the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Jul 2026
Towards quantifying contrails of fuel cell powered aircraft: first in-flight measurements of a fuel cell exhaust emulator
Abstract. Aviation forces the climate via multiple pathways: The emission of carbon dioxide (CO2) from conventional propulsion technologies has a warming effect, but also non-CO2 effects like contrail cirrus play a crucial role. For new hydrogen (H2) based propulsion concepts there are no direct CO2 and soot emissions, but higher water vapor emissions, which can affect the formation, micro-physical properties and the climate impact of contrails. In-flight measurements of relevant contrail and emission parameters of fuel cell propulsion systems are non-existent. In order to study contrail formation in the wake of fuel cell powered aircraft, a fuel cell exhaust emulator (FCEE) was designed by Airbus to produce the low temperature and high humidity exhaust conditions expected from these systems. Due to their unique exhaust composition and thermodynamic emission characteristic, processes such as plume mixing with ambient air are crucial to understand contrail formation. Further, to compare engine and fuel-dependent contrail properties, in situ methods derive a dilution corrected apparent ice emission index (AEI) using conservative tracers, like CO2 or nitrogen oxides (NOX). As the FCEE emits no such inert trace gas, a newly developed artificial dilution tracer system was used. The design and characterization of this system, as well as the significance and application for exhaust dilution and emission index assessment of fuel cell-powered aircraft in flight, are described.
As part of the Nephele flight test campaign, in December 2023 in Minnesota, USA, the prototype FCEE and artificial dilution tracer release system were installed on board a Convair C131. The sampling system was installed on a Piper Cheyenne 400LS chaser aircraft, together with instrumentation for emission and contrail measurement, both aircraft operated by AV Experts LLC. A proof-of-concept of the artificial tracer system was demonstrated in contrail and emission flights. Contrails formed on aerosol particles emitted by the FCEE near the homogeneous freezing threshold (at ambient temperatures around 230 K) were sampled at 400 hPa in distances between 50 and 1500 m behind the aircraft. An increase of AEI from 1.4·1013 to 4·1015 kg-1H2 with increasing temperature difference of 3 to 5 K below the homogeneous freezing temperature was observed. As contrails from fuel cell systems already form at higher ambient temperatures compared to conventional propulsion systems, these are the first contrail measurements in this temperature range. Complementary model simulations suggest that droplet freezing is responsible for the evolution of ice crystal numbers in this temperature range.
Competing interests: J.C., F.T., C.M., F.L. and C.R. are employed by Airbus. R.V. and A.V. are employed by AV Experts LLC. All the other authors declare that they have no conflict of interest.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(6681 KB) - Metadata XML
-
Supplement
(80807 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2887', Anonymous Referee #1, 08 Sep 2026
-
RC2: 'Comment on egusphere-2026-2887', Anonymous Referee #2, 29 Sep 2026
General comments:
This paper considers how exhaust conditions which may be typical of future fuel cell powered aircraft affect the formation of contrails from such aircraft. Direct atmospheric measurements were undertaken through flight campaigns where a chase aircraft sampled the exhaust plume of an aircraft which was being studied. It is a scientifically robust and valuable manuscript.
There are two key novel aspects in this work. Firstly, the development and evaluation of an artificial dilution tracer system (ATReUS), which will be key to measuring the emission index (EI) from flights which do not produce CO2. The authors use N2O and conscientiously assess the potential climate implications of this tracer.
Secondly, the authors studied how the apparent emission index of ice (AEI) and the size of the ice crystals depends on the ambient conditions for temperatures close to the homogeneous freezing temperature (Thf) in the case of high humidity and low temperature exhaust provided by their fuel cell exhaust emulator, concluding that small changes in ambient conditions near Thf can lead to vastly different microphysical properties of the contrail, although experimental results around this point are limited.
My main concerns revolve around the instrumentation and particle characterisation, as well as assumptions made while assessing the microphysical properties of the contrail.
Specific comments:
Section 2 (the development of ATReUS):
In this section, the authors do not describe the sampling system fully. As a result, the reader is left without knowledge of any of the potential losses or interactions within the system. Some more details would be worthwhile.
The EI calculation is clearly laid out. As this manuscript aims to correct to H2 burnt, it is important to lay out how the effective work performed by a unit mass of kerosene compares to that performed by a unit mass of H2. This would to allow more direct comparison between fuel cells and current engines. In fact, the authors allude to this (l215-217), but I don’t see that such a comparison was actually made.
The environmental assessment in section 2.5 is very valuable, with the conclusion made that a CO2e of 3140 kg resulted from the use of N2O. I would like to see that put into context with an estimated amount of CO2 from the fuel burnt during the campaign.
The authors note that N2O, the tracer gas they have used, can be ‘produced in small quantities during combustion processes (Santoni et al 2011)’. This should come with a value so that the reader doesn’t need to delve into the paper to understand this fully. A similar criticism can be made of many comparisons throughout the manuscript which are begun but not brought to their conclusion.
Section 3:
Instrumentation:
The description of the instrumentation in Table 1 is incomplete. For some instruments, the range and the uncertainty are quoted (e.g. CAPS PT100) whereas others do not have a range or an uncertainty at all (aMCPC). This should be completed. In the text (l103-105), the LI-7820 is described as measuring both N2O and H2O. However, in Table 1, only its N2O measurement ability is described. Likewise for the LI-7000 (l 314).
Furthermore, the SEMS is quoted as measuring the particle size distribution (PSD) between 5-350 nm, but as this must be coupled with an aMCPC to produce the PSD, the SEMS cannot measure particles below 7 nm.
The table also suggests that the aMCPC alone can segregate the nvPM counts from the vPM counts. I suggest that the thermodenuder should be included in the table for greater clarity rather than as a footnote. There should also be additional details about the thermodenuder; if this does not include a catalytic element, it may not be removing the volatile particles in the way that is expected. In some cases, a thermodenuder could potentially even create new particles <10 nm from the evaporated volatile materials (Amanatidis et al, 2018). This needs to be addressed or discussed.
The aMCPC has a cutoff diameter of 7 nm. Yet, with the close chase distances, volatile particles may not have grown large enough to be observed at 7 nm, so there should be some discussion of the growth time of the volatiles, particularly as it seems that the authors suspect that many of the particles present could be oil. That being said, this attribution to oil assessment does not seem satisfactory as the nvPM concentrations are high (fig. 7). Additionally, it seems as though the particle concentrations in the FCEE were not measured during the relevant flights, but only later (l480). The assumption that the concentration was the same during this later flight and the earlier flights is not necessarily correct. Overall, a discussion of the particle concentrations and their nature should be frontloaded and condensed into one area rather than requiring the reader to piece it together.
The particle measurement payload may not be well adapted to future measurements for hydrogen engines where it is likely that small volatile particles play the largest role in contrail formation due to this 7 nm cutoff.
Fig 6, right: both ends of the temperature scale are impossible to resolve. Why is it not resolved? If this is impossible, would it not be more beneficial to have this as >241 K, <230 K? In the text you say you account for uncertainties by relaxing the RHi criterion to 90%. Does this mean that the uncertainty is typically biased in a certain direction? This merits a clarifying sentence or two.
How long must a contrail survive for you to qualify it as persistent? I think this should be more clearly defined as this can mean different things depending on the precise research field.
Fuel cell exhaust emulator:
The FCEE is innovative and a good tool to study the effect of future propulsion technologies on contrail formation, provided issues can be addressed. Most important is the high concentration of particles in the exhaust which is unlikely to be present in a true fuel cell experiment. The fact that these particles were not fully characterised or identified serves to exacerbate the problem. As it stands, this is inadequately discussed.
Some aspects of the FCEE could be expanded further, particularly how it is ensured that 100% of the water droplets evaporate. If any droplets remain at the exhaust plane, these could strongly influence contrail formation.
The statement on deployment is confusing. L359 refers to the system only being deployed once, however the manuscript later refers to 3 test flights where the FCEE appears to be deployed.
Values of the humidity in the fuel cell exhaust are only given much later in the manuscript. It would be clearer to have them here as well.
Similarly, to improve clarity, the authors could consider the units and standards they use, for example at lines 146-147, gas flows are quoted in both kg/s and slpm. Shortly after, the reader must compare a nitrogen equivalent mass flow to an N2O mass flow, which leads to some of the values initially seeming inconsistent.
The description of the variables in Eq. 6 and 7 (l 239 – 245) is unclear as the order in the text does not appear to be the same as the order in which the variables appear in the equation.
Section 4:
The values quoted l 409 cannot be found in fig. 7. I believe this is because they were measured in different flight, but if so, this should be specified to prevent confusion. L412, 90% of the maximum N2O flow rate is referred to with a different value to that previously (9.585 slpm v 9.33 slpm on l 156). Again, this removes clarity and confidence.
The dilution and mixing of the plumes are very valuable measurements to have come from this work, as they give an idea of how exhaust plumes spread out behind an aircraft. This warrants more discussion.
The final sections in fig. 7 show high nvPM, but low ice. I think this is because the gasoline engine was sampled, with lower humidity and therefore less favourable conditions for contrail formation, but this should be laid out explicitly, perhaps even within the figure.
The models are used before they are introduced and therefore the reader does not really understand them when it is relevant. For example, it is not clear what the water droplets are forming on when they are allowed to form, or if they result from homogeneous nucleation (l459). How the peak supersaturation is actually obtained (why does it start decreasing when it is shown to in fig. 9?) is also unclear. A more in depth and earlier discussion of the model, including details on the dependence of Thf on particle size and solubility and any quenching of the supersaturation is needed.
The discussion of the peak supersaturation in these models is also unsatisfactory; are these values likely to truly be reached when particles are present, activating and thus quenching the supersaturation?
The calculation of the fraction of frozen droplets relies on an assumption that the maximum number of ice crystals corresponds to 100% frozen fraction. This assumption should be more clearly validated. I would also like this treatment to be performed with respect to the number of particles sampled in the exhaust rather than just water droplets to understand how the activation conditions required for aircraft emissions to activate. I think it a shortcoming of this paper that needs to be addressed for future studies that you have so little knowledge about the particulate emissions from the FCEE.
The behaviour measured near Thf does not closely follow the model, partly due to the large uncertainties on the measurements, but it seems as though the increase in AEI only occurs around ΔThf > 4.5 Κ. Does this agreement get better with higher aerosol loadings in the model as the trend seems to go that direction? There are also very few points in the model showing this increase (only one or two before reaching a relative plateau). The model data should be extended to ΔThf = 0 K to make this relationship clearer and could also include more aerosol loading values (e.g. the 3e6 cm-3 that was established as the minimum aerosol loading).
l 618 refers to vertical colored lines in fig. 11d, but I do not see these.
The final statement l 625 is phrased too strongly, although I agree with it. The model run with and without HDN, and in the absence of measurements directly at the exhaust plane we cannot strictly know what supersaturations were reached.
Appendix C:
This cross-contamination aspect was a key question in my mind throughout this manuscript. As such, I think this appendix should be in the main text. The final section of this appendix is worded confusingly, and the data do not seem to agree with the text. Fig. A1 does not show a difference in the distance at which the max ΔrN2O and ΔrCO2 were reached. Perhaps the use of different color scales hides this as it makes it more difficult to compare the two plots. The authors also state that the lowest cross-contamination was measured up to 58 m behind the C131, but there are almost no data shown at closer distances, so this conclusion cannot be supported.
Technical corrections:
Generally, many sentences in the manuscript are very long (e.g. l337-338, l 430-432), preventing clarity and marring the good science which was done. I would recommend the authors proofread the manuscript once more and try to address this.
At many points variables are referred to, but numbers are not included. Doing so would help prevent the reader from searching for those values later in the manuscript and make the discussion flow more clearly (e.g. l363 – 365, what are the number of particles, the humidity and the temperature?).
At times, abbreviations for quantities are introduced, but these are then not always used (e.g. RHw, RHi, Thf) leading to longer and less clear sentence structure.
Comparisons to literature are often not brought to their conclusion, with values or the conclusion of the comparison left out, requiring the reader to finish making the comparison themselves, and potentially coming to the wrong conclusion. This should be corrected (e.g l392-393 comparison to ERA5, l. 456 does the exponent increase or decrease with jet size, l 563)
The authors could consider using AEIice for greater clarity. This would also match Eq. 2 better.
L 75: was lead -> was led
L 182: did not effect -> did not affect
L230 done, by -> done by
L 258: 2023 lead by -> 2023 led by
L288: is utilizing -> utilizes
L337-338 I do not understand
L370: saturation versus liquid wate -> saturation with respect to liquid water
L378 Global Forcast System -> Global Forecast System
L397: is analyzed -> are analyzed
l 401: exemplary -> example
References
Amanatidis et al, Aerosol Science and Technology, 2018: https://doi.org/10.1080/02786826.2017.1422236
Citation: https://doi.org/10.5194/egusphere-2026-2887-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 300 | 101 | 43 | 444 | 58 | 43 | 33 |
- HTML: 300
- PDF: 101
- XML: 43
- Total: 444
- Supplement: 58
- BibTeX: 43
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This paper deals with a concept to mimic hydrogen combustion engine and with a proposal for a tracer allowing dilution factor determination. Authors bring proposals to investigate airbone emission without a real engine.
Hydrogen engine substitute:
Technology use is smart and simple. Well describe in the manuscript. Appreciated the aim to emit the right amount of water vapor. Nevertheless, using non-filter air coming from aircraft engine bring a non negligeable amount of particles that allow contrail formation (ice nuclei) that should not appear with real hydrogen burning engine. Moreover, present of oil bring as well ice nuclei that can modified physico-chemistry. This concept is interesting but need few upgrade to help to assess impact of hydrogen engine
Tracer for dilution:
Having a tracer for hydrogen combustion emission is a great challenge. The proposal of this paper offers a clever approach. Authors well describe the process, the potential drawbacks (including environmental greenhouse price) and well evaluate evolution with a crossing experiment using CO2 of the first aircraft engine. Calibration and limit of dection of the Licor device is well manage considering airborne using. Into the paper, there is a strong explenation about all calculation and validation of determine dilution factors.
Airbone set-up
The suit of devices used are a little limited . For exemple, particle counters have a cutof at 7nm and may miss the main volatile particle matter part when they are close to the first aircraft. It is true that with the amount of particles coming from the emission, nucleation/adspotion will occur fast and the cut oof don't need to be small but with a real hydrogen experiment, it will be a must to have the possibility to assess volatile particles parts.
Using a thermodenuder can be not enough to remove volatile part and as the non-volatile particle matter measurement requierement from SAE E31, a catalytic stripper may be better.
Regarding best practice on particle measurement, the article lack description or at least reference about process to assess and correct particle lost into the sampling measurement system.
Airbone measurement
Authors bring some results of flight measurement at the end of the paper. Understanding that the main contribution of the article is FCEE and the tracer, some more input of experiment will be appreciate as do contrail formed persist or not? size of particles and ice cristal. Perhaps a second article is under writing.
Conclusion
This article provide more than a proof of concept of a device that aims to substitut full hydrogen engine. It can help to perform basic experiment but with limitations. The tracer approach can be develop on flight measurement on emission without CO2.
Authors provide clear description of FCEE and N2O tracer, more information can be provide on chasing aircraft measurement kit regarding last state of art on particle measurements