Shimming, Part II
Basics of Shimming
Progress during shimming can be monitored in various ways, the most common being those listed below:
- Swept NMR resonance.
- Lock level.
- Free induction decay (FID).
Choice of the most useful method depends on the instrument and its current condition. The swept NMR resonance can be a signal from a proton, deuterium from the lock solvent, or another NMR-active nucleus which is swept by either field sweep or frequency sweep. To be used for shimming, the signal should have sufficient signal-to-noise (S/N) that the height and ringing pattern can be observed. The sweep repetition rate should be fast enough to give a “real time” response as the shims are adjusted. This method of homogeneity measurement is best used for initial adjustment of the shims from a raw magnetic field. It is also commonly used when the spectrometer has no internal lock. The swept response should be adjusted for pure absorption phase and the rf power should be adjusted to avoid saturation of the signal. The height of the initial signal response should be used to determine the best response while shimming. The ringdown pattern can be used for the final adjustment of Zl and Z2. Care should be taken not to overemphasize the ringdown pattern in the initial stages of shimming since varying lineshapes can produce better ringing, but less desirable lineshape and resolution.
The lock level is probably the most common method of adjusting homogeneity. The lock level can be displayed by an analog meter on the NMR spectrometer or as a level on a CRT display. The lock signal can be digitized and displayed as a number or level on a computer display. The NMR lock nucleus must not be partially saturated with rf power and the phase of the NMR signal MUST be adjusted carefully. Care should be taken if the lock phase is adjusted by maximizing the lock level. An asymmetric NMR lineshape leads to a lock maximum at a phase that is not at pure absorption. Therefore, when the lock phase is adjusted by maximizing the lock level, improper lock phase can lead to an asymmetric lineshape, or asymmetric lineshape can lead to an improper lock phase. This problem does not prevent the phase adjustment of the lock using the lock level, but care should be taken that whenever an even-order axial shim is significantly changed the lock phase is readjusted.
Another problem with using the lock level for homogeneity adjustment is the relative sensitivity to field inhomogeneities of the NMR lock signal and the signal under observation. Many spectrometers use an internal deuterium lock. If the nuclei under observation are protons, then the lock signal’s sensitivity to inhomogeneities is one-sixth that of a proton (the ratio of the magnetogyric ratios of the nuclei). This is further complicated by a deuterium resonance usually being broader than its proton counterpart. Again, this does not mean that the lock signal cannot be used for homogeneity adjustment, but means that care should be taken not to be misled by the lock level. In general, good shimming results can be obtained by using a lock solvent with a sharp lock signal to do shimming for basic lineshape and minimum SSB ( Spinning SideBands ) followed by a touch up adjustment of Zl and Z2 on the FID signal.
Using the FID response is probably the most difficult method of shimming and should probably be left for the final “touch-up” of Zl and Z2. However, many spectroscopists prefer this method to the other two methods. To the trained eye, the FID yields information about the lineshape and resolution similar to the ringdown pattern from a swept signal. Some excellent examples of FID shape and resulting lineshape are given in an article by Chmurny and Hoult. Briefly, the longer the FID rings, the better the resolution. The closer the FID’s shape to a perfect exponential decay, the better the lineshape. A FID that falls sharply to another level and then decays more slowly either has poor lineshape displaying considerable hump or contains the resonances from two or more signals with different line widths. Some NMR computer systems can integrate the total area of the FID and yield a number indicative of the resolution similar to the lock level. FID shimming is the most sensitive method for shimming and has the advantage of using the observed nucleus as the shimming criterion. This means that shimming of the FID does not suffer from the loss of sensitivity to resolution caused by different magnetogyric ratios of the nuclei that is often seen when using a lock signal and it also removes all doubt that what the lock system “sees” is not what the observe system “sees”.
All three of the above methods for measuring relative magnetic homogeneity of NMR instruments can be successfully used. Some areas to be concerned with in each method have been pointed out. The shimming sequences described later refer only to response. Except where otherwise indicated, the operator should choose his/her own method. It is probably best to use all of them in some combination during the shimming process.
If the shimming operation were a simple maximization of each shim gradient, then shimming would be quick and easy. However, several things prohibit this.
By their design some shims generate some gradients of other shims. This means that Z4 is expected to generate some Z2 and Z0.
Each shim has impurities of other shims. For example, when the Z4 gradient current is adjusted it not only generates a Z4 gradient but also some expected Z2 and some unexpected Z3 and Z1 gradients.
As the shimming process progresses, the response being used to adjust the shims becomes more sensitive. For example, if the magnetic field has a 20% Z2 gradient and an 80% Z4 gradient, adjustment of Z2 first will be insensitive. Then after Z4 is adjusted, Z2 will be more sensitive and might adjust to a different value. This is an over-simplified explanation and is used only as an example.
Shims have, by nature of their design, levels of interactions with other shims. See the table Equations for Common Room Temperature Shims.
Zero Order Shim Adjustment
Shims having no interactions (zero-order interaction) can be adjusted for best homogeneity by one simple adjustment. However, most shims have interactions that complicate the process. The other two types of shim interaction are called first-order and second-order interactions.
First Order Shim Adjustment
First-order interactions are the type that allow a true minimization of magnetic inhomogeneity by repeated maximization of the individual shims. After the complete set of shims is adjusted, readjustment of the first shim of the set leads to a different optimum. Successive iterations then lead to less and less change on readjustment until finally no further change is observed.
An example of this first-order interaction is that of Z1 and Z2. If the Zl shim is adjusted for optimum response followed by adjustment of Z2, then Zl has a new optimum when readjusted. If the process is repeated, the amounts by which Z1 and Z2 change on each iteration decrease until the optimum values for Z1 and Z2 are found. The process can often be accomplished faster by noting the direction the Z1 shim is changing and moving slightly too far on the early corrections. The ability to make educated guesses based on a knowledge of the shims leads to a much faster shimming process.
Second Order Shim Adjustment
Second-order interactions are of the type where a given shim must first be moved and then others adjusted before any determination of improvement can be evaluated. Successive optimizations (first-order process) of this shim type and other shims do not necessarily lead to the best homogeneity. The process employed for second-order interactions is usually to change this shim a measured amount and reoptimize another set of shims. If this leads to a better response, then the shim is changed another measured amount and the process repeated until the response starts to decline. If the initial response is worse, the other direction is tried.
Which Shims are which Order
A classification of each shim as to the type of interactions to be expected is shown in the table. The interaction order assigned to each shim is a general description of the expected type of response. In some circumstances, a simpler type of interaction is observed. However, the operator should not expect a simpler interaction, but proceed on the worst case assumption. It can often be noted that shims displaying a complex type of interaction in the early stages of the shimming process display a much simpler type of interaction as the homogeneity improves. This simplification process often yields zero-order interactions for most shims in the later stages of shimming. Under these conditions, the shims appear to “drive” directly to their final positions with almost no interactions.
Open All Senses
As can be deduced from the summary just given, shimming can be a complex undertaking. It is important that the NMR operator use all tools and techniques available to him/her in the process. In addition, he/she needs to be very observant. The NMR instrument is giving lots of feedback at all times. These should be taken as clues leading the operator to the successful conclusion of the shimming process. The operator should, however, be wary of drawing conclusions from too little information. He/she will often be receiving contradictory information. Tests should be formulated to expand and explore contradictory clues. The operator needs to avoid becoming frustrated by contradictory clues, but also not avoid frustration by ignoring some information. Note all information the NMR instrument is providing, even if it is not immediately understood. The entire shimming process has a lot in common with a detective story with the NMR operator being the detective and the NMR instrument providing the mystery of what are the best shim values.
When an NMR instrument is first installed, the setup is usually done by a trained service engineer. Most do a good job, but turnover is high in the service area and sometimes mistakes are made. The ease which the NMR operator can operate his instrument over the instrument’s lifetime is often directly related to the skill of the installation service engineer. If the superconducting magnet is properly setup and a set of shim libraries established for all the probes, routine operation of the instrument can be very easy. In fact, when problems are encountered with resolution or lineshape with the instrument even years after the installation, the problems can often be improved by going back to the shim values left by the installation engineer. A common problem in NMR laboratories is that these shim values are not kept or are stored where no one can find them. It’s a little like computer disk backups: they are not needed until it’s too late to do them.
Making sure the room temperature shim assembly is installed in the correct position in the magnet is very important. Often the best position for the room temperature shim assembly is with the center of the room temperature shim assembly at the center of the superconducting shims. The center of the superconducting shims is usually taken from the magnet manufacturer’s data. This can be confirmed with a technique very similar to that used to establish the center of the room temperature shim assembly. There are conditions where the center of the RT shims should not be coincident with the center of the superconducting shims. In general, these arise when a different area of the magnet has a better overall field shape (on and off axis) than the area around the center of the superconducting shims. However, when the center of the RT shims is not the center of the superconducting shims it is more often than not a mistake.
One problem encountered in the superconducting magnet shimming process is the interaction of the various shims. When shims interact, shimming becomes a tedious process requiring many repetitive passes through the shimming sequence. Shim interactions become much greater when the center of the shim set and the center of the sample are not coincident. Since the Z1 gradient has a zero contribution to the field at its center (z = 0), it can be used to locate the center of the shim set. One or both of the processes described below can be used to ensure the probe and shim alignment. If the sample and shim alignment is unknown and the alignment process is not followed, the shimming sequences described later still optimize the field homogeneity, but there are considerably more shim interactions. These interactions make the shimming process more difficult than necessary.
Most NMR systems do not allow the probe position and the room temperature shim assembly position to move independently. For this reason, if one of the following tests indicates that the probe center and the room temperature shim center are not the same, the problem probably resides in the placement of the probe coil in the probe. This is not usually an NMR operator adjustable parameter. If any of the tests reveal a misalignment, check several different probes and see if they are all misaligned. If the problem exists in only one probe, consider having the probe manufacturer check the coil placement. If all the probes are misaligned, something fundamental in the spectrometer installation or design is wrong.
If the NMR spectrometer can display a swept lock signal, the easiest method to check the probe and room temperature assembly alignment is to observe the swept lock display while adjusting the Z1 shim control. The first step is to display the swept lock signal. Then while noting the lock signal display position, adjust the Z1 shim current in one direction and then the other. If the probe is in alignment with the room temperature shim assembly, the lock signal should not move, but should become wider and shorter while staying in the same position. If it does move while becoming broader and shorter, the direction should change as Z1 is moved in different directions. If the peak moves then there is some degree of misalignment between the probe position and the room temperature shim alignment.
A value judgment is called for if the lock signal does change position with Z1 current. In general, if the peak moves less than it broadens then the misalignment is small and is usually best ignored. If the peak moves more than it broadens then corrective action, if possible, would make shimming easier.
If a swept lock signal is not available, an alternative, albeit slightly more involved, technique for determining the relative probe position and room temperature shim alignment is the effect of Z1 current on a plot of probe coil response. While being more involved, this technique also gives more information which is useful for understanding the overall NMR instrument’s operation. Place a drop of water in the bottom of an NMR tube. The smaller the drop the better. Also be as careful as possible to keep water from the sides of the NMR tube. Put this sample in the NMR spinner at a depth such that the drop of water is in the center of the probe’s coil and insert the sample into the spectrometer. The sample should not be spinning. Set up the spectrometer for 1H observation and acquire a spectrum with a sweep width appropriate for the experiment. What is appropriate comes with experience and you will know more about how large a sweep width to use when the experiment is over, but for now use something like +/-2000 Hz. After doing the experiment you may want to go back and collect the data again using a different sweep width.
Process the spectrum using a magnitude calculation and whatever processing parameters are necessary to keep the computer scaling constant for the duration of this experiment (e.g., Bruker or Nicolet AI command). Take an integral of the entire sweep width and set up whatever integral corrections are necessary. If possible set this integral value to 100 and record all future integrals relative to this one. Record the integral value and the frequency position of half height on the integral line. An eyeball guess at the frequency position is usually more than good enough for this experiment. Now adjust the Z1 current in the positive direction an appropriate amount. What is appropriate again comes with experience, but it needs to be a large amount and you should note that you are going to change the Z1 current the same amount in the negative direction next, so do not change the current more than you have available in the opposite direction. Now record the integral value and the frequency position of the half height of the integral with the Z1 current change in the positive direction and again with the same Z1 current change in the negative direction. A table similar to the one below should result:
|Z Position||Z1+ Frequency||Z1+ Integral||Z1=0 Frequency||Z1=0 Integral||Z1- Frequency||Z1- Integral|
+ 90 Hz
– 3 Hz
– 104 Hz
+ 50 Hz
+ 2 Hz
– 65 Hz
– 10 Hz
– 7 Hz
– 10 Hz
– 60 Hz
+ 3 Hz
+ 30 Hz
– 90 Hz
– 5 Hz
+ 102 Hz
Do not be concerned at this stage if the frequencies or integral values are changing by small amounts. If they are changing by large amounts, make sure you did this right by reproducing the values, but if they are changing by small amounts, complete the data collection.
Now move the position of the water drop up 5mm in the spinner and repeat the above experiment. Record the data and move the water drop position up in 5mm steps until the water signal can no longer be observed. Then move the water drop 5mm below the original position. Record the data and move the water drop position down in 5mm steps until the water signal can no longer be observed. Plot the data of the frequency versus Z position of the water drop for the three values of Z1 current. You should get a graph something like the one shown below.
NMR Frequency Versus Position in the Probe for Different Values of Z1
The three lines should cross at one point, but usually you get a small triangle. For this purpose, the center of the triangle is the center of the room temperature shim assembly.
Now plot the data of the integral versus Z position when the water drop is at the Z1 = 0 current. NOTE: the Z1 = 0 current value is not zero current but the starting value for Z0 on your instrument. You should get a graph something like shown below:
NMR Signal Integral Versus Position in Probe
The plot of NMR signal versus water drop position lets you see over what height of sample the probe coil is active. This should describe roughly a rectangular region. The center of this rectangular region should be the region on the previous graph where the Z1 current lines crossed. This method tells the NMR operator where the active coil region is located relative to the room temperature shim assembly. In addition, it tells the NMR operator how long the probe coil is. This information is very useful when determining how long a sample is required in the NMR instrument for shimming. It also detects probe coil lead pickup which can make shimming more difficult. Techniques for shimming different sample lengths will be discussed in the chapter on The Sample and Shimming.
If the plot is roughly what is shown above with the ends approaching zero but with increased intensity further out, then you have detected coil lead pickup. This phenomenon comes from the probe coil’s leads being at a high voltage potential and not being shielded from the sample. If this occurs, it is often the case that the lead pickup is occurring at a region of the RT shims that is not very homogeneous, which leads to lineshape problems.
The unshielded lead problem often reveals itself in another way. When you have a normal single resonance close to the carrier frequency and keep increasing the pulse length past a 90 degree flip to a 180 degree flip, the signal should pass through a null. With lead pickup, most of the signal goes through a null but a small portion of it remains positive even when the rest of the signal is becoming negative with increasing pulse length. The small positive portion of the signal is coming from lead pickup. The sample seen by the leads is outside the main coil region is seeing a much smaller effective field than the main region of the sample. The effective 180 degree flip for this region can be many times that for sample in the main coil region. Often, since this sample can be in a region of different field homogeneity and therefore have a different field value, the small peak which remains positive will be to one side or another of the main peak which is going through a null with a 180 degree pulse.
If either of these symptoms of probe coil lead pickup are observed, the probe should be modified to have a lead shield. This will lead to better lineshapes, pulse homogeneity and solvent suppression experiments.
A shimming procedure follows. It should not be taken as the gospel, but as a suggestion of where to start the learning process about the relationship of shimming procedures, your samples, your instrument and the NMR data acquisition process in general as it relates to your laboratory. Not every instrument will shim the same, nor will every sample. This procedure is a starting place. Modify it at will for your environment. Later discussions will talk about what to look for and do when this process is done. For this procedure the optimization processes are:
Zero Order – This is a straightforward process of adjusting the control for the best response.
First Order – Adjust one control, then the next control, then the next until all controls in the set have been adjusted. Repeat the process until no further response improvement can be obtained.
Second Order – Note the current response level. Adjust a shim control from the current value a defined amount (a rule of thumb would be enough to change the response to between 50% and 75% of the original value) to a new value. Optimize all other shims in the set with a first order process. If the new response is better than the previous response, note the new response level and adjust the shim to a new value in the same direction and repeat the process. If the new response is less than the original response, adjust the shim control a defined amount in the opposite direction and repeat the process. Continue until the best value is clearly determined. This means it is necessary to go too far to make sure no further improvement can be made and then return to the optimum value. It is often useful in this process to plot the results of the shim of interest and the response value after the other shims have been optimized. The operator can then make sure that the changes being observed are real and significant and can better determine the optimum shim value. A typical plot where the Z4 shim is set to a series of values (at which Z1, Z2 and Z3 are optimized) versus the lock response is shown below. This plot gives the operator confidence that Z4 is at its optimal position.
Z4 Plot Technique
If the NMR spectrometer is in a state of unknown homogeneity or is known to have poor homogeneity, a simple optimization of certain shims is the first step in the shimming sequence. A swept NMR signal is the recommended method of judging the response for this operation:
- Spin the sample (20-30 Hz) and adjust the Z1 and the Z2 shims interactively to produce the tallest swept signal response (first-order process).
- Stop the spinner and adjust X and Y for the tallest swept signal response (first-order process).
- Adjust X and ZX for the tallest swept signal response (second-order process).
- Adjust Y and ZY for the tallest swept signal response (second-order process).
- Adjust XY and X2-Y2 for the tallest swept signal response (first-order process).
- If any large shim changes were observed in the above process then the process should be repeated from Step 1.
After the above procedure, the NMR instrument should be capable of a field/frequency lock. Any one of the three methods described above for measuring homogeneity can be used in the following sequence steps.
The adjustment procedure for the spinning shims should be conducted with the sample spinning at more than 10 Hz. Care should be taken at all times to avoid a vortex. A vortex leads to a false shim optimum, with Z2 usually being the most misadjusted. If the lock signal is being used for shimming, then ensure that the lock signal is not being partially saturated with rf power and that the lock phase is correctly adjusted. The lock phase should be reexamined each time a large change is made in an even-order Z shim.
In the adjustment of Z3, Z4, and Z5 described below (steps 2, 3, and 4), it is best to make a plot of the response level versus the shim under adjustment. If the operator is careful to proceed far enough past the maximum response position for the shim under adjustment, then the plot should reveal a broad curve. The best position for the shim can then be determined even if an interpolation between two sample positions is necessary. With experience, this plotting becomes an automatic mental process. Also, confidence is gained that all the shims were correctly optimized when a broad smooth curve is obtained as a result of this process.
- Use the first-order process to optimize Zl and Z2.
- Use the second-order process to optimize Z3. Note the position of Z3 and the response. Change Z3 enough to degrade the response by 20-30%. Repeat the process in Step 1. If the new position for Z3 has yielded a better response, then continue in the same direction. If the new response is poorer, then try the other direction for Z3.
- Use the second-order process to optimize Z4. Note the position of Z4 and the response. Change Z4 enough to change the response by 30-40%. Repeat the process in Step 1. Adjust Z3 to provide the optimum response. If the Z3 shim changes considerably, then repeat Step 1 again and readjust Z3 again for maximum response. If, after optimizing Z3, Z2, and Zl, the new response is better than the previous response, then continue in the same direction. If the response is worse, then try the other direction.
The Z5 shim is difficult to adjust for two reasons. First, only probes with longer coils give significant response change with Z5 owing to its strong dependence on distance. Second, the Z5 shim often has more Zl, Z2, Z3, and Z4 components than a Z5 component in its correction. The Z5 shim normally needs to be adjusted only with wide-bore magnet systems with large-diameter tubes or with longer coil probes. To adjust Z5, note its position and the response. Change Z5 enough to lower the response by 30-50%. Repeat Step 1. Adjust Z3 for the maximum response. Adjust Z4 for the maximum response. If either Z3 or Z4 changed a considerable amount, repeat Step 1 and reoptimize Z3 and Z4. If the new response obtained after this procedure is better than before, continue in the same direction. If the response is worse, try the other direction with Z5.
The nonspin shim set should be adjusted while the sample is not spinning. Changing the nonspin shims which have Z components causes changes in the spinning shim set. If any of these shims change significantly, then the spinning shim sequence should be repeated after completion of the nonspinning sequence. With all shims involving a second-order process, the technique described under the spinning shim sequence of plotting the result and interpolating the shim position should be followed either on paper or mentally.
- Adjust X and Y interactively using the first-order process for maximum response.
- Use a second-order process to adjust ZX. Note the position of ZX and the response. Change ZX by enough to lower the response 10% and adjust X for a maximum response. If the new response is better, continue in the same direction with ZX. If the response is less, try the opposite direction with ZX.
- Repeat Step 2 but using the Y and ZY shims.
- Adjust XY and X2-Y2 interactively using the first-order process for maximum response. If either XY or X2-Y2 changed significantly, then repeat Steps 2 and 3.
- Use a second-order process to adjust Z2X. Note the position of Z2X and the response. Change Z2X by enough to decrease the response 30%. Maximize the response with ZX. Maximize the response with X. If the new response is larger than the initial response, then continue with Z2X in the same direction. If the response is less, then try the opposite direction.
- Repeat Step 5 but using Z2Y, ZY, and Y.
- Use a second-order process to adjust ZXY. Note the position of ZXY and the response. Change ZXY enough to decrease the response by 20%. Maximize the response with XY. If the new response is larger than the initial response, continue with ZXY in the same direction. If the response is less, try the other direction.
- Repeat Step 7 but using Z(X2-Y2) and X2-Y2.
- Adjust X3 and X interactively for maximum response (first-order process).
- Adjust Y3 and Y interactively for maximum response (first-order process).
If the nonspin shim settings have significantly changed, then the spinning shim sequence should be repeated. If there are significant changes in the spin set after optimization, repeat the nonspin set also.
After the spinning and nonspinning sequences have been conducted, the NMR instrument should be delivering less than 0.5 Hz line width with good lineshape and minimum spinning sidebands. This is all that is required for most NMR experiments. If better resolution is desired, then a first-order adjustment of Zl and Z2 using the FID can be helpful. Maximize the response by adjusting Z1 and Z2 for maximum ringout of the FID. If the FID is not observable then the swept lock signal ringout is the best second choice. The ringout of the FID or the lock signal is a very sensitive measure of resolution. Adjusting only Zl and Z2 prevents changes in the lineshape.
Return to Part I, Introduction
Proceed to Part III, Symptoms of Inhomogeneity
Skip to Part IV, The Z1 Profile
Skip to Part V, Effect of Sample and Coil
Last updated: 01/22/03