A classic method is the Bechet-Beaujard etchant, which is an acqueos solution of picric acid and a surfactants. In the original work (S. Bechet, L. Beaujard, Revue de Metallurgie, 52, 830-836, 1955), they used a commercial surfactant (Teepol); however, others used sodium tridecylbenzene sulfonate, and I personally used sodium dodecylbenzene sulfonate. More on this etchant can be found for example in the ASTM E407 int. standard. However, it must be noted that the PAG boundaries can only be directly evidenced if they still exist as high-angle grain boundaries, which happens if the austenite was transformed by displacive (i.e., bainitic or martencitic) transformations. If, on the contrary, the austenite was trasformed by reconstructive transformations, e.g. to ferrite and perlite, the previous grain size can only be estimated indirectly by the general microstructure (e.g. from the size of the ferrite network) after a standard nital etching.
Bechet Beajard is in fact the most simple way to highlight prior austenite grains. From a practical point of view, this etching is not very relialable. Some external factors can affect its efficiency : humidity, old of the solution...
Another efficient way is to perform thermal etching in a dilatometer. See the details in the paper from Garcia de Andres et al, Materials Characterization, vol.49, issue 2, 2002 (available in sciencedirect).
Finally, a team from Metz University has developed a reconstruction algorithm for the prior austenite grains based on crystallographic data obtained by EBSD. This method is highly reliable but is only working on martensite and lath bainite structure (because based on varient selection). Other teams have developed similar approach but this one is to my opinion the most revelant (see the paper from L. Germain et al., Acta Materialia, Volume 60, Issue 11, June 2012).
Yes, an acqueos solution of picric acid with a surfactant and a few drops of HCl typically the most effective etchant. We have found that filtering the solution through fine filter paper gives better results.
Alternative method is via heat treatments, if you are able to simulate your samples thermal history, to develop the right size austenite, you can cool to just below to austenite/ferrite transformation temperature then quench, so only the PAGB are deliniated with ferrite (Nital etch). A dilatometer or Gleeble can do this.
Polish your sample and then austenitize it to form a surface oxide layer. Then quench it and repolish (very lightly) it again. Etch with Nital. The PAGS will be more oxidized and succeptible to etcihing than the areas within the grain. The grain boundaries are then revealed.
Please try to emerge your sample(s) in a heated solution at 60 °C for 15 min, combining 200 mL of saturated picric acid and 50 mL of xylen±ethanol solvent system, and then proceed to etching operation with 3 % nital solution.
There is a very comprehensve article about this isssue in George Vander Voort's site, here http://vacaero.com/information-resources/metallography-with-george-vander-voort/1405-revealing-prior-austenite-grain-boundaries.html , please check it.
the article that you have recommended through the as-site is actually very interisting but it was already mentioned by sir Srivatsa Kulkarni (please see above)