We were intrigued by a poster presented at the 1994 ENC by Ken Metz (Poster # WP111) entitled "Simple Technique for Improving Resolution in Heteronuclear NMR Spectra by Deconvolution with the Measured B0 Field Distribution". The poster demonstrated a processing technique for removing lineshape distortions, based on an earlier paper by Morris (J.Magn.Reson., 80, 547, 1988 ) and showed some impressive improvements in lineshape.
The basic idea is that if you know the shape of the distortion, you should be able to correct for it. To measure the distortion, you need a "reference" spectrum of a single, isolated peak whose ideal lineshape is known, eg., chloroform or water. A comparison of the reference spectrum and the ideal lineshape characterizes the shape of the distortion. Both ideal and reference peaks are mathematically adjusted to be at zero frequency and inverse FTed. An apodization function is created by dividing the ideal time-domain function by the reference FID. This apodization function is then applied to a real FID. The resulting lineshapes are substantially improved. The whole approach, of course, relies on the assumption that all peaks have the same distortion.
To implement Reference Deconvolution with NUTS:
The first step in creating the function is to have a spectrum containing an isolated singlet peak. This serves as a reference peak which characterizes the distorted lineshape. Use Zoom to expand the spectrum to a small region with only this singlet displayed. Estimate the real linewidth, in the absence of distortions, and set LB to this value (Type LB and enter the chosen value in the highlighted box.) Suggested values are 0.3 to 1 Hz. Typing CF creates the convolution function in 3 steps: It creates an FID from the displayed singlet via an inverse FT (removing all other peaks that were in the spectrum). It also creates an ideal FID corresponding to a Lorentzian line with width equal to LB. Finally, it creates the convolution function as the result of dividing the ideal FID by the "reference" FID created from the singlet peak in the real spectrum.
To apply this function, read in (with File/Open or GA) the FID to be corrected, and type CA (Convolution function Apply), then FT and proceed as usual. If the base of the peak appears distorted, it may help to repeat the entire process and use a larger value for LB. The convolution function has a shape somewhat like other resolution enhancement functions. It can be viewed (after being created by CF) by typing CV. The function can be saved by selecting the Save Convolution Filter option under the Tools/Convolution option. Similarly, a previously saved function can be recalled from the same menu.
Refs: K.Metz, Poster # WP111 presented at the 35th ENC, 1994
G.A.Morris, J.Magn.Reson., 80, 547 (1988).
A.Gibbs and G.A.Morris, J.Magn.Reson., 91, 71-83 (1991).
Last update for this page: 05/25/02