ABSTRACT

This paper describes tests conducted using lock-in and pulse thermography on carbon fibre reinforced composites with implanted delamination defects in varying in size from 1mm to 11 mm in diameter and of depths 0.3mm to 3.0mm.

Detection sensitivity charts were produced for all procedures which showed the reduction in sensitivity with depth below the surface. A 1 mm defect was detectable at 0.3 mm depth. All techniques were inferior to c-scan ultrasound but this latter procedure requires considerably greater testing time and is not a remote procedure. The next most sensitive procedures were lock-in thermography and then laser shearography.

Theoretical and finite element evaluations by Bai, W on the lock-in reflection procedure showed good correlation with the experimental data. The defect contrast versus frequency is therefore predictable for any defect at any depth in any material providing the latter’s thermal properties are known. Therefore it is possible to predict the optimum frequency for any defect at any depth prior to a test and determine the contrast to be expected. Also the predicted blind frequency could be avoided. If only one frequency is selected for scanning the predicted reductions in contrast of other defects in the specimen at different depths can obtained thus enabling the user to determine if this one frequency is suitable for the test. The predictions remove the current lock-in procedure of trial and error to determine the optimum frequency.

Lock-in thermographic procedures using a micro lens showed good sensitivity to delamination defects between Si and Copper layers as found in integrated circuit package specimens. Defects of about 100 micron in diameter were detectable.

INTRODUCTION

Carbon fibre composites are now fairly widely used in civilian and military aircraft structures. Common defects found in these materials are delaminations and the their presence will lead to structural weaknesses which could lead failure of the airframe structures. It is important to develop effective non-destructive testing procedures to identify these defects and increase the safety of aircraft travel. This paper describes tests conducted with various non-destructive testing procedures and gives some comparison between the procedures.

Lock-in thermography direct a sinusoidal heat wave to the surface of the specimen. At the surface the wave then propagates into the specimen and is reflected by internal defects. The reflected wave interferes with the incoming waves. The interference pattern has characteristics such as phase delays which can be measured with the thermographic camera system. A theoretical model predicting phase delays is also presented in this paper.

Digital laser shearography consists of causing two coherent images of the object surface to be formed simultaneously, but with one shifted with respect to the other and then making a recording. After some deformation has taken place, a second recording is made in the same way in a double exposure manner. By optical processing, fringes are observed which can then indicate the presence of internal defects like delaminations. The surface above the defect needs to be deformed e.g. heating the specimen surface which causes a bulge above the defective area.

2. EXPERIMENTAL DETAILS

2.1. EQUIPMENT
2.1.1 LOCK-IN THERMOGRAPHY EQUIPMENT
The lock-in thermography system used in this work is an AGEMA Thermovision 900 system. It consists of an infrared camera, a system controller (a computer), a heat source (1 kW-halogen lamp with an infrared filter) and a lock-in module. The heat source is driven by the lock-in module that is controlled by the system controller. As the heat source is modulated digitally, the intensity of the heat source is an approximate sine function. The lock-in module controls the inspection frequency which ranges from 0.0037Hz to 3.75Hz.

During the inspection, the IR camera is used to record the oscillating surface temperature of the inspected object. The image recording is synchronized with the modulation frequency and the IR camera takes 4 images within one cycle (Figure 1). The lock-in system gets 4 signal values S1, S2, S3 and S4 in every pixel of the image the indices refer to recording time. From these values the system calculates a phase (F) image according to the following basic equation:

(1)

Fig 1: Signal acquisition during thermal wave cycle.

2.1.2 LASER SHEAROGRAPHY
The optical set-up for the speckle shearing interferometer is shown in Figure 2.

Fig 2: A schematic diagram of the shearing interferometer.

A single mode and single frequency HeNe laser of 60 mW output power with an emission wavelength of 632 nm is used as the coherent source in the set-up. The laser light illuminates the object to be measured via a single mode fibre. The diffusely scattered light from the object passes through a beam splitter and is imaged at the plane of the CCD camera by the two mirrors, which are orthogonally placed at an equal distance of 15 mm from the beam splitter. The mirror 1 acts as the shearing mirror and the amount of shear can be adjusted by shifting the angle of the mirror. After passing through the prism, the two laterally sheared wave fronts interfere and overlap each other at the plane of the camera and produce the resultant speckle pattern. The light intensity of the speckle pattern is converted to an electric video signal and this is sent ot frame grabber board where it is sampled to yield a digital image. The whole optical set-up was mounted on a vibration isolation table.

2.2 SPECIMEN
A specimen was fabricated with 30 layers of Carbon Epoxy laminates (Figure 3).

Fig 3: Carbon fibre reinforced composite specimen with implanted delamination defects in.

The specimen had implanted defects (three layers of Teflon film) of various sizes at different layers. The specimen was cured under high pressure and high temperature in an autoclave. The implanted defects consist of five layers: three layers of teflon film and two layers of epoxy resin. The three Teflon layers have the same thickness of 0.06 mm and the two resin layers have the same thickness of 0.01 mm. The total thickness of the specimen is about 4.2 mm. The diameters of defects were 1, 2, 6, 8 and 11 mm respectively. The defects with the same diameter were assigned at different depths below the surface. The depths were 0.28 mm, 0.56 mm, 0.84 mm, 1.12 mm, 1.4mm, 2.1 mm and 2.8 mm respectively. The dimension of the specimen was 300x300x4.2 mm. The thermal properties of the involved materials are listed in Table. 1. The parameters were substituted into the photothermal model described in the Appendix to predict the phase differences between the defective areas and non-defective areas. The specimen was fabricated under high pressure (7 bars) and the epoxy was melted at high temperature (120 ºC).

Fig 4: The geometric scheme of cross-section of the implanted defects.

	Density (kg m-3)	Thermal conductivity (W m-1 K-1)	Specific heat (kJ kg-1 K-1)
CFRP	1600	0.67	1200
Teflon (FEP)	2150	0.209	1100
Epoxy resin	1300	0.20	1700
Table 1: Thermal properties of CFRP, Teflon and Epoxy

2.3 EXPERIMENTAL PROCEDURE FOR LOCK-IN THERMOGRAPHY
The experiments were performed in a large room and the room temperature was about 23ºC. The infrared camera of the lock-in system was positioned at 0.6 metre and perpendicular to the test specimen. The heat source was positioned at 0.5 metro from the specimen. The lens was then focused on the specimen. The specimen was heated to about 80° C with maximum power to shorten the time required to reach the steady state. Subsequently, the specimen was heated periodically at the selected frequency. A steady state was established when the variation of the peak value of the oscillating temperature of the specimen surface was less than 0.5° C. The thermal wave data of the object surface was then collected to produce phase images. The specimen was detected at different modulation frequencies ranging from 0.0037 Hz to 0.93 Hz.

In the room, there was no obvious airflow. Hence the convection occurred at the specimen surface was approximately considered as free convection. In steady state, the average temperature of the front surface was about 75ºC and the average temperature of the rear surface was about 71ºC. In this situation, convection and radiation occurred simultaneously at the surface of the specimen.

3. RESULTS AND DISCUSSION

3.1 LOCK-IN THERMOGRAPHY - EFFECT OF MODULATION FREQUENCY ON PHASE DIFFERENCE PRODUCED BY DEFECTS
Figure 5 shows some phase images (left) and phase profile plots (right) of the 11 mm defects of the specimen. In the phase images, each pixel represents a phase value related to the phase difference between the oscillating surface temperature and the heat source. From left to right, the depths of the defects are 1.4 mm, 1.12 mm, 0.84 mm, 0.56 mm and 0.28 mm, respectively. The right side images of Figure 4 plot the phase values of a line, which crosses the central points of the 11 mm defect images in the left side phase image. It can be seen that there are phase differences between the defective areas and non-defective areas. Hence the subsurface defects can be detected.

(a) Frequency = 0.0037 Hz

(b) Frequency = 0.0146 Hz

(c) Frequency = 0.0296 Hz

Fig 5: Thermograms and phase profile plots of 11 mm defects.

Figure 6 (a) - (c) show the differences between the central point phase values of the 11 mm defects and the average phase values of the non-defective areas. The figures plot both the results obtained from the photothermal model (See Appendix) and the experiment. The figures indicate that the theoretical and experimental results have similar trends. The phase differences obtained experimentally are larger than that obtained theoretically. This is because that there are some air holes in the epoxy resin between the Teflon films. During the fabrication of the specimen, although the autoclave was maintained at high vacuum and the specimen was under high pressure, the air between the Teflon films could not completely escape from the melt epoxy resin. The effective thermal conductivity and density of the epoxy resin with air bubbles are lower than that of pure epoxy.

(a) Defect depth=0.56 mm

(b) Defect depth=0.84 mm

(c) Defect depth=1.4 mm

Fig 6: Phase differences between defective areas and non-defective areas produced by 11 mm defects (theoretical results and experimental results for defects with different depths).

From Figure 6, it can be observed that there exists a frequency for a specific defect at a certain depth such that there is no phase difference or a very small phase difference produced by the defect. Therefore the defect cannot be detected at this frequency, which can be called ‘the blind frequency’. There are another two special frequencies at which a certain defect produces maximum positive or negative phase differences. The two frequencies are named as the ‘optimum frequencies’. Figure 6 also indicates that the blind frequencies and optimum frequencies change with the defect depths. The deeper the defect depth, the smaller the blind frequency and optimum frequency are.

In practical inspection, the blind frequency should be avoided and the optimum frequencies should be selected. The ‘blind frequencies’ and ‘optimum frequencies’ obtained theoretically are very close to those obtained experimentally. Also, when the size of a laminate shape defect is much larger than the thermal diffusion length, the defect size does not affect the blind frequency and optimum frequency. Hence, the photothermal model described in the Appendix can be used to predict the optimum frequency and blind frequency.

It can also be observed from Figure 6 that the maximum phase differences produced by the defects with the same size decrease with the defect depth. Both the experimental result and theoretical results indicate that the maximum difference produced by an 11-mm defect, which is at a depth of 1.4 mm below the surface, is less than 5 degrees. When the defect depth increases further, the defect cannot be detected.

Finite Element Modeling was also conducted to support the theoretical model. Figure 7 shows the close correlation between the two procedures so FEM provides an effective modeling procedure for Lock-in thermography.

Fig 7: Comparison between the theoretical and finite element plot.Defect depth=1.4 mm (ref. Figure 7 (c))

3.2 SHEAROGRAPHY
Figure 8 shows a typical shearography result showing the distinctive bulls eye characteristics above a defect. The deformation was caused by a heat pulse applied to the surface of the specimen.

Fig 8: The shearographic fringe characteristics of the 11 mm diameter defect 0.28 mm from the surface of the specimen.

3.3 INTEGRATED CIRCUIT PACKAGE INSPECTION
Figure 9 shows an image of voids between the interface of a layer of silicon and copper as may be found in an integrated circuit package. The area covered in Figure 9(a) is approximately 3mm by 5mm on the surface of the Silicon. The light areas indicate the small voids which can be seen.

a)	b)
Fig 9: (a) Lock-in thermographic image of voids between a Si and Copper layer (b) Contrast profile across a void across the line shown in (a).

3.4 A COMPARISON OF VARIOUS NON-DESTRUCTIVE TESTING TECHNIQUES
Figure 10 shows a comparison of various non-destructive testing techniques on the specimen. The chart shows an approximate experimental comparison only based on whether a defect was detectable or not. The smaller and deeper a defect is, the more difficult it is to detect. Ultrasonic "C" scan is still the most sensitive technique followed lockin phase thermography and then shearography. The reflection results refer to a procedure which applied a heat pulse to the specimen surface and the defect detected by the heat profile above it

Fig 10: A comparison of various non-destructive testing techniques.

4. CONCLUSIONS

Generally the order of defect sensitivity found was "C" scan ultrasonics, lock-in thermography and then shearography. However for near surface defects and for testing thin materials, the latter two techniques are approaching ultrasound's sensitivity. They have the additional advantage of being remote faster techniques and also would be able to test specimens with some slight contours, as in aircraft structures, more effectively.

A theoretical model, finite element modeling and experimental results of lock-in phase thermography showed reasonable close correlation with each other,

The model developed for lock-in phase thermography enables the the optimum frequency for an application to be predicted (i.e. the frequency giving the maximum defect contrast). The blind frequency when no defect can be detected also be detected. This removes the current trial and error process involved to determine the optimum frequency for testing.

APPENDIX: THEORY FOR LOCKIN PHASE THERMOGRAPHY [2]

An infinitely large multi-layer plate with finite thickness is considered. The plate consists of n thermally different media 1 to n and it is surrounded by air. The different media are separated by planar interfaces (at x = x_1, x₂...x_n-1). The thermal properties of the system are labeled: k_i thermal conductivity, r_i density, c_i specific heat and a_i thermal diffusivity with i=1, 2..., n. A modulated planar heat source is applied to the free surface of medium 1. Medium 1 is the opaque front surface of the plate. The distributed heat flux of the heat source is (Q₀/2)[1+cos(wt)], where Q₀ is the intensity of the heat source, w is the angular modulation frequency and t is time.

In each layer, the temperature distribution obeys the general heat transfer equation

(A1)

where T_i is the temperature in medium i; the parameter x is perpendicular distance from the front surface and at the front surface x equals to 0.

The front surface is heated by the heat source. At both the front surface and the rear surface convection and radiation occur due to the surface temperature is higher than the environment temperature. Hence, at the two surfaces the boundary conditions are:

(A2)

(A3)

T_f = the temperature of the front surface
T_r = the temperature of the rear surface
T_¥ = the air temperature measured at a point far away from the surfaces
h_f = the heat transfer coefficient of the front surface
h_r = the heat transfer coefficient of the rear surface

At the interfaces between two media, the heat flux is continuous and there exist a thermal contact resistance between the two media. As such, the boundary conditions are:

(A4a)

(A4b)

where, R_i,i+1 is the thermal contact resistance between medium i and medium i+1.

The heat flux is divided into two parts, Q₀/2 and (Q₀/2)exp(jwt), which produce a dc temperature increase and an ac thermal modulation respectively. When a steady state is reached, the solution of Equation (1A) has the form:

(A5)

where T_di(x) and T_ai(x)exp(jwt) are the dc component and the ac component of the temperature in layer i, respectively. In practical inspection, the variation of surface temperature is seldom larger than 10ºC. Hence, it is reasonable to assume the heat transfer coefficient as a constant.

As the dc component of the temperature does not change with time, it obeys the following equation:

(A6)

subject to the boundary conditions:

(A7)

(A8)

(A9a)

(A9b)

T_df = the dc temperature component of the front surface
T_dr = the dc temperature component of the rear surface

Because only the ac temperature components are used in lock-in thermography, the primar interest is in the ac component. Substituting Equation (5A) and Equation (6A) into Equation (1A) and omitting the dc component, the following equation is obtained:

(A10)

and the boundary conditions of equation are:

(A11)

(A12)

(A13a)

(A13b)

where

T_af = the spatial dependence of the ac temperature component of the front surface
T_ar = the spatial dependence of the ac temperature component of the rear surface

The solution of Equation 10A has the following form:

(A14)

Considering the boundary conditions Equation (11A) to Equation (13A), A_i and B_i can be obtained from the following equation:

(A15)

At the heated surface (front surface), the spatial dependence of ac temperature component is:

(A16)

(A17)

Here, A₁+B₁ is a complex quantity. F is the phase angle of A₁+B₁, i.e. the phase difference between the surface temperature and heat source.

For a plate that only has one layer, Equation (15A) becomes

(A18)

(A19)

(A20)

and the phase difference between the surface temperature and the heat source is:

(A21)

This model can be used to predict the phase value caused by a laminate defect. In a plate the area without defect can be considered as a homogeneous plate and the phase value of non-defective area (F_{non-defective} ) can be calculated with Equation (21A). On the other hand, the area with a laminate defect can be considered as a multi-layer structure. The internal layers can be air gaps or other inclusions. The phase value of the defective area (F_defective ) can be calculated with Equation (17A). The phase difference between the defective area and non-defective area is:

(A22)

h_f, the heat transfer coefficient of the front surface and h_r, the heat transfer coefficient of the rear surface are assumed to be equal for the specimen in this paper and its determination is described in reference (3).

REFERENCES

1. Wong, B. S., Tui, C. G., Low, B. S. and Bai, W. "Thermographic and Ultrasonic Evaluation of Composite Materials", Insight, UK, Vol. 41, No. 8, pp. 504-509, 1999.
2. Bai. W, and Wong, B. S. "Evaluation of defects in composite plates under convective environments using lock-in thermography", Measurement Science and Technology, UK, Vol. 12, pp. 142-150, 2000.
3. Bai. W, and Wong, B. S. "Photothermal models for lock-in thermographic evaluation of plates with finite thickness under convection conditions", Journal of Applied Physics, USA, Vol. 89, No. 6, pp. 3275-3282, 2001.

© NDT.net - info@ndt.net

|Top|