I'm not sure why you need the line width. For a normal spectrum run on a high field instrument most of the linewidth is down to magnetic field inhomogeneity

basically how well you have managed to shim.

To get the true line width you need to do a spin-echo experiment.

If the peaks are broad due to some other effect such as exchange a line-shape fitting is the best way to go.

COVID-19 like SARS before it has a myriad of systemic effects on top of the acute respiratory symptoms that most obvious but are probably only the tip of the iceberg for this disease. SARS created long term metabolic effects- including dyslipidaemias detectable up to 12 years after infection. In most cases novel biomarkers are sought that are specific and sensitive for the particular disease - but COVID-19 has diverse effects that are often overlaid on top of existing conditions such as diabetes, hypertension etc... and it also causes an acute inflammatory response that can result in death by cytokine storm... thus there will be a diversity of molecular responses that emerge through the disease trajectory and that may reverse on recovery- though this is certainly not a given. So NMR is one of several tools that can be used to study COVID-19 -and it is very powerful for several major metabolic classes- especially lipoproteins- MS gives other metabolic windows- but no one technology platform can get even close to giving a comprehensive picture of the disease- which will also change and evolve with time and be complicated by a series of systemic changes brought about by microvascular damage that can affect any organ system... this is now slowly emerging from the clinical data. The most significant healthcare questions are yet to be asked let alone answered- especially about long term disease risk changes from these called "recovered" patients.. and indeed do such patients ever really recover? So yes NMR can definitely be used to look into COVID related biochemical changes (and we are doing that right now!) but it is only really useful when placed in a broader analytical context and in the context of a good experimental design. Analytical procedures are also confounded by the absolute requirement to handle virus deactivated samples in certain countries - this creates massive chemical damage to the samples. There is a broader requirement that live samples must be analysed in category 2 level biosecurity labs - this is uncommon in most chemical or spectroscopic facilities leading again to regulatory requirements for virus inactivation. A number of papers have already been published that have been on cooked or denatured samples- treat all such data generated in this way with caution unless the method has been specifically validated for such procedures.

WTF Jerry!

First you need to find appropriate shim for the experiments. Then if you expect difficulties related to SNR you may let the experiments be conducted by increasing the number of averages.

The spin echo sequence helps you get rid of the field inhomogeneities. However, it basically depends on the experiments you like to perform (if you are going to measure diffusion that would be another story). Finally, carefully apply baseline correction functions in order to process the data. Are you processing the data via a home made code? or commercial softwares like Mnova?

Dear Bahwna,

I guess what you want to know is probably not the line-width (affected by shim substantially as was pointed out), but T2 of your magnetisation/coherence.

Here the answer is not trivial, because due to cross-correlated relaxation in many cases different lines of a multiplett will have different T2 (which also may not be mono-exponential if different relaxation mechanisms contribute). So - if you are happy with an "overall" T2 of the multipletts I would do what was recommended and simply use a series of spin-echo experiments and fit a decay on the integral of the NMR signal.

If you are however really interested in all details of the relaxation mechanism it gets difficult - and space is to small here to provide details - through literature study can be recommended (e.g. Segawa and Bodenhausen in JMR 2013: Determination of transverse relaxation rates in systems with scalar-coupled spins: The role of antiphase coherences, or Jeff Peng in JMR: Cross-Correlated 19F Relaxation Measurements for the Study of Fluorinated Ligand–Receptor Interactions).

If you are however (for what reason soever) interested in line-width larger than shim effects (to be sure that your signals are in this regime - you may want to compare your lines of relevance with a reference [e.g. deuterated solvent]) you can deconvolute the lines - such tools are available in several NMR software packages (e.g. Topspin, ACD).

I hope this helps - kind regards

Alfred

Thanks a lot Prof Alfred Ross. It was a great help. I will go through these papers :)

Thanks a lot Dr. David Stephenson . But I was interested in intrinsic linewidth due to actual transverse relaxation not by field inhomogenity.

Thank you Dr. Mojtaba Mirdrikvand , I am using Topsin 3.2 for processing the data.

You can also use spin simulation programs such as DAISY in Topspin to simulate your spectrum,. In the program you can enter shifts, coupling constants and line-width as parameters and create a synthetic spectrum, where you can manually or by iteration adjust the parameters for a best match with the experimental data. You can then read the optimal line-width from the refined parameters.

Thanks a lot Dr. Clemens Aklin. I will try

The approaches via line deconvolution and spin analysis do not necessarily provide an unambiguous answer because the line width of coupled nuclei may also correlate with small long range couplings if present. Also this may lead to non-uniform line widths within a “multiplet”, especially when the signal is not of completely first order. The result for the line width then depends on how many lines or couplings you fit. In fact, large differences in line widths for nuclei that have very similar chemical environments often indicates underlying small couplings, which often can be revealed by resolution enhancement. In general, calculating a spectrum and comparing it to the experimental is always a good test for the quality of the interpretation because if the interpretation does not return a calculated spectrum that resembles the experimental one, the interpretation is flawed.

Each "line in the multiplet" is simply a "line" and its linewidth needs to be measured separately.

As Matthias Niemitz says, each line in a multiplet, in principle and often also in practice, may have a different linewidth, with differences sometimes reaching a factor of 2. I will describe somewhere the reasons for, together with some qualitative rules that people might like interesting and use in spectra interpretation.

I believe that your uncertainty is due to the recent (last 10 years) corruption of the terminology of NMR spectroscopy according to which a "spectrum is composed of peaks" and "peaks have multiplicity". This is an oversimplification due to teaching NMR to chemists (sorry, guys, I know you are bright, it is the teachers).

The correct definition is:

A spectrum is a set of multiplets,

A multiplets is a set of peaks,

A peak is a set of [unresolved] quantum transitions.

Definition: Two quantum transitions belong to the same multiplet if the nucleus with the largest |delta S_z| in each is the same or at least if the two belong to the same magnetic equivalence group.

Definition: Two quantum transitions belong to the same peak if (a) they belong to the same multiplet and (b) given the current magnetic field homogeneity as a threshold, the difference between their offsets (frequencies) is smaller.

For nitpickers: yes, I know, the second definition requires ad addendum in case of overlapping multiplets, but the space for an answer is limited.

Let us discuss it somewhere else.

Thank you Matthias Niemitz and Stan Sykora ... What I understood that determining a uniform linewidth in case of strongly coupled peaks of multiplet is cumbersome..Relying only on experimental data will not reveal the reliable value. One needs to have simulation support...