From the access road the array looks clean.
Up close there is a grey haze on the glass, thin enough that you would rub a finger through it without thinking.
Each grain sits a few tens of microns deep and weighs nothing.
Together they can take 30 percent off a module’s output inside a month, with no storm involved.
The usual answer arrives on a truck with a tank on it.
The other answer arrives with no tank at all.
Why the grit holds on, and what shifts it
Dust does not sit on glass. It grips it.
The forces are van der Waals attraction and electrostatic adhesion, the same family of surface effects that lets a gecko hang off a ceiling, and at that scale they beat gravity easily.
Water breaks those bonds, which is why washing works. It also has to be clean water or it leaves mineral spots behind, and it has to be brought in.
The alternative is to put a charge on the particle and push it off with a matching one.
Which should not work. Desert dust is mostly silica, an insulator, and insulators do not take an induced charge easily.
The way around it is water again, but a trace of it. Silica is hydrophilic, so above roughly 30 percent relative humidity every grain wears a shell of adsorbed water molecules, and that shell conducts.
Charge the shell and you have charged the grain. That is the whole trick.
The rig, which is almost nothing
Underneath sits a glass plate carrying a transparent conducting layer of aluminium doped zinc oxide, laid down by atomic layer deposition at five nanometers.
That is thinner than the wall of a soap bubble, and it does not get in the way of the light.
Above it a plain metal electrode passes across at a gap of about half an inch, carrying 12 kilovolts.
The field induces charge on the grains, the coating underneath takes the opposite sign, and the particles leave the surface.
Nothing touches the glass, so nothing scratches it, which is the failure mode of every brush based system ever fitted to solar panels.
And no current flows between the two electrodes, so the electrical draw is essentially zero. The only energy going in moves the bar along a rail on a small stepper motor.
What the bench actually returned
The dust was Arizona test dust, a crystalline silica blend that stands in for desert mineral grit.
Humidity was run from 5 to 95 percent.
Below about 30 percent the effect falls away, because there is not enough adsorbed water on the grains to carry a charge. Above it, almost everything comes off.
For particles larger than roughly 30 micrometers, about a thousandth of an inch, the rig recovered 95 percent of the power the dust had cost.
The work was published in March of 2022 by Sreedath Panat and Kripa Varanasi at MIT, funded through an Italian energy company’s research arrangement with the university.
Varanasi’s own estimate is that the approach could take 80 to 90 percent out of a farm’s cleaning bill.
The parts that are not solved
Below that 30 micrometer line the physics turns against you. Van der Waals attraction scales differently from the electrostatic push, and the finest grains stay where they are.
The same group has since gone after those with nano textured conductive surfaces, which is a second paper rather than a fix inside this one.
Then there is the voltage. A specialist in electrostatic panel cleaning has warned in print that a conducting plate at 10,000 volts of direct current can create a life threatening hazard, which is a genuine constraint on hardware crews work around.
The coating also has to get onto the glass. Five nanometers of that oxide by atomic layer deposition is routine in a factory and an open question on panels already bolted down in a field.
Soiling varies enormously by site as well, so what any farm would save depends heavily on where it sits.
And all of this is bench work. No field trial has published numbers.
What it would be worth
Washing panels consumes about 10 billion gallons of fresh water a year worldwide, which is drinking water for two million people.
It also runs to roughly 10 percent of what it costs to operate an installation, largely because the water has to be trucked to it.
And the places with the most panels are the places with the least water. Nearly four million modules cover almost eight square miles of Abu Dhabi desert, and the same pattern runs through the Atacama, Rajasthan and the American Southwest.
A one percent output loss on a 150 megawatt plant costs 200,000 dollars a year. Across the industry the soiling bill runs into billions a year.
Against numbers like that, a metal bar on a rail and a coating nobody can see is a cheap thing to try.
What it has not yet done is spend a year in the actual desert.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
