The MUSE: A multi- start optimization algorithm for surface complexation parameters in complex systems

Maria Chrysochoou, Nefeli Bompoti, Michael Machesky

Research output: Contribution to conferencePaperpeer-review


Surface complexation models (SCMs) provide a thermodn. framework to describe adsorption processes, and can, in principle, replace empirical distribution factors for fate and transport modeling of contaminants. A major drawback of SCMs is the high degree of parameterization, even when a pure sorbate-​sorbent is considered. Certain parameters can be measured, while others are either calcd., assumed or fitted, rendering the applicability of SCMs to complex systems quite limited. Although there is a plethora of software for estn. of intrinsic equil. consts., the variability and inconsistency of the extd. parameters indicates that better tools are needed to address these limitations. To this end, a hybridized optimization approach, based on a multi - start algorithm tied to a local optimizer, has been developed to allow the simultaneous optimization of several SCM parameters. The MUSE algorithm can be applied to any modeling formulation and parameter choice, while remaining cognizant of overfitting and phys. constraints. To further facilitate the application of sorbent-​specific SCMs to complex systems, we propose the application of MUSE to ext. common thermodn. consts. for classes of surfaces, similar to the surface chem. assemblage model previously proposed by Lofts and Tipping (1998)​. In this study, we demonstrate the MUSE application to chromate adsorption on iron oxides and specifically ferrihydrite, goethite and hematite. The results indicate that differences in intrinsic equil. consts. between the three minerals are driven primarily by the sp. surface area value adopted, and that it is possible to account for this effect. In addn., similar surface coverage effects are obsd., specifically higher equil. consts. for very low surface coverages, an effect that has been previously obsd. calorimetrically by Kabengi et al. (2017)​. We propose that this effect corresponds to the "weak" and "strong" sites typically adopted for Generalized Composite models and that the MUSE algorithm allows for a systematic examn. of these site classes for different surfaces. Kabengi N., Chrysochoou M., Bompoti N. and Kubicki J. 2017, Chem. Geol., 464, 23-​33. Lofts S. and Tipping E. 1998. Cosmochim. Acta 62(15)​, 2609-​2625.
Original languageEnglish (US)
StatePublished - 2018


  • ISWS


Dive into the research topics of 'The MUSE: A multi- start optimization algorithm for surface complexation parameters in complex systems'. Together they form a unique fingerprint.

Cite this