A girl rides a donkey carrying jerry cans filled with water from a cistern at a make-shift camp for displaced Yemenis in severe shortage of water, in the northern Hajjah province on March 24, 2020.

A girl rides a donkey carrying jerry cans filled with water from a cistern at a make-shift camp for displaced Yemenis in severe shortage of water, in the northern Hajjah province on March 24, 2020. Photo by ESSA AHMED / AFP) (Photo by ESSA AHMED/AF

The Military Wants To Produce Water From Air. Here’s the Science Behind It

The key is the right combination of elements in porous crystal structures, and the AI-powered search is on.

A promising new way to extract drinking water from air could change how U.S. troops hydrate in remote locations — and might even relieve regional water-scarcity tensions. 

On Thursday, General Electric announced that it has received $14 million under DARPA’s Atmospheric Water Extraction project to continue developing a device that can produce enough water to support 150 troops while remaining small enough to be lifted by just four people.

Extracting water from air is the sort of elusive challenge that has frustrated scientists for decades. Even in desert climates, air contains water vapor. But condensing it to fill a tank or a glass requires cool temperatures. Think of walking out into a grassy field at dawn when the grass has small droplets of water on it. The temperature at which that water in the air condenses into those little droplets is the dew point, which typically occurs at night when the air is the coolest. In arid regions with low relative humidity, that dew point is very low. 

Today’s water-condensing systems require too much power to be worth the effort in arid climates. Some salts or other materials work but the water harvesting takes too long to be worthwhile. 

UC Berkeley chemistry professor Omar Yaghi, who is part of GE’s team on the project, has developed a new family of highly porous crystal-like substances, good for capturing carbon and catalyzing reactions.  Yaghi and his team began to experimenting  with these metal-organic frameworks, or MOFs, in the desert in 2014

“We designed several generations of prototypes employing kilograms of MOFs and showed that these water harvesters work well in some of the driest deserts in the world,” he told Defense One in an email. “We tested these MOF water harvesters in Arizona and the Mojave deserts and found that significant amounts of water can be harvested from air.” 

Specifically, he found that they can absorb water from the air at relatively high temperatures of 25 Celsius and releasing it at 45 Celsius.

Water vapor binds to pores in his device much the same way as water bonds in ice, Yaghi said.

“The weak bonds formed by those ‘seed’ water molecules further attract other water molecules from air thereby helping to concentrate water inside the pores, especially in low-humidity environments,” he wrote. “In essence, inside these MOFs, one has solid water (i.e. ice fragments) in hot weather. To get the water out, as the goal is liquid water, one simply heats the materials to a mere [45 Celsius]. This provides enough energy to break those weak bonds and release the water from the pores to make clean drinking water.”

That means you can trap water in the pores in the desert at night, when it’s a bit cooler (but not cold) and capture it as liquid during the day in the heat. With power, it can go through several water-producing cycles per day. 

So where to go from here? The three-year DARPA program will proceed along two tracks. GE won the award in the “stabilization track” to develop materials to provide water to a 150-person unit. The second track seeks to build a canteen-sized air-to-water device for individual soldiers. 

Yaghi said the key will be using artificial intelligence to find the right combinations of elements.

“There are over 70,000 different MOFs…being invented so far since we first pioneered them, and the possibilities of making new ones are endless,” he says. “This is really an infinite chemistry with infinite materials. Thus being able to develop and search this vast MOF space will require computations and machine learning. Identifying a specific MOF with favorable water harvesting abilities and optimizing its behavior in devices will be some of the outcomes of the computation using AI methods.”