In lieu of presenting an absolute, unwavering view of how one should assess
the merits of a given spectrometer, I will instead just present the issues
that were mentioned to me in the e-mail I received in response to my request
for 'Standardized NMR Tests.' At the end of this message I have also added
something regarding the applicability of these sorts of tests.
I received some advocacy of Varian's Autotest protocol. The problem with this
test is that Varian has designed it, so the other manufacturers may not be so
keen to do it Varian's way. At the very least, Varian has the jump on the
others with this if they decide to adopt it as well. This protocol is more
rooted in putting the hardware through its paces than allowing a chemist to
discern how well a spectrometer will perform on a real sample. This touches
on the general issue of how different people understand the workings of NMR
spectrometers and to what level of sophistication they will benefit from
seeing, for example, the 13 degree test.
The argument can be made that if the spectrometer satisfies all of the
manufacturer's tests, then it will also be able to do fine on a real sample,
but unless the other manufacturers adopt this protocol, comparisons are
difficult to make.
It is important to first make sure the hardware is working properly before
testing it on a real sample. That is, make sure the manufacturer has
satisfied all of their specs before you look at how a spectrometer performs
on a real sample. As far as demos go, the instrument should be performing
within specs when you see it run, so run your particular test sample(s) and
see how it looks.
The argument was made that it is better to test sensitivity and lineshape/
resolution separately.
A general take-home message that I get from the responses is that 'real' sample
tests are not good for diagnosing spectrometer problems (a lousy s/n or
inaccurate result may arise from a number of different problems). When an
experiment fails to give the proper or desired result, don't look to a
real-world sample to tell you exactly what is wrong. I wrote my original e-mail
on this subject thinking in terms of what one might want to see during a
demo, not what one would use to assess an installed spectrometer that is
acting up. With this in mind, I present the following list of issues one
might consider when one is contemplating the purchase of an instrument.
Unless there is a major initiative on the matter, I think we will always have
to do our own demo trips to assess spectrometers.
General issues:
for broad band probes, look at quoted sensitivity in the middle and at both
ends of the tuning range (e.g., 15N, 13C, 31P)
Water suppression - look at a 1mM sample of phenylalanine in 10% D2O/90% H2O.
sucrose was also suggested.
Consult Chapter 3 entitled "Standard Tests" in "100 and More Basic NMR
Experiments," VCH, 1996, ISBN 3-527-29091-5
Support was expressed for looking at a sample with fixed volume at a given
concentration. Maybe 500ul and 700ul both are appropriate (use the same
kind of NMR tube, e.g., Wilmad 535pp).
don't forget to degas your sample.
Beware of looking at residual CHCl3 in CDCl3 in a 600MHz NMR or higher, as
the peak is split due to partial alignment of the molecules in B0. Acetone
and benzene are better for this.
dielectric resonance
Changes due to rf sample heating should be addressed. One might want to make
sure the lock is in the linear range (far from saturation) when addressing
how much the lock level might move when the decoupler is cranked up.
radiation damping may also be an issue, although it may be more
dependent on sample than on spectrometer.
baseline flatness
background signals
decoupler isolation from lock
vibration isolation
RFI isolation
shim stability and drift
digital filtering
receiver noise figure
RF Homogeneity (810/90)
Cancellation
Quad Image Rejection
Small-angle phase (accuracy of small phase adjustments)
30 deg pulse stability (amplitude stability)
pulse turnon (rise time)
pulse ideality (tests rectangularity of pulse shape)
glitch (noise/glitches with wide sweep width)
phase stability (13 deg test)
rf amplitude predictability (tests linearity)
rf excitation predictability (tests pulse shaping)
shaped rf amplitude stability
gradient profile (gradient calibration and linearity)
gradient recovery stability (similar to 30 deg test with grad)
gradient recovery (tests time for gradient recovery)
gradient echo (tests quality of bipolar grad echo)
gradient effect on cancellation (cancellation test with grad)
amplifier droop (requires high speed scope)
salt tolerance (sucrose in D2O with/without NaCl)
variable temp test (tuning stability, temp stability, and
temp gradients)
channel isolation (decoupler isolation from lock)
vibration and electrical noise (no vib or 60Hz, etc., in spectra)
shim stability and drift (line shape stability and magnet drift)
digital filtering
receiver noise figures
-------------------------------------------------
Jeff
Dr. Jeffrey H. Simpson jsimpson@mit.edu
Director of the Spectrometry Laboratory 617/253-1812 office
Department of Chemistry, 18-082 617/253-1806 lab
Massachusetts Institute of Technology 617/253-0873 fax
77 Massachusetts Avenue, Cambridge, MA 02139-4307 USA