Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Journal of Crystal Growth 318 (2011) 230–233

Contents lists available at ScienceDirect

Journal of Crystal Growth


journal homepage: www.elsevier.com/locate/jcrysgro

Improved fracture strength of multicrystalline silicon by germanium doping


Peng Wang, Xuegong Yu, Zhonglan Li, Deren Yang n
State Key Laboratory of Silicon Materials and Department of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, People’s Republic of China

a r t i c l e i n f o a b s t r a c t

Available online 24 November 2010 The impact of germanium doping on the fracture strength of multicrystalline silicon (mc-Si) has been
Keywords: investigated by three-point bending testing. It is found that after the damaged layer removal by chemical
A1. Germanium doping etching, germanium doped multicrystalline silicon (Gmc-Si) wafers show significantly improved fracture
A1. Fracture strength strength compared to conventional mc-Si ones. Moreover, the improvement of the percentage of the
A2. Multicrystalline silicon fracture strength increases with decrease in thickness of the etched wafers. This suggests that the fracture
B3. Solar cells toughness of mc-Si wafers is enhanced by germanium doping. The results are of interest for solar cells
production yields improvement in the photovoltaic industry.
& 2010 Elsevier B.V. All rights reserved.

1. Introduction In this work, we have investigated the effect of germanium


doping on the fracture strength of multicrystalline silicon (mc-Si)
With the trend towards using thinner and larger area silicon wafers. By comparing to conventional mc-Si, the impact of
wafers by the photovoltaic (PV) industry, a significant loss in germanium doping on the breakage of wafers during multi-wire
production yields is unavoidable, due to an increase in wafer and sawing is presented. Then the influence of germanium doping on
solar cell breakage [1,2]. The latter problem is particularly impor- the fracture strengths of as-sawn and chemically etched mc-Si
tant because current PV companies try to lower the production cost wafers is well demonstrated. Finally, the Weibull analysis is used to
by achieving higher conversion efficiency with decreasing silicon describe the statistical fracture strength data.
wafer thickness simultaneously. Consequently, it is essential to
understand the fracture behavior of thin silicon wafers and to
develop ways for their strength improvement. At room tempera- 2. Experiment
ture, silicon is a brittle material and neither plastic deformation nor
dislocation generation is observed when the applied stress exceeds One conventional casting mc-Si ingot and one Gmc-Si ingot with
a certain limit [3]. Its fracture strength is limited by the structural additional germanium doping of about 1019 cm  3 were grown
defects, which can serve as stress concentrators [4–6]. For PV under almost the same conditions. Then, both multicrystalline
silicon wafers, microcracks are the most frequent stress concen- silicon ingots were cut into 156  156 mm2 blocks, followed by
trators and the fracture strength of silicon is dominated by the wafering using multi-wire sawing. The wafer thickness was about
density and size of microcracks [5]. Therefore, most studies focused 220 mm. The percentage of breakage of wafers during the multi-
their efforts on the impact of surface or edge damage, i.e., biaxial wire sawing process was statistically monitored.
flexure test [7] is used to investigate the quality of the wafers For clarifying the influence of germanium on wafer mechanical
surface, and four-point bending test [3] is applied to probe both strength, the surface roughness of the as-sawn samples was first
surface and edges of the wafers. Moreover, fracture of brittle measured by employing a Veeco WykoTopo-3D optical profil-
material exhibits statistical behavior and the Weibull analysis is ometer (NT-9300). Note that for each kind of samples, 20 areas have
applied to demonstrate the statistical fracture data [3,7,8]. been measured to determine the average roughness by the Vertical
Recently, it was reported that germanium doping could improve Scanning Interferometry (VSI) mode. The average roughness (Ra) is
the mechanical strength of the Czochralski silicon [9,10] and calculated over the entire measured area and is commonly used as a
suppress dislocation generation and mobilization [11]. However, surface condition parameter. Then, the fracture strengths of thin
little attention was given to the impact of germanium doping on the mc-Si and Gmc-Si wafers from comparable positions (top, middle,
brittle fracture behavior of silicon, especially for multicrystalline and bottom) in the ingots were investigated. Note that rectangular
silicon, which is of relevance for the PV industry. samples of 10  20 mm2 were sliced by laser cutting from those
wafers. Parts of them were chemically etched to different thick-
nesses with 160 and 110 mm in the white etchant (HNO3:HF¼3:1),
n
Corresponding author. Tel.: + 86 571 8795 1667; fax: + 86 571 8795 2322. respectively. Before and after etching, the thickness of each wafer
E-mail address: mseyang@zju.edu.cn (D. Yang). was measured. The fracture strengths of the as-sawn and

0022-0248/$ - see front matter & 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jcrysgro.2010.11.081
P. Wang et al. / Journal of Crystal Growth 318 (2011) 230–233 231

Table 1
Statistical breakage rate of mc-Si and Gmc-Si wafers during multi-wire sawing
process.

Bach no. Total number (broken number) Breakage rate

mc-Si Gmc-Si mc-Si (%) Gmc-Si (%)

1 8313 (9) 7315 (2) 0.11 0.03


2 6610 (29) 7019 (25) 0.44 0.36
3 1142 (7) 2095 (16) 0.61 0.76
Fig. 1. Three-point bending test setup used for measuring the fracture strength of Overall 16,065 (45) 16,429 (43) 0.28 0.26
wafers.

chemically etched samples were measured by a universal testing


machine (Zwick/Roell Z2.5) using the three-point bending method.
The load span L ¼12 mm. The tests were carried out using a loading
speed of 0.5 mm/min. The three-point bending test setup is shown
in Fig. 1.

3. Results and discussion

The breakage rates for mc-Si and Gmc-Si ingots during sawing
process were statistically compared, and listed in Table 1. It was
found that the breakage rate shows a large scatter and the breakage
rates overall for both ingots were nearly the same. The wafer cut
during sawing can be regarded as a series of micro-indentations
accompanying with cracks generation, and materials is continu-
ously removed when these cracks propagate, meanwhile, the
cracks reaching into the surface form part of the saw damage
[12]. Many sawing parameters such as SiC particle size, sawing
load, and sawing speed determine the maximum depth of micro-
cracks [5], therefore, it is deduced that the strength of as-sawn Fig. 2. Example of experimental force–displacement curves of as-sawn and
wafers is mainly related to such parameters and germanium chemically etched (160, and 110 mm) samples of mc-Si and Gmc-Si wafers.

doping hardly plays a role at this condition.


To further clarify the impact of germanium on the mechanical
strength of mc-Si, the fracture strengths of as-sawn and chemically
etched samples from mc-Si and Gmc-Si wafers were investigated. A
typical force–displacement curves per sample type investigated is
shown in Fig. 2. The breakdown of the samples almost behaves
linearly; hence, linear elastic stress analysis using the three-point
bending test was performed. The fracture strength is obtained via
the equation,

3FL
s¼ ð1Þ
2bh2
where F is the peak force (at which, cracks propagate and the
sample breaks), L is the span, b is the width and h is the thickness of
the sample, s is the maximum tensile stress.
Due to the random distribution of defects in brittle materials,
the measured strengths show large variability. Hence, the Weibull
theory [13] is commonly used to analyze the experimental results.
The Weibull theory is applied to describe the probability of failure
by using two parameters, and is given by
  m Fig. 3. Weibull’s plot showing the fracture strength distribution of as-sawn samples
s
Pf ¼ 1exp  ð2Þ from different locations (Top-T, Middle-M, and Bottom-B) of mc-Si and Gmc-Si
sy ingots under testing.

where Pf is the probability of failure at the applied stress, s, and sy is


the characteristics strength and m is the Weibull modulus. A number of samples have been tested directly after sawing
Pf ¼(n  1/2)/N dependent on its ranking number, n (N is the total from comparable positions of the Gmc-Si and mc-Si ingots. The
number of samples). The characteristic strength, sy , is the stress at results of the fracture strength distribution with the Weibull
which 63% of all samples fail. The Weibull modulus, m, describes plotting are shown in Fig. 3. The experimental data can be fitted
the variability of strength data. The higher the value of the Weibull well using a Weibull plot, and the Weibull parameters obtained and
modulus, the smaller the scattering of the measured strengths. By the number of samples measured are summarized in Table 2. The
plotting In (  In(1 Pf)) versus In(s), a linear fit will allow the fracture strength of each position of both ingots is very small for as-
characteristic strength and the Weibull modulus to be determined. sawn wafers. The three-point bending test generates high tensile
232 P. Wang et al. / Journal of Crystal Growth 318 (2011) 230–233

Table 2
Weibull’s parameters and number of samples measured (N).

Location Wafer As-sawn 160 mm 110 mm

s0 (MPa) m N s0 (MPa) m N s0 (MPa) m N

Top mc 145.2 5.8 12 501.7 2.1 12 547.3 3.8 12


Gmc 133.8 6.6 12 575.4 4.6 12 696.5 4.0 11
Middle mc 103.8 4.2 12 476.3 4.0 12
Gmc 95.1 5.0 12 591.7 2.7 12
Bottom mc 121.4 15.2 12 500.7 2.5 11
Gmc 115.4 20.7 12 526.4 3.7 11
Average mc 125.8 4.9 36 491.8 2.7 35 547.3 3.8 12
Gmc 117.6 5.3 36 565.1 3.6 35 696.5 4.0 11

Fig. 5. Weibull’s plot showing the fracture strength distribution of chemically


etched (160 and 110 mm) samples from different locations (Top-T, Middle-M, and
Bottom-B) of mc-Si and Gmc-Si ingots under testing.

sawing, more or less existing on the surface of every wafer. Even


though these stripes can cause some disturbing for the determina-
tion of surface roughness, an average roughness of 20 areas used
here should improve the accuracy of present data. Nevertheless, an
average roughness (Ra) of 317.8 nm has been obtained for mc-Si
and 499.8 nm for Gmc-Si wafers. Therefore, a larger surface
roughness could contribute to the smaller characteristic strength
for as-sawn Gmc-Si wafers.
After removing about 60 and 110 mm of as-sawn samples by
etching, the thickness of each sample is measured and this
thickness is used for determining the fracture strength. The fracture
Fig. 4. Interferometric images of the surface roughness of as-sawn wafers. (a) mc-Si strength distribution of samples from comparable positions of the
wafer and (b) Gmc-Si wafer.
two ingots is shown in Fig. 5 and Table 2. In this case, the Weibull
modulus decreases, which indicates that an increase in scatter of
stress in the middle of the surface as well as at the edge of the strength values around the mean strength is obtained. The
sample. As a result, these low values of fracture strengths indicate characteristic strengths are improved significantly compared to
that the strengths of as-sawn silicon wafers are to a large extent that of as-sawn samples. Moreover, Gmc-Si samples exhibit
determined by surface and edge damage introduced during the notably larger fracture strength compared to mc-Si ones, and this
multi-wire sawing process. Moreover, the relatively high Weibull enhancement of strength increases with a decrease of wafer
modulus implies that the distribution of microcracks is narrow. The thickness. As shown in Table 2; however, no significant difference
characteristic strength of Gmc-Si seems to be smaller than that of between each position of mc-Si ingot is found. Larger strength is
mc-Si for as-sawn wafers at each position of both ingots, also observed in the top and middle of the Gmc-Si ingot. The segregation
shown in Fig. 3. This should be attributed to a difference in surface coefficient of germanium in silicon is lower than unit (K ¼0.33 [14],
roughness of mc-Si and Gmc-Si wafers. Fig. 4 shows the interfero- and 0.68 [15] reported), causing an accumulation of germanium
metric images of the surface roughness of as-sawn mc-Si and Gmc- towards the last solidified part of the ingot. This confirms that
Si wafers. Note that some stripes appear on both sample surfaces, germanium doping can improve the fracture strength of mc-Si. To
which are so called ‘‘saw marks’’. They are actually the traces our knowledge, this is the first time that this strengthening effect of
caused by wire sawing, and generally cannot be avoided during germanium doping is reported for mc-Si. However, it should be
P. Wang et al. / Journal of Crystal Growth 318 (2011) 230–233 233

noted that in Table 2, the characteristic strength of Gmc-Si wafers wafers, the characteristic fracture strength is quite small and is
at the middle is larger than that at the top. The reason is not clear for determined mainly by the surface and edge defects introduced
us. Since it has been reported that the germanium with a high during sawing process. While after chemically etching down to 160
concentration above a critical value will cause a cellular growth in and 110 mm thickness, respectively, the characteristic strength
CZ silicon [16]. Thus, we suppose that the concentration of Ge increases significantly and Gmc-Si exhibits an improved fracture
segregating at the last casting part could be too high. This could strength compared to conventional mc-Si. Furthermore, the
cause sufficient defects, which will to some extent weaken the improvement of the percentage of the fracture strength increases
mechanical strength of Gmc-Si wafers at the top. with decrease in thickness of the etched wafers. The observations
Silicon shows brittle fracture behavior at room temperature. indicate that Gmc-Si has higher fracture strength than conven-
The generation of a crack is accompanied by breaking atomic bonds tional mc-Si and thus might have promising applications in the PV
and the formation of new surfaces. Both simulation and theoretical industry for improvement in the production yield of solar cells.
predictions verify that the fracture toughness is equal to the energy
consumed by creating surfaces and lattice defects [17]. For the
same material, the fracture strength is ultimately controlled by Acknowledgement
the presence of defects that act as stress concentrators [18]. The
relation between fracture strength, the fracture toughness and the This work is supported by National Natural Science Foundation
size of the critical defects is given by the Griffith relation [19] of China (Nos. 60906002 and 50832006), ‘973 Program’ (No.
2007CB613403) and the Fundamental Research Funds for the
K
sF ¼ pICffiffiffi ð3Þ Central Universities (2009QNA4007). The first author also thanks
Y a
Prof. Jan.Vanhellemont for his valuable suggestions.
where KIC is the fracture toughness of the material, Y is a
geometrical factor, and a is the size of the critical defect. References
During the three-point bending test, the specimen is subjected
to both compressive and tensile stresses. Both the stresses cause [1] X.F. Brun, S.N. Melkote, Analysis of stresses and breakage of crystalline silicon
the silicon to bend downwards with a tensile stress at the bottom of wafers during handling and transport, Sol. Energy Mater. Sol. Cells 93 (2009)
silicon, and a compressive stress at the upper surface. Fracture 1238–1247.
[2] P.A. Wang, Industrial challenges for thin wafer manufacturing, in: Proceedings
strength is the maximum stress at the moment of fracture. As a
of the Fourth World Conference on Photovoltaic Energy Conversion, Waikoloa,
result, the fracture strength is dictated by the surface quality on the HI, USA, 2006, pp. 1179–1182.
backside of silicon. According to the Griffith relation, the larger the [3] O. Borrero-Lopez, T. Vodenitcharova, M. Hoffman, A.J. Leo, Fracture strength of
polycrystalline silicon wafers for the photovoltaic industry, J. Am. Ceram. Soc.
size of the critical defect is, the smaller the fracture strength.
92 (2009) 2713–2717.
Therefore for as-sawn Gmc-Si and mc-Si wafers, their strengths are [4] I. Paul, B. Majeed, K.M. Razeeb, J. Barton, Statistical fracture modelling of silicon
mainly dependent on surface and/or edge defects, such as micro- with varying thickness, Acta Mater. 54 (2006) 3991–4000.
cracks. They initiate crack growth much easier, which leads to the [5] P. Rupnowski, B. Sopori, Strength of silicon wafers: fracture mechanics
approach, Int. J. Fract. 155 (2009) 67–74.
small fracture strengths of the as-sawn samples. [6] C. Funke, S. Wolf, D. Stoyan, Modeling the tensile strength and crack length of
After removing the damaged layer by etching, Gmc-Si wafers wire-sawn silicon wafers, J. Sol. Energy Eng. 131 (2009) 011012.
exhibit significantly larger fracture strengths compared to con- [7] C. Funke, E. Kullig, M. Kuna, H.J. Moller, Biaxial fracture test of silicon wafers,
Adv. Eng. Mater. 6 (2004) 594–598.
ventional mc-Si ones. A similar enhancement effect of germanium [8] H.J. Moller, C. Funke, M. Rinio, S. Scholz, Multicrystalline silicon for solar cells,
doping on the fracture strength in Cz-Si was also reported [10]. Thin Solid Films 487 (2005) 179–187.
Because the influences of surface and edge defects are negligible, it [9] T. Fukuda, A. Ohsawa, Mechanical strength of silicon crystals with oxygen and
or germanium impurities, Appl. Phys. Lett. 60 (1992) 1184–1186.
is thus inferred that the fracture toughness of mc-Si wafers might [10] J.H. Chen, D.R. Yang, X.Y. Ma, Z.D. Zeng, D.X. Tian, L.B. Li, D.L. Que, L.F. Gong,
be improved by germanium doping. The fracture toughness is Influence of germanium doping on the mechanical strength of Czochralski
related to the surface energy, which is consumed to break atom silicon wafers, J. Appl. Phys. 103 (2008) 123521.
[11] I. Yonenaga, T. Taishi, X. Huang, K. Hoshikawa, Dynamic characteristics of
bonds and form new surfaces. However, the surface energy
dislocations in Ge-doped and (Ge+ B) codoped silicon, J. Appl. Phys. 93 (2003)
depends not only on the crack plane, but also on the propagation 265–269.
direction, since the cracks propagate easily on (1 1 1) and (1 1 0) [12] H.J. Moller, Basic mechanisms and models of multi-wire sawing, Adv. Eng.
planes in the [1 1 0] direction [20]. In multicrystalline silicon the Mater. 6 (2004) 501–513.
[13] W.A. Weibull, A statistical theory of the strength of materials, Ingeniors
crack inclines to go the way energy favored, which might pass Vetenskaps Akademien Handlingar 151 (1939) 5–45.
through the grain boundaries. It is possible that germanium atoms [14] F.A. Trumbore, Solid solubilities of impurities elements in germanium and
accumulate at the grain boundaries, and therefore hinder the crack silicon, Bell Syst. Tech. J. 39 (1960) 205–233.
[15] S.N. Rea, J.D. Lawrence, J.M. Anthony, Effective segregation coefficient of
to propagate over the grain boundaries. However, further study is germanium in Czochralski silicon, J. Electrochem. Soc. 134 (1987) 752–753.
needed to investigate the distribution of germanium and the [16] T. Taishi, X. Huang, I. Yonenaga, K. Hoshikawa, Dislocation behavior in heavily
mechanism for the improvement of the fracture strength of mc- germanium-doped silicon crystal, Mater. Sci. Semicond. Process 5 (2003)
409–412.
Si wafers by germanium doping. [17] J.G. Swadener, M.I. Baskes, M. Nastasi, Molecular dynamics simulation of
brittle fracture in silicon, Phys. Rev. Lett. 89 (2002) 085503.
[18] B.R. Lawn, Fracture of Brittle Solids, Cambrige University Press, Cambrige,
4. Conclusion 1993.
[19] A.A. Griffith, The phenomena of rupture and flow in solids, Philos. Trans. R. Soc.
A. 221 (1920) 163–198.
We have demonstrated the impact of germanium doping on the [20] R. Perez, P. Gumbsch, Directional anisotropy in the cleavage fracture of silicon,
fracture strength of multicrystalline silicon wafers. For as-sawn Phys. Rev. Lett. 84 (2000) 5347–5350.

You might also like