Accurate chemical speciation models of solutions containing the ions of seawater have applications in the calculation of carbonate system equilibria and trace metal speciation in natural waters, and the determination of pH. Existing models, based on the Pitzer formalism for the calculation of activity coefficients, do not yet agree with key experimental data (potentiometric determinations of H⁺ and Cl⁻ activity products in acidified artificial seawaters) and, critically, do not include uncertainty estimates. This hampers applications of the models, and also their further development (for which the uncertainty contributions of individual ion interactions and equilibrium constants need to be known). We have therefore implemented the models of Waters and Millero (Mar. Chem. 149, 8-22, 2013) and Clegg and Whitfield (Geochim. et Cosmochim. Acta 59, 2403-2421, 1995) for artificial seawater, within a generalised treatment of uncertainties, as a first step towards a more complete model of standard seawater and pH buffers. This addition to the models enables both the total uncertainty of any model-calculated quantity (e.g., pH, speciation) to be estimated, and also the contributions of all interaction parameters and equilibrium constants. Both models have been fully documented (and some corrections made). Estimates of the variances and covariances of the interaction parameters were obtained by Monte Carlo simulation, with simplifying assumptions. The models were tested against measured electromotive forces (EMFs) of cells containing acidified artificial seawaters. The mean offsets (measured – calculated) at 25 °C for the model of Waters and Millero are: 0.046 ± 0.11 mV (artificial seawater without sulphate, 0.280 mol kg⁻¹ to 0.879 mol kg⁻¹ ionic strength); and −0.199 ± 0.070 mV (artificial seawater, salinities 5 to 45). Results are similar at other temperatures. These differences compare with an overall uncertainty in the measured EMFs of about 0.04 mV. Total uncertainties for calculated EMFs of the solutions were dominated by just a few contributions: mainly H⁺-Cl⁻, Na⁺-Cl⁻, and H⁺-Na⁺-Cl⁻ ionic interactions, and the thermodynamic dissociation constant of HSO₄⁻. This makes it likely that the accuracy of the models can readily be improved, and recommendations for further work are made. It is shown that standard EMFs used in the calibration of the marine ‘total’ pH scale can be accurately predicted with only slight modification to the original models, suggesting that they can contribute to the extension of the scale to lower salinities.

含有海水离子的溶液的准确化学形态模型可用于计算天然水中的碳酸盐系统平衡和痕量金属形态，以及测定 pH 值。现有模型，基于用于计算活度系数的 Pitzer 形式，尚未与关键实验数据一致（H ⁺和 Cl ^{-的电位测定）}酸化人工海水中的活度产物），而且至关重要的是，不包括不确定性估计。这阻碍了模型的应用，也阻碍了它们的进一步发展（需要了解单个离子相互作用和平衡常数的不确定性贡献）。因此，我们实施了 Waters 和 Millero (Mar. Chem. 149, 8-22, 2013) 和 Clegg 和 Whitfield (Geochim. et Cosmochim. Acta 59, 2403-2421, 1995) 的模型，用于人工海水，在广义处理中不确定性，作为迈向更完整的标准海水和 pH 缓冲液模型的第一步。对模型的这种添加既可以估计任何模型计算量（例如，pH、物种形成）的总不确定性，也可以估计所有相互作用参数和平衡常数的贡献。两种模型都已完整记录（并进行了一些更正）。交互参数的方差和协方差的估计是通过蒙特卡罗模拟获得的，并简化了假设。这些模型针对含有酸化人造海水的电池的测量电动势 (EMF) 进行了测试。Waters 和 Millero 模型在 25 °C 时的平均偏移量（测量 - 计算）为：0.046 ± 0.11 mV（不含硫酸盐的人造海水，0.280 mol kg^-1至 0.879 mol kg ^-1离子强度）；和 -0.199 ± 0.070 mV（人造海水，盐度 5 至 45）。其他温度下的结果相似。这些差异与测量的 EMF 的总体不确定性相比，约为 0.04 mV。解决方案的计算 EMF 的总不确定性仅由几个贡献支配：主要是 H ⁺ -Cl ^-、Na ⁺ -Cl ^-和 H ⁺ -Na ⁺ -Cl ^-离子相互作用，以及 HSO ₄ ^{-的热力学解离常数}. 这使得模型的准确性很可能很容易提高，并提出了进一步工作的建议。结果表明，只需对原始模型稍作修改，即可准确预测用于校准海洋“总”pH 标度的标准 EMF，这表明它们有助于将标度扩展到更低的盐度。