webmaster Posted November 5, 2004 Share Posted November 5, 2004 From: Sadoon Morad Subject: two tables for ionic activities in React! When I calculated the ionic activities and mineral saturation indices in React for a water sample I have, I faced two problem for which I have no explanation: (1) I obtained two different tables, in both of which the ionic concentrations are different from my original water composition. It is mentioned for the first table that there are "no minerals in the system" and for the second table some minerals, which were calculated by the software to be in saturation with the water, were used. Which of these tables are correct and relevant to my task (i.e., obtaining free ionic activities and saturation indices)? (2) why are there composite ions in the tables, e.g. NaSO4-, if I am concerned with calculating the free ionic activities and saturation indices? From: Mark Logsdon Subject: two tables for ionic activities in React! You will want to find a copy of Dr. Bethke's text, Geochemical Reaction Modeling (1996: Oxford University Press), which in his messages Craig always refers to as "the Green Book" because of the color of the cover. Much of what you need to understand what the code is trying to say is in Chapters 3 and 5. When one enters a water chemistry as input to React, he (or at least I) generally uses the analytical data from, say a ground water. In most cases, this would use units of mg/L, unless you already have converted that to some other set of units. But in any event, if they are the analytical data, they will represent the full "dissolved" concentrations of all the aqueous species that actually are present. If the sample is a natural water of any degree of complexity (versus a prepared laboratory solution of known composition), you have no idea (nor way of knowing in advance) what those species are. because the linear algebra to sort this out is much more complicated than we can sort out in our heads, we use the computer to calculate the distribution of species from the analytical data and some other constraints that the modeler stipulates (explicitly or implicitly through the code itself). For example, you specify T (or React assumes 25C) and P (1 atm, again by default). So, you now have used up your 2 degrees of freedom. You must handle the proton condition in one way or another (by specifying pH or changing the basis in such a way that the proton condition will be defined; React won't run unless you do. In addition, the code assumes that the solution must be electro-neutral. Furthermore, by selecting (actively or by default) a thermodynamic database, you are choosing (a) the components and ( the phases from which React will choose to do the distribution of species calculations, and also © the equilibrium constants that it will use in those calculations. So, once you enter the complete input, you have defined the problem in such a way that solutions will be constrained to a framework that relates to equilibrium in terms of the Gibbs free-energy minimum for the system. (This is true even for kinetic simulations, because they will be developed in terms of reaction progress, and the distribution of species/mineral-phase saturation indices described in relation to an equilibrium condition.) The basic theory is in Bethke (1996) Chapter 3; the linear algebra in Chapter 5. So, to get on to your specific questions. The first output that you see (the system with no minerals) is the aqueous distribution of species for the defined T,P (and based on the chosen thermo set) assuming that no minerals precipitate and no gases exsolve (to change the overall solution chemistry). Conceptually, this is equivalent to you removing a measurable aliquot of fluid from whatever mineral context in which it existed, magically closing that fluid as a new system in which no new phases are allowed to form; then inferring the ionic composition of that new, homogeneous aqueous phase from the thermodynamic constraint that the ensemble ionic assemblage must have the minimum free energy for any possible set of components. The second set of output begins calculating the same way, but (as always, constrained by the thermo set you have chosen) computes a stable phase assemblage, allowing minerals to precipitate from the solution is they were supersaturated. See Bethke (1996, Section 5.4, p. 75). Because there may be alternative phases (e.g., as between polymorphs such as quartz, cristobalite etc for SiO2) with different free energies of formation, the "stable mineral assemblage" is chosen by React from the possible mineral phases, again, to minimize the free energy of the system as defined. You have the option in React (during problem definition) of restricting the active portion of the database to limit the choices of phases that the code can make). For example, if you are considering a low-T system, you may make the judgment that you believe that neither hematite nor goethite would precipitate directly from solution, so you can "suppress" them, causing Fe(OH)3, a "ferrihydrite"-like phase to be the minimum delta-G ferric oxide/hydroxide. If you do not make this choice, React will assume that it is a conscious decision, and will calculate the stable phase assemblage using the lowest free-energy phase, hematite, to control iron. The reason that the distribution of species looks different in this output is that that model "removes" aqueous Fe when the ferric "precipitates" and then re-calculates the entire distribution of species again. Your second major question: You may be interested only in the free-ion activities, but there is no way for React to tell you what those are unless it does the full distribution of species calculation. (This is why a complete chemical analysis is so important: if you do not include, for example, F concentrations, there is no way for the model to consider F-complexes, and the distribution of species, including the apparent free-ion activities, will be imprecise to the extent that F actually is important in the water.) Because it cannot foresee what specific information you wish to use, it prints out the full distribution of species, including all the aqueous complexes. If you are interested in the free ions, just scan the output for them and make whatever use you wish (e.g., calculate how much of the total analytical Na is present as free-Na+, as opposed to complexed with sulfate or whatever other ligands may be present). The model calculates the saturation indices for you (as it also calculates the ionic strength, activity coefficient and other factors), so you can use them without having to separately take the free-ion activities, look up the equilibrium constants, and do the calculation by hand. From: Sadoon Morad Subject: In case I want to report the saturation indices, which ones shall I use: the ones from the first or second output? From: Mark Logsdon Subject: I'm not sure there is a simple answer to your question. At least I would probably answer it differently depending on what I was trying to do. Here's the strategy I follow most often:1. Look at the first list (with no minerals present) and examine the saturation indices critically in terms of the hydrogeologic environment in which I am working and the specific problem I am trying to address. I am a ground-water geochemist working most often with relatively short-term issues in mining environmental problems. For example, the list may show supersaturation with 3 SiO2 polymorphs (Quartz, Cristobalite, Chalcedony), but not with SiO2(am). Now, at a field GW temperature of say 10C, I don't believe that Qtz, Crist or Chalcedony would precipitate directly from solution, so there's hardly any point to my interpreting the SI values to mean that either (a) Qtz would be a meaningful solid phase in the short-run, or ( the steady-state SiO2(aq) should be reduced to a value consistent with Qtz. Therefore, I need to re-enter the simulation and suppress Quartz, Cristobalite and Chalcedony and reconsider the second list. I'd do the same (at the same time, so as not to spend too much time going back to the beginning) for iron-oxides. My personal view is that - again at near-surface temperatures and for short-term evaluations - all the complex alumino-silicates that will appear if you have both Al+++ and SiO2(aq) in the analyses, also will not precipitate from solution, so I suppress them. (This can be a time-consuming exercise because the databases carry a very large number of these low-T phases [smectites, zeolites etc], and you will suppress one such phase and find it replaced in the output by another anfd so on for a longish time. Eventually, I just decide to suppress all the related phases (e.g., Mordenite-Ca, Mg, ...]. Again, if you look at the initial output, you'll see the phases listed and you can just compile long lists and make the suppression choices once or a few times.) Now, perhaps you are working on a problem in sediment diagenesis or nuclear-waste management, where the temperatures are apt to be higher and the time-frames of evaluation are long enough that more complex relationships - e.g., "solid-state" transformations - may be geochemically plausible. In such cases, you might very reasonably choose chalcedony or even quartz as the silica polymorph and goethite for ferric oxyhydroxide. There are well-done reports of sepiolite in marine evaporites, so perhaps if one is looking at a long-term problem there, sepiolite is a credible Mg-control. I'm sure you see the implications.2. Once I have decided on the full list of credible solid-phase controls for my problem, then I would use the two list together more or less as follows: Cite the values from the first list together with my other geologic or project knowledge as the rationale for having selected the equilibrium assemblage that will be used for my final list two. Effectively, this becomes part of documenting the assumptions of your modeling exercise. Then I would use the saturation indices from List 2 to evaluate how for the "steady state" solution is from equilibrium with other phases - for examples (a) to imagine what would happen if this solution ran through a bed of limestone; ( to imagine what would happen of CO2 degassed (or was added); © to imagine what would happen to the metals if they reacted with a ferric hydroxide phase along the flow path ... Where by "to imagine" I might mean anything from either a conceptual model through to a new sequence of numerical simulations. From: Sadoon Morad Subject: Saturation and precipitation The issue of activity and saturation indices calculations seem to be far more complicated than what imagined. I have seen plenty of SI values published in the literature, but have not, so far, seen any mention of this complexity. Does it mean that there other easier ways of calculating SI? Now to SI with respect to minerals. If we have waters calculated to be saturated with respect to e.g. quartz, can't we then use the value to indicate that quartz in my rock will not dissolve? Rather than to indicate that precipitation should be expected? From: Mark Logsdon Subject: Re: Saturation and precipitation The saturation index concept is basically just a definition that is useful (empirically) for aqueous geochemists: SI = log (IAP/Ksp), where IAP is the ion activity product and Ksp is the solubility product at the temperature of interest. When IAP = Ksp, then delta Gr = 0, meaning that the dissolution - precipitation reaction is at equilibrium. The water would be saturated with the solid phase being considered. If IAP Ksp, the water is supersaturated and the solid phase also would not dissolve. Therefore, your interpretation of the behavior of quartz is correct: a water supersaturated (according to its SI) with respect to quartz would not dissolve further quartz. If you are not concerned about what, if any, phases would precipitate, then you have solved the problem. Stepping back another second, because of the definition of SI, there are no reliable shortcuts. It is the free-ion activity that determines the status with respect to equilibrium. If one were to take the total analytical concentration, or even to use an activity-coefficient to scale the concentration, in most natural waters one would overestimate the IAP (by not accounting for ion-pairs that reduce the free-ion activity), and therefore mis-calculate the SI. If the water is very dilute, the difference may be small, but by the time the total dissolved solids reaches say 200 mg/L, the differences may be important for some minerals. That said, there is an exception: up to significant ionic strengths, neutral species have activity coefficients very close to 1 (so ai ~ mi), and SiO2(aq) does not form significant ion pairs with other dissolved species. So, for dissolved Si, one could simply look at the many tabulations of silica solubility, compare those to the observed water chemistry, and decide whether the silica polymorph you have is or is not close to saturation. Because the distribution of species calculation is difficult to do by hand, everyone now uses one or another of the available computer models to do it, then finds a convenient way to summarize the results, often without explaining how they were derived. I think this is because lots of people think that the matter is so "simple" that it is not worth discussion. My view - and I think yours, now, too - is that like many elementary matters in science, there is nothing much "simple" and certainly not "easy" about this. As you express the issue in the most recent e-mail, I probably would use REACT as Craig advised in his answer: set "precip off" and just examine the output (which would be that of the first list in the default mode). If you then want to do something more where you allow phases to precipitate, remember that you will have to re-enter the command line to set "precip". Quote Link to comment Share on other sites More sharing options...
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.