- Email: [email protected]
Challenges for photovoltaic silicon materials
Solar Energy Materials & Solar Cells 130 (2014) 629–633
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Editorial
Challenges for photovoltaic silicon materials$
This special issue is dedicated to the 2nd workshop on Silicon Materials. These Si Materials workshops address a specific issue in the field of crystalline silicon for photovoltaic applications. The first workshop, organized in 2008, focused on specifications for solar-grade feedstock as a response to the polysilicon shortage [1]. A definition of Solar Grade (SoG) Silicon categories based on the usage for solar cell manufacturing was proposed in 2008 and a tentative and approximate standard set of specifications aimed to the fabrication of solar cells was provided. The workshop also gave input to start an initiative [2] for the industrial standardization of silicon feedstock specifications with the institution by SEMI of the silicon feedstock task force. In the last 5 years the global situation changed considerably from several perspectives [3,4]. The feedstock shortage regime for the PV industry ended. There was a delay of a few years before extra polysilicon capacity could enter into the market because of the large capital expenditure and the time needed for the construction and ramping up of poly-silicon production plants. Despite the credit crunch hitting the financial sector, PV flourished with record installations in 2010 and 2011. Eventually, the economic stagnation and the abrupt reduction of subsidies affected the PV market causing an oversupply of PV modules. As a consequence, profit margins strongly reduced and the market stabilized in 2012. All these factors initiated an even stronger drive towards cost reduction pushing the PV industry in a consolidation period. Currently (mid-June 2014), we observe a mild readjustment of the module price towards a more healthy economy of profit and a market that is growing again significantly. Initially triggered by the shortage of feedstock and by the need of a clear understanding of the polysilicon limitation factors, research on silicon materials and on new silicon purification methods multiplied in the past years. However, due to the rapid change in the economic and market situations, an alignment between the short-term industrial needs and the long-term research performed by research institutes is not always straightforward. Despite the proliferation of conferences related to photovoltaics, it was felt that a dedicated event in the form of an actual workshop with ample discussions on “Silicon material” research was desirable. Therefore, the aim of this 2nd Silicon Materials workshop was to provide a worldwide platform where industry and research groups would discuss their programs and future targets with the ambition to find new synergies for the development of PV solar energy. For these reasons, the second edition of the workshop was dedicated to the definition of future challenges
☆ Special issue for the 2nd Silicon Materials workshop, Rome, Italy, October 7th– 8th 2013.
http://dx.doi.org/10.1016/j.solmat.2014.07.045 0927-0248/& 2014 Elsevier B.V. All rights reserved.
for Silicon material research and to discuss industrial needs versus research efforts. 1. Online survey In preparation of the workshop discussions and to sketch a roadmap for Si material research, a survey was carried out among a large number of international experts. The result of the survey was presented during the workshop. This allowed us to identify the most crucial topics that were then discussed in detail during the workshop. The survey focused on the properties and challenges of Si materials for photovoltaics. The input for this survey came from 44 experts working in the industry, research institutes and universities worldwide (Fig. 1) [5]. The respondents do not form a representative sample of the entire scientific community but they did include a significant share of the experts in Si materials for photovoltaics. This is complementary to other initiatives like the ITRPViii and it is expected to give a more visionary interpretation of enabling technologies and challenges for Si materials as seen from the expert point of view. 1.1. Methodology The survey contained open questions and multiple choice questions. This was chosen in order to obtain an easy aggregation of the input data for the discussion and at the same time keeping a certain freedom for out of the box thinking and to have room for identifying new trends. The survey was structured in 4 sections listed in Table 1. The main question concerned the identification of technology combinations that will enable the next steps towards PV cost reduction. Short term refers to about 1 year and longer term to the 5–10 year range. The idea to specify tentative cost targets arose from the need to select feasible technologies combining high performance together with cost effectiveness. The costs are merely indicative and should be read as a qualitative indicator for the impact of the technology. Figs. 2–6 1.1.1. Roadmap of technologies that will enable cost reduction In this section we report on the main results of the survey together with part of the discussion that happened at the workshop. 1.2. Feedstock The large majority of the respondents opted for the optimized Siemens process as the technology suitable to reach a PV module cost reduction down to 50$c/Wp in the very near future. Major challenges identified here are the large investment cost, the fact
630
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
Australia, 2%
Belgium, 5% Australia
Canada, 5% UK, 2%
China, 5%
USA, 14%
the Netherlands, 9%
Denmark, 2%
Belgium Canada China
France, 7% Denmark France
Spain, 2% Singapore, 2%
Germany
Germany, 19%
Norway, 12%
Italy Japan Norway
Japan, 7%
Italy, 7%
Fig. 1. Participant country of origin.
Table 1 Structure of the survey. Topic
Question
1
Which are the technology combinations enabling the next steps towards PV cost reduction? (e.g. towards 50$c/Wp and 20–30$c/Wp)
2
What is the key parameter to enable the next step towards PV cost reduction?
3
Tentative outline of technology roadmap
4
Open questions
Siemens
Siemens
Optimised Siemens
Optimised Siemens
FBR
FBR
UMG
UMG
Other
Other 0%
20%
40%
60%
0%
Feedstock Crystallization Wafering Cell technology Wafer physical parameter Other Wafer thickness Kerf loss Minority carrier diffusion length Efficiency % n-type versus p-type % mono/multi/cast mono wafers Wafer production cost Expected module production cost Other items or parameters Topics for R&D How should a roadmap look like
10%
20%
30%
40%
Fig. 2. Feedstock frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
that the absolute cost limit may be approaching, and the need to check if maximization of by-product reuse can give room to further cost reduction. On the longer term, the respondents were almost equally distributed between Fluidized Bed Reactor (FBR) and Upgraded Metallurgical Grade (UMG) feedstock to reach further cost reductions
in the next five years. Among the challenges identified for FBR are the difficulty to scale up and the monopolistic situation including IP protection. For UMG, the main unknowns are whether the cost target for this type of technology is really competitive with respect to FBR and Siemens, the demonstration of a clear performance validation and showing that there is no significant efficiency penalty.
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
Standard casting
Standard casting
Seedless casting
Seedless casting
Cast mono
Cast mono
Cz
Cz
CCz
CCz
Direct wafering
Direct wafering
FZ
FZ
Other
Other 0%
5%
10%
15%
20%
25%
30%
0%
631
5%
10%
15%
20%
25%
30%
35%
Fig. 3. Crystallization frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
Mono p-type
Mono p-type
Mono n-type
Mono n-type
Mul p-type
Mul p-type
Mul n-type
Mul n-type 0%
10%
20%
30%
40%
0%
10%
20%
30%
40%
50%
60%
Fig. 4. Cell technology frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
Mono
Multi
Mono
24%
7x
Multi
22%
Efficiency
Diffusion length [thickness]
9x
5x
3x
20%
18%
16%
1x Now
2014
2015
2020
2025
Now
2014
2015
2020
2025
Fig. 5. Which will be the average mainstream diffusion length for mono and multicrystalline Si solar cells?
Fig. 6. Which will be the average mainstream efficiency for mono and multicrystalline Si solar cells?
1.3. Crystallization
dominant role although overall keeping the same current market share (more than 40%). A large share, about a third, is expected to be fulfilled by other technologies like direct wafering and FZ. Multicrystalline silicon ingots (including cast mono) will still be present but its share will be strongly reduced (down to 25%). These changes are quite surprising and, probably, they have to do with the expectation for an increase in the efficiency gap between mono and multicrystalline solar cells in the future (see Cell technologies).
At present, both multicrystalline and monocrystalline silicon have a substantial market share. In the short term it is expected that cast mono and advanced casting (aka high performance multi) will dominate with about 50% of the share. The respondents clearly indicated that to further reduce the cost in the longer term, monocrystalline ingots, and in particular continuous Cz ingots, will play a
632
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
The main challenges indicated for cast mono material are the yield and the quality of the wafers. This refers to the multicrystalline fraction at the ingot edges and to the presence of highly dislocated areas in the ingot core towards the top [6]. Development on continuous Cz material should mainly address oxygen reduction since oxygen content is still relatively high compared to multicrystalline ingots and detrimental in p-type devices due to the Light Induced Degradation (LID). Also the segregation of impurities will play an important role with accurate specifications needed to balance the yield or ingot length with the cell performance distribution. For FZ, the extent of capital and operating costs (e.g. feeding rods) are the main challenge. Alternative technologies like direct wafering including exfoliation or lift-off techniques and CVD/ epitaxial growth should mainly show in the coming years how to reach economics of scale, how they can be implemented (integrated) in the value chain and be competitive with current main stream technology and its development. 1.4. Wafering According to our survey, the wafer thickness is expected to continuously reduce down to 120 μm by 2020. Kerfloss is also expected to decrease continuously down to 100 μm. Fixed abrasives will dominate the coming years but on the longer term kerfless wafering will need to play a bigger role. The main challenges identified for fixed abrasive based wafering are the wire cost and wire breakage together with the post-sawing surface preparation. Another challenge, and potential opportunity, is the application to multicrystalline blocks. A larger step down to 50–100 μm in wafer thickness is anticipated by 2025 with the introduction of kerf-free advanced technologies like exfoliation, epi-growth and/or lift-off technologies. However for their introduction, production cost and cell process technologies still need to be demonstrated. 1.5. Cell technology On the short time, our respondents indicate that focus should be on further developing cost-effective p-type multicrystalline devices and n-type monocrystalline devices. On the longer term monocrystalline technology on n-type wafers will take over with a share of about 50%. The process complexity versus performance is the main challenge foreseen for monocrystalline cells. In p-type cells, LID mitigation is identified as another major challenge. This can be achieved on wafer level by reducing the oxygen concentration or using alternative doping. On cell level, the boron-oxygen defects can be stabilized using regeneration and hydrogenation, however the industrial feasibility of these techniques has to be proven yet. It is quite surprising that the multi share reduces so dramatically. The extent of multi reduction, as confirmed also in the crystallization section, was not expected and also not predicted in other roadmaps. In term of material requirements, the diffusion length required should increase steadily up to 10 times the wafer thickness for monocrystalline silicon wafers. Essentially this should be achieved by decreasing the thickness since at the current thickness the diffusion length is approaching the Auger limit. The requirements for diffusion length depend strongly on the cell architecture and the operational injection level as well. This also explains the large error bar in the response. For multicrystalline wafers, diffusion lengths of 5 (max 7) times the wafer thickness seem to be the limit due to the nature of the growth. However, on the long term, one can argue whether cast mono or other non-contact crucible techniques can push this limit higher.
For mono- and multicrystalline wafers, the expected efficiency increases up to 23% and 20% respectively. Therefore the efficiency gap between mono and multi is expected to double from the current 1.5%abs to 3%abs. In term of $/Wp at module level, the cost advantage of multi wafers will be overcompensated by the 3% mono versus multi efficiency gap (14%rel to multi) and even more on system level due to the balance of system costs. Indeed, according to the respondents, both mono- and multicrystalline wafer costs are expected to decrease 5% per year on average and the cost gap of mono versus multi wafers shrinks from the current 13% down to 5% in 2025. The outcome of this analysis is in agreement with the earlier expectation for monocrystalline wafers to be the dominant enabling technology for dramatic module cost reduction. The respondents indicated also a wafer cost target of 60 $c in 2025 which would require significant innovations and also architecture changes when e.g. considering the implementation of very thin wafers. Regarding the solar cell technologies there is large agreement on the long term for an IBC-type of structure (homo- or heterojunction not specified). Instead for the medium term there is no differentiation with PERC, PERL, PERT and heterojunction structures being equally accepted together with the standard Al-BSF.
2. Main conclusions During the workshop several discussions were held on the main question: “Which are the challenges for Si material research”. The discussion focused on the results of the online survey and on the outstanding contributions of experts and presenting authors. One of the most debated points was how to select technologies that would enable cost reduction. There is general agreement that a new technology cannot be properly evaluated without analyzing its impact on the entire value chain. Therefore any alternative technology step aiming to replace a conventional one should be evaluated within a certain scenario for wafer, cell and module production. A few possible scenarios to reduce costs were proposed. No consensus was reached with respect to the best scenario enabling the strongest cost reduction. This makes sense since opinions are quite different, reflecting the wide spectrum of research currently carried out by research groups and companies. During the discussion, there was not enough time to collect the details of the technologies within each scenario, and the market/ economy indicators which would be necessary in order to make a correct evaluation of these options. However, a guideline to look at the different options was generally accepted: Any new or alternative technology has to demonstrate its efficacy (in performance and/or cost reduction terms) not only in comparison to the one in direct competition (horizontally on the value chain) but also vertically along the value chain. In fact, any variation in one process would have an effect on the previous and following production steps and therefore on the final product and its performance. It has been shown already that technologies belonging to the mainstream state-of-the-art have a much steeper learning curve and rapid cost reduction since they benefit from massive development efforts from different players. For example, when research institutes, producers, equipment and material suppliers work on the same technology, the impact and outcome of their R&D multiplies thanks to a better synergy of their efforts. The farther a new technology is from the mainstream technology, the more likely it will fail to benefit from these global development efforts. This is especially evident if a new technology needs modifications of the preceding or following processes in the value chain. In this case, R&D efforts also have to be directed to these other processes. However, especially in the initial development phase, other players might not necessarily deviate or invest their
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
R&D resources on these processes. The responsibility of demonstrating the potential benefit of the new technology lies with the innovator alone, who has to blend his own R&D efforts and compete with players even outside his area of expertise. The concept of the moving target [7] applies within crystalline silicon technologies as well. This however does not mean that innovative technologies would have no room to further develop PV. On the one hand we have to accept that it is getting more and more difficult to have a revolutionary change in the PV sector due to the large capacity and assets already in place including R&D and know-how resources. On the other hand a technology which requires a change not only horizontally in the value chain but also modifications up- and down-stream (vertically) will have to face larger challenges, having to compete not only with direct competitors but with the entire sector and also it will miss the beneficial support from the mainstream R&Ds. Hence industrial cooperation, alliances and shared development programs help to overcome these issues to the benefit of the whole PV sector. For this to happen, a common technology roadmap needs to be developed, shared and agreed upon to some extent. The market expansion also plays an important role. In the future, decisions will be even more market-driven rather than technology-dominated. If the market growth is limited, a drastic change of the mainstream technology is more difficult. Once expansion reaches a critical volume however, revolutionary changes might still have an important chance of success. In this context the idea of a technological roadmap is a useful tool to coordinate the R&D across multiple technologies. In this way critical technology as well as technology gaps can be identified. In this editorial, at the workshop and in these special issue papers a series of options have been discussed and future trends have been indicated. The coordination of such a roadmap is difficult in a competitive sector. However, the information exchange and sharing of R&D insights achieved within our workshop was very useful even when an agreement on a particular technology of choice was not reached. The lack of compelling arguments for excluding a certain technology at present is also interesting by itself: such an agreement on a particular technology might not even be the best path for the moment and a thorough collection of arguments for several scenarios may provide the most valuable information for market players and thus would be a task worthwhile for the near future. If one general conclusion can be derived from this workshop, it is that there is a general trend towards higher efficiencies and, to support this, Si materials technologies need further development
633
in order to match the new tighter requirements for advanced devices. For the future, we aim to hold another workshop and to form a working group that can update and further analyze these trends. Please contact the authors for further information and in case you want to contribute. Finally, during the workshop 27 contributions in addition to four personal statements from invited experts were given by industry, research institutes and universities. We decided to publish the highest rated contributions in this special issue of Solar Energy Materials and Solar Cells. References [1] Advanced Silicon Materials for Photovoltaic Applications, Edited by S. Pizzini, John Wiley & Sons, 2012. [2] J. Nyhus, Solar Grade Silicon – History, Chemistry and Standardization, Silicon for the chemical and solar industry X, Norway, 2010 (June 28–July 2). [3] SEMI, International Technology Roadmap for Photovoltaic (ITRPV), fifth ed., March 2014. [4] EPIA, Global Market Outlook for Photovoltaic 2014–2018. http://www.epia.org/ news/publications/global-market-outlook-for-photovoltaics-2014–2018. [5] For access to the aggregated data, contact Gianluca Coletti at [email protected]. [6] See contributions to this special issue. [7] R.M. Swanson, A vision for crystalline silicon photovoltaics, Prog. Photovolt: Res. Appl 14 (2006) 443–453.
Gianluca Coletti ECN Solar Energy, the Netherlands Ivan Gordon IMEC, Belgium Martin C. Schubert, Wilhelm Warta Fraunhofer ISE, Germany Eivind Johannes Ovrelid Sintef, Norway Anis Jouini CEA-INES, France Mario Tucci ENEA, Italy Giampiero de Cesare University of Rome “La Sapienza”, Italy
Available online 21 August 2014
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Editorial
Challenges for photovoltaic silicon materials$
This special issue is dedicated to the 2nd workshop on Silicon Materials. These Si Materials workshops address a specific issue in the field of crystalline silicon for photovoltaic applications. The first workshop, organized in 2008, focused on specifications for solar-grade feedstock as a response to the polysilicon shortage [1]. A definition of Solar Grade (SoG) Silicon categories based on the usage for solar cell manufacturing was proposed in 2008 and a tentative and approximate standard set of specifications aimed to the fabrication of solar cells was provided. The workshop also gave input to start an initiative [2] for the industrial standardization of silicon feedstock specifications with the institution by SEMI of the silicon feedstock task force. In the last 5 years the global situation changed considerably from several perspectives [3,4]. The feedstock shortage regime for the PV industry ended. There was a delay of a few years before extra polysilicon capacity could enter into the market because of the large capital expenditure and the time needed for the construction and ramping up of poly-silicon production plants. Despite the credit crunch hitting the financial sector, PV flourished with record installations in 2010 and 2011. Eventually, the economic stagnation and the abrupt reduction of subsidies affected the PV market causing an oversupply of PV modules. As a consequence, profit margins strongly reduced and the market stabilized in 2012. All these factors initiated an even stronger drive towards cost reduction pushing the PV industry in a consolidation period. Currently (mid-June 2014), we observe a mild readjustment of the module price towards a more healthy economy of profit and a market that is growing again significantly. Initially triggered by the shortage of feedstock and by the need of a clear understanding of the polysilicon limitation factors, research on silicon materials and on new silicon purification methods multiplied in the past years. However, due to the rapid change in the economic and market situations, an alignment between the short-term industrial needs and the long-term research performed by research institutes is not always straightforward. Despite the proliferation of conferences related to photovoltaics, it was felt that a dedicated event in the form of an actual workshop with ample discussions on “Silicon material” research was desirable. Therefore, the aim of this 2nd Silicon Materials workshop was to provide a worldwide platform where industry and research groups would discuss their programs and future targets with the ambition to find new synergies for the development of PV solar energy. For these reasons, the second edition of the workshop was dedicated to the definition of future challenges
☆ Special issue for the 2nd Silicon Materials workshop, Rome, Italy, October 7th– 8th 2013.
http://dx.doi.org/10.1016/j.solmat.2014.07.045 0927-0248/& 2014 Elsevier B.V. All rights reserved.
for Silicon material research and to discuss industrial needs versus research efforts. 1. Online survey In preparation of the workshop discussions and to sketch a roadmap for Si material research, a survey was carried out among a large number of international experts. The result of the survey was presented during the workshop. This allowed us to identify the most crucial topics that were then discussed in detail during the workshop. The survey focused on the properties and challenges of Si materials for photovoltaics. The input for this survey came from 44 experts working in the industry, research institutes and universities worldwide (Fig. 1) [5]. The respondents do not form a representative sample of the entire scientific community but they did include a significant share of the experts in Si materials for photovoltaics. This is complementary to other initiatives like the ITRPViii and it is expected to give a more visionary interpretation of enabling technologies and challenges for Si materials as seen from the expert point of view. 1.1. Methodology The survey contained open questions and multiple choice questions. This was chosen in order to obtain an easy aggregation of the input data for the discussion and at the same time keeping a certain freedom for out of the box thinking and to have room for identifying new trends. The survey was structured in 4 sections listed in Table 1. The main question concerned the identification of technology combinations that will enable the next steps towards PV cost reduction. Short term refers to about 1 year and longer term to the 5–10 year range. The idea to specify tentative cost targets arose from the need to select feasible technologies combining high performance together with cost effectiveness. The costs are merely indicative and should be read as a qualitative indicator for the impact of the technology. Figs. 2–6 1.1.1. Roadmap of technologies that will enable cost reduction In this section we report on the main results of the survey together with part of the discussion that happened at the workshop. 1.2. Feedstock The large majority of the respondents opted for the optimized Siemens process as the technology suitable to reach a PV module cost reduction down to 50$c/Wp in the very near future. Major challenges identified here are the large investment cost, the fact
630
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
Australia, 2%
Belgium, 5% Australia
Canada, 5% UK, 2%
China, 5%
USA, 14%
the Netherlands, 9%
Denmark, 2%
Belgium Canada China
France, 7% Denmark France
Spain, 2% Singapore, 2%
Germany
Germany, 19%
Norway, 12%
Italy Japan Norway
Japan, 7%
Italy, 7%
Fig. 1. Participant country of origin.
Table 1 Structure of the survey. Topic
Question
1
Which are the technology combinations enabling the next steps towards PV cost reduction? (e.g. towards 50$c/Wp and 20–30$c/Wp)
2
What is the key parameter to enable the next step towards PV cost reduction?
3
Tentative outline of technology roadmap
4
Open questions
Siemens
Siemens
Optimised Siemens
Optimised Siemens
FBR
FBR
UMG
UMG
Other
Other 0%
20%
40%
60%
0%
Feedstock Crystallization Wafering Cell technology Wafer physical parameter Other Wafer thickness Kerf loss Minority carrier diffusion length Efficiency % n-type versus p-type % mono/multi/cast mono wafers Wafer production cost Expected module production cost Other items or parameters Topics for R&D How should a roadmap look like
10%
20%
30%
40%
Fig. 2. Feedstock frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
that the absolute cost limit may be approaching, and the need to check if maximization of by-product reuse can give room to further cost reduction. On the longer term, the respondents were almost equally distributed between Fluidized Bed Reactor (FBR) and Upgraded Metallurgical Grade (UMG) feedstock to reach further cost reductions
in the next five years. Among the challenges identified for FBR are the difficulty to scale up and the monopolistic situation including IP protection. For UMG, the main unknowns are whether the cost target for this type of technology is really competitive with respect to FBR and Siemens, the demonstration of a clear performance validation and showing that there is no significant efficiency penalty.
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
Standard casting
Standard casting
Seedless casting
Seedless casting
Cast mono
Cast mono
Cz
Cz
CCz
CCz
Direct wafering
Direct wafering
FZ
FZ
Other
Other 0%
5%
10%
15%
20%
25%
30%
0%
631
5%
10%
15%
20%
25%
30%
35%
Fig. 3. Crystallization frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
Mono p-type
Mono p-type
Mono n-type
Mono n-type
Mul p-type
Mul p-type
Mul n-type
Mul n-type 0%
10%
20%
30%
40%
0%
10%
20%
30%
40%
50%
60%
Fig. 4. Cell technology frequency distribution: Which key technology will allow us to reach rapid module cost reduction on the short term (left) and on the longer term (right)?
Mono
Multi
Mono
24%
7x
Multi
22%
Efficiency
Diffusion length [thickness]
9x
5x
3x
20%
18%
16%
1x Now
2014
2015
2020
2025
Now
2014
2015
2020
2025
Fig. 5. Which will be the average mainstream diffusion length for mono and multicrystalline Si solar cells?
Fig. 6. Which will be the average mainstream efficiency for mono and multicrystalline Si solar cells?
1.3. Crystallization
dominant role although overall keeping the same current market share (more than 40%). A large share, about a third, is expected to be fulfilled by other technologies like direct wafering and FZ. Multicrystalline silicon ingots (including cast mono) will still be present but its share will be strongly reduced (down to 25%). These changes are quite surprising and, probably, they have to do with the expectation for an increase in the efficiency gap between mono and multicrystalline solar cells in the future (see Cell technologies).
At present, both multicrystalline and monocrystalline silicon have a substantial market share. In the short term it is expected that cast mono and advanced casting (aka high performance multi) will dominate with about 50% of the share. The respondents clearly indicated that to further reduce the cost in the longer term, monocrystalline ingots, and in particular continuous Cz ingots, will play a
632
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
The main challenges indicated for cast mono material are the yield and the quality of the wafers. This refers to the multicrystalline fraction at the ingot edges and to the presence of highly dislocated areas in the ingot core towards the top [6]. Development on continuous Cz material should mainly address oxygen reduction since oxygen content is still relatively high compared to multicrystalline ingots and detrimental in p-type devices due to the Light Induced Degradation (LID). Also the segregation of impurities will play an important role with accurate specifications needed to balance the yield or ingot length with the cell performance distribution. For FZ, the extent of capital and operating costs (e.g. feeding rods) are the main challenge. Alternative technologies like direct wafering including exfoliation or lift-off techniques and CVD/ epitaxial growth should mainly show in the coming years how to reach economics of scale, how they can be implemented (integrated) in the value chain and be competitive with current main stream technology and its development. 1.4. Wafering According to our survey, the wafer thickness is expected to continuously reduce down to 120 μm by 2020. Kerfloss is also expected to decrease continuously down to 100 μm. Fixed abrasives will dominate the coming years but on the longer term kerfless wafering will need to play a bigger role. The main challenges identified for fixed abrasive based wafering are the wire cost and wire breakage together with the post-sawing surface preparation. Another challenge, and potential opportunity, is the application to multicrystalline blocks. A larger step down to 50–100 μm in wafer thickness is anticipated by 2025 with the introduction of kerf-free advanced technologies like exfoliation, epi-growth and/or lift-off technologies. However for their introduction, production cost and cell process technologies still need to be demonstrated. 1.5. Cell technology On the short time, our respondents indicate that focus should be on further developing cost-effective p-type multicrystalline devices and n-type monocrystalline devices. On the longer term monocrystalline technology on n-type wafers will take over with a share of about 50%. The process complexity versus performance is the main challenge foreseen for monocrystalline cells. In p-type cells, LID mitigation is identified as another major challenge. This can be achieved on wafer level by reducing the oxygen concentration or using alternative doping. On cell level, the boron-oxygen defects can be stabilized using regeneration and hydrogenation, however the industrial feasibility of these techniques has to be proven yet. It is quite surprising that the multi share reduces so dramatically. The extent of multi reduction, as confirmed also in the crystallization section, was not expected and also not predicted in other roadmaps. In term of material requirements, the diffusion length required should increase steadily up to 10 times the wafer thickness for monocrystalline silicon wafers. Essentially this should be achieved by decreasing the thickness since at the current thickness the diffusion length is approaching the Auger limit. The requirements for diffusion length depend strongly on the cell architecture and the operational injection level as well. This also explains the large error bar in the response. For multicrystalline wafers, diffusion lengths of 5 (max 7) times the wafer thickness seem to be the limit due to the nature of the growth. However, on the long term, one can argue whether cast mono or other non-contact crucible techniques can push this limit higher.
For mono- and multicrystalline wafers, the expected efficiency increases up to 23% and 20% respectively. Therefore the efficiency gap between mono and multi is expected to double from the current 1.5%abs to 3%abs. In term of $/Wp at module level, the cost advantage of multi wafers will be overcompensated by the 3% mono versus multi efficiency gap (14%rel to multi) and even more on system level due to the balance of system costs. Indeed, according to the respondents, both mono- and multicrystalline wafer costs are expected to decrease 5% per year on average and the cost gap of mono versus multi wafers shrinks from the current 13% down to 5% in 2025. The outcome of this analysis is in agreement with the earlier expectation for monocrystalline wafers to be the dominant enabling technology for dramatic module cost reduction. The respondents indicated also a wafer cost target of 60 $c in 2025 which would require significant innovations and also architecture changes when e.g. considering the implementation of very thin wafers. Regarding the solar cell technologies there is large agreement on the long term for an IBC-type of structure (homo- or heterojunction not specified). Instead for the medium term there is no differentiation with PERC, PERL, PERT and heterojunction structures being equally accepted together with the standard Al-BSF.
2. Main conclusions During the workshop several discussions were held on the main question: “Which are the challenges for Si material research”. The discussion focused on the results of the online survey and on the outstanding contributions of experts and presenting authors. One of the most debated points was how to select technologies that would enable cost reduction. There is general agreement that a new technology cannot be properly evaluated without analyzing its impact on the entire value chain. Therefore any alternative technology step aiming to replace a conventional one should be evaluated within a certain scenario for wafer, cell and module production. A few possible scenarios to reduce costs were proposed. No consensus was reached with respect to the best scenario enabling the strongest cost reduction. This makes sense since opinions are quite different, reflecting the wide spectrum of research currently carried out by research groups and companies. During the discussion, there was not enough time to collect the details of the technologies within each scenario, and the market/ economy indicators which would be necessary in order to make a correct evaluation of these options. However, a guideline to look at the different options was generally accepted: Any new or alternative technology has to demonstrate its efficacy (in performance and/or cost reduction terms) not only in comparison to the one in direct competition (horizontally on the value chain) but also vertically along the value chain. In fact, any variation in one process would have an effect on the previous and following production steps and therefore on the final product and its performance. It has been shown already that technologies belonging to the mainstream state-of-the-art have a much steeper learning curve and rapid cost reduction since they benefit from massive development efforts from different players. For example, when research institutes, producers, equipment and material suppliers work on the same technology, the impact and outcome of their R&D multiplies thanks to a better synergy of their efforts. The farther a new technology is from the mainstream technology, the more likely it will fail to benefit from these global development efforts. This is especially evident if a new technology needs modifications of the preceding or following processes in the value chain. In this case, R&D efforts also have to be directed to these other processes. However, especially in the initial development phase, other players might not necessarily deviate or invest their
Editorial / Solar Energy Materials & Solar Cells 130 (2014) 629–633
R&D resources on these processes. The responsibility of demonstrating the potential benefit of the new technology lies with the innovator alone, who has to blend his own R&D efforts and compete with players even outside his area of expertise. The concept of the moving target [7] applies within crystalline silicon technologies as well. This however does not mean that innovative technologies would have no room to further develop PV. On the one hand we have to accept that it is getting more and more difficult to have a revolutionary change in the PV sector due to the large capacity and assets already in place including R&D and know-how resources. On the other hand a technology which requires a change not only horizontally in the value chain but also modifications up- and down-stream (vertically) will have to face larger challenges, having to compete not only with direct competitors but with the entire sector and also it will miss the beneficial support from the mainstream R&Ds. Hence industrial cooperation, alliances and shared development programs help to overcome these issues to the benefit of the whole PV sector. For this to happen, a common technology roadmap needs to be developed, shared and agreed upon to some extent. The market expansion also plays an important role. In the future, decisions will be even more market-driven rather than technology-dominated. If the market growth is limited, a drastic change of the mainstream technology is more difficult. Once expansion reaches a critical volume however, revolutionary changes might still have an important chance of success. In this context the idea of a technological roadmap is a useful tool to coordinate the R&D across multiple technologies. In this way critical technology as well as technology gaps can be identified. In this editorial, at the workshop and in these special issue papers a series of options have been discussed and future trends have been indicated. The coordination of such a roadmap is difficult in a competitive sector. However, the information exchange and sharing of R&D insights achieved within our workshop was very useful even when an agreement on a particular technology of choice was not reached. The lack of compelling arguments for excluding a certain technology at present is also interesting by itself: such an agreement on a particular technology might not even be the best path for the moment and a thorough collection of arguments for several scenarios may provide the most valuable information for market players and thus would be a task worthwhile for the near future. If one general conclusion can be derived from this workshop, it is that there is a general trend towards higher efficiencies and, to support this, Si materials technologies need further development
633
in order to match the new tighter requirements for advanced devices. For the future, we aim to hold another workshop and to form a working group that can update and further analyze these trends. Please contact the authors for further information and in case you want to contribute. Finally, during the workshop 27 contributions in addition to four personal statements from invited experts were given by industry, research institutes and universities. We decided to publish the highest rated contributions in this special issue of Solar Energy Materials and Solar Cells. References [1] Advanced Silicon Materials for Photovoltaic Applications, Edited by S. Pizzini, John Wiley & Sons, 2012. [2] J. Nyhus, Solar Grade Silicon – History, Chemistry and Standardization, Silicon for the chemical and solar industry X, Norway, 2010 (June 28–July 2). [3] SEMI, International Technology Roadmap for Photovoltaic (ITRPV), fifth ed., March 2014. [4] EPIA, Global Market Outlook for Photovoltaic 2014–2018. http://www.epia.org/ news/publications/global-market-outlook-for-photovoltaics-2014–2018. [5] For access to the aggregated data, contact Gianluca Coletti at [email protected]. [6] See contributions to this special issue. [7] R.M. Swanson, A vision for crystalline silicon photovoltaics, Prog. Photovolt: Res. Appl 14 (2006) 443–453.
Gianluca Coletti ECN Solar Energy, the Netherlands Ivan Gordon IMEC, Belgium Martin C. Schubert, Wilhelm Warta Fraunhofer ISE, Germany Eivind Johannes Ovrelid Sintef, Norway Anis Jouini CEA-INES, France Mario Tucci ENEA, Italy Giampiero de Cesare University of Rome “La Sapienza”, Italy
Available online 21 August 2014