The 1984 solar cell debate that still shapes space arrays

What the work claims

John A. Scott-Monck of the Jet Propulsion Laboratory argues that the next decade of space photovoltaics will be defined by a competition between silicon and gallium arsenide (GaAs) solar cells, with silicon holding the advantages of flight heritage, lower cost, and lighter mass, while GaAs offers higher conversion efficiency and better radiation resistance. He further contends that concentrating arrays and cascade or thin-film cells could eventually change the terms of that competition, but only if manufacturing, substrate cost, and proton-radiation characterization problems are solved. The paper treats specific power (W/kg) and power per unit area (W/m²) as the two figures of merit that dominate array configuration.¹

How it works

The paper compares cells on three physical parameters that propagate through the entire power system: conversion efficiency at beginning of life, degradation under radiation by end of life, and mass per unit power. Silicon space cells in 1984 averaged about 11 percent efficiency at 28°C for a 250 μm thick cell, an improvement from the 300-350 μm, 10-12 percent cells of the 1960s. GaAs cells then offered 16-18 percent efficiency, but were 300-350 μm thick and roughly 2.3 times as dense as silicon, with substrate costs between $2 and $5 per cm² depending on quantity and quality. The mass and cost penalties meant that GaAs could not simply replace silicon cell-for-cell.¹

Scott-Monck identifies two routes to make GaAs competitive. The first is to reduce substrate material. The CLEFT peeled-film technique demonstrated greater than 15 percent efficiency in cells only 5 μm thick, while allowing the substrate to be reused. A second route used a silicon substrate with a germanium buffer layer grown by organo-metallic chemical vapor deposition, so the cell weight would be set by the silicon substrate thickness rather than a thick GaAs wafer. For panel-level cost parity, the analysis concluded that GaAs substrate cost would need to fall well below $1 per cm² even with a 50 percent substrate recovery yield.¹

Concentrator arrays offered a different escape from the GaAs cost-mass trap. Because concentration raises the effective irradiance on each cell, it reduces the cell area needed, and the higher operating temperature tolerance of GaAs mitigates the thermal penalty. The paper notes that concentrator designs also sidestep the unresolved proton-radiation equivalence question for GaAs, because shielding is easier to provide in a concentrator system. Cascade cells and thin-film options such as amorphous silicon and copper-indium diselenide were discussed as longer-term paths to efficiencies above 25 percent or to high radiation tolerance at low mass.

Where a skeptic should push

The paper is an engineering assessment, not a controlled experiment, and several of its judgments are explicitly conditional. The author notes that the influence of the space proton environment on GaAs "has not been completely resolved," and that the rapidly evolving cell structure meant 1 MeV electron data could not yet be extrapolated to predict proton behavior. That uncertainty directly affects end-of-life power calculations, which the paper calls the dominant factor in array design.¹

The cost figures are also tied to a specific manufacturing moment. The $2-5/cm² GaAs substrate cost and the $1/cm² target were 1984 economics; modern epitaxial growth, larger wafers, and terrestrial LED and photonics markets have changed both numbers. A skeptic should also ask whether the CLEFT and silicon-substrate transfer techniques, which looked attractive in the laboratory, could scale to the yields and areas needed for multi-kilowatt arrays. The paper itself flags these as works in progress rather than mature processes.

What this means for solar power in space

The non-obvious implication is that the paper's central trade-off has persisted because it is rooted in physics, not in a temporary market condition. Silicon remains cheaper and lighter per watt at the cell level; III-V and perovskite alternatives still compete on efficiency and radiation tolerance but struggle to match silicon on cost and scalable flight heritage. Modern programs such as roll-out solar arrays and flexible substrate blankets are still optimizing the same three variables the paper named: efficiency, end-of-life degradation, and specific power.¹

The genuine threat is that the community keeps rediscovering the same cost-mass floor. Every new high-efficiency cell technology faces the same questions GaAs faced in 1984: can it be made thin enough, cheap enough, and radiation-hard enough to displace silicon across the full range of missions? Without a clear answer, missions continue to default to silicon, which starves the flight-data feedback loop that would mature alternatives.

The opportunity lies in missions where the operating environment changes the value of each parameter. Deep-space missions far from the Sun, high-radiation belts, and compact stowage-constrained spacecraft all shift the balance toward the technologies the paper flagged: thin-film transfer, concentrators, and cascade cells. A mission that explicitly optimizes for end-of-life power per unit stowed volume, rather than beginning-of-life cost, could finally justify the manufacturing investments that silicon alternatives have needed for four decades.

The bottom line

This 1984 assessment is a snapshot of a technology transition that is still underway. Its specific cost and efficiency numbers are dated, but its framing of the solar-array design problem is not. The paper correctly identifies that cell efficiency alone is insufficient; what matters is the system-level trade among efficiency, radiation degradation, mass, and cost. Its cautions about proton-radiation data for GaAs and about substrate economics remain relevant warnings for any new cell technology aiming to displace silicon in space.