Assessment of Thermodynamic Data

Assessment of Thermodynamic Data

Any computational thermodynamics software would be useless without high-quality databases containing thermodynamic data. For this reason, a lot of resources are devoted to assessing, optimizing, and validating individual thermodynamic systems and then using this data to construct internally consistent CALPHAD databases.

A Collaborative Effort

At Thermo-Calc Software, databases are developed using the proven CALPHAD methodology. The most critical and time-consuming steps in the process are the capture and assessment of thermodynamic data, as shown in Figure 1.

Figure 1. Our databases are developed using the proven CALPHAD methodology, as shown in the graphic. The most time-consuming and critical steps in the process are the capture and assessment of data, which we devote tremendous resources and expertise to through in-house experts and international collaborations.

Since the early development of Thermo-Calc, large efforts have been made with respect to thermodynamic assessments of individual systems through a great number of national and international collaboration projects. These efforts have resulted in several comprehensive databases of high quality. However, there is a continuous strong demand for this kind of activity.

To meet this demand, we employ a highly-skilled team of experts to continuously develop and improve our range of databases, both internally and through international collaboration projects such as The ADVANCE Project.

The Assessment Process

In order to be successful in assessing thermodynamic data for the development of thermodynamic databases, one must follow a rigorous set of steps to ensure quality and consistency in the final product.

Data Capture

For a high-quality assessment of thermodynamic data, all available experimental data for the system is captured and studied, such as phase equilibria and thermochemical properties like enthalpy of mixing or formation energies. As much as possible, the original publication is used since critical information about the experimental setup may be lacking in data that is cited in later articles.

Capturing experimental information is an extremely important step and it is crucial that the “assessor” has good knowledge in different experimental techniques to critically judge the uncertainty in the experimental values. In this step it is also important to capture phase data, like crystal structure, point-defects, ordering, and so on. This data is important for the selection of models to be used for each phase later in the assessment procedure.

Supplemental Data

For many of our databases, we have continuous development projects with external partners. If key experimental data is lacking for a system, tailor-made experiments can be carried out to strengthen the assessment.

We also perform ab-initio calculations using an in-house computer cluster to support the assessments. Additionally, we make use of machine learning, empirical relations, and/or generally accepted conventions to generate data for systems with little or uncertain data, based on the abundance of data that we have for experimentally well characterized systems.

In addition to the above, we continuously screen the literature, attend conferences, and are a member of several larger research centers to be up-to-date with recent research.

Model Selection based on Critical Assessment

Once all of the data is captured, a model is assigned based on critical assessment. The selection of the model for a phase must be based on the physical and chemical properties of the phase, for example crystallography, type of bonding, order-disorder transition, and magnetic properties, that have been captured during the literature review, as discussed above. A complete description of the models used in CALPHAD assessments can be found in the book by Lucas et. al1.

When selecting a model, it is also of great importance to consider which other systems the phase is stable in and whether the model chosen can be expanded into those systems as well. For example, the Mu-phase (Fe7W6,hR13, R-3m) is stable in Fe-W, Fe-Mo, Fe-Mo-W, and many more systems, so a model selected for this should be able to expand to these systems.

Free Pure Element Database to Ensure Compatibility

In order for different thermodynamic assessments to be compatible with each other within a database, the Scientific Group Thermodata Europe (SGTE) has established a set of reference data for pure elements, also known as unary systems, which they recommend to be used for all new assessments. The SGTE Pure Element Database is provided for free with Thermo-Calc and is referred to as PURE5 inside Thermo-Calc. The database can also be downloaded for free from the Resources page on our website.

Download the PURE5 database (found under the Free Databases menu)

It is also crucial that the same end-member descriptions for a phase are used throughout a database. For instance, if a previous assessment in the A-B-C system uses parameter G(Phase_X,A:B;0) for Phase_X, then the same parameter value must be used for Phase_X in the A-B-D system. If, during the assessment of A-B-D, it is found that this parameter should be slightly changed, then all A-B-X systems that uses this parameter should be reoptimized.

Published assessments can usually not be incorporated directly. Either the model selected for a phase is different from what is used in the database or an end-member description for a phase is different. In such cases, a re-optimization of the phase is made with compatible model and end-member descriptions. Sometimes a full reassessment of the system is necessary.

PARROT Module for Optimization

Once the data has been captured and assessed, it must be optimized to fit the model parameters. This is done from within Thermo-Calc using a powerful module called the PARROT Module, also known as the Data Optimization Module.

You can read about the PARROT Module and the rest of the process of developing a CALPHAD database on the Data Optimization Module (PARROT) page.

References

1Hans Leo Lucas, Suzana G Fries, Bo Sundman, Computational Thermodynamics “The CALPHAD Method”. University Press, Cambridge.  ISBN 978-0-521-86811-2

Let Us Help You

Find the Right Products