1. Introduction

The CIGS solar cell represents an exciting approach to address the cost issue in solar devices with relatively moderate efficiency. The current high value for solar conversion efficiency devices had been reported by using multi-junction tandem cell approach in III-V semiconductors [1, 2]. Recent progress on GaInNAs-containing quantum wells active regions grown on GaAs template had resulted in superior device characteristics [3, 4], which has led to its implementation in four-junction solar cells with 43.5% solar conversion efficiency [5].

To enhance the efficiency, the processes used in production are asked to provide ultimate performances in terms of control of the action and absence of side effects. Among these processes, the thin films scribing plays a critical role in multilayer solar cells patterning. In the PV modules based on this thin film technology, lasers are able to scribe the metal back contact isolating pattern, a process known as P1, or to eliminate cleanly the semiconductor absorbing layer (P2) or to ablate the front contact isolation (P3) patterns. A relevant example, which will be addressed in the present work is the CIGS cell, in which the P2 process is required to ablate the CIGS layer without damaging the molybdenum layer beneath. This action is known to be critical for providing a high conversion efficiency.

Several groups have already addressed laser P2 process: using pulses in nanosecond regime from a Nd:YAG laser (both 1064 and 532 nm wavelengths) [6–8] a Cu vapor (511/578 nm), an XeCl excimer (308 nm), and a KrF excimer (248nm) [9, 10], in ultrafast regime with a Ti:sapphire laser at 800 nm, using a Nd:YVO₄ picosecond laser system at 1064 nm [11–13]. These works demonstrated that laser scribing provides processing advantages, but it introduces undesired effects, which are not present in mechanical scribing. Indeed, melting of CIGS at the edges of the grooves could cause short-circuiting and shunting of cells was reported [14]. Moreover, cracks in the molybdenum layer or partial removing of the semiconductor layer can increase the interconnect resistivity [11] and a large heat affected zone (HAZ) can increase the shunt resistance [15].

In this work we present an approach named Laser Induce Strain Ablation (LISA) [16] which differs from the previous works for the type of physical processes exploited to obtain P2 [17]. The differences in the value of the optical absorption and the various mechanical properties, including density, thermal conductivity, surface adhesion energy and stress-strain curve, allow for the exploitation of laser driven non-contact removal processes based on differential principles.

The minimization of the mechanical stress induced on the underlying molybdenum layer and the thermal damage to the rims of the remaining CIGS layer is also addressed. By using a suitable laser wavelength, power and pulse duration, to match transiently a strong absorption in the semiconductor layer, we obtain the mechanical stress of the CIGS layer that eventually results in its ejection without melting. This approach is attractive because the low power involved is insufficient to melt the layer and the controlled strain gives the opportunity to avoid cracks, burrs and residues that can affect the total efficiency of the panel. The thermal exchange is very limited, so that the HAZ is very restricted and similar to the best cases obtained with ultrafast ablation. The LISA approach could be extended to the other type of layer and is also suitable for parallelization due to the low cost source used.

2. Laser induced thermal gradient and strain in CIGS

In this work, we have considered the cell structure as follows: a glass substrate of 2 mm, a 500 nm molybdenum film, a 2 μm CIGS film and, in the case of process P3 only, a 50nm TCO film [17]. LISA aims are the separation of a selected stripe of CIGS layer from the molybdenum one and that of TCO for P3, without inducing the phase change in to the CIGS. This separation is obtained by exploiting the induced thermal gradient that causes the material expansion. The volume expansion creates strains and induces detachment when the strains reach certain threshold.

The LISA model is present here and also to gain insight on the assessment of the robustness, in term of the variation of the parameters that are effective in LISA.

In accordance with the Lambert-Beer law the energy absorbed from an incident beam of intensity I(z) in material is described in terms of the attenuation of the incident intensity along the propagation direction z, $\partial I((z)=-\frac{\lambda }{4\pi k}I(z)\partial z$, where λ is wavelength and k is imaginary part of refractive index [18].

The energy distribution is analyzed using Fourier equations, without phase transition. This latter condition has been verified numerically from the model. The thermal distribution is then as follows [19–21]:

The boundary conditions for the thermal flux have only one non-adiabatic face, in which the laser impinges: n_z · (k∇T)|_surface = S(0)|_surface. For all the other faces the condition is then n· (k∇T) = 0, with gradient n indicating the lateral and bottom faces. The initial temperature is set to 293 K.

The strain conditions are defined by the deformation components u, v, and w and their derivatives. The precise relation between strain and deformation depends on the relative magnitude of the displacement [22]. The strain in the layer is only caused by thermal gradient, (ε_th), with its initial ε₀ = 0.

Under the assumption of small displacements, the normal strain components are ε_i, i = x, y, z, and the shear strain components ε_ij are related to the deformation [22]. For example: ${\epsilon }_x=\frac{\partial u}{\partial x}$ and ${\epsilon }_{xy}=\frac{1}{2}\frac{\partial u}{\partial y}+\frac{\partial v}{\partial x}$.

From the solution of Eq.(1) we calculated the relevant thermal strain components using ${\epsilon }_{\mathbf{ij}}^{\mathbf{th}}={\alpha }_{ij}\left(T-T_{\mathit{ref}}\right)$, where α_ij are the thermal expansion coefficients, which is in our case symmetric.

The stress-strain relationship is the final step, and it reads [23]

The quantities λ_L and μ_L are the Lamè’s first and second parameters, they are strongly dependent on the type of material and its deposition technique.

The boundary conditions imposed for the mechanical strain depends on the structure, and permits to decide if they are fixed or if it is free to move in one or more directions. In the model only the top face is free. For the remaining surfaces u=0, because there is no displacement.

3. FEM results

Figure 1 shows the temporal evolution of temperature in the center of the laser spot, in different points inside the target. The target, or almost the volume of the film, didn’t exceed the melting temperature (1600 K). On the bottom face, the temperature reached was lower, thus far below the melting temperature of molybdenum (2896 K). From the figure we may verify that the temperature of about 500 K reached inside the film, results significantly lower than the melting point.

 

Fig. 1 Temperature distribution along z axis in the center of the spot and pulse shape (black) x₀=100 μm, y₀=5 μm, τ =0.5μs, λ =808 nm; k=0.27 silica glass ρ =2203 Kg/cm³, C_p=703J/(Kg °K), λ_T =1.38 W/m°K CIGS ρ =5770 Kg/cm³,C_p=300J/(Kg°K), λ_T =3.7W/m°K molybdenum ρ =10200 Kg/cm³, C_p=255J/(Kg °K), λ_T =138W/m °K [25][17][26][18]The legend shows different line types for different point in z axis from 0.2 to 1.8 μm

Download Full Size | PDF

Figure 2 shows the value of first principal stress, from which we may see that it is increasing toward the lower film interface. The peak value reaches several tens of MPa. In Fig. 3 is plotted the second principal stress (the third was symmetric for the geometry of the laser spot). Here the peak value was larger, of about several hundreds of MPa. As this effect induces a strain orthogonal to the layer larger than the adhesion force of the CIGS/Mo interface, it would result in the detachment of the irradiated layer.

 

Fig. 2 First principal stress with the detachment threshold [27] and its directions silica glass α = 0.55e − 6[1/°K] E = 73.1GPa, ν = 0.17 CIGS α = 8e − 6[1/°K] E = 50GPa, ν = 0.4 molybdenum α = 4.8e − 6[1/°K] E = 312GPa, ν = 0.3 The legend shown different types of line for different point in z axis from 0.2 to 1.8 μm

Download Full Size | PDF

 

Fig. 3 Second principal stress and directions of second and third principal stress The legend shown different line types for different point in z axis from 0.2 to 1.8 μm

Download Full Size | PDF

Typically the yield strength of a single crystal CIS (about 0.91 GPa [28]) is higher than the adhesion threshold and we can’t assume it as threshold. At the same time not exceeding the yield strength of the crystal is important to avoid cracks out of the channel. In practice, the adhesion force depends on several parameters of the deposition method. However, we can analyse some adhesion tests used to measure peel resistance for CIGS. From these tests the strength of adhesion of CIGS/MO is one of the lower compared with many others kinds of solar films [27]. It is clear that CIGS/MO patterning by induced stress is a good solution. Unfortunately peel test physical implications are quite different to LISA physical implications and it couldn’t be compared directly, but it is relevant to know the characteristic of the interface compared with other solar cells type.

4. Experimental results

LISA was realized with a pulsed laser diode with a high current capacity driver at 1 kHz. The average power varied from 1 to 10 mW. The wavelength (808 nm) was chosen in order to obtain strong absorption. A precise translation stage was used to scan the cell under the laser focus. In Fig. 4 are shown some frames from the video taken during LISA ablation that clarify the process: the light grey channel is the underlying molybdenum layer, the black part is the CIGS layer. In particular in this image there are 2 important phenomena: the lifting of the CIGS and its detachment in rectangular solid fragments of the size of the laser focus.

 

Fig. 4 Images of the LISA process on CIGS, channel width 180 μm. From left to right there are 3 different frames captured during the process, in the first and in the second there is a partial lift off and in the third a fragment is detached

Download Full Size | PDF

The experimental results of laser scribing are analyzed with a scanning electron microscope (SEM). Figure 5(a) shows a relevant example, from which we may draw the following observations: 1) the irradiated area is clear from CIGS residuals; 2) the removal left clean rims of the CIGS edges, with only sporadic deformation; 3) no cracks are induced in the remaining CIGS layer. Figure 5(b) (4000× magnification) shows an example of residual defects at the channel rim in the case where only some tens nanometers are melted. In this case the defect spans a small fraction of the channel width, that was set to 180 μm. There isn’t slope on the edges and the shape of the channel is clean and vertical, also nearby the defect. This means that there isn’t a region where the pulse energy shape affects the layer as opposed to the classical scribing with gaussian pulse. There isn’t debris near the edge as well as in the center of the channel Fig. 5(a). Using low thermal gradient we can reach the total absence of micro cracks because the area interested by the cracks is only the removed area. The HAZ is limited only to some micrometers from the channel because a low peak power was used for inducing stress.

 

Fig. 5 SEM images of the channel. Scale bars are 100 and 10 μm respectively on the left and on the right.

Download Full Size | PDF

5. Conclusion

To sum up, LISA appears to be an effective method for the removal of an adsorbing layer such as CIGS on molybdenum. The very low power used prevents the induction of permanent damages and the melting of the layer. The induced stresses are localized in the channel and without any cracks out of it. These latter advantages are crucial for preserving the efficiency of the cell. The experimental results were confirmed by a numerical calculation, thus fostering its introduction as an effective process for the solar technology.