In 2D NMR spectroscopy there is no detector in the F; dimension; therefore the signs of the frequencies in F, must be determined beforehand during the acquisition in F,. A specific phase cycling for the individual FIDs obtained in F, for each 4, increment provides the necessary basis for this.

There are two different approaches. In the phase-sensitive mode sine and cosine components with respect to the ¢, evolution are created by the phase cycle and stored separately. Subsequent real Fourier transformation detects the signs of the frequencies and the phase of the signals. This is almost equivalent to the procedure used in 1D NMR spectroscopy, and can be performed in either the simultaneous or the sequential mode; the former is called the Ruben–States-Haberkorn procedure and the latter the TPPI (Time Proportional Phase Increment) or Redfield method; a combination of both is called States-TPPI.

The other method is based on subtracting or adding the sine and cosine components (N- or P-type, or echo/anti-echo selection) created by the phase cycle within the FIDs recorded in f; thus they are not stored separately. The subsequent complex Fourier transformation again detects the signs of the frequencies. However, here the signal shapes are skewed, consisting of cosine and sine components in both dimensions. These signals are usually processed in magnitude mode.

There have been numerous debates about the advantages and disadvantages of the two principal approaches. The phase-sensitive approach is clearly the more modem concept and yields Lorentzian line-shapes in both dimensions. However, it requires larger data matrices for processing and, since the full phase information is retained, twice the amount of time. The N-or P-type mode gives crude information in half the time and requires a smaller data matrix for transformation. However, exact spin coupling constants cannot be extracted from the resulting spectra due to the skewed line-shape. Clearly, the choice of method depends upon the information required. In the following examples we use both methods.

Pulsed field gradients can select the coherence pathways directly without the need for phase cycling. If the field gradients are applied during the f, period, either echo or anti-echo selection results, and this is a priori not phase-sensitive. Phase-sensitive approaches using field gradients have also been developed by alternate selection of the N- or P-type pathway, and this is called the echo/anti-echo, or with an additional sign shift, the echo–States method (see Ch. 12).

For a phase-sensitive experiment the NMR spectroscopist has to set the f, increment to 1/[2-sw1], where sw1 is the spectral width [Hz] in the F, dimension. If the f, period is split by a 180° pulse this increment is set to 1/[4-sw1]. For non-phase-sensitive measurements these increments are 1/sw1 and I/[2-sw1!] respectively. In the following experiments (except the J-resolved methods) we have given the parameter sw! in ppm so that it does not depend on the field strength. Of course, the software calculates these increments from sw] values given in Hz.

Before processing 2D NMR data, one first has to consider the data matrix. Usually one performs a zero-filling in F, by a factor of at least 2, in order to give nearly symmetrical data matrices. Zero-filling in F, is uncommon; it is always better to use more data points in F> from the beginning.

The processing of 2D NMR spectra must take into account the phase mode under which the recording was performed, and the correct FT procedure has to be chosen.