I have just carried out an exploration of the effect of different solvents and conditions on the 1H caffeine relaxation by benchtop NMR spectroscopy.

When collecting the 1D 1H NMR spectra of the caffeine in CDCl3 solution, the parameter for the number of points was set up, but when the relaxation run for T1 and T2 were done, the parameter was replaced by the value of time delay, tau.

I think I kind of get a clear view from your discussion, as the number of points is needed in determining the right peaks for 1D 1H spectra. But, as for the relaxation time, the intensities behaviour during the time delay is now being the subject of interest, hence the np parameter is neglected.

This is what I understand by far.


They are the number of points (np) and tau.

I think I kind of understand how np works;

it's the discrete points of analogue FID that are being converted to digital points along with the FID signal envelope. Therefore it's associated with the resolution of the spectra which the lesser np used to define the FID, the lower the resolution. *Let me know if I interpreted this wrongly.

This question raised when I'm analysing my data though, would really appreciate it if someone can enlighten me.

  • 1
    $\begingroup$ tau could be anything (usually a name reserved for a time delay, therefore greek "t") for all we know. Provide more info regarding the pulse sequence, maybe a link? $\endgroup$ – Buck Thorn Dec 2 '20 at 17:41
  • $\begingroup$ Possibly you apply a technique (equally known as «pulse sequence») already literature known; if this is the case, add this information. Examples of widely used techniques are DEPT, INADEQUATE, APT; keywords like these help to communicate and are used in the literature, too. They may help you to trace information on-line, too (e.g., chm.davidson.edu/vce/NMR/InversionRecovery.html, ucl.ac.uk/nmr/sites/nmr/files/L5_3SH_web_shortened.pdf) and $\tau$ may be the intentionally varied vdlist parameter on page 4 on nmr.chm.jhu.edu/Material/Notes/JHU-NMR-Relaxation.pdf. $\endgroup$ – Buttonwood Dec 2 '20 at 19:21
  • $\begingroup$ Like described here: nanalysis.com/nmready-blog/…? images.squarespace-cdn.com/content/v1/5707ede0d210b8708e037a1e/… $\endgroup$ – Karsten Theis Dec 2 '20 at 19:45
  • $\begingroup$ @KarstenTheis For my gusto, the example does not show all, because the integrated signal intensities are all for one type of (aromatic) C(sp^2)-bound H. It would have been much more instructive if it were about H's bound differently to C(sp3), C(sp2) and C(sp) because recording quantitative 1H-NMR often requires to adjust the delay times normally set for recording routine 1H-NMR spectra to check the identity of a product in a synthesis lab. (Of course, recording quantitative 13C-NMR [like pubs.acs.org/doi/10.1021/ol403776k ] is a different story.) $\endgroup$ – Buttonwood Dec 2 '20 at 21:52
  • 1
    $\begingroup$ @Buttonwood What the question is sorely lacking is the model of the bench top NMR in question. Some are so simple that you can run a couple of canned experiments only, and then it would be a matter of looking those up in the user manual. In the example I found, I'm not sure if you have to run multiple spectra setting the delay time, or if it runs a series of experiments based on a minimum, maximum and increment for the delay time. $\endgroup$ – Karsten Theis Dec 2 '20 at 22:33

AS Karl said read the instrument's manual and add the meaning of the symbols. However there is one important point in discrete Fourier transform (DFT).

I think I kind of understand how np works; it's the discrete points of analogue FID that are being converted to digital points along with the FID signal envelope. Therefore it's associated with the resolution of the spectra. *Let me know if I interpreted this wrongly.

I have been recently discussing this exact point with signal processing engineers and spectroscopist friend with a lot of good constructive discussions. No, the number of points cannot enhance the intrinsic resolution of the signals whether it is NMR or FTIR or any other FT spectroscopic technique. What you increase, when you collect more points, is the frequency resolution on the x-axis, and consequently more points per peak. That is it.

For an N-point DFT, the frequency step is

$$\frac{1}{N\Delta t}$$

For example if your sampling frequency is fixed to 1 s (an arbitrary number), in one experiment you collect 1000 points, the frequency step on the axis is

$$\frac{1}{1000 . 1 s}=0.001 Hz $$

In another experiment, you only collect 250 points with a time step of 1 s

$$\frac{1}{250 . 1 s}= 0.004Hz$$

Note the time to sample the signal in each case is different, because it is given by $N\Delta t$. So by chance, if your original experimental spectrum was "limited" by the number of points per peak by increasing the number of points, you might see an improvement in the peak shape and apparent resolution when you increase the number of points in an FID.


Not the answer you're looking for? Browse other questions tagged or ask your own question.