Overlooked organic vapor emissions from thawing Arctic permafrost

The frozen peat and active layer soils were collected in early September 2019 from a palsa mire (Kevo, 68°53'N, 21°03'E), located in the sporadic permafrost zone of Finnish Lapland. The peatland was characterized by isolated permafrost mounds (palsas) a few meters high and surrounded by wet flarks (Zhang et al 2018). Collapse features are common along the palsa edges. The climate of the Kevo area is subarctic and relatively continental (Drebs et al 2002). The Kevo research station permafrost monitoring site reported a mean annual ground temperature of −0.2 °C and a mean annual air temperature of −1.0 °C during 2008–2009 (Christiansen et al 2010). The active layer thickness was ca. 60 cm. Borehole measurements revealed existence of warm permafrost close to 0 °C in this region, making it highly sensitive to climate change (Christiansen et al 2010). According to long-term observations of up to 65 yr, the mean annual air temperature in Kevo has shown a statistically increasing trend of +0.4 °C/decade (Merkouriadi et al 2017).

The vegetation at the site is dominated by dwarf shrubs, such as Empetrum nigrum ssp. hermaphroditum, Vaccinium vitis-idaea, Betula nana and Rhododendron tomentosum and lichens and bryophytes (Cladonia spp. and Dicranum sp.). Patches of bare peat occur on top of the palsa mounds.

The permafrost soils were sampled by hammering down a steel pipe with an inner diameter of 30 mm into the permafrost and pushing out the frozen core with a wooden rod. For permafrost samples, a 10 cm-long section from around 100–110 cm depth was collected and kept frozen throughout the transportation to Helsinki for further analysis. In Helsinki, the samples were stored in a freezer at −20 °C until the start of analysis.

For the active layer samples, a 50 cm-long section was collected from the active layer using a box corer. In the laboratory in Helsinki, the layer extending from ∼15 cm to ∼25 cm depth was chosen for further comparison measurements and stored in the freezer (−20 °C).

The permafrost and active layer samples in this study came from a single palsa. Three frozen peat replicates of the same core, P1, P2, and P3, from around 100–110 cm depth were used for the permafrost peat property analysis, while four active layer replicates of the same core, AL1, AL2, AL3, and AL4, from around 15–25 cm depth were used to study active layer peat properties. Sample masses are provided in table S1.

The schematic figure of the experimental setup is provided in figure S1 (https://stacks.iop.org/ERL/15/104097/mmedia). The experiments were performed in a climate-controlled chamber with a stainless steel interior (Aralab FitoClima D1200; internal dimensions (H × W × D) (mm): 1330 × 600 × 680), providing constant air temperature and the ability to regulate humidity and light level. Lights were off during the whole experiment to mimic natural soil conditions. Inside the chamber, soil samples were placed in 500 ml glass jars sealed with PTFE-coated screw caps. The glass jars and caps were carefully cleaned with Milli-Q water before the experiments. To assess the response of VOC release to temperature increase, the incubation temperature cycle for each sample was set as follows: 0 °C for 18 h, raised to 5 °C for 48 h, and then further increased to 15 °C for another 48 h.

VOC fluxes from soil samples were measured through a dynamic flow-through system. Ambient air was pressurized and passed through a zero-air generator (Parker Balzon HPZA-3500-220). A flow of 0.8 L min⁻¹ set and monitored by a mass flow controller was then fed into the glass jar through a PTFE tubing (6 mm o.d., 4 mm i.d.). Finally, the sample flow from the glass jar was passed to the instrument using the same size of PTFE tubing and an overflow exhaust through T-connectors. To reduce evapotranspiratory losses from the soils, the inflow air was humidified through a bubbler before feeding to the glass jar. The experimental setup included an empty jar that contained no soil sample. Empty jar measurements were conducted before and after the experiments as blank samples to account for potential VOC contamination from the materials used or zero air.

Exactly the same procedures were applied to all experiments.

A Vocus PTR-TOF (Tofwerk AG/Aerodyne Research Inc.) was deployed to quantify the VOC fluxes from soil samples. The PTR instruments have been widely used to measure most VOCs in air. Compared with traditional PTR instruments, the Vocus PTR-TOF implements an enhanced inlet and source design and minimizes contact between analyte molecules and inlet/source walls, enabling detection of semi- and low-volatility compounds (Riva et al 2019, Li et al 2020). With the improved detection efficiency, the Vocus PTR-TOF provides a much lower detection limit for various VOCs (<1 pptv for 1 min integration time). The instrument mass resolving power can reach up to 15 000 m dm⁻¹ for m/Q > 100 Th. In this study, a high total sample flow of 5 L min⁻¹ was used to draw the air to the entrance of the Vocus, which helped to reduce inlet wall losses. The instrument ionization source was operated at a low pressure of 1.4 mbar. At this pressure, around 0.1 L min⁻¹ of the total sample flow went into the Vocus PTR-TOF, while the remainder was directed to other instruments and the exhaust. Data were recorded with a time resolution of 5 s. More details about the Vocus PTR-TOF technique are available in previous studies (Krechmer et al 2018).

Data were processed using the software package 'Tofware', which runs in the Igor Pro environment (WaveMetrics Inc. Lake Oswego, OR). Tofware enables the time-dependent mass calibration, baseline subtraction, and assignment of a molecular formula to the identified ions by high-resolution analysis. Signals were averaged over 15 min before data processing.

The instrument calibrations were conducted before and after the experiments using a calibration gas tank (Apel Riemer Environmental Inc.) containing 16 different compounds. It has been shown that the sensitivity of VOCs measured by PTR instruments is closely related to their elemental composition and functionality (Sekimoto et al 2017). The calibration gas tank included standards with the molecular formulas of C_xH_y, C_xH_yO_z, and C_xH_yN_z. Compounds with the formula of C_xH_y but not included in the standard mixture were quantified with the average sensitivity of the standards C_xH_y. For compounds with the formula of C_xH_yO_z and C_xH_yN_z but not included in the calibration tank, the average sensitivities of the standards C_xH_yO_z and C_xH_yN_z were used for them, respectively. There was no standard available for the compounds C_xH_yO_zN_p. In this study, the average sensitivity of the C_xH_yO_z and C_xH_yN_z standards was used for compounds with the formula of C_xH_yO_zN_p.

The emission rates (E, ng g⁻¹ dw soil h⁻¹) of the investigated VOCs were calculated using the following equation:

$$\begin{equation}E = Q*60*\frac{{\left( {{C_{{\text{out}}}} - {C_{{\text{in}}}}} \right)}}{{dw}}\end{equation} \tag{ 1 }$$

where $Q$ (in L min⁻¹) is the gas flow rate through the glass jar, ${C_{{\text{out}}}}$ and ${C_{{\text{in}}}}$ (in ng L⁻¹) are the VOC concentrations in the glass jar with thawing soils (sample measurements) and empty glass jar without soils (background measurements), and $dw$ (in g) is the dry weight of soil samples. The number of 60 is used to convert minute to hour. ${C_{{\text{out}}}} - {C_{{\text{in}}}}$ is calculated by subtracting the background measurements with empty glass jar from the measured VOC concentrations with soil samples.

While the focus of this work is on non-methane VOC emissions, CH₄ was measured in this study to provide a favorable benchmark regarding the behavior of soil processes during incubations. CH₄ concentrations were measured by cavity ring-down spectroscopy (Picarro G2301). The analyzer was alternatingly connected to measure the air entering and leaving the sample container for 5 min each. The emission rate of CH₄ was also calculated using equation 1.

The bulk soil samples were ¹⁴C dated at the Laboratory of Chronology in Helsinki. The permafrost sample age 5665 BP (before present, when present is 1950 AD) was calibrated using the Calib 704 program (Stuiver and Reimer 1993) and the active layer age −228, pMC (% modern carbon) = 102.88 ± 0.25, indicating a modern age, was calibrated using the CALIBomb program (Reimer et al 2004).

After VOC flux measurements, samples were put into the oven and dried at 70 °C for 24 h. The gravimetric soil water content was determined based on the water loss after drying. Before SOM, total carbon, and total nitrogen content analysis, the dry materials were further put into the oven and dried at 105 °C for 12 h. Approximately 1.0 g of dry soils was used for SOM estimation based on weight loss on ignition at 550 °C for 4 h. Approximately 0.3 g of dry materials was used to determine total carbon and nitrogen content with an automated LECO 828 series elemental analyzer (LECO Corporation, St. Joseph, MI).

It is generally acknowledged that increasing carbon and oxygen numbers lower the volatility of organic molecules. To estimate the volatilities of individual organic compounds, we used a volatility parameterization based on the number of carbon ($n_C^{\,i},$) oxygen ($n_O^{\,i},$) and nitrogen atoms ($n_N^{\,i}$) of the specific compound i (Donahue et al 2011, Mohr et al 2019). The volatilities are expressed as the logarithm of the saturation mass concentration $lo{g_{10}}C_{{\text{sat}}}^{\,i}$ for compound i:

$$\begin{align*}lo{g_{10}}C_{{\text{sat}}}^{\,i} & = \left( {n_C^0 - n_C^{\,i}{\text{ }}} \right){b_C} - \left( {n_O^{\,i} - 3n_N^{\,i}} \right){b_O}\\ & \quad - 2\frac{{(n_O^{\,i} - 3n_N^{\,i})n_C^{\,i}}}{{(n_C^{\,i} + n_O^{\,i} - 3n_{N)}^{\,i}}}{b_{CO}} - n_N^{\,i}{b_N}\end{align*}$$

where $n_C^0$=25, ${b_C}$=0.475, ${b_O}$=0.2, ${b_{CO}}$=0.9, and ${b_N}$ = 2.5. To investigate how the emissions of different compounds respond to temperature variations, the detected molecules in this study were further divided into three groups based on their volatility: VOCs, intermediate volatility organic compounds (IVOCs), and semi-volatile organic compounds (SVOCs).

During the warming of the active layer and permafrost soils (soil characteristics of each sample in figure 1 and table S2), even at a low temperature of 0 °C, moderate amounts of VOCs were released (figure S2(a)). When the temperature was increased to 5 °C, the emission rate of VOCs peaked within one hour and then subsequently declined. The peak emission rate of total VOCs reached >2 µg g⁻¹ dw soil h⁻¹. To see the full potential of VOC emissions from thawing soils, we further raised the incubation temperature to 15 °C, when another rapid although relatively low VOC pulse was observed. The high emission pulse occurring immediately after thaw and its rapid increase and decline within several hours suggest that the direct release of old, trapped gases from buried and subsequently frozen peat layers is the main source of VOCs. This was further confirmed by the timing and shape of CH₄ emissions upon permafrost thaw (figure S2(b)). We observed no detectable CH₄ emissions from active layer soils, whereas the permafrost soils emitted a strong CH₄ pulse immediately after thawing. According to the chronology analysis, the age of the permanently frozen soil sample was dated to 6450 cal. BP (calibrated years before present) and the active layer soil sample to 1955 AD.

Zoom In Zoom Out Reset image size

As shown in figure 1, the average cumulative flux of total VOCs from the thawing permafrost during the whole observations is 21.4 ± 3.4 µg g⁻¹ dw soil (mean ± SD), around four times as high as that from the active layer (5.1 ± 2.5 µg g⁻¹ dw soil). Compared to total VOCs, the cumulative flux of CH₄ emissions from the thawing permafrost is 7.3 ± 0.8 mg g⁻¹ dw soil.

After subtracting the background signals measured with blank samples, 165 organic ions detected by Vocus PTR-TOF were identified as VOCs released from soil samples. Generally, the most abundant compounds showed both overlapping and differences between permafrost and active layer (figure 2). The cumulative fluxes of acetic acid (C₂H₄O₂) and acetone (C₃H₆O) were the highest for both permafrost and active layer, on average accounting for 41.2% (32.9% ∼ 45.9%) and 18.5% (13.1% ∼ 23.2%) of total VOC release for the thawing permafrost and 19.9% (12.0% ∼ 25.3%) and 15.6% (8.9% ∼ 18.2%) for the active layer. In addition, acetaldehyde (C₂H₄O) and ethanol (C₂H₆O) also showed high abundances in total VOC emissions from the thawing permafrost and active layer soils.

Zoom In Zoom Out Reset image size

We furthermore observed the release of large quantities of aromatic compounds, such as benzene (C₆H₆), toluene (C₇H₈), xylene (C₈H₁₀), and phenol (C₆H₆O), which are listed in figure 2 as the highly emitted compounds from thawing soils. Among the emitted aromatic compounds, phenol and toluene were the most abundant species. For thawing permafrost, phenol on average contributed 6.9% (4.0% ∼ 8.5%) to total VOC emissions and toluene contributed 2.3% (1.5% ∼ 2.8%).

After thawing, emissions of isoprene, p-cymene, monoterpenes, sesquiterpenes, and diterpenes were measured from permafrost and active layer soils. For the thawing permafrost, the average emission rates of isoprene and monoterpenes during the whole measurements were 1.7 ± 0.6 ng g⁻¹ dw soil h⁻¹ and 0.4 ± 0.2 ng g⁻¹ dw soil h⁻¹, respectively (figure S3). While the emission rate of monoterpenes at 15 °C was lower than that at 5 °C, isoprene emissions were still high at 15 °C and comparable to its emissions at 5 °C. In particular, the less volatile sesquiterpenes and diterpenes were emitted with high fluxes. The average emission rate of sesquiterpenes from the thawing permafrost was 0.7 ± 0.3 ng g⁻¹ dw soil h⁻¹, higher than that of monoterpenes. Diterpenes were released with the average emission rate of 0.11 ± 0.03 ng g⁻¹ dw soil h⁻¹ for permafrost soils. As shown in figure S3, when the temperature was increased to 15 °C, sesquiterpenes and diterpenes were observed with higher amounts than at lower temperatures.

Figure 3 displays the overview of the speciated VOCs emitted from the thawing permafrost at different temperatures, with the mass defect (the difference between the exact mass and the nearest integer mass) of the measured compound plotted against the exact molecular mass of the compound. Based on the elemental composition, compounds of four categories were observed including CH, CHO, CHN, and CHON. The only compound in CHN group was acetonitrile (C₂H₃N), with the average emission rate of 0.1 ± 0.1 ng g⁻¹ dw soil h⁻¹ from the thawing permafrost soils. Due to the improved capacity of the Vocus PTR-TOF, larger molecules of the CH, CHO and CHON groups were measured from the thawing soils and were characterized with relatively lower volatility. More details can be found below.

Zoom In Zoom Out Reset image size

At higher temperatures, more compounds with larger molecular weight and lower volatility were released from the thawing permafrost. As shown in figure 3, when the temperature increased from 0 °C to 15 °C, much higher signals were observed for larger hydrocarbons (CH) and larger oxygenated molecules (CHO and CHON). Different VOCs showed their preferential responses to the varying temperature. For instance, aromatic compounds like toluene (C₇H₈), xylene (C₈H₁₀), and trimethylbenzenes (C₉H₁₂) were mostly emitted at 5 °C, and only a much smaller pulse was observed when the temperature rose to 15 °C (figure S4). However, for terpenes, particularly sesquiterpenes and diterpenes, much higher amounts were released with temperature increasing from 5 °C to 15 °C (figure S3). Figure S5 displays the correlation matrix among diverse VOCs for the variations in their emission rates during the whole incubation experiments with temperature varying from 0 °C to 5 °C and further to 15 °C, which illustrates whether the temperature-dependences of different volatile species are similar or not. The time series of the emission rates of all aromatic compounds correlated well with each other, consistent with the results in figure S4. For sesquiterpenes, diterpenes, and some larger oxygenated molecules like C₉H₁₄O, C₈H₁₄O₃, C₁₀H₁₄O₂, and C₁₀H₁₆O₃, the variations in their emission rates during the whole temperature cycle correlated tightly with each other but showed no correlations with those of aromatic compounds.

Based on the volatility parameterization, the compounds in this study were categorized into VOC, IVOC, and SVOC, with IVOC and SVOC being less volatile than VOC. Figure 4 illustrates the average cumulative fluxes of VOC, IVOC, and SVOC from the thawing permafrost samples at 0 °C, 5 °C, and 15 °C. The total volatile emissions at different temperatures were largely dominated by VOCs, especially at 0 °C. The relative contributions of different species in the VOC group can be seen in figure 4, such as acetic acid, acetone, aromatic hydrocarbons, isoprene, and monoterpenes. When the temperature increased, the mass contributions of IVOCs and SVOCs were substantially enhanced (figures 4 and S6). The mass fraction of IVOCs in total volatile emissions rose from 0.01 to 0.18 with the temperature varying from 0 °C to 15 °C. In addition to sesquiterpenes, the IVOC group in this study included the compounds of C₆₋₈H_yO₁₋₂, C₉₋₁₂H_yO₁₋₂, and C₁₂₋₁₅H_y. The SVOC group contributed a small fraction of <0.02 in total volatile emissions, which comprised the compounds of C₁₃₋₁₆H_yO₁₋₂ and C₂₀H_y.

Zoom In Zoom Out Reset image size