A composite is a structure made of materials with different physical properties. When combined, a material with characteristics different from the individual components is obtained. The properties of the composite structure depend on the matrix and the reinforcement materials and their setups. Polymer matrix composites (PMC) play important roles in modern industry because of their high stiffness, corrosion resistance, and strength to weight ratio [1
]. Increasing the number of such structures in aerospace, construction, and automotive applications, as well as in lightweight structures in the wind energy industry, enforces continuous monitoring of their condition. Due to high anisotropy, nondestructive inspection of layered composite materials is a much more complicated process than the evaluation of homogenous, (mostly metallic) structures [2
]. Several nondestructive methods are utilized in this case (ultrasonics, shearography, tap testing, acoustic emission, digital radiography, infrared thermography) but none of them gives a full description of the evaluated structures. Thus, further development of NDT techniques should be studied.
The use of terahertz inspection of dielectric or semiconductive structures is attracting increasing attention in the research community and industry. Recently, various applications have been demonstrated by many research groups. Continuous-wave imaging (CW), frequency domain measurement using a vector network analyzer (VNA), and pulsed time domain spectroscopy (TDS) are utilized in this frequency range [9
]. The first group of applications refers to spectroscopic evaluation of materials or potential inclusions compositions, i.e., polymer evaluation [16
], drugs identification [19
], explosives detection [19
], biological tissue identification in the case of biotechnological and medical applications [21
], hydration/moisture level determination [23
], polymerization state monitoring [25
], chemical mixtures evaluation [26
], and determination of tea geographical origin [27
]. The second group of applications refers to observations of the inner structure of an object under test (OUT), similar to ultrasonic testing or the pulsed ground penetrating radar (GPR) technique. In this case homogenous or layered dielectric materials are examined in order to detect irregularities or defects [14
]. Non-conductive coatings of dielectric or conductive materials, as well, as layers thicknesses can be also evaluated [28
]. In the case of integrated circuits the inner structure is monitored [31
]. TDS is also utilized in order to evaluate the integrity of layered tablet structures. Another group of applications refers to terahertz tomographic imaging [32
]. Tomography refers to the cross-sectional imaging of an object by measuring transmitted or/and reflected waves (ray). There are various terahertz imaging options: time of flight reflection tomography (THz ToFRT), diffraction tomography (THz DT), computed tomography (THz CT), tomography with a binary lens, and digital holography.
Continuous wave terahertz imaging [37
] is a generally faster and less complicated technique compared to TDS. In this case the source (typically horn antenna, photoconductive antenna or far infrared laser) illuminates object under test and detector (photoconductive, pyroelectric, or bolometric) or a matrix of detectors receive the electromagnetic wave. In most cases this technique provides power or intensity information and phase data is not available. Higher power of sources and utilization of detector matrices enables inspection of thicker and bigger structures in comparison to TDS.
Main advantages associated with terahertz imaging technique are:
non-contact measurement in reflection and transmission arrangement;
inner structure and spectral information is obtainable; and
fraction of a millimeter resolution.
The main disadvantages of terahertz inspection are:
low power of THz emitters;
low speed of examination (need of raster scanning in case of VNA and TDS solutions);
restriction to nonconductive materials (because of high frequency and skin effect); and
high cost of VNA and TDS solutions.
A pulsed terahertz method seems to be a good candidate for layered PMC inspection. It is based on picosecond electromagnetic pulses interacting with evaluated structure. Differences of dielectric parameters enables the detection of particular layers in a layered material and any defect which disturbs the distribution of the refractive index, e.g., inclusion, delamination, void, material inhomogeneities (fiber/matrix distribution), and internal interfaces between layers (in layered structures).
In the case of multilayered structures, only layers close to the surface can be effectively detected. The response of deeper layers is averaged because of multiple reflections. In this paper a new inspection algorithm based on double-sided measurement, acquired signal deconvolution, and data combining based on weighting functions is proposed. In order to verify the application of algorithm stress subjected glass fiber reinforced polymer (GFRP) was evaluated. Obtained results enabled detection and detailed analysis of delaminations introduced by stress treatment and proved the applicability of the proposed algorithm.
3. Results and Discussion
The proposed algorithm was experimentally validated. A double-sided inspection of a polymer composite was carried out. During the measurements a commercially-available material which consists of 30 plies of glass fiber fabric with regularly-oriented fibers subjected to stress forces below 140 MPa was utilized. The photography of the exemplary specimen is presented in Figure 9
. After applying mechanical excitation, several delaminations were created on various depths of the layered structure and both surfaces were perforated. Material in this condition has significantly worse mechanical properties.
The exemplary A-scan signals gmCH1
) and −gmCH2
) obtained using the proposed double-sided inspection and signal processing are presented in Figure 10
. One can observe constructive interference of FSRCH1
and defect/layer reflection pulses, respectively. Similarly, B-scan signals obtained during a single line scan along the sample are presented in Figure 11
(deconvolved) and Figure 12
(channel 2 signal time-reversed). In each case BSR pulse has smaller amplitude and frequency content in comparison with FSR pulse. This is caused by attenuation of the pulse energy during propagation through multilayered structures and defocusing. As mentioned before, layers’ reflection pulses (in the case of layers close to back surfaces) are blurred, as well. Based on Figure 12
, one can see that gmCH1
) and gmCH2
) signals contain complementary information about the layers.
The B-scan signal combined using the proposed algorithm is presented in Figure 13
. In comparison to the results of standard, single-side inspection (shown in Figure 11
), here both material-air interfaces have “sharp” indications and the layers’ response is noticeable over the whole cross-section of the examined structure (with maximum values in the vicinity of material boundaries). Moreover, the pulses caused by reflection from defects (mainly delaminations) are detectable even on the background of the layers’ responses.
In this work an improvement of a relatively new nondestructive testing technique—pulsed terahertz inspection—is presented. An algorithm of double-sided inspection and obtained data treatment was proposed in order to preserve information about the layer arrangement in a multilayered composite structures. The reasons for the loss of this information in the case of the standard pulsed THz inspection technique were briefly characterized. The proposed algorithm is based on double-sided reflection measurement, deconvolution, and weighted summing. The weighting functions were obtained as envelopes of analytically-calculated signals. In this step some knowledge about examined structure is needed (material type—permittivity, thickness, number of layers), but even if not all of this information is available—a normalized weighting functions wnCH1(t) and wnCH2(t) can be calculated based on approximation Equation (2) and measured signal, without application of analytical model. It is possible, because the character of the layer reflection response (exponential decay) is similar in the case of all multilayered materials as long as there is one type of layer within the whole thickness.
The algorithm was validated using glass fiber-reinforced composite consisting of 30 layers and subjected to stress. As was expected, the application of the proposed algorithm enabled the preservation of weak signals of layer reflections in the case of layers situated over opposite sides of the examined structure. Retaining the layers’ reflection response has no negative influence on defects’ responses and detection. In the case of very thick or lossy samples, the problem with BSR pulse detection may appear, which can prevent proper matching of gmCH1
signals. Additionally visible layer response (Figure 6
) can be attenuated before reaching the central area of the material cross-section. In this case, in the center of the combined signal gcmb
) the amplitude and SNR are very low.
An application of proposed technique is possible only if there is an access to both sides of examined structure (restriction as in case of transmission inspection) and two channel measurements are available. If the latter condition is not fulfilled, inspection can be performed with two separate single-side measurements, but it doubles the examination time. Moreover, the proposed technique can be utilized for nonconductive materials, like polymer composites reinforced with glass, basalt, Kevlar, and natural fibers. Thus, polymer, ceramic or composite coatings of metallic objects cannot be examined using the proposed algorithm.