Welcome to the ICME community discussion concerning thermodynamic software development.

There is considerable interest in open source thermodynamics, primarily for research and education. In education, a lightweight and flexible platform for illustrating basic thermodynamic principles with a very basic graphical interface would be very helpful. And in research, the ability to extend and customize software is very important.

In particular, some research extension ideas have included:
- Equilibria under fields, i.e. the effect of electric, magnetic and strain fields on phase equilibria.
- Automatic generation of data using ab initio and statistical mechanics techniques, such as the ATAT code which links with abinit or VASP.
- Statistical estimation of inputs for phase equilibria.
- Estimation of free energy using phase diagrams using optimization, which is tricky because the mapping is not deterministic.

Toward that end, a first step is to put together a simple library of functions for thermodynamic calculations, and a couple of simple front ends. Some work on these pieces has already begun.

- Craig Carter at MIT has shown that the minimum energy surface for a system is simply the convex hull of the free energy functions of the various phases. This presents a straightforward route toward rapid calculation of complex multicomponent equilibria all at once, instead of one-by-one root finding for each individual composition and tie line. He implemented this in mathematica for a binary system as an educational illustration.

- Edwin Garcia at Purdue has extended these python scripts, and written a set of python scripts which find phase boundaries in binary systems.

- Adam Powell of Opennovation (me) has a simple ternary free energy visualization program using Geomview as its front end. It should not take long to link this with qhull to generate the convex hull, and visualize this along with the phase diagram. From there, one need only add sliders to show changes in shapes and topologies with free energy parameters, a ternary equivalent to Edwin and Craig's binary visualization scripts.

These efforts are largely useful for educational purposes, so they are good candidates for entries in the courseware section of the Materials Digital Library Teaching Archive http://teaching.matdl.org/ . This can all happen very soon.

In the "medium term", a platform for research should start with components such as:

- An object library for sublattices, phases, energies, systems, etc. The design of these objects will be crucial to the flexibility of the future code. This should be written in an interface definition language such as SIDL or CORBA, for automatic generation of multiple language front ends and flexibility in back ends.

- Some provision for data import and export. The data format will be important here, it should include references such that one can easily trace data back to its source, methods, etc.

- A data entry front-end to facilitate entry of data into an online database using this format. [Perhaps there could be some automation of this process for harvesting data from literature.]

These objects, and the resulting libraries, should go into a code repository such as MatForge on the Materials Digital Library, http://www.matforge.org/ .

This is a start, I welcome comments on the algorithm, design, data, applications, or any other aspect of this work.

I spent a couple of hours this afternoon "packaging" an old and simple ternary free energy visualization program for the Teaching Archive on MatDL. You can see it at: http://teaching.matdl.org/teachingarchives/wiki/Ternary .

Share and enjoy.

Version 0.2 of Ternary now links with the qhull library to calculate the convex hull of a free energy function. It uses this to visualize 2- and 3-phase regions of a ternary phase diagram isotherm. (Same URL.)

Ternary version 0.3 is now up, same website. This version features support for multiple phases with different free energy functions, and now properly calculates temperature dependence of the relative free energy curves. Well, almost properly -- it assumes free energy decreases linearly with temperature...

This version is also nearly sufficiently abstracted to wrap its functions in a graphical interface.

The next version will improve the accuracy of the 2- and 3-phase region boundaries, and add visualization of the spinodal region.

Plans moving forward:

Version 0.4 is nearing completion on teaching.matdl.org . This version will much more accurate in its calculation of the edges of two- and three-phase regions. Instead of using only compositions on the grid to form the convex hull, it will use a Newton method to search for lower-energy compositions in multi-phase triangles.

When this is complete, the next steps will be:
- Visualize the accurate ternary phase diagram in temperature-composition space.
- Create an SIDL interface using Babel so other languages (e.g. python) can call its methods. Babel is at: http://www.llnl.gov/casc/components/babel.html
- Make a nice interactive demonstration.
- Move the back-end from the Teaching Archive to a new thermo section of MatForge, keeping the interactive demo(s) on the Teaching Archive.
- Expand the phase diagram engine to arbitrary components.
- Include more phenomena in the free energy function, e.g. pressure, stress, electric and magnetic fields, free surface shape.
- Invite data contribution from academics, government, and industry.
- Use ATAT/abinit to estimate free energies in unknown systems.

This is roughly the plan for future development. Anyone is welcome to participate in the programming and applications, and in this discussion.

One of my students showed me this link:
http://serc.carleton.edu/resea...ria/perplex_vs_tc.html . Perplex seems to be free: http://www.perplex.ethz.ch/ . May be interesting for some of you on the list.

Bob