SystemPitzer is an electrolyte-GE model. Its interaction coefficients must not be placed in reaction-equilibrium, mineral-solubility, SIT, eNRTL, Extended-UNIQUAC, or electrolyte-EOS tables. Pitzer results may validate another model, but direct coefficient transfer requires a separately derived and reviewed equation mapping.

Current dataset and coverage contract

New SystemPitzer phases use a catalog-first default: on the first Pitzer activity/property evaluation, NeqSim selects the complete bundled PHREEQC topology when every required active aqueous interaction exists. EOS-role pure hydrocarbons are identified by molecular formula as well as component-type metadata and do not create aqueous neutral-interaction requirements; this includes normal database components such as methane whose GE-phase type is not reliably HC. Other non-hydrocarbon aqueous neutrals still require explicit complete lambda and zeta families. If the catalog is incomplete, loading falls back to the historical binary table; mixed primary-salt coverage then still fails closed on missing binary, theta, or psi rows. No missing interaction is silently converted to zero. Call SystemPitzer.useLegacyPitzerParameters() before the first property evaluation only to reproduce a historical result.

The compatibility table is identified as neqsim-legacy-pitzer-parameters-v1. It stores molality-scale binary coefficients at 298.15 K and a NeqSim three-term temperature expression:

p(T) = p(298.15 K) + pT1 (1/T - 1/298.15 K) + pT2 ln(T/298.15 K).

This identity documents the existing table; it does not certify every legacy row. The table has no usable tuple identity for same-sign theta or ternary psi, and it has no qualified lambda, zeta, mu, or eta rows. Mixed-ion and neutral-solute calculations therefore require explicit parameter definitions.

PhasePitzer.getPitzerParameterCoverage() returns active primary-salt cations and anions plus every missing binary, theta, and psi key. For a mixed primary-salt topology, the automatic gate fails before activity or water-osmotic evaluation when that coverage is incomplete. Established single-cation/single-anion calculations retain their binary behavior; callers can explicitly audit their binary row with requireCompletePitzerParameterCoverage(). An explicit zero set through setBinaryParameters, setTheta, or setPsi counts as a definition; an absent value does not. The neutral families are sparse and opt-in. Once any neutral interaction is configured, every active neutral-neutral (including repeated species) and neutral-ion lambda pair and every neutral-cation-anion zeta tuple must be explicit. If a dataset enables mu or eta, its active neutral or same-sign-ion topology is audited too. Missing rows fail before activity or osmotic evaluation; an explicitly defined zero remains distinct from an absent row. Acid-base species created as reaction-solver trial variables (H3O+, OH-, HCO3-, and CO3--) are outside this first automatic gate because their transient iteration amounts are not a stable input topology. After a reactive calculation converges, call getPitzerReactionParameterCoverage() to audit every active ion, or requireCompletePitzerReactionParameterCoverage() to fail closed. The complete reactive diagnostic is calculated on demand, so it adds no work to ordinary Pitzer properties or non-electrolyte models. On the current carbonate reproducer, the legacy dataset reports the active H3O+|HCO3- binary as missing; the bundled PHREEQC 3.9.0 catalog also has no explicit H+|HCO3- binary row. No zero or name-based proton convention is inferred. A qualified source row and an independently reviewed H3O+/PHREEQC-H+ standard-state mapping remain adoption dependencies. The primary diagnostic is cached by active primary-salt topology and parameter revision, and the automatic gate runs once after each standard level-zero state initialization. Call init(0) after changing composition, as required by the general NeqSim state contract. This work is confined to PhasePitzer and ComponentGePitzer; neutral PR, SRK, CPA, Electrolyte-CPA, and Fürst electrolyte-EOS paths do no new work.

Complete-dataset qualification publication gate

Interaction coverage and scientific qualification remain independent contracts. A complete topology may still use a dataset whose full species/range matrix lacks held-out evidence. SystemPitzer.getPitzerParameterQualification() completes lazy dataset selection and returns the immutable metadata for the exact selected dataset. The opt-in requireCompletePitzerDatasetQualification() first requires complete active-ion coverage and then rejects any named dataset below VALIDATED_WITHIN_DECLARED_ENVELOPE. Because an overall level cannot distinguish activity, water-property, reaction, mineral, and VLE evidence, property publication should use requirePitzerDatasetValidationFor(ValidationTarget) instead. The target gate requires complete topology and fails unless that exact observable is independently qualified.

This gate is intentionally stricter than subsystem validation. The broad PHREEQC catalog remains PARTIALLY_EXPERIMENTALLY_VALIDATED, even when the current topology is one of its independently checked binaries; callers that need a complete named-dataset gate must select a separately qualified subset such as Na-K-Cl. A successful dataset gate does not prove that the current temperature, pressure, molality, or composition lies inside its evidence envelope. The applicable isWithin...ValidationRange helper remains a separate mandatory state check. Diagnostics include dataset identity, level, validated systems, machine-readable validation targets, and limitations in deterministic order. Existing callers of the complete-dataset gate retain source compatibility, but must not infer a property target from the overall level.

The accessor and gate are explicit setup/publication operations. They perform no flash, change no parameter, and add no work to Pitzer activity/property kernels or neutral PR/SRK/CPA calculations.

Pitzer reaction-data validation boundary

Reaction-equilibrium constants and Pitzer ion-interaction coefficients remain separate datasets. The PITZER reaction source now treats VALIDATED as an initialization requirement for every active row that is relevant to the selected feed. The retained MDEAprot (Huttenhuis2005) and DEAprot (Austgen1989) rows are both active compatibility rows but are explicitly UNVALIDATED; their activation flags therefore do not authorize use by SystemPitzer. Initialization rejects either row before it can add MDEA+ or DEA+. Validated carbonate, water-dissociation, and first H2S-dissociation rows are unchanged, as are the STANDARD and KENT_EISENBERG sources.

No equilibrium coefficient, Pitzer interaction coefficient, table row, or thesis value is copied, changed, or adopted by this gate. Kaasa (1998), Appendix F, printed pp. 260–267 remains a provenance index only; its table-redistribution terms and row-level independent qualification remain unresolved. Enabling Pitzer amine chemistry requires an independently sourced and redistributable molality-standard-state amine equilibrium correlation, the complete applicable binary, same-sign, ternary, and neutral-ion interaction family, an explicit species/standard-state mapping, and held-out speciation plus gas-liquid-equilibrium validation.

Qualified public-domain CO2-Na2SO4 subset

PitzerParameterDatasets.applyPhreeqcCo2SodiumSulfate installs one coherent subset identified as usgs-phreeqc-pitzer-b0b3be767158ccc3322d2c816625cf470045e67e-co2-na2so4-v1. It is copied exactly from public-domain PHREEQC commit b0b3be7, database blob 324f852, and it marks the phase as manually populated so the legacy table cannot be mixed into the selected set. The species use the molality standard state, pressure is not an argument of these interaction functions, the reference temperature is 298.15 K, and the six coefficients are stored in PHREEQC order. The Na+/SO4– 1-2 binary uses alpha1=2 and has no beta2 term. Java and Python callers can use SystemPitzer.applyPhreeqcCo2SodiumSulfateParameters() after adding the required species; cloning and subsequent init(0) calls preserve the selected dataset.

Species tuple Family PHREEQC coefficients [a0,a1,a2,a3,a4,a5] Source-lineage note
Na+ / SO4– beta(0) [2.73e-2, 0, -5.8, 9.89e-3, 0, -1.563e5] PHREEQC pitzer.dat; exact current public-domain row
Na+ / SO4– beta(1) [0.956, 2.663e3, 0, 1.158e-2, 0, -3.194e5] same coherent binary row
Na+ / SO4– Cphi (C0) [3.418e-3, -384, 0, -8.451e-4, 0, 5.177e4] unchanged C0 value; downstream charge normalization is in the equation
CO2 / CO2 lambda [-1.34e-2, 348, 0.803, 0, 0, 0] PHREEQC comment limits this temperature function to 150 °C
CO2 / Na+ lambda 0.085 constant row in the same database
CO2 / SO4– lambda 0.075 constant row in the same database
CO2 / Na+ / SO4– zeta -0.015 required ternary companion row

This is a deliberately closed species system, not a general brine database. Any additional active neutral or ion forces the ordinary binary/mixed-ion and neutral-family coverage checks; a missing interaction is never silently replaced by zero. isWithinCo2SodiumSulfateValidationRange records the independently checked envelope of 303.15–423.15 K and 1–2 mol/kg Na2SO4. It is an evidence envelope for this subset, not an extrapolation guarantee.

Independent calculation evidence uses IPhreeqc 3.7.3 with the exact database blob above and MacInnes scaling disabled. Four CO2 states at 298.15, 373.15, and 423.15 K and 0.1–2 mol/kg Na2SO4 match PHREEQC natural-log activity contributions within 1.5e-5. At 298.15 K and approximately 1 mol/kg Na2SO4, NeqSim gives mean ionic activity coefficient 0.20016248 versus PHREEQC 0.20137297 (−0.60%) and osmotic coefficient 0.64159881 versus 0.64224702 (−0.10%). The remaining difference is retained as an explicit tolerance because NeqSim’s water electrostatic correlation is not PHREEQC’s optional database Aphi function.

The held-out qualitative gate is Dos Santos et al. (2020), DOI 10.1021/acs.jced.0c00230: 48 CO2-solubility points at 303.15–423.15 K and 1–2 mol/kg Na2SO4 report increased salting-out with salt molality. The qualified subset independently produces a larger CO2 activity coefficient from 1 to 2 mol/kg at 373.15 K. The publication is used only as cited validation evidence; no copyrighted table is stored. A second outside-range comparison is Zhao et al. (2015), DOI 10.1002/aic.14825, covering Na2SO4 ionic strength 1–6 mol/kg at 15 MPa and 323–423 K; it is evidence for later full-VLE work, not grounds to widen this subset’s accepted range.

The hybrid EOS-GE phase-boundary calculation now converges for the selected gas-forming CO2/brine states, so VLE accuracy can be assessed independently of flash closure. Six held-out bubble pressures from Bermejo et al. (2005), DOI 10.1016/j.fluid.2005.10.006, were selected from NIST ThermoML dataset 2. They cover 307.78–340.10 K, 0.986703–0.996313 mol/kg Na2SO4, liquid CO2 mole fractions 0.00773–0.01161, and 43.5–118.2 bara. The archived 95% expanded pressure uncertainties are 0.5–2.0 bara. The publisher-permitted ThermoML JSON has SHA-256 aeeff43c1aa196a749b01dbb31d22bf502fe140a85e4823ecea8544f06dd3897.

Pressure was converted from kPa to bara. On a one-kilogram-water basis, molecular salt molality was mapped to 2 Na+ + SO4--, and the reported liquid CO2 mole fraction was converted with nCO2=x/(1-x)*(nH2O+nNa2SO4). No parameter was fitted. Exact-master calculations predict 29.306, 43.165, 53.064, 74.040, 45.715, and 58.930 bara: underprediction of 32.6–43.8%, outside each experimental uncertainty. Treating the reported fraction with an explicit-ion denominator reduces the central-point residual only to 31.5%, so the composition-basis sensitivity does not change the rejection. Material-balance residuals are at most order 1e-16, logarithmic fugacity residuals are below 1e-10, and aqueous charge is zero. The fixture is therefore held-out scientific evidence, not a flash convergence failure or a training-data regression.

The subset remains qualified for aqueous activity and water/osmotic targets only. GAS_AQUEOUS_VLE fails closed through the observable-specific gate. No Pitzer coefficient, Henry correlation, Poynting correction, or reaction row is changed to fit these six points.

Qualified public-domain Na-K-Cl subset

PitzerParameterDatasets.applyPhreeqcSodiumPotassiumChloride installs the coherent subset usgs-phreeqc-pitzer-b0b3be767158ccc3322d2c816625cf470045e67e-na-k-cl-v1. The selected rows were read twice and checked against public-domain PHREEQC commit b0b3be7, pitzer.dat blob 324f852. The subset keeps the molality standard state and PHREEQC six-term temperature expression at 298.15 K. Both binaries use alpha1=2 and have no beta(2) term. It is separate from, and does not overwrite, neqsim-legacy-pitzer-parameters-v1.

Species tuple Family PHREEQC coefficients [a0,a1,a2,a3,a4,a5] Source-lineage note
Na+ / Cl- beta(0) [0.07534, 9598.4, 35.48, -0.058731, 1.798e-5, -5.0e5] exact public-domain pitzer.dat row
Na+ / Cl- beta(1) [0.2769, 13770, 46.8, -0.069512, 2.0e-5, -748230] same coherent binary row
Na+ / Cl- Cphi (C0) [0.00148, -120.5, -0.2081, 0, 1.166e-7, 11121] raw C0; charge normalization remains in the equation
K+ / Cl- beta(0) [0.04808, -758.48, -4.7062, 0.010072, -3.7599e-6, 0] exact public-domain pitzer.dat row
K+ / Cl- beta(1) [0.2168, 0, -6.895, 0.02262, -9.293e-6, -1.0e5] same coherent binary row
K+ / Cl- Cphi (C0) [-0.000788, 91.27, 0.58643, -0.001298, 4.9567e-7, 0] raw C0; charge normalization remains in the equation
K+ / Na+ theta [-0.012, 0, 0, 0, 0, 0] required same-sign companion
Cl- / K+ / Na+ psi [-0.0015, 0, 0, 1.8e-5, 0, 0] required ternary companion

The PHREEQC source evaluation also requires two common single-ion terms. The binary derivative contribution to F is summed over every active cation-anion pair, not only the pair containing the target ion. The normalized C0 sum is likewise global and contributes |z_i| sum_c sum_a m_c m_a C0_ca/(2 sqrt(|z_c z_a|)) to every ion. These terms are enabled only by an explicitly mapped PHREEQC dataset. The legacy Pitzer path retains its historical result, and all non-Pitzer models bypass this code.

Independent evidence uses IPhreeqc 3.7.3 with MacInnes scaling disabled and the same Na/K/Cl rows. Eight charge-balanced states cover pure 1 mol/kg NaCl, pure 1 mol/kg KCl, mixed 0.05/0.05/0.1, 0.5/0.5/1.0, 0.2/0.8/1.0, and 1.5/1.5/3.0 mol/kg Na/K/Cl, plus 0.5/0.5/1.0 at 373.15 and 423.15 K. At 298.15 K the maximum absolute natural-log single-ion residual is below 0.002; it is below 0.008 at 373.15 K and 0.019 at 423.15 K. Across the matrix, water-activity and osmotic-coefficient residuals are below 0.0007 and 0.007, respectively. The growing high-temperature difference is retained rather than fitted away: PHREEQC may use its database Aphi function, while NeqSim retains its existing water dielectric/density correlation. The checked implementation-comparison envelope is 298.15–423.15 K and 0.1–3.0 mol/kg chloride; it is not a pressure or universal experimental validity claim. Composition sweeps separately check continuity, repeated execution, normalization, non-negative phase state, and electroneutrality. Additional active species fail closed until all required binary and mixed-ion companions are qualified.

An independent experimental-data synthesis checks the NaCl binary without reusing PHREEQC output. Partanen and Partanen (2020), DOI 10.1021/acs.jced.0c00402, Tables 6 and 10, recommend mean activity and osmotic coefficients from a traceable extended-Huckel analysis of vapor-pressure, electrochemical, cryoscopic, and solubility evidence. The article and tables are licensed CC BY 4.0 and postdate the adopted PHREEQC parameter lineage. At 298.15 K and 0.2, 0.5, 1.0, and 2.0 mol/kg NaCl, NeqSim’s maximum relative mean-activity residual is 1.36%, the maximum absolute osmotic-coefficient residual is 0.0053, and the maximum water-activity residual derived from the reported osmotic coefficient is below 0.0004. The published recommendations are rounded to 0.001; acceptance limits are 1.5%, 0.006, and 0.0005, respectively. No coefficient was refitted.

The KCl binary has a separate held-out check against the critically evaluated 298.15 K values of Hamer and Wu (1972), DOI 10.1063/1.3253108, prepared at the U.S. National Bureau of Standards and published by the U.S. Secretary of Commerce. Four values at 0.0982, 0.5001, 0.9926, and 1.9895 mol/kg are independently reproduced in Table 2 of the open-access primary experiment by Dash et al. (2012), DOI 10.5402/2012/730154, under its CC BY terms. NeqSim’s maximum relative mean-activity residual is 0.110%; the acceptance limit is 0.20%, allowing for the source’s 0.001 rounding without refitting. Dash et al.’s own single-ion-selective-electrode means are 10.7–13.4% higher at these four states. They are retained as a conflicting method result, not used to refit or relax the critically evaluated thermodynamic gate. Together with the independent IPhreeqc water/osmotic checks, this qualifies both binary boundaries of the Na/K/Cl family.

Independent mixed NaCl+KCl measurements exist in Rodil et al. (2009), DOI 10.1021/je800389q, with a mandatory correction at DOI 10.1021/je9002674. The ACS supporting information is publicly downloadable, but the article and correction are copyright ACS and no redistribution license was established in this audit. No table value is stored. The exact-composition IPhreeqc matrix therefore remains the mixed-family numerical gate; licensing-clear held-out mixed-salt experimental data are still the next evidence dependency.

Versioned PHREEQC catalog and Ca-Mg-Cl-SO4 scale-brine qualification

PhreeqcPitzerParameterCatalog lazily reads the exact PITZER block from public-domain USGS PHREEQC 3.9.0-17591 (released 2026-05-13), commit b0b3be767158ccc3322d2c816625cf470045e67e, database blob 324f852784be84650b77bd7f07f8316aafd8188b. The 3.9.0 release notes identify revisions to Mg++, Ba++, CO3–, HCO3-, CO2, and related bicarbonate and neutral-hydrogen Pitzer interactions. The bundled catalog contains 54 B0, 48 B1, 8 B2, 32 C0, 30 theta, 59 psi, 27 lambda, and 10 zeta rows. No reaction constant, mineral log K, gas binary, SIT, eNRTL, Extended-UNIQUAC, or electrolyte-EOS coefficient is imported. The resource is parsed lazily on the first catalog-eligible Pitzer evaluation and cached as one immutable catalog. Neutral EOS paths do no catalog I/O or lookup work; a Pitzer phase performs automatic topology selection only once when its parameter dataset is first loaded.

The default selector and explicit applyCompletePhreeqcPitzerCatalog API select rows by the active aqueous species. Every active opposite-sign pair requires explicit B0, B1, and C0; every same-sign pair requires theta; every mixed-ion triple requires psi; and active neutral solutes require their complete lambda and zeta topology. An absent required row fails before phase mutation. A B2 row is optional only in the strict sense that it is applied whenever the source defines it; the loader never creates one or silently substitutes an unqualified zero. This makes a broad source database available without a manual selection call while preserving a closed, reviewable calculation topology. Explicit subset APIs remain available for audited reproduction and validation work.

The first fully exercised four-ion family is Ca++/Mg++/Cl-/SO4–. It includes all four binaries, both same-sign pairs, and all four ternary compositions:

Tuple Family PHREEQC coefficients [a0,a1,a2,a3,a4,a5]
Ca++ / Cl- B0; B1; B2; C0 [0.3159,0,0,-3.27e-4,1.4e-7,0]; [1.614,0,0,7.63e-3,-8.19e-7,0]; [-1.13,0,0,-0.0476,0,0]; [1.4e-4,-57,-0.098,-7.83e-4,7.18e-7,0]
Ca++ / SO4– B0; B1; B2; C0 [0,0,0,0,0,0]; [3.546,0,0,5.77e-3,0,0]; [-59.3,0,0,-0.443,-3.96e-6,0]; [0.114,0,0,0,0,0]
Mg++ / Cl- B0; B1; C0 [0.351,0,0,-9.32e-4,5.94e-7,0]; [1.65,0,0,-1.09e-2,2.6e-5,0]; [0.00651,0,0,-2.5e-4,2.418e-7,0]
Mg++ / SO4– B0; B1; B2; C0 [0.2135,-951,0,-2.34e-2,2.28e-5,0]; [3.367,-5.78e3,0,-0.148,1.576e-4,0]; [-32.45,0,-3.236e3,21.812,-1.8859e-2,0]; [2.875e-2,0,-2.084,1.1428e-2,-8.228e-6,0]
Ca++ / Mg++; Cl- / SO4– theta [0.007,0,0,0,0,0]; [0.03,0,0,0,0,0]
Ca++ / Cl- / Mg++; Ca++ / Cl- / SO4– psi [-0.012,0,0,0,0,0]; [-0.122,0,0,-1.21e-3,0,0]
Ca++ / Mg++ / SO4–; Cl- / Mg++ / SO4– psi [0.024,0,0,0,0,0]; [-0.008,32.63,0,0,0,0]

PHREEQC source assigns alpha1=2 and alpha2=12 to pairs containing a monovalent ion, alpha1=1.4 and alpha2=12 to 2-2 pairs, and alpha1=2, alpha2=50 to the remaining charge orders. Current-master NeqSim previously discarded B2 for every non-2-2 pair, which omitted the explicit CaCl2 term. The catalog path now maps this PHREEQC convention in activity, common B-prime, and both water/osmotic calculations; legacy rows with zero B2 are unchanged.

Held-out CaCl2 mean ionic activity coefficients come from Partanen (2013), DOI 10.1021/je300852v, through the NIST ThermoML archive. The archived states are on the molality scale at 298.15 K and 101 kPa with reported uncertainty 0.001. At 0.1, 0.5, 1.0, and 2.0 mol/kg, NeqSim predicts 0.515385, 0.444320, 0.497186, and 0.798718 versus 0.517, 0.449, 0.502, and 0.790; the maximum relative residual is 1.104%, without parameter fitting. NIST identifies the ThermoML archive as publisher-permitted and its Data.gov catalog applies the NIST public-data license. Exact-version mixed-family PHREEQC activity/water comparisons remain a validation dependency.

Held-out MgCl2 mean ionic activity coefficients come from Partanen (2013), DOI 10.1016/j.jct.2013.06.016, through the same publisher-permitted NIST ThermoML archive. At 0.1667, 0.3333, 1.0, and 2.0 mol/kg and 298.15 K, NeqSim predicts 0.493757, 0.469707, 0.564503, and 1.051132 versus 0.5018, 0.4796, 0.5776, and 1.0695. The maximum relative residual is 2.27%, without parameter fitting. The binary CaCl2 and MgCl2 evidence is acceptable for screening inside this narrow envelope, but it does not qualify the four-ion mixture.

Independent mixed CaCl2+MgCl2 observable evidence comes from Robinson and Bower (1966), DOI 10.6028/jres.070A.026, Table 2. The National Bureau of Standards article is a U.S. Government work and is public domain. Its isopiestic correlation is

R = M_B / m_total = 1 - a y_Mg - b y_Mg^2,

where M_B and M_B phi_B describe the CaCl2 reference solution and y_Mg is the MgCl2 fraction of total formula-unit molality. The stored regression states are transparently derived as m_total=M_B/R, m_MgCl2=y_Mg m_total, m_CaCl2=(1-y_Mg)m_total, and phi=(M_B phi_B)R/M_B. Water activity follows ln(a_w)=-0.01801528*3*m_total*phi; both salts dissociate into three ions per formula unit. Nine points cover the three tabulated isopiestic levels and MgCl2 fractions 0.25, 0.50, and 0.75, for total salt 0.548-2.161 mol/kg water at 298.15 K. The source reports no pointwise uncertainty for Table 2, so the independent regression uses explicit engineering gates of 0.04 maximum absolute osmotic-coefficient residual and 0.004 maximum water-activity residual. It also checks the published preprocessing equation, formula balance, electroneutrality, normalized/non-negative phase state, water-activity/osmotic identity, repeated evaluation, and changed-state freshness. No parameter is fitted or adopted. This validates mixed chloride observables only; it does not qualify sulfate- bearing quaternary mixtures or mineral standard states.

The independent MgCl2+MgSO4 potentiometric study at 298.15 K, DOI 10.1016/j.jct.2011.08.020, reports theta(Cl,SO4)=0.0252 +/- 0.0042 and psi(Mg,Cl,SO4)=-0.0049 +/- 0.0003. The PHREEQC catalog values are 0.03 and -0.008, respectively. Theta agrees within about 1.14 reported standard uncertainties, while psi differs by about 10.3. These are fitted parameters from a different experimental/model lineage, not direct transferable validation values. NeqSim keeps the coherent PHREEQC family unchanged and records the disagreement; it does not mix the two parameter sets.

Independent observable validation now uses the 28 mixed MgCl2+MgSO4 water activities of Guendouzi and Errougui (2007), DOI 10.1021/je7002176. The hygrometric measurements are at 298.15 K and 101 kPa for MgCl2 ionic-strength fractions 0.20, 0.50, and 0.80, with total formula-unit molality from 0.35 to 3.80 mol/kg water. The NIST ThermoML record specifies the pure-water standard state and 95% combined expanded water-activity uncertainties of 0.001 to 0.004. The checked JSON has MD5 29969a5f814eb0ad6d7ca5b4f093b658; the unchanged numerical subset is stored with NIST acknowledgment under the NIST public-data license. Composition is mapped without speciation or fitting as Mg++ = m(MgCl2)+m(MgSO4), Cl- = 2m(MgCl2), and SO4– = m(MgSO4).

Across all 28 states, NeqSim’s maximum absolute water-activity residual is 0.002985 and its RMSE is 0.001275; 26 states fall inside their individual reported 95% expanded uncertainty. Both calculated and experimental water activity decrease monotonically with total molality on each composition line. The regression also checks formula-unit material balance, electroneutrality, normalized and non-negative phase state, water-activity/osmotic-coefficient identity, deterministic repetition, changed-state equivalence, and complete parameter coverage. No coefficient was fitted or changed. This qualifies only mixed Mg-Cl-SO4 water activity at the three documented composition lines and 298.15 K; it does not qualify Ca-bearing mixtures, pressure/temperature extrapolation, individual ion activities, mineral saturation, or precipitation.

PitzerParameterDatasets.getQualification(datasetId) exposes this distinction to Java and Python callers. Complete PHREEQC catalog topology reports PARTIALLY_EXPERIMENTALLY_VALIDATED, lists the CaCl2 and MgCl2 binaries, mixed CaCl2-MgCl2 osmotic/water properties, and mixed MgCl2-MgSO4 water activity as checked, and explicitly leaves quaternary Ca-Mg-Cl-SO4 activity and sulfate-mineral thermodynamics unresolved. Separate range diagnostics expose the mixed-chloride envelope and only the three experimentally checked MgCl2-MgSO4 ionic-strength-fraction lines with their line-specific molality intervals. Unknown or private dataset identities report UNQUALIFIED. This metadata lookup is outside the activity/property kernels and adds no non-electrolyte overhead.

The catalog is an aqueous-GE parameter source, not a complete scale or process model by itself. SystemPitzer keeps SRK gas and oil role phases and exposes catalog application on its Pitzer aqueous role. Hybrid gas-oil-water phase stability, model-specific aqueous reactions, mineral solubility/precipitation complementarity, and process mass/energy/property closure retain their separate gates and database semantics.

Default BaCl2 and SrCl2 binary qualification

The catalog-first default also activates the complete PHREEQC binary rows for charge-balanced Ba++/Cl- and Sr++/Cl- aqueous topologies; users do not select a dataset. At 298.15 K the audited six-term rows are:

Species tuple Family PHREEQC coefficients [a0,a1,a2,a3,a4,a5] Qualification
Ba++ / Cl- B0; B1; C0 [0.5268,0,0,0,0,4.75e4]; [0.687,0,0,0,0,0]; [-0.143,-114.5,0,0,0,0] Source-mapped only; the high-concentration Partanen (2014) table is not stored because a redistribution route was not established.
Sr++ / Cl- B0; B1; C0 [0.2858,0,0,7.17e-4,0,0]; [1.667,0,0,2.8425e-3,0,0]; [-0.0013,0,0,0,0,0] Independently validated against all 58 publisher-permitted NIST ThermoML values described below.

The SrCl2 evidence is Partanen (2013), DOI 10.1021/je400472v, NIST ThermoML archive MD5 0f26a7668fdc333b8a9d6f35223d3fbd. It contains 22 mean ionic activity coefficients at 298.15 K and 0.01-3.52 mol/kg plus 36 temperature-dependent values at 283.15-333.15 K and 0.01-0.30 mol/kg, all at 101 kPa with reported 95% expanded uncertainty 0.001. NeqSim uses every point without refitting: maximum relative mean-activity residual is 3.11%, relative RMSE 1.58%, and mean absolute relative residual 1.35%. The 3.2% engineering gate is deliberately separate from the source’s point uncertainty because it tests the independently fitted PHREEQC family and NeqSim’s unchanged water electrostatic correlation rather than reproducing the source Hückel equation.

Partanen (2014), DOI 10.1016/j.jct.2014.02.023, reports traceable BaCl2 activity and osmotic equations from 273 to 333 K and to 1.7884 mol/kg at 298.15 K. The article confirms the relevant observable and range, but no publisher-permitted numerical archive or explicit table-reuse license was found in this audit. BaCl2 therefore remains source-mapped rather than experimentally qualified; no value is inferred or copied. Hamer’s public NBS NSRDS 24 report, DOI 10.6028/NBS.NSRDS.24, is retained as low-ionic-strength theoretical context, not substituted for the missing high-concentration experiment.

A mixed Ba++/Sr++/Cl- topology requires a same-sign theta(Ba,Sr) row and the corresponding psi(Ba,Sr,Cl) companion. The bundled PHREEQC release does not provide that complete family, so NeqSim fails closed rather than treating either coefficient as zero. Sulfate-bearing Ba/Sr scale chemistry remains a separate boundary because its binary C0 and mixed-brine companions are also incomplete.

Mineral-solubility and precipitation boundary

ThermodynamicOperations.precipitateScale now couples the selected aqueous activity model to the existing COMPSALT solubility-product correlations for one named pure mineral. At 298.15 K and 1.01325 bara, the unmodified COMPSALT correlations give log10 Ksp values -9.934923 for BaSO4, -6.631329 for SrSO4, and -4.355596 for anhydrite (CaSO4_A). The exact public-domain PHREEQC 3.9.0-17591 database at commit b0b3be767158ccc3322d2c816625cf470045e67e gives -9.97, -6.63, and -4.362, respectively. These are independent mineral-constant comparisons; no PHREEQC mineral constant was fitted or copied into COMPSALT.

The complete PHREEQC Ca-Mg-Cl-SO4 interaction family is therefore used for activity-consistent anhydrite precipitation validation. The same catalog contains no explicit Ba++/SO4– B0 row, so a fully qualified barite Pitzer precipitation calculation still fails closed. The close barite Ksp agreement does not qualify its missing aqueous interaction family. A redistributable, coherent Ba/Sr sulfate family with binary and all required mixed-brine terms plus held-out solubility or saturation evidence is the next parameter dependency.

ThermodynamicOperations.precipitateScales adds a model-neutral orchestration layer for competing pure COMPSALT minerals. It does not merge parameter semantics: SystemPitzer continues to evaluate molality-scale activities from the explicitly selected PHREEQC family, while electrolyte EOS systems continue to use their own aqueous fugacity/activity implementation. No Pitzer interaction is copied into an EOS or reaction table.

The active set owns an external, non-negative pure-solid ledger, precipitates supersaturated minerals, and redissolves undersaturated present minerals until the maximum complementarity violation is at most 1e-6 log10-SR. It fails closed on non-convergence and on a dissolved-plus-solid component residual above 1e-10 mol. A continuation API accepts the previous ledger after a T/P or composition change, preventing a stale solid state in process sequences.

The algorithmic regression uses a charge-balanced Na/Ca/Mg/Cl/SO4 brine with nonzero amounts of all four Ca/Mg/Cl/SO4 ions and the CaSO4_A and CaSO4_G correlations. Their unmodified COMPSALT correlations at 298.15 K give log10 Ksp values -4.355596 and -4.578529, respectively. Because both consume one Ca++ and one SO4–, the lower-Ksp correlation must be present and the higher-Ksp phase absent, independent of input order. The regression closes the Ca/SO4 dissolved-plus-solid ledger and proves that Mg/Cl spectators are unchanged, then carries the same four-ion fluid through the gas-oil-aqueous process-property path. This is an internal complementarity/topology regression, separate from the PHREEQC anhydrite Ksp comparison and from fitted-data validation. The CaSO4_G COMPSALT row does not explicitly represent crystallization-water activity or mass; therefore the regression does not qualify gypsum hydration thermodynamics, and that standard-state extension remains a distinct parameter/model boundary.

PitzerCatalogPerformanceBenchmark measures nine fixed-work batches after warmup. On OpenJDK 17 in the development container, the median four-ion activity/osmotic kernel was 10.542 microseconds and the complete aqueous init(3) plus physical-property calculation was 0.168756 milliseconds. The same neutral SRK property control measured 31.940 microseconds before catalog loading and 22.576 microseconds after loading (ratio 0.707, reflecting additional JIT warmup rather than a regression). No neutral EOS production class is changed, and catalog parsing is absent until its explicit API is called. These figures are diagnostic wall-clock evidence, not portable hardware guarantees.

The H2S reaction increment extended the same benchmark with nine ten-calculation median batches. Against exact base a85d255, the complete aqueous reaction initialization and equilibrium solve decreased from 26.824 ms to 22.645 ms (15.6% faster), while the equilibrium-constant kernel decreased from 51 ns to 43 ns. The changed solver branch is reached only by reactive systems that explicitly select SOLUTE_MOLALITY; nonreactive Pitzer properties and neutral SRK/PR/CPA calculations do not call it. At 298.15 K the benchmark recorded maximum absolute ln(Q/K) 3.55e-14, maximum element residual 8.39e-14 mol, charge -1.00e-10 mol, and normalized charge residual 1.47e-6. HS- molality increased physically from 3.3995e-5 at 298.15 K to 4.3531e-5 at 318.15 K.

Source-comparison matrix

The exact PHREEQC Pitzer block is now stored as a versioned catalog. A calculation may activate only an explicit, complete topology; the CO2-Na2SO4, Na-K-Cl, Ca-Mg-Cl-SO4, and SrCl2 families above have dedicated evidence. BaCl2 is source-mapped and automatically available as a complete binary, while other catalog rows remain source-available rather than scientifically qualified until their complete topology and validation matrix are documented.

Source / version Species and parameter families Conditions, equation, scale, and alpha Evidence, range, uncertainty Lineage and reuse status Mapping decision / conflict
NeqSim legacy CSV, audited 2026-08-27 30 cation-anion rows; beta0, beta1, beta2, Cphi; columns named theta and psi_common_ion contain only zero and no tuple identity. The converged carbonate reproducer activates H3O+ and HCO3-, but the CSV has no row for that pair. 298.15 K reference; CSV Tmin/Tmax; NeqSim three-term T expression; molality activity path; hard-coded alpha 2.0 for pairs containing a univalent ion and 1.4/12.0 for 2-2 References vary from publications to “Estimated”; per-row residuals and uncertainty absent Repository data under NeqSim’s license; legacy lineage is not a coherent qualified release Preserve as dataset v1. Binary 298.15 K NaCl is numerically consistent with the PHREEQC example, but temperature and electrostatic mappings are not equivalent. The new reactive audit reports H3O+|HCO3- as missing; it does not infer zero.
Pitzer (1973), general equations, and Pitzer–Mayorga (1973), strong electrolytes Foundational binary virial families for strong electrolytes; individual tuples must be read from the original article Molality-scale Pitzer formulation near 298.15 K; exact parameter-specific alpha and fitted range must be taken from the paper Mean activity and osmotic-coefficient fits; no value transcribed in this review Primary publications; ACS copyright, so values require row-level reuse review or redistributable corroboration Equation lineage accepted; no direct adoption pending exact NeqSim term-by-term mapping.
Harvie, Møller and Weare (1984), natural-water model Na/K/Mg/Ca/H with Cl/SO4/OH/HCO3/CO3/CO2/H2O; binary and mixed interactions 25 °C, high ionic strength; molality-scale Pitzer model Fitted isopiestic, EMF, and solubility data; multicomponent comparisons outside subsystems are reported Primary Elsevier article; numerical-table redistribution not established here High-priority scientific comparison. No coefficient copied; full family and electrostatic convention must be mapped together.
Pitzer (1975), higher-order electrostatic mixing Nonsymmetric same-sign electrostatic terms Etheta and its ionic-strength derivative for unequal charge pairs; no fitted short-range coefficient is supplied by this term Molality-scale Pitzer formulation; the term is zero for equal charges and depends on charge tuple, ionic strength, and Aphi Primary equation source; it establishes model structure rather than a parameter fit Primary Springer article; no numerical table copied Equation structure accepted in this increment: the public PHREEQC recurrence and its ion/activity/osmotic placement are mapped directly. No fitted coefficient is copied; short-range tuple coverage remains independently fail-closed.
USGS PHRQPITZ (Plummer et al. 1988), WRIR 88-4153, PHREEQC 3.8.6 tag commit 74cdaf0, and PHREEQC 3.9.0-17591 commit b0b3be7 Na/K/Mg/Ca/H plus major anions and extended Fe/Mn/Sr/Ba/Li/Br; B0, B1, B2, C0, theta, psi, lambda, zeta, and six temperature coefficients. Release 3.9.0 revises Mg/Ba/carbonate/bicarbonate/CO2 species and related HCO3 and neutral-H2 Pitzer interactions. Exact catalog inspection on 2026-08-27 found no H+|HCO3- B0, B1, or C0 row. Molality scale; PHREEQC enables nonsymmetric electrostatic mixing by default. Default alpha is charge dependent: 2/12 when a pair contains a monovalent ion, 1.4/12 for 2-2, and 2/50 otherwise. Pressure is absent from these interaction functions. PHREEQC names the proton H+; NeqSim’s reacting species is H3O+, and no standard-state alias is assumed. Appelo (2015), DOI 10.1016/j.apgeochem.2014.11.007, documents database principles and calculations from 0–200 °C and 1–1000 atm; applicability remains parameter/system specific. Exact fitted observables, residuals, and uncertainty remain row/source specific. USGS software and data are public domain. Audited release date: 2026-05-13. Audited blobs: pitzer.cpp 1f32a08 and pitzer.dat 324f852. The exact interaction block is stored as a lazy catalog and selected automatically for complete active topologies. Missing mixed families fail closed rather than receiving zero or legacy coefficients. CO2-Na2SO4, Na-K-Cl, Ca-Mg-Cl-SO4, and SrCl2 have dedicated mappings/evidence. H3O+|HCO3- remains unqualified; no value is adopted.
Bermejo et al. (2005), DOI 10.1016/j.fluid.2005.10.006, NIST ThermoML dataset 2 CO2 + H2O + Na2SO4 bubble pressure; validation observable only, with no beta, Cphi, theta, psi, lambda, zeta, mu, eta, alpha, reaction log K, or temperature coefficient adopted Reported solvent Na2SO4 molality, liquid CO2 mole fraction, temperature in K, and pressure in kPa; six selected states span 307.78–340.10 K, 0.986703–0.996313 mol/kg, and 43.5–118.2 bara. The molecular-salt mole-fraction mapping and kPa-to-bara conversion are explicit. Source contains 112 bubble-pressure points and 95% expanded uncertainties. Six held-out states have 0.5–2.0 bara uncertainty; exact-master residuals are −32.6% to −43.8% without refitting. The experiment is independent of the PHREEQC parameter lineage. Publisher-permitted NIST ThermoML numerical archive; JSON SHA-256 aeeff43c1aa196a749b01dbb31d22bf502fe140a85e4823ecea8544f06dd3897. Six observations and metadata are stored, not an article table. Reject GAS_AQUEOUS_VLE qualification for the existing CO2-Na2SO4 family. Activity and water-property targets remain qualified. The data cannot be used to mix model families or infer a missing pressure correction.
Xia, Maurer and coworkers (2000), DOI 10.1021/ie990416p, as cited on PHREEQC H2S rows Neutral H2Sg with Na+/Cl-: lambda(Cl-,H2Sg)=-0.005, lambda(H2Sg,Na+)=[0.1047,0,-0.0413], and zeta(H2Sg,Cl-,Na+)=-0.0123; PHREEQC also defines separate (H2Sg)2 rows and a dimerization reaction Molality-scale PHREEQC neutral-ion convention; Xia measurements cover H2S solubility in 4-6 mol/kg NaCl from 313-393 K and total pressures to 10 MPa. Pressure dependence is not encoded in the listed Pitzer functions. Primary H2S solubility observables; public abstract does not provide row-wise parameter uncertainty. The PHREEQC row is redistributable public-domain data, while the ACS article remains copyright. No primary table is copied. The exact PHREEQC values and source comment are recorded for comparison. Rejected for activation in NeqSim: the available family lacks an H2Sg-H2Sg self interaction and instead relies on an explicit (H2Sg)2 species that NeqSim does not model. Partial aliasing would violate fail-closed topology and change the source species model.
Hershey, Pleše and Millero (1988), DOI 10.1016/0016-7037(88)90183-4 First H2S dissociation only; pK1 and HS- interactions with Na+, K+, Mg+2, and Ca+2. It supplies no second-dissociation or neutral-H2S self/dimer family. Molality scale; NaCl from 0.1 mol/kg to saturation at 5, 25, and 45 °C, KCl at 5 and 25 °C, and selected MgCl2/CaCl2 additions to ionic strength 6 mol/kg. Infinite-dilution pK1=-98.080+5765.4/T+15.0455 ln(T). EMF measurements; replicate uncertainty is reported in the paper and the infinite-dilution correlation provides an independent check of PHREEQC pK1. Publisher-controlled primary article; only its published equation, DOI, scope, and residual summary are recorded, not its tables. Accepted solely as independent validation of the Pitzer reaction-table pK1. No interaction coefficient is adopted or transferred.
Partanen (2013), traceable SrCl2 values, NIST ThermoML MD5 0f26a7668fdc333b8a9d6f35223d3fbd Recommended SrCl2 mean ionic activity coefficients; no Pitzer parameter is supplied or adopted 283.15-333.15 K, 101 kPa, molality scale; 22 points at 298.15 K from 0.01-3.52 mol/kg and 36 temperature states from 0.01-0.30 mol/kg Rounded values have 95% expanded uncertainty 0.001; all 58 are checked without refitting; maximum relative residual 3.11%, relative RMSE 1.58% Primary ACS article; numerical record distributed through the publisher-permitted NIST ThermoML archive and stored with exact archive checksum Accepted as held-out binary validation of the PHREEQC Sr++/Cl- B0/B1/C0 temperature family and NeqSim equation convention. It does not qualify mixed Sr brines, chloride minerals, or sulfate scale.
Partanen (2014), traceable BaCl2 study, and Hamer (1968), public NBS theoretical report BaCl2 mean activity/osmotic evidence and low-I theoretical context; no coefficient is adopted from either source Primary study covers 273-333 K and 298.15 K to 1.7884 mol/kg; molality scale. NBS report covers low ionic strength from 0-100 C Primary study reports transparent equations but its numerical table was not available through a permission-clear archive in this audit; NBS values are theoretical, not a replacement experiment Elsevier article reuse not established; NBS report is a U.S. Government work and public use is permitted PHREEQC Ba++/Cl- rows remain source-mapped and automatically usable as a complete binary, but not independently qualified. No table value is copied or inferred.
Partanen and Partanen (2020), traceable NaCl values Recommended NaCl mean activity and water osmotic coefficients; no Pitzer parameter is supplied or adopted 273.15–373.15 K, 0.2 mol/kg to saturation; molality scale; extended-Huckel model fitted independently of the PHREEQC implementation Traceable synthesis of vapor-pressure, electrochemical, cryoscopic, and solubility evidence; tabulated values are rounded to 0.001 Primary open-access article and numerical tables under CC BY 4.0 Four 298.15 K points from 0.2–2.0 mol/kg are held-out validation only. Max NeqSim residuals are 1.36% in mean activity, 0.0053 in osmotic coefficient, and below 0.0004 in derived water activity; no coefficient was refitted.
Hamer and Wu (1972), NBS/NIST critical evaluation, independently reproduced by Dash et al. (2012), open-access table Critically evaluated KCl mean activity coefficients; no Pitzer parameter is supplied or adopted 298.15 K, molality scale, 1:1 electrolyte; four selected states span 0.0982–1.9895 mol/kg Hamer-Wu evaluates thermodynamic literature; reproduced values are rounded to 0.001. Dash et al.’s direct single-ion-selective-electrode means are 10.7–13.4% higher at the same states and are treated as a method conflict. U.S. NBS authorship and U.S. Secretary of Commerce publication; the four checked values are also reproduced in a CC BY open-access primary experiment Held-out binary validation only. Max NeqSim mean-activity residual is 0.110% against a 0.20% gate; no coefficient was refitted. The critically evaluated values take precedence over the conflicting single-ion method.
Partanen (2013), traceable MgCl2 values Recommended MgCl2 mean activity and water osmotic coefficients; no Pitzer parameter is supplied or adopted 298.15 K, 0–3 mol/kg, molality scale; four checked states span 0.1667–2.0 mol/kg Traceable synthesis and tests against isopiestic and vapor-pressure literature; tabulated values are rounded Journal of Chemical Thermodynamics data distributed through the publisher-permitted NIST ThermoML archive; NIST public-data terms and citation request recorded Held-out binary validation only. Max NeqSim mean-activity residual is 2.27% without refitting; this does not qualify mixed Mg chloride/sulfate brines.
Robinson and Bower (1966), NBS mixed CaCl2+MgCl2 isopiestic data Table 2 R=1-a*yMg-b*yMg^2 correlations for osmotic coefficient and derived water activity; no Pitzer coefficient is supplied or adopted 298.15 K; three CaCl2 reference levels; checked MgCl2 fractions 0.25/0.50/0.75; total salt 0.548–2.161 mol/kg; molality and pure-water activity standard state Nine checked states; source gives no pointwise uncertainty. Gates are max absolute 0.04 in osmotic coefficient and 0.004 in water activity, without refitting. U.S. National Bureau of Standards authorship; U.S. Government work, public domain; exact table inputs and transparent derived values are stored Accepted as independent mixed-chloride observable validation. It does not qualify sulfate-bearing quaternary mixtures or mineral parameters.
Dinane, Messnaoui and Abou Nohra (2012), mixed MgCl2+MgSO4 experiment Mean MgCl2 activities over total ionic strength 0.001–8 mol/kg; fitted theta(Cl,SO4) and psi(Mg,Cl,SO4) 298.15 K; salt ratios 2.5, 5, 7.5, 10, and 15; potentiometric Mg-ISE/Ag-AgCl method and Pitzer fit theta=0.0252 +/- 0.0042, psi=-0.0049 +/- 0.0003; PHREEQC has 0.03 and -0.008 Primary Elsevier article; no table is copied and numerical-table redistribution was not established Theta broadly agrees, but psi materially conflicts. No coefficient is adopted or mixed; the coherent PHREEQC family is assessed against separate observable data.
Guendouzi and Errougui (2007), mixed MgCl2+MgSO4 water activity, NIST ThermoML MD5 29969a5f814eb0ad6d7ca5b4f093b658 28 measured water activities; no interaction parameter is adopted. The paper also fits mixing parameters, but they are not transcribed or mixed with PHREEQC. 298.15 K, 101 kPa; MgCl2 ionic-strength fractions 0.20, 0.50, 0.80; total formula-unit molality 0.35–3.80 mol/kg; molality composition and pure-water activity standard state Hygrometric method; per-state 95% combined expanded uncertainty 0.001–0.004. NeqSim max absolute residual 0.002985, RMSE 0.001275, and 26/28 states inside individual uncertainty without refitting. Primary article; numerical record is publisher-permitted NIST ThermoML. Stored unchanged subset follows the NIST public-data license and acknowledgment terms. Accepted as held-out mixed Mg-Cl-SO4 water-activity evidence only. It supports PHREEQC B0/B1/B2/C0, Cl-/SO4– theta, Mg++/Cl-/SO4– psi, charge-dependent alpha, and electrostatic conventions as one coherent calculation; it does not transfer a fitted parameter or qualify Ca/minerals.
Rodil, Arce, Wilczek-Vera and Vera (2009), mixed NaCl+KCl experiment and mandatory correction Individual Cl-/Na+/K+ activity evidence in mixed NaCl+KCl at cation molal fractions 0.75, 0.5, and 0.25 298.15 K; total chloride to 4 mol/kg; Henderson liquid-junction correction and single-ion convention Original paper reports the experiment; the correction replaces erroneous Na+/K+ tabulations without changing conclusions ACS supporting information is publicly downloadable, but copyright and no explicit data-redistribution license were established Candidate held-out mixed-family validation only. No value is stored; licensing-clear data and convention mapping remain the blocker. Exact-composition IPhreeqc checks are retained as implementation evidence, not a substitute for experiment.
Kaasa (1998), Prediction of pH, mineral precipitation and multiphase equilibria during oil recovery, National Library item, supplied scan inspected 2026-08-23 Appendix F printed pp. 260–263: binary beta(0), beta(1), occasional beta(2), and Cphi for oil-recovery-brine ions; pp. 264–266: same-sign theta and ternary psi; pp. 266–267: neutral lambda and ksi (zeta) for CO2, H2S, and CH4 systems Molality-scale Pitzer model; 298.15 K reference. Appendix F equation (F.1) uses six coefficients in raw order [a,b,c,d,e,f]; NeqSim/PHREEQC order is [a,d,e,b,c,f]. Pressure effects are handled separately in chapter 4. Appendix F cites sources 1–16, including primary literature and rows marked “This work.” Chapter 4 §4.3.2, printed pp. 100–101, states that source ranges differ and some high-temperature fits were forced toward zero where extrapolation became nonphysical, with a poorer fit to experiments. Row uncertainty and complete validity are not tabulated. User-supplied scan was visually inspected against its text layer. Thesis-table redistribution terms remain unclear; no numerical row was copied. Original cited publications require independent row-level checks and their own reuse review. Formula, page/family inventory, and coefficient-order permutation accepted. PitzerTemperatureFunction.fromKaasa1998 performs only the tested order mapping. No coefficient, forced extrapolation, or “This work” estimate was adopted.
THEREDA release 2026-01, official database Quality-controlled high-salinity Pitzer datasets, including variable-temperature sets Official site reports an oceanic set from 0 to 110 °C; formulation and scale are dataset specific Official site reports 425 tests and more than 3200 results; row-level uncertainty remains dataset specific Current release and licensing/reuse terms must be confirmed before storage Candidate validation and provenance source only; no value adopted while row-level lineage, reuse, range, and complete family compatibility remain unresolved.

Kaasa 1998 supplied-scan audit

The supplied scan removed the access blocker but not the scientific or licensing gates. Selected Appendix F citations were checked against their primary publisher records before any adoption decision:

Kaasa reference Primary source and verified scope Agreement / conflict decision
[8] He and Morse (1993), carbonate/calcite brines: CO2, bicarbonate, carbonate, and calcite evidence in Na/K/Ca/Mg chloride/sulfate solutions from 0 to 90 °C at about 0.1032 MPa Supports the thesis family and temperature lineage, but does not establish redistribution rights or validate every transcribed coefficient. No row adopted.
[13] Harvie, Møller and Weare (1984), natural-water model: mixed Na/K/Mg/Ca/H chloride/sulfate/carbonate system at 25 °C and high ionic strength Family topology agrees with Appendix F. The primary model is a 25 °C reference, so it cannot independently qualify Kaasa’s later temperature fits. No row adopted.
[14] Pabalan and Pitzer (1987), high-temperature mixed electrolytes: Na/K/Mg chloride/sulfate/hydroxide mixtures and mineral solubility Supports temperature-dependent theta/psi lineage for its stated system. Species coverage is narrower than the full thesis table and reuse remains publisher-controlled. No row adopted.
[16] Duan et al. (1992), methane in brines: CH4 solubility in 0–6 molal brines, 0–250 °C, and 0–1600 bar Supports the neutral-methane interaction lane only. NeqSim represents neutral Pitzer families, but no complete, redistributable CH4 family with the required hybrid EOS-GE validation is adopted, so this topology still fails closed.

Appendix F printed page 259 defines

p(T) = a + b(T-Tr) + c(T²-Tr²) + d(1/T-1/Tr) + e ln(T/Tr) + f(1/T²-1/Tr²),

with a=p(298.15 K). This is algebraically the PHREEQC six-term function, but its printed coefficient columns are not in PHREEQC storage order. The exact permutation is [a,b,c,d,e,f] -> [a,d,e,b,c,f]. The setup-only PitzerTemperatureFunction.fromKaasa1998 factory makes that permutation explicit, validates the six finite coefficients through the immutable value object, and performs no unit, standard-state, family, or numerical-table conversion.

The visual audit found the full implemented binary/theta/psi families and additional neutral-ion families that NeqSim does not yet represent. The tables cover H/Na/K/Mg/Ca/Sr/Ba/Fe with major chloride, sulfate, bisulfate, hydroxide, carbonate, bicarbonate, bromide, bisulfide, and acetate species, followed by CO2/H2S/CH4 neutral interactions. This is a provenance inventory, not a usable dataset: source references and experimental ranges vary by row, some rows are thesis estimates, and chapter 4 documents high-temperature zero-forcing used to suppress nonphysical extrapolation. Therefore all numerical rows remain rejected until an original redistributable source independently agrees, the complete interacting family is available, row ranges and uncertainty are recorded, and held-out activity, osmotic/water-activity, speciation, mineral, mixed-brine, continuity, balance, and performance validation pass. No thesis coefficient is stored in NeqSim.

Adoption gate

A parameter-set PR must preserve a versioned dataset identity and per-row provenance. It must map all applicable binary, same-sign, ternary, neutral-ion, alpha, electrostatic, and temperature terms; document standard state, species charge, units, pressure assumptions, fitted observables, validity, uncertainty, preprocessing, and license; and validate against data outside the fit. Required checks include mean activity and osmotic coefficients or water activity, mixed-brine speciation and mineral saturation, electroneutrality and material balance, derivative continuity, boundary behavior, determinism, and an independently configured exact-version implementation such as PHREEQC.

The supplied Kaasa scan was inspected at chapter 4 §4.3.2 (printed pp. 100–101) and Appendix F (printed pp. 259–267). No Kaasa numerical parameter was transcribed or adopted because reuse, row lineage, ranges, and independent agreement remain unresolved. The exact public-domain PHREEQC CO2-Na2SO4 and Na-K-Cl subsets are separately versioned: their complete companion families, equation mappings, range evidence, and independent IPhreeqc gates are recorded above. Conflicting NeqSim legacy NaCl/KCl rows remain available only under the unchanged legacy identity; no family is silently mixed across formulations.

For the CO2-Na2SO4 VLE qualification increment, no new Kaasa coefficient was found or needed: the already audited Appendix F CO2 neutral-family inventory supplies provenance leads, not an independent bubble-pressure dataset or a redistributable coefficient family. Consequently there is no new Kaasa/PHREEQC numerical agreement or disagreement and no convention mapping beyond the previously verified six-term coefficient-order permutation. No coefficient was adopted or rejected anew. The next missing neutral interaction remains a complete redistributable H2S family, including the self/dimer semantics required by the source model plus all active ion companions; partial H2Sg-Na+, H2Sg-Cl-, and ternary rows remain fail-closed.

Unequal-charge electrostatic mixing mapping

The unequal-charge mapping keeps Pitzer’s parameter-free nonsymmetric electrostatic term separate from the fitted short-range theta coefficient. For same-sign ions $j$ and $k$, NeqSim evaluates the public-domain PHREEQC 3 ETHETAS recurrence at the current molal ionic strength and $A_\phi$. Equal-charge pairs return exactly zero. Unequal-charge pairs contribute $\theta_{jk}+E\theta_{jk}$ to the direct single-ion term, $m_jm_k\,dE\theta_{jk}/dI$ to the common activity-coefficient function, and $m_jm_k(E\theta_{jk}+I\,dE\theta_{jk}/dI)$ to both water/osmotic-coefficient implementations.

This is the term-by-term mapping in PHREEQC source commit b0b3be7. The recurrence coefficients are USGS public-domain source values, not fitted thermodynamic data. Regression values for charges +1/+2 at independently fixed $A_\phi=0.392$ cover ionic strengths 0.01, 0.1, 1, and 5 mol/kg; a centered finite-difference check separately gates the derivative. A per-thread recurrence workspace preserves concurrency without per-call recurrence allocations, and a cached component-topology gate bypasses the kernel for equal-charge and ordinary binary salts.

The electrostatic mapping itself does not qualify or adopt any binary, theta, or psi coefficient. Mixed-brine coverage still fails closed until every short-range tuple is explicitly defined. The later sections resolve the PHREEQC six-term temperature representation, neutral families, and the C0/Cphi equation convention. Parameter-source conflicts outside the qualified CO2-Na2SO4 and Na-K-Cl subsets and reaction-species coverage remain unresolved. The supplied Kaasa scan was inspected during this mapping, but no Kaasa numerical value was copied or adopted. Its same-sign and ternary tables confirm the required family topology; source lineage, reuse, validity, and independent numerical agreement remain row-level blockers.

The full versioned PHREEQC catalog is selected automatically whenever it contains the complete active Pitzer topology; users do not choose a parameter subset. Missing mixed-ion, neutral-gas, sulfate/carbonate, or reaction companions are never inferred or borrowed from another formulation. The legacy single-salt set remains a compatibility fallback only and is not mixed into a catalog calculation.

Six-term temperature-function mapping

NeqSim also supports the PHREEQC six-coefficient temperature function as an explicit, sparse, dataset-owned override:

p(T) = a0 + a1(1/T - 1/Tr) + a2 ln(T/Tr) + a3(T - Tr) + a4(T² - Tr²) + a5(1/T² - 1/Tr²)

The mapping follows calc_pitz_param in the audited public-domain PHREEQC source (commit b0b3be7). PHREEQC fixes (T_r=298.15) K in PTEMP; the NeqSim value object stores the reference temperature explicitly so imported datasets cannot silently assume a different reference. Values within 0.001 K of the reference reproduce PHREEQC’s exact (a_0) branch. Temperature is absolute kelvin, and the returned parameter retains the source parameter’s units and standard-state convention.

Family PHREEQC six-term support NeqSim mapping in this increment Adoption decision
beta(0), beta(1), C B0, B1, C0 use all six coefficients atomic binary setter for beta0, beta1, and NeqSim Cphi Function adopted; PHREEQC C0 maps identically to NeqSim Cphi; qualified CO2-Na2SO4 and Na-K-Cl coefficients adopted
beta(2) B2 uses all six coefficients explicit beta2 temperature setter and temperature-aware activity/osmotic calls Function adopted; no coefficient adopted
theta THETA uses all six coefficients sparse same-sign-pair function, including both ion and water paths Function and qualified K+/Na+ coefficient adopted for the Na-K-Cl subset
psi PSI uses all six coefficients sparse ternary function, including both ion and water paths Function and qualified Cl-/K+/Na+ coefficient adopted for the Na-K-Cl subset
lambda, zeta, mu, eta all four use six coefficients sparse immutable neutral-family tuples in activity and both water/osmotic paths Function and equation placement adopted; qualified CO2-Na2SO4 lambda/zeta rows adopted
Aphi optional database function NeqSim retains its existing water dielectric/density correlation Not replaced; separate validation is required

Neutral-solute interaction mapping

The aqueous Henry-reference record is the durable source-comparison and runtime-boundary matrix for the IAPWS gas-reference increment. It does not adopt a Kaasa or PHREEQC Pitzer coefficient: reactive CO2/H2S Pitzer calculations retain their established reference until this matrix contains a complete, independently qualified neutral-interaction family for their exact species topology.

The neutral layer follows the public PHREEQC PITZER contract and exact source commit b0b3be7. The parameter families remain distinct:

For a distinct lambda pair, each species receives 2 m_other lambda and the osmotic sum receives m_neutral m_ion lambda. Zeta and eta use the direct derivative of their three-species molality product. Mu preserves PHREEQC’s permutation multiplicities: 1 for three identical neutrals, 3 when two are identical, and 6 when all are distinct; repeated component slots are differentiated before their contributions are accumulated. All four families use the same six-term temperature function documented above. The water-activity denominator includes neutral-solute molality only after this opt-in layer is configured, which preserves the legacy result exactly when its sparse map is empty.

PitzerNeutralParameterCoverage distinguishes absent rows from explicit zero values. Lambda and zeta are mandatory for every active neutral/ion topology once the layer is enabled. Mu and eta become topology-complete gates when their family is present. This prevents a partly imported gas/brine dataset from silently treating an unqualified missing cross-interaction as zero.

The equations are mapped on the molality scale and contribute to natural-log activity coefficients and the common Pitzer osmotic sum. They do not convert activity scales, standard states, pressure corrections, or species definitions. The Kaasa Appendix F lambda and ksi (zeta) rows for CO2, H2S, and CH4 are now structurally representable, but no thesis value is copied or adopted. Row-level lineage, reuse rights, range, uncertainty, complete companion interactions, and held-out validation remain mandatory before a versioned parameter dataset can use them.

PHREEQC C0 and NeqSim Cphi convention

PHREEQC parses the database option -C0 as TYPE_C0. In both its complete and optimized parameter loops, the public-domain implementation contributes C0/(2*sqrt(abs(zM*zX))) to the Pitzer binary term (exact source). NeqSim stores the corresponding database field as Cphi and applies the same Cphi/(2*sqrt(abs(zM*zX))) normalization in single-ion activity and both independent water/osmotic paths. PHREEQC C0 is therefore identical to NeqSim Cphi; it is not the already charge-normalized C term.

The explicit setPhreeqcBinaryTemperatureCoefficients API passes every PHREEQC -C0 coefficient unchanged. Callers must not pre-divide or multiply a source value by 2*sqrt(abs(zM*zX)). Synthetic 1-1 and 2-1 tuple regressions pin the identity mapping and the distinct downstream charge normalization at 298.15 and 373.15 K. The setter executes only while a parameter dataset is configured; runtime activity equations, the legacy Pitzer dataset, electrolyte EOS models, reaction tables, and all neutral models are unchanged. Explicit PHREEQC datasets additionally enable the model’s global normalized C0 sum and global binary-derivative F sum; these opt-in terms are required for mixed-ion single-ion activity equivalence.

The equation-convention mapping alone does not qualify a PHREEQC row or establish pressure or temperature validity. PHRQPITZ and the exact PHREEQC source are USGS public-domain evidence. The Na-K-Cl rows above are adopted only as their complete named family with independent IPhreeqc checks; other rows remain unqualified. The supplied Kaasa scan independently confirms the printed Cphi family name and six-term algebra, but no thesis value is reused because its table-redistribution terms and row-level qualification remain unresolved.

The legacy NeqSim database columns *_25, *_T1, and *_T2 keep their established three-coefficient behavior. The six-term layer is empty by default and guarded before map lookup, so legacy binary systems pay only one predictable boolean branch and non-Pitzer models execute no new code. Setter calls are explicit and retain dataset identity and mixed-brine coverage diagnostics. Functions survive serialization; clones copy only the sparse map and share immutable function values.

Synthetic coefficients continue to isolate the public equation mapping at 298.15 and 373.15 K. Separately, the named PHREEQC CO2-Na2SO4 and Na-K-Cl helpers embed only their qualified rows; they do not bulk-import pitzer.dat, change the reaction database, or select reactions. Their explicit temperature/molality evidence envelopes do not make pressure-dependent VLE, extrapolation, or any additional species topology validated.

The supplied Kaasa (1998) scan was visually inspected at chapter 4 §4.3.2 and Appendix F. Its family inventory and coefficient-order convention are recorded above, but no numerical table value was copied or adopted. The next parameter dependency is a redistributable, independently checked mixed oilfield-brine family with row-level lineage and held-out mean-activity, osmotic, water-activity, speciation, and mixed-brine validation.