1 Introduction

The effects caused by coherent light with sufficient energy to alter material parameters permanently or temporarily have been in the focus of physical and engineering research for many years [1]. One of these effects is the generation of ultrasound waves by short-pulsed lasers [2, 3]. Laser ultrasound generation depends on the material parameters (density, structure, anisotropy, thermal conductivity, photoplasticity, strain tensor—and therefore the direction-dependent Poisson ratio), as well as on the laser parameters (wavelength, pulse energy, pulse shape and pulse length, energy distribution) and their interrelationship with the material (absorption). Since carbon-fiber-reinforced plastics (CFRP) is built up of a resin with inclusion of fiber layers in varying orientations, all material parameters and the absorption tend to be nonlinear. Therefore, it is generally difficult to excite a reproducible ultrasound wave by means of a laser.

For laser ultrasound measurements of carbon-fiber-reinforced composite materials, it is favorable to excite the photo-acoustic absorption in the polymer and not in the carbon fiber to avoid optical damage of the material. Most epoxy compounds have strong absorption at wavelengths between 3.3 and 3.4 μm [4]. The CH-bond in the resin is resonant at 87.5 THz which corresponds to a wavelength of 3.4 μm. On the other hand, the OH-absorption is relatively high at 2.9 μm [4–6]. In between these wavelengths is a gap where the absorption varies strongly [7, 8] and therefore also the ultrasound amplitude which can be generated non-destructively by using a short-pulsed laser [5]. Possible solid-state laser systems to excite these photo-acoustic absorptions are either a ZGP OPO pumped by a 2.09 μm Ho:YAG laser [9] or a KTA OPO pumped by a 1.064 μm Nd:YAG laser [10]. We decided to build an all-solid-state KTA OPO system because of the availability of reliable optical components and the favorable shorter pulse length of such a system in the region of 7–10 ns. A short pulse length is essential to achieve a high spatio-temporal resolution for the laser ultrasound measurement.

2 Experimental setup

The experimental setup is shown in Fig. 1. It consists of a Q-switched, end-pumped Nd:YAG master laser oscillator and a two stage, end-pumped power amplifier (MOPA configuration) to minimize the effects of thermal depolarization and lensing during amplification. The laser system includes a Pockels cell for Q-switching. It bases on earlier works on compact MOPA systems with two amplifier stages [11], which are pumped by fiber-coupled high-power diode lasers. The Nd:YAG rods of the laser system are doped with 0.9 at.% of Neodymium and have a diameter of 3 mm and a length of 30 mm. The end faces of the rods are slightly wedged (0.5°) and on one side anti-reflection coated for 1,064 nm to prevent parasitic lasing effects. The pumped side of each laser rod is high-reflection coated for 1,064 nm and anti-reflection coated for 808 nm, which is the wavelength of the pump diodes.

Fig. 1
figure 1

Schematic of the experimental setup

Full size image

The master laser oscillator is pumped by a fiber-coupled, quasi-continuous wave (QCW) laser diode module with a peak power of 400 W. The fiber has a diameter of 600 μm and a numerical aperture of 0.22. By using a collimation lens and a focusing lens, the pump radiation is directly imaged into the laser rod, which is conductively cooled by means of a copper heat sink with an indium foil to ensure thermal contact. The end facet of the laser crystal forms one end mirror of the laser resonator. The other end mirror of the laser resonator, serving as output coupler (OC), has a radius of curvature of 50 cm and a reflectivity of 80 %. The plano-concave laser resonator has a length of 25 cm. Calculation of the TEM00 beam radius, utilizing the formalism of Kogelnik and Li [12], results in a beam radius smaller than 350 μm for a thermal lens smaller than 3 dioptres at the position of the laser crystal. The corresponding calculation result is depicted in Fig. 2. For a beam radius of 350 μm, the imaged pump beam, with a beam size as large as the fiber diameter, generates a gain profile which acts as a soft mode aperture within the laser resonator. This results in a near TEM00 beam profile of the master oscillator.

Fig. 2
figure 2

Calculated TEM00 beam radius at the laser crystal in dependency of the thermal lens of the laser crystal

Full size image

An optical isolator is installed between the master oscillator and the power amplifier to prevent parasitic lasing of the power amplifier. Otherwise, parasitic lasing could happen when a parasitic resonator is formed by the input coupling mirror of the subsequent optical parametric oscillator and the output coupling mirror of the master oscillator. The amplifiers are pumped by a 808 nm QCW fiber-coupled laser diode module capable of 800 W peak power with a duty cycle of 1 %. The fiber diameter of this module is 600 μm with a numerical aperture of 0.22. The pump radiation is collimated and split into two beams by means of a 50 % beam splitter and then imaged into the amplifier rods. The thermal lens in the amplifier rods leads to a relay imaging (corresponding to 2 times focal length) between both amplifiers at an optimized pump power.

The optical parametric oscillator is a signal resonant, walk-off compensated, critically phase-matched (cpm) KTA OPO. The used combinations of KTA crystals have an aperture of 6x8 mm and individual lengths of 15 and 20 mm. Type II phase matching \((\hbox{o}\rightarrow\hbox{oe})\) for a 3.3 μm idler wavelength was obtained at a phase-matching angles of θ = 74.3° and ϕ = 0° [13], with a walk-off angle of 22 m rad for the idler radiation. The estimated threshold of this setup was calculated using the formulas presented by Brosnan and Byer [14]. Figure 3 shows the calculated dependence of the OPO threshold on the reflectivity (1-losses) of the OC and the crystal length for a single-resonant KTA OPO for 3.3 μm. The effect of walk-off (0.4 mm for the 20 mm crystal) was neglected in the calculation, because of the walk-off-compensated setup of the crystals in the OPO and the relatively small size compared to the pump beam diameter at the position of the OPO. If the walk-off is neglected, the threshold scales as the reciprocal square of the crystal length. Related to the threshold of the OPO with two 20-mm crystals, the combination of two 15-mm crystals leads to a 1.8-times higher threshold, and the combination of a 15- and a 20-mm crystal increases the threshold by a factor of 1.3. To prevent parasitic oscillations, the crystals are wedged and, to reduce resonator losses, they are anti-reflection coated for 1,064, 1,560–1,580 and 3,300 nm. The reflectivity at 1,064 and 1,560–1,580 nm is lower than 0.5 %, whereas the residual reflectivity at 3,300 nm is approximately 6 %. Due to the required wavelength combination, the coatings cannot be optimized further.

Fig. 3
figure 3

Results from threshold calculations for a 3.3 μm singly resonant KTA OPO with different crystal lengths

Full size image

3 Performance of the laser

The laser was quasi-cw pumped at a pump duration of 280–300 μs, which is slightly above the inversion time of Nd:YAG of 240 μs. Figure 4 depicts the dependence of the output power of the master oscillator on the pump power at a repetition rate of 20 Hz. The laser showed an optical–optical slope efficiency of 14 % in Q-switched operation, whereas a slope efficiency of 34 % was measured when the laser was operated without Pockels cell and polarizer.

Fig. 4
figure 4

Output power of the master oscillator as a function of the pump power

Full size image

The dependence of the output power of the MOPA on the pump power is shown in Fig. 5. The power amplifier rods were pumped synchronously to the master oscillator. A slope efficiency of 25 % was achieved for the power amplifier.

Fig. 5
figure 5

Output power of the MOPA as a function of the average pump power. The energy of the master oscillator was 5 mJ at 20 Hz

Full size image

The stability of the temporal pulse profile is demonstrated in Fig. 6. This measurement was performed using a fast photo diode and taking a short-term overlay of 121 pulses with the oscilloscope. The laser exhibits an excellent short-term stability and has a pulse length (full width at half maximum) of approximately 7 ns. Longitudinal mode beating was not observed due to the bandwidth limit in the order of magnitude of 200–400 MHz of the photo diode and the oscilloscope, and the larger longitudinal mode spacing of the short resonator. A maximum pulse energy of about 50 mJ was achieved when the laser system was operated at a repetition rate of 20 Hz.

Fig. 6
figure 6

Overlay of 121 oscilloscope trace of laser pulses from the MOPA system, recorded at maximum pulse energy at a timescale of 10 ns

Full size image

Figure 7 depicts the spatial laser beam profile after amplification by the two amplifier stages, measured at the position of the OPO crystals.

Fig. 7
figure 7

Beam profile of the MOPA output at the position of the OPO. In addition, image cuts through the center of the profile are plotted in comparison with a Gaussian profile

Full size image

4 Performance of the OPO

The performance of the OPO for the different combinations of KTA crystals with lengths of 15 and 20 mm is depicted in Fig. 8. The relation between the thresholds of 11, 13 and 19 mJ corresponds well with the predicted relation (see Fig. 3).

Fig. 8
figure 8

Idler energy of the OPO plotted as function of the output energy of the MOPA for different crystal lengths

Full size image

By scaling the data presented in Fig. 8 to the threshold power P th of the OPO and calculate the conversion efficiency P Idler/P Pump one observes the characteristic behavior of an OPO shown in Fig. 9. The conversion efficiency achieved by an OPO depends mainly on the number of times the pump power exceeds the threshold of the OPO and on the quantum efficiency (Manley Rowe relation). For example, if we consider a continuous wave OPO pumped by a Gaussian beam, the theoretical limit for the conversion efficiency of approximately 70 % is reached for a pump energy of five times the pump energy at the OPO threshold [15]. For an OPO pumped by a pulsed laser, this limit is reduced to approximately 50 % on account of the temporal pulse shape of the pump light. If we assume a maximum efficiency of 50 % for the conversion of the pump radiation to signal and idler radiation, and a quantum efficiency of 32 % for the conversion of 1.064–3.3 μm, a total efficiency limit of 16 % results for the conversion to idler radiation. The investigated OPO reached a conversion efficiency of 9 % at a pump energy of 4.5 times above threshold energy. Extrapolated to 5–6 times above threshold, a conversion efficiency of up to 9.5–10.5 % is expected. This reduced conversion efficiency, compared to the theoretical limit, is possibly a result of the coating losses for the idler radiation within the resonator.

Fig. 9
figure 9

Conversion efficiency of the OPO idler radiation related to the pump energy, plotted as a function of the number of times the pump power exceeds the threshold of the OPO, \(N_{\rm th}=\frac{P_{\rm Pump}}{P_{\rm th}}\), for different crystal lengths. The data are derived from the measurements displayed in Fig. 8

Full size image

5 Photo-acoustic measurements

In order to test the laser system and to verify its suitability, we used a setup which allows for normalizing of ultrasound measurements with different beams. The principal setup was developed during the INCA-Project (funded by the European Community GRD1-2000-25309). It has of two 2-mm pinholes separated by a certain distance in front of a specimen. On the rear side of the specimen, a water tank with a calibrated immersion tester (Ge/Krautkramer, Alpha series, 5 MHz) is used to detect the ultrasound wave. The principle setup is illustrated by the inset of Fig. 10. The energy of the laser beam is measured by a removable optical power sensor (OPHIR Typ 3a-ROHS) placed behind the first pinhole. Since the diameter of the collimated laser beam (2.5 mm FWHM) is slightly larger than 2 mm, only the central part of the Gaussian pulse contributes to the generation of ultrasound waves. The efficiency then is defined as the ratio of the ultrasound amplitude measured in mV and the laser energy measured in mJ.

Fig. 10
figure 10

Signal efficiency for the conversion of the energy of the incident laser pulse into an ultrasound wave, measured for three specimens at different laser energies. The inlay illustrates the measurement setup

Full size image

In our experiments, three different specimens were used, with thicknesses of 3.2 mm (P1), 1.5 mm (P2) and 4.9 mm (P3), consisting of multi-layered CFRP (see Fig. 11). A typical signal obtained from the generated ultrasound is shown in Fig. 10. The ultrasound wave is remarkably high even though low excitation energies were applied, resulting in a high acousto-optical conversion efficiency for the described laser system.

Fig. 11
figure 11

Oscilloscope trace of a laser-exited ultrasound wave detected at the rear side of the specimen P2 (1.5 mm CFRP, see text)

Full size image

6 Summary and outlook

We demonstrated a 3.3 μm KTA OPO laser system for the generation of laser ultrasound in CFRPs. The system is compact and shows a high efficiency for the excitation of laser ultrasound waves, even at low-pulse energies. Since the laser crystals were conductively cooled, the operation frequency was limited to 20 Hz in the experimental investigations, although successful operation of the laser at 100 Hz was demonstrated for a short period of time. Compared to CO2-lasers [16, 17], which are currently used for laser ultrasound excitation, the presented laser concept provides compactness and the advantage to scale the pulse rate to higher operating frequencies, and thus increase the inspection speed significantly. Works to scale the system to 1 kHz repetition rate have been started.

Additional improvement of the conversion efficiency should be achievable by using a dual stage OPO, which uses the generated signal radiation to further amplify the idler radiation [18]. This could be performed, e.g., by a RISTRA [19] type of setup to achieve an optimized output beam shape by means of acceptance angle filtering.

Investigations regarding a quantitative comparison between the laser reported in this paper and other systems will be carried out in short.