| 27 Feb 2026
Continuous analysis of N2O isotopic composition during biological nitrogen removal in wastewater treatment to disentangle production and reduction processes
Abstract. Nitrous oxide (N2O) is a potent greenhouse gas, and emissions from wastewater treatment plants (WWTPs) represent a significant and highly variable source. Understanding the dynamics in microbial pathways of N2O formation and reduction during biological nitrogen removal is essential for targeted mitigation strategies. Stable isotope analysis of N2O (δ15Nα, δ15Nβ, δ18O, and 15N site preference) provides a powerful tool to disentangle and quantify N2O production and reduction processes, yet conventional analytical approaches lack temporal resolution. Here, we present the first long-term application of an off-axis integrated cavity output spectrometer for real-time N2O isotopic analysis at a pilot-scale WWTP over one year of operation. We developed a dynamic dilution system and implemented correction protocols for drift, N2O mole fraction dependence, and gas matrix effects on isotopic results, achieving uncertainties of 0.85 ‰ (δ15Nα), 1.08 ‰ (δ15Nβ), 0.81 ‰ (δ15Nbulk), 0.48 ‰ (δ18O) and 1.09 ‰ (15N site preference). Representative datasets demonstrate the system’s capability to (i) identify dominant N2O production pathways under standard operation, (ii) quantify N2O reduction in relation to dissolved oxygen concentration, and (iii) trace nitrogen transformation during low-level 15N-labelling experiments. Our results indicate nitrifier or heterotrophic denitrification as the main source of N2O, and that N2O reduction efficiency is strongly controlled by oxygen availability. This study highlights the potential of laser spectroscopy for continuous isotopic monitoring in real-world engineered systems and provides practical guidelines for uncertainty reduction and data interpretation. More specifically, our work forms a foundation for further investigations of the operational factors controlling N2O formation and N2O reduction in biological WWTPs and other complex anthropogenically-perturbed settings.
General comments
Keck et al. report methodological advances in monitoring isotope ratios of N2O emitted from waste water treatment processes. They conducted a detailed investigation of the dependence of N₂O isotope ratios on N₂O and oxygen concentrations and long-term stability of the measurements, and established an automated method for obtaining accurate analytical values using infrared absorption spectroscopy for sources with high variability.
The strength of this paper is that the developed system allows for frequent measurement of N₂O isotope ratios at the source site. This is particularly useful for assessing anthropogenic sources that fluctuate significantly over time. I hope that this method will provide information on the hour-by-hour processes of N₂O formation and dissipation, thereby contributing to proposals aimed at reducing N₂O emissions from various N2O sources including WWTP.
However, I am concerned that the authors have made little mention of the shortcomings and limitations of this method compared to conventional mass-spectroscopic or infrared-absorption-spectroscopic measurements using an intermittent gas sampling method. The following disadvantages can be cited. First, the system requires a large amount of dilution gases and calibration gases to maintain constant N2O concentration of the analyte gas fed to the laser spectrometer and to correct for temporal drift of the spectrometer responses, respectively. Although the cost of synthetic air to dilute raw sample gases may be low, it is necessary to install pure air cylinders or pure air generators at the site, and one has to consider logistics. Preparation of calibration standards will likely require both money and effort, as some of the authors have already reported. Second, as stated in conclusion, the developed system actually allows three 5-minute averaged measurements per hour. I expect the authors’ further efforts to improve the measurement frequency, but at this stage, I think it would be an exaggeration to state that this method allows “continuous analysis” of N2O in source gases.
In summary, this paper is worth publishing in Atmospheric Measurement Techniques after revision regarding above points and minor comments below.
Specific comments
Title. “continuous” would be better replaced with “high-frequency” or “on-line”.
L62. There is a paper (Toyoda et al., 2011) published earlier than Wunderlin et al. (2013) that reported isotope ratios of N2O emitted from WWTP.
L74 & L80. “resN2O” and “redN2O” look very similar and are easy to confuse. I suggest the authors to use a single character (with sub/super scripts, if any) to represent a variable. Please also refer to this journal’s submission guidelines.
L120. From this point onward, measurements have been conducted with the N₂O concentration set to 12 ppm. Please explain why this value was chosen—was it set to match the lowest expected N₂O concentration in WWTP emissions, or is there another reason?
L137. Although two-point calibration was conducted, neither the isotopic values of N2O in target gas nor sample gas fall within the range of the two standard gas values. This means the authors are extrapolating; please explain whether this is acceptable and justify the validity of this approach.
L150. Do “blocks” mean the four dates described in the next line?
L159. It would be helpful for readers unfamiliar with this method if the authors explain following points in the introduction or discussion section: the reasons and principles behind why isotope ratio measurements using infrared absorption spectroscopy depend so heavily on concentration of target gas and coexisting gases.
L168–169. It is unclear how the authors are using the term “strong”. Does it mean that the slope of delta value versus N2O (d(delta)/d(N2O)) is large? If so, it would be better to show the concentration ranges lower than 6 ppm more closely, as in the case of 8-16 ppm range. The horizontal scale of Figure 2 is too coarse, making it difficult to read the data. Simply adding a secondary scale would help. Also, because the inset figure of 8-16 ppm range hides the plots of d15Nbulk, d18O, and SP at >50 ppm, one cannot examine the relationship for high concentration ranges.
L178. I think air or oxygen is injected not only to oxidize ammonium to nitrite, but also oxidize organics using microorganisms.
L213–214. Notation of variables are not fully explained and are partly inconsistent or confusing. For example, is the integral sign attached to delta_cal1 in line 213 the same as “Int” attached to the parameters in equation 3? Moreover, eq. 3 itself is not correct if I follow the explanation described in L211–213. It seems that the coefficients of the time-difference terms in the denominator are reversed.
L215. I think subscript “sa” in equation 4 denotes sample, but the authors have already used “SA” to mean synthetic air. To avoid confusion, replace with other expression.
L237–238. I think estimating the isotope ratios at the time of formation based on the distribution of observed values is a second-best approach, but it is not clearly explained how the reduction line was estimated. To calculate the “perpendicular distance” of the data point and the reduction line, the latter must be estimated independently with the observation. While the slope can be determined using literature values, how was it decided which points on the graph the line passes through (or what the y-intercept should be)?
L242. I agree that d18O of N2O is related to d18O of water, but do the authors argue that N2O and water is under full isotopic equilibrium? I wonder the changes in relative contributions from Hy and nD/hD or the changes in dissolved oxygen isotope ratios due to partial consumption might affect d18O of N2O or precursors like NO2-.
L248. Does “repeatability of the instrument” include uncertainty of drift correction?
L260. It would be helpful if the authors explain derivation of equation 9. What does “n” mean?
Figure 4. I would like to confirm the accuracy of the description in this figure regarding the gas sampling method from the reactors. A portion of the water’s surface is covered, and something resembling a chimney is depicted. Is the gas being actively drawn from the water surface, or is it passively carried by the airflow generated by aeration? This question concerns whether the representativeness of the collected gas is ensured.
L380–382. It seems the authors assume that isotopic signature of produced N2O is constant over time and that the variation in dO and SP is simply caused by difference in the progress of N2O reduction. This assumption should be clearly stated along with its basis, together with the reason why the slope of the line in Figure 6 is different from that in Figure 7 or Figure A1. To my eyes, variation around the regression line seems to indicate variation in the produced N2O.
L391–393. This statement is difficult to confirm looking at Figure 7B, mainly because the color gradients corresponding to the time are difficult to distinguish.
L417–421. Although the caption of Figure 8 mentions grey dots representing the results from standard operation, I cannot see the data points. If I estimate the d15Nbulk values for the standard operation of ca. -50‰ from Figure 5, d15Nbulk obtained for Cycle 1 here is higher by 40-60‰. Assuming the d15N of NH4 during the standard operation of 6-24‰ based on literature (Toyoda et al. 2011), the enhancement of d15N of N2O due to the labeled NH4 is significantly smaller than the difference in d15N of NH4+ (100 – 6 or 24 = 94 to 76‰). If the measured d15N of N2O here is correct, what would be the reason for the discrepancy?
P28. Fix the citation of Morley et al..
Figure A1. I can see only a single “blue area”. If the authors really plotted both light blue and dark blue areas, please differentiate them by improving the color contrast.
References
Toyoda, S., Suzuki, Y., Hattori, S., Yamada, K., Fujii, A., Yoshida, N., et al. (2011). Isotopomer analysis of production and consumption mechanisms of N2O and CH4 in an advanced wastewater treatment system. Environmental Science and Technology, 45, 917–922. https://doi.org/10.1021/es102985u