Today’s conventional triple-junction cells seem to offer little room for improvement, with average production efficiencies hovering just below 30 percent. This can be increased by ten percentage points with a switch to six junction cells incorporating two dilute nitride layers, say Simone Missirian, Jeff Allen, Vijit Sabnis and Homan Yuen from Solar Junction.

Growth of the terrestrial solar industry hinges on reductions in energy costs. One way to do this is to manufacture cells that are more efficient than their predecessors, but cost no more to make. This can either boost the power produced by a solar power system of a given price – or allow a smaller, cheaper-to-build system to generate the equivalent power of its forerunners.

At Solar Junction, headquartered in San Jose, California, we have increased the efficiency of triple-junction cells for concentrating photovoltaic (CPV) systems, which are designed to track the position of the sun across the sky and focus sunlight on cells with mirrors or lenses. Our cells feature high-quality InGaAsNSb junctions, and can deliver a record-breaking efficiency of 43.5 percent at 925 suns (figures that have been verified by the National Renewable Energy Laboratory). This new benchmark for photovoltaic performance is the result of our novel materials technology, which we call A-SLAM (adjustable spectrum lattice matched). It will achieve far more than just netting an efficiency record – it is already the foundation for a flexible multi-generational roadmap to reach 50 percent conversion efficiency and beyond.

The core technology that we have developed can also bring a new level of efficiency to solar cells deployed on commercial and military satellites. We are adapting our technology to address these markets, which place a premium on high efficiency and high reliability.

Today’s satellite market

The vast majority of satellites launched today are powered by triple-junction solar cells with average efficiencies of just below 30 percent. These feature three junctions, connected in series, with progressively narrower bandgaps from the top to the bottom junction. This arrangement – traditionally a germanium bottom junction, and middle and top cells made from InGaAs and InGaP, respectively (see Figure 1) – divides the solar spectrum, making good use of the available energy by increasing the voltage delivered by the solar cell, and significantly improving the efficiency when compared to a single junction solar cell.
Figure 1. Conventional triple junction cells (left) have a germanium bottom junction, an InGaAs middle junction and an InGaP top junction. Solar Junction’s triple-junction cells (right) replace the germanium cell with one built from GaInNAsSb

Increasing the efficiency of solar cells for space applications is quite beneficial, as it enables higher power production and ultimately allows the satellite manufacturer to make a number of favourable economic trade-offs.  For instance, higher efficiency solar cells allow the use of a smaller, lower-mass solar array to meet a given power requirement.  The cost to launch a satellite is of the order of $10,000 per pound, so this reduction in solar array mass leads to a significantly lower launch cost. Alternatively, higher efficiency cells can generate more power and enable the satellite to carry a higher-value payload, such as a greater number of transponders. This leads to a higher revenue stream for the satellite operator and improved return-on-investement for the asset. 

Unfortunately, the traditional architecture of space cells (the combination of InGaP, GaAs and germanium) has reached technological maturity, and during the last few years conversion efficiencies have stagnated at the 28.5-29.5 percent level. A new materials system is required to meet the ever-increasing power requirements of next generation satellites and resume the historical rate of progress – an annual increase in absolute efficiency of about 1 percent.

Dilute nitrides

The superior efficiencies of our cells stem from the switch from a germanium junction to one built from InGaAsNSb (see Figure 1). This increases the bandgap of the low energy cell from 0.67 eV to 1.0 eV and leads to a greater energy harvest of incoming photons. Higher efficiency results from an increase in output voltage while maintaining adequate current through the junction.  

Building a bottom junction from a dilute nitride subcell is not a new idea. Throughout the previous decade numerous research institutes and commercial entities tried to do just this, but they were never able to grow dilute nitride junctions with sufficient material quality for high-performance photovoltaics. These failures drove the industry to abandon dilute nitrides in favour of the inverted metamorphic (IMM) approach.

The IMM has showed promise in terms of efficiency improvements. However, that is only one part of what makes a technology attractive for space. Unlike CPV, lower system costs are achieved with large area cells, and the cells must also withstand exposure to ionizing radiation without suffering catastrophic loss in performance. What the space market does share with the CPV industry is sensitivity to cost and low tolerance for technology risk. This presents several challenges for IMM technology; it is tricky to produce, involving the growth of epitaxial layers in reverse order, incorporation of thick strain-relief layers, and epitaxial lift-off and wafer bonding to realize the 1eV subcell.

This complex fabrication adds significantly to manufacturing costs, and the space industry is concerned over use of lattice-mismatched materials that may exhibit accelerated degradation under the extreme thermal stress environments in space. Solar arrays on commercial satellites can experience temperatures below –170 °C and as high as 100 °C for thousands of cycles. What’s more, there are complications associated with integrating the cell into industry standard solar arrays, and this creates a large and potentially unacceptable risk for the space power market.

We have succeeded where so many others have failed, thanks to a breakthrough in materials technology. This has resulted from exploiting the advantages of MBE, combined with proprietary hardware modifications and process IP. Our efforts at developing processes to grow high-quality InGaAsNSb have borne much fruit – this dilute nitride is truly a remarkable material. When properly grown, it possesses a continuously variable bandgap ranging from 0.8 eV to 1.4 eV while remaining lattice matched to the materials above and below. Lattice matching is an incredibly important attribute for cells targeting the space environment. This industry is very conservative, with long duration on-orbit performance of deployed cells used as the ultimate assessment for reliability. As of today, all radiation-hard multi-junction solar cells that power spacecraft are lattice matched.

The variable bandgap of InGaAsNSb enables the design of a device with up to six junctions, including two from different compositions of dilute nitride. Extending the number of subcells increases the device’s overall efficiency, and has enabled us to put together a multi-generational technology roadmap to higher and higher efficiencies for cells designed for use in space (the AM0 spectrum).



Figure 2. Solar Junction’s wafers processed with CPV cells have an area of 0.7 cm2, far less than that of space cells, which are 26 cm2 in size. The CPV wafer has approximately 100 cells while the space configuration has only two cells per wafer (or in some cases, one cell per wafer)

Manufacturers of solar arrays for satellites use large area cells that are 26 cm2to 30 cm2in size, which are shaped in a manner that maximizes wafer utilization while maintaining the highest cell efficiency (see Figure 2). These cells have a relatively large active area, and in order to combine acceptable yields with competitive costs, manufacturers must use very well controlled production techniques that deliver incredibly low defect densities.  If there is one single catastrophic defect present on the wafer, due to either epitaxial growth or subsequent processing, the space format will suffer a 50 or 100 percent yield loss, compared to a hit of just 1 percent for the smaller area cell built for concentrating photovoltaic systems.



Figure 3. Biasing the top junction so that it produces electroluminescence enables assessments of the uniformity and material quality of this junction

High quality epitaxial growth is essential for the production of uniform, consistent multi-junction cells. MBE has many strengths when compared to MOCVD, including the ability to produce better layer uniformity, fewer defects, and more abrupt interfaces with lower impurity levels. Biasing one of our 24 cm2space cells so that it produces electroluminescence from the top junction highlights the qualities of MBE (see Figure 3). This layer produces highly uniform emission, and there are very few dark spots, which are indicative of shunts in the epitaxial layers. Our cells have a conversion efficiency of 29-30 percent (AM0 spectrum), and their voltage output is higher than that of the incumbents’ present-day production cells, thanks to the higher voltage of the dilute nitride sub-cell (see Figure 4). This plot also highlights opportunities for further cell improvement, such as optimisation of the fill factor, which for the cells produced to date were impaired by a sub-optimum top grid metal design.



Figure 4. Current-voltage sweeps for two Solar Junction 24 cm2 cells fabricated on the same wafer reveal the improvement in voltage afforded by the dilute nitride subcell. This pair of novel triple-junction cells have an open circuit voltage of 2.85 V, about 200 mV higher than that of present day space cells. The current-voltage plots for the two cells are virtually indistinguishable, illustrating the high degree of intra-wafer uniformity achieved in the epitaxial growth and cell processing. Electroluminescence images also reveal the high degree of uniformity within the cells, and from cell to cell

To be considered for use on spacecraft, solar cells must demonstrate reliable, high efficiency operation in a space environment. This means that multi-junction stacks must be able to withstand constant exposure to radiation over the 15-year operational life of the satellite.  Device degradation in this environment largely results from bombardment by ionizing radiation, causing dislocations in the lattice structure. Designers of solar arrays account for this degradation by using thicker radiation shields and adding more cells to the array, increasing both weight and cost of the satellite. This is required to ensure the array can deliver enough power to operate the payload and bus subsystems through the entire operating life of the satellite.  What these designers are looking for are solar cells that not only exhibit a high initial efficiency, but also show a minimal and predictable shift in performance with accumulated radiation exposure. In addition, they want to employ devices that still deliver high power at the end of 15 years on orbit. Any solar cell technology that exhibits rapid performance loss when exposed to a high radiation environment is simply unacceptable.

Extensive testing reveals that our cells exhibit excellent tolerance to high radiation exposure. They appear to perform as well as those of commercially available triple-junction solar cells from Boeing-Spectrolab and Emcore Corporation under bombardment from doses of radiation commonly used to evaluate solar cells destined for use on spacecraft (see Figure 5). This exposure exceeds the damage expected from 15 years of operation in a geostationary orbit. These results demonstrate that the radiation performance of our dilute-nitride-containing solar cells is more than sufficient for long-term operation in the harsh environment of space. In addition, they reveal that at the lower fluence, our device retains more of its initial power than that of the incumbent’s solar cells.

Figure 5. Results for cells from Solar Junction (SJAM01), Boeing-Spectrolab (XTJ) and Emcore Corporation (ZTJ) under exposure to proton radiation at 1 MeV energy and several fluences, the highest being 3x1011p+/cm2(note that the data for Emcore and Boeing Spectrolab is taken from published radiation performance for commercially available triple-junction solar cells). Exposure to the cells is more than the equivalent damage expected from 15 years of operation in a geostationary orbit.

The normalized remaining power factor for an early development-stage, dilute-nitride-containing solar cell that is not fully optimized for radiation exposure is measured at 76 percent after an exposure of 3x1011cm-2, and is equivalent to the best available technology in the market

We are now pursing several low risk pathways to optimize the radiation performance of our cell. Our expectation is that the normalized remaining power factor will quickly improve to in excess of 85 percent when exposed to the equivalent fluence of a 15-year geostationary orbit mission. Hitting this goal will make our technology far superior to any other high efficiency approach.

Pathways to higher efficiency 

It has taken us little more than a year to produce cells that deliver an equivalent performance to the products of leading triple-junction suppliers. This rapid, tremendous progress stems from the incorporation of a high quality 1eV subcell, which we expect to soon enable an average production efficiency of 32 percent (AM0) for the InGaP/GaAs/InGaAsNSb device.  Although impressive, this superior triple junction is just the beginning of what is possible with our dilute nitride technology. We have developed a roadmap that will enable sustained performance improvement by incorporating additional subcells into the lattice-matched architecture. This will ultimately lead to a six-junction configuration with anticipated AM0 efficiency of over 40 percent. 

All of the steps required to go from a triple-junction cell to a six-junction variant are relatively simple and straightforward (see Figure 6 for an overview). They begin by building the existing design on a germanium substrate with an active 0.67eV subcell. This active junction, which will be similar to those presently used in commercially available triple-junction products, will increase the voltage produced by the solar cell and propel efficiency to the 35-36 percent range. One tremendously attractive feature of this particular design is that it is a drop-in replacement (equivalent form, fit and function) for today’s 29 percent efficient triple-junction cells. This will be our ‘go to market’ product for the space segment.



Figure 6. Solar Junction has a roadmap that promises to propel cell efficiency to 40 percent through the addition of extra junctions

Our five-junction design involves the addition of aluminium to the InGaAs subcell to shift the bandgap slightly, plus a switch from one dilute nitride junction to two – the 0.99 eV junction used in the four-junction cell is replaced by two with energy gaps of 0.93 eV and 1.18 eV. Forming these should be very straightforward, because today we regularly grow photovoltaic-quality InGaAsNSb materials with bandgaps from 0.88 eV through 1.4 eV.

The final node of this roadmap focuses on the addition of a higher bandgap top cell, alongside adjustments to the bandgap of several others to optimise overall performance. This enables proper allocation of the spectrum throughout the device. Modifications include the incorporation of aluminium in the top two junctions to enable better harvesting of the high-energy portion of the solar spectrum. This should enable cells for space to hit 40 percent, an efficiency milestone for the photovoltaic industry that will equip satellites with more power and allow them to carry out more tasks than they do today.