Thermal conductivity & acoustic (speed of sound) sensors

Summary

Hydrogen is the lightest molecule found in the field. It carries heat away from a hot surface roughly seven times faster than air, and sound travels through it nearly four times faster than through air. Both properties convert directly into a sensor: a heated wire or membrane read for its cooling rate (thermal-conductivity, or katharometer, sensing), or a path or cavity read for the speed of sound passing through it (acoustic sensing). Neither needs a catalyst, oxygen, or any chemistry, so the family resists poisoning and drift, and has been used this way since 1915 in hydrogen-cooled generator purity meters and industrial safety sensors.

Both methods report a single scalar, a bridge voltage or a resonance frequency, and convert it to concentration assuming a fixed two-gas background. Real soil gas, mud gas, and generator housings routinely carry five or six gases at once. Helium, a common geologic companion of natural hydrogen, pushes that same number in the same direction hydrogen does, so a helium-rich, hydrogen-free sample can read as a large hydrogen anomaly.

For Hestia: strong for percent-level safety monitoring and mud-gas-return analysis at surface. Weak for the ppm-level soil-gas reconnaissance surveys, against a ~0.5 ppm atmospheric background, where most exploration decisions get made.

At a glance — published performance envelope

Detection limit

1–2,000 ppm

Range

500–1,000,000 ppm

Response t₉₀

1–60 s

Recovery

1–60 s

Selectivity

low

Drift

low

Operating T

−40 … 90 °C

Humidity

0–100 % RH

Power

0.05 mW – 5 W

Unit cost

$10–$10k

TRL

9

Main interferents: CO₂, CH₄, He, H2O vapor · poisons: . Ranges span the published devices/variants covered below — per-claim citations are in the body; data: src/data/modalities/thermal-conductivity-acoustic.yaml.

How it works

Hydrogen’s low molar mass (2.016 g/mol) pushes two transport properties, thermal conductivity and speed of sound, to extremes at once. No other common exploration gas (air, N₂, O₂, CO₂, CH₄, H₂S) comes close on either axis except helium, which shares hydrogen’s light-molecule character and is the chief interferent for both modalities [1][2].

Thermal conductivity. For a dilute gas, $\lambda \approx \frac{1}{3} n \bar{c} \, \ell \, c_v$; the more practical Eucken-type relation for polyatomic gases is $\lambda = \left(c_v + \frac{5}{4}\frac{R}{M}\right)\eta$, where $\eta$ is viscosity [3]. Hydrogen’s low mass drives a high mean molecular speed and low viscosity-normalized momentum transfer, giving it roughly seven times air’s thermal conductivity:

Gas	λ (W·m⁻¹·K⁻¹), ≈290–300 K	Source
H₂	0.180–0.187	[1][3][4]
He	~0.142–0.151	[5]
Air	0.024–0.026	[1][3]
N₂	0.0234–0.0258	[1]
CH₄	0.030–0.033	[1]
CO₂	0.0146–0.0166	[1][3]
H₂S	~0.0146 (approx., low-T value; unverified at 300 K)	[6]

This is the largest thermal-conductivity contrast of any common exploration gas relative to air, which is why G. A. Shakespear invented the katharometer in 1915 specifically to detect hydrogen in air mixtures [7][8].

Sensor equation. A thermal-conductivity element (wire, thermistor, or MEMS membrane) held at $T_h$ above ambient $T_a$ loses heat mainly by conduction through the gas, plus usually-small convection, radiation, and lead conduction:

$$I^2 R_h = h(\lambda_{\text{gas}}) \, A \, (T_h - T_a) + \varepsilon\sigma A(T_h^4-T_a^4) + Q_{\text{leads}}$$

Two matched elements in a Wheatstone bridge, one in sample gas and one in reference gas, give a bridge output $\Delta V \propto \Delta R \propto \frac{\partial R}{\partial \lambda}(\lambda_{\text{sample}} - \lambda_{\text{reference}})$. This differential architecture makes classical TCDs binary-mixture instruments: the physics cleanly resolves only a two-component system, analyte plus one known balance gas, which breaks down in real multi-gas soil gas (see Cross-sensitivity below).

Transient and 3ω variants. The transient hot-wire method reads the time constant of a heating step, inversely related to $\lambda_{\text{gas}}$. The 3ω method drives the heater with an AC current at frequency $\omega$ and reads the 3ω harmonic via lock-in, extracting both $\lambda$ and thermal diffusivity $D = \lambda/(\rho c_p)$ from the amplitude and phase versus $\omega$ [9][10]. Because $D$ and $\lambda$ do not scale identically across gases, 3ω partially discriminates interferents (H₂ vs. CO₂ against N₂) that a single DC bridge cannot [9][10].

Speed of sound. $c = \sqrt{\gamma R T/M}$, where $\gamma = c_p/c_v$ [11]. Hydrogen’s molar mass sits a factor of ~14.4 below air’s (2.016 vs. ~28.97 g/mol); with $\gamma_{H_2}\approx1.41$ close to $\gamma_{air}\approx1.40$, the speed-of-sound ratio works out to $\sqrt{14.4}\approx3.8$. Concretely, $c_{\text{air}}\approx343$ m/s at 20 °C versus $c_{\text{H}_2}\approx1270$–1320 m/s depending on temperature and ortho-para composition [11][12]; a 2024 metrology paper still refines $c_{H_2}(T,P)$ for industrial flow-measurement accuracy [13].

For a binary mixture, $M_{\text{mix}} = x_1 M_1 + x_2 M_2$, so a single sound-speed or resonance-frequency measurement of a known two-gas system yields the molar mixing ratio directly, no chemistry, catalyst, or optical absorption line needed [14][15][16]. Two acoustic transducers a known distance $L$ apart give transit time $t$ and $c=L/t$; differencing upstream/downstream time of flight also yields flow velocity, which is why ultrasonic flow meters often report hydrogen purity alongside flow [17][18]. A resonant cavity’s standing-wave frequencies $f_n = nc/(2L)$ track $c$, and hence composition, via a phase-locked loop without needing absolute-timing precision. Steven Garrett’s sonic gas analyzer used a self-resonant tube (2 cm diameter × 3 cm long) capped by two small speakers, deriving mean molecular weight six times per minute from resonance frequency; ~0.02 Hz frequency stability corresponds to ~30 ppm concentration resolution, with overall system stability better than 350 ppm [14][19].

Ratios and limits worth carrying forward:

Quantity	Value	Note	Ref
Thermal-conductivity contrast, H₂/air	≈7	Largest of any common exploration gas	[1][3]
Speed-of-sound ratio, H₂/air	≈3.7–3.8	—	[11][12]
Knudsen-regime crossover	—	Pirani-style thermal sensors: once mean free path $\ell$ approaches the sensor-to-wall gap $d$, $\lambda_{\text{eff}}$ becomes pressure-dependent, so uncompensated thermal sensors are implicitly barometric	[20]
Forced convection (King’s Law term)	—	Ambient flow across a hot wire adds a velocity-dependent heat-loss term that can swamp the composition signal; practical sensors are housed to minimize it or correct for it	[9]

Device classes

• Classical katharometer / hot-wire TCD. Four matched filaments (tungsten, platinum, platinum-alloy, or thermistor beads) in a Wheatstone bridge, two in sample flow and two sealed in reference gas. The architecture behind GC-TCD detectors, hydrogen-purity katharometers (Eaton/MTL K522, ABB 6517), and Baker Hughes/Panametrics process analyzers [21][22][23][24].
• MEMS hot-plate / suspended-membrane thermopile. A heater or thermopile pair on a thin SiN/SiO₂ membrane over a micromachined cavity, reducing thermal mass and substrate losses versus a free-standing wire. Differential designs add a matched reference thermopile on an unexposed membrane for common-mode temperature rejection. Examples: Xensor XEN-TCG3880 [25][26], Sensirion STC31/STC31-C [27][28], Posifa PGS1000/PGS4100 [29][30], SGX/Neroxis TC-1326/MTCS-2601 [31][32], and a research Pt/Al₂O₃-catalyst thermopile pair reaching 1 ppm LOD [33][34].
• Transient hot-wire / 3ω sensors. AC or pulsed heater excitation, using a helix-wound 8 µm tungsten microwire or a suspended microheater bridge, extracts both $\lambda$ and $D$ as a second, partially independent discrimination axis [9][10][35].
• MEMS Pirani-style / thermal-flow variants. Same structure as the MEMS TC sensor, operated where molecular-flow (Knudsen-regime) heat transfer dominates, a vacuum gauge rather than a gas-composition sensor. Recent work uses porous-silicon thermal isolation to reach the fine/rough vacuum range [20][36][37]; the commercial Neroxis MTCS-2601 covers 10⁻⁴–1000 mbar at <6 mW [32]. Thermal-flow (hot-wire anemometry) sensors use the same physics to read mass flow instead, often on the same die.
• Multi-property sensors (NevadaNano MPS). Combines a resistive-heater thermal measurement with other thermodynamic properties (heat capacity, thermal diffusivity) and onboard classification to report %LEL across ~14 combustible gases including H₂ from one factory calibration, marketed as non-cross-sensitive and immune to drift, poisoning, and decay [38].
• Ultrasonic time-of-flight (ToF) analyzers. Paired piezoelectric or MEMS (CMUT/FBAR) transducers a fixed path apart measure transit time; used in industrial ultrasonic flow meters that report gas velocity, volumetric flow, and, as an optional derived variable, hydrogen purity for hydrogen-natural-gas blends [17][18][39][40].
• Acoustic resonator / standing-wave analyzers. A closed cavity (tube, Helmholtz resonator, or piezoelectric tube) driven at its natural resonance; resonance-frequency shift tracked with a phase-locked loop yields mean molecular weight and hence composition. Implemented in hydrogen-cooled turbogenerator purity meters and demonstrated as a compact sonic gas analyzer for hydrogen/methane leak and mine-safety detection [14][19][41].
• MEMS ultrasonic (CMUT/FBAR) sensors. Capacitive micromachined ultrasonic transducers and film bulk acoustic resonators as chip-scale acoustic transducers; uncoated CMUTs operated broadband have measured sound velocity and attenuation together to discriminate H₂, CO₂, and CH₄ against N₂, a candidate solid-state, non-thermal acoustic gas sensor on a chip [17][42][43].
• GC-TCD. The standard laboratory and wellsite (mud-gas logging) reference method for quantifying H₂ alongside CH₄, CO₂, He, and N₂. Chromatographic separation ahead of the TCD cell sidesteps the multi-component binary-mixture problem, at the cost of a carrier gas, a column, minutes-not-seconds cycle time, and a carrier-gas choice that must avoid a thermal conductivity too close to H₂’s (see Cross-sensitivity, below) [21][44][45][46].

H₂ and He are the only two gases far from the rest on both transport properties, so a thermal-conductivity or acoustic cell reads them strongly but cannot tell them apart (see helium & co-tracers).

Performance envelope

Every cell keeps its source citation; entries without a directly verifiable number are marked “(unverified)” rather than invented.

Parameter	Classical katharometer (industrial TCD)	MEMS TC (Xensor/Sensirion/Posifa/SGX class)	3ω / transient thermal	Acoustic resonator / ToF
Practical LOD (H₂ in air)	~0.1–1 vol% typical process use; some lab TCDs to low-ppm with chromatographic pre-separation	1 ppm demonstrated in research thermopile design [33][34]; ~10 ppm analog / 2 ppm digital resolution, ~1200 ppm accuracy floor in commercial PGS4100 [47]; SAW-augmented TC design ~6 ppm [48]	0.062% (620 ppm) H₂ at 1 Hz in a robust tungsten-microwire design [9]; suspended-microheater theoretical floor ~1400 ppm at reduced uncertainty [35]	~30 ppm concentration resolution demonstrated in a resonant-tube sonic analyzer [14][19]; ±30 ppm H₂ accuracy quoted for the same device [49]
Measurement range	0–100 vol% (full binary range) [22][23]	0–4 vol% (PGS4100) [47] to 0–100 vol% (Xensor XEN-5310/5320, STC31 CO₂ mode 0–100 vol%) [26][27]	Demonstrated to ~1% H₂ (safety-limited in testing), 100% CO₂ [9]	0–100% H₂ in binary systems (turbogenerator purity meters); up to 30–40% H₂ blends in NG pipelines for flow meters [17][40]
Response time (t90)	Seconds to ~1 min depending on cell volume/flow	1 s (PGS4100) [47]; ~9 ms thermal time constant, chip-level (XEN-TCG3880 in air) [26]; 1.9 s (research thermopile) [34]	~28 s response / 23 s recovery (tungsten 3ω design) [9]	Sub-100 ms reported for some ultrasonic ToF hydrogen-sensor demonstrations (unverified precise figure); resonant sonic analyzer samples composition ~6×/min [14]
Recovery time	Seconds–minutes	1.4 s (research thermopile) [34]	~23 s [9]	Comparable to response time (no chemical desorption step)
Selectivity	Low: bulk property, responds to any Δλ from reference gas [8][21]	Same physical limitation; compensated variants (STC31-C) add RH/O₂ correction [28]	Improved via λ+D dual read [9][10]	Low: responds to mean molar mass, not gas identity
Operating temperature	Typically −20 to +85 °C sensor head; katharometer analyzer cabinets to ~−40 to +60 °C ambient (process-analyzer class)	−40 to +85/90 °C (Posifa PGS4100 op/storage) [47]; −40 to +85 °C (Xensor integrated systems) [26]	Demonstrated near ambient / ~137 °C heater DC operating point [9]	Ambient-rated for most commercial units; no fundamental barrier to higher T with appropriate transducer materials
RH range	0–100% RH with compensation (PGS4100 has onboard RH+P compensation) [47]	0–100% RH nominal, accuracy budget affected [27][47]	Not well characterized (unverified)	Humidity changes local sound speed and must be compensated (handled electronically in the Garrett design via a local temperature sensor) [19]
Long-term drift	Katharometer/E-One GGA: <0.2%/month (best-in-class, drift-corrected design) [24]	Xensor XEN-5310: ~1200 ppm/yr typical transfer drift [26]; Posifa PGS4100: 1200 ppm over 5 years long-term stability spec [47]	Not yet established at commercial maturity (unverified)	Garrett design: <350 ppm total system stability [19]
Power	Watts (heated filament, continuously powered) for classical katharometers	~5–8 mA at 5 V (Xensor XEN-5310, room temp) [26]; 50 µA average / 5 mA peak (Sensirion STC31, pulsed) [27]; 8 mA (Posifa PGS1000) [50]; <6 mW (Neroxis Pirani-regime MTCS-2601) [32]	mW-range heater drive, higher than ultra-low-power NDIR alternatives (unverified precise number)	mW-class electronics; acoustic driver power small but nonzero
O₂ required	No: purely physical, non-catalytic [8][21][51]	No	No	No
Poisoning susceptibility	None: no catalyst, no chemisorption step; “intrinsically immune to chemical poisoning” [51]	None	None	None
Lifetime	Effectively unlimited (no consumable, no catalyst degradation), years to decades in process service [21][24]	Years (no reported wear-out mechanism beyond membrane/heater electromigration at extremes)	Not yet field-proven at scale (unverified)	Years; no consumable
Cost class	Industrial process analyzer (katharometer + cabinet + electronics): low-thousands to ~$10⁴ USD class (qualitative, no verified list price found)	Chip/die level: $10–70 USD (Xensor XEN-TCG3880 ~$10–20 on distributor listings [52]; Sensirion STC31-C-R3/R5 $24–65 depending on distributor/qty [53]; Posifa PGS1004 $35–45 in small qty [54])	Research/prototype stage, not yet a commercial line item (unverified)	Sonic/resonator analyzer modules: research-grade prototypes to industrial purity-meter systems (E/One GGA, Panametrics XMTCpro); cost class not independently verified
TRL (qualitative)	9, mature industrial product line (GC-TCD since 1950s–60s; turbogenerator purity meters decades in service) [7][8][22][24]	7–9, Sensirion STC31/Posifa PGS4100 are shipping commercial parts; research thermopile/SAW variants at TRL 3–5	3–4, laboratory demonstrations only [9][35]	6–9, turbogenerator/flow-meter purity applications mature; MEMS CMUT/FBAR gas-discrimination variants at TRL 2–4

Figures of merit

Specializations of the universal definitions on the Figures of merit page; units follow Concentration units.

Figures-of-merit table for this modality (12 rows)

Figure of merit	Symbol / formula	What it means here	Typical value for this technology	How to read it
Thermal-conductivity contrast	$\Delta\lambda/\lambda_{ref} = (\lambda_{H_2}-\lambda_{ref})/\lambda_{ref}$	How far H₂’s thermal conductivity sits above the air/N₂ reference a Wheatstone-bridge or MEMS TC cell is balanced against, the raw signal-generating quantity for the entire TC sensor family	$\lambda_{H_2}\approx0.180$–0.187 W·m⁻¹·K⁻¹ vs. $\lambda_{air}\approx0.024$–0.026 W·m⁻¹·K⁻¹ → ratio ≈ 7, the largest contrast of any common exploration gas [1][3]	Bigger ratio means a bigger raw bridge signal per ppm H₂, but the same physics gives helium (λ≈0.15) a large signal on the same side, which is why this ratio alone says nothing about selectivity
Speed-of-sound ratio (acoustic variant)	$c=\sqrt{\gamma RT/M}$; ratio $c_{H_2}/c_{air}$	The acoustic analog of the row above: how far H₂’s sound speed is shifted from the air/N₂ baseline a resonator or ToF cell is calibrated against	$c_{air}\approx343$ m/s (20 °C) vs. $c_{H_2}\approx1270$–1320 m/s → ratio ≈ 3.7–3.8 [11][12]	The same trade-off as thermal contrast: a large, easily-measured raw shift, but molar-mass-driven, so light helium produces a comparably large, indistinguishable shift without a second (e.g., attenuation) measurement axis
Sensitivity	$\Delta V \propto \partial R/\partial\lambda\,(\lambda_{sample}-\lambda_{ref})$ (bridge) or $\partial f/\partial c$ (resonator)	Signal moved per unit Δλ or Δc, the general sensitivity definition specialized to a bridge voltage or resonance-frequency slope rather than a chemical current	No single vendor-quoted nA/ppm-style figure exists for this family; effective sensitivity is expressed indirectly through resolution, e.g. 10 ppm (analog) / 2 ppm (digital) per count on the Posifa PGS4100 [47]	Because $\lambda(C)$ and $c(C)$ are smooth, nearly linear functions of composition over most of the binary range, sensitivity is close to constant across decades, unlike log-linear MOX sensors, which is why this family’s dynamic range (below) is so wide
Limit of detection (LOD)	$\mathrm{LOD}=k\,\sigma_{blank}/S$ (see definitions page)	Smallest ΔH₂ the bridge or resonance readout can distinguish from a zero-gas baseline	Classical katharometer ~0.1–1 vol%; commercial MEMS ~few hundred ppm to ~1200 ppm accuracy floor (Posifa PGS4100) [47]; 3ω tungsten-microwire 620 ppm at 1 Hz [9]; research thermopile 1 ppm (not yet commercial) [33][34]; acoustic resonator ~30 ppm resolution in a lab demonstration [14][19]	Every one of these numbers implicitly assumes a specific, often-unstated integration time and bandwidth (1 Hz for the 3ω figure, ~6 measurements/min for the acoustic resonator), per the Allan-deviation caveat on the definitions page, a faster reading rate on the same hardware will show a worse LOD, so these are not directly comparable without matching dwell time
Selectivity / cross-sensitivity	$K_{ij}=S_j/S_i$; overall selectivity $1/\max_j(K_{ij})$	Both families measure one scalar bulk property and cannot distinguish which gas produced a given Δλ or Δc	Low by construction: “responds to any Δλ from reference gas,” not to H₂ specifically [8][21]; even the 3ω method’s λ+D dual read only partially discriminates interferents [9][10]	This is the family’s structural weak point: unlike an electrochemical cell’s CO cross-sensitivity (a few percent, correctable), a TC/acoustic sensor has no chemistry to make selective in the first place
Interferent equivalent-ppm (He/CH₄/CO₂)	$\Delta_{H_2,eq}\approx x_i\times\dfrac{\lambda_i-\lambda_{air}}{\lambda_{H_2}-\lambda_{air}}\times10^6$	Converts a percent-level co-present gas into how many ppm of false H₂ signal it produces on an uncompensated bridge, the field-relevant, quantified form of the row above	First-order estimates: 1% CO₂ ≈ −650 ppm H₂-equivalent (masks a real anomaly); 1% CH₄ ≈ +350 ppm H₂-equivalent (false-high); 1% He ≈ +7,700 ppm H₂-equivalent (dominant risk), order-of-magnitude illustrative, not vendor-published (unverified) [1][3]	Helium is the most dangerous interferent because it sits on the same high-conductivity, high-sound-speed side as H₂ and is frequently co-generated with it geologically, exactly the exploration targets of interest; CO₂ biases readings low, CH₄ and He bias them high
Binary-mixture validity	Wassiljewa/Mason-Saxena mixing rule vs. a real n-component gas	Both families’ inversion from Δλ or Δc to concentration assumes exactly two gases are present; this row quantifies the error when that assumption breaks	<3–4% error for nonpolar-nonpolar binary pairs; >5–8% error once three or more gases (or a polar species like H₂O or CO₂) are present [57][58]; real soil gas is essentially never binary [55][56]	Not a spec-sheet number but a structural validity bound: treat any TC/acoustic reading in real multi-gas field air as carrying this extra, uncorrectable error on top of the interferent bias above; GC-TCD sidesteps this problem by separating gases in time before they reach the detector
Response / recovery time	$t_{90}=2.303\,\tau$ (first-order)	How fast the bridge voltage or resonance frequency settles after a concentration step, and returns to baseline after	MEMS ~1 s (Posifa PGS4100) [47], ~9 ms chip-level thermal time constant (Xensor XEN-TCG3880) [26]; classical katharometer seconds to ~1 min; 3ω ~28 s response / 23 s recovery [9]; acoustic resonator samples composition ~6×/min [14]	Among the fastest-responding H₂ sensor families in this wiki because there is no diffusion-limited chemistry or catalytic surface step, and, unlike MOX/catalytic sensors, recovery is comparably fast to response since there is no desorption tail
Dynamic / linear range	$N_{decades}=\log_{10}(C_{max}/C_{min})$	How wide a concentration span one sensitivity slope covers	0–100 vol% (full binary range) for classical katharometers and many acoustic resonators [22][23]; MEMS commonly 0–4 vol% (PGS4100) [47] up to 0–100 vol% (Xensor XEN-5310/5320, STC31) [26][27]	One of the widest, most genuinely linear ranges of any H₂ sensor family covered in this wiki, a real strength for percent-level and LEL work, but irrelevant to the ppm-level floor (LOD row) that actually limits soil-gas use
Drift / long-term stability	$R_0(t)=R_0(0)+d_Zt$; $S(t)=S(0)(1+d_St)$	Baseline and slope creep over months to years	Best-in-class katharometer/E-One GGA <0.2%/month [24]; MEMS ~1200 ppm/yr (Xensor XEN-5310) [26] or 1200 ppm over 5 yr (Posifa PGS4100) [47]; acoustic resonator <350 ppm total system stability [19]	This is the family’s main strength: no catalyst and no chemisorption step to degrade, so drift is close to instrumentation-only, roughly an order of magnitude better than catalytic or MOX competitors, and the main reason this modality is trusted for unattended safety monitoring
Power	Continuous vs. pulsed heater/driver draw	Electrical power needed to hold the sensing element at temperature (TC) or drive the transducer (acoustic)	Watts, continuously, for classical hot-wire katharometers; MEMS ~5–8 mA at 5 V (Xensor XEN-5310) [26], 50 µA avg/5 mA peak pulsed (Sensirion STC31) [27], <6 mW in Pirani-regime devices (Neroxis MTCS-2601) [32]; acoustic driver electronics mW-class	MEMS variants are compatible with battery-powered field stations; classical hot-wire katharometers are not, which is why they show up as powered process/safety analyzers rather than portable survey instruments
Operating envelope	$C_{reported}=C_{raw}\cdot f_T(T)\cdot f_{RH}(RH)\cdot f_P(P)$	The nested storage ⊃ operating ⊃ compensated domains from the definitions page, applied to a bulk-property sensor whose governing equations ($\lambda$, and especially $c=\sqrt{\gamma RT/M}$) have temperature baked in directly	−40 to +85/90 °C operating for MEMS parts [47][26]; 0–100% RH with onboard compensation on RH/P-equipped parts (PGS4100) [47]; acoustic sensors require real-time temperature (and often humidity) compensation because $c$ depends explicitly on $T$ and local $M$ [19]	A 60 °C ambient swing changes air’s own thermal conductivity enough to compete with a 1,000 ppm H₂ signal in an uncompensated design (unverified precise multiplier; not confirmed in the underlying 3ω source; see dossier), outside the compensated range, both families keep producing numbers with no signal that the correction has stopped applying

What dominates

Selectivity decides the field verdict. Thermal conductivity and speed of sound are non-selective bulk properties, so a helium-rich, hydrogen-free sample can read as a multi-thousand-ppm hydrogen anomaly, which rules the family out for ppm-level soil-gas reconnaissance regardless of raw LOD.

This sits at one extreme of the trade-off triangle on the Figures of merit page: near-unmatched drift immunity and lifetime, bought by giving up selectivity almost entirely. The trade pays off only once concentrations are at the percent level, safety/LEL monitoring and mud-gas returns, where the interferent bias is a small fraction of the signal rather than dominating it.

Cross-sensitivity & environment

Both families measure one scalar bulk property and invert it to concentration assuming a fixed two-gas background. Real soil gas, mud gas, and generator atmospheres carry N₂, O₂, CO₂, CH₄, H₂S, He, and water vapor at once [55][56], so any co-present gas outside the assumed balance gas folds into the reported concentration, undetected. This is the family’s central limitation, quantified in the table below using $\Delta_{H_2,eq}\approx x_i\times\frac{\lambda_i-\lambda_{air}}{\lambda_{H_2}-\lambda_{air}}\times10^6$ with $\lambda_{H_2}=0.187$, $\lambda_{air}=0.026$ W/m·K; treat the figures as order-of-magnitude estimates from cited pure-gas conductivities, not vendor error-budget numbers.

The Wassiljewa/Mason-Saxena mixing rule is the standard engineering approximation for real (three-plus gas) mixtures. It runs under 3–4% error for nonpolar-nonpolar binary pairs but over 5–8% once a polar species (H₂O, CO₂) is present [57][58], which is the normal soil-gas case.

Interferent	Effect on TC sensor	Effect on acoustic sensor	Effect on GC-TCD
CO₂	Negative bias, ≈−650 ppm H₂-equivalent at 1% CO₂: λ_CO2 < λ_air, opposite sign from H₂, masks real anomalies [1][3]	Lowers apparent sound speed if misattributed to H₂	Resolved chromatographically, but co-elution / peak overlap possible
CH₄	Positive bias, ≈+350 ppm H₂-equivalent at 1% CH₄: λ_CH4 > λ_air, same sign as H₂, false-high risk [1][3]	Positive bias (lower molar-mass increase vs. air, though far smaller effect than H₂)	Resolved chromatographically
He	Large positive bias, ≈+7,700 ppm H₂-equivalent at 1% He: dominant risk, and geologically co-occurring with H₂ [1][3]	Very large positive bias (light monatomic gas)	Can require carrier-gas change (below); often the actual carrier gas
H₂S	Small, uncertain-sign bias (λ lower than air; unverified magnitude)	Minor (heavier molecule, small effect at typical soil-gas ppm levels)	Resolved chromatographically; TCD not poisoned by H₂S unlike some electrochemical cells
RH/H₂O vapor	Moderate positive bias, worsens with T	Directly changes local $\gamma$, $M$ of sampled gas, must be compensated	Managed via sample conditioning/drying upstream
Temperature	Can dominate signal by ~100× if uncompensated (unverified; not confirmed in [9])	Directly enters $c=\sqrt{\gamma RT/M}$; must be measured/compensated in real time	Column/detector thermostatting standard practice
O₂	None required, no cross-sensitivity mechanism	None	Carrier-gas dependent
Pressure/elevation	Matters for Knudsen-regime (Pirani) devices; compensated in modern MEMS parts	Enters via density/real-gas effects; managed by calibration envelope	Standard GC pressure control

First-order estimates

The ppm-equivalent figures above are first-order linear estimates from cited pure-gas thermal conductivities, not vendor-published error-budget numbers. The direction is robust regardless of the exact λ values used: CO₂ biases TC-based H₂ readings low, CH₄ and He bias them high, and all three routinely co-occur with H₂ in the settings Hestia surveys.

Oxygen and poisoning. Neither family needs oxygen, unlike pellistors, which need ~10%+ O₂ for complete catalytic oxidation (see catalytic pellistors) [51]. With no catalyst and no chemisorption step, they cannot be poisoned by silicones, sulfur compounds, lead, or halogens the way catalytic and some electrochemical/semiconductor sensors can, an advantage in H₂S-co-evolving geologic settings [51]. One exception: strongly halogenated compounds can attack the passivation layer of classical hot-wire TCD filaments over long exposure, a GC-TCD operational note rather than a field soil-gas concern [8].

GC-TCD carrier-gas trap. Using helium as GC carrier gas is a documented failure mode directly relevant to wellsite mud-gas-logging GC systems. H₂ and He have similar, both very high, thermal conductivities, and the mixture’s conductivity-vs-mole-fraction curve is non-monotonic, so high H₂ concentrations with a He carrier can produce signal inversion, s-shaped, or negative chromatographic peaks; H₂ typically cannot be quantified reliably above roughly 1% with He carrier [59][60][61]. Standard mitigations: switch carrier gas to argon or nitrogen (at some cost to overall TCD sensitivity for other analytes), or dope the He carrier itself with ~8.5% H₂ so the analyte peak stays on the linear side of the crossover, a technique specifically validated for Agilent Micro-GC hydrogen analysis [59][62][63].

Carrier-gas check

Confirm any Hestia wellsite GC-TCD deployment’s carrier-gas choice against this known artifact before trusting low-level H₂ quantification.

Use for geologic H₂

Field verdict

Treat TC and acoustic sensors as percent-range confirmatory and safety instruments, not exploration-grade ppm detectors. They are a good second, physically orthogonal check on electrochemical/semiconductor front-line soil-gas sensors once a site shows strong (>0.1%) seepage, the primary sensor for mud-gas-logging return-gas analysis at surface, and the sensor of choice for fixed LEL/safety monitoring around H₂ handling equipment. They should not be relied on as the sole or primary sensor for reconnaissance-level soil-gas surveys, where the target signal is single- to low-double-digit ppm above a ~0.53 ppm background.

Soil-gas surveys. Weakest use case. Atmospheric H₂ background is ~0.53 ppm (530 ppb) [64][65], and “anomalous” near-surface soil-gas readings at many reported natural-seep sites range from a few ppm to a few thousand ppm, with point surveys showing 0 to >1,000 ppm variability across sampling days [56]. Every realistic TC-based LOD sits at ≥1–500 ppm at best: commercial MEMS parts read a few hundred ppm to low-thousands-ppm accuracy floor [47], and the best research thermopile designs reach 1 ppm but are not yet commercial [33][34]. Acoustic/resonant methods show ~30 ppm resolution in a controlled lab demonstration [14][19] but have no demonstrated field soil-gas deployment. TC and acoustic sensors can resolve the upper end of soil-gas anomalies (hundreds of ppm to percent level) but cannot reliably distinguish 1–10 ppm anomalies from atmospheric background, and the cross-sensitivity quantified above can bias even in-range readings in either direction from gases routinely present at exactly the sites of interest. The portable analyzer most commonly cited in the natural-hydrogen soil-gas literature, the Geotech GA5000, uses an electrochemical, not thermal-conductivity, cell for its H₂ channel [66][67][68] (see electrochemical cells), independent confirmation that the field-standard instrument has already moved away from TC sensing for low-ppm soil-gas work.

Downhole / mud-gas logging. Surface mud-gas logging is established practice: post-degassing concentrations are frequently well above the percent-level range where TC/acoustic sensing is strong, and GC-TCD is the standard method, including a 2025 SPE paper specifically on mud gas logging for hydrogen exploration [21][44][45][46][69]. True downhole (in-borehole, at the bit or on wireline) sensing has no path yet: heated-wire and MEMS thermopile elements are active electronic devices needing continuous power and signal conditioning, and standard downhole logging electronics (Si- or SiC-based) are qualified only to ~175 °C against a ~400 °C target for deeper wells [70][71]. No literature surveyed describes a purely passive downhole TC or acoustic hydrogen sensor. Fiber-optic sensing is the established all-optical alternative for HTHP wells because it avoids downhole electronics entirely [71][72][73], though hydrogen-darkening of the fiber core is itself a known failure mode there [72].

Continuous monitoring stations. Strongest fit. Near or above the 4 vol% LEL, TC sensors offer no O₂ requirement, immunity to H₂S/silicone/lead poisoning, minimal drift (<0.2%/month best-in-class [24]), effectively unlimited service life, and mature certified (ATEX/IECEx-class) hardware from multiple vendors, Eaton/MTL, ABB, Baker Hughes/Panametrics [22][23][24]. Acoustic/ultrasonic composition sensing is likewise mature for hydrogen-blended natural-gas pipeline metering and hydrogen-cooled generator purity monitoring, with commercial ultrasonic flow meters certified for custody-transfer-grade H₂ measurement up to 30% blends [39][40].

See why H₂ is hard for why ppm-level H₂ sensing is hard in general, and electrochemical cells for the alternative that fills the soil-gas reconnaissance role.

Vendors & products

Vendor	Product / line	Technology	Key specs found	Price class	Source
Xensor Integration (NL)	XEN-TCG3880 (die) / XEN-5310 (RS-232/analog readout) / XEN-5320 (USB/WiFi readout)	MEMS SiN-membrane thermal conductivity gauge	Time constant 9 ms (air); 1–3% FS inaccuracy; −40 to +85 °C; ~1200 ppm/yr drift; 5–8 V, ~20 mA	Die ~$10–20 (distributor listings) [52]; module pricing not independently verified	[25][26][52]
Sensirion (CH)	STC31 / STC31-C	CMOS-integrated thermal conductivity, binary CO₂-in-air or CO₂-in-N₂ (designed and marketed for CO₂, not H₂, quantification; included because it is the same physical sensing principle and is sometimes proposed for H₂ use by analogy)	0–100 vol% range; accuracy ±(0.5 vol%+3% MV); response <2 s; 2.7–5.5 V, 50 µA avg / 5 mA peak; I²C	$24–65 (Digikey, qty-dependent) [53]	[27][28][53]
Posifa Technologies (US)	PGS1000 / PGS1004 / PGS4100	MEMS thermal-conductivity H₂ sensor w/ onboard RH + barometric compensation	PGS4100: 0–4 vol% range, 10 ppm (analog)/2 ppm (digital) resolution, ~1200 ppm accuracy 0–20,000 ppm, 1 s response, −40 to 85/90 °C, long-term stability 1200 ppm/5 yr [47]; PGS1004: ±6% accuracy, 8 mA, −25 to 85 °C, ~$35–45 [50][54]	$35–45 (small qty, PGS1004) [54]	[29][30][47][50][54]
SGX Sensortech / Neroxis (CH/UK)	VQ series (katharometer beads), TC-1326/TC-1326-A MEMS, Neroxis MTCS-2601	Matched-element katharometer beads (VQ) and MEMS Ni-Pt four-resistor Pirani/TC die (TC-1326, MTCS-2601)	VQ: 0–100 vol% H₂/He/CH₄ vs. sealed reference [74]; MTCS-2601: Pirani range 10⁻⁴–1000 mbar, <6 mW [32]	Not independently verified	[31][32][74]
Eaton / MTL (Crouse-Hinds)	K522 katharometer / K1550 gas analyser	Classical hot-wire katharometer, DIN-rail electronics + remote flameproof sensor head	0–100% range (programmable), 4–20 mA + RS-232 output	Process-analyzer class; not independently verified	[22][75]
ABB	Katharometer Series 6517	Classical katharometer for continuous industrial use	Title/series confirmed; detailed specs not extracted	Not verified	[76]
Baker Hughes / Panametrics	XMTCpro	Online thermal-conductivity transmitter for binary mixtures (H₂, CO₂, CH₄, He, etc.)	SIL-certified, contamination-resistant, no moving parts	Industrial process-analyzer class; not independently verified	[23][77]
E/One Utility Systems (GE lineage)	Generator Gas Analyzer (GGA)	Triple-range online TCD for H₂-cooled turbogenerators	70–100% H₂ in air (±0.5% FS); 0–100% H₂/air in CO₂ (±1.0% FS); linearity ±1.0% FS; drift <0.2%/month; 4–20 mA + relay outputs; 500 cc/min flow	Industrial capital-equipment class; not independently verified	[24]
NevadaNano	MPS (Molecular Property Spectrometer) Flammable Gas Sensor	Multi-property thermal MEMS sensor (extends TC concept), 0–100% LEL, 14 gases incl. H₂	No field recalibration claimed; immune to drift/poisoning per vendor	Not independently verified	[38]
GC vendors (Agilent et al.)	Micro-GC with TCD, mixed-carrier-gas H₂ method	GC-TCD, chromatographic separation ahead of TC detection	Mixed-carrier-gas method resolves H₂/He crossover artifact (above)	Lab/wellsite capital-instrument class	[21][59][62]
Research (Zhang et al., SIMIT/CAS)	Differential P⁺/N⁺ single-crystalline Si thermopile pair	Differential MEMS thermopile with Pt@Al₂O₃ catalyst	1 ppm LOD; 1 ppm–2% linear range; response 1.9 s / recovery 1.4 s; ~40 mW/thermopile; 120 °C optimum	Research prototype (not yet commercial)	[33][34]
Research (3ω tungsten-microwire group)	Helix-wound 8 µm tungsten wire 3ω sensor	Transient AC thermal conductivity + diffusivity	0.062% (620 ppm) H₂, 0.62% CO₂ LOD at 1 Hz; response 28 s / recovery 23 s	Research prototype	[9][10][35]
Research (Garrett, Penn State)	Resonant-tube sonic gas analyzer	Acoustic resonance / mean molecular weight	~30 ppm concentration resolution; <350 ppm system stability; 6 measurements/min	Research/demonstration device	[14][19][49]
Industrial ultrasonic flow/purity	Endress+Hauser FLOWSIC610/Prosonic Flow G300; Emerson; general ultrasonic flow meters	ToF ultrasonic, optional H₂ purity/composition output	Certified for custody-transfer H₂ measurement up to 30% blends; ≤±0.2–0.5% accuracy classes	Industrial flow-meter capital-cost class	[39][40]

Open questions

• No downhole-viable architecture exists. No literature describes a passive or minimal-electronics TC/acoustic hydrogen sensor rated for HTHP borehole deployment. Worth scoping: a fiber-optic-coupled remote thermal-conductivity cell (heater/readout at surface, only a passive optical or capillary-sampling element downhole), given fiber hydrogen-darkening constraints [72].
• The CO₂/CH₄/He ppm-equivalent derivation needs vendor validation. The cross-sensitivity numbers above are first-order linear estimates from cited pure-gas thermal conductivities, not manufacturer-published interference tables. Requesting Sensirion/Posifa interference/error-budget tables directly would let those estimates be replaced with vendor-verified ones [28].
• H₂S thermal conductivity at 300 K is unverified. The value used above (~0.0146 W/m·K) was only confirmed near 100 K in the source compilation retrieved; a verified 300 K value is still needed [6].
• The field “gold standard” (GA5000) is not TC-based; it uses electrochemical sensing for H₂ [66][67]. TC/acoustic sensors should be positioned internally as a complementary confirmatory/safety layer, not a replacement, for soil-gas reconnaissance.
• Pricing gaps for process-analyzer-class instruments. MTL K522, ABB 6517, Panametrics XMTCpro, and E/One GGA have no independently verified public pricing; direct vendor quotes are needed for realistic capital-cost benchmarking against electrochemical/optical alternatives.
• 3ω and SAW-hybrid devices are not yet purchasable. Worth tracking the 2026 SAW/thermal-conductivity hybrid sensor [48] and the 3ω tungsten-microwire design [9] for commercialization; either reaching TRL 7+ would materially close the ppm-level gap that is this modality’s core weakness.
• The Wassiljewa-based multicomponent error bound (3–8%) needs validation against a real 4+ component soil-gas standard (H₂/CH₄/CO₂/N₂/H₂O) in a controlled lab test before trusting any TC sensor’s readout on real multi-gas field samples.
• Confirm Hestia’s contracted mud-logging vendors’ carrier-gas practice. Whether their GC-TCD systems use He or N₂/Ar carrier gas matters given the crossover artifact described above, an operationally actionable, cheap-to-fix item if not already addressed.

Related pages

• Why H₂ is hard to sense
• Catalytic pellistors
• Work-function devices
• Palladium-resistive sensors
• MOX sensors
• Electrochemical cells
• Fiber-optic sensors
• Geologic context

References

Full reference list (77 entries)

1. Wikipedia, “List of thermal conductivities,” https://en.wikipedia.org/wiki/List_of_thermal_conductivities (accessed 2026-07-04).

2. General synthesis from thermal-conductivity and speed-of-sound values cited throughout this dossier (H₂ and He as shared outliers on both axes); see [1], [3], [11], [12].

3. Engineering-oriented thermal conductivity discussion, ScienceDirect Topics, “Thermal Conductivity of Gases — an overview,” https://www.sciencedirect.com/topics/chemical-engineering/thermal-conductivity-of-gases; general Eucken-relation context.

4. “Micromachined Thermal Gas Sensors — A Review,” PMC9860855, https://pmc.ncbi.nlm.nih.gov/articles/PMC9860855/ (accessed 2026-07-04).

5. Search synthesis on helium gas thermal conductivity (~0.142–0.151 W/m·K near 300 K, NSRDS-NBS-8 table referenced), via WebSearch of “helium gas thermal conductivity 0.151 W/m·K 300K value table,” citing Electronics Cooling (“The thermal conductivity of gases,” https://www.electronics-cooling.com/1998/09/the-thermal-conductivity-of-gases/) and NIST Huber & Harvey compilation (https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=907540).

6. Search snippet, NIST Huber & Harvey thermal conductivity compilation, value for H₂S ~14.6 mW/m·K at 100 K noted via search synthesis; marked unverified at 300 K in this dossier. https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=907540.

7. Taylor & Francis, “TCD – Knowledge and References,” https://taylorandfrancis.com/knowledge/Engineering_and_technology/Chemical_engineering/TCD/ (accessed 2026-07-04).

8. Wikipedia, “Thermal conductivity detector,” https://en.wikipedia.org/wiki/Thermal_conductivity_detector (accessed 2026-07-04).

9. “A Robust Miniaturized Gas Sensor for H2 and CO2 Detection Based on the 3ω Method,” PMC8780541, https://pmc.ncbi.nlm.nih.gov/articles/PMC8780541/ (accessed 2026-07-04); also MDPI Sensors 22(2):485, https://www.mdpi.com/1424-8220/22/2/485. (Note: does not support the 60 °C/two-orders-of-magnitude temperature-interference figure cited alongside it in the Environmental covariance & cross-sensitivity section.)

10. “Minimum Detection Concentration of Hydrogen in Air Depending on Substrate Type and Design of the 3ω Sensor,” PMC10648048, https://pmc.ncbi.nlm.nih.gov/articles/PMC10648048/ (accessed 2026-07-04).

11. Wikipedia, “Speed of sound,” https://en.wikipedia.org/wiki/Speed_of_sound (accessed via search synthesis, 2026-07-04).

12. Search synthesis on speed of sound in hydrogen at room temperature (~1270–1320 m/s range, T-dependent), via WebSearch “speed of sound hydrogen 1284 OR 1310 OR 1320 m/s 20C room temperature value.”

13. “Experimental speed-of-sound data and a fundamental equation of state for normal hydrogen optimized for flow measurements,” ScienceDirect, https://www.sciencedirect.com/science/article/pii/S0021961424000946 (title/abstract-level; full text not retrievable in this session — 403).

14. S. Garrett, “Sonic gas analyzer for hydrogen and methane,” Acoustics’08 Paris proceedings paper 000614, http://www.conforg.fr/acoustics2008/cdrom/data/articles/000614.pdf (fetch unsuccessful — 404; content sourced via Acoustical Society of America press summary, see [19]); also indexed at J. Acoust. Soc. Am. 123, 3372 (2008), https://pubs.aip.org/asa/jasa/article/123/5_Supplement/3372/715641/Sonic-gas-analyzer-for-hydrogen-and-methane and ResearchGate https://www.researchgate.net/publication/5326551_Sonic_gas_analyzer_for_hydrogen_and_methane.

15. Search synthesis on acoustic resonator/binary gas mixture composition principles, via WebSearch “acoustic resonator gas analyzer hydrogen binary mixture composition MEMS ultrasonic gas sensor.”

16. Academia.edu, “An ultrasonic sensor for the analysis of binary gas mixtures,” https://www.academia.edu/48172175/An_ultrasonic_sensor_for_the_analysis_of_binary_gas_mixtures (title-level reference).

17. Nature Scientific Reports, “Gas discrimination by simultaneous sound velocity and attenuation measurements using uncoated capacitive micromachined ultrasonic transducers,” https://www.nature.com/articles/s41598-021-04689-4 (title/topic confirmed via search synthesis; full text behind login redirect in this session).

18. arXiv:1210.4835, “A combined ultrasonic flow meter and binary vapour mixture analyzer for the ATLAS silicon tracker,” https://arxiv.org/pdf/1210.4835 (title-level; PDF text not extractable in this session).

19. Acoustical Society of America, press summary of Garrett sonic gas analyzer, https://acoustics.org/pressroom/httpdocs/135th/golden.htm (fetched 2026-07-04); resonance frequency stability ~0.02 Hz / ~30 ppm resolution and <350 ppm system stability figures via search synthesis of the same work (ScienceDaily, https://www.sciencedaily.com/releases/2005/10/051024084141.htm; phys.org, https://phys.org/news/2005-10-pitch-sonic-gas.html).

20. “A MEMS Pirani Vacuum Gauge Based on Porous Silicon,” MDPI Micromachines / PMC11946229, https://pmc.ncbi.nlm.nih.gov/articles/PMC11946229/ (title/topic-level via search synthesis).

21. Wikipedia, “Thermal conductivity detector,” applications section, https://en.wikipedia.org/wiki/Thermal_conductivity_detector (see also [8]).

22. Eaton / MTL, “K522 Thermal Conductivity (TCD) Gas Sensor for Hydrogen (H2), Helium (He), Nitrogen and Argon,” https://www.mtl-inst.com/product/k522_-_thermal_conductivity_tcd_gas_sensor_for_hydrogen_h2_helium_he_argona/ and https://www.eaton.com/us/en-us/catalog/intrinsic-safety/mtl-k522-thermal-conductivity-tcd-gas-sensor-for-hydrogen-helium.html (accessed 2026-07-04).

23. Baker Hughes / Panametrics, “XMTCpro,” https://www.bakerhughes.com/panametrics/process-analyzers/industrial-oxygen-gas-analyzers/xmtcpro (accessed 2026-07-04).

24. E/One Utility Systems, “Hydrogen Gas Purity Analyzer (Generator Gas Analyzer, GGA),” https://eone.com/hydrogen-systems/products/generator-gas-analyzer-gga (fetched 2026-07-04).

25. Xensor Integration, “Thermal Conductivity Gauge XEN-TCG3880” datasheet, https://www.isweek.com/Uploads/20180604/5b14e8a49967c.pdf (title/summary via search synthesis; PDF text not machine-readable in this session).

26. Xensor Integration, XEN-5310/XEN-5320 datasheets, https://xensor.nl/images/pdf-files/XEN-5320-HP-USB.pdf, https://xensor.nl/images/pdf-files/XEN-5320/Thermal_Conductivity_Gas_Sensor_XEN-5320-ALU-USB.pdf, and https://xensor.nl/products/thermal-conductivity-gauge (specs via search synthesis of these documents, including 9 ms/36 ms time constants, 1–3% FS accuracy, −40 to +85 °C, ~1200 ppm/yr drift, 5–8 V/~20 mA).

27. Sensirion, “Datasheet STC31 - CO2 Sensor Based on Thermal Conductivity Measurement Principle,” https://sensirion.com/media/documents/7B1D0EA7/61652CD0/Sensirion_Thermal_Conductivity_Datasheet_STC31_D1_1.pdf; product page https://sensirion.com/products/catalog/STC31 (fetched 2026-07-04).

28. Sensirion, STC31-C product/design-guide references, https://sensirion.com/products/catalog/STC31-C and https://sensirion.com/media/documents/45B45DA4/666C0401/Sensirion_Thermal_Conductivity_AN_STC31-C_Design_Guide.pdf (low-cross-sensitivity mode, RH/O₂ compensation, per search synthesis).

29. Posifa Technologies, “PGS1000 Hydrogen Sensor,” https://posifatech.com/pgs1000/ (accessed 2026-07-04).

30. Posifa Technologies, “PGS4100 MEMS Hydrogen Sensor,” https://posifatech.com/pgs4100/ (fetched 2026-07-04).

31. SGX Sensortech, “TC-1326-A/TC-1326-AS MEMS Thermal Conductivity Sensor Elements,” https://www.sgxsensortech.com/content/uploads/2020/09/SPF-2210-Data-Sheet-TC1326_V4.0.pdf (title-level; PDF text not machine-readable in this session).

32. Neroxis, “MTCS 2601 — Gas — Micro Thermal Conductivity Sensor” brochure, https://www.neroxis.ch/wp-content/uploads/2020/04/DPM-18-098-002-E-MTCS-2601-Thermal-Conductivity-Sensor-for-Gas_Brochure_EN.pdf (specs via search synthesis: Pirani range 10⁻⁴–1000 mbar, <6 mW).

33. H. Zhang et al., “1ppm-detectable hydrogen gas sensors by using highly sensitive P+/N+ single-crystalline silicon thermopiles,” Microsystems & Nanoengineering (2023), https://www.nature.com/articles/s41378-023-00506-2.

34. Same paper, open-access PMC version, https://pmc.ncbi.nlm.nih.gov/articles/PMC10027663/ (fetched 2026-07-04).

35. “Minimum Detection Concentration of Hydrogen in Air Depending on Substrate Type and Design of the 3ω Sensor,” PMC10648048 (see [10]) — substrate comparison data (SiO₂ ~18%, polyimide 3–4%, suspended 0.3–0.35%, theoretical floor ~1400 ppm).

36. “A MEMS Pirani Vacuum Gauge Based on Porous Silicon,” MDPI Micromachines 16(3):296, https://www.mdpi.com/2072-666X/16/3/296 (title-level).

37. “An Ultra-Compact MEMS Pirani Sensor for In-Situ Pressure Distribution Monitoring,” PMC9611872, https://pmc.ncbi.nlm.nih.gov/articles/PMC9611872/ (title-level).

38. NevadaNano, “MPS Flammable Gas Sensor,” https://nevadanano.com/products/mps-flammable-gas-sensor/ and https://nevadanano.com/technology/molecular-property-spectrometer/ (accessed 2026-07-04, via search synthesis).

39. Emerson, “Hydrogen Measurement with Ultrasonic Flow Meters,” https://www.emerson.com/en-us/automation/measurement-instrumentation/common-applications/hydrogen-measurement-with-ultrasonic-flow-meters (title-level via search).

40. Endress+Hauser, “Hydrogen-enriched gas measurement with ultrasonic flowmeters,” https://www.endress.com/en/support-overview/learning-center/ultrasonic-flow-measurement-gas-hydrogen; FLOWSIC610 product info via https://www.us.endress.com/en/field-instruments-overview/flow-measurement-product-overview/ultrasonic-flowmeters (accessed 2026-07-04, via search synthesis).

41. Wikipedia / HandWiki mirror, “Hydrogen-cooled turbo generator,” https://en.wikipedia.org/wiki/Hydrogen-cooled_turbo_generator and https://handwiki.org/wiki/Physics:Hydrogen-cooled_turbo_generator (λ_H2 ≈ 0.168 W/(m·K) cited in this source, slightly different from the 0.187 brief value, reflecting normal cross-source spread — accessed via search synthesis 2026-07-04).

42. IOPscience, “Discrimination of Binary Gas Mixture Using CMUT Based Sound Attenuation Spectrum Gas Sensor,” https://iopscience.iop.org/article/10.1149/MA2021-01591592mtgabs (title-level).

43. “A review of piezoelectric MEMS sensors and actuators for gas detection application,” PMC9971541, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9971541/ (title-level, CMUT/FBAR context).

44. Agilent, “Hydrogen Detection with a TCD using Mixed Carrier Gas on the Agilent Micro GC,” application note 5991-3199EN, https://www.agilent.com/cs/library/applications/5991-3199EN.pdf (title/topic via search synthesis).

45. “Thermal Conductivity Detector Analysis of Hydrogen Using Helium Carrier Gas and HayeSep® D Columns,” Journal of Chromatographic Science 36(4):191, https://academic.oup.com/chromsci/article-pdf/36/4/191/742027/36-4-191.pdf (title-level via search).

46. “The evolution of the hydrogen signal during GC-TCD analysis with helium carrier gas and PLOT columns,” International Journal of Hydrogen Energy, https://www.sciencedirect.com/science/article/abs/pii/S0360319925058409 (title-level via search).

47. Posifa Technologies, PGS4100 datasheet specifications via search synthesis (resolution 10 ppm analog/2 ppm digital, range 0–4 vol%, accuracy 1200 ppm 0–20,000 ppm and 6% above, temperature drift 1000 ppm, long-term stability 1200 ppm/5 yr, response 1 s, warm-up 1 s, op. temp −40 to 85 °C), https://posifatech.com/wp-content/uploads/2025/08/Datasheet_PGS4100_MEMS_TC_H2_RevA_C1.7.pdf (link updated 2026-07-04; original 2023/03 path returned 404) and https://posifatech.com/pgs4100/ (accessed 2026-07-04).

48. “High sensitivity SAW hydrogen gas sensor based on thermal conductivity effect,” Microsystems & Nanoengineering (2026), https://www.nature.com/articles/s41378-026-01199-z (title/abstract via search synthesis).

49. Search synthesis of Garrett sonic gas analyzer accuracy figures (“accuracy of better than +/-30 ppm” for hydrogen), via WebSearch “Sonic gas analyzer hydrogen methane ResearchGate speed of sound sensor,” citing http://www.conforg.fr/acoustics2008/cdrom/data/articles/000614.pdf title record.

50. DigiKey Marketplace, Posifa PGS1000/PGS1004 listing and specifications, https://www.digikey.com/en/products/detail/posifa-technologies/PGS1000/15761969 (fetched 2026-07-04): I2C interface, −25 to 85 °C, 4.75–5.25 V, 8 mA, ±6% accuracy, $35–45 (1–100 unit price breaks).

51. Nano Environmental Technology S.r.l., “Katharometer gas technology,” https://www.nenvitech.com/technologies/katharometer-gas-technology/ (accessed via search synthesis, poisoning-immunity claim).

52. Alibaba distributor listing, XEN-TCG3880, price range referenced via search synthesis (US$10–20/unit, MOQ 2), https://www.alibaba.com/product-detail/Thermal-Conductivity-Sensor-XEN-TCG3880_1601000219005.html.

53. DigiKey, STC31/STC31-C product listings, https://www.digikey.com/en/products/detail/sensirion-ag/STC31-C-R3/24394079 and https://www.digikey.com/en/products/detail/sensirion-ag/STC31-C-R5/24394081 (prices via search synthesis: STC31-C-R3 $33.41, STC31-C-R5 $23.91, STC31-R3 $64.25).

54. DigiKey Marketplace, PGS1004 pricing (see [50]).

55. “Exploring natural hydrogen hotspots: a review and soil-gas survey design for identifying seepage,” Geoenergy / Lyell Collection, https://www.lyellcollection.org/doi/full/10.1144/geoenergy2023-014 (title-level; full text 403 in this session; multi-gas co-occurrence context via ResearchGate summary https://www.researchgate.net/publication/372872189_Exploring_Natural_Hydrogen_Hotspots_A_Review_and_Soil-Gas_Survey_Design_for_Identifying_Seepage).

56. Search synthesis on soil-gas H₂ survey variability (“point surveys show significant variability… ranging from 0 ppm to >1,000 ppm”; multi-gas interpretation need), via WebSearch “geologic hydrogen exploration soil gas survey sensor thermal conductivity limitations ppm background,” referencing [55] and USGS “Prospectivity mapping for geologic hydrogen,” https://pubs.usgs.gov/publication/pp1900/full.

57. Search synthesis, Wassiljewa equation performance for binary vs. multicomponent mixtures, via WebSearch “Wassiljewa equation thermal conductivity gas mixture binary assumption error,” citing ADS 1965JChPh..42.2533B, https://ui.adsabs.harvard.edu/abs/1965JChPh..42.2533B/abstract.

58. IDAES documentation, “Wassiljewa-Mason-Saxena Method for Low-pressure Gas Mixture Thermal Conductivity,” https://idaes-pse.readthedocs.io/en/2.5.0/explanations/components/property_package/general/transport_properties/thermal_conductivity_wms.html (accessed via search synthesis).

59. Search synthesis on GC-TCD H₂/He carrier-gas crossover, via WebSearch “GC-TCD hydrogen helium carrier gas blind spot thermal conductivity crossover problem,” citing academic.oup.com/chromsci [45] and ResearchGate discussion threads (e.g., https://www.researchgate.net/post/How-to-analyse-Hydrogen-Gas-from-a-GC-with-TCD-using-Helium-as-the-carrier-gas-with-a-capillary-column).

60. Wikipedia, “Thermal conductivity detector,” hydrogen-detection/negative-peak note (see [8]).

61. Lotus Instruments, “Operating Parameters for the Thermal Conductivity Detector,” https://lotusinstruments.com/wp/wp-content/uploads/Operating-Parameters-for-TCD.pdf (title-level via search).

62. Agilent application note 5991-3199EN (see [44]) — 8.5% H₂-in-He carrier-gas mitigation method, via search synthesis.

63. University of Toronto Scarborough TraceS Lab, “Thermal Conductivity Detector (TCD),” https://www.utsc.utoronto.ca/~traceslab/PDFs/GC%20TCD.pdf (title-level via search).

64. “First detection of industrial hydrogen emissions using high precision mobile measurements in ambient air,” Scientific Reports / PMC11480439, https://pmc.ncbi.nlm.nih.gov/articles/PMC11480439/ (fetched 2026-07-04): background H₂ mole fraction ~0.5 ppm / ~530 ppb, GC-PDHID precision <2 ppb.

65. Search synthesis on NOAA/AGAGE atmospheric H₂ background (~552.8 ppb MBL mean, 2021), via WebSearch “atmospheric hydrogen background 0.53 ppm NOAA AGAGE mole fraction,” referencing NOAA Global Greenhouse Gas Reference Network context.

66. Search synthesis, Geotech GA5000 sensor technology (“uses electrochemical sensors, specifically for measuring H2 gas concentrations”), via WebSearch “GA5000 gas analyzer hydrogen sensor technology electrochemical thermal conductivity specifications,” citing https://www.chromservis.eu/en/ga5000-analyser-for-measurement-of-ch4-co2-o2-h2-1000-ppm-bluetooth and https://www.envirotecnics.com/en/product/ga5000-gas-analyzer/.

67. “Sensing Hydrogen Seeps in the Subsurface for Natural Hydrogen Exploration,” MDPI Applied Sciences 12(13):6383, https://mdpi.com/2076-3417/12/13/6383/htm (title-level; full text 403 in this session); selectivity (K > 1000) and stability (<1 ppm/day) criteria and GA5000 field methodology via search synthesis, also summarized at ResearchGate https://www.researchgate.net/publication/363006767_Sensing_Hydrogen_Seeps_in_the_Subsurface_for_Natural_Hydrogen_Exploration.

68. Search synthesis on GA5000 accuracy specifications (±0.5% CH4/CO2, ±1% O2 after calibration), same sources as [66].

69. “Advanced Mud Gas Logging Techniques for Hydrogen Exploration,” SPE Middle East Oil and Gas Show and Conference, OnePetro, https://onepetro.org/SPEMEOS/proceedings-abstract/25MEOS/25MEOS/D011S018R002/790177 (title-level; abstract 403 in this session); “Key Role of Regearing Mud Gas Logging For Natural H2 Exploration,” SPWLA Annual Logging Symposium, https://onepetro.org/SPWLAALS/proceedings-abstract/SPWLA22/3-SPWLA22/D031S001R003/487788 (title-level).

70. “Geologic hydrogen: a review of resource potential, subsurface dynamics, exploration, production, transportation, and research opportunities,” Energy & Environmental Science, RSC Publishing, DOI:10.1039/D5EE02910D, https://pubs.rsc.org/en/content/articlehtml/2025/ee/d5ee02910d (title-level; full text 403 in this session; downhole electronics temperature-limit claim, ~175 °C Si/SiC vs. ~400 °C target, via search synthesis of this source).

71. “Sensing Electronics for Downhole Drilling in Extreme Environments,” labpartnering.org, https://labpartnering.org/technologies/3f72efda-2142-450a-aa56-547d82b5ed4b (title-level via search synthesis).

72. Search synthesis on fiber-optic downhole sensing and hydrogen darkening, via WebSearch “downhole tool no electronics fiber optic sensor high temperature hydrogen exploration well logging preference,” citing Fibercore https://fibercore.humaneticsgroup.com/perspectives/fiber-optic-sensing-downhole-monitoring-oil-gas and Laser Focus World https://www.laserfocusworld.com/fiber-optics/article/16551514/fiber-for-remote-sensing-downhole-sensing-applications-enhanced-by-specialty-optical-fibers.

73. “Thermal Management of Downhole Oil and Gas Logging Sensors for HTHP Applications Using Nanoporous Materials,” ASME ENIC2007, https://asmedigitalcollection.asme.org/ENIC/proceedings-abstract/ENIC2007/47993/27/327004 (title-level via search).

74. SGX Sensortech, VQ series thermal conductivity sensor family, https://sgx.cdistore.com/products/detail/vq31mb-sgx-sensortech/337610/ and related VQ-series distributor listings (title-level via search synthesis).

75. Eaton/MTL, K1550 hydrogen/helium/argon/xenon gas analyser and general gas analyser brochure, https://www.eaton.com/gb/en-gb/catalog/intrinsic-safety/mtl-k1550-hydrogen-helium-argon-and-xenon-gas-analyser.html and https://www.mtl-inst.com/images/uploads/Eaton_Gas_Analysers_Brochure_010916.pdf (title-level).

76. ABB, “Katharometers Series 6517,” https://library.e.abb.com/public/d6a48d9b3db3ac48c1257b0c0053e5e5/SS_6517_4.pdf (title/series confirmed; PDF text not machine-readable in this session).

77. Baker Hughes, “Hydrogen purity in hydrogen-cooled generators,” https://www.bakerhughes.com/panametrics/success-stories/hydrogen-purity-hydrogencooled-generators (accessed via search synthesis 2026-07-04).