1.

Introduction

Repair and maintenance of structures demand a considerable amount of resources to be used each year. Embedded sensors as a component of smart structures technology can provide structural health information. Embedded sensors can give quality assurance in new structures or an estimate of remaining utility in aging structures. This capability provides an opportunity for preventive measures, e.g., performing repairs in time to prevent any major damage. Embedded fiber optic sensors have been used for cure monitoring, fatigue detection, strain profiling, and temperature measurement in the fields of aerospace, energy, and infrastructure.^1–4 Optical-fiber-based sensors have gained wide interest for structural health monitoring applications due to their compact size, immunity from electromagnetic interference, multiplexing capabilities, and so on. Different types of optical-fiber sensors such as fiber Bragg gratings,¹ extrinsic Fabry–Perot interferometric (EFPI) sensors,² intrinsic Fabry–Perot interferometric (IFPI) sensors,⁵ long-period fiber gratings,⁶ and combinations of these sensors¹ have been used in the field of structural health monitoring. High-temperature and embedded applications, however, impose significant limitations on the types of sensors that may be used. Optical-fiber sensors may be embedded in composite materials and structures with little material degradation. The embedded sensors must survive both the curing process and the environmental operating conditions.

In the work presented in this paper, high-performance carbon/bismaleimide (BMI) composite laminates are used. BMI-based composites are candidates for aerospace applications due to their superior strength and mechanical performance at elevated temperatures.⁷ They are also easier to process as compared to other high-temperature materials such as polyimides. The composite laminates are manufactured using an out-of-autoclave (OOA) process. OOA processing has been shown to be capable of fabricating high quality epoxy laminates, but OOA processing of BMI is relatively new.^8,9 Previous work had established that laboratory scale OOA cured BMI composites can have properties similar to that of autoclave cured composites.¹⁰ OOA BMI composites were also evaluated as part of a NASA Composites for Exploration program.^11,12

Composites in aerospace applications can be exposed to harsh service conditions for high strain and fluctuating temperature. Optical-fiber sensors have gained wide interest for in-situ monitoring of the composite laminates and the most common parameter to be monitored for evaluating the in-service condition of the composite laminates is strain.^1,2 Among various optical-fiber-based sensors, EFPI-based sensors are best suited for strain monitoring applications under high ambient temperatures. Bragg gratings and IFPI sensors have been demonstrated for strain applications over the years,^5,13 but are very sensitive toward the ambient temperature as well. EFPI-based optical-fiber sensors have a low sensitivity toward the ambient temperature when compared to their sensitivity toward the strain. In addition to strain monitoring, EFPI sensors have been used to estimate the size of damage under impact loading.¹⁴ Composite components used in airframes are also susceptible to barely visible impact damage (BVID), which are not easy to locate. Data from an array of fiber optic sensors can be used to locate the site of BVID, which can be of great importance in maintenance of composite structures.¹⁵ However, many types of EFPI sensors cannot be used at high temperatures due to limitations of bonding epoxy. They are also difficult to calibrate. A microcavity EFPI strain sensor, fabricated using femtosecond (fs)-laser micromachining, avoids these limitations.¹⁶ Cavity formation with fs-laser process yields stable structures that can withstand higher temperatures. Precise cavity dimensions can be used as compared to the traditional EFPI optical-fiber sensors or chemically etched, microcavity EFPI structures. Microcavity EFPI sensors with fs-laser fabrication have been shown to reliably operate up to 800°C.¹⁶

The microcavity EFPI strain sensor is applied for embedded, high-temperature operation to investigate a new composite material and test the sensor’s embeddability. The sensor has a very low sensitivity toward the ambient temperature as compared to its strain sensitivity thus making it an ideal choice for strain monitoring at high temperatures. The sensor performance is shown for an embedded application in BMI composite laminates. The microcavity sensor survived the curing process at 190.56°C and provided reliable strain measurements at operating temperatures up to 225°C. The glass transition temperature of the BMI resin limited the highest tested temperature limit for the embedded sensor. Measurement performance, strain transfer, and thermal behavior are discussed. Analytical work done using the OOA BMI laminates was limited in nature until now. The work presented in this paper offers a new insight into the strain and temperature interactions of the embedded sensors with the BMI and demonstrates the embeddability of the microcavity EFPI sensor in composite materials.

2.

Microcavity Extrinsic Fabry–Perot Interferometer Sensor Design and Instrumentation

The EFPI sensor consists of two reflective surfaces created by fabricating a cavity. Figures 1(a) and 1(b) show a traditional cavity-based EFPI sensor design and the microcavity EFPI design, respectively. Input light traveling in the core of the fiber reflects from the two walls of the cavity and the difference (induced by the cavity length change) between these two reflections, in turn, causes a wavelength shift in the original input signal. This wavelength shift is a measure of change in the cavity length. In a strain sensor with an air gap of length $d$, the gauge length is approximately the tube length $L$, and the measured strain is $\mathrm{\Delta }L/L=\mathrm{\Delta }d/d$.

Fig. 1

Sensors: (a) traditional tube-based EFPI and (b) microcavity EFPI.

Traditionally, the cavity in an optical-fiber sensor is formed using two pieces of optical fiber, aligned and bonded to a capillary tube using an epoxy adhesive.^17,18 The use of epoxy introduces a temperature limitation, e.g., the maximum temperature for Loctite epoxy is 225°C once cured. Moreover, the tube component causes the design to be bulkier and difficult to fabricate. The exact gauge length and the initial gap length are difficult to determine for the traditional design; hence, calibration is an issue.

The microcavity sensor has a low sensitivity to the thermal environment. Thermal expansion of the EFPI depends upon the thermo-optic coefficient and the coefficient of thermal expansion (CTE). Thermo-optic coefficient further depends upon the refractive index of the medium of the cavity (air). Thus, the low CTE of silica ($0.55\times {10}^{-6}/°\mathrm{C}$) makes the single-mode silica fiber an ideal candidate to be used for high-temperature applications.¹⁶

3.

Manufacturing

3.2.

Bismaleimide Composite Sample Fabrication

The host material for the embedded testing was carbon fiber/BMI composite (laminates). For results presented here, six-layer unidirectional laminates were fabricated ($12\mathrm{in}.\times 1\mathrm{in}.$) using IM7G/AR4550 prepreg by OOA process. The ply orientations of the six layers were ${[0\mathrm{deg}/90\mathrm{deg}/0\mathrm{deg}]}_{\mathrm{s}}$. A microcavity EFPI sensor was placed between the central layers. Figure 2 shows the arrangement of the BMI laminate layers and the sensor placement. The approximate location of the sensor was at [6 in., 0.5 in.]. A heat shrinking tube was used at the egress point to prevent damage to the optical fiber due to high stress concentrations. The prepreg layup was cured at 190.56°C for 2 h. The samples were then postcured at 200°C.

Fig. 2

Embedded sample layout with the sensor embedded in the middle.

4.

Experimentation

Mechanical and thermal tests were conducted on free sensor as well as sensor embedded in BMI composites. In case of embedded sensor, strain transfer was evaluated after green body cure at a curing temperature of 190.5°C and after a freestanding postcure at 200°C. Term green body¹⁹ is used for the fiber-reinforced polymers or ceramics that are held in shape by mechanical pressing. Details of testing parameters are shown in Table 1.

Table 1

Testing parameters for embedded and free sensors.

Figure 3 shows the instrumentation used for the EFPI sensor testing. A 100-nm (1520 to 1620 nm) broadband source (B&W Tek Inc.) with a resolution of 0.01 nm was used as input, a 3-dB coupler was used to send the signal to the sensor and receive the reflected signal back. This signal was then recorded using an optical spectrum analyzer (OSA). The wavelength shift in the return spectra is linearly dependent upon the strain, $\mathrm{\Delta }d/d$ where $d$ is the cavity length and $\mathrm{\Delta }d$ is the change in the cavity length resulted either from thermal expansion or applied strain.¹⁶ With the application of axial strain, the cavity length $d$ will change inducing a change in the central wavelength, $\lambda$, resulting in a wavelength shift, $\mathrm{\Delta }\lambda$. For a precise measurement, a sharp valley (or dip) was tracked for the wavelength shift. The OSA used was capable of storing the reflected spectra at regular intervals. A MATLAB code was then used for signal processing (in CSV format) and determining the wavelength shift for strain as well as temperature applications.

Fig. 3

The instrumentation setup for sensor testing and response monitoring.

4.1.

Temperature Response

To test the temperature response of the free sensor, the optical fiber was placed inside a box furnace (Lindberg/Blue M) and heated from room temperature to 800°C in steps of 50°C. The resultant wavelength shift in the spectrum was recorded.

The temperature response of the embedded sensor was evaluated in a similar manner. A sample with the embedded sensor was placed in the box furnace and heated to 225°C in steps of 25°C. The glass transition temperature (271°C) of the BMI resin limited the upper limit of the temperature tested. Table 2 lists some of the properties of the BMI resin used in the carbon-reinforced fiber laminates used for embedded sensor testing. Figure 4 shows the microscopic image of the embedded sensor in a tested sample. Note the quality of the interface between the BMI structure and the optical-fiber sensor.

Table 2

Properties of the BMI composite.

Parameter	Value
Sample size used for embedding the sensor	$12\mathrm{in}.\times 1\mathrm{in}.$ ($304.8\mathrm{mm}\times 25.4\mathrm{mm}$)
Glass transition temperature	271°C
Cure temperature	190.55°C
Tensile modulus	142 GPa
Major Poisson ratio (${\nu }_{12}$)	0.29

Fig. 4

Microscopic image of the embedded sensor.

4.2.

Strain Response

For the strain testing of the free sensor at room temperature, the sensor was fixed to translation stages (Newport) at both the ends. A certain amount of prestrain was applied in order to remove slack in the optical fiber. An axial strain was induced along the optical fiber/sensor axis resulting in a wavelength shift in the reflection spectra.

Tensile tests on embedded sensor samples were conducted on an Instron 5985 universal testing machine. A strain controlled test was performed, and the wavelength shift was recorded every $500\mu ϵ$. Figure 5 shows a composite sample with embedded sensor between the grips of the test equipment. A longitudinal split, i.e., sample failure can also be seen; optical-fiber sensor also failed when the BMI laminate sample failed. For testing the sensor’s strain performance at elevated temperatures, an environmental test chamber was used.

Fig. 5

INSTRON machine used for strain testing of the embedded sensor.

5.

Results and Discussions

In this section, the experimental results for temperature and strain responses of the sensor are presented and discussed. Results for testing in free and embedded environments are presented. Similar EFPI sensors were used to produce the results presented; fringe visibility of the sensors ranged from 12 to 20 dB and the cavity length was $35\mu \mathrm{m}$.

5.1.

Temperature Response

Figure 6 shows the wavelength shift and the thermal strain resulting from the increasing ambient temperature as observed by using the free and embedded EFPI sensor.

Fig. 6

Response of the embedded (in carbon fiber-reinforced composites) and free EFPI sensors toward the ambient temperature.

For the free sensor, the slope of the response was calculated to be $0.6\mathrm{pm}/°\mathrm{C}$ and CTE of the silica was calculated to be $0.715\times {10}^{-6}/°\mathrm{C}$ which is 1.3 times larger than that of the actual CTE. The slope and CTE calculated from the sensor embedded in BMI composites were calculated to be $1.7\mathrm{pm}/°\mathrm{C}$ and $1.615\times {10}^{-6}/°\mathrm{C}$, respectively. The composite CTE can be calculated to be $0.028\times {10}^{-6}/°\mathrm{C}$ using²⁰

Eq. (1)

$${\alpha }_{\mathrm{c}}=\frac{E_{\mathrm{f}}{V}_{\mathrm{f}}{\alpha }_{\mathrm{f}}+E_{\mathrm{m}}{V}_{\mathrm{m}}{\alpha }_{\mathrm{m}}}{E_{\mathrm{f}}{V}_{\mathrm{f}}+E_{\mathrm{m}}{V}_{\mathrm{m}}}\mathrm{.}$$

The terms in Eq. (1) are shown in Table 3.

Table 3

Properties used in Eq. (1).

Parameter	Value
Fiber elasticity, ${\mathbf{E}}_{\mathbf{f}}$	4830 MPa (material datasheet)
Matrix elasticity, ${\mathbf{E}}_{\mathbf{m}}$	48.26 MPa (transverse tensile strength of composite, material datasheet)
Fiber volume fraction, ${\mathbf{V}}_{\mathbf{f}}$	0.5766
Matrix volume fraction, ${\mathbf{V}}_{\mathbf{m}}$	0.4344
Composite fiber CTE, ${\mathbf{\alpha }}_{\mathbf{f}}$	$-0.6\times {10}^{-6}/°\mathrm{C}$ (material datasheet)
Matrix CTE, ${\mathbf{\alpha }}_{\mathbf{m}}$	$44\times {10}^{-6}/°\mathrm{C}$21

The composite CTE calculated using the embedded optical fiber was higher than the CTE calculated using Eq. (1), and about 2.3 times larger than the calculated CTE of silica. For the free sensor, the thermal strain resulting from the cavity expansion was $0.37\mu ϵ/°\mathrm{C}$ whereas the strain exerted on the embedded sensor was calculated to be $1.11\mu ϵ/°\mathrm{C}$. Larger strain in the case of the embedded sensor is due to the effect of the interfacial bond between optical fiber and the surrounding composite material. The sensor is not very sensitive toward the temperature, but it does respond very well toward the thermal strain of the host structure. For both the embedded and free sensors, the correlation coefficient was 0.99, demonstrating a highly linear response.

5.2.

Strain Response

Strain response of the free and the embedded sensors at room temperature is presented in Fig. 7. The response for both the sensors toward strain is highly linear with a correlation coefficient of 0.99. The free sensor had a strain transfer of $0.98\mu ϵ/\mu ϵ$ indicating a good agreement between measured strain and applied strain. For the free sensor, the strain was applied until the fusion joint broke loose from the cavity; it was verified under a microscope that the cavity broke at the fusion joint. The breaking point for the EFPI was observed to be at $3800\mu ϵ$.

Fig. 7

Measured strain response of the embedded and the free sensors at room temperature.

The strain response of the embedded sensor was evaluated after green body cure and a postcure cycle. The strain transfer for the embedded sensor after green body was $38\mu ϵ/\mu ϵ$, while the strain transfer of the embedded sensor after postcuring was $1.28\mu ϵ/\mu ϵ$. The embedded sensors in green samples underpredict the strain, while the sensors in postcured samples overpredict the applied strain levels. Postcuring results in a change in degree of cure and crosslinking in thermosetting resins. Residual stresses are also developed due to cure shrinkage and the difference in CTE between the carbon fibers and matrix. These can result in a difference in strain response between a green body and a postcured sample.

The sensor was also used to observe the strain response in an embedded environment at an elevated temperature of 104.4°C. The results are presented in Fig. 8. The ambient temperature of 104.4°C yielded a response with a slightly higher slope. Because of the low sensitivity of the sensor toward the temperature, the difference in the slopes was $1\mathrm{pm}/\mu ϵ$. For electrical strain gauges and strain applicator, strain on the order of $100\mu ϵ$ is considered as noise. As demonstrated in the Fig. 8, apparent strain added by a temperature raise of 104.4°C is less than $200\mu ϵ$, which is acceptable for most regularly monitored civil and aerospace infrastructures. Due to low temperature sensitivity of the microcavity sensor, cross-sensitivity does not pose a high risk for previously mentioned applications. If the temperature variations and strain variations are larger, cross-sensitivity and material dependent strain transfer should be considered.

Fig. 8

Measured strain response of the embedded sensor at room temperature and at 104.4°C.

6.

Conclusions

A microcavity EFPI fiber optic sensor for the measurement of strain in BMI composites is presented in this paper. The silica material of the optical-fiber sensor and the thermal stability resulting from the fs-laser fabrication enable the sensor to have excellent properties for embedded applications and for high-temperature applications. Although the sensor itself has been shown to handle temperatures up to 800°C, it was limited in this work to the glass transition temperature (271°C) of the BMI composite. A single fusion joint allows for better structural integrity and ease-of-fabrication over tube-encapsulated designs. The inherent properties of the fused silica and the EFPI cavity structure enable the sensor to be used as an efficient strain sensor. The responses of a free sensor as well as a sensor embedded in OOA cured carbon/BMI composite laminates were evaluated. The microcavity EFPI sensor survived the composite cure process.

Temperature and strain responses are demonstrated for the free and the embedded sensors. The sensors provided reliable thermal strain measurements at operating temperatures up to 225°C. The sensors demonstrated a highly linear response toward temperature and strain; the coefficient of linearity was 0.98 or higher for both responses. The temperature sensitivity of the free and embedded sensors was 0.6 and $1.7\mathrm{pm}/°\mathrm{C}$, respectively. The resulting cavity expansion exerted a thermal strain amounting to $0.37\mu ϵ/°\mathrm{C}$ for the free sensor and $1.11\mu ϵ/°\mathrm{C}$ for the embedded sensor. There were expected differences in the slopes of the responses and the CTE values calculated for silica were higher than the actual CTE of the silica. However, the sensor responds linearly toward the temperature increase and the wavelength shift is small at temperatures as high as 800°C as compared to its strain sensitivity. The fitting linearity for both the temperature and strain response was 0.99.

Strain response of the free and the embedded sensors was also evaluated. The strain sensitivity of the free sensor was 1.5 and $0.6\mathrm{pm}/\mu ϵ$ for the embedded sensor for green samples. Incorporating a freestanding postcure during manufacturing of embedded samples resulted in an increase in strain transfer from 0.38 to $1.28\mu ϵ/\mu ϵ$. This improvement can be a result of increased degree of cure and a change in residual stresses in the composite. Ongoing work is examining the repeatability of the embedded sensor performance. Due to some little material properties (of BMI composites), more work needs to be done in order to explain conclusively the residual strain behavior in the data obtained from repeatability tests.

An ease-of-fabrication, small size, embedding capability, low temperature, and high strain sensitivity for carbon fiber composite laminates makes this sensor a very good candidate for strain monitoring applications in high ambient temperatures. The work also provides new insight into the behavior of BMI composite laminates, especially the strain transfer interaction between the sensors and the laminates. The strain endured by commonly monitored structures such as airframes, bridges, and so on, is above $500\mu ϵ$ and the temperature variation for regular operations is not more than a few tens of degrees Celsius. Due to low temperature sensitivity of the microcavity sensor, cross-sensitivity does not pose a high risk for previously mentioned applications. If the temperature variations and strain variations are larger, cross-sensitivity and material-dependent strain transfer should be considered.