The total analytical concentration of Fe and the distribution of Fe between (II) and (III) in solution cannot be determined without additional information.

•You need to assume that thermodynamic equilibrium is achieved.

• You need the T.

•You need the pH.

In addition, if you need to assume that there is available Fe, e.g. as pyrite. (It seems self-evident, but a rock system without Fe cannot generate Fe in solution - it has to have a source. The observed Fe can be limited by the source or by the chemical reactions that transfer Fe from the solid phases) to the aqueous phase.) . Finally, if your measured "Redox potential" is an ORP measurement, using an electrode, you need to convert that measured ORP to Eh (or pe).

I'll assume the simplest form of your question: you have a solution chemistry that includes total Fe, and you want to know Fe2+ and Fe3+ in solution. The "best" (I mean the computationally simplest and most complete) way is to take a measured solution and determine the distribution of species and saturation indices for the water using PHREEQC , REACT or another thermodynamic model. This will provide you, along with other information, the distribution of Fe2+ and Fe3+ species in solution, and the final aqueous concentration of total Fe, assuming equilibrium. Depending on what you want to do with the information, you can directly read the a(Fe2+) and a(Fe3+) , or you can sum the Fe2+-complexes and the fe3+ complexes and get a total distribution of Fe2+ and Fe3+.

From your question, you may be after a different matter: to predict the final concentration of Fe(total) and Fe2+/Fe3+ as a (for example) pyrite-bearing rock undergoes oxidation under some defined redox state. This is a different modeling problem, a *forward* model of a reaction path. Again you use a reaction-path model, but now you define the problem differently, starting with some defined mass of reactant and tracking the reaction. The User's Manual for the computer program you use will explain how this is done, and probably provide an example problem (or more than one). Depending on how you set up the problem there likely is no unique solution, the answer depending on how far along the reaction path you go. The best way to do this is using a kinetic reaction, which will allow you to see what happens to the solution chemistry that is generated by your defined reaction as a function of time. The alternative is to examine the reaction path in abstract terms of "reaction progress", which for an engineer seems unlikely to be as useful as you need.

Finally, if you are actually interested in the forward-model ("What iron concentration and Fe2+/Fe3+ can I expect if the pyritic material weathers at a constant Eh of *** mV?"), you will need to struggle with how in the real world one would maintain that restrictive Eh. If the Eh is not constant in reality, then the actual outcomes will differ from what you have modeled - maybe quite greatly.

There is a very good, clear discussion of this in Craig Bethke's textbook, Geochemical and Biogeochemical Reaction modeling (2nd Ed.) In his Chapter 31 on Acid drainage. Even if you use PHREEQC instead of the Geochemist's Workbench models, his presentation is clear and easy to follow and will help you understand some of the underlying issues (for example, whether the system is open or closed to O2). The graphics that go with React in that suite of models readily produce graphical output that will help you visualize what is happening in the Fe-O-H system, including the effects of complex-ion formation when you include a realistic water-rock reaction model that accounts for ion-association. On the other hand, you can't beat the price of PHREEQC, and it does the underling thermodynamics in the same way.

If you are assaying similar water repeatedly and the matrix of other ions, pH, etc doesn't change tremendously, then you can get rough Fe(II) v Fe(III) numbers by correlating your field measured redox potential with a redox titration in the lab. By redox titration I mean start with the water you are analyzing and measure the ORP response as you drip in oxidant/reductant (permanganate or peroxide and SMBS respectively). Pure ferrous will be on the order of 200-300 mV Ag/AgCl and pure ferric will be 600-700 mV Ag/AgCl. This will not be as accurate as the more involved computational methods Mark describes, but if higher error is tolerable then it will give you an indication of your speciation fairly rapidly.

Alternatively, if you are interested in the Fe+2 and Fe+3 in existing waters where you can measure redox potential, then you could just measure them. There are a range of field-measurement kits that you can use to measure Total Fe and Fe.

If you want to characterize the specific AMD you have to deal with, the simplest way is to titratethe seepage as you will get a reasonable picture of what is in your AMD including the metals. Each pH in your curve will give you some formation about what is precipitating. I have a publication on the use of titration in AMD from a former PhD student of mine.

Many Thanks to everybody, your comments are very useful for me.