Hi Joseph,

Did you try linear prediction and zero filling yet? Does the problem persist?

How many dummy scans did you use? HSQCTOCSY and other experiments can warm up the sample quite a bit. With insufficient dummy scans part of the equilibration period will be during the experiment. This means peaks will shift during the first few FIDs. This can lead to line shape distortions. This is most pronounced in D2O or H2O.

Dear Joseph,

frequently (not always) I observe a similar effect with standard HSQC spectra, Phase correction along F1 cannot be done perfectly, in consenquence you have a "tailing" of some peaks, best visible at large signals on the left and the right end of the scale. But since this does in no way affect the evaluation and interpretation of the spectra, I never thought much about it. See the attached example (Instrument: AV III HD 300 MHz, pulprog=hsqcetgp, ds=16, duty cycle below 10%, chloroform).

Rudraksha, 

I have tried zero filling and forward LP through Mestrenova, and it does not seem to help. The largest issue is the phasing in F1 seems almost to be non-uniform, so phasing in one direction will make the upfield region correct, but completely distort the downfield region. 

Clemens, 

This makes a whole lot of sense, and I acquired using 16 dummy scans for a 16 scan experiment. I figured this would be enough, but I am also acquiring the data in NUS mode, and I haven't thought of how the collection may effect the number of required dummy scans. Do you have a recommendation for this? Also, these data were collected in a standard 5mm tube, if a shigemi tube were used, I can imagine this problem may be exacerbated due to the lack of convection. 

Wolfgang,

I too have seen this in a standard HSQC, but it seems to be exacerbated by the HSQC-TOCSY combination. For single compounds, it doesn't seem to be too big of an issue, but with mixtures or extracts with variable relative amounts, the tailing seems to be a huge issue for judging if a peak tail is hiding another signal. However, I found that this is less pronounced with the HSQC pulse sequence I've been using (hsqcedetgpsisp2.3). 

Thanks for your help everyone, I will see if extending the DS will help and will update. 

Joseph, my guess would be DS=16 is not enough. NUS will further distort the line shape as there will be FID's with shifted lines at random positions. Whether you use a shigemi or not is not going to make a differnce. This is not caused by convection but by simple heating. I often see that 64 to 128 dummy scans are needed for temperature to be stable. 

HI Joseph (and Clemens can feel free to chime in here),

I have seen similar things as well. Is the phase distortion you see a frequency-dependent phase distortion? If so, it may be that theDIPSI-2 pulses were not properly set.

There are a few steps we take in my group in order to minimize these. First, we calibrate the power needed for 50 us to be a 90 deg pulse in the DIPSI-2. That keeps the power lower and we therefore minimizes the effects of sample heating that Clemens mentioned in an earlier post. Once we have the experiment set up, we run the longest mixing time that we plan to use (typically 120 ms, but up to 150 ms), in GS mode and use "self-tune" in the EDTE panel. You can stop GS once the self-tune is complete. That will help you with stabilizing temperature fluctuations, which can lead to shimming changes, under experimental conditions. You can also turn on auto-shim so that environmental changes in the room don't impact your spectra as much. 

I hope this helps,

Daron

Thanks again for everyones feedback. 

I went through the old data and found that the heating was the most likely cause. I went back to the FID, FT in only one dimension, and observed the effects from scan to scan, and sure enough, the signal drifts from the position in the early FIDs. 

I was able to try out a few things that I think have really improved the spectra. I ran a study of the effects of the DS (16,32,64) for each of the different inversion pulses (Chirp60,Chirp80,Hard Pulse) and found that I saw the least effects in all of the 64 DS variants, and the overall quality of the spectra were best in the Chirp80. That's something that I totally forgot to consider until it was brought up here, but now it makes a lot of sense. I imagine that doing 128 or more DS would improve this even further, but as you can see in the attached, the effects in the carbon dimension are greatly reduced. 

All spectra were 100ms mixing time, NUS sampled at 50%

Thanks again everyone! 

Hello, I have encountered a peculiar phenomenon as well. In my polysaccharide spin system, peaks in the HSQC-TOCSY exhibit splitting of this nature.