Advances in Diffusion-Ordered Spectroscopy

  • Jamie Guest

Student thesis: Master of Science by Research


The factors which affect precision in the diffusion coefficients in the DOSY experiment are examined. Specifically, to what extent does the signal-to-noise ratio limit precision in the diffusion domain and how do variations in signal amplitude express the systematic errors at play in these experiments? The PSYCHE-iDOSY and Oneshot-PSYCHE experiments, previously developed by the NMR Methodology group at the University of Manchester, are to be re-examined to determine if the resolution gain in the NMR domain really does translate into an improvement in precision in diffusion domain compared to a standard Oneshot45 experiment. Mathematical simulation was performed using MATLAB to establish an empirical relationship between certain experimental parameters, analogous to those used in an actual DOSY experiment, and the resulting precision of diffusion coefficients produced, which was expressed as a simple equation. The experimental parameters examined included the number of gradient levels used, the maximum value of the exponent of the Stejskal-Tanner equation and the signal-to-noise ratio. Through simulation and the derivation of an equation followed by comparison with experimental data to which noise had been added using Mathematica, precision in D (where D is the diffusion coefficient) was observed to have an initially linear relationship with signal-to-noise before plateauing out after a given SNR, demonstrating the limit to which increasing SNR alone can further improve precision in D, before other sources of error, such as those caused by temperature fluctuations in the room in which the spectrometer is housed, begin to become dominant. This work also verified that for simple samples there is a loss of precision in accordance with a reduction in signal-to-noise ratio using the PSYCHE-iDOSY sequence compared to the standard Oneshot45. However, in the analysis of relatively more complex mixtures, a gain in precision in the diffusion domain was observed, in accordance with previous observations, which resulted from the removal of erroneous peaks at intermediate positions between signals corresponding to actual diffusion coefficients.
Date of Award1 Aug 2021
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorGareth Morris (Supervisor) & Mathias Nilsson (Supervisor)

Cite this