| 01 Oct 2025
Experimental determination of the global warming potential of carbonyl fluoride
Abstract. Carbonyl fluoride (COF2) has recently attracted attention as a potential low-global-warming-potential (GWP) replacement for high-GWP fluorinated gases (F-gases) used in semiconductor and display manufacturing, such as HFCs, PFCs, SF6, and NF3, because of its proven efficacy as a chamber-cleaning gas and rapid hydrolysis in moist air. In this study, the infrared absorption cross-section (ACS) of COF2 was measured using Fourier-transform infrared spectroscopy, and its radiative efficiency (RE) was calculated using a revised form of the Pinnock curve that incorporates stratospheric temperature adjustment, yielding 0.1413 W·m⁻²·ppb⁻¹. Atmospheric lifetimes of COF2 determined from kinetic decay profiles were 7.56 h, 36.67 min, and 54.86 min for dry synthetic air (O2-only), high-humidity, and low-humidity conditions, respectively, corresponding to GWP100 values of 0.1018, 0.0082, and 0.0117, respectively. Accordingly, in moist tropospheric air, COF2 exhibited GWP100 <1. These results demonstrate that water vapor-driven hydrolysis overwhelmingly governs COF2 removal in the atmosphere, leading to a substantially shorter lifetime and far lower GWP than conventional F-gases. Furthermore, since CO2 is the confirmed terminal degradation product, the ultimate climate impact of COF2 is equivalent to that of CO2 on a molar basis. This study presents one of the most comprehensive experimental analyses of COF2 and offers a robust evaluation of its GWP and its potential as a sustainable alternative for reducing the climate footprint of semiconductor and display manufacturing processes.
On one level this is a very nice systematic study of the GWP due to emissions of carbonyl fluoride. It is very clearly written and well presented. To my knowledge, no other paper has presented a GWP for CF2O emissions, and so this part of the paper is clearly original and worthy of publication. The overall conclusion that the GWP is very low, is likely robust for all conditions.
Unfortunately, the paper has a significant flaw. It fails to connect with the previous literature and so confuses what is already known about CF2O with what is original in this paper. In places it also presents information (e.g. on CH2O) that is not relevant to the paper and I believe it can be significantly shortened while retaining the key messages.
I recommend that the paper needs a major revision to clearly separate out what is already known about CF2O, the extent to which the new observations and calculations support/challenge what is already in the earlier literature, and what is truly original. The paper also lacks any uncertainty budget for the measurements and often presents values with an inappropriate level of precision.
Major comments
Other comments – those that start with * are particularly important.
18: “kinetic decay” – in atmospheric science it is more common to call this the e-folding lifetime
*32-34: This definition is incomplete – it is the time-integrated radiative forcing of a 1 kg pulse emission at time t=0 compared to the same mass emission of CO2
58-61: Very minor, but I am not sure this text is needed. The ACS is the prime property.
77: No explanation is given for the lower wavenumber (649 cm-1) lower limit. Presumably determined by the infrared detector?
*79: No source is given for these CO2 reference values and the specification of a single lifetime for CO2 is inappropriate for reasons explained in all the IPCC WG1 assessments. I recommend that the authors simply adopt and cite the IPCC AR6 AGWP(100) for CO2 which is in the supplementary information for https://doi.org/10.1017/9781009157896.009 at https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter07_SM.pdf
88: “potent” – do you mean in terms of radiative efficiency?
91: “emphasize the need” – this isn’t quite accurate. These protocols, amendments and agreements were posed in terms of CO2-equivalence (and hence why GWP-100 is widely used to provide that equivalence). They did not specifically focus on the need to reduce emissions of particular gases, but provided signatories with the option of doing so. It can also be noted that the Kigali Amendment to the Montreal Protocol https://ozone.unep.org/treaties/montreal-protocol/amendments/kigali-amendment-2016-amendment-montreal-protocol-agreed is perhaps also relevant, although not specifically relevant to the gases mentioned on line 85.
98: “moderately toxic” – all references that I looked at seem to say it is highly toxic. I am not in a position to judge.
110: The wavenumber range of the measurements needs to be stated. Presumably this is determined by the detector?
*157-163: Although quite interesting, no motivation is given for presenting these DFT calculations. Why are they needed, when you have measurements and, in terms of integrated cross-sections, how do they compare with the measurements? If it is to allow attribution of bands to specific modes of vibration, that is interesting, but I think this has already been established in the earlier literature – for example, see the papers that are referred to by the HITRAN database for CF2O. So, again, more references to the older literature are necessary and the text can be shortened so that it doesn’t appear as if the results presented here are original.
*168-181: Again, this section is presented without any references to earlier work, implying that this is new knowledge. As noted above, the molecular structure presented here is very little different to the CF2O entry on Wikipedia. I am not sure why the authors choose to compare with formaldehyde. This seems to lengthen the paper unnecessarily and can be removed.
Figure 2a: Although I believe Figure 2a is not necessary, note that the angles in the figure and in the caption are slightly different. There is also no need, in my view, to repeat information that is in the figure in the caption.
*Figures 2b and 3b: the authors need to revisit how that they handle the low wavenumber noise in the FTIR spectra. The fact that the noise sends the ACS negative is completely unphysical but it is not recommended that these are set to zero whilst retaining positive noise, as this introduces a bias. Many studies simply exclude regions between bands (i.e. set the ACS to zero) where the signal to noise ratio is too low.
184-189: Some of this discussion repeats information provided already in Section 2.
*190-204: Again results are presented without a single reference to previous work in this area. See my comment on 157-163 for where some of earlier papers can be found.
*183: This section needs to report the integrated absorption cross-section and to discuss uncertainties in the final value.
*217: Is 4 decimal places justified? How does this value compare with previous literature and what are possible reasons for the difference? And what is the estimated uncertainty in the IAC and RE.
*218: “measurable contribution to RF”. Without estimates or measurements of the abundance of C2FO it is not possible to make this statement, and I doubt if the RF can be measured as it is so small. The statement should either be better justified, or the authors should stick to RE, which is what they calculate and present.
*221-273: Again there is not a single reference to the older literature in the whole of this section, and so the impression is given that this is all new. I believe that the reactions given in Scheme 1 are already well established.
226: “negligible”. I am not sure what this means. Its presence is very clear in Figure 4a, but the reader is not told this.
*Table 2: Some values are quoted to 6 significant figures (see also other tables in the paper) without any accompanying uncertainty estimate.
*225: More care is needed about such statements as “typical outdoor conditions”. While they may be correct for near-surface conditions where the measurements were made, they may not be appropriate elsewhere. Despite the short lifetimes found in this study, it may be possible for convection to transport some surface emitted C2FO to reach high altitudes and hence lower humidities. Similarly, emissions in high latitudes may not experience such moist conditions. The authors could simply point out that these lifetimes are appropriate to the conditions in which measurements are made, but they may not be appropriate for all locations.
*275: The reader should be reminded that these estimates of RE and GWP use the assumption that CF2O is well-mixed, which is highly unlikely given the short lifetime. Although not fully appropriate to the destruction processes for CF2O, the Hodnebrog papers present simple methods to adjust the RE (and hence GWP) for gas lifetime. Consequently, the values presented here are likely overestimates. In addition, it should be noted that the GWP for very short lived species is dependent on both the location and time of year of the emissions. A short statement on this may be useful.
299-291: This statement is not correct for the AGWP of CO2.
Supplementary Material: I feel that all the information concerning COH2 should be removed, as its relevance to this study is never established. It is also presented without any reference to the older literature which is extensive for this molecule.