NF3: The Greenhouse Gas That Nobody Knew
NF3, which is used in hi-tech manufacturing, was hailed as a way to fight global warming. But research shows that it has 17,000 times the warming potential of carbon dioxide.
When industry began using NF3 in
high-tech manufacturing, it was hailed as a way to fight global
warming. But new research shows that this gas has 17,000 times the
warming potential of carbon dioxide and is rapidly increasing in the
atmosphere - and that's turning an environmental success story into a
public relations disaster.
Hypothetical question: You're
heartsick about global warming, so you've just paid $25,000 to put a
solar system on the roof of your home. How do you respond to news that
it was manufactured with a chemical that is 17,000 times stronger than
carbon dioxide as a cause of global warming?
It may sound like
somebody's idea of a bad joke. But last month, a study from the Scripps
Institution of Oceanography reported that nitrogen trifluoride (NF3),
with a global warming potential of 17,000, is now present in the
atmosphere at four times the expected level and rapidly rising. Use of
NF3 is currently booming, for products from computer chips and
flats-screen LCDs to thin-film solar photovoltaics, an economical and
increasingly popular solar power format.
Moreover, the Kyoto
Protocol, which limits a half-dozen greenhouse gases, does not cover
NF3. The United Nations Framework Convention on Climate Change now
lists it among five major new greenhouse gases likely to be included in
the next phase of global warming regulation, after 2012. And while that
may be reassuring, it also suggests the complicated character of the
global warming problem.
In fact, NF3 had become popular largely
as a way to reduce global warming. The U.S. Environmental Protection
Agency began actively encouraging use of NF3 in the 1990s, as the best
solution to a widespread problem in making the components for
everything from cell phones to laptop computers. Manufacturers in the
electronics industry all use a vacuum chamber to etch intricate
circuitry and to deposit a thin layer of chemical vapor on the surface
of a product. Some of the vapor inevitably builds up instead as glassy
crud on the interior of the chamber.
To tear apart that layer of
crud and clean the vacuum chamber, manufacturers were using powerful
fluorinated greenhouse gases. The usual choice, hexafluorethane, or
C2F6 sounds better at first than NF3. In global warming terms, it's
only about 12,000 times worse than carbon dioxide. But C2F6 is
difficult to break down, and roughly 60 percent of what goes into the
vacuum chamber ends up in the atmosphere. With NF3, estimates suggested
that under optimal conditions, roughly 98 percent of what goes into the
vacuum chamber is destroyed there.
So when the semiconductor
industry announced a voluntary partnership with the EPA to reduce
greenhouse-gas emissions by 10 percent from 1995 levels between 1999
and 2010, NF3 became the replacement technology of choice. Makers of
flat-screen displays soon announced a similar program. In 2002, the EPA
gave a Climate Protection Award to the largest NF3 producer,
Pennsylvania-based Air Products and Chemicals Inc., for its work in
Then last summer, a paper calling NF3 "the
greenhouse gas missing from Kyoto" attracted widespread press
attention. Co-authors Michael J. Prather and Juno Hsu of the University
of California at Irvine noted that NF3 is one of the most potent
greenhouse gases known and persists in the atmosphere for 550 years.
back in the 1990s when the Kyoto Protocol was being negotiated, NF3 was
a niche product of unknown global warming potential (GWP). [In
calculating GWP, carbon dioxide is the basic unit, with a GWP of one.
For other gases, scientists measure infrared-absorption, the spectral
location of the absorbing wavelengths, and the atmospheric lifetime of
the gas to determine its global warming effect relative to carbon
dioxide.] So NF3 got left out, meaning no requirement for industry to
track emissions, or even to report how much NF3 is actually being
That left room for what felt to Prather like a
"flimflam." In an interview with Yale Environment 360, he estimated
that 20 or 30 percent of total NF3 production ends up in the atmosphere
- not the two percent industry had seemed to suggest. He and Hsu
characterized Air Products, the same NF3 producer that the EPA had
honored, as producing the annual global warming equivalent of one of
the world's largest coal-fired power plants.
A new paper,
published in Geophysical Research Letters in October, filled in gaps in
this glum picture - and threatened to turn the NF3 emissions success
story into a public relations disaster. Ray Weiss and his research team
at the Scripps Institution of Oceanography reported that NF3 is now
present in the atmosphere at four times the expected amount, with
atmospheric concentrations rising 11 percent a year. Working from
annual production estimates of 4,000 metric tons, Weiss figured that
about 16 percent of current production is ending up in the atmosphere.
Painter, a vice president at Air Products, praised the Weiss paper but
argued that "in terms of order of magnitude the numbers are relatively
close" to earlier estimates. In a letter to New Scientist magazine this
summer, Painter had seemed to give the impression that overall
emissions were in the two percent range. "More than 20 years of
research and work with our customers finds that less than 2 percent of
NF3 is released into the atmosphere," he wrote.
But in an
interview with Yale Environment 360, Painter said Air Products has a
two percent emissions rate just in producing and packaging the gas,
though he said that rate continues to go down. He said global NF3
production is actually 7,300 tons annually. Given Weiss's figures for
atmospheric concentrations, he said, that would translate to an overall
emissions rate closer to 8 percent, including manufacturing,
transportation, and end-use.
Getting the advertised results with
NF3 always hinged on an expensive new technology called remote plasma
cleaning. It breaks up the gas in a remote container, then injects the
active ingredient, fluorine, together with nitrogen, into the vacuum
chamber. With the optimal configuration, the process destroys almost
all the NF3. Bigger companies made the change to remote plasma cleaning
when they switched to newer fabrication tools, often at great expense.
"You can hear guys saying, 'I've gone from a Hummer to a Prius. I've
met all my voluntary commitments,'" said Painter.
companies stuck with older tools, simply replacing C2F6 with NF3. This
Band-Aid approach still releases about 20 percent of the NF3 into the
atmosphere. Painter argued that the struggling economy will force
manufacturers to shut down these less efficient production lines,
reducing overall emissions. But in October, Global Industry Analysts
estimated that over the next four years NF3 production will increase to
almost 20,000 tons, because of growing demand in the electronics
Moreover, even the latest equipment does not guarantee
that a company will achieve the optimal emissions rates - for instance,
in the solar cell industry. Amorphous silicon thin-film solar
photovoltaic cells, manufactured using NF3, are slightly less efficient
than crystalline silicon solar cells, the dominant technology. But they
are cheaper to produce and expected to supply a rapidly increasing
share of the solar market, for both large-scale and domestic
Because thin-film is a new technology,
manufacturers generally use the latest equipment. But a knowledgeable
source, who asked to remain unidentified, recently visited thin-film
solar researchers in Asia. "They were unaware of the NF3 issue. They
were using a remote plasma, but they were also using quite a bit of
NF3. They weren't sure they had it set up right for 98 percent
destruction. It wasn't really on their radar."
The bottom line,
said UC Irvine's Prather, is that "industry really cannot be trusted
for self-regulation." We will not know the extent of the problem "until
we have honest, legally required reporting." The other important lesson
from the NF3 case, according to Scripps's Weiss, is that the bottom-up
measurements required by some global warming regulations aren't enough.
Figuring out how much methane a cow produces, then adding up the cows,
may not give you ground truth when it comes to global warming. "You
have to measure from the top down, and see what's actually going into
A practical alternative to NF3 already exists.
According to Paul Stockman of Munich-based Linde Gas, fluorine has zero
global warming potential and no atmospheric lifetime. But it's also
highly toxic and reactive. So instead of being shipped in bottles like
NF3, it must be generated on site using special equipment. Stockman,
whose company manufactures NF3, said fluorine will become essential in
thin-film solar manufacturing, because faster cleaning times mean a
substantial boost in productivity.
Meanwhile, Air Products says
it supports adding NF3 to the list of regulated greenhouse gases in the
Kyoto Protocol's second commitment period, beginning in 2012. But
Prather believes industry needs to get more honest about NF3 production
and emissions before then. Solar cells are like any other product, he
said, in that the manufacturing process has a global warming footprint.
But solar buyers are likely to be particularly concerned with the size
of that footprint - and not so pleased to find out that what they
thought was a Prius is really just a Hummer on the inside.