The site and the test
A CSP plant runs on the sunlight its mirrors reflect onto the receiver, so dust on the glass costs output directly. Chemitek developed Mirtek, an antistatic coating for CSP mirrors, with Atlantica and Rio Glass Services, and tested it on the parabolic-trough mirrors of the Solaben plants in Extremadura, Spain. The question was simple: after a wash, does coated glass lose reflectivity more slowly than uncoated glass?
Chemitek’s report covers two trials. In each, Mirtek went on in the wash water during a normal clean, and reflectivity was read at fixed mirror positions on set dates afterwards.
| Trial | Application | Coated mirrors | Uncoated mirrors |
|---|---|---|---|
| October 2021 | Jetwash | 4 positions, read on 8, 13, 18 and 25 Oct | 2 positions, read on 14, 18 and 25 Oct |
| July 2022 | Brush | 8 positions, read on 14, 19 and 25 Jul | 8 positions, read on the same dates |
The losses are in percentage points of reflectivity: an uncoated mirror that read 92.73% on 14 October and 90.29% on 25 October lost 2.44 points, shown in the report’s table as −2.44.
Results: reflectivity lost after the wash
- Brush, uncoated2.17%
- Brush, with Mirtek0.16%
- Jetwash, uncoated2.44%
- Jetwash, with Mirtek0.4%
Brush wash July 2022, jetwash October 2021; lower is better (Chemitek case).
| Trial | Uncoated mirrors | Mirtek-coated mirrors | Chemitek’s result |
|---|---|---|---|
| Brush wash, July 2022 | 2.17% lost, 14–25 July | 0.16% lost, 14–25 July | 13 times slower |
| Jetwash, October 2021 | 2.44% lost in 11 days (0.22% a day) | 0.40% lost in 10 days (0.04% a day) | 6 times slower |
| Jetwash, October 2021, day 17 | – | 1.34% lost in 17 days (0.08% a day) | still 2.8 times slower |
Chemitek adds that in several cases the coated mirrors’ reflectivity recovered to its starting level, and concludes that applying Mirtek with a brush gives the better result. The report also claims faster recovery after abnormal weather, such as sandstorms, and easier cleaning than on uncoated mirrors, but gives no separate data for either.
How to read it
July is the cleaner comparison. Coated and uncoated mirrors were read on the same days, eight positions each. Three details matter:
- Five days in, on 19 July, both groups had lost reflectivity: the coated mirrors about 1.5 points on average, the uncoated about 2.8. By 25 July both had regained some, and the coated mirrors nearly all of it. The report doesn’t say whether wind, dew or a wash did that; Chemitek’s reading is that Mirtek lets reflectivity recover faster.
- Chemitek’s 0.16% equals the sum of the eight coated positions’ changes. Their average was close to zero (−0.02 points), so on the table’s own numbers the 13 times is conservative.
- The dates span 11 days, although the summary says 10. Both groups share the dates, so the ratio holds.
October is the weaker comparison. The coated mirrors were read from 8 October and the uncoated ones from 14 October, so the two windows saw different weather. In the week both groups were read, 18 to 25 October, the coated mirrors lost about half as much reflectivity as the uncoated ones. Chemitek’s 2.8 times at day 17 is the more cautious reading of this trial.
For both trials: a handful of mirror positions, windows of 10 to 17 days, reflectivity rather than plant output, and no instrument named. In July the coated and uncoated mirrors also sit in areas with different labels (SB10 and SB30), and the report doesn’t say how the two areas compare.
What this means for South African CSP plants
- The fleet. South Africa has 600 MW of CSP in seven plants, all in the Northern Cape (CSIR, 2025; van Zyl et al., 2025). Mirror soiling matters more there than panel soiling does on PV; see CSP.
- Dust loads vary enormously. Near a ferromanganese smelter at Emalahleni, test mirrors lost on average 32.6% of their reflectance per 14 dry-season days (Swart et al., 2023), against about 2.2–2.4% in 11 days on Solaben’s uncoated mirrors. A result on Spanish dust has to be proven again on yours.
- Troughs and heliostats. This trial was on parabolic troughs. Chemitek sells Mirtek for all types of CSP mirror, but a heliostat field would need its own test.
- Water. Mirtek is dosed into distilled or demineralised wash water (Chemitek), so it goes on during a wash the plant already runs. South African CSP plants wash within water-use licences of 1,397–17,900 m³ a year per MW (van Zyl et al., 2025); if the coating slows soiling on your field, the saving shows up as longer intervals between washes.
- Measure before and after. Reflectance readings on coated and uncoated mirrors, on the same days, show what Mirtek does on your field and when a wash pays. See CSP mirror reflectivity measurement and CSP plant O&M.
Want proof on your own mirrors first? We’ll trial Mirtek on sample mirrors beside matched uncoated ones and read both with a reflectometer; pilots are free for utility-scale plants. If your wash crews are ready, order just Mirtek.
