It is very important that your NMR samples are mixed well before NMR data are acquired. If the sample has a concentration gradient (i.e. more concentrated at one end of the sample column compared to the other) it will be very difficult to shim the magnet over the entire volume of the sample as the magnetic susceptibility is not constant over the sample volume. As a result the NMR lines may be skewed and will be much broader than necessary. This will lead to a much lower signal-to-noise ratio based on signal heights. The figure below shows partial 300 MHz 1H NMR spectra for 2-bromobutane in CDCL3. The spectrum on the left was acquired on a sample where 1 drop of 2-bromobutane was added to CDCL3 in an NMR tube. The tube was gently swirled but not shaken. The magnet was shimmed using a gradient shimming routine and the data collected. The spectrum on the right was acquired on the same sample except the tube was removed from the magnet, shaken and reinserted. The magnet was reshimmed with the same gradient shimming routine and the same number of scans were collected. The difference in the quality of the NMR data is obvious.
