In response to my post regarding calibrating gradient strengths for DOSY
experiments, I wanted to report that I had a total of 27 responses and I
wish to thank you all for your emails. In particular, I want to acknowledge
Stephane Viel, Ross Mair, Todd Alam, Mike Ditty, Claudia Krywiak, Gerry
Griffith, Glenn Facey, Doug Harris, Steve Huhn, and Markus Hoffman who sent
me EXCEL worksheets, manuscripts, datasets, PP presentations, references,
etc, etc.
It is not surprising that there are many different opinions out there. There
are some who think samples with known diffusion constants should be used in
a diffusion measurement for the coil calibration. Many use water for this
purpose but everyone reading this knows about the temperature dependence
issue, which is a whole separate matter. Issues mentioned for improving the
accuracy of this approach included (a) a probehead T calibration first
(necessary!) (b) using water but then testing the robustness of the
calibration by using slower diffusing samples. 10mM beta cyclodextrin in H2O
and frozen cyclohexane were mentioned by different people (c) putting the
water in a sealed capillary tube inside a 5mm NMR tube to minimize
convection.
There were others who think a phantom is the superior approach. Several have
just used Shigemi tubes or susceptibility matched plugs but here the
accuracy in the precise distance is a concern. More accurate ideas for
phantoms were (a) D2O susceptibility matched plug of known thickness witha
drilled hole (b) solid glass cylinder sandwiched between two glass tubes
inside a regular nmr tube filled with CHCl3 (c) fiberglass rod with 2
drilled holes and immersed in copper sulphate doped water (d) a 2mm slice of
teflon with a hole drilled in the center fixed by a thin glass capillary on
the tube's bottom in D2O (e) buy a phantom from Bruker. It was mentioned to
me numerous times to perform this measurement using different gradient
strengths rather than at a single strength and extrapolating.
I heard from several who use a phantom but then check this result using a
standard sample. And I also heard the opinion that both of the above methods
are very prone to errors and one is best advised to use the Bruker
theoretical value of 5.35. There were constant warnings regarding the
non-linearity of the gradient over the coil length. If one is using a
standard compound with known D, this problem will result in the coil
constant being a function of the viscosity of whatever sample is used.
Lastly, I heard about two ideas new to me (a) using relative rather than
absolute diffusivity and (b) using a water sample and making use of the
integral of the HOD signal at 2% and 95% gradient strengths to calculate the
coil constant.
Once again, thanks to all!
Mike
----------------------------------------------------------------------------
----------
Michael D. Lumsden, Ph.D.
NMR Facility Coordinator
Room 428, Atlantic Region Magnetic Resonance Centre
Department of Chemistry, Dalhousie University
Halifax, Nova Scotia, Canada
B3H 4J3
phone: 902-494-1635
FAX: 902-494-1310
http://armrc.chemistry.dal.ca
Received on Tue Oct 31 2006 - 18:02:34 MST