Résumé
Cette thèse explore la résonance magnétique nucléaire (RMN) du solide hyperpolarisée par polarisation dynamique nucléaire (DNP) à ultra-basse température, en se concentrant sur le problème du bruit-t1 apparaissant dans l'étude des composés à abondance isotopiques naturelle. Une caractérisation théorique et expérimentale de ce bruit met en évidence sa forte dépendance aux conditions expérimentales et sa variabilité temporelle. Le développement d'un "filtre zz", combiné à un filtre z, est exploré afin de supprimer efficacement le bruit-t1 directement au niveau de la séquence d'impulsion et de révéler des corrélations à longue distance. Cette approche est validée dans les conditions d'ULT MAS-DNP et est utilisée pour étudier le PROTAC ACBI1, une molécule pharmaceutique sous forme amorphe. Ces expériences montrent le potentiel de l’ULT MAS-DNP pour étudier des systèmes complexes et révéler des informations structurales inaccessibles par MAS-DNP conventionnelle à 100 K.​​

Title: Enhancing the sensitivity of multidimensional DNP-enhanced MAS NMR with cryogenic Helium spinning: from t1-noise artefacts suppression to the study of amorphous pharmaceuticals.

Abstract
This thesis aims to extend the applicability of natural-abundance solid-state NMR by exploiting ultra-low-temperature (ULT) magic-angle spinning dynamic nuclear polarization (MAS-DNP) while addressing one of its major practical limitations: t1-noise. Although ULT MAS-DNP provides unprecedented sensitivity enhancements, which are particularly valuable for low-abundance and low-gamma spin systems, these gains are often accompanied by a dramatic amplification of multiplicative noise arising from hardware instabilities, severely degrading spectral quality in multidimensional experiments.​​
A central objective of this work is therefore to characterize, understand, and suppress t1-noise in the context of natural-abundance solid-state NMR, with particular emphasis on experiments performed on the ULT MAS-DNP system in Grenoble. To this end, a comprehensive framework for the analysis of t1-noise is established, combining theoretical descriptions of noise processes, numerical simulations, and experimental observations. This study reveals that t1-noise strongly depends on experimental conditions and can vary significantly over time, even under nominally identical settings. Building on this understanding, the thesis reviews existing strategies for t1-noise suppression and evaluates their applicability to ULT MAS-DNP experiments. Both experimental approaches and post-processing methods are investigated, highlighting their respective strengths and limitations when applied to highly sensitive, noise-prone datasets.​​
The main methodological contribution of this work is the development of an experimental zz-filter designed to suppress t1-noise directly at the pulse-sequence level. By selectively converting double-quantum coherences into longitudinal two-spin order while dephasing unwanted single-quantum coherences, this approach efficiently reduces multiplicative noise without relying on post-acquisition processing. Combined with a pre-detection z-filter to form a z3-filter, the method enables the recovery of weak long-range correlations that would otherwise remain obscured by noise. The performance of the filter is demonstrated for both isotropic (J-based) and anisotropic (dipolar-based) polarization-transfer schemes, in both homo- and heteronuclear cases.​​
Finally, the utility of the proposed approach is demonstrated under ULT MAS-DNP conditions on systems ranging from model compounds to complex biological materials. In particular, the potential of ULT MAS-DNP is illustrated on an amorphous pharmaceutical: the PROTAC ACBI1. The enhanced sensitivity and improved system stability achieved under these conditions enable the acquisition of 13C-13C and 13C-15N correlation spectra within practical experimental times. These experiments allow the assignment of most ACBI1 sites and reveal long-range correlations that provide valuable structural information for future investigations of the material. Altogether, these results demonstrate that ULT MAS-DNP opens new opportunities for the study of highly challenging materials that remain inaccessible with state-of-the-art 100 K MAS-DNP experiments.​

Direction
Gaël De Paepe

Co-direction
Sabine Hediger et ​Paul Subhradip

​