## Ft

F(Fl9F2) Imag.,Imag.

### Figure 9.53

through a peak and phase correct that 1D spectrum to get the F\ phase correction parameters. This works well for 2D experiments with strong crosspeak intensity, such as TOCSY, but for others we may see only one peak in a given 1D slice. How do you come up with a chemical-shift dependent (first-order) phase correction of a spectrum with only one peak? The solution is to use more than one slice to generate multiple 1D spectra with peaks in different parts of the spectrum. Bruker software uses a display with three 1D windows, all controlled by the same two phase parameters (phc0 and phcl). Three different slices of the same type (e.g., rows) are loaded into the windows, the pivot peak is selected in one of the windows and the two phase parameters are adjusted, with all three 1D slices responding in real time to the adjustments. When the optimal parameters are found, this phase correction is applied to all rows of the data matrix. Varian software will only do horizontal slices ("traces") so the matrix has to be turned on its side (command: trace = "/l") in order to load an F1 slice. The phase correction parameters (rp and Ip) are determined by treating the slice as a 1D spectrum. If you need more than one slice, you can adjust rp with one slice (with a peak on the right) and Ip with another (with a peak on the left side). Other software packages (e.g., Felix) construct a 1D spectrum as a sum of several slices from the 2D matrix. For example, three columns (F1 slices) can be selected and summed to give a 1D spectrum with three peaks, which is then phase corrected and the parameters are applied to all the columns of the 2D matrix. Regardless of the software used, it can be tricky if there are significant phase errors in both dimensions because the F1 phase errors can "flip" peaks upside down in the F2 slices. When making multiple F2 slices (rows) of the 2D matrix, select a row just above or below the center of a peak (Fig. 9.52, upper right) if there is a significant F1 phase error. Try to be consistent in making all slices on the same side (e.g., all rows just above the center or all columns just to the left of the center) to keep the phase errors in the other dimension from interfering.

2D spectra with very weak crosspeak intensities (e.g., NOESY) require very careful phase correction because dispersive "tails" (Fig. 9.39) can extend far outwards from the intense diagonal peaks to obscure weak crosspeaks. Sometimes you will also have to "flatten" the baseline to be able to lower the contour threshold enough to see the weak crosspeaks. If there is curvature in the baseline (in two dimensions that would be like a rug held up at the corners and sagging in the middle) you cannot see the weak peaks because the threshold plane cuts through the noise in some places. Baseline errors appear as streaks extending in both directions (up/down or left/right) from a peak with the same sign (same color code) on both sides. There are many techniques to do 1D baseline flattening. The most common method is to generate a 1D spectrum and manually set up a series of baseline "points" which represent specific chemical-shift positions where there is only baseline noise and no peaks.

The program fits the intensity at these point values to a polynomial (up to 5th order) function and then subtracts the polynomial function from the whole dataset. This is repeated for each 1D slice (row or column) of the 2D data matrix. More sophisticated methods calculate the baseline points automatically and use functions other than polynomials. For example, a program called FLATT (by Kurt Wiithrich) is very effective at removing horizontal or vertical streaks resulting from baseline curvature in rows or columns of the data matrix. Especially with NOESY and ROESY data baseline correction is essential to getting "clean" 2D displays and plots.

The following table summarizes the 2D parameters for Bruker and Varian.

 Varian Bruker Varian Bruker
0 0