Potassium Vapor Spectra Turn a Blue Laser Into a Precise Atomic Thermometer

Temperature is one of those quantities that seems easy to measure until the thing you need to measure is a cloud of atoms inside a heated glass cell. A thermocouple strapped to the outside reads the wall, not the vapor, and thermal gradients between the copper heater block, the glass, and the gas mean the number on the display can quietly drift away from the temperature the atoms actually experience. A study published in Applied Physics B reports a way around this problem: reading the temperature directly from the atoms themselves, using nothing more exotic than a blue diode laser, a 10 centimeter vapor cell, and a well-tested model of how light is absorbed by moving atoms.

The work, led by Prosenjit Majumder, focuses on a comparatively neglected pair of transitions in potassium: the 4S to 5P lines at around 405 nanometers, known as the D1 and D2 lines. Unlike the heavily studied 4S to 4P D lines used in nearly every potassium magneto-optical trap, the higher-lying 5P transitions have received far less experimental attention, particularly when it comes to temperature-dependent spectroscopy and quantitative modeling. According to the paper, a systematic treatment of the 4S to 5P line with full multi-isotope structure, verified across a broad temperature range, had not previously been reported to the authors’ knowledge.

The experimental recipe is deliberately simple. A home-built extended-cavity diode laser in the Littrow configuration, built around a 405 nanometer laser diode and a 3600-lines-per-millimeter holographic grating, delivered roughly 1.2 milliwatts of single-mode light. Achieving stable, mode-hop-free operation at this wavelength is notably harder than at the near-infrared wavelengths familiar from rubidium or cesium experiments, partly because the diode facet lacks an anti-reflection coating and multimode behavior emerged at higher drive currents. The probe beam was attenuated, focused through the heated cell containing potassium at natural isotopic abundance, and detected with a photodiode, while a scanning Fabry-Pérot interferometer with a 1.50 gigahertz free spectral range provided frequency calibration for each laser scan of roughly 6 gigahertz.

A central experimental question was where the so-called weak-probe regime begins: the range of intensities where the transmission spectrum no longer depends on how bright the light is. Below that threshold, the absorption reflects the intrinsic atomic response; above it, saturation and optical pumping distort the profile. In alkali vapors this boundary is genuinely non-trivial, because atoms excited by the probe can decay into hyperfine ground states that the probe cannot address, trapping them where they stop absorbing. The measurements placed the onset of the plateau at approximately 1 × 10⁻² times the saturation intensity for the D2 line and about 5 × 10⁻² for D1, a difference the authors attribute to the open nature of the D1 transition, which is more susceptible to optical pumping, whereas the D2 line includes a partially closed cycling transition from F=2 to F=3 that keeps atoms resonant and produces stronger absorption.

To turn spectra into temperatures, the team fitted the Doppler-broadened transmission profiles with ElecSus, a theoretical framework originally developed for rubidium and extended here to potassium. The model computes the complex electric susceptibility of the vapor by summing over all allowed electric-dipole couplings between hyperfine sublevels, with each transition convolved with a temperature-dependent Gaussian to capture Doppler broadening. Crucially, the fits incorporated experimentally determined isotope shifts from high-resolution Doppler-free measurements by Behrle and colleagues: 235.0 megahertz for potassium-40 and 454.2 megahertz for potassium-41, relative to potassium-39, held as fixed inputs. The authors report that using these measured values enabled accurate reproduction of the absorption spectra, validating their incorporation into the framework.

The temperature dependence enters the model in two coupled ways, and this is what makes the method more than a linewidth measurement. The Doppler width of each transition grows with temperature, but the atomic number density, calculated from a Nesmeyanov-type vapor-pressure relation, rises exponentially, so the depth of the absorption feature is itself a strong thermometer. The temperature determination is therefore constrained simultaneously by the spectral linewidth and the temperature-dependent optical depth, providing what the authors describe as a self-consistent spectroscopic determination of the vapor temperature rather than an estimate based on the Doppler width alone.

Measurements were taken across the range from 300 to 403 kelvin, with transmission spectra recorded at cell temperatures between roughly 93.5 and 123.8 degrees Celsius for both transitions, at probe intensities near 2 × 10⁻³ of the saturation intensity, comfortably inside the weak-probe regime. The ElecSus fits reproduced the measured spectra with root-mean-square errors on the order of 10⁻³. The residuals showed smooth systematic structure, which the authors trace to nonlinearities in the frequency axis calibration, since the limited scan range provides only a handful of Fabry-Pérot markers, and possibly to weak etalon effects from residual reflections. These distortions slightly affect the fitting but remain small compared with the overall absorption signal, and no significant influence on the extracted temperatures was observed within the experimental uncertainty.

The spectroscopic temperatures were benchmarked against an RS PRO Type-K thermocouple mounted between the cell and its copper heating blocks, specified to about ±1.5 degrees Celsius. Fitting the relationship between the two datasets with a linear regression gave a slope consistent with unity, and the mean temperature difference between the spectroscopic and thermocouple values was approximately 2.5 degrees Celsius over the investigated range, less than 1 percent of the absolute temperature. The authors caution that the regression intercept should not be read as a constant offset; instead, the discrepancy likely reflects a combination of systematic effects on both sides, including thermal gradients, uncertainties in the potassium vapor-pressure model, frequency calibration, and transmission normalization, rather than a bias in either method alone.

The precision of the technique varies with temperature in a predictable way. From the covariance matrix of the fits, the typical one-sigma uncertainties were ±1.2 degrees Celsius below about 330 kelvin, ±0.6 degrees in the mid-range, and ±0.4 degrees above 370 kelvin. The pattern follows directly from the physics: at low temperatures the atomic density is small, absorption is weak, and the slope of the transmission minimum with respect to temperature is shallow, so small errors in the signal translate into larger errors in temperature. At higher temperatures the vapor becomes optically thicker, spectral features sharpen, and the method becomes most precise, capable under favorable conditions of resolving sub-degree temperature variations. The dominant experimental noise source was electronic, arising from laser intensity fluctuations, photodiode dark noise, and oscilloscope readout in a nanowatt-scale weak-probe beam, and the authors note that better detectors could push the uncertainties lower still.

Beyond the immediate result, the study quantifies how temperature sensitivity is distributed across the spectrum, identifying the detuning regions and temperature regimes where transmission changes most per degree, which the authors say enables identification of optimal operating conditions for practical thermometry. The two lines emerge as complementary tools: the D2 transition, with its higher degeneracy and stronger absorption, offers superior temperature sensitivity, while the open D1 line provides a stringent test of the weak-probe limit. The authors also point toward applications beyond thermometry, since the same weak-probe characterization and optical-pumping dynamics underpin work on all-optical switching, quantum memory schemes, and precision measurements in thermal alkali vapors. For experiments that need accurate atomic densities, from nonlinear optics to magnetometry, the message is that the atoms will tell you their own temperature, if you ask in the right spectral language.

Subject of Research: Doppler-broadened absorption spectroscopy of potassium 4S to 5P transitions for spectroscopic thermometry using the ElecSus model

Article Title: Atomic thermometry using 4S \(\rightarrow \) 5P absorption spectroscopy in potassium vapor

Article References: Majumder, P. (2026). Atomic thermometry using 4S $$\rightarrow $$ 5P absorption spectroscopy in potassium vapor. Applied Physics B, 132(10), Article 106. https://doi.org/10.1007/s00340-026-08716-y

Image Credits: AI Generated

DOI: 10.1007/s00340-026-08716-y

Keywords: atomic thermometry, potassium vapor, absorption spectroscopy, Doppler broadening, ElecSus, weak-probe regime, D1 and D2 lines, 405 nm diode laser, isotope shifts, optical pumping, vapor cell, quantum technologies