From the rooftop of Klaus Lackner's seven-storey building on the Arizona State University campus, photovoltaic panels seem to glisten in every direction.
The school claims to have more solar capacity installed than any other university in America - part of a plan to offset the carbon emissions of this institution of more than 80,000 students.
But the odd little box Lackner has come to check might someday take things a big step further. If it works on a larger scale, what's in the box could make the university a negative emitter - more than offsetting the amount of carbon it releases into the air.
Lackner's box is part of a new wave of technology aimed at heading off climate change.
Such devices were considered a pipe dream until recently, but more and more, they are seen as indispensable. That's because a goal set by last year's Paris accord on climate change - holding global warming to "well below" 2C more than pre-industrial levels - may not be achievable unless such technology comes to fruition.
Solar power and electric cars won't be enough, scientists say. Humans may have to engineer ways to draw carbon out of the air, the way trees do naturally, and on a gigantic scale.
"If you want to balance the books at this point," Lackner says, "I don't think you have a choice but to pull CO2 back that has already made it out, or is about to make it out, because we are not overnight shutting down all the coal plants."
But that turns out to be very hard to do.
There are a wide variety of ideas for getting carbon out of the atmosphere, including Lackner's boxes and specialised power plants that burn renewable biomass and stow away the waste carbon dioxide. But they all have high costs and other problems to overcome in the search for technology big enough, and cheap enough, to make a difference.
The device that Lackner is using to capture carbon dioxide doesn't look like much: a transparent box containing two stacks of something that resembles pasta.
The "pasta" actually contains a sorbent, or absorptive material, that has been crushed into tiny pieces and embedded in strips of softer plastic. Lackner has shown that, in the open air, the sorbent, a synthetic equivalent of natural substances such as amber, binds with carbon dioxide, acting as a kind of sponge to pull the greenhouse gas out of the atmosphere.
The need for "negative emissions" technologies such as the one being tested at Arizona State - one of the first US universities to establish a centre for research in this area of science - comes down to a simple matter of mathematics.
With the planet already about 1C warmer than in pre-industrial times, scientists have estimated the amount of carbon we can emit and still stay below a 2C increase.
The carbon budget is extremely tight - well under 1000 additional gigatonnes (1000 billion or 1 trillion tonnes) of carbon dioxide - and human beings emit about 32 gigatonnes annually through energy use alone.
Moreover, the Paris accord contains an even more ambitious target: holding the increase in global warming to 1.5C, which some scientists say might be impossible unless massive amounts of carbon can be removed from the air.
There is no shortage of ideas on how to do that, but "all of them face significant challenges," says Noah Deich, executive director of the Centre for Carbon Removal at the University of California at Berkeley. "There's no silver bullet. There's no clear winner."
The organisations trying to develop ways to suck carbon dioxide directly out of the air include a number of start-ups. One firm, Carbon Engineering, has a plant in British Columbia, Canada, that is capturing a tonne a day of atmospheric carbon dioxide using a liquid rather than a solid sorbent, according to Geoff Holmes, the company's business development manager. (One funder of this effort is Bill Gates.)
Direct air capture has advantages: it's passive and can be located virtually anywhere. And the carbon being captured doesn't come from a single source, such as burning coal. It can come just as well from the exhaust of a car or an aeroplane.
But critics contend that the costs will be prohibitive.
"I'm sceptical there is a technology that will cheaply capture CO2 at 400 parts per million when it's expensive to do at 400,000 parts per million in a smokestack," says Rob Jackson, a professor of earth sciences at Stanford University who has researched negative emissions technology. "It's tougher thermodynamically.
Carbon dioxide in air is a thousand times less abundant." Jackson and other researchers have studied what is currently perhaps the leading idea for achieving negative carbon emissions, and it's very different. It's called bioenergy with carbon capture and sequestration, or BECCS.
The idea pairs two existing technologies: growing corn, trees or other forms of biomass to provide fuel for power plants and then capturing the waste carbon dioxide and storing it in the ground, instead of letting it escape into the air.
In repeated cycles, growing plants for biomass takes carbon dioxide out of the air, and capturing and storing the waste carbon dioxide when the biomass is burned for fuel yields a net reduction in the amount of carbon in the atmosphere.
The BECCS technology appears to have been advanced by a project involving the US Department of Energy, agribusiness giant Archer Daniels Midland and the University of Illinois.
About a million tonnes of carbon dioxide from a corn-to-ethanol production facility has been captured and sequestered in Illinois sandstone layers 2100m below the surface, according to Sallie Greenberg, a scientist with the Illinois State Geological Survey who works on the project.
But Lackner and other researchers think BECCS may have an Achilles heel - namely, that the amount of land required to grow enough trees or corn to reduce atmospheric carbon on a planetary scale would be enormous.
One recent study found that to offset about a third of current global carbon emissions, you would need to cover 48 US states with BECCS projects.
A related idea - planting huge numbers of trees where they do not grow now - faces a similar hurdle. No doubt, more trees would mean less carbon dioxide in the atmosphere. But once again, vast areas could be required, and in the future, humanity will need more land, not less, to grow food.
Some researchers advocate yet another approach: "enhanced weathering".
Scientists have long known that over vast periods, carbon is removed from the atmosphere by becoming embedded in rock.
Here's how it works: rainwater containing atmospheric carbon in the form of carbonic acid falls on certain types of rock. The rock breaks down, or "weathers", and the interaction leads eventually to the formation of a carbonate - such as limestone - with the carbon locked inside.
However, the process is very slow. "You can wait about 100,000 years, and then nature mops up all of our CO2," Lackner says.
The idea is to speed up this process by crushing silicate rock, such as olivine, and spreading it over the ground to be rained on. Sounds simple, but it takes a lot of money and energy to mine and crush rock, and the process would produce fine dust that could prove a major health hazard if inhaled.
It would also be expensive. A recent study estimated that it would cost US$60 trillion to US$600 trillion to remove 50 parts per million of carbon dioxide from the planet's atmosphere in this manner.
Lackner and his colleagues are grappling with similar issues. Their box works well as a demonstration model, but how do you pull carbon dioxide from the air affordably at a much larger scale? And what do you do with all that gas to keep it from getting back into the atmosphere?
A key property of Lackner's device is that the plastic sorbent gives back CO2 readily if you simply get it wet. If you lower the plastic strips into the box and spray them with water, computer readings show carbon dioxide levels spiking.
"That's more than a cup of coffee of pure CO2 in there," Lackner says, after the process has been under way for more than an hour.
In Lackner's laboratory, the sorbent has been formed into objects of many shapes, including a rough doormat, with lots of bristles.
When the carbon-gobbling doormat is relatively full of CO2, the researchers place it in a chamber with a large indoor plant called devil's ivy, a fan, and plenty of humidity.
The sorbent will feed the plant carbon dioxide and help it grow.
Lackner says the experiment is about "finding a temporary use" for some of the carbon dioxide that has been pulled from the air.
But in the long term, captured carbon dioxide will have to be stored underground or, perhaps, in rock. "My personal view is the true long-term storage is mineral carbonates, which is some form of accelerated weathering," Lackner says. But he says it will be necessary to store carbon in the ground, too.
Underground storage is being explored at the US$6.6 billion Kemper Plant, a 582-megawatt power plant in Mississippi expected to start sequestering carbon this year.
The plant is designed to strip 65 per cent of the carbon dioxide out of coal through a gasification process. The CO2 is then piped miles away to oil fields and injected into the ground as part of a process called "enhanced oil recovery".
The pressurised gas will make it possible to extract more oil from the wells and will be sequestered underground when the wells are sealed.
But even if storage can be worked out, many studies suggest Lackner's air-capture approach to removing carbon dioxide from the atmosphere would be extremely expensive.
Over all these technologies and ideas hang questions about how to lower the cost and about when the world will truly start investing. A boon to all of them, of course, would be the setting of a global or national price on carbon emissions, thus making their removal more valuable.
Lackner, for his part, thinks that with a US$10 million R&D investment, he could scale up to a device that would remove a tonne of carbon dioxide from the air per day. Such devices would become cheaper if they were mass-produced, he says.
A tonne a day may not sound like much, but as Lackner points out, it's much more than this World Bank estimate of US carbon dioxide emissions: 17 tonnes per person per year. The device would, in effect, offset the emissions attributed to about 20 people.
It may be that no single approach is going to provide the answer. With the clock ticking on the goals of the Paris accord, and with scientists disturbed by the accelerating rate of climate change, the important thing, Lackner says, is simply to act.
"We already delayed, and we waited," he says, "and now is the time to clean up."