Photonica

Tunable diode laser absorption spectroscopy (TDLAS)

A gas-sensing method that tunes a narrow-linewidth laser across a single molecular absorption line and measures the dip in transmitted power. With wavelength modulation and 2f detection it resolves fractional absorbances of about 10⁻⁵ or better, enough for ppm-level methane near 1.65 µm and ppb levels with mid-infrared lasers.

Lab practiceUpdated October 2026

Tunable diode laser absorption spectroscopy (TDLAS) measures the concentration, temperature or pressure of a gas by sweeping a single-frequency laser across one rotational-vibrational absorption line and recording how much light the gas removes. The laser linewidth, typically a few MHz for a DFB laser, is far narrower than the gas line, which at atmospheric pressure has a width of a few GHz, so the instrument traces the line shape directly. Common targets are methane near 1653.7 nm (about 6047 cm⁻¹), carbon dioxide near 1572 nm and 2004 nm, water near 1392 nm and oxygen near 760 nm, all reachable with near-infrared DFB or VCSEL diode lasers; stronger fundamental bands in the mid-infrared use quantum cascade and interband cascade lasers.

Direct absorption

The transmitted power follows the Beer–Lambert law. Written with natural logarithms, the absorbance at optical frequency ν\nu is

A(ν)=−ln⁡I(ν)I0(ν)=S ϕ(ν) N L,A(\nu) = -\ln\frac{I(\nu)}{I_0(\nu)} = S\,\phi(\nu)\,N\,L,

where SS is the line strength (tabulated in the HITRAN database), ϕ(ν)\phi(\nu) the normalized line shape, NN the number density of the absorbing molecule and LL the path length. Integrating AA over the line removes the dependence on the line shape, so the integrated absorbance gives NN directly once SS and LL are known; this is the calibration-free character of the method.

Worked example. A peak absorbance of 10−310^{-3} corresponds to a transmission of 0.9990, a dip of 0.1% on the detector. Measuring that absorbance to 2% requires the baseline power to be known to 2×10−52 \times 10^{-5} of its value, a level at which laser intensity noise, optical fringes and the slope of the laser's power during the sweep all matter. For methane at 1 atm, the strongest lines near 1653.7 nm give a peak absorbance of order 10−510^{-5} per ppm·m, so ambient methane (about 1.9 ppm) in a 30 m multipass cell produces a peak absorbance between 10−410^{-4} and 10−310^{-3}.

The baseline I0(ν)I_0(\nu) is obtained by fitting a polynomial to the wings of the scan on either side of the line. In practice the sweep is produced by ramping the DFB injection current, which tunes the frequency through heating of the active region, while the heatsink temperature sets the center wavelength; a temperature tuning coefficient of about 0.1 nm/K corresponds to about 11 GHz/K at 1653.7 nm.

Line widths

The absorption coefficient profile has a Gaussian Doppler component and a Lorentzian collisional component, combined in a Voigt profile. For methane at 1653.7 nm and 296 K the Doppler full width is about 0.56 GHz (0.019 cm⁻¹). Air-broadening coefficients for small molecules are typically 0.05–0.1 cm⁻¹ per atmosphere (half width), so at 1 atm collisional broadening dominates: a half width of 0.07 cm⁻¹ is about 2.1 GHz. Reducing the pressure in a sample cell narrows the line toward the Doppler limit, separates overlapping lines and increases the peak absorbance per molecule.

Wavelength modulation spectroscopy and 2f detection

Most field instruments add a sinusoidal modulation at frequency ff, typically tens of kHz, to the slow current ramp. The absorption line converts this frequency modulation into intensity modulation at harmonics of ff; a lock-in amplifier demodulates the detector signal at 2f2f. The 2f signal is proportional to the second derivative of the line shape for small modulation and peaks at the line center, while a linear baseline slope contributes almost nothing at 2f2f. Shifting detection from DC to tens of kHz also moves it away from the low-frequency excess noise of the laser and electronics. For a Lorentzian line the 2f peak is largest at a modulation amplitude of about 2.2 times the line half width. Normalizing the 2f signal by the 1f signal removes the dependence on received power, which is important when windows foul or the beam path changes.

Reported minimum detectable absorbances are of order 10−310^{-3}–10−410^{-4} for simple direct absorption and 10−510^{-5}–10−610^{-6} with wavelength modulation in well-engineered systems; residual etalon fringes from windows and lenses usually set the floor, well above the detector noise.

Detectors and wavelength regions

Near-infrared systems use InGaAs photodiodes at room temperature. In the mid-infrared, fundamental vibrational bands are typically two to four orders of magnitude stronger than the near-infrared overtones and combination bands, so quantum cascade lasers (above about 4 µm) and interband cascade lasers (3–6 µm) combined with HgCdTe or InAsSb detectors reach ppb concentrations, for example methane at 3.3 µm or 7.7 µm and carbon monoxide at 4.6 µm.

Common questions

What is TDLAS used for?

Industrial process and combustion monitoring, natural-gas leak detection, environmental monitoring of greenhouse gases, breath analysis and in-situ measurements of gas temperature from the ratio of two lines with different lower-state energies.

How is TDLAS different from FTIR?

FTIR spectroscopy records a broad spectrum covering many molecules at once with a thermal source; TDLAS examines one or a few lines with a laser, giving higher spectral resolution, faster response and lower detection limits for the chosen species.

Does TDLAS need calibration gas?

Direct absorption with known line strength, path length, pressure and temperature is in principle calibration-free. Wavelength modulation systems are usually checked against reference gases, although calibration-free 1f-normalized methods exist.

References: R. K. Hanson, R. M. Spearrin, C. S. Goldenstein, Spectroscopy and Optical Diagnostics for Gases (Springer, 2016); J. Reid, D. Labrie, Appl. Phys. B 26, 203 (1981); D. S. Bomse, A. C. Stanton, J. A. Silver, Appl. Opt. 31, 718 (1992); P. Werle, Spectrochim. Acta A 54, 197 (1998); I. E. Gordon et al., J. Quant. Spectrosc. Radiat. Transf. 277, 107949 (2022).