Eli Calls Rabett Droppings

UPDATE: Thinking assignment: Eli, has well, been thinking. NF3 has a lot of advantages for semiconductor manufacturers which outweigh its cost. It is replacing CF4, which in addition to leaving carbon residues, requiring more energy and other problems is produced almost unavoidably in Aluminum manufacture. OTOH, CF4 used in plasma etching is to some extent destroyed. Should we insist on continuing to use CF4 in plama etching to destroy some fraction of it?

FURTHER UPDATES: click read more at the bottom

BEGIN READING HERE: Well, actually something much stronger, but this is a family blog and the bunnies are, well they have long ears and good memories. The story starts with a piece of cow dung dropped into Geophysical Research Letters by the tag team of Michael Prather and Juno Hsu. Let us first read the abstract

Nitrogen trifluoride (NF₃) can be called the missing greenhouse gas: It is a synthetic chemical produced in industrial quantities; it is not included in the Kyoto basket of greenhouse gases or in national reporting under the United Nations Framework Convention on Climate Change (UNFCCC); and there are no observations documenting its atmospheric abundance. Current publications report a long lifetime of 740 yr and a global warming potential (GWP), which in the Kyoto basket is second only to SF₆. We re-examine the atmospheric chemistry of NF₃ and calculate a shorter lifetime of 550 yr, but still far beyond any societal time frames. With 2008 production equivalent to 67 million metric tons of CO₂, NF₃ has a potential greenhouse impact larger than that of the industrialized nations' emissions of PFCs or SF₆, or even that of the world's largest coal-fired power plants. If released, annual production would increase the lower atmospheric abundance by 0.4 ppt, and it is urgent to document NF₃ emissions through atmospheric observations.

Pay careful attention to that last sentence which Eli has conveniently bolded while we explore what NF3 is used for and why there is no chance in hell that even 1% of the annual production would ever be released to the atmosphere including in case of nuclear war. NF3 is used for plasma etching and high temperature thermal cleaning in the electronics industry.

That means that NF3 is run into RF/microwave/electron discharges (plasmas) or into systems held at very high temperatures (thermal cleaning), which decompose the molecule, leaving principally nitrogen and fluorine atoms or molecules, and various small radical species such as NF. The very aggressive reactants etch surfaces in the production of electronics or clean out an apparatus that has been used for some process. NF3 IS DESTROYED IN ALL OF THESE PROCESSES.

The Press Association figured out what NF3 is used for

Nitrogen trifluoride, or NF3, is used in the electronics industry mainly to flush out the by-products of chemical vapour deposition. This is the process by which thin films are deposited for liquid crystal displays (LCDs) - used in flat screen TVs - or silicon chips.

but did not realize the implications, that NF3 is DESTROYED in the process. They went on to note that

Ironically Air Products' developed NF3 as an alternative to perfluorocarbons (PFCs), greenhouse gases which are subject to the Kyoto protocol.

The protocol covers six man-made greenhouse gases, carbon dioxide, methane, nitrous oxide, hydrofluorocarbons, PFCs and sulphur hexafluoride. NF3 was one of more than a dozen less prominent greenhouse gases that were excluded when the protocol was agreed.

The Telegraph, predictably screwed the pooch, seeing the word plasma, they thought that it referred to plasma screen TVs, and predictably people panicked thinking that their new plasma TV had nasty chemicals (actually NF3 is, as these things go, easy to handle and not terribly dangerous, not nearly as dangerous as NH3, ammonia, for example) and the web spread the fertilizer.

Eli, who dabbles in materials science, pretty much knew this was crap the minute he read it, but surely he said, someone has thought of this issue before, and sure enough, if you google NF3 and cleaning, you toss up "Environmentally friendly wafer production: NF3 remote microwave plasma for chamber cleaning" by H. Reichardt, A. Frenzela, and K. Schoberb in Microelectronic Engineering 56 (2001) 73–76, and you don't even have to read the paper, the abstract is enough

For NF3 remote microwave plasma chamber cleaning, compared with CF4 cleaning processes, a reduction of emission of gases relevant to global warming is observed. At the same time a reduction of operating costs for the abatement is possible. The presented data show a very high destruction and removal efficiency for NF3 and its major decomposition product F2 in an ESCAPE abatement system.

which describes how what NF3 survives the plasma process can be scrubbed. It damn well better be because F2 and its atmospheric decomposition product HF are no walk in the park, not as greenhouse gases, but as aggressive toxic gases. Any fab using NF3, is scrubbing the effluent.

If Prather and Hsu had honestly read this paper, even the introduction, they would have reported something like

with new production lines being build newer — generally more severe— environmental statutes apply. On top of that, there is the world-wide goal of PFC emission reduction. Regarding the consumption of PFC etch gases, chamber cleaning processes are the major contributor. Since usually the utilisation of etch gas in these processes is less than 50%, the remaining gas has to be destroyed and removed by a waste gas abatement system. Generally, for CVD and etch processes, waste gas abatement is necessary for three reasons:

· Environmental concern and legal restrictions on emissions.
· Safety within the fabrication area.
· To prevent corrosion or clogging of exhaust lines and thus to guarantee process up-times.

Recently, the use of NF3 as an etch gas for chamber cleaning processes is reported to give promising results. Besides less wear of the tool, gas consumption is lower, since the utilisation of NF3 is very high (85–99%). At the same time NF3 has a much lower atmospheric lifetime (740 years) than standard etch gases like CF4 and C2F6 (estimated atmospheric lifetimes of 50 000 and 10 000 years, respectively).

MOREOVER in 2001 Reichardt, Frenzela, and Schoberb showed that with proper scrubbing NF3 and F2 emissions were BELOW the detection limits of 3 and 1 ppm respectively. It also takes a lot less energy to decompose NF3 rather than SF6 or CF4, so there are considerable energy, and even greenhouse emission savings.

Eli was busy last week and only was able to sneak a comment about this nonsense into Real Climate. About the only blog that got this right is Physical Insights (prop. Luke Weston)

I must say, this looks like more biased “You’ve got a TV? You’re guilty of climate change!” baloney from the “green” fanatics in the press who like spinning scientific papers out of context.

Who also spotted the Reichardt et al. paper

Whilst these inorganic fluorine compounds and perfluorocarbons have large global warming potentials, which make for dramatic media headlines, their atmospheric abundances and mixing ratios are very small, and hence their contributions to radiative forcing in the atmosphere and hence to anthropogenic forcing of climate processes are very small by comparison to carbon dioxide, methane and water vapor.

Carbon dioxide is responsible for an increased radiative forcing term of 1.66 W/m2, according to up-to-date IPCC data, along with 0.5 W/m2 for methane and 0.16 W/m2 for nitrous oxide. For comparison, sulfur hexafluoride is associated with a far smaller increased radiative forcing term of 0.002 W/m2, along with 0.001 W/m2 for perfluoroethane. We can reasonably expect that the contribution from nitrogen trifluoride is similar, at around 0.001 to 0.002 W/m2. Whilst nitrogen trifluoride is certainly worthy of inclusion under the Kyoto protocol, along with perfluorocarbons and the like, especially as worldwide consumption of the gas grows, it is however nothing worth making a huge irrational fuss in the media about.

and the Wikipedia which fairly quickly pretty much got this right (ear tip to Wm)

Eli thinks that Prather and Hsu should be turned over to Dr. Motl in the operating room with the rusty machete.

UPDATE: So anyhow Eli went and RTFR and RTFRR and. . .


and what do you think he saw. Well in Prather and Hsu, you see the conclusion:

[10] Current production, if released, would add about 0.4 ppt (picomoles per mole) of NF3 to the lower atmosphere annually. There appear to be no reported atmospheric measurements in the peer-reviewed literature [Harnisch et al., 2000], and it is important to document the rise in atmospheric NF3 that is almost certainly occurring today.

If you read "Revised IR spectrum, radiative efficiency and global warming potential
of nitrogen trifluoride" J. I. Robson, L. K. Gohar, M. D. Hurley, K. P. Shine, and T. J. Wallington GEOPHYSICAL RESEARCH LETTERS, VOL. 33, L10817 you get a reasonable conclusion:

NF3 is a potent greenhouse gas. The upper limit of atmospheric concentration of NF3 is estimated to be 0.10 pptv and its contribution to radiative forcing of climate change is negligible (<10^4>preceded by an reasonable estimate of what is happening

Some PFCs and SF6 are included in the Kyoto Protocol and have been used extensively within the semiconductor industry as plasma etchants. In the plasma, the etching gas decomposes to give F atoms which react with the semiconductor surface. A proportion of the etching gas is not converted and is vented to the atmosphere. . . .

One motivation for the industry’s increasing interest in NF3 as a source of fluorine, as opposed to the PFCs or SF6, is that for NF3 there is a higher percentage conversion to fluorine in the cleaning processes (greater than 90–95% breakdown) [e.g., Rink et al., 2005]. Hence, emissions of NF3 are expected to be smaller than for alternative sources of fluorine, such as PFCs. The semiconductor industry’s usage of NF3 has therefore increased in recent years. . . .

The current atmospheric concentration of NF3 is probably at a level which is below that which can be detected. We provide a rough estimate of current concentrations based on industry figures (P. J. Maroulis, personal communication, 2006) but stress that data on actual usage and emissions is very uncertain. Current global production levels are believed to be about 2300 metric tonnes per year. An estimated 85% of these 2300 metric tonnes are used in processes which release an upper limit of 2% to the atmosphere. The remaining 15% are used in processes which release an estimated 30% to the atmosphere. However, these latter processes are currently being phased out. Using these values, the current day emission is then estimated to be 140 metric tonnes per year. Assuming these emissions have been sustained for 10 years and using the MWM lifetime of 740 years, yields an upper limit of the current mixing ratio of 0.10 pptv. Despite the large GWP of NF3, the amount currently in the atmosphere is so small that the contribution of this molecule to overall radiative forcing is very minor (<10^4>

Interestingly, it appears that the fraction of NF3 that remains undecomposed in a plasma can be reduced to practically zilch by operating at high RF voltage "Rf discharge dissociative mode in NF3 and SiH4 V Lisovskiy, J-P Booth, K Landry, D Douai, V Cassagne and V Yegorenkov J. Phys. D: Appl. Phys. 40 (2007) 6631–6640

This new dissociative δ-mode is characterized by a high dissociation degree of gas molecules (actually up to 100% in NF3 and up to 70% in SiH4), higher resistivity and a large discharge current.At rather high rf voltage when a sufficiently large number of high energy electrons appear in the discharge, an intense dissociation of gas molecules via electron impact begins, and the discharge experiences a transition to the dissociative δ-mode. The dissociation products of NF3 and SiH4 molecules possess lower ionization potentials, and they form an easily ionized admixture to the main gas.

Prather and Hsu play the old three card monte:

[7] While NF3 is ostensibly destroyed during the manufacture of flat screen displays,

apologies from the bunny to the Telegraph, Prather clearly planted the seed about killer flat screens. They continue by showing the queen

this destruction cannot be complete. We expect that some fraction of the NF3 produced will escape to the atmosphere during production, transport, use, or disposal.

without, of course, estimating the size of emissions as Robson did and then they spring the trap

The maximum potential release of NF3, assumed here to be its production, is equivalent to approximately 67 MMTCO2 (million metric tons of CO2, see Table 2). Thus, in terms of climate change, annual production of NF3 is now larger than emissions of PFCs or SF6 reported by the developed nations (Annex I) for 2005. (See Table 2; note that these emissions are generally decreasing since 1990.)

They are comparing the TOTAL PRODUCTION OF NF3 to the EMISSIONS of SF6 and the perflurocarbons. Comparing apples and baseballs as it were. A better comparison might be production of NF3 to the production of SF6 and PFC, but that would still overestimate the emissions of NF3, because SF6 and PFCs have uses and are produced in processes where they are not being destroyed. For example CF4 is produced in making aluminum and SF6 is used to prevent arcs in large electrical switches. Prather justifies this on the basis that

Experience with the ozone-depleting gas CFC-12 [Rowland et al., 1982] has shown that emission inventories from the chemical industry cannot be relied upon. Once released to the atmosphere, gases like CFC-12 and NF3 will take centuries to clean out. Given this potential, the production of high-GWP, long-lived, greenhouse gases like NF3 should be included in the national greenhouse gas inventories once global usage exceeds a threshold, e.g., 5 MMTCO2, no matter what the claim for containment.

Prather needs to make a case on emissions which he has not made rather than trying to scare the kids.

ZUPDATE: But wait, if we RTFR we see that in 1982 Rowland and friends attempted to fit estimates of emissions to observations of emissions and CFC-12 production. They concluded that

The CMA (Chemical Manufacturers Association) estimated atmospheric release of CCl2F2 (CFC-12) for the period 1976-1979 is 1.46 megatons, with an accuracy stated as +/- 5%. However, patterns A, B and D all exceed this estimate by 0.5 megatons, corresponding to an excess release of about 35% for this four year period.

One of the useful things one can do is to search forward by looking at papers that cite a paper. If you do this you come across (eventually) "Releases of refrigerant gases (CFC-12, HCFC-22 and HFC-134a) to the atmosphere" by Archie McCulloch, Pauline M. Midgley, and Paul Ashford Atmospheric Environment 37 (2003) 889–902 which has a neat table of CFC 12 emissions from 1930, where we read that the emissions in metric tons (tonnes) were

1974 435,100
1975 423,700
1976 413,600
1977 401,300
1978 376,700
1979 375,900
1980 379,900

and the total between 1976 and 1979 was 1.57 megatonnes). Now we have the issue of whether Sherry Rowland was being a good scientist and using SI (megatonnes) or casual tons. If he was being a bad boy that would make the CMA estimate 1.33 megatonnes with a difference of about 0.24 megatonnes between them. If he were doing right that would make the difference 0.11 megatonnes (e.g. a 7 or 8% difference or about twice that if Rowland used customary units), but curiously while emphasizing the Rowland et al. result as a sign of the perfidious CMA, Prather and Hsu ignore McCullocha and friends (who have several similar articles).

Eli would say that continued monitoring is a good place to start before having a panic.