The software THERMOCALC calculates model phase equilibria involving the HPx-eos and/or individual end-members from the Holland & Powell dataset. Here we provide an overview of the program’s purposes and history. For practical information and downloads, see Getting started with THERMOCALC.

On this page:

Paradigm and use

THERMOCALC reads in thermodynamic data in the form used by the HPx-eos. Using the data, it constructs and solves statements of equality of chemical potential among the end-members of the phases, supplemented where necessary by mass-balance or direct constraints on compositional variables. In this way it is able to find the equilibrium compositions of phases in an assemblage provided by the user, under the conditions specified by the user.

Three main calculation facilities are publicly available at present:

  • Phase diagram calculations (forward modelling)
  • Multiple-reaction thermobarometry (inverse modelling)
  • Tabulating thermodynamic data extracted from the HP dataset and HPx-eos

The most common use of THERMOCALC is to calculate stable phase diagrams at constrained bulk composition, using a bulk composition derived from a rock sample. The aim is to predict what mineral and fluids would coexist in the rock if the rock were subjected to a range of pressures and temperatures. The user then hopes to gain insight into the rock’s history by comparing the predictions with petrographic observations and perhaps analysis of phase compositions.

Past and future

Roger Powell began to write THERMOCALC (unrelated to the commercial software, Thermo-Calc) in the early 1970s, initially as part of his PhD research, and later through his collaboration with Tim Holland on what was to become the Holland & Powell dataset. Details of its operation were first published in Powell & Holland (1988). At first the program was primarily intended to calculate equilibrium among pure end-members, or impure end-members at fixed activity. This made it a powerful tool for optimal thermobarometry (Powell & Holland, 1988, 1994), a multiple-reaction approach to thermobarometry that formally accounts for correlated uncertainties in the input. For this purpose, the analysed compositions of minerals in a presumed equilibrium were converted into activities via simple activity-composition relations.

Compatibility triangles for metapelites calculated with THERMOCALC

Powell, Holland & Worley (1998) used THERMOCALC to calculate phase diagrams at fixed bulk composition, modelling the thermodynamics of solid solutions as if they underwent ideal mixing-on-sites. Over the next twenty years the solution models, now for both solid solutions and fluids, expanded drastically in complexity, becoming today’s HPx-eos.

THERMOCALC is evolving rapidly at present. Recent and imminent changes to the user’s experience are aimed at:

  • Modifying the output to
    • make the HPx-eos easier to understand
    • give access to the more of the calculated thermodynamic properties of phases and assemblages
    • produce .csv output that can be read by external applications such as Excel, in order to facilitate manipulation and plotting of output.
  • Facilitating open-system calculations, a step towards developing integrated simulations of open-system Earth processes.
  • Expanding the options for applying multiple-reaction thermobarometry, and making this more user-friendly.
  • Making some uncertainty-estimation tools available to users.
  • Developing co-functionality with the ENKI environment (via Jupyter workbooks).

THERMOCALC as the definitive HPx-eos implementation

THERMOCALC is the primary tool for calibrating the HPx-eos, including the underlying Holland & Powell dataset. In effect, therefore, THERMOCALC’s implementation defines the correct version of these models – overriding any typos in the papers where they are published!

Both the Holland & Powell dataset and the HPx-eos have been implemented in a variety of other software, in particular Perple_X (Connolly, 2005) and Theriak/Domino (de Capitani & Petrakakis, 2010). Although for recent families of HPx-eos – those of Green et al (2016) and Holland et al (2018) – there has been an effort to achieve near- equivalence between THERMOCALC and Perple_X, there is in general no guarantee that the dataset or HPx-eos are implemented correctly outside THERMOCALC. Any differences are liable to vary with the chemical system and phases present. In Perple_X, precise comparison with THERMOCALC is limited since the software digitises the Gibbs energy surfaces rather than solving the equilibrium equations exactly.


De Capitani & Petrakakis (2010) Am Mineral 95 1006-1016. Connolly (2005) Earth Planet Sci Lett 236 524-541. Green et al (2016) J Metamorph Geol 34 845-892. Holland et al (2018) J Petrol 59 881-900. Powell & Holland (1988) J Metamorph Geol 6 173-204. Powell & Holland (1994) Am Mineral 79 120-133. Powell et al (1998) J Metamorph Geol 16 577-588. Ziberna et al (2017) Am Mineral 102 2349-2366.

Get started with THERMOCALC ⟶