It's better to use a more corrosion resistant impeller -- mixing in a solid that maybe doesn't dissolve quickly will wear off a coating.  Plastics (PE can be applied by powder coating) don't stick very well to 3xx SS, and will peel as soon as the first hole or tear develops.  
HNO₃ itself isn't too corrosive toward 304 SS (which is used for 55-gallon drums of nitric) but the metals & impurities in the nitrate may contribute to a bad interaction.
What metal nitrate, quality of water, what color is formed, have you analyzed the raw materials & solution for Fe, Cr & Ni?

Consider using a 316 (CF8M if cast) SS impeller. Also, check any setscrews if used; they are usually a harder but less corrosion resistant SS.

If you have a high speed mixer, maybe erosion is the problem. A slower but larger diameter mixer may help, then.

Actually, I run the mixer at lower speed (let's say 100-150 rpm). Solution becomes grayish when it's finished. Doing the same thing in a lab (magnetic stirrer/beaker) results in a clean fluid).
Isn't there some kind of hard, dense PE formulation, if I remember I ran onto it somewhere (trade magazine)?

I will definitely have to do analysis on metal ions in the product.

You can do a somewhat crude analysis for [Fe⁺³] by a simple pH titration.  Slowly add 1 N NaOH while stirring and observe the pH at which reddish-brown Fe(OH)₃ precipitates.  Then either use K_sp or a solubility curve to get [Fe⁺³].
Solubility curves vs. pH for various metals (only a partial one for Fe⁺³): http://www.finishing.com/Library/metalpH.html

Check both the lab prepared sample & the bulk.  A faint color in a bulk tank usually appears colorless in a beaker.