Isotopically characterised N2 O reference materials for use as community standards.

RATIONALE: Information on the isotopic composition of nitrous oxide (N2 O) at natural abundance supports the identification of its source and sink processes. In recent years, a number of mass spectrometric and laser spectroscopic techniques have been developed and are increasingly used by the research community. Advances in this active research area, however, critically depend on the availability of suitable N2 O isotope Reference Materials (RMs).
METHODS: Within the project Metrology for Stable Isotope Reference Standards (SIRS), seven pure N2 O isotope RMs have been developed and their 15 N/14 N, 18 O/16 O, 17 O/16 O ratios and 15 N site preference (SP) have been analysed by specialised laboratories against isotope reference materials. A particular focus was on the 15 N site-specific isotopic composition, as this measurand is both highly diagnostic for source appointment and challenging to analyse and link to existing scales.
RESULTS: The established N2 O isotope RMs offer a wide spread in delta (δ) values: δ15 N: 0 to +104‰, δ18 O: +39 to +155‰, and δ15 NSP : -4 to +20‰. Conversion and uncertainty propagation of δ15 N and δ18 O to the Air-N2 and VSMOW scales, respectively, provides robust estimates for δ15 N(N2 O) and δ18 O(N2 O), with overall uncertainties of about 0.05‰ and 0.15‰, respectively. For δ15 NSP , an offset of >1.5‰ compared with earlier calibration approaches was detected, which should be revisited in the future.
CONCLUSIONS: A set of seven N2 O isotope RMs anchored to the international isotope-ratio scales was developed that will promote the implementation of the recommended two-point calibration approach. Particularly, the availability of δ17 O data for N2 O RMs is expected to improve data quality/correction algorithms with respect to δ15 NSP and δ15 N analysis by mass spectrometry. We anticipate that the N2 O isotope RMs will enhance compatibility between laboratories and accelerate research progress in this emerging field.

INTRODUCTION

Since its first application by Sakae Toyoda and Naohiro Yoshida in 1999, site‐specific N2O isotope analysis has been applied by many research groups to differentiate N2O source and sink processes at different spatio‐temporal scales (see reviews by Toyoda et al, Ostrom et al, Decock et al, Denk et al, and Yu et al ). Likewise, dual‐isotope plots (e.g. δ 15NSP/δ 15N) or so‐called “isotope mapping” approaches have been used to constrain the contributions of specific pathways, and the effect of isotope fractionation during N2O reduction. , The informative value of N2O isotope data has been markedly increased by using the data to inform biogeochemical models, providing regional and global patterns of N2O losses and independent process information. , , , Advances in applications have been accompanied and accelerated by progress in analytics, complementing the traditional high‐precision isotope‐ratio mass‐spectrometry (IRMS) , by laser spectroscopic techniques, with the potential for field applicability and real‐time data coverage. , , , , ,

The isotopic composition of a sample is reported using the delta (δ) notation, which is the relative difference in isotope ratio (R) between a sample P and a reference material, i.e. δ(P/ref) = R P/R ref − 1. For nitrogen, the 15N/14N isotope ratio is used, R(15N/14N) = x(15N)/x(14N), where x is the isotopic abundance and tropospheric N2 is the international reference material for the Air‐N2 scale. For oxygen, the 18O/16O and 17O/16O ratios are used, which are related to the Vienna Standard Mean Ocean Water (VSMOW) scale. In addition, we adopt the following notation conventions: δ 15N = δ(15N/14N, P/Air‐N2) (average of both nitrogen atoms) and δ 18O = δ(18O/16O, P/VSMOW). The 15N site preference (SP) is defined by the predominance of 15N substitution in the central (α) position as compared to the terminal (β) position, and calculated accordingly as δ 15NSP = δ 15Nα − δ 15Nβ. All δ values in this paper are reported against Air‐N2 (for 15N/14N ratios) and against VSMOW (for 18O/16O and 17O/16O ratios).

Further progress in N2O isotope research critically depends on the compatibility of laboratory results. To achieve this, individual laboratories have to implement a traceability chain, i.e. a hierarchy of reference materials which descends with increasing uncertainty, linking the isotopic composition of primary RMs used to realise the respective scale, through secondary standards and working laboratory standards to a sample. Generally, two RMs with distinct δ values should be used for calibration purposes, following the two‐point data normalisation requirement. However, primary RMs and secondary scale anchors for δ 15N (ammonium sulfate, potassium nitrate) as well as δ 17O and δ 18O (water) have a different chemical identity than N2O sample gas. Thus, a chemical conversion reaction has to be implemented prior to analysis, which requires specialised laboratories.

The synthesis of N2O by thermal decomposition of isotopically characterised ammonium nitrate (NH4NO3) has been suggested as an approach to link the position‐dependent nitrogen isotopic composition of N2O to the Air‐N2 scale. The basic concept of this technique is that the nitrogen atom at the α‐position of of the formed N2O originates from NO3 −, while the β‐nitrogen comes from NH4 +. The validity of the NH4NO3 decomposition technique has been confirmed, , but its accuracy for the calibration of δ 15Nα and δ 15Nβ was found to be limited by non‐quantitative NH4NO3 decomposition in combination with substantially different isotope enrichment factors of −4 or −19‰ for the conversion of the NO3 − or NH4 + nitrogen atom into the α‐ or β‐position of the N2O molecule. To overcome such difficulties, two new N2O reference gases, USGS51 and USGS52, recently became available with assigned δ values based on a preliminary assessment by Naohiro Yoshida and Sakae Toyoda (Tokyo Institute of Technology). , However, the two standards offer only a small range of δ 15N and δ 18O values (< 1‰), which is not suitable for a two‐point calibration approach.

In the present study, we report the development of additional N2O RMs within the framework of the European Metrology Programme for Innovation and Research (EMPIR) 16ENV06 project ‘Metrology for Stable Isotope Reference Standards (SIRS)’. The target isotopic composition of N2O RMs was selected according to discussions at a stakeholder workshop at the 19th GGMT conference at Empa (29 August 2017). The focus of this study is to extend the range of isotopic composition of N2O RMs compared to RMs presented in Ostrom et al and to provide additional δ 17O data in order to improve data quality/correction algorithms with respect to δ 15NSP and δ 15N analysis by mass spectrometry. In addition, the link of δ values to the international isotope‐ratio scales was revisited.

EXPERIMENTAL

The main purpose of this study is the provision of isotopically characterised N2O RMs, covering an extended range of delta values as compared to existing gases. Figure 1 provides a schematic overview on the links established within this study between existing international RMs and the novel gaseous N2O RMs.

FIGURE 1

Schematic overview on the traceability chain applied in this study to propagate 15N/14N, 18O/16O and 17O/16O isotope ratios from international RMs to δ 15N, δ 15NSP, δ 18O and δ 17O in the novel N2O RMs

In section 2.1 (“left branch” of Figure 1), 15N/14N isotope ratios on the Air‐N2 scale were propagated from NH4 + and NO3 − salts supplied by IAEA/USGS, through isotopic analysis of gravimetrically prepared NH4NO3 salts (section 2.1.2) and their thermal decomposition (section 2.1.3), to δ 15Nβ(N2O) /δ 15Nα(N2O) in the novel N2O RMs. The international RMs applied in this study are listed in Table 1. To provide a reliable link between the Air‐N2 scale and the N2O site‐specific isotopic composition, the NH4NO3 decomposition reaction was optimised for high yield, reproducibility, and N2O purity (see section 2.1.3). Following the recommended two‐point calibration approach, a number of NH4NO3 salts, ranging from 15N‐depleted to 15N‐enriched, were prepared (see section 2.1.1), decomposed, and analysed.

TABLE 1

International RMs applied in this study for the analysis of δ 15N(NH4NO3), δ 15N(NH4 +) and δ 15N(NO3 −) in NH4NO3 salts (section 2.1.2) and δ 15N as well as δ 18O in N2O RMs (section2.2). Values are taken from Brand et al and Ostrom et al and reported in ‰

		δ 15NAir‐N2	σ	δ 18OVSMOW	σ
IAEA‐N‐1	NH 4 SO 4	+0.43	0.07	‐	‐
IAEA‐N‐2	NH 4 SO 4	+20.41	0.12	‐	‐
USGS25	NH 4 SO 4	−30.41	0.27	‐	‐
USGS26	NH 4 SO 4	+53.75	0.24	‐	‐
IAEA‐NO‐3	KNO 3	+4.72	0.13	+13.2	‐
USGS32	KNO 3	+180	0	+25.4	0.2
USGS34	KNO 3	−1.8	0.1	−27.78	0.37
USGS35	NaNO 3	+2.7	0.1	+56.81	0.31
USGS40	L‐glutamic acid	−4.52	0.06	‐	‐
USGS51	N 2 O	+1.21	0.21	+41.45	0.34
USGS52	N 2 O	+0.29	0.25	+40.80	0.40

In section 2.2 (“right branch” of Figure 1), preparation of N2O RMs and analysis by expert laboratories for δ 15N(N2O), δ 18O(N2O), δ 17O(N2O) and δ 15NSP(N2O) is described. In one laboratory (Empa), δ 15NSP(N2O) in the N2O RMs was linked to the Air‐N2 scale making use of the traceability chain established in section 2.1. Links to scale applied in the other laboratories are independent and are described in detail in the respective experimental sections.

Re‐evaluation of NH4NO3 thermal decomposition technique to propagate δ 15N(NO3 −)/δ 15N(NH4 +) to δ 15Nα(N2O)/δ 15Nβ(N2O)

Preparation of NH4NO3 salts

Six NH4NO3 salts (S1–S6), covering a wide range of δ 15N(NH4 +) and δ 15N(NO3 −) values, were produced by gravimetric mixing of five commercially available NH4NO3 salts (A–E). A: unlabelled NH4NO3 (purity >98%, K299.1, Carl Roth GmbH, Karlsruhe, Germany), B: 15NH4NO3 (>98% 15NH4 +, NLM‐711‐1, Cambridge Isotope Laboratories Inc., Tewksbury, USA), C: NH4 15NO3 (>98% 15NO3 −, NLM‐712‐1, Cambridge Isotope Laboratories Inc., Tewksbury, USA), D: 15NH4 +‐depleted NH4NO3 (0.306% 15NH4 +, Shoko Science Co., Ltd, Japan), E: 15NO3 −‐depleted NH4NO3 (0.306% 15NO3 −, Shoko Science Co., Ltd, Japan).

For preparation of these six NH4NO3 salts (Table 2), approximately 110 g of unlabelled NH4NO3 (A) was ground to a fine powder using a mortar and pestle and then dried at 120°C for 1 h (a temperature low enough to avoid triggering decomposition). From this, around 100 g (S1–S5) or around 40 g (S6) were gravimetrically (XP205, Mettler Toledo GmbH, Greifensee, Switzerland) mixed with appropriate amounts of salts B, C, D, and E to obtain the desired isotopic composition. The salt mixtures were dissolved in deionised water (Milli‐Q Advantage A10, Millipore AG, Switzerland), recrystallised, dried, and then stored in air‐tight sample containers. The isotopic homogeneity of S1–S6 was confirmed by repeated IRMS analysis (MPI‐BGC), demonstrating δ 15N(NH4NO3) values within <0.2‰ (σ, n = 10).

TABLE 2

Overview of NH4NO3 salts (S1–S6) prepared from commercially available NH4NO3 (A–E) and covering a wide range of δ 15N(NH4 +) and δ 15N(NO3 −) values

	Characteristic	A (unlabelled)	B (15NH4NO3)	C (NH4 15NO3)	D (15NH4 +‐depleted)	E (15NO3 −‐depleted)
S1	Unlabelled NH 4 NO 3	X
S2	15 NH 4 , 15 NO 3 ‐low enriched	X	X	X
S3	Ambient isotopic composition	X	X	X
S4	15 NH 4 , 15 NO 3 ‐enriched	X	X	X
S5	15 NH 4 , 15 NO 3 ‐high enriched	X	X	X
S6	15 NH 4 , 15 NO 3 ‐depleted	X			X	X

Analysis of NH4NO3 salts for δ 15N(NH4NO3), δ 15N(NH4 +) and δ 15N(NO3 −) against IAEA and USGS RMs

Subsamples of the prepared NH4NO3 salts (S1–S6) were sent together with international reference materials ((NH4)2SO4, NaNO3, KNO3) provided by the IAEA (International Atomic Energy Agency, Vienna, Austria) and by USGS (U.S. Geological Survey, Reston, USA) (Table 1) to eight isotope laboratories. Table 3 provides basic information on the analytical techniques applied by the laboratories. Details on the analytics are given in the supporting information (Supplementary Method 1).

TABLE 3

Analytical techniques applied by the involved isotope laboratories for the analysis of δ 15N(NH4NO3), δ 15N(NH4 +) and δ 15N(NO3 −) in NH4NO3 salts (S1–S6). Details on the analytics are given in the supporting information (Supplementary Method 1)

Laboratory	Measurand	Technique
MPI‐BGC Lab (1)	δ 15 N(NH 4 NO 3 )	NH 4 NO 3 analysis by elemental analyser (EA)/IRMS
UC Davis Lab (2)	δ 15 N(NH 4 NO 3 )	NH 4 NO 3 analysis by EA/IRMS
University of Ghent Lab (3)	δ 15 N(NH 4 NO 3 ) δ 15 N(NH 4 + ) δ 15 N(NO 3 − )	NH 4 NO 3 analysis by EA/IRMS 30 NH 4 + oxidation with BrO − to nitrite (NO 2 − ), reaction with hydroxylamine (NH 2 OH) to N 2 O; purge‐and‐trap (PT)‐IRMS analysis 31 NO 3 − conversion into N 2 O by denitrifier method; PT‐IRMS analysis 32, 33
University of Pittsburgh Lab (4)	δ 15 N(NH 4 NO 3 ) δ 15 N(NO 3 − )	NH 4 + oxidation with BrO − to nitrite (NO 2 − ), NO 2 −  + NO 3 − conversion into N 2 O by denitrifier method; PT‐IRMS analysis 34 NO 3 − conversion into N 2 O by denitrifier method; PT‐IRMS analysis 32, 33
UEF‐BGC Lab (5)	δ 15 N(NH 4 + ) δ 15 N(NO 3 − )	NH 3 microdiffusion on acid‐impregnated glass fibre filter, followed by EA/IRMS analysis 35 NO 3 − reaction with vanadium(III) chloride (VCl 3 ) and sodium azide (NaN 3 ) under acidic conditions to N 2 O; PT‐IRMS analysis 35
University of Vienna Lab (6)	δ 15 N(NH 4 + ) δ 15 N(NO 3 − )	NH 3 microdiffusion on acid‐impregnated glass fibre filters, followed by EA/IRMS analysis 35 NO 3 − reaction with VCl 3 and NaN 3 under acidic conditions to N 2 O; PT‐IRMS analysis 35
Tokyo Institute of Technology Lab (7)	δ 15 N(NH 4 + ) δ 15 N(NO 3 − )	NH 3 distillation into acid solution, NH 4 + oxidation with KBrO to N 2 ; IRMS analysis 36 After removal of NH 4 + , NO 3 − reduction by Devarda's alloy to NH 4 + and NH 3 distillation; IRMS analysis as above 36
Hydroisotope Lab (8)	δ 15 N(NH 4 + ) δ 15 N(NO 3 − )	NH 4 + oxidation with LiBrO to N 2 ; IRMS analysis NH 4 + removal by ion exchange; residual measured by EA/IRMS 37, 38

N(NH NO )

N(NH )

N(NO )

NH NO analysis by EA/IRMS

NH oxidation with BrO to nitrite (NO ), reaction with hydroxylamine (NH OH) to N O; purge‐and‐trap (PT)‐IRMS analysis

NO conversion into N O by denitrifier method; PT‐IRMS analysis 32, 33

N(NH NO )

N(NO )

NH oxidation with BrO to nitrite (NO ), NO  + NO conversion into N O by denitrifier method; PT‐IRMS analysis

NO conversion into N O by denitrifier method; PT‐IRMS analysis 32, 33

N(NH )

N(NO )

NH microdiffusion on acid‐impregnated glass fibre filter, followed by EA/IRMS analysis

NO reaction with vanadium(III) chloride (VCl ) and sodium azide (NaN ) under acidic conditions to N O; PT‐IRMS analysis

N(NH )

N(NO )

NH microdiffusion on acid‐impregnated glass fibre filters, followed by EA/IRMS analysis

NO reaction with VCl and NaN under acidic conditions to N O; PT‐IRMS analysis

N(NH )

N(NO )

NH distillation into acid solution, NH oxidation with KBrO to N ; IRMS analysis

After removal of NH , NO reduction by Devarda's alloy to NH and NH distillation; IRMS analysis as above

N(NH )

N(NO )

NH oxidation with LiBrO to N ; IRMS analysis

NH removal by ion exchange; residual measured by EA/IRMS 37, 38

δ 15N(NH4NO3), δ 15N(NH4 +) and δ 15N(NO3 ) results from all laboratories were calibrated using the provided international IAEA and USGS reference materials, with δ 15N values and uncertainties according to Brand et al and references cited therein. The uncertainty of laboratory results (σ cal) was estimated from the uncertainty (σ a, σ b) in the linear calibration function (Equation 1), considering the uncertainty in IAEA and USGS standards and their analyses, as well as the uncertainty (σ meas) in δ 15Nmeas, following the law of error propagation (Equation 2). , , Results (δ 15Ncal,, σ cal,) from individual laboratories i were combined to a weighted mean value (δ 15Nweighted, Equation 3) with an uncertainty (σ weighted, Equation 4) :

NH4NO3 (S1–S6) thermal decomposition to N2O (S1‐N2O–S6‐N2O)

Aliquots of approximately 1.0 g (12.5 mmol) of NH4NO3 salts (S1–S6) were weighed into round‐bottomed glass flasks with a break‐seal (150 mL, borosilicate glass, Willi Möller AG, Zürich, Switzerland). In a variant of the NH4NO3 decomposition reaction according to Szabó et al, 1.4 g NH4HSO4 (>99.99%, Art. No. 455849‐100G, Sigma Aldrich GmbH, Buchs, Switzerland) and 0.2 g (NH4)2SO4 (>99.5%, Art. No. 09978‐500G, Sigma Aldrich GmbH, Buchs, Switzerland) were added. Adding surplus NH4 + salt will lead to a loss in δ 15Nβ information but was included to test if very high reaction yields can be achieved, which might still be attractive. Therefore, for S1, both variants (with/without NH4HSO4/(NH4)2SO4) were tested, while for S2–S6 only decomposition without NH4 + addition was performed. Thereafter, the flasks were evacuated (<10−1 mbar) and flame‐sealed. The sealed flasks were placed in a circulating‐air oven (model TSW 120 ED, Salvis AG, Reussbühl Switzerland) and heated to 270°C for 24 h.

After the decomposition reaction, the N2O product gas, e.g. S1‐derived‐N2O (here: S1‐N2O) or S6‐derived‐N2O (S6‐N2O), was purified on a vacuum manifold by cryogenic distillation. Reaction by‐ and side‐products (e.g. H2O, HNO3, NH3) were trapped at −78°C (dry ice/ethanol bath); N2O was trapped at −196°C (liquid N2) in a coiled stainless‐steel tube, while N2 and O2 (side products) were removed by evacuation with an oil‐sealed rotary vane pump (RV3, Edwards Ltd, Crawley, UK). Thereafter, the N2O product was condensed into 10 mL stainless‐steel flasks (CS‐20181323‐ARBOR, ARBOR Fluidtec AG, Wohlen, Switzerland) under liquid‐nitrogen cooling. The cryogenic extraction was repeated five times to fully capture the produced N2O. Finally, the N2O yield was determined gravimetrically (XP205 analytical balance, Mettler Toledo AG, Greifensee, Switzerland). The N2O purity, i.e. the absence of IR‐active impurities (<5 μmol mol−1 NO, <1 μmol mol−1 NO2, and <0.5 μmol mol−1 NH3), was confirmed by FTIR spectroscopy (Gasmet CX4000 FTIR gas analyser, Temet Instruments Oy, Helsinki, Finland). The distillation procedure (e.g. the trap size and the timing) was optimised for quantitative removal of N2 (<0.01%) and N2O recovery (>99.4%), using different gravimetric mixtures of high‐purity N2O and N2 (Messer Schweiz, Lenzburg, Switzerland).

Test for consistency of NH4NO3 decomposition reaction

First, the consistency of the NH4NO3 decomposition reaction across the large range of δ values (15N‐depleted to highly 15N‐enriched in S1–S6 for both salts and N2O) was tested. In detail, such tests were made by comparing δ 15Nα of NH4NO3‐derived N2O gases (S1‐N2O–S6‐N2O) with the δ 15N(NO3 −) of substrate NH4NO3 salts (S1–S6) and δ 15Nβ with δ 15N(NH4 +), respectively. While the link provided by the NH4NO3 decomposition reaction was assumed to be valid across a wide range of δ values, the analytics involved in δ 15Nα, δ 15Nβ or δ 15N(NO3 −), δ 15N(NH4 +) analysis might display non‐linearities.

For this consistency test, the N2O gases S2‐N2O, S3‐N2O, S5‐N2O and S6‐N2O were analysed together with S1‐N2O and S4‐N2O using the QCLAS analyser (section 2.2.2). S1‐N2O and S4‐N2O were selected as calibration gases, as they differ substantially in delta values (>50‰ in δ 15N) and in preliminary experiments displayed a consistent offset between δ 15Nα(N2O), δ 15Nβ(N2O) and δ 15N(NO3 −), δ 15N(NH4 +) values (data not shown). For actual δ 15Nα and δ 15Nβ of S1‐N2O and S4‐N2O, known δ 15N(NO3 −) and δ 15N(NH4 +) values of the respective NH4NO3 salts were adopted and no correction for fractionation effects due to incomplete decomposition or branching isotope effects due to N2 production was applied. The uncertainty of actual δ 15Nα and δ 15Nβ for S1‐N2O and S4‐N2O was estimated from the uncertainty of weighted mean δ 15N(NO3 −) and δ 15N(NH4 +) values (Table 5) and the standard deviation of δ 15Nα and δ 15Nβ analysis for repeated decomposition experiments using the law of error propagation.

TABLE 5

δ 15N(NH4NO3) (top), δ 15N(NO3 −) (middle), and δ 15N(NH4 +) (bottom) of prepared NH4NO3 salts (S1–S6) analysed by different laboratories using techniques described in Table 3 and the supporting information (Supplementary Method 1). Results from individual laboratories were calibrated using international (IAEA, USGS) standards and their uncertainties (σ) calculated following the law of error propagation. Laboratories: (1) MPI‐BGC, (2) UC Davis, (3) University of Ghent, (4) University of Pittsburgh, (5) UEF‐BGC, (6) University of Vienna, (7) Tokyo Tech, (8) Hydroisotop

δ 15N(NH4NO3)/‰	Lab (1)	Lab (2)	Lab (3)	Lab (4) a			σ (1)	σ (2)	σ (3)	σ (4) a			Weighted mean ± σ
S1	−0.60	−0.70	+0.05	+0.64			0.11	0.07	0.09	0.08			−0.44 ± 0.05
S2	+13.77	+13.73	+14.48	+15.11			0.17	0.15	0.10	0.07			+14.14 ± 0.08
S3	+7.23	+7.02	+8.11	+8.22			0.13	0.07	0.11	0.06			+7.31 ± 0.06
S4	+52.65	+52.42	+53.27	+54.55			0.17	0.41	0.24	0.11			+52.81 ± 0.13
S5	+107.56	+107.61	+108.21	+110.58			0.24	0.32	0.19	0.18			+107.90 ± 0.13
S6	−49.91	−50.00	−49.37	−49.25			0.13	0.14	0.24	0.13			−49.87 ± 0.09

Results were not considered for calculation of weighted mean values as the applied technique is associated with a higher uncertainty.

Measured δ 15Nα values of S1‐N2O and S4‐N2O and actual values, i.e. δ 15N(NO3 −) of the educt NH4NO3 salts S1/S4, were used to define a linear calibration function (Equation 1). Then, δ 15Nα cal values were calculated from measured δ 15Nα values of S2‐N2O, S3‐N2O, S5‐N2O and S6‐N2O using this correction function. The combined uncertainty in δ 15Nα cal values was calculated from the uncertainty in the actual δ 15Nα values and the analyses of S1‐N2O and S4‐N2O, as well as the uncertainty in the measured δ 15Nα of the N2O gases S2‐N2O, S3‐N2O, S5‐N2O and S6‐N2O, in accordance with Equation 2. Finally, the agreement of δ 15Nα cal values (Equation 1) of the individual N2O gases (S1‐N2O–S6‐N2O) was tested against the actual δ 15Nα values, i.e. the δ 15N(NO3 −) of the respective NH4NO3 salts (S1–S6). The same procedure was applied to δ 15Nβ and δ 15N(NH4 +).

Preparation of N2O RMs and analysis for δ15N(N2O), δ18O(N2O), δ17O(N2O) and δ15NSP(N2O)

Preparation of N2O RMs

Currently available commercial N2O gases offer only limited isotopic variability. Therefore, high‐purity N2O (99.999%, Linde, Germany) was supplemented with defined amounts of 15N‐enriched/15N‐depleted and 18O‐enriched N2O dopant gas using a ten‐port two‐position valve (EH2C10WEPH, Valco Instruments Inc., Schenkon, Switzerland) with sample loops of different volumes (Table 4). The gas was transferred into evacuated Luxfer aluminium cylinders (3 L, 10 L, 20 L) with ROTAREX valves (Matar, Mazzano, Italy) to a final filling pressure below 45 bar to avoid condensation, given that the cylinder temperature remains above 15°C.

TABLE 4

Overview of N2O RMs produced from high‐purity N2O supplemented with N‐enriched/ N‐depleted and O‐enriched N2O

	Characteristic	High‐purity N2O	15N14NO	14N15NO	NN18O	15Nβ‐depl. N2O
RM1A/RM1B	High‐purity N 2 O	X
RM2	Ambient isotopic composition	X		X	X	X
RM3A/RM3B	15 N‐/ 18 O‐enriched; no SP	X	X	X	X
RM4	15 N‐ / 18 O‐highly enriched; no SP	X	X	X	X
RM5	15 N‐enriched; SP	X	X	X

The dopant gases were commercial 15N14NO and 14N15NO (isotopic purity of >98%, Cambridge Isotope Laboratories Inc., Tewksbury, USA), as well as 18O‐enriched N2O ((36.25 ± 0.10)% NN16O, (63.75 ± 0.76)% NN18O) and 15Nβ‐depleted N2O (δ 15Nα = (−2.54 ± 0.005)‰, δ 15Nβ = (−162.21 ± 0.03)‰, δ 18O = (+38.92 ± 0.003)‰), both produced and characterised at Empa. Details on the production and analysis of 18O‐enriched N2O and 15Nβ‐depleted N2O are provided in the supporting information (Supplementary Method 2). N2O RMs were provided to laboratories in 50 mL (Lab TT, Lab UEA) or 150 mL (Lab MPI) stainless‐steel flasks (CS‐07291113‐ARBOR, Arbor Fluidtec AG, Wohlen, Switzerland) for isotopic analysis.

Analysis of N2O RMs for δ 15Nα and δ 15Nβ by QCLAS at Empa (Lab Empa)

For analysis of δ 15Nα, δ 15Nβ and δ 18O in the N2O gases, a QCLAS spectrometer (Aerodyne Research Inc., Billerica, MA, USA) equipped with a continuous‐wave quantum cascade laser (cw‐QCL) with spectral emission at 2203 cm−1 and an astigmatic Herriott multi‐pass absorption cell (204 m path length) was applied. Prior to analysis, pure N2O gases (e.g. RM1–RM6, S1‐N2O–S6‐N2O) were diluted to around 50 μmol mol−1 using one cylinder of synthetic air ((20.5 ± 0.5)% O2 in N2, Messer Schweiz AG, Switzerland) into 2 L high‐pressure stainless‐steel cylinders (Luxfer, Messer Schweiz AG, Switzerland) using a ten‐port two‐position valve (EH2C10WEPH with a 1 mL sample loop, Valco Instruments Inc., Schenkon, Switzerland). A singular cylinder of synthetic air was used for all experiments to minimise differences in the oxygen content, which would otherwise affect pressure broadening of absorption lines, result in differences in apparent isotopologue mole fractions and increase uncertainties. The selection of synthetic air as diluent is somewhat arbitrary and not meant to represent an alternative for a full‐air matrix for high‐accuracy ambient N2O isotope analysis, which would enclose noble and trace gases depending on the analytics and accuracy requirements.

The spectroscopically determined isotope ratios were related to the isotope‐ratio scales realised by Toyoda et al through the analysis of calibration gases CG1 (δ 15Nα = (+25.73 ± 0.24)‰, δ 15Nβ = (+25.44 ± 0.36)‰, δ 18O = (+35.86 ± 0.22)‰) and CG2 (δ 15Nα = (−48.59 ± 0.25)‰, δ 15Nβ = (−46.11 ± 0.43)‰, δ 18O = (+27.37 ± 0.11)‰). The isotopic composition of the calibration gases had been previously analysed by Sakae Toyoda at the Tokyo Institute of Technology using their analytical technique as a link to the international scales.

For the analysis of N2O RMs by QCLAS, the site‐specific isotopic information provided by NH4NO3‐derived N2O gases S1‐N2O (δ 15Nα = (−1.41 ± 0.21)‰, δ 15Nβ = (+0.33 ± 0.12)‰) and S4‐N2O (δ 15Nα = (+52.36 ± 0.15)‰, δ 15Nβ = (+53.06 ± 0.16)‰) was propagated to the N2O RMs (RM1–RM5). For this, the N2O RMs were analysed together with S1‐N2O and S4‐N2O, as described in the preceding section, to propagate the moiety‐specific isotopic composition defined by S1 and S4 to the novel RMs (Equation 1). An uncertainty assessment was conducted according to Equation 2 including uncertainties of S1‐N2O and S4‐N2O, as discussed above, their analyses, and the analyses of RMs.

Analysis of N2O RMs for δ 15Nα, δ 15Nβ and δ 18O by DI‐IRMS and δ 17O by HR‐IRMS at Tokyo Institute of Technology (Lab TT)

N2O RMs were analysed for δ 15N, δ 15Nα, δ 15Nβ, δ 18O and δ 15NSP values with a dual‐inlet (DI) MAT 252 mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) against an isotopically characterised laboratory tank of pure N2O (N2O‐5N, Showa Denko, >99.999% chemical purity); C1: δ 15N = (−2.4 ± 0.4)‰, δ 15Nα = (−4.5 ± 0.4)‰, δ 15Nβ = (−0.3 ± 0.8)‰, δ 18O = (+23.3 ± 1.2)‰. IRMS analysis of the N2O intramolecular 15N distribution was based on the quantification of the fragment NO+ (m/z 30 and 31) and molecular N2O+ (m/z 44, 45 and 46) ions to calculate isotope ratios for the entire molecule and the central (α) and terminal (β) N atom. Analysis of δ 15N (45/44) and δ 15Nα involves correction for interfering 14N2 17O+ (m/z 45) and 14N17O+ (m/z 31) using actual Δ17O values analysed at the University of East Anglia (UEA). For the analysis of δ 15Nα and δ 15Nβ, rearrangement of N atoms (Nα and Nβ) in the ion source was considered. The δ 15N, δ 15Nα and δ 15Nβ values of the local reference gas were previously anchored to Air‐N2 by NH4NO3 decomposition, whereas the δ 18O value was anchored to VSMOW by converting N2O into CO2 with graphite and a platinum foil (Yoshida, unpublished data). The analytical uncertainties were calculated from the uncertainty of the in‐house working N2O standard gases and the standard deviation for repeated measurements of the sample gas (N2O RM) and the in‐house working N2O standard following the law of error propagation. Specifically, the uncertainty of the in‐house working N2O standard gas for δ 15N, δ 15Nα and δ 15Nβ values comprises both the uncertainty in the δ 15N(NH4 +) and δ 15N(NO3 −) analysis and the repeatability of the NH4NO3 decomposition reaction. For δ 18O, the uncertainty of the in‐house working N2O standard gas includes the repeatability of the conversion reaction of N2O into CO2 with graphite. δ 17O signatures of three N2O RMs (RM1A, RM3A, RM4) were analysed by high‐resolution IRMS (MAT 253 Ultra, Thermo Scientific, Bremen, Germany). Experimental details of this prototype analyses are provided in the supporting information (Supplementary Method 3).

Analysis of N2O RMs for δ 15N and δ 18O by EA/IRMS and DI‐IRMS at MPI‐BGC (Lab MPI)

Analysis for δ 15N by EA/IRMS (MPI‐I)

δ 15N values of the N2O RMs were determined using a modified EA/IRMS system (EA 1110 CHN combustion analyzer, CE Instruments Ltd, Wigan, UK; Delta plus isotope ratio mass spectrometer, Thermo Fisher Scientific, Bremen, Germany). The system and the method used have been described by Sperlich et al. The δ 15N values of the sample N2O were scaled to IAEA‐N‐1 and USGS32. In addition to the sample gases, an in‐house standard N2O gas NINO was analysed in each sample run, which was used as an anchor for δ 15N measurements by DI‐IRMS. USGS40, and the in‐house standards Ali‐j3 (δ 15N = (−1.51 ± 0.1)‰; acetic anilide) and Caf‐j3 (δ 15N = (−15.46 ± 0.1)‰; caffeine), were analysed in each daily run as quality controls, but not used for data correction.

Analysis for δ 15N and δ 18O by dual‐inlet IRMS (MPI‐II)

The N2O RMs were analysed twice (September 2019, February 2021) on a DI‐IRMS system (MAT253, Thermo Fisher Scientific, Bremen, Germany) using separately subsampled flasks. We note that the published δ values for USGS51 and USGS52 are average values with a rather large deviation between laboratories. Therefore, we scaled the DI‐IRMS δ 15N analyses to the in‐house standard NINO using the value reported for the primary calibration using EA/IRMS (δ 15N = (+0.54 ± 0.21)‰). The δ 18O value of NINO (δ 18O = (+39.94 ± 0.34)‰) was determined by setting the δ 18O of USGS51 equal to the average value from the interlaboratory comparison (δ 18O = (+41.45 ± 0.34)‰) published by Ostrom et al. In addition, USGS52 was analysed to test the consistency of the results (shown in Table 9) but not used for correction.

TABLE 9

DI‐IRMS analyses of RMs, USGS51, USGS52 and NINO at MPI‐BGC (MPI‐II). Analyses were conducted in two campaigns in September 2019 and February 2021 on individual sample flasks. For RM1A, in each campaign three flask samples were analysed; for RM2, two flask samples were analysed in 2021. Referencing and 17O corrections considered actual Δ17O values: δ values were referenced to Air‐N2 and VSMOW using the in‐house working standard NINO (δ 15N) and USGS51 (δ 18O) and calculated according to Kaiser et al. n indicates the number of repeated analyses per campaign. Uncertainties for individual campaigns are calculated following the law of error propagation. For the uncertainty of the weighted mean, the uncertainty of the working standard was applied, which was considered as a conservative approach. All values are reported in ‰

	Sep 2019					Feb 2021					Weighted mean ± σ
	δ 15N	σ	δ 18O	σ	n	δ 15N	σ	δ 18O	σ	n	δ 15N	δ 18O
RM1A	+0.30	0.22	+39.51	0.35	9	+0.29	0.23	+39.48	0.38	9	+0.29 ± 0.21	+39.50 ± 0.34
RM1B	+0.20	0.23	+39.14	0.36	3	+0.19	0.24	+39.10	0.38	3	+0.20 ± 0.21	+39.12 ± 0.34
RM2	+6.96	0.23	+44.37	0.35	3	+6.94	0.24	+44.32	0.38	6	+6.95 ± 0.21	+44.35 ± 0.34
RM3A	+53.12	0.25	+103.60	0.37	3	+53.09	0.26	+103.50	0.41	3	+53.11 ± 0.21	+103.55 ± 0.34
RM3B	+16.10	0.23	+55.58	0.35	3	+16.08	0.24	+55.55	0.39	3	+16.09 ± 0.21	+55.57 ± 0.34
RM4	+104.33	0.28	+154.93	0.38	3	n.a.	n.a.	n.a.	n.a.	n.a.	+104.33 ± 0.21	+154.93 ± 0.38
RM5	+33.48	0.23	+39.77	0.35	3	+33.44	0.25	+39.74	0.38	3	+33.46 ± 0.21	+39.76 ± 0.34
USGS51	+0.92	0.22	+41.45 a	0.35 a	6	n.a.	n.a.	n.a.	n.a.	n.a.	+0.92 ± 0.22	+41.45 ± 0.35 a
USGS52	+0.07	0.22	+40.89	0.35	6	n.a.	n.a.	n.a.	n.a.	n.a.	+0.07 ± 0.22	+40.89 ± 0.35
NINO	+0.54 b	0.22 b	+39.90	0.35	6	+0.53 c	0.23 c	+39.90 c	0.38 c	9	+0.54 ± 0.21 b	+39.90 ± 0.34

n.a. not analysed.

Average of laboratory results from Ostrom et al taken for referencing of δ 18O

Value provided by EA/IRMS analysis (Table S4, supporting information), value taken for referencing of δ 15N

Analysed as quality control.

In contrast to the EA/IRMS technique, where δ 15N is measured from N2 gas, the DI‐IRMS method allows the analyses of δ 15N and δ 18O values by simultaneously recording m/z 44 (14N14N16O+), 45 (15N14N16O+, 14N15N16O+, 14N14N17O+) and 46 (14N14N18O+, 15N15N16O+, 14N15N17O+) ion currents. δ 15N and δ 18O values for N2O RMs were calculated according to Kaiser et al to correct for isobaric interferences, for which the Δ17O values determined by UEA were used.

The uncertainty of the analyses was calculated from the uncertainty of δ 15N and δ 18O measurements of the N2O standard gases (NINO, USGS51) and from the standard deviation for repeated measurements of the sample gas (N2O RM) and the N2O standards, following the law of error propagation.

Analysis of N2O RMs for δ 15N, δ 18O and δ 17O by IRMS at UEA (Lab UEA)

Analysis for δ 15N, δ 18O and δ 17O by GC/IRMS (UEA‐I)

The N2O RM samples and a N2O‐MG‐6.0 working reference (99.9999% chemical purity, N2O‐MG‐6.0, Messer‐Griesheim, Krefeld, Germany) were diluted to 0.09 mmol mol−1 in N2 (zero grade, BOC, UK), filled into 20 mL serum vials (Wheaton, Fisher Scientific, Loughborough, UK) and analysed for δ(Ν2Ο) and δ(Ν2Ο) on a custom‐built automated cryogenic extraction and purification system comprised of an autosampler, a valve system, and PoraPLOT Q pre‐ and main columns (Agilent Technologies, Santa Clara, USA), coupled to a GEO 20‐20 isotope ratio mass spectrometer (Sercon Ltd, Crewe, UK).

Using the same mass spectrometer, these samples were also analysed for δ(O2) = δ 17O,  δ(O2) ≈ δ 18O (the error of this approximation is <0.01‰) and 29 δ(Ν2) = δ 15Ν after cryogenic N2O extraction and decomposition to N2 and O2 with a 500 mm long pure gold tube (1.6 mm OD, 0.6 mm ID; Heimerle & Meule, Pforzheim, Germany) held at 854°C. N2 and O2 were separated directly (without further cryofocussing) on a molecular‐sieve 5‐Å PLOT main column (Restek, Bellefonte, USA, 30 m × 0.32 mm, 30 μm, 30°C, 1.3 mL min−1 (at 20°C and 1 bar, standard temperature and pressure)). The quantitative conversion of N2O was verified by swapping the molecular‐sieve main column for the PoraPLOT Q main column and testing for residual N2O with the mass spectrometer. The raw δ 17O and δ 18O measurements were affected by scale compression. To correct for this, a logarithmic scale normalisation , was applied using the δ 18O value of +112.4‰ (relative to N2O‐MG‐6.0) derived from the δ(Ν2Ο) measurements of the diluted RM4 sample measured on the GEO 20‐20 mass spectrometer. The same normalisation was used for δ 17O as for δ 18O because no N2O reference material with a calibrated δ 17O value was available. No scale‐normalisation was applied to the δ 15N measurements. Uncertainties were calculated using the law of error propagation from the standard deviations of replicate measurements against the working reference gas and the calibration uncertainties of the working reference gas against Air‐N2 and VSMOW.

Analysis for δ 15N, δ 18O and δ 17O by dual‐inlet IRMS (UEA‐II)

The N2O RM samples were analysed for δ(Ν2Ο) and δ(Ν2Ο) with respect to the N2O working reference N2O‐MG‐6.0 using the dual‐inlet system of a Finnigan MAT 253 isotope ratio mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The N2O‐MG‐6.0 working reference has been calibrated by Kaiser et al, who reported values of δ 15N = (+1.01 ± 0.06)‰ with respect to Air‐N2, as well as δ 18O = (+38.45 ± 0.22)‰ and δ 17O = (+19.66 ± 0.11)‰ with respect to VSMOW. Actual δ 17O values of N2O RMs analysed with the Sercon GEO 20‐20 were used for the data correction according to Kaiser et al. Uncertainties were calculated using the law of error propagation from the standard deviations of replicate measurements against the working reference gas and the calibration uncertainties of the working reference gas against Air‐N2 and VSMOW.

RESULTS AND DISCUSSION

Re‐evaluation of the NH4NO3 thermal decomposition technique to propagate δ 15N(NO3 −)/δ 15N(NH4 +) to δ 15Nα(N2O)/δ 15Nβ(N2O)

In the following sections, the main procedures for anchoring of δ 15Nα and δ 15Nβ in N2O to the Air‐N2 scale and calculating uncertainties are described. Section 3.1.1 details results of δ 15N(NH4 +) and δ 15N(NO3 −) analyses in NH4NO3 salts (S1–S6) by eight isotope laboratories against international IAEA and USGS standards. Section 3.1.2 informs about optimal conditions for NH4NO3 decomposition at high yield, repeatability, and N2O purity. To enable the two‐point calibration, a number of NH4NO3 salts with different isotopic composition were produced and decomposed and the consistency of δ 15Nα and δ 15Nβ of the N2O gases (S1‐N2O–S6‐N2O) and the δ 15N(NH4 +) and δ 15N(NO3 −) of NH4NO3 salts (S1–S6) was tested (section 3.1.3).

Isotopic composition of NH4NO3 salts for δ 15N(NH4NO3), δ 15N(NO3 −) and δ 15N(NH4 +)

The isotopic composition of the prepared NH4NO3 salts (S1–S6), as analysed by the eight isotope laboratories and calibrated to Air‐N2 by analysis of IAEA and USGS standards, is indicated in Table 5. The uncertainty (σ) of individual laboratory results was estimated using the law of error propagation including the uncertainty in the international standards, their analyses, and the analyses of the NH4NO3 samples (Equations 1 and 2).

For δ 15N(NH4NO3), all results obtained by EA/IRMS were included for calculation of the weighted mean value except for results by one laboratory (Lab 4), as this laboratory used a more complicated analytical procedure with higher uncertainties. For δ 15N(NO3 −) and δ 15N(NH4 +), all laboratory results were included to calculate the weighted mean values, irrespective of the applied analytical technique.

A comparison of δ 15N(NH4NO3) with average δ 15N(NH4 +)/δ 15N(NO3 −) values indicates a good agreement to within <0.2‰. Nonetheless, results by individual laboratories for moiety‐specific δ values deviate substantially from the weighted mean. As an example, δ 15N(NO3 −) results from Lab 8 are substantially higher than those from other laboratories by an average of (+2.15 ± 0.58)‰. This may be due to the specific preparation technique applied, NH4 + removal by ion exchange, a technique which is prone to preferential retention/elution of 15NO3. In contrast, microdiffusion methods tend to underestimate δ 15N values of both NO3 − and NH4 +, which may be reflected in the δ 15N(NH4 +) values of Lab 6 but not those of Lab 5, where a similar technique was used. Conversely, systematic fractionation effects by preparation techniques should be accounted for by identical treatment (IT) of the provided IAEA and USGS standards used for data correction. In summary, analysis of δ 15N(NH4 +) and δ 15N(NO3 −) is still challenging; however, the ensemble of techniques applied in this study provides good agreement with δ 15N(NH4NO3) values.

Optimal reaction conditions for NH4NO3 thermal decomposition to N2O

Under optimised reaction conditions (270°C, 24 h) and distillation procedure, an average N2O yield of 93–95% was achieved for the decomposition of NH4NO3 salts S1–S6 (Table S1, supporting information). The yield and repeatability of the decomposition reaction are somewhat better than reported in our earlier study (91.2–93.5%) and published by Toyoda et al ((90.1 ± 3.7)%, n = 3), surpasses results by Westley et al ((65.6 ± 5.1)%, n = 20). A further increase in the yield of the NH4NO3 decomposition was achieved by conducting the reaction in a NH4HSO4–(NH4)2SO4 melt (around 2%), as suggested for industrial applications by Szabó et al. This variant displayed comparable δ 15Nα values but a loss in the δ 15Nβ information due to NH4 + salt addition, and was thus not continued.

No correction was applied to δ 15Nα and δ 15Nβ for the loss in N2O (around 5–7%), mainly due to uncertainties in the reaction mechanisms (incomplete decomposition or side‐reaction), which makes it difficult to estimate the effect on δ values. Assuming incomplete reaction accompanied by fractionation effects, according to our earlier study, a 5% reduction in yield for S1–S6 should result in 0.7/3.0/1.8‰ lower δ 15Nα/δ 15Nβ/δ 15N values, respectively. However, a much smaller difference in δ 15N was observed when comparing results of N2O RMs analysed by QCLAS (calibrated by NH4NO3 decomposition) with IRMS analyses. Therefore, our assumption is that the decrease in yield is at least partly caused by a “branching” side reaction, e.g. nitrogen gas (N2) production, which was observed to display higher δ 15N(N2) values. We speculate that N2 production has a minor effect on δ 15Nα, δ 15Nβ and δ 15NSP, but the effect is expected to depend on the timing of N2 generation, which is not known.

Consistency of isotopic composition of S1‐N2O–S6‐N2O

A general goal of the current project was to provide a link to the Air‐N2 scale and to determine the N2O site‐specific isotopic composition across a wide range of δ values. Therefore, the consistency of the isotopic composition of the N2O gases (δ 15Nβ and δ 15Nα, S1‐N2O–S6‐N2O) and the NH4NO3 salts (δ 15N(NH4 +) and δ 15N(NO3 −), S1–S6) was tested. The detailed procedure is described in section 2.1.3. In short, assuming the validity of the NH4NO3 decomposition reaction, measured δ 15Nα values of S1‐N2O/S4‐N2O and actual δ 15Nα values, i.e. δ 15N(NO3 −) of the educt NH4NO3 salts S1/S4, were used to define a linear calibration function. δ 15Nα cal values of S2‐N2O, S3‐N2O, S5‐N2O and S6‐N2O were calculated from measured δ 15Nα using this correction function and compared against actual values (Table 6).

TABLE 6

Consistency check for δ 15Nα cal, δ 15Nβ cal, δ 15NSP cal and δ 15Ncal of N2O gases (S2‐N2O, S3‐N2O, S5‐N2O, S6‐N2O) as analysed by QCLAS and referenced to the actual isotopic composition of S1‐N2O and S4‐N2O; against the actual isotopic composition of the same gases, expressed by δ 15N(NO3 −), δ 15(NH4 +), δ 15N(NO3 −)–δ 15N(NH4 +) and δ 15N(NH4NO3) of the respective NH4NO3 substrates (S2, S3, S5, S6). For details see section 2.1.3. The number of repetitions (n) for S2‐N2O/S3‐N2O analysis is 3, for S5‐N2O and S6‐N2O it is 10. All values are reported in ‰

	Isotopic composition of N2O as analysed by QCLAS (Sx‐N2O)
	δ 15Nα cal	σ	δ 15Nβ cal	σ	δ 15NSP cal	σ	δ 15Ncal	σ
S2‐N 2 O	+13.20	0.23	+13.99	0.29	−0.79	0.37	+13.60	0.37
S3‐N 2 O	+13.70	0.17	+0.36	0.24	+13.34	0.30	+7.03	0.30
S5‐N 2 O	+113.53	0.24	+103.67	0.32	+9.78	0.41	+108.60	0.41
S6‐N 2 O	−51.26	0.17	−50.03	0.24	−1.27	0.30	−50.60	0.30

Results of δ 15Nα cal/δ 15Nβ cal/δ 15NSP cal for S2‐N2O and S3‐N2O agree within expanded uncertainties (2 x σcal, Equation 2) with the isotopic composition of the substrate NH4NO3 (S2, S3; Table 5). In contrast, for S5‐N2O and S6‐N2O, δ 15Nα cal/δ 15Nβ cal/δ 15NSP cal values of the N2O gases show a significant deviation from δ 15N(NO3 −)/δ 15N(NH4 +)/δ 15N(NO3 −) – δ 15N(NH4 +) of the respective salts (S5, S6). Assuming similar fractionation effects for decomposition of all NH4NO3 salts (S1–S6), provided the comparable decomposition yield (Table S1, supporting information), we conclude that the deviation is caused by non‐linearities either in N2O isotope analysis by QCLAS or in δ 15N(NO3 −) and δ 15N(NH4 +) analyses of the NH4NO3 salts. The latter is more plausible, as the QCLAS analyses using the same calibration approach showed good agreement with independent IRMS measurements for N2O RM with high 15N enrichment (see RM4, Table S2, supporting information). The observed deviations were highest for δ 15Nβ cal to δ 15N(NH4 +) (e.g. S5), which agrees with earlier studies indicating challenges in δ 15N(NH4 +) analysis, but this may also be due to the lack of available international standards for δ 15N(NH4 +) that cover δ values above (+53.75 ± 0.24)‰ (USGS26) and below (−30.41 ± 0.27)‰ (USGS25).

In summary, our results demonstrate consistency of the isotopic composition of the N2O gases from around zero (S1‐N2O) to 15N‐enriched (S4‐N2O) and of the substrate NH4NO3 salts (S1–S4). Thereby, our study covers a much larger range of δ values (> 50‰ in δ 15Nα cal and δ 15Nβ cal) than earlier studies, , and provides a robust link to the Air‐N2 scale. At very high and low 15N enrichment (S5‐N2O, S6‐N2O), the calibration approach using NH4NO3 decomposition is more challenging, probably due to less satisfying analytical accuracy of δ 15N(NH4 +) measurements to date. As the N2O gases S5‐N2O and S6‐N2O were not included in the analysis of N2O RMs, their enhanced uncertainty in δ 15Nα cal and δ 15Nβ cal does not affect the data quality of N2O RMs.

Isotopic composition of N2O RMs

Isotopic composition of N2O RMs for δ 15NSP by QCLAS and IRMS

The novel N2O RMs (RM1–RM5) were calibrated against Air‐N2 by both QCLAS (Lab Empa) and IRMS (Lab TT) analyses. For QCLAS analyses, two N2O gases produced by NH4NO3 decomposition (S1‐N2O, S4‐N2O) were applied to define a calibration function and propagate the isotopic information of the NH4NO3 salts (δ 15N(NO3 −), δ 15N(NH4 +)) to the N2O RMs (δ 15Nα, δ 15Nβ). δ 15NSP and δ 15N were calculated using definitions and their uncertainty estimated using the law of error propagation. In Table 7, δ 15NSP values acquired by QCLAS (Lab Empa) using the calibration approach established in this study are compared with results provided by DI‐IRMS (Lab TT) using a previously published link to the Air‐N2 scale. The complete QCLAS and DI‐IRMS datasets for N2O RMs are shown in Tables S2 and S3 (supporting information).

TABLE 7

δ 15NSP analyses of N2O RMs by QCLAS (Lab Empa) referenced to Air‐N2 by NH4NO3 decomposition as performed in this study (S1‐N2O/S4‐N2O) and by DI‐IRMS (Lab TT). All values are reported in ‰

	Lab Empa (QCLAS)		Lab TT (DI‐IRMS)		Difference δ 15NSP (Lab Empa – Lab TT)
	δ 15NSP	σ	δ 15NSP	σ		σ
RM1A	+0.47	0.26	−1.04	0.91	+1.51	0.95
RM1B	+0.30	0.30	−1.19	0.91	+1.49	0.96
RM2	+18.92	0.24	+17.00	0.91	+1.92	0.94
RM3A	−2.13	0.37	−4.13	0.93	+2.00	1.00
RM3B	+1.01	0.23	−0.68	0.91	+1.69	0.94
RM4	+0.00	0.60	−2.75	0.93	+2.75	1.11
RM5	+21.96	0.33	+20.20	0.91	+1.76	0.97

Results in Table 7 indicate a 1.5–2.7‰ offset in δ 15NSP measurements by DI‐IRMS (Lab TT, Tokyo Institute of Technology) and QCLAS (Lab Empa) across all N2O RMs. This is most likely attributable to the calibration of the position‐dependent δ values with respect to Air‐N2 via the NH4NO3 decomposition technique, which were performed independently for the two labs. Incidentally, for the NH4NO3 salts S1–S4, the δ 15N(NO3 −) results provided by Lab TT were always lower ((−0.63 ± 0.59)‰), while δ 15N(NH4 +) values were higher than the respective weighted mean values ((+0.49 ± 0.25)‰), which would lead to 1.12‰ lower δ 15NSP values (Table 5).

A similar 1.5–2.0‰ difference in δ 15NSP results was recently detected by Kantnerová et al using an independent approach, equilibrating N2O at 200°C over a catalyst and comparing theoretical predictions with analytical results traceable to the δ 15NSP scale of Lab TT. One previous comparison using an independent link to the Air‐N2 scale also indicated 1.5‰ higher δ 15NSP values: (+20.2 ± 2.1)‰ vs. (+18.7 ± 2.2)‰ for ambient tropospheric N2O. Other studies confirmed the δ 15NSP measurements by the Tokyo Institute of Technology, using the NH4NO3 decomposition technique. , The uncertainty of both approaches, however, was quite high.

We conclude that the realisation of the link between δ 15NSP and the Air‐N2 scale with high accuracy is still challenging and the current realisation of the Air‐N2 scale for δ 15NSP provided by USGS51 and USGS52 may lead to too low δ 15NSP values and should be revisited in future studies.

Isotopic composition of N2O RMs for δ 15N by IRMS

δ 15N values of N2O RMs were analysed by IRMS in three different laboratories using independent links to the Air‐N2 scale (Table 8). Results display a consistent offset of (+0.22 ± 0.05)‰ and (+0.46 ± 0.14)‰ between Lab MPI‐I (EA/IRMS, Thermo Delta Plus, MPI‐BGC) and Lab UEA‐I (Sercon GEO 20‐20, UEA) versus Lab TT (Thermo MAT252, Tokyo Institute of Technology). A slightly larger offset was detected for the N2O RMs USGS51 and USGS52 (Table 9) by a comparison of published provisional values (Lab TT) and results of MPI‐II, with δ 15N values of (+0.92 ± 0.22)‰ (USGS51) and (+0.07 ± 0.22)‰ (USGS52). These values fall between results published by laboratories 7 and 8 (USGS and BGC‐IsoLab) in Ostrom et al, and are lower than results of the other laboratories, highlighting an ongoing scaling problem in δ 15N(N2O) measurements. A similar offset between laboratories had already been detected earlier and was attributed to differences in the propagation of the Air‐N2 scale to δ 15N(N2O). , , To account for differences between individual approaches to anchor laboratory results to scales, a weighted mean value was calculated for N2O RMs.

TABLE 8

δ 15N analyses of N2O RMs by IRMS at the Tokyo Institute of Technology (Lab TT: Thermo MAT252), MPI‐BGC (Lab MPI‐I: Thermo Delta Plus, Lab MPI‐II: Thermo MAT 253), and UEA (Lab UEA‐I: Sercon GEO 20‐20, Lab UEA‐II: Finnigan MAT 253) using independent calibration approaches. The 17O correction of DI‐IRMS was conducted using actual Δ17O values. All values are reported in ‰. The full set of analyses for all laboratories is provided in Table 9 (Lab MPI‐II) and in the supporting information (Lab TT: Table S3, Lab MPI‐I: Table S4, Lab UEA‐I: Table S5, Lab UEA‐II: Table S6)

	Lab TT	Lab MPI‐I	Lab MPI‐II	Lab UEA‐I	Lab UEA‐II	σ TT	σ MPI‐I	σ MPI‐II	σ UEA‐I	σ UEA‐II	Weighted mean ± σ
RM1A	+0.67	+0.44	+0.29	+0.29	+0.28	0.45	0.16	0.21	0.13	0.06	+0.30 ± 0.05
RM1B	+0.53	+0.33	+0.20	+0.24	+0.19	0.45	0.14	0.21	0.10	0.06	+0.22 ± 0.05
RM2	+7.31	+7.09	+6.95	+6.73	+6.94	0.45	0.16	0.21	0.07	0.06	+6.88 ± 0.04
RM3A	+53.41	+53.25	+53.11	+52.69	+53.09	0.47	0.15	0.21	0.11	0.07	+53.02 ±   0.05
RM3B	+16.45	+16.14	+16.09	+15.96	+16.08	0.46	0.14	0.21	0.17	0.06	+16.08 ±   0.05
RM4	+104.54	+104.39	+104.33	+104.18	+104.30	0.50	0.37	0.28	0.13	0.08	+104.28 ±   0.06
RM5	+33.76	+33.52	+33.46	+33.38	+33.45	0.46	0.21	0.21	0.10	0.06	+33.44 ± 0.05

In contrast, differences between analytical techniques applied within one lab, thus using the same link to the scale, were smaller than offsets between laboratories: (+0.10 ± 0.04)‰ for Lab MPI and (+0.12 ± 0.14)‰ for Lab UEA. This indicates that both EA/IRMS and GC‐IRMS and DI‐IRMS can achieve high accuracy, provided that an accurate link to the scale and Δ17O data are available. Consistency of N2O RM flask subsamples was demonstrated using DI‐IRMS (Lab MPI‐II, Table 9) by replicate sampling and analysis in two campaigns in September 2019 and February 2021. For RM1A, a total of six independent flask samples were analysed; the results agreed to within 0.02‰ for δ 15N(N2O) and 0.03‰ for δ 18O(N2O) (2σ, data not shown).

Isotopic composition of N2O RMs for δ 18O and δ 17O by IRMS

δ 18O values of N2O RMs were analysed by IRMS in three different laboratories (Table 10). Results show deviations of (+0.30 ± 0.13)‰, (+0.22 ± 0.24)‰ and (+0.07 ± 0.38)‰ between Lab TT and Lab MPI‐II, Lab UEA‐I and Lab UEA‐II, respectively. Differences were highest for N2O RMs with high δ 18O values (RM3A, RM4), indicating a potential scaling or scale compression issue. Measurements were not completely independent for all laboratories, as the results for Lab MPI‐II were referenced to average δ 18O values of USGS51 in Ostrom et al, which in turn was determined by seven laboratories.

TABLE 10

δ 18O analyses of N2O RMs by IRMS at the Tokyo Institute of Technology (Lab TT: Thermo MAT252), MPI‐BGC (Lab MPI‐II: Thermo MAT 253), and UEA (Lab UEA‐I: Sercon GEO 20‐20, Lab UEA‐II: Finnigan MAT 253). All values are reported in ‰. The full set of analyses for all laboratories is provided in the supporting information (Lab TT: Table S3, Lab UEA‐I: Table S5)

	Lab TT	Lab MPI‐II	Lab UEA‐I	Lab UEA‐II	σ TT	σ MPI‐II	σ UEA‐I	σ UEA‐II	Weighted mean ± σ
RM1A	+39.37	+39.50	+39.06	+39.22	1.24	0.34	0.25	0.22	+39.22 ±   0.15
RM1B	+38.86	+39.12	+38.77	+38.83	1.24	0.34	0.24	0.22	+38.86 ±   0.15
RM2	+44.08	+44.35	+43.69	+44.02	1.25	0.34	0.24	0.22	+43.96 ±   0.15
RM3A	+103.21	+103.55	+103.04	+102.78	1.30	0.34	0.27	0.24	+103.04 ±   0.16
RM3B	+55.28	+55.57	+54.98	+55.13	1.26	0.34	0.26	0.22	+55.17 ±   0.15
RM4	+154.35	+154.93	+155.17	+153.63	1.36	0.38	0.39	0.24	+154.25 ±   0.18
RM5	+39.50	+39.76	+39.43	+39.50	1.29	0.34	0.24	0.22	+39.52 ±   0.15

δ 17O values were determined by GC/IRMS at UEA (Lab UEA‐I) and showed a (0.98 ± 0.27)‰ offset to prototypical measurements by HR‐IRMS (MAT253 ULTRA) at the Tokyo Institute of Technology (Lab TT; Table 11). Consistency of GC/IRMS results agreed with an approximation, where the δ 17O was calculated from the 17O content of the 18O‐labelled H2O used for 18O‐labelled NH4NO3 and N2O production (certificate of analysis provided by Medical Isotopes Inc., USA; see Supplementary Method 2, supporting information). A 1‰ error in δ 17O results in around −0.1‰ error in δ 15Nα, when used for correction of DI‐IRMS measurements.

TABLE 11

δ 17O analyses of N2O RMs by GC/IRMS at UEA (Lab UEA‐I), HR‐IRMS at the Tokyo Institute of Technology (Lab TT), and predictions based on mixing of 18O‐labelled N2O with commercial N2O. All values are reported in ‰

	Lab TT	Lab UEA‐I	Predicted	σ TT	σ UEA‐I
RM1A	+21.60	+20.33	+20.4	0.08	0.59
RM1B		+20.88	+20.2		0.56
RM2		+20.87	+20.8		0.40
RM3A	+24.40	+23.78	+24.5	0.21	0.54
RM3B		+21.22	+21.5		0.24
RM4	+27.75	+26.71	+27.6	0.35	0.83
RM5		+20.90	+20.6		0.44

CONCLUSIONS

Within the SIRS project, we established seven pure N2O isotope RMs, which were analysed by specialised laboratories against the international isotope‐ratio scales. The established N2O isotope RMs offer a wide coverage of δ values (Table 12) beyond the currently available standards USGS51 and USGS52. This will enable future users to implement the recommended two‐point calibration approach for IRMS instrumentation, and, upon dilution with an appropriate gas matrix, for laser spectroscopic techniques as well. , In addition, the gases have been characterised for their δ 17O signatures in order to improve data quality/correction algorithms with respect to δ 15NSP and δ 15N analysis by mass spectrometry. In summary, the novel N2O isotope RMs are expected to improve compatibility between laboratories and accelerate the progress in this emerging field of research.

TABLE 12

Weighted mean δ values for the N2O RMs. All values are reported in ‰

	δ 15NSP	σ	δ 15N	σ	δ 18O	σ	δ 17O	σ
RM1A	−1.04	0.91	+0.30	0.05	+39.22	0.15	+20.33	0.59
RM1B	−1.19	0.91	+0.22	0.05	+38.86	0.15	+20.88	0.56
RM2	+17.00	0.91	+6.88	0.04	+43.96	0.15	+20.87	0.40
RM3A	−4.13	0.93	+53.02	0.05	+103.04	0.16	+23.78	0.54
RM3B	−0.68	0.91	+16.08	0.05	+55.17	0.15	+21.22	0.24
RM4	−2.75	0.93	+104.28	0.06	+154.25	0.18	+26.71	0.83
RM5	+20.20	0.91	+33.44	0.05	+39.52	0.15	+20.90	0.44

PEER REVIEW

The peer review history for this article is available at https://publons.com/publon/10.1002/rcm.9296.

Table S1. Minimum, maximum, and average reaction yield for NH4NO3 thermal decomposition at 270°C for salts S1 – S6. S1* indicates the decomposition of S1 in a NH4HSO4‐(NH4)2SO4 melt. n indicates the number of decomposition experiments.

Table S2. Isotopic composition of RMs, analysed by QCLAS at Empa versus S1‐N2O and S4‐N2O to calculate δ 15Nα, δ 15Nβ, δ 15NSP, and δ 15N values. n indicates the number of analyses. Uncertainties are calculated using the law of error propagation involving the uncertainties in S1‐N2O and S4‐N2O, their analyses, and the analyses of N2O RMs but do not enclose deviations due to fractionation or branching effects during NH4NO3 decomposition. All values are reported in ‰.

Table S3. Isotopic composition of RMs, analysed by IRMS at Tokyo Institute of Technology (Lab TT) versus an isotopically characterised in‐house working standard to calculate δ 15Nα, δ 15Nβ, δ 15NSP, δ 15N, and δ 18O values on the “Tokyo Tech scale”. n indicates the number of analyses. Uncertainties are calculated using the law of error propagation. All values are reported in ‰.

Table S4. δ 15N of RMs, the in‐house N2O standard gas (NINO), and a number of quality control standards, analysed by Lab MPI (EA‐IRMS, Thermo Delta plus, MPI‐I) versus primary reference materials and second scale anchor of the Air‐N2 scale (IAEA‐N1, USGS32). n indicates the number of analyses. Expanded uncertainties are calculated following the law of error propagation. For the quality control, standards target values and references are provided as well. All values are reported in ‰.

Table S5. Isotopic composition of RMs, analysed as N2O diluted to 0.09 mmol mol−1 on Sercon GEO 20–20 IRMS (UEA‐I) after gold decomposition, scale‐normalised to the δ 18O value of RM4. n indicates the number of analyses. Uncertainties are calculated using the law of error propagation from the standard deviations of replicate measurements against the working reference gas and the calibration uncertainties of the working reference gas against Air‐N2 and VSMOW.14 All values are reported in ‰.

Table S6. Isotopic composition of RMs, analysed as pure N2O on Finnigan MAT 253 IRMS (UEA‐II) using the actual Δ 17O measurements. n indicates the number of analyses. Uncertainties are calculated using the law of error propagation from the standard deviations of replicate measurements against the working reference gas and the calibration uncertainties of the working reference gas against Air‐N2 and VSMOW.15 All values are reported in ‰.

Click here for additional data file.