Efficiency Enhancement of Silicon Photovoltaic Solar Cells by Passivation

Increasing energy demands, exhaustion of easily accessible oil resources and fears of climate change make renewable energy sources a necessity. Although it is evident that future power generation will result from a wide mix of technologies, photovoltaic cells have made astounding technical and commercial progress in recent years. Over the last decade renewable energy generation has been stimulated by tax concessions and feed-in tariffs. Large scale manufacturing of photovoltaics has benefited from this and progress along the learning curve necessary to achieve economies of scale in manufacture has been very rapid. However like all renewable energy sources today the cost per kWh of electricity from photovoltaics is greater than that generated by fossil fuels, although the gap has reduced quite dramatically in the last two years. The cost reductions in generation from photovoltaics have been achieved through innovative cell design, the use of lower cost materials, advances in power management electronics and lower profit margins. At the moment, >85% of new installations use wafered silicon cells of multi-crystalline or single crystal material. In these cases a key issue has been developing technologies which use thinner slices (using less silicon for a given area of solar panel) and moving to "solar grade" silicon. This type of silicon is less pure than the electronic grade used for integrated circuits and is cast into multi-crystalline ingots but it is very much cheaper. This is an important issues because before these developments as much as 50% of the cost of a cell could be attributed to the silicon material. An important cost reduction per kWh delivered has been achieved in this way despite solar grade silicon producing cells of lower conversion efficiency than electronic grade material. Further substantial reductions in cost could be achieved by using silicon produced by less energy hungry metallurgical processes, for example starting the manufacturing process by the reduction of quartz with carbon and applying low energy purification processes. This type of silicon, known as upgraded metallurgical silicon, is even less pure containing compensated dopants and metals which can act as important recombination centres so reducing the efficiency further. The aim of this proposal is to develop methodologies which are able to bring the efficiency of cells made from these cheap forms of silicon close to the efficiencies achieved from the higher cost electronic grade material. This could increase the efficiency of multi-crystalline solar grade silicon by around 5% absolute and even more in the case of upgraded metallurgical silicon. Current silicon cell structures work well because hydrogen (usually from the silicon nitride antireflection layer) passivates surfaces and bulk defects. In electronic grade single crystal this reduces recombination to insignificant levels. It doesn't work as well in solar grade multi-crystalline silicon or upgraded metallurgical silicon because there are regions, sometimes entire crystal grains, which are not passivated by the hydrogen. However other regions are of very high quality often as good as electronic grade silicon. We associate the resistance to passivation with specific types of defect observed in lifetime maps of slices. In this project we plan to identify the defects which show resistance to hydrogen passivation by using electronic and chemical techniques (carrier lifetime, Laplace deep level transient spectroscopy, SIMS, Raman spectroscopy and defect modeling). The key part of the proposal is to use our knowledge of defect reactions in silicon to develop alternative passivation chemistries which can be applied, during slice or cell production, to those defect species resistant to hydrogen passivation. In this way we would expect to make a very important improvement to the efficiency of the dominant solar PV technology.

The overall aim is to increase the efficiency of cheap silicon solar cells. If this is successful to the extent described in the case (5% absolute) it will move silicon PV generation into grid parity for domestic users for most of Europe (ie the cost of PV generation by domestic users without subsidy will be less than the price paid for electricity delivered to their premises by the utility supplier). This would have widespread economic and social consequences. Below we put these into context.

The cumulative capacity of photovoltaics installed in Europe (EU) at the end of 2010 was 30GW which represents 75% of the world's PV capacity. New installations in 2010 constituted 13GW, much more than any other form of EU generation installations in 2010 with the exception of gas at 20GW (reliable 2012 figures are not yet available). 90% of the EU PV capacity is wafered silicon. Estimates for 2020 are an EU cumulative capacity of 200GW ("Global Market Outlook for PV", European PV Industry Association May 2011 http://www.epia.org/ ). To put this into context the cumulative capacity of all the UK's nuclear power stations is < 9GW. In total, the European PV market is estimated to be 39 billion euro with 68% of the value currently created by European manufacturing. From a world view, more than 53% of the value of the global PV supply chain is created in Europe. Germany is dominant with 17GW of installations and the largest share of the manufacturing, however a UK manufacturer of multi-crystalline silicon with a subsidiary in Germany is a major player in the market (PV Crystalox, Abingdon, UK http://www.pvcrystalox.com).

So what impact could our objective of an improvement of mc-Si solar cell efficiency by 5% have? Estimating the market share of mc-Si to be 75% and an improvement in efficiency from 12% to 17% if our new passivation technology could be applied to 50GW of PV produced between 2015 and 2020 a way of visualising the impact would be that it would eliminate the need for 10 nuclear stations the size of Sizewell B across Europe (load factors of nuclear plant are ~65% only slightly more than solar PV). This rather simplistic view ignores network stability (a generic problem with wind and solar micro-distributed generation). We have not factored in any increase in manufacturing cost because defect engineering of this type in silicon usually has an insignificant impact on cost. We have considered how such passivation might be applied during the cell production process. If we can use fluorine in the way envisaged the technique could be incorporated into existing technologies with only small modifications making the implementation time dependent more on commercial negotiations and cell testing than on plant changes. It might be possible to see adoption by 2018.

The increased efficiency would also decrease the energy pay-back time (EPBT). Depending on the location of the installation, the EPBT of wafered silicon technologies is between 0.6 and 1.4 years, the lower figure is that for solar grade fluidised bed material in southern Europe. Our work could reduce this to 0.4 years ... the proven operating life of wafered silicon cells is >30 years. It would have little effect on meeting the UK 2020 target of the 2008 Climate Change Act (34% reduction in greenhouse gases) but could be an important contributor to the 2050 target of 80% reduction.

What is quite certain, and evidenced by our track record, is that we will make a major contribution to the understanding of passivation mechanisms in silicon. This has academic impact described in the academic beneficiaries section but is also significant in several other commercial processes. In particular in extremely scaled CMOS the issue of degradation due to hydrogen loss at the gate is increasingly important. Substitution of hydrogen by deuterium or fluorine looks interesting but in the latter case lacks underlying science. Other cases are enumerated in Academic Benificaries.