BPASS: Binary Population and Spectral Synthesis

• BPASS is a binary-inclusive stellar population synthesis framework that models both single and binary star evolution to predict composite observables.
• It employs detailed grids and physical ingredients such as Roche-lobe overflow, common-envelope evolution, and metallicity-dependent winds to simulate stellar outcomes.
• BPASS outputs include spectral energy distributions, ionizing photon fluxes, and transient rate predictions that inform studies of high-redshift galaxies and compact-object mergers.

Binary Population and Spectral Synthesis (BPASS) is a stellar population synthesis framework explicitly built to include the effects of binary stellar evolution when predicting the properties of stellar populations. Rather than assuming that each star evolves in isolation, BPASS evolves grids of single and binary stellar models and combines them into composite stellar populations characterized by an initial mass function, metallicity, age, and, when required, a star formation history. In this form it predicts spectral energy distributions, ionizing photon production, transient rates, stellar number counts in different evolutionary states, nucleosynthetic yields, and compact-object merger rates, and it has served as a binary-inclusive alternative to standard single-star population models for more than a decade (Stevance et al., 2020).

1. Scientific motivation and historical role

BPASS is motivated by the empirical fact that a large fraction of stars are born in multiple systems and that more than 70% of massive stars are found in close binary systems, meaning that they will interact over the course of their lifetime (Stevance et al., 2020). In this setting, binary interactions alter stellar lifetimes, spectral types, final core masses, supernova classes, remnant masses, and the hardness and duration of ionizing radiation fields. A concise formulation from early BPASS work is that “the general effect of binaries is to cause a population of stars to look bluer at older ages than predicted by single-star models” (Eldridge, 2010).

The scientific significance of this shift is broad. Binary-inclusive synthesis has been used to reinterpret Hertzsprung–Russell diagrams, Wolf–Rayet to red-supergiant number ratios, core-collapse supernova subtype fractions, ionizing photon production efficiencies, and gravitational-wave progenitor populations (Eldridge et al., 2020). BPASS therefore occupies a central position in the broader move from single-star evolutionary synthesis to population models in which interacting binaries are treated as a first-order ingredient rather than a perturbation.

2. Physical ingredients and synthesis formalism

BPASS constructs stellar populations from detailed stellar evolution tracks, including both single stars and binaries, and follows binary interactions such as Roche-lobe overflow mass transfer, common-envelope evolution, binary mergers, and tidal effects in an approximate way; it also incorporates metallicity-dependent stellar winds, core-collapse supernova prescriptions, compact remnant formation, stripped helium stars, and the extended lifetimes of hot, luminous objects produced by binary mass gain (Stevance et al., 2020). In the v2.x series, the underlying stellar models are built with a customized version of the Cambridge STARS code, and the population layer weights them with observationally motivated distributions of mass, mass ratio, and orbital period (Stevance et al., 2022).

At the level of formal synthesis, the basic BPASS building block is a single stellar population formed in an instantaneous burst. For a population of age $t$ and metallicity $Z$, the single-burst spectrum is written as

$$L_\lambda^{\mathrm{SSP}}(t, Z) = \int_{M_{\min}}^{M_{\max}} \xi(M)\, l_\lambda(M,t,Z)\, dM,$$

where $\xi(M)$ is the IMF and $l_\lambda(M,t,Z)$ is supplied by the stellar and binary evolution tracks together with atmosphere models (Stevance et al., 2020). For extended star formation histories, BPASS outputs are convolved as

$$L_\lambda(t,Z)=\int_0^t \mathrm{SFR}(t')\,L_\lambda^{\mathrm{SSP}}(t-t',Z)\,dt'.$$

The framework also predicts ionizing photon budgets and transient rates from the same underlying grids. For example, the hydrogen-ionizing photon rate is obtained from the integrated spectrum, while supernova rates follow from time-dependent mappings between initial mass, age, and explosion outcome (Stevance et al., 2020). This structure makes BPASS simultaneously a stellar-evolution library, a population-synthesis engine, and a source of synthetic observables.

3. Outputs, grids, and data products

BPASS outputs are distributed as large tabulated grids over age, metallicity, IMF choice, and binary parameters. Publicly available products include spectral energy distributions, ionizing photon fluxes, supernova rates by subtype, stellar numbers and Hertzsprung–Russell diagram positions, compact-object and merger predictions, and nucleosynthetic yields (Stevance et al., 2020). In practical use, these outputs are notable not only for their breadth but for their scale: the data products are described as large, varied, and complex, with thousands of files and more than 100,000 individual stellar and binary models (Stevance et al., 2020).

Data product	Content
Spectral energy distributions	$L_\lambda(t,Z)$ or $L_\nu(t,Z)$ over UV, optical, and IR
Ionizing photon fluxes	Hydrogen-ionizing and analogous He-ionizing photon rates
Transient and rate tables	Time-dependent supernova and merger rates
Stellar catalog products	Stellar properties, HR-diagram positions, evolutionary-state counts
Yield tables	Elemental yields from supernovae and winds

These grids are designed to support both direct inspection and downstream modeling. Typical tasks include extracting a spectrum at fixed age and metallicity, integrating above the Lyman edge to obtain ionizing photon rates, comparing supernova rates to transient surveys, and constructing synthetic HR diagrams or colour–magnitude diagrams from the stellar catalogs (Stevance et al., 2020). A practical consequence is that BPASS is powerful precisely because it exposes population-level outputs derived from internal binary microphysics while leaving interpolation, convolution, and observational forward modeling to the analyst.

4. Observational domains and astrophysical applications

BPASS has been used extensively in the interpretation of classical star-formation rate indicators. In early work on H$\alpha$ and far-UV diagnostics, binary-inclusive populations retained significant ionizing output beyond $\sim 8$ Myr, making H$Z$0 luminosity decline more slowly with age than in single-star populations and reducing the susceptibility of star-formation indicators to stochastic high-mass sampling (Eldridge, 2010). This logic extends directly to nebular-line modeling: when BPASS stellar spectra are coupled to CLOUDY, binary-star models continue to produce strong flux and high [O III]/H$Z$1 ratios at ages above 10 Myr, whereas single-star models rapidly decrease in flux and ionization strength (Xiao et al., 2018).

The framework has also become standard in studies of high-redshift star-forming galaxies and cosmic reionization. Binary interactions extend and harden the ionizing phase, increase $Z$2 at low metallicity, and help account for the strong nebular emission and hard UV continua observed in young, metal-poor systems (Eldridge et al., 2020). A later BPASS release, v2.3, introduced α-enhanced synthetic spectra while retaining the established BPASS stellar evolution tracks, enabling explicit exploration of how $Z$3 affects UV and optical diagnostics in young binary-rich populations (Byrne et al., 2022).

Compact-object applications are equally prominent. BPASS predicted that GW150914-like binary black-hole mergers are most likely in populations below $Z$4 and have very low probability above $Z$5, with low-metallicity channels including both standard binary evolution and quasi-homogeneous evolution (Eldridge et al., 2016). In a different regime, BPASS v2.2.1 fiducial models natively produce VFTS 243-like black-hole plus O-star systems and favor extremely small natal kicks, while the reduced Hobbs kick distribution is strongly disfavoured in that specific context (Stevance et al., 2022). BPASS has also been used to model core-collapse gamma-ray burst progenitors, where binary stripping and angular-momentum retention produce type Ic progenitors consistent with observed GRB rate evolution and host metallicity distributions (Chrimes et al., 2019).

The same formalism has been adapted to other classes of binary products. It has been used to interpret massive-star number-count diagnostics and their dependence on age, metallicity, and binary fraction (Dorn-Wallenstein et al., 2018), to study helium-burning blue large-amplitude pulsators with predicted Galactic populations and binary properties (Zhang et al., 26 Sep 2025), and, through X-BPASS, to incorporate self-consistent X-ray binary emission into BPASS stellar populations, with the conclusion that X-ray binaries contribute toward powering nebular He II emission without causing significant overestimates of hydrogen ionization (Bray et al., 26 Aug 2025).

5. Software ecosystem and model extensions

The complexity of BPASS outputs led directly to the development of supporting software. Hoki is an open-source Python package designed to make BPASS data easily accessible and facilitate analysis by automatically loading outputs into pandas.DataFrame objects or specialized Python structures, and by providing infrastructure for stellar models, HR diagrams, colour–magnitude diagrams, plotting, and access to standard BPASS constants (Stevance et al., 2020). The use of Python is deliberate because it is ubiquitous within Astronomy, allowing BPASS results to fit naturally into Jupyter-based workflows and reproducible analysis pipelines.

BPASS has also continued to expand in physical scope. BPASS v2.3 added α-enhanced atmosphere grids based on C3K spectra, making it possible to study the dependence of UV and optical line indices, colours, and ionizing continua on $Z$6 in young binary stellar populations while still using Solar-scaled stellar evolution models (Byrne et al., 2022). X-BPASS extends the framework to X-ray binaries by calculating the accretion-disc evolution, luminosity, and spectral energy densities of individual accreting compact objects and combining them into total X-ray spectra for stellar populations over age and metallicity (Bray et al., 26 Aug 2025).

Recent comparison work also places BPASS within a wider ecosystem of binary-inclusive synthesis tools. A POSYDON-based solar-metallicity spectral synthesis framework confirms BPASS’s main qualitative conclusion that binaries substantially harden and prolong the ionizing output of stellar populations, while also showing that quantitative predictions, especially for He II-ionizing photons, depend sensitively on stripped-star populations, atmosphere assignments, and late hot evolutionary phases (Kasdagli et al., 11 Jun 2026). This suggests that BPASS now functions not only as a standalone model suite but also as a benchmark against which alternative detailed binary frameworks are evaluated.

6. Limitations, uncertainties, and active debates

BPASS carries the standard uncertainties of detailed binary population synthesis. The public documentation emphasizes empirical uncertainties in binary parameter distributions, approximations in common-envelope and mass-transfer physics, atmosphere models and bolometric corrections, discretization in metallicity, and assumptions about stellar rotation and magnetic fields that are not always fully modeled (Stevance et al., 2020). In practice, these uncertainties mean that users must track IMF choices, binary assumptions, and metallicity mapping carefully, and avoid over-interpreting small differences between BPASS and other synthesis codes without accounting for the underlying model choices.

Several active debates concern compact remnants and late binary evolution. For Galactic double white dwarfs, BPASS predicts 20 to 40 times fewer detectable systems than SeBa or BSE in some LISA forecasts, and comparison with electromagnetic samples indicates that BPASS underpredicts the number of short-period double white dwarfs by at least an order of magnitude, implicating differences in the treatment of mass transfer and common-envelope events (Zeist et al., 27 May 2025). In high-energy spectral synthesis, comparison with POSYDON shows that BPASS and alternative detailed frameworks agree on the importance of binaries and stripped stars for UV and ionizing output but diverge in the number of H-poor stars produced and in the treatment of He II-ionizing emission (Kasdagli et al., 11 Jun 2026). For black-hole natal kicks, BPASS-based modeling of VFTS 243 argues against reduced Hobbs kicks in that channel and favors very low-kick black-hole formation (Stevance et al., 2022).

A common misconception is that BPASS is simply a single-star spectral synthesis code with an added binary fraction. The research record instead presents it as a framework in which binary interactions reshape the population-level predictions themselves: they change the duration of ionizing phases, alter supernova subtype fractions, shift HR-diagram occupancy, modify compact-remnant populations, and introduce evolutionary channels that do not exist in isolated-star models (Eldridge et al., 2020). A plausible implication is that BPASS is best understood not as a minor extension of classical synthesis, but as a modeling paradigm in which interacting binaries are treated as constitutive components of stellar populations.