Hello Mr Pier

I do not know the concentration of interest. However, graphite tube AAS may help you. I've got good results with a continuum source AAS spectrometer. Matrix can be mainly destroyed during thermal pretreatmant.

To my mind, you should try it. Best regards and good luck.

Additional comment:

The proposed method only counts the content of the element P and cannot distinguish different phosphate species, such as acid or salts.

An other phosphate identification method is provided for water analysis. Applying Mo ions yields in blue MoP-compounds, which can be easily measured by an UV-Vis spectrometer.

Thank you Gregor for the answer, but my problem is the analysis of PO33- (should be around 500mg/Kg) in (NH4)2HPO4 matrix,

I'm looking for a method that can discriminate the two species of P.

You should try suppressed anion Ion Chromatography... it will discriminate between phosphite and phosphate

I tried to do  the IC analysis for see if it was possible to have the separation of the two species, but unfortunately I had only a column for normal ions analysis (AS14) that is not able to do the separation.

May be with a IC that can work in gradient, a different column (like AS9) and with electrochemical detector, I think I can achieve my target.

Also a voltammetry technique should be able to performer the analysis.

The problem is that I don't have this instruments :( .

In this case you can try easily with colorimetric methods. First you determine phosphate (by molibdenum blue method or by ammonium molbdate vanadate method) an then you oxidize phosphite to phosphate by 20mM KMnO4/1M H2SO4 and dtermine total phosphate again. Then you can calculate phosohite by substraction. 

The ratio PO4 / PO3 that I have in my sample doesn't permit this method.

The sample is a bag of 18Kg of Diammonium Phosphate pure and the client wants know if the PO3 is over the specification (0,01%).

The method that I used it is based on the reaction I2 / PO3, but i can not be sure at 100% that my result is PO3 or another reducing substance.

http://onlinelibrary.wiley.com/doi/10.1002/1098-2728(2000)12:6%3C286::AID-LRA3%3E3.0.CO;2-D/pdf

Maybe this could suit

If you force esterification with methanol and purge with nitrogen, you might be able to enrich or recover the trimethyl phosphite selectively from trimethyl phosphate -- the bp's are far apart (~120 vs. 190C if I recall). Then perhaps do oxidation using I2 such as in a coulometric KF, if you have such available (the methanol is harmless in this instance). It's kind of roundabout, but even incomplete recoveries could be assessed kinetically, as is possible with moisture analysis, for example. Your fertilizer would need to be quite anhydrous, first.

Try 31P-NMR of sufficiently concentrated solutions a) in H2O and b) in D2O. You will not see signals for [PO3]3- but for [HPO3]2- or [DPO3]2- . Those signals might be very weak beside a strong signal for the phosphate anion. Hence take care of good signal to noise in your spectra: large number of scans is necessary. This quantitativ determination might be a the limits of the NMR methods.

I have been able to separate several species of P containing substances including PO4 / PO3  in the past.  If memory serves me right I used a gradient anion IC and an AS11 anion column with hydroxide eluent (up to 35mM NaOH) using conductivity detector.  I used the IC in chemical suppression mode.  You may be able to add some methanol (up to 10-15%) to your eluent to aid elution.  Typical runs would have taken ~ 30 minutes.

Try a second method based upon 1H NMR: serach for the typical doublet of [HPO3]2- in 1H NMR. If you find it, than use standard addition techniques: Make a standard solution of CH3PO3H2 in H20 (in proper concentration) and add . known amounts of this standard to your solution holding [HPO3]2-. Compare integrals for doubletts from  HPO3 and CH3PO3 in 1H NMR.