Compressive creep of SiC whisker/Ti
 3
 SiC
 2
 composites at high temperature in air

MAX phases are a class of ternary carbides or nitride ceramic materials characterized by nanolaminated layers.1 The uniqueness of MAX phases lies in the fact that they can be easily machined and they are categorized as ceramic materials with the advantage of low density unlike metals and good damage tolerance unlike other conventional ceramics.2 Ti3SiC2 is the most studied MAX phase because of its superior mechanical properties, both at room and high temperature, compared to other MAX phases.3-6 Since Ti3SiC2 is stable at high temperature (<1000°C in air, <1600°C in vacuum), it is a potential candidate for applications where thermomechanical stability is required such as parts of the turbo jet engines.7 The availability of Ti3SiC2 powders has limited the investigation on ceramic matrix composites (CMCs) with Ti3SiC2 or any other MAX phase matrix, however, Lenz et al fabricated Cf/Ti3SiC2 composite by reactive melt infiltration of Si into Cf preforms impregnated with a mix of phenolic resin and TiC particles.8 Mu et al reported an increase in flexural strength with the addition of only 9 wt% of Ti3SiC2 particles as filler in SiCf/SiC composites. 9 Yin et al manufactured C-SiC-Ti3SiC2 hybrid composites by the combination of TiC Received: 2 April 2020 | Revised: 3 June 2020 | Accepted: 3 June 2020 DOI: 10.1111/jace.17323


| INTRODUCTION
MAX phases are a class of ternary carbides or nitride ceramic materials characterized by nanolaminated layers. 1 The uniqueness of MAX phases lies in the fact that they can be easily machined and they are categorized as ceramic materials with the advantage of low density unlike metals and good damage tolerance unlike other conventional ceramics. 2 Ti 3 SiC 2 is the most studied MAX phase because of its superior mechanical properties, both at room and high temperature, compared to other MAX phases. [3][4][5][6] Since Ti 3 SiC 2 is stable at high temperature (<1000°C in air, <1600°C in vacuum), it is a potential candidate for applications where thermomechanical stability is required such as parts of the turbo jet engines. 7 The availability of Ti 3 SiC 2 powders has limited the investigation on ceramic matrix composites (CMCs) with Ti 3 SiC 2 or any other MAX phase matrix, however, Lenz et al fabricated C f /Ti 3 SiC 2 composite by reactive melt infiltration of Si into C f preforms impregnated with a mix of phenolic resin and TiC particles. 8 Mu et al reported an increase in flexural strength with the addition of only 9 wt% of Ti 3 SiC 2 particles as filler in SiC f /SiC composites. 9 Yin et al manufactured C-SiC-Ti 3 SiC 2 hybrid composites by the combination of TiC | 5953 DASH et Al. slurry and liquid silicon infiltration. 10 The final composite had 4 vol% of Ti 3 SiC 2 which resulted in the improvement of both flexural strength and fracture toughness. Lagos et al fabricated Ti 3 SiC 2 -based CMCs by densification with spark plasma sintering (SPS) of mixed chopped carbon fiber and Ti 3 SiC 2 powder. 11 There was an interfacial reaction between carbon fiber and Ti 3 SiC 2 which might lead to the degradation of mechanical properties.
It was reported by Spencer et al that Ti 3 SiC 2 does not react with SiC fibers even at higher temperatures of 1550°C and, hence, SiC w was selected as the reinforcement in the present work. 12 The inertness of SiC with respect to Ti 3 SiC 2 provides the possibility of SiC w /Ti 3 SiC 2 CMCs. Gonzalez et al fabricated SiC fiber containing CMCs with Cr 2 AlC and Ti 2 AlC MAX phases as the matrix. 13,14 A reaction was observed in the interface of Ti 2 AlC and SiC fiber, whereas no reaction was observed in Cr 2 AlC-SiC fiber interface. Considering the better creep resistance of Ti 3 SiC 2 reported in literature as compared to other MAX phases, Ti 3 SiC 2 was used as the matrix in the present study. Moreover, SiC particles were also used to improve the tribological properties of Ti 3 SiC 2 . 15 Wan et al improved the fracture toughness of Ti 3 SiC 2 by the addition of 30 vol% of SiC particles. 16 Zhou et al and Li et al used SiC reinforcement to improve the oxidation resistance of Ti 3 SiC 2 at high temperature and these results are indicative of better oxidation resistance of the composites fabricated in the present work. 17,18 Recently, we also explored the effect of short SiC fibers 1 mm on the creep behavior of Ti 3 SiC 2 . 19 The evaluation of mechanical properties of MAX phasebased composites is majorly done at room temperature. 20,21 The high-temperature mechanical property evaluation of short-fiber reinforced MAX phase has not been reported up to now and, hence, this work presents the evaluation of compressive creep of SiC whisker (SiC w ) reinforced Ti 3 SiC 2 MAX phase. For the usage of components at a high temperature for a prolonged time under a constant stress they should have good creep resistance at the operation temperature. Creep resistance is an important aspect, which needs to be considered for thermomechanical applications. 22 The mechanical properties like flexural strength (room and high temperature) 20 and tribological behavior 23 of MAX phase are well documented and reported, however, creep investigations of MAX phase materials are very limited. 4,5,[24][25][26] Radovic et al measured the tensile creep of both fine-and coarsegrained Ti 3 SiC 2 in the temperature range 1000-1200°C. The coarse-grained Ti 3 SiC 2 had higher creep rate since large grains are more prone to delamination and kink-band formation thereby absorbing the stresses. The activation energies (~460 kJ/mol) and stress exponent (1.5-2) for creep were similar for both coarse-and fine-grained Ti 3 SiC 2 and dislocation creep was the proposed dominant mechanism. 5,24 The overall creep response was attributed to the formation and relaxation of internal stresses. Zhen et al investigated the creep response of fine-and coarse-grained Ti 3 SiC 2 in compressive mode. 4 Although the creep rate was found to be a magnitude lower than that in the tension mode, the creep mechanism was attributed to dislocation with the possibility of subcritical crack growth and grain boundary sliding (GBS) because of similar stress exponent values as that in tension mode.
Since the deformation of Ti 3 SiC 2 is only based on basal slip both at room and high temperature, the understanding of the creep behavior of Ti 3 SiC 2 reinforced with SiC w will provide more insights about thermomechanical properties of MAX phase-based CMCs, hence, being the focus of the current work. In the present work, we have used SiC w as the reinforcing phase in Ti 3 SiC 2 MAX phase matrix. The motivation of using Ti 3 SiC 2 as the matrix material is the pseudoductility of the same provided by a combination of mechanisms like kinking, shear band, and basal slip. 27

| EXPERIMENTAL PROCEDURE
Ti 3 SiC 2 MAX phase was synthesized by starting with elemental reactants (Ti, Si, Al, and C) and following a process named as molten shielded salt synthesis in air. The details of the synthesis process are given elsewhere. 28 Briefly, the elemental precursors were mixed in stoichiometric ratio with equal weight of KBr and further encapsulated with KBr to form a gas-tight cladding. Then, the pellets were placed in a KBr salt bed for high-temperature processing in air without the need of protective atmosphere. The molten KBr at high temperature serves as a barrier between the material and ambient atmosphere. The synthesis of Ti 3 SiC 2 was done at 1250°C for 1 hour, followed by cooling and washing the KBr content to obtain free flowing powder. The resultant powder was agglomerated hence it was milled on a roller bench in a slurry form for 24 hours with zirconia balls (3 mm diameter) in ethanol medium to break the loosely bound agglomerates. The milled slurry was dried in a rotary evaporator at 60°C to obtain soft agglomerates. The dried powder was then passed through a 300 µm sieve to obtain homogeneously agglomerated particles of Ti 3 SiC 2 MAX phase.
As-received β-SiC w (length: 10 µm, diameter of 1 µm; Haydale Technologies) were processed to remove the metal catalyst used for the growth of whiskers. Ten grams of SiC w was mixed with 100 mL of 0.1 mol/L HNO 3 with continuous magnetic stirring at a temperature of 70°C for 1 hour. The suspension of SiC w in diluted HNO 3 was filtered and repeatedly washed with de-ionized water to confirm the removal of added HNO 3 . Acid-treated SiC w was finally rinsed with ethanol and dried at 60°C to avoid agglomeration. The characterization of starting materials for the fabrication of CMC was done in a scanning electron microscope (SEM; Zeiss Ultra55).

DASH et Al.
The synthesized Ti 3 SiC 2 was mixed with 10 and 20 vol% of SiC w in ethanol with high shear mixer in an ultrasonic bath for 1 hour immediately followed by drying in a rotary evaporator to avoid segregation of phases due to density differences. The dried powder mix of Ti 3 SiC 2 and SiC w was loaded into a graphite die (inner diameter of 20 mm) with a graphite foil. The weight of powder was so calculated to have a sintered block with 20 mm as diameter and 10 mm as height upon full densification by SPS (HPD5; FCT System). The graphite die with the powder was heated at a rate of 100°C/min with a holding time of 35 minutes and 80 MPa at the peak temperature of 1250°C followed by rapid cooling. Additionally, a blank specimen based on monolithic Ti 3 SiC 2 was sintered under the same conditions to be compared with the composites. The density was measured by Archimedes method of water displacement. The sintered dense blocks were sand blasted to remove the graphite foil used to separate the contact between MAX phase and graphite die wall. Further the sintered block was ground plane-parallel and cylinders with a diameter of 5 mm and a height of 10 mm were obtained by electrodischarge machining. Three different blocks were sintered with monolithic Ti 3 SiC 2 , and 10 and 20 vol% of SiC w which is referred to as TSC, TSC-10SiC w , and TSC-20SiC w , respectively. The fabricated CMCs were analyzed for identifying crystalline phases by X-ray diffraction (XRD; D4 Bruker). The quantitative analysis from XRD data was obtained by Rietveld analysis using Topas software. The microstructure of the composites was observed in a SEM (Zeiss Ultra55).
The creep tests followed testing procedures reported in the work by, for example, Stournari et al. 29 The creep testing was done in air at temperatures of 1100, 1200, and 1300°C using constant loads corresponding to stress levels of 20-120 MPa. The creep data were not analyzed below 1100°C since the deformation was too small (1 µm/d) and dominated by primary creep. The samples were heated to the test temperature at 5 K/ min and a holding time of 1 hour was allowed for equilibrium to be achieved between the furnace and sample temperature before the application of constant load by an electromechanical machine (INSTRON 1362). The creep tests were performed for a specific temperature and stress level until a deformation of 100 µm after which the tests were aborted. The deformation was measured by a linear variable differential transformer (Solartron Metrology). The same sample was subjected to a higher temperature and a higher stress after the deformation at a corresponding lower temperature and lower stress exceeded 100 µm. A total of 12 combinations of temperature and stress were used to measure the creep deformation on the single cylindrical specimen. The creep at lower temperatures (1100-1200°C) might affect the creep at 1300°C, hence creep measurements were in addition carried out on a pristine specimen by directly applying a higher load at a higher temperature (eg, 1200°C/80 MPa). The values for the total deformation over time and the minimum creep rate obtained in these tests were in agreement with the creep testing done by sequential increase in temperature and stress on the same sample, hence, verifying the procedure.
The samples before and after creep tests were ground and polished in both parallel and perpendicular direction of the applied load for electron backscattered diffraction (EBSD, Aztec 2.4; Oxford Instruments) analysis. To avoid the usage of highly corrosive hydrofluoric acid (HF) for etching the grain boundaries, band contrast from kikuchi patterns was used for the reconstruction of an image which is a representation of the microstructure.
Samples for transmission electron microscopy (TEM) were prepared by Focused Ion Beam (FIB, Zeiss Crossbeam 540) equipped with scanning transmission electron microscopy detector (4-Channel aSTEM Detector) for STEM analysis as well as electron backscatter diffraction detector for EBSD analysis on TEM lamellas (transmission kikuchi diffraction). The lamella was obtained from the center of the crept sample as it is believed to have undergone maximum deformation during creep. The lamella was characterized by TEM (FEI Tecnai G 2 20S-TWIN).

| RESULTS
The synthesized Ti 3 SiC 2 powder ( Figure 1A) exhibits a platelet-like morphology with a mean particle size (d 50 ) of 6 µm. The SiC w ( Figure 1B) particles have a diameter of 0.5 µm and an approximate length of 5 µm. Figure 1C shows the XRD of Ti 3 SiC 2 powder, SPS-sintered Ti 3 SiC 2 , and 20 vol% SiC wreinforced Ti 3 SiC 2 . TSC-10SiC w is not shown because it is same as that of TSC-20SiC w . The XRD peaks correspond to Ti 3 SiC 2 , SiC w , and other phases like TiC and TiSi 2 . There is no direct implication of any interfacial chemical reaction between SiC w and Ti 3 SiC 2 that could have compromised the structural integrity of the final CMC. Although the starting Ti 3 SiC 2 powder presented only 3 wt% of TiC and 1 wt% of Al 2 O 3 , the sintered Ti 3 SiC 2 has a higher content of TiC (8.8 wt%) and Al 2 O 3 (4.6 wt%). The higher content of TiC might have resulted from the thermal decomposition of Ti 3 SiC 2 during sintering, whereas the increase in the Al 2 O 3 content from 1 to 4.6 wt% after sintering can be related to the crystallization of amorphous aluminum hydroxide which could not be detected by XRD in the powder stage. The referred aluminum hydroxide might have formed due to the reaction of unreacted Al with water during the washing step after the molten salt-based synthesis process. 28,30 The proposed mechanism of decomposition is by deintercalation of Si-from Ti 3 SiC 2 and subsequent formation of TiC x . 31,32 Traces of TiSi 2 were observed which might have formed due to the reaction between deintercalated Si and Ti 3 SiC 2 .
The density of the CMCs was measured to be above 99% of the theoretical value as calculated by considering the initial volume fraction of Ti 3 SiC 2 and SiC w . The formation of TiC (4.93 g/cm 3 ) and TiSi 2 (4.02 g/cm 3 ) by apparent decomposition does not affect the calculated density because of the countercompensation of new phases resulting in an aggregate with the same density of Ti 3 SiC 2 . The presence of TiC and Al 2 O 3 might affect the creep behavior of the CMCs, but the content of TiC and Al 2 O 3 is same in samples with different volumes of SiC w , hence a comparison can be made. The change in the intensity ratio of the peaks at 39.5 and 40.8 2θ positions is due to the evolution of a crystallographic texture during sintering of Ti 3 SiC 2 . As pointed out earlier, the present study is about the influence of SiC w on the creep resistance of Ti 3 SiC 2 -based CMCs and hence we will ignore here the effect of texture on the creep of Ti 3 SiC 2 and the creep is measured only in the direction perpendicular to the basal plane orientation. Figure 2A-C show the microstructure of SPSed TSC, TSC-10SiC w , and TSC-20SiC w after polishing a face perpendicular to the direction of external pressure applied during sintering. The dispersion of SiC w is homogeneous in TSC-10SiC w , whereas an increase in SiC w resulted in agglomeration in TSC-20SiC w . However, the Ti 3 SiC 2 matrix phase was infiltrated into the SiC w agglomerates. There were no pores observed in any of the composition and this was beneficial for the evaluation of creep in further investigation. Alumina inclusions can be seen as the dark grains and may result in the increase in the hardness and decrease in the fracture toughness and flexural strength of Ti 3 SiC 2 . 33 The effect of alumina on the creep response is unknown at the moment, however, it is expected to improve the creep resistance. Figure 3 reveals the grain shape and SiC w orientation in the direction perpendicular to the direction of uniaxial pressure during sintering. The high vacuum during sintering might be responsible for the evaporation of Si from the crystal structure of Ti 3 SiC 2 -forming TiC. Figure 3C,D depicts the phase map of TSC and TSC-20SiC w corresponding to Figure 3A,B. The presence of TiC on the grain boundaries supports the assumption of Si depletion from the grain boundaries in vacuum at high temperature. Small amount of TiSi 2 resulting from the decomposition of Ti 3 SiC 2 is visible in the grain boundaries of TSC, whereas it is not visible in TSC-20SiC w . This can be related to the reaction of TiSi 2 with SiC w to form Ti 3 SiC 2 as SiC is also used as a precursor for the synthesis of Ti 3 SiC 2 . 2 The presence of TiC may have an effect on the creep of the material, but it is believed to behave similar to that of pure Ti 3 SiC 2 because of the continuous percolation of Ti 3 SiC 2 . Similar results were reported by Spencer et al with 16 vol% of TiC in Ti 3 SiC 2 as compared to 8.8 wt% TiC in the present case. 34 The orientation of SiC w may be beneficial for the creep resistance because the deformation due to compressive stress is in lateral direction which is same as that of whisker alignment. The microstructures of the CMCs on the perpendicular face are equiaxed with an average grain size of 1.2 ± 0.6 µm of the Ti 3 SiC 2 phase, irrespective of the SiC w content. The grain size is much smaller than the initial particle size (6 µm) of the Ti 3 SiC 2 powder, this is probably due to the measurement of agglomerate size instead of the primary particle size in the case of the powder.
Examples of creep curves of SiC w -reinforced Ti 3 SiC 2 -based CMCs for temperatures of 1200 and 1300°C under 20 MPa are shown in Figure 4A,B, respectively. The creep curves 4 can be separated into three stages being representative of (a) primary creep regime with a rapidly decreasing creep rate also known as transient creep stage; (b) secondary creep regime (steady state creep) with a constant creep rate, it is this regime which is of interest and is considered for evaluation of the creep properties of the CMCs fabricated in this work. The tests were aborted before reaching the (c) tertiary creep regime.
Since the creep testing was performed in air, the surface was oxidized because of very long dwell time at high temperature (>1000°C). However, the reduction in cross-section area is not significant and is same for all the CMC with different SiC w content. Hence, we have neglected the effect of oxide scale (~200 µm) on the evaluation of creep data. 4,24 To assess the effect of oxide scale on creep, some creep tests were conducted in the cooling cycle for the sake of comparison. For example, the creep rate of TSC-10SiC w at 1200°C and 40 MPa during heating from room temperature to 1300°C was compared with the same creep conditions but during cooling from 1300°C to room temperature. At 1300°C the oxidation of the sample is more imminent and might influence the creep but the creep rates during heating and cooling were approximately the same. It is worth noting that the formation of oxide scales on the surface plays an important role in the healing of the surface defects and scratches which in turn may result in a better creep resistance. [35][36][37] The addition of SiC w has a profound effect on the creep resistance of Ti 3 SiC 2 . TSC-20SiC w is much more resistant to deformation as compared to monolithic Ti 3  or TSC-10SiC w . The inset shows magnified creep curves during the primary stage. TSC-10SiC w and TSC-20SiC w have similar deformation at 1200 and 20 MPa, whereas TSC and TSC-10SiC w have their creep curves close to each other at 1300°C and 20 MPa. The deformation of TSC-10SiC w is not intermediate to that of TSC and TSC-20SiC w and, hence, the addition of SiC w leads to a nonlinear effect on the increase in creep resistance, being an indication of different activation energies and/or stress exponents as will be discussed further.
The creep curves of pure and SiC w -reinforced Ti 3 SiC 2 in the temperature range 1100-1300°C at 40 and 80 MPa are shown in the Figure S1. At higher temperature (1300°C, 40 MPa, Figure S1C), the primary creep regime is indistinguishable for TSC and at higher loads with high temperature (1300°C, 80 MPa, Figure S1F), the primary creep is concealed even for TSC-20SiC w .
The steady-state creep rate ̇ was obtained by measuring the slope of the fitted straight line in the steady-state regime of the creep curves with time as the x-axis and creep strain as the y-axis. The activation energy and the stress exponent values were calculated by using the following relation 38 where A, n, and Q are dimensionless constant, stress exponent, and activation energy, respectively. ̇ o and o are 1 s −1 and 1 MPa used for defining the dimensions of strain rate and stress. The slope of straight lines plotted by fitting the data of minimum creep rate with respect to inverse of absolute temperature and ln of the applied stress gives the activation energy and stress exponent, respectively. Equation (1) is simplified into straight line equation (y = mx + c) by applying the natural logarithm on both sides of the equation, where m accounts for the slope of the fitted line. Figure 4A presents Arrhenius plots of the steady-state creep rate vs the reciprocal absolute temperature for different applied stresses. Regression analysis of the data was carried out via a global fit of all data using the software Origin to obtain the activation energies for the considered temperature range 1100-1300°C.
A simplified bootstrap type analysis was done by deleting individual data points and performing a fit of the rest of the data points to identify if specific data points cause an increase in scatter of the activation energy. It was found that the data points for TSC-10SiC w at 1100°C and 20 MPa corresponded to a creep of 1 μm/d and is probably still in primary creep regime, hence, it was excluded for the calculation of activation energy, the same was done for TSC-20SiC w at 1100°C and 40 MPa.
The calculated activation energies for the CMCs are summarized in Table 1. Upon the addition of SiC w there is reduction in activation energy from 696 ± 16 kJ/mol for monolithic Ti 3 SiC 2 to 665 ± 31 kJ/mol for TSC-20SiC w with a similar value for the material with 10 vol% whisker, so the addition of SiC w results in ~10% lower activation energy, which is, however, within the limits of experimental uncertainty. A lower activation energy simply means a change in creep rate controlling process and has no strict implication toward the resistance to creep of a material, however, more indications with respect to the active creep mechanism is provided by the stress exponent values. Hence, the overall activation energies of three compositions are similar and the rate controlling factor can be assumed to be almost the same as well.
Ln-ln plots of steady-state creep as a function of applied stress for different temperatures are given in Figure 5. The stress exponents (n) were calculated by a linear regression analysis of the data. A global fit of the creep rates of TSC and TSC-10SiC w was performed, whereas, for TSC-20SiC w , individual fits were performed since the global fit with TSC-20SiC w resulted in large scatter.
The stress exponents of every composition are compared in Table 1 for the different temperatures. The increasing values of stress exponents of TSC-20SiC w clearly indicate a transition of the creep mechanism for TSC-20SiC w at different temperatures.
With the addition of 10 vol% of SiC w addition there is a reduction in the creep rate by two orders of magnitude. There is no temperature dependency of the stress exponent for TSC and TSC-10SiC w , hence, the creep mechanism is expected to be same over the whole temperature range. On the other hand, TSC-20SiC w has a strong temperature dependency and the stress exponent increases from 3.4 at 1100°C to 4.6 at 1300°C.
Fractured surfaces of creep-tested samples are shown in the Figure S2. The TSC samples show delamination, whereas CMCs with SiC w show rather kinking and bending of grains ( Figure S2A-C).

| Microstructure development of MAX phase-based CMCs
Pressure-assisted sintering of hexagonal materials normally yields an anisotropic microstructure. Whether the c-axis is aligned perpendicular 39 or parallel 40 to the pressing direction is material dependent. It was reported by Tang et al that the basal plane of Ti 3 SiC 2 grains prefers a nonparallel alignment with the pressing surface, which implies that the c-axis is preferably aligned perpendicular to the pressing direction, 41 however, the opposite is observed in the present case. Since the starting powder had platelet morphology, a plate-like microstructure was to be expected. TSC-10SiC w was selected for microstructural investigation assuming TSC-20SiC w will have similar microstructure with a higher content of SiC w only. There was no difference in average grain size expected because the sintering was performed at a low temperature (1250°C). Figure 6A shows the microstructure of TSC-10SiC w perpendicular to the pressing direction. Clearly, the sintered TSC-10SiC w has a texture due to the applied uniaxial external pressure and the rearrangements of the platelets under high temperature and pressure to attain a high density. The c-axis is parallel to the pressing direction ( Figure 6B). As expected, the SiC w is also oriented perpendicular to the pressing direction and the orientation of the whiskers can be considered as random within the perpendicular plane. 42 In Figure 7A, only the tips of SiC w are visible, hence are not identifiable as entities with high aspect ratio (white arrows). The Ti 3 SiC 2 grains on the face perpendicular to pressing direction are equiaxed, whereas on the face parallel to pressing direction the grains were elongated with an aspect ratio ~10.
Since the microstructure is anisotropic, the grain size and morphology of parallel and perpendicular face with respect to pressing direction are also different. The equivalent grain size is calculated by Relation (2). 42 The effective grain size can be derived from: where d ⊥ and d ∥ are the average grain size of perpendicular and parallel face, respectively. Although the grain morphology was different comparing the parallel and perpendicular face, the average grain size calculated by the projected area diameter was similar for both faces and the effective grain size was always calculated to be 1.2 ± 0.6 μm. Unlike expected, the addition of SiC w did not result in any grain size refinement due to the absence of grain growth of the starting Ti 3 SiC 2 powder owing to the low sintering temperature. For comparison, the orientation of monolithic Ti 3 SiC 2 in the parallel direction of applied stress is shown in Figure 7C. It can be noted that the anisotropy of monolithic Ti 3 SiC 2 after sintering is higher than TSC-10SiC w . The Ti 3 SiC 2 grains are aligned in the direction perpendicular to the applied uniaxial stress for sintering. The degree of alignment of Ti 3 SiC 2 platelets is higher because of absence of any SiC w which might have restricted grain rearrangement during sintering.

| Creep deformation mechanism
The creep deformation mechanisms of composite materials are difficult to decipher by only measuring activation energies and stress exponents. Activation energies of all compositions agree within the limits of experimental uncertainty. The stress exponent of TSC and TSC-10SiC w is around 3 and is constant over the whole temperature range (1100-1300°C), whereas for TSC-20SiC w the stress exponents increase with increasing temperature.
Stress exponents of 3-5 indicate dislocation climb as the rate controlling mechanism. In addition, the deformation process also involves dislocation glide due to shear stress acting normal to the applied uniaxial stress during creep test. 43 The rate control mechanism of both monolithic and reinforced Ti 3 SiC 2 MAX phases is similar due to similar activation energies, but the exact rate controlling mechanism of the CMCs is not clear.
In case of TSC-10SiC w , GBS is evident by the change in the grain direction and alignment of the basal plane normal to the applied stress during creep (compare Figure 7A,B). Some basal planes which were visible in Figure 7A (TSC-10SiC w precreep) disappeared after creep in Figure 7B (TSC-10SiC w postcreep). There appears not to be any GBS in case of TSC, which can be concluded from the grain shape before and after creep, which is further supported by grain orientation data by EBSD ( Figure 7C,D).
The evidence of the presence of GBS in TSC-10SiC w is also provided by the evolution of low-angle grain boundary (LAGB) in TSC-10SiC w after creep ( Figure S3). 44 This appears to be a clear case of grain rotation during high-temperature deformation. 45 In the case of TSC, the misorientation angle distribution remains the same after creep.
It is now well known that dislocations exist in basal planes of MAX phases and these dislocations also form a LAGB normal to the basal plane and this is observed in the current work as well. 21 The SiC w may limit GBS by interlocking the Ti 3 SiC 2 grains. 46 With the increase in SiC w content, the creep resistance simply increases because of reduction in the amount of plastic phase in the composite.
The basal plane is aligned perpendicular to the pressing direction and so is the <0001> direction. Although finegrained Ti 3 SiC 2 is not susceptible to deformation as their coarse-grain counterparts, during creep kinking also might have played a role to absorb the mechanical stress during creep. The shear of (0001) plane is difficult because it involves the breaking of Ti-C bonds, whereas the Ti-Si bond is more susceptible to break. Hence, a preferential dislocation glide can be the mechanism favored along the basal plane. 47 As the material has a high fraction of grains with their basal plane aligned perpendicular to the compression direction, there is also a higher possibility of basal plane shear.
For temperatures < 1300°C and low stresses (<40 MPa), the presence of SiC w reduces the creep, whereas the increase in the content of SiC w (from 10 to 20 vol%) has only a subtle decrease in the creep rates (Figure 4), suggesting the extra 10 vol% SiC w does not efficiently inhibit GBS rather, this might imply dislocation creep as one of the mechanism in case of SiC w -reinforced Ti 3 SiC 2 . The presence of rigid SiC w may have restricted GBS of Ti 3 SiC 2 partially or completely and hence resulted in a smaller deformation and creep rate. 42 Since the aspect ratio of the SiC w is ~10, the critical percolation limit volume fraction (ϕ pcp ) is around 7% as calculated by the following relation: 48 where ϕ pcp is the limit above which the whiskers can form an interconnected network. Since the materials are sintered under uniaxial pressure, there is a lack of three-dimensional connectivity (percolation) of SiC w , rather there are two-dimensional arrays of SiC w perpendicular to the uniaxial stress during sintering. ϕ pcp for such materials can be larger by a factor of 2 because Expression (2) holds true for only randomly oriented whiskers. 49 Considering a ϕ pcp of 14%, the content of SiC w used in the present study is both below (TSC-10SiC w ) and above (TSC-20SiC w ) percolation limit. The fact that the stress exponent of TSC-20SiC w becomes larger at high temperature indicates that the property for this material under this condition is not a general material representative rather representative for material under specific conditions only.
The increasing stress exponents of TSC-20SiC w provide the evidence of a change in creep mechanism over the temperature range 1100-1300°C, whereas for TSC and TSC-10SiC w the creep mechanism is same across the entire temperature range. At 1200°C, 20 MPa, the addition of SiC w may have restricted GBS and hence the lower deformation and creep rates of TSC-10SiC w and TSC-20SiC w . Although fine-grained Ti 3 SiC 2 has a brittle to plastic (BTP) transition at ~1200°C, the creep resistance is enhanced by the addition of SiC w . 50 Since the addition of SiC w has an increase in the creep resistance of Ti 3 SiC 2 , grain boundary diffusion may have been hindered.
At a temperature of >1200°C, both lattice and grain boundary diffusion may have become significant as indicated by the BTP transition of Ti 3 SiC 2 and hence pile-up of dislocations can be reduced efficiently and so the addition of SiC w did not reduce the creep rates considerably at 1300°C. At temperature above 1200°C, where the Ti 3 SiC 2 has a BTP transition, the creep of CMCs (specifically TSC-20SiC w ) can be viewed as a sequential viscoplastic and viscoelastic deformation. The viscoplastic creep refers to the approach of SiC w randomly oriented perpendicular to the creep direction. The matrix phase (Ti 3 SiC 2 ) is squeezed out from between the in-plane SiC w network. The hard SiC w acts as lines of stress concentration which cuts through the soft Ti 3 SiC 2 grains.
The large arrows in Figure 6C show a grain with straightedge grain boundary because of SiC w sinking into it and a cut Ti 3 SiC 2 grain with SiC w in the grain boundary. The distance between the randomly SiC w is stochastic and the lack of three-dimensional SiC w network is due to the perpendicular alignment of SiC w to the stress during sintering. For interconnectivity there should be a random angular distribution of SiC w , whereas it is not in the present case. The viscoplastic deformation ceases once the SiC w have point contacts within the perpendicularly oriented planes.
The elastic bending of SiC w with the deformation of Ti 3 SiC 2 in the immediate vicinity of the SiC w to accommodate the elastic deformation of SiC w is referred to as viscoelastic creep. The viscoplastic deformation is only if ϕ ≤ ϕ pcp , hence this is valid for TSC-10SiC w , whereas in the case of TSC-20SiC w , the SiC w network is responsible for decreased creep rate. The probable absence of any viscoplastic creep above 1200°C results in an increase in creep resistance of TSC-20SiC w . Since TSC-20SiC w has a SiC w content higher than ϕ pcp , the SiC w network inhibits GBS and the creep is governed by only diffusion, hence, it has lower creep rates as compared to TSC and TSC-10SiC w . However, beyond a certain stress TSC-20SiC w behaves different and the creep rates are not reduced to the same degree as at low stresses. This might be due to the interaction of agglomerates of SiC w present in the material.

| Transmission electron microscopy
The TEM micrographs of the crept specimen demonstrate various defects indicating the creep mechanism which it underwent. The defects, dislocations, as well as grain structures visible in the microstructures have been elucidated to explain the formation of the dislocation structures and their mechanism. The most common defects visible in crept MAX phases are edge dislocations and stacking faults. The layered inherent character of MAX phases makes it prone to stacking faults during processing as well as deformation. 51 Stacking faults are diffusion-driven phenomenon which are evident in the vicinity of grain boundaries ending with partials. 52 Triangular cavitation can be observed, which is an implication of GBS. Polygonal cavitation is also visible on triple junctions of the specimen in Figure 8A. Size of cavities ranges from 50 to 60 nm; basal dislocation arrays are seen to be surrounding the cavitation. Dislocation pile up at the whiskers boundary is visible here; cavitation (triangular ones) was also observed by Barsoum et al in tensile crept specimen. 24,44 The faint grey contrast throughout the specimen is due to bending, curling, and kinking of the layers in the crept sample. Due to angular field of view, we can see the kinks and bends as small curly entities. Stacking faults ( Figure 8B) pertaining to basal plane are very often visible in MAX phase. Basal dislocation glide gives rise to formation of grain boundary steps. GBS is a failure mode which is evident from Figure 8C during creep, the layered nature of Ti 3 SiC 2 promotes sliding action of layers which also gives rise to GBS of the material. A curved grain with a kink band in the entre is shown in Figure 8C suggesting the presence of kinking during creep, although not in abundance. The basal planes generally form a curved surface to accommodate the grain boundary step formation. Grain boundary steps can be seen in Figure 8D which is a result of grain boundary glide planes during creep of the specimen which is an identified phenomenon in the past. 44 Hexagonal networks can be seen in Figure 8E along with complex dislocation network. Dislocation-dislocation interaction generally leads to dislocation reactions forming planar hexagonal networks which can be seen in Figure 8E. Figure 8F reveals whiskers embedded in the matrix surrounded by dislocation loop. Figure 8G demonstrates defects which are non-planar in nature. These defects are unusual to MAX phase and are a result of complex dislocation interactions and superimpositions. The inset shows selected area diffraction (SAD) pattern pertaining to basal plane of the MAX phase showing sixfold symmetry and some amount of streaking due to layer morphology of Ti 3 SiC 2 .
The SiC w (bright white areas) can be seen impeding the dislocation movement, the Figure 9A shows the pile up of dislocations near the SiC w . Figure 9B,C shows a dislocation network in the vicinity of the SiC w entities. The dislocation pile up around the whiskers proves that they impede the dislocation movement in the Ti 3 SiC 2 matrix. Edge dislocations are visible in ample number in the matrix, interacting with each  Figure  8A, 8B, 8C and 8D denotes cavitation due to creep, stacking faults, cavitation due to GBS and grain boundary steps due to creep respectively. The curved grain in Figure 8C is a result of kinking during creep and is denoted by curved white dashed lines. The black arrows in Figure 8G Figure 9D shows dislocation cells formation which is a sign of recovery during creep. There were no features identified in the case of TSC-20SiC w implying GBS as in the case of TSC-10SiC w . Transmission kikuchi diffraction of the lamella in Figure S4 reveals that the grains with nonbasal orientation also have dislocations present in them. As compared to the data reported in the literature (Table 2), the activation energy and stress exponents of Ti 3 SiC 2 are higher, this might be attributed to the high anisotropy and texture present in the currently investigated monolithic Ti 3 SiC 2 . Monolithic Ti 3 SiC 2 is believed to undergo pure dislocation creep, whereas TSC-10SiC w undergoes a combination of dislocation creep, GBS, and viscoplastic creep. This has been shown and correlated by the postcreep EBSD and TEM analysis. TSC-20SiC w can be assumed to have undergone a combination of dislocation creep, viscoplastic creep, and viscoelastic creep. Although the proposed mechanism for monolithic material and composites with partial evidences are different from each other, the rate controlling creep mechanism is believed to be mainly dislocation based and hence all the materials has similar activation energies. Moreover, the transition in the creep mechanism of TSC-20SiC w is supported by the change in stress exponent over the temperature range.

| CONCLUSION
Ti 3 SiC 2 -based CMCs were fabricated by reinforcing with SiC w and sintering the same to full density by SPS. The grain size of the CMCs irrespective of SiC w content was measured to be 1.2 ± 0.6 µm. The dispersion of SiC w was uniform for TSC-10SiC w, whereas for TSC-20SiC w there were agglomerations of SiC w . The compressive creep of the composites was carried out in air in the temperature  T A B L E 2 Summary of creep activation energy and stress exponents for Ti 3 SiC 2 in compression with comparison to the previously investigated results range 1100-1300°C and stress level from 20 to 120 MPa. The creep rates were a function of the SiC w content and decreased with increasing SiC w . The stress exponent of TSC and TSC-10SiC w was constant over the range of temperatures but that of TSC-20SiC w increased with increasing temperature from 3.4 at 1100°C to 4.6 at 1300°C. The creep mechanism of monolithic Ti 3 SiC 2 is believed to be dislocation creep, whereas for composites, a combination of dislocation creep and GBS is believed to be the deformation mechanism. SiC w inhibited the GBS by anchoring grains together; a higher content of SiC w in Ti 3 SiC 2 might be responsible for creating a network of particles resulting in higher creep resistance. A combination of viscoplastic and viscoelastic creep can be operating during the creep of TSC-20SiC w .