Narrow band photoacoustic lamb wave generation for nondestructive testing using candle soot nanoparticle patches

Industrial facilities, such as nuclear power plants and chemical plants, contain numerous metallic pipe lines and pressure vessels that require continuous health monitoring for instant detection of structural flaws since their failure can result in severe collateral destruction, environmental damage, and great human cost such as genetic diseases or cancer. Therefore, it is necessary to develop nondestructive testing and evaluation (NDT&E) methods to assess the reliability of pipe structures and components. The generation of ultrasonic surface waves with lasers has become a useful noncontact NDT&E tool.^1–3 This technique provides a very flexible and simple method for the noncontact and remote generation of ultrasonic surface waves. So, this method has advantages of temperature-independent measurement and a wide range of monitoring areas by easily changing the position of devices. Also, the laser-based NDT method has the advantage that various wave shapes can be easily produced by changing the shape of the beam illuminating the target.⁴ However, the laser ultrasonic method has a limitation that the illuminating power on the structures should be carefully adjusted. At high energy densities, typically of the order 10¹²–10¹⁴ W/m² on metals, a pulsed laser beam will ablate the surface of structures, which causes some damage on the surface.^5,6

In addition, typical laser generated surface waves have a broad frequency bandwidth.⁷ A broad bandwidth has advantages in laboratory investigations because much information can be gleaned from a single acoustic pulse. However, when complex interactions occur between a pulse and the material in which it propagates, the interactions are often frequency dependent. When this occurs, the signal analysis procedures required to extract useful information from a broadband signal become time consuming and computationally intensive. Also, since the energy is distributed throughout all the generated frequencies, the amplitude of individual frequencies is low and the signal to noise ratio (SNR) is reduced over the whole spectrum.³ Therefore, several laser based methods for generating narrow band ultrasound waves and reinforcing the acoustic amplitude in a particular direction of propagation have been developed using array techniques, including slit masks,^8,9 lenticular arrays,¹⁰ optical diffraction gratings,¹¹ multiple lasers,¹² and interference patterns.^13–15 Of these techniques, using a line-arrayed slit to mask an expanded laser beam is the most effective way to produce line arrayed illumination sources to excite surface waves with narrowband frequencies. McKie et al.¹⁰ developed a model in order to help predict the nature of a generated acoustic signal, which would arise as a result of varying the pulse duration of a laser-generated line array. Huang et al.¹¹ provided a brief theoretical principle of narrow-band signal generation using a laser line array source in the thermoelastic regime. They demonstrated that the generation of narrow-band surface waves could be controlled by adjusting the line array parameters. Lee and Lee⁹ proposed a noncontact system for defect localization by using an ultrasonic guided wave from the line array laser source and a dual air-coupled transducer.

However, a line-arrayed slit method is also limited by direct laser illumination on the structure which causes the ablating effect on the structure to generate a sufficient amplitude of surface waves.⁴ To overcome the ablating method, a line-arrayed patch made with a candle soot nanoparticle-polydimethylsiloxane (CSNP-PDMS) composite was adopted and characterized in this paper. Candle soot (CS) is a type of carbon nanoparticle and has been attracting increasing attention in applications such as in optoelectronic devices and ultrasound transducers.¹⁶ Each CSNP is about 30–40 nm in diameter. CSNPs can provide an efficient light absorption and a high energy conversion coefficient (4.41 × 10⁻³) in a broad frequency range (21 MHz). Moreover, CSNPs can be deposited on substrates with any size and shape due to a simple process for fabrication.

Therefore, in this paper, we propose a hybrid method to combine the merits of a line-arrayed slit method and the carbon soot CSNP-PDMS patch in order to generate a surface wave with a selective range of frequencies and high amplitudes while avoiding the ablating effect.

Lamb waves are guided surface acoustic waves propagating in thin plates, which have free traction forces at the top and bottom surfaces.¹⁷ In brief, the solution for Lamb wave propagation can be expressed in terms of symmetric modes represented by cosine (even) functions and antisymmetric modes given by sine (odd) functions, which is given as follows:⁶ 

where $p2=(w2/cL2−k2), q2=(w2/cT2−k2),$h is the half-thickness of the plate, k is the wave number, $w$ is the angular frequency, and C_L and C_T are the longitudinal and transverse wave velocities, respectively.

A Q-switched Nd:YAG pulsed laser was used to generate Lamb waves in aluminum plates. The mechanism of ultrasound waves generated by the pulsed laser can be explained by the thermoelastic regime.^18–20 When a solid surface of a target structure is illuminated by a high-power pulsed laser, a localized temperature gradient is formed at an infinitesimal area of the surface. Then, the temperature gradient induces elastic expansion and compression over the illuminated area, acting as an ultrasound source. Finally, the laser-generated Lamb wave propagates in the target structure as an elastic wave. The most popular method for the photoacoustic Lamb wave generation is the point or circle source illumination. However, it has a low directivity due to the omnidirectional feature of the laser source. The line source type has been proposed to improve the low directivity of the point source type.²¹ However, both types have a wide frequency bandwidth, and dispersion can then complicate signal processing. The line-arrayed source type, also referred to as the wavelength matching method, has been developed to generate a single wavelength Lamb wave with a narrow bandwidth.²² In this method, a higher concentration of wave energy results in better defect-detection ability and spatial resolution.

The line-arrayed patch shown in Fig. 1(a) was used to achieve linearly arrayed light illumination that acts as the line-arrayed source of ultrasonic guided waves on the target surface. This linear array source dominantly generates an axial longitudinal mode that propagates mainly along the axial direction of the pipe. The spacing of the array (Ds) coincided with the wavelength (λ) of the guided wave that was generated. Therefore, the specific mode required to be generated can be selected by simply changing the array spacing. Figure 1(b) shows the dispersion curves for the phase velocity (Cp) of Lamb wave modes in a 1 mm thick aluminum plate, where S is the symmetric and A is the asymmetric mode. Using Eq. (3), the diagonal line with a slope of D_s/t is illustrated in the dispersion curves in Fig. 1(b). The active modes of guided waves were determined by the intersectional points between the diagonal line and the dispersion curves.^4,9 Therefore, various modes with different frequencies were generated simultaneously,

There are several considerations for designing the CSNP-PDMS patch for NDT applications: (1) target frequency of the receiver transducer, (2) target Lamb wave mode, (3) duty ratio (D_s/w), and (4) number of arrays (N). A plate type aluminum nitride (AIN) single crystal (10 × 10 × 0.5 mm³) was used as the receiving transducer in this work due to the high enough sensing efficiency, i.e., piezoelectric voltage constant (g), compared to the PZT ceramic material.^23,24 The lateral mode resonance frequency of the transducer was about 600 kHz, so the target phase velocity of the A₀ mode was approximately 2 km/s, and the slope (D_s/t) and intersectional point were determined as shown in Fig. 1(b). In the ultrasound NDT applications, the A₀ mode was usually selected as an active Lamb wave mode because the asymmetric mode of the Lamb wave has out-of-plane displacement and exhibits higher sensing ability for the defect on the structure than the symmetric mode of the Lamb wave.⁴ In the line-arrayed laser source system, the harmonic frequency component is excited simultaneously with the fundamental frequency wave, and its magnitude is dependent on the duty ratio. In order to suppress the second-order harmonic wave, the duty ratio needs to be 0.5, in which the fundamental wave is able to have a narrower bandwidth and higher magnitude.³ Finally, the number of arrays needs to be considered based on the size of the laser beam (beam diameter, 10 mm). Table I shows the design parameters for the patterned patch used in this study.

The fabrication process of the patterned CSNP-PDMS patch is shown in Fig. 2. First, the glass plate was treated with silane coating for 6 h to become antiadhesive. The PDMS base and curing agent (Sylgard 184) were mixed at a ratio of 10:1 and then placed in a vacuum chamber to degas for 30 min. The mixture was then poured onto the antiadhesive glass plate, followed by spin coating at 1000 rpm. At the same time, two different types of patterned aluminum sheets were prepared (patterned and nonpatterned) for comparison. Then, the patterned Al sheet was placed on the PDMS cured glass plate, which was then positioned within the flame core. After the 15–30 s growth process, this glass plate was coated with a uniform CSNP layer. The degassed PDMS was poured onto the CSNP layer, owing to the porous structure of the CSNP layer, and PDMS penetrated into the CS layer easily, forming the CSNP-PDMS composite. After PDMS was fully cured at 65 °C for 1 h, the antiadhesive glass plate was removed, and only the CSNP-PDMS composite patch remained.

The experimental setup for the laser ultrasound generation with the line-arrayed CSNP-PDMS patch is shown in Fig. 3. The excitation laser source was the 532 nm Q-switched Nd:YAG pulsed laser (Minilite I, Continuum Inc., Santa Clara, CA) with a pulse duration of 6 ns and a pulse repetition frequency of 10 Hz. The laser beam size was about 10 mm in diameter. The laser energy was measured by using a pyroelectric energy sensor (J-50MB-YAG, Coherent, Portland, OR). The external trigger from the laser system was synchronized with an oscilloscope for data acquisition. The laser beam illuminated the surface of the aluminum plate (300 × 250 × 1 mm³) via various CSNP-PDMS patches (no patch, bulk patch, and patterned patch). The distance between the laser source and the sensor is constant (D = 200 mm), but various sensor's positions with different angles (θ = 0°, 45°, and 90°) were imposed to characterize the directivity and bandwidth of patch-induced photoacoustic Lamb waves [Fig. 3(b)]. The signal from the AlN sensor was obtained using a pulser-receiver (5077PR, Olympus, Tokyo, Japan) and an oscilloscope (DSO7104B, Agilent, Santa Clara, CA). The data were transferred to a computer in Excel format for signal processing and analysis. For the comparison of CSNP-PDMS patch efficiency, three types of patch applications were adopted: no patch, bulk patch, and line-arrayed patch as shown in Fig. 3(b). Different thicknesses of CSNPs, 3 ± 0.5 μm and 6 ± 0.8 μm, were coated on the patch by controlling the growth time on the flame with 15 s and 30 s, respectively.

The morphology and structure of CSNPs were studied by using an emission scanning electron microscope (SEM, FEI Verios 460L, OR), as shown in Fig. 4. The CSNPs are homogeneous, and the particle size is uniform with a diameter of around 45 ± 5 nm, forming a loose and branchlike network. The surface roughness of CSNPs was measured to be about 200 nm by using a scanning probe microscope (Dimension Icon, Bruker, Santa Barbara, CA). As reported in our previous paper,¹⁶ the optical properties, including the optical reflection, transmission, and absorption coefficients, were determined using a spectrometer (Agilent Cary 5000 UV-VIS-NIR, Santa Clara) with the measured area of 10 mm². The optical transmission and reflection coefficients of the CSNP-PDMS composite were 0.015% and 3.659%, respectively. The optical absorption coefficient was 96.24 ± 0.31% by averaging the measurements at different spots on 20 mm × 25 mm samples, which implied the great laser absorption capability of CSNPs.

The position of the Nd:YAG pulsed laser beam was fixed from the AlN sensors at a distance of 150 mm. The laser energy level of 30 mJ (energy density of 6.375 × 10¹⁰ W/m²) was adopted for whole tests to prevent the ablation effect. Once the stainless steel plate was excited by the pulsed laser, the S₀ and A₀ modes propagated in the plate, as shown in Fig. 5. The velocity was measured at D = 150 mm from TOF analysis, which resulted in clear identification of the S₀ and A₀ modes from the dispersion characteristics, i.e., the S₀ mode had a greater velocity than the A₀ mode. The first task was to explore the characterization of the CSNP-thickness effect on the sensitivity and selectivity of the AlN sensor. It can be seen that the signal amplitudes have a higher value in the thick CSNP-PDMS patch than in the thin one [Fig. 5(a)], which explains the proportional relationship between the heat absorption rate and the thickness of CSNPs. It is worth noting that the S₀ mode was reported to show a greater amplitude than the A₀ mode signal in a relatively high frequency range (>200 kHz) with a normal ultrasound sensor system.²⁵ However, the A₀ mode signal shows higher amplitude than the S₀ mode due to the narrow bandwidth induced by the line-arrayed patch design. The time domain signal of the A₀ mode was further analyzed in the frequency domain by performing a Fast Fourier Transform [Fig. 5(b)]. The A₀ mode lamb wave has the main frequency of about 600 kHz, which is in good accordance with the lateral resonance frequency of the AlN sensor. In Fig. 5(b), the signal magnitude (dB) from the thick patch exhibits a higher value than the one from the thin patch, which can be explained with the difference of signal amplitudes. For the rest of the tests, we used the thick-type patch to maximize the patch effect on the sensitivity and selectivity. Compared with the results in Figs. 5(c) and 5(d), the clear efficiencies of the CSNP-PDMS patch are represented in terms of amplitude and frequency selectivity. The patterned patch shows 2.3 times higher amplitude than no patch condition and a narrower bandwidth than other conditions.

The directivity of the line-arrayed CSNP-PDMS patch was also investigated by adjusting the sensor's positions (parallel, 45° and 90°) as shown in Fig. 3(b). The reduced signal amplitude and unclear Lamb waves were detected in the 45° and 90° conditions [Figs. 5(e) and 5(f)]. A maximum Lamb wave signal can be detected in a clear manner at the condition of parallel alignment, and FFT results also show the highest magnitude (in dB), which is attributed to the interferences from mixed modes occurred due to the pattern. So, the line-arrayed patch can make the most of its performances at the well-aligned conditions. So, the alignment of the sensor to the arrayed pattern needs to carefully be accomplished to get the maximum and clear wave signal, which would be the issue to overcome in the future work.

In this work, a hybrid ultrasound NDT method using a photoacoustic-laser-source as a noncontact Lamb wave generator and a line-arrayed patterned CSNP-PDMS patch as the signal amplifier and narrow bandwidth selector was demonstrated. Photoacoustic Lamb waves were propagated in a thin aluminum plate and detected by using the AlN sensor system. It was verified that the CSNP-PDMS patch can improve the signal strength by absorbing the laser beam energy and the line-arrayed patch design also helps to generate the narrow bandwidth photoacoustic waves, which is able to filter the unwanted wave modes and enhance the convenience and accuracy of Lamb wave analysis. However, the current patch design can be useful only at the parallel condition with the sensor. So, different designs of the patch to overcome the limitation of the current patch will be developed and adopted to the defect mapping even at high temperature in future work.

This work was partly supported by the DOE Nuclear Energy Enabling Technologies (NEET) Program under Contract No. DE-NE0008708 and the 2019 research fund of Korea Military Academy (Hwarangdae Research Institute). The authors would like to thank Dr. Tiegang Fang for technical support on laser instruments.