Final Report Summary - POLYSIMODE (Improved polycrystalline-silicon modules on glass substrates)
Executive summary:
The aim of this project is to improve the efficiency and the cost effectiveness of thin-film polysilicon solar modules. Thin-film polysilicon solar cells have recently emerged as a promising thin-film alternative to bulk crystalline silicon. With solid phase crystallisation (SPC) of amorphous Si, CSG Solar AG achieved in the past mini-modules with an efficiency of around 10 %, matching the efficiencies of the best European micromorph mini modules. The main goals of POLYSIMODE are to have large-area polysilicon modules with an efficiency of 12 % and with a cost of 0.7 euro per Watt peak at the end of the project. The expected impact of the proposed project is to enhance the efficiency of polysilicon modules, thereby increasing their cost effectiveness. At the start of the project, the partnership consisted of five research institutes (IMEC, Fraunhofer ISE, CNRS-InESS, Czech Academy of Science - FZU, and HZB) and one company (CSG Solar AG). At M21, the industrial partner CSG Solar AG became insolvent and had to stop all activities and leave the consortium. To be able to continue the project and achieve all technical and valorisation goals, two new industrial partners were included into the consortium from M27 onwards, namely Evonik Industries and Suntech R&D Australia (SRDA).
Within the POLYSIMODE project we worked on the improvement of the crystallographic and electronic quality of the polysilicon material and on the development of advanced new methods for light confinement. By in-depth characterisation of the polysilicon material, a better understanding of the relationship between the processing parameters, the electrical and optical properties of the material and the resulting device properties was obtained. The most promising technique to grow the polysilicon layers was identified as being the e-beam crystallisation approach in which a layer of amorphous silicon is crystallised by scanning an e-beam across. This method enables the largest grain sizes, the best material quality in terms of open circuit voltage, and a significant increase of throughput. Moreover, two fast and effective methods to improve the electrical material quality of Si absorber materials have been established: a lamp-field and a laser based annealing technology. The two methods are advantageous over the standard rapid thermal annealing (RTA) as the heating energy is introduced by radiation alone which enables us to keep the thermal load of the substrates at a very low level. The beneficial use of amorphous silicon layers as hetero back surface field (BSF) and as hetero-emitter was shown. A new glass texturing process in which the glass is abraded with a bead blaster and glass damage is removed by an etching step has been successfully introduced into the CSG production process, leading to a substantial gain in absolute efficiency on 1.4 m2 panels. Plasma texturing of the rear silicon surface has been developed to create light trapping in devices that need to be grown on flat or slightly textured glass substrates. Moreover, a novel use of Raman spectroscopy was developed to evaluate the light trapping capabilities in Si thin films. Microwave-detected photoconductance decay (MWPCD), Raman microspectroscopy and terahertz spectroscopy were identified as candidates for in-line characterisation methods that could be used to monitor material quality during production. Overall, a cumulated cost saving potential of about 1 EUR / Wp was identified for a device process based on steps developed within POLYSIMODE compared to the original CSG Solar AG process.
Project context and objectives:
Thin-film polysilicon solar cells have recently emerged as a promising thin-film alternative to bulk crystalline Si. With SPC of amorphous Si, CSG Solar AG achieved devices with efficiencies of around 7 % for module areas of 1.4 m2 and around 8 % for module areas of 30 x 39 cm2. With this technique, CSG also achieved mini-modules (10 x 10 cm2) with an efficiency of more than 10 %, matching the efficiencies of the best European micromorph mini-modules. Moreover, simulations point out that with the proper crystallographic quality of the polysilicon material and with advanced device processing, single-junction polysilicon solar cells with efficiencies above 15 % are feasible. The aim of this project is to increase the efficiency of large-area polysilicon modules to more than 12 % and thus to improve their cost-effectiveness. This project aims to reduce the module cost of goods sold per Watt peak (COGS / Wp) of these modules to 0.7 EUR / Wp by the end of the project. This will be achieved by the improvement of the crystallographic and electronic quality of the active polysilicon material and by the implementation of new, advanced methods for light confinement. The device structure and material of the CSG modules at the beginning of this project are taken as the starting point. By advanced characterisation of the polysilicon material and related interfaces, a better understanding of the relationship between the growth parameters, the electrical and optical properties of the material and the resulting device properties will be obtained. This information will be used to enhance the material quality by the development of more effective post-deposition treatments (defect annealing and hydrogen passivation) for polysilicon and by the development of improved material growth processes leading to a much lower number of defects in the material. Besides by an improved material quality, the efficiency of polysilicon modules will also be increased by the development of advanced methods for light confinement. We note that there is no need for transparent conductive oxide (TCO) layers in the thin-film polysilicon solar cell technology, which is, in addition to the long-term stability in performance, a clear advantage of polysilicon modules compared to other thin-film solar cell technologies like the micromorph and CIGS technologies. The main concrete scientific and technological objectives of the project are:
(a) the development of alternative silicon deposition methods, having increased growth rates and / or leading to higher material quality;
(b) optimised in-line post-deposition treatments (defect annealing and hydrogen passivation) leading to improved polysilicon material quality;
(c) improved light trapping schemes that enhance the efficiency of polysilicon modules;
(d) improved understanding of the relationship between silicon growth parameters, post-deposition treatments and material quality;
(e) thin-film polysilicon modules (39 x 30 cm2) produced in-line with efficiencies above 12%;
(f) a COGS / PW value for polysilicon modules below 0.7 EUR / Wp at the end of the project.
It should be stressed that POLYSIMODE fits very well in the Strategic Research Agenda (SRA) defined by Europe's photovoltaic (PV) technology platform. Thin-film polycrystalline silicon solar modules are listed as one of the priorities in the section 'Emerging and novel PV technologies'. The targets for this topic are listed in Table 1. The topics for the period 2007 - 2013, namely improving the material quality of the polysilicon layers, and upscaling of the deposition are covered in POLYSIMODE. The POLYSIMODE efficiency target of 12 % for a rather large monolithic module of 30 x 39 cm2 by 2012 (end of the project) is in line with the SRA document (14 %) that does not specify a size. Finally the industrial implementation in the period 2014 - 2020 (12-14 % efficiency at a cost of 0.5 - 0.8 EUR / Wp) corresponds well to the cost target and the time horizon of POLYSIMODE. In fact the aim of our project is to have the technology ready to produce polysilicon modules with a cost of 0.7 EUR / Wp by the end of the project. This means that by 2014 industrial production of polysilicon modules with a cost of 0.7 EUR / Wp should be a realisable.
The expected impact of the proposed project is to enhance the efficiency of large-area polysilicon modules, thereby increasing their cost effectiveness. Since all the main European institutes working on thin-film polysilicon solar cells are joining forces within this project, a substantial acceleration in the improvement of the cost-effectiveness of polysilicon modules and therefore the market development of this type of thin-film solar cells is expected. The combination of a real thin-film concept (using only a few micrometres of active material on glass), with its large potential for cost reduction, and of crystalline Si, is a highly innovative approach. The fact that crystalline Si is used for the active material in our module concept means that we can tap from the wealth of knowledge and expertise that is available from the bulk crystalline Si PV industry and from the microelectronic industry. This gives our approach an excellent starting point for a successful development.
In addition, the development of device parts such as heterojunctions (emitters / BSF), temperature stable passivation layers and texturisation processes for the polysilicon material under investigation can also be of considerable interest for related technologies (such as thin silicon ribbons or silicon films made by a layer transfer process).
The targeted cost reduction to a value below 0.7 EUR / Wp will also have a positive impact on the acceptance of PV by the public and by politics. Moreover, since 'energy efficiency' is currently of great interest in public discussion, PV will be an example of one of the highest electricity production efficiencies that have been achieved of all power generators (with the additional advantage of using cost-free energy from the sun on the input side, of course).
At the start of the project, the partnership consisted of five research institutes (IMEC, Fraunhofer ISE, CNRS-InESS, Czech Academy of Science - FZU, and HZB) and one company (CSG Solar AG). The active participation of CSG within this project targeted the following benefits:
- It would allow the consortium to produce in-line module demonstrators (39 x 30 cm2) by application of the developed process steps in the pilot line of CSG.
- It would allow the consortium to accurately determine the influence of the developed technologies on the cost effectiveness of polysilicon modules.
- It would ensure that obtained results and technologies developed within the project could be transferred to production lines at the end of the project.
At M21, the industrial partner CSG Solar AG however became insolvent and had to stop all activities and hence leave the consortium. To be able to continue the project and achieve all the technical and valorisation goals described above, two new industrial partners were included into the consortium from M27 onwards, namely Evonik Industries and SRDA.
Project results:
This part describes the main results achieved per work package (WP).
WP1:
In WP1, the POLYSIMODE consortium has pursued four approaches to produce polysilicon layers for thin film solar cell devices: SPC of amorphous silicon, a seed layer approach based on aluminium-induced crystallisation of amorphous Si and epitaxial thickening, direct growth, and electron-beam crystallisation of amorphous Si. While SPC and directly grown layers yield rather small grains of maximum 3 µm size, the seed layer approach and, in particular, the crystallisation by an electron beam allow for much larger grains up to 30 µm (seed layers) or even one millimetre (e-beam crystallisation).
By the SPC approach no further improvement of the material quality in terms of open circuit voltage and of the efficiency could be achieved. However, electron-beam evaporation of silicon was established as new deposition method exceeding the rates achievable with conventional plasma-enhanced chemical vapour deposition (PECVD) techniques by more than a factor of 20, thus offering a further substantial cost reduction potential in industrial production. It has been shown that polysilicon deposited by electron-beam evaporation followed by SPC exhibits a slightly better solar cell performance compared to PECVD-grown devices if planar substrates were used. As a 'side' result coming along with the gained knowledge about growth and crystallisation of electron-beam evaporated silicon on textured surfaces, a fabrication process for large-area periodic polysilicon nanostructures was developed with applications in the fields of PV and photonics.
A new efficiency record was achieved for the seed layers approach. By epitaxial thickening of seed layers prepared by aluminium-induced crystallisation of amorphous silicon a solar cell efficiency of 8.5 % was achieved, coming along with an excellent material quality reflected in an open circuit voltage of 522 mV.
Two high-rate silicon deposition methods have been identified to be able to deliver the up to 15 µm thick polysilicon layers by direct crystalline growth within a few minutes: Atmospheric pressure chemical vapour deposition (APCVD) and the above mentioned electron-beam evaporation of silicon. Although the electrical material quality of such directly grown polysilicon layers is limited, they are a perfect precursor material for the electron-beam crystallisation technique.
The most promising technique turned out to be the e-beam crystallisation approach enabling the largest grain sizes, the best material quality in terms of open circuit voltage, and a significant increase of throughput by fast liquid phase fabrication processes. Efficiencies up to 5.9 % and a reproducibly high polysilicon material quality reflected in open circuit voltages above 530 mV could be achieved. A potential analysis of these solar cells based on available cell data, particularly the excellent open circuit voltages, predicts an achievable efficiency over 11 % with improved device concept.
WP2:
The main goal of WP2 was to improve the electrical quality of the polysilicon thin films after deposition and / or crystallisation.
In the course of the project, two fast and effective methods to improve the electrical material quality of Si absorber materials with grain sizes in the range of tens to hundreds of micrometres have been established: a lamp-field and a laser based annealing technology. The two methods are advantageous over the standard RTA as the heating energy is introduced by radiation alone. That enables us to heat up mainly the absorber material and keeping the thermal load of the substrates at a very low level. Several combinations of glasses and glass surfaces have been tested in the two new set ups. The processes were optimised so that we were able to apply the maximal temperature load to the absorber before corrupting the substrates. Thereby several processes have been established both with the lamp-field and the laser set up that have been found to be superior in terms of material quality to the standard RTA process.
The second part of this WP dealt with the improvement of the material quality by atomic hydrogenation defect passivation. Several set-ups for hydrogenation processing have been tested and compared in order to find an optimal process. Furthermore different types of material quality have been characterised and the main achievement was the fundamental understanding and the successful adaptation of the process sequence with annealing and hydrogen passivation processes. For each specific material another specialised sequence has been established. We therefore found that the best process parameters for thermal annealing and hydrogen passivation can be quite different depending on the crystallisation procedure although the materials look the same at first glance. Our understanding of defect development during the processes has hereby also been improved significantly.
An additional part of WP2 was the introduction of amorphous silicon layers as hetero BSF as well as emitter layers. This was done successfully for lab type processing. Unfortunately we were not successful in adapting a newly developed low-cost amorphous silicon: H deposition process. Although first solar cells led to some success, technological difficulties which had to be met for very high Voc values, could not be overcome due to a lack of time.
WP3:
WP3 focused on advanced light trapping. Since the JSC of state-of-the-art polysilicon devices is pretty high (round 30 mA cm-2 for a 2-micron thick absorber) due to the very effective light confinement developed by CSG Solar AG in the past, the POLYSIMODE project focused more on the improvement of the material quality of the absorbers (improvement in Voc) than on a further improvement of the already very effective light trapping scheme available.
Nevertheless, the following achievements were accomplished:
- A new glass texturing process in which the glass is abraded with a bead blaster and glass damage is removed by an etching step has been successfully introduced into the CSG production process, leading to a substantial gain in absolute efficiency on 1.4 m2 panels.
- Plasma texturing of the rear silicon surface has been developed to create a Lambertian reflection and oblique coupling of the light into the absorber. Moreover, a significant improvement in the fill factor of devices has been observed after plasma texturing. This type of texturing is especially suited for absorbers made by electron beam evaporation of amorphous silicon and crystallisation, since these absorbers need to be grown on flat or only slightly textured glass substrates to achieve a high material quality. Plasma texturing the rear silicon is then crucial to obtain a high current density.
- Optical simulations allowed identifying the optimal geometrical and optical parameters of the anti-reflective coating (ARC) layer between the glass and the polysilicon layer in our solar cell devices. An estimated gain of 10 % in current density can be achieved by using this optimised ARC layer compared to the standard SiNx layer currently used in state-of-the-art devices.
WP4:
In WP4 the POLYSIMODE consortium has developed new approaches to microscopic characterisation of electronic properties of microcrystalline and polycrystalline silicon thin films, with resolution down to a few nanometres. The methods were mainly based on point contact represented by the tip of a conductive cantilever in an atomic force microscopy (C-AFM) or by multiple contacts navigated by scanning electron microscopy. The local conductivity maps were used to characterise polycrystalline silicon layers prepared by different methods and their changes during the technological steps.
It was found that under certain conditions the tip modifies the sample surface by local anodic oxidation. The resulting oxide may even fundamentally change the character of the local conductivity map, leaving the grain boundaries more conductive than the oxidised grains. This research led to guidelines how to avoid oxidation and even how to remove oxide from aged and oxidised samples.
Another important advance was the clear distinction between the dark conductivity and photoconductivity maps. This progress allows for measurement of local photoresponse at different grains or at different positions on the sample.
We found that locating the grains, grain boundaries or defects in various microscopes (using electron or optical beams or scanning probe) requires a precision on the order of 100 nm or better after repeated mounting of the sample. This surpasses the current capabilities of microscopic tables. We have thus developed a localisation procedure which uses a combination of scratches using a diamond stylus and indentation marks prepared by a microhardness tester. The procedure was tested for locating the same spot in Raman microspectroscopy (optical microscopy), electron backscatter diffraction (EBSD) and C-AFM. The technique allows for repeated sample mounting even after various technological steps and may find wider use in materials research.
Important advances were made in using the Raman microspectroscopy, a fast, non-destructive and contactless tool, which is widely used to assess the crystalline nature of the polycrystalline silicon films and even its electronic properties. We have developed a novel use of Raman spectroscopy to evaluate the light trapping capabilities in Si thin films. Typical thickness of the silicon thin films used for solar cells is between 0.3 - 3 µm, which is not enough for complete absorption of near infrared light. In order to improve the light absorption, silicon films are deposited on rough substrates. We have found that the absolute intensities of the Raman spectra excited by a 785-nanometre laser are strongly affected by light scattering on a rough microcrystalline surface. This effect is due to the proportionality of the Raman intensity to the laser path length in the silicon film, which may be increased several times by light trapping. Correlation between the absolute Raman intensity and the final properties of the cell (IQE and JSC) was demonstrated and Raman spectroscopy thus became also a candidate for use as an inline characterisation tool described in WP5.
Amongst the other characterisation approaches we want to point out the use of microwave detected photoconductance decay method (MWPCD) which was adapted to the specific properties of polycrystalline Si layers and which is another candidate for inline characterisation.
WP5: Demonstration and cost calculation
The main objectives were to implement process steps developed within this project at module (30 x 39 cm2) level, to develop a standard in-line characterisation procedure for polysilicon modules, and to determine the cost effectiveness of new process steps when implemented in production.
Effect of new process steps on the efficiency at 30 x 39 cm2 module level
Due to the limited resources after the exit of CSG from the consortium and the shift of the project focus to e-beam recrystallised samples, a further fabrication into larger modules has not been pursued. However, line focus RTA (LF-RTA) was found to lead to an increase in quality of cells and minimodules, measured by Voc. Voc values after the hydrogenation ranged up to 400 mV. Laser thermal annealing (LTA) of CSG produced polycrystalline silicon multilayer samples was done at InESS to reduce the defects and dopant activation. Before laser treatment, Voc of the samples are around 185 ± 5 mV. Maximum Voc obtained is 288 ± 5 mV, i.e. an improvement of 100 mV before plasma hydrogenation.
Direct comparison of LTA and conventional RTA on planar 'SUNTECH samples' showed a relative improvement of 3.6 % in Voc, 1.4 % in pseudo fill factor (pFF) and 4.3 % in pseudo efficiency (pEff), when LTA is used in place of RTA. Laser Crystallisation with a dual wave length laser was evaluated at InESS in collaboration with HZB. 2 µm thick a-Si films were crystallised with the help of 40 nm highly phosphorus doped polysilicon layer deposited on SiN coated borosilicate glass. Raman spectra proved successful crystallisation, but with a grain size of less than 100 nm and no preferential orientation. A decrease of FWHM was observed with increase in laser power and increase of substrate temperature. The sample processing temperature was limited due to melting of the borosilicate glass.
Unfortunately, no working cells were obtained from these samples due to damage of the underlying n+ layer.
In order to further increase the short circuit current of CSG samples a plasma texturing process was introduced at ISE. Jsc of up to 19.3 mA / cm2 and a Voc of 486 mV were measured compared to 17.7 mA / cm2 and 466 mV for unstructured samples. Although these new processes could not be applied to the module line at CSG, the knowledge gained from these experiments has significantly enhanced the process know-how at the institutions involved. Pathways for new processing routes have been investigated and can be made available to the interested commercial customer.
Development of in-line characterisation and monitoring
At the PV production lines each module is characterised upon completion by a flasher, which means the measurement takes place at the end of the production line. For improved process control and optimisation and implementation of non-contact inline measurement tools is desirable.
MWPCD measurements on polycrystalline silicon thin-film material on glass have been carried out by ISE. MWPCD is a well-established characterisation technique in the semiconductor and PV fields. Excess charge carriers are generated in the sample by a short laser pulse and the decay of photoconductance is detected by following the microwave reflectance. The time constant of the decay is extracted and associated with the effective minority carrier lifetime. In the POLYSIMODE project, a commercially available MWPCD setup WT2000 (Semilab Co. Ltd.) was used and adapted for polycrystalline silicon films. MWPCD was proven to be a likely online measurement tool for polycrystalline silicon.
The time-resolved terahertz spectroscopy is a contact-free method able to follow the local (round 10 nm) in-plane motion of charges with a subpicosecond to nanosecond time resolution. Furthermore, different conductivity mechanisms may then appear simultaneously as unique fingerprints in the time resolved terahertz spectra (Drude-like conductivity, hopping, etc). At FZU experiments were done to test the THz spectroscopy for polycrystalline silicon films at a home-built system. A lot of further effort is required to adapt it for poly-Si measurements. However, due to rapid progress of the THz technology motivated by detection of explosives for security purposes, turn-key commercial systems are becoming available for use in industrial process control. To our knowledge, we obtained the first results in THz domain for polycrystalline silicon thin films. It remains to be seen whether the technique would be sensitive to passivation or rapid-thermal annealing of the samples.
Raman microspectroscopy is contactless and non-destructive measurement method which does not require special sample preparation. At FZU a RenishawInVia with 365, 442 and 785 nm laser excitation wavelengths is used.
The Raman spectra may be used for several diagnostics purposes: crystallinity of the film, analyse light trapping, evaluate intrinsic stress and holes density (or doping efficiency - crucial for open circuit voltage VOC values) and carrier lifetime at high injections levels. In order to increase sensitivity of the method to carrier's density, 785 nm laser should be used for excitation of the Raman spectra, which also allows to get as much information as possible. In order to provide information about the sample homogeneity, Raman systems needs to allow recording multiple spots, which is available on the market.
The feasibility of integrating the mentioned techniques depends strongly on the compatibility with the conditions that govern typical production lines: number of necessary measurements per module, number of pulses per measurement and integration times, acceptable spot sizes, precision of position of substrate with respect to the measurement head (focal depth), resistance to vibrations etc.
In batch processing the total measurement time for a 1 x 1.4m2 substrate should not exceed 1 - 2 minutes. This means only 1 - 2 measurements in the case of Terahertz spectroscopy, while Raman and microwave PCD could perform a larger number (> 25) of measurements per substrate. When equally distributed over 1 x 1.4 m2, each of 25 measurements would represent an area of 24 x 24 cm2.
Under these basic considerations, Terahertz Spectroscopy appears to comply easily with the restrictions on minimum focal depth and could deliver of the order 1 - 2 measurements per substrate in a batch processing scenario.
Raman and microwave photoconductive detection techniques would be able to deliver a much more detailed map of the poly-Si properties of the module. Commercial systems allow reducing the measurement time to the sub-second regime and are able to perform mapping of areas. Also microwave based techniques for the detection of photoconductivity are commercially available for measuring the photoconductivity.
For some production lines it might be attractive to characterise the modules while they are moving on a conveyor belt. As times for laboratory measurements are not compatible with typical belt speeds an adaption is needed like moving the measurement heads at the same speed as the belt for the duration of the measurement, which leads to some inaccuracy. But the most promising path to reliable inline measurement on a moving substrate would be the simplification of the data collection.
Another objective of the project was to improve the calibration of module IV-measurement under illumination. The investigations of polycrystalline silicon on glass devices showed that there was an improvement in uncertainty of the mismatch correction thanks to the matched reference cell made during the project. The determination of the spectral response out of a module is challenging and specially prepared modules with single contacted cells are necessary. Also the resolution of the spectral response is critical due to interferences. The comparison between dc- and flash simulator measurements showed good congruence, so that the larger flash simulator could be used for module characterisation.
Demonstration of a 30 x 39 cm2 module with an efficiency of at least 12 %
A device stack built on borofloat glass using liquid phase crystallisation by means of a line shaped electron beam was produced. The device evolves a modified version of the CSG contact system. The first attempt to transfer crystallised layers to SRDA was not successful in terms of device performance but showed the applicability of the SRDA contact scheme on liquid phase crystallised absorbers as a module cell voltage of 372 mV per cell was achieved. Parallel to these experiments device processing was continued at HZB using a MESA etch based cell architecture relying on an a-Si:H hetero-junction. As the design suffers from a large metallised area - covering 16% of the cells front side - the active area values JSC = 19.2 mA / cm2, ETA = 5.9 % provide a realistic view of the absorbers quality.
Besides this result using high rate electron beam evaporated silicon as precursor, a comparison was realised using ISE's high-rate APCVD process for absorber deposition. The I(V) curves of the fabricated c-Si(p)/a-Si:H(n+) heterojunction cells show a high and homogenous Voc up to 532 mV as well as an increased fill factor of 61 % mostly caused by a reduced series resistance due to higher absorber dopant level. However, a low short circuit current density of 13.7 mA / cm3 remains due to various loss mechanisms.
While the deliverable of 12 % module efficiency was not reached, progress has been made in terms of material quality as reflected in the reproducibly achieved open circuit voltages quite above 500 mV with a record of 532 mV. The applicability of the original CSG contact scheme was demonstrated.
An optimised cell contacting scheme as well as post crystallisation treatments are expected to enhance the short circuit current density as a potential analysis of these cells performed based on available cell data predicts an achievable JSC above 28.7 mA /cm2, thus boosting the device performance to over 11 %. These values dont account for any improvement related gain in Voc.
Cost benefit analysis of new module process developed within the project
Within POLYSIMODE potential cost savings for alternative processes and parameters were estimated.
Substrate: So far, borosilicate glass supplied by Schott (BOROFLOAT 33) was used as the reference substrate material. The development of new processes for crystallisation allows usage of other substrates that do not require high-temperature stability. In this case a price reduction for the substrate up to 80 % is possible.
Silicon deposition: CSG used for production an Oerlikon KAI 1200 PECVD equipment. Today more efficient silicon deposition techniques are available on the market. Besides PECVD which is also offered by Leybold, APCVD, which is developed at Fraunhofer ISE and Ebeam deposition are the most promising future deposition techniques. Although Ebeam deposition is not commercially available at the moment, first cost estimations predict a cost saving potential of up to 80 % compared to the CSG standard process.
Crystallisation: CSG used a 24-hour oven process for SPC of silicon, which is a cost driver due to its long processing time. Within this project the cost saving potential of Zone-melt-recrystallisation (ZMR) and Electron-beam crystallisation (EBC) were reviewed. ZMR is also a solid-phase crystallisation process, but reduces the necessary crystallisation time to minutes. First cost estimations predict a cost reduction about one order of magnitude.
EBC is a liquid phase crystallisation process. HZB is leading the development of EBC for silicon PV. Unfortunately, this process is so far only available at a laboratory scale and reliable cost estimation for a pilot or production scale would require upscaling experiments and theoretical simulations. These were beyond the scope of this project. However, there is a strong interest in the scientific community and at commercial tool manufacturers to investigate this technology further.
Overall, a cumulated cost saving potential of about 1 EUR / Wp was identified for the 'updated' CSG process, upon implementation of the three steps described above.
Potential impact:
The European PV technology platform has defined a SRA with several research challenges that, if properly addressed, will provide significant advances to the European industry. Thin-film polysilicon devices were mentioned in this agenda in 2009 at the start of this project as among the most promising candidates for advanced inorganic thin film technologies. Therefore, POLYSIMODE has contributed to the improvement of the European technological base by the development and demonstration of thin-film polysilicon cells and modules.
Since all the main European institutes working on thin-film polysilicon solar cells were joining forces within this project, a substantial acceleration in the improvement of the cost-effectiveness of polysilicon modules and therefore the market development of this type of thin-film solar cells has been achieved.
Moreover, the combination of a real thin-film concept (using only a few micrometres of active material on glass), with its large potential for cost reduction, and of crystalline silicon, is a highly innovative approach, giving the European industry a clear advantage.
Communication and dissemination
A public website of the POLYSIMODE project is available at http://www.polysimode.eu(si apre in una nuova finestra). Its content includes a general presentation of the project. Updates concerning public dissemination and events were made on a regular basis.
The consortium has published 20 scientific papers in peer-reviewed journals with results obtained within POLYSIMODE. Moreover, the consortium has presented 54 contributions at international conferences and workshops in the form of invited presentations, regular oral presentations and posters. Conferences included top conferences in the field such as the European Photovoltaic Solar Energy Conference, the IEEE Photovoltaic Specialists conference and the E-MRS spring meeting. The POLYSIMODE project also resulted in 10 PhD and Master theses.
Work done within the frame of POLYSIMODE was also presented at the IUVSTA summer school on Physics at Nanoscale, which was organised with support of the POLYSIMODE project in June 2011 in Devet Skal, Czech Republic.
Exploitation
The POLYSIMODE project led to 3 patent applications. The main exploitable results generated by POLYSIMODE include:
(a) large-area silicon nanoarchitectures by combining nanoimprint-lithography and silicon electron-beam evaporation;
(b) electron-beam evaporation as high-rate deposition technique for silicon thin films;
(c) electron-beam crystallisation as high-throughput fabrication technique for high-quality polysilicon thin films;
(e) continuous wave laser processing for defect passivation and for crystallisation;
(f) definition of a procedure on how to locate the same spot in various microscopic measurement techniques in order to characterise the same sample spot during different stages of the production process;
(g) the use of Raman spectroscopy for light-trapping characterisation;
(h) halogen lamp induced RTA and rapid SPC;
(i) plasma texturisation of polysilicon for light trapping purposes;
(j) amorphous-silicon / crystalline-silicon heterojunctions for emitter and BSF;
(k) improved hydrogen plasma passivation.
Within POLYSIMODE, several processes have been developed and optimised for an enhanced thin-film polysilicon module technology. It is now up to the European actors to capitalise on these results to contribute to the European leadership by providing innovative PV solutions to the market.
List of websites: http://www.polysimode.eu(si apre in una nuova finestra)
The list of main project contacts is:
IMEC - Coordinator: Dr Ivan Gordon, gordoni@imec.be, Tel: +32 (0)16 28 82 49
Fraunhofer ISE: Dr Stefan Janz, stefan.janz@ise.fraunhofer.de
FZU: Dr Antonin Fejfar, fejfar@fzu.cz
HZB: Dr Christiane Becker, christiane.becker@helmholtz-berlin.de
CNRS-InESS: Dr Abdelilah Slaoui, abdelilah.slaoui@icube.unistra.fr
SRDA: Mr Jonathon Dore, jonathon.dore@suntech-power.com.au
Evonik: Dr Paul Woebkenberg, paul.woebkenberg@evonik.com
The aim of this project is to improve the efficiency and the cost effectiveness of thin-film polysilicon solar modules. Thin-film polysilicon solar cells have recently emerged as a promising thin-film alternative to bulk crystalline silicon. With solid phase crystallisation (SPC) of amorphous Si, CSG Solar AG achieved in the past mini-modules with an efficiency of around 10 %, matching the efficiencies of the best European micromorph mini modules. The main goals of POLYSIMODE are to have large-area polysilicon modules with an efficiency of 12 % and with a cost of 0.7 euro per Watt peak at the end of the project. The expected impact of the proposed project is to enhance the efficiency of polysilicon modules, thereby increasing their cost effectiveness. At the start of the project, the partnership consisted of five research institutes (IMEC, Fraunhofer ISE, CNRS-InESS, Czech Academy of Science - FZU, and HZB) and one company (CSG Solar AG). At M21, the industrial partner CSG Solar AG became insolvent and had to stop all activities and leave the consortium. To be able to continue the project and achieve all technical and valorisation goals, two new industrial partners were included into the consortium from M27 onwards, namely Evonik Industries and Suntech R&D Australia (SRDA).
Within the POLYSIMODE project we worked on the improvement of the crystallographic and electronic quality of the polysilicon material and on the development of advanced new methods for light confinement. By in-depth characterisation of the polysilicon material, a better understanding of the relationship between the processing parameters, the electrical and optical properties of the material and the resulting device properties was obtained. The most promising technique to grow the polysilicon layers was identified as being the e-beam crystallisation approach in which a layer of amorphous silicon is crystallised by scanning an e-beam across. This method enables the largest grain sizes, the best material quality in terms of open circuit voltage, and a significant increase of throughput. Moreover, two fast and effective methods to improve the electrical material quality of Si absorber materials have been established: a lamp-field and a laser based annealing technology. The two methods are advantageous over the standard rapid thermal annealing (RTA) as the heating energy is introduced by radiation alone which enables us to keep the thermal load of the substrates at a very low level. The beneficial use of amorphous silicon layers as hetero back surface field (BSF) and as hetero-emitter was shown. A new glass texturing process in which the glass is abraded with a bead blaster and glass damage is removed by an etching step has been successfully introduced into the CSG production process, leading to a substantial gain in absolute efficiency on 1.4 m2 panels. Plasma texturing of the rear silicon surface has been developed to create light trapping in devices that need to be grown on flat or slightly textured glass substrates. Moreover, a novel use of Raman spectroscopy was developed to evaluate the light trapping capabilities in Si thin films. Microwave-detected photoconductance decay (MWPCD), Raman microspectroscopy and terahertz spectroscopy were identified as candidates for in-line characterisation methods that could be used to monitor material quality during production. Overall, a cumulated cost saving potential of about 1 EUR / Wp was identified for a device process based on steps developed within POLYSIMODE compared to the original CSG Solar AG process.
Project context and objectives:
Thin-film polysilicon solar cells have recently emerged as a promising thin-film alternative to bulk crystalline Si. With SPC of amorphous Si, CSG Solar AG achieved devices with efficiencies of around 7 % for module areas of 1.4 m2 and around 8 % for module areas of 30 x 39 cm2. With this technique, CSG also achieved mini-modules (10 x 10 cm2) with an efficiency of more than 10 %, matching the efficiencies of the best European micromorph mini-modules. Moreover, simulations point out that with the proper crystallographic quality of the polysilicon material and with advanced device processing, single-junction polysilicon solar cells with efficiencies above 15 % are feasible. The aim of this project is to increase the efficiency of large-area polysilicon modules to more than 12 % and thus to improve their cost-effectiveness. This project aims to reduce the module cost of goods sold per Watt peak (COGS / Wp) of these modules to 0.7 EUR / Wp by the end of the project. This will be achieved by the improvement of the crystallographic and electronic quality of the active polysilicon material and by the implementation of new, advanced methods for light confinement. The device structure and material of the CSG modules at the beginning of this project are taken as the starting point. By advanced characterisation of the polysilicon material and related interfaces, a better understanding of the relationship between the growth parameters, the electrical and optical properties of the material and the resulting device properties will be obtained. This information will be used to enhance the material quality by the development of more effective post-deposition treatments (defect annealing and hydrogen passivation) for polysilicon and by the development of improved material growth processes leading to a much lower number of defects in the material. Besides by an improved material quality, the efficiency of polysilicon modules will also be increased by the development of advanced methods for light confinement. We note that there is no need for transparent conductive oxide (TCO) layers in the thin-film polysilicon solar cell technology, which is, in addition to the long-term stability in performance, a clear advantage of polysilicon modules compared to other thin-film solar cell technologies like the micromorph and CIGS technologies. The main concrete scientific and technological objectives of the project are:
(a) the development of alternative silicon deposition methods, having increased growth rates and / or leading to higher material quality;
(b) optimised in-line post-deposition treatments (defect annealing and hydrogen passivation) leading to improved polysilicon material quality;
(c) improved light trapping schemes that enhance the efficiency of polysilicon modules;
(d) improved understanding of the relationship between silicon growth parameters, post-deposition treatments and material quality;
(e) thin-film polysilicon modules (39 x 30 cm2) produced in-line with efficiencies above 12%;
(f) a COGS / PW value for polysilicon modules below 0.7 EUR / Wp at the end of the project.
It should be stressed that POLYSIMODE fits very well in the Strategic Research Agenda (SRA) defined by Europe's photovoltaic (PV) technology platform. Thin-film polycrystalline silicon solar modules are listed as one of the priorities in the section 'Emerging and novel PV technologies'. The targets for this topic are listed in Table 1. The topics for the period 2007 - 2013, namely improving the material quality of the polysilicon layers, and upscaling of the deposition are covered in POLYSIMODE. The POLYSIMODE efficiency target of 12 % for a rather large monolithic module of 30 x 39 cm2 by 2012 (end of the project) is in line with the SRA document (14 %) that does not specify a size. Finally the industrial implementation in the period 2014 - 2020 (12-14 % efficiency at a cost of 0.5 - 0.8 EUR / Wp) corresponds well to the cost target and the time horizon of POLYSIMODE. In fact the aim of our project is to have the technology ready to produce polysilicon modules with a cost of 0.7 EUR / Wp by the end of the project. This means that by 2014 industrial production of polysilicon modules with a cost of 0.7 EUR / Wp should be a realisable.
The expected impact of the proposed project is to enhance the efficiency of large-area polysilicon modules, thereby increasing their cost effectiveness. Since all the main European institutes working on thin-film polysilicon solar cells are joining forces within this project, a substantial acceleration in the improvement of the cost-effectiveness of polysilicon modules and therefore the market development of this type of thin-film solar cells is expected. The combination of a real thin-film concept (using only a few micrometres of active material on glass), with its large potential for cost reduction, and of crystalline Si, is a highly innovative approach. The fact that crystalline Si is used for the active material in our module concept means that we can tap from the wealth of knowledge and expertise that is available from the bulk crystalline Si PV industry and from the microelectronic industry. This gives our approach an excellent starting point for a successful development.
In addition, the development of device parts such as heterojunctions (emitters / BSF), temperature stable passivation layers and texturisation processes for the polysilicon material under investigation can also be of considerable interest for related technologies (such as thin silicon ribbons or silicon films made by a layer transfer process).
The targeted cost reduction to a value below 0.7 EUR / Wp will also have a positive impact on the acceptance of PV by the public and by politics. Moreover, since 'energy efficiency' is currently of great interest in public discussion, PV will be an example of one of the highest electricity production efficiencies that have been achieved of all power generators (with the additional advantage of using cost-free energy from the sun on the input side, of course).
At the start of the project, the partnership consisted of five research institutes (IMEC, Fraunhofer ISE, CNRS-InESS, Czech Academy of Science - FZU, and HZB) and one company (CSG Solar AG). The active participation of CSG within this project targeted the following benefits:
- It would allow the consortium to produce in-line module demonstrators (39 x 30 cm2) by application of the developed process steps in the pilot line of CSG.
- It would allow the consortium to accurately determine the influence of the developed technologies on the cost effectiveness of polysilicon modules.
- It would ensure that obtained results and technologies developed within the project could be transferred to production lines at the end of the project.
At M21, the industrial partner CSG Solar AG however became insolvent and had to stop all activities and hence leave the consortium. To be able to continue the project and achieve all the technical and valorisation goals described above, two new industrial partners were included into the consortium from M27 onwards, namely Evonik Industries and SRDA.
Project results:
This part describes the main results achieved per work package (WP).
WP1:
In WP1, the POLYSIMODE consortium has pursued four approaches to produce polysilicon layers for thin film solar cell devices: SPC of amorphous silicon, a seed layer approach based on aluminium-induced crystallisation of amorphous Si and epitaxial thickening, direct growth, and electron-beam crystallisation of amorphous Si. While SPC and directly grown layers yield rather small grains of maximum 3 µm size, the seed layer approach and, in particular, the crystallisation by an electron beam allow for much larger grains up to 30 µm (seed layers) or even one millimetre (e-beam crystallisation).
By the SPC approach no further improvement of the material quality in terms of open circuit voltage and of the efficiency could be achieved. However, electron-beam evaporation of silicon was established as new deposition method exceeding the rates achievable with conventional plasma-enhanced chemical vapour deposition (PECVD) techniques by more than a factor of 20, thus offering a further substantial cost reduction potential in industrial production. It has been shown that polysilicon deposited by electron-beam evaporation followed by SPC exhibits a slightly better solar cell performance compared to PECVD-grown devices if planar substrates were used. As a 'side' result coming along with the gained knowledge about growth and crystallisation of electron-beam evaporated silicon on textured surfaces, a fabrication process for large-area periodic polysilicon nanostructures was developed with applications in the fields of PV and photonics.
A new efficiency record was achieved for the seed layers approach. By epitaxial thickening of seed layers prepared by aluminium-induced crystallisation of amorphous silicon a solar cell efficiency of 8.5 % was achieved, coming along with an excellent material quality reflected in an open circuit voltage of 522 mV.
Two high-rate silicon deposition methods have been identified to be able to deliver the up to 15 µm thick polysilicon layers by direct crystalline growth within a few minutes: Atmospheric pressure chemical vapour deposition (APCVD) and the above mentioned electron-beam evaporation of silicon. Although the electrical material quality of such directly grown polysilicon layers is limited, they are a perfect precursor material for the electron-beam crystallisation technique.
The most promising technique turned out to be the e-beam crystallisation approach enabling the largest grain sizes, the best material quality in terms of open circuit voltage, and a significant increase of throughput by fast liquid phase fabrication processes. Efficiencies up to 5.9 % and a reproducibly high polysilicon material quality reflected in open circuit voltages above 530 mV could be achieved. A potential analysis of these solar cells based on available cell data, particularly the excellent open circuit voltages, predicts an achievable efficiency over 11 % with improved device concept.
WP2:
The main goal of WP2 was to improve the electrical quality of the polysilicon thin films after deposition and / or crystallisation.
In the course of the project, two fast and effective methods to improve the electrical material quality of Si absorber materials with grain sizes in the range of tens to hundreds of micrometres have been established: a lamp-field and a laser based annealing technology. The two methods are advantageous over the standard RTA as the heating energy is introduced by radiation alone. That enables us to heat up mainly the absorber material and keeping the thermal load of the substrates at a very low level. Several combinations of glasses and glass surfaces have been tested in the two new set ups. The processes were optimised so that we were able to apply the maximal temperature load to the absorber before corrupting the substrates. Thereby several processes have been established both with the lamp-field and the laser set up that have been found to be superior in terms of material quality to the standard RTA process.
The second part of this WP dealt with the improvement of the material quality by atomic hydrogenation defect passivation. Several set-ups for hydrogenation processing have been tested and compared in order to find an optimal process. Furthermore different types of material quality have been characterised and the main achievement was the fundamental understanding and the successful adaptation of the process sequence with annealing and hydrogen passivation processes. For each specific material another specialised sequence has been established. We therefore found that the best process parameters for thermal annealing and hydrogen passivation can be quite different depending on the crystallisation procedure although the materials look the same at first glance. Our understanding of defect development during the processes has hereby also been improved significantly.
An additional part of WP2 was the introduction of amorphous silicon layers as hetero BSF as well as emitter layers. This was done successfully for lab type processing. Unfortunately we were not successful in adapting a newly developed low-cost amorphous silicon: H deposition process. Although first solar cells led to some success, technological difficulties which had to be met for very high Voc values, could not be overcome due to a lack of time.
WP3:
WP3 focused on advanced light trapping. Since the JSC of state-of-the-art polysilicon devices is pretty high (round 30 mA cm-2 for a 2-micron thick absorber) due to the very effective light confinement developed by CSG Solar AG in the past, the POLYSIMODE project focused more on the improvement of the material quality of the absorbers (improvement in Voc) than on a further improvement of the already very effective light trapping scheme available.
Nevertheless, the following achievements were accomplished:
- A new glass texturing process in which the glass is abraded with a bead blaster and glass damage is removed by an etching step has been successfully introduced into the CSG production process, leading to a substantial gain in absolute efficiency on 1.4 m2 panels.
- Plasma texturing of the rear silicon surface has been developed to create a Lambertian reflection and oblique coupling of the light into the absorber. Moreover, a significant improvement in the fill factor of devices has been observed after plasma texturing. This type of texturing is especially suited for absorbers made by electron beam evaporation of amorphous silicon and crystallisation, since these absorbers need to be grown on flat or only slightly textured glass substrates to achieve a high material quality. Plasma texturing the rear silicon is then crucial to obtain a high current density.
- Optical simulations allowed identifying the optimal geometrical and optical parameters of the anti-reflective coating (ARC) layer between the glass and the polysilicon layer in our solar cell devices. An estimated gain of 10 % in current density can be achieved by using this optimised ARC layer compared to the standard SiNx layer currently used in state-of-the-art devices.
WP4:
In WP4 the POLYSIMODE consortium has developed new approaches to microscopic characterisation of electronic properties of microcrystalline and polycrystalline silicon thin films, with resolution down to a few nanometres. The methods were mainly based on point contact represented by the tip of a conductive cantilever in an atomic force microscopy (C-AFM) or by multiple contacts navigated by scanning electron microscopy. The local conductivity maps were used to characterise polycrystalline silicon layers prepared by different methods and their changes during the technological steps.
It was found that under certain conditions the tip modifies the sample surface by local anodic oxidation. The resulting oxide may even fundamentally change the character of the local conductivity map, leaving the grain boundaries more conductive than the oxidised grains. This research led to guidelines how to avoid oxidation and even how to remove oxide from aged and oxidised samples.
Another important advance was the clear distinction between the dark conductivity and photoconductivity maps. This progress allows for measurement of local photoresponse at different grains or at different positions on the sample.
We found that locating the grains, grain boundaries or defects in various microscopes (using electron or optical beams or scanning probe) requires a precision on the order of 100 nm or better after repeated mounting of the sample. This surpasses the current capabilities of microscopic tables. We have thus developed a localisation procedure which uses a combination of scratches using a diamond stylus and indentation marks prepared by a microhardness tester. The procedure was tested for locating the same spot in Raman microspectroscopy (optical microscopy), electron backscatter diffraction (EBSD) and C-AFM. The technique allows for repeated sample mounting even after various technological steps and may find wider use in materials research.
Important advances were made in using the Raman microspectroscopy, a fast, non-destructive and contactless tool, which is widely used to assess the crystalline nature of the polycrystalline silicon films and even its electronic properties. We have developed a novel use of Raman spectroscopy to evaluate the light trapping capabilities in Si thin films. Typical thickness of the silicon thin films used for solar cells is between 0.3 - 3 µm, which is not enough for complete absorption of near infrared light. In order to improve the light absorption, silicon films are deposited on rough substrates. We have found that the absolute intensities of the Raman spectra excited by a 785-nanometre laser are strongly affected by light scattering on a rough microcrystalline surface. This effect is due to the proportionality of the Raman intensity to the laser path length in the silicon film, which may be increased several times by light trapping. Correlation between the absolute Raman intensity and the final properties of the cell (IQE and JSC) was demonstrated and Raman spectroscopy thus became also a candidate for use as an inline characterisation tool described in WP5.
Amongst the other characterisation approaches we want to point out the use of microwave detected photoconductance decay method (MWPCD) which was adapted to the specific properties of polycrystalline Si layers and which is another candidate for inline characterisation.
WP5: Demonstration and cost calculation
The main objectives were to implement process steps developed within this project at module (30 x 39 cm2) level, to develop a standard in-line characterisation procedure for polysilicon modules, and to determine the cost effectiveness of new process steps when implemented in production.
Effect of new process steps on the efficiency at 30 x 39 cm2 module level
Due to the limited resources after the exit of CSG from the consortium and the shift of the project focus to e-beam recrystallised samples, a further fabrication into larger modules has not been pursued. However, line focus RTA (LF-RTA) was found to lead to an increase in quality of cells and minimodules, measured by Voc. Voc values after the hydrogenation ranged up to 400 mV. Laser thermal annealing (LTA) of CSG produced polycrystalline silicon multilayer samples was done at InESS to reduce the defects and dopant activation. Before laser treatment, Voc of the samples are around 185 ± 5 mV. Maximum Voc obtained is 288 ± 5 mV, i.e. an improvement of 100 mV before plasma hydrogenation.
Direct comparison of LTA and conventional RTA on planar 'SUNTECH samples' showed a relative improvement of 3.6 % in Voc, 1.4 % in pseudo fill factor (pFF) and 4.3 % in pseudo efficiency (pEff), when LTA is used in place of RTA. Laser Crystallisation with a dual wave length laser was evaluated at InESS in collaboration with HZB. 2 µm thick a-Si films were crystallised with the help of 40 nm highly phosphorus doped polysilicon layer deposited on SiN coated borosilicate glass. Raman spectra proved successful crystallisation, but with a grain size of less than 100 nm and no preferential orientation. A decrease of FWHM was observed with increase in laser power and increase of substrate temperature. The sample processing temperature was limited due to melting of the borosilicate glass.
Unfortunately, no working cells were obtained from these samples due to damage of the underlying n+ layer.
In order to further increase the short circuit current of CSG samples a plasma texturing process was introduced at ISE. Jsc of up to 19.3 mA / cm2 and a Voc of 486 mV were measured compared to 17.7 mA / cm2 and 466 mV for unstructured samples. Although these new processes could not be applied to the module line at CSG, the knowledge gained from these experiments has significantly enhanced the process know-how at the institutions involved. Pathways for new processing routes have been investigated and can be made available to the interested commercial customer.
Development of in-line characterisation and monitoring
At the PV production lines each module is characterised upon completion by a flasher, which means the measurement takes place at the end of the production line. For improved process control and optimisation and implementation of non-contact inline measurement tools is desirable.
MWPCD measurements on polycrystalline silicon thin-film material on glass have been carried out by ISE. MWPCD is a well-established characterisation technique in the semiconductor and PV fields. Excess charge carriers are generated in the sample by a short laser pulse and the decay of photoconductance is detected by following the microwave reflectance. The time constant of the decay is extracted and associated with the effective minority carrier lifetime. In the POLYSIMODE project, a commercially available MWPCD setup WT2000 (Semilab Co. Ltd.) was used and adapted for polycrystalline silicon films. MWPCD was proven to be a likely online measurement tool for polycrystalline silicon.
The time-resolved terahertz spectroscopy is a contact-free method able to follow the local (round 10 nm) in-plane motion of charges with a subpicosecond to nanosecond time resolution. Furthermore, different conductivity mechanisms may then appear simultaneously as unique fingerprints in the time resolved terahertz spectra (Drude-like conductivity, hopping, etc). At FZU experiments were done to test the THz spectroscopy for polycrystalline silicon films at a home-built system. A lot of further effort is required to adapt it for poly-Si measurements. However, due to rapid progress of the THz technology motivated by detection of explosives for security purposes, turn-key commercial systems are becoming available for use in industrial process control. To our knowledge, we obtained the first results in THz domain for polycrystalline silicon thin films. It remains to be seen whether the technique would be sensitive to passivation or rapid-thermal annealing of the samples.
Raman microspectroscopy is contactless and non-destructive measurement method which does not require special sample preparation. At FZU a RenishawInVia with 365, 442 and 785 nm laser excitation wavelengths is used.
The Raman spectra may be used for several diagnostics purposes: crystallinity of the film, analyse light trapping, evaluate intrinsic stress and holes density (or doping efficiency - crucial for open circuit voltage VOC values) and carrier lifetime at high injections levels. In order to increase sensitivity of the method to carrier's density, 785 nm laser should be used for excitation of the Raman spectra, which also allows to get as much information as possible. In order to provide information about the sample homogeneity, Raman systems needs to allow recording multiple spots, which is available on the market.
The feasibility of integrating the mentioned techniques depends strongly on the compatibility with the conditions that govern typical production lines: number of necessary measurements per module, number of pulses per measurement and integration times, acceptable spot sizes, precision of position of substrate with respect to the measurement head (focal depth), resistance to vibrations etc.
In batch processing the total measurement time for a 1 x 1.4m2 substrate should not exceed 1 - 2 minutes. This means only 1 - 2 measurements in the case of Terahertz spectroscopy, while Raman and microwave PCD could perform a larger number (> 25) of measurements per substrate. When equally distributed over 1 x 1.4 m2, each of 25 measurements would represent an area of 24 x 24 cm2.
Under these basic considerations, Terahertz Spectroscopy appears to comply easily with the restrictions on minimum focal depth and could deliver of the order 1 - 2 measurements per substrate in a batch processing scenario.
Raman and microwave photoconductive detection techniques would be able to deliver a much more detailed map of the poly-Si properties of the module. Commercial systems allow reducing the measurement time to the sub-second regime and are able to perform mapping of areas. Also microwave based techniques for the detection of photoconductivity are commercially available for measuring the photoconductivity.
For some production lines it might be attractive to characterise the modules while they are moving on a conveyor belt. As times for laboratory measurements are not compatible with typical belt speeds an adaption is needed like moving the measurement heads at the same speed as the belt for the duration of the measurement, which leads to some inaccuracy. But the most promising path to reliable inline measurement on a moving substrate would be the simplification of the data collection.
Another objective of the project was to improve the calibration of module IV-measurement under illumination. The investigations of polycrystalline silicon on glass devices showed that there was an improvement in uncertainty of the mismatch correction thanks to the matched reference cell made during the project. The determination of the spectral response out of a module is challenging and specially prepared modules with single contacted cells are necessary. Also the resolution of the spectral response is critical due to interferences. The comparison between dc- and flash simulator measurements showed good congruence, so that the larger flash simulator could be used for module characterisation.
Demonstration of a 30 x 39 cm2 module with an efficiency of at least 12 %
A device stack built on borofloat glass using liquid phase crystallisation by means of a line shaped electron beam was produced. The device evolves a modified version of the CSG contact system. The first attempt to transfer crystallised layers to SRDA was not successful in terms of device performance but showed the applicability of the SRDA contact scheme on liquid phase crystallised absorbers as a module cell voltage of 372 mV per cell was achieved. Parallel to these experiments device processing was continued at HZB using a MESA etch based cell architecture relying on an a-Si:H hetero-junction. As the design suffers from a large metallised area - covering 16% of the cells front side - the active area values JSC = 19.2 mA / cm2, ETA = 5.9 % provide a realistic view of the absorbers quality.
Besides this result using high rate electron beam evaporated silicon as precursor, a comparison was realised using ISE's high-rate APCVD process for absorber deposition. The I(V) curves of the fabricated c-Si(p)/a-Si:H(n+) heterojunction cells show a high and homogenous Voc up to 532 mV as well as an increased fill factor of 61 % mostly caused by a reduced series resistance due to higher absorber dopant level. However, a low short circuit current density of 13.7 mA / cm3 remains due to various loss mechanisms.
While the deliverable of 12 % module efficiency was not reached, progress has been made in terms of material quality as reflected in the reproducibly achieved open circuit voltages quite above 500 mV with a record of 532 mV. The applicability of the original CSG contact scheme was demonstrated.
An optimised cell contacting scheme as well as post crystallisation treatments are expected to enhance the short circuit current density as a potential analysis of these cells performed based on available cell data predicts an achievable JSC above 28.7 mA /cm2, thus boosting the device performance to over 11 %. These values dont account for any improvement related gain in Voc.
Cost benefit analysis of new module process developed within the project
Within POLYSIMODE potential cost savings for alternative processes and parameters were estimated.
Substrate: So far, borosilicate glass supplied by Schott (BOROFLOAT 33) was used as the reference substrate material. The development of new processes for crystallisation allows usage of other substrates that do not require high-temperature stability. In this case a price reduction for the substrate up to 80 % is possible.
Silicon deposition: CSG used for production an Oerlikon KAI 1200 PECVD equipment. Today more efficient silicon deposition techniques are available on the market. Besides PECVD which is also offered by Leybold, APCVD, which is developed at Fraunhofer ISE and Ebeam deposition are the most promising future deposition techniques. Although Ebeam deposition is not commercially available at the moment, first cost estimations predict a cost saving potential of up to 80 % compared to the CSG standard process.
Crystallisation: CSG used a 24-hour oven process for SPC of silicon, which is a cost driver due to its long processing time. Within this project the cost saving potential of Zone-melt-recrystallisation (ZMR) and Electron-beam crystallisation (EBC) were reviewed. ZMR is also a solid-phase crystallisation process, but reduces the necessary crystallisation time to minutes. First cost estimations predict a cost reduction about one order of magnitude.
EBC is a liquid phase crystallisation process. HZB is leading the development of EBC for silicon PV. Unfortunately, this process is so far only available at a laboratory scale and reliable cost estimation for a pilot or production scale would require upscaling experiments and theoretical simulations. These were beyond the scope of this project. However, there is a strong interest in the scientific community and at commercial tool manufacturers to investigate this technology further.
Overall, a cumulated cost saving potential of about 1 EUR / Wp was identified for the 'updated' CSG process, upon implementation of the three steps described above.
Potential impact:
The European PV technology platform has defined a SRA with several research challenges that, if properly addressed, will provide significant advances to the European industry. Thin-film polysilicon devices were mentioned in this agenda in 2009 at the start of this project as among the most promising candidates for advanced inorganic thin film technologies. Therefore, POLYSIMODE has contributed to the improvement of the European technological base by the development and demonstration of thin-film polysilicon cells and modules.
Since all the main European institutes working on thin-film polysilicon solar cells were joining forces within this project, a substantial acceleration in the improvement of the cost-effectiveness of polysilicon modules and therefore the market development of this type of thin-film solar cells has been achieved.
Moreover, the combination of a real thin-film concept (using only a few micrometres of active material on glass), with its large potential for cost reduction, and of crystalline silicon, is a highly innovative approach, giving the European industry a clear advantage.
Communication and dissemination
A public website of the POLYSIMODE project is available at http://www.polysimode.eu(si apre in una nuova finestra). Its content includes a general presentation of the project. Updates concerning public dissemination and events were made on a regular basis.
The consortium has published 20 scientific papers in peer-reviewed journals with results obtained within POLYSIMODE. Moreover, the consortium has presented 54 contributions at international conferences and workshops in the form of invited presentations, regular oral presentations and posters. Conferences included top conferences in the field such as the European Photovoltaic Solar Energy Conference, the IEEE Photovoltaic Specialists conference and the E-MRS spring meeting. The POLYSIMODE project also resulted in 10 PhD and Master theses.
Work done within the frame of POLYSIMODE was also presented at the IUVSTA summer school on Physics at Nanoscale, which was organised with support of the POLYSIMODE project in June 2011 in Devet Skal, Czech Republic.
Exploitation
The POLYSIMODE project led to 3 patent applications. The main exploitable results generated by POLYSIMODE include:
(a) large-area silicon nanoarchitectures by combining nanoimprint-lithography and silicon electron-beam evaporation;
(b) electron-beam evaporation as high-rate deposition technique for silicon thin films;
(c) electron-beam crystallisation as high-throughput fabrication technique for high-quality polysilicon thin films;
(e) continuous wave laser processing for defect passivation and for crystallisation;
(f) definition of a procedure on how to locate the same spot in various microscopic measurement techniques in order to characterise the same sample spot during different stages of the production process;
(g) the use of Raman spectroscopy for light-trapping characterisation;
(h) halogen lamp induced RTA and rapid SPC;
(i) plasma texturisation of polysilicon for light trapping purposes;
(j) amorphous-silicon / crystalline-silicon heterojunctions for emitter and BSF;
(k) improved hydrogen plasma passivation.
Within POLYSIMODE, several processes have been developed and optimised for an enhanced thin-film polysilicon module technology. It is now up to the European actors to capitalise on these results to contribute to the European leadership by providing innovative PV solutions to the market.
List of websites: http://www.polysimode.eu(si apre in una nuova finestra)
The list of main project contacts is:
IMEC - Coordinator: Dr Ivan Gordon, gordoni@imec.be, Tel: +32 (0)16 28 82 49
Fraunhofer ISE: Dr Stefan Janz, stefan.janz@ise.fraunhofer.de
FZU: Dr Antonin Fejfar, fejfar@fzu.cz
HZB: Dr Christiane Becker, christiane.becker@helmholtz-berlin.de
CNRS-InESS: Dr Abdelilah Slaoui, abdelilah.slaoui@icube.unistra.fr
SRDA: Mr Jonathon Dore, jonathon.dore@suntech-power.com.au
Evonik: Dr Paul Woebkenberg, paul.woebkenberg@evonik.com
