Pivot peak

876543210 ppm Figure 3.40

876543210 ppm Figure 3.40

(Varian parameter lp, Brnker parameter PHC1) to zero and start over using manual phase correction. This time adjust the first-order phase correction by looking at a peak that is close to the pivot peak, and then move to peaks farther and farther away.

Whether you are adjusting the b (zero-order) or the m (first-order) parameter, when you get close to the correct phase setting, focus your attention on the baseline (or noise line) on either side of the peak in question (Fig. 3.41). This should be at the same (vertical) level on each side of the peak. Expand the peak horizontally and increase the vertical scale first so that these baseline differences are greatly exaggerated. If there are distortions to the baseline (curvature), try to imagine a smooth curve between the noise on one side of the peak and the noise on the other side of the peak. Then make the peak blend smoothly into this imaginary curve on both sides of the peak with neither side extending higher over the curve than the other. Sometimes it is necessary to exaggerate the phase error in both directions, especially with noisy data, to clearly see the phase error (one side of the peak extending below the baseline) and then create the same phase error on the other side of the peak. The correct phase setting will then be somewhere near the middle of these two settings.

3.12.2 Setting the Reference

This is a simple procedure whereby a reference peak (e.g., TMS in organic solvents) is selected with a cursor (Bruker uses a triangle or vertical arrow, Varian a vertical red line) and given a specific chemical-shift value. Without this reference, the chemical-shift scale

Fine tuning the phase correction Figure 3.41

of your spectrum will be meaningless. The primary reference for XH and 13C NMR is TMS at zero ppm, but in many cases this fails or is not practical. Referencing in D2O or 90% H2O/10% D2O presents a challenge because the solvent contains no 13C and TMS is not water soluble. Adding a water-soluble form of TMS (Me3SiCD2CD2CO2- Na+ or TSP) or a small amount of acetonitrile, dioxane, or methanol can give a sharp peak of known chemical shift.

An alternative to the added standards is to use the solvent peak as a chemical-shift reference. If you forgot to add TMS, or the TMS peak is obscured by other peaks, you can use this (residual) solvent peak as the reference peak. This is only valid in dilute solutions where there is only one solvent. Each solvent gives a characteristic residual 1Hpeak in the 1H spectrum due to the 0.2% or so of the solvent molecules that contain 1H, and a solvent13 C peak in the 13C spectrum. For example, CDCl3 solvent is typically 99.8% CDCl3 and 0.2% CHCl3, since it is impossible to get 100% incorporation of 2H into the chloroform molecule. In the 1H spectrum one sees a small singlet peak at 7.26 ppm due to the 0.2% of CHCl3. This can be used as a chemical-shift reference if the normal reference compound (added TMS) is not present, provided there are no solute peaks at 7.26 ppm. In aqueous solutions (D2O or H2O/D2O) the solvent peak (HOD/H2O) chemical shift depends on temperature: 5(H2O) = 7.83 - T/96.9, where T is the absolute temperature in kelvin (°C + 273).

In organic solvents the solvent peak is almost always used as the reference in 13 C NMR. For 13 C spectra in CDCl3 solvent, we observe a "triplet" pattern (1:1:1 intensity ratio) at 77.0 ppm due to the 13C in the CDCl3 solvent. There are three peaks because 2H is a spin-1 nucleus with three spin states possible: spin 1,0, and -1. Just as a single spin-/ nucleus like 1H will split the NMR signal of a directly bonded 13 C into a doublet (1:1 ratio, J ~ 150 Hz), the 2H nucleus splits the 13 C signal into three equally spaced peaks (1:1:1 ratio due to the nearly equal populations of the three 2H spin states). Because the magnet strength of the deuterium nucleus is about 1/7 of the strength of the 1H nucleus (yH/yD ~ 7), the coupling constant is reduced by a factor of 7, and the separation between peaks is around 20 Hz. This solvent 13C peak is usually used for a chemical-shift reference since the tiny amount of TMS added (typically 0.02%) does not give a strong enough peak in the 13C spectrum to be observed over the noise level. You might think that this solvent 13C peak would be enormous compared to the solute peaks due to the preponderance of solvent molecules, but the relaxation of 13 C is very slow if it has no 1H atoms attached, so the peak is usually similar in height to the solute peaks.

Solvents with more than one 2H give more complicated patterns in both 1H and 13C spectra. The 13 C and residual 1H chemical shifts and coupling patterns of all deuterated solvents can be found in charts provided by the solvent manufacturers (isotope companies). For example, CD2Cl2 (d2-dichloromethane) gives a residual 1H peak (from the 0.2% of CHDCl2 present), which is a 1:1:1 "triplet" (J = 1.1 Hz) at 5.32 ppm, and a solvent 13C peak (from the 99.8% of CD2Cl2 present), which is a 1:2:3:2:1 "quintet" (J = 27 Hz) at 54.0 ppm. The "quintet" pattern is due to splitting first by one deuterium into three equally spaced peaks and then splitting each of these by the second deuterium, resulting in a pattern of five peaks. The geminal 1H-2H splitting in the residual 1H solvent peak is quite small due to the reduced magnet strength (y) of 2H, so that often these splittings are barely resolved or not resolved at all depending on the quality of shimming. Even if poorly resolved, the shape of these peaks can be a dead giveaway in identifying them and using them as a chemical-shift reference, or at least for ignoring them in interpreting the solute 1H spectrum. The 13C solvent peak splitting patterns can be quite complicated. For example, d6-acetone (CD3COCD3) gives a "septet" (1:3:6:7:6:3:1 ratio, J = 19 Hz) at

29.9 ppm and a singlet (C=O peak) at 206.7 ppm. The terms triplet, quintet, and so on are placed in quotations because these are not the classical spin-'/2 splitting pattern intensities we will concentrate on in this book. One can diagram these splitting patterns as long as each splitting is diagramed as a division into three equally spaced peaks of equal intensities. You can even draw a "spin-1 Pascal's triangle" as follows:

OneD 111 (CDCI3, C6D6)

Note that each number is the sum of three numbers: the number directly above it, the number above it to the right, and the number above it to the left. The long-range (2 or 3 bond) couplings between 2H and 13C are usually not resolved (~1 Hz) so we do not need to worry about these. Keep in mind that the XH to 13 C couplings are not observed in ordinary 13 C spectra because we are using *H decoupling to actively suppress these couplings. Because 2H has a completely different resonant frequency than *H, the XH decoupling does not affect the 2H to 13C couplings at all.

3.12.3 Peak Lists

You will often want a printed list of chemical shifts for all the major peaks of your spectrum. First you have to set a threshold intensity (Bruker minimum intensity MI, Varian threshold th) below which a peak is not included in the list. If you set the threshold too low, you will get a very long list that includes many noise intensities; if you set it too high, you will miss real peaks. Peak lists can be displayed on the screen next to each peak, plotted on the spectrum next to each peak, or printed out as a list on a printer. With a list showing both ppm and hertz values for each peak, simple subtraction gives the J values in hertz. Be careful of using subtraction of ppm values to get J couplings: these are often not accurate enough. For example, even if ppm values are printed with four digits after the decimal point (e.g., 7.3293 ppm) the precision is 0.0001 ppm or (on a 600 MHz instrument) 0.6 Hz. Subtracting another ppm values increases this error to 1.2 Hz. Much more accurate J values can be measured by printing out peak lists in hertz or by using the software to visually position two cursors and compute the separation in hertz. Another common error is to measure J couplings directly between peaks when the J value is similar to or not much more than the linewidth. If the peaks are not resolved to baseline (intensity dropping to the baseline between the peaks), the distance between peaks is less than the J value because one peak "rides up" on the other, skewing the peak shape and shifting the maximum of the peak toward the other peak. In the extreme of a single peak with a slight "notch" at the top, the difference between the two maxima may be a small fraction of the true coupling. In this case, a resolution-enhancing window function (e.g., an unshifted sine bell) can be used to sharpen the peaks, or a nonlinear least squares fit can be performed to extract both the peak width and the J coupling independently.

3.12.4 Baseline Correction

The "baseline" is the average of the noise part of your spectrum. Ideally, this would be a straight, horizontal line representing zero intensity. In the real world it can drift, roll, and wiggle like a drunken sailor. These errors generally result from erroneous data that are collected at the very beginning of the FID, when the electronics is still recovering from the shock of the RF excitation pulse. This becomes a problem when you try to measure peak areas (see Section 3.12.5). On the Varian, you indicate where you have peaks and where you have noise in your spectrum by indicating integral regions as part of the integration process. Then the command bc(1) fits the noise portions of the spectrum to a smooth function, which is then subtracted from the whole spectrum including peaks. Bruker uses the command abs (automatic baseline straightening) to accomplish the same thing. There are also a variety of more sophisticated baseline correction methods, such as mathematically or visually fitting the noise points to polynomial functions.

3.12.5 Integration

To get quantitation of peak areas (numbers of protons), you need to plot an integral. In the old days before Fourier transform NMR, the plotter was set to integral mode and the pen was swept through the peak as the pen level rose with the integrated intensity. For this reason, integrals are still presented as lines that start at the left-hand side of a peak and rise vertically as they pass through the peak. In FT NMR there is often a problem with baseline "wiggle" and this will lead to inaccurate integration of proton peaks (carbon peaks are essentially never integrated because their peak areas are determined more by differences in relaxation rates than by differences in the number of carbons). To get good integrals, you may need to correct the baseline first (see above). Other causes of inaccurate integration include low pulse power (poor excitation of peaks at the edge of the spectral window), "droop" at the edges due to the response of audio filters, and incomplete relaxation due to short relaxation delays. The first and last become more important for higher field instruments because the spectral width (in hertz) and the relaxation times (r1) increase as the field strength (Bo) increases. Integration generally involves adjusting the display height of the integrals, indicating the start and end points of each peak integral, correcting the drift and curvature so that noise regions give a horizontal line in the integral, and normalizing the peak areas so that the number of protons can be read directly. All processing software allows you to plot the integral area numbers directly on the spectrum next to each integral or to print out a list of integral values. Details of how these steps are accomplished are specific to the software being used and will not be dealt with here.

3.12.6 Plotting

A hardcopy of your spectrum can be obtained using a pen plotter, an inkjet, or laser printer. There are a number of things you can include in your plot:

• integral areas (numerical values)

• parameters

• text describing sample, experiment

Plots can be made on "normal" paper (8.5" x 11") or "large" paper (11" x 17"). Multiple spectra can also be plotted on the same paper, either side-by-side or one above the other (horizontal or vertical "stacked" plots). The plotting procedures for Varian, Bruker, and "third" party software packages are quite different, so these details will not be covered here. Usually the software will also plot to a file, in PostScript format or the more PC-friendly TIFF or JPEG formats. These files can then be introduced into PC drawing programs and annotated with structure diagrams, text, arrows, and lines for use in publications and posters. For example, most of the spectra in this book were processed using Felix software and "plotted" to a graphics file in PostScript format. This text file is converted to a bitmap and imported into a drawing program on a PC.

3.12.7 Archiving (Saving) Your Data

If you had the foresight to save your data on the NMR instrument's hard disk, you will find that these data must be periodically "purged" as the disk gets full. Why save it forever? Someday you will be writing up a paper or thesis and will ask the inevitable question, "what was the coupling constant for that triplet at 3.5 ppm?" Since you probably did not anticipate this question when you plotted your spectrum, you will need to get the data back and reprocess it. At that point you will thank yourself profusely for having the foresight to archive your data. NMR data are generally archived in the raw (FID) form so that you have the maximum flexibility in processing it. Data can be transferred to a PC via the internet or a local network using file transfer protocol (ftp) or the more secure and modern version SSH secure file transfer. From the PC, it can be saved on a CD-ROM or DVD. Bruker and Varian software give data file sizes in terms of the number of data points (Bruker TD, Varian NP). But because each data point uses 2 bytes (Varian with dp = '«'), 3 bytes (Bruker AM), or 4 bytes (Bruker AMX, DRX or Varian with dp = '/) of data, the actual file size of a 16 K FID (16,384 data points) can be a little more than 32 or 64 kB on Varian (depending on the dp setting), a little more than 48 kB on the AM or 64 kB on the DRX (the "little more" is for a file header). When you transfer it to a PC (Windows), you will see the file size in actual bytes. On modern UNIX-based NMR instruments, the NMR data file is actually a directory that contains a number of files in addition to the FID binary data file, and even may contain other directories. When transferring these data by ftp or SSH secure file transfer, be sure to specify that directories as well as files are to be transferred. Sometimes long file names will be truncated by Windows when the files are transferred to the PC, and restoring files can cause problems because the names are not correct when they return to the UNIX environment. This problem can be fixed on a case-by-case basis and is never disastrous.

0 0

Post a comment