Try a solvent like deuterated Methanol or water, that would form H-bonds with the analyte instead of the analyte forming bonds among themselves.

Diluting the sample might help depending on the strength of association. Or add some more polar solvent.

You may try to add an small amount of a H-bond donor/acceptor competing solvent (e. g. MeOD) or directly measure your sample in such kind of solvents

Agree with Jimmy, but use DMSO to break up hydrogen bonds in CDCl3 solution.

thanks for answer Jimmy. The problem is with my compounds that I can solubilise them just in low-polar solvents. Methanol or water are not suitable deuterated solvents.

This is a very interesting question. Besides the heating, have you tried to decrease the analyte concentration? This could lead to higher analysis time, nevertheless the J coupling can be revealed instead.

Decreasing the concentration and polarity of the solvent could help to avoid such self-aggregation...

You could try a higher boiling solvent like TCE for high temperature measurements, this helps with my compounds.

CDCl3 + 1 % d6 DMSO NMR solvent mixture might help to disrupt your self aggregation problems.

Try to use a NMR solvent mixture (even dough maybe you will have undesirable NMR solvent signals) and try to use Ultra-Sounds instead of increasing the temperature by heating. Why methanol, dmso or water are not suitable?

Adding or measuring in a hydrogen bond breaking solvent tends to help a lot. Adding methanol or trifluoroacetic acid to chloroform usually does the trick. If that is insufficient, use DMSO-d6 or deuterated HFIP as solvent. In the case of HFIP I usually add 5% acetone-d6 to facilitate locking/shimming.

Higher boiling solvents that are useful are trichloroethane-d3, o-dichlorobenzene-d4, again DMSO, and perhaps dioxane.

Keep in mind that if you use protic polar deuterated solvents to break the hydrogen bonds, the NH/OH signals in your analyte wil exchange and disappear.

5 mg analyte / 5 mL CDCl3 is a concentration that often provides acceptable signal to noise ratio (for 16 scans on a 400 MHz instrument) without much/any aggregation. You could try decreasing the concentration to 1 mg / 5 mL, 0.5 mg / 5 mL, etc. This, in addition to the above mentioned 1 % d6 DMSO in CDCl3 trick, could dramatically reduce the H-bonding.

Add some amount of a compound that can compete with the formation of the H-bond, like MeOH or trifluoroethanol, into your sample.

enhance dilution of the analyte until 10-5 Molar or use another deuterated solvent as dimethylsulfoxide sometimes its works

One way to address this challenge is to add strong acids such as TFA, provided that your compound is stable under acidic conditions. Add only 2-3 drops of TFA (you do not need to add much). You can add it even in the CDCl3 solution. I use this trick whenever I need to find out the coupling constant of a particular proton in the sample.

I concur with Serge Söntjens. I prefer to add a small amount of methanol (CD3OD) to chloroform (CDCl3) solutions. This works miracles with lipids for instance, but I have seen sharper spectra with other substrates as well.

I agree with Conner. When running a chemistry NMR facility, the most common cause of broad lines/aggregation I saw was high sample concentration. If you're primarily interested in 1D 1H data, you can get by with much less than 1mg of sample; simply use more scans to improve your S/N.

Also, why do you need to measure your coupling constants? If you need to determine which 1H's are scalar-coupled to one another, a simple COSY will provide answers even on aggregated samples.

Both sample dilution and using H-bonding solvents, as noted by others in this forum, are excellent suggestions. If your problem persists, however, you also can also run a 1D Hahn spin-echo CPMG sequence that will enable you to remove the broad lines from your spectrum (see Sanders and Hunter's book "Modern NMR Spectroscopy" for details).

Thank you very much to all of you. Your informations were very good. I succeed to perform H-NMR in a mixture THF-d8: D2O in volume ratio 5:1.

By doing the NMR at high temperature like 330 K

Use high polar deuterated solventes, like DMSO-d6, acetone-d6 or acetonitrile d6 etc.

Magic angle spinning (MAS ~ HR-MAS setup) can be used to average out the dipole-dipole interactions giving rise to line-broadening. This is often used nowadays to analyze eg. tissue samples or swollen polymer networks. Several options for small molecules in solvents also exist.

Note also that the solvent (as well as the concentration) commonly influences shifts and J couplings. If you want to compare the spectral data of several related compounds, you should record each in the same solvent mixture and at the same concentration.

Prepare a diluted sample, using ultrasonic string and running more scans

Use a solvent that can compete with the hidrogen bond formed by your molecules, or add some water or methanol to your sample

Thank you all for your useful answers. The NMR analysis worked well (without broadening) after I used a mixture THF:H2O (5:1 vol).

I had a similar problem with phenylboronic acids which form a mixture of cyclic dimers and trimers in DMSO-d6. Taking proton NMR in CD3OD solved the problem instantly.

The life time of the excited state should be 0.1-10 s (moderate).

We have solved the same like problem by using CDCl3/CD3OD:10/1, v/v.

The only solution is to do measurement at higher temperature and higher magnetic field magnitude

Use high temp and low concentration and increase the number of scans. I hope it can work.

Try add one or two drops of deuterated trifluoroacetic acid

I work with dendrimers and I have the same problem. A small amount of deuterated methanol can help a lot

I remember, I have met such probleme few year ago when I worked on porphyrin derivatives. so, of course you should make a very diluted sample; increasing the number of scans and more important, you try an ultrasonic stirring about 15 min at warm temperature just before making your NMR tube in the machine, so I think you will be able to get nice spectrum

You can use MAgic Angle spinning to minimize the broadening peaks and also, sometimes, Spin-echo works to remove background peak.