pyEQL 1.4.0rc9__cp312-cp312-win_amd64.whl

pyEQL/engines.py

@@ -0,0 +1,1153 @@
+"""
+pyEQL engines for computing aqueous equilibria (e.g., speciation, redox, etc.).
+
+:copyright: 2013-2024 by Ryan S. Kingsbury
+:license: LGPL, see LICENSE for more details.
+
+"""
+
+import copy
+import logging
+import math
+import os
+import warnings
+from abc import ABC, abstractmethod
+from pathlib import Path
+from typing import TYPE_CHECKING, Literal
+
+from phreeqpython import PhreeqPython
+
+import pyEQL.activity_correction as ac
+from pyEQL import ureg
+from pyEQL.presets import ATMOSPHERE, EQUILIBRIUM_PHASE_AMOUNT
+from pyEQL.utils import FormulaDict, standardize_formula
+
+# These are the only elements that are allowed to have parenthetical oxidation states
+# PHREEQC will ignore others (e.g., 'Na(1)')
+SPECIAL_ELEMENTS = ["S", "C", "N", "Cu", "Fe", "Mn"]
+
+logger = logging.getLogger(__name__)
+
+if TYPE_CHECKING:
+    from pyEQL import solution
+
+
+"""
+We determine if PhreeqPython from the phreeqpython package is actually
+invocable on this platform, regardless of whether it is installed.
+This flag helps us selectively skip phreeqpython-related tests inside
+manylinux containers, among other uses.
+"""
+
+
+def _phreeqpython_available():
+    try:
+        PhreeqPython()
+    except:  # noqa: E722
+        return False
+    else:
+        return True
+
+
+PHREEQPYTHON_AVAILABLE = _phreeqpython_available()
+
+
+class EOS(ABC):
+    """
+    Abstract base class for pyEQL equation of state classes.
+
+    The intent is that concrete implementations of this class make use of the
+    standalone functions available in pyEQL.activity_correction and pyEQL.equilibrium
+    as much as possible. This facilitates robust unit testing while allowing users
+    to "mix and match" or customize the various models as needed.
+    """
+
+    @abstractmethod
+    def get_activity_coefficient(self, solution: "solution.Solution", solute: str) -> ureg.Quantity:
+        """
+        Return the *molal scale* activity coefficient of solute, given a Solution
+        object.
+
+        Args:
+            solution: pyEQL Solution object
+            solute: str identifying the solute of interest
+
+        Returns:
+            Quantity: dimensionless quantity object
+
+        Raises:
+            ValueError if the calculation cannot be completed, e.g. due to insufficient number of parameters.
+        """
+
+    @abstractmethod
+    def get_osmotic_coefficient(self, solution: "solution.Solution") -> ureg.Quantity:
+        """
+        Return the *molal scale* osmotic coefficient of a Solution.
+
+        Args:
+            solution: pyEQL Solution object
+
+        Returns:
+            Quantity: dimensionless molal scale osmotic coefficient
+
+        Raises:
+            ValueError if the calculation cannot be completed, e.g. due to insufficient number of parameters.
+        """
+
+    @abstractmethod
+    def get_solute_volume(self, solution: "solution.Solution") -> ureg.Quantity:
+        """
+        Return the volume of only the solutes.
+
+        Args:
+            solution: pyEQL Solution object
+
+        Returns:
+            Quantity: solute volume in L
+
+        Raises:
+            ValueError if the calculation cannot be completed, e.g. due to insufficient number of parameters.
+        """
+
+    @abstractmethod
+    def equilibrate(self, solution: "solution.Solution") -> None:
+        """
+        Adjust the speciation and pH of a Solution object to achieve chemical equilibrium.
+
+        The Solution should be modified in-place, likely using add_moles / set_moles, etc.
+
+        Args:
+            solution: pyEQL Solution object
+
+        Returns:
+            Nothing. The speciation of the Solution is modified in-place.
+
+        Raises:
+            ValueError if the calculation cannot be completed, e.g. due to insufficient number of parameters or lack of convergence.
+        """
+
+
+class IdealEOS(EOS):
+    """Ideal solution equation of state engine."""
+
+    def get_activity_coefficient(self, solution: "solution.Solution", solute: str) -> ureg.Quantity:
+        """
+        Return the *molal scale* activity coefficient of solute, given a Solution
+        object.
+        """
+        return ureg.Quantity(1, "dimensionless")
+
+    def get_osmotic_coefficient(self, solution: "solution.Solution") -> ureg.Quantity:
+        """
+        Return the *molal scale* osmotic coefficient of solute, given a Solution
+        object.
+        """
+        return ureg.Quantity(1, "dimensionless")
+
+    def get_solute_volume(self, solution: "solution.Solution") -> ureg.Quantity:
+        """Return the volume of the solutes."""
+        return ureg.Quantity(0, "L")
+
+    def equilibrate(self, solution: "solution.Solution") -> None:
+        """Adjust the speciation of a Solution object to achieve chemical equilibrium."""
+        warnings.warn("equilibrate() has no effect in IdealEOS!")
+        return
+
+
+class NativeEOS(EOS):
+    """
+    pyEQL's native EOS. Uses the Pitzer model when possible, falls
+    back to other models (e.g. Debye-Huckel) based on ionic strength
+    if sufficient parameters are not available.
+    """
+
+    def __init__(
+        self,
+        phreeqc_db: Literal[
+            "phreeqc.dat", "vitens.dat", "wateq4f_PWN.dat", "pitzer.dat", "llnl.dat", "geothermal.dat"
+        ] = "llnl.dat",
+    ) -> None:
+        """
+        Args:
+            phreeqc_db: Name of the PHREEQC database file to use for solution thermodynamics
+                and speciation calculations. Generally speaking, `llnl.dat` is recommended
+                for moderate salinity water and prediction of mineral solubilities,
+                `wateq4f_PWN.dat` is recommended for low to moderate salinity waters. It is
+                similar to vitens.dat but has many more species. `pitzer.dat` is recommended
+                when accurate activity coefficients in solutions above 1 M TDS are desired, but
+                it has fewer species than the other databases. `llnl.dat` and `geothermal.dat`
+                may offer improved prediction of LSI but currently these databases are not
+                usable because they do not allow for conductivity calculations.
+        """
+        self.phreeqc_db = phreeqc_db
+        # database files in this list are not distributed with phreeqpython
+        self.db_path = (
+            Path(os.path.dirname(__file__)) / "database" if self.phreeqc_db in ["llnl.dat", "geothermal.dat"] else None
+        )
+        # create the PhreeqcPython instance
+        # try/except added to catch unsupported architectures, such as Apple Silicon
+        try:
+            self.pp = PhreeqPython(database=self.phreeqc_db, database_directory=self.db_path)
+        except OSError:
+            logger.error(
+                "OSError encountered when trying to instantiate phreeqpython. Most likely this means you"
+                " are running on an architecture that is not supported by PHREEQC, such as Apple M1/M2 chips."
+                " pyEQL will work, but equilibrate() will have no effect."
+            )
+        # attributes to hold the PhreeqPython solution.
+        self.ppsol = None
+        # store the solution composition to see whether we need to re-instantiate the solution
+        self._stored_comp = None
+
+    def _setup_ppsol(self, solution: "solution.Solution") -> None:
+        """Helper method to set up a PhreeqPython solution for subsequent analysis."""
+        self._stored_comp = solution.components.copy()
+        solv_mass = solution.solvent_mass.to("kg").magnitude
+        # inherit bulk solution properties
+        d = {
+            "temp": solution.temperature.to("degC").magnitude,
+            "units": "mol/kgw",  # to avoid confusion about volume, use mol/kgw which seems more robust in PHREEQC
+            "pH": solution.pH,
+            "pe": solution.pE,
+            "redox": "pe",  # hard-coded to use the pe
+            # PHREEQC will assume 1 kg if not specified, there is also no direct way to specify volume, so we
+            # really have to specify the solvent mass in 1 liter of solution
+            "water": solv_mass,
+        }
+        if solution.balance_charge == "pH":
+            d["pH"] = str(d["pH"]) + " charge"
+        if solution.balance_charge == "pE":
+            d["pe"] = str(d["pe"]) + " charge"
+
+        # add the composition to the dict
+        # also, skip H and O
+        for el, mol in solution.get_el_amt_dict().items():
+            # CAUTION - care must be taken to avoid unintended behavior here. get_el_amt_dict() will return
+            # all distinct oxi states of each element present. If there are elements present whose oxi states
+            # are NOT recognized by PHREEQC (via SPECIAL_ELEMENTS) then the amount of only 1 oxi state will be
+            # entered into the composition dict. This can especially cause problems after equilibrate() has already
+            # been called once. For example, equilibrating a simple NaCl solution generates Cl species that are assigned
+            # various oxidations states, -1 mostly, but also 1, 2, and 3. Since the concentrations of everything
+            # except the -1 oxi state are tiny, this can result in Cl "disappearing" from the solution if
+            # equlibrate is called again. It also causes non-determinism, because the amount is taken from whatever
+            # oxi state happens to be iterated through last.
+
+            # strip off the oxi state
+            bare_el = el.split("(")[0]
+            if bare_el in SPECIAL_ELEMENTS:
+                # PHREEQC will ignore float-formatted oxi states. Need to make sure we are
+                # passing, e.g. 'C(4)' and not 'C(4.0)'
+                key = f"{bare_el}({int(float(el.split('(')[-1].split(')')[0]))})"
+            elif bare_el in ["H", "O"]:
+                continue
+            else:
+                key = bare_el
+
+            if key in d:
+                # when multiple oxi states for the same (non-SPECIAL) element are present, make sure to
+                # add all their amounts together
+                d[key] += str(mol / solv_mass)
+            else:
+                d[key] = str(mol / solv_mass)
+
+            # tell PHREEQC which species to use for charge balance
+            if solution.balance_charge is not None and solution._cb_species in solution.get_components_by_element()[el]:
+                d[key] += " charge"
+
+        # create the PHREEQC solution object
+        try:
+            ppsol = self.pp.add_solution(d)
+        except Exception as e:
+            print(d)
+            # catch problems with the input to phreeqc
+            raise ValueError(
+                "There is a problem with your input. The error message received from "
+                f" phreeqpython is:\n\n {e}\n Check your input arguments, especially "
+                "the composition dictionary, and try again."
+            )
+
+        self.ppsol = ppsol
+
+    def _destroy_ppsol(self) -> None:
+        """Remove the PhreeqPython solution from memory."""
+        if self.ppsol is not None:
+            self.ppsol.forget()
+            self.ppsol = None
+
+    def get_activity_coefficient(self, solution: "solution.Solution", solute: str):
+        r"""
+        Whenever the appropriate parameters are available, the Pitzer model [may11]_ is used.
+        If no Pitzer parameters are available, then the appropriate equations are selected
+        according to the following logic: [stumm96]_.
+
+        I <= 0.0005: Debye-Huckel equation
+        0.005 < I <= 0.1:  Guntelberg approximation
+        0.1 < I <= 0.5: Davies equation
+        I > 0.5: Raises a warning and returns activity coefficient = 1
+
+        The ionic strength, activity coefficients, and activities are all
+        calculated based on the molal (mol/kg) concentration scale. If a different
+        scale is given as input, then the molal-scale activity coefficient :math:`\gamma_\pm` is
+        converted according to [rbs68]_
+
+        .. math:: f_\pm = \gamma_\pm * (1 + M_w \sum_i \nu_i m_i)
+
+        .. math:: y_\pm = \frac{m \rho_w}{C \gamma_\pm}
+
+        where :math:`f_\pm` is the rational activity coefficient, :math:`M_w` is
+        the molecular weight of water, the summation represents the total molality of
+        all solute  species, :math:`y_\pm` is the molar activity coefficient,
+        :math:`\rho_w` is the density of pure water, :math:`m` and :math:`C` are
+        the molal and molar concentrations of the chosen salt (not individual solute), respectively.
+
+        Args:
+            solute: String representing the name of the solute of interest
+            scale: The concentration scale for the returned activity coefficient.
+                Valid options are "molal", "molar", and "rational" (i.e., mole fraction).
+                By default, the molal scale activity coefficient is returned.
+
+        Returns:
+            The mean ion activity coefficient of the solute in question on  the selected scale.
+
+        Notes:
+            For multicomponent mixtures, pyEQL implements the "effective Pitzer model"
+            presented by Mistry et al. [mistry13]_. In this model, the activity coefficient
+            of a salt in a multicomponent mixture is calculated using an "effective
+            molality," which is the molality that would result in a single-salt
+            mixture with the same total ionic strength as the multicomponent solution.
+
+            .. math:: m_{effective} = \frac{2 I}{(\nu_{+} z_{+}^2 + \nu_{-}- z_{-}^2)}
+
+        References:
+            .. [may11] May, P. M., Rowland, D., Hefter, G., & Königsberger, E. (2011).
+                     A Generic and Updatable Pitzer Characterization of Aqueous Binary Electrolyte Solutions at 1 bar
+                     and 25 °C. *Journal of Chemical & Engineering Data*, 56(12), 5066-5077. doi:10.1021/je2009329
+
+            .. [stumm96] Stumm, Werner and Morgan, James J. *Aquatic Chemistry*, 3rd ed,
+                       pp 165. Wiley Interscience, 1996.
+
+            .. [rbs68] Robinson, R. A.; Stokes, R. H. Electrolyte Solutions: Second Revised
+                     Edition; Butterworths: London, 1968, p.32.
+
+        See Also:
+            :attr:`pyEQL.solution.Solution.ionic_strength`
+            :func:`pyEQL.activity_correction.get_activity_coefficient_debyehuckel`
+            :func:`pyEQL.activity_correction.get_activity_coefficient_guntelberg`
+            :func:`pyEQL.activity_correction.get_activity_coefficient_davies`
+            :func:`pyEQL.activity_correction.get_activity_coefficient_pitzer`
+        """
+        # identify the predominant salt that this ion is a member of
+        salt = None
+        rform = standardize_formula(solute)
+        for d in solution.get_salt_dict().values():
+            if rform == d["salt"].cation or rform == d["salt"].anion:
+                salt = d["salt"]
+                break
+
+        # show an error if no salt can be found that contains the solute
+        if salt is None:
+            logger.error(f"No salts found that contain solute {solute}. Returning unit activity coefficient.")
+            return ureg.Quantity(1, "dimensionless")
+
+        # use the Pitzer model for higher ionic strength, if the parameters are available
+        # search for Pitzer parameters
+        param = solution.get_property(salt.formula, "model_parameters.activity_pitzer")
+        if param is not None:
+            # TODO - consider re-enabling a log message recording what salt(s) are used as basis for activity calculation
+            logger.info(f"Calculating activity coefficient based on parent salt {salt.formula}")
+
+            # determine alpha1 and alpha2 based on the type of salt
+            # see the May reference for the rules used to determine
+            # alpha1 and alpha2 based on charge
+            if salt.z_cation >= 2 and salt.z_anion <= -2:
+                if salt.z_cation >= 3 or salt.z_anion <= -3:
+                    alpha1 = 2.0
+                    alpha2 = 50.0
+                else:
+                    alpha1 = 1.4
+                    alpha2 = 12
+            else:
+                alpha1 = 2.0
+                alpha2 = 0.0
+
+            # determine the average molality of the salt
+            # this is necessary for solutions inside e.g. an ion exchange
+            # membrane, where the cation and anion concentrations may be
+            # unequal
+            # molality = (solution.get_amount(salt.cation,'mol/kg')/salt.nu_cation+solution.get_amount(salt.anion,'mol/kg')/salt.nu_anion)/2
+
+            # determine the effective molality of the salt in the solution
+            molality = salt.get_effective_molality(solution.ionic_strength)
+
+            activity_coefficient = ac.get_activity_coefficient_pitzer(
+                solution.ionic_strength,
+                molality,
+                alpha1,
+                alpha2,
+                ureg.Quantity(param["Beta0"]["value"]).magnitude,
+                ureg.Quantity(param["Beta1"]["value"]).magnitude,
+                ureg.Quantity(param["Beta2"]["value"]).magnitude,
+                ureg.Quantity(param["Cphi"]["value"]).magnitude,
+                salt.z_cation,
+                salt.z_anion,
+                salt.nu_cation,
+                salt.nu_anion,
+                str(solution.temperature),
+            )
+
+            logger.debug(
+                f"Calculated activity coefficient of species {solute} as {activity_coefficient} based on salt"
+                f" {salt} using Pitzer model"
+            )
+            molal = activity_coefficient
+
+        # for very low ionic strength, use the Debye-Huckel limiting law
+        elif solution.ionic_strength.magnitude <= 0.005:
+            logger.debug(
+                f"Ionic strength = {solution.ionic_strength}. Using Debye-Huckel to calculate activity coefficient."
+            )
+            molal = ac.get_activity_coefficient_debyehuckel(
+                solution.ionic_strength,
+                solution.get_property(solute, "charge"),
+                str(solution.temperature),
+            )
+
+        # use the Guntelberg approximation for 0.005 < I < 0.1
+        elif solution.ionic_strength.magnitude <= 0.1:
+            logger.debug(
+                f"Ionic strength = {solution.ionic_strength}. Using Guntelberg to calculate activity coefficient."
+            )
+            molal = ac.get_activity_coefficient_guntelberg(
+                solution.ionic_strength,
+                solution.get_property(solute, "charge"),
+                str(solution.temperature),
+            )
+
+        # use the Davies equation for 0.1 < I < 0.5
+        elif solution.ionic_strength.magnitude <= 0.5:
+            logger.debug(
+                f"Ionic strength = {solution.ionic_strength}. Using Davies equation to calculate activity coefficient."
+            )
+            molal = ac.get_activity_coefficient_davies(
+                solution.ionic_strength,
+                solution.get_property(solute, "charge"),
+                str(solution.temperature),
+            )
+
+        else:
+            logger.error(
+                f"Ionic strength too high to estimate activity for species {solute}. Specify parameters for Pitzer "
+                "model. Returning unit activity coefficient"
+            )
+
+            molal = ureg.Quantity(1, "dimensionless")
+
+        return molal
+
+    def get_osmotic_coefficient(self, solution: "solution.Solution") -> ureg.Quantity:
+        r"""
+        Return the *molal scale* osmotic coefficient of solute, given a Solution
+        object.
+
+        Osmotic coefficient is calculated using the Pitzer model. [may11]_ If appropriate parameters for
+        the model are not available, then pyEQL raises a WARNING and returns an osmotic
+        coefficient of 1.
+
+        If the 'rational' scale is given as input, then the molal-scale osmotic
+        coefficient :math:`\phi` is converted according to [rbs68]_
+
+        .. math:: g = - \phi M_{w} \frac{\sum_{i} \nu_{i} m_{i}}{\ln x_{w}}
+
+        where :math:`g` is the rational osmotic coefficient, :math:`M_{w}` is
+        the molecular weight of water, the summation represents the total molality of
+        all solute  species, and :math:`x_{w}` is the mole fraction of water.
+
+        Args:
+            scale: The concentration scale for the returned osmotic coefficient. Valid options are "molal",
+                "rational" (i.e., mole fraction), and "fugacity".  By default, the molal scale osmotic
+                coefficient is returned.
+
+        Returns:
+            Quantity:
+                The osmotic coefficient
+
+        See Also:
+            :meth:`pyEQL.solution.Solution.get_water_activity`
+            :meth:`pyEQL.solution.Solution.get_salt`
+            :attr:`pyEQL.solution.Solution.ionic_strength`
+
+        Notes:
+            For multicomponent mixtures, pyEQL adopts the "effective Pitzer model"
+            presented by Mistry et al. [mistry13]_. In this approach, the osmotic coefficient of
+            each individual salt is calculated using the normal Pitzer model based
+            on its respective concentration. Then, an effective osmotic coefficient
+            is calculated as the concentration-weighted average of the individual
+            osmotic coefficients.
+
+            For example, in a mixture of 0.5 M NaCl and 0.5 M KBr, one would calculate
+            the osmotic coefficient for each salt using a concentration of 0.5 M and
+            an ionic strength of 1 M. Then, one would average the two resulting
+            osmotic coefficients to obtain an effective osmotic coefficient for the
+            mixture.
+
+            (Note: in the paper referenced below, the effective osmotic coefficient is determined by weighting
+            using the "effective molality" rather than the true molality. Subsequent checking and correspondence with
+            the author confirmed that the weight factor should be the true molality, and that is what is implemented
+            in pyEQL.)
+
+        Examples:
+            >>> s1 = pyEQL.Solution({'Na+': '0.2 mol/kg', 'Cl-': '0.2 mol/kg'})
+            >>> s1.get_osmotic_coefficient()
+            <Quantity(0.923715281, 'dimensionless')>
+
+            >>> s1 = pyEQL.Solution({'Mg+2': '0.3 mol/kg', 'Cl-': '0.6 mol/kg'},temperature='30 degC')
+            >>> s1.get_osmotic_coefficient()
+            <Quantity(0.891409618, 'dimensionless')>
+
+        References:
+            [may11]
+
+            May, P. M., Rowland, D., Hefter, G., & Königsberger, E. (2011).
+            A Generic and Updatable Pitzer Characterization of Aqueous Binary Electrolyte Solutions at 1 bar
+            and 25 °C. Journal of Chemical & Engineering Data, 56(12), 5066-5077. doi:10.1021/je2009329
+
+            [rbs68]
+
+            Robinson, R. A.; Stokes, R. H. Electrolyte Solutions: Second Revised Edition; Butterworths: London, 1968,
+            p.32.
+
+            [mistry13]
+
+            Mistry, K. H.; Hunter, H. a.; Lienhard V, J. H. Effect of composition and nonideal solution
+            behavior on desalination calculations for mixed electrolyte solutions with comparison to
+            seawater. Desalination 2013, 318, 34-47.
+
+        """
+        ionic_strength = solution.ionic_strength
+
+        effective_osmotic_sum = 0
+        molality_sum = 0
+
+        # loop through all the salts in the solution, calculate the osmotic
+        # coefficint for each, and average them into an effective osmotic
+        # coefficient
+        for d in solution.get_salt_dict().values():
+            item = d["salt"]
+            # determine alpha1 and alpha2 based on the type of salt
+            # see the May reference for the rules used to determine
+            # alpha1 and alpha2 based on charge
+            if item.z_cation >= 2 and item.z_anion <= -2:
+                if item.z_cation >= 3 or item.z_anion <= -3:
+                    alpha1 = 2.0
+                    alpha2 = 50.0
+                else:
+                    alpha1 = 1.4
+                    alpha2 = 12.0
+            else:
+                alpha1 = 2.0
+                alpha2 = 0
+
+            # set the concentration as the average concentration of the cation and
+            # anion in the salt, accounting for stoichiometry
+            # concentration = (solution.get_amount(Salt.cation,'mol/kg')/Salt.nu_cation + \
+            # solution.get_amount(Salt.anion,'mol/kg')/Salt.nu_anion)/2
+
+            # get the effective molality of the salt
+            concentration = ureg.Quantity(d["mol"], "mol") / solution.solvent_mass
+
+            molality_sum += concentration
+
+            param = solution.get_property(item.formula, "model_parameters.activity_pitzer")
+            if param is not None:
+                osmotic_coefficient = ac.get_osmotic_coefficient_pitzer(
+                    ionic_strength,
+                    concentration,
+                    alpha1,
+                    alpha2,
+                    ureg.Quantity(param["Beta0"]["value"]).magnitude,
+                    ureg.Quantity(param["Beta1"]["value"]).magnitude,
+                    ureg.Quantity(param["Beta2"]["value"]).magnitude,
+                    ureg.Quantity(param["Cphi"]["value"]).magnitude,
+                    item.z_cation,
+                    item.z_anion,
+                    item.nu_cation,
+                    item.nu_anion,
+                    str(solution.temperature),
+                )
+
+                logger.debug(
+                    f"Calculated osmotic coefficient of water as {osmotic_coefficient} based on salt "
+                    f"{item.formula} using Pitzer model"
+                )
+                effective_osmotic_sum += concentration * osmotic_coefficient
+
+            else:
+                logger.debug(
+                    f"Returning unit osmotic coefficient for salt {item.formula} because Pitzer parameters are not"
+                    "available in database."
+                )
+                effective_osmotic_sum += concentration * 1
+
+        try:
+            return effective_osmotic_sum / molality_sum
+        except ZeroDivisionError:
+            # this means the solution is empty
+            return ureg.Quantity(1.0)
+
+    def get_solute_volume(self, solution: "solution.Solution") -> ureg.Quantity:
+        """Return the volume of the solutes."""
+        # identify the predominant salt in the solution
+        salt = solution.get_salt()
+        solute_vol = ureg.Quantity(0, "L")
+
+        # use the pitzer approach if parameters are available
+        pitzer_calc = False
+        param = None if salt is None else solution.get_property(salt.formula, "model_parameters.molar_volume_pitzer")
+
+        if param is not None:
+            # determine the average molality of the salt
+            # this is necessary for solutions inside e.g. an ion exchange
+            # membrane, where the cation and anion concentrations may be
+            # unequal
+            molality = (solution.get_amount(salt.cation, "mol/kg") + solution.get_amount(salt.anion, "mol/kg")) / 2
+
+            # determine alpha1 and alpha2 based on the type of salt
+            # see the May reference for the rules used to determine
+            # alpha1 and alpha2 based on charge
+            if salt.z_cation >= 2 and salt.z_anion <= -2:
+                if salt.z_cation >= 3 or salt.z_anion <= -3:
+                    alpha1 = 2.0
+                    alpha2 = 50.0
+                else:
+                    alpha1 = 1.4
+                    alpha2 = 12
+            else:
+                alpha1 = 2.0
+                alpha2 = 0.0
+
+            apparent_vol = ac.get_apparent_volume_pitzer(
+                solution.ionic_strength,
+                molality,
+                alpha1,
+                alpha2,
+                ureg.Quantity(param["Beta0"]["value"]).magnitude,
+                ureg.Quantity(param["Beta1"]["value"]).magnitude,
+                ureg.Quantity(param["Beta2"]["value"]).magnitude,
+                ureg.Quantity(param["Cphi"]["value"]).magnitude,
+                ureg.Quantity(param["V_o"]["value"]).magnitude,
+                salt.z_cation,
+                salt.z_anion,
+                salt.nu_cation,
+                salt.nu_anion,
+                str(solution.temperature),
+            )
+
+            solute_vol += (
+                apparent_vol
+                * (
+                    solution.get_amount(salt.cation, "mol") / salt.nu_cation
+                    + solution.get_amount(salt.anion, "mol") / salt.nu_anion
+                )
+                / 2
+            )
+
+            pitzer_calc = True
+
+            logger.debug(f"Updated solution volume using Pitzer model for solute {salt.formula}")
+
+        # add the partial molar volume of any other solutes, except for water
+        # or the parent salt, which is already accounted for by the Pitzer parameters
+        for solute, mol in solution.components.items():
+            # ignore water
+            if solute in ["H2O", "HOH", "H2O(aq)"]:
+                continue
+
+            # ignore the salt cation and anion, if already accounted for by Pitzer
+            if pitzer_calc is True and solute in [salt.anion, salt.cation]:
+                continue
+
+            part_vol = solution.get_property(solute, "size.molar_volume")
+            if part_vol is not None:
+                solute_vol += part_vol * ureg.Quantity(mol, "mol")
+                logger.debug(f"Updated solution volume using direct partial molar volume for solute {solute}")
+
+            else:
+                logger.warning(
+                    f"Volume of solute {solute} will be ignored because partial molar volume data are not available."
+                )
+
+        return solute_vol.to("L")
+
+    def equilibrate(
+        self,
+        solution: "solution.Solution",
+        atmosphere: bool = False,
+        solids: list[str] | None = None,
+        gases: dict[str, str | float] | None = None,
+    ) -> None:
+        """
+        Adjust the speciation of a Solution object to achieve chemical equilibrium.
+
+        Args:
+            atmosphere:
+                Boolean indicating whether to equilibrate the solution
+                w.r.t atmospheric gases.
+            solids:
+                A list of solids used to achieve liquid-solid equilibrium. Each
+                solid in this list should be present in the Phreeqc database.
+                We assume a target saturation index of 0 and an infinite
+                amount of material.
+            gases:
+                A dictionary of gases used to achieve liquid-gas equilibrium.
+                Each key denotes the gas species, and the corresponding value
+                denotes its concentration, as a log partial pressure value or
+                other interpretable pressure units. For example, the following
+                are equivalent (log10(0.000316) = -3.5)
+                {"CO2": "0.000316 atm"}
+                {"CO2": -3.5}
+        """
+        if self.ppsol is not None:
+            self.ppsol.forget()
+        self._setup_ppsol(solution)
+
+        # store the original solvent mass
+        orig_solvent_moles = solution.components[solution.solvent]
+
+        # store the original solvent elements and components with that element
+        # that we may need to add back later.
+        orig_el_dict = solution.get_el_amt_dict(nested=True)
+        orig_components_by_element = solution.get_components_by_element(nested=True)
+
+        # Use supplied gases, merged with atmospheric gases
+        # if atmosphere == True
+        gases = (ATMOSPHERE if atmosphere else {}) | (gases or {})
+
+        # Mapping from phase name to:
+        #   (<saturation_index>, <amount_in_moles>) tuples (for solids).
+        #   (<log_partial_pressure>, <amount_in_moles>) tuples (for gases).
+        phases = {}
+        if solids is not None:
+            # Assume saturation index of 0 for all solids.
+            phases |= dict.fromkeys(solids, (0, EQUILIBRIUM_PHASE_AMOUNT))
+
+        for k, v in gases.items():
+            v_quantity = ureg.Quantity(v)
+            if v_quantity.dimensionless:
+                log_partial_pressure = v_quantity.magnitude
+            else:
+                log_partial_pressure = math.log10(v_quantity.to("atm").magnitude)
+            phases |= {f"{k}(g)": (log_partial_pressure, EQUILIBRIUM_PHASE_AMOUNT)}
+
+        if phases:
+            phase_names = list(phases.keys())
+            saturation_indices = [v[0] for v in phases.values()]
+            amounts = [v[1] for v in phases.values()]
+
+            try:
+                self.ppsol.equalize(phases=phase_names, to_si=saturation_indices, in_phase=amounts)
+            except Exception as e:
+                # Re-raise exception due to unrecognized phases as ValueError.
+                if "Phase not found in database" in str(e):
+                    raise ValueError(str(e))
+                raise e
+
+        solution.components = FormulaDict({})
+        # use the output from PHREEQC to update the Solution composition
+        # the .species_moles attribute should return MOLES (not moles per ___)
+        for s, mol in self.ppsol.species_moles.items():
+            solution.components[s] = mol
+
+        # log a message if any components were not touched by PHREEQC
+        # if that was the case, re-adjust the charge balance to account for those species (since PHREEQC did not)
+        missing_species = set(self._stored_comp.keys()) - {standardize_formula(s) for s in self.ppsol.species}
+        if len(missing_species) > 0:
+            logger.warning(
+                f"After equilibration, the amounts of species {sorted(missing_species)} were not modified "
+                "by PHREEQC. These species are likely absent from its database."
+            )
+
+        # tolerance (in moles) for detecting cases where an element amount
+        # is no longer balanced because of species that are not recognized
+        # by PHREEQC.
+        _atol = 1e-16
+
+        new_el_dict = solution.get_el_amt_dict(nested=True)
+        for el in orig_el_dict:
+            orig_el_amount = sum([orig_el_dict[el][k] for k in orig_el_dict[el]])
+            new_el_amount = sum([new_el_dict[el][k] for k in new_el_dict.get(el, [])])
+
+            # If this element went "missing", add back all components that
+            # contain this element (for any valence value)
+            if orig_el_amount - new_el_amount > _atol:
+                logger.info(
+                    f"PHREEQC discarded element {el} during equilibration. Adding all components for this element."
+                )
+                solution.components.update(
+                    {
+                        component: self._stored_comp[component]
+                        for components in orig_components_by_element[el].values()
+                        for component in components
+                        if component not in solution.components
+                    }
+                )
+
+        # re-adjust charge balance for any missing species
+        # note that if balance_charge is set, it will have been passed to PHREEQC, so the only reason to re-adjust charge balance here is to account for any missing species.
+        solution._adjust_charge_balance()
+
+        # rescale the solvent mass to ensure the total mass of solution does not change
+        # this is important because PHREEQC and the pyEQL database may use slightly different molecular
+        # weights for water. Since water amount is passed to PHREEQC in kg but returned in moles, each
+        # call to equilibrate can thus result in a slight change in the Solution mass.
+        solution.components[solution.solvent] = orig_solvent_moles
+
+    def __deepcopy__(self, memo) -> "NativeEOS":
+        # custom deepcopy required because the PhreeqPython instance used by the Native and Phreeqc engines
+        # is not pickle-able.
+
+        cls = self.__class__
+        result = cls.__new__(cls)
+        memo[id(self)] = result
+        for k, v in self.__dict__.items():
+            if k == "pp":
+                result.pp = PhreeqPython(database=self.phreeqc_db, database_directory=self.db_path)
+                continue
+            setattr(result, k, copy.deepcopy(v, memo))
+        return result
+
+
+class PhreeqcEOS(NativeEOS):
+    """Engine based on the PhreeqC model, as implemented via the phreeqpython package."""
+
+    def __init__(
+        self,
+        phreeqc_db: Literal[
+            "phreeqc.dat", "vitens.dat", "wateq4f_PWN.dat", "pitzer.dat", "llnl.dat", "geothermal.dat"
+        ] = "phreeqc.dat",
+    ) -> None:
+        """
+        Args:
+        phreeqc_db: Name of the PHREEQC database file to use for solution thermodynamics
+                and speciation calculations. Generally speaking, `llnl.dat` is recommended
+                for moderate salinity water and prediction of mineral solubilities,
+                `wateq4f_PWN.dat` is recommended for low to moderate salinity waters. It is
+                similar to vitens.dat but has many more species. `pitzer.dat` is recommended
+                when accurate activity coefficients in solutions above 1 M TDS are desired, but
+                it has fewer species than the other databases. `llnl.dat` and `geothermal.dat`
+                may offer improved prediction of LSI but currently these databases are not
+                usable because they do not allow for conductivity calculations.
+        """
+        super().__init__(phreeqc_db=phreeqc_db)
+
+    def get_activity_coefficient(self, solution: "solution.Solution", solute: str) -> ureg.Quantity:
+        """
+        Return the *molal scale* activity coefficient of solute, given a Solution
+        object.
+        """
+        if (self.ppsol is None) or (solution.components != self._stored_comp):
+            self._destroy_ppsol()
+            self._setup_ppsol(solution)
+
+        # translate the species into keys that phreeqc will understand
+        k = standardize_formula(solute)
+        spl = k.split("[")
+        el = spl[0]
+        chg = spl[1].split("]")[0]
+        if chg[-1] == "1":
+            chg = chg[0]  # just pass + or -, not +1 / -1
+        k = el + chg
+
+        # calculate the molal scale activity coefficient
+        # act = self.ppsol.activity(k, "mol") / self.ppsol.molality(k, "mol")
+        act = self.ppsol.pp.ip.get_activity(self.ppsol.number, k) / self.ppsol.pp.ip.get_molality(self.ppsol.number, k)
+
+        return ureg.Quantity(act, "dimensionless")
+
+    def get_osmotic_coefficient(self, solution: "solution.Solution") -> ureg.Quantity:
+        """
+        Return the *molal scale* osmotic coefficient of solute, given a Solution
+        object.
+
+        PHREEQC appears to assume a unit osmotic coefficient unless the pitzer database
+        is used. Unfortunately, there is no easy way to access the osmotic coefficient
+        via phreeqcpython
+        """
+        # TODO - find a way to access or calculate osmotic coefficient
+        return ureg.Quantity(1, "dimensionless")
+
+    def get_solute_volume(self, solution: "solution.Solution") -> ureg.Quantity:
+        """Return the volume of the solutes."""
+        # TODO - phreeqc seems to have no concept of volume, but it does calculate density
+        return ureg.Quantity(0, "L")
+
+
+class Phreeqc2026EOS(EOS):
+    """Engine based on the PhreeqC model, as implemented in the pyphreeqc
+    module of pyEQL."""
+
+    def __init__(
+        self,
+        phreeqc_db: Literal[
+            "phreeqc.dat", "vitens.dat", "wateq4f_PWN.dat", "pitzer.dat", "llnl.dat", "geothermal.dat"
+        ] = "phreeqc.dat",
+    ) -> None:
+        """
+        Args:
+            phreeqc_db: Name of the PHREEQC database file to use for solution thermodynamics
+                and speciation calculations. Generally speaking, `llnl.dat` is recommended
+                for moderate salinity water and prediction of mineral solubilities,
+                `wateq4f_PWN.dat` is recommended for low to moderate salinity waters. It is
+                similar to vitens.dat but has many more species. `pitzer.dat` is recommended
+                when accurate activity coefficients in solutions above 1 M TDS are desired, but
+                it has fewer species than the other databases. `llnl.dat` and `geothermal.dat`
+                may offer improved prediction of LSI but currently these databases are not
+                usable because they do not allow for conductivity calculations.
+        """
+
+        from pyEQL.phreeqc import IS_AVAILABLE, Phreeqc  # noqa: PLC0415
+
+        if not IS_AVAILABLE:
+            raise RuntimeError("pyEQL phreeqc support is not available in this installation")
+
+        self.phreeqc_db = phreeqc_db
+        # database files in this list are not distributed with phreeqpython
+        self.db_path = (
+            Path(os.path.dirname(__file__)) / "database" if self.phreeqc_db in ["llnl.dat", "geothermal.dat"] else None
+        )
+        # create the PhreeqcPython instance
+        self.pp = Phreeqc(database=self.phreeqc_db, database_directory=self.db_path)
+        # attributes to hold the PhreeqPython solution.
+        self.ppsol = None
+        # store the solution composition to see whether we need to re-instantiate the solution
+        self._stored_comp = None
+
+    def _setup_ppsol(self, solution: "solution.Solution") -> None:
+        """Helper method to set up a PhreeqPython solution for subsequent analysis."""
+
+        from pyEQL.phreeqc import PHRQSol  # noqa: PLC0415
+
+        self._stored_comp = solution.components.copy()
+        solv_mass = solution.solvent_mass.to("kg").magnitude
+        # inherit bulk solution properties
+        d = {
+            "temp": solution.temperature.to("degC").magnitude,
+            "units": "mol/kgw",  # to avoid confusion about volume, use mol/kgw which seems more robust in PHREEQC
+            "pH": solution.pH,
+            "pe": solution.pE,
+            "redox": "pe",  # hard-coded to use the pe
+            # PHREEQC will assume 1 kg if not specified, there is also no direct way to specify volume, so we
+            # really have to specify the solvent mass in 1 liter of solution
+            "water": solv_mass,
+        }
+        if solution.balance_charge == "pH":
+            d["pH"] = str(d["pH"]) + " charge"
+        if solution.balance_charge == "pE":
+            d["pe"] = str(d["pe"]) + " charge"
+
+        # add the composition to the dict
+        # also, skip H and O
+        for el, mol in solution.get_el_amt_dict().items():
+            # CAUTION - care must be taken to avoid unintended behavior here. get_el_amt_dict() will return
+            # all distinct oxi states of each element present. If there are elements present whose oxi states
+            # are NOT recognized by PHREEQC (via SPECIAL_ELEMENTS) then the amount of only 1 oxi state will be
+            # entered into the composition dict. This can especially cause problems after equilibrate() has already
+            # been called once. For example, equilibrating a simple NaCl solution generates Cl species that are assigned
+            # various oxidations states, -1 mostly, but also 1, 2, and 3. Since the concentrations of everything
+            # except the -1 oxi state are tiny, this can result in Cl "disappearing" from the solution if
+            # equlibrate is called again. It also causes non-determinism, because the amount is taken from whatever
+            # oxi state happens to be iterated through last.
+
+            # strip off the oxi state
+            bare_el = el.split("(")[0]
+            if bare_el in SPECIAL_ELEMENTS:
+                # PHREEQC will ignore float-formatted oxi states. Need to make sure we are
+                # passing, e.g. 'C(4)' and not 'C(4.0)'
+                key = f"{bare_el}({int(float(el.split('(')[-1].split(')')[0]))})"
+            elif bare_el in ["H", "O"]:
+                continue
+            else:
+                key = bare_el
+
+            if key in d:
+                # when multiple oxi states for the same (non-SPECIAL) element are present, make sure to
+                # add all their amounts together
+                d[key] += str(mol / solv_mass)
+            else:
+                d[key] = str(mol / solv_mass)
+
+            # tell PHREEQC which species to use for charge balance
+            if solution.balance_charge is not None and solution._cb_species in solution.get_components_by_element()[el]:
+                d[key] += " charge"
+
+        try:
+            ppsol = self.pp.add_solution(PHRQSol(d))
+        except Exception as e:
+            # catch problems with the input to phreeqc
+            raise ValueError(
+                "There is a problem with your input. The error message received from "
+                f" phreeqc is:\n\n {e}\n Check your input arguments, especially "
+                "the composition dictionary, and try again."
+            )
+
+        self.ppsol = ppsol
+
+    def _destroy_ppsol(self) -> None:
+        if self.ppsol is not None:
+            self.pp.remove_solution(0)  # TODO: Are we only expecting a single solution per wrapper?
+            self.ppsol = None
+
+    def get_activity_coefficient(self, solution: "solution.Solution", solute: str) -> ureg.Quantity:
+        """
+        Return the *molal scale* activity coefficient of solute, given a Solution
+        object.
+        """
+        if (self.ppsol is None) or (solution.components != self._stored_comp):
+            self._destroy_ppsol()
+            self._setup_ppsol(solution)
+
+        # translate the species into keys that phreeqc will understand
+        k = standardize_formula(solute)
+        spl = k.split("[")
+        el = spl[0]
+        chg = spl[1].split("]")[0]
+        if chg[-1] == "1":
+            chg = chg[0]  # just pass + or -, not +1 / -1
+        k = el + chg
+
+        # calculate the molal scale activity coefficient
+        # act = self.ppsol.activity(k, "mol") / self.ppsol.molality(k, "mol")
+        act = self.ppsol.get_activity(k) / self.ppsol.get_molality(k)
+
+        return ureg.Quantity(act, "dimensionless")
+
+    def get_osmotic_coefficient(self, solution: "solution.Solution") -> ureg.Quantity:
+        osmotic = self.ppsol.get_osmotic_coefficient()
+        return ureg.Quantity(osmotic, "dimensionless")
+
+    def get_solute_volume(self, solution: "solution.Solution") -> ureg.Quantity:
+        """Return the volume of the solutes."""
+        # TODO - phreeqc seems to have no concept of volume, but it does calculate density
+        return ureg.Quantity(0, "L")
+
+    def equilibrate(
+        self,
+        solution: "solution.Solution",
+        atmosphere: bool = False,
+        solids: list[str] | None = None,
+        gases: dict[str, str | float] | None = None,
+    ) -> None:
+        """
+        Adjust the speciation of a Solution object to achieve chemical equilibrium.
+
+        Args:
+            atmosphere:
+                Boolean indicating whether to equilibrate the solution
+                w.r.t atmospheric gases.
+            solids:
+                A list of solids used to achieve liquid-solid equilibrium. Each
+                solid in this list should be present in the Phreeqc database.
+                We assume a target saturation index of 0 and an infinite
+                amount of material.
+            gases:
+                A dictionary of gases used to achieve liquid-gas equilibrium.
+                Each key denotes the gas species, and the corresponding value
+                denotes its concentration, as a log partial pressure value or
+                other interpretable pressure units. For example, the following
+                are equivalent (log10(0.000316) = -3.5)
+                {"CO2": "0.000316 atm"}
+                {"CO2": -3.5}
+        """
+        if self.ppsol is not None:
+            self.ppsol.forget()
+        self._setup_ppsol(solution)
+
+        # store the original solvent mass
+        orig_solvent_moles = solution.components[solution.solvent]
+
+        # store the original solvent elements and components with that element
+        # that we may need to add back later.
+        orig_el_dict = solution.get_el_amt_dict(nested=True)
+        orig_components_by_element = solution.get_components_by_element(nested=True)
+
+        # Use supplied gases, merged with atmospheric gases
+        # if atmosphere == True
+        gases = (ATMOSPHERE if atmosphere else {}) | (gases or {})
+
+        # Mapping from phase name to:
+        #   (<saturation_index>, <amount_in_moles>) tuples (for solids).
+        #   (<log_partial_pressure>, <amount_in_moles>) tuples (for gases).
+        phases = {}
+        if solids is not None:
+            # Assume saturation index of 0 for all solids.
+            phases |= dict.fromkeys(solids, (0, EQUILIBRIUM_PHASE_AMOUNT))
+
+        for k, v in gases.items():
+            v_quantity = ureg.Quantity(v)
+            if v_quantity.dimensionless:
+                log_partial_pressure = v_quantity.magnitude
+            else:
+                log_partial_pressure = math.log10(v_quantity.to("atm").magnitude)
+            phases |= {f"{k}(g)": (log_partial_pressure, EQUILIBRIUM_PHASE_AMOUNT)}
+
+        if phases:
+            phase_names = list(phases.keys())
+            saturation_indices = [v[0] for v in phases.values()]
+            amounts = [v[1] for v in phases.values()]
+
+            try:
+                self.ppsol.equalize(phases=phase_names, saturation_indices=saturation_indices, amounts=amounts)
+            except Exception as e:
+                # Re-raise exception due to unrecognized phases as ValueError.
+                if "Phase not found in database" in str(e):
+                    raise ValueError(str(e))
+                raise e
+
+        solution.components = FormulaDict({})
+        # use the output from PHREEQC to update the Solution composition
+        # the .species_moles attribute should return MOLES (not moles per ___)
+        for s, mol in self.ppsol.species_moles.items():
+            solution.components[s] = mol
+
+        # log a message if any components were not touched by PHREEQC
+        # if that was the case, re-adjust the charge balance to account for those species (since PHREEQC did not)
+        missing_species = set(self._stored_comp.keys()) - {standardize_formula(s) for s in self.ppsol.species}
+        if len(missing_species) > 0:
+            logger.warning(
+                f"After equilibration, the amounts of species {sorted(missing_species)} were not modified "
+                "by PHREEQC. These species are likely absent from its database."
+            )
+
+        # tolerance (in moles) for detecting cases where an element amount
+        # is no longer balanced because of species that are not recognized
+        # by PHREEQC.
+        _atol = 1e-16
+
+        new_el_dict = solution.get_el_amt_dict(nested=True)
+        for el in orig_el_dict:
+            orig_el_amount = sum([orig_el_dict[el][k] for k in orig_el_dict[el]])
+            new_el_amount = sum([new_el_dict[el][k] for k in new_el_dict.get(el, [])])
+
+            # If this element went "missing", add back all components that
+            # contain this element (for any valence value)
+            if orig_el_amount - new_el_amount > _atol:
+                logger.info(
+                    f"PHREEQC discarded element {el} during equilibration. Adding all components for this element."
+                )
+                solution.components.update(
+                    {
+                        component: self._stored_comp[component]
+                        for components in orig_components_by_element[el].values()
+                        for component in components
+                        if component not in solution.components
+                    }
+                )
+
+        # re-adjust charge balance for any missing species
+        # note that if balance_charge is set, it will have been passed to PHREEQC, so the only reason to re-adjust charge balance here is to account for any missing species.
+        solution._adjust_charge_balance()
+
+        # rescale the solvent mass to ensure the total mass of solution does not change
+        # this is important because PHREEQC and the pyEQL database may use slightly different molecular
+        # weights for water. Since water amount is passed to PHREEQC in kg but returned in moles, each
+        # call to equilibrate can thus result in a slight change in the Solution mass.
+        solution.components[solution.solvent] = orig_solvent_moles