The challenge
To reduce measurement time and solvent consumption while maintaining or improving accuracy with respect to standard methods (SEC/GPC), and to do so in a way that is independent of solvent and sample concentration.
Outcome of the FUNPOLYMER project · PID2021-126445OB-I00
DiffAtOnce: diffusion NMR and AI for polymer molecular weight
Molecular weight prediction in any solvent and at any concentration
DiffAtOnce is a platform developed within the FUNPOLYMER project to process diffusion NMR experiments and estimate molecular weight distributions in polymers. It implements the new Arrabal-Fernandez scaling law \( D\eta_{c} = D\eta_{1/\infty}\,\exp(-\kappa C^{\nu}) \) ,, which makes it possible to compute molecular weights independently of solvent and concentration, combining advanced numerical algorithms, inverse Laplace transforms and artificial intelligence.
Developed at the University of Almería in collaboration with the Institute of Physical Chemistry, Polish Academy of Sciences – “Nuclear Hyperpolarization of Molecular Systems and Nanomaterials” group (leader: Dr. Tomasz Ratajczyk; main contact: Dr. Mateusz Urbańczyk) and PhD student Marek Czarnota.
FUNPOLYMER: from diffusion experiments to molecular weight distributions
FUNPOLYMER addresses the fast and traceable determination of molecular weight distributions in polymers in solution by diffusion NMR, as an alternative to classical size-exclusion chromatography. The key breakthrough of the project is a methodology that allows molecular weight to be estimated in any solvent and at any concentration, based on universal κ(Mw) and Dη|1/∞(Mw) curves.
The solution
To integrate in a single platform advanced ILT algorithms, universal calibration (UCC) and the new κ(Mw) and Dη|1/∞(Mw) curves, together with extended diffusion methodologies (ediffNMR) and artificial intelligence, turning diffusion decays into robust molecular weight distributions.
The outcome
DiffAtOnce, a research software platform that implements the scaling law Dη|c = Dη|1/∞·exp(−κCν) and the universal curves developed in FUNPOLYMER, enabling polymer molecular weight prediction in any solvent and at any concentration.
What DiffAtOnce provides
The platform combines numerical, physical and AI tools to cover the full workflow: from diffusion NMR acquisition to the final report with molecular weight distributions comparable to SEC/GPC, without solvent or concentration restrictions.
Specialized ILT algorithms
A library of inverse Laplace transform methods tailored to diffusion NMR, including discrete and continuous approaches with advanced regularization, designed to provide reliable diffusion coefficients on which the universal calibration laws are applied.
Solvent-independent universal calibration
Universal calibration curves that treat viscosity as an independent parameter, together with κ(Mw) and Dη|1/∞(Mw) curves, allowing molecular weight estimation for multiple polymer families (PS, PPG, PMMA, PE, dextran, polyisoprene…) over a broad range of solvents and concentrations.
Artificial intelligence and HPC
Deep-learning models and deployment on high-performance hardware to accelerate the analysis of large diffusion NMR datasets and to explore extensions to ultrafast and time-resolved methodologies.