Synchrotron-based microanalysis of iron distribution after
thermal processing and predictive modeling of resulting
solar cell efficiency
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Fenning, D. P. et al. “Synchrotron-based Microanalysis of Iron
Distribution After Thermal Processing and Predictive Modeling of
Resulting Solar Cell Efficiency.” 2010 35th IEEE Photovoltaic
Specialists Conference (PVSC), 2010. 000430–000431.
CrossRef. Web. © Copyright 2010 IEEE.
As Published
http://dx.doi.org/10.1109/PVSC.2010.5616767
Publisher
Institute of Electrical and Electronics Engineers
Version
Final published version
Accessed
Wed May 25 22:03:13 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/78325
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
SYNCHROTRON-BASED MICROANALYSIS OF IRON DISTRIBUTION AFTER THERMAL
PROCESSING AND PREDICTIVE MODELING OF RESULTING SOLAR CELL EFFICIENCY
D. P. Fenning1 , J. Hofstetter2 , M. I. Bertoni1 , J. F. Lelièvre3 , C. del Cañizo2 , T. Buonassisi1
1
Laboratory for Photovoltaic Research, Massachusetts Institute of Technology, Cambridge, MA
2
Instituto de Energı́a Solar, Universidad Politécnica de Madrid, Madrid, Spain
3
Centro Tecnológico de Silicio Solar CENTESIL, Madrid, Spain
ABSTRACT
Synchrotron-based X-ray fluorescence microscopy is applied to study the evolution of iron silicide precipitates during phosphorus diffusion gettering and low-temperature
annealing. Heavily Fe-contaminated ingot border material
contains FeSi2 precipitates after rapid in-line P-diffusion
firing, suggesting kinetically limited gettering in these regions. An impurity-to-efficiency (I2E) gettering model is
developed to explain the results. The model demonstrates
the efficacy of high- and medium-temperature processing
on reducing the interstitial iron population over a range of
process parameters available to industry.
INTRODUCTION
As an impurity in silicon, iron is known to have a strong
negative influence on minority carrier lifetime, and hence
solar cell efficiency. However, the relationship between
as-grown iron content and final cell efficiency is not
straightforward - the ultimate impact of iron contamination
depends on its chemical state and spatial distribution,
which in turn are a function of processing conditions and
cell architecture.
To accurately predict how thermal history will affect solar
cell efficiency in both traditional and non-traditional processing, we have developed a simulation tool to predict
final solar cell efficiency using measurable material and
process parameters as inputs [1]. These inputs include
concentration and distribution of iron in as-grown wafers,
solar cell processing conditions, and cell architecture.
We test our model by examining low-temperature annealing (LTA) after P-diffusion as an example of a materialsdriven process step that might be added to traditional
silicon solar cell manufacturing to enhance performance.
Using synchrotron based micro X-ray fluorescence (µXRF), we determine the iron distribution as a function of
annealing condition and compare it to the model results.
EXPERIMENT AND RESULTS
Synchrotron-based µ-XRF investigations were performed
on samples from higher locations of the same corner brick
from Rinio et al. [2] to determine metal nanoprecipitate
distributions along grain boundaries at the Advanced Photon Source Beamline 2-ID-D at Argonne National Laboratory, utilizing zone plate lenses to achieve an X-ray spot
size between 150 and 200 nm in diameter [3]. A hightemperature stage was used to perform an in-situ low978-1-4244-5892-9/10/$26.00 ©2010 IEEE
temperature anneal in order to allow for the direct observation of changing metal distribution. The high-temperature
sample stage and associated in-situ measurement are
described in further detail in [4]. Iron-rich precipitates
are detected in all samples, suggesting that the thermal
budget of the phosphorus diffusion was insufficient to
dissolve all metals. Further analysis and discussion of the
precipitate distribution will be published in an upcoming
journal article [5].
A second synchrotron-based experiment studying iron
distribution was conducted using samples that were characterized as part of the Coletti et al. study of the effect of
iron in mc-Si solar cells [6]. Sister wafers were removed
from the solar cell line in three states: after saw damage
etch, after phosphorus-diffusion, and after phosphorus
diffusion with a faux firing step conducted with no metal
contacts present. An article regarding the results of this
µ-XRF experiment concerning the distribution of iron in asgrown and phosphorus-diffused multi-crystalline silicon is
also forthcoming [7].
The experimental determination of iron distribution allows
for careful selection of boundary conditions for modeling
of the time-temperature transformation of iron in silicon
solar cell manufacturing. The I2E diffusion-gettering simulator we have developed solves coupled partial differential equations describing the diffusion of phosphorus
and iron within silicon, segregation of iron to the heavily
doped near-surface region, and growth of iron silicide
precipitates. The solution scheme of the I2E model and
verification of its predictions against experimental results
will be published elsewhere [1], [8], [9].
Using our diffusion-gettering simulator, we can study the
effect of alternative processing, like low-temperature annealing, without conducting large-scale experiments. We
simulated a P-diffusion plus low-temperature anneal for
highly contaminated silicon and observed the dissolved
Fe concentration when P-diffusion is followed by a variety
of LTAs. The simulation showed that the external gettering
efficiency strongly depends on the initial metal distribution,
not simply on the total metal content.
Using a Shockley-Read-Hall lifetime model, we calculate
resulting lifetime profiles as a function of the processed
iron distribution and introduce them into PC1D [10] to
calculate the estimated solar cell efficiencies.
000430
CONCLUSIONS AND OUTLOOK
µ-XRF experimental results of P-diffused and LTA samples
reveal the precipitated iron content of processed wafers,
confirming our I2E diffusion-gettering model results. The
integrated model allows us to investigate the influence
of different input parameters (e.g. average precipitate
size) on final solar cell performance. Furthermore, the
I2E-simulator enables us to tailor a solar cell fabrication
process for a given starting iron contamination level.
ACKNOWLEDGEMENTS
This work has been supported by the U. S. Department
of Energy, contract number DE-FG36-09GO1900. D. P.
Fenning acknowledges the support of the NSF Graduate
Research Fellowship. The support of the MIT-Spain/La
Cambra de Barcelona Seed Fund is also acknowledged.
REFERENCES
[1] J. Hofstetter, D. P. Fenning, M. I. Bertoni, J. F. Lelièvre,
T. Buonassisi, and C. del Cañizo, “Impurity-to-efficiency
simulator: Predictive simulation of silicon solar cell
performance based on iron content and distribution,”
submitted, 2010.
[2] M. Rinio, A. Yodyunyong, M. Pirker, S. Keipert, P. Wang, D.
Borchert, and T. Buonassisi, “Defect redistribution by low
temperature anneal in ingot silicon solar cells,” 23rd EU
PVSEC, p. 34, 2008.
[3] Z. Cai, B. Lai, W. Yun, I. McNulty, A. Khounsary, J. Maser,
P. Ilinski, D. Legnini, E. Trakhtenberg, S. Xu, B. Tieman,
G. Wiemerslage, and E. Gluskin, “Performance of a highresolution x-ray microprobe at the advanced photon source,”
Synchrotron Radiation Instrumentation: Eleventh US National
Conference, 2000.
[4] S. Hudelson, B. K. Newman, S. Bernardis, D. P. Fenning,
M. I. Bertoni, T. Buonassisi, M. A. Marcus, S. C. Fakra, and
B. Lai, “Retrograde melting and internal liquid gettering in
silicon,” Advanced Materials, 2010, in press.
[5] D. P. Fenning, J. Hofstetter, M. I. Bertoni, J. F. Lelièvre, S.
Hudelson, B. Lai, M. Rinio, C. del Cañizo, and T. Buonassisi,
to be submitted, 2010.
[6] G. Coletti, R. Kvande, V. D. Mihailetchi, L. J. Geerligs,
L. Arnberg, and E. J. Ovrelid,“Effect of iron in silicon
feedstock on p- and n-type multicrystalline silicon solar cells,”
Journal of Applied Physics, 104, no. 10, p. 104913, Jan 2008.
[7] D. P. Fenning, J. Hofstetter, M. I. Bertoni, J. F. Lelièvre, G.
Coletti, C. del Cañizo, and T. Buonassisi, to be submitted
2010.
[8] J. Hofstetter, D. P. Fenning, M. I. Bertoni, J. F. Lelièvre,
T. Buonassisi, and C. del Cañizo, “Impurity-to-efficiency
simulator: Predictive simulation of solar cell efficiencies
based on measured metal distribution and cell processing
conditions,” 25th EU PVSEC, Valencia, Spain, 2010.
[9] J. Hofstetter and et al., “Enhanced impurity reduction during
short extended p-diffusion gettering for industrial solar cell
processing,” submitted to Physica Status Solidi.
[10] D. Clugston and P. Basore, “Pc1d version 5: 32-bit solar cell
modeling on personal computers,” 26th IEEE PVSC, 1997.
978-1-4244-5892-9/10/$26.00 ©2010 IEEE
000431