1 Introduction

The application of laser processing technology is highly desired and critical in certain areas of photovoltaics manufacturing such as drilling, scribing, ablation, welding, etc. [1–9]. Especially, there have been tremendous research and development effort recently to establish thin film solar cell (TFSC) technologies in place of the more expensive crystalline silicon-based solar cells. TFSCs can be manufactured at much lower cost by adopting simple manufacturing processes and easily integrated onto the exterior of building structures.

Among various types of TFSC, CuIn1−x Ga x Se2 (CIGS) solar cell has received much attention due to its record high cell efficiency of 20.3 % [10] and stable performance. For the development of laser manufacturing technology, proper understanding of the ablation characteristics of CIGS solar cell is crucial. For example, stable laser scribing technology is a prerequisite to realize the benefits of cost-effective mass production of CIGS TFSCs. The module of CIGS solar cell consists of many small cells interconnected in series, known as the monolithic interconnects, and its fabrication includes deposition of metal back electrode layer (typically Mo), CIGS absorber layer, transparent front electrode layer, and three scribing processes known as P1, P2, and P3 corresponding to the isolation of back electrodes, formation of connection channel between the front and back electrodes of adjacent cells, and isolation of front electrodes, respectively [11–14]. Researches about laser scribing of P2 and P3 patterns proved that avoiding melting is a challenging problem [15, 16]. As an alternative example, the photoconversion efficiency of CIGS TFSCs is known to vary sensitively to the elemental composition of CIGS absorber film. Accordingly, a rapid and reliable technique for monitoring the composition of CIGS thin films during manufacturing is essential to ensure a desired product quality. Recently, laser-induced breakdown spectroscopy (LIBS) has been proposed for compositional analysis of CIGS [17]. For LIBS elemental analysis, the characteristics of laser ablation plasma of CIGS thin films need to be thoroughly understood.

The ablation characteristics of a CIGS absorber film depend not only on the laser parameters such as pulse energy, wavelength, and pulse duration but also strongly on the elemental composition of Cu, In, Ga, and Se. The elemental composition of CIGS solar cells is designed in principle approximately in the ratio of Cu: (In + Ga): Se = 1:1:2, where Ga is a substituent element for In. Depending on the Ga ratio over the sum of Ga and In, denoted by x Ga ≡ Ga/(In + Ga), band gap and efficiency of CIGS solar cells vary significantly.

In this work, the ablation characteristics of thin CIGS films with different x Ga values are reported. The changes in ablation mechanism with respect to Ga concentration and laser pulse energy were investigated by taking shadowgraph images of the ablation plume and scanning electron micrographs of craters. The effects of laser wavelength were also discussed.

2 Experimental

The details of sample preparation, their composition, and band gap measurements are described elsewhere [18]. Shortly, the CIGS thin films were fabricated by co-evaporation process on Mo-coated glass substrates. Since the experiments were intended to investigate the ablation characteristics of CIGS layer, the samples were prepared in the form of CIGS/Mo/glass structure. The band gap of the CIGS samples was estimated by measuring ellipsometry and fitting the measured spectra into the Tauc–Lorentz model to obtain absorption coefficient. The elemental composition and band gap of the CIGS samples in experiments are shown in Table 1. Note that, the estimated band gap of these samples was slightly lower as compared with the values reported in other study [19].

Table 1 Composition, thickness, and band gap of the CIGS absorber films
Full size table

A Q-switched Nd:YAG laser (New Wave, τ = 8 ns, λ = 1,064 and 532 nm, multimode) was used for single-shot irradiation of the CIGS samples in air. The diameter of near top-hat laser beam was about 150 μm. Shadowgraph images were taken for the 1,064-nm-laser irradiations using a probe laser (Nd:YAG, τ = 5 ns, λ = 532 nm). The measurement of emission intensity of laser plasma was carried out using a commercial LIBS system (Applied Spectra Inc., RT250-EC) which is equipped with a six-channel CCD spectrometer (gate width = 1.05 ms, spectral window = 187–1,045 nm, resolution = ~0.1 nm).

3 Results and discussion

Shock wave formation is a characteristic of pulsed laser ablation of solids. For opaque targets, the incident laser light is strongly absorbed within a thin layer of material, which results in a rapid increase of surface temperature over the boiling point and an intense vaporization of the target material. The expansion of vapor into the surrounding atmosphere then results in the generation of a shock wave. Figure 1 shows the shadowgraph images of the CIGS samples with varying x Ga values taken at varying laser fluences (F) at the delay time of about 470 ns. For the sample with x Ga = 0.028 in Fig. 1a, shock wave was observed for all laser energies in the experiments. The average shock wave velocity of this sample exceeds 1,100 m/s (Fig. 2), comparable to that of typical opaque samples [20]. The high shock wave velocities are considered to imply that strong laser evaporation had occurred on the x Ga = 0.028 sample. When the Ga concentration was increased to x Ga = 0.192 at F = 0.60 J/cm2 condition (Fig. 1b), however, shock wave velocity dropped almost by half (~560 m/s) from that of the x Ga = 0.028 sample, implying that vaporization of this sample was much weaker despite the same laser energy. When the Ga concentration was further increased to x Ga = 0.408 (Fig. 1c), no shock wave was produced at the F = 0.60 J/cm2 condition. On the other hand, at the elevated laser fluence of 2.34 J/cm2 or higher, shock wave was generated in all samples with slight decrease of velocity for increasing x Ga samples.

Fig. 1
figure 1

Shadowgraph images of the a x Ga = 0.028, b 0.192, and c 0.408 at varying laser fluence at delay time of about 470 ns. (The arrows indicate the shock wave front)

Full size image
Fig. 2
figure 2

Average velocity of the shock waves produced at varying laser fluence

Full size image

The changes in shock velocity of the CIGS samples are summarized in Fig. 2 which shows that (1) strong evaporation takes place at all laser fluences if x Ga is nearly zero but (2) evaporation diminishes rapidly for increasing Ga concentration at the low fluence condition. The observed changes in shock wave velocity or equivalently in CIGS film evaporation are due to the changes in absorption characteristics of the CIGS films with respect to Ga concentration. For low Ga concentration, the band gap of CIGS film is much smaller than the photon energy (1.2 eV) so that strong surface absorption takes place, leading to rapid evaporation and strong shock wave formation. As Ga concentration increases, the difference between the band gap and photon energy decreases and almost disappears at x Ga = 0.408. As a result, the penetration depth of incident laser light will gradually increase with x Ga, whereas surface evaporation diminishes, more rapidly at low energy.

The characteristics of material removal of these samples could be observed with the shadowgraphs taken over extended delay times. First, the sample with low Ga concentration (x Ga = 0.028) showed no observable changes for all delays at the low fluence condition of F = 0.60 J/cm2, Fig. 3a. The crater of this sample revealed a resolidified surface, and energy dispersive X-ray spectroscopy (EDX; Horiba, EX-250) measurements verified that the elemental composition of the resolidified surface is almost identical to that of non-irradiated surface. The shadowgraphs, crater morphologies, and EDX measurement results of the x Ga = 0.028 sample showed the same results for higher laser fluences. On the other hand, for the sample with x Ga = 0.192, an ejection of material was observed at delays over 2,400 ns as shown in Fig. 3b at the low laser fluence of F = 0.60 J/cm2. The SEM image of the crater in Fig. 3b verified the removal of CIGS film, and the EDX results reconfirmed that only the CIGS layer was removed by revealing a strong Mo signal from the exposed surface. However, when the laser fluence was increased to F = 2.34 J/cm2 or higher, shadowgraph images of this sample showed shock wave only without material ejection at all delays, and the crater morphology was nearly the same as that in Fig. 3a, molten and resolidified surface. When the Ga concentration was further increased to x Ga = 0.408, the ejection of film material was also observed at the low fluence of F = 0.60 J/cm2, Fig. 3c. The crater morphology and EDX data again confirmed the removal of CIGS layer only. The shadowgraph images of the ejected material in Fig. 3c show that the ejected materials are large fragments in polygonal shapes of about tens of micrometer size (see the image at delay time of 7,000 ns). In contrast, the shadowgraphs of the same sample at an elevated laser fluence of F = 8.18 J/cm2 in Fig. 3d showed that the material was ejected in fine fragments. At this high fluence, the SEM image of the crater showed a morphology with significant melting and resolidification of CIGS material, especially in the middle of laser beam where temperature was relatively higher. However, since the temperature along the edge of laser beam would be lower than the middle due to heat loss toward the surrounding film, thermal-stress-induced film removal as described below had occurred rather than melting and resolidification. Based on the observation that the Mo layer in the SEM image in Fig. 3d appears undamaged, it is considered that the fine fragments observed in the shadowgraphs are CIGS particles and droplets, and possibly solid fragments from the edge.

Fig. 3
figure 3

Shadowgraphs at varying delay times and SEM images of the a x Ga = 0.028, b 0.192, c 0.408 samples at F = 0.60 J/cm2 and d x Ga = 0.408 sample at F = 8.18 J/cm2

Full size image

The selective removal of CIGS films at the low laser fluence of F = 0.60 J/cm2 for samples with x Ga ≥ 0.192 in Fig. 3 is understood due to thermal expansion of the films rather than ablative vaporization. The decrease of shock velocity for these samples is the evidence that vaporization is not the driving mechanism of CIGS film removal. The sharp crater edge with no evidence of melting in the SEM images further confirms that melting was not significant during the film removal. A numerical simulation of CIGS film temperature and the estimation of induced thermal stress showed that thermal stress of the film could be in the range of 0.74–1.06 GPa, close to the yield strength of CIGS thin film (0.91 GPa) [21]. Therefore, it is concluded that delamination of the CIGS film from the Mo layer due to laser-induced thermal expansion and eventual fracture as the thermal stress exceeded the yield strength was the responsible mechanism for the selective removal of CIGS films with x Ga ≥ 0.192 at the low laser fluence. The CIGS film underwent this process would be removed in solid fragments with little melting, revealing the underlying Mo layer clearly.

For the samples with x Ga ≥ 0.458, there was little difference in crater morphologies between samples provided that the laser fluence was the same. Specifically, at the low energy of F = 0.60 J/cm2, only CIGS was removed as shown in Fig. 4a like the x Ga = 0.192 and 0.408 cases. However, as the laser energy was increased to F = 2.34 J/cm2 or higher, both the CIGS and Mo layers were ablated over the entire laser spot and glass melting was occurred, implying that transmission of the incident laser light thorough the CIGS layer was significant due to the high band gap of these samples.

Fig. 4
figure 4

a SEM images at the laser fluences of F = 0.60 and 3.42 J/cm2 and b shadowgraphs of varying delay time at the laser fluence of F = 3.42 J/cm2 (x Ga = 0.516)

Full size image

Figure 4b shows the shadowgraph images of the x Ga = 0.516 sample at the laser fluence of F = 3.42 J/cm2 condition where it is seen that the material was ejected initially in fine fragments and then in large fragments of polygonal shapes. The fine fragments appear similar in size to those of x Ga = 0.408 sample at F = 8.18 J/cm2 condition, Fig. 3d, in which significant melting of CIGS film had occurred without Mo layer damage. Thus, it is considered that the initial fine fragments were due to ablation of CIGS layer, whereas the following large fragments are likely to be pieces of fractured Mo layer. The EDX measurement of the crater produced at this condition showed that the glass substrate was exposed.

Figure 5 shows the craters produced on the x Ga = 0.458 sample using the second harmonic Nd:YAG laser (λ = 532 nm) for which the photon energy (2.34 eV) is far greater than the band gap of all samples in experiments. Accordingly, under this irradiation condition, all the CIGS films should strongly absorb the incident laser light as was the x Ga = 0.028 sample under 1,064 nm laser light. In fact, the craters in Fig. 5 show exactly the same morphology as the x Ga = 0.028 sample under 1,064 nm case regardless of laser fluence, confirming that ablative evaporation was the main mechanism of laser–film interaction at this short wavelength. The crater morphology showed no difference for samples with different Ga concentration.

Fig. 5
figure 5

SEM images of the craters produced using the second harmonic Nd:YAG laser (λ = 532 nm) at varying laser fluence on the x Ga = 0.458 sample

Full size image

Figure 6 shows the intensity of plasma emission of the samples measured with a commercial LIBS system as a function of laser fluence. Since Cu content within the samples was almost constant at about 25 %, while Ga and In percentages varied from sample to sample, a Cu emission line at 327.4 nm was selected for plasma intensity comparison. First, plasma intensity of all samples at F = 0.60 J/cm2 condition was nearly zero, meaning that ionization was insignificant for all samples at this energy level, even for the x Ga = 0.028 sample for which vaporization was clearly observed. The emission intensity increased with laser fluence as expected. Note, however, that, the strongest emission was detected for the sample of x Ga = 0.028. Since plasma emission is the result of vapor breakdown, plasma intensity should become higher when strong vaporization takes place as in opaque solids. Although the entire CIGS film was removed from the samples with x Ga ≥ 0.192 at F = 0.60 J/cm2, little emission was detected because the material removal took place mostly in solid fragments at this energy level.

Fig. 6
figure 6

Intensity of laser plasma emission of the samples with respect to laser fluence

Full size image

In summary, the ablation characteristics of CIGS TFSCs can be categorized by two main parameters, namely, Ga concentration and laser fluence. If the Ga concentration is too low or equivalently the band gap is well below the laser wavelength, laser–film interaction is more like the ablation of opaque solids in which melting and evaporation dominate. When Ga concentration is increased to a level at which the band gap is close to or exceeds laser wavelength, the ablation phenomena differ with laser fluence; stress-induced film fracture is the primary mechanism of material removal at low fluence, whereas melting and vaporization are dominant at elevated laser fluence.

4 Conclusion

From shadowgraphy imaging of the ablation of thin CIGS solar cell films with a 1,064 nm Nd:YAG laser, it is shown that the laser–material interaction differs significantly with respect to Ga concentration of the CIGS film and laser energy. When x Ga ≈ 0, ablative evaporation is the dominant mechanism irrespective of laser energy because the band gap of CIGS is well below the incident photon energy and thus laser light absorption occurs as in opaque solids. This will also be true for irradiation with a shorter wavelength laser such as the second harmonic Nd:YAG laser. If x Ga > ~0.2, however, the laser–film interaction differs completely depending on the laser energy level. At low laser energy approximately near or below melting threshold, laser heating dominates the interaction which can lead to solid phase removal of CIGS film without melting. This process could be very useful for laser scribing. At elevated laser fluence, the film undergoes ablative evaporation which is typically accompanied by simultaneous increase of plasma emission. For the application of laser technology for CIGS TFSCs, these differences in ablation characteristics should be considered for successful implementation of the technology.