Reference Deconvolution: an alternative recipe for Resolution Enhancement
reference: KR Metz, MM Lam, and AG Webb. 2000. "Reference Deconvolution: A Simple and Effective Method for Resolution Enhancement in Nuclear Magnetic Resonance Spectroscopy". Concepts in Magnetic Resonance. 12:21-42.
Reference Deconvolution is extremely similar, in concept, to the Lorentz-to-Gauss transformation (more commonly performed) and quite different in practice. This is why they often give equivalent result and often give strikingly different result, and apparently in an unpredictable way. The procedural difference is that traditional weighting is performed in the true, experimental, time domain, while reference deconvolution is performed in an artificial time domain. The theoretical difference is that weighting assumes that all the lines have a Lorentzian shape, while reference deconvolution allows any (?) initial shape. It still assumes, however, that all the lines in the spectrum share the same shape.
Reference Deconvolution works better with a large number of points. Remember to zero-fill your spectra, if they are to undergo this treatment.
First select a 1D peak from a perfectly processed spectrum (phase and baseline corrected). The success of the procedure depends a lot on the goodness of this selection. Try selecting all the intense part of the peak, where the noise is not evident, and avoid the regions dominated by noise. (Tails are not essential).
Issue the command “Process/Reference Deconvolution”. You will see both an interactive dialog and the deconvoluted spectrum. You can use the keyboard to navigate through the spectrum or to adjust the intensity, as if the dialog were not present.
With the top slider you can adjust the width of the selected region, to monitor what happens with a different selection. Use the bottom slider to vary the amount of gaussian broadening that smoothes the spectrum. This broadening is exactly the same used in the Lorents-to-Gauss transformation.
Use the “Toggle” button to undo/redo the deconvolution. You can also close the dialog, reopen it and this button will still work. You can use it to compare the spectrum before and after deconvolution.
Only if you have selected the proton resonance of TMS (or TSP or DSS), check the option “Si satellites selected”. The satellites flank the TMS signals at a disctance of 3.33 Hz and their height is the 2.46% of that of the main peak.
As long as iNMR remains open, it remembers the selected peak (your “reference”). You can therefore reuse it. You do NOT select anything and re-issue the command “Process/Reference Deconvolution” with the same or with another spectrum (if the total number of points remains constant). For example, you can deconvolute a row of a 2D spectrum, using as a reference the peak contained into a different row.
The deconvoluted spectrum is not saved, even if you use the Save command. To preserve it use the command “File/Export”.