The amount of sample you need for an NMR experiment is realted to many factors:

1) The magnetic field (and the related Larmor frequence) of your NMR instrument: for a 14 Tesla (600 MHz on proton) you need half of the amount of a 7 Tesla (300 MHz on proton).

2) The molecular weight of your compound: 10 mg of methanol have more proton concentration compared to 10 mg of taxol

3) The equivalence of the protons in your compound: benzene has 6 equivalent hydrogens, but i.e. toluene not...

All this suggestions are valid for 13C, too

Good luck for your experiments!

Roberto

1 mg is enough for proton and carbon NMR. Solubility of your compounds in the deuterated solvents also should be considered. 13C-NMR always need long running time especially for less amount of compound. So, you may increase the time for carbon.

1. You must verify full compound solubility in deuterated solvent selected.

2. NMR signals are associated to NMR frequencies, as follow approx:

300 MHz: 2-5 mg or higher for a good proton and carbon signals.

500 MHz: 1-2 mg.

600 MHz: 0.5 mg, etc, etc.

This values are for a sampler NMR volume of 0.8 mL.

Devices with cryosonde are more sensitive and it demands less sample.

Firstly, you have to consider about the solubility of your compounds in the deuterated solvent of your choice. Usually 1 to 2 mg of compound should be enough for NMR analysis. As for 13C NMR analysis, it usually require a longer period of time to collect the spectrum if the amount of your sample is small (e.g less than 1 mg or so). Perhaps, an overnight run of your NMR sample (more scans) would be needed for this case in order to obtain a relatively good NMR spectrum. Thus, I would suggest for longer running time for carbon NMR and check your compound solubility at first.

The NMR analysis depends mainly on solubility of compound. It should be completely soluble in suitable deuterated solvent. Secondly, in 0.8 ml solvent (600 MHz spectrometer) 20mg of compound give excellent results.    

For proton 0,5 till 10 mg are good amounts, for carbon a fast analysis with 50 mg or more can give a good spectrum in 128 scans. With less just keep in mind that you have to increase the analysis time 4 times to double the peak size. This makes it always faster to add more compound in the tube, as long as it dissolves.

thanks all for theses valuable information. It seems that C13 NMR needs really a large amount to get a good analysis and needs also more analysis time.

Why wouldn't you be able to recover all of your product after measuring any kind of NMR? If you choose an appropriate NMR solvent you can simply distill it off (evaporate in vacuo) after the measurement. As you have complex structures (of natural products) I would use a little more product for 13 C NMR than Elkin suggested above. If you have de facto only 1-2 mg product and you can get acces to 2D NMR spectra, I'd run an HMQC or HSQC measurement (C-H correlation), which is much more sensitive for carbon than a simple  1D 13C NMR (with low abundance of 13C) as carbon is measured indirectly through proton and it holds also more information.

 8 mg for 1H NMR & above 12 mg  for 13C NMR and also depends molecular  weight of your compd

I agree with Mr. Imre suggestion. It is better to run HMQC or HSQC first than 13C-NMR which will give more valuable information about the structure of your compound. If possible for you to run HSQC, that will be better because you can determine CH3, CH2 or CH rather than use DEPT (13C) data. And of course can save money and time.

Good luck.

Firstly, you must consider the solubility of the compound in the solvent and dissolve completely your compound in the deuterated solvent. I would suggest for running 1H spectrum around 5 mg of the compound. Good 13C can be obtained using long running times and appropriate quantity (mass) of your compound, which actually depends on the molecular weight of your compound. As an example, I got good 13C NMR spectra with excellent S/N ratio of clarithromycin derivatives with 30-35 mg of the compound with an overnight scanning. 

I agree with its  said by  Biljama Arsic. I would add that the molar ratio is a factor to be considered to obtain a spectrum. If you are collecting samples  direct from HPLC and taking  to NMR, adding only the deuterated solvent, dilution must be very low and the signal strength of the solvent must  hide the signs of the sample even obtaining a spectrum of hydrogen. My suggestion is remove the HPLC solvent and so have a more concentrated sample to obtain the NMR spectrum. Moreover, an apparatus is 400 MHz or higher with a reverse probe helps to improve in the quality spectrum. It is also possible to use micro cells. See the website of Wilmad to obtain details about the micro cells.

I agree to the answers given by both Arsic, Miguez. But additionally, i think complete removal of solvents used in Preprative HPLC must be ensured. 

I agree with has been said by other colleagues, perhaps 5 mg will be enough to do a decent 1H NMR, regardless of the molecular weight of the compound. however, for 13C NMR a number of factors need to be considered. The molecular weight of the compound, the instrument, time allocated for the experiment, and the mass available. I do think, that about 30 mg will be enough for a good 13 C NMR. If the amount is less, perhaps doing a large number of scans (overnight scanning) should be considered.

The amount of sample you need for an NMR experiment is realted to many factors:

1) The magnetic field (and the related Larmor frequence) of your NMR instrument: for a 14 Tesla (600 MHz on proton) you need half of the amount of a 7 Tesla (300 MHz on proton).

2) The molecular weight of your compound: 10 mg of methanol have more proton concentration compared to 10 mg of taxol

3) The equivalence of the protons in your compound: benzene has 6 equivalent hydrogens, but i.e. toluene not...

All this suggestions are valid for 13C, too

Good luck for your experiments!

Roberto