As so often, this code started out as a small project in 1987. Back then, Claus Leitherer wanted to have a little computer program to calculate ionizing fluxes for individual O stars. The program evolved from handling a single star to a whole population. At the same time, there was the desire to assign positions in the HRD for comparison with observations. Results from this rather simple code were reported in ApJS, 73, 1 (1990). By 1990 the program had grown to compute the output of the mechanical luminosity of winds and of supernovae. A paper on that is in ApJ, 401, 596 (1992). The "real" spectral synthesis capability was implemented in 1991, when we added Kurucz and Schmutz model atmospheres from the far-UV to the near-IR. This allowed us to calculate spectra and colors for young populations. Carmelle Robert included an IUE spectral library in the code in order to compute synthetic lines at 0.75 A resolution (ApJ, 418, 749 , ApJS, 99, 173 ).
By 1994, we had computed so many models, synthesizing all kinds of galaxy parameters that the suggestion was made to combine everything, plot it in a homogeneous way, and publish it (ApJS, 96, 9 ). Who would have predicted that quite a few astronomers found this useful? Later we began to add a few more specialized routines, like the computation of O VI line profiles by Rosa Gonzalez-Delgado (ApJ, 489, 601 ). The most recent subroutine was written by Jeff Goldader. This part calculates the strengths of selected stellar features in the near-IR.
In 1997 it become clear that the code needed a major upgrade. Daniel Schaerer replaced the ancient Maeder (1990) tracks by the most recent (1992-94) model series of Geneva. He also implemented the capability to perform isochrone synthesis in addition to classical evolutionary synthesis. One of the subroutines used for this was written by Georges Meynet. We also decided to take advantage of the homogeneous atmosphere grid compiled by Lejeune et al. (1997) and replaced the original Kurucz models by the new set. The updated code was made available to the community via a spiffy website called Starburst99 (Leitherer et al. 1999, ApJ, 123, 3).
Since then, quite a few important improvements were made. First, Duilia de Mello added a high-resolution spectral library of B stars in fall 1999. A description is in ApJ, 530, 251 (2000). This library replaced the low-resolution B-star library we used before. (O stars have always been from a high-dispersion library.)
Alessandra Aloisi added supernova yields to the code. Before that, only yields from stellar winds were taken into account, and the supernova yields of those elements which undergo no nuclear processing during a supernova explosion. Now the models are up-to-date with respect to the nucleosynthesis in type II supernovae (and only those -- we don't do type I's yet). This change was made in August 2000.
Claus Leitherer and summer student Joao Leao Souza added a spectral library of LMC and SMC stars in August 2000 (Leitherer et al. 2001, ApJ, 550, 724). This allows calculations of models for UV spectra with sub-solar metallicities. Only O-stars are used for SMC/LMC metallicity, everything else is solar. This is not perfect -- but it is a starting point.
Miguel Cervino, in collaboration with Daniel Schaerer substantially improved several numerical aspects of the code. In particular he wrote new code to handle the calculation of the supernova rate in isochrone synthesis. Previously, all SN related parameters showed rather ugly discontinuities due to numerical instabilities (see Leitherer et al. 1999). This was fixed in October 2000.
Anne Pellerin and Claus Leitherer added a high-resolution spectral library of hot stars observed with FUSE. The library stars are in the Galaxy, the LMC, and the SMC. The spectral range is from 1000 to 1183 A, and the structure corresponds to that of the IUE/HST library used at longer wavelengths. The library is discussed in Robert et al. (2003).
A major limitation of the original Starburst99 models was removed by Richard Norris, Linda Smith, and Paul Crowther. They replaced the unblanketed WR atmospheres of Schmutz et al. (1992) by a fully blanketed set calculated with John Hillier's code. The updated Starburst99 code, released as v4.0 in July 2002, gives more realistic EUV fluxes during WR dominated phases.
Starburst99 v5.0 in December 2004 was the most significant upgrade since the original Starburst99 code was released. Gerardo Vazquez worked hard to add the Padova tracks including thermally pulsing AGB stars to Starburst99. We always considered our primary application domain young, OB-star dominated starbursts. Most astronomers prefer modeling such phases with the evolutionary tracks from the Geneva group. Nevertheless, young and old stars often come together, and the Geneva models are not optimized for reproducing low-mass stars. Therefore we opted for adding the Padova models as well (Vazquez & Leitherer 2005). A second major addition was the implementation of a high-resolution optical library by Lucimara Martins. This library has a resolution of 0.3 A and complements the previous low-resolution library when line profile studies are desired (Martins et al. (2005).
On August 25, 2010 we released Starburst99 v6.0. This version includes a new subroutine to calculate fully theoretical UV spectra using a library generated with the WM-Basic code (Leitherer et al. 2010). These models complement and extend the existing library of empirical spectra. Additional changes in this version include: replacement of the old Schmidt-Kaler spectral-type calibration of OB stars by the more modern work of Martins et al. (2005); update of the subroutine nucleo to account for the SN yields of very massive stars (>40) in an approximate way. Finally, the code was restructured quite a bit and many questionable fortran issues were fixed.
Version 7.0.0 of Starburst99 was released in March 2014. This version was developed in close collaboration with the Geneva stellar evolution group and offers a set of tracks with rotation. The models are discussed in Levesque et al. (2012) and Leitherer et al. (2014). Other major new features of the package are the inclusion of the Potsdam Wolf-Rayet library for the computation of the theoretical UV line spectra, and a subroutine to calculate the stellar Lyman-alpha line (Pena-Guerrero & Leitherer 2013).
More to come in the future --- stay tuned!