Where Solar Efficiency Actually Stands

Headlines quote lab records. Roofs get something else. This is the size of the gap, and what causes each step down it.

Author
Samhita research
Figures as of
2026-08-21
Scope
Silicon and silicon tandems, one-sun sunlight
Confidence
Mixed — see below

Summary

A solar cell cannot convert all the sunlight that hits it. Physics sets a hard ceiling near 34 percent for a simple cell. Silicon, the material almost every panel is made from, has a lower ceiling of its own: about 29 percent.

The best silicon cell ever measured sits a little under that. The panel you can buy sits several points lower again. Most of that last drop is not physics. It is glass, wiring, and the empty space between cells.

Bottom line: silicon is close to finished. Meaningful gains now come from stacking a second material on top of it, not from improving silicon further.

The numbers

33.7%
Theoretical ceiling, any single-junction cell
Fixed by physics, not by engineering
27.3%
Best silicon cell measured in a lab
Within about 2 points of silicon's own limit
~23%
Good commercial panel you can order today
Roughly 4 points below the lab record

Figures as of 2026-08-21. Record cell efficiencies move; re-check them against the source before quoting this page.

Four ceilings, one after another

33.7 Physics limit any single cell 29.4 Silicon limit material bound 27.3 Best lab cell one small square ~23 Panel on sale whole module percent of sunlight converted to electricity
Each bar is a different kind of limit. The first two are set by nature; the last two are set by manufacturing.

Findings

Finding 1

The famous 33.7 percent limit is not the one that binds silicon.

The Shockley–Queisser calculation gives the best possible result for any single-material cell under normal sunlight. Silicon's own properties cut in earlier. Charge carriers inside silicon destroy each other in a process called Auger recombination, which caps a perfect silicon cell near 29.4 percent.

Confidence: high — both figures are calculations from established physics, not measurements that drift.

Finding 2

Silicon research has almost nowhere left to go.

The best laboratory silicon cell sits about two points below silicon's own ceiling. Twenty years of work bought the last few points. The remaining two will be slower and worth less.

Confidence: moderate — the record figure is current as of the date above and will be beaten, but only by small amounts.

Finding 3

The drop from lab cell to sold panel is mostly not physics.

A record cell is a small square measured under ideal light. A panel is dozens of cells behind glass, joined by wires, with gaps between them and a frame around the edge. Reflection off the glass, resistance in the wiring, and area that holds no cell at all account for most of the difference.

Confidence: high on the cause, moderate on the exact split — the breakdown varies by manufacturer.

Finding 4

Stacking is the only route past the single-junction ceiling.

Put a second material over silicon and each layer takes the part of the spectrum it handles best. Perovskite-on-silicon cells have already been measured above the 33.7 percent single-junction limit in the lab. They are not yet proven to last twenty-five years on a roof.

Confidence: provisional — laboratory records in this area move several times a year, and long-term durability data is still thin.

Supporting data

Figure Percent Kind of number
Shockley–Queisser limit, one sun33.7Calculated
Silicon Auger-limited ceiling29.4Calculated
Best laboratory silicon cell27.3Certified measurement
Best perovskite-on-silicon tandem cell34.6Certified measurement
Good commercial module23.0Product datasheets
Typical mainstream module21.0Product datasheets

What would change this

A tandem panel with a proven twenty-five year warranty would overturn the summary above. The ceiling stops mattering the moment two layers become ordinary.

Two limits of this report: it covers ordinary sunlight only, so concentrator systems and space cells are excluded, and it treats module efficiency as a single number when in practice it varies with temperature and light level. A panel rated 23 percent in a laboratory produces less on a hot afternoon.

Sources

  1. Shockley, W. and Queisser, H., Detailed Balance Limit of Efficiency of p-n Junction Solar Cells, 1961. Source of the 33.7 percent single-junction limit.
  2. Richter, A. et al., Reassessment of the Limiting Efficiency for Crystalline Silicon Solar Cells, 2013. Source of the 29.4 percent silicon ceiling.
  3. National Renewable Energy Laboratory, Best Research-Cell Efficiency Chart. Source of all certified record cell figures. This chart is updated continuously; verify before quoting.
  4. Manufacturer product datasheets. Source of the commercial module range.