redox couple and element charge

[darcy]

Dear Dr Parkhurst,
first thanks for your dedication to replying to so many questions.
Regarding some water elemental concentrations, I realised that in order to keep the same concentrations as the one measured, it is necessary no to input equilibrium phases, otherwise, the software would equilibrate the phases with the solutions giving a new solution composition as the measured one.
And just to assume equilibrium and simple speciation, I wonder what is more correct to input for Redox and for charging purposes. To assume that the element with the biggest concentration to charge balances the equation?  to use the redox couple?
I did use one solution and tried with different combinations of redox couple (Fe(2)/Fe3), O(-2)/O(0)
, pe) and element used for charge.  When charging with Fe for instance, the concentration changes the most, using Na, K,  please refer to an excel file that I am attaching with a sheet using only the first site (from sheet data) in the sheet trial I plot the species concentration against variations on the charge element and Redox( pe or Fe). The differences On species concentrations are large when Fe is used.
Thanks for your time
Kind regards
Darcy

[dlparkhurst]

My preference is not to charge balance the solution unless you suspect one of the analyses is likely the reason for the imbalance. (If the charge balance is very large, I'd probably ignore the analysis.) If you do not know the reason for the charge imbalance, then it makes sense to me to use the analytical data without changing anything, that is accept the charge imbalance. In any subsequent reaction calculations using the solution, PHREEQC will include the charge imbalance in the charge balance equation, as if the charge imbalance is a completely inert ion that does not react with any element and does not contribute to the ionic strength or activity of water.

As for redox, note that a pe is used for redox elements defined as total element. For example, if Fe(2) and Fe(3) are defined, then no pe is used for iron. If Fe (no parentheses) is used, then a pe is used to split Fe among the Fe(2) and Fe(3) redox states. Often, pe has a relatively small effect on the calculations.

[darcy]

When not charging balancing the solution the error and the charge imbalance is large. When I use Na, K or Fe some of my samples fail converge to a numerical solution. So I opted for doing one by one trying different charging elements and results in smaller errors and charge imbalances. The problem is that the charging element gets adjusted and conc is different depending on the charge element used. So which is the best fit?. All elements have a particular mobility/ transport number so they influence differently the whole solution, some are more important than others or am I confused?.
 In McCleskey, 2012 from the USGS, compares electrical conductivities from experiments and calculated with Phreeqc and the percentage difference together with the imbalance charge gives a possible quality control of the results. However, the equations have limited possible electrolytes. The method by Appelo relies on diffusion coefficients which are difficult to find. Waterq doesnt count with all the possible precipitation phases I am suppose to have.


RB McCleskey, DK Nordstrom, JN Ryan: Comparison of electrical conductivity calculation methods for natural waters, Limnol. Oceanogr.: Methods 10, 952–967 (2012)

CAJ Appelo: Specific conductance – how to calculate the specific conductance with PhreeqC (2010), http://www.hydrochemistry.eu/exmpls/sc.html

[dlparkhurst]

You haven't said what you are trying to do with your calculations. Are you simply calculating saturation indices? Are you trying to match specific conductance? Are you planning to do reaction or transport calculations? If you can, use analyses that balance within 5 or 10 percent (although rain samples are so dilute you might relax your criteria).

If you charge balance with an element, then yes, in general the charge imbalance will be reduced to essentially zero. The calculation can fail if, for example there is a positive charge imbalance and you chose potassium for adjustment. In this case, if the charge imbalance is greater than  potassium concentration, it is not possible to remove enough potassium to achieve charge balance.

Again, unless you know the source of the errors, I don't think the calculations are any more accurate if you force charge balance. If you are really into the charge balance errors, plot the charge imbalances to see if they are systematically positive or negative, or if they are correlated with any major element  and measured specific conductance.

If you used phreeqc.dat or pitzer.dat, the specific conductance will appear in the output, and can be written to a selected output file by using the Basic function SC. I don't know anything about your samples, but normally, the elements included in phreeqc.dat include the major contributors to specific conductance. Trace elements are not going to make a significant contribution.

[darcy]

Thank you sir for your reply. I am just trying to establish the conditions in which my samples were, I have measures of elemental composition and conductivity. The problem is that I have some organic matter and as I know it is not likely to be CH4 or C2H2O, it makes more sense to me attribute it to the usual culprits that are found in my systems. Just to say, that this would be the worst-case scenario.
So if I could calculate the individual electric conductivity and attribute the other to the unknown, it would take trial and error but eventually, it will become closer to represent my systems. I have done some data conductivity contribution plots and It is noticed for instance that S(6)>Cl>Ca>HCO3>Na>Mg>K. So a way to calculate my conductivity would be very helpful. However, I am limited to use sit.dat because of the many precipitated phases that are supposed to be found in my systems.

[dlparkhurst]

The additional elements in sit.dat probably do not contribute much to the specific conductance. So, I think you could use phreeqc.dat for your specific conductance investigations by considering only the major ions. You can still use sit.dat calculate saturation indices and reactions.

Organic acids may contribute a bit to the ionic composition of your water, but usually not too much, except in really organic rich environments. PHREEQC does not have a good ligand-binding model, although the Tipping_Hurley.dat database does represent the WHAM model. If you are really interested in the organics, you will need some more specialized analytical techniques.

I'm not sure you can really improve on your analyses by these extra manipulations. Remember that you have to justify your results when you write your paper.

[darcy]

Thank you sir, this is a very good idea. I have additionally to fulvic/humic substances some small carboxylic acids, so I suppose I could take the Master Species from the Tipping and Hurley model and make assumptions regarding an average composition.