At the end of last month – in May – the 2018 edition of the “PERC Solar Cell Technology” report appeared as published by Taiyang news. It is the third edition of this interesting report and it starts by stating that “we have entered the PERC era in the solar cell technologies segment” with “today, PERC is the new standard”.
Cover of the 2018 edition of the “PERC Solar Cell Technology” report as published by Taiyang news (May 28, 2018).
There are many things to say (and explain) about the rise of PERC and its manufacturing process and this is something I will leave for another blog post for now. But one thing is evident as also is stated clearly in the report: “The key to PERC manufacturing is rear passivation, whereas the unanimous material of choice for this purpose is aluminum oxide, which can be deposited using PECVD machines, well-known from applying silicon nitride, or Atomic Layer Deposition (ALD) tools”. I want to connect to this aspect as our research at the Eindhoven University of Technology has greatly contributed to the exploration of the surface passivation by Al2O3 (ALD and PECVD), to the investigation of fundamental aspects and materials properties underlying the high-level of surface passivation, as well as to the demonstration of Al2O3 in solar cell devices.
I thought about addressing some important aspects of Al2O3 surface passivation and its deposition processes but then I remembered that I had written down many of these aspects in 2011 when preparing a conference paper for the 21st NREL Workshop on Crystalline Silicon Solar Cells & Modules: Materials and Processes organized in Breckenridge Colorado in 2011. I was invited to this conference (taking place annually, see https://siliconworkshop.com) because our work on Al2O3 had attracted a lot of attention by that time. Rereading the conference paper, I found that many of the aspects described in the paper still hold and were quite prescient. Therefore I have decided to copy the text of the whole paper below and to just add some small comments to it. By the way, the paper was based on 10 questions whose answers should give a good idea about “the prospects for the use of Al2O3 for high efficiency solar cells” as this was the title of the paper.
I would like to add here that I also gave a plenary talk at the 25th European PV Solar Energy Conference and Exhibition in Valencia in 2010. This was at the time that the interest in Al2O3 in the solar cell industry really started to take off. I recorded that presentation and you can listen it back here . It should give you a quick overview about all relevant aspects related to Al2O3 in 20 min. Moreover, I want to note that much more information is provided in the review paper that my former PhD student and I wrote in 2012: Status and prospects of Al2O3-based surface passivation schemes for silicon solar cells (link). If you are involved or interested in Al2O3 for solar cells, this is probably a must-read.
Finally I want to mention that many things have happened since these days but as said, this will be addressed in another blog post soon!
Conference paper 21st Workshop on Crystalline Silicon Solar Cells & Modules: Materials and Processes – Breckenridge Colorado – 2011 *
Review on the prospects for the use of Al2O3 for high-efficiency solar cells
Al2O3 is a material that has rapidly gained in popularity in the past years as thin film passivation material for c-Si photovoltaics (PV). In this contribution ten questions will be addressed as might exist in the solar cell community.
1) – Surface passivation by Al2O3, what’s the story?
Already in 1989 Hezel and Jaeger reported on the passivation properties of Al2O3 films at that time prepared by pyrolysis . Although this paper reports on the very interesting properties of the material in terms of surface passivation of c-Si (e.g., the presence of a high density of negative charges), there was more interest for a-SiNx:H thin films at that time and the material remained basically unnoticed in the PV community. This changed however around 2005 when research groups at IMEC  and the Eindhoven University of Technology (TU/e)  showed that Al2O3 films prepared by atomic layer deposition (ALD) – a particular form of chemical vapor deposition (CVD)  – lead to excellent levels of surface passivation of n-type and p-type c-Si. After these initial reports the interest in Al2O3 grew rapidly, especially when it was demonstrated that Al2O3 also leads to an excellent passivation of p+-type surfaces  and after reporting on the performance of solar cells in which the Al2O3was incorporated to passivate rear and front side surfaces of p-type  and n-type  solar cells.
2) – What are the basic material properties of Al2O3 films used for Si passivation?
Al2O3 is a wide bandgap (~8.8 eV for bulk material) dielectric which consists in various crystalline forms. However, for passivation layers amorphous Al2O3 films are used with a somewhat lower bandgap (~6.4 eV) and with a refractive index of ~1.65 at a photon energy of 2 eV. The films are therefore fully transparent over the wavelength region of interest for solar cells. The films are typically quite stoichiometric ([O]/[Al] ratio = ~1.5) although there can be a slight excess of O in the film. When prepared by CVD-based techniques, the films exhibit also a low hydrogen content (typically 2-3 at.%) and this hydrogen is mostly bonded to the (excess) O as –OH groups. It has however been observed that the excellent passivation properties do not depend sensitively on the Al2O3 properties such as stoichiometry and material purity . The hydrogen content of the Al2O3 films is however found to be very important for the chemical passivation of c-Si obtained from the Al2O3 films. This holds also to the interfacial layer of SiOx (1-2 nm thickness) that is (always) formed between the Al2O3 and the Si when applying CVD-based techniques [3,9].
The refractive index n and extinction coefficient k of a 30 nm Al2O3 film deposited by ALD .
3) – Which techniques can be used to prepare Al2O3 thin films?
Al2O3 films for c-Si surface passivation have been deposited by thermal and plasma-assisted ALD employing Al(CH3)3 precursor dosing together with different oxidant sources (H2O, O3 and O2 plasma) [8,11]. Plasma-enhanced CVD (PECVD, from Al(CH3)3 and N2O or CO2 mixtures) has also been employed to deposit Al2O3 [8,12,13] as well as the physical vapor deposition (PVD) technique of sputtering . In the early days (1989) Hezel and Jaeger used pyrolysis of Al(OiPr)3 for the deposition of Al2O3 which were the first results on Al2O3-based passivation of c-Si ever reported . Also sol-gel processes have been investigated for Al2O3 synthesis for c-Si passivation [15,16]. In all these cases annealing of the films at ~400 ºC is beneficial or even required to achieve a high level of surface passivation.
Different reactor configurations for thermal ALD: (a) single-wafer reactor, (b) batch reactor, and spatial ALD reactor. In (a) and (b) the ALD cycles are carried out in the time-domain and in (c) the ALD cycles are carried out in the spatial domain .
4) – What makes Al2O3 so unique for surface passivation?
Two passivation mechanisms can be discerned for Si surfaces. The first mechanism is the reduction of interface state density Dit at the Si surface, e.g., through the passivation of Si dangling bonds by H atoms. This mechanism is referred to as “chemical passivation”. The second mechanism is the reduction of the density of the minority charge carriers present at the Si surface through a built-in electrical field at the surface. This so-called “field effect passivation” can be achieved by doping profiles or by fixed charges Qf present in a thin film deposited on the Si. The excellent passivation by Al2O3 is typically a combination of both mechanisms.
The fact that Al2O3 can contain a very high density (up to 1013 cm-3) of negative charges makes the material unique . Nearly almost all other materials (in particular SiO2 and a-SiNx:H) contain positive fixed charges and at a lower density. For Al2O3 the fixed charges are located at the interface between the Al2O3 and the interfacial SiOx on the Si . Furthermore, it is interesting to note that the density of fixed charges in the Al2O3 depends on the preparation method of the Al2O3. For films prepared by plasma-assisted ALD and PECVD generally a higher Qf is found as for films prepared by thermal ALD. In the later case the excellent level of passivation can mainly be attributed to a low Dit level.
A second key aspect of Al2O3, an aspect that has received less attention so far, is the fact that Al2O3 also acts an effective hydrogen reservoir providing hydrogen to the Si interface during thermal treatments (during annealing and during the firing step). This has recently unambiguously been established  and explains the fact that such an excellent level of chemical passivation can be achieved by Al2O3 films, either deposited directly on H-terminated Si or on Si containing a deposited SiOx layer (e.g., by PECVD or ALD) which is passivating relatively poorly by itself (i.e., when no Al2O3 capping layer is applied) .
Surface recombination velocity Seff,max for plasma-assisted and thermal ALD Al2O3 films as a function of the corona charge density deposited on the Al2O3. This plot reveals that both films contain a fixed negative charge density but with less charge in the thermal ALD sample. The thermal ALD has a higher level of chemical passivation as revealed by the lower value of Seff,max at the point where the fixed charges are compensated by the corona charges.
Note 2018: Recent follow-up research on the passivation of silicon surfaces by various metal oxides has revealed that many of these metal oxides are negative charge dielectrics, e.g., HfO2, Ga2O3, TiO2, Nb2O5, etc.
5) – What is the performance of (industrial-type) solar cells with Al2O3?
Considering the enthusiasm about Al2O3 within the PV community [21,22], it is very likely that the performance of solar cells containing Al2O3 passivation layers is being tested extensively. However as it concerns valuable and proprietary information for PV companies the outcome of these tests are not disclosed or not explicitly reported as such. The first results on solar cells with Al2O3 set however the stage and were crucial in triggering the interest of the PV industry. The first solar cell results were reported for p-type PERC cells in which ALD Al2O3 was used for rear-surface passivation, as a single layer and in a stack combined with PECVD-SiOx (collaboration ISFH – TU/e) . The best efficiency in this first report was 20.6% and in later work for similar solar cells an efficiency of 21.5% was obtained . Another important early achievement was an efficiency of 23.2% for n-type PERL cells in which ALD Al2O3 combined with PECVD a-SiNx:H were used for front-surface passivation (collaboration Fraunhofer ISE – TU/e) . At a later stage an efficiency of 23.5% was achieved for this kind of solar cells . Other solar cell results have been reported by ITRI , ECN  and the University of Konstanz .
PERL solar cell with n-type Si base and a front-surface passivation layer of Al2O3 (30 nm) together with an a-SiNx:H (40 nm) antireflection coating .
Note 2018: Obviously, the industrial breakthrough of Al2O3has been in the PERC technology.
6) – What are the requirements on the film and processing conditions?
Many technical questions need to be addressed in order to implement Al2O3 in solar cells. The answers on these questions evidently depend on the solar cell type and configuration envisioned but some general insights have been obtained from the studies carried out in the last few years. For ALD-deposited films the minimum thickness has been found to be 5 nm and 10 nm for plasma-assisted and thermal ALD, respectively . The difference is expected to originate from the lower importance of field-effect passivation by thermal ALD. The optimum deposition temperature is within the range of 150-250 oC . Although the passivation level is not very sensitive to the deposition temperature, the optimum is ruled by the chemical passivation . At lower temperatures, the Al2O3 film density is not high enough whereas at higher temperatures the Al2O3 has a too low hydrogen content. In both cases, the Al2O3 cannot provide sufficient hydrogen to passivate the Si dangling bonds on the interface (during annealing), either due to too large out-diffusion of hydrogen into the ambient or a too small reservoir of hydrogen to start with. Considering the annealing of Al2O3 – a step that is essential to activate the surface passivation to the full extent – the optimum temperature is around 400 oC . At this temperature sufficient hydrogen is liberated from the film. The fact that the hydrogen from the film reduces the interface state density is also confirmed by the fact that an anneal in N2 gas works well, no forming gas anneal is required. The duration of the annealing step can be as short as 1 min. to provide excellent levels of surface passivation. The Al2O3 is also sufficiently stable during the firing step as used in industrial-type solar cells with screen-printed metallization. The level of passivation however deteriorates during this high temperature step (typically 800 – 900 oC for a few seconds) [28,29] but the remaining level of passivation is by far sufficient for such industrial type solar cells. The Al2O3was also found compatible with a-SiNx:H in stack systems and even an improved thermal stability was reported . Also stacks of Al2O3 with low-temperature-synthesized SiO2 were found to be firing stable .
Surface recombination velocity Seff,max for plasma-assisted and thermal ALD Al2O3 films after annealing at different temperatures in N2 for 10 min. Data is given for p- and n-type Si. The data at 200 oC concern as-deposited films (the deposition temperature was 200 oC for all films) .
Note 2018: In PERC, a stack of Al2O3/a-SiNx:H is used and this stack allows for thinner Al2O3 films. The thickness of the Al2O3 in PERC is 4-10 nm.
7) – Are the methods for the deposition of Al2O3 scalable?
The deposition methods of PECVD [13,31] and sputtering [14,32] are certainly scalable and they are already implemented in c-Si solar cell manufacturing. The company Roth & Rau has adapted their microwave PECVD technique for Al2O3 deposition and good passivation results were reported . The competitive edge of this technology is that existing PECVD systems can quite easily be modified avoiding large investments in developing new technologies and/or reducing large capital expenditures. For sputtering the passivation results reported so far are not as good as for PECVD and ALD although they might be sufficient for commercial solar cell manufacturing.
Conventional ALD is unsuitable for high-throughput industrial solar cell production. Throughput can however be increased by going to batch processing in which multiple (100+) wafers are coated at once in a single reactor chamber. This route is pursued by the companies Beneq [33,34] and ASM  Another approach is undertaken by two Dutch companies. Both Levitech [36-38] and SolayTec [39-41] have developed spatial-ALD equipment in which the ALD cycles are not carried out in the time domain but in the spatial domain. This should allow high throughput processing of more than 3,000 wafers per hour per tool.
Comparison of c-Si passivation results for spatial-ALD, PECVD and sputtering . ALD typically yields the best passivation performance although PECVD comes very close [8,43].
Note 2018: In 2011, Roth & Rau was acquired by Meyer Burger and this is the current name of the company. In the last few years, a lot has happened in the field of Al2O3 deposition and the companies providing the tools. See the follow up blog.
8) – Spatial-ALD for high volume manufacturing, what are the benefits?
The two most important benefits of spatial-ALD are that it allows for inline atmospheric ALD processing and that the cycles are not carried out in the time domain but in the spatial domain. The latter means that the precursor and reactant injection takes places in different compartments or zones in which the gas phase species are confined. These zones are separated by inert gas barriers created by purge zones in between. To have the substrate being exposed to the different zones alternately, the substrate surface is translated through the different zones. This translation can be linearly by moving the substrate through many repeated zones (approach pursued by Levitech [36-38]) or it can be periodically by moving the substrates relative to a deposition head hence-and-forth (approach pursued by SolayTec [39-41,44]). Other benefits for inline spatial ALD are the fact that single-side deposition can easily be achieved, the absence of moving parts (apart from the wafers), and the fact that no deposition takes place at the reactor walls. Also the use of precursors is efficient.
The spatial ALD system “Levitrack” of Levitech for inline processing of solar cell wafers at atmospheric pressure [36-38]. The wafers are propelled at the track inlet and they “float” on bearings of gas created by the gases injected: Al(CH3)3 precursor, N2 purge, H2O reactant, and N2 purge etc. The position of the wafers is self-stabilizing in the middle of the track and also the distance between adjacent wafers of a few centimeters is self-regulating. In the current configuration the system yields ~1 nm Al2O3 per 1 m system length.
9) – What about the production costs per wafer for Al2O3 passivation layers?
This question is difficult to answer at this moment. Some equipment manufacturers of Al2O3 deposition systems report a few cents per wafer. However, the implementation of for example rear-surface passivation schemes has major consequences for the total process flow of the solar cell manufacturing and the cost-of-ownership will therefore depend largely depend on the details of the rear-surface passivation scheme chosen. Also the integration of Al2O3 with other materials and processing steps is a major challenge which is currently addressed by the PV industry.
One important finding so far is the fact that the passivation of solar cells by Al2O3 does not require semiconductor grade purity of the Al(CH3)3 precursor. It was found that the passivation performance obtained by solar grade Al(CH3)3 is also excellent . This is only one of the important cost-related aspects that need to be considered. Another interesting observation was that a very good passivation performance can also achieved by other, somewhat less pyroforic precursors than Al(CH3)3, for example ALD of Al2O3 from Al(CH3)2(OiPr) and O2 plasma revealed also a very good passivation performance .
Effective lifetime for plasma-assisted and thermal ALD Al2O3 films deposited from semiconductor and solar grade Al(CH3)3 . The corresponding Seff,max values are as low as = 1-2 cm/s for injection levels of 1014-1015 cm-3. From this figure it can be concluded that there is no need to use very expensive precursors to reach excellent levels of surface passivation
Note 2018: Clearly, the use of Al2O3 nanolayers for passivation pays off. The use Al(CH3)3 as precursor is a very significant cost factor so an optimized and efficient precursor usage is key.
10) – What are the overall prospects for the use of Al2O3 in PV?
The question is probably not whether Al2O3 will be used in commercial solar cells but when Al2O3 will be applied. The question is also in which type of solar cells the Al2O3 will be applied. It might not only be in high-end, high-efficiency, monocrystalline Si solar cells. Al2O3 thin films might also be interesting for more mainstream solar cell production. It can therefore be concluded that the overall prospects are very bright.
Note 2018: Al2O3 nanolayers have been enabling the PERC technology which appeared on the market around 2014. This year the global cell factories’ output could reach close to 50%.
- R. Hezel et al., J. Electrochem. Soc 136 518-523 (1989)
- G. Agostinelli et al., Sol. Energy Mater. Sol. Cells 90 3438-3443 (2006)
- B. Hoex et al., Appl. Phys. Lett. 89 042112 (2006)
- S.M. George et al., Chem. Rev. 110 111-131 (2010)
- B. Hoex et al., Appl. Phys. Lett. 91 112107 (2007)
- J. Schmidt et al., Prog. Photovoltaics Res. Appl. 16 461-466 (2008)
- J. Benick et al., Appl. Phys. Lett. 92 253504 (2008)
- G. Dingemans et al., Electrochem. Solid-State Lett. 13 H76-H79 (2010)
- G. Dingemans et al., Appl. Phys. Lett. 97 152106 (2010)
- G. Dingemans and W. M. M. Kessels, 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- G. Dingemans et al., Electrochem. Solid-State Lett. 14 H1-H4 (2011)
- S. Miyajima et al., Appl. Phys. Express 3 012301 (2010)
- P. Saint-Cast et al., IEEE Electron Device Lett. 31 695-697 (2010)
- T.-T. Li et al., Phys. Status Solidi RRL 3 160-162 (2009)
- P. Vitanov et al., Thin Solid Films 517 6327-6330 (2009)
- H.-Q. Xiao et al., Chin. Phys. Lett. 26 088102 (2009)
- D.H. Levy et al., J. Disp. Technol. 5 484-494 (2009)
- B. Hoex et al., J. Appl. Phys. 104 113703 (2008)
- N.M. Terlinden et al., Appl. Phys. Lett. 96 112101 (2010)
- G. Dingemans et al., Phys. Status Solidi RRL 5 22-24 (2011)
- Sun & Wind Energy, November (2010)
- Photon International, March (2011)
- J. Benick et al., 35th IEEE Photovoltaic Specialists Conference, Honolulu (2010)
- W.C. Sun et al., Electrochem. Solid-State Lett. 12 H388-H391 (2009)
- I.G. Romijn et al., 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- J. Ebser et al., 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- G. Dingemans et al., Phys. Status Solidi RRL 4 10-12 (2010)
- G. Dingemans et al., J. Appl. Phys. 106 114907 (2009)
- J. Benick et al., Phys. Status Solidi RRL 3 233-235 (2009)
- J. Schmidt et al., Phys. Status Solidi RRL 3 287-289 (2009)
- Roth&Rau, http://www.roth-rau.de
- J. Liu et al., 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- J.I. Skarp, 218th Electrochemical Society Meeting, Las Vegas (2010)
- Beneq, http://www.beneq.com
- ASM, http://www.asm.com
- E.H.A. Granneman et al., 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- V.I. Kuznetsov et al., 218th Electrochemical Society Meeting, Las Vegas (2010)
- Levitech, http://www.levitech.nl
- B. Vermang et al., Prog. Photovoltaics Res. Appl. (2011)
- P. Poodt et al., Adv. Mater. 22 3564-3567 (2010)
- SoLayTec, http://solaytec.org
- J. Schmidt et al., 25th European Photovoltaic Solar Energy Conference and Exhibition, Valencia (2010)
- P. Saint-Cast et al., Appl. Phys. Lett. 95 151502 (2009)
- P. Poodt et al., Phys. Status Solidi RRL 5 165-167 (2011)
* Conference paper by W.M.M. Kessels, J.A. van Delft, G. Dingemans, and M.M. Mandoc