Thank you for your reply Rao sir. These papers are very useful, but I would like to know in between NDA and OPA derivatization of sphingosine which is more stable. I  have done OPA derivatization even though it is very rapid, the OPA adduct is highly unstable and difficult analyse by ESI-mass. 

The instability of OPA adducts from mercaptoethanol is easily overcome by switching to 3-mercapto-1-propanol or another thiol, such as an acylated cysteine (used for chiral amine analysis).  Mercaptopropanol is ridiculously priced - ~ $75/g from Aldrich - but blown experiments are much more expensive.  About 25 years back I got a 500 mL sample from a manufacturer (industrial scale producer), distilled a few mL and that was all I ever needed.   

B.t.w.  Did RG mean to write napthalenedicarboxaldehyde rather than ----dicarboxylate?  For either dialdehyde the issue is the same - the -S-CH2CH2OH loses hydroxide and the resulting thiiranium proceeds to react with any nucleophile in sight.  OPA gives excellent sensitivity so I suspect that the additional sensitivity for NDA -derived fluorophores is not needed and doesn't justify dealing with the latter's lower aqueous mixture solubility.  (Separation is another matter.) Can't seem to find the file for a paper of mine (J. Chem. Soc., Perkin Trans. 1, 1991, 79-84) using mercaptopropanol OPA analysis, but if you find it, it should have the reference for the decomposition of the adduct.  If mercaptopropanol isn't readily available, search for OPA-acetylcysteine analyses - there are several procedures published starting about 1985.

Thank you James. Your answer was really useful. I will try this method but In my case I am not using amino acids, my compounds are sphingoid bases their reactivity were different from amino acids so I am little worried about that.

Like any good derivatization reaction, the aryldialdehyde-thiol system is very general - it only requires a primary amino group.  I've seen acetylcycteine/OPA derivatization used to separate enantiomers of chiral benzylamines and many other compounds.   As to quantities, buffer, time, etc., use the conditions in the  papers for sphingoids first, just substitute the thiol.  Also, my experience was with fluorescence detection only - is LCMS required for confirmation?  Don't see why the OPA/thiol adducts should be  a problem, though you might see better sensitivity in positive ion detection for mercaptopropanol or mercaptobutanol-derived adducts than with Ac-cysteine.

James, could you send me the article concerning enantiomer separation. The problem is we don't have LC-MS system. I have to do with HPLC and ESI mass separately. Is it possible to check the NMR of the OPA adduct?

I only brought up LC/MS because you mentioned ESI when you replied to RG's suggestion.  I am not currently associated with a university and have no library access so I can't provide the articles you ask for.  (That's only important if you can't get mercaptopropanol.)  The article about OPA-mercaptoethanol adduct instability referenced in my Perkin paper was probably Stobaugh, et al, Anal Bioch, 1983, 135(2) 495–504 - these authors suggested the use of mercaptopropanol.  Additional optimization work on OPA derivatization reaction conditions can be found in Cooper et al, ibid, 1984, 142(1), 98–102.

OPA and NDA derivatives can be used for UV detection, though they give even better results with fluorescence detectors if you have one.  (Sub-picomol LODs were reported by the early 1980s).   As to isolating the OPA adduct and characterizing it by NMR, etc., that should be feasible and I'd recommend mercaptopropanol for that.  (If you use N-Accys, diazomethane esterification of the acidified reaction mixture should simplify the isolation.)  Since you are working on sphingoid compounds, you can probably find papers on isolation and/or synthesis of other sphingoid derivatives with little difficulty - extrapolate isolation procedures from there. 

Thanks James for your valuable suggestions. I will try with mercaptopropanol.