|
Introduction to practical NMR at UNSW
These notes are designed for newcomers to practical NMR in the UNSW NMR Facility. Their aim is to answer questions such as "Why am I twiddling this knob and what happens when I do?" which all beginners should have! Much emphasis is placed on the procedures of locking and shimming, as these are the least understood and most troublesome areas for learners. With the high degree of automation now present on the 300 MHz instruments it is tempting to pay no attention to these matters. Ignore learning these basics at your own peril!! Although written with DMX/DPX instruments in mind, the basic procedure is general to using any NMR instrument for acquiring routine solution state spectra. These notes are meant as an accompaniment to the Facility User Notes which contain step by step procedures for acquiring data but with minimal explanation as to what is happening during these processes.
More detailed essential reading for practical NMR can be found in:
1) "Modern NMR spectroscopy : a guide for chemists" - chapter 1
Jeremy K.M. Sanders and Brian K. Hunter, 2nd edition. Oxford University Press, 1993.
Which has a workbook associated with it:
"Modern NMR spectroscopy : a workbook of chemical problems" - section 1
Jeremy K.M. Sanders, Edwin C. Constable, Brian K. Hunter and Clive M. Pearce. 2nd edition. Oxford University Press, 1993.
Or better still:
2) "Modern NMR techniques for chemistry research" - chapters 1-3 and 5.3.4
Andrew E. Derome, 1st edition. Pergamon Press, 1987.
Acquiring routine spectra
Basically, this can be broken down into ten sections:
- Sample preparation
- Putting your sample in the magnet!
- Locking
- Shimming
- Setting up acquisition parameters
- Acquiring the data
- Transferring to the PC/workstation
- Processing
- Plotting
- Archiving
1. Sample Preparation
This is a much neglected area! If you make the sample up poorly it is impossible to get a good spectrum. There are numerous factors to consider:
Should be deuterated to provide a lock signal (see below). It is possible to run in protio solvents if it is impossible to find an affordable deuterated solvent but this complicates acquisition. Compound must dissolve - 5 mg in 0.6 ml is plenty for protons (about 0.02 - 0.05 M). Other nuclei may require maximum amount of sample dissolved, e.g., 13C.
For proton work, too much sample can be bad as lines can become broader - use a dilute solution for best results (probably up to about 30 mg of medium molecular weight (~500) compounds OK on the 300's).
| Cost |
Solvent |
| Cheap |
CDCl3, D2O |
| Moderate |
Acetone-d6, C6D6 |
| Expensive |
Toluene-d8, CD2Cl2 |
| Very Expensive |
THF-d8 |
Viscosity - lower viscosity solvent => sharper lines
e.g.
acetone - low viscosity
DMSO - very viscous
Must be liquid over range of interest for variable temperature experiments. Beware using solvents such as D2O and methanol-d4 if you wish to see protons that can exchange with deuterium in these solvents, e.g., OH and NH protons. Peaks due to solvent overlapping signals of interest are avoidable with judicious choice of solvent.
Use approximately 0.6 ml for good/easy shimming. This corresponds to a sample depth of 3.5 - 4cm Use approximately 0.4 ml with limited sample requiring 13C, for example, to maximise concentration. Special tubes (Shigemi) are available to restrict volume to 0.25 ml in cases where lack of sample is severe.
For best results filter the sample! Solid particles can degrade resolution drastically! Filtering can be done through a small plug of cotton wool, etc., wedged into a Pasteur pipette. Avoid introducing/remove paramagnetic impurities (e.g., certain metal ions introduced by corroded syringe needles) as these broaden lines. On occasion degassing is required to remove O2 which is paramagnetic.
- Tubes - Wilmad preferable.
Hot ovens can distort tubes so they will no longer spin.
Avoid cleaning with paramagnetic materials e.g., chromic acid.
- Money saving tip - for X-nucleus observation only, use only 10-20% deuterated solvent (sufficient to lock) - particularly important if using 10 mm tubes on Pulse.
2. Putting your sample in the magnet!
This seems like a ridiculously obvious thing to include, but it is one of the few chances you have to do serious damage to the instrument. The first rule of NMR is MAKE SURE THE EJECT AIR IS ON BEFORE YOU PLACE YOUR SAMPLE ON THE MAGNET BORE!!
Before placing your sample (in the spinner!) in the magnet, the depth of the sample tube in the spinner must be set correctly. Too far down and the sample will not spin and may even break in the instrument. Not deep enough and it will be difficult/impossible to shim (see below).
Wash your hands first if you've just had a lardfest lunch at McDonalds or the like! Always clean the sample tube with tissue before and after placing in the spinner. Handle the spinner with tissue as much as possible, as grease from hands accumulates on the spinner and eventually in the probe, which may cause spinning to stop.
3. Locking
The function of the lock is to prevent drift of the magnetic field over time.
Magnetic fields produced by NMR magnets will fluctuate slightly over time if left unchecked. The frequency of an NMR line is directly proportional to the applied magnetic field. This is the most important relationship to remember in NMR! Usually, the NMR experiment is repeated several or many times over a period of a few seconds to many hours and the data are added. Therefore, the field must be held very constant over this period, otherwise the resonant frequency may slowly change which would cause broadening of the resonance.
- How good does it need to be?
Under favourable circumstances, the width of a proton NMR line at half height may be 0.3 Hz, compared with its actual frequency of say 300,000,000 Hz. So, the field must be held constant to 1 part in 1,000,000,000 over time to prevent broadening!
It turns out then, that NMR itself is a most sensitive measurement of field. The lock is effectively an extra NMR spectrometer that works at the frequency of the deuterium in the solvent and operates continuously. It monitors the frequency of the deuterium resonance against a fixed frequency produced by the instrument. This fixed frequency (and indeed all frequencies generated in NMR machines), in contrast to an uncompensated magnetic field, is highly stable, reproducible and invariant over time - the frequencies are produced from oscillators similar to those found in (very accurate) digital watches.
In addition to the large magnetic field produced by the superconducting magnet, the instrument contains a small solenoid electromagnet which has a coaxial field and adjustable strength. So, for example, if the field is tending to increase, the frequency of the deuterium line will tend to increase as well. This increase is detected by the lock system and compensated for by a small decrease in the strength of the field produced by the electromagnet. This is known as a constant feedback mechanism. So, the lock system works by monitoring the frequency of the deuterium resonance of the solvent and making tiny adjustments to the field to hold the frequency as constant as possible.
Practical aspects of locking:
- Locating and centring the signal
Immediately after placing the sample in the instrument, it is necessary to first locate and centre the lock signal. Nowadays, the spectrometer usually does this automatically via the lock command. It's useful to know how to do things manually when things go wrong though! When the lock is switched off, the machine will, by default, "sweep" the field around a central value to facilitate location of the lock signal, raising then lowering the magnetic field in cycles. This sweeping of the field occurs when the light on the button labelled "SWEEP" is on (it automatically switches on/off when the "LOCK" or "AUTOLOCK" buttons are deactivated/activated respectively). In the unlocked mode, as the lock display travels across the screen this corresponds to the field in the electromagnet being increased in one direction of travel or decreased in the other. The centre of the screen represents where the lock signal should be located in order for the field to be such that the resonant frequency of the solvent matches the lock frequency produced by the lock system.
If the solvent being used is the same as the previous one, then at some point near the centre of this field sweep, the correct field to match the resonance condition for the deuterium lock signal is obtained and the lock signal will appear on the screen, followed by a succession of "beats" known as a ringing pattern. However, if a different solvent is being used, it is likely that the lock signal will not appear anywhere within the current window of sweeping. This is because different types of deuterium in different solvents naturally have different chemical shifts, i.e., resonance frequencies. For example, the resonant frequency of acetone-d6 is lower than that of chloroform-d by 5.21 ppm. On DMX/DPX series instruments, the observation windows are fixed, so to make the resonance condition again, the lock frequency must be lowered by 5.21 ppm if we change from chloroform-d solvent to acetone-d6. This is controlled by the lock software which has a frequency programmed for each solvent. This means that the instrument always places your residual proton solvent signal at the correct chemical shift in the proton spectrum providing the spectrometer is told correctly what the solvent shift is. Conversely, if you tell the spectrometer the solvent is acetone-d6 but in reality the solvent is chloroform-d, the residual chloroform peak will appear at the acetone frequency of 2.04 ppm instead of the correct chloroform shift of 7.26 ppm. This can also be a problem for lock solvents with multiple types of deuterium sites as care must be taken to lock on to the correct deuterium signal.
The lock signal may be located manually by using the "FIELD" button, which controls the strength of the central value of the field produced by the electromagnet. There is also a button labelled "SWEEP AMPL" which controls the amplitude of sweeps made. It is often useful to increase the sweep amplitude in order to locate the lock signal initially. After centring the lock signal with the "FIELD" button, the sweep amplitude may be reduced again.
(NB. In practice, decreasing the value of the "FIELD" on the display, which is in arbitrary units, actually causes an increase in the real magnetic field)
Once the "FIELD" parameter has been correctly adjusted, the lock can be activated by pressing the "LOCK" button. Note that as soon as the lock is activated the light on the "SWEEP" button light automatically goes off, since now the objective is to hold the field as constant as possible. Once locked ("LOCK" light on continuously), the lock signal display has a different appearance and function - the line will stay basically level (except for some noise) and the height of the signal on the screen represents the height of the lock signal.
Prior to activating the lock, the correct lock power should be set. The lock command does this to a first approximation automatically. The lock power actually controls the amount of radio frequency power that continuously irradiates the deuterium nuclei. Some of this radio frequency is absorbed by the sample, since there will be more spins in the lowest energy spin state of the deuterium at thermal equilibrium. This net absorption of radio frequency energy tends to equalise the populations of spins in the different energy levels of the deuterium. With the correct amount of power, this tendency to equalise the populations in energy levels is offset by relaxation processes which tend to restore the thermal equilibrium. However, if too much lock power is put into the system, relaxation is insufficiently fast to prevent the energy levels approaching the situation where they have equal populations. This process of equalising populations in energy levels is termed saturation and results in the lock signal disappearing since the size of the NMR signal is directly proportional to the population difference.
It is quite easy to saturate the lock solvent. If saturation occurs, then the lock level will visibly fluctuate up and down and performance is wrecked. Since the deuterons in different samples relax at different rates, the amount of lock power that can applied is also solvent dependent. Acetone-d6 for example relaxes slowly, saturates easily and so must be used with little power.
Too little lock power is also undesirable as the signal becomes smaller and so the signal to noise ratio is lowered which also degrades performance. If set really low, the lock will not operate at all.
The lock gain controls the amplification of the detected lock signal. At first glance, the lock gain appears to have the same effect as the lock power in that increasing it increases the lock level, just as increasing the lock power does (until saturation begins). However, increasing the lock gain also amplifies the noise and so the lock level becomes noisier as well as higher. Unlike lock power, there are no major problems caused by turning up the lock gain too much. The lock gain should be adjusted such that the lock level is visible in the top third of the display.
The lock phase must be correctly set for efficient operation of the lock. Use of the "lock" command automatically reads in an approximately correct value. The lock system feedback mechanism actually monitors the dispersion mode signal of the deuterium, thus correct phase adjustment is required to make this so. (For a description of phase and dispersion mode lines see section on processing). The lock phase does not vary from sample to sample as such, but does vary slightly with lock power. The easiest way to adjust the lock phase is to adjust the lock phase to give maximum lock level after locking, since an optimum phase setting gives maximum lock level (note that the lock level monitors the absorption component of the lock signal).
If the phase is grossly miss-set, the instrument will not lock at all. Indeed, if the phase is out by 180 degrees, a situation of "anti-lock" is produced. To tell if this is the case, it is easiest to look at the lock display in unlocked, sweeping mode:
It may be necessary to slow down the sweep rate in order to reproduce these figures exactly.
The loopadj auto program will automatically set the lock gain and phase as well as some other “hidden” lock parameters (loop gain, time, filter). It is highly recommended for long term acquisitions. The procedure is:
- Adjust Lock Power for saturation. To do this, lock and shim the spectrometer then SLOWLY increase the lock power well above its normal value until the lock signal starts to oscillate up and down
- Then adjust the Lock power 2 to 3 dB below that point.
- Start the AU program with xau loopadj from the TOPSPIN command line.
This au program will automatically adjust the lock phase and lock gain to appropriate values. The lock gain value is read into the equations to calculate the correct loop time, filter and gain for that lock gain value. The results will give optimal stability for long term experiments.
4. Shimming
Shimming is performed to make the magnetic field as constant as possible throughout the sample.
For the same reasons that it was desirable to keep the magnetic field constant with respect to time, i.e., to keep the lines as sharp as possible, the magnetic field must be made as constant or homogenous as possible throughout sample volume. This, again, is because small differences in field in different parts of the sample lead to different resonant frequencies. A whole range of different regions with different frequencies leads to lines which appear broad or have distorted shape. In extreme cases, resonances which should be single lines may appear as 2,3 or more distinct lines.
- How good does it need to be?
On a 300 MHz spectrometer, field inhomogeneity of the order of less than 1 part in 109 is required to give 0.3 Hz proton lines. On a 600 MHz spectrometer it needs to be of the order of less than 5 parts in 1010. That's not very much!
A series of adjustable magnetic field gradients with various spatial shapes called shims (for traditional reasons) are located around the sample. The strengths of these gradients are altered in order to "fudge" the magnetic field to be as homogeneous as possible.
Each sample actually distorts the magnetic field differently, since each sample has different magnetic susceptibility and sample length. This means that the shims must be readjusted for each sample.
Practical aspects of shimming
- How do you know when the shimming is improving?
There are several methods to test whether the shim adjustments are improving the homogeneity of the field. The simplest is to monitor the height of the lock level when the machine is locked. The lock level represents the height of the lock signal. Since there is a fixed amount of deuterium in the sample, the area of the resonance is fixed, so if the height of the lock signal increases as a shim is adjusted, this means that (usually) the width of signal is getting less and hence the homogeneity is improving.
Another trick used to remove the effects of field inhomegeneity is spinning the sample. For superconducting magnets, the field direction is perpendicular to the floor along the length of the sample tube. The field direction is referred to as the Z direction. The plane parallel to the floor is the XY plane as far as the shims are concerned. Spinning the sample about the Z axis at a rate typically around 20 - 30 Hz moves each part of the sample through a variety of environments in the XY plane and, to a large extent, averages out small local field variations in these directions. This means, that when the sample is spun, small field inhomogeneities in the X and Y directions do not need to be "shimmed out."
Spinning is becoming much less used these days, especially on high fields (500+) where the spinning often makes spectra look worse due to artifact lines called rf sidebands. High quality shim sets mean that off axis shims (one's with components in the XY plane) produce highly homogeneous fields without spinning. Spinning is not used in most multipulse experiments e.g., 2D measurements either.
The routine shimming method, with the sample spinning involves selecting shims that contain only a Z component - i.e., Z, Z2 and sometimes Z3 and adjusting till maximum lock signal is obtained. The shims containing only Z components are referred to as the "on axis" or "spinning" shims, as these may be adjusted with the sample spinning.
Unfortunately, the shim settings are not independent of each other and so adjustment of one shim may require readjustment of another one. For example, the two most important shims, Z and Z2, interact extensively and several repetitions of adjusting each one in turn may be required to find a maximum. A general rule is that when a spinning shim is adjusted, all spinning shims of lower order (index) should be readjusted. So, for example, whenever a change in Z3 is made, both Z and Z2 should be readjusted/checked. Shimming is a skill that requires practice. The key to good shimming is patience!
During shimming, the lock level often increases above the top of the display. The lock gain should then be reduced so that the level is in the top third of the screen. Once the shim is close to being optimised, it is a good policy to check the lock phase as described above, as it will be more sensitive when the shims are close. A correct lock phase is essential when using lock level as the shimming indicator.
The routine shimming methods only work well if the shims are close to start with. Since the shim currents are created digitally, it is possible to store shim files which contain all the shim values obtained by an experienced operator shimming on a high quality test sample. In general, making the assumption that no matter who was on the machine before you, the shims have in fact been messed up and that the appropriate shim file should be read in (e.g., "rsh SHIMQNP" on DPX-1 spin) is the way to go! Note that the closer the properties of your sample, particularly the length of sample (ideally 3.5 - 4 cm deep) to that of the standard sample used for the shim file creation, the easier the shimming job will be. Short samples (< 3 cm deep) generally require extensive re-shimming. With a sample changer, this corresponds to more time spent on your sample!
Gradient Shimming
See the separate notes here
5. Setting up acquisition parameters
- Start with a parameter file
After locking and shimming, the instrument must be set up to acquire the spectra of interest. For routine spectra this is most easily done by reading in a parameter file containing sensible default values to do the job. For example, on the DMX 500, there are standard parameter files for acquiring proton (1Htbi.500) or carbon (13Ctbi.500) spectra. The "rpar" command is used in TopSpin to call in the stored parameters. Using these parameters as is will undoubtedly get you some sort of spectrum. However to obtain optimum spectra, as often as not, it is necessary to modify some of these parameters. To be able to sensibly modify the parameters, some understanding of the experiment is required.
- The "one pulse experiment"
The simplest NMR experiment, which is used for collecting proton spectra is the "1 pulse sequence":
The detailed theory is not dealt with here. Basically, the radio frequency pulse (PW), delivered through coils around the sample tube, excites the spins in the sample. After the pulse, the net effect of the spins is that they behave like tiny bar magnets rotating inside the same coils. Just like a dynamo, these "rotating bar magnets" induce an oscillating voltage in the coils. This is the NMR signal, called the free induction decay (fid). The frequencies of oscillation are indicative of their chemical shifts. Over a period of time, typically several seconds for protons, this signal dies away due to relaxation processes. During this period (AQ), the voltage is sampled at regular intervals, amplified, digitised then stored in computer memory. After the signal has decayed, it is often necessary to wait a further amount of time, the relaxation delay (RD), before the experiment is repeated the prescribed number of scans, defined by the "NS" parameter.
The centre of the spectrum is defined by the frequency used for the pulse, so its frequency must be set to be in the middle of the expected chemical shift range prior to acquisition. On the DMX/DPX, the parameter which controls the centre of the window of the observed nucleus is "o1" (units of Hz) or " o1p" (units of ppm), the later being easier to use when a precise frequency is not required.
Most simple 1D NMR experiments use a pulse width (PW) which gives significantly less than a 90° pulse, typically 30° . This usually has a duration of several microseconds. A 90° pulse gives maximum signal excitation. However, the system takes longer to recover after a 90° pulse, so a smaller flip angle allows for faster repetition (shorter relaxation delay, RD).
The magnetisation is sampled at fixed intervals after the pulse. The time between sampling is the dwell time, DW. In fact, the NMR signal is detected along two perpendicular "directions" by two phase sensitive "detectors." This is termed quadrature detection and is required to determine if frequencies are positive or negative with respect to the centre of the spectrum.
There are 2 methods of sampling the data:
- Collect the data points in both phase sensitive detectors simultaneously.
- Collect the data points in both phase sensitive detectors sequentially.
The DMX/DPX can use either method.
How fast should we sample the signal? The rate of sampling governs the spectral width in Hz (SWH) that is covered (Nyquist Theorem). It can be shown (Derome p. 14) that for sequential sampling (method b):
SWH = ½*DW
So if we wish to cover a range of 10 PPM for protons, on a 300 MHz Machine (i.e. 3000 Hz), DW should be 167 μs.
Signals which lie outside of the sweep width can fold back into the spectrum. Folded signals can be spotted due to their dispersive phase properties. On the DMX/DPX however, oversampling and digital filtering means that there are no folded peaks.
- Acquisition times and digital resolution
The overall length of the acquisition time (AQ) is the product of the dwell time and the number of points (TD on Bruker systems) collected:
AQ = DW * TD
The overall length of the acquisition time determines the resolution of the final transformed spectrum. The longer the signal is sampled, the better its frequency may be defined and the more closely spaced the points will be in the final spectrum. In general, if we wish to distinguish 2 lines separated by n Hz, the acquisition time should be at least 1/n seconds.
Digital resolution, Rd = 1/AQ
(N.B. the lower the value of Rd in Hz/point, which is the separation of data points in the spectrum, the better the digital resolution.) It is always worth remembering that the fid and the transformed spectrum are actually just a series of dots, which the display joins together to make visualisation easier.
So, too short an acquisition time can lead to loss of information. Alternately, acquiring much longer than the signal lasts collects only noise. A useful rule of thumb is to have the acquisition last between 1.5 and 3 times as long as the signal endures (if distinguishable from noise). This means that, often, the number of points collected for proton NMR may need to be increased - particularly if the sample was prepared correctly and the instrument was shimmed well! Note that reducing the spectral width also increases the acquisition time (for fixed TD) since the dwell time, by definition, must increase.
- Receiver gain, dynamic range and digitizer resolution
Each time a data point is sampled, the voltage is amplified then the analog to digital converter (ADC) converts the value of the voltage for storage as a number in the computer. The amplification is controlled by the receiver gain (RG). The receiver gain should be adjusted to be as high as possible without overloading the receiver. The ADC converts the signal to a binary number. The maximum number depends on the ADC (digitizer) resolution, i.e., the computer word length used by the ADC. This is not to be confused with digital resolution! For the DMX/DPX this is usually 16 bits. If the signal is amplified too much (receiver gain too high) then the data point value is bigger than the maximum number the computer can store. This leads to a "clipped" fid which manifests itself as baseline distortions after Fourier transformation:
Usually, the receiver gain is set automatically with the RGA command, but may be set manually using the GS command. Note that the fid is usually most intense at its start (left hand side on the screen) so close attention should be paid to the first few data points to ensure there is no receiver overload in the first few data points.
The recycling (or relaxation) delay is adjusted to optimise the repetition rate of the experiment.
If accurate integration is required, which is often the case for proton NMR, the total recycling time i.e., time between repetitions (AQ + RD) should be long enough for complete relaxation of all protons (> 5 times longest T1). Relaxation to restore the equilibrium is an exponential process, usually defined in terms of a time constant, T1. T1 is basically the time taken for ~64% of the difference of the populations from their equilibrium values to return towards these equilibrium values.
For optimum signal to noise, usually required by 13C NMR, RD (and PW) are adjusted according to the Ernst angle equation:
cos α = exp( -τ / T1 )
Where τ is the repetition rate = (AQ + RD) and α the pulse angle.
The problem is that different resonances in the same sample can have very different relaxation rates and are obviously unknown when running the spectra. Knowledge of the relaxation rates of similar systems is often a good starting point though.
Repeating the experiment multiple times improves the ratio of the signal to noise (S/N) ratio. The amount of signal is proportional to NS, but unfortunately the noise also increases as the square root of NS. The net result is:
S/N is proportional to √NS
For simple spectra on the DMX/DPX systems, it is usually desirable to collect a multiple of 4 scans.
6. Acquiring the data
There is little to say about this as the quality depends upon the prior setup. Keep away from the magnet while experiments are running, especially sensitive 2D experiments such as NOESY.
7. Transferring to the PC/workstation
Spectra should be processed on the PCs as much as possible to avoid wasting instrument time. You don’t need the $300,000 instrument, complete with magnet, to do the processing - the $2,500 PC does a better job anyway! Notes for the transfer procedure are in a separate guide.
8. Processing
- Time and Frequency Domains
The fid is called the time domain response (i.e., measure response as function of time).
Normally, we wish to examine the frequency domain data (response as a function of frequency). Convert between the two using the Fourier Transform. Fortunately, typing FT usually accomplishes this task!
Before Fourier transformation (FT), it is usual to apply some sort of window function to the fid aimed at either enhancing sensitivity or resolution. It is often a good policy to process a proton spectrum twice or more - once with a small amount of sensitivity enhancement to improve the quality of the integrals and detect low intensity peaks and once with resolution enhancement to resolve the maximum number of lines to extract coupling constants. A good way to learn about window functions is to play with the "winfunc" command in TopSpin - this is not implemented in TopSpin yet.
- Sensitivity Enhancement - Exponential Multiplication, EM
The fid is multiplied by an exponentially decaying function:
This function emphasises the signal at the beginning of the fid, where signal to noise (S/N) is highest. However, the apparent rate of signal decay is increased, so lines appear broader. This property of the lines being broader in the frequency domain the faster the signal decays in the time domain can be thought of as an expression of the uncertainty principle - the less time we can measure the signals frequency for, the more uncertain we are of its frequency, hence broader.
The decay constant of the exponential is usually entered as a line broadening LB parameter. Optimum S/N is achieved when LB is set to the observed linewidth at half height for the signal of interest in the case of no window function being applied. For example, if after processing by just using FT a line is measured as being 2 Hz wide at half height, then maximum signal to noise can be achieved by setting LB to 2 Hz and then processing with EM prior to FT. This will result in an apparent line width of 4 Hz. When a range of linewidths is present, multiple processing with a range of LB values is in order.
- Resolution Enhancement - Gaussian Multiplication, GM
Frequently, especially for proton spectra, more information can be obtained if the lines are made narrower (e.g., resolve more couplings). The most appropriate window function is applied using Gaussian multiplication (GM):
This function has 2 phases. The initial rising phase suppresses the natural decay of the fid signal, apparently sharpening the lines. The maximum is positioned approximately where the signal of interest is no longer visible. Thereafter, the decreasing function attenuates noise. In practice, two parameters determine shape of this function (with Bruker software):
LB - should be set negative to decrease linewidth. Usually start with LB set to minus one half times the natural linewidth at half height. If LB made too negative severe lineshape distortions will eventually occur.
GB - determines position of maximum as fraction of acquisition time (0 < GB < 1 in TopSpin; 0 < GB < 100% in WINNMR). Normally set to fraction of fid where signal "disappears" into the noise.
After 'GM' + 'FT' the lines are sharper, but s/n is much poorer and integration is no longer reliable. Usually zero fill before applying GM.
Adding zeroes to the end of an fid before FT gives a cosmetic improvement in spectral appearance. The extra zeroes make the apparent acquisition time longer, resulting in apparent better digital resolution. Zero filling beyond a factor of 4 will have negligible effect however. Zero filling is effected in practice by making the transform size greater than the number of points acquired. In TopSpin used on the DPX/DMX, the size parameter, SI, is automatically twice the size of the TD parameter. SI must be a power of 2. For example, if we collected 16K points (TD) and later transformed with SI set to 32K, we have zero filled by a factor of four.
Quadrature detection requires collecting 2 components of fid with 90° phase difference. The 90° phase difference means that one component should be the sine component of the oscillating signal and the other the cosine component (sine and cosine differ in phase by 90° but are otherwise identical.)
Ideally, after FT one component would be absorption mode (cosine) or "real" and one dispersion mode (sine) or "imaginary":
For analysis purposes, the dispersive line is ignored. To convince yourself that there really are two components, click on the "Im" button (imaginary) in TopSpin, which toggles the display to showing only the dispersion mode line. Click on the "Re" button (real) to switch back.
Normally after FT the 2 components are neither pure sine or cosine components, but a mixture of the two:
To compensate, make a phase correction, which is expressed in degrees. In practice, make a zero and first order phase correction. The zero order is independent of frequency due to difference in pulse phase and receiver phase.
Simply take linear combination of real (R) and imaginary (I) components:
A = R cos α + I sin α
D = R sin α - I cos α
A - true absorptive component, D - true dispersive component
Adjusting the PH0 button in the phase window of TopSpin or the zero order phase slider in WINNMR adjusts the value of a . Usually use largest peak in spectrum as reference point.
First order - Varies linearly with offset from carrier (centre) frequency
Corrects for errors due to off resonance and delayed sampling of first data point.
First order correction = β ν
ν - offset from carrier
Adjusting the PH1 button in the phase window of TopSpin or the first order phase slider in WINNMR adjusts the value of b. Usually use lines as far away as possible from zero order reference point to monitor the first order correction.
If the reference substance (TMS for 1H and 13C) is present in the sample, this is simply set to the appropriate chemical shift value (0 for TMS). Usually, there is no reference substance present and so the solvent, which has a known chemical shift, is referenced instead. However, it often happens that the solvent resonance is overlapped and cannot be identified. In this case, the machine can simply be given the correct spectral referencing value, SR, which will calibrate the spectrum very accurately. The instrument/workstation references spectra by storing a frequency (in MHz), SF, the frequency of 0 PPM. So, changing SF or SR simply resets the zero PPM point. By definition, SF = BF1 + SR.
SR values for each nucleus are stored in the parameter files. On the DMX/DPX, the SR values should be very similar for different solvents. They can simply be entered to reference the spectrum. Using the SR values works because the frequency used for the lock signal is invariant. This means that the magnetic field for a given lock solvent and hence the frequency of 0 PPM for any nucleus is always exactly known.
Most NMR users consider this a straightforward procedure. In fact, it is far more complex than most people realise. A separate set of notes on integration is available.
9. Plotting
In my humble opinion, plotting is often the hardest thing about doing NMR! This is because so many parameters are used to create plots and usually you get one wrong along the way. Because of this, several large trees worth of paper is wasted in every NMR lab every year. The trick is to use the plot previewers that are invariably available, (X)WINNMR being no exception, before plotting. Unfortunately, there is no substitute for reading the manufacturers manual to find out about all the parameters.
10. Archiving
Always store all of your data - you never know when you might need it again!
Store the raw fid data and not the spectrum or processed data, as the fid may need to be massaged in different ways during data processing. This also uses half (or less) space. Zip disks are convenient for storing many 1D and small 2D data sets. Writeable CDs are good for larger amounts of data.
Back to Local Information
|
Privacy Policy
email Enquiry