Hi AMMRL,
Thanks to Dave, Celine, Jay, Chris, Glenn, Nanette, WLJ, Jochem, Jeff, Bob & Adam, Darcy, and James for sharing their experiences. Sample dilution with silica ended up working in our case!
I’ve summarized the shorter responses into two areas: diagnosing the problem and reducing it. I’ve copied two of the longer responses at the bottom of this e-mail (with the authors’ permission).
Diagnosing the problem:
Before preparing a rotor, check if the sample is magnetic. Samples may have picked up magnetic contaminants during processing (e.g. steel filings from machinery, etc.). Multiple people reported difficulty ejecting rotors containing magnetic samples.
Tune the probe to glycine, then insert the conductive sample. Conductive samples can cause a big change in the probe tuning. Increases in the width of the tuning dip are indicative of conductivity.
Conductive samples may resist spinning or even refuse to spin due to eddy currents (see
https://www.youtube.com/watch?v=Qy8AGTbmNd4 at the 13 minute mark). Spinning samples may be warmed as a result of the eddy currents, and one person reported a sample exploding. The bearing and drive pressures (as compared to glycine at the same speed) can provide a measure of resistance to spinning.
One person reported destabilization of the MAS in a periodic fashion as a response to a pulse.
Reducing the problem:
Multiple people reported diluted the sample with an inert powder (Teflon, MgSO4, silica, KBr, NaCl, PbNO3, and sulphur were all reported/suggested), or by using a rotor insert. One report mentioned melting the sulphur to better distribute the sample. Ratios of up to 4:1 filler:sample were reported.
Two people reported grinding samples to a finer powder (e.g. less than 44 microns / 325 mesh) to reduce the RF skin depth effect and to reduce arcing.
Conflicting reports on the benefit of reducing 1H decoupling power: one person reported that once a sample arced, it would continue to arc no matter how low the decoupling power, while another reported some success by reducing decoupling power.
Two people mentioned that samples which arc under 1H decoupling were amenable to direct detection of 13C (spin-echo without decoupling). One person reported a sample where they had measured 1H T1 at ~ 100 msec and estimated 13C T1 at ~ 30 sec; he successfully recorded a 13C spin-echo (no 1H RF) using 10 sec D1.
It was suggested to try an “efree” probe.
Detailed response #1.
For semiconductive/conductive samples, there are several issues related to MAS experiments. First, conductive samples may have eddy currents induced by the magic-angle spinning. This may lead to "magnetic braking" in which the eddy currents work in opposition to the spinning. This presence and magnitude of this phenomenon may be checked on your Bruker MAS probe simply by comparing the bearing and especially the drive gas pressures for the rotor with the conductive sample with either that of an empty rotor or a rotor containing a non-conductive sample. If present, this effect can often be mitigated by the use of ever smaller particles.
MAS of conductive samples can also lead to an issue of sample heating beyond that of the usual rotor heating from atmospheric drag. This issue can be checked by mixing in some lead nitrate with the conductive sample and using the 207Pb signal to see if increased (from that of simple spinning the sample) sample temperature is present.
If the sample is conductive, then powdering the sample and mixing with a non-conductive sample can reduce conductivity resulting from particle-to-particle contact. Depending upon the sample, compounds such as sodium chloride or lead nitrate can be used. Sulphur also works but is not particularly nice to work with.
If the sample is conductive, there is also the issue of the radiofrequency skin depth effect, i.e., how far into the conductive material that the RF can penetrate. For many semiconductive/conductive materials, a particle size of less than 44 microns (less than 325 mesh) generally works well. RF skin-depth effects can produce issues with phasing for the spectrum for those nuclei that are observed.
If you do acquire a MAS spectrum, even with 1H decoupling, the resulting spectrum may still be quite broad. This holds for 13C of graphene oxide in addition to the usual semiconductors and conductors. There are several reasons or contributions to the broadening. First, interaction with the conduction charge carrier can cause a very short spin-lattice relaxation time. T1s on the order of 1 millsecond to 10 milliseconds are not unusual in cases in which the s-orbital is involved. The lifetime broadening of the resonance is then quite large as compared with what might be usually expected or found in a sample with less conductivity. Second, the interaction with the conduction charge carrier can also induce an additional shift in the resonance. Samples in which the charge carriers are not uniformly distributed can then result in broaden resonance due to the different shifts and may also show different spin-lattice relaxation rates.
NMR can be done on may conductive sample under the appropriate conditions.
Detailed response #2.
You're right to be cautious, because the last thing you want is the probe arcing & going off for a holiday to the repair shop.
No reliable metrics but some 'rules of thumb' below:
First thing is to check the wob curve for 1H & 13C; with respect to a standard organic eg glycine. Usually it's width increases for both nuclei is indicative of conductivity & whether it's worth going further.
As you remark, dilution of the sample with some inert powder prior to packing may help also
Eg MgSO4; silica etc
Next, it seems to spin ok? Sometimes they are so resistive(conductive) that that's the end of it. Again dilution may help.
If spinning & wob curve is not too broad & 'detuned' then you may find the 1H T1 to be very short say 100ms, so it's worth checking by a one pulse one scan test with variable d1 or by a spin echo ( ksr pulse programs).
Difficult to repeat this with 13C because the signal is too weak.
Direct 13C detection can be done without decoupling but maybe with spin echo. 13C T1 may be much longer however.
Had a borderline char recently that was 1H ~100ms & 13C 30s est. T1. But a spin echo with 10s recycle delay allowed accumulation of signal without harm.
You're right to be cautious & try soft decoupling first & watch carefully for spikes. Keep aq short either way.
Best regards,
Sameer
--
Sameer Al-Abdul-Wahid, Ph.D.
Manager, Nuclear Magnetic Resonance Centre
University of Guelph
50 Stone Road East, Guelph, ON N1G 2W1, Canada
Tel: 519-824-4120 x58914
Web: http://nmr.uoguelph.ca/
Received on Wed Sep 19 2018 - 09:48:20 MST