# A largely factual and incomplete history of MELTS

**1978:** **Mark S. Ghiorso**, a graduate student at UC Berkeley, begins work on silicate liquid/mineral geothermometers. The project is prompted by** Ian Carmichael**, who wants to know how to make better geothermometers for igneous systems. Ghiorso uses the state-of-the-art CDC6400 computer at Berkeley and spends countless hours with punch cards.

**1980:** Ghiorso and Carmichael publish the geothermometers based on the first silicate-liquid model. Carmichael hires **Richard Sack** (Harvard) as a postdoc to generate data for a better model. For the next two years, Sack completes countless experiments, spends hundreds of hours on the microprobe, and assembles extensive data.

**1983:** Ghiorso (now at University of Washington) completes a new silicate liquid model incorporating Sack's work. The new model is based on more data and more sophisticated regression techniques. Consequently, it is more comprehensive and more robust than the previous effort. The resulting paper is the most lengthly one every published in Contributions to Mineralogy and Petrology. Ghiorso also completes a computer program (written in FORTRAN 77) - SILMIN -that can be distributed (via 9-track magnetic tape ) to others so that they too can run silicate melt/mineral mass transfer calculations.

**1985:** Ghiorso and Sack realize that, for the mass transfer models to work better, they need to develop more sophisticated descriptions of the thermodynamics of igneous minerals. They work on this subject for the next decade, and publish a series of papers on pyroxenes, olivine, oxides, feldspars, and feldspathoids.

**1988:** Ghiorso receives a generous grant from Digital Equipment Corporation (DEC). This is the first grant from any funding source awarded for the MELTS project. At the time, NSF did not think proto-MELTS was a very interesting or useful idea. Ghiorso obtains a MicroVax from DEC, the first semi-portable personal mainframe computer.

**1989-1993:** Ghiorso works on computational thermodynamics routines (which eventually become the MELTS package) and rewrites the original FORTRAN mass transfer program into C, adding a graphical user interface based on DEC's new windowing system. (That windowing system later became Motif.)

**Mid 1990s:** **Mark Hirschmann**, a graduate student of Ghiorso, adds an improved olivine model to the code base.

**1994:** Ghiorso unveils a preliminary version of the new model, now called "MELTS," at a poster session at the GSA meeting in Boston.

**1995:** Ghiorso and Sack publish the MELTS model and release a version that users can download and run on a UNIX workstation. The MELTS model paper, which appears in Contributions to Mineralogy and Petrology, later becomes the most highly cited paper in that journal.

**1995:** **Edward Stolper** (Caltech) becomes interested in using MELTS to model melting of the upper mantle. He hires Hirshmann as a postdoc to work on this project. Stolper's student **Paul Asimow** is given this project while a first year graduate student. Hirshmann and Asimow generate some unusual modeling results that are later confirmed by experiment, and so emboldened go off where no one has gone before. Hirschmann discovers a fundamental problem when using MELTS at higher pressures than originally intended; there are workarounds to the situation, but these are unsatisfying and will not be fixed without recalibration of the silicate liquid model (see pMELTS below).

**1995:** Asimow adds code to MELTS to extend calculations into the subsolidus.

**Late 1990s**: With the growing popularity of the World Wide Web, Ghiorso releases a MELTS applet written in Java; one of the first web-based client-server apps in the Earth sciences.

**1998:** Ghiorso and Hirshmann (now at University of Minnesota) resolve to fix the problem manifest for mantle melting calculations at higher pressures and receive NSF funding from OCE to do so. This is the first NSF grant (of several to follow) secured for MELTS. A better liquid equation of state, more high-pressure data and a recalibration does the trick. Washington graduate student **Pete Reiners** is involved in this effort as is postdoc **Victor Kress**.

**2001**: A revised MELTS model, called "pMELTS," is released and published. pMELTS is designed specifically to melt mantle-like bulk compositions in the pressure range 1-3 GPa and is not intended for other applications.

**2001:** After Apple Computer moves Mac OS to a Linux-based system, Ghiorso ports MELTS to the Macintosh (in version 10.1, or Puma) and adds numerous numerical improvements. He focuses on the Macintosh for all subsequent program development.

**2002 -2004:** Ghiorso resolves to make an even better model for melting the mantle. Unable to find a suitable equation of state, he creates a new one. The work is published in *American Journal of Science.*

**2005-2006**: Ghiorso (now at OFM Research) teams up with Hirschmann and **Tim Grove** (MIT) to help gather more data to create a better model for melting the mantle. The LEPR database is initiated.

**2009-2015:** Ghiorso teams up with Asimow and staff member **Paula Antoshechkina** in hopes to create better models for high-pressure mineral phases.

**2011-12:** Ghiorso and **Gil Gualda** (Vanderbilt) modify MELTS to work better in high-silica rhyolitic systems. Rhyolite-MELTS is released.

**2012:** While working on PhasePlot, Ghiorso develops new numerical algorithms that he adds to rhyolite-MELTS, greatly improving its numerical stability.

**2013:** Asimow and Antoshechkina release alphaMELTS, which is built upon the pMELTS compute engine.

**2015:** Ghiorso and Gualda incorporate a new H_{2}O-CO_{2} fluid saturation model into MELTS.

**2017:** Ghiorso makes MELTS open source.