Surrey researchers price silicon satellite arrays 85 to 90 percent below triple-junction

The gap between the two cell prices is far wider than the saving on the finished array. Take the cheapest triple-junction figure in the review, $250 per watt, against the most expensive of the three silicon designs, heterojunction at $0.39 per watt ^[5][6]. The cell is roughly 640 times cheaper ^[1]. The array modelled by the Surrey team is about 6.7 to 10 times cheaper ^[2]. Everything that is not the cell absorbs the difference, and most of that is the space-qualified glass ^[9].

"The interesting finding for us was not that silicon is cheaper but where the remaining cost sits. Once you put silicon cells behind space-qualified glass, the glass is what you are paying for," said Tommy Richards, the Surrey PhD student who is first author of the review ^[10][22].

The other penalty is area. Silicon delivers about 28 percent less power per unit area at the start of a mission, so matching a triple-junction panel's output takes roughly 39 percent more panel ^[13][3]. The review scaled the arrays to equal output, added the cost of the extra structure, and reports silicon still came out several times cheaper on beginning-of-life performance ^[13].

Mass runs the other way: silicon heterojunction cells came in at about 920 to 1,000 watts per kilogram against 455 to 505 for the triple-junction option ^[12][2]. That is where the halved cell mass comes from.

Radiation decides who should care. For the same coverglass thickness, silicon is around 2.6 times less resistant than triple-junction ^[14], and the review expects triple-junction cells to hold more of their performance after five years in orbit ^[15]. Every cost number in the comparison is stated at beginning of life ^[1]. How much of the saving survives depends on how long the spacecraft is meant to work, and the review does not settle that.

Richards frames the fix as a shielding problem. "If we can make the cell itself tougher against radiation, we can use thinner glass or substrates, and we cut cost and weight at the same time," he said ^[11]. Record efficiencies give some room to work with: 27.8 percent for silicon heterostructure cells and 34.85 percent for perovskite/silicon tandems ^[4].

Nothing about volume depends on orbit. Objects launched per year rose from around 120 in 2010 to more than 2,800 in 2024, about a twenty-three-fold increase, and an average satellite needs close to a kilowatt ^[16][17][4]. Proposed orbital solar power stations would need structures kilometres across delivering gigawatts, and the review argues that demand would consume the entire triple-junction market while barely registering against silicon output ^[18]. "Silicon is the only material with factories already running at that size," Richards said ^[20].

Two caveats on provenance. The comparison is modelled on one face of a 3U CubeSat and on a Micro Sat format ^[8], and Surrey Satellite Technology Limited co-funds Richards's PhD and supplied the spacecraft data used in the study ^[21].