Downstream: End-of-life and recycling performance

Silicon's downstream performance is, by a considerable margin, the weakest examined in this study. Chart 40 reveals that the collection rate remains low throughout the projection period, peaking at almost 18% in the early 2030s, with a collection rate of 17% at the end of the projection period. This reflects, in part, the physical difficulty of recovery: semiconductor chips are typically highly integrated with other components on circuit boards, making separation challenging without damaging the material, while only small quantities of silicon can realistically be reclaimed from individual products such as smartphones.¹⁵⁹ Uncollected and disposed of material remains by far the largest end-of-life destination across the entire period, growing substantially from 1.5 million tonnes in 2020 to 4.3 million in 2050 as end-of-life volumes increase.

The more revealing pattern lies in what happens to the material that is collected. By the end of the projection period, recycled volumes are equivalent in size to just 2% of downcycled volumes. This aligns closely with what is observed in practice: silicon in ferroalloys is recovered when steel and iron are recycled, but it is recovered as part of the steel stream rather than as silicon. The material remains in the economy, but its original function is lost entirely. It does not return to service as silicon and cannot subsequently be recovered for silicon applications. Similar logic applies to silicon contained in aluminium alloys, where functional recycling is not currently possible. Even where producers have an economic incentive to recover value from side-streams—as some metallurgical-grade silicon producers do—low material concentrations often make recovery uneconomic in practice, so little is ultimately cleaned and returned to the market as usable silicon.¹⁶⁰ Silicon's headline losses therefore obscure the underlying challenge: even the material that enters the recovery system is rarely reclaimed as silicon itself.

Chart 41 below shows that new scrap contributes nothing to secondary supply essentially across the entire projection period. This is consistent with the physical and economic realities of silicon processing: kerf powder from wafer slicing is finely dispersed and easily contaminated, purification residues are chemically transformed, and even where recovery is technically possible, it is often uneconomic compared with primary production.¹⁶¹ The limited secondary supply that does exist comes almost entirely from old scrap, and its absolute volume remains negligible relative to total realised supply throughout the projection period, just 70 thousand tonnes in 2050, compared to almost 12 million tonnes of primary supply.

Chart 42 quantifies the challenge described above. Both the End-of-Life Recycling Rate and the Recycling Input Rate remain effectively negligible throughout the projection period, with the former reaching just 1.4% by 2050, and the latter only 0.6% by 2050. This is well below the levels observed for any other material in this study. Silicon's circularity is therefore not simply underperforming relative to a stronger benchmark—under current trajectories, it is barely established.

However, the headline recycling indicators do not capture where silicon's most meaningful circularity opportunities actually lie. As the midstream section highlights, the strategically important silicon streams are small in mass but high in value. These streams support a range of circular strategies, from reuse through recycling to downcycling, each retaining a different share of silicon's original value and embodied energy.

Wafer reclaim—where used wafers are processed through layer removal and polishing for reuse—is already a functioning commercial activity and is expanding as new wafer production costs increase.¹⁶² Dedicated recovery of silicon from end-of-life photovoltaic panels is operating at limited scale but is nonetheless underway.¹⁶³ Silicon sludge and wafer waste from semiconductor fabrication can also be converted into engineered ceramics.¹⁶⁴

None of these pathways will substantially change silicon's aggregate Recycling Input Rate because the streams involved represent only a small fraction of total silicon demand. But that is precisely the point: for silicon, circularity measured by mass may offer limited insight, while circularity measured by strategic value and supply security may reveal a considerably more promising opportunity than the headline indicators suggest.