Publications

2026

• Probabilistic Computing with Neuromorphic Elements
  
  • Li Zonglong
  • Calvet Laurie E
  , 2026. The paper introduces a new low power and low transistor count approach to realizing a Bayesian engine that is simulated in TSMC 130 nm technology. Analog spiking neuron circuits produce deterministic streams that represent probabilities for computation. Synapse-like circuits encode the conditional probability distribution of features directly into their parameters and enable the determination of likelihood probabilities. The paper explores Gaussian and arbitrary distribution circuits, showing how a variety of different features can be considered. The calculation of the posterior is done using either an analog AND or a Muller C-element. While both are found to produce similar accuracies, the Muller C-element implementation enables faster computations. The final inference is accomplished with a time-tofirst-spike neuron that determines the most likely class. The architecture eliminates the need for random number generators, memory elements, and analog-to-digital conversion. It is validated on a breast cancer detection task, achieving classification accuracies of 73-90%, comparable to equivalent implementations in software. The system exhibits a compact footprint and operates in the μW range. This work provides a proof-of-concept for scalable, interpretable, compact and low-power probabilistic inference framework for edge computing.
• Refractive indices of photochemical haze analogs for Solar System and exoplanet applications: A cross-laboratory comparative study between the PAMPRE and COSmIC experimental setups
  
  • Drant Thomas
  • Sciamma-O’brien Ella
  • Jovanovic Lora
  • Perrin Zoé
  • Maratrat Louis
  • Vettier Ludovic
  • Garcia-Caurel Enric
  • Brubach Jean-Blaise
  • Wooden Diane H
  • Roush Ted L
  • Ricketts Claire L
  • Rannou Pascal
  Astronomy & Astrophysics - A&A, EDP Sciences, 2026, 706, pp.A167. Previous observations of Titan, Pluto, and Solar System gas giants, along with recent observations of exoplanet atmospheres with the James Webb Space Telescope, have taught us that photochemical hazes are ubiquitous and form in a variety of temperature, gas composition, and irradiation environments. Despite their crucial role in understanding their impact on observations and on the planetary radiative budget, the refractive indices of these haze particles remain unknown and are strongly influenced by changes in gas-phase chemistry. In this study, we performed a cross-laboratory investigation to assess the effect of the experimental setup and gas composition on the refractive indices of Titan, Pluto, and exoplanet haze analogs. We report new data in a broad spectral range from UV to far-IR (up to 200 µm) for future use in climate models and retrieval frameworks. We compare the refractive indices of laboratory haze analogs produced from six different gas compositions, in which we varied the relative abundances N 2 /CH 4 and CH 4 /CO in the initial gas mixture, using the PAMPRE (LATMOS, France) and COSmIC (NASA Ames Research Center, USA) experimental setups. We observed strong variations in the k values in the spectral range from UV to near-IR between the different analogs, which are caused by both the experimental setup and changes in the gas N 2 /CH 4 ratio. We find that the gas N 2 /CH 4 ratio has a stronger influence on the haze refractive indices in the entire spectral range compared to the gas CH 4 /CO ratio. The experimental setup is the primary factor affecting the refractive indices, confirming that the gas residence time, irradiation, pressure, and gas temperature are important parameters influencing the composition of the solid analog. The higher n and k values in the UV-visible range, along with the stronger amine, alkene, aromatic, and/or hetero-aromatic signatures in the mid-IR for the COSmIC analogs, are consistent with a greater incorporation of nitrogen into the COSmIC solid analogs compared to the PAMPRE analogs, even at similar nitrogen abundances in the gas phase. Haze analogs produced in gas mixtures without nitrogen, similar to the stratospheres of Solar System gas giants and the H 2 -dominated atmospheres of sub-Neptunes, are generally more transparent with lower n values across the entire spectral range from UV to mid-IR and should therefore be carefully considered in climate and observational applications. The variations in IR absorption features between hazes produced with and without nitrogen could help constrain the presence of N 2 in exoplanet atmospheres. (10.1051/0004-6361/202555916)
  DOI : 10.1051/0004-6361/202555916
• Scalable fabrication of high-quality WS2 thin films via solution processing for NO2 sensing
  
  • Li Ran
  • Maina Elmehdi Ould
  • Dembélé Kassiogé
  • Denawi Adam Hassan
  • Vach Holger
  • Bouanis Fatima
  Materials Today Communications, Elsevier, 2026, 51, pp.114748. (10.1016/j.mtcomm.2026.114748)
  DOI : 10.1016/j.mtcomm.2026.114748
• Live-cell SICM imaging: An Introductory Guide for New Users
  
  • Papa Martina
  • Treussart François
  • Mothet Jean-Pierre
  • Güell Aleix G
  , 2026. High-resolution, minimally invasive imaging of live cells is essential for investigating cellular morphology and its dynamic changes. Among available approaches, Scanning Ion Conductance Microscopy (SICM) offers a unique combination of precise topographic imaging down to nanoscale with experimental conditions that preserve true livecell behavior. Here, we present a practical guide based on our own experience and experiments, intended to be a resource to help first-time SICM users, covering critical aspects from instrumentation, probe characterization, to cell preparation and morphometric data extraction and presentation, to accelerate their implementation, learning and to gain confidence in exploring live-cell structure and dynamics.
• A Bayesian System with Neuron Clocks for Biosignal Classification
  
  • Li Zonglong
  • Calvet Laurie E
  , 2026. A compact Bayesian system for end-to-end inference that uses time-encoded probabilistic computing is presented. The architecture integrates (1) a neuron clocking scheme for timing sequential phases of the system, (2) a feature-extraction module based on rate-coding information with neuron circuits, (3) a circuit able to extract likelihoods from a probability distribution and then calculate Bayes' rule and (4) a race-to-threshold winner-take-all circuit that can drive downstream actuation circuits. All circuits were implemented and simulated at the transistor level in TSMC 130 nm CMOS using a 1.0 V supply. The system was benchmarked using a two-category sleep-stage classification task, and achieved an accuracy of 80.8%, which closely matches the 81.5% result of an equivalent inference in software. The complete architecture uses less than 150 transistors, making it suitable for ultra-lowpower edge biosignal processing.
• Tunable spin-crossover in 2D ruthenium metal–organic frameworks based on hexahydroxybenzene ligand
  
  • Denawi Adam Hassan
  Journal of Physics and Chemistry of Solids, Elsevier, 2026, 208, pp.113154. (10.1016/j.jpcs.2025.113154)
  DOI : 10.1016/j.jpcs.2025.113154