Bill Gates is no stranger to Washington. His foundation has a big office here. And last March he met with President Trump and talked about preventing disease and spreading vaccines.

Recently Gates visited some key members of Congress. This time he was talking about what he sees as a key part of the answer for combating climate change: a return to nuclear power. And the Microsoft co-founder was trying to persuade Congress to spend billions of dollars over the next decade for pilot projects that would test two or three new designs for nuclear power reactors.

Gates has been thinking about nuclear power for a while. He founded TerraPower in 2006 and during his recent visits to Capitol Hill, he told lawmakers that he personally would invest $1 billion and raise $1 billion more in private capital to go along with federal funds for a pilot of his company's never-before-used technology, according to congressional staffers.

"Nuclear is ideal for dealing with climate change, because it is the only carbon-free, scalable energy source that's available 24 hours a day," Gates said in his year-end public letter. "The problems with today's reactors, such as the risk of accidents, can be solved through innovation."

Gates's latest push comes at an important turn in climate politics. Nuclear power has united both unpopular industry executives and a growing number of people - including some prominent Democrats - alarmed about climate change.

"I remain convinced that advancing nuclear . . . remains in our interest for both climate and national security reasons, and in the end we will not get across the finish line without direct government engagement," former energy secretary Ernest J. Moniz said in an email. "This is not special to TerraPower - it's a generic challenge."

But many nuclear experts say that Gates is the wrong messenger and that his company is pursuing a flawed technology. They say that any new nuclear design is likely to come at a prohibitive economic cost and take decades to perfect, market and construct in any significant numbers.

Lawmakers are listening to him, though. Through the Energy Department, Congress approved $221 million to help companies develop advanced reactors and smaller modular reactors in fiscal 2019, above the budget request. But Gates and TerraPower, which received a $40 million Energy Department research grant in 2016, are looking for more.

With some Democrats reconsidering opposition to nuclear energy dating back to the Three Mile Island accident 40 years ago, Gates met with lawmakers from both parties, including Sens. Lamar Alexander, R-Tenn., and Dianne Feinstein, D-Calif., both senior members of the Senate Appropriations Committee. Last month, he had dinner with Sen. Lisa Murkowski, R-Alaska, and three other senators.

Edwin Lyman, a nuclear expert at the Union of Concerned Scientists, said TerraPower is one of many companies that is raising the public's hopes for advanced nuclear reactor designs even though they're still on the drawing boards and will remain unable to combat climate change for many years.

"We think the vendors of advanced nuclear power designs are saying they can commercially deploy them in a few years and all over the world," Lyman said. "We think that is counterproductive because it is misleading the public on how fast and effective these could be."

Gates is a believer in tapping innovation on the climate front. He has invested heavily in other nascent technologies - much of it related to energy storage - in search of the sort of breakthrough he hopes will slow global warming. And Gates also gave $1 million to the campaign to approve a ballot initiative for a carbon "fee" in the state of Washington. (The effort failed.) But he has warned that a focus on solar and wind would be "dangerous."

Gates, who declined interview requests, won't say how much he has invested in TerraPower, but the Bellevue, Wash.-based company has about 150 employees.

Many nuclear power experts say that the technology Gates is promoting - called a "traveling wave reactor" - does not work as advertised, at least not yet. "These designs . . . require advances in fuel and materials technology to meet performance objectives," a Massachusetts Institute of Technology report said last year.

In his letter, Gates praised TerraPower's "traveling wave" technology. He said it "is safe, prevents proliferation, and produces very little waste" - important selling points in Congress, which has not settled on the location of a site for long-term waste storage.

Gates has compared the technology to a candle. He said that uranium-235, which is burned in conventional light-water reactors, would be used to ignite the rest of the candle, burning up depleted uranium-238 that is treated as waste.

And instead of water, it would use liquid sodium to cool the plant, which TerraPower said would be more efficient.

Gates has said the reactor could be placed in a vessel underground and left there for 60 years without refueling. That would reduce chances for human error and defuse concerns about long-term spent fuel storage or the theft of nuclear material during refueling or fuel reprocessing, the company said.

But critics say TerraPower has been stumbling over a handful of obstacles.

First, TerraPower has discovered that the traveling wave didn't travel so well and that it would not evenly burn the depleted uranium in the "candle."

Second, and partly as a result, it needed to change the design to reshuffle the fuel rods - and do that robotically while keeping the reactor running. Third, it has struggled to find a metal strong enough to protect the fuel rods from a bombardment of neutrons more intense than those commonly used in reactors - and for a much longer period of time.

TerraPower's Marcia Burkey said in an email that the company has been researching new steel alloys. It has sent ingots to a unique Russian test reactor and brought them back for examination. She said the company had made important advances in that and other areas.

In many ways, TerraPower's design resembles fast-breeder reactors. Fast breeders have faster-moving neutrons, the subatomic particles that trigger fission.

Allison Macfarlane, former chairman of the Nuclear Regulatory Commission, said earlier versions of fast-breeder reactors have turned in a "dismal performance." The United States built two small reactors at a government laboratory in Idaho, Japan built a commercial unit called Monju, and France built two called Phenix and Superphenix - and all of them have been shut down.

TerraPower also suffered a setback in October when the Energy Department effectively killed any chance of building a demonstration reactor in China. The department announced measures to prevent "China's illegal diversion" of U.S. civilian nuclear technology for military purposes.

Three years earlier, TerraPower had unveiled an agreement to establish a joint venture with China National Nuclear Corp. to build a pilot reactor. But the Energy Department, in a move that seemed aimed directly at TerraPower, said it would deny new license applications or extensions to existing authorizations related to the Chinese state-owned company.

TerraPower has been working on "advanced" nuclear technology for a decade, and it remains far from filing a final proposal for review by the Nuclear Regulatory Commission.

In fact, a small modular reactor design by NuScale Power is the application before the NRC. The commission is expected to complete its review by September 2020, NuScale says.

Since 2016, TerraPower has also been working with the Energy Department and Southern Co. on another reactor design. That one would rely on molten salt as both coolant and fuel. TerraPower believes an advanced molten salt reactor could be more efficient and produce less waste than current models.

However, that technology was examined in different countries 60 years ago - and abandoned. Lyman said the molten salt was "highly corrosive, so you need special materials for the reactor. That's an engineering problem they still have to confront."

The political engineering problem still needs work, too, though some surprising bipartisanship has taken place over the past year.

An unusual coalition of Sens. Sheldon Whitehouse, D-R.I., Cory Booker, D-N.J., Tom Carper, D-Del., James M. Inhofe, R-Okla., John Barrasso, R-Wyo., and Mike Crapo, R-Idaho, sponsored legislation aimed at speeding up NRC reactor approvals and capping company costs. The Senate passed the bill Dec. 20 and the House on Dec. 21. The measure was seen as a triumph for the industry group, the Nuclear Energy Institute, which called it "a significant, positive step toward reform."

The Union of Concerned Scientists took a neutral stance on the measure, which was altered before passage to protect the latitude of the NRC. But Lyman said the group still did not like the "idea of Congress micromanaging NRC licensing activities."

For all nuclear designs, both new and old, the colossal expense of nuclear construction and the absence of a carbon tax remain obstacles.

In the United States, only one new nuclear reactor has been completed in three decades. Two were shelved in 2017. Two others in Georgia are running wildly behind schedule and over budget with costs running around $27 billion, more than double the original estimate.