Variable temperature NMR spectra

Some types of dynamic NMR questions can be answered with little fuss. For example, amides often show cis-trans isomerism such that two sets of broad peaks will be seen in the 1H spectrum at room temperature. The spectrum simplifies if the sample is warmed a bit and the two sets of peaks coalesce. For some amides, 40-50 C is sufficient to fix up the proton spectrum (though it may actually make the carbon spectrum worse, if the amide carbonyl signals broaden more but aren't yet coalesced). If all we want is a simplified proton spectrum, we may not care what the sample temperature actually is or how different it is from the displayed temperature, as long as the solvent doesn't boil in the process.

However, for accurate determination of activation parameters, or kinetics experiments, the actual temperature must be known. It can be quite challenging to get good quality spectra at temperatures far from ambient, especially if the acquisition times are long. Variable temperature experiments also raise additional experimental safety considerations. Bear the following in mind:

If you change the probe temperature, or switch from house air to nitrogen, you are responsible for changing it back in time for the next user of the instrument.
The geometry of an NMR sample--a long, narrow cylinder--is optimized for magnetic field homogeneity and ease of changing samples. It is a really unfortunate choice for variable temperature experiments, and necessarily involves some experimental compromises.
Different probes have different temperature limits, depending on the thermal expansion of the materials used in their construction, the types of adhesives being used, and how well the probe insulates the shims and the magnet bore from the extremes of the sample temperature. Some of the newer probes have a sticker on the base giving the probe's temperature limits; for others, the information is given in the manual. We have summarized the characteristics of all the probes in the facility, including temperaature limits, on our website at http://mangia.caltech.edu/NMR_inst.html. Please respect and plan your experiments around these limits. If a probe change is required for your experiment, please check with facility staff in advance.
Know the boiling and freezing point of your solvent. Approach this temperature with caution, because the displayed temperature may be off, and as you raise or lower the temperature, it is common to overshoot the set point by a few degrees before coming in to regulation.
The temperature displayed is measured by a thermocouple wire placed in the stream of air entering the probe, typically from below the sample. This wire has to be outside the NMR coil when we are delivering radio pulses to the sample, for the same reason that aluminum foil should be removed before heating leftovers in a microwave oven. Since the temperature is measured at a distance, the displayed temperature is likely to be at least slightly different from the actual sample temperature. For precise work, the actual sample temperature should be calibrated. Two methods for doing that are to use a separate sample of an alcohol (methanol for low temperature, ethylene glycol for high temperature), where peaks in the spectrum show a temperature dependent chemical shift offset, which has previously been calibrated against temperature; or to use a thermocouple suspended in a sample in the NMR coil, which is removed prior to acquiring the spectrum. With either method, the experimental setup should be changed as little as possible when the calibration sample is switched for the experimental sample.
One of the variables in the behavior and calibration of the probe is the air flow rate through the VT unit. Varian states that their hardware is designed for optimal agreement between displayed and actual sample temperature with an air flow of 10 SCFM. This flow is set via a flow guage in the magnet leg. We recommend that you run at that flow rate, unless you cannot reach the low temperature you need without increasing the air flow. But be advised, changing the air flow will invalidate any previous calibration and will increase the deviation between displayed and actual temperature. Using a VT flow below 10 SCFM is likely to make temperature gradients worse (see below).
There is another flow gauge in the magnet leg controlling cooling air. This is air circulated through the probe body to protect the shims against extremes of hot or cold probe temperatures. A normal setting is 5-10, but according to the Daytona AutoX probe manual, it should be set to 10 when Daytona is operating at the extremes of temperature. Daytona's AutoX has an extra wide temperature range of 150 to -150 C.
Any time you start heating or cooling a sample, or you insert a room temperature sample into a heated or cooled probe, temperature gradients naturally occur in the sample, from top to bottom, and outside to inside. Static temperature gradients will normally relax out after a few minutes, if the airflow through the probe is sufficient.

Make your sample as short as possible consistent with still being able to get good homogeneity, in order to limit vertical temperature gradients. 4 cm is a good value. Center the column of liquid in the tube around the center line shown in the probe depth gauge, instead of pushing the sample all the way to the bottom. A short sample normally causes the Z2 and Z4 shims to move toward more positive values. If you are shimming by hand, move those shims that way. If you are gradient shimming, it may be a good idea to use a narrower section of the profile on a shorter sample. If you shim through the gradient shim setup panel, a.k.a. gmapsys, and the normal window size is 40, you might cut it to 30. Also, if you normally shim 5 shims, you may wish to reduce this to 4.
If the displayed sample temperature is not totally stable, but oscillates up and down, you probably have a dynamic temperature gradient in the sample, which lags the probe temperature.
By definition, if you have a temperature gradient in the sample, you can't know its temperature, because different parts of the sample are at different temperatures.
Temperature gradients make it very hard to get good homogeneity. Wait to shim unti the sample temperature is stable. One way to judge this is watching the lock: you may see the lock level drop quickly when you change temperature, then slowly rise. Don't do anything until the lock level is constant. (A temperature gradient can be masked by applying extra Z2 shim correction, but when the gradient relaxes out, Z2 will be wrong and all your peaks will tail either left or right.)
Changing the temperature will cause the shims to change, and if the temperature change is large enough, also change the lock phase. Periodically check that the lock phase you are using corresponds to a maximum lock level.
A big temperature gradient may cause convection cells to form in the NMR tube. Convection degrades resolution, and totally kills gradient shimming, and diffusion experiments. The lower the viscosity of your solvent, the more likely it is to break into convection. (Most organic solvents have low viscosity, DMSO being a notable exception).
Spinning the sample prevents convection cells from getting established, and also helps remove temperature gradients. Unfortunately, most 2D experiments, as well as gradient shimming, should be performed on a nonspinning sample.

Varian recommends that nitrogen gas should be used in VT experiments over 100 C, because of potential oxidation of the radiofreqency coils. Nitrogen should also be used for experiments where the VT gas goes through the liquid nitrogen bucket, because liquid oxygen would condense out in the coils otherwise. However, Varian probes are susceptibility matched to air, which contains paramagnetic oxygen molecules. When you switch from air to nitrogen, your shims will change noticeably, and even after you compensate for that, the resolution will be degraded.

We have two sources of nitrogen gas, one being the high pressure liquid nitrogen dewars near the lab entrance, and the other being a passive nitrogen separator. The dewars are the only option for low temperature experiments using liquid nitrogen in the heat exchanger, because the gas must be completely free of water vapor, as well as oxygen. For high temperature experiments, the separator is an attractive alternative, because you don't have to worry about running out of nitrogen partway through your experiment, and we don't have to pay for the high pressure nitrogen dewars. Below 100 degrees, switching to nitrogen is optional but recommended. At or above 100 degrees, it is mandatory.
For low temperature experiments not far from ambient temperature, the FTS air chillers are useful. Oxygen will not condense in these, but water may, at very low temperature. Our house air is dried to a dew point of -40 C. We have run the FTS chillers using house air at a set point of -20 for extended periods without icing up the FTS bath. The probe can get to a temperature around -5 with the FTS at -20. Note that the low temperature limit for Fid's triax probe is 0 degrees. On Daytona, the FTS bath can reach -80, giving a minimum probe temperature around -40. However, you should use the dry nitrogen gas from the liquid nitrogen tanks for these low temperatures; ice will form with the house air.

For any experiment where a sample must be kept cold and transferred to an already cold probe, plan how the transfer will be accomplished, and how frost buildup on the sample can be minimized. It is very hard to avoid having the sample warm up significantly while it is being inserted into the probe, as the eject air is at room temperature. Carefully consider any safety risks associated with reactive compounds if they do warm up.
For any NMR samples that may generate internal pressure during the VT experiment, carefully consider any safety risks. It is advisable to test any such experiment beforehand, in a hood with suitable protective equipment such as a polycarbonate shield, with the sample exposed to at least as much stress as it will experience during the NMR experiment.

Troubleshooting problems during VT experiments:

Any interruption to the VT gas flow can create a potentially dangerous situation if the probe heater keeps heating but this heat cannot be carried away. You may get a message in software saying "HW error" if the system senses something wrong, or you may not. If you move or handle the air connections to the probe, make sure they are on firmly, and don't become dislodged while you add liquid nitrogen to the dewar, or make other adjustments. If the airflow has been interrupted, and you fix it, the sample temperature may quickly rise to dangerous levels--past the boiling point of an organic solvent, or hot enough to denature a protein sample. A symptom of an air leak or disconnected hose may be that the system responds very slowly to a command to increase the temperature, or during a low temperature experiment, the temperature starts rising out of regulation. If you find a problem with the air flow, the best steps to take are:

Immediately cut off the heat if you are doing an above ambient temperature experiment.
Then remove the sample from the probe, so it will not experience a temperature spike.
If possible, wait for a time to allow the probe to cool/warm passively.
Finally, reconnect the VT air supply.

If you are using the FTS air chiller with house compressed air, and you initially have the temperature regulated, but then the temperature starts rising, the FTS may be getting blocked with ice. The FTS will have to be warmed up, so the ice melts, and the acculated water flushed from the system. If you see this problem, please let lab staff know. It may indicate our compressed air dryer needs service.
If you are using nitrogen from the high pressure tanks, and your temperature moves out of regulation, it may indicate the tank(s) are running out. Treat this like an interruption to the VT gas flow, above, because suddenly restoring the nitrogen flow can cause a temperature spike.

Dave 7/23/09