You should try to do both things simultaneously. If your impurity is at a low level, it will not hinder interpretation of the initial spectra and while you do that, you continue your purification in order to get 2D spectra of yourpure compounds.. Good luck

Why you do not try to purify if you have possibility to do it. I understand that you have a very little amount of your compound. But when you receive results of NMR if you see interference signals what you think - is it because of your compound or because of impurities? You will come all the same to confusing about the structure.

I think you can do NMR first then try to recover your sample from NMR tube and later see if you can purify it.

Since you have a limited amount of compound, make sure the that solvents you use for whatever subsequent steps (e.g. purification/NMR pre-test) you perform are really pure, so that this doesn´t become a source of contamination as well. I know this may sound trivial, but solvents can be of so-so quality, or the last user of a bottle may not have been careful.

I think it's better to purify it before proceed to any determination of the structure

Thank you for all the suggestions given by RG professionals. I will consider about my next step to be taken wisely. I am greatly appreciate for all the answers provided.

You should do both - often 2D spectra do not suffer from impurities, and in fact you can more easily identify impurity peaks in the 2D spectra. So structure elucidation may not be hindered if the compound is not 100% pure. However, the purer the compound is the easier it is to convince your colleagues (and referees!) that you have done a good job in identifying the compound. Of course if you plan to use the pure material in bioassays then it needs to be as pure as possible.

You need to purify before going to the spectal analyses

I would at least go for 1H-NMR early on, it is non-destructive (unless you compound is labile, of course) and fully quantitative, so might already give you an idea on the actual purity of the sample, especially if you are looking at a small organic compound (as opposed to a peptide). Plus, instrument time for 1H-NMR is usually quite moderate compared to 13C- or 2D-NMR experiment (which you can still go for if your 1H-NMR looks good), and it might give a clue as to the nature of the impurity as well.

Thank you for all the helpful advice and suggestion regarding my question. I will try to go for 1H-NMR at first to get a rough idea about the nature of impurity which might contain in my compound and subsequently proceed to further purification if it is required to be done. 

If you have about 80% purities, you can identified the compound, but you have to conduct 2D NMR. I agree with several comments that suggest you to perform 1H NMR first

2 mg is just to small. I will advise you go ahead if the proton NMR look good enough, then run C-13 and then the 2D experiments.After all this, you can recover your sample and try to purify. Depending on the nature of the compound, I will implore you to pack a small glass column of preferably fresh(new) sephadex LH-20. 

I would always run NMR on the sample first, for several reasons. Firstly, minor impurities (10-20%) rarely seriously hinder structure determination, so you should be able to work out what the compound is (or at least develop a working hypothesis). Secondly, compounds sometimes are lost during purification due to instability or accidents, so if you get NMR data now then---if you lose the compound---you will still have data that you probably can use later in a publication. Thirdly, even preliminary NMR data will give you some idea as to what type of compound you have, and what type of contaminant is present. This information is invaluable for designing a final purification step that will separate the two form each other.

Oh, and 2 mg should be plenty for NMR on a medium to high field instrument. Another point is that you are relying on LCMS to know how much impurity you have. That is very uncertain, because the LCMS response for your compound and any impurities present can vary enormously (and some impurities may not be detected at all, depending on your LC-MS method e.g. mass range, eluent polarity range, positive or negative ion mode etc.).

Of course you should run a 1H NMR and this will clearly show a set of signals for the main compound, and only weak signals for impurities.

All other comments given above here fit also to my experience.

It is possible to recover your compound from the NMR solution by evaporating it in vacuo. No problem for CDCl3, even possible for high boiling  DMSO-d6 if you let it overnight at a "vac-line" in fine vacuum.

The only thing that may occur is deuteration but this will not disturb you because you have a molecular ion peak already from your MS experiment.

I would suggest running all needed NMR experiments for structure elucidation before you continue with the purification. If the sample contain only small amount of impurity it want be a problem to solve the structure and you will have to purify it only before you conduct the bioassay.

Thank you so much for all the helpful suggestions provided by RG professionals or researchers. Currently, I am trying to obtain the NMR data for this compound prior to further purification as the amount of my compound is relatively low. So, it would be a wiser and appropriate step to obtain the preliminary NMR spectra initially to confirm whether impurities exist in my compound before any purification effort taken place.

I agree with Christopher regarding impurity levels.  The LC/MS response depends on ionisation efficiency.  So  a significant impurity may have poor ionisation efficiency and SEEM  to be only present at  only low level.  However, proton NMR  response depends on the number of moles of hydrogen and so will reflect true impurity levels.  This can be very different from the indication inferred from LC/MS.  Still worth doing the NMR but be aware of these complications.  It is often possible with the aid of 2D NMR to determine the structure of both the major compound and the impurity!

I agree with idea of obtain 1H-NMR spectra and proceed to purify, but, I run also 13C-NMR; although do you have already  any NMR spectroscopy? because you affirm to had obtained a novel compound...