Solar, Explained
No quotes, no installer referrals, no advertising. Rules and rates vary by state and utility.

Multijunction Solar Cells: Stacked Junctions Explained

Multijunction cells stack junctions with different bandgaps so each absorbs part of the spectrum. How current matching, III-V materials and tandems work.

A single silicon junction converts light into current across one bandgap, and that one bandgap sets a ceiling on how much of the solar spectrum can become electricity. A multijunction solar cell raises that ceiling by stacking two or more junctions, each with a different bandgap, so that photons a lower junction cannot use are absorbed and converted by a higher one. The idea is old and the physics is settled; what changes between designs is the material system, the number of junctions, and how the layers are joined.

The single-junction ceiling and why stacking raises it

A photon with less energy than the bandgap passes through without creating an electron-hole pair. A photon with more energy than the bandgap creates a pair, but the excess above the bandgap is lost as heat as the carrier relaxes to the band edge. Those two losses bracket the single-junction problem: too little energy is not absorbed, too much is absorbed and then wasted.

The fraction of the spectrum that falls into each loss depends on the bandgap. A lower bandgap absorbs more photons but wastes more energy per photon; a higher bandgap wastes less per photon but absorbs fewer. A single junction is a compromise between the two, and the compromise is what limits conversion efficiency.

Stacking junctions removes the compromise. A high-bandgap junction on top absorbs the short-wavelength photons and lets the rest through. A lower-bandgap junction underneath absorbs the longer wavelengths that pass. Each junction operates near its own bandgap, so the energy lost per absorbed photon is smaller and the range of the spectrum that is used is wider.

How the junctions are arranged in a multijunction cell

The junctions are connected in series, one on top of another, usually grown as epitaxial layers. The top junction faces the light and has the largest bandgap; each junction below has a smaller bandgap than the one above it. Between junctions sit tunnel junctions, which are heavily doped and thin, and which allow current to pass between the subcells without a significant voltage drop.

Because the subcells are in series, the same current flows through all of them. That is the central design constraint. If one junction produces less current than the others, it limits the whole cell, and the excess current the other junctions could have produced is not collected. The subcells must be current-matched: the thicknesses and bandgaps are chosen so that, under the intended spectrum, each junction generates the same current.

The intended spectrum matters. A design current-matched for the air mass zero spectrum above the atmosphere is not matched for the air mass 1.5 spectrum at the surface, because the atmosphere removes some wavelengths and changes the balance between junctions. A cell optimised for space is not the same cell as one optimised for a terrestrial concentrator.

III-V materials and the lattice-matching problem

The junctions in most high-efficiency multijunction cells are made from III-V compound semiconductors: alloys of elements from groups III and V of the periodic table, such as gallium indium phosphide, gallium arsenide and germanium. These materials can be grown with the bandgaps needed for a stack, and they can be grown epitaxially on a substrate.

Growing one crystal on another requires the atomic spacing, the lattice constant, to be close. A large mismatch introduces defects that act as recombination centres and degrade the junction. The classic three-junction cell is lattice-matched: a gallium indium phosphide top junction, a gallium arsenide middle junction and a germanium bottom junction, grown on a germanium substrate. The lattice constants are close enough that the stack can be grown with low defect density.

Metamorphic designs deliberately grade the lattice constant between junctions, which allows a wider choice of bandgaps at the cost of a more complex growth process. Wafer bonding joins subcells that cannot be grown monolithically, which opens bandgap combinations that lattice matching would not allow.

Current matching under a real spectrum

Current matching is not a fixed property of the cell. It is a relationship between the cell and the light falling on it. Change the spectrum and the current balance between junctions changes.

A hazy day shifts the spectrum toward longer wavelengths, because scattering removes more of the blue end. A cell current-matched for a clear spectrum may find its bottom junction producing more current and its top junction less, and the series connection then limits the cell to the smaller current. The same happens with the angle of the sun, with the air mass, and with the altitude. In space, the spectrum outside the atmosphere is different again, and the cell is designed for that spectrum rather than for the surface.

This is why multijunction cells are rated under a stated spectrum and why a rating cannot be transferred between environments without adjustment. It is also why concentrator systems, which use optics to focus light onto a small cell, must preserve the spectrum they concentrate: a concentrator that filters or shifts the spectrum changes the current balance.

Space and concentrator applications

Multijunction cells are used where the cost of the cell is small compared with the cost of the area or the mass it occupies. A satellite has a fixed launch mass and a fixed array area, so the value of a higher efficiency is high. A concentrator system uses lenses or mirrors to focus light onto a small cell, so the cell area is small and the cell efficiency matters more than the cell area cost.

The trade-off is that multijunction cells are more complex to manufacture than silicon cells, and the manufacturing complexity is what limits their use in flat-plate terrestrial modules. The cell itself may be more efficient, but the process, the substrate and the current matching all add constraints that a silicon module does not have.

Perovskite-silicon tandems

A tandem cell is a multijunction cell with two junctions. The perovskite-silicon tandem pairs a perovskite top junction with a silicon bottom junction. The perovskite has a wider bandgap than silicon and absorbs the short-wavelength photons, while the silicon absorbs the longer wavelengths that pass through.

The attraction is that silicon is the established terrestrial cell material and perovskite can be deposited on top of it, so the tandem adds a junction without replacing the bottom one. The constraints are the same as for any series-connected tandem: the two junctions must be current-matched, and the perovskite must be stable enough to survive the service life of the module. The current matching is affected by the perovskite bandgap, which can be tuned by composition, and by the silicon bottom cell.

A perovskite-silicon tandem is not a different kind of physics from a III-V multijunction cell. It is the same series-connected, current-matched stack, with different materials and a different manufacturing route.

Where the current-matching rule does not apply

A multijunction cell can be connected with independent terminals rather than in series, which removes the current-matching constraint but adds wiring and power electronics. The table below sets out when the series rule applies and when it does not.

SituationSeries current matching appliesWhy
Monolithic two-terminal stackYesOne current path through all junctions
Three-terminal or four-terminal tandemNoJunctions are contacted separately
Cell measured under a spectrum it was not designed forYes, and the match is wrongThe current balance shifts with the spectrum
Single-junction cellNot applicableOnly one junction
Concentrator cell with a filtered spectrumYes, and the filter changes the balanceThe optics alter the spectrum reaching the cell

What changes when the light changes

The behaviour of a multijunction cell under a changed spectrum follows from the series connection. If the top junction receives less short-wavelength light, its current falls and the cell current falls with it, even if the bottom junction is receiving more long-wavelength light than before. If the top junction receives more, the bottom junction may become the limit.

Temperature affects the junctions differently. The voltage of each junction falls with temperature, and the temperature coefficients of the junctions are not identical, so the operating voltage of a multijunction cell under a given spectrum changes with temperature in a way that is not simply the sum of the single-junction coefficients. The current is less temperature-sensitive than the voltage, as in a single junction, but the current balance between junctions can shift slightly with temperature because the bandgaps change with temperature at different rates.

The practical consequence is that a multijunction cell’s rated efficiency is tied to a stated spectrum and temperature, and the rating is not a prediction of what the cell will do under a different spectrum or at a different temperature. That is true of any cell, but the series connection makes the spectrum dependence stronger in a multijunction cell than in a single junction.

Reading a multijunction datasheet

The datasheet states the spectrum and the temperature at which the efficiency was measured, and for a multijunction cell those conditions are part of the specification rather than a formality. A cell measured under the air mass 1.5 spectrum is not directly comparable with one measured under air mass zero. A concentrator cell is rated at a stated concentration ratio, and the efficiency at that ratio is not the efficiency at one sun.

The number of junctions and the bandgaps are usually given, and the current-matching condition is implied by the design. The temperature coefficients are given per junction or for the cell as a whole, and the measurement conditions for those coefficients matter as much as the coefficients themselves.

A multijunction cell is a stack of junctions in series, and the series connection is what makes the spectrum and the current balance central to its behaviour. The physics is the same whether the materials are III-V compounds grown on germanium or a perovskite deposited on silicon; what differs is the manufacturing route and the bandgaps that route makes available.

Common questions

What is a multijunction solar cell?

A multijunction solar cell is a cell with two or more junctions stacked on top of each other, each with a different bandgap. The top junction absorbs the short-wavelength photons and the lower junctions absorb the longer wavelengths that pass through, so more of the spectrum is converted.

Why are multijunction cells more efficient than silicon cells?

A single junction loses the energy of photons above its bandgap as heat and does not absorb photons below it. Stacking junctions with different bandgaps reduces both losses, because each junction converts the part of the spectrum near its own bandgap.

What does current matching mean in a multijunction cell?

The junctions are connected in series, so the same current flows through all of them. The cell is designed so that each junction produces the same current under the intended spectrum. If the spectrum changes, the balance can shift and the junction with the lowest current limits the whole cell.

What are III-V solar cells?

III-V solar cells use compound semiconductors made from elements in groups III and V of the periodic table, such as gallium indium phosphide, gallium arsenide and germanium. These materials can be grown with the bandgaps and lattice constants needed for a multijunction stack.

What is a perovskite-silicon tandem cell?

A perovskite-silicon tandem is a two-junction cell with a perovskite top junction and a silicon bottom junction. The perovskite has a wider bandgap and absorbs short-wavelength light, while the silicon absorbs the longer wavelengths that pass through. The two junctions are in series and must be current-matched.

Read next