Monitoring Damage in Nonoxide Composites at High Temperatures Using Carbon-Containing CVD SiC Monofilament Fibers as Embedded Electrical Resistance Sensors

Ceramic matrix composites (CMCs) especially nonoxide composites evolved as the candidate materials for future propulsion systems for improved thermal efficiency with lower exhaust emissions because of their low weight, high temperature capability, and high strength at high temperatures [1]. However, the mechanism for CMC toughness and strength involves matrix microcracking which will result in loss of load bearing capability when subject to oxidation at elevated temperatures, ultimately reducing the life of the composite. As a result, an increasing importance is given to monitor and predict the damage initiation and propagation before the occurrence of a catastrophic failure.

Conventional nondestructive evaluation techniques such as ultrasonic scan, X-ray, thermal diffusivity, and micro-CT can be used to assess the integrity of the material, damage detection, and for quality control [2–5]. However, these methods are used for postdamage inspection and require the removal of the structure from operation and in some cases can only be used on relatively small specimens (micro-CT). Some of these nondestructive evaluation techniques (X-ray, ultrasonic, and thermal diffusivity) are excellent at detecting interlaminar (out-of-plane) damage but are incapable of detecting in-plane damage such as transverse cracking which often occurs under tensile loads [6]. Acoustic emission (AE) is a well-established technique which can monitor the damage initiation and propagation in CMCs at room temperature [7–12]. However, one major limitation of AE is its effectiveness during high temperature testing.

Electrical resistance (ER) is a simple technique which can monitor the damage in the CMCs at ambient and high temperature [13–20]. The main aim of this study is to improve the sensitivity of ER in damage detection at high temperatures by using carbon via CVD SiC monofilament fibers. The monofilaments consist of carbon core and carbon outer layer. Carbon is more conductive than the other constituents of the CMC (SiC and/or Si), and it oxidizes around 450 °C. Thus, any damage to the carbon core in the form of cracks or other open access channels to the environment at elevated temperatures will result in an increase in resistance when exposed to oxidizing species. Also, the conductivity of carbon increases with temperature (similar to SiC but opposite to Si) but not to the degree that SiC or Si do. For example, the resistance of carbon is less temperature sensitive. Therefore, two approaches were taken as to placement of C. The first approach is to place the monofilaments so that they extend from the cold section to just outside the region of interest (notched region), i.e., using C as a lead that extends into the hot zone of the region of interest. The second approach is for the monofilament to extend from one end of the specimen to the other so that if C is penetrated, oxidation of C will result.

SiC matrices can vary considerably for different composite systems [21]. Si-containing matrix (melt-infiltrated or MI) composites can be two orders of magnitude more conductive than chemically vapor infiltrated or polymer infiltrated pyrolysis (PIP) matrix composites due to the presence of a continuous Si phase in MI composites. As mentioned, the resistivity of Si increases with temperature significantly compared to C which decreases with temperature. Thus, with increasing temperature, the contribution of current being carried by C will increase. Therefore, two matrix types were used in this study: a high conductivity Si-containing matrix composite (prepreg MI) and a low conductivity PIP matrix to assess the effects of relative constituent conductivities and different temperature dependences of the different current carriers.

The specimens considered were subjected to a step fatigue condition with a selected stress condition at a stress ratio of 0.1 and frequency of 1 Hz and specimen surface temperature of 815 °C. AE sensors were placed in the colder region of the specimens as well to assist in monitoring damage and to interpret the change in resistance. Fracture surfaces of the failed specimens were observed under scanning electron microscope (SEM) to understand the effect of oxidation and carbon recession length.

Polymer infiltration and pyrolysis composites were fabricated using a polymer-derived process with a boron nitride (BN) interphase and Hi-Nicalon (Nippon Carbon, Tokyo, Japan) fibers by COI ceramics (San Diego, CA). An older version of SCS-6 prototype CVD monofilament (Specialty Materials, Lowell, MA) [21] with a single outer C layer was uniformly spaced and aligned unidirectionally along the tensile loading direction in the center of the composite so that the woven plies of the composite were balanced on both sides of the SCS-6 monofilaments. The SCS-6 fibers consist of 33 μm turbostatic carbon core, 108 μm of CVD SiC deposited on the C core, and a final carbon outer layer of 1.5 μm thickness. Further details of the SCS-6 fiber can be found elsewhere [22]. Three different PIP composite panels were fabricated: a standard panel with no monofilaments, a single layer continuous monofilament, and a single layer with discontinuous monofilaments (Fig. 1). The microstructures with and without the SCS-6 fibers are shown in Fig. 2.

The MI composite was fabricated using a proprietary prepreg melt infiltration process. The prepreg system consists of Hi-Nicalon Type S (Nippon Carbon) fibers coated with BN interphase. The coated fiber tows were arranged in the form of unidirectional prepreg tapes in the form of [0/90]_2S configuration. The SCS-Ultra (Specialty Materials) monofilaments used in the prepreg composite have a 33 μm carbon core, a 108 μm CVD SiC sheath, and a 1.5 μm pyro-carbon coating. The same configurations of monofilaments were fabricated as in the PIP system. An additional panel was fabricated where a continuous monofilament layer not only was placed on the interior of the composite but also two other layers near both surfaces of the composite. Therefore, prepreg composites consisted of a standard [0/90]_2 s with no monofilaments, a single center layer continuous SCS-Ultra, single center layer discontinuous SCS-Ultra, and a three-layer continuous SCS-Ultra. The typical microstructure of SCS-Ultrafiber, standard material, and continuous SCS-Ultraspecimen is shown in Fig. 3.

All the specimens tested in this study were straight sided. The dimensions of the PIP system are 152.4 mm length, 12.8 mm in width, and 2.6 mm in thickness, and the MI system are 152.4 mm in length, 12.4 mm in width, and 2.0 mm in thickness.

All the specimens considered in this study were subjected to a tension–tension fatigue loading at a selected stress condition, frequency of 1 Hz, stress ratio of 0.1, and a test temperature of 815 °C on a specimen with a single-edge notch of ∼1.5 mm. A resistive heat furnace (AMTECO) with 15 mm hot zone length was used to raise and maintain the test temperature. Temperature was raised at a rate of 40 °C/min and held for 10 min prior to the fatigue condition in order for the temperature to equilibrate on the specimen. During the temperature ramp up, the MTS machine (MTS 8100 system) was set to maintain a minimum load of 50 N to prevent the specimen going into compression due to thermal expansion. Fatigue loading was performed using a step approach. Initially, a peak stress condition was chosen, and specimen was subject to fatigue at that peak stress for 24 h. If the specimen survives at the initial stress, the peak stress was raised (maintaining an R = 0.1) and tested for another 24 h. This process was repeated until the specimen fails. Details of test setup are shown in Fig. 4. It should be noted that only one specimen from each panel of the two composite systems was evaluated under fatigue loading.

Electrical resistance was measured using the four-probe method. A high conductive silver paste was used to attach the current leads in the electrically insulated and water-cooled grip region to protect them from high temperatures. A constant current source digital multimeter (Keithley 2450 Source Meter) was used to send constant current. The potential drop was recorded using an Agilent (model 34980A).

Two high temperature AE sensors (D9215, 50–650 kHz) were clamped on the face of the specimen. Vacuum grease was used as a couplant for PIP composite. However, due to the MI more highly conductive matrix, current was leaking through the metal AE sensors. A thin alumina cement layer (AREMCO, Ceramabond 671) was applied to the surface of the composite to electrically insulate the sensor which also provided a mechanical couple for sound transmission to the sensor. Pencil lead breaks were performed prior to the start of the test to check if the sensors were properly attached.

To assess the effect of the C containing monofilament fibers on the electrical properties of the different composite systems and monofilament configurations, room temperature resistance measurements of the monofilaments and different composites specimens were performed using the four-point probe technique. The measured resistance and calculated resistivity of the individual monofilaments, the PIP composites, and the MI composites based on the cross section and length of the inner lead distance are shown in Tables 1, 2, and 3, respectively.

There is a strong effect of monofilament content on the resistance and resistivity of PIP composites demonstrating the effect of a small amount of high conductivity continuous carbon phase (Table 2). For the MI composites (Table 3), negligible change in resistance was observed with the addition of a single layer of monofilaments. There was a considerable reduction in resistance for the triple layer composites. It should be noted that for the triple layer MI composites, a thicker layer of matrix was added to the surface to encapsulate the fibers which was probably the major contributor to the reduced resistivity.

Based on this analysis, approximately 75% of the current would be carried by the monofilament in the continuous PIP composite, whereas only 0.8% of the current would be carried by the monofilament in the continuous MI composite.

where ρ_scs and L_scs refers to the resistivity and length of SCS region, ρ_{no scs} and L_{no scs} refers to the resistivity and length of no SCS region and A refers to the cross-sectional area of the composite.

There was good agreement between the estimated resistivity of the as-produced composites based on monofilament content except for the triple layer MI. This composite had excess matrix and free Si on the surface which would reduce resistivity and was not accounted for in the analysis.

where ΔR is the measured R at each notch increment minus the initial R_o of the pristine composite. The R_o for the PIP composite was 24.71 ohms, and for the MI composite was 0.54 Ω. The % change in ER is considerably greater for the PIP system compared to the MI system.

A significant % change in ER was observed for the PIP system compared to MI with increasing notch length. The primary current carrier in the PIP composite is the high conducting carbon phase of the monofilaments. An assumption was made that only the SCS monofilaments carried the current. Then, when a monofilament is cut by the notch, then monofilament would no longer be capable of carrying any current and the current was forced to flow through the other intact monofilaments resulting in an overall increase in the resistance. Figure 5 also plots this assumption. Initially, the resistance of the single SCS monofilament for the given composite length was estimated based on Table 1. There were ten SCS fibers in the composite, and the composite resistance was calculated by dividing the estimated SCS fiber resistance with the number of monofilaments in the composite. With every 1 mm increment in notch length, one monofilament was removed, and the composite resistance was calculated. The overall trend observed based on this is in line with the measured change in resistance. This indicates that in a PIP system when the current carrying SCS fiber is cut it is no longer capable of carrying any curren,t and the remaining current is primarily forced through the remaining monofilaments.

For MI less than 1% of current was carried by the SCS monofilaments compared to the high conducting matrix. Therefore, a smaller change in resistance was observed with increasing notch length (Fig. 5) since current flows through the larger area matrix.

The specimens were subjected to a tension–tension fatigue loading using a step approach at a frequency of 1 Hz and a stress ratio of 0.1. For the PIP system the initial fatigue stress was selected as 100 MPa. The change in ER and AE energy for each peak stress step for the standard, continuous SCS and dis continuous SCS composites are shown in Figs. 6–8. For all the PIP composites, resistance decreased with temperature before loading as one would expect for SiC and/or C.

For the standard PIP, with loading and continued fatigue at a peak stress of 100 MPa, a gradual increase in resistance was observed (Fig. 6(a)) with time/cycles indicating some kind of change to the circuit. No AE events were recorded due to bottom sensor malfunction for this step. The specimen survived the 24 h, and the peak stress was increased to 120 MPa (Fig. 6(b)). The overall ER trend observed at 120 MPa (Fig. 6(b)) stress condition was similar to the 100 MPa stress. However, AE monitoring was successful for this step and showed a good correspondence with the change in ER (Fig. 6(c)). AE events are expected to correspond to crack formation and propagation. In other SiC/SiC systems, matrix cracking can be directly related to AE energy [23]. Note the excellent correlation of cumulative AE energy (the summation of energy of each event) with change in resistance. At failure, an increase in AE activity and ER was observed.

The change in electrical resistance and AE cumulative energy for the three fatigue stress-steps of the single layer continuous SCS specimen is shown in Fig. 7. For 100 MPa peak stress fatigue (Fig. 7(a)), a change in resistance, ΔR (resistance at time minus resistance at 815 °C) of 8 Ω was observed during the fatigue part of the test with no AE activity. No AE activity at this stress level suggests no crack initiation/growth. At 120 MPa and 140 MPa stress levels (Figs. 7(b) and 7(c)), increase in ER and AE activity were observed, indicating some crack initiation and propagation.

The ER data for the discontinuous SCS system is shown in Fig. 8 for two fatigue stress-steps (unfortunately AE monitoring was unsuccessful for these tests). For the initial 100 MPa peak stress fatigue (Fig. 8(a)), a change in resistance of 53 Ω is observed during the fatigue part of the test. This was much larger compared to the standard specimen (14 Ω) and continuous SCS specimen (8 Ω) for the same condition.

Two observations are significant from the PIP fatigue crack growth experiments. First, there was an excellent correlation between ER and AE when AE could be captured effectively. This implies that for the standard material, at least, ER is a good monitor of damage. Second, the change in resistivity for C containing composites was quite different for discontinuous compared to continuous composites. Figure 9 shows the change in resistance for the first two fatigue steps for the three different PIP composites. Note that the discontinuous specimen had the largest change in resistance for each step and the continuous specimen had the least change. For the 100 MPa peak stress fatigue step, the ΔR for the discontinuous composite is nearly five times that of the no SCS specimen. It is hypothesized that the large ΔR for the discontinuous specimen must be due to the oxidation of carbon core from the cut end of the monofilament in the furnace. There should be no matrix cracking near the cut end of the monofilament since the cut end is ∼12 mm from the notch region. Oxidation of the C core must occur via oxygen ingress through the porous PIP matrix. For the continuous specimen, no AE was observed for the 100 MPa step but was observed for the 120 and 140 MPa steps and the ΔR for each fatigue step was less than that of the No SCS specimen. For this, it is hypothesized that no cracking occurred for 100 MPa fatigue; however, the change in ER was due to oxidation of the monofilament outer layer C coating. During the fatigue steps where matrix cracking occurred (120 and 140 MPa), a relatively small ΔR occurred which implies that matrix cracks debonded around the monofilaments leaving the C core intact through the length of the composite until near failure.

To confirm this, the fracture surface of the continuous SCS specimen is shown in Fig. 10. Only the monofilament nearest the notch showed removal of the carbon core whereas the other monofilament fracture surfaces did not. This implies that only one monofilament failed some time before the ultimate failure event. For the same continuous specimen, polished cross section was obtained starting at 2 mm away from the fracture surface (Fig. 11(a)). Note that the C core was in-tact; however, the carbon outer layer was removed. Sections were subsequently cut and polished further lengths from the fracture surface. At a length of 22 mm from the fracture surface, the outer carbon layer was completely oxidized on all the monofilaments. At 23 mm, the outer carbon layer was intact (Fig. 11(b)). The distance of 22 mm corresponds to the distance from the notch to the top of the furnace. If oxidation occurred through a propagating matrix crack, one would expect carbon recession along the outer C layer to be deeper nearer to the notch due to oxidation kinetics (longer time of oxidation from the exposed surface carbon layer in the matrix crack). However, equal recession lengths of carbon outer layers of all monofilaments suggest that oxidation was occurring through the porosity of the matrix rather than from an exposed end. This observation indicates that increase in resistance with no AE activity at initial stress level of the continuous specimen is due to recession of carbon outer layer. Microscopy is currently underway for the discontinuous specimen to affirm or not the recession of cut-end C core.

For the MI system, an initial peak stress of 180 MPa was selected which proved effective for eventual failure. The initial resistance of the standard material was 2.3 Ω whereas the initial resistance of triple layer continuous SCS was 0.5 Ω and discontinuous SCS was 2.4 Ω. The change in resistance along with AE cumulative energy for the as produced standard material, triple layer continuous SCS, and discontinuous SCS is shown Fig. 12 (the single layer continuous specimen ER short-circuited). The overall ER trend for all three systems was very similar where a sharp increase in resistance was observed upon loading which was accompanied by significant AE events. Presumably, this corresponds to initial crack growth from the notch primarily in the form of tunnel cracking along the interior 90 plies [23]. The total change in resistance during fatigue to failure at temperature for standard and discontinuous SCS material (Figs. 12(a) and 12(b)) was comparable (0.2 Ω). This behavior corresponds with the similarity in resistivity at room temperature for both systems (∼0.35 Ohm mm). The change in resistance observed during the entire test correlated well with the cumulative AE energy which has to be associated with crack growth similar to the No SCS PIP composite for these two specimens as well. Evidently, since there were no monofilaments in the region of the notch for these two specimens, crack growth and ER response were very similar.

For the triple layer continuous SCS specimen, a much smaller change in resistance was observed over the entire test (0.08 Ω). Sharp increases in ER were observed when AE events occurred, indicating crack growth. After initial loading ER was relatively constant, even decreasing slightly. However, at ∼16,000 s and ∼23,000 s after a number of AE events and sharp increase in ER, there was observed a continuous increase in ER with little or no AE activity (Fig. 12 enhanced). It is hypothesized that this increase in ER without AE implies that some monofilaments fractured during the prior crack growth event which led to carbon removal from the core and an increase in ER. The fact that no increase in ER was observed after initial loading implies that the monofilaments initially bridged the propagating crack.

Even though the triple-layer SCS test had a lower absolute change in resistance than the other two specimens which had no monofilaments in the notch section, the percent change in resistance was twice that for the triple layer composite (16% for the triple layer compared to 8% for the other two composites). This result suggests that increasing the volume fraction of carbon containing SCS fibers will enhance the sensitivity of ER to damage. It should be noted that the fraction of current carried by the C will increase with temperature since the resistivity of C decreases with temperature and that of Si increases.

To confirm this hypothesis, the triple layer SCS specimen fracture surface was examined under the SEM. A map of the monofilament placement in the cross section is shown in Fig. 13. It can be seen from the image that the carbon core was oxidized up to the sixth monofilament in the center layer, whereas only the first monofilament carbon core was oxidized in the edge layers. Oxidation of carbon core in the first six fibers is due to tunnel cracking in the interior 90 plies [23]. In brittle laminate that are subjected to tensile loading, crack tunneling occurs in the weak 90 deg ply, as crack tip grows further across the width of the specimen the stress intensity for the crack to propagate in the thickness direction will increase so that a through-thickness crack develops behind the leading inner ply tunnel crack tip [24]. Since oxidation of the carbon core did not occur in most of the fibers, it can be assumed that the position of the crack just prior to failure was something like the triangular region highlighted in Fig. 13.

To apply damage detection concept to more complex structures and closer to the hot zone section, a new method of incorporating features into the composite themselves such as SCS monofilaments with carbon core shows some promising results to improve sensitivity of ER to damage. The two different matrix-types showed different behaviors which may be able to be exploited for damage monitoring.

For the PIP system, the general resistance properties of the composites were significantly affected by the small content of carbon in the composite due to the poor conductivity of the PIP composite itself. At elevated temperature during fatigue crack growth, it was observed that the most prevalent effect of carbon was the oxidation of the outer carbon layer for continuous monofilament composites and the inner carbon core for the discontinuous composite via oxidation through the porous matrix. Though this was not the initial objective of this study, it may be a useful property for monitoring oxidation effects in this composite system. Due to the porous nature of the matrix, monofilaments did not fail easily in the matrix crack, and the desired effect of carbon removal from the matrix crack was not achieved. For this, fabricating similar composite systems with a chemically vapor infiltrated matrix might be more advantageous.

For the MI system, the presence of carbon had a minimal effect on general electrical resistivity of the composite because of the relatively high conductivity of Si and small volume fraction of carbon. However, for crack growth where the fibers were continuous, the effect of carbon removal was apparent due to notch crack growth over time.

One other outcome of this work was the excellent correlation between cumulative AE energy and change in resistance during fatigue at temperature. Since the measurement of resistance is much easier and more practical than AE, this correlation may prove useful enabling ER to be a much more practical and effective health monitoring technique than AE for elevated temperature conditions.

• Dr. David Shiffler, Office of Naval Research (Grant No. N00014-18-1-2646; Funder ID: 10.13039/100000006).