Available online at www.sciencedirect.

com

Physics Procedia 12 (2011) 207–213

LiM 2011

A Modelling Approach for an Edge Deletion Process of Thin-Film

Solar Cells with High-Power Nanosecond Laser Radiation
René Sieberta,c,*, Severin Luziusb, Andreas Wetzigc, Sascha Weilerb, Eckhard Beyera,c
a
University of Technology Dresden, Institute of Surface and Manufacturing Technology, Dresden 01069, Germany
b
TRUMPF Laser, Aichhalder Straße 39, Schramberg 78713, Germany
c
Fraunhofer IWS, Winterbergstrasse 28, Dresden 01277, Germany

Abstract

The effects of near-infrared laser induced ablation on thin-film solar cells were investigated. As a result, a modelling approach
for the described material deletion process is suggested. Analyses of optical behaviour of the irradiated layers and their interfaces
as well as experiments on ablation threshold determining parameters to validate the modelling approach were observed. Thereby
the interaction of pulse duration and overlapping with respect to the threshold behaviour has been examined. The submitted
contribution promotes laser process development and laser resonator optimisation as well.

Keywords: Thin-films; Solar cells; amorphous Silicon; high-power nanosecond laser, edge deletion

1. Motivation / State of the Art

The electricity production through photovoltaic (PV) has increased continuously within the last years, which is
associated with enormous growth rates of the photovoltaic industry. Solar cell manufacturers are constrained to
increase cell efficiency and productivity in order to reduce costs steadily of each single solar module. The result is
an ongoing variation of layer configuration in regards to the optimisation of optical and electrical behaviour with
changing demands on manufacturing techniques. While modelling approach for layer design [1] and e.g. array
systems [2] is state of the art, ordinary laser processing of prototype solar cells still might be characterised as try-
and-error proceeding. This method can be seen as unfavourable with respect to time orientation and process
development. Therefore basic assumptions are helpful to determine process limits and parameters or as proof of
concept. Indeed suitable and cell type independent tools are needed.

One of those processes is Edge Deletion used to be carried out with high power nanosecond laser radiation that
entails electrical insulation of the current-carrying layers. Furthermore, it is a necessary preparation process for
hermetical encapsulation of the whole module to provide protection against environmental influences by lamination
with another sheet of glass [3]. Typical applications are thin-film solar cells consisting of amorphous silicon (a-Si),
cadmium telluride (CdTe) or copper indium gallium selenide (CIGS) as direct-bandgap material. These materials

* Corresponding author. Tel.: +49-351-83391-3066; Fax: +49-351-83391-3300.

E-mail address: rene.siebert@iws.fraunhofer.de.

1875-3892 © 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.phpro.2011.03.124
208 René Siebert et al. / Physics Procedia 12 (2011) 207–213

were deposited on low-cost substrates at moderate temperatures by chemical (CVD) or physical (PVD) vapour
deposition, and processed by laser radiation [4, 5]. Thus in the literature almost all values required for calculation
can generally be found. An exception might be the influence of the material doping agent variation. Referring to the
described requirements, these deviations are acceptable.

Edge Deletion by laser irradiation has been processed successfully for some years. Indeed, laser processing
competes chiefly with the conventional technique of plasma etching and sandblasting, which implicates some
disadvantages like micro-cracks induced by mechanical stress while processing and handling the modules [3].

2. Experimental

The amorphous silicon solar cell samples consist of six layers deposited on a glass substrate (Figure 1.a). The
first layer, which is a transparent conductive Fluorine-doped tin oxide (SnO2: F), has to fulfil following eligibility
criteria: high optical transmission and at the same time high electrical conductivity. To improve light trapping a
textured surface is generated by wet chemical etching. The following layer is composed of the semiconductor
material amorphous silicon, which is doped in the upper and lower part and undoped inside. These two layers as
well as the substrate material define the requirements for possible laser sources and the processing direction, which
is seen in Figure 1.a as well. So a new generation of pulsed and cavity dumped laser system (TruMicro7025) was
chosen with an optical system manufactured by SCANLAB. The chosen high power nanosecond solid-state laser
system delivers pulse energy up to 80 millijoules and repetition rates up to 100 kilohertz depending on the maximum
laser power.

Figure 1.b shows the square beam profile, an advantage of fibre-coupled multi-mode solid state lasers that
enables high efficiently processing. The beam diagnostic was carried out with Primes FocusMonitor, a tool used in
the field of laser processing for high power and large focus diameter applications. In cooperation with TRUMPF
R&D department experiments with varied pulse durations could also be carried out. For application development, it
is recommended to use straight lines as scanned sample contours to obtain basic understanding on occurring
phenomena. To guarantee a complete layer deletion above the substrate the sheet resistance between the lines was
measured and the result evaluated by Scanning Electron Microscopy (SEM). The determination of the chemical
composition defined by Energy Dispersive X-ray (EDX) analysis allowed the identification of the material debris on
the substrate.

Layer Thickness
glass substrate 2000µm
doped tin oxide 800nm
p-doped amorphous silicon 15nm
amorphous silicon 400nm
n-doped amorphous silicon 25nm
doped tin oxide 800nm
aluminium 1000nm

(a) (b)

Figure 1. (a) Layer configuration of the sample; (b) Beam profile in isometric illustration

As a result of optical process consideration, Figure 2.a shows the intensity loss percentage due absorption and
reflection based on the maximum laser intensity for each layer. It is evident that only the tin oxide layer is
 René Siebert et al. / Physics Procedia 12 (2011) 207–213 209

responsible for the absorption, reflectance varies in the range around two percent, except for the air-substrate-
interface, it is approximately 4 percent. At the interface of tin oxide and amorphous silicon the radiation output is
reduced by 74.2 percent. In Figure 2.b this intensity loss is additionally divided into the ratio for absorption,
reflection and transmission. Corresponding material properties can be found in the literature, because of the
development of new layer configuration, as already mentioned, extensive research were done [e.g. 6, 7, 8].

1,9%
11,8%
74,2% 2,0%
80,4%
2,2%
6,0% 17,7%

3,8%
optics transmission
glass substrate absorption
doped tin oxide reflection
amorphous silicon
doped tin oxide
aluminium
(a) (b)

Figure 2. (a) Representation of intensity loss due reflection and absorption in the entire layer composite; (b) normalized representation of
individual components in the doped tin oxide layer

It was not possible to ascertain the optical behaviour of the composite layer while manufacturing after each
coating step. In this regard, the only chance is to analyse direct reflectance and diffuse reflectance of the whole layer
composite through infrared spectroscopy. At least 17 percent of laser intensity at 1030nm wavelength is not usable
for processing. The direct reflectance behaviour (6 percent at 1030nm) is shown as black coloured graph in Figure 3,
plotted against the radiation wavelength. It correlates with the values taken from the literature for optical process
consideration. Furthermore, the diffuse reflection ratio can not be detected with the mentioned directional
measurement at the same time. Therefore, it has been determined in a subsequent experiment by integrating sphere
with the result of 11 percent for corresponding laser wavelength (red coloured graph in Figure 3). The accumulated
reflectance is mainly caused by interface interactions between air, glass substrate and the first layer.

Figure 3. Optical analysis of reflectance and diffuse reflectance (red)

210 René Siebert et al. / Physics Procedia 12 (2011) 207–213

The energy, absorbed by the solid state Eabs through phonon coupling, defines the vaporisation process, excluding
the energy ratio that is reflected at the interface of the layers. Hence, thermal conduction Qcon causes energy losses
that reduce the energy in the irradiated area, which finally is available for vaporisation. According to the first law of
thermodynamics, the physical model was set.

GEabs  GQcon dH (1)

In order to induce layer deletion, a certain enthalpy difference dH is required that causes temperature increasing
dT to the boiling point and energy consumption for melting and evaporation D H°evap. In dependence of the layer
material density ȡ, the specific heat capacity cv(T) and the boiling point Tevap, a rough prediction can be calculated to
give an overview of the required heating energy. The amount of energy that is needed for heating up is usually
approximately 14 to 18 percent of the total sum that is required to induce ablation. Here, possible metals consume
more energy for melting than semiconductor materials. Relevant information is taken from the literature [cf. 9, 10,
11].

dH = ³³ U ˜ c ( T ) ˜ dT ˜ dV+'H q
v evap (2)

The layer material properties define the reflection (through reflectance coefficient R) and absorption capability
(through absorption coefficient Į) depending on the radiation wavelength and the layer thickness z. The laser fluence
F(x,y) is affected by the beam profile and the magnification of the optical system. The equation for the absorbed
energy per pulse with respect to the focussed spot size is given as;

³
Eabs = F ( x , y ) ˜ ( 1  R ) ˜ e D ˜z ˜ dA
A
(3)

On the dependence of the average power on the pulse energy and the repetition rate frep, the required energy to
induce ablation is calculated as

Pmid = ³³ F ( x , y ) ˜ dA ˜ dt
f A
(4)

Depending on pulse duration, laser irradiation with multiple or overlapping pulses can cause chemical or
structural changes in the surface and facilitate or impede the ablation of subsequent pulses. Furthermore, a surface
modification can be achieved by thermally or electronically induced refractive index change. This allows explaining
the formation of some defects in dielectrics [12]. The phenomenon of ablation threshold by pulse train at ultra short
pulse duration was investigated on metals [13], ceramics [14], polymers [15] and semiconductors [16]. These
analyses base on a model, developed by JEE et al., to obtain the multi-shot damage threshold at specific pulse
overlay. The determined single-shot damage threshold can be used to predict the extent of accumulation. According
to the power law, the ablation threshold at certain fluence Fth for N pulse shots on the same spot area is related to the
threshold fluence at one shot through the certain exponent ȟ [17].

Fth( N ) Fth( 1 ) ˜ N [ 1 (4)

The exponent ȟ is an inversely proportional index for the degree of incubation. The greater the exponent is the
less the dependence of the number of pulses on the ablation threshold, i.e. incubation is absent when ȟ =1.
 René Siebert et al. / Physics Procedia 12 (2011) 207–213 211

3. Results and Discussion

The influence of laser pulse energy and duration on the ablation process, thus the initiation of insulation in
regards to remaining debris, was investigated and the experimental results were listed in Table 1. It was noticed that
at 25ns pulse duration 20 percent less power was needed to obtain the same residue-free ablation in comparison with
experiments at 65ns. Independently from applied pulse duration, insulation occurs at the same fluence of
approximately 2.8J/cm².

Table 1. Experimental result overview

Layer Pulse duration tp [ns] Fluence F [J/cm²] Resistance RΩ [GΩ] Material debris
SnO2: F 25 2.7 0 -
25 2.8 11 -
25 5.6 11 +
65 2.5 0 -
65 2.8 11 -
65 7.0 11 +

One of the conclusions is that the experimental results could be characterised by using the basic model approach,
regardless of thermal conduction processes in the first step. Calculated fluence values appear to be comparable to the
experimental results when insulation was observed. Nevertheless, insulation is not a criterion for a complete layer
removal that is shown in Figure 4. In spite of insulation, SnO2: F debris was detected. Therefore, it is recommended
to include the influence of thermal conduction with a certain parameter that depends on the used laser pulse
duration. The thermal losses reduce the effective process energy by more than a half. In conclusion, the submitted
modelling approach allows forecasts on the process and the calculation of laser parameters.

(a) (b)

Figure 4. (a) SEM picture of insulating but not completely removed layer; (b) magnified picture of white circle marked area: SnO2 debris

Process forecasting becomes more important for samples with different layer material or thickness, if maximum
laser power and pulse energy might restrict the ablation process. In this regard, processing with pulse trains could be
a matter of interest on the ablation behaviour.

Therefore, samples at constant pulse duration and varied power were enquired about the size of the ablated area and
the width of the margins through light microscopy. Despite the beam profile being relatively homogeneous, even in
the marginal area, the irradiated area did not correlate with the effectively ablated area. The margin width decreased
exponentially and the ablated area rose logarithmically with increasing laser power. Further experiments,
212 René Siebert et al. / Physics Procedia 12 (2011) 207–213

additionally with varied pulse overlay, were done on these samples. It was also noticed that by increasing pulse
overlay, and consequently the energy input per scanned length, the ablation threshold for insulation could be
reduced. In Figure 5.a the minimum pulse energy for achieving insulation is plotted against the pulse overlay.
According to the model by JEE et al.[17] the pulse overlay was converted into the number of pulses per irradiated
spot area. A regression analysis of the ascertained data through graphical method led to an exponential behaviour
with the coefficient of -0.15 (Figure 5.b) and correlate with the size of the ablated area. This coefficient shows
similarities to the ratio of the amount of energy needed for heating up compared to the total energy sum that is
required to induce ablation, which is described in chapter 2. The same effect was revealed for the usage of longer
pulse duration (65ns) but with reduced extend, i.e. attended by a lower capability to level down the ablation
threshold.

(a) (b)

Figure 5. (a) Minimum pulse energy for achieving insulation plotted against the pulse overlay; (b) Minimum pulse energy plotted against the
number of pulses per irradiated spot area with regression analysis

The incubation exponent was determined by insulation measurements and is flawed by the selected increment of
power increase. However, in a series of experiments to obtain residue-free ablation, an equivalent behaviour has
been detected. Here, the peak amplitude equated to the ablation threshold for residue-free ablation (cf. Table 1) and
with same incubation exponent, i.e. the regression curve in Figure 5.b was shifted up consistently on the y-axis.
Nevertheless, an evaluation of debris is usually a subjective grading.
 René Siebert et al. / Physics Procedia 12 (2011) 207–213 213

References

[1] Létay, G.: Modellierung von III-V Solarzellen. Ph.D. Thesis, University of Konstanz, Germany, 2003
[2] Liu, G.; Nguang, S. K.; Ren, Y.: The Simulated Dynamical Photovoltaic Array Systems. In: Information and Automation (ICIA), 2010 IEEE
International Conference on, (2010), pp. 344–349
[3] Vogt, H.: DE 199 33 703 B4 2005.07.28, Patent specification, 2005
[4] Oertel, M.; Dimmler, B.: TCO-Schichten für die CIGS-Solarmodulproduktion. In: Workshop 2005: TCO für Dünnschichtsolarzellen,
ForschungsVerbund Erneuerbare Energien, (2005), 61–65
[5] Powalla, M.; Dimmler, B.; Gross, K.-H.; Siebentritt, S.; Reineke-Koch, R.; Probst, V.: CIS-Dünnschichtsolarmodule. In: FVS Themen
(2000), pp. 22–27
[6] Schmid, M.: Optik der CuGaSe2-Solarzelle für hocheffiziente Tandemkonzepte. Ph.D. Thesis, Freie Universität Berlin, Germany, 2009
[7] Sun, B. K.; Xiang, Z.; Grigoropoulos, C. P.: Spectral optical functions of silicon in the range of 1.13-4.96 eV at elevated temperatures. In:
International Journal of Heat and Mass Transfer, 40 (1997), pp. 1591–1600
[8] Nagasawa, M.; Shionoya, S.: Electrical and Optical Properties of Reduced Stannic Oxide Crystals. In: Japanes Journal of Applied Physics, 10
(1971), pp. 473–475
[9] Lide, D. R.; Haynes, W. M.: Handbook of Chemistry and Physics. CRC Press, Gaithersburg, 2009
[10] Binnewies, M.; Milke, E.: Thermochemical Data of Elements and Compounds.Wiley-VCH Verlag, Weinheim, 1999
[11] Cordfunke, E. H. P.; Westrum Jr., E. F.: The heat capacity and derived thermophysical properties of In2O3 from 0 to 1000 K. In: J. Phys.
Chem. Solids, 53 (1992), pp. 361–365
[12] Horn, A.; Weichenhain, R.; Kreutz, E.-W.; Poprawe, R.: Dynamics of laser-induced cracking in glasses at a picosecond time scale. In: Proc.
SPIE 4184, (2001), pp. 539–544
[13] Güdde, J.; Hohlfeld, J.; Müller, J.G.; Matthias, E.: Damage threshold dependence on electron–phonon coupling in Au and Ni films. In: Appl.
Surf. Sci., 127–129 (1998), pp. 40–45
[14] Bonse, J.; Rudolph, P.; Krüger, J.; Baudach, S.; Kautek, W.: Femtosecond pulse laser processing of TiN on silicon. In: Applied Surface
Science, 154–155 (2000), pp. 659–663
[15] Baudach, S.; Bonse, J.; Kautek, W.: Ablation experiments on polyimide with femtosecond laser pulses. In: Applied Physics, A 69 (1999),
pp. 395–398
[16] Bonse, J.; Wrobel, J. M.; Krüger, J.; Kautek, W.: Ultrashort-pulse laser ablation of Indium Phosphide in air. In: Applied Physics, A 72
(2001), pp. 89–94
[17] Jee, Y.; Becker, M. F.; Walser, R. M.: Laser-induced damage on single-crystal metal surfaces. In: J. Opt. Soc. Am., B 5 (1988), pp. 648–659