NMR Fingerprinting of Complex Mixtures

Complex mixtures, whose composition is unknown or only partially known, are widespread in chemical and biotechnological processes. In practice, a comprehensive analysis of such mixtures is often too time-consuming and labor-intensive.

At the Laboratory of Engineering Thermodynamics (LTD), methods for automated NMR fingerprinting are therefore being developed based on simple NMR experiments and machine learning (ML) algorithms. The aim of my work is to further develop these methods to enable the analysis and thermodynamic modeling of complex mixtures.

Hybrid Models for Predicting Diffusion Coefficients

Diffusion is a fundamental process in almost all technical and natural systems. Diffusion coefficients are essential for describing mass transport processes. However, the experimental determination of diffusion coefficients is extremely labor-intensive, making reliable prediction methods indispensable.

In my work, I develop hybrid models that combine physical laws with ML algorithms to enable the accurate prediction of diffusion coefficients.

Information on potential topics for student research projects supervised by me or my colleagues can be found at:: http://https://mv.rptu.de/fgs/ltd/jobs/studentische-arbeiten.

Teaching

• Thermodynamik der Transportprozesse (SS 26)
• Thermodynamik I (WS 25/26)
• Modellierung, Simulation und Optimierung in der Verfahrenstechnik (WS 24/25)

• Maschinelles Lernen in der Verfahrenstechnik (SS 24)

   

Office Hours

By appointment

Publications

• J. Wagner, Z. Romero, K. Münnemann, S. Schmitt, T. Specht, H. Hasse, F. Jirasek: Hybrid Machine Learning for Enhanced Prediction of Diffusion Coefficients in Liquids, Physical Chemistry Chemical Physics 28 (2026) 25407-25416. [doi]
• J. Wagner, K. Münnemann, T. Specht, H. Hasse, F. Jirasek: Deep Set Model for the Automated NMR Fingerprinting of Unknown Mixtures, Digital Discovery (2026) Advance Article. [doi]
• J. Wagner, Z. Romero, K. Münnemann, T. Specht, F. Jirasek, H. Hasse: Thermodynamic Modeling of Poorly Specified Mixtures Using NMR Fingerprinting and Group-Contribution Equations of State, Fluid Phase Equilibria 596 (2025) 114446. [doi]

Presentations

• J. Wagner, S. Schmitt, H. Hasse, F. Jirasek; A Unified Hybrid Model for Predicting All Diffusion Coefficients in Multicomponent Mixtures, Thermodynamik-Kolloquium, Dortmund, 30.09.-02.10.2026.
• J. Wagner, T. Specht, H. Hasse, F. Jirasek: Physically Consistent Prediction of Diffusion Coefficients with Machine Learning, European Conference on Thermophysical Properties, Gouvieux, France, 21.-24.06.2026.
• J. Wagner, T. Specht, H. Hasse, F. Jirasek: Hybrid Models for Physically Consistent Diffusion Coefficient Prediction, Annual Meeting of DECHEMA/VDI Subject Group Heat and Mass Transfer, Horw LU, Schweiz, 09.-11.03.2026.
• J. Wagner, T. Specht, H. Hasse, F. Jirasek; Boosting the Prediction of Diffusion Coefficients with Machine Learning, Thermodynamik-Kolloquium, Hamburg, 22.09.-24.09.2025.
• J. Wagner, T. Specht, F. Jirasek: Automated NMR Fingerprinting of Unknown Mixtures using Machine Learning, NMRPM Symposium, Kaiserslautern, 31.03.-02.04.2025.
• J. Wagner, J. Arweiler, Z. Romero, T. Specht, K. Münnemann, H. Hasse, F. Jirasek: Automated NMR Fingerprinting of Poorly Specified Mixtures with Machine Learning, qNMR Summit, Buchs SG, Schweiz, 18.-20.09.2024.

Posters

• J. Wagner, Z. Romero, T. Specht, K. Münnemann, H. Hasse, F. Jirasek: NMR Fingerprinting and Thermodynamic Modeling of Poorly Specified Mixtures (Poster), International Conference on Properties and Phase Equilibria for Product and Process Design (PPEPPD), Bad Gögging, 11.05.-15.05.2025.
• J. Wagner, T. Specht, K. Münnemann, H. Hasse, F. Jirasek: Machine Learning-Driven Characterization of Unknown Mixtures with NMR Spectroscopy (Poster), NMRPM Symposium, Kaiserslautern, Germany, 31.03.-02.04.2025.
• J. Wagner, T. Specht, K.Münnemann, H. Hasse, F. Jirasek: Thermodynamik Modeling of Poorly Specified Mixtures Using NMR Fingerprinting and Group-Contribution Equations of State (Poster), Thermodynamik-Kolloquium, Stuttgart, 25.-27.09.2024.
• J. Wagner, T. Specht, K.Münnemann, H. Hasse, F. Jirasek: Deep Set Models for Elucidating Unknown Mixtures with NMR Spectroscopy (Poster), European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), Vilnius, Litauen, 09.-13.09.2024.