# Phase Correction in Multi-Dimensional Spectroscopy

This is where it is easy to get lost. Some indications are really needed

First Thing First: Memorize (or annotate) ! Despite appearances, phasing a 2D data-set is an extremely repetitive task. What you learn from your first experience will also work in future.

It is possible to suggest a workflow that should always work, like the one below. More than with any manual, however, you will learn more quickly with the assistance of an expert. To understand what follows, you should already be familiar with mark creation and with the operation of extracting rows and columns. You should also be familiar with FT options and manual phase-correction with iNMR. We suppose you have a 2D phase-sensitive spectrum to process.

- Have with you a 1D spectrum of your sample, so you know where the peaks must follow. In this way you can discover errors in your FT parameters. There is no hope to phase-correct after a wrong FT.
- Perform the first FT. Extract the bottom row. Check that the frequencies are correct.
- Phase the extract, but remember that not all peaks may points upwards. In TOCSY and ROESY spectra, for example, peaks in the right half of the spectrum will point upwards, while peaks on the left will point downward. (We are speaking of the first row, not of the final spectrum).
- Close the extract. Perform the final FT, with the “hypercomplex” option ON, "phase-sensitive" shuffling and whatever else is necessary.
- Phase the 2D matrix along rows (please note that iNMR has just transposed the spectrum, and now the rows contain f1). Use known values for the phase correction. For example, if the spectrum was acquired on a Bruker Avance, the first order correction is (in our experience) 180°, while the zero-order correction can only have two values (0° or 90°). Which of the two you need only depends on the pulse program. In other words, if your first ROESY requires 0°, that value will be OK for all your future ROESYs.
- At this point you will probably see vertical stripes (parallel to f2). If the stripes are perfectly monochromatic, then phase is correct and you can skip other points (7, 8, 9) here below. Otherwise there is extra-work to do.
- Extract two or three of these stripes simultaneously and cumulatively (they should be quite far apart).
- Phase the extract so all diagonal peaks are positive (or negative, it's mainly a matter of tastes).
- Close the extract. Now the matrix should be in phase.

Phase correction in 2D is important because it is fundamental to know the sign of the peaks, but phase is also the main factor that affects resolution. You can experiment by yourself: take a perfectly phased spectrum and apply a 90°, zero-order, phase shift along any dimension. How many times the peaks are larger now?