Impact of iron contamination in multicrystalline silicon solarcells: origins, chemical states, and device impacts Page: 3 of 4
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to UNT Digital Library by the UNT Libraries Government Documents Department.
Extracted Text
The following text was automatically extracted from the image on this page using optical character recognition software:
nm across and 900-1200 nm in the vertical direction, i.e.
not elongated along the crystal growth direction.
4.2 Discussion
The origins of these particles appear to be similar to
the case of sheet material, with a few key differences:
Firstly, the large Type 2 clusters are oxidized. It is rather
unusual and unexpected to find oxidized Fe particles in
silicon, since they are not thermodynamically favored to
form. Silicon, which has a higher binding energy to
oxygen than iron, will be a strongly reducing atmosphere
for any oxidized iron. It is thus more likely, given the
oxidized chemical state and presence with other metals
reminiscent of stainless steels or ceramics, that the Type
2 particles are the remnants of foreign particles included
in the melt, from the feedstock or possibly from the
crucible walls, as was suggested by Rinio et al. [13].
Because the cast me-Si process maintains a liquid melt
for many hours, these oxidized particles should survive
only a rather short time before complete dissolution in
the melt, i.e. one would expect to find them only
towards the bottom of the ingot.
However, as these foreign particles are dissolving in
the melt, they most certainly can increase the Fe content
of the melt considerably. To reduce the Fe content in the
melt and avoid supersaturation, atomically dissolved Fe
is likely to precipitate in the form of FeSi2.
Iron silicide clusters such as the Type 1 particles (as
well as those below the current detection limit of the p-
XRF technique) are likely to make the greatest
contribution to the recombination activity of grain
boundaries in this material, due to their higher spatial
density following from arguments presented in Sec. 3.2.
As can be seen from the p-XRF map in Fig. 3b, the
distances between Type 1 particles can be as small as 20
pm, which is predicted to severely limit the minority
carrier diffusion length [10]. As shown by ILT and
described in Sec. 3.2, recombination at grain boundaries
may make a large contribution to the total power loss of
an illuminated me-Si solar cell. It is thus of interest to
identify means of reducing the impact of these defects on
minority carrier diffusion length. Three possibilities are
identified and discussed forthwith: (a) gettering existing
Type 1 defects, (b) passivating Type 1 defects, and (c)
limiting the dissolution of Type 2 particles in the melt.
(a) Gettering has shown mixed results when applied
to mc-Si materials [11]. This is believed to be due to the
fact that large Fe-rich particles do not fully dissolve
during normal gettering sequences, because of the
relatively low diffusivity and solubility of Fe in Si [7].
Thus, when the gettering sequence ends, there are still
dissolving Fe-rich particles, and thus there will still be
Fe dissolved within the silicon bulk at its solubility
limit. A slow cool is thus advised at the end of the
gettering step, to allow for the dissolved Fe to diffuse to
the most energetically favorable sites, thus giving
Ostwald ripening a chance to occur. If cooling from the
high gettering temperature occurs too quickly, then
supersaturated Fe dissolved in the bulk will be forced to
precipitate at the nearest available precipitation site or
to remain dissolved within the grains as point defects,
resulting in a greatly increased recombination activity.
(b) Passivation, presumably with hydrogen, may be a
good alternative for reducing the recombination activityof FeSi2 clusters. There is already evidence that iron
clusters can be rendered recombination-inactive as the
result of hydrogen passivation [12], although many
properties of this reaction are poorly understood and
warrant additional investigation.
(c) By restricting the dissolution of foreign particles
within the melt, it is perhaps possible to limit the
dissolved Fe concentration within the melt, and thus the
abundance of resulting Type 1 particles. This may be
achieved by reducing the time the feedstock remains in
liquid form, while ensuring that large second-phase
particles become included into the crystal rather than
segregated from it. Unfortunately, for block cast me-Si,
the feedstock remains in liquid form in contact with the
crucible for a considerable length of time, as long as
several hours, and the growth speed is rather slow,
several microns/s. Novel methods, such as
electromagnetic casting, reduce considerably the time
the feedstock is in liquid form, by increasing the pull
rate and reducing the size of the molten zone. In
addition, the increased pull rate allows for larger
second-phase particles to be incorporated wholly into the
forming crystal. It is thus not surprising that in me-Si
grown by this technique, large (12 - 30+ pm in
diameter) oxidized Fe particle clusters were identified
via p-XRF/XAS by McHugo et al. [5]. Also, LBIC
measurements by P6richaud et al. [8] have indicated that
aside from isolated regions of recombination activity at
metallic inclusions, the recombination activity of most
structural defects is rather low at room temperature, in
contrast to recombination-active structural defects in
block cast me-Si. The drawback of EMCP is that faster
pull rates produce lower effective segregation
coefficients for Fe, resulting in a greater point defect
concentration. Thus, a gettering or passivaiton step is
required, with temperatures low enough as not to
significantly dissolve the metallic inclusions.
5 CONCLUSIONS
Synchrotron-based microprobe techniques have been
applied to investigate the distribution, size, chemical
state, and recombination activity of Fe-rich clusters in
two me-Si materials: sheet material and cast me-Si. In
the former, a high concentration (-2x107 cm2) of rather
large (23 5 nm radius) iron silicide particles were
observed along the grain boundary. Metals were also
found concentrated at intragranular clusters, which in as-
grown material have a composition similar to stainless
steel. It is believed that these clusters, likely introduced
into the melt from the feedstock, rapidly dissolve during
crystal growth and during high-temperature device
processing, leading to the contamination of nearby
structural defects such as grain boundaries, and to the
formation of smaller Fe silicide particles. Materials
improvements can be expected by increasing grain size
near the pn junction, reducing the density of
intragranular metallic clusters near the pn junction, and
by retarding the dissolution of these foreign particles in
the feestock, perhaps by oxidizing them. In cast me-Si,
two types of clusters were observed: Type 1, consisting
of sub-micron FeSi2 particles elongated along the crystal
growth direction, and Type 2, consisting of micron-sized
Upcoming Pages
Here’s what’s next.
Search Inside
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Buonassisi, Tonio; Heuer, Matthias; Istratov, Andrei A.; Marcus,Matthew A.; Jonczyk, Ralf; Lai, Barry et al. Impact of iron contamination in multicrystalline silicon solarcells: origins, chemical states, and device impacts, article, November 8, 2004; (https://digital.library.unt.edu/ark:/67531/metadc794796/m1/3/: accessed April 23, 2024), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.