These are my crude notes if they are helpful
Notes from Paul Bowyer talk on DOSY nnmr.
Things on my side that was clarified is:
-How important lockphase is. I never adjust that. Only when I do a shim map.
-More scans to improve the phase cycle
-Higher diffusion time can increase convection problems
-Keep delta <= 4 ms
-Try not to use vt.
Varian Performa II gradients were stronger than the JEOL equivalent ( 0.5 Tau/cm cf 0.3 Tau/cm )
JEOL software problem is that signals of different compounds line up together giving the wrong diffusion result. ( and not separating compounds ). Overlap will show an intermediate coefficient value.
VT will affect the results as it sets up convection. If lab temperature is stable then avoid the VT. If you need to use VT then you must use a temperature compensated cc experiment.
The best quality data is imperative, apart from line shape there are other complicated reasons that compound problems eg non uniformity of gradients.
Optimise the Lock Phase as the filed lurches during the pulses and a poor lock phase can show up as distortion on higher molecular weight peaks.
3 main parameters to changes are
Diffusion time ie del on Varian.
Delta
G = gradient strength ( to be arrayed ).
Start with the bpp_led_dosy sequence
Start with 2 values of g first 50 mT and 280 mT, use at least 8 scans ( or multiple of 8 ) for the full phase cycle. Not doing this can lead to out of phase peaks.
Use the Y data processing button to process the data
Use options connect all
Use Home button to rescale
Check attenution of signal and adjust diffusion delay and gradiant duration ( no more than 4 ) so that you get an attenuation of 50% in the 2 spectra of the peaks of interest.
Delta is more effective than diffusuion time at changing the attenuation. But be aware that a longer diffusion time introduces convection.
Create a new “ g “ array use the logarithmic option and a minimum of 15 points
Software
JASON – needs licence ( affordable )
GNAT has reference deconvolution, download the latest version
Open file in JEOL
Remove all processing steps
Save as JEOL generic
Reference deconvolution in GNAT using a reference signal eg TMS however would need a larger molecule for larger masses.
Information for a publication reference:
Diffusion measurements (DOSY)
Diffusion measurements were conducted on a Jeol type JNM-ECZ500S NMR spectrometer operating at 11.747 Tesla and 500.16 MHz for proton resonance. The NMR is equipped with a 5 mm Inverse detection proton sensitive auto tuneable ROYALPROBE™.
Parameter optimization was carried out using the bpp_led_dosy pulse programme (Bipolar Pulse Pairs stimulated echo Longitudinal Eddy current Delay selected from homodecoupling, presaturation). The parameters were further optimised at an interval time of 4 milli seconds, diffusion encoding pulse width of 2 milli seconds and a large Delta of 0.1 seconds. The Eddy current delay was 20 ms. A relaxation delay of 10 seconds was used.
The gradient strength applied during the diffusion experiment of 32 increments ranged logarithmically from 0.03 T/m to a maximum gradient strength of 0.3 T/m (3G/mm) at 10 Amps with 16 scans for each increment.
Data processing was performed with the Jeol Delta™ software, version 5.3.3, which uses multivariate statistical analysis to plot the results.
New Auto DOSY setup:
Work presented at https://sites.google.com/view/dosyday2022/home
DOSY automation that automatically estimate the best experiment parameters to automatically run a DOSY given an estimated molecular weight using the SEGWE model( https://nmr.chemistry.manchester.ac.uk/?q=node/432). It calibrates the 90 pulse and checks for convection,
choosing automatically a convection compensated DOSY experiment if needed.
Estimate_diffusion coefficient method can be used as a calculator, it does not run any experiment.
Setup for small molecules, 0.1[s] diffusion time fixe and up to 0.3[T/m].
This may need to be adjusted when setting it up for larger molecules.
Estimations other than SEGWE may be preferred for polymers.