Dissertation:
Alternative materials for crystalline silicon solar cells - Risks and implications

No Thumbnail Available
Date
2010
Editors
Kwapil, Wolfram
relationships.isEditorOf
Contact
Journal ISSN
Electronic ISSN
ISBN
Bibliographical data
Publisher
Series
URI (citable link)
DOI (citable link)
ArXiv-ID
International patent number
Link to the license
Project
EU project number
Open Access publication
Collections
Restricted until
Title in another language
Research Projects
Organizational Units
Journal Issue
Publication type
Dissertation
Publication status
Abstract
This thesis considers the use of alternative silicon materials for photovoltaics – often termed “upgraded metallurgical grade” silicon – from different angles and evaluates the risks and implications for the wafer and solar cell properties at selected steps along the entire process chain.
The properties of the alternative, upgraded metallurgical grade silicon materials analyzed in the course of this thesis were governed by the simultaneous presence of boron and phosphorus in high concentrations (in the order of >2x1016 at/cm3) which has fundamen-tal consequences for the base resistivity, the carrier mobility and the light-induced degra-dation in UMG-Si wafers.

Aiming to compare the experimental data to existing models, the majority carrier con-ductivity and Hall mobility of UMG-Si and intentionally compensated wafers doped with various B- and P-concentrations was investigated. In compensated silicon, the mobility decreases significantly firstly with growing sum of both dopants (NA+ND), hence with in-creasing density of ionized scattering centers, and secondly with decreasing net doping concentration p0=(NA-ND), attributed to a weakening of the Coulombic screening of ion-ized atoms. Our measurements suggest that in the commonly accepted model for com-pensated silicon published by Klaassen, the modified Coulombic screening is not correctly implemented. Hence, Klaassen’s model can currently be used safely only in a limited range of NA- and ND-values when regarding compensated silicon (e.g. UMG-Si).
A thorough investigation of the light-induced degradation in compensated and UMG Cz-silicon wafers revealed that in general the normalized Cz-defect concentration is a func-tion of the compensation ratio RC, which surprisingly holds for p- as well as for n-type material. For the special case of UMG-Si wafers with p-type net doping concentrations usually used for solar cells, the Cz-defect density can alternatively be described by a lin-ear dependence on the net doping concentration, supporting results obtained by other groups. Therefore, for oxygen-rich UMG-Si the net doping concentration of the feedstock should be reduced in order to minimize the light-induced degradation. This can only be realized by simultaneously decreasing the B- and P-concentration.
Although the concentration of transition metals in UMG-Si is not as high as previously expected, it has been worthwhile to test the feasibility of the concept of “defect engineer-ing” via intrinsic gettering and co-precipitation (intentional addition of e.g. Cu to the melt). It was shown that an interaction between the total transition metal content in the melt and the crystallization process has a negative impact on the silicon crystal quality, the harmfulness depending on the impurity elements. Moreover, in carefully prepared NAA measurements the total metal concentration was observed to depend (super-) line-arly on the dislocation density. Both observations point towards a non-linear crystal de-fect generation process being started if the metal content in the melt exceeds a critical level, prohibiting the exploitation of co-precipitation.
The optimal design of temperature ramps aimed at intrinsic gettering was demonstrated to depend on the metal species: While intermediate diffusers (Fe, Cr,…) re-precipitate at 600 °C after a high-temperature step (>800 C), the same temperature ramp leads to the dissolution of the large clusters of fast diffusers (Ni,…) being present after mc-Si so-lidification. Subsequently, the fast diffusers spread along grain boundaries and disloca-tions, thus presenting a larger recombination active effective surface than before. Our experiments indicate that for intrinsic gettering of fast diffusers like Ni, lower tempera-tures (around 500°C-550°C) and longer gettering times (12 h) are necessary.
However, in NAA measurements the fast diffusers Ni and Cu, and to some extent also Co, were seen to be easily gettered during P-emitter diffusion even at high initial concentra-tions from highly dislocated wafer regions. At the same time, P-diffusion gettering of the intermediate diffusers Fe or Cr from high dislocation density areas is not effective, allow-ing for a significant reduction only in less defected crystal regions.
These results implicate that the interplay of extrinsic and intrinsic gettering can be aimed at and optimized for the specific group of intermediate diffusers, since fast diffusers are externally gettered while slow diffusers cannot be mobilized in the interesting tempera-ture and time scale.
Reports on inferior reverse characteristics of UMG mc-Si solar cells, possibly posing a danger to the solar modules by generating “hot spots”, incited detailed investigations of the diode breakdown behavior of mc-Si solar cells in general. In this thesis, it was shown that in these devices breakdown happens in three stages termed “early”, “soft” and “hard” breakdown which can be clearly distinguished by their breakdown voltage and reverse I-V characteristics resulting from (at least) three different physical reasons.
All three breakdown types have in common that white light is emitted from the break-down sites, pointing at the occurrence of “hot” electrons. It can be concluded that elec-tron multiplication due to avalanching is always involved in a way, possibly initiated by tunneling processes at the beginning.
“Early breakdown” sets on at around -4 to -5 V and is presumably induced by surface defects including paste particles from the front as well as from the rear side metallization and small pits in the surface structure. Early breakdown can be explained by a modifica-tion of the emitter leading to a local increase of the electric field in the space charge re-gion. Since early breakdown is highly localized and starts at very low reverse voltages, the local heat generation can become large, possibly being harmful to solar modules. However, in most practical cases the heat remains relatively low.
The “soft breakdown” happens exclusively at recombination active crystal defects. XRF-measurements at soft breakdown sites provided strong evidence that it is caused by metal precipitates close to the solar cell surface, i.e. close to or in the space charge re-gion. The onset voltage is influenced by several factors: Both a high impurity concentra-tion and a high base net doping concentration decrease the soft breakdown voltage, as does deep etching of dislocations and grain boundaries during the wet chemical texturiza-tion. Depending on the solar cell properties, the onset voltage of soft breakdown can therefore vary between -8 to -14 V. Due to the wide lateral distribution of recombination active defects and the soft reverse I-V characteristics, it is expected that soft breakdown does not pose any danger to solar modules.
To explain the physical mechanism of soft diode breakdown found at the sites of metallic precipitates, a numerical simulation describing the internal Schottky junction between silicide clusters with the surrounding silicon was set up. Under reverse bias in thermody-namic equilibrium, the electric fields around the metal clusters can reach very large val-ues (>5x105 V/cm) which are generally assumed to lead to avalanche electron multiplica-tion.
The “hard breakdown” sets on around -14 to -16 V, happening at deeply etched disloca-tions at which no recombination active impurities have accumulated. Due to fast increase of the reverse current with increasing reverse voltage, this breakdown type can be detri-mental if the necessary high reverse voltage is reached.
Having clarified the physical mechanisms of diode breakdown in mc-Si solar cells in gen-eral, the relatively high net doping concentration in the base of UMG-Si solar cells was identified to be the main reason for the large reverse currents at low breakdown volt-ages. Since thus, the UMG-Si reverse bias behavior can be attributed to the soft break-down type, we expect that the risk of including UMG-Si solar cells in solar modules is not larger compared to modules consisting of conventional mc-Si cells.
In summary, important properties of UMG-Si wafers (low conductivity mobility and mi-nority carrier lifetime, high Cz-defect concentration as well as relatively high reverse cur-rents) are related to the high concentration of boron and phosphorus. Hence, by further reducing their amount in the silicon feedstock, which is currently being done already quite successfully by several feedstock producers, upgraded metallurgical grade silicon is expected to constitute an interesting, cost-effective alternative to conventional material at present and in the near future. However, as optimizations of the solar cell processes have been yielding significantly increasing efficiencies, in the long term, the fate of UMG-Si will depend firstly on its ability to keep up with future process developments, and sec-ondly on the future trend of silicon feedstock prices.
Summary in another language
Die vorliegende Arbeit betrachtet aus verschiedenen Blickwinkeln die Verwendung alter-nativer Siliciummaterialien für die Photovoltaik – häufig „upgraded metallurgical grade“ (UMG)-Silicium genannt –und bewertet seine Auswirkungen auf die Wafer- und Solarzell-eigenschaften an ausgewählten Punkten entlang der Prozesskette bis hin zum Modul.
Die Eigenschaften der hier untersuchten Siliciummaterialien werden hauptsächlich durch die hohe Konzentration sowohl an Bor als auch an Phosphor (beide etwa >2x1016 at/cm3) bestimmt. Dies hat fundamentale Auswirkungen u.a. auf die Basiswiderstandsverteilung durch elektrische Kompensation, die Ladungsträgermobilität und die lichtinduzierte De-gradation.
Wegen der Verwendung von Mobilitätsmodellen in vielen Analysetechniken und Simula-tionen ist es von großer Bedeutung, die Ladungsträgermobilität auch in kompensierten Materialien modellieren zu können. Daher wurden in dieser Arbeit umfangreiche Messun-gen der Majoritätenleitfähigkeitsbeweglichkeit und Hall-Beweglichkeit in Siliciumwafern mit unterschiedlichen B- und P-Konzentrationen durchgeführt: In kompensiertem Si ver-ringert sich die Ladungsträgermobilität deutlich sowohl mit zunehmender Summe beider Dotierstoffe (NA+ND), also mit wachsender Streuzentrendichte, als auch mit abnehmen-der Nettodotierkonzentration p0=(NA-ND), was auf eine abnehmende Coulomb-Abschirmung ionisierter Streuzentren zurückgeführt wird. Unsere Messungen weisen dar-auf hin, dass das für kompensiertes Si am ehesten geeignete Mobilitätsmodell von Klaas-sen die Coulomb-Abschirmung nicht korrekt beschreibt. Im Moment kann es daher die Ladungsträgermobilität nur in einem eingeschränkten Bereich von NA- und ND-Werten treffend vorhersagen.

Eine sorgfältige Analyse der lichtinduzierten Degradation in kompensiertem und UMG Czochralski-Silicium ergab, dass die normalisierte Cz-Defektkonzentration nicht, wie bis-her angenommen, in allen Dotierbereichen linear von der Nettodotierkonzentration ab-hängt, sondern vielmehr eine Funktion der Kompensation RC ist. Dies gilt sowohl für p- als auch für n-Typ-Silicium, welches in kompensiertem Si ebenfalls unter Lichteinfluss degradiert. Für die in der Praxis relevanten Nettodotierkonzentrationen reicht das lineare Modell jedoch aus. Um die lichtinduzierte Degradation in sauerstoffreichem UMG-Si zu minimieren, sollten die B- und P-Konzentrationen gleichermaßen verringert werden.
Obwohl die Konzentration der Übergangsmetalle in UMG-Si geringer ist als ursprünglich angenommen, wurden die Konzepte des „Defect Engineering“, die eine Materialverbesse-rung durch intrinsisches Gettern zum Ziel haben (theoretisch verbessert durch Ko-Präzipitation mit beweglichen Verunreinigungen wie z.B. Cu), unter die Lupe genommen. Es wurde gezeigt, dass die Kristallqualität unter einer Wechselwirkung zwischen metalli-schen Verunreinigungen und der Kristallisation leidet; die Stärke der Wechselwirkung hängt von der Zusammensetzung und Konzentration der Verunreinigungselemente ab. Dies wird auf einen erhöhten Einbau von Metallen in Kristalldefekt-reichen Gebieten zu-rückgeführt: In sorgfältigen Messungen der Verunreinigungskonzentration in Abhängig-keit der Versetzungsdichte konnte ein (super-)linearer Zusammenhang nachgewiesen werden, weshalb die absichtliche Zugabe schnell diffundierender Metalle für das „Defect Engineering“ über Ko-Präzipitation keinen Vorteil bringen dürfte.
Die optimale Ausgestaltung von Hochtemperaturschritten für das intrinsische Gettern von Metallen hängt von der dominierenden Verunreinigung ab. Während mittelschnell diffun-dierende Metalle (Fe, Cr,…) re-präzipitieren, wenn nach einem Hochtemperaturprozess ein Temperschritt bei 600°C durchgeführt wird, verursacht die gleiche Temperaturführung eine Auflösung der großen Präzipitate von schnell diffundieren, hoch löslichen Metallen wie Ni. Wie Synchrotronmessungen belegen, verteilen sich diese Atome anschließend entlang der Korngrenzen und Versetzungen und bieten daher eine größere, rekombinati-onsaktive Grenzfläche als zuvor. Um für Ni ein intrinsisches Gettern zu bewirken, werden niedrigere Temperaturen (ca. 500°C-550°C) über eine längere Zeit (12 h) benötigt.
In vergleichenden Messungen der Verunreinigungskonzentration in hoch-versetzten Ge-bieten von Nachbarwafern, wovon die Hälfte einer Phosphordiffusion (externes Gettern) unterzogen wurden, konnte jedoch gezeigt werden, dass die schnell diffundierenden Ele-mente Ni und Cu, teilweise auch Co, trotz hoher Versetzungsdichte äußerst effektiv ge-gettert werden. Die mittelschnell diffundierenden Elemente Fe und Cr hingegen werden hier so gut wie nicht entfernt, weshalb sich für diese Spezies das Gettern auf Gebiete geringerer Kristalldefektdichte beschränkt.
Zusammengenommen bedeuten diese Ergebnisse, dass das Zusammenspiel zwischen externem und intrinsischem Gettern für die spezielle Elementgruppe der mittelschnell diffundierenden Metalle optimiert werden kann, da leicht bewegliche Elemente gegettert werden, während sich die langsam diffundierenden Metalle im relevanten Temperatur- und Zeitbereich kaum bewegen.
Berichte über schlechtere Sperreigenschaften der Solarzellen aus UMG-Si, welche mögli-cherweise zur Entstehung sogenannter „Hot Spots“ im Modul beitragen können, haben eine Reihe detaillierter Untersuchungen der Rückwärtseigenschaften von multikristallinen Si-Solarzellen im Allgemeinen ausgelöst. Im Rahmen dieser Arbeit wurde erstmals ge-zeigt, dass der Diodendurchbruch in mc-Si in drei Klassen („früh“, „weich“ und „hart“) unterteilt werden kann. Sie lassen sich anhand ihrer Durchbruchspannung und ihrer Kennlinie in Sperrrichtung unterscheiden und können durch drei verschiedene physikali-sche Vorgänge beschrieben werden.
Alle drei Klassen haben gemeinsam, dass der Durchbruch in zahlreichen, mikrometer-großen Bereichen stattfindet, welche daraufhin weißes Licht aussenden. Dies deutet auf die Anwesenheit „heißer“ Elektronen und damit auf Elektronenmultiplikation durch einen Avalanche-Vorgang hin.
Der „frühe” Durchbruch beginnt bei -4 bis -5 V und wird wahrscheinlich durch eine be-schädigte Solarzellenoberfläche verursacht, wozu eingefeuerte Pastenpartikel der Vorder- und Rückseitenmetallisierung gehören als auch kleine Risse und Löcher. Diese Beschädi-gungen verändern die lokalen Eigenschaften des Emitters; infolgedessen werden vermut-lich hohe elektrische Felder in der Raumladungszone induziert, die zu einem Durchbruch führen. Da der frühe Durchbruch stets auf eine sehr kleine Fläche beschränkt ist und be-reits bei einer relativ niedrigen Sperrspannung einsetzt, kann es zu einer großen Hitze-entwicklung kommen. In den meisten beobachteten Fällen blieb die Erwärmung jedoch unkritisch.
„Weicher“ Durchbruch findet ausschließlich an rekombinationsaktiven Kristalldefekten statt. Wie Synchrotronuntersuchungen zeigen, wird er mit hoher Wahrscheinlichkeit durch Metallpräzipitate in der Raumladungszone verursacht. Dabei beeinflussen mehrere Faktoren die Durchbruchspannung: In Wafern mit einer hohe Metallverunreinigungskon-zentration oder einer hohen Dotierkonzentration (d.h. mit einem niedrigen Basiswider-stand) setzt der weiche Durchbruch bei niedrigeren Spannungen ein. Ebenfalls reduziert wird die Durchbruchspannung durch eine tiefe Anätzung der Waferoberfläche bevorzugt an Korngrenzen und Versetzungen, z.B. durch die Texturierung. Daher variiert der Beginn des weichen Durchbruchs je nach Solarzelleigenschaften zwischen -8 und -14 V. Weil re-kombinationsaktive Defekte im Allgemeinen über ein großes Gebiet verteilt sind und weil der weiche Durchbruch eine weiche Rückwärtskennlinie besitzt, wird davon ausgegangen, dass die Wärmeentwicklung stets ungefährlich für ein Modul bleibt.
Um die Entstehung des weichen Durchbruchs erklären zu können, wurde ein Metallpräzi-pitat, welches sich in der Raumladungszone befindet, mittels eines internen Schottky-Übergangs beschrieben und die Verteilung des elektrischen Feldes numerisch berechnet. Es zeigte sich, dass das Feld schon bei geringen Sperrspannungen große Werte erreichen kann, welche gewöhnlich mit der Entstehung von Lawinenmultiplikation in Verbindung gebracht werden.
Der “harte” Durchbruch setzt erst bei etwa -14 bis -16 V an tief geätzten Versetzungen ein, an welche sich keine Verunreinigungen angelagert haben. Wegen der starken Zu-nahme des Sperrstroms innerhalb eines relativ kleinen Spannungsintervalls und weil er sich häufig auf ein kleines Gebiet konzentriert, kann der harte Durchbruch für das Modul gefährlich werden, wenn die erforderlichen Sperrspannungen erreicht werden.
Indem die allgemeinen Mechanismen des Diodendurchbruchs in multikristallinen Si-Solarzellen weitestgehend aufgeklärt werden konnten, konnte auch die Ursache der rela-tiv hohen Sperrströme in UMG-Si Solarzellen ermittelt werden: die verhältnismäßig hohe Nettodotierkonzentration in der Waferbasis. Da es sich hier um den weichen Durchbruch-typ handelt, kann man davon ausgehen, dass durch die Verwendung von UMG-Si Solar-zellen im Modul nicht das Risiko von Hot Spot-Entstehung steigt.
Wie in der vorliegenden Arbeit gezeigt wird, werden wichtige Eigenschaften der UMG-Si Wafer vorwiegend durch hohe Bor- und Phosphorkonzentrationen bestimmt. Indem der Gehalt an beiden Dotierstoffen im Feedstock reduziert wird, was bereits gegenwärtig recht erfolgreich bei einigen Herstellern erfolgt, bietet UMG-Silicium in naher Zukunft eine interessante, kosteneffektive Alternative zu herkömmlichen Materialien. Ob dies je-doch so bleiben wird, hängt entscheidend davon ab, ob UMG-Si Solarzellen mit zukünfti-gen Wirkungsgradverbesserungen mithalten können.
Subject (DDC)
530 Physics
Keywords
solar cells , silicon , upgraded metallurgical grade , dopant compensation , metallic impurities
Published in
Conference
Review
undefined / . - undefined, undefined. - (undefined; undefined)
Cite This
ISO 690KWAPIL, Wolfram, 2010. Alternative materials for crystalline silicon solar cells - Risks and implications [Dissertation]. Konstanz: University of Konstanz
BibTex
@phdthesis{Kwapil2010Alter-12892,
  year={2010},
  title={Alternative materials for crystalline silicon solar cells - Risks and implications},
  author={Kwapil, Wolfram},
  address={Konstanz},
  school={Universität Konstanz}
}
RDF
<rdf:RDF
    xmlns:dcterms="http://purl.org/dc/terms/"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    xmlns:bibo="http://purl.org/ontology/bibo/"
    xmlns:dspace="http://digital-repositories.org/ontologies/dspace/0.1.0#"
    xmlns:foaf="http://xmlns.com/foaf/0.1/"
    xmlns:void="http://rdfs.org/ns/void#"
    xmlns:xsd="http://www.w3.org/2001/XMLSchema#" > 
  <rdf:Description rdf:about="https://kops.uni-konstanz.de/server/rdf/resource/123456789/12892">
    <dcterms:rights rdf:resource="https://rightsstatements.org/page/InC/1.0/"/>
    <dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2011-05-05T06:46:17Z</dcterms:available>
    <dspace:hasBitstream rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/12892/1/Kwapil_PhDthesis_Alternative_materials.pdf"/>
    <dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/41"/>
    <dc:creator>Kwapil, Wolfram</dc:creator>
    <dcterms:hasPart rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/12892/1/Kwapil_PhDthesis_Alternative_materials.pdf"/>
    <dc:language>eng</dc:language>
    <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2011-05-05T06:46:17Z</dc:date>
    <foaf:homepage rdf:resource="http://localhost:8080/"/>
    <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
    <dc:rights>terms-of-use</dc:rights>
    <dc:contributor>Kwapil, Wolfram</dc:contributor>
    <dcterms:issued>2010</dcterms:issued>
    <dcterms:abstract xml:lang="eng">This thesis considers the use of alternative silicon materials for photovoltaics – often termed “upgraded metallurgical grade” silicon – from different angles and evaluates the risks and implications for the wafer and solar cell properties at selected steps along the entire process chain.&lt;br /&gt;The properties of the alternative, upgraded metallurgical grade silicon materials analyzed in the course of this thesis were governed by the simultaneous presence of boron and phosphorus in high concentrations (in the order of &gt;2x1016 at/cm3) which has fundamen-tal consequences for the base resistivity, the carrier mobility and the light-induced degra-dation in UMG-Si wafers.&lt;br /&gt;&lt;br /&gt;Aiming to compare the experimental data to existing models, the majority carrier con-ductivity and Hall mobility of UMG-Si and intentionally compensated wafers doped with various B- and P-concentrations was investigated. In compensated silicon, the mobility decreases significantly firstly with growing sum of both dopants (NA+ND), hence with in-creasing density of ionized scattering centers, and secondly with decreasing net doping concentration p0=(NA-ND), attributed to a weakening of the Coulombic screening of ion-ized atoms. Our measurements suggest that in the commonly accepted model for com-pensated silicon published by Klaassen, the modified Coulombic screening is not correctly implemented. Hence, Klaassen’s model can currently be used safely only in a limited range of NA- and ND-values when regarding compensated silicon (e.g. UMG-Si).&lt;br /&gt;A thorough investigation of the light-induced degradation in compensated and UMG Cz-silicon wafers revealed that in general the normalized Cz-defect concentration is a func-tion of the compensation ratio RC, which surprisingly holds for p- as well as for n-type material. For the special case of UMG-Si wafers with p-type net doping concentrations usually used for solar cells, the Cz-defect density can alternatively be described by a lin-ear dependence on the net doping concentration, supporting results obtained by other groups. Therefore, for oxygen-rich UMG-Si the net doping concentration of the feedstock should be reduced in order to minimize the light-induced degradation. This can only be realized by simultaneously decreasing the B- and P-concentration.&lt;br /&gt;Although the concentration of transition metals in UMG-Si is not as high as previously expected, it has been worthwhile to test the feasibility of the concept of “defect engineer-ing” via intrinsic gettering and co-precipitation (intentional addition of e.g. Cu to the melt). It was shown that an interaction between the total transition metal content in the melt and the crystallization process has a negative impact on the silicon crystal quality, the harmfulness depending on the impurity elements. Moreover, in carefully prepared NAA measurements the total metal concentration was observed to depend (super-) line-arly on the dislocation density. Both observations point towards a non-linear crystal de-fect generation process being started if the metal content in the melt exceeds a critical level, prohibiting the exploitation of co-precipitation.&lt;br /&gt;The optimal design of temperature ramps aimed at intrinsic gettering was demonstrated to depend on the metal species: While intermediate diffusers (Fe, Cr,…) re-precipitate at 600 °C after a high-temperature step (&gt;800  C), the same temperature ramp leads to the dissolution of the large clusters of fast diffusers (Ni,…) being present after mc-Si so-lidification. Subsequently, the fast diffusers spread along grain boundaries and disloca-tions, thus presenting a larger recombination active effective surface than before. Our experiments indicate that for intrinsic gettering of fast diffusers like Ni, lower tempera-tures (around 500°C-550°C) and longer gettering times (12 h) are necessary.&lt;br /&gt;However, in NAA measurements the fast diffusers Ni and Cu, and to some extent also Co, were seen to be easily gettered during P-emitter diffusion even at high initial concentra-tions from highly dislocated wafer regions. At the same time, P-diffusion gettering of the intermediate diffusers Fe or Cr from high dislocation density areas is not effective, allow-ing for a significant reduction only in less defected crystal regions.&lt;br /&gt;These results implicate that the interplay of extrinsic and intrinsic gettering can be aimed at and optimized for the specific group of intermediate diffusers, since fast diffusers are externally gettered while slow diffusers cannot be mobilized in the interesting tempera-ture and time scale.&lt;br /&gt;Reports on inferior reverse characteristics of UMG mc-Si solar cells, possibly posing a danger to the solar modules by generating “hot spots”, incited detailed investigations of the diode breakdown behavior of mc-Si solar cells in general. In this thesis, it was shown that in these devices breakdown happens in three stages termed “early”, “soft” and “hard” breakdown which can be clearly distinguished by their breakdown voltage and reverse I-V characteristics resulting from (at least) three different physical reasons.&lt;br /&gt;All three breakdown types have in common that white light is emitted from the break-down sites, pointing at the occurrence of “hot” electrons. It can be concluded that elec-tron multiplication due to avalanching is always involved in a way, possibly initiated by tunneling processes at the beginning.&lt;br /&gt;“Early breakdown” sets on at around -4 to -5 V and is presumably induced by surface defects including paste particles from the front as well as from the rear side metallization and small pits in the surface structure. Early breakdown can be explained by a modifica-tion of the emitter leading to a local increase of the electric field in the space charge re-gion. Since early breakdown is highly localized and starts at very low reverse voltages, the local heat generation can become large, possibly being harmful to solar modules. However, in most practical cases the heat remains relatively low.&lt;br /&gt;The “soft breakdown” happens exclusively at recombination active crystal defects. XRF-measurements at soft breakdown sites provided strong evidence that it is caused by metal precipitates close to the solar cell surface, i.e. close to or in the space charge re-gion. The onset voltage is influenced by several factors: Both a high impurity concentra-tion and a high base net doping concentration decrease the soft breakdown voltage, as does deep etching of dislocations and grain boundaries during the wet chemical texturiza-tion. Depending on the solar cell properties, the onset voltage of soft breakdown can therefore vary between -8 to -14 V. Due to the wide lateral distribution of recombination active defects and the soft reverse I-V characteristics, it is expected that soft breakdown does not pose any danger to solar modules.&lt;br /&gt;To explain the physical mechanism of soft diode breakdown found at the sites of metallic precipitates, a numerical simulation describing the internal Schottky junction between silicide clusters with the surrounding silicon was set up. Under reverse bias in thermody-namic equilibrium, the electric fields around the metal clusters can reach very large val-ues (&gt;5x105 V/cm) which are generally assumed to lead to avalanche electron multiplica-tion.&lt;br /&gt;The “hard breakdown” sets on around -14 to -16 V, happening at deeply etched disloca-tions at which no recombination active impurities have accumulated. Due to fast increase of the reverse current with increasing reverse voltage, this breakdown type can be detri-mental if the necessary high reverse voltage is reached.&lt;br /&gt;Having clarified the physical mechanisms of diode breakdown in mc-Si solar cells in gen-eral, the relatively high net doping concentration in the base of UMG-Si solar cells was identified to be the main reason for the large reverse currents at low breakdown volt-ages. Since thus, the UMG-Si reverse bias behavior can be attributed to the soft break-down type, we expect that the risk of including UMG-Si solar cells in solar modules is not larger compared to modules consisting of conventional mc-Si cells.&lt;br /&gt;In summary, important properties of UMG-Si wafers (low conductivity mobility and mi-nority carrier lifetime, high Cz-defect concentration as well as relatively high reverse cur-rents) are related to the high concentration of boron and phosphorus. Hence, by further reducing their amount in the silicon feedstock, which is currently being done already quite successfully by several feedstock producers, upgraded metallurgical grade silicon is expected to constitute an interesting, cost-effective alternative to conventional material at present and in the near future. However, as optimizations of the solar cell processes have been yielding significantly increasing efficiencies, in the long term, the fate of UMG-Si will depend firstly on its ability to keep up with future process developments, and sec-ondly on the future trend of silicon feedstock prices.</dcterms:abstract>
    <dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/41"/>
    <dcterms:title>Alternative materials for crystalline silicon solar cells - Risks and implications</dcterms:title>
    <bibo:uri rdf:resource="http://kops.uni-konstanz.de/handle/123456789/12892"/>
  </rdf:Description>
</rdf:RDF>
Internal note
xmlui.Submission.submit.DescribeStep.inputForms.label.kops_note_fromSubmitter
Contact
URL of original publication
Test date of URL
Examination date of dissertation
December 16, 2010
Method of financing
Comment on publication
Alliance license
Corresponding Authors der Uni Konstanz vorhanden
International Co-Authors
Bibliography of Konstanz
Refereed
Link to research data
Description of supplementary data