A new portable sampler of atmospheric methane for radiocarbon measurements

Radiocarbon (¹⁴C) is an optimal tracer of methane emissions, as ¹⁴C measurements enable distinguishing between fossil methane and biogenic methane (CH₄). However, ¹⁴C measurements in atmospheric methane are still rare, mainly because of the technical challenge of collecting enough carbon for ¹⁴C analysis from ambient air samples. In this study, we address this challenge by advancing the system in Zazzeri et al. (2021) into a much more compact and portable sampler and by coupling the sampler with the MICADAS (MIni CArbon DAting System) accelerator mass spectrometry (AMS) system at ETH Zurich, using a gas interface.

Here, we present the new sampler setup, the assessment of the system contamination and a first inter-laboratory comparison with LARA (Laboratory for the Analysis of Radiocarbon with AMS) at the University of Bern.

With our sampling line, we achieved a very low blank, 0.7 µgC compared to 5.5 µgC in Zazzeri et al. (2021), and a sample precision of 0.9 %, comparable with other measurement techniques for ¹⁴CH₄, while reducing the sample size to 60 L of air. We show that this technique, with further improvements, will enable routine ¹⁴CH₄ measurements in the field for an improved understanding of CH₄ sources.

1 Introduction

Understanding the methane (CH₄) budget and identifying methane sources have become a priority to tackle global warming, as methane is the second-most-important anthropogenic greenhouse gas after carbon dioxide (CO₂) and because the dynamics that led to the CH₄ increase in the last decade have not been fully unravelled. Tracing CH₄ sources and monitoring mitigation strategies are urgently needed.

¹⁴C measurements of atmospheric methane can advance our knowledge of methane production processes by differentiating fossil vs. biogenic sources. This is because fossil CH₄ is depleted in ¹⁴C and, when emitted into the atmosphere, exerts a dilution effect on the ¹⁴C in the atmosphere that can be quantified. However, this research field is still underexplored, as ¹⁴C measurements of atmospheric methane are challenging.

One of the main challenges is sampling enough air for ¹⁴C analysis via accelerator mass spectrometry (AMS), as the atmospheric methane concentration is low (∼ 2 ppm). Here, we build on recent advances that have been made in the analysis of ¹⁴C in atmospheric methane. Traditionally, air was collected in pressurized cylinders using high-pressure pumps, followed by an extraction procedure in the laboratory (Eisma et al., 1995; Townsend-Small et al., 2012). Zazzeri et al. (2021) developed a new technique that separates methane carbon from ambient air while sampling, simplifying the transportation of collected samples in a small trap and minimizing the laboratory processing needed. The F¹⁴C measurement precision achieved is between 0.5 % and 1.2 %, comparable to the best precisions of alternative but more lab-intensive techniques. The laboratory-based system developed by Zazzeri et al. (2021) was applied in the quantification of fossil and biogenic proportions of CH₄ in London (Zazzeri et al., 2023). A portable system using a similar technique was demonstrated by Palonen et al. (2017), but in that study they only used samples with enriched methane concentrations of > 100 ppm, e.g. for CH₄ emissions from wetlands. Another promising recently developed technique applies chromatographic separation of CH₄ from air as it requires only 60 L of atmospheric air to be sampled in a bag (Espic et al., 2019), still achieving precisions of 1.2 %.

In this study, we advance the sensitive though simple methane sampling system in Zazzeri et al. (2021) with the portability of the system by Palonen et al. (2017), requiring as little air as demonstrated by Espic et al. (2019). The result is a compact and portable system that can be deployed in field campaigns. We present the technology advancement and the assessment of the system efficiency by quantifying the amount of extraneous carbon introduced during sample preparation and ultimately the measurement precision to be achieved. We demonstrate the method by comparing ¹⁴C measurements made by the new portable system at the Laboratory of Ion beam Physics (LIP) at ETH Zurich and by the system using bag sampling and chromatographic separation at the Laboratory for the Analysis of Radiocarbon with AMS (LARA) at the University of Bern.

2 Method

2.1 The sampling setup

The sampling system is based on four main steps as in Zazzeri et al. (2021): (1) H₂O removal with a Nafion dryer, (2) CO and CO₂ removal, (3) combustion of CH₄ to CO₂, and (4) adsorption of the combustion-derived CO₂ onto a molecular sieve sample trap. Figure 1 shows the system schematic. Ambient air is sampled through a Nafion dryer (Perma Pure gas dryer, PD-50-24) at up to 500 cm³ min⁻¹ with a KNF membrane pump (“Pump 2” in Fig. 1), controlled by a mass flow controller. The Nafion filter enables reduction of the water content to levels of 0.01 % (Zazzeri et al., 2021). Downstream of the pump, CO is oxidized to CO₂ using the Sofnocat^® catalyst before all CO₂ (from ambient air and from oxidation of CO) is removed by a trap containing 14 g of 13X molecular sieve in 1 mm pellets. This amount of molecular sieve has been found to be sufficient to trap atmospheric CO₂ in ∼ 300 L of air (see Fig. A1). After collection of the three samples, this trap is disconnected from the system via two Swagelok ball valves and then removed and regenerated by heating at 500 °C with a high-purity nitrogen back-flush step for at least 3 h, in a similar manner as in Zazzeri et al. (2021).

Figure 1 (a) Schematic of the sampling system. First, the filtered air is dried with a Nafion dryer; then, any CO₂ from ambient air and from oxidation of CO is removed by a trap. The CO₂ derived from the combustion of CH₄ is collected onto a final sample trap. Dark green lines in the 6-port valve indicate the active flow direction. In the indicated configuration, the CO₂ sensor measures the CO₂ level after the sample trap, enabling us to check for the trap breakthrough. In the alternative configuration, complete CO₂ removal prior to CH₄ oxidation can be checked. (b) Sample trap filled with 0.250 g of 13X.

After the sample air passes through the CO₂ trap, CH₄ is combusted at 800 °C in a small furnace comprising a 22 cm long quartz tube with 1 g of platinized quartz wool (Sigma-Aldrich) acting as catalyst (Petrenko et al., 2008). The H₂O derived from the CH₄ oxidation is trapped onto a magnesium perchlorate trap (“H₂O trap” in Fig. 1), while the CO₂ derived from combustion of CH₄ is collected on the sample trap (13X, 45–60 mesh) for subsequent ¹⁴C measurement. A non-dispersive infrared CO₂ sensor (NDIR FLOW^EVO from SmartGas) monitors both the completeness of CO₂ removal from air prior to methane combustion and the completeness of CO₂ collection on the sample trap. If regularly calibrated and run at constant temperature and pressure, the sensor can measure CO₂ concentrations in a range of 0 to 100 ppm with a precision of ±1 ppm. The sample trap, minimized in size for low cross contamination, can be cooled with Peltier coolers to maximize trapping efficiency and avoid sample loss.

The whole system runs either on 115/230 V AC or 48 V DC provided by a battery pack. Two 72 V 30 Ah (2160 Wh) lithium batteries are sufficient to run the sampling system for 10 h.

2.2 Sample trap and cooling system

The sample trap consists of 0.250 g of 13X 45–60 mesh molecular sieve packed in a 4 cm long 1/4 in. o.d. stainless-steel tube. The trap tube is welded to stainless-steel capillary tubing and attached to a VICI 4-port valve which can be disconnected from the sampling system (Fig. 1b) in order to release the sample for ¹⁴C analysis in the AMS. Before its first use, the sample trap is heated gradually to 650 °C in a customized oven while flushing with high-purity nitrogen. The NDIR FLOW^EVO CO₂ sensor is used to check when the trap is not releasing CO₂ anymore and does not contain any residual carbon, typically after 1 h. During sampling, the sample trap is cooled down to −10 °C using two Peltier elements (part (6) in Fig. 2). This maximizes the trapping efficiency such that 0.250 g of 13X can adsorb ∼ 60 µgC (methane carbon from ∼ 60 L of ambient air at 2 ppm), before the CO₂ breakthrough happens. The adsorption capacity can be enhanced by lowering the temperature even further.

Figure 2 Setup of the atmospheric methane sampling device. The whole system fits well into a box of 80 × 40 × 30 cm. It runs either on 115/230 V AC or 48 V DC provided by a battery pack. Major parts of the system are (1) the Nafion dryer (mostly hidden underneath), (2) pumps (partially hidden), (3) CO₂ trap, (4) flow controller, (5) furnace for CH₄ combustion, (6) sample trap with Peltier coolers, and (7) CO₂ sensor.

After sampling, the sample trap is disconnected from the system and heated at 450 °C for 10 min for sample desorption using the TSE (Tube Sealing Equipment) system (https://www.ionplus.ch/tse, last access: 8 January 2025), which enables us to measure the pressure of the desorbed gases and to quantify the amount of the CO₂ released. The desorbed CO₂ is cryogenically sealed into a glass ampule to be used in the gas interface system of the MIni CArbon DAting System (MICADAS) (Wacker et al., 2013).

Before the next sample is collected, the sample trap is cleaned of remaining CO₂ by flushing with high-purity nitrogen while heating at 550 °C for 30 min. Such a long procedure compared to other cleaning processes for other applications is only precautionary; a shorter procedure might be sufficient to remove any residual carbon from previous sampling.

2.3 AMS analysis

The sample is measured with the MICADAS accelerator mass spectrometry facilities for radiocarbon measurements in the Laboratory of Ion Beam Physics, ETH (Wacker et al., 2010). The ¹⁴C analysis using the gas interface of the MICADAS takes about 20 min and achieves a measurement precision of less than 1 % for modern samples (Wacker et al., 2013) for a sample containing 20 µg of carbon. Precisions down to 0.5 % can be achieved when measurements are repeated on ≥ 50 µg of carbon (Fahrni et al., 2013).

The combusted NOX standard (SRM-4990C, Mann, 1983; Wacker et al., 2019) and the ¹⁴C-free CO₂ premixed with helium in gas bottles were measured for standard normalization or blank correction. Measured data (Zazzeri, 2025) were evaluated with the BATS computer program, where the samples were fractionation-corrected, blank-subtracted and normalized with the NOX standard (Wacker et al., 2010) to obtain F¹⁴C values (Reimer et al., 2004).

2.4 Characterization of the extraneous contaminant carbon within the sampler

Extraneous contaminant carbon in the sample trap after a sample collection might derive from intrusion of lab air into the system, from incomplete removal of atmospheric CO₂, from residual carbon on the sample trap prior to sampling or from impurities within the combustion column.

To check for and quantify the contaminant carbon, we collected CH₄ samples of different sizes, and we followed the relationship between the measured fraction modern (F¹⁴C) vs. the sample masses given by the mass balance in Eq. (1):

$$\begin{array}{}\text{(1)}& \begin{array}{rl}{F}^{\mathrm{14}}{\mathrm{C}}_{\mathrm{meas}.}& =F^{\mathrm{14}}{\mathrm{C}}_{\mathrm{true}}\\ & +{\displaystyle \frac{\mathrm{1}}{{\mathrm{\mu }\mathrm{C}}_{\mathrm{meas}.}}}\left[{\mathrm{\mu }\mathrm{C}}_{\mathrm{add}}\times \left(F^{\mathrm{14}}{\mathrm{C}}_{\mathrm{add}}-F^{\mathrm{14}}{\mathrm{C}}_{\mathrm{true}}\right)\right],\end{array}\end{array}$$

where “meas.” indicates the measured value, F¹⁴C_true indicates the F¹⁴C value of the sampled air, µC_add indicates the carbon added into the system and F¹⁴C_add indicates its F¹⁴C value. If assuming a constant contamination, the contaminant carbon added to the system is given by the µC_add value that produces the best fitting curve through the F¹⁴C_meas. values plotted against the measured sample masses (µC_meas.). We assess the goodness of fit using reduced chi-squared statistics (i.e. “curve_fit” function from the “scipi.optimize” Python package).

To quantify the modern contaminant carbon, we collected seven samples from 10 to 70 µgC from a 2 ppm mixture of fossil methane and synthetic air with no CO₂, CO or hydrocarbons (Fossil Ref).

To check for any fossil contaminant, we collected seven samples, from 10 to 75 µgC, from a cylinder of pressurized ambient air (Ref 1), with a CH₄ mole fraction of 2040 ppb. Note that in this case, the F¹⁴C of the reference gas (F¹⁴C_true) is unknown. The amount of fossil contaminant carbon and the F¹⁴C value of the reference gas are calculated by tweaking µgC_add and F¹⁴C_true in Eq. (1) to produce the best fitting curve.

In order to verify the source of the contaminant carbon, we collected five blanks. Three blanks were collected by running the system with lab air and without combustion for 3 h. This enabled verification of any contaminant carbon deriving from atmospheric CO₂ that was not trapped in the CO₂ trap and from residual carbon in the sample trap. Two samples were collected by flushing the system with nitrogen and with the combustion furnace at 800 °C to verify that additional carbon was not produced within the combustion process. No carbon was extracted from these five blanks.

2.5 Comparison with chromatographic extraction procedure

Three samples transferred in sampling bags from the cylinder of pressurized ambient air Ref 1 were extracted at LARA, University of Bern, using 60 L of air and following the chromatographic extraction procedure in Espic et al. (2019). CO₂ derived from the sample extraction in Bern was measured using the gas interface system of the MICADAS AMS system at ETH, in the same manner as the samples collected with our portable sampler.

3 Results

Table 1 shows the F¹⁴C values and masses of the samples collected.

Table 1 Mass, F¹⁴C values and uncertainty of samples collected. Modern samples were collected from Ref 1, fossil samples were collected from Fossil Ref and “Bern” represents samples extracted following the chromatographic procedure at LARA.

Assuming a F¹⁴C_true value for Fossil Ref of 0.01 and a F¹⁴C value of the modern contaminant (F¹⁴C_add) of 1, the best fitting curve through the Fossil Ref samples indicates a constant level of modern contamination (µgC_add) of 0.5 ± 0.1 µgC. Larger samples (> 50 µgC) show an offset that can be explained with a size-dependent contamination, an additional 0.1 µgC every 10 µgC collected, which we can correct for (Fig. 3).

Figure 3 F¹⁴C values against the measured mass of the samples collected from the Fossil Ref. The grey bands represent the 1σ uncertainty bar on the curve fit.

The data collected from Ref 1 best fit onto a curve with an F¹⁴C value of 1.38 ± 0.01 for the reference gas. However, by considering only the quantified modern contamination of 0.5 µgC, we do not achieve the best fitting curve, and we need to add approximately 0.2 ± 0.1 µg of contaminant fossil carbon. Zazzeri et al. (2021) indicated that some fossil carbon might be produced within the combustion furnace; therefore, it is likely that even with our setup the combustion process led to the production of some fossil carbon.

Ref 1 samples extracted in Bern, blue markers in Fig. 4 (not included for determining the constant contamination), agree well with the F¹⁴C values for Ref 1 samples with the same mass (60 µgC) extracted at ETH, indicating that the two extraction methods are comparable. Samples of 60 µgC are equivalent to 2 or 3 h of sampling of ambient air with our portable system at 500 sccm or 60 L of ambient air with the extraction line in Bern.

Figure 4 Measured F¹⁴C values against the measured mass of the samples collected from Ref 1. Symbols in blue are the Ref 1 samples extracted in Bern and are not included for the curve fitting. The grey bands represent the 1σ uncertainty bar on the curve fit.

4 Discussion

In order to make the sampler portable, we have reduced the size of the sampler components compared to the system in Zazzeri et al. (2021). The main changes include the following:

• a smaller CO₂ trap placed before the combustion furnace, with 14 g against 60 g of molecular sieve; its adsorption capacity is demonstrated by the very low modern blank, which indicates that all the ambient CO₂ is captured while sampling;

• a new design of the sample trap, with 0.250 g of molecular sieve against 1 g, accommodated in a 4 cm long tube and connected to a single 4-port VICI valve; collection of 60 µgC (2 or3 h of sampling at 500 sccm) has been achieved by cooling down the sample trap using two Peltier elements;

• a smaller combustion furnace built at the Laboratory of Ion Beam Physics;

• connections and tubing of 1/8 in. size instead of 1/4 in.

All these modifications led to an important reduction in the level of constant modern contamination, from 5.5 ± 1.1 µgC in Zazzeri et al. (2021) down to 0.5 ± 0.1 µgC. According to the F¹⁴C measurements of our modern reference cylinder (Ref 1), we have an extra 0.2 ± 0.1 µgC of fossil contamination, leading to 0.70 ± 0.14 µgC total amount of contaminant carbon with an averaged F¹⁴C value of 0.71. We also found a size-dependent contamination of 1 %, which can be explained either with a tiny leak within the sampler or with some outgassing.

To further test the system contamination and demonstrate full separation of CH₄ and CO₂, more gas mixtures could be used for F¹⁴C measurements. For example, the system could be run using a mixture made of 2 ppm of fossil CH₄ and 400 ppm of CO₂ from combustion of OXII (oxalic acid II), diluted in N₂ or He.

The overall uncertainty for individual samples of 60 µgC, calculated by propagating the error from counting statistics and background uncertainty, is 0.9 %, comparable with other measurement techniques for ¹⁴CH₄, demonstrating that a larger sample, and therefore a longer sampling time, is not needed.

The main benefit of a portable system that needs only 60 L of air for one sample is the important time saving both in the field and in the laboratory. The sample processing time in the laboratory has been reduced significantly, and so has the likelihood of contamination or mistakes by the operator. The system, given its small size, could be placed in a vehicle, enabling sampling in a source area (such as a landfill site or an urban environment) or performing a mapping of isotopic signatures in a region.

5 Conclusions

We have advanced the CH₄ sampling system from Zazzeri et al. (2021) to a portable system that can be used in field campaigns while also reducing the contamination in the system. Further improvements could be made to automate the system so that the valve and pump switching and the flow rate are computer-controlled, making the whole sampling procedure more consistent. More samples could be collected in parallel and at the same time. In addition, the CO₂ desorbed from the sample trap is presently cryogenically trapped in glass ampules sealed for offline ¹⁴C measurements, but a direct coupling of the zeolite trap to the gas interface (Wacker et al., 2013) connected to the MICADAS AMS system could be implemented, avoiding the additional step of using glass ampules.

Full assessment of the fossil carbon contamination in ¹⁴CH₄ measurements is still challenging, because there is no modern reference material available for CH₄. The production of a modern CH₄ standard for ¹⁴C analysis, followed by an inter-laboratory comparison, should be pursued.

Overall, the combination of a selective and field deployable CH₄ sampler and sensitive AMS analysis provides a unique technology that can expand the use of ¹⁴CH₄ measurements.

Appendix A

Figure A1 CO₂ concentrations (in ppm) measured with the NDIR sensor downstream the CO₂ trap while flushing the trap with lab air. The trap starts saturating after 4 h and 30 min of flushing at a 1 L min⁻¹ flow rate. Note that the NDIR sensor measures negative values for a gas stream with no CO₂.