Exotic optical approaches: LIBS, interferometry, flame UV & imaging

Summary

This page is a grab-bag on purpose: six physically unrelated optical (and one acoustic) techniques that keep coming up in vendor conversations and safety-engineering discussions but don’t fit cleanly into any other page in this compendium. They split into two very different buckets, and mixing them up is a common mistake.

The first bucket is bulk-property physics that cannot tell gases apart: laser-induced breakdown spectroscopy (LIBS) zaps a sample with a laser and reads the atomic hydrogen line in the resulting spark — but that line lights up for any hydrogen atom, whether it came from H₂, water vapor, a hydrocarbon, or a wet rock, so it cannot answer “is this H₂?” on its own.

Refractive-index and dielectric-constant (“how much a gas bends light” and “how much it resists an electric field”) sensing have the same problem one level down: they measure a bulk mixture property that shifts with humidity, methane, helium, temperature, and pressure just as much as it shifts with hydrogen.

The second bucket is mature, catalog-buyable safety and visualization instruments: cameras that make invisible gas jets visible by their effect on light (schlieren imaging) or by the ultrasonic hiss a pressurized leak makes (acoustic imaging), and UV flame detectors that catch a hydrogen fire — which burns nearly invisibly to the naked eye — by its faint ultraviolet glow.

These are genuinely useful, field-proven products, but they are wellsite safety tools for pressurized infrastructure, not ppm-sensitive exploration sensors for a diffuse soil-gas seep. For a non-sensing colleague: everything on this page is real physics and, in several cases, a purchasable product, but none of it is a substitute for the electrochemical, catalytic, metal-oxide, or laser-absorption sensors that should carry Hestia’s actual H₂ concentration measurements.

At a glance — published performance envelope

Detection limit

≥ 20 ppm

Range

≤ 1,000,000 ppm

Response t₉₀

0.001–60 s

Recovery

0.001–60 s

Selectivity

none

Drift

medium

Operating T

−55 … 200 °C

Humidity

0–100 % RH

Power

500 mW – 500 W

Unit cost

$1k–$150k

TRL

9

Main interferents: H₂O, CH4, CO2, He, H2S, N2, O2: all have distinct but broadly similar-order molar refractivity, Rock, Wind, thermal plumes, sun-warmed surfaces: produce refractive-index gradients indistinguishable from a real gas-composition gradient in schlieren · poisons: . Ranges span the published devices/variants covered below — per-claim citations are in the body; data: src/data/modalities/exotic-optical.yaml.

Mechanism & physics

Laser-induced breakdown spectroscopy (LIBS): plasma emission, not absorption

LIBS focuses a pulsed laser (typically Q-switched Nd:YAG, ~1064 nm, tens to hundreds of mJ, nanosecond pulse width) onto a solid, liquid, or gas target. The focal-volume irradiance (commonly >1 GW/cm²) exceeds the optical breakdown threshold of the target/ambient gas, producing a micro-plasma with initial electron temperatures of several eV and electron densities in the 10¹⁶–10¹⁸ cm⁻³ range.

As the plasma cools over the following microseconds, ions and atoms recombine and de-excite, emitting a spectrum of sharp atomic/ionic lines on top of a decaying bremsstrahlung/recombination continuum — this is atomic emission spectroscopy of a laser-ablated micro-plasma, fundamentally different in mechanism from every absorption-based (TDLAS/CRDS) or Raman-based technique covered elsewhere in this compendium [23][24].

For hydrogen, the diagnostic line is the Balmer-α (Hα) transition, n=3→n=2, at 656.28 nm — a fully dipole-allowed atomic transition, not a molecular one. LIBS never “sees” H₂ the molecule: the plasma’s extreme temperature dissociates and ionizes every hydrogen-bearing species in the ablated/sampled volume (H₂O, OH, hydrocarbons, hydroxylated minerals, and any free H₂) into atomic H, and the Hα emission intensity tracks the total elemental hydrogen population in the plasma, with no memory of which molecule that hydrogen came from [21][22][25].

This point applies throughout this section: LIBS senses total hydrogen, not H₂ selectively, precisely analogous to how a combustion-based sensor can’t distinguish CH₄ from H₂ by heat alone (see catalytic pellistors).

Because Hα also carries the largest Stark broadening coefficient of any commonly used LIBS diagnostic line, it doubles as the standard plasma electron-density diagnostic: the Stark-broadened linewidth of Hα (or Hβ, 486.1 nm) is related to electron density $n_e$ by

$\Delta\lambda_{1/2} \approx 2\,\alpha_{1/2}(n_e, T_e)\left(\frac{n_e}{10^{16}\,\mathrm{cm^{-3}}}\right)^{2/3}$

(the numerical Stark-width parameter $\alpha_{1/2}$ is tabulated per Balmer line and depends weakly on $T_e$), and Hα is routinely used to validate the local-thermodynamic-equilibrium (LTE) assumption underlying calibration-free LIBS (CF-LIBS) quantification schemes [26][27].

At high hydrogen concentration the same line suffers self-absorption/self-reversal — cooler, less-excited outer plasma re-absorbs photons emitted from the hotter core, flattening or dipping the line center and biasing naive intensity-based quantification low, an effect that worsens as the hydrogen fraction rises and remains a standing challenge for isotopic (H/D/T) LIBS work resolving fine structure on top of this same broadened line [21][22][28].

Refractive index and interferometry: Lorentz–Lorenz and the phase-shift equation

A gas’s refractive index departs from unity because its constituent molecules polarize in the optical field; the classical Lorentz–Lorenz relation (the optical-frequency form of the Clausius–Mossotti relation) connects the two:

$\frac{n^2-1}{n^2+2} = \frac{N\alpha}{3\varepsilon_0}$

which for a dilute gas ($n\approx 1$) reduces to the familiar linear form $n - 1 \approx \dfrac{N\alpha}{2\varepsilon_0}$ — refractivity is directly proportional to number density $N$ and molecular polarizability $\alpha$. H₂’s low polarizability (established elsewhere in this compendium as roughly a third of N₂’s — see why H₂ is hard to sense) gives it one of the lowest refractive indices of any common gas: the standard textbook figures, at 0 °C, 1 atm, sodium D line (589.29 nm), are

$n(\mathrm{H_2}) = 1.000132, \qquad n(\mathrm{air}) = 1.000293$

[8], consistent with the classic Peck & Huang metrological dispersion formula for hydrogen, fitted to interferometric measurements at eighteen wavelengths from 0.4047 to 1.6945 µm: $10^6(n-1) = 21.113 + 12723.2/(111-\sigma^2)$ with $\sigma$ the wavenumber in µm⁻¹ [7].

(Evaluating this formula at the sodium D line gives $n-1 \approx 1.39\times10^{-4}$, close to but not numerically identical to the textbook figure — a reminder that “the” refractive index of a gas is itself a function of the exact temperature/pressure/wavelength convention used, which matters when a spec sheet quotes RIU resolution without stating equivalent gas-concentration resolution.)

A two-beam interferometer converts this tiny index difference into an observable optical phase shift over pathlength $L$:

$\Delta\Phi = \frac{2\pi L}{\lambda}\,\Delta n$

For a classic mine gas interferometer with a ~10 cm gas cell and visible light, replacing air with pure H₂ ($\Delta n \approx 1.6\times10^{-4}$) produces a phase shift of order tens of radians — many fringes, trivially visible by eye, which is why interferometric methanometers (and, by the same physics, hydrogen-sensitive designs) were viable with 1950s optics using nothing more than a hand-squeezed rubber bulb, a lamp, a beamsplitter, and an eyepiece graticule reading fringe position on a scale [6].

Modern instantiations replace the bulk optical bench with guided-wave photonics: fiber or on-chip Mach–Zehnder interferometers (MZI), capillary-based MZIs exciting core and cladding modes simultaneously, and whispering-gallery-mode microresonators/microring resonators, all reading out the same $\Delta n$ as a wavelength or phase shift with far better mechanical stability and far smaller sample volumes than a 1950s bulk interferometer [9][12][13].

Dielectric constant (permittivity) sensing: the same molecular handle, at lower frequency

Static/low-frequency relative permittivity $\varepsilon_r$ is governed by the same Clausius–Mossotti physics as optical refractive index (the two coincide, $n^2\approx\varepsilon_r$, whenever the polarization mechanism — here, electronic/molecular polarizability with no significant permanent-dipole or ionic contribution — is frequency-independent across the gap between DC/RF and optical frequencies, essentially true for a nonpolar gas like H₂).

A parallel-plate or coaxial-cylinder capacitor reads this out directly via $C = \varepsilon_0\varepsilon_r A/d$; precision metrology uses cross-capacitor geometries specifically because they are insensitive to small dimensional distortions, and NIST-grade measurements of eight common gases (He, Ar, N₂, O₂, CH₄, C₂H₆, C₃H₈, CO₂) with a four-ring toroidal cross capacitor illustrate how small — and how similar in order of magnitude across species — these permittivity departures from unity are [17].

A dedicated NIST/NPL-class measurement of hydrogen’s dielectric permittivity used a quasi-spherical microwave cavity resonator (2.4–7.3 GHz, up to 6.5 MPa, 273–293 K) to extract H₂’s molecular electric polarizability to metrology-grade precision, $\alpha_{\varepsilon,\mathrm{H_2}} = (8.9568\pm0.0008)\times10^{-41}\ \mathrm{F\,m^2}$ at 273 K [16] — a precise value for a physics reference table, and at the same time a demonstration of how much specialized apparatus (a temperature- and pressure-controlled microwave resonant cavity) is required to extract a gas’s identity from its bulk permittivity at all.

The selectivity problem is therefore identical in kind to refractive-index sensing, because it is the same physics: any gas-mixture change that alters mean molecular polarizability-times-density — humidity, CH₄, CO₂, He, pressure, temperature — moves the reading indistinguishably from an H₂ change.

Rayleigh-scattering deficit and dispersion: theoretical curiosities, not sensors

Rayleigh scattering cross-section scales as $\sigma_R \propto (2\pi/\lambda)^4\alpha^2$ — the square of molecular polarizability. Because H₂ is one of the least polarizable common molecules, air with a trace admixture of H₂ scatters very slightly less visible light than pure air — a “Rayleigh deficit.”

This exact physics is exploited at the opposite extreme of pathlength and optical depth in exoplanet atmospheric science, where a hydrogen-dominated exoplanet atmosphere produces a characteristic, shallow, blue-steepening “Rayleigh slope” in transmission spectroscopy, diagnostic of an H₂/He-dominated envelope with mean molecular weight $\mu_m \approx 2.3$–2.4. That measurement is only possible because the light path integrates through an entire planetary atmosphere (thousands of kilometers) during transit against a photon-noise-limited but otherwise very clean baseline.

Scaled down to a 0.5–2 vol% H₂ soil-gas anomaly over a few meters of open air, the corresponding scattering-intensity deficit is many orders of magnitude below the shot-noise floor of any deployable instrument, and would in any case be swamped by aerosol/particulate Mie scattering, which dwarfs molecular Rayleigh scattering under real field conditions.

A related idea — using multi-wavelength dispersion (different gases have subtly different Cauchy/Sellmeier coefficients, so $n(\lambda)$ curves differ in shape, not just magnitude) to identify a gas from a two-color interferometric measurement — is a real, demonstrated laboratory technique for characterizing gas dispersion itself [40], and is used practically in an entirely different context as dispersion interferometry for absolute electron-density measurement in fusion plasmas.

Applied to identifying trace H₂ in air, the difference in dispersion between H₂ and its most likely soil-gas confounders (CH₄, CO₂, H₂O vapor) is itself a small effect riding on top of an already-small primary refractive-index signal — a second-order handle on a first-order non-selectivity problem.

Both effects are included for completeness, as examples of physics that is measurable in principle but not usable in the field; this research found no evidence of either being pursued as a practical trace-H₂ sensing route.

Schlieren and background-oriented schlieren (BOS): imaging the density gradient, not the gas

Schlieren imaging exploits the same Gladstone–Dale relation that ties refractive index to gas density, $n - 1 = k\rho$, where $k$ is the (wavelength- and gas-composition-dependent) Gladstone–Dale coefficient — for air near visible wavelengths, $k \approx 2.3\times10^{-4}\ \mathrm{m^3/kg}$ (shock-tube measurement, high-T/P conditions rather than ambient near-surface conditions — see Open Questions).

A light ray traversing a refractive-index gradient bends toward higher density; classic (Z-type/Toepler) schlieren optics use a knife edge at a focal point to convert this deflection into a brightness modulation, while background-oriented schlieren (BOS) — the modern, camera-only variant — instead photographs a fixed speckle/dot background through the flow and cross-correlates the apparent pixel displacement between a reference (no-flow) image and a live image to recover the same path-integrated density-gradient field, with no laser, knife edge, or precision optical bench required [34][35][36][37][38].

Both techniques are fundamentally density-gradient imagers, not chemical-species sensors: a hot air plume and a cold, dense H₂-depleted-air jet edge can, in the right geometry, produce visually similar schlieren signatures, and neither can on its own distinguish a hydrogen leak from a helium leak, a hot-air thermal, or a compressed-air leak of any composition — species identity is inferred contextually (a known gas line, a known process), not measured [34][35].

Why thermal-IR cameras cannot see H₂ — and the tracer-gas/catalytic-window workarounds

As derived fully in why H₂ is hard to sense and applied to gas-imaging optics in IR absorption and proxies, an infrared camera detects photons absorbed or emitted through a molecule’s electric-dipole-allowed vibrational transitions; H₂’s homonuclear symmetry forbids this to leading order, so H₂ gas is, to an excellent approximation, transparent and featureless to every optical-gas-imaging (OGI) thermal camera on the market — the same physical fact that makes NDIR structurally inapplicable to H₂.

Two commercially deployed workarounds sidestep, rather than solve, the physics: (1) tracer-gas OGI, in which a small admixture (commonly 3–5%) of a strongly IR-active gas such as CO₂ is deliberately added to a hydrogen-filled system (e.g., generator-cooling H₂ at >95% purity), so the OGI camera images the co-flowing CO₂, not the H₂ itself, as a proxy for where the H₂ is also leaking [30];

and (2) catalytic/reactive IR imaging, a 2025 research demonstration in which a Pt-WO₃ catalytic window plate sits in front of an ordinary IR camera, so the camera images the heat released by catalytic H₂+O₂ combustion at the window, converting the imaging problem back into a calorimetric one (reported 17 s/24 s response/recovery time) rather than a true spectroscopic one [31].

Acoustic leak imaging: turbulent/choked-flow noise and beamforming

A pressurized gas escaping through an orifice into a lower-pressure ambient produces broadband turbulent and (for a sufficiently large pressure ratio) choked-jet acoustic noise that extends well into the ultrasonic band (tens of kHz and above), well above the frequency range of most industrial background noise [3][32]. This physics has nothing to do with the gas’s chemical identity — it depends on the pressure differential, orifice geometry, and flow regime, not on whether the escaping gas is H₂, N₂, compressed air, or steam.

Acoustic imaging cameras exploit this using a phased microphone array (Distran’s instruments use a 124-element ultrasound sensor array; Fluke’s ii900/ii910 use a comparable multi-element MEMS microphone array) to perform delay-and-sum or similar beamforming, localizing the ultrasound source and overlaying it on a co-registered optical image in real time [11][32][33].

Angular resolution scales, as with any phased array, roughly as $\theta \sim \lambda_{ac}/D$ for array aperture $D$ and acoustic wavelength $\lambda_{ac}$ — since ultrasonic wavelengths at 20–100 kHz in air are on the order of a few mm to 1.7 cm, a compact handheld array aperture (tens of cm) already gives useful angular resolution at multi-meter working distances, which is why these are handheld, not vehicle- or building-scale, instruments.

H₂ flame detection: OH* chemiluminescence and the solar-blind UV window

A hydrogen–air flame is nearly invisible in the visible band (it lacks the black-body-glowing soot particles that give hydrocarbon flames their characteristic yellow-orange color) but is not radiatively silent: excited-state hydroxyl radical, OH*, formed transiently in the reaction zone, relaxes via the $A^2\Sigma^+ \to X^2\Pi$ electronic transition, radiating a distinctive chemiluminescence band peaking near 306–310 nm (reported ranges 305–325 nm) [1][2].

This band sits inside a genuinely useful atmospheric coincidence: solar UV below ~300–305 nm is almost completely removed by stratospheric ozone absorption before it reaches ground level, so a detector responsive only to ~185–260 (or up to ~280–300) nm sees essentially zero background from the sun — the “solar-blind” window that makes UV flame detection practical outdoors without heavy optical filtering [2][3][4].

The workhorse solar-blind UV sensing element, exemplified by Hamamatsu’s UVTron series, is a gas-filled photoelectric/Geiger-Müller-type tube: incident UV photons within the ~185–260 nm response band eject photoelectrons from a metal cathode via the photoelectric effect, and the tube’s internal gas amplifies this into a detectable discharge pulse, with essentially zero sensitivity to visible or longer-UV light by construction (no interference filter required) [3].

Commercial fixed-point hydrogen flame detectors (MSA FL500-H2, Net Safety/Emerson UVS-H2 and UV/IRS-H2) pair this UV channel with a co-located IR channel (commonly ~2.9–4.5 µm) specifically to reject false alarms from arc-welding, lightning, X-rays, and hot surfaces that can also trigger a bare UV tube; commercial detectors ship with dual-band UV/IR logic, not UV alone [4][5].

Variants & subtypes

LIBS

• Gas-phase LIBS: laser breakdown initiated directly in the gas stream; demonstrated for H₂-in-N₂ binary mixtures with excellent signal linearity on Hα [21].
• Solid-substrate-initiated LIBS: plasma initiated on a metal (stainless steel found optimal over Al, Cu, Mo, Ti, W) or rock substrate, with the expanding plume sampling the surrounding gas; gave marginally better hydrogen sensitivity than pure gas-phase breakdown in the one head-to-head study found [21].
• Handheld/portable LIBS: battery-powered field units (SciAps Z-series) built primarily for solid-sample elemental/geochemical analysis, with H detection one line among dozens, not a dedicated gas-sensor mode [14][15][19][20].
• Downhole/subsurface LIBS: wireline logging tools that lower the laser-ablation head (and, in the fiber-coupled architecture, only the optical head) into a borehole while keeping the laser source and spectrometer at surface [18][44].
• Calibration-free LIBS (CF-LIBS): quantification via measured line intensities plus an assumed LTE plasma composition/temperature (validated via Hα Stark width), without matrix-matched calibration standards [26][27].
• Femtosecond/isotopic LIBS: fs-pulse LIBS and multivariate chemometric regression on the Hα fine structure/isotope shift, developed for H/D/T isotopic-ratio determination in fusion-materials and nuclear-forensics contexts [23][24][25].
• Standoff LIBS: double-pulse and long-focal-length configurations achieving breakdown and useful spectra at tens of meters (to ~50–70 m in explosives/hazardous-chemical detection work); not, in the sources found here, specifically applied to H₂ [29].

Refractive-index / interferometric

• Classic bulk-optic gas interferometer (Rayleigh/Jamin/Zeiss designs): historical mine-safety instrument exemplified by the Zeiss ZGL 2505, manually operated with a rubber-bulb gas sampler and eyepiece fringe-counting scale [6].
• Fiber/guided-wave Mach–Zehnder interferometer (MZI): in-fiber or capillary-based designs exciting core and cladding (or two-arm) optical paths that interfere at a coupler, converting $\Delta n$ into a fringe-phase shift; a capillary design demonstrated $5.7\times10^{-8}$ RIU resolution using only air-pressure changes as the test stimulus [9].
• Photonic ring resonator / whispering-gallery-mode (WGM) sensor: a microring, microbottle, or microsphere resonator whose resonance wavelength shifts with the surrounding medium’s refractive index; when coated with a hydrogen-selective transducer layer (Pd, Pd–WO₃, Pt-loaded WO₃) rather than left as a bare-RI sensor, this becomes a chemically selective device — really a Pd/WO₃ optical hydrogen sensor riding on an RI/whispering-gallery readout, cross-linked to fiber-optic sensors [10][13].
• Comb-mode-resolved / Fabry–Pérot absolute gas refractometry: frequency-comb-referenced spectral interferometry and Fabry–Pérot-cavity methods pushing absolute gas-refractivity metrology toward $10^{-9}$–$10^{-10}$ RIU, used for fundamental-constant and pressure-metrology work rather than field trace-gas sensing [39].
• Two-wavelength dispersion interferometry: measures the wavelength-dependence of $n$ (rather than $n$ at one wavelength) to separately determine physical path length and mean refractivity, or — in the fusion-plasma variant — free-electron density; demonstrated for several molecular gases (CH₄, CHF₃, CH₃F, CF₄, SF₆) but not H₂-in-air discrimination specifically [40].

Dielectric / capacitance

• Bulk cross-capacitor / coaxial-cylinder metrology cells: precision physics instruments measuring a pure gas’s or calibrated mixture’s dielectric permittivity to parts-per-million accuracy, used for equation-of-state and fundamental-constant work, not field gas identification [16][17].
• Coaxial-cylinder capacitive H₂ sensor with volumetric analysis: a research device reporting hydrogen response from 0.15–1500 ppm using transparent coaxial capacitive electrodes; the described mechanism (volumetric/optical-transparency-assisted capacitive readout) appears to rely on more than bulk gas permittivity alone, but independent replication/selectivity data were not accessible in this research pass (publisher paywall) — flag as unverified before citing this figure anywhere load-bearing.
• Functionalized interdigitated-electrode (IDE) chemocapacitor arrays: polymer- or zeolite/MOF-coated capacitive chips (e.g., an 8-coating, 10-chip array reported for natural-gas composition analysis) that regain selectivity by absorbing target gases into a coating whose own dielectric constant then changes — conceptually a chemocapacitor, not a bulk-gas-permittivity sensor, and demonstrably still cross-sensitive between chemically similar species (methane/ethane/propane cross-slopes of order −1) [43].

Rayleigh scattering / dispersion

• Single-wavelength Rayleigh-scattering lidar: an atmospheric-science tool for aerosol/density profiling, not gas-species identification; included only to distinguish it from Raman lidar (see Raman spectroscopy), which is used for quantitative H₂ gas detection at the 1–100% level over tens of meters via inelastic (vibrational) Raman shift — a completely different and far more selective mechanism [41].
• Exoplanet-atmosphere Rayleigh-slope retrieval: the astrophysical analogue described above, included purely as a physics-education cross-reference, not a field technique.

Leak/flow visualization

• Z-type (Toepler) schlieren: classic two-mirror or two-lens folded optical system with a knife-edge cutoff, laboratory/wind-tunnel standard [34][38].
• Color schlieren: replaces the knife edge with a color filter/graded-color source so deflection direction and magnitude map to hue, aiding visual interpretation [38].
• Background-oriented schlieren (BOS): camera-and-background-pattern only, no laser or precision optical bench, making it field-deployable at wellsites and pipeline yards; demonstrated for hydrogen-jet leak visualization and, with cross-correlation algorithms, for quantitative concentration/pressure-field estimation [34][35][36][37].
• Adaptive stream-stripe BOS (ASS-BOS): a refinement using an adaptively generated striped background pattern to better resolve the concentration-distribution field of a hydrogen jet [37].
• Deep-learning-enhanced BOS (e.g., dual-channel “MGR-Net”): convolutional/generative-network post-processing of raw BOS image pairs for real-time hydrogen leakage diagnosis and density/pressure field reconstruction [36].

Acoustic leak imaging

• Handheld ultrasonic phased-array cameras: Distran Ultra/Ultra Pro/Ultra Pro Max/Ultra Pro X and Fluke ii900/ii910, combining a microphone array with a visible-light camera and a real-time acoustic-overlay display [11][32][33].
• Single-point (non-imaging) ultrasonic detectors: fixed or handheld single-transducer devices (e.g., MSA/Emerson Observer-i) that alarm on ultrasonic energy in a zone but do not localize the leak visually — a lower-cost, lower-information alternative to the imaging cameras [45].

Flame UV/UV-IR detectors

• Solar-blind UV-only detectors: gas-discharge tube sensors (Hamamatsu UVTron family) responsive to ~185–260 nm, used as OEM components inside larger fixed-point detector housings [3].
• UV/IR dual-band fixed-point detectors: the dominant commercial product class for hydrogen service (MSA FL500-H2, Net Safety/Emerson UVS-H2 and UV/IRS-H2), combining the UV OH* channel with an IR channel and a coincidence/logic algorithm for false-alarm rejection [4][5].
• Multispectral/imaging OH* research cameras: sCMOS-based imaging systems with narrowband 310 nm bandpass filters (stacked for high out-of-band rejection), used in hydrogen-safety research to image, not just alarm on, flame location and size, and directly compared against thermal-IR imaging of the same flames [1].

Performance envelope

Reading this table

All figures are as reported in the cited sources; manufacturer figures should be read as best-case, controlled-condition numbers unless a field/independent validation is also cited. Because this page spans genuinely different instrument classes, “range” and “LOD” are not always meaningful in the same units across rows; that is noted explicitly where it occurs.

Technique	Best-demonstrated sensitivity / range	Response	Environmental rating	Selectivity for H₂	Cost class	Refs
Gas-phase LIBS (Hα)	~20 ppm (mass) LOD, N₂–H₂ binary mixture, atmospheric pressure lab demo	Single ns laser shot; ms-scale full readout	Lab-demonstrated only at ambient T/P	None — total elemental H, confounded by any H-bearing species	Research-grade laser + spectrometer	[21]
LIBS on hydrous rock/mineral matrices (e.g. basalt, ChemCam-relevant)	±~1.5 wt% H₂O prediction error across normalization schemes; separates high-H₂O clay/gypsum phases (12–20+ wt% H₂O) from low-H₂O basalt but not finer distinctions	Single-shot per analysis point	Demonstrated on Mars (ambient CO₂, low pressure) via ChemCam	None — total structural/bound water + OH, strong matrix effects	Integrated into a multi-hundred-million-dollar planetary rover instrument (not a Hestia-relevant cost class)	[22]
Handheld LIBS (SciAps Z-903 class)	Full periodic table incl. H (extended 190–950 nm spectrometer); Li LOD 2–5 ppm cited as a reference point; no published H-in-gas ppm LOD	Seconds per analysis spot	Field-rugged handheld (IP-rated, battery)	None for H specifically; vendor markets it for solids (Li, carbon, metals), not gas-phase H₂	Rental ~$2,000/week; purchase implied in the tens of thousands to ~$100k+ range	[14][15][19][20]
Fiber-coupled subsurface LIBS (dissolved-metals proof of concept)	LOD 0.009–0.10 ppm for K, Sr, Ca in solution; not a gas-phase H₂ demonstration	ns pulses, 5 Hz rep rate	Lab prototype; stated path to pressure-vessel ruggedization for downhole use	N/A (dissolved metals, not H₂)	Prototype/research stage	[44]
Downhole wireline LIBS (elemental geochemistry logging)	ppm-level for rock-forming/trace elements; <100 µm spot; 60–300 analyses/m at 2–10 m/min logging speed	~10 Hz shot rate	Pre-commercial; field trials only as of the source date	N/A — elemental rock geochemistry tool, not gas sensing	Pre-commercial; price not disclosed	[18]
Capillary Mach–Zehnder gas refractometer	$5.7\times10^{-8}$ RIU resolution (demonstrated on air pressure changes, not H₂)	Not stated; interferometric readout is inherently fast (optical)	Lab demonstration	None (bulk RI, no selective coating tested)	Research-grade fiber-optic bench setup	[9]
Photonic ring-resonator hydrogen sensor (Pd/WO₃-functionalized)	0.03–1 vol% (312–10,000 ppm) range; up to 140 pm/% sensitivity; ~60 ppm LOD reported for one design	Not separately stated in sources found	Lab demonstration	Medium–high (via Pd/WO₃ chemistry, not bulk RI)	Research-grade	[10]
Classic Zeiss-type mine gas interferometer	Sub-percent CH₄/CO₂ resolution by eye (historical); precision limited by manual technique	Tens of seconds (manual bulb pumping + visual reading)	Underground coal-mine service, 1957–1960s	None; later assessed as unreliable owing to sensitivity to pressure, temperature, RH, and interfering gases	Expensive/specialist for its era; discontinued	[6][42]
Schlieren / BOS hydrogen-jet imaging	Qualitative leak visualization; quantitative concentration/pressure-field retrieval demonstrated in specific lab jet geometries (e.g., ~23° average jet expansion angle)	Video-rate (tens of fps)	Lab/test-rig demonstrations; some outdoor field use	None (density-gradient imager, not species-selective)	Camera + optics/background pattern; order $1,000s–10,000s for a research rig; no packaged commercial “H₂ BOS camera” product identified	[34][35][36][37]
Acoustic imaging camera — Fluke ii910	2–100 kHz band; detection range up to ~120 m (spec’d “up to 393 ft”)	Real-time video-rate imaging	General industrial (not explicitly ATEX in base spec; check model variant)	None (any pressurized-gas or electrical-discharge ultrasound source)	~$28,550 street price (new)	[32][33]
Acoustic imaging camera — Distran Ultra Pro / Ultra Pro Max	Minimum detectable leak ~0.3 L/h (Ultra Pro) down to ~0.15 L/h (Ultra Pro Max) at ~30 cm; detection range 0.3–100 m; requires gas under pressure (effective threshold commonly cited ≳2 bar / 29 psi)	Real-time video-rate imaging	ATEX-rated variants available for hazardous areas	None	Price not published; comparable cost class to Fluke ii-series	[11][32][46][47]
Solar-blind UV flame tube (Hamamatsu UVTron R2868)	Detects a cigarette-lighter flame at >5 m; hydrogen burner flame at ~10 m in application testing	”A few ms” to signal output	185–260 nm spectral response; no visible-light interference by design	High for any UV-emitting flame (not H₂-selective vs. other flames) once combustion has started	OEM component, low unit cost (tens of USD class for the tube itself; system cost is in the housing/electronics)	[2][3]
UV/IR fixed-point hydrogen flame detector (MSA FL500-H2)	UV 185–260 nm + IR ~2.95 µm; detects H₂ flame at up to 18 m (60 ft), <3 s response	−55 to +85 °C; 130° field of view; 2-min automatic optical-path self-check (COPM)	SIL 3-suitable, CSA/FM/ATEX/IECEx/CE	High for flame presence once ignited; not a pre-ignition gas sensor	Certified safety-instrumented-system component; price not confirmed in sources found	[4]
UV/IR fixed-point hydrogen flame detector (Net Safety/Emerson UVS-H2)	24-inch flame plume detected at 55 ft (~16.8 m) in 1.8 s; 120° horizontal FOV	−50 to +75 °C	CSA, ANSI-UL/ISA, ATEX	Same as above	Similar class to MSA FL500-H2	[5]
OGI thermal camera + CO₂ tracer (FLIR GF343, discontinued)	Images the co-added 3–4% CO₂ tracer, not H₂ itself; effective for >95%-purity H₂ systems (e.g. generator cooling gas)	Video-rate	InSb detector, 3–5 µm cooled, ~4.3 µm passband	None for H₂ directly; entirely dependent on deliberately introduced tracer gas	Discontinued product; was a premium OGI camera (tens of thousands USD class typical of this product family)	[30]
Catalytic Pt-WO₃ IR-imaging window (2025 research demo)	Converts standard IR camera into an H₂-selective calorimetric imager	17 s response / 24 s recovery	Lab demonstration	Medium (via Pt-WO₃ catalytic chemistry, not IR spectroscopy)	Research-stage add-on component	[31]

Figures of merit

The figures of merit below are defined in full on the Figures of merit page; this section only specializes them to the seven technique families covered here and flags which axis actually gates field use. Where a figure is a concentration or a bulk physical unit (RIU, vol%, ppm), see Concentration units for how they convert into one another.

Figures-of-merit table for this modality (11 rows)

Figure of merit	Symbol / formula	What it means here	Typical value for this technology	How to read it
Sensitivity ($S$)	$\partial(\text{signal})/\partial(C_{H_2})$	Signal units differ completely across this page’s seven families: LIBS reports an atomic emission-line intensity, interferometric/ring-resonator designs report a phase or wavelength shift, dielectric sensors report a capacitance change	Ring-resonator (Pd/WO₃-functionalized): up to 140 pm per vol% H₂ [10]; LIBS’s best-demonstrated “sensitivity” is actually to humidity: the Hα/N I intensity ratio fits relative humidity with $R^2=0.9997$ [25]	Raw sensitivity numbers cannot be compared across rows without first dividing by each technique’s own noise floor (next row) — and for LIBS, the strongest fit found is to the wrong variable
Limit of detection (LOD)	$k\,\sigma_{blank}/S$, $k=1$–$3$	Smallest resolvable total-H (LIBS) or $\Delta n$/$\Delta\varepsilon_r$ (RI/dielectric) signal; none of the sources here state which coverage factor $k$ or averaging time was used	Gas-phase LIBS Hα: ~20 ppm (mass), single laser shot, N₂–H₂ lab binary mixture [21]; ring resonator: ~60 ppm for one design [10]; capillary Mach–Zehnder interferometer: $5.7\times10^{-8}$ RIU, not converted to an H₂ ppm-equivalent in the source [9]	Integration-time caveat: unlike TDLAS or an electrochemical cell, none of these LOD figures come with a stated $\tau$ or Allan-deviation floor to trade against response time — a single ns laser pulse (LIBS) or an optical readout (interferometric) has no simple “average longer, see smaller” story published here. Treat every LOD in this table as a one-shot lab number, not a metrologically derived detection limit
Selectivity / cross-sensitivity	$1/\max_j(K_{ij})$	Whether the signal can be attributed to H₂ specifically, versus any hydrogen-bearing species (LIBS), any polarizable/dense gas (RI, dielectric), any density gradient (schlieren/BOS), any pressurized leak (acoustic), or any flame (UV/IR)	None for LIBS (total elemental H), bare RI/interferometry, dielectric sensing, schlieren/BOS, and acoustic imaging; high only post-ignition for UV/IR flame detectors; medium–high for ring-resonator/MZI designs specifically functionalized with a Pd or WO₃ layer [10][12][13]	Most rows here are, by the underlying physics, non-selective bulk-property or total-elemental figures of merit — selectivity is not a maturity gap that better engineering closes, it is architecturally absent unless a chemically selective transducer has been bolted on top
LIBS Hα line intensity / detection limit	$I_{H\alpha}$ (656.28 nm, Balmer-α)	Atomic hydrogen emission intensity from the laser-induced micro-plasma; tracks total elemental hydrogen in the ablated/sampled volume (H₂O, OH, hydrocarbons, hydroxylated minerals, and any free H₂ all indistinguishably) — never H₂ the molecule	~20 ppm (mass) LOD, H₂-in-N₂ gas-phase lab demo [21]; ±~1.5 wt% H₂O prediction error on hydrous rock/mineral matrices (ChemCam-relevant) [22]	Read every LIBS Hα number as “total hydrogen present,” never as “H₂ concentration”: humidity alone reproduces the same line with $R^2=0.9997$ correlation to RH [25], and self-absorption biases the reading low at high hydrogen fraction [21][22][28]
Refractive-index resolution (RIU)	$\Delta n_{\min}$ (dimensionless, refractive-index units)	Smallest resolvable change in bulk refractive index — the fundamental sensitivity floor of every interferometric/RI design, before it is converted into an equivalent gas concentration	Capillary Mach–Zehnder interferometer: $5.7\times10^{-8}$ RIU, demonstrated with air-pressure changes rather than H₂ [9]; comb-mode-resolved/Fabry–Pérot absolute gas refractometry: $10^{-9}$–$10^{-10}$ RIU, metrology-grade rather than field-deployable [39]	An impressive RIU figure does not automatically imply an impressive H₂ ppm figure: converting the two requires $dn/dC_{H_2}$ at the actual gas composition, and H₂’s own $\Delta n$ from air is only $\approx1.6\times10^{-4}$ at 100% H₂ [7][8] — a step most RIU spec sheets omit entirely
Standoff distance	$d_{\text{standoff}}$	Distance from instrument to target/leak/flame at which a usable signal is still obtained; relevant to standoff LIBS, UV/IR flame detection, and acoustic imaging, all line-of-sight instruments rather than in-situ probes	Standoff LIBS: demonstrated to ~50–70 m in explosives/hazardous-chemical detection work, unverified for H₂ specifically [29]; UV/IR flame detectors: 16.8–18 m (Net Safety UVS-H2, MSA FL500-H2) [4][5]; bare UVTron tube: ~10 m for a hydrogen burner flame [2][3]; acoustic imaging cameras: 0.3–120 m depending on model [11][32][33][46][47]	Vendor standoff figures are clean-air, controlled-condition numbers; fog, dust, and water vapor measurably attenuate the weak 306–310 nm flame signal over path length, a degradation the datasheet range does not capture [4][5]
Response / recovery time	$t_{50}$, $t_{90}$, or family-specific	How fast a usable signal appears once the precondition (laser pulse, ignition, pressurized leak) has already occurred	LIBS: single ns pulse, ms-scale full readout [21]; UVTron: “a few ms” [3]; MSA FL500-H2: <3 s [4]; Net Safety UVS-H2: 1.8 s [5]; catalytic Pt-WO₃ IR window: 17 s response / 24 s recovery [31]; schlieren/BOS and acoustic imaging: real-time video rate [32][33][34][35][36][37]	Fast response is close to irrelevant to this page’s central problem: none of these techniques give a pre-ignition or diffuse-seep warning, so a sub-second number only matters once the underlying event (a flame, a pressurized jet, a laser shot on a rock) has already happened
Dynamic range	$\log_{10}(C_{\max}/C_{\min})$	Full span of usable output; stated for only a minority of rows on this page	Ring resonator: 0.03–1 vol% (312–10,000 ppm) [10]; coaxial-cylinder capacitive H₂ sensor: 0.15–1,500 ppm claimed, but the underlying mechanism and selectivity are unverified — full text was paywalled and no independent replication was found	Where no explicit decades figure is stated (LIBS, schlieren/BOS, acoustic imaging), that is a genuine gap in the sources, not a deliberate omission — do not assume a wide range in the absence of a published one
Drift / stability	$R_0(t)$, $S(t)$	Long-term baseline/sensitivity creep; almost never quantified for the families on this page outside the certified flame-detector product class	Classic Zeiss-type mine gas interferometer: assessed as unreliable in field service owing to pressure/temperature/RH sensitivity [6][42]; MSA FL500-H2: 2-minute automatic optical-path self-check (COPM) substitutes for a drift number by continuously verifying the optical path is still clear [4]	Absence of a published drift spec is itself informative here — for LIBS, ring resonators, schlieren/BOS, and acoustic imaging, treat long-term stability as unverified rather than assuming it is good because no bad number was reported
Power	Family-specific	Electrical/optical power draw; not itemized as a simple mW figure for most product classes on this page	LIBS: pulsed Nd:YAG laser (tens–hundreds of mJ/pulse) plus spectrometer, research-grade, continuous draw unverified; acoustic cameras and UV/IR flame detectors are line- or battery-powered catalog products with no published mW spec in the sources found [4][5][32][33]	This page’s cost and power questions are better framed as capital cost class than a power budget: handheld LIBS rents around $2,000/week with purchase implied in the tens of thousands to $100k+ range [14][15][19][20], and a Fluke ii910 acoustic camera lists around $28,550 new [32][33]
Operating envelope	$C_{\text{raw}}\cdot f_T(T)\cdot f_{RH}(RH)\cdot f_P(P)$	Where each instrument survives, produces output, and — separately — where its stated accuracy actually holds	MSA FL500-H2: −55 to +85 °C [4]; Net Safety UVS-H2: −50 to +75 °C [5]; Distran Ultra Pro: ATEX-rated variants for hazardous areas [11][46][47]; LIBS, RI/interferometric, dielectric, and schlieren/BOS rows: lab-demonstrated only, no stated field operating envelope [21][9][34][35]	For the safety-instrument product class (flame detectors, acoustic cameras) the operating envelope is genuinely wide and certified; for every physics-demonstration row (LIBS, RI, dielectric, schlieren) no compensated envelope has been established at all — there is no field accuracy claim yet to void

What dominates. Selectivity/cross-sensitivity is the figure of merit that decides whether any technique on this page is usable for a geologic-H₂ job, and for six of the seven families the answer is architecturally “none,” not “not yet good enough” — LIBS reads total elemental hydrogen, RI/interferometric and dielectric sensing read a bulk mixture property, and schlieren/BOS and acoustic imaging respond to density gradients or pressure-driven noise that have nothing to do with gas identity. No improvement to LOD, response time, or standoff distance changes that conclusion.

The one place these numbers become operationally decisive is the mature safety-instrument tier — UV/IR flame detectors and acoustic imaging cameras — where standoff distance and response time genuinely are the design variables, but only after ignition or a pressurized leak has already occurred.

In trade-off-triangle terms, this page sits at a degenerate corner: selectivity is fixed at zero (or fixed at “post-event only”) for most of these techniques, so sensitivity and stability improvements cannot buy back the one axis that would make any of them a soil-gas or downhole exploration sensor.

Environmental covariance & cross-sensitivity

LIBS

LIBS. The Hα signal is a direct, strong function of ambient/sample humidity: a controlled study fit linear relationships between relative humidity and the Hα-to-N/O emission-line intensity ratio (e.g., Hα/N I 746.8 nm) with $R^2 = 0.9997$, and validated the resulting humidity estimate against a reference hygrometer [25]. This is both a demonstration that LIBS can be turned into a quantitative humidity sensor and a warning that any attempt to read LIBS Hα as an H₂-gas proxy in the field would be dominated by moisture rather than trace H₂.

On solid/rock targets, classic matrix effects — ablation efficiency and plasma coupling depending on the sample’s mechanical/chemical/mineralogical properties, not just its elemental composition — mean the same true hydrogen content can produce different Hα intensities in different rock types, which is why a Mars-relevant basaltic-materials calibration study reported an irreducible ±1.5 wt% H₂O error across every normalization and regression scheme tried [22].

At high hydrogen fraction, self-absorption/self-reversal further biases naive intensity-based readings downward, and isotopic (H/D/T) work must explicitly correct for fine/hyperfine structure and line-distortion effects riding on top of this same broadened, sometimes self-reversed line [21][23][24]. Laser-target coupling itself is sensitive to surface roughness, moisture films, and standoff-distance jitter in any field deployment, compounding the matrix-effect problem.

Refractive index / interferometry and dielectric/capacitance sensing

Refractive index / interferometry and dielectric/capacitance sensing. Both share the same fundamental cross-sensitivity: any change in bulk gas composition or density that shifts mean molecular polarizability-weighted number density is indistinguishable from an H₂ change. In practice this means CH₄, CO₂, He, H₂S, and — dominating everything at typical soil-gas/atmospheric timescales — ambient temperature and pressure fluctuations (which change gas density directly, with no composition change at all) are all confounders of equal or greater magnitude than a single-digit-to-low-double-digit ppm H₂ anomaly.

This is precisely the failure mode documented for interferometric mine methanometers: a peer-reviewed comparison of mine methane-sensor technologies concluded that “devices with an interferometric sensor are unreliable since their readings are very sensitive to changes in environmental conditions (pressure, temperature, and relative humidity) and to atmospheric factors such as the presence of other interfering gases and the variation in oxygen concentration” [42] — a direct, quantitative-in-spirit confirmation of the theoretical selectivity problem above, from the exact application domain (underground mine gas monitoring) most analogous to a soil-gas or borehole H₂ survey.

Functionalized capacitive-sensor arrays partially escape this by absorbing gases selectively into a coating, but even there, cross-sensitivity between chemically similar small hydrocarbons (methane/ethane cross-slope ≈ −1.16 in one reported array) persists, and no hydrogen-specific validation data were found [43].

Rayleigh scattering / dispersion

Rayleigh scattering / dispersion. Not a practical field technique (see above); the dominant “environmental” confound is that aerosol/particulate Mie scattering exceeds molecular Rayleigh scattering by many orders of magnitude under real atmospheric conditions, burying any hydrogen-composition signal entirely.

Schlieren / BOS

Schlieren / BOS. Cannot distinguish a compositional refractive-index gradient from a thermal one — a sun-warmed rock surface, a hot compressor housing, or simple convective air currents produce schlieren/BOS signals of the same character as a real gas jet, so species identity must come from context (a known process line), not from the image itself [34][35]. Outdoor deployments are additionally sensitive to wind and atmospheric turbulence, which perturb the background pattern independent of any leak.

Acoustic leak imaging

Acoustic leak imaging. The central operational limitation is that ultrasonic emission requires a pressure differential driving turbulent/choked flow through an orifice; industry guidance places the effective threshold for reliable detection at roughly ≥2 bar (29 psi) gauge pressure [46][47]. Below that, or for a diffuse, low-driving-pressure seep (exactly the geometry of a natural soil-gas H₂ anomaly), there is no meaningful ultrasonic signal to image, which rules the technique out for exploration-style diffuse-flux measurement even though it works well for a pressurized wellhead valve or flange leak.

False positives arise from any other ultrasonic-band noise source in the environment: compressed-air tools, corona/partial electrical discharge, and mechanical friction are all commonly cited confounders, which is also why some of these same cameras double as partial-discharge detectors for electrical equipment [32][33].

Flame UV/UV-IR detectors

Flame UV/UV-IR detectors. The UV channel is close to immune to solar interference by design, but long-path detection distance (10–18 m in the cited specs) can be degraded by fog, water vapor, or dust attenuating the weak 306–310 nm signal over that path — a genuine field limitation not fully captured in clean-air datasheet numbers.

The co-located IR channel exists specifically to reject false triggers from arc-welding flashes, lightning, X-ray sources, and hot/reflective surfaces that could otherwise fire a bare UV tube; conversely, a UV-only design lacking this logic is more prone to nuisance alarms in an active industrial environment with welding or electrical-switching activity nearby [4][5].

Pre-ignition blind spot

None of these detectors respond to H₂ before ignition — they are strictly post-combustion safety instruments, offering zero warning of a pre-ignition leak or accumulating explosive atmosphere (that job belongs to the electrochemical/catalytic/MOX/laser-absorption sensors covered elsewhere in this compendium).

Use for geologic H₂

Field verdict

None of the six technique families on this page is a drop-in replacement for the electrochemical, MOX, catalytic, Pd-resistive, or laser-absorption sensors that should form the core of Hestia’s ppm-level soil-gas and downhole H₂ sensing stack. Their correct role is (a) LIBS as an opportunistic solid-sample mineralogical/elemental-hydrogen screening adjunct where a handheld unit is already in hand for other purposes, and (b) schlieren/BOS, acoustic imaging, and UV/IR flame detection as mature wellsite safety and leak-management instrumentation once Hestia has physical H₂ infrastructure to protect.

Refractive-index and dielectric/capacitance sensing are ruled out for exploration use by the underlying physics, not by a maturity gap that better engineering will close.

LIBS

LIBS — not currently a soil-gas or downhole H₂ exploration tool, but a plausible adjunct for cuttings/core screening. No source found in this research describes LIBS being used for natural-hydrogen exploration specifically; the closest analogues are (a) handheld LIBS’s established role screening drill cuttings and outcrop samples for lithium and other economic elements in mineral exploration [14][19], and (b) the demonstrated (if matrix-effect-limited, ±1.5 wt% H₂O) ability of LIBS to flag hydrous alteration phases (clays, gypsum) in rock via total-structural-hydrogen content [22].

Because natural H₂ generation and migration is frequently associated with specific alteration mineralogy (serpentinization, clay-cap formation — see geologic context for H₂ exploration), a case could be made for using existing field-deployed handheld LIBS units — many exploration geologists already carry one for lithium/critical-minerals screening — as a secondary, opportunistic mineralogical-alteration proxy on cuttings or core, but this would be an indirect geochemical inference, not a direct H₂ gas measurement, and no published validation of that specific workflow was found.

For free-gas H₂ quantification, the one directly relevant number — 20 ppm(mass) LOD for H₂ in an N₂ background via gas-phase LIBS [21] — was demonstrated as an industrial hydrogen-leak-detection concept, not a soil-gas or downhole exploration instrument, and would need substantial re-engineering (open-path or extractive gas-cell geometry, ruggedization, matrix validation against real soil-gas backgrounds containing CH₄/CO₂/H₂S/water vapor) before it could be considered field-ready.

Downhole LIBS specifically remains pre-commercial even for its primary (solid-rock elemental geochemistry) application [18], and the one fiber-coupled “no downhole electronics” architecture found in this research was validated for dissolved metals in groundwater, not gas-phase species [44] — a promising architecture (laser and spectrometer stay at surface; only a passive optical head goes downhole, echoing the passive-fiber philosophy in fiber-optic sensors) that has not yet been demonstrated for H₂.

Refractive-index and dielectric sensing

Refractive-index/interferometric and dielectric/capacitance sensing — ruled out for exploration use by the physics. Bulk RI and permittivity sensing cannot distinguish an H₂ excess from co-occurring CH₄/CO₂/He/H₂S variation or from ordinary T/P/RH swings of the same or larger magnitude as a single-digit-ppm exploration anomaly, and the mine-methanometer field history above [42] confirms this is not a theoretical worry — it played out as an operational reliability problem in the one directly analogous historical deployment.

This is not a maturity problem that better photonics will fix; it is intrinsic to measuring a bulk, additive physical property (polarizability × density) in a real multi-component gas matrix with an unknown, time-varying background.

The one way this physics becomes exploration-relevant is when it is paired with an independent, selective transducer (Pd, Pd-alloy, or WO₃) that converts a chemically selective hydrogen-uptake event into an RI or capacitance change — at which point the sensor is properly classified (and should be selected/evaluated) as a Pd-hydride or metal-oxide device, covered in palladium-resistive sensors, fiber-optic sensors, and MOX chemiresistors, not as a general-purpose RI/dielectric sensor. Bare RI/dielectric sensing has no role in soil-gas surveys, downhole logging, or continuous monitoring for Hestia.

Schlieren/BOS

Schlieren/BOS — a wellsite safety and leak-visualization tool, not an exploration sensor. These techniques image macroscopic, pressure-driven jets and plumes; a diffuse, low-flux natural soil-gas seep does not present the sharp refractive-index gradient these methods are built to see, and no source found here applies schlieren/BOS to soil-gas or seep detection. Their legitimate role for Hestia is post-discovery infrastructure safety: visualizing and triaging leaks from compressors, wellheads, flanges, and process piping once a site has moving/pressurized H₂ to manage.

Acoustic leak imaging

Acoustic leak imaging — likewise a wellsite/facility safety and leak-triage tool, not an exploration or diffuse-seep sensor. The ≥2 bar operational threshold [46][47] directly excludes the low-driving-pressure, diffuse-flux regime of natural soil-gas seepage and shallow borehole monitoring. It is, however, a strong candidate for continuous monitoring stations and wellsite HSE programs once Hestia has pressurized H₂ infrastructure (wellheads, compression, pipeline tie-ins) to inspect — a genuinely useful, mature, catalog-buyable instrument in that specific role. See downhole deployment and calibration & QA/QC for how a safety-instrument layer like this fits alongside the primary sensing stack.

Flame UV/UV-IR detectors

Flame UV/UV-IR detectors — mandatory safety instrumentation, not an exploration sensor, but directly relevant to continuous monitoring station design. Any Hestia facility that vents, flares, compresses, or electrolyzes hydrogen should budget for certified UV/IR flame detection (MSA FL500-H2-class or equivalent) as part of its fire-and-gas safety system, precisely because H₂ flames are otherwise nearly invisible to personnel and to ordinary IR-only flame detectors calibrated for hydrocarbon fires.

This sits alongside, not instead of, pre-ignition gas detection (electrochemical/catalytic/MOX/pellistor sensors covered elsewhere in this compendium) — the LEL of 4 vol% H₂ is the threshold a pre-ignition sensor network should catch well before a UV/IR flame detector ever needs to fire.

Procurement guidance

For procurement: none of the six technique families here is a substitute for the core ppm-level soil-gas and downhole H₂ sensing stack. Budget LIBS as an opportunistic solid-sample screening adjunct where a unit is already in the truck for other purposes, and budget schlieren/BOS, acoustic imaging, and UV/IR flame detection as mature wellsite safety and leak-management instrumentation once Hestia has physical H₂ infrastructure to protect — not as line items competing with the electrochemical, MOX, catalytic, Pd-resistive, or laser-absorption sensing budget.

Vendors & products

Category	Vendor / Product	Notes	Refs
Handheld LIBS	SciAps Z-903 (distributed by Malvern Panalytical)	190–950 nm triple-spectrometer, measures H via extended-range spectrometer; full periodic table H→U; marketed for lithium/geochem, PMI, scrap sorting	[14][19][20]
Handheld LIBS	SciAps Z-901	Lower-cost model; cannot measure H, Li, Na, C, N, O, Cl, Br, F, Ce, Rb, S per vendor spec — a procurement-relevant distinction from the Z-903	[15]
Handheld LIBS	Thermo Fisher Scientific Niton Apollo	Competing handheld LIBS analyzer product line (identified via cross-reference; not independently verified in this research pass)	—
Downhole LIBS	MinEx CRC (Australia) pre-commercial LIBS logging tool	75 mm OD, ~1.7 m, <5 kg, wireline-deployed, NQ-borehole-compatible, 60–300 analyses/m at 2–10 m/min; field trials only as of the source brochure	[18]
Downhole LIBS (concept/academic)	LIBSLog project	Academic AGU-presented concept for a LIBS-based borehole exploration logging tool; existence/title confirmed via literature search, full technical detail not accessible in this research pass	—
Historical gas interferometer	Carl Zeiss ZGL 2505	1957-era portable methane/CO₂ interferometer used at Teralba Colliery (NSW), 1957–1960s; no longer manufactured	[6]
Acoustic leak-imaging camera	Fluke ii910 / ii900	2–100 kHz, up to ~120 m range, 7” touchscreen, LeakQ/PDQ/MecQ modes; ii910 ≈ $28,550 new	[32][33]
Acoustic leak-imaging camera	Distran Ultra / Ultra Pro / Ultra Pro Max / Ultra Pro X	124-element ultrasound array; Ultra Pro sensitivity ~0.3 L/h @ 30 cm (4-bar leak), range 0.3–100 m; Ultra Pro Max improves to ~0.15 L/h; ATEX variants available	[11][46][47]
Acoustic point detector	MSA/Emerson Observer-i	Fixed-point (non-imaging) ultrasonic gas-leak detector for continuous-monitoring installations	[45]
Solar-blind UV flame sensor (component)	Hamamatsu UVTron R2868 (and related R-series)	185–260 nm gas-discharge UV tube; OEM component inside larger detector housings; detects a lighter flame >5 m, H₂ burner flame ~10 m in vendor application testing	[2][3]
Fixed-point UV/IR hydrogen flame detector	MSA FL500-H2	UV 185–260 nm + IR ~2.95 µm; 18 m detection range, <3 s response, −55…+85 °C, SIL 3-suitable	[4]
Fixed-point UV/IR hydrogen flame detector	Net Safety / Emerson Rosemount UVS-H2, UV/IRS-H2	120° FOV, 55 ft/1.8 s response spec, −50…+75 °C, CSA/ANSI-UL/ATEX certified	[5]
OGI thermal camera (tracer-gas workflow)	FLIR GF343 (discontinued)	InSb 3–5 µm cooled detector filtered to ~4.3 µm; used with 3–4% CO₂ tracer gas to visualize H₂ leaks in >95%-purity generator-cooling systems	[30]
Research imaging (not a product)	EHD MaxCam-400UV-TE + 310 nm bandpass; FLIR T640 (comparison)	Used in a published UV-vs-IR hydrogen-flame-imaging comparison study; not a packaged commercial “hydrogen flame camera” product	[1]

Open questions

1. No literature was found combining LIBS with natural-hydrogen exploration directly (soil gas, seep detection, or targeted downhole H₂ gas quantification) — this is either a genuine white space or a sign that search terms need refinement against domain-specific French/Australian/Latin-American natural-hydrogen literature (much of the active field literature is French-language, from the Mali/Bourakébougou and French natural-hydrogen research community). Worth a dedicated follow-up pass.
2. Verify the SciAps Z-903 hydrogen detection limit/precision directly with the vendor — no public LOD or precision figure for H specifically (as opposed to Li, C, and metals) was found; procurement decisions should not proceed on the assumption that “H is on the element list” implies useful quantitative sensitivity.
3. Get current, official pricing for the SciAps Z-903 (only rental rates found), Distran Ultra Pro/Ultra Pro Max (no public pricing found), and MSA FL500-H2/Net Safety UVS-H2 per-unit costs before using any cost figures here for budget planning.
4. Confirm the Gladstone–Dale coefficient used above against a primary, non-shock-tube source for ambient near-surface conditions relevant to a wellsite BOS rig — the cited measurement was made behind reflected shock waves at high temperature/pressure, not at 1 atm/near-room-temperature.
5. Chase full text of the mine-methanometer comparison study [42] — the interferometric-sensor-unreliability claim is currently sourced from a search-engine-synthesized snippet rather than a directly read passage; get the original paper to confirm exact wording and any quantitative RH/T/P sensitivity coefficients.
6. Investigate whether any research group has combined the “no downhole electronics, fiber-delivered laser, all-optical trigger” LIBS architecture demonstrated for dissolved metals [44] with gas-phase H₂ sensing — this architecture is unusually well-matched to Hestia’s stated downhole-electronics preference and appears to be a natural, currently-unexplored extension.
7. Independently verify the coaxial-cylinder capacitive hydrogen sensor’s selectivity mechanism — a publisher login wall blocked full-text access, and the “0.15–1500 ppm” figure and “transparent coaxial cylinder” architecture deserve a closer read to determine whether this is truly a bulk-permittivity measurement (which the physics above says should not be H₂-selective) or a functionalized/optical-hybrid device that regains selectivity by some other means.
8. Quantify realistic UV/IR flame-detector detection range and false-alarm rate under Hestia’s specific candidate wellsite conditions (dusty, arid soil-gas exploration sites vs. the humid/coastal or indoor industrial settings most manufacturer datasheets and the one hysafe field study were validated in) before finalizing a fire-and-gas system spec.
9. Track the “catalytic Pt-WO₃ IR-imaging window” line of research [31] (published January 2025) — if response/recovery times and selectivity continue to improve, this could mature into a genuinely useful visual H₂-leak-imaging tool that does not require a tracer gas, a meaningful upgrade over the current FLIR-GF343-style CO₂-tracer workaround for any Hestia facility handling high-purity H₂.

Related pages

• Why H₂ is hard to sense — the compendium-wide framing of the symmetry/selectivity problem underlying the RI/dielectric section
• Raman spectroscopy — the selective optical technique this page’s Rayleigh/dispersion detour explicitly contrasts against
• Enhanced Raman — cavity/fiber/plasmonic tricks for ppb-level H₂ sensitivity, the productive cousin of the RI/interferometric techniques here
• IR absorption and proxies — the fuller derivation of why thermal-IR/NDIR cannot see H₂, referenced by the tracer-gas OGI discussion
• UV/EUV and fluorescence — the native optical UV/EUV handle on H₂ itself, distinct from the flame-emission UV covered here
• Fiber-optic sensors — the passive, no-downhole-electronics sensing family that Pd/WO₃-functionalized ring resonators and subsurface LIBS architectures both echo
• MOX chemiresistors — where a WGM/RI sensor lands once it is functionalized with a selective coating
• Palladium-resistive sensors — the selective Pd-hydride chemistry that must pair with bare RI/dielectric readout to make it exploration-relevant
• Thermal conductivity & acoustic sensors — the real physics behind most “IR hydrogen sensors” actually sold, relevant background for the OGI/tracer-gas discussion
• Catalytic pellistors — another non-selective, total-combustible-load technique analogous in spirit to LIBS’s total-elemental-hydrogen signal
• Electrochemical cells — the field-proven non-optical workhorse for the H₂ channel these techniques cannot replace
• Geologic context for H₂ exploration — the alteration mineralogy (serpentinization, clay caps) that motivates LIBS as an opportunistic cuttings/core screening adjunct
• Downhole deployment — where the “no downhole electronics” architecture shared by fiber-coupled LIBS and passive fiber-optic sensing matters most
• Calibration & QA/QC — where safety-instrument layers like UV/IR flame detection and acoustic imaging fit alongside the primary sensing stack
• Environmental covariance — the compendium-wide treatment of T/P/RH confounders that dominate the RI/dielectric and LIBS-humidity discussions here

References

Full reference list (0 entries)

[1] International Association for Hydrogen Safety, “Small Burning Hydrogen Leaks in Industrial Plants: A Comparison of UV and IR Remote Imaging Detection,” Hydrogen Safety journal, article 10.58895/hysafe.39. https://hysafejournal.org/articles/10.58895/hysafe.39

[2] Hamamatsu Photonics, “Topic: Hydrogen flame detection,” UVTRON application note. https://www.hamamatsu.com/jp/en/product/optical-sensors/uv_flame-sensor/flame-sensor_uv-tron/application-hydrogen-flame.html

[3] Hamamatsu Photonics, “Flame sensor (UV TRON) R2868” product/datasheet page. https://www.hamamatsu.com/us/en/product/optical-sensors/uv_flame-sensor/flame-sensor_uv-tron/R2868.html

[4] MSA Safety, “Hydrogen Flame Detector — FL500-H2 (UV/IR Flame Detector for Hydrogen Applications)” product and specifications pages. https://us.msasafety.com/Fixed-Gas-%26-Flame-Detection/Flame-Detectors/FL500-H2-UV-IR-Flame-Detector-for-Hydrogen-Applications/p/000014000600011019; specifications: https://docs.msasafety.com/fl500/en-us/FL500%20UV_IR%20FL500-H2%20Flame%20Detector/Specifications.htm

[5] Net Safety Monitoring / Emerson Rosemount, “UV/IRS-H2 & UVS-H2 Flame Detectors,” product data sheet BCR-0040-00-011; Instrumart product listing. https://www.instrumart.com/products/38383/net-safety-uvs-h2-hydrogen-flame-detector; datasheet: https://www.instrumart.com/assets/UV_IRS-H2_UVS-H2-datasheet.pdf

[6] Storyplace.org.au, “A Necessary Invention” (history of the Zeiss ZGL 2505 interferometer at Teralba Colliery, NSW). https://storyplace.org.au/story/a-necessary-invention/

[7] E. R. Peck, S. Huang, “Refractivity and dispersion of hydrogen in the visible and near infrared,” J. Opt. Soc. Am. 67(11), 1550 (1977); dispersion formula summarized via refractiveindex.info. https://opg.optica.org/josa/abstract.cfm?uri=josa-67-11-1550; https://refractiveindex.info/?shelf=main&book=H2&page=Peck

[8] Wikipedia, “List of refractive indices” (n(H₂) = 1.000132, n(air) = 1.000293, both at 0 °C/1 atm, 589.29 nm). https://en.wikipedia.org/wiki/List_of_refractive_indices

[9] “Ultrasensitive Gas Refractometer Using Capillary-Based Mach–Zehnder Interferometer,” Sensors 20(4), 1191 (2020); PMC7070851. https://pmc.ncbi.nlm.nih.gov/articles/PMC7070851/

[10] “Highly Sensitive Optical Sensor for Hydrogen Gas Based on a Polymer Microcylinder Ring Resonator,” Sens. Actuators B (2020), and related microbottle-resonator Pd-WO₃/PDMS hydrogen-sensor synthesis (0.0312–1% H₂ range, up to 140 pm/% sensitivity, ~60 ppm LOD reported for one design). https://www.sciencedirect.com/science/article/abs/pii/S0925400520301532

[11] Distran, “Ultrasound Imaging Technology for Gas Leak Detection” and company technology pages. https://distran.swiss/en/technology/ ; https://distran.swiss/en/home/

[12] “Highly sensitive hydrogen sensor based on an in-fiber Mach-Zehnder interferometer with polymer infiltration and Pt-loaded WO₃ coating,” Opt. Express 29(3), 4147 (2021). https://opg.optica.org/oe/fulltext.cfm?uri=oe-29-3-4147&id=446897

[13] “In-line Mach-Zehnder Interferometer and FBG with Pd film for simultaneous hydrogen and temperature detection,” Sens. Actuators B 202, 893–896 (2014). https://www.sciencedirect.com/science/article/abs/pii/S0925400514006819

[14] Malvern Panalytical, “SciAps Z-903 Handheld LIBS Analyzer” product page. https://www.malvernpanalytical.com/en/products/product-range/z-series/z-903

[15] SciAps / Malvern Panalytical Z-901 product documentation (element exclusion list including H). https://www.sciaps.com/products/libs/z-901 (link fixed — original German SciAps Z-series URL no longer resolves)

[16] “Reference measurements of hydrogen’s dielectric permittivity,” Molecular Physics 107(15) (2009); NIST publication record. https://www.nist.gov/publications/reference-measurements-hydrogens-dielectric-permittivity; https://www.tandfonline.com/doi/abs/10.1080/00268970902980052

[17] NIST, “Dielectric Permittivity of Eight Gases Measured with Cross Capacitors,” Int. J. Thermophys. https://www.nist.gov/publications/dielectric-permittivity-eight-gases-measured-cross-capacitors

[18] MinEx CRC, “Pre-commercial product brochure: Laser-induced Breakdown Spectroscopy (LIBS)” downhole logging tool. https://minexcrc.com.au/news/pre-commercial-product-brochure-laser-induced-breakdown-spectroscopy-libs/

[19] SciAps, “Making a case for handheld LIBS in mineral exploration” (company blog/application article). https://www.sciaps.com/post/making-a-case-for-handheld-libs-in-mineral-exploration

[20] SciAps Z-903+ handheld LIBS rental listing (pricing context: $1,990/week, $3,990/2 weeks, $5,990/month). https://shop.xrfcompany.com/products/metal-fabrication-handheld-libs-rental-analyzer-sciaps-z903-for-carbon-applications

[21] “Hydrogen Leak Detection Using Laser-Induced Breakdown Spectroscopy,” Appl. Spectrosc. 59(3), 348–353 (2005); PubMed 15901317. https://pubmed.ncbi.nlm.nih.gov/15901317/

[22] N. H. Thomas et al., “Characterization of Hydrogen in Basaltic Materials With Laser-Induced Breakdown Spectroscopy (LIBS) for Application to MSL ChemCam Data,” J. Geophys. Res. Planets 123(8) (2018). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2017JE005467

[23] S. S. Harilal, E. J. Kautz, “Analysis of hydrogen and lithium isotopes using Laser-Induced Breakdown Spectroscopy,” Proc. SPIE High-Power Laser Ablation VIII (2024); PNNL publication record. https://www.pnnl.gov/publications/analysis-hydrogen-and-lithium-isotopes-using-laser-induced-breakdown-spectroscopy

[24] “Novel Calibration Approach for Monitoring Aerosol Hydrogen Isotopes Using Laser-Induced Breakdown Spectroscopy for Molten Salt Reactor Off-Gas Streams,” Sensors 23(24), 9797 (2023); PMC10747517. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10747517/

[25] “Laser-induced breakdown spectroscopy for the measurement of air humidity using spectral line intensity ratio,” PMC12308682. https://pmc.ncbi.nlm.nih.gov/articles/PMC12308682/

[26] “On the determination of plasma electron number density from Stark broadened hydrogen Balmer series lines in Laser-Induced Breakdown Spectroscopy experiments,” Spectrochim. Acta B (2013). https://www.sciencedirect.com/science/article/abs/pii/S0584854713001535

[27] “Laser plasma diagnostics and self-absorption measurements of the Hβ Balmer series line,” Spectrochim. Acta B (2015). https://www.sciencedirect.com/science/article/abs/pii/S002240731530042X

[28] C. G. Parigger, “Self-Absorption Corrections to Hydrogen and Aluminum…” Univ. of Auburn/IRAMP. https://www.auburn.edu/cosam/departments/physics/iramp/6_2/surmick_parigger.pdf

[29] Search-result synthesis on standoff LIBS demonstrations (double-pulse configurations to ~50–70 m range; frequency-doubled Nd:YAG remote-detection work to ~120 m for hazardous-chemical LIF), incl. Oxford Instruments/Andor, “Stand-off LIBS for remote detection of explosives.” https://andor.oxinst.com/learning/view/article/stand-off-libs

[30] FLIR/Teledyne, “FLIR GF343 Optical Gas Imaging Camera” (discontinued) and “Avoid an outage with new hydrogen leak detection technology” application note (CO₂ tracer-gas method for H₂ leak visualization). https://www.flir.com/products/gf343/ ; https://www.flir.com/discover/instruments/gas-detection/avoid-an-outage-with-new-hydrogen-leak-detection-technology/

[31] “Catalytic IR thermal imaging-based hydrogen detection,” Sens. Actuators B (2025). https://www.sciencedirect.com/science/article/abs/pii/S0925400525000115

[32] Fluke, “ii910 Acoustic Imaging Camera: Precision Acoustic Imager” product page and technical datasheet. https://www.fluke.com/en-us/product/industrial-imaging/precision-acoustic-imager-ii910; datasheet: https://www.fluke-direct.com/pdfs/cache/www.fluke-direct.com/ii900/datasheet/ii900-datasheet.pdf

[33] Fluke ii910 retail pricing (new unit, $28,549.99). https://business.walmart.com/ip/Fluke-ii910-Precision-Acoustic-Imager/1839040869

[34] “Hydrogen leakage diagnosis and visualization of density and pressure using Background Oriented Schlieren with Dual-Channel MGR-Net,” Int. J. Hydrogen Energy (2025). https://www.sciencedirect.com/science/article/abs/pii/S036031992504594X

[35] “Visualization of hydrogen leak for electro-hydrogen coupled system based on Background Oriented Schlieren,” Process Saf. Environ. Prot. 175, 437 (2023). https://www.sciencedirect.com/science/article/abs/pii/S0957582023003890

[36] Search-result synthesis on deep-learning-enhanced BOS methods for hydrogen leak imaging (dual-channel network architectures), consistent with ref [34].

[37] “Measurement of the concentration distribution of hydrogen jets using adaptive stream stripe-background oriented schlieren (ASS-BOS),” Int. J. Hydrogen Energy (2024). https://www.sciencedirect.com/science/article/abs/pii/S0360319924022985

[38] “Review of the Application of Schlieren Systems in the Field of Hydrogen and Hydrogen Blends,” Energies 19(7), 1691 (2026), doi:10.3390/en19071691 (full text behind access restrictions in this research pass; cited for title/existence and topical scope only). https://www.mdpi.com/1996-1073/19/7/1691

[39] “High-precision gas refractometer by comb-mode-resolved spectral interferometry,” PMC6219595. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6219595/

[40] “Measurements of refractive-index dispersion in gases with a two-wavelength one-path interferometer,” J. Opt. Soc. Am. 73(12), 1836 (1983). https://opg.optica.org/josa/abstract.cfm?uri=josa-73-12-1836

[41] “Hydrogen Gas Detection by Mini-Raman Lidar,” IntechOpen chapter (the chapter’s DPSS-laser compact-lidar sub-system is reported at 1–100% H₂ concentration range and <50 m detectable distance; the mini-Raman lidar variant specifically is reported at 20–30 m range) — cited here only to distinguish Raman lidar (selective, vibrational) from Rayleigh-scattering approaches (non-selective); full Raman physics covered in Raman spectroscopy. https://www.intechopen.com/chapters/59801

[42] “Experimental Comparison of Functionality between the Main Types of Methane Measurement Sensors in Mines,” Energies 16(5), 2207 (2023), doi:10.3390/en16052207 (interferometric-sensor reliability/cross-sensitivity claim sourced from search-result snippet of this paper; direct full-text fetch was blocked in this research pass). https://www.mdpi.com/1996-1073/16/5/2207

[43] “A Versatile Capacitive Sensing Platform for the Assessment of the Composition in Gas Mixtures,” Micromachines 11(2), 116 (2020); PMC7074232. https://pmc.ncbi.nlm.nih.gov/articles/PMC7074232/

[44] “Development of a subsurface LIBS sensor for in situ groundwater quality monitoring with applications in CO2 leak sensing in carbon sequestration,” PMC6418306. https://pmc.ncbi.nlm.nih.gov/articles/PMC6418306/

[45] MSA Safety, “Ultrasonic Leak Detector Observer-I — Acoustic Gas Detection” product page. https://us.msasafety.com/Fixed-Gas-%26-Flame-Detection/Gas-Detectors/Observer%C2%AE-i-Ultrasonic-Gas-Leak-Detector/p/000140006900001005

[46] Distran / distributor product pages for Ultra Pro and Ultra Pro Max sensitivity and range specifications. https://keysfortomorrow.com/distran-ultrasonic-cameras-gas-leak-detection/ ; https://en.gazdetect.com/distran-ultra-pro-max/

[47] MSA Safety, “Rethinking Gas Leak Detection with Ultrasonic Technology” (operational pressure threshold context, ≥2 bar/29 psi for reliable ultrasonic leak generation). https://blog.msasafety.com/fgfd-rethinking-gas-leak-detection-with-ultrasonic-technology/