Processing HMQC spectra of codeine
There are several variations on ^{1}Hdetected onebond CH correlation experiments. Magnitude mode experiments have the advantage of simplified processing, because no phase correction is required. The last step in processing is a magnitude calculation. Choice of window function dramatically affects the quality of the resulting spectrum.
The sample used here was 3.3 mg codeine in CDCl_{3}, acquisition time 2.5 hrs.
A 90degree shifted sine squared function was used above. Note the broad "wings" on the peaks that result from the subsequent magnitude calculation. This is most severe for the intense methyl peak in the center of this plot.
A 0degree shifted sine squared function is commonly used to avoid the broad wings in magnitude spectra. However, in this case, use of this function led to complete loss of one peak! The missing peak is broad in the ^{1}H spectrum, and is severely attenuated by the sine function.
A 30degree (above) or 45degree (below) shifted sine squared function both seem to be good compromises.
The bottom line is, you need to match the window function to the shape of the fid. This is a case where the NUTS "window view" command (WV) is helpful, because it allows you to compare the shape of the fid and window function.
For example, below is the first slice of an HMBC FID (green), with grey lines showing the shape of different window functions:
sine squared window function, 0 degree phase shift
It is clear that this results in loss of signal near the beginning of the FID. For a broad line, which decays quickly, signal can be lost. 

sine squared window function, 90 degree phase shift
This doesn’t discard signal, but also does nothing to attenuate the early time points which give rise to the broad "wings" 

sine squared window function, 45 degree phase shift
A reasonable compromise… 

…but this might be the better choice.
This is a LorentzGauss window function applied twice, with

Below are expansions of the same codeine HMQC spectrum shown above, processed with 45 degree sine squared window function in both dimensions.
The plot on the left was processed using Magnitude Calculation (NUTS command MC) as the last step. For the plot on the right, a Power Spectrum (NUTS command M2) was used instead of MC.
Below are slices through the methoxy peak of the HMQC data shown above. MC used in the left spectrum, M2 used in the right spectrum. Compare the signaltonoise.
MC uses this equation: SQRT( Re2 + Im2 )
M2 uses: Re2 + Im2
See also: Explanation of HMQC, comparison of HMQC and HSQC, HMBC
Last updated: 01/22/2003