Stable Isotope Analysis of Greenhouse Gases Requires Analyte Preconcentration

: Nitrous oxide (N 2 O) is an important trace gas contributing to global warming and depletion of ozone in the stratosphere. Its increasing abundance is caused mainly by anthropogenic sources, such as application of fertilizers in agriculture or emissions from industry. To understand the N 2 O global budget, its sources and sinks need to be well-described and quantified. In this project, a new method for N 2 O source appointment was developed that can help with this task. The method is based on analysis of the eight most abundant isotopic molecules of N 2 O, using quantum cascade laser absorption spectroscopy (QCLAS). The applicability of the method towards the N 2 O biogeochemical cycle was demonstrated on a prominent N 2 O source (bacterial denitrification) and the most important N 2 O sink (UV photolysis) on samples prepared in laboratory experiments. An extension of the QCLAS method to natural samples can be achieved by hyphenation with a preconcentration technique that increases concentration of the analyte and standardizes the sample matrix. This article provides an overview of currently applied preconcentration techniques in the field of greenhouse-gas analysis and a description of the preconcentration device TREX that will be employed in future projects with the developed QCLAS method.


Introduction
Greenhouse gases (GHGs) have been present in the atmosphere since the formation of the Earth approx. 4.6 billion years ago. [1] The composition of the atmosphere has been changing ever since, first due to geological activity, later through the appearance of living organisms (e.g. plant photosynthesis), and nowadays due to anthropogenic activities. While the atmospheric composition evolved slowly in the past thousands of years, the recent human-induced changes have occurred very rapidly.
The most important GHGs are CO 2 , CH 4 , and N 2 O (Fig. 1). The change in their atmospheric abundance can be clearly linked to human-induced emissions, starting in the second half of the eighteenth century (Industrial Revolution), mainly with burning of fossil fuels and later introducing fertilizers in agriculture. These three substances are the main cause of GHG-radiative forcing [3] causing global warming, due to their ability to absorb infrared radiation exiting the atmosphere and emit it back towards the Earth. This so-called greenhouse effect is a natural process; without any GHGs in the atmosphere, the surface temperature on the Earth would be only around −18°C. [4] However, the rising abundances of the GHGs have disrupted the sensitive system of our living environment, causing increasing surface temperatures, thinning of the ozone layer, rising of sea levels, and alterations of natural biogeochemical processes.
To mitigate anthropogenic GHG emissions, their source and sink processes and magnitudes must be well-understood. Aside from many atmospheric monitoring programs (e.g. Global Atmosphere Watch GAW, the European ICOS program, or the American NOAA program), there have been numerous smallscale studies focusing not only on the GHG abundances, but also abundances of their isotopically substituted molecules (isoto-For both IRMS and LAS techniques, there are specific requirements on sample preparation. First, ambient concentrations of GHGs are too low for direct isotope analysis as the required amount of the analyte is substantially higher (cf .  Table 1). Second, both techniques are affected by characteristic interference issues. In the case of IRMS, isobaric interferences of species with similar nominal masses can hinder the analysis, unless sufficiently high mass resolution is achieved. For LAS, there can be spectral overlaps when species have rovibrational transitions at nearly the same absorption frequencies. Therefore, an online or offline separation of the target substance from a larger sample amount is usually required to preconcentrate the analyte into a smaller volume and remove potential interferents.

Preconcentration Te chniques
Common preconcentration techniques for isotope analysis of GHGs take advantage of different boiling points of the analytes and interferents. Their separation is achieved by trapping analytes in a cold trap (cryofocusing), while non-condensable components such as N 2 or O 2 pass the trap and are removed by evacuation. IRMS preconcentration methods apply trapping of the analyte at liquid-nitrogen temperature (−196°C) followed by separation of the GHGs by gas chromatography (GC) before entering an IRMS instrument. [23,24] Purification of CH 4 samples for clumped-isotope IRMS analysis has been done using a helium-cooled cryostat alternating between different temperature settings to remove interferents (from −253.15°C to −203.15°C).
Prior to cryofocusing, water vapor is usually removed using chemical adsorbents, e.g. magnesium perchlorate (Mg(ClO 4 ) 2 ) in an inline trap through which the gas is sampled. [10] Alternatively, water vapor can be separated cryogenically using a stainlesssteel coil immersed into an ethanol-dry ice bath (−78°C) [20] or with a Nafion permeation dryer [25] (a co-polymer membrane with high permeability for water). Such a preconcentration routine before IRMS analysis can be performed offline using a manual procedure, but automated instruments interfaced with water removal, cryofocusing and GC are usually employed. [10] Preconcentration before LAS analysis has been mostly achieved with fully automated liquid-N 2 -free devices. [14,26,27] The core part of these devices is a cryogenic refrigeration system or Stirling cooler that actively cools a stainless-steel trap filled with an adsorbent through which dried air or sampled gas are purged. A more detailed description follows in Section 4. The hyphenation of the automatic preconcentration device and a LAS instrument is suitable for unattended field observations, delivering high-precision isotope data in nearly real time. [28] pocules). [5] It has been shown that GHGs emitted from specific sources possess unique isotopic signatures that can be applied to semi-quantification of sector-specific fluxes and construction of global budgets of the GHGs. [6] Isotopic composition of GHGs is commonly analyzed by isotope-ratio mass spectrometry (IRMS). [7] The technique separates ionized molecules using electric and magnetic fields based on differing mass-to-charge ratios (m/z). Another, less prevalent technique for isotope-ratio analysis of GHGs is laser absorption spectroscopy (LAS). This technique detects rotationalvibrational motions of gaseous or volatile molecules, where frequencies of the internal molecular motions depend on their reduced masses and rotational constants. Therefore, in contrast to IRMS, LAS techniques can directly distinguish isotopocules with the same nominal mass but different molecular position of the isotopic substitution. Such species are called isotopic isomers (isotopomers); an example for N 2 O with one 15 N substitution are the isotopomers 14 N 15 N 16 O and 15 N 14 N 16 O. In addition, in the last decade, the techniques have been used to develop new methods for analysis of multiply substituted isotopocules (clumped isotopes). More details on this emerging field and the methods can be found in our previous CHIMIA publication [8] and references therein. Ta ble 1. Current ambient concentrations of the three most abundant greenhouse gases (after water vapour, H 2 O), their global warming potential, and molar amounts required for isotope analysis of their singly substituted isotopocules and clumped isotopes. greenhouse gas ambient concentration [3] global warming potential (GWP-100) a [9] sample amounts required for analysis of: singly substituted isotopic species clumped isotopes CO 2 413.2 ± 0.2 ppm 1h undreds of nmol or more [10,11] tens of μmol [12] 100 μmol [13] CH 4 1889 ± 2 ppb 27.9 hundreds of nmol [14] tens of μmol [15,16] hundreds of μmol [17,18] N 2 O3 33.2 ± 0.1 ppb 273 tens of nmol [19] 4 μmol [20] 10 μmol [21] In the second application, the QCLAS method was employed to study the most important N 2 O sink that occurs in the stratosphere -photolysis by UV light. Aliquots of 1.70% N 2 O in N 2 were gradually photolyzed by UV light in a custom-made photoreactor at two wavelengths, 200 nm and 214 nm, as UV light in this spectral region has the largest contribution to N 2 O photolysis. The gas was directly sampled every 12 hours from the photoreactor to the QCLAS instrument using an automated gas inlet system. [32] Unique isotope fractionation constants for the N 2 O isotopocules were retrieved from the obtained temporally resolved data and compared with theoretical fractionation constants that were calculated based on previously published absorption cross sections derived from first principles.
The experiments showed that stratospheric photolysis has a low impact on the clumped isotope signatures of atmospheric N 2 O. Therefore, variations in the N 2 O isotopic composition are expected to be mostly affected by sources and not by photolysis. More details on the results and their discussion can be found in ref. [32].

TREX -A Preconcentration Device for N 2 O Analysis
Sufficiently high N 2 O concentrations for direct QCLAS analysis can be obtained under laboratory settings, as was shown in the two previous applications. However, in most natural samples, N 2 O concentrations are too low for direct analysis. Therefore, as described in Section 2, a preconcentration technique must be employed to study singly substituted molecules and also the more rare clumped isotopes with sufficient accuracy.
For QCLAS analysis of the singly substituted N 2 O species, automated preconcentration systems have been implemented, e.g. in field studies of ecosystem- [33] or soil-emitted [28] N 2 O. Advantages of using such an automated preconcentration device include higher temporal data resolution and better data quality (no sampling effects) with less manual work, less coordination effort, and long-term unattended operation. In addition, the preconcentration procedure automatically removes gas matrix effects (changes in the bulk gas composition) and spectral interferences that would otherwise interfere with the QCLAS analysis. These effects occur whenever concentration of N 2 , O 2 , argon, or an IR-active compound deviate between sample and reference gases. [34] Fig. 4 shows an automated preconcentration device developed at Empa, called TRace gas EXtractor (TREX). The device can cryogenically adsorb N 2 O (or another trace gas) from sampled gas or air, separate co-adsorbed species applying stepwise temperature-controlled heating (e.g. CO), and finally desorb the purified N 2 O into the optical cell of a QCLAS instrument. The core part of the device is a stainless-steel trap filled with HayeSep D adsorbent. The trap is attached to a linear driver and thus can be pressed against a copper base plate that is cooled down by a Stirling cooler to −210°C. Upon thermal contact with the base plate, the trap is cooled down to approx. −150°C and N 2 O is adsorbed. After sufficient trapping time, the trap is moved away from the base plate and heated by active resistive heating. In the final heating step, N 2 O is desorbed, transferred to the QCLAS optical cell and diluted with a defined matrix gas prior to the QCLAS analysis. A more detailed description of the operation 3. Applications in the Award-winning Project N 2 O is an isotopically complex molecule that carries a wealth of information, through the abundance of 18 O, the site-specific distribution of 15 N in the asymmetric N 2 O molecule, and the abundance of clumped isotopes. [29,30] The award-winning PhD work focused on developing a method with quantum cascade LAS (QCLAS) for the analysis of clumped isotopic species of N 2 O and exploring isotope signatures of source and sink processes. The developed QCLAS method provides simultaneous high-precision isotope data of eight N 2 O isotopocules -the most abundant species 14 [20] The method requires samples of N 2 O of approx. 1.5% in pure N 2 .
During the PhD project, the QCLAS method, together with a new mole-fraction calibration scheme for quantification of the clumped N 2 O species, was applied to explore clumped isotope signatures as potential tracers for the global cycle of N 2 O in two projects. In the scope of the first application, Kristýna spent a 3-month research stay in the group of Prof. Naohiro Yo shida at Tokyo Institute of Technology. There, she prepared samples of N 2 O from bacterial incubations.
The majority of natural N 2 O emissions is generally attributed to bacterial denitrification and nitrification. During the research stay, the denitrifying bacteria Pseudomonas aureofaciens and the nitrifying bacteria Nitrosococcus oceani were cultivated in nitrogen-containing media and the N 2 O produced was purified using a preconcentration line built in-house. [25] Upstream of the preconcentration, CO 2 and H 2 O were removed before N 2 O was cryofocused manually in a stainless-steel loop in a liquid-N 2 bath. The purified N 2 O samples were collected into stainless-steel cylinders and shipped to Empa for QCLAS analysis. The samples were diluted with N 2 to 1.55% N 2 O prior to analysis.
Incubations of the denitrifying bacteria in a nitrate (NO 3 -) medium provided large amounts of N 2 O, as N 2 O is the final product of NO 3 − reduction for the cultivated bacterial strain of Pseudomonas aureofaciens (Fig. 2). N 2 O was sampled one day, three days, and seven days after the incubation started, to obtain temporal trends of isotopic composition of the produced gas. From the temporal trends, isotope enrichment factors during NO 3 − reduction to N 2 O were obtained. Isotopic signatures of the singly substituted species and the clumped isotopocules suggest a trans-isomeric structure of a reaction intermediate that is involved in the enzymatically catalyzed NO reduction, where two N atoms coming from two different NO 3 − species combine and the N-N bond in the final N 2 O is formed. A more detailed interpretation of the obtained data requires theoretical models, which have not been developed yet; for more details see ref. [25].
In the nitrification process, N 2 O is produced only as a side product during oxidation of hydroxylamine (NH 2 OH; Fig. 3). [31] For several repetitions of the QCLAS analysis and purging of gas lines, around 50 -100 μmol of pure N 2 O are required. However, due to relatively short lifetimes of the cultivated nitrifying bacteria Nitrosococcus oceani and low N 2 O yields, data from the QCLAS analysis of the nitrification-derived N 2 O could not be evaluated in this project. we have opened unexplored possibilities for research in the N 2 O biogeochemical cycle, demonstrated in laboratory settings on microbial production and UV photolysis.
For future projects targeting time-resolved data of clumped N 2 O isotopes, the developed QCLAS method is to be connected with an automated preconcentration unit of the described TREX design. This will be a part of a follow-up project funded by the Swiss National Science Foundation (SNF; a collaboration between Empa and University of Basel) that is going to start in summer 2022. The aim of the project is to extend the database of prominent N 2 O microbial and abiotic production pathways in pure culture and develop a theoretical framework for data interpretation. In addition, under a new SNF project at the Sustainable Agroecosystems group at ETH Zurich, automated gas chambers will be coupled to their TREX-QCLAS system. The analysis setup will be used to obtain the first online measurements of N 2 O fluxes and its isotopic composition from tropical agricultural soils in the Eldoret region of Kenya. The isotopic measurements will be used to infer N 2 O production and consumption rates for different pathways to advance our understanding of the major drivers of variability in tropical N 2 O fluxes. Funding  can be found in our previous CHIMIA publication [8] or in the publications cited in this section.
The design of such an automatic preconcentration device was first applied for analysis of ambient N 2 O by Mohn et al. [26] Based on that, Eyer et al. [14] developed a more compact device with higher cooling power, capable of extracting CH 4 from ambient air. Currently, the Sustainable Agroecosystems group of Prof. Six at ETH Zurich is the first group outside of Empa running a TREX-QCLAS system (a postdoctoral project of the main author). The instrument is based on a replicate of the TREX device that was used in the work of Ibraim et al. for field experiments. [35] The new TREX device was constructed at ETH Zurich by E. Ibraim, and finalized and tested by the main author of this contribution.
An important parameter of a preconcentration device is the capacity of the trap for the trace gas adsorption. The capacity of the ETH TREX device was tested by continuous adsorption of N 2 O from a mixture of N 2 O in synthetic air (N 2 and O 2 in 70:30 ratio) while analyzing the gas at the outlet of the trap by the QCLAS instrument. In this so-called breakthrough experiment, the TREX device trapped approx. 14 μmol of N 2 O, corresponding to the amount of N 2 O in more than 1,000 L of ambient air, and no breakthrough was observed in the experimental timeframe. In a breakthrough experiment performed at Empa on the TREX device that was originally employed for CH 4 adsorption, [14] the breakthrough of the trap capacity was observed after adsorbing 8.2 mmol of N 2 O (2,170 L of gas at 90.9 ppm N 2 O; 45.3 hours).
The trap capacity is therefore large enough to preconcentrate sufficient amounts of N 2 O even for clumped isotope analysis, provided that interfering CO 2 is removed from the sampled gas prior to adsorption (e.g. using an Ascarite trap). [35] In this application, approx. 360 L of ambient air would be processed (approx. 5 μmol of pure N 2 O per analysis). To achieve high output concentration, the current design of the TREX device requires certain modifications, mainly introducing a second, smaller trap to focus the N 2 O gas into smaller volume after desorption from the first trap. This also holds advantages with respect to separating gases with similar boiling points, e.g. O 2 and CH 4 . [16]

Conclusions
Within the award-winning PhD project, a new analytical method based on the QCLAS technique was developed. Together with a new calibration scheme, it can be directly employed for simultaneous analysis of the eight most abundant N 2 O isotopocules, including the three most abundant clumped isotopic species 14 Fig. 4. Scheme of the TREX-QCLAS system. In the first step, the analyte (in this case N 2 O) is cryofocused in an adsorbent-filled trap cooled by a Stirling cooler in the preconcentration unit TREX and the target substance is separated from potential interferents. In the second step, the analyte is purged out of the trap by a matrix gas to the optical cell of a QCLAS instrument, where the isotopic composition of the analyte is measured.

License and Terms
This is an Open Access article under the terms of the Creative Commons Attribution License CC BY 4.0. The material may not be used for commercial purposes. The license is subject to the CHIMIA terms and conditions: (https://chimia.ch/chimia/about).
The definitive version of this article is the electronic one that can be found at https://doi.org/10.2533/chimia.2022.656