Comparative analysis of mechanical properties of Si substrates processed by different routes

Abstract

The purpose of this work is to check the potential of innovative processes for the Si wafers production toward the solar cell industry. Studies have been focused on a comparative analysis of mechanical properties of such wafers, since: (i) reduced wafer strength leads to a high breakage rate during subsequent handling and solar cell processing steps, (ii) cracking of solar cells has become one of the major sources of solar module failure and rejection. Therefore while developing new types of wafer materials and processing, it is essential to assess the mechanical strength of the wafers. Mechanical properties of several innovative Si based substrates are estimated. The bending strength measurements of the silicon wafer are performed using the ring-on-ring set-up coupled with a numerical model to obtain estimate of the fracture stress and the Weibull parameters of the fracture distribution. Results are presented for five different materials: sintered Si powder, standard multi-crystalline Si, Czochralski monocrystalline Si, and two types of thermal sprayed Si wafers.

Full text

Comparative analysis of mechanical
properties of Si substrates processed
by different routes
Sylvain Gouttebroze
1
, Hans Ivar Lange
2
, Xiang Ma
1
, Ronny Gløckner
3
, Behnaz Emamifard
3
,
Martin Syvertsen
2
, Michalis Vardavoulias
4
, and Alexander Ulyashin*,
1
1
SINTEF MK, Forskningsveien 1, 0314 Oslo, Norway
2
SINTEF MK, Trondheim, Norway
3
Elkem Solar, Kristiandsand, Norway
4
Pyrogenesis, Lavrion, Greece
Received 10 January 2013, revised 11 February 2013, accepted 12 February 2013
Published online 28 March 2013
Keywords mechanical strength, silicon substrate, Si powder
*
Corresponding author: e-mail AlexanderG.Ulyashin@sintef.no, Phone: þ47 93 00 22 24, Fax: þ47 22 06 73 50
The purpose of this work is to check the potential of innovative
processes for the Si wafers production toward the solar cell
industry. Studies have been focused on a comparative analysis
of mechanical properties of such wafers, since: (i) reduced
wafer strength leads to a high breakage rate during subsequent
handling and solar cell processing steps, (ii) cracking of solar
cells has become one of the major sources of solar module
failure and rejection. Therefore while developing new types of
wafer materials and processing, it is essential to assess the
mechanical strength of the wafers. Mechanical properties of
several innovative Si based substrates are estimated. The
bending strength measurements of the silicon wafer are
performed using the ring-on-ring set-up coupled with a
numerical model to obtain estimate of the fracture stress and
the Weibull parameters of the fracture distribution. Results are
presented for five different materials: sintered Si powder,
standard multi-crystalline Si, Czochralski monocrystalline Si,
and two types of thermal sprayed Si wafers.
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1 Introduction Currently, the PV market is dominated
by crystalline silicon solar cells, and about 40% of the silicon
module cost is from the silicon wafers. Therefore, a major part
of current research activity is concentrated on a search for
alternative silicon based solar cell concepts with reduced
consumption of high-purity silicon. In conventional Si wafer
based solar cell technology, most of the Si material acts as a
mechanical carrier for the solar cell structures. However, since
most of the optical absorption in Si takes place in the upper 15–
30 mm, it is sufficient to use only thin Si layers with thicknesses
in this range. Indeed, if special optical confinement schemes are
implemented, even thinner layers can be used [1]. In general,
thin-film photovoltaics are assumed to become a market
dominating technology in the long term and any development
in this field is extremely important for the PV industry [2].
To be able to reduce thin Si film solar cell cost, both
material and material fabrication costs must be reduced. This
can be achieved by growing a high quality ‘‘expensive’’ thin
active crystalline Si layer onto a less expensive substrate.
Ceramic or glass based materials have been proposed as such
substrates. Such option is still envisaged to be cheaper than
the use of conventional Si-wafer substrates, as thin film
PECVD and PVD deposition processes, for example, have
been routinely used to produce high quality thin films for
electric and optoelectronic devices at accepted consumer
costs. Depending on process conditions, thin Si based-layers
of amorphous (a-Si), hydrogenated amorphous (a-Si:H),
microcrystalline (mc-Si), or polycrystalline (poly-Si) silicon
can be grown on such substrates. For mc-Si-based solar cell
structures deposited on glass substrates, energy conversion
efficiencies up to 10% have been demonstrated [3]. At
the same time solar cells utilizing thin-film polycrystalline
Si with an optimum thickness about 20 mm can achieve
photovoltaic power conversion efficiencies greater than 19%
[4]. However, the use of ceramic and glass substrates for
thin Si solar cells have some problems:
Conductivity: Glass and ceramic substrates are non-
conductive. Thus, a lot of attention has been directed towards
Phys. Status Solidi A 210, No. 4, 777–784 (2013) / DOI 10.1002/pssa.201300003 pss
applications and materials science
a
status
solidi
www.pss-a.com
physica
Part of Special Issue on
Advanced Concepts for Silicon Based Photovoltaics
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

making a highly conductive back side electrode as well as a
system for contacting the electrode after deposition of silicon
[5]. Diffusion of aliovalent dopants from the electrode is a
big problem. Conducting SiC is being developed as an
alternative [6]. However, still a number of problems have to
be solved in this approach, since crystallization of Si layers
on SiC substrates is rather problematic.
Lattice matching: Glass and ceramic substrates have no
crystallographic relation to silicon to aid crystallization at
lower temperatures and/or into higher crystallographic
quality. Thus, higher temperatures are needed when using,
e.g., a higher cost single crystal Si wafer.
Purity: Care must be taken to avoid diffusion of dopants
into Si during deposition and post-processing.
Temperature stability: One needs >700–1000 8C for
growth of poly-Si. This is marginal for the case of glass
substrates. Therefore lower temperatures are used for such
processes.
It can be concluded that there is a demand for advanced
low-cost substrates, which can be used for thin Si-based
solar cell structures. So far such low-cost Si supporting
substrates have been based on highly doped Si wafers
processed in the same way as conventional Si wafers,
i.e., by crystallization of ingots and wafering. Such
conventional processing of Si wafers can be substituted
by a cost-effective powder-to-wafer processing using
ceramics technology, hence avoiding costly crystallization
and wafering steps. Such ‘‘powder-to-wafer’’ approach,
can simplify the wafer based processing of the supporting
Si substrates and therefore relevant thin Si-based solar
cells, Usage of a low grade Si feedstock can reduce the
cost of Si wafers even further. These low purity substrates
can be processed from a low-quality Si powder. When
the powder is shaped and sintered into an appropriate
substrate, it can be considered as a poly-Si seeding material,
which can provide good crystallization conditions for
any Si-based layers deposited and annealed at appropriate
conditions.
Such substrates have several advantages compared to
ceramic or glass substrates: (i) highly doped silicon is
conductive. Hence, the substrate can be used as electrode,
avoiding any contact problems; (ii) a perfect lattice match
will lead to the lowest possible crystallization temperatures
for the deposited high purity thin film, (iii) silicon-based
low-cost supporting substrates are fully compatible with the
deposition/crystallization processes of thin Si layers on top
of such substrates.
The concept of crystalline silicon thin-film solar
cells on low-purity substrates provides a hope to reduce
substantially the consumption of high-purity silicon
material and has at the same time potential to reach high
efficiencies comparable to wafer silicon solar cells. The
goal of this article is to test mechanical properties of Si
wafers, processed in frame of an innovative approach
basedonathermalsprayofSipowderandsinteredSiand
to compare both Si powder-based substrates with those,
which are processed by casting or CZ growth methods.
This article firstly describes the material and samples
preparation. Secondly the electrical properties are quickly
studied. The measurements and the associated numerical
model are presented in the next section. Finally the results
are discussed as well as possible explanations for the
discrepancies in the material properties and their link to
processing routes.
2 Samples material and preparation
2.1 Materials and processing techniques The
original powder (Si feedstock) for the substrates comes from
a by-product stream of the Elkem Solar metallurgical process
route to produce Elkem Solar Silicon
1
in a form of Si
powder. The quality is in-between the process starting-point:
metallurgical grade silicon and the end-point: solar grade
silicon. Several types of Si powders in the wide range of
particle size distribution (PSD) have been tested. The PSD
values were varied in the range of 1–150 mm and were
adjusted to the special requirements of each processing
techniques used. The following techniques, which utilized
as produced Si powder directly without any additional
processing or cleaning (like HF dip) steps have been tested:
(i) casting of Si ingot from Si powder using a pilot scale
vertical gradient freeze (VGF) furnace and (ii) thermal spray
of Si powder. Important to underline that in both cases
Si powder was covered by native oxide and processing
conditions were optimized to reduce this oxide upon the
fabrication. In case of a thermal spray technique PSD values
were below 100 mm. As-processed Si powder was used
for sintering of Spray-1 Si substrates. For the other type
of sprayed Si substrates 3% (weight) of Al powder was
mixed with the Si powder-based feedstock prior the spraying
(Spray 2 substrates).
In case of the casting process, which was carried out
using a VGF furnace at SINTEF, Si powders with PSD
values up to 150 mm have been used. The silicon powder-
based feedstock was melted and solidified. The thermal
conditions and mass transfer in this furnace have been
studied thoroughly the latest years [7, 8]. This required
special care during furnace operation.
Since the resistivity of the final substrates were
supposed to be relatively low (<0.01 Vcm), 1.5 g of 95–
97% pure boron was added to the charge before melting.
Even though – according to phase Si-boron diagrams [9] the
melting point of pure boron is about 2100 8C, the boron will
dissolve completely in the liquid silicon at 1430 8Csince
the solubility is about 10% and the added amount is merely
160 ppm.
The logged data from the casting are shown in Fig. 1.
A complete description of the operation of the furnace
under normal conditions has been given earlier [10]. The
main differences between standard furnace operation and
the one done in this study are:
1. Double layer of coating was used to ensure no sticking
between ingot and crucible even with long time at high
temperature.
778 S. Gouttebroze et al.: Mechanical properties of Si substrates processed by different routes
physica
ssp
status
solidi
a
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-a.com

2. Since the feedstock was in the form of powder, the initial
pump-down phase was done very gentle in order to avoid
sucking powder out of the crucible.
3. Gas flow after the vacuum phase was reduced in order not
to blow feedstock out of the crucible.
4. After complete melting, the melt was held for 6.5 h longer
than normal in order to secure complete dissolution of
the boron.
2.2 Wafer samples preparation Highly conduc-
tive pþSi wafers from casted Si ingots were obtained by
wire sawing.
Additionally series of Si wafers were produced by thermal
spray route, with thicknesses between 300 and 1000 mm, and
dimensions 65 50 mm
2
have been prepared from nominally
un-doped low-cost Si powder. Silicon layers (wafers) were
detached from specially selected and prepared substrates/
moulds without breaking (Fig. 2). For the mechanical
properties measurements sintered wafers were cut on
11cm
2
smaller pieces by laser. Summary of samples
material and process routes can be seen from Table 1.
3 Measurement of electrical properties Resis-
tivity measurements have been performed by the four-probe
method. It has been found that Si wafers fabricated from
the casted Si ingot have resistivity <0.01 Vcm. Such
wafers can be used as highly supporting ones for thin
Si-based solar cells. Si wafers processed by thermal spray
from nominally non-doped low-cost Si powder demon-
strated resistivity in the range of 1–10 Vcm, which shows
that sintering process upon thermal spray has been done
properly and no barriers between grains have been created.
Since this work is focused on mechanical properties of Si
powder-based substrates, doping issues for thermal sprayed
substrates were not addressed. However, it can be noted
that processing of such wafers using highly doped Si
powder should result in highly doped and therefore highly
conductive substrates.
4 Measurement of mechanical properties
4.1 Experimental set-up In order to assess the
mechanical strength of the wafer produced by the new
process routes, the ring-on-ring test was used. Based on the
recommendations of the ASTM standard C1499-08 [11],
a smaller experimental set-up has been designed as
illustrated in Fig. 3. This set-up has been selected to measure
the material intrinsic strength and avoid the effect of edge
defects.
Phys. Status Solidi A 210, No. 4 (2013) 779
Original
Paper
Figure 1 (online color at: www.pss-a.com) Data from the
Crystalox furnace during casting. Shown are the heater temperature,
pedestal temperature (temperature beneath crucible), heater power,
and shutter opening during crystallization. M, D, and S denote
melting, dissolution of boron and mixing, and solidification
phases, respectively.
Figure 2 (online color at: www.pss-a.com) Si wafers (65 50 mm
2
)
fabricated by thermal spray.
Table 1 Summary of samples material and process routes.
sample name material process route
Si powder-based
multi
Elkem powder ingot sintering/VGF
furnace
P-type multi solar grade Si conventional casting
as cut P-type C7 solar grade Si Czochralski growth
Spray 1 Elkem powder/100% Si thermal spray
Spray 2 Elkem powder
3% Al/97% Si
thermal spray
Figure 3 (online color at: www.pss-a.com) Schematic view of the
experimental set-up.
www.pss-a.com ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

4.1.1 Testing equipment The first tests were per-
formed in the Bose Electroforce 250N test machine on
lower speeds (i.e., 0.1–0.2 mm min
1
). Due to irregular
loading rate at the lower speeds, the speed was increased to
1 mm min
1
. Nevertheless irregular jumps of the machine
remains and it led to machine change for the rest of the study.
Therefore only Si powder-based multi samples have been
tested using Bose Electroforce 250N machine. An Instron
2kN test machine was used instead in all the remaining
experimental work presented in this paper (P-type multi,
As cut P-type C7, Spray 1 and Spray 2).
The fixture was made of heat-treated steel, hardened to
about 40 HRC.
The data acquisition sampling rate was 100 Hz for the tests
done in the Bose machine and 15 Hz for the Instron machine.
4.1.2 Calibration Both test machines were recently
calibrated by the manufacturer. Due to limited utilization of
the load cell of the Instron machine an additional control was
performed with one of the reference load cells of the test
laboratory. Check of coaxiality of the test fixture was done
during mounting, both by visual inspection and by using pre-
machined pin holes in the center of the upper and lower
fixture and a guiding pin. The plane parallelism of the upper
and lower ring was corrected before each test by visual
inspection of a light gap of about 0.25 mm between the
upper and lower ring. The light gap was checked in two
perpendicular directions. The fixture was thoroughly cleaned
before each test in order to prevent silicon debris deposition
on the ring surfaces.
4.2 Numerical model A numerical model of the ring-
on-ring mechanical test has been established. The main
objectives were to assess the uniformity of the stress
distribution inside the loading ring area (especially at the
contact point) and develop a formula to compute directly
the maximal principal stress for the displacement and the
sample thickness.
The model was developed using Abaqus/Standard 6.11
with implicit time integration. Only a quarter of the sample
and rings is modeled (see Fig. 4). The sample mesh size is
0.05 mm. The material properties are indicated in Table 2.
The dimensions and element type are provided in Table 3.
As illustrated in Fig. 5, considering the thinnest sample
which is the most critical in terms of stress distribution, the
stress inside the loading ring is quite homogeneous and
a relatively small increase (5%) is observed only in the
contact area with the loading ring. Therefore the dimensions
of the sample and the rings are adequate for wafer strength
measurements.
By applying the model to various thicknesses (0.20, 0.39,
0.6, and 0.8 mm) and performing a regression analysis, we
obtained the following formula:
s¼2616hþ3067ðÞd2þ14 830h465ðÞd;(1)
780 S. Gouttebroze et al.: Mechanical properties of Si substrates processed by different routes
physica
ssp
status
solidi
a
Figure 4 (online color at: www.pss-a.com) Abaqus model of the
ring-on-ring test.
Table 2 Silicon properties for the numerical model.
material silicon
model isotropic elasticity
Young’s modulus 162.5 GPa
Poisson’s ratio 0.223
Table 3 Numerical model dimensions and element type.
part shape finite element model
silicon specimen square
a
(10 10) 8-nodes brick element
loading ring ring
b
(4.5 0.5) shell, discrete rigid
support ring ring
b
(9 0.5) shell, discrete rigid
a
Length and width in mm.
b
Diameter and curvature radius in mm.
Figure 5 (online color at: www.pss-a.com) Stress distribution
along the cross-section of the sample (thickness: 0.2 mm).
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-a.com

where sis the stress at the sample center in MPa, his sample
thickness in mm, and dis the displacement in mm.
5 Results and analysis The raw data from the series
of mechanical tests using the ring-on-ring set-up previously
described are presented in Figs. 6–10. The displacement of
the ring is given in the x-axis while the y-axis provides the
applied force measured by the load cell. As mentioned in
the Section 4.1, the curves of the initial set-up (Fig. 6) using
the Bose Electroforce 250N test machine has a stair case
appearance. Nevertheless it does affect neither the curve
shape nor the displacement at fracture. The main issue in this
series of test is the large variation in the slope of the curves
before the peak force value is reached. We would expect for
such material (except maybe for the spray samples) a very
reproducible elastic response. Indeed we have previously
measured the strength of the Si powder based multi samples
using the 4pt-bending test and the curves were overlapping.
The reason for this discrepancy is not clearly identified but
there are two possible explanations. First, as mentioned by
Cotterell et al. [12], the proper positioning of the samples for
the ring-on-ring test could be an issue. Friction between the
ring and the sample can lead to inaccurate results. Therefore,
Wasmer et al. [13] recommend the use of carbon paper.
Please note that the Weibull fit of the experimental data
without carbon paper from Wasmer et al. shows a similar
staircase aspect as some of our results (especially Fig. 12
where one can observe a slope change for the experimental
curves between a probability of 0.4 and 0.5). It is also
important to note that this friction effect might be dependent
on sample roughness which varies between our materials.
Nevertheless we assume that these effects induce only a
systematic error on the force measurements. Therefore the
analysis of the results based on the displacement should
Phys. Status Solidi A 210, No. 4 (2013) 781
Original
Paper
Figure 6 Force displacement curves for the Si powder-based
multi samples (28 samples).
Figure 7 Force displacement curves for the P-type multi samples
(23 samples).
Figure 8 Force displacement curves for the As cut P-type C7
samples (25 samples).
Figure 9 Force displacement curves for the Spray 1 samples
(16 samples).
www.pss-a.com ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

provide good estimate of the samples strength for compari-
son between different materials.
The raw data has been post-processed using Eq. (1).
Then the stress values at fracture were ordered and related to
a probability in order to build the Weibull distribution.
Indeed the standard parameters in the literature to assess
the wafer strength are the parameters based on the Weibull
distribution.
The probability of fracture is described by an exponen-
tial function with two parameters:
PðsÞ¼1eðs=s0Þm:(2)
From the ordered values, a curve fitting procedure has
been applied in Excel. The results for the three first materials
are presented in Figs. 11–13. In addition, previous results
with Si powder based multi samples obtained with 4pt-
bending set-up and larger samples (5 5mm
2
) are also
presented for comparison.
As illustrated by Figs. 9 and 10, the material behavior of
the spray samples is significantly different. Therefore the
direct application of the previous procedure is not possible
and the study of the sprayed samples requires a more detailed
analysis of the material. In all the previous cases we assume
that the material properties are similar to bulk silicon
which is a quite good approximation. Nevertheless for
the sprayed samples, the microstructure of the material is
non-homogeneous which is mainly due to the presence of
porosity and layers.
In order to obtain a better estimate of the equivalent
fracture stress for this material, we will need to first estimate
the porosity and subsequently estimate the relative correc-
tion of the stress value resulting from this porosity. As
illustrated on Fig. 14, the metallography image for Si sample
Spray 1 is processed using a simple threshold method in
order to detect the holes and estimate the porosity.
This procedure is not the most advanced and provides only
an estimate of the porosity. Using this technique we obtain a
782 S. Gouttebroze et al.: Mechanical properties of Si substrates processed by different routes
physica
ssp
status
solidi
a
Figure 10 Force displacement curves for the Spray 2 samples
(16 samples).
Figure 12 (online color at: www.pss-a.com) Strength measure-
ments and Weibull fit for P-type multi samples (s
0
¼264 MPa and
m¼7.5).
Figure 11 (online color at: www.pss-a.com) Strength measure-
ments and Weibull fit for Si powder based multi samples using ring-
on-ring test (s
0
¼158 MPa and m¼5.9) compared with 4pt-bending
experimental measurements (s
0
¼133 MPa and m¼14).
Figure 13 (online color at: www.pss-a.com) Strength measure-
ments and Weibull fit for P-type multi samples (s
0
¼283 MPa and
m¼6.4).
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-a.com

value of 11% porosity, which agrees with the results obtained
for other type of powder based materials (yttria-stabilized
zirconia coatings) processed by thermal spray [14].
In the work by Wang et al. [14], different spraying
techniques are used with increasing level of porosity and the
effect on the elasticity modulus for each sample is measured.
Two values are closed to our porosity estimate: 7.6% and
12.2% porosity for which the elasticity modulus is reduced to
45% and 25% of the bulk value, respectively. From these
values, we obtain by simple interpolation a decrease to 30%
of the elasticity modulus of bulk silicon. Various materials
sprayed using high velocity oxygen fuel technique (HVOF)
ranged also between 30% and 40% of the bulk elastic
modulus [15].
Therefore, in order to estimate the fracture stress using
the previous formula, we apply a corrective multiplicative
factor of 0.3 to the fracture stress computed from the
displacement at rupture (Eq. (1)). The results are presented in
Figs. 15 and 16. Please note that these values are sensitive to
the sample thickness (not precisely measured over the whole
surface of the sample and to the porosity (estimated for only
one sample) and therefore this analysis will provide only
a coarse estimate of the material strength. In any case, as
presented by Margadant et al. [16], it is very difficult to
obtain an accurate value of the elastic modulus of a coating as
it depends on the measurement and spraying technique used.
For comparison the results for the five materials are
summarized in Table 4. As expected the variance for all
materials is in the same range except for the 4pt-bending test
where usually a larger variance is obtained.
It is also worth mentioning the significant increase
between Spray 1 and Spray 2 samples. This might be
explained by an influence of 3% of Al powder, which is
mixed with Si powder based feedstock on mechanical
properties of sprayed Si wafers. More detailed studies are
required to investigate this effect.
6 Conclusions Comparative analysis of mechanical
properties of Si samples obtained from new process routes
for the production of silicon wafers has been presented.
Characterization of the mechanical strength of the different
silicon samples was investigated using a ring-on-ring test.
The maximum principal stresses at failure during bending
were calculated to indicate the fracture strength and fitted to a
Weibull distribution. The study showed that:
In spite of some experimental challenges, the ring-on-ring
set-up provided acceptable estimate of the material for
various materials and process routes.
Phys. Status Solidi A 210, No. 4 (2013) 783
Original
Paper
Figure 14 Comparison between initial metallography image
(500) on the left and processed image for porosity estimate on
the right.
Figure 15 (online color at: www.pss-a.com) Strength measure-
ments and Weibull fit for P-type multi samples (s
0
¼290 MPa and
m¼4.8).
Figure 16 (online color at: www.pss-a.com) Strength measure-
ments and Weibull fit for P-type multi samples (s
0
¼303 MPa
and m¼9).
Table 4 Summary of the Weibull fit of the ring-on-ring tests
(or 4pt-bending when specified).
material Weibull mean
stress, s
0
Weibull
variance, m
Si powder based multi 158 5.9
Si powder based multi
(4pt bending)
133 14
P-type multi 264 7.5
as cut P-type C7 283 6.4
Spray 1 290 4.8
Spray 2 303 9
www.pss-a.com ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

The Si powder based multi samples were the weaker but
still with a reasonable strength comparable to standard
wafer without etching.
The spray samples display a specific material behavior due
to porosity and layering. Nevertheless the final material
strength is high.
As a summary, this work has demonstrated the potential of
new process routes leading to the production of Si wafers with
adequate material properties both electrical and mechanical.
It can be concluded that Si powder-to-substrate approach
can be utilized for the processing of Si based supporting
substrates, which potentially are fully compatible with the
deposition/crystallization processes of thin Si layers on top
of such substrates and posses comparable with multi-Si/Cz-
Si substrates mechanical properties.
Further work should be performed to analyze in more
details the microstructure of the new materials and assess
their downstream integration in the solar cell production.
Acknowledgements The authors have received funding
from the European Community’s Seventh Framework Program
(FP7/2010-2013) under grant agreement number 241281 (ThinSi).
References
[1] R. Brendel and D. Scholten, Appl. Phys. A 69, 201 (1999).
[2] www.eupvplatform.org.
[3] A. V. Shah, J. Meier, E. Vallat-Sauvain, N. Wyrsh, U. Kroll,
C. Droz, and U. Graf, Sol. Energy Mater. Sol. Cells 78, 469
(2003).
[4] A. M. Barnett, J. A. Rand, R. B. Hall, J. C. Bisaillon, E. J.
DelleDonne, B. W. Feyock, D. H. Ford, A. E. Ingram, M. G.
Mauk, J. P. Yaskoff, and P. E. Sims, Sol. Energy Mater. Sol.
Cells 66, 45 (2001).
[5] E. Schmich, N. Schillinger, and S. Reber, Surf. Coat. Technol.
201, 9325 (2007).
[6] S. Janz, S. Reber, F. Lutz, and C. Schetter, Thin Solid Films
511–512, 271 (2006).
[7] E. A. Meese, E. J. Ovrelid, H. Laux, and M. M. Hamdi, 20th
European Photovoltaic Solar Energy Conference, Barcelona,
Spain (2005), pp. 909–913.
[8] H. Laux, E. A. Meese, E. J. Ovrelid, and M. M. Hamdi, 20th
European Photovoltaic Solar Energy Conference, Barcelona,
Spain (2005), pp. 1090–1094.
[9] T. B. Massalski, H. Okamoto, P. R. Subramanian, and
L. Kacprzak, ASM Binary Phase Diagrams, 2nd ed. (ASM
International, Materials Park, Ohio, USA, 1990).
[10] Ø. Mjøs, Directional solidification of silicon for solar cells,
PhD thesis, NTNU, Norway, 2006.
[11] Standard Test Method for Monotonic Equibiaxial Flexural
Strength of Advanced Ceramics at Ambient Temperature,
ASTM C1499-08.
[12] B. Cotterell, Z. Chen, J. B. Han, and N. X. Tan, J. Electron.
Packag. 125, 114–119 (2003).
[13] K. Wasmer, A. Bidiville, J. Michler, C. Ballif, M. Van der
Meer, and P. Nasch, 22nd European Photovoltaic Solar
Energy Conference, Milan, Italy (2007).
[14] Z. Wang, A. Kulkarni, S. Deshpande, T. Nakamura, and
H. Herman, Acta Mater. 51, 5319 (2003).
[15] J. Pina, A. Dias, J. L. Lebrun et al., Mater. Sci. Eng. A 267,
130 (1999).
[16] N. Margadant et al., Surf. Coat. Technol. 200, 28050 (2006).
784 S. Gouttebroze et al.: Mechanical properties of Si substrates processed by different routes
physica
ssp
status
solidi
a
ß2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-a.com