Free tools for sensor developers
Tools
Calculators for people who build gas sensors. The first one answers the question that comes before any design work: which kind of sensor should detect your target at all? The rest answer the design questions every metal-oxide sensor raises: how big the grains should be, how thick the film can be, whether the additive is a metal or an oxide at operating temperature, and how much of the surface it covers.
Which detection technology fits your target?
Pick a molecule and tell us where it is and how little of it you need to see. You get a ranked list of sensor technologies, with the reason for each verdict and the catch to watch for. Every verdict comes from a stated property of the molecule, and when a property is missing the tool says “can’t tell” instead of guessing.
Molecule not on the list?
Tell us what you need to detect. We look up its properties, run it through our calculations, add it to this tool and email you the result. Requests also tell us which sensors the market is missing, so they shape what we build next.
How the ranking works, and where the numbers come from
Vapor pressure decides whether a molecule can reach a gas sensor at all: curated measured values where we hold them, otherwise standard vapor-pressure correlations (the open-source thermo/chemicals libraries, which bundle CRC and NIST data). Ionization energy and proton affinity come from the NIST Chemistry WebBook and link to the entry. Where NIST has no value, Sniffi computed one and labels it so. Ionization energies come from density-functional theory (ωB97X/def2-TZVP), checked against 39 molecules NIST has measured: typical error 0.15 eV. Proton affinities come from the same method on fast semi-empirical geometries, checked against 27 NIST molecules: typical error 8 kJ/mol. For aromatic molecules a ring carbon can take the proton more strongly than the site we modelled, so those values are shown as lower bounds (≥). A photoionization detector sees a molecule only if its ionization energy is below the lamp’s photon energy (10.6 eV for the standard lamp). Electro-active groups and “can hot surface oxygen react with it?” come from substructure rules on the molecule. Infrared bands are textbook group frequencies; a band inside the water or CO2 absorption regions is flagged. Flammability limits come from the IEC 60079-20-1 and NFPA 497 tables, and exposure limits from the Ontario occupational limits table.
Typical detection ranges and prices are order-of-magnitude figures for commodity parts. A specific product can beat them.
For every molecule this tool knows, the same rules run for every technology. A missing property gives “can’t tell yet”, never a default. Treat the output as a first screen. The next step is a lab test or a conversation with us.
Film designer: grain size, thickness and electrodes
Three questions every metal-oxide film has to answer. Are the grains small enough to be fully depleted? How thick can the film be before the gas stops reaching the bottom? And will the resistance be something your electronics can read?
A Grain size against the Debye length
Method
Debye length LD = √(εrε0kT / q²Nd). Regimes after Xu, Tamaki, Miura and Yamazoe (1991). When D < 2LD the whole grain is depleted and every adsorbed molecule moves the resistance (grain control, highest sensitivity). From about 2LD to 6LD the necks between grains set the resistance. Above that the grain boundaries dominate and sensitivity falls. The 6LD boundary is approximate. The permittivity presets are typical literature values, so use your own if you have them.
B How thick before the gas stops reaching the bottom?
Method
Gas diffuses into the pores in the Knudsen regime, DK = (2/3)·r·v̄ with v̄ = √(8RT/πM), and reacts on the pore wall as it goes (Sakai and Yamazoe diffusion–reaction model). The chance of reacting per wall collision is s = θ·exp(−Ea/kT). For cylindrical pores the Thiele modulus is then m = L·√(3s/4)/r, and the fraction of the film that does useful work is U = tanh(m)/m. Past the half-utilization thickness L½, extra thickness mostly adds baseline conductance. The counterintuitive part is that a more reactive film is used less deeply, because it consumes the gas near the surface.
C Interdigitated electrodes: what resistance will you read?
Method
Each gap is a resistor of length g, width equal to the finger overlap and height equal to the film thickness, and the N−1 gaps sit in parallel: R = ρ·g / (t·L·(N−1)). The cell constant K = g/(L·(N−1)) multiplies the film’s sheet resistance ρ/t. This holds while the film is thinner than the gap. Thicker films also conduct through the material above the fingers, which lowers R.
Metal or oxide at your operating temperature?
A “Pd-doped” sensor at 300 °C in air may actually hold PdO, and the two behave very differently. Pick an additive, a temperature and an oxygen level to see which form is thermodynamically stable.
Method, accuracy and limits
Dopant loading converter
Papers quote loadings in wt%, at% or mol%, and none of those tells you how much of the surface the additive actually covers. Convert between them, and see the coverage the number implies for your grain size or measured surface area.
Method
Spherical grains of diameter D give a specific surface area of SSA = 6/(ρ·D). Surface cation-site densities were measured on Sniffi’s own relaxed surface models (one face; SnO2 is rutile (110)). Coverage = dispersion × additive atoms per gram ÷ surface sites per gram. Real powders cluster, so true coverage is lower than this upper bound. A result above 100% means the additive cannot all be single atoms: it has to form particles or a shell.
These tools are free to use and give engineering estimates. They are not a substitute for measurement. Found an error, or want a tool we don’t have? Tell us.