Supplement of Atmos. Meas. Tech., 13, 2797–2831, 2020 https://doi.org/10.5194/amt-13-2797-2020-supplement © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. Supplement of N2O isotopocule measurements using laser spectroscopy: analyzer characterization and intercomparison Stephen J. Harris et al. Correspondence to: Joachim Mohn (joachim.mohn@empa.ch) The copyright of individual parts of the supplement might differ from the CC BY 4.0 License. 5 Section S1 – IRMS Methodology 10 IRMS analyses were conducted at ETH Zürich using a gas preparation unit (Trace Gas, Elementar, Manchester, UK) coupled to an IsoPrime100 IRMS (Elementar, Manchester, UK). The gas preparation unit was modified with an additional chemical trap (0.5 in diameter stainless steel), located immediately downstream from the autosampler to pre-condition samples. This pre-trap was filled before each run with new NaOH, Mg(ClO4)2, and activated carbon in the direction of flow and was designed as a first step to 15 scrub CO2 and H2O. After pre-scrubbing, the samples were purged through a second set of chemical traps (NaOH, Mg(ClO4)2) before cryogenic trapping and focusing in liquid N2. Before final injection into the IRMS, the purified gas sample was directed through a permeation drier and subsequently separated in a gas chromatograph column (5 Å molecular sieve). The IRMS is equipped with five Faraday cups with m/z of 30, 31, 44, 45, 46, measuring δ15Nbulk and δ18O of N2O and δ15N of the NO+ molecule dissociated 20 from N2O during ionization in the source. The 15N/14N ratio of the NO molecule reflects the α (central) position N of the initial N2O, thus allowing the measurement of the site-specific isotopic composition of N2O. During each run three sets of two working standards (∼ 3 ppm N2O mixed in synthetic air) with different isotopic composition (δ15Nα = 0.954 ± 0.123 ‰ and 34.446 ± 0.179 ‰; δ15Nβ = 2.574 ± 0.086 ‰ and 35.98 ± 0.221 ‰; δ18O = 39.741 ± 0.051 ‰ and 38.527 ± 0.107 ‰) were included with a batch of 25 22 samples at the beginning, middle and end of each run. Sample peak ratios are initially reported against a N2O reference gas peak (100% N2O, Carbagas, Gümligen, Switzerland) and are subsequently corrected for drift and span using the working standards. Instrument linearity and stability was frequently checked by injection of 10 reference gas pulses of either varying or identical height respectively, with accepted levels of < 0.03 ‰ nA−1. Since instrument linearity could only be achieved for either N2O or NO, the 30 instrument was tuned for the former and δ15Nα non-linearity subsequently corrected using sample peak heights by determining non-linearity of δ15Nα with diluted working standards. 1 Section S2 – Analysis of high [N2O] reference gases, ambient reference gasses, PA1 and PA2 As detailed in Sect. 2.2.2, the isotopic composition of high [N2O] isotope reference gases in synthetic air (S1-a90ppm, S2-a90ppm) was analyzed in relation to N2O isotope standards (Cal1, 2 and 3) in the same gas 5 matrix (matrix a) using laser spectroscopy (CW-QC-TILDAS-200; ARI, Billerica, USA). Ambient mole fraction N2O isotope reference gases (S1-c330ppb, S2-c330ppb) and PA1 and PA2 were analyzed by TREX- QCLAS (Sect. 2.1.4) using N2O isotope standards (Cal1 – Cal5) shown in Table S2-1. Cal1 – Cal5 have been previously measured by Sakae Toyoda at Tokyo Institute of Technology.” 10 Table S2-1. N2O isotope standards (Cal1 – Cal5) used for the analysis of reference gases (S1, S2) and pressurized air (PA1, PA2). The standards (Cal1 – Cal5) used for analysis of the respective gases are indicated by a tick () N2O isotope δ15Nα vs δ15Nβ vs δ18O vs S1- S2- S1- S2- standard used for PA1 PA2 AIR-N2 [‰] AIR-N2 [‰] VSMOW [‰] a90ppm a90ppm c330ppb c330ppb calibration Cal1 in matrix a 2.06±0.05 1.98±0.20 36.12±0.32   Cal2 in matrix a -48.59±0.25 -46.11±0.43 27.37±0.11     Cal3 in matrix a 25.73±0.24 25.44±0.36 35.86±0.22   Cal4 in matrix a 16.29±0.07 -2.59±0.06 39.37±0.04   Cal5 in matrix a -51.09±0.07 -48.12±0.04 30.81±0.03   15 20 25 30 2 Section S3 – Experimental setups Figs. S3-1 to S3-10 depict the setup for the gas matrix, trace gas and two end-member mixing experiments undertaken in this study. Experiments were performed simultaneously for all analyzers, with the exception of the TREX-QCLAS, which requires an extensive measurement protocol to trap and measure 5 N2O (Ibraim et al., 2018) and thus could not be integrated concurrently with the other analyzers. Fig. S3-1. Experimental setup for the O2 dependence testing performed in Section 2.4.5. 10 Fig. S3-2. Experimental setup for the Ar dependence testing performed in Section 2.4.5. 3 Fig. S3-3. Experimental setup for the H2O dependence testing performed in Section 2.4.6. Fig. S3-4. Experimental setup for the CO2 dependence testing performed in Section 2.4.6. 4 Fig. S3-5. Experimental setup for the CH4 dependence testing performed in Section 2.4.6. 5 Fig. S3-6. Experimental setup for the CO dependence testing performed in Section 2.4.6. 5 Fig. S3-7. Experimental setup for the Ascarite and Sofnocat trap testing performed in Section 2.4.7. Fig. S3-8. Experimental setup for Experiments 1 and 2 performed in Section 2.4.8. 6 Fig. S3-9. Experimental setup for Experiments 3 and 4 performed in Section 2.4.8. 5 Fig. S3-10. Experimental setup for Experiments 5 and 6 performed in Section 2.4.8. 10 7 Section S4 – Complete datasets Due to the large number of results acquired in this Section, only selected results are shown in Figs. 3 to 14. The complete datasets (including [N O], δ15Nα, δ15 β2 N and δ18O acquired by all instruments tested) are shown below. 5 Fig. S4-1. Allan deviation (square root of Allan Variance) plots for the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green), QCLAS II (purple) and QCLAS III (brown) at different N2O mole fractions (~327, 1000 and 10000 ppb). The dashed lines represent a slope of -0.5 (log-log scale) and indicate the expected behavior for Gaussian white noise in each analyzer. The Allan deviations of all 10 analyzers tested were reproducible on three separate occasions prior to the test results presented here. 8 Fig. S4-2. Dependency of the measured [N 15 α2O], δ N , δ15Nβ and δ18O values on laboratory temperature (ºC) for OA-ICOS I (blue), CRDS I (red), CRDS II (black) and QCLAS I (green). The complete dataset is provided in Supplementary Material 4 (Fig. S4-2). The laboratory temperature is indicated by a solid 5 orange line and was allowed to vary over time. Cell temperatures for each instrument are also plotted for comparison. The analyzers began acquiring measurements at 00:00 on 8/07/2018, capturing the end of the rising limb of the laboratory temperature. Results are plotted as the deviation from the mean, without any anchoring to reference gases. 10 9 Fig. S4-3. Deviations of the measured δ15Nα, δ15Nβ and δ18O values according to 1/[N2O] for the OA-ICOS I (blue), CRDS I (red), CRDS II (black) and QCLAS I (green). Measurements span the manufacturer-specified operational ranges of the analyzers. The experiment was repeated on three separate days. A linear regression is indicated by the solid line, and a 5 residual plot is provided above each plot. Individual linear equations, coefficients of determination (r2) and p-values are indicated above each plot. 10 Fig. S4-4. Deviations of the measured [N 152O], δ Nα, δ15Nβ and δ18O values according to ΔO2 (%) at different N2O mole fractions (330, 660 and 990 ppb) for the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and TREX- QCLAS I (brown). The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. Data points represent the 5 mean and standard deviation (1σ) of triplicate measurements. Dependencies are best-described using linear regression, which are indicated by a solid line. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot for the 330 ppb N2O data only. 11 Fig. S4-5. Deviations of the measured [N2O], δ15Nα, δ15Nβ and δ18O values according to ΔAr (%) for the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and TREX-QCLAS I (brown). Data points represent the mean and standard deviation (1σ) of triplicate measurements. Dependencies are best-described by polynomial fits, which are 5 indicated by solid lines. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot. 12 Fig. S4-6. Deviations of the measured [N O], δ152 Nα, δ15Nβ and δ18O values according to ΔCO2 (ppm) at different N2O mole fractions (330, 660 and 990 ppb) for the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and TREX- QCLAS I (brown). The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. Data points represent the 5 mean and standard deviation (1σ) of triplicate measurements. Dependencies are best-described by linear fits, which are indicated by solid lines. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot for the 330 ppb N2O data only. 13 Fig. S4-7. Deviations of the measured [N 15 α 15 β2O], δ N , δ N and δ18O values according to ΔCH4 (ppm) at different N2O mole fractions (330, 660 and 990 ppb) for the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and TREX- QCLAS I (brown). Data points represent the mean and standard deviation (1σ) of triplicate measurements. Dependencies 5 are best-described by linear fits, which are indicated by solid lines. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot for the 330 ppb N2O data only. 14 Fig. S4-8. Deviations of the measured [N O], δ152 Nα, δ15Nβ and δ18O values according to ΔCO (ppm) at different N2O mole fractions (330, 660 and 990 ppb) for OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and TREX- QCLAS I (brown). The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. Data points represent the 5 mean and standard deviation (1σ) of triplicate measurements. Dependencies are best-described by linear fits, which are indicated by solid lines. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot for the 330 ppb N2O data only. 15 Fig. S4-9. Deviations of the measured [N O], δ15Nα2 , δ15Nβ and δ18O values according to ΔH2O (ppm) for OA-ICOS I (blue), CRDS I (red), CRDS II (black) and QCLAS I (green). The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. Data points represent the mean and standard deviation (1σ) of triplicate measurements. Dependencies are 5 best-described by either linear or polynomial fits, which are indicated by solid lines. Individual equations, coefficients of determination (r2) and p-values are indicated above each plot. 16 Fig. S4-10. Correlation diagrams for [N O], δ15Nα, δ15Nβ, δ15Nbulk2 , SP and δ18O measurements at various ΔN2O mole fractions analyzed by the OA-ICOS I plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed blue line represents a linear fit to the data. Individual 5 equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. 17 Fig. S4-11. Correlation diagrams for [N O], δ15Nα, δ15Nβ, δ15Nbulk2 , SP and δ18O measurements at various ΔN2O mole fractions analyzed by the CRDS I plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed red line represents a linear fit to the data. Individual 5 equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. 18 Fig. S4-12. Correlation diagrams for [N O], δ15Nα, δ15Nβ, δ15Nbulk2 , SP and δ18O measurements at various ΔN2O mole fractions analyzed by the CRDS II plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed black line represents a linear fit to the data. Individual 5 equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. 19 Fig. S4-13. Correlation diagrams for N2O mole fractions, δ15Nα, δ15Nβ, δ15Nbulk and SP measurements at various ΔN2O mole fractions analyzed by the QCLAS I plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed green line represents a linear fit to the data. 5 Individual equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. Results for Exp. 5-6 are highlighted in red, with the dashed red line indicating a linear fit to this data. 20 Fig. S4-14. Correlation diagrams for [N2O], δ15Nα, δ15Nβ, δ15Nbulk, SP and δ18O measurements at various ΔN2O mole fractions analyzed by the TREX-QCLAS I plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed green line represents a linear fit to the data. 5 Individual equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. 21 Fig. S4-15. Δδ15Nα, Δδ15Nβ, Δδ15Nbulk, ΔSP and Δδ18O (EstimatedSource – TrueSource) values derived from the OA-ICOS I (blue), CRDS I (red), CRDS II (black), QCLAS I (green) and IRMS (purple) via Keeling analysis of the two end-member 5 mixing scenario. EstimatedSource = TrueSource is indicated by a solid black line at y = 0, and the dotted lines indicated ± 2‰ deviation from y = 0. The change in concentration exceeding that of the background gas is indicated for experiments 1-2 (ΔN2O = ~30 ppb), 3-4 (ΔN2O = ~700 ppb) and 5-6 (ΔN2O = ~10000 ppb). Note: the QCLAS I results for experiments 1 and 2 are not depicted to maintain clarity, as they exceed the selected y-axis scale. 22 Section S5 – Application of an automatic spectral correction method for QCLAS measurements The primary cause of the observed excess drift in QCLAS I isotopocule measurements was fluctuating spectral baseline structure. That baseline structure, at an absorbance level of ~5 × 10-5 , was likely due to 5 interference fringes that were enhanced by somewhat dirty absorption cell mirrors (pers. comm. Aerodyne Research Inc.). One way to reduce the effects of excess baseline structure is to employ the instrument operation mode “Automatic Background”, wherein a spectrum is recorded while zero-air is injected, and subsequent spectra are divided by the recorded background spectrum. Periodic zero-air injections with refreshed background spectra can reduce the effect of changing baseline structure. Aerodyne recommends 10 that “Automatic Background” be used in measurements with very weak absorptions, such as in the measurements reported in this paper. However, Aerodyne Research Inc. recognizes that frequent backgrounds may not be convenient or practical in all circumstances. Also, instrument operators may not observe a gradual increase in baseline structure over time. 15 Recently, Aerodyne has been developing a new method of correcting data that is influenced by changing baseline structure. In this new method, injections of zero air and calibrations may be less frequent than would otherwise be necessary. An outline of the correction method provided by Aerodyne Research Inc. is as follows. Mixing ratio (MR) and spectral data are grouped according to type: background gas, calibration gas (Cal), and air sample measurements. Averages for the MR’s and spectra for the groups are 20 subtracted. An aggregate subtracted spectrum array is smoothed to the absorption linewidth. A principal component analysis [PCA] is performed, so that spectral fluctuations are represented as a sum of vectors, Vi(x), times amplitude histories, Ai(t). For the “Cal” segments, we fit the recorded MR’s with a sub-set of {Ai(t)}. If the specific linear combination of {Ai(t)} from the MR fit well-represent the “Cal” segments, we apply that linear combination of {Ai(t)} to the full measured data-set to determine MR adjustments. 25 In the Allan variance experiments performed on QCLAS I in this study (Sect. 3.1), the instruments were continuously sampling from a tank, so fictive hourly calibration periods (5 m duration) were created and used to generate mixing-ratio corrections. This method is computationally rapid, as ~8 hr of MR data can 23 be adjusted in seconds. A summary of the corrected data is provided in Table S5-1. This new method is not yet published or broadly verified, but Aerodyne Research Inc. intends, after further validation, to share the methodology with the measurement community. 5 Table S5-1. Summary of the corrected QCLAS I Allan variance data acquired in Sect. 3.1. 1σ data refers to Allan deviation (square root of Allan variance). N2O [ppb] δ15Nα [‰] δ15Nβ [‰] N2O [ppb] 1σ (1s) 1σ (300s) 1σ (600s) 1σ (1s) 1σ (300s) 1σ (600s) 1σ (1s) 1σ (300s) 1σ (600s) 326.5 0.062 0.021 0.024 1.2 0.39 0.37 1.7 0.42 0.55 1000 0.165 0.14 0.10 1.4 0.19 0.23 1.4 0.20 0.22 10000 2.9 0.46 0.38 0.33 0.027 0.029 0.25 0.024 0.028 10 15 20 24 Section S6 – Short-term repeatability To test the short-term repeatability of the instruments, sample gas at different at different [N2O]: ambient (PA1), 1 ppm , 10 ppm N2O was repeatedly analyzed 10 times for 15 min, intercepted by dry ambient air for 5 min. Gas mixtures with 1 and 10 ppm N2O were prepared by dynamically diluting S1-c90ppm with 5 matrix gas c.. All data were corrected for drift using a linear interpolation of two bracketing anchor gas measurements (cf. Lebegue et al., 2016). In Table S6-1 and S6-2 the short–term repeatability is expressed as standard deviation (1σ) of 10 repeated measurements using averaging times of 300 s and 600 s, respectively. At ~327 ppb N2O, the best 10 repeatability for δ values was achieved by TREX–QCLAS I, with around 0.3 ‰ for δ15Nα, δ15Nβ, and δ18O at 300 s averaging. The CRDS analyzers showed a similar repeatability level of 0.25–0.40 ‰ for all δ values, at 600 s averaging, somewhat worse 0.35–0.6 ‰ at 300 s, but without the requirement for preconcentration. OA-ICOS I achieved a repeatability of 0.75 ‰ (1.1 ‰) for δ15Nα and δ15Nβ, and 1.3 ‰ (2.4‰) for δ18O at 600 s (300 s) averaging. QCLAS analyzers without preconcentration (QCLAS I) 15 showed poor repeatability for δ values regardless of averaging time, in-line with results acquired in Sect. 3.1. In contrast to δ values, the best [N2O] repeatability was achieved by both the OA-ICOS I and QCLAS I (0.02 ppb), while the repeatability was worst for TREX–QCLAS I (0.48 ppb), due to the multiple parameters involved in [N2O] analysis. 20 At 1 ppm N2O, the repeatability of δ-measurements improved for all instruments, with the exception of CRDS I, compared to ~327 ppb N2O. CRDS II achieved the best repeatability of 0.15 – 0.24 ‰ (0.20 – 0.26 ‰) for δ values at 600 s (300 s) averaging, while the repeatability for CRDS I did not profit from an increase in [N2O] but degraded to 0.39 – 0.49 ‰ (0.65 – 0.91 ‰). OA-ICOS I achieved a repeatability of 0.28–0.37 ‰ (0.21–0.27 ‰) for δ15Nα and δ15Nβ and 0.69 ‰ (0.54 ‰) for δ18O at 300 s (600 s) averaging. 25 QCLAS I showed again the weakest repeatability for δ values of 2–7 ‰, irrespective of the averaging time. Again the best repeatability for [N2O] was achieved by OA-ICOS I and QCLAS I but worsened for all instruments compared to ~327 ppb N2O. 25 At 10 ppm N2O, δ-measurement repeatability improved further compared to both near-atmospheric and 1 ppm measurements. The best repeatability was demonstrated by OA-ICOS I with 0.09‰, 0.16‰ and 0.20‰ at 300 s averaging, and 0.06‰, 0.13‰ and 16‰ at 600 s averaging for δ15Nα, δ15Nβ and δ18O, respectively. The repeatability of QCLAS I was also largely improved to 0.3–0.45 ‰ for δ15Nα and δ15Nβ. 5 In contrast, QCLAS I achieved the best [N2O] repeatability of 1.05 ppb (1.45 ppb) at 600 s (300 s) averaging. Table S6-1. Repeatability acquired for [N2O], δ15Nα, δ15Nβ and δ18O as standard deviation (1σ) for 10 repeated measurements at 300 s averaging time. Measurements were made at 326.5 ppb (PA1), 1000 ppb and 10000 ppb N2O. 1σ N O 1σ δ15Nα 1σ δ15 β2 N 1σ δ18O Instrument n N O range δ15Nα range δ15Nβ2 range δ18O range [ppb] [ppb] [‰] [‰] [‰] [‰] [‰] [‰] 326.5 ppb N2O CRDS I 10 0.17 0.50 0.48 1.82 0.38 1.07 0.59 2.03 CRDS II 10 0.05 0.13 0.58 1.64 0.37 1.14 0.35 0.94 OA-ICOS I 10 0.02 0.06 1.12 3.32 1.14 3.26 2.40 7.06 QCLAS I 10 0.02 0.05 6.60 19.19 9.84 31.35 - - TREX-QCLAS I 10 0.48 1.36 0.27 0.68 0.31 0.97 0.29 0.73 1000 ppb N2O CRDS I 10 0.87 2.78 0.75 2.37 0.91 2.82 0.65 1.95 CRDS II 10 0.45 1.15 0.20 0.70 0.26 0.91 0.23 0.61 OA-ICOS I 10 0.24 0.88 0.37 1.40 0.28 0.83 0.69 2.28 QCLAS I 10 0.31 0.87 2.38 7.47 6.62 16.75 - - 10000 ppb N2O OA-ICOS I 10 2.13 6.65 0.09 0.26 0.16 0.49 0.20 0.64 QCLAS I 10 1.45 4.77 0.28 0.99 0.43 1.54 - - 10 26 Table S6-2. Repeatability acquired for [N2O], δ15Nα, δ15Nβ and δ18O as standard deviation (1σ) for 10 repeated measurements at 600s averaging time. Measurements were made at 326.5 ppb (PA1), 1000 ppb and 10000 ppb N2O. 1σ N O 1σ δ15Nα 2 1σ δ15Nβ 1σ δ18O Instrument n N2O range δ15Nα range δ15Nβ range δ18O range [ppb] [ppb] [‰] [‰] [‰] [‰] [‰] [‰] 326.5 ppb N2O CRDS I 10 0.17 0.48 0.39 1.12 0.23 0.84 0.37 1.43 CRDS II 10 0.05 0.13 0.34 1.04 0.29 1.04 0.30 1.07 OA-ICOS I 10 0.01 0.04 0.77 2.33 0.75 2.05 1.29 3.45 QCLAS I 10 0.02 0.06 6.99 17.63 11.66 32.97 - - 1000 ppb N2O CRDS I 10 0.66 2.05 0.49 1.72 0.48 1.47 0.39 1.15 CRDS II 10 0.39 1.15 0.24 0.70 0.15 0.46 0.16 0.50 OA-ICOS I 10 0.23 0.69 0.27 0.88 0.21 0.61 0.54 1.84 QCLAS I 10 0.21 0.62 2.21 7.19 6.34 16.30 - - 10000 ppb N2O OA-ICOS I 10 1.61 4.79 6.3∙10-2 0.17 0.13 0.47 0.16 0.56 QCLAS I 10 1.05 3.09 0.28 0.89 0.46 1.64 - - 5 10 15 27 Section S7 – Scaling of the signal-to-noise ratio Expanding on the N2O mole fraction dependence testing performed in Sect. 2.4.4, we plotted the uncertainty (1σ) of the single 5 min-averaged measurements performed across the three days of mole fraction dependence experimentation acquired by each analyzer, as shown in Fig. S7-1. Because the 5- 5 min averaged data is acquired at the frequency rates as outlined in Table 1, individual points in Fig. S7-1 broadly correspond to the initial starting point of each Allan deviation plot (1s for OA-ICOS I, QCLAS I, II and III; 3.41s for CRDS I; 2.54s for CRDS II) for each analyzer measuring at a given N2O mole fraction. Fig. S7-1 is therefore a useful guide for determining the theoretical optimum measurement precision range for each analyzer. The behavior of this δ-measurement uncertainty as a function of N2O 10 mole fraction is also of great interest, as this will have important implications for situations where measurements cannot be repeated in order to increase certainty in the measurement (e.g. low sample volume or soil flux chamber measurements). For δ measurements, the greatest precision is likely to achieved by OA-ICOS I between ~6000–20000 ppb N2O, CRDS I and II between ~500–1000 ppb N2O, and QCLAS I between ~30000–90000 ppb. 15 28 Fig. S7-1. Scaling of single measurement standard deviation (1σ; 300 second averaging time) as a function of N2O mole fraction for the OA-ICOS I (blue), CRDS I (red), CRDS II (black) and QCLAS I (green). Data was acquired during the mole fraction dependence testing depicted in Fig. 5. 5 29 Section S8 – Continuity of gas matrix and trace gas corrections at higher N2O mole fractions Gas matrix (O2) and trace gas (CO2, CH4 and CO) experiments conducted at 660 and 990 ppb N2O showed that the interference effects on N2O mole fraction and delta values is also dependent on N2O mole fraction 5 (Tables S8-1 and S8-2). Figs. S8-1 to S8-4 show all data (330, 660 and 990 ppb N2O) acquired during O2, CO2, CH4 and CO dependence testing, and shows data corrected using Eqs. (7-8) for O2 and Eq. (9) for CO2, CH4 and CO. Corrected data is provided if the linear regression conducted at 330 ppb N2O for the interference effect was statistically significant at p < 0.05. The similarity between the CH4 and CO dependencies for N2O mole fraction measurements across all instruments suggests that the apparent 10 effects may be due to the dynamic dilution process, rather than a discrete spectral interference effect. Therefore, data has not been corrected for these effects. Correction using Eqs. (7-8) and Eqs. (11-12) removes the matrix and trace gas effects to the extent that corrected measurements are typically within the uncertainty bounds of the anchor. The O2 constants A and B, and a, b and c estimated for each analyzer are provided in Table S8-3, while the approximated trace gas constant values of 𝐴 , 𝐵 , 𝑎 and 𝑏 for 15 each analyzer are provided in Table S8-4. 30 Fig. S8-1. Dependency of the measured [N2O], δ15Nα, δ15Nβ and δ18O values on changing O2 content (%) for OA-ICOS I, CRDS I, CRDS II, and QCLAS I. Non-corrected data is shown in red for various N2O 5 mole fraction testing (circle = 330 ppb; square = 660 ppb; diamond = 990 ppb). Corrected data using Eqs. (7-8) is shown in blue. The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. 31 Fig. S8-2. Dependency of the measured [N2O], δ15Nα, δ15Nβ and δ18O values on CO2 (ppm) for OA-ICOS I, CRDS I, CRDS II, and QCLAS I. Non-corrected data is shown in red for various N2O mole fraction testing (circle = 330 ppb; square = 660 ppb; diamond = 990 ppb). Corrected data using Eqs. (11-12) is 5 shown in blue. The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. 32 Fig. S8-3. Dependency of the measured N2O mole fraction, δ15Nα, δ15Nβ and δ18O values on CH4 (ppm) for OA-ICOS I, CRDS I, CRDS II, and QCLAS I. Non-corrected data is shown in red for various N2O mole fractions tested (circle = 330 ppb; square = 660 ppb; diamond = 990 ppb). Corrected data using Eqs. 5 (11-12) is shown in blue. The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. 33 Fig. S8-4. Dependency of the measured N2O mole fraction, δ15Nα, δ15Nβ and δ18O values on CO (ppm) for OA-ICOS I, CRDS I, CRDS II, and QCLAS I. Non-corrected data is shown in red for various N2O mole fractions tested (circle = 330 ppb; square = 660 ppb; diamond = 990 ppb). Corrected data using Eqs. 5 (11-12) is shown in blue. The standard deviation of the Anchor gas (±1σ) is indicated by dashed lines. 10 34 Table S8-1. Summary of regression slopes and coefficients of determination (r2) for O2 interferences performed at different N2O mole fractions (330, 660 and 990 ppb) for OA-ICOS I, CRDS I and II, and QCLAS I. OA-ICOS I CRDS I CRDS II QCLAS I Co- ΔO2 [%] measured Slope r2 Slope r2 Slope r2 Slope r2 N2O [ppb] 330 -0.044 0.84 0.24 1.00 0.305 1.00 0.351 1.00 N2O [ppb] 660 -0.273 0.98 0.478 0.99 0.608 1.00 0.690 1.00 990 -0.570 0.99 0.61 0.99 0.859 1.00 0.979 1.00 330 1.146 0.99 -1.364 1.00 -0.888 0.99 n.s. n.s. δ15Nα [‰] 660 1.116 1.00 -1.387 1.00 -0.874 0.99 n.s. n.s. 990 1.204 1.00 -1.326 1.00 -0.891 0.99 0.374 0.90 330 1.270 1.00 -0.642 1.00 -0.279 0.95 -1.111 0.87 δ15Nβ [‰] 660 1.282 1.00 -0.580 0.98 -0.303 0.96 n.s. n.s. 990 1.361 1.00 -0.319 0.96 -0.273 0.99 n.s. n.s. 330 0.874 0.97 -0.577 0.98 -0.304 0.99 n.d. n.d. δ18O [‰] 660 1.419 0.99 -0.621 0.98 -0.267 0.94 n.d. n.d. 990 1.446 1.00 -0.507 0.97 -0.256 0.98 n.d. n.d. n.d. not determined 5 n.s. not statistically significant at p < 0.05 35 Table S8-2. Summary of regression slopes and coefficients of determination (r2) for trace gas interferences (CO2, CH4 and CO) performed at different N2O mole fractions (330, 660 and 990 ppb) for OA-ICOS I, CRDS I and II, and QCLAS I. OA-ICOS I CRDS I CRDS II QCLAS I Co- ΔCO2 [ppm] measured Slope r2 Slope r2 Slope r2 Slope r2 N2O [ppb] 330 1.12⋅10-3 0.96 4.90⋅10-4 0.88 4.94⋅10-4 0.83 n.s. n.s. N2O [ppb] 660 1.50⋅10-3 0.99 6.57⋅10-4 0.75 6.45⋅10-4 0.73 -1.64⋅10-4 0.70 990 1.92⋅10-3 0.99 n.s. n.s. 1.37⋅10-3 0.87 n.s. n.s. 330 -8.70⋅10-3 0.95 -1.58⋅10-3 0.88 -6.83⋅10-4 0.66 n.s. n.s. δ15Nα [‰] 660 -5.01⋅10-3 0.99 -6.95⋅10-4 0.76 n.s. n.s. n.s. n.s. 990 -3.80⋅10-3 1.00 n.s. n.s. n.s. n.s. n.s. n.s. 330 2.55⋅10-2 1.00 n.s. n.s. n.s. n.s. 1.54⋅10-2 0.59 δ15Nβ [‰] 660 1.11⋅10-2 1.00 n.s. n.s. n.s. n.s. 5.50⋅10-3 0.70 990 6.94⋅10-3 0.99 n.s. n.s. n.s. n.s. 6.64⋅10-3 0.76 330 -2.65⋅10-2 0.99 -1.88⋅10-3 0.86 -1.20⋅10-3 0.72 n.d. n.d. δ18O [‰] 660 -1.66⋅10-2 1.00 n.s. n.s. n.s. n.s. n.d. n.d. 990 -1.18⋅10-2 1.00 n.s. n.s. 5.09⋅10-4 0.72 n.d. n.d. ΔCH4 [ppm] 330 -4.86⋅10-2 0.67 -3.91⋅10-2 0.48 -5.56⋅10-2 0.57 -4.15⋅10-2 0.45 N2O [ppb] 660 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 36 990 -1.28⋅10-1 0.92 n.s. n.s. -1.06⋅10-1 0.69 -1.14⋅10-1 0.88 330 n.s. n.s. 2.50⋅100 1.00 2.49⋅100 1.00 n.s. n.s. δ15Nα [‰] 660 n.s. n.s. 1.29⋅100 0.98 1.22⋅100 1.00 n.s. n.s. 990 n.s. n.s. 8.26⋅10-1 0.95 8.13⋅10-1 0.99 -7.59⋅10-1 0.93 330 1.73⋅10-1 0.29 n.s. n.s. 8.47⋅10-2 0.23 n.s. n.s. δ15Nβ [‰] 660 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 990 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 330 n.s. n.s. 1.14⋅100 0.98 1.29⋅10-1 0.97 n.d. n.d. δ18O [‰] 660 -2.21⋅10-1 0.74 7.34⋅10-1 0.99 6.36⋅10-1 0.99 n.d. n.d. 990 n.s. n.s. 3.76⋅10-1 0.79 3.55⋅10-1 0.98 n.d. n.d. ΔCO [ppm] 330 -2.90⋅10-1 0.72 -1.49⋅10-1 0.30 -2.39⋅10-1 0.59 -1.90⋅10-1 0.51 N O [ppb] 660 -5.29⋅10-12 0.76 n.s. n.s. n.s. n.s. n.s. n.s. 990 -7.72⋅10-1 0.76 n.s. n.s. n.s. n.s. n.s. n.s. 330 n.s. n.s. -1.50⋅100 0.82 -1.64⋅100 0.77 -4.04⋅100 0.76 δ15Nα [‰] 660 n.s. n.s. n.s. n.s. -6.49⋅10-1 0.81 n.s. n.s. 990 n.s. n.s. n.s. n.s. -4.16⋅10-1 0.98 -4.17⋅100 0.73 330 n.s. n.s. -2.19⋅100 0.95 -2.41⋅100 0.92 -4.26⋅100 0.29 δ15Nβ [‰] 660 n.s. n.s. -1.31⋅100 0.67 -1.44⋅100 0.89 n.s. n.s. 990 n.s. n.s. n.s. n.s. -7.86⋅10-1 0.95 n.s. n.s. δ18O [‰] 330 n.s. n.s. -5.31⋅10-1 0.32 n.s. n.d. n.d. 37 660 n.s. n.s. n.s. n.s. n.s. n.s. n.d. n.d. 990 n.s. n.s. n.s. n.s. n.s. n.s. n.d. n.d. n.d. not determined n.s. not statistically significant at p < 0.05 Table S8-3. The fitted constants A, B, and a, b and c for OA-ICOS I, CRDS I, CRDS II and QCLAS I, derived using Eqs. 5 and 6. For values of 0, there was either no apparent trend or the fitted values fell 5 within the uncertainty of the measurements. The units for each constant are A: 1/(ppbN2O∙%O2), B: 1/%O2, a: ‰/(ppb 2N2O ∙%O2), b: ‰/(ppbN2O∙%O2), and c: ‰/%O2. N O [ppb] δ15Nα [‰] δ15Nβ2 [‰] δ18O [‰] OA-ICOS I A -5.97∙10-7 n.a. n.a. n.a. a n.a. -1.37∙10-8 -2.57∙10-7 -2.93∙10-6 B 7.09∙10-5 n.a. n.a. n.a. b n.a. 2.07∙10-4 5.39∙10-4 4.82∙10-3 c n.a. 1.00∙100 1.07∙100 -4.40∙10-1 CRDS I A -2.40∙10-7 n.a. n.a. n.a. a n.a. 7.74∙10-7 7.71∙10-7 7.55∙10-7 B 8.67∙10-4 n.a. n.a. n.a. b n.a. -9.55∙10-4 -4.71∙10-4 -8.51∙10-4 c n.a. -1.18∙100 -6.70∙10-1 -4.40∙10-1 CRDS II A -2.38∙10-7 n.a. n.a. n.a. a n.a. -3.11∙10-8 -7.71∙10-8 -8.34∙10-8 B 1.08∙10-3 n.a. n.a. n.a. b n.a. 6.71∙10-7 1.93∙10-4 2.03∙10-4 c n.a. -9.09∙10-1 -4.06∙10-1 -3.92∙10-1 38 QCLAS I A -2.20∙10-7 n.a. n.a. n.a. a n.a. 0 0 n.d. B 1.21∙10-3 n.a. n.a. n.a. b n.a. 0 0 n.d. c n.a. 0 0 n.d. n.a. not applicable n.d. not determined Table S8-4. The fitted constants Ax, Bx, ax and bx, for OA-ICOS I, CRDS I, CRDS II and QCLAS I, derived using Eqs. 9 and 10. For values of 0, there was either no apparent trend or the fitted values fell 5 within the uncertainty of the measurements. The unit for Ax and ax is ppmN2O/ppmtrace gas, and the unit for Bx and bx is 1/ppmtrace gas. [N O] δ15Nα2 δ15Nβ δ18O A B a δ15Nα bδ15Nα a δ15Nβ b δ15Nβ a δ18O b δ18O OA-ICOS I CO2 -3.35∙10-4 2.181∙10-3 -1.50∙10-3 -2.49∙10-3 8.91∙10-3 -1.86∙10-3 -6.53∙10-3 -6.25∙10-3 CH4 0 0 0 0 5.71∙10-2 0 0 0 CO 0 0 0 0 0 0 0 0 CRDS I CO 9.44∙10-5 2.63∙10-4 -7.780∙10-42 -8.35∙10-4 0 0 -1.32∙10-3 -1.84∙10-3 CH4 0 0 8.42∙10-1 0 2.651∙10-2 0 3.99∙10-1 0 CO 0 0 -4.05∙10-1 0 -7.88∙10-1 0 -1.34∙10-1 0 CRDS II CO2 -5.04∙10-4 2.07∙10-3 3.20∙10-4 2.30∙10-4 0 0 -1.06∙10-3 1.59∙10-3 39 CH4 0 0 7.94∙10-1 0 3.71∙10-2 0 4.27∙10-1 0 CO 0 0 -4.66∙10-1 0 -7.83∙10-1 0 -2.03∙10-1 0 QCLAS I CO2 0 0 -8.56∙10-6 0 -5.73∙10-6 0 n.d. n.d. CH4 0 0 0 0 0 0 n.d. n.d. CO 0 0 -1.41∙100 0 -1.57∙100 0 n.d. n.d. n.d. not determined Section S9 – Comparison with GC-IRMS Fig. S9-1 shows the triplicate measurements (mean ± 1σ) obtained using the GC-IRMS plotted against 5 expected mixing values calculated using MFC flow rates and the mole fraction and isotopic composition of background and source. A comparison between laser spectrometer and GC-IRMS measurements are presented in Figs. S9-2 to S9-5 using the mean ± 1σ of triplicate measurements. TREX-QCLAS I measurements were undertaken separately, and therefore were not directly compared to IRMS measurements. 10 There was excellent agreement between the IRMS and calculated expected isotope values (all r2 > 0.99). Measurements for δ15Nα, δ15Nβ and δ15Nbulk were all typically within ± 1 ‰ of expected values, while SP was within ± 0.7 ‰ of expected values. δ18O measurements were the poorest performing and were typically within ± 3 ‰ of expected values. The standard deviations of triplicate isotope measurements 15 were typically between 0.1 – 1 ‰. Generally, there is good agreement between the laser spectrometers and GC measurements, but disagreement between the two techniques becomes pronounced at higher ΔN2O for OA-ICOS I and QCLAS I during experiments 5 and 6. All analyzers, with the exception of one triplicated measurement 20 taken from CRDS I, showed better 1σ repeatability than those acquired using GC. There was excellent 40 agreement for δ15Nα, δ15Nβ and δ15Nbulk between the IRMS and both CRDS I and II at all concentrations tested. δ15Nα, δ15Nβ and δ15Nbulk measurements for these analyzers were typically within ± 1 ‰ of those acquired from IRMS, while SP measurements were typically within ± 1.3 ‰. OA-ICOS I δ15Nα, δ15Nβ and δ15Nbulk measurements were typically within ± 2.5 ‰ of IRMS, while SP were in slightly better 5 agreement (± 1.7 ‰). Conversely, QCLAS I showed good agreement with IRMS only at higher ΔN2O (> 1,000 ppb), presumably due to less precise measurements that are expected (based on results acquired in Section 3) to be acquired at the lower concentrations tested. For OA-ICOS I, the repeatability of the triplicate δ15Nα, δ15Nβ, δ15Nbulk and SP measurements was typically better than the IRMS exclusively at higher ΔN2O (>1,000 ppb). CRDS I and II had comparable repeatability to the IRMS, and there was no 10 clear distinction based on ΔN2O. QCLAS I showed comparable or better repeatability to IRMS only at higher ΔN2O (>1,000 ppb). δ18O measurements from OA-ICOS I, and CRDS I and II showed good agreement with IRMS results (all r2 > 0.98), with the vast majority of measurements similarly within ± 2‰. OA-ICOS I δ18O measurements 15 had typically better repeatability compared to IRMS at higher ΔN2O, while both CRDS analyzers showed varied repeatability regardless of ΔN2O. 41 Fig. S9-1. Correlation diagrams for [N2O], δ15Nα, δ15Nβ, δ15Nbulk, SP and δ18O measurements at various ΔN2O mole fractions analyzed by GC-IRMS plotted against expected values. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed black line represents a linear fit to the data. Individual 5 equations, coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O mole fractions, and the 1:1 line is similarly a solid line. 42 Fig. S9-2. Correlation diagrams for [N2O], δ15Nα, δ15Nβ, δ15Nbulk, SP and δ18O measurements at various ΔN2O concentrations analyzed by both OA-ICOS I and GC-IRMS. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed blue line represents a linear fit to the data. Individual equations, 5 coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O concentrations, and the 1:1 line is similarly a solid line. 43 Fig. S9-3. Correlation diagrams for [N2O], δ15Nα, δ15Nβ, δ15Nbulk, SP and δ18O measurements at various ΔN2O concentrations analyzed by both CRDS I and GC-IRMS. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed red line represents a linear fit to the data. Individual equations, 5 coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O concentrations, and the 1:1 line is similarly a solid line. 44 Fig. S9-4. Correlation diagrams for [N2O], δ15Nα, δ15Nβ, δ15Nbulk, SP and δ18O measurements at various ΔN2O concentrations analyzed by both CRDS II and GC-IRMS. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed black line represents a linear fit to the data. Individual equations, 5 coefficients of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O concentrations, and the 1:1 line is similarly a solid line. 45 Fig. S9-5. Correlation diagrams for [N O], δ15Nα, δ15Nβ, δ15Nbulk2 and SP measurements at various ΔN2O concentrations analyzed by both QCLAS I and GC-IRMS. The solid black line denotes the 1:1 line, while the dotted line indicates ±1σ of the residuals from the 1:1 line. The dashed green line represents a linear fit to the data. Individual equations, coefficients 5 of determination (r2) and p-values are indicated above each plot. Each data point represents the mean and standard deviation (1σ) of triplicate measurements. The inset plots indicate the standard deviation (1σ) of the triplicate measurements achieved at different ΔN2O concentrations, and the 1:1 line is similarly a solid line. 10 46 Section S10: Extrapolated source intercept values In Sect. 3.7.2, the extrapolated source intercept values acquired using Keeling analysis showed large standard errors, especially for Experiments 1 and 2 (Table S10-1). This was mostly due to the small mole fraction range (i.e. large inverse mole fraction range) over which the regression line was extrapolated in 5 order to acquire the intercept value. 47 Table 10. Intercept values obtained by the four analyzers from the Keeling analysis. The error reported is 1 standard error. A 3-point concentration correction was applied to the data. Error (SourceEstimated – SourceTrue) represents the error (‰) between the estimated source values and the true source values. N2O represents the average concentration difference (ppb) between the highest concentration in each experiment and the background based on the measurements recorded by 5 each analyzer. Experiment N2O Instrument Error (SourceEstimated – SourceTrue) sequence [ppb] OA-ICOS I δ15Nα δ15Nβ δ15Nbulk SP δ18O δ15Nα [‰] δ15Nβ [‰] δ15Nbulk [‰] SP [‰] δ18O [‰] [‰] [‰] [‰] [‰] [‰] 1 – Enriched 31.42 71.15 ± 8.44 80.04 ± 7.93 75.62 ± 6.97 -8.84 ± 8.74 131.2 ± 16.16 0.89 -5.84 -2.45 6.79 2.35 2 – Depleted 31.26 -37.4 ± 10.91 -37.19 ± 9.34 -36.59 ± 8.78 0.28 ± 10.02 58.65 ± 8.78 7.27 -4.32 2.18 12.08 33.40 3 – Depleted 694.22 -18.53 ± 1.19 -22.7 ± 0.56 -20.56 ± 0.60 4.27 ± 1.46 26.49 ± 1.41 7.21 0.92 4.12 6.39 -4.85 4 – Enriched 701.46 54.64 ± 1.03 57.21 ± 0.63 55.94 ± 0.56 -2.51 ± 1.30 107.3 ± 2.06 1.43 -0.75 0.35 2.24 4.75 5 – Enriched 9831.51 53.3 ± 0.07 56.62 ± 0.49 54.96 ± 0.26 -3.33 ± 0.48 105.1 ± 1.17 2.88 2.54 2.71 0.33 5.88 6 – Depleted 9933.59 -26.46 ± 0.16 -24.02 ± 0.70 -25.26 ± 0.42 -2.35 ± 0.56 31.49 ± 1.21 -2.11 -1.08 -1.62 -0.94 -0.30 CRDS I 1 – Enriched 29.37 55.66 ± 4.27 86.39 ± 5.76 70.75 ± 4.15 -30.73 ± 6.05 134.5 ± 3.81 -14.60 0.51 -7.32 -15.10 5.65 2 – Depleted 29.21 -61.43 ± 8.86 -40.32 ± 5.10 -49.69 ± 5.57 -22.37 ± 8.29 22.32 ± 6.54 -16.76 -7.45 -10.92 -10.57 -2.93 3 – Depleted 646.99 -23.14 ± 0.86 -26.89 ± 0.73 -25.05 ± 0.59 3.72 ± 1.15 26.89 ± 0.78 2.60 -3.27 -0.37 5.84 -4.45 4 – Enriched 651.93 56.83 ± 0.74 57.41 ± 0.8 57.07 ± 0.53 -0.69 ± 1.21 103.8 ± 0.95 3.62 -0.55 1.48 4.06 1.27 CRDS II 1 – Enriched 29.55 68.45 ± 3.15 103.53 ± 5.16 86.01 ± 3.92 -34.52 ± 3.7 140.05 ± 3.52 -1.81 17.65 7.94 -18.89 11.21 2 – Depleted 29.41 -52.51 ± 7.94 -29.32 ± 5.47 -40.65 ± 5.72 -23.1 ± 6.84 25.76 ± 7.31 -7.84 3.55 -1.88 -11.30 0.51 3 – Depleted 620.28 -28.49 ± 0.30 -24.8 ± 0.22 -26.59 ± 0.19 -3.61 ± 0.35 28.23 ± 0.24 -2.76 -1.18 -1.91 -1.49 -3.11 4 – Enriched 626.24 50.94 ± 0.30 58.78 ± 0.21 54.87 ± 0.21 -7.76 ± 0.29 104.75 ± 0.25 -2.27 0.82 -0.72 -3.01 2.25 QCLAS I 1 – Enriched 30.52 185.3 ± 84.81 157.7 ± 74.91 172.3 ± 68.04 26.11 ± 59.38 - 115.0 71.81 94.27 41.74 - 2 – Depleted 30.48 106.5 ± 68.41 103.3 ± 111.5 102.7 ± 82.67 2.23 ± 76.5 - 151.1 136.2 141.5 14.03 - 3 – Depleted 648.35 -32.8 ± 4.83 -27.92 ± 5.12 -30.33 ± 4.67 -3.3 ± 2.74 - -7.07 -4.30 -5.65 -1.18 - 4 – Enriched 654.63 46.36 ± 5.03 34.08 ± 2.84 40.29 ± 3.55 12.85 ± 3.19 - -6.85 -23.88 -15.30 17.60 - 5 – Enriched 9231.17 56.8 ± 0.17 49.17 ± 0.63 52.99 ± 0.33 7.51 ± 0.61 - 6.38 -4.91 0.74 11.17 - 6 – Depleted 9323.02 -27.44 ± 0.48 -22.49 ± 0.53 -25.02 ± 0.39 -4.96 ± 0.61 - -3.09 0.45 -1.38 -3.55 - 48 TREX-QCLAS I 1 – Enriched 32.38 73.95  1.71 86.49  1.80 80.20  1.27 -12.54  1.85 134.3  1.81 0.48 -4.64 -2.10 5.12 0.55 2 – Depleted 34.14 -44.84  0.83 -41.78  1.83 -43.31  1.47 -3.02  2.08 25.96  1.33 3.55 -7.09 -1.79 10.64 1.90 GC-IRMS 1 – Enriched 30.14 75.31  8.78 88.49  13.71 81.93  10.19 -13.24  10.73 144.2  12.88 5.06 2.61 3.86 2.39 15.35 2 – Depleted 30.43 -40.50  4.63 -36.49  5.61 -38.48  2.34 -4.02  9.17 24.12  3.46 4.17 -3.62 0.27 7.77 -1.13 3 – Depleted 668.62 -24.00  0.23 -24.55  0.25 -24.28  0.12 0.55  0.40 29.77  0.27 1.74 -0.93 0.40 2.67 -1.57 4 – Enriched 674.33 53.48  0.19 56.05  0.23 54.76  0.16 -2.57  0.27 106.7  0.26 0.26 -1.91 -0.82 2.17 4.24 5 – Enriched 8540.91 55.62  0.33 59.25  0.73 57.44  0.44 -3.63  0.72 111.5  0.51 5.20 5.17 5.19 0.03 12.37 6 – Depleted 8958.53 -25.35  0.29 -24.43  0.32 -24.89  0.28 -0.92  0.24 31.38  0.39 -1.00 -1.49 -1.24 0.49 -0.41 49 Section S11: Lower state energies of probed N2O isotopocule lines The lower state energies of probed N2O isotopocule lines are provided in Table S11-1. Differences in the rotational quantum numbers for a pair of isotopocules (such as 14N15N16O / 14N14N16O) lead to changes in 5 N2, O2 and Ar broadening parameters (Henry et al., 1985). If the sample gas matrix is different to that of the reference gas, deviations in the apparent delta values will arise. Table S11-1. Wavenumber positions, line strength, branch / rotational quantum numbers and lower state energies of selected N2O isotopocule lines applied for different laser spectrometers as retrieved from 10 HITRAN2016 database. Line positions Line strength Branch / rotational Lower-state (cm-1) (cm-1 /(molecule cm- quantum number energy (cm-1) 2)) OA-ICOS I 14N14N16O 2192.401 4.92E-20 P / 19 748.33 2192.436 4.92E-20 P / 19 748.03 2192.483 3.38E-19 P / 33 469.91 14N15N16O 2192.309 3.31E-21 R / 18 143.27 15N14N16O 2192.330 2.97E-21 P / 11 53.44 14N14N18O 2192.133 1.11E-21 P / 28 321.10 CRDS I & II 14N14N16O 2196.21 5.16E-20 P / 15 689.55 2196.24 5.16E-20 P / 15 689.36 14N15N16O 2195.762 2.73E-21 R / 23 231.22 15N14N16O 2195.796 2.20E-21 P / 7 22.67 14N14N18O 2195.951 1.43E-21 P / 24 237.29 50 QCLAS I, II & III 14N14N16O 2188.045 2.60E-21 P / 9 1205.92 14N15N16O 2187.943 3.29E-21 R / 12 65.36 15N14N16O 2187.846 3.27E-21 P / 16 110.11 14N14N18O 2203.281 1.79E-21 P / 16 107.59 TREX- QCLAS I 14N14N16O 2203.100 2.71E-21 R / 8 1198.37 2203.114 1.44E-21 R / 8 1314.95 14N15N16O 2203.359 9.80E-22 R / 35 527.64 15N14N16O 2203.205 7.02E-22 R/ 1 0.81 14N14N18O 2203.281 1.79E-21 P / 16 107.59 51