PP1.3a Silicon Tandem Cells (monolithic)

Project Coordinator: 
UNSW
Chief Investigators: 
Martin Green
Associate Investigators: 
UNSW
Other Team Members: 
UNSW

The limiting efficiency for the single junction silicon wafer cells is 29%, with commercial devices expected to approach the world-best laboratory value of 25.6% by 2020. The main reason silicon’s limiting efficiency is less than the theoretical photovoltaic limit of 33% is unavoidable recombination of photo-generated carriers by Auger processes in the silicon. Stacking cells on silicon in a tandem structure brings the limiting performance even closer to the theoretical limits for such stacks, due to both silicon’s near ideal bandgap in this role and a reduction in the significance of Auger effects as generation is split between multiple devices. For three cells stacked on silicon, the limiting efficiency exceeds 50%, over 70% relatively higher than for the limit for a single cell. 

The large and vigorous silicon photovoltaic industry will continue to perfect the production and reduce the cost of high quality wafers, as well as other aspects of silicon cell processing. Combined with the large potential efficiency gains outlined above, this suggests that one possible evolutionary path for silicon wafer-based cells would be to use a silicon cell
as a substrate for the deposition of thin, high performance, wide-bandgap cells on its top surface, much the same way as an anti-reflection coating layer or a heterojunction emitter structure is deposited in present commercial cell sequences. 

The AUSIAPV work programs address silicon tandem cell technologies as monolithic devices (built on the same substrate) or with cells mechanically stacked.

Five strands under Program Package 1.3a aim, for the first time, to successfully mate the commercially dominant PV technology based on silicon solar cells with other promising PV materials, including the III-V semiconductors, the chalcogenides and perovskite technologies. 

Of the Group III-V materials of interest, only GaP offers a good lattice match to silicon, with the other III-V semiconductors as used in high performance III-V cells having about 4% mismatch, being better matched to Ge. One strategy of PP1.3a, inplemented in PP1.3a(i), is to take advantage of the miscibility between Si and Ge to grow a series of SixGe1-x buffer layers on Si, with x steadily decreasing. In this way, the lattice constant can be changed from that of Si to that of Ge after growth of a micron or more of buffer material. High quality III-V cells can then be grown on the Ge surface. 

A parallel approach PP1.3a(ii) with some success demonstrated in 2013 and 2014 is to take advantage of the similar miscibility of GaP and GaAs by growing a series of GaPxAs1-x buffer layers on Si with x again steadily decreasing, allowing a transition from the Si lattice constant to that of GaAs. An advantage in this case is that the material in the buffer layer has a much higher bandgap than silicon, allowing the silicon substrate to participate as an active cell in the stack

Working with US- and Australian-based collaborators, with additional project support leveraged beyond that able to be provided from SRI funding, the aim of an efficiency of over 24% with one or more of these approaches was achieved in the first 4 years of the Australian Centre operation. Options for reducing the thickness and cost of any buffer layers or otherwise reducing costs will be the target of the second phase of activities during years 5-8, with a targeted cell efficiency of 32%.

Two more adventurous approaches to building high quality tandem cells on silicon wafers are also being explored. PP1.3a(iii) targets an atomically abrupt Si/Ge transition, where the lattice mismatch is taken up in a single atomic
layer, which is thermodynamically feasible since it is a low energy configuration. UNSW has filed patent applications on approaches that have given promising results of this type. This would allow the Ge layer to be very thin, creating negligible absorption loss or, alternatively, thick enough to be used as an active cell in a novel “out-of-sequence tandem”.

The second of these more adventurous approaches (PP1.3a(iv)) is the investigation of silicon tandem cells using chalcogenides as the upper cells in the stack. Although the established chalcogenide cell materials (copper indium gallium selenide [CIGS] and CdTe) have shown high efficiency potential, they are not lattice-matched to silicon and have problems for longterm use arising from the use of toxic and/or scarce materials.

More promising for the long term are devices made from  materials based on the CZTS (CuZnSnS) system. Despite the relatively small effort so far devoted to the development of this material, solar cells using it have already demonstrated energy conversion efficiency above 12%. Moreover, the lattice constant of CZTS and that of related alloys are a close match to silicon and the CZTS bandgap, at circa 1.5 eV, is almost ideal for the lower cell in a two-cell stack on silicon. Moreover, alloying with related compounds to replace, for example, Zn or Sn by lighter elements, such as Fe(ii) or Si, will increase the bandgap making values such as the 1.7 eV required for a one-cell stack or the circa 2 eV required for the top cell in a two-cell stack on silicon also accessible in a highly compatible materials system.

The final strand of activity (PP1.3a(v)) involves investigation of materials systems that do not require lattice matching to silicon. Work in 2015 has achieved a very encouraging energy conversion efficiency of 23.4% when a perovskite solar cell is coupled with a 22.7% efficient silicon, passivated emitter, rear locally diffused solar cell by spectrum splitting.