Aleksandar Ćirić

Here, we evaluate a full-spectrum, data-driven readout framework, termed Luminescence Thermometry 2.0, in which normalized emission spectra are used directly as inputs to temperature regression. We benchmark this approach against conventional luminescence-intensity ratio thermometry using the same experimental dataset obtained from the upconversion and downshifting emissions of Er³⁺,Yb³⁺-co-doped yttrium oxyfluoride nanoparticles. Relative to the best-performing conventional LIR calibration, full-spectrum regression reduces the absolute prediction bias by up to approximately 2.5-fold and the prediction spread by 24% for visible green upconversion. For near-infrared emission, the corresponding reductions reach approximately 2.2-fold and 8.6-fold, respectively. The approach reduces feature-selection bias and exploits temperature-encoded information that conventional analysis discards. Shapley additive explanations for the best-performing Gaussian process regression model identify the Er³⁺ spectral regions that dominate temperature prediction, linking the improved performance to physically meaningful emission features. These results demonstrate improved temperature readout using full-spectrum regression for YOF:Er³⁺/Yb³⁺ under the investigated conditions. Matched synthetic-background tests showed that training-set augmentation reduced background-induced set-point bias, with effects that depended on the regression method and normalization.

Luminescence thermometry has evolved through decades of research focused on optimising materials and on extracting temperature information from isolated spectral features such as luminescence intensity ratios, bandwidth, line shift and excited-state lifetime. Despite extensive material development, these conventional methods remain fundamentally limited by construction: only a small subset of pre-selected spectral features is exploited, while the bulk of the temperature-relevant information encoded in the full spectrum is systematically discarded. A paradigm shift is presented here: Luminescence Thermometry 2.0 (LT 2.0), implemented through the newly developed SpectraSensML platform, in which machine learning regression operates on the entire emission spectrum to deliver temperature readout. The approach is demonstrated on a Yb³⁺-doped phosphor emitting in the near-infrared biological transparency window across 100 to 700 K. Yb³⁺ is a particularly demanding case: only the single ²F_5/2 multiplet emits, and its weakly thermally coupled Stark sub-levels yield modest sensitivity under conventional intensity-ratio thermometry. A total of 27 regression algorithms drawn from four families, namely tree ensembles, physics-aware regression models, kernel and instance methods, and neural networks, are systematically benchmarked. A sensor-fusion estimator that combines the first three principal components reaches an average root-mean-square error of 0.36 K on an unseen-temperature test set, a seven-fold improvement over the best luminescence intensity ratio variant. Standard normal variate (SNV) normalisation is identified as the most effective preprocessing strategy because it isolates the band-shape deformations that encode temperature. Single-component approaches that rely on the first principal component alone are shown to be quantitatively sub-optimal: multi-component regressors that exploit the first three principal components reduce the temperature uncertainty by close to an order of magnitude. The structural reason behind the failure of decision-tree ensembles on unseen temperatures is explained: their piecewise-constant predictions cannot interpolate beyond training set-points. The open-source SpectraSensML application used to obtain the results is released alongside the manuscript to enable reproducible community benchmarks.

Al₂O₃:Cr³⁺ coatings were synthesized on tungsten substrates by plasma electrolytic oxidation in a phosphate-aluminate electrolyte containing dispersed Cr₂O₃ nanoparticles, and their structural, photoluminescent, and temperature-sensing properties were investigated. The coatings exhibited a typical porous PEO morphology with a uniform thickness of approximately 31 μm, and EDS analysis confirmed the incorporation of Cr species from the electrolyte, with Cr content increasing with the concentration of Cr₂O₃ particles. XRD analysis showed that the coatings were composed predominantly of α-Al₂O₃, with minor contributions from metastable γ-Al₂O₃, confirming that our previously established process for forming the thermodynamically stable α-Al₂O₃ phase directly on a non-aluminum substrate remains robust upon the introduction of dopant nanoparticles. The Al₂O₃:Cr³⁺ coatings displayed characteristic ruby-like photoluminescence, including broad excitation bands associated with the ⁴A₂⟶⁴T₁ and ⁴A₂⟶⁴T₂ transitions and sharp R-line emission arising from the spin-forbidden ²E⟶⁴A₂ transition. The strongest emission was obtained for coatings prepared with 0.05 g/L Cr₂O₃, while higher concentrations resulted in concentration quenching. Temperature-dependent photoluminescence revealed two complementary thermometric mechanisms: R-line spectral shifting and thermally induced redistribution between the ²E and ⁴T₂ emissions. The deconvolution-based intensity-ratio approach provided a stronger temperature response than simple spectral partitioning, demonstrating the potential of PEO-derived Al₂O₃:Cr³⁺ coatings on tungsten as robust luminescent temperature-sensing layers.

The Judd–Ofelt theory governs the intensity of 4fn ↔ 4fn transitions of rare earth ions in terms of an admixed induced electric dipolar character from higher energetic even parity states by the crystal field and successfully reproduces experimental absorption and emission line strengths. Traditionally, the Judd–Ofelt theory requires three intensity parameters that have to be estimated from an experimental absorption spectrum. That approach is fundamentally problematic with non-transparent or polycrystalline materials, and particularly with Pr3+ ions due to the low energy difference between the excited 4f15d1 configuration and the 4f2 spin–orbit levels, which is in contrast with the closure approximation necessary for the validity of the Judd–Ofelt theory and usually leads to non-physical intensity parameters. In this work, an algorithm based on experimental emission spectra is presented that resolves these limitations. For that purpose, steady-state and time-resolved emission spectra at liquid nitrogen temperatures (T = 77 K) as well as decay times of the 1D2 and 3P0 levels were measured. The method was tested on hydrothermally prepared Sr2LaF7:Pr3+ as a representative example as this compound shows emission from both the 3P1,0 and 1D2 levels at room temperature with measured decay times of the 3P0 and 1D2 level of 59 and 225 μs, respectively. From the intensities of the 3P0 → 3H4,6, 3F2-based emission, and the 1D2-based decay time, Judd–Ofelt parameters of Ω2 = 2.34 × 10−20, Ω4 = 5.84 × 10−20, and Ω6 = 21.68 × 10−20 cm2 were obtained. The calculated values showed very good agreement with the diffuse reflectance measurements and allowed the prediction of Pr3+ near infrared emission intensities.

Advancing measurement precision and extending the temperature range are key goals in luminescence thermometry. Traditional single‐parameter methods often underperform. Sensor fusion (SF), a statistical tool widely used in fields like autonomous vehicles and medical imaging, is applied to luminescent thermometry by combining multiple sensor probes or treating each temperature‐dependent parameter as a separate sensor. This approach consistently enhances precision and extends the temperature range, with fused precision equaling the sum of individual precisions. SF using inverse variance weighting surpasses traditional linear regression models due to its adaptability, achieving maximum performance with any sensor material. It works with both time‐resolved and steady‐state readouts, using single or multiple excitation sources. Computer simulations and experiments validate this method. For Sm ²⁺ , combining lifetime and intensity ratio measurements significantly improves precision across the entire range. Fusion of Mn ⁴⁺ , Ho ³⁺ , and Cr ³⁺ lifetimes expands the temperature range to 300–650 K. For Yb ³⁺ /Er ³⁺ upconversion green and red emission lifetimes, precision improves across all temperatures. However, Mn ⁵⁺ shows limited improvement due to the dominance of precision in line‐shift measurements, highlighting a limitation of the approach. Overall, SF demonstrates its potential to revolutionize luminescence thermometry by enhancing precision and usability across diverse conditions.