FAQ 2001-02.1: 2D techniques, why do they occasionally take longer at 600 MHz than at 300 MHz?
Essentially all NMR users know very well that increasing field strength means higher sensitivity and thus substantially shorter experimental times to reach a certain signal to noise (s/n) ratio. But under certain circumstances a 2D experiment, a GCOSY for example, can take double as long on the i600 compared to the i300. Why and when does this happen?

The time it takes to record a 2D spectrum is determined mainly by the parameters nt (number of transients) and ni (number of increments, i.e. how many 1D spectra will be recorded to make up the 2D data matrix). The number of transients (also called scans) determines the s/n of the spectrum (it increases by the square root of nt).

How many ni are needed depends on the width of the spectrum. Suppose we take the usual 10 ppm width. At 300 MHz this is equivalent to 3000 Hz while at 600 MHz it is 6000 Hz. If equal digitization and consequently equal digital resolution is desired, then double as many ni are needed on the 600 to cover the entire range: 3000 Hz/256 points and 6000 Hz/512 points result in the same digital resolution and therefore same quality spectra. If the sample concentration requires more than one scan at 300 MHz then the experimental time at 600 MHz will be shorter or equal at the worst. But if the concentration is high enough that even the 300 can record the GCOSY with one scan, then the 600 will naturally also do it with one scan (less than one is not possible!), but with ni=512 and consequently it will take about 18 instead of 9 minutes to complete.

One easy way to reduce the experimental time, is to decrease the sweep width by using the EZ NMR S+A button called Set ppm Range sw|sw1 if not the full 10 ppm range is needed. Note, however, that this is only worthwhile when a substantial reduction can be made. To reduce from 10 to 9 ppm would hardly be worth the effort but a reduction from 10 to 6 ppm would nearly save half of the experimental time.

All this applies also to heteronuclear correlation experiments (GHMQC, GHSQC, GHMBC). Reduction of the sweep width in C13 from 200 to say 100 ppm, if the sample allows it, can be particularly time saving as those experiments in general run longer than the homonuclear ones. When using the sweep width reduction, be careful not to cut out signals as they will be folded into the region of interest and can be very confusing unless done in a controlled way.