Today, most solar cells are made with a process that turns almost half of the raw material—highly refined and processed crystalline silicon—into sawdust. A new process developed by startup Astrowatt aims to eliminate most of this waste while making solar cells more efficient.

Conventional solar manufacturing requires sawing a block of crystalline silicon into wafers about 180 micrometers thick. As the saw cuts through the silicon, it turns almost the same amount of silicon (a layer 100 to 150 micrometers thick) into sawdust that can’t typically be reused.

With the conventional process, a millimeter of silicon can produce about three solar-cell wafers. Astrowatt says it can make five or more wafers from the same amount of material by mostly replacing the sawing with a technique that allows it to peel thin layers of silicon away from a thick silicon wafer.

Astrowatt is one of several companies hoping to substantially reduce the amount of silicon needed to make solar cells. Although the price of silicon has dropped in recent years, it’s still the most expensive item in solar-panel manufacturing.

The Astrowatt process begins by sawing a block of silicon into relatively thick wafers, each nearly a millimeter thick. The company then modifies the top of each wafer so that it can act as the back of a solar cell—a process that ends with depositing a layer of metal onto the wafer.

Next, the wafer is heated, causing stress within the material because the metal and silicon expand at different rates. Applying a wedge to the edge of the stressed silicon starts a crack that propagates from one edge to the other, allowing the engineers to finally peel away the metal film along with a thin, 25-micrometer layer of silicon. Crucially, the crystalline structure of silicon allows the crack to propagate evenly across the entire wafer, and the silicon is flexible, so it won’t shatter as it’s peeled off.

The resulting metal-silicon film is then further processed to form the front of a solar cell. The entire process is repeated, with successive 25-micrometer layers being peeled off the original thick wafer. Once this is finished, what’s left is a wafer that’s still relatively thick, ranging from 180 micrometers to a few hundred. It can either be used to make a conventional solar cell, or it can be recycled by dropping it back into the furnace that produces blocks of silicon. (Unlike the sawdust, the wafer remains of high enough quality to be recycled.)

Other companies have made thin wafers of silicon using new sawing techniques or other methods. But other approaches tend to produce fragile wafers, and they can’t be used with existing cell manufacturing equipment. Rajesh Rao, the company’s director of technology, says the metal backing on Astrowatt’s cells makes them more durable.

The company has demonstrated the technology in the lab, making large, eight-inch-wide wafers and small solar cells that are nearly 15 percent efficient. That’s slightly less efficient than conventional crystalline silicon solar cells, but the researchers haven’t yet applied all of the known methods for increasing solar-cell efficiency. In fact, the cells could theoretically reach higher efficiencies than conventional silicon solar cells because they’re thinner, which makes it easier for electrons to exit the cell to generate electricity.

The next step is to demonstrate the process on commercial-scale equipment. Almost all of the steps in the process can be done on machines already found in solar-cell factories.

So far, Astrowatt has raised an undisclosed amount in an initial round of venture capital investment, along with $1.5 million under the U.S. Department of Energy’s Sunshot initiative, which aims to make solar power competitive with electricity from fossil fuels.

The approach does have some drawbacks. The metal-silicon films curl up slightly, which makes them somewhat difficult to handle on a conventional production line. Also, unlike some other approaches, this one doesn’t completely eliminate the need to make a block of crystalline silicon and saw into it, although it does greatly reduce the amount of sawing needed.

Other approaches, such as one being developed by startup 1366 Technologies, aim to eliminate these steps altogether, which could further reduce manufacturing costs. That technology, however, has challenges of its own, including achieving high yields and producing high-quality silicon.