GaN-based solar cells for integration to multi-junction photovoltaics could raise concentrated photovoltaic system efficiencies to more than 50%. Compound Semiconductor talks to the researchers designing the devices.

Concentrated photovoltaic systems: will next generation multi-junction cells take efficiencies over 50%?

Without a doubt, concentrated photovoltaics using multi-junction solar cells can reach efficiencies that leave other photovoltaic technologies floundering.

Today's typical three junction devices have achieved efficiencies as high as 44.4% under concentration while next generation four- and five-junction architectures are approaching 50% conversion efficiencies.

However, these leaps in efficiency come at a cost. MOCVD deposition of the III-V layers that make up a p-n junction is complex as manufacturers strive to minimise the lattice mismatches between neighbouring layers that induce strain, degrade materials and stymie device performance.

And crucially, the cost of a CPV system using III-V cells is yet to be truly competitive with industry-standard silicon-based photovoltaic systems. Factor in falling silicon prices and it's easy to see why several key CPV system makers have folded.

One clear way to cut the cost of a CPV system is to boost the efficiency of its solar cells by adding another junction. However, the constraints associated with lattice-matching makes the epitaxial growth of a fourth or fifth junction a real challenge.

As a result, some researchers aim to grow high-energy junctions separately, and then add these to an underlying multi-junction cell. One such group is based at the University of California, Santa Barbara, and has fabricated an InGaN/GaN multi-quantum well solar cell with a wide bandgap of 2.65 eV.

Serving as the 5th junction, the researchers intend to optically bond this 'sub-cell' to an underlying GaAs multi-junction solar cell. This move will side-step the lattice mismatch issues that prohibit InGaN growth on GaAs structures, and could catapult overall device efficiency to a breathtaking 50% or more.

 

Proposed cross-section of the device structure including contacts and optical coatings.

As Nathan Young from the Materials Department at UCSB explains, the InGaN/GaN solar sub-cell produces a relatively small current, and efficient current matching with the underlying junctions simply isn't possible. Consequently the bonding layer - a transparent and non-conducting polymer - is critical to isolate the this cell from the underlying multi-junction cell.

"There's no electrical connection [between the junctions], we're just keeping the device optically coupled," he says. "We basically place the InGaN/GaN sub-cell onto the multi-junction cell and bond it with a transparent polymer. We then take out an extra contact so we can link the structures electrically."

According to Young, while the additional contact introduces an extra step to device manufacture, depositing one more epitaxial, current matched junction simply isn't an option.

As he explains: "This inability to current match an InGaN cell with a 2.65eV bandgap is unavoidable... InGaN is the only material that can perform efficiently at this wide bandgap."

Broadband coatings

A crucial part of the research has been to add optical coatings - as is standard practice in today's multi-junction cells - to the InGaN/GaN cells. As Young highlights: "Before [coatings] were added to the cell, it just wasn't performing that well as it wasn't absorbing enough light."

In this case, the challenge was to develop coatings that exhibited anti-reflection properties across sufficiently wide wavelength ranges, and, for the first time, apply these to InGaN/GaN cells.

"We're trying to get as much light as possible into our device, without sacrificing any light that would not be absorbed by the InGaN multi-quantum well layer," says Young's colleague, Emmett Perl, from the Department of Electrical and Computer Engineering at UCSB.

To this end, the team designed a system of high performance broadband optical coatings - a front-side anti-reflective coating and back-side dichromic mirror - for the InGaN/GaN solar cell. These actually minimise front surface reflections across the broad spectral range while maximising rear surface reflection only in the spectral range absorbed by the InGaN. The dichromic mirror also allows for the InGaN/GaN cell to absorb additional light on a second pass.

So far, the results are good. Application of the coatings increased the peak external quantum efficiency of the InGaN/GaN cell by 56%, and conversion efficiency by 37.5%, relative to an uncoated structure.

The team is now looking to boost InGaN device efficiency. Young reckons the sub-cell will need to be at least 50% more efficient than current demonstrations, if integration to a five junction stack is to make sense.

Work is also underway to develop a final bonding process for integration. In other multi-junction architectures, the bonding layer has to allow electrical conduction and optical transmission, but as Perl highlights: "Our optical coatings will provide excellent optical transmission into the underlying four junction structure."

"Our electrically isolated terminal configuration also means electrical conduction across this bonding interface isn't necessary," he adds.

Perl is now optimising thermal conduction across the layer as well as its fabrication processes, while at the same time improving the quality of the broadband coatings. Currently investigating the use of anti-reflective nanostructures in the coatings, he believes this additional work will reduce reflections to near zero, for GaN, and is confident any increase in manufacturing costs will be offset by the gains in cell efficiency.

"With the right resources in place, I think an integrated device could be demonstrated in two to three years, with a 50% [efficient] device in the next five to seven years," says Young. "With time and effort, we believe the inherent scalability of this technology provides good commercial opportunities for CPV."