Soil-gas surveys: hardware, protocols & survey design

If you want to know whether hydrogen gas is leaking up from underground rock, the cheapest and fastest way to check is to poke a shallow hole in the ground, stick a thin tube into it, and pump some of that soil air through a handheld gas detector. That is, in essence, everything a “soil-gas survey” is: a hole, a tube, a pump, and a number.

The trouble is that the number you get depends enormously on things that have nothing to do with the rock a kilometer below — how long you pumped before reading, how recently it rained, what the barometric pressure did overnight, and how hard you had to drill to make the hole in the first place (drilling itself can chemically manufacture hydrogen at the drill bit, in amounts that can dwarf a real signal). Most of the craft of soil-gas surveying is therefore about controlling all of that noise well enough that a reading means something, rather than about the gas detector itself.

This page is Hestia’s field-practice reference for that work: what hardware to drive into the ground, how long to purge it, how deep to go, how to lay out a survey grid, which published instruments have actually been used for this in the geologic-hydrogen literature, and where the state of the art runs out (there is, for instance, no off-the-shelf continuous H₂ monitoring station or H₂ flux chamber you can simply buy today). It is the operational companion to Geologic context, which derives the underlying transport physics and the 2023–2025 drilling-artifact controversy in full; this page assumes that background and focuses on what to actually do in the field.

Context numbers used throughout

Atmospheric H₂ background ≈ 0.53 ppm (552.8 ppb marine boundary layer mean, 2021) [30]. Published soil-gas H₂ surveys report ranges spanning zero to >1,000 ppm within tens of meters at a single site [2]. Drilling-induced hydrogen artifacts have been measured up to 464 ppm and accounted for ~85% of “anomalies” at a systematically studied set of sites [27]. Standard probe depth across nearly every source reviewed here is ~0.75–1.0 m.

Mechanism & physics

What a soil-gas survey is actually measuring

A soil-gas H₂ survey samples a single, noisy, spatiotemporally autocorrelated realization of a concentration field $C(x,y,z,t)$ that is simultaneously controlled by a deep source term, near-surface transport, and a biological sink — and then asks an analytical instrument with its own noise floor and cross-sensitivities to report a number. Every methodological choice in this dossier — probe depth, purge volume, grid spacing, replicate policy — is an attempt to make that one noisy draw representative of the underlying field rather than of the measurement act itself.

The governing 1-D transport equation, derived in full in Geologic context, combines Fick’s first law of diffusion with an advective (pressure-driven, Darcy) term and a first-order microbial sink:

$\frac{\partial (\Theta_a C)}{\partial t} = \frac{\partial}{\partial z}\!\left(D_s \frac{\partial C}{\partial z}\right) - \frac{\partial (vC)}{\partial z} - k_s \Theta_a C + Q(z,t)$

where $D_s$ is the effective soil-gas diffusivity (a function of air-filled porosity $\Theta_a$ and tortuosity), $v = -(k/\mu)\nabla P$ is the Darcy advective velocity driven by a barometric pressure gradient, $k_s$ is a microbial first-order removal-rate constant, and $Q(z,t)$ is a deep source flux. Every term is weather-modulated: $D_s$ and $\Theta_a$ track soil moisture, $k_s$ tracks soil temperature and moisture, and $v$ tracks transient barometric pressure — which is exactly why survey design has to treat weather as a covariate to be logged and regressed, not noise to be averaged away.

The purge-volume mass balance behind the “3–5 volumes” rule of thumb

The near-universal soil-gas field rule of thumb — purge 3 to 5 internal (dead) volumes of the probe/tubing system before drawing a sample — falls directly out of a first-order, well-mixed-tank mass balance. If a probe and its connecting tubing have combined dead volume $V_d$, are purged at flow rate $Q$, start filled with a gas of concentration $C_0$ (commonly atmosphere, if the tubing was open to air), and are fed by the true soil-gas concentration $C_\infty$ at the probe tip, then for a well-mixed volume:

$\frac{C(t)-C_\infty}{C_0-C_\infty} = \exp\!\left(-\frac{Qt}{V_d}\right) = \exp(-N)$

where $N = Qt/V_d$ is the number of dead-volume exchanges. At $N=3$, the sample is 95% of the way from $C_0$ to $C_\infty$; at $N=5$, it is >99% of the way there. This is precisely the number reported by EPA method-development work: “removal of three to five internal volumes of a sample system should ensure that vapor concentration entering a sampling vessel is 95% to greater than 99%… of vapor concentration in surrounding soil” [1].

The same EPA study found that purging up to 12 L from a sub-slab probe or 103 L from a dedicated vapor probe produced “little impact” on the reported concentration once true equilibrium had been reached [1] — i.e., past $N\approx5$, further purging on an undisturbed probe is wasted pump time, not a source of additional artifact.

The second-order effect, and where the “over-pumping creates artifacts” controversy sits, is that a freshly drilled hole behaves not as a well-mixed tank with a fixed $C_\infty$ but as a second, disturbed reservoir with its own equilibration time. The same EPA study found that “stabilization of vapor concentration at a direct-push location sampled immediately after probing required 8 to 12 internal volume exchanges,” indicating non-attainment of equilibrium and the need for materially more purging soon after probe installation than the simple dead-volume model alone predicts [1].

Physically, this is a diffusive-relaxation problem layered on top of the pumping-washout problem: the disturbed zone around a fresh probe tip has a characteristic re-equilibration time $\tau \sim L^2/D_s$, where $L$ is the radius of soil mechanically or chemically perturbed by drilling. Because H₂’s diffusivity in air-filled soil pore space is roughly 3–4× that of CH₄ (full treatment in Why H2 is hard), this relaxation is faster for H₂ than for heavier co-evolved gases — but it is not instantaneous, and pumping through it too soon draws down whatever transient, drilling-generated gas is sitting in that disturbed shell before it has re-equilibrated with the undisturbed formation.

Statistical framing: a spot reading is one draw from a skewed, autocorrelated field

Soil-gas H₂ values are not normally distributed in space: published surveys report ranges spanning zero to >1,000 ppm within tens of meters at a single site [2], and background-versus-anomaly discrimination in geochemical exploration generally is done in log space because concentration data of this kind is closer to log-normal than normal.

This has two direct design consequences: (i) a single spot reading cannot distinguish a real but transient pulse from a stable seep from a drilling artifact — separating the population’s shape from a single sample’s noise requires either spatial replication (a grid or transect) or temporal replication (repeat visits or continuous monitoring); and (ii) survey statistics should be computed and reported in a way that is robust to the small number of very large outlier readings that dominate raw arithmetic means — the Harvey, Perth Basin study explicitly used interquartile-range and outlier-flagged box plots rather than simple means to characterize its point-survey population for exactly this reason [2].

Diurnal and continuous covariate-regression framing

Two distinct survey designs follow from the transport equation above. Fixed-time spot sampling assumes the covariates ($T$, $P$, RH) are approximately constant across the (short) duration of a survey day, and tries to hold sampling time-of-day constant across a campaign to cancel the diurnal term.

Continuous covariate regression instead measures $C(t)$, $T(t)$, $P(t)$, RH$(t)$ simultaneously and fits the relationship explicitly — the Harvey Perth Basin study computed a full Pearson correlation matrix between cH₂, cCO₂, atmospheric temperature, atmospheric pressure, and relative humidity from its autonomous monitoring record, finding a weak temperature correlation ($r^2=0.13$) but strong, sign-opposite correlations of cH₂ and cCO₂ with relative humidity (−0.57 and +0.54 respectively) [2].

The Brazilian cratonic-environment monitoring network similarly resolved two superimposed signal classes only by continuous, multi-week logging: sporadic large pulses lasting 1–2 days, and a smaller daily periodic signal lasting about 6 hours and peaking near midday, “due to pressure-driven gas release” [3][4]. Neither signal class is resolvable from a fixed-time spot-sampling design; this is the theoretical argument for the continuous-monitoring station hardware described below.

Variants & subtypes

Probe hardware taxonomy

• Driven, temporary steel probes (direct-push style). A hollow steel rod is driven into the soil with a slide hammer or rotary/percussion hammer drill to a target depth, then sampled directly through the rod or via a separate tube inserted through a pilot hole. The AMS Gas Vapor Probe (GVP) product line is the vendor-named example: a stainless-steel probe driven with a slide hammer or electric rotary drive hammer and GVP drive extensions, with fluoropolymer, nylon, or polyethylene tubing run to the surface for remote sampling [5].
• Retract-a-tip design. A refinement in which the sampling tip is driven closed, then the outer drive-rod string is retracted a short distance to expose sampling ports (protected by a stainless mesh screen) without pulling the point out of the ground; the point can then be driven deeper and reopened for a second reading from the same hole [5][6]. The stated advantage is reduced equilibration time (the tip is only opened to the formation at the moment of sampling) and easy decontamination/reuse between survey points [6]. AMS’s commercial Retract-A-Tip is built from 300-series stainless steel (body) and 400-series stainless steel (tip), SKU 210.01, 3/16″ ID × 1/4″ OD tubing in 50–500 ft lengths, list price roughly $303/unit [5].
• Percussion hammer-drill-and-tube (the method actually used in most published H₂ soil-gas campaigns). Rather than a dedicated retractable-tip sampler, most published geologic-H₂ surveys use a simple handheld or vehicle-mounted hammer drill to bore a temporary hole to ~0.75–1 m, then insert a thin perforated steel or PVC tube and connect it to a handheld analyzer’s pump inlet. This is explicitly the method described by Gold Hydrogen/CSIRO for the 2023 Yorke Peninsula survey: “a small handheld hammer drill… will drill a hole up to a metre in depth into the soil… the drill bit will then be retracted, and an 80cm stainless steel tube will be inserted into the hole and connected to a handheld multi-gas analyser device” [7]. The historical USGS Kansas survey used the same logic at industrial scale in 1984: “soil gas was sampled by driving a hollow steel probe into the ground to a depth of .75 m… five cm³ of soil gas were withdrawn from that depth with a hypodermic syringe and needle and immediately analyzed” in a truck-mounted mass spectrometer [8].
• Permanent buried probes. A probe is installed once and left in place for weeks to months of repeat or continuous measurement, trading spatial coverage for temporal resolution. The MONHyTOR probe (Université de Pau et des Pays de l’Adour/CNRS, France) is a purpose-built example: 1028.8 mm long, 30 mm diameter, with its electrochemical sensor positioned 950 mm down the shaft, gas entering only through a hydrophobic membrane at the lower tip, installed via a rotary hammer with 36–40 mm stainless drill bits [9]. The 950 mm depth is explicitly justified on the same thermal-stability argument used throughout this literature (see Depth selection, below): at 1 m, “soil temperature is relatively stable, i.e., not or minimally affected by daily air temperature variation” [9].

Sampling-mode taxonomy

• Spot/point pumped grab sampling. A handheld multigas analyzer’s internal pump draws gas through the probe for a purge interval, then a reading is logged — the dominant published method (GA5000, iBrid MX6) [2][10]. It is fast, requires an operator on-site, and is the mode most exposed to the purge-volume and drilling-artifact issues above.
• Continuous pumped autonomous monitoring. An unattended, battery/solar-powered unit pumps on a duty cycle and logs continuously. The WHALI system (Axiom Sensing, Australia) used in the Harvey Perth Basin study “was programmed to pump for 3.5 min every 30 min… which averaged 300 mL/min pump volume,” evacuating roughly 2 L of soil gas per hour of monitoring [2].
• Non-pumped (diffusive) continuous monitoring. A sensor is buried with only a passive membrane or diffusion path connecting it to the formation, with no active pump — MONHyTOR is the clearest published example, relying purely on diffusion through its hydrophobic membrane [9].
• Accumulation/flux chambers (static and dynamic). A chamber is placed on the soil surface to trap the gas emerging by diffusion/advection, either allowed to accumulate statically and be sampled once, or continuously circulated through an on-board analyzer that computes flux from the rate of concentration rise (dynamic/“non-stationary” accumulation). This is a mature, commercialized method for CO₂, CH₄, and H₂S soil flux — the West Systems Portable Fluxmeter explicitly measures “CO2, CH4, H2S, and VOC” soil diffuse flux via the accumulation-chamber method [11] — but as of the sources reviewed here, no commercial off-the-shelf flux-chamber product is marketed with an H₂ channel. Researchers investigating candidate H₂ seeps (e.g., the South Gironde circular depressions) have instead built ad hoc permanent-pumping flux setups paired with electrochemical H₂ sensors rather than adapting a certified commercial fluxmeter [12]; this is an instrumentation gap (see Open questions).
• Headspace equilibration (water-logged sites). Where a dry soil probe cannot be driven (flooded terrain, standing water, drilling mud), a sealed headspace is equilibrated against a water or drilling-fluid sample and the gas phase analyzed — exploiting H₂’s low aqueous solubility to concentrate it preferentially into the gas phase. This method is developed specifically for natural-hydrogen exploration in water-logged environments in the peer-reviewed literature [13].
• Borehole/mud-gas logging during drilling. Distinct from soil-gas surveying proper, but relevant context for campaigns that graduate from soil gas to drilling: gas-chromatograph mud-logging probes measure the composition of gas liberated from drilling fluid in real time as a well is advanced — this is how the Bourakebougou, Mali wells were characterized down-hole (up to 98% H₂ at the Bougou-1 wellhead) [14][15]. This is a fundamentally different measurement regime (percent-level, continuous downhole gas rather than ppm-level near-surface diffusive gas), and instrument selection must not conflate the two — see Mud-gas logging and Downhole deployment.

Depth selection — why ~1 m is the field convention

The ~1 m probe-depth convention recurring across nearly every source reviewed here — USGS Kansas (0.75 m) [8], MONHyTOR (0.95–1.0 m) [9], South Gironde (1 m) [12], Perth Basin point surveys (0.80–1.0 m) [2], and the Yorke Peninsula protocol (“a hole up to a metre in depth”) [7] — is set by the depth at which diurnal surface-temperature swings are damped to a level that no longer dominates the measured signal, so that a spot reading reflects gas transport rather than the time of day it was taken.

MONHyTOR’s own siting rationale states this explicitly: 950 mm was chosen because soil temperature at that depth is “not or minimally affected by daily air temperature variation” [9]. The Harvey Perth Basin study makes the complementary point from the other direction: it found only a weak correlation between atmospheric temperature and cH₂ ($r^2=0.13$) and explicitly notes this is “consistent with previous studies that demonstrated that temperature fluctuations at depths of 80–100 cm are largely seasonal rather than diurnal” [2] — i.e., the ~1 m convention is empirically validated by the fact that diurnal thermal noise is largely gone by that depth, even though barometric-pressure-driven advective signal is not (see Environmental covariance, below).

A secondary practical constraint is also visible across the sources: 1 m is roughly the maximum depth conveniently reached by a handheld or vehicle-mounted percussion/rotary hammer drill without escalating to a drill rig, which is also why it is the depth ceiling used in resource-constrained reconnaissance surveys such as Yorke Peninsula [7].

Permanent vs. roving probes

Roving (point-survey) probes maximize spatial coverage per unit field-day at the cost of temporal resolution; permanent probes invert that trade. The Harvey Perth Basin study is the most direct head-to-head comparison published to date: it ran both approaches at the same site and found that “repeated point surveys successfully captured the overall range of emissions but failed to resolve dynamic processes, such as abrupt flux changes,” while continuous (permanent-probe) monitoring “provided additional insights into these transient phenomena” but obviously could not, in the study’s timeframe and budget, cover more than two fixed locations 400 m apart [2].

The same study quantifies how many point surveys are actually needed: of five point-survey days conducted, the authors found “the full range of hydrogen concentrations observed during the 31st March 2022 survey was already encompassed within the interquartile range… of the data collected between… 17th May 2022,” concluding that “four well-timed point surveys were sufficient to characterize soil gas hydrogen variability… during the autumn 2022 monitoring period” for that specific site and season [2] — with the explicit caveat that this number is season-specific, the authors recommending “either four discrete point measurements or at least four weeks of autonomous monitoring, per season” for future work [2].

Survey design taxonomy — grid, transect, and ring layouts

• Linear reconnaissance traverses. The historical USGS Kansas survey ran two ~40–60-mile east–west traverses at 1/2-mile (≈800 m) sample spacing, explicitly designed to test whether a known deep hydrogen-rich gas field produced a detectable near-surface signature over a wide area, rather than to delineate a specific target precisely [8].
• Dense point grids over a candidate structure. The Perth Basin pilot study drilled 91 holes over five separate days within a 0.3 km² area around a single well site and known fault trace [2] — roughly two orders of magnitude denser than the Kansas reconnaissance spacing, appropriate for characterizing a specific already-identified feature rather than searching a wide area.
• Opportunistic/access-constrained road-easement sampling. The Gold Hydrogen/CSIRO Yorke Peninsula survey identified “approximately 80 survey points… in a wide locality,” sited along public road verges/easements for two weeks, explicitly trading a formal regular grid for land-access practicality [7].
• Nested “ring” designs around a candidate seep/fairy circle. A dedicated review of natural-hydrogen soil-gas survey design concludes that, because “there is currently a lack of data regarding the optimal sampling step or sample density for hydrogen soil-gas surveys” and lateral variability patterns reported in the literature vary widely, “a regular sampling step, ranging from tens to hundreds of meters, in each of [several concentric] rings” around a candidate feature is the most defensible design given present knowledge [16]. This is, to this page’s knowledge, the only published source that explicitly frames H₂ soil-gas grid-spacing as an open methodological question rather than a solved one, which is a reason not to over-specify Hestia’s own survey grids on borrowed non-H₂ convention.
• Conventional (non-H₂) environmental soil-gas practice, for contrast. Standard VOC/contaminant soil-gas delineation surveys typically run far denser grids than any H₂ campaign reviewed here — one published U.S. Army site characterization used grids “established at intervals of 50 to 100 feet” (≈15–30 m), tightening to 25 ft (≈8 m) immediately around a suspected source [17]. The order-of-magnitude gap between this and the “tens to hundreds of meters” H₂ guidance above reflects both cost (H₂ campaigns are reconnaissance-stage; contaminant delineation is typically post-detection) and the fact that H₂ seepage features (fairy circles, sub-circular depressions) are reported at tens-to-hundreds-of-meters scale to begin with, so sub-10 m spacing would oversample within a single feature rather than resolve it.

Method comparison & quantitative performance envelope

Field instruments used in published H₂ soil-gas campaigns

Instrument	Class	H₂ range	H₂ precision/accuracy	Response (T90)	Pump flow	Notable use	Ref
Geotech/QED GA5000	Handheld multigas, electrochemical H₂ cell (optional)	0–1,000 ppm	±2.5% FS (manufacturer, after calibration)	≤90 s	550 mL/min typical (5–5,000 mL/min settable)	Most widely used single instrument in this literature; Perth Basin, ISTerre chamber-tested	[18][2][19]
iBrid MX6 (Industrial Scientific)	Handheld multigas, electrochemical H₂ cell	0–2,000 ppm	1 ppm precision (per comparison table)	not stated in sources reviewed	not stated in sources reviewed	Used alongside GA5000 at Harvey, Perth Basin	[2]
Axiom Sensing WHALI	Autonomous pumped multigas logger, electrochemical H₂/H₂S	0–2,000 ppm	10 ppm precision	not stated	~300 mL/min average (3.5 min pumped every 30 min)	First autonomous H₂ microseepage monitoring in Perth Basin; solar-capable	[2]
Univ. Pau et Pays de l’Adour/CNRS MONHyTOR	Buried, non-pumped diffusive probe, electrochemical	0–1,000 ppm	0.1 ppm display resolution, ±5% FS error	adjustable to 1 s sampling	none (passive diffusion through membrane)	Deployed as a research “fleet” at 3 French sites since 2024	[9]
Dräger X-am 8000 (XXS H2 HC sensor)	Handheld multigas, electrochemical/catalytic combustible-gas sensor	0–4 vol% (0–100% LEL)	T90 ≈ 20 s	20 s	n/a (diffusion or pump-accessory)	Safety-class LEL monitoring; not seen used in a published H₂ exploration soil-gas paper in sources reviewed	[20]
Geotech BIOGAS 5000	Handheld multigas, optional electrochemical H₂ cell	0–1,000 ppm	not independently verified in sources reviewed	not stated	not stated	Biogas/AD-process instrument marketed with optional H₂ channel	[21]
SRI Instruments portable GC (TCD)	Field GC, thermal-conductivity detection	percent-level down to TCD floor	~200–300 ppm LOD (manufacturer)	minutes per run (batch)	n/a	Compositional confirmation, not spot-survey speed	[22]
INFICON Micro GC Fusion	Portable micro-GC, MEMS micro-TCD	ppm–percent	down to ~1 ppm LOD (manufacturer, optimized method)	minutes per run	n/a	Marketed for pipeline-quality natural-gas H₂ monitoring; not seen deployed in a published soil-gas paper in sources reviewed	[23]
USGS mobile mass spectrometer (historical)	Truck-mounted quadrupole MS	relative peak height (not absolute ppm calibrated for H₂ in this study)	n/a (relative units)	~40 s per AMU scan; ~10 min/site; 40–50 samples/day	n/a (syringe grab, not continuous pump)	Landmark 1984 Kansas reconnaissance survey	[8]

Every H₂-specific accuracy figure in this table is the manufacturer’s or study’s own stated number; none should be read as an independently validated field detection limit relative to the 0.53 ppm atmospheric background — see below and Calibration, drift & QA/QC for why realistic field LOD is materially worse than headline sensor precision.

The gap between sensor precision and usable field LOD

The GA5000’s own datasheet states ±2.5% of full-scale (i.e., ±25 ppm on a 0–1,000 ppm cell) as “typical accuracy… after calibration plus accuracy of calibration gas used” [18] — a number that already sits nearly 50× above the 0.53 ppm atmospheric background this page uses as the exploration zero-reference.

Independent controlled-chamber testing of the GA5000 sharpens this further: sweeping ambient pressure from 500–1400 mbar around a fixed 50 ppmv H₂ calibration standard, researchers found that “depressurization causes an overestimation of the H2 concentration… at 500 mbar, the H2 concentration is overestimated by 13% compared to ambient conditions: 62 ppmv versus 55 ppmv” [19].

That is an instrument artifact of several ppm riding on top of the manufacturer’s own stated accuracy band, confirming the general point made in Geologic context that realistic field LOD for the electrochemical-cell class sits in the 1–10 ppm range even for the most widely used instrument in this literature — not at the sub-ppm level implied by raw display resolution.

Purge protocol quantitative summary

Parameter	Value	Basis	Ref
Purge volumes for 95% equilibration (undisturbed probe)	3 dead-volume exchanges	Exponential washout mass balance, confirmed by EPA field trials	[1]
Purge volumes for >99% equilibration (undisturbed probe)	5 dead-volume exchanges	Same	[1]
Purge required for a freshly drilled/direct-push hole	8–12 dead-volume exchanges	EPA field trials found non-attainment of equilibrium below this	[1]
Extra purging beyond equilibrium, sub-slab probe	up to 12 L, “little potential for impact”	EPA air-flow modeling	[1]
Extra purging beyond equilibrium, dedicated vapor probe	up to 103 L, “little impact on vapor concentration”	EPA sequential sampling	[1]
WHALI autonomous duty cycle	pump 3.5 min every 30 min, ~300 mL/min	Field-deployed autonomous logger	[2]
Time-to-peak, drilling-disturbed-chamber extraction	~26 s of pumping to reach peak cH₂; ~120 cc extracted at peak	Direct instrumented observation at AXH2-13, Perth Basin	[2]
Volume-vs-assembly-volume at peak extraction	~284 cc drawn after 60 s ≈ 2× the downhole probe-assembly internal volume	Same study; interpreted as direct extraction from soil, not just clearing the probe	[2]

Sample-container stability for lab-chain sampling

Container	H₂ stability finding	Mechanism	Ref
Standard 12 mL Exetainer, butyl rubber septum, untreated	H₂ increased to ~16 ppm after 9 days from a reference gas (order-of-magnitude contamination); near-linear (zero-order) increase	Septum material degradation releasing H₂, not stored-gas leakage	[24]
Exetainer, NaOH-treated butyl septum	Contamination reduced ~4×, but H₂ still reached ~3 ppm after 9 days; extrapolated to “double approximately every 2 weeks”	Reduced but not eliminated degradation	[24]
Modified Exetainer (silicone plug + stainless bolt + Viton O-ring)	H₂ stable within <5% deviation for >60 days at both ambient and 10 ppm spiked levels	Silicone/steel sealing eliminates rubber-degradation source	[24]
Modified Exetainer, CO channel	Still increases linearly but ~9× slower than treated conventional Exetainers; correctable with reference standards	Residual small degradation source, well-characterized	[24]
Tedlar (PVF) gas sampling bags	Vendor literature describes good stability for low molecular weight compounds such as CO, CO2, methane, hydrogen and SF6, but this specific claim is not supported by the cited source (unverified)	Low permeation/adsorption for small nonpolar molecules	[25]
Summa (passivated stainless steel) canisters	Gold standard for GC-MS air analysis generally; carbonic-acid/metal reactions inside a steel canister have been reported elsewhere to generate H₂ in situ, confounding natural H₂ measurement, but this is not supported by the cited source (unverified)	Chemical artifact intrinsic to the container material, not a storage-stability issue per se	[26]

Sample-to-lab chain

The practical conclusion for a Hestia sample-to-lab chain: off-the-shelf butyl-septum Exetainers are not fit for purpose for trace H₂/CO work without the silicone-and-steel modification published by Nauer et al. [24][28]; Tedlar or equivalent fluoropolymer bags are a plausible default for short-hold-time transport (the specific H₂/SF₆ stability claim for Tedlar is unverified [25]); and steel canisters, despite being the GC-MS gold standard for most trace gases, are reported elsewhere to carry a chemically mechanistic H₂-generation risk that must be controlled for or avoided in an H₂-focused program, though this is unverified against the cited source [26].

Environmental covariance & cross-sensitivity

Barometric pressure — the confounded variable

Pressure is simultaneously a real geologic transport mechanism (barometric pumping) and a documented instrument artifact, and the two are not separable from a single spot measurement. The Harvey Perth Basin autonomous record found pressure only weakly linked to cH₂ directly ($r^2=0.11$), but noted that “the broad relationship cannot capture the daily scale pressure pumping” and that “falling atmospheric pressure during the day facilitated increased H2 gas flux, while rising nighttime pressure inhibited emissions” [2] — a genuine transport signal.

Simultaneously, the ISTerre/GA5000 chamber study demonstrates a real electrochemical-cell measurement artifact under depressurization [19]. Any Hestia field protocol should log barometric pressure at every reading specifically so this confound can be regressed out rather than mistaken for signal in either direction — see Environmental covariance for the regression approach.

Relative humidity and soil moisture

Relative humidity showed the strongest single-covariate correlation with cH₂ in the Harvey Perth Basin dataset: −0.57 (H₂ readings dropped as humidity/soil moisture rose) [2], consistent with the diffusion-limiting effect of water filling soil pore space and reducing $\Theta_a$ and $D_s$ in the governing transport equation above.

The South Gironde permanent-monitoring result is a useful complementary data point in the opposite direction: after a full year of continuous monitoring at 10-second intervals following an initial elevated spot reading (H₂ up to 3.5‰ = 3,500 ppm), “no H2 flux was detected, neither in the soil, nor in the surface aquifer” [12] — i.e., the site’s initial spot anomaly did not survive extended, moisture/pressure-covariate-aware monitoring, a cautionary data point for how easily a single high spot reading can be a transient (possibly drilling-related) artifact rather than a stable seep.

Drilling-induced hydrogen (“drill bit metamorphism”)

A severe and well-quantified artifact in this literature is mechanochemical hydrogen generation at the drill bit during hole preparation. A systematic study across 24 sub-circular depressions in the Yilgarn Craton and Perth Basin, drilling 83 holes with repeated (non-redrilled) measurements, found “drilling induced hydrogen artifacts… measured at up to 464 ppm in 85% of locations, primarily in high-strength, low-moisture soils, while only 15% represented naturally occurring hydrogen” [27].

The consequence for survey design is that a protocol that drills once and reads once, in high-strength, low-moisture ground, is measuring drill-bit chemistry rather than geology roughly six times out of seven at these sites.

The Harvey Perth Basin study’s own instrumented extraction-curve data is consistent with a slow-recharging, drilling-perturbed local gas reservoir rather than a stable formation concentration: peak cH₂ at the AXH2-13 site was reached only after 26 s of active pumping, with an exponential decline afterward that “implies a much slower hydrogen recharge rate into the soil compared to the pump rate” [2] — precisely the signature predicted by the disturbed-zone diffusive-relaxation argument above.

Cross-sensitivity of the dominant sensor class (electrochemical)

Because nearly every fielded soil-gas instrument reviewed here (GA5000, iBrid MX6, WHALI, MONHyTOR, BIOGAS5000) uses an electrochemical H₂ cell, the cross-sensitivities of that transduction class — detailed fully in Electrochemical cells — dominate the field error budget. The manufacturer-stated cross-sensitivity most relevant to a co-located CH₄/CO/H₂/H₂S survey is the reverse direction — hydrogen’s effect on the carbon monoxide channel — which the GA5000 datasheet quantifies at “approximately 1%… do not use where hydrogen is in excess of 10,000ppm” [18], and which MONHyTOR’s manufacturer states is the only gas its electrochemical sensor cross-responds to at all [9].

This is a comparatively well-behaved cross-sensitivity profile relative to MOX or catalytic-bead sensors, but it does mean a co-located CO survey (e.g., for shallow combustion or biomass-burning confounds) needs H₂-compensation, which several vendors (GA5000, Dräger) sell as an explicit optional cell configuration [18][20].

Container/handling artifacts downstream of the probe

The Exetainer degradation data above is itself a cross-sensitivity-like artifact worth restating in field-protocol terms: an untreated butyl-septum vial can inflate an H₂ reading by a factor of ~3 in as little as 30 days of storage purely from septum off-gassing, with no change in the true sample composition [24][28] — a lab-chain artifact that looks, on paper, exactly like a real anomaly if the QA chain doesn’t track container hold-time.

Use for geologic H2

Field summary

Soil-gas surveys are, and should remain, Hestia’s primary reconnaissance tool — they are the only method validated at ppm-level sensitivity to real seepage in multiple independent basins (Mali, Brazil, France, Western Australia) [8][2][12][3], deployable at tens of readings per field-day with off-the-shelf handheld hardware [2][7], and cheap enough to run reconnaissance-density campaigns before committing to drilling. But a single spot reading, taken once, immediately after drilling, is not a defensible unit of evidence on its own: at a systematically studied set of sites, genuine hydrogen was found in only 15% of drilling-induced-artifact-prone locations [27], and even the best published survey design needed four repeat visits across a season, cross-checked against multi-week continuous monitoring, before its authors were willing to characterize a site’s baseline [2]. Continuous monitoring stations and a hard repeat-measurement/settling-time protocol are the minimum discipline this literature supports for calling any reading “signal”; treat them as required, not as optional refinements.

Point surveys

Soil-gas surveys are Hestia’s workhorse reconnaissance method. They are simultaneously:

• validated at ppm-level sensitivity to real seepage in multiple independent basins [8][2][12][3]
• deployable at the pace of tens of readings per field-day with off-the-shelf handheld hardware [2][7]
• cheap enough (a percussion drill, a handheld multigas analyzer, and a technician) to run reconnaissance-density campaigns before committing to drilling

Continuous monitoring

Continuous monitoring stations complement point surveys rather than replacing them. Sporadic pulses (1–2 days, large amplitude) and daily periodic emissions (~6 hours, peaking near midday) documented in Brazil [3][4], and transient pulses up to 598 ppm documented in Western Australia [2], are by construction invisible to any survey design that visits a point once.

The state of the art here is research-instrument-grade rather than commercial-off-the-shelf: WHALI [2], EVA, and MONHyTOR [9] are all bespoke builds by small research groups or a single instrumentation company, not catalog products with a datasheet and a distributor network the way the GA5000 is. This is a procurement gap for Hestia: a field campaign wanting permanent, telemetered H₂ monitoring today is choosing between building an in-house electrochemical-probe station (leveraging the sensor-selection guidance in Electrochemical cells) or pursuing an academic collaboration/license with one of these groups, not buying a station off a shelf.

Where a station’s channel count grows beyond a single H₂ cell — co-tracer gases, T/P/RH covariates, or a deliberately redundant multi-sensor node — pattern-recognition approaches applied across that whole channel set are a promising way to separate genuine transient pulses from covariate-driven noise algorithmically rather than by manual regression alone; see Sensor arrays & machine learning.

Flux chambers

Flux chambers are the least-supported method in this dossier for H₂ specifically. The technique is mature and commercialized for CO₂/CH₄/H₂S [11], but no source reviewed here shows a certified, commercial H₂ flux-chamber product; H₂ flux work in the literature is done with ad hoc accumulation-chamber builds paired to an electrochemical sensor [12]. Hestia should not expect to buy a “West Systems for H₂”; building or commissioning one is itself a small instrumentation project.

Known limitations

1. Manufacturer-stated electrochemical H₂ accuracy (±2.5% FS, i.e. tens of ppm) is already coarser than the 0.53 ppm background this page treats as the exploration zero-reference [18], so “detect an anomaly against background” overstates what these instruments can do; “detect a high-tens-to-thousands-of-ppm anomaly against a noisy few-ppm floor” describes them more accurately.
2. Pressure is a confounded variable that is simultaneously real geology and real instrument artifact, and no source reviewed here has published a field-deployable correction that fully separates the two [19][2].
3. Drilling-induced artifacts dominate single-spot surveys in high-strength/low-moisture ground at a rate the WA literature quantifies as ~6-in-7 false positives at artifact-prone sites [27], meaning Hestia’s field SOP must build in repeat-measurement or settling-time discipline as a hard requirement rather than an optional refinement.
4. There is no off-the-shelf continuous-monitoring station or H₂ flux chamber to buy — both require either a research collaboration or an in-house build.

Instruments & vendors used in published campaigns

Vendor	Product / line	Category	Key specs (as sourced)	Price class	Ref
AMS, Inc. (American Falls, ID)	Gas Vapor Probe (GVP), Retract-A-Tip (SKU 210.01)	Driven steel soil-gas probe hardware	300/400-series SST; 3/16″ ID tubing; slide-hammer or rotary-hammer driven	~$300/unit (probe only; kits with drive rods extra)	[5][6]
QED Environmental Systems / Geotech	GA5000	Handheld multigas analyzer, electrochemical H₂ (optional cell)	H₂ 0–1,000 ppm ±2.5% FS; CH₄/CO₂ by dual-wavelength IR 0–100% ±0.5–1.5%; pump 550 mL/min; IP65; ATEX/IECEx/MCERTS/UKAS certified	Mid-range professional handheld (list price not disclosed in sources reviewed)	[18]
QED Environmental Systems / Geotech	BIOGAS 5000	Handheld multigas analyzer, optional H₂ cell	H₂ 0–1,000 ppm (optional); primary channels CH₄/CO₂/O₂	Mid-range professional handheld	[21]
Industrial Scientific	iBrid MX6	Handheld multigas analyzer, electrochemical H₂	H₂ 0–2,000 ppm, 1 ppm precision (per third-party comparison)	Mid-range professional handheld	[2]
Dräger	X-am 8000 (with XXS H2 HC sensor)	Handheld combustible-gas/multigas detector	H₂ 0–4 vol% (0–100% LEL), T90 ≈ 20 s	Mid-range professional handheld	[20]
Axiom Sensing (Australia)	WHALI (pumped autonomous logger), EVA (passive logger)	Continuous autonomous soil-gas monitoring station	H₂ 0–2,000 ppm, 10 ppm precision; solar-capable; ~300 mL/min average pumped duty cycle	Research/instrument-grade custom build; also offered as a commercial survey service	[2]
Université de Pau et des Pays de l’Adour / CNRS (France)	MONHyTOR	Buried non-pumped diffusive H₂ probe	0–1,000 ppm, 0.1 ppm resolution, ±5% FS, 1 Hz capable, months of battery autonomy	Research-grade; deployed as a fleet in France since 2024	[9]
SRI Instruments (USA)	Natural Gas Analyzer GC; 310 Soil-Gas GC System	Field-portable GC, TCD (+FID option)	TCD-only LOD ~200–300 ppm; field/battery-portable	~$5,000–20,000+ (per prior sensing-dossier vendor research; not independently re-verified in this session)	[22]
INFICON	Micro GC Fusion	Portable micro-GC, MEMS micro-TCD	~1 ppm LOD (optimized method); minutes/run; marketed for pipeline-quality natural-gas H₂	Professional field-analyzer class	[23]
West Systems (Italy)	Portable Fluxmeter	Soil diffuse-flux accumulation chamber	Measures CO2, CH4, H2S, VOC; no H₂ channel identified in sources reviewed	Professional field-instrument class	[11]
Labco / generic	Exetainer (12 mL, butyl septum)	Gas sample vial for lab-chain transport	Not H₂/CO-stable untreated; stable >60 days when modified per [24]	Consumable, low cost	[24]
SKC / generic	Tedlar (PVF) gas sampling bags	Gas sample bag for lab-chain transport	Vendor literature claims good stability for H₂, CO, CH₄, CO₂, SF₆ (unverified — not supported by cited source)	Consumable, low cost	[25]
Entech Instruments / generic	Summa (passivated stainless) canisters	Whole-air gas sample canister	GC-MS gold standard generally; H₂-generation risk from canister-wall chemistry reported elsewhere (unverified — not supported by cited source)	Consumable/rental, moderate cost	[26]
Gold Hydrogen Ltd / CSIRO (Australia)	Yorke Peninsula / Ramsay Project soil-gas survey program	Operator-run soil-gas survey campaign	~80 survey points, hammer-drilled to ≤1 m, handheld multigas spot reading + 24 h buried follow-up probe where H₂ detected	Operator/survey program, not a purchasable product	[7]

Open questions

The most rigorous instrument- and protocol-level work in this literature comes from a small set of groups:

• the Grenoble ISTerre group (Auclair, Lefeuvre, Dutoit, Donzé, Truche) produced the controlled-chamber GA5000 characterization [19]
• a Université de Pau et des Pays de l’Adour/CNRS team (Adjie, Bordes, Brito, Nasri, Normandin) produced the MONHyTOR probe-design paper [9]
• the Edith Cowan University/CSIRO Western Australia cluster (Davies, Haines, Thomas, Normore, Josse, Frery, Esteban, Keshavarz, Iglauer) produced the Harvey point-versus-autonomous comparison and the 83-hole drilling-artifact quantification [2][27]
• Moretti, Prinzhofer and collaborators established the francophone long-term monitoring tradition in Brazil and Mali [3][14][15]

The historical foundation for essentially the entire method is the 1986 USGS Kansas study [8], and a broader sensing-technology review of subsurface H₂-seep detection methods synthesizes how soil-gas hardware fits alongside the other transduction classes covered elsewhere in this wiki [29]. Specific open items for Hestia’s field program:

1. Commission or license a continuous H₂ soil-gas monitoring station design. No commercial off-the-shelf product exists; WHALI, MONHyTOR, and PARHyS are all bespoke research builds. Evaluate whether Axiom Sensing’s WHALI/EVA service offering, or an academic collaboration with Université de Pau et des Pays de l’Adour/CNRS (MONHyTOR), is faster to field than an in-house build using the electrochemical-cell guidance in Electrochemical cells.
2. Build (or commission) a purpose-built H₂ flux chamber. The mature commercial CO₂/CH₄/H₂S accumulation-chamber ecosystem (West Systems) has no H₂ analog identified in this research [11]; published H₂ flux work uses ad hoc builds [12]. Determine whether an existing West Systems-class chamber body can simply be re-plumbed to an electrochemical or MOX H₂ cell rather than designing a chamber from scratch.
3. Adopt a hard repeat-measurement / settling-time SOP as a non-negotiable field protocol, not an optional refinement: percussion or hammer drilling to ~1 m, an explicit purge-volume target (3–5 dead volumes for an already-equilibrated probe, escalating toward 8–12 for a freshly drilled hole), a settle interval, then a repeat reading to check for the exponential decay signature of a drilling artifact versus a stable or rising reading consistent with active seepage.
4. Pilot the modified-Exetainer protocol (silicone plug + stainless bolt) [24] before committing to any lab-confirmation sample chain. Off-the-shelf butyl-septum vials risk manufacturing a false positive purely from storage time; this is a cheap, already-published fix worth adopting immediately rather than discovering the problem in Hestia’s own QA/QC data.
5. Resolve Hestia’s own grid-spacing convention empirically rather than borrowing one. The one review that addresses this directly for H₂ soil gas [16] states plainly that the field does not yet know the optimal sampling step; a pilot nested-ring design (tens of meters near a candidate feature, widening outward) at an early Hestia site, with the point-density-vs-information-gain analysis the Harvey Perth Basin paper ran for temporal replication [2], would let Hestia establish this for its own target geology rather than defaulting to Kansas-scale (800 m) or Perth-Basin-scale (tens of meters) spacing without local justification.
6. Quantify diurnal/covariate regression requirements for Hestia’s specific climate. The Harvey Perth Basin correlation structure (RH the dominant single covariate, pressure a real but noisy driver, temperature weak) [2] and the Brazilian daily-periodic-emission timing (midday peak) [3] may not transfer directly to Hestia’s candidate basins; a short (2–4 week) continuous-logger pilot at a first site, following the WHALI duty-cycle model, would let the covariate regression above be fit locally before designing a full fixed-time spot-survey campaign around it.
7. Cross-reference Geologic context before reporting any single-visit anomaly internally. That page’s synthesis of the artifact-vs-signal controversy, and this page’s instrument- and protocol-level detail, should be read together by any Hestia geoscientist evaluating a candidate soil-gas result before it is escalated to a drilling decision.

Related pages

• Geologic context — the transport physics and the 2023–2025 drilling-artifact-vs-signal controversy behind every protocol choice on this page
• Why H2 is hard — the selection-rule and diffusivity framing behind H₂’s field-detection difficulty
• Calibration, drift & QA/QC — reference-gas traceability, drift, and the uncertainty budget behind every ppm figure quoted here
• Environmental covariance — the regression pipeline for separating pressure/humidity/temperature confounds from real signal
• Downhole deployment — where soil-gas prospecting hands off to drilling and downhole measurement
• Electrochemical cells — the dominant transduction physics behind nearly every fielded soil-gas instrument
• Thermal conductivity & acoustic sensors — the TCD physics behind field-portable GC detection
• Gas chromatography — compositional confirmation of candidate anomalies
• Mass spectrometry — the historical USGS Kansas detection method and modern isotope/source-diagnosis work
• Mud-gas logging — the percent-level, downhole-during-drilling regime that soil-gas survey must not be conflated with
• Sensor arrays & machine learning — multivariate pattern-recognition approaches for separating covariate-driven noise from real signal in a multi-channel continuous-monitoring dataset

References

Full reference list (0 entries)

[1] U.S. EPA (National Risk Management Research Laboratory / National Exposure Research Laboratory). “Leak, Purge, and Gas Permeability Testing to Support Active Soil Gas Sampling.” https://cfpub.epa.gov/si/si_public_file_download.cfm?p_download_id=537688&Lab=NRMRL (see also related EPA record “Evaluation of Vapor Equilibration and Impact of Purge Volume on Soil-Gas Sampling Results,” https://cfpub.epa.gov/si/si_public_record_Report.cfm?Lab=NRMRL&dirEntryID=167246).

[2] Davies, K., Haines, P.W., Thomas, C., Normore, L. (2025). “Natural hydrogen soil gas emissions near Harvey, Perth Basin: a comparative study of survey methods.” Science and Technology for Energy Transition 80, 46. DOI: 10.2516/stet/2025026. https://www.stet-review.org/articles/stet/pdf/2025/01/stet20250031.pdf

[3] Moretti, I., Prinzhofer, A., Françolin, J., Rosanne, M., Rupin, F., Mertens, J. (2021). “Long-term monitoring of natural hydrogen superficial emissions in a brazilian cratonic environment. Sporadic large pulses versus daily periodic emissions.” International Journal of Hydrogen Energy 46(5), 3615–3628. DOI: 10.1016/j.ijhydene.2020.11.026. https://www.sciencedirect.com/science/article/abs/pii/S0360319920341987

[4] (Same study as [3]; daily periodic midday-peak flux finding cross-confirmed via search-result synthesis of the abstract.)

[5] AMS, Inc. “Gas Vapor Probe Retract-A-Tip” product page (materials, tubing dimensions, SKU 210.01, price). https://www.ams-samplers.com/gvp-retract-a-tip/

[6] ENVCO Global. “AMS Retract-A-Tip” catalog page (retract mechanism, mesh-screen description, reuse/decontamination). https://envcoglobal.com/catalog/soil/soil-gas-sampling/reusable-soil-gas-samplers/ams-retract-tip; AMS, Inc. “Gas Vapor Probe Samplers” overview. https://www.ams-samplers.com/gas-vapor-probe-samplers/

[7] Gold Hydrogen Ltd / CSIRO (2023). “Soil Gas Survey Information Sheet — Yorke Peninsula Soil Gas Survey.” https://www.goldhydrogen.com.au/docs/GH_Soil_Gas_Survey_Information_Sheet.pdf

[8] McCarthy, J.H. Jr., Cunningham, K.I., Roberts, A.A., Dietrich, J.A. (1986). “Soil gas studies around hydrogen-rich natural gas wells in northern Kansas.” U.S. Geological Survey Open-File Report 86-461. https://pubs.usgs.gov/of/1986/0461/report.pdf

[9] Adjie, N., Bordes, C., Brito, D., Nasri, D., Normandin, É. (Université de Pau et des Pays de l’Adour, CNRS/LFCR-IPREM) (2025). “In-soil hydrogen concentration monitoring using MONHyTOR: application for natural hydrogen exploration.” Advances in Geosciences 67, 57. https://adgeo.copernicus.org/articles/67/57/2025/

[10] (iBrid MX6 range/precision figure sourced from the comparison table in Davies et al. 2025 [2].)

[11] West Systems s.r.l. “The West Systems Portable Fluxmeter — Diffuse Flux.” https://www.portablefluxmeter.com/the-west-systems-portable-fluxmeter-diffuse-flux-2/

[12] Halas, P., Dupuy, A., Franceschi, M., Bordmann, V., Fleury, J-M., Duclerc, D. (2021). “Hydrogen gas in circular depressions in South Gironde, France: Flux, stock, or artefact?” Applied Geochemistry 127, 104928. DOI: 10.1016/j.apgeochem.2021.104928. https://www.sciencedirect.com/science/article/abs/pii/S0883292721000603

[13] Davies, K., Esteban, L., Keshavarz, A., Iglauer, S. (2024). “Advancing Natural Hydrogen Exploration: Headspace Gas Analysis in Water-Logged Environments.” Energy & Fuels. DOI: 10.1021/acs.energyfuels.3c04562. https://pubs.acs.org/doi/10.1021/acs.energyfuels.3c04562

[14] Prinzhofer, A., Tahara Cissé, C.S., Diallo, A.B. (2018). “Discovery of a large accumulation of natural hydrogen in Bourakebougou (Mali).” International Journal of Hydrogen Energy 43(42), 19315–19326. DOI: 10.1016/j.ijhydene.2018.08.193. https://www.sciencedirect.com/science/article/abs/pii/S0360319918327861

[15] Prinzhofer, A., Moretti, I., Francolin, J., Pacheco, C., D’Agostino, A., Werly, J., Rupin, F. (2019). “Natural hydrogen continuous emission from sedimentary basins: The example of a Brazilian H2-emitting structure.” International Journal of Hydrogen Energy 44, 5676–5685. DOI: 10.1016/j.ijhydene.2019.01.119. https://www.sciencedirect.com/science/article/abs/pii/S0360319919302514

[16] Frery, E. et al. “Exploring natural hydrogen hotspots: a review and soil-gas survey design for identifying seepage.” Geoenergy, Geological Society of London / Lyell Collection. DOI: 10.1144/geoenergy2023-014. https://www.lyellcollection.org/doi/full/10.1144/geoenergy2023-014

[17] Range Complex / U.S. Army Environmental Command (2001). “Soil Gas Survey Introduction” (Camp Stanley Storage Activity, TX). https://www.stanley.army.mil/Volume3-1/Soil-Gas-Surveys/Intro.htm

[18] QED Environmental Systems Ltd. “GA5000 Technical Specifications” datasheet, DS45-Issue.15. https://www.wiserenvironment.co.uk/wp-content/uploads/2020/11/21167-QED-Geotech-GA5000-Datasheet-v4.pdf

[19] Auclair, S., Lefeuvre, N., Dutoit, H., Donzé, F-V., Truche, L. (2024). “Assessing the Portable Gas Analyzer GA5000 for Accurate H₂ Monitoring in Geological Exploration.” Geonum Ed., ISRN GEONUM-NST—2024-01—ENG. https://hal.science/hal-04861335

[20] Dräger. “X-am 8000” product page and “Sensor XXS H2 HC 0-4 Vol.-%” product listing. https://www.draeger.com/en-us_us/Products/X-am-8000; https://hazmatresource.com/product/gas-detection/drager-sensors/draeger-sensor-xxs-h2-hc-0-4-vol-6812025/

[21] Geotech / QED. “BIOGAS 5000” product pages (optional H₂ 0–1,000 ppm cell). https://norrscope.com/product/geotech-biogas-5000/ ; https://diamondsci.com/products/biogas-gem%E2%84%A25000-with-h2s-portable-methane-detector

[22] SRI Instruments. “Natural Gas Analyzer GC” and “310 Soil Gas GC System” product pages (TCD detection limit ~200–300 ppm). https://srigc.com/product/gas-chromatographs/natural-gas-analyzer-gc; https://srigc.com/product/gas-chromatographs/310-soil-gas-gc-system

[23] INFICON. “Micro GC Fusion Gas Analyzer” product page and “Monitoring H2 in Pipeline-Quality Natural Gas with Micro GC Fusion” application note. https://www.inficon.com/en/products/chemical-detection-and-utility-monitoring/micro-gc-fusion-gas-analyzer; https://www.inficon.com/en/news/monitoring-h2-in-pipeline-quality-natural-gas-with-micro-gc-fusion

[24] Nauer, P.A., Chiri, E., Jirapanjawat, T., Greening, C., Cook, P.L.M. (2021). “Technical note: Inexpensive modification of Exetainers for the reliable storage of trace-level hydrogen and carbon monoxide gas samples.” Biogeosciences 18, 729–737. DOI: 10.5194/bg-18-729-2021. https://bg.copernicus.org/articles/18/729/2021/

[25] EAS Laboratories. “Tedlar Bag Information.” https://easlab.com/tedlarbagsinfo.htm (does not support the claim)

[26] Entech Instruments. “SUMMA Canisters” product page. https://www.entechinst.com/summa-canisters/ (does not support the claim)

[27] Davies, K., Josse, R., Frery, E., Esteban, L., Keshavarz, A., Iglauer, S. (2025). “Distinguishing drilling-induced artifacts from naturally occurring hydrogen in soil gas surveys: Insights from sub-circular depressions.” International Journal of Hydrogen Energy 109, 1230–1240. https://ro.ecu.edu.au/ecuworks2022-2026/5818/ ; https://www.sciencedirect.com/science/article/pii/S0360319925006597

[28] (Zero-order-release mechanism finding is part of the same Nauer et al. 2021 study, [24].)

[29] Mainson, M., Heath, C., Pejcic, B., Frery, E. (2022). “Sensing Hydrogen Seeps in the Subsurface for Natural Hydrogen Exploration.” Applied Sciences 12(13), 6383. DOI: 10.3390/app12136383. https://www.mdpi.com/2076-3417/12/13/6383

[30] Petron, G. et al. (2024). “Atmospheric H2 observations from the NOAA Cooperative Global Air Sampling Network.” Atmospheric Measurement Techniques 17, 4803. Marine boundary layer global mean H₂ = 552.8 ppb in 2021 (≈0.553 ppm). https://ui.adsabs.harvard.edu/abs/2024AMT…17.4803P/abstract