.
`
`GE-1018.008