journal of materials ELSEVIER Journal of Nuclear Materials 307-311(2002)1057-1072 www.elsevier.com/locate/jnucma Section 11. Structural ceramics and graphite Promise and challenges of Sic /Sic composites for fusion energy app. lications R.H. Jones a,, L. Giancarli A. Hasegawa, Y. Katoh d ohyama B. Riccardi. LL. Snead w.J. Weber a Pacific Northwest National Laboratory, MS P8-15, P.O. Box 999, Richland, WA 99352, US.A b CEA, Centre d Etudes de saclay, F-9119 Gif sur Yvette cedex, france Tohoku Unirersity, Aoba-ku, Sendai 980-8579, Japan d Institute of Adranced Energy, Kyoto University, Gokasho, Uji, Kyoto 611-00l1, Japan r Oak Ridge National Laboratory, Oak Ridge, TN37831,USA abstract Silicon carbide fiber/silicon carbide matrix composites have been specified in several recent fusion power plant design studies because of their high operating temperature (1000-1100oC)and hence high energy conversion efficiencies Radiation resistance of the B-phase of Sic, excellent high-temperature fracture, creep, corrosion and thermal shock resistance and safety advantages arising from low induced radioactivity and afterheat are all positive attributes favoring the selection of Sicr/SiC composites. With the promise of these materials comes a number of challenges such as their thermal conductivity, radiation stability, gaseous transmutation rates, hermetic behavior and joining technology Re- cent advances have been made in understanding radiation damage in Sic at the fundamental level through MD sim- ulations of displacement cascades. Radiation stability of composites made with the advanced fibers of Nicalon Type s and the UBE Tyranno SA, where no change in strength was observed up to 10 dpa at 800C, in the development of materials with improved thermal conductivity, modeling of thermal conductivity, joining techniques and models for ife-prediction. High transmutation rates of C and Si to form H, He, Mg, and Al continue to be a concern. c 2002 Elsevier science bv all rights reserved 1. Introduction led to their being considered in the TAURO, arIes and dream power plant designs. Sicr/Sic composites offer the promise of a high Challenges for these materials include their thermal temperature fusion reactor design because of the radia- conductivity, radiation stability, gaseous transmuta- tion resistance of the cubic, B-phase SiC matrix, their tion rates, hermetic behavior and joining technology excellent high-temperature fracture, creep, corrosion Their radiation stability is dominated by the differential and thermal shock resistance and safety advantages swelling between the Sic fibers, that are not fully dense arising from their low induced radioactivity and after- or crystalline, carbon interphases and B Sic matrices. heat. Also, developments for other applications, such as Within limits, these materials can be engineered to have aerospace, have driven improvements in material per- select properties; therefore, much of the research in formance that are beneficial to fusion applications. understanding their behavior and improving their per These positive attributes of Sic/Sic composites have formance has been focused on this aspect of their character. An overview of new understanding of the radiation behavior of Sic and SiCr/Sic composites Corresponding author. Tel +1-509 376 4276: fax:+1-509 will be given. This includes fundamentals of radiation 3760418 damage in SiC, advances in new composite materials E-mail address: rh. jones@pnl. gov (R H. Jones). experiments and modeling of thermal conductivity 0022-3115/02/. see front matter e 2002 Elsevier Science B v. All rights reserved PI:S0022-3115(02)00976-5
Section 11. Structural ceramics and graphite Promise and challenges of SiCf/SiC composites for fusion energy applications R.H. Jones a,*, L. Giancarli b , A. Hasegawa c , Y. Katoh d , A. Kohyama d , B. Riccardi e , L.L. Snead f , W.J. Weber a a Pacific Northwest National Laboratory, MS P8-15, P.O. Box 999, Richland, WA 99352, USA b CEA, Centre dEtudes de Saclay, F-9119, Gif sur Yvette cedex, France c Tohoku University, Aoba-ku, Sendai 980-8579, Japan d Institute of Advanced Energy, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan e ENEA-CR Frascati, via E. Fermi, 27, I00044 Frascati (Roma), Italy f Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA Abstract Silicon carbide fiber/silicon carbide matrix composites have been specified in several recent fusion power plant design studies because of their high operating temperature (1000–1100 C) and hence high energy conversion efficiencies. Radiation resistance of the b-phase of SiC, excellent high-temperature fracture, creep, corrosion and thermal shock resistance and safety advantages arising from low induced radioactivity and afterheat are all positive attributes favoring the selection of SiCf /SiC composites. With the promise of these materials comes a number of challenges such as their thermal conductivity, radiation stability, gaseous transmutation rates, hermetic behavior and joining technology. Recent advances have been made in understanding radiation damage in SiC at the fundamental level through MD simulations of displacement cascades. Radiation stability of composites made with the advanced fibers of Nicalon Type S and the UBE Tyranno SA, where no change in strength was observed up to 10 dpa at 800 C, in the development of materials with improved thermal conductivity, modeling of thermal conductivity, joining techniques and models for life-prediction. High transmutation rates of C and Si to form H, He, Mg, and Al continue to be a concern. 2002 Elsevier Science B.V. All rights reserved. 1. Introduction SiCf/SiC composites offer the promise of a hightemperature fusion reactor design because of the radiation resistance of the cubic, b-phase SiC matrix, their excellent high-temperature fracture, creep, corrosion and thermal shock resistance and safety advantages arising from their low induced radioactivity and afterheat. Also, developments for other applications, such as aerospace, have driven improvements in material performance that are beneficial to fusion applications. These positive attributes of SiCf/SiC composites have led to their being considered in the TAURO, ARIES and DREAM power plant designs. Challenges for these materials include their thermal conductivity, radiation stability, gaseous transmutation rates, hermetic behavior and joining technology. Their radiation stability is dominated by the differential swelling between the SiC fibers, that are not fully dense or crystalline, carbon interphases and b SiC matrices. Within limits, these materials can be engineered to have select properties; therefore, much of the research in understanding their behavior and improving their performance has been focused on this aspect of their character. An overview of new understanding of the radiation behavior of SiC and SiCf /SiC composites will be given. This includes fundamentals of radiation damage in SiC, advances in new composite materials, experiments and modeling of thermal conductivity, Journal of Nuclear Materials 307–311 (2002) 1057–1072 www.elsevier.com/locate/jnucmat * Corresponding author. Tel.: +1-509 376 4276; fax: +1-509 376 0418. E-mail address: rh.jones@pnl.gov (R.H. Jones). 0022-3115/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 3 1 1 5 ( 0 2 ) 0 0 9 7 6 - 5
RH. Jones et al. I Journal of Nuclear Materials 307-311(2002)1057-1072 ransmutation rates, chemical compatibility a which are an anticipation of successful future R&D. The sion, irradiation creep and crack growth, join most significant assumptions are the following ology and thermal and mechanical transient Recent experimental results and mechanistic and mod- (i The Sic/Sic thermal conductivity at 1000C and materials. predictions show promise of new, improved at end-of-life(EOL)conditions is 20 W/m K. This value is considerably higher than that shown by present-day SiC/SiC. In fact, ent available data on existing industrial 3D Sicr/SiC indicate 2. Current design possibilities and needs at 1000C a value of 15 W/m K in the plane and of 9 W/m K through the thickness [1], without tak- Silicon carbide composites (SiCr/SiC), are being ing into account the effect of irradiation which are considered in future fusion power reactors because their expected to decrease by about a factor 3 the out high temperature properties (=1000C), offer the po of-pile value tential of very high energy conversion efficiency (50% or (in The maximum and minimum acceptable tempera- more). SiCr/SiC composite has been proposed as struc- tures are respectively 1100C and 600C; these tural material for the first wall and blanket in several values need to be confirmed under irradiation for conceptual design studies EOL conditions (iii)Compatibility between Pb-17Li and Sicr/Sic is ac- ceptable at 800C; this statement should be valid 2.1. Proposed blanket concepts after irradiation and at Pb-17Li velocity of few m/s and should also be valid for any brazing mate- The most recent proposals are TAURO in the Eu- ropean Union, ARIES-AT in the United States, and rials in contact with Pb-17Li: available dat firm a good compatibility for static Pb-17Li at DREAM in Japan. The first two concepts are Pb-17Li 800oC for 3000 h self-cooled blankets, while Dream is cooled by 10 MPa (iv) Use of preliminary SicrSiC models and design Helium [1]. Both TAURO and Aries-AT blankets are criteria are not yet validated by experiments: mod- essentially formed by a SiC/Sic box with indirectly els and criteria currently used for metals and de- cooled Fw that acts as a container for the pb-17Li fined in industrial design codes (.g, ASME, which has the simultaneous functions of coolant tritium RCC-MR) are not applicable for Sic/sic struc- breeder, neutron multiplier and, finally, tritium carrier tures Because of the relatively low SiC/sic electrical con (v) The electrical conductivity of Sic /Sic is about 500 ductivity, high Pb-17Li velocity is allowed without Q2m this value would allow sufficiently low needing large coolant pressures (<1.5 MPa). TAURO MHD effects for self-cooled Pb-17Li blanket and blanket is characterized by 2m-high single modules correspond to the presently measured out-of-pile which are reinforced by SiCr-SiC stiffeners ARIES-AT data. This result could, however, be jeopardized is characterized by a coaxial Pb-17Li flow, which occurs by Pb-17Li infiltration in the top layer of SiC: in two 8 m-high boxes inserted one into the other. The SiC: this infiltration could dramatically increase DREAM blanket is characterized by smaller modul the wall electrical conductivity which could quickly (0.5 m of height), each divided in three zones: FW reeding zone and shield; neutron multiplier material become unacceptably high. A SiC coating on SiCr/ Sic is probably sufficient to avoid this kind of effect (Be), tritium breeding material (LiO or other lithium (vi) Acceptably low coolant leakage in case of He-cool- ceramics)and shielding material(Sic)are packed in the g(10 MPa of pressure); very low quantity of He module as small size pebbles of I mm-diameter for Be tolerated in the plasma so SiCr/Sic hermeticity and Li2O, and 10 mm for SiC. The He coolant path need to be ensured by a reliable coating which includes a flow through the pebble beds and a porous should have the same irradiation resistance of the partition wall. These blankets allow very high coolant main structures; no experimental results are ye outlet temperatures and therefore a high energy con version efficiency. The maximum coolant outlet tem- (vii) Possibility of manufacturing relevant shapes with perature is 1100C obtained in the Pb-17Li of the appropriate thickness ranging between I and 6 ARIES-AT blanket which lead to a thermal efficiency of mm. Present requirements appear achievable in 58.5% present day industrial composites; however, mate- rial properties in these conditions need to be exper 2.2. Main assumptions for blanket de imentally verified. vili)Existing methods of joining finite components The TAURO. ARIEs-AT and dream designs have with characteristics similar to the base material been performed g optimistic SiCr/Sic properties good results are already available
transmutation rates, chemical compatibility and corrosion, irradiation creep and crack growth, joining technology and thermal and mechanical transient behavior. Recent experimental results and mechanistic and modeling based predictions show promise of new, improved materials. 2. Current design possibilities and needs Silicon carbide composites (SiCf /SiC), are being considered in future fusion power reactors because their high temperature properties (’1000 C), offer the potential of very high energy conversion efficiency (50% or more). SiCf /SiC composite has been proposed as structural material for the first wall and blanket in several conceptual design studies. 2.1. Proposedblanket concepts The most recent proposals are TAURO in the European Union, ARIES-AT in the United States, and DREAM in Japan. The first two concepts are Pb–17Li self-cooled blankets, while DREAM is cooled by 10 MPa Helium [1]. Both TAURO and ARIES-AT blankets are essentially formed by a SiC/SiC box with indirectlycooled FW that acts as a container for the Pb–17Li which has the simultaneous functions of coolant, tritium breeder, neutron multiplier and, finally, tritium carrier. Because of the relatively low SiCf /SiC electrical conductivity, high Pb–17Li velocity is allowed without needing large coolant pressures (<1.5 MPa). TAURO blanket is characterized by 2m-high single modules which are reinforced by SiCf–SiC stiffeners. ARIES-AT is characterized by a coaxial Pb–17Li flow, which occurs in two 8 m-high boxes inserted one into the other. The DREAM blanket is characterized by smaller modules (0.5 m of height), each divided in three zones: FW, breeding zone and shield; neutron multiplier material (Be), tritium breeding material (Li2O or other lithium ceramics) and shielding material (SiC) are packed in the module as small size pebbles of 1 mm-diameter for Be and Li2O, and 10 mm for SiC. The He coolant path includes a flow through the pebble beds and a porous partition wall. These blankets allow very high coolant outlet temperatures and therefore a high energy conversion efficiency. The maximum coolant outlet temperature is 1100 C obtained in the Pb–17Li of the ARIES-AT blanket which lead to a thermal efficiency of 58.5%. 2.2. Main assumptions for blanket designs The TAURO, ARIES-AT and DREAM designs have been performed assuming optimistic SiCf /SiC properties which are an anticipation of successful future R&D. The most significant assumptions are the following: iiii(i) The SiCf /SiC thermal conductivity at 1000 C and at end-of-life (EOL) conditions is 20 W/m K. This value is considerably higher than that shown by present-day SiCf/SiC. In fact, present available data on existing industrial 3D SiCf/SiC indicate at 1000 C a value of 15 W/m K in the plane and of 9 W/m K through the thickness [1], without taking into account the effect of irradiation which are expected to decrease by about a factor 3 the outof-pile value. iii(ii) The maximum and minimum acceptable temperatures are respectively 1100 C and 600 C; these values need to be confirmed under irradiation for EOL conditions. ii(iii) Compatibility between Pb–17Li and SiCf/SiC is acceptable at 800 C; this statement should be valid after irradiation and at Pb–17Li velocity of few m/s and should also be valid for any brazing materials in contact with Pb–17Li; available data con- firm a good compatibility for static Pb–17Li at 800 C for 3000 h. ii(iv) Use of preliminary SiCf /SiC models and design criteria are not yet validated by experiments; models and criteria currently used for metals and de- fined in industrial design codes (e.g., ASME, RCC-MR) are not applicable for SiCf /SiC structures. iii(v) The electrical conductivity of SiCf /SiC is about 500 X1 m1; this value would allow sufficiently low MHD effects for self-cooled Pb–17Li blanket and correspond to the presently measured out-of-pile data. This result could, however, be jeopardized by Pb–17Li infiltration in the top layer of SiCf/ SiC; this infiltration could dramatically increase the wall electrical conductivity which could quickly become unacceptably high. A SiC coating on SiCf / SiC is probably sufficient to avoid this kind of effect. ii(vi) Acceptably low coolant leakage in case of He-cooling (10 MPa of pressure); very low quantity of He is tolerated in the plasma so SiCf /SiC hermeticity need to be ensured by a reliable coating which should have the same irradiation resistance of the main structures; no experimental results are yet available to give indication about this requirement. i(vii) Possibility of manufacturing relevant shapes with appropriate thickness ranging between 1 and 6 mm. Present requirements appear achievable in present day industrial composites; however, material properties in these conditions need to be experimentally verified. (viii) Existing methods of joining finite components with characteristics similar to the base material; good results are already available. 1058 R.H. Jones et al. / Journal of Nuclear Materials 307–311 (2002) 1057–1072
RH. Jones et al. I Journal of Nuclear Materials 307-311(2002)1057-1072 59 x) Acceptable structure lifetime in terms of sumed as the maximum allowable compressive stress fluence and plasma-first wall interaction;a since no damage is observed under compression. The cant experimental campaign is required to number of fibers through the thickness of the composite the sice/Sic limits is usually lower and their arrangement different. So, if one can accept the uncoupling of stresses in plane and 2.3. SiC /SiC thermo-mechanical model and design crite stresses through the thickness, the former can be eva rId uated using the von mises criteria, while the latter corresponds to the measured rupture value. Taking into The proposal for using SiC/SiC as structural mate- account the above remarks, for compressive stresses the rial for a nuclear component with a reasonably long limit is the rupture limits while for tensile stresses the lifetime is fairly recent, therefore no adequate modeling limit is the elastic limit. a margin of about 20% may be and design criteria are available yet. Some preliminary allowable on the elastic limit. work has recently been performed [2] aiming both to For example, in the case of the TaURo blanket identify appropriate models available in the aerospace based on the CERaSEP@ composites produced by research field and to theoretically define sound design SNECMA, the following limits have been assumed criteria to improve the design thermo-mechanical ana for normal stresses through the thickness, 110 MPa for tensile stresses(roughly corresponding to the ma- 2.3.1. Modeling trix tensile resistance limit outside the composite)and SiCr/Sic composites exhibit a complex nonlinear 420 MPa for compressive stresses(rupture limit for behavior combining brittle damage, residual strains CERASEP N2-1) and opening-closing of microcracks. These composites for shear stresses through the thickness, 44 MPa(as- present different properties, and therefore different sumed rupture limit, to be confirmed ) strengths, for different loading directions; moreover, for stresses in plane, 145 MPa for tensile stresses(be- tensile and compression strengths are very different. ginning of fiber/matrix debonding) and 580 MPa for Under loading, the interaction of fibers and matrix lead compressive stresses(rupture limit measured on the at first to matrix microcracking. then to matrix/fiber CERASEP N2-1). It appears clear that this design decohesion, followed by opening of the microcrack and criteria proposal is very preliminary; it needs to finally to fiber failure. This sequence corresponds to an further evaluated both theoretically and, more im- initial isotropic behavior in plane and then to a crack portant, experimentally through a systematic specific growth perpendicular to the fibers depending on the load direction. A relatively simple model, able to take into account such a behavior and based on continuum damage mechanics which consider the composites as a 3. Promise of Sicr/Sic composites continuous media, has been implemented in the FEM code CASTEM. Significant improvements on the results Composite materials made from continuous fibers of have been obtained when compared with models used SiC, can be woven into several variant fabric architec- for metals [1]. On the other end, damage description has tures and the matrix formed with a variety of infiltration been limited to scalar variables that is appropriate when methods. The se of Sic has been shown by damage is oriented in the fiber direction but not satis- merous studies [l] to have a saturation swelling value of factory when damage is loading oriented. A substantial about 0.1-0.2% at 800-1000oC. This suggests that effort is still required to develop and implement this for a composites of Sicr/Sic have the potential for excellent radiation stability. The continuous fiber architecture coupled with engineered interfaces between the fiber and 2.3.2. Design criteria matrix, provide excellent fracture properties and frac- To avoid degradation of the composite physical ture toughness values on the order of 25 MPam /.The properties, the elastic limit must be used as the maxi- strength and fracture toughness are independent of mum allowable stress. On the other hand, one of the temperature up to the limit of the fiber stability. With most attractive characteristics of SiCr/Sic composites is aprovements in fiber stability these materials exhibit that they are damage-tolerant, that is, they are capable. excellent mechanical properties to at least 1200C. Als of accommodating a high degree of deformation because these fiber/matrix microstructures impart excellent of crack arrest phenomena driven by the interface be- thermal shock and thermal fatigue resistance to these ween fibers and matrix. In principle, it can be assumed materials so plasma discharge and start-up and shut the limit for matrix microcracking saturation(beginning lown cycles should not induce significant structural of fiber/matrix debonding) as the maximum allowable damage In oxygen bearing environments, Sic will form tensile stress. The actual failure limit can instead be as- a protective layer of Sio, that greatly retards further
ii(ix) Acceptable structure lifetime in terms of neutron fluence and plasma-first wall interaction; a signifi- cant experimental campaign is required to define the SiCf/SiC limits. 2.3. SiCf /SiC thermo-mechanical model and design criteria The proposal for using SiCf /SiC as structural material for a nuclear component with a reasonably long lifetime is fairly recent, therefore no adequate modeling and design criteria are available yet. Some preliminary work has recently been performed [2] aiming both to identify appropriate models available in the aerospace research field and to theoretically define sound design criteria to improve the design thermo-mechanical analyses. 2.3.1. Modeling SiCf/SiC composites exhibit a complex nonlinear behavior combining brittle damage, residual strains and opening-closing of microcracks. These composites present different properties, and therefore different strengths, for different loading directions; moreover, tensile and compression strengths are very different. Under loading, the interaction of fibers and matrix lead at first to matrix microcracking, then to matrix/fiber decohesion, followed by opening of the microcrack and finally to fiber failure. This sequence corresponds to an initial isotropic behavior in plane and then to a crack growth perpendicular to the fibers depending on the load direction. A relatively simple model, able to take into account such a behavior and based on continuum damage mechanics which consider the composites as a continuous media, has been implemented in the FEM code CASTEM. Significant improvements on the results have been obtained when compared with models used for metals [1]. On the other end, damage description has been limited to scalar variables that is appropriate when damage is oriented in the fiber direction but not satisfactory when damage is loading oriented. A substantial effort is still required to develop and implement this for a complete design model. 2.3.2. Design criteria To avoid degradation of the composite physical properties, the elastic limit must be used as the maximum allowable stress. On the other hand, one of the most attractive characteristics of SiCf /SiC composites is that they are damage-tolerant, that is, they are capable of accommodating a high degree of deformation because of crack arrest phenomena driven by the interface between fibers and matrix. In principle, it can be assumed the limit for matrix microcracking saturation (beginning of fiber/matrix debonding) as the maximum allowable tensile stress. The actual failure limit can instead be assumed as the maximum allowable compressive stress since no damage is observed under compression. The number of fibers through the thickness of the composite is usually lower and their arrangement different. So, if one can accept the uncoupling of stresses in plane and stresses through the thickness, the former can be evaluated using the Von Mises criteria, while the latter corresponds to the measured rupture value. Taking into account the above remarks, for compressive stresses the limit is the rupture limits while for tensile stresses the limit is the elastic limit. A margin of about 20% may be allowable on the elastic limit. For example, in the case of the TAURO blanket, based on the CERASEP composites produced by SNECMA, the following limits have been assumed: • for normal stresses through the thickness, 110 MPa for tensile stresses (roughly corresponding to the matrix tensile resistance limit outside the composite) and 420 MPa for compressive stresses (rupture limit for CERASEP N2-1); • for shear stresses through the thickness, 44 MPa (assumed rupture limit, to be confirmed); • for stresses in plane, 145 MPa for tensile stresses (beginning of fiber/matrix debonding) and 580 MPa for compressive stresses (rupture limit measured on the CERASEP N2-1). It appears clear that this design criteria proposal is very preliminary; it needs to be further evaluated both theoretically and, more important, experimentally through a systematic specific experimental campaign. 3. Promise of SiCf/SiC composites Composite materials made from continuous fibers of SiC, can be woven into several variant fabric architectures and the matrix formed with a variety of infiltration methods. The b-phase of SiC has been shown by numerous studies [1] to have a saturation swelling value of about 0.1–0.2% at 800–1000 C. This suggests that composites of SiCf /SiC have the potential for excellent radiation stability. The continuous fiber architecture, coupled with engineered interfaces between the fiber and matrix, provide excellent fracture properties and fracture toughness values on the order of 25 MPa m1=2. The strength and fracture toughness are independent of temperature up to the limit of the fiber stability. With improvements in fiber stability these materials exhibit excellent mechanical properties to at least 1200 C. Also, these fiber/matrix microstructures impart excellent thermal shock and thermal fatigue resistance to these materials so plasma discharge and start-up and shutdown cycles should not induce significant structural damage. In oxygen bearing environments, SiC will form a protective layer of SiO2 that greatly retards further R.H. Jones et al. / Journal of Nuclear Materials 307–311 (2002) 1057–1072 1059
RH. Jones et al. I Journal of Nuclear Materials 307-311(2002)1057-1072 oxidation. Therefore, SiCrSic composites have the po Table 2 tential for excellent oxidation resistance in He +O, en- Defect formation energies in 3C-SiC [4] Defect type Formation energy (el 4. Challenges of engineering the properties of Sic / Sic Given the positive attributes of Sic /SiC composites, they would be the obvious first choice for structural applications in fusion energy systems if there were no issues in their use. However there are some unresolved 71 issues associated with their use as outlined in table l Considerable progress has been made in understanding these issues and in some cases improvements have beer C-Si(100) made. Since the properties of these composites are 9.32 engineerable, there is the potential, to some extent,to (110 engineer around these issues. The purpose of this paper is to summarize some of the understanding and im provements made in these materials suggest that the c interstitials should migrate from sublattice to sublattice. The most stable configuration for Si interstitials is in a tetrahedrally-coordinated in- 5. Fundamentals of radiation damage in SiC terstitial site surrounded by c atoms on the C sublattice Ongoing theoretical and computational 5.1. Defect formation energi migration of these stable defect configurations will yield the necessary parameters to model radiation damage Density functional theory dFT), based on the processes at higher temperatures and over longer time pseudopotential plane-wave method within the frame- scales in SiC using rate-theory approaches or kinetic york of the local density approximation (LDA), has Monte Carlo methods been used to study the formation and properties of na tive defects in 3C-SiC (cubic SiC), as described in detail 5.2. Damage production and accumulation elsewhere [3-5]. The formation energies for vacancie antisite defects and interstitials in 3C-SiC are summa Molecular dynamics(MD) simulations of displace rized in Table 2. Two types of vacancies form, namely C nent cascades and cascade overlap events have been and si vacancies. In addition, two types of antisite de performed using a modified version of the code mOl fects are formed by atoms located on the wrong sub- DY [6]. with either constant volume or constant pressure lattice. For interstitial defects, there are ten possible and periodic boundary conditions. Details of the md configurations, four tetrahedral and six dumbbell(split) simulations and interatomic potentials employed are configurations. It is found that the most stable config urations for C interstitials are C-C and C-Si split in The short cascade lifetime in SiC is illustrated in Fig. terstitials along the(100)and(110) directions, which 1, where the numbers of interstitials and antisite defects produced in a 10 keV Si cascade are shown as a function Table I of time. The number of interstitials and vacancies(not Critical issues associated with the use of SiC SiC composites in shown) reaches a peak at about 0. 1 ps and then de- nuclear environments creases due to defect recombination [7, 9, 10]. The defect Priman Secondary concentrations attain steady state values after about 0.4 Thermal conductivity Chemical compatibility(He) ps. The cascade lifetime has been found to be slightly Radiation stability Carbon interfaces longer(about 0.7 ps)for a 50 keV Si PKa in SiC [7, 91 Fibers-polymer derived Thermal fatigue and shock These lifetimes are about an order of magnitude smaller Lack of a database than the values reported for metals using similar PKA + Matrices-CVI and polymer Long-term thermal stability energies [6, 11]. impregnated Design codes The results from MD simulations [10 for the ne Transmutations displacements and antisite defects produced by a 10 keV Hermetic behavior Si primary knock-on atom(PKA) are shown in Fig. 2 as Joining technology a function of PKA energy. The number of net dis- Chemical compatibility placements is defined as the sum of the total number of interstitials (or vacancies) and antisite defects. The
oxidation. Therefore, SiCf/SiC composites have the potential for excellent oxidation resistance in He þ O2 environments. 4. Challenges of engineering the properties of SiCf/SiC Given the positive attributes of SiCf/SiC composites, they would be the obvious first choice for structural applications in fusion energy systems if there were no issues in their use. However, there are some unresolved issues associated with their use as outlined in Table 1. Considerable progress has been made in understanding these issues and in some cases improvements have been made. Since the properties of these composites are engineerable, there is the potential, to some extent, to engineer around these issues. The purpose of this paper is to summarize some of the understanding and improvements made in these materials. 5. Fundamentals of radiation damage in SiC 5.1. Defect formation energies Density functional theory (DFT), based on the pseudopotential plane-wave method within the framework of the local density approximation (LDA), has been used to study the formation and properties of native defects in 3C–SiC (cubic SiC), as described in detail elsewhere [3–5]. The formation energies for vacancies, antisite defects and interstitials in 3C–SiC are summarized in Table 2. Two types of vacancies form, namely C and Si vacancies. In addition, two types of antisite defects are formed by atoms located on the wrong sublattice. For interstitial defects, there are ten possible configurations, four tetrahedral and six dumbbell (split) configurations. It is found that the most stable configurations for C interstitials are C–C and C–Si split interstitials along the h100i and h110i directions, which suggest that the C interstitials should migrate from sublattice to sublattice. The most stable configuration for Si interstitials is in a tetrahedrally-coordinated interstitial site surrounded by C atoms on the C sublattice. Ongoing theoretical and computational studies of the migration of these stable defect configurations will yield the necessary parameters to model radiation damage processes at higher temperatures and over longer time scales in SiC using rate-theory approaches or kinetic Monte Carlo methods. 5.2. Damage production and accumulation Molecular dynamics (MD) simulations of displacement cascades and cascade overlap events have been performed using a modified version of the code MOLDY [6], with either constant volume or constant pressure and periodic boundary conditions. Details of the MD simulations and interatomic potentials employed are described elsewhere [7–10]. The short cascade lifetime in SiC is illustrated in Fig. 1, where the numbers of interstitials and antisite defects produced in a 10 keV Si cascade are shown as a function of time. The number of interstitials and vacancies (not shown) reaches a peak at about 0.1 ps and then decreases due to defect recombination [7,9,10]. The defect concentrations attain steady state values after about 0.4 ps. The cascade lifetime has been found to be slightly longer (about 0.7 ps) for a 50 keV Si PKA in SiC [7,9]. These lifetimes are about an order of magnitude smaller than the values reported for metals using similar PKA energies [6,11]. The results from MD simulations [10] for the net displacements and antisite defects produced by a 10 keV Si primary knock-on atom (PKA) are shown in Fig. 2 as a function of PKA energy. The number of net displacements is defined as the sum of the total number of interstitials (or vacancies) and antisite defects. The Table 1 Critical issues associated with the use of SiCf /SiC composites in nuclear environments Primany issues Secondary issues Thermal conductivity Chemical compatibility (He) Radiation stability þ Carbon interfaces þ Fibers-polymer derived Thermal fatigue and shock Interphases-C, porous Lack of a database þMatrices-CVI and polymer impregnated Long-term thermal stability Design codes Transmutations Hermetic behavior Joining technology Chemical compatibility (Pb–Li) Table 2 Defect formation energies in 3C–SiC [4] Defect type Formation energy (eV) Vc 5.48 VSi 6.64 CSi 1.32 SiC 7.20 CTC 6.41 CTS 5.84 SiTC 6.17 SiTS 8.71 Cþ–Sih100i 3.59 Cþ–Ch100i 3.16 C–Siþh100i 10.05 Siþ–Sih100i 9.32 Cþ–Ch110i 3.32 Cþ–Sih110i 3.28 1060 R.H. Jones et al. / Journal of Nuclear Materials 307–311 (2002) 1057–1072
RH. Jones et al. /Jounal of Nuclear Materials 307-311(2002)1057-1072 10 keV Si PKA Ct9 60 40 o Antisite Defects 0.001 0.01 Time( ps) Fig. 1. Number of interstitials and antisite defects produced in a 10 keV Si cascade as a function of time [10] 98 3C-SiC Net Displacements Fig. 3. MD simulation of primary damage state in SiC at 300K Si PKA Damage Energy(kev) due to a 10 keV Si PKA 8. The Si and c defects are dark and light gray, respectively, and the interstitials, antisite defects and Fig. 2. Net displacements and antisite defects produced as a vacancies are given by large, medium, and small spheres, unction of Si PKA damage energy [10 spectively number of C displacements is much larger than the (dpa)for irradiation with 550 keV Si* ions at 190 K[12 number of Si displacements, which is consistent with 14. The solid curve (Fig. 4) is based on the direct recent experimental observations [5]. Similar behavior is impact/defect-stimulated model for amorphization [15]. observed for C PKAs. Antisite defects are produced by where point defects, such as interstitials and antisite nearest-neighbor replacements during the collisional defects, stimulate the growth of amorphous nuclei (or phase and some random interstitial-vacancy recombi- defect clusters) produced directly in a displacement nation during the subsequent relaxation phase cascade. As the dose increases, cascade superposition MD simulations, as illustrated in Fig. 3, have also and defect-stimulated growth at crystalline-amorphous shown that Si PKAs generate only small interstitial interfaces become more probable. The relative ratio of clusters, with most defects being isolated single inter- direct- impact and defect-stimulated cross sections from stitials and vacancies distributed over a large region the model fit to the data for Si are consistent with those [8, 12, 13]. These predictions are in agreement with the derived from the MD simulations based on relative interpretation of the experimental results on disordering cluster distributions [121 behavior in SiC, as shown in Fig. 4, where the relative MD methods with 10 key si pkas have been em- order on the Si sublattice in Sic at the damage peak ployed to simulate cascade overlap, damage accumula shown as a function of dose in displacements per atom tion and amorphization processes in 3C-SiC. In this
number of C displacements is much larger than the number of Si displacements, which is consistent with recent experimental observations [5]. Similar behavior is observed for C PKAs. Antisite defects are produced by nearest-neighbor replacements during the collisional phase and some random interstitial-vacancy recombination during the subsequent relaxation phase. MD simulations, as illustrated in Fig. 3, have also shown that Si PKAs generate only small interstitial clusters, with most defects being isolated single interstitials and vacancies distributed over a large region [8,12,13]. These predictions are in agreement with the interpretation of the experimental results on disordering behavior in SiC, as shown in Fig. 4, where the relative disorder on the Si sublattice in SiC at the damage peak is shown as a function of dose in displacements per atom (dpa) for irradiation with 550 keV Siþ ions at 190 K [12– 14]. The solid curve (Fig. 4) is based on the directimpact/defect-stimulated model for amorphization [15], where point defects, such as interstitials and antisite defects, stimulate the growth of amorphous nuclei (or defect clusters) produced directly in a displacement cascade. As the dose increases, cascade superposition and defect-stimulated growth at crystalline-amorphous interfaces become more probable. The relative ratio of direct-impact and defect-stimulated cross sections from the model fit to the data for Si are consistent with those derived from the MD simulations based on relative cluster distributions [12]. MD methods with 10 keV Si PKAs have been employed to simulate cascade overlap, damage accumulation and amorphization processes in 3C–SiC. In this Fig. 3. MD simulation of primary damage state in SiC at 300 K due to a 10 keV Si PKA [8]. The Si and C defects are dark and light gray, respectively, and the interstitials, antisite defects and vacancies are given by large, medium, and small spheres, respectively. Fig. 1. Number of interstitials and antisite defects produced in a 10 keV Si cascade as a function of time [10]. Fig. 2. Net displacements and antisite defects produced as a function of Si PKA damage energy [10]. R.H. Jones et al. / Journal of Nuclear Materials 307–311 (2002) 1057–1072 1061