Initial-Final Mass Relation (IFMR)

• Initial-Final Mass Relation (IFMR) is the mapping between a star's birth mass and its white-dwarf remnant mass, encapsulating critical mass-loss and dredge-up processes.
• Empirical studies from open clusters, wide binaries, and field populations reveal a nonlinear IFMR with distinct kinks and steep segments, challenging simple linear models.
• The IFMR informs stellar evolution models by constraining physical processes like TP-AGB mass loss, rotational mixing, and overshooting, thereby refining age estimates and population synthesis.

Searching arXiv for recent and foundational papers on the white-dwarf Initial–Final Mass Relation. The Initial–Final Mass Relation (IFMR) is the empirical and theoretical mapping between a star’s zero-age main-sequence mass, $M_{\rm initial}$ or $M_i$, and the mass of the compact remnant it leaves, most commonly a white dwarf with mass $M_{\rm final}$ or $M_f$. In the white-dwarf context, the IFMR quantifies the net outcome of mass loss over the red giant branch and asymptotic giant branch, and thereby encodes late-stage core growth, dredge-up, and envelope ejection. It is foundational for modeling white-dwarf mass distributions, stellar feedback and return fractions in chemical evolution, asymptotic-giant-branch mass-loss prescriptions, stellar population synthesis, the white-dwarf luminosity function, and the age-dating of stellar populations (Cummings et al., 2015). Modern work shows that the IFMR is not adequately described by a single global linear law: cluster studies, field white-dwarf population analyses, and wide double-white-dwarf inference all point to structure in the relation, including a steep segment near $M_i \sim 3$–$4\,M_\odot$, a low-mass kink near $M_i \sim 1.65$–$2.10\,M_\odot$, and likely regime changes associated with second dredge-up and carbon-star evolution (Cummings et al., 2018, Marigo et al., 2020, Hollands et al., 2023).

1. Definition and astrophysical role

In its standard white-dwarf formulation, the IFMR links the birth mass of a star to the mass of the white-dwarf remnant it leaves after completing single-star evolution. Operationally, semi-empirical determinations use either cluster chronometry or coeval binary chronometry to infer the progenitor lifetime, then convert that lifetime to $M_i$ using stellar evolution tracks, while $M_f$ is obtained from atmospheric fits and white-dwarf cooling models (Cummings et al., 2015, Andrews et al., 2014).

The importance of the IFMR extends beyond remnant demographics. Because the returned mass is $M_i$0, the relation directly affects chemical-evolution return fractions, stellar population synthesis, the normalization of massive white-dwarf tails, and applications in white-dwarf cooling-age dating (Cummings et al., 2015). A steeper local IFMR implies that intermediate-mass progenitors near $M_i$1 produce more massive white dwarfs than predicted by broader linear fits, while progenitors near $M_i$2 produce slightly less massive remnants, altering inferred mass return to the interstellar medium (Cummings et al., 2015).

The same conceptual framework is now used more broadly for neutron stars and black holes, where the IFMR becomes a mapping from initial stellar properties to remnant type and remnant mass. In that regime, the relation is probabilistic and depends on metallicity, explosion physics, fallback, rotation, and binarity (Lu et al., 2019). The white-dwarf IFMR therefore serves both as an empirical benchmark and as a methodological precedent for compact-remnant population inference more generally.

2. Semi-empirical construction from open clusters

Open clusters remain the most direct route to an empirical white-dwarf IFMR because cluster ages provide a common clock. For a cluster white dwarf, the progenitor lifetime is inferred from

$M_i$3

or equivalently, in some formulations, $M_i$4 (Cummings et al., 2015, Barnett et al., 2021). The observed white dwarf is spectroscopically fitted to obtain $M_i$5 and $M_i$6, these are mapped to $M_i$7 and $M_i$8 via white-dwarf cooling sequences, and the progenitor lifetime is inverted through stellar-evolution models to recover $M_i$9 (Cummings et al., 2015, Cummings et al., 2018).

A representative example is the NGC 2099 study, which used Keck/LRIS spectroscopy for 20 white-dwarf candidates. Nineteen were bona fide white dwarfs, 14 had sufficiently high S/N for reliable masses and cooling ages, and 11 were consistent with singly evolved cluster membership after photometric and spectroscopic vetting (Cummings et al., 2015). With an adopted cluster age of $M_{\rm final}$0 Myr, $M_{\rm final}$1, $M_{\rm final}$2, and solar metallicity, the member sample spanned $M_{\rm final}$3–$M_{\rm final}$4, with the main locus at $M_{\rm final}$5–$M_{\rm final}$6 once a likely low-mass outlier was excluded, corresponding to progenitors of $M_{\rm final}$7–$M_{\rm final}$8 (Cummings et al., 2015).

The spectroscopic procedure in such cluster work is highly standardized. In NGC 2099, five Balmer lines, H$M_{\rm final}$9, H$M_f$0, H$M_f$1, H$M_f$2, and H8, were fitted simultaneously with DA atmosphere models of Tremblay et al. (2011), using Levenberg–Marquardt $M_f$3 minimization; external uncertainties of $M_f$4 in $M_f$5 and $M_f$6 dex in $M_f$7 were adopted (Cummings et al., 2015). Masses and cooling ages were then derived from Montreal CO-core thick-H-layer cooling models (Cummings et al., 2015). A similar self-consistent program across 13 clusters and Sirius B yielded a directly measured IFMR spanning $M_f$8 to $M_f$9 in progenitor mass with a scatter of $M_i \sim 3$0 (Cummings et al., 2018).

A central result of this cluster-based literature is that model consistency matters. When the same evolutionary model family is used both to derive the cluster age and to map progenitor lifetimes back to $M_i \sim 3$1, the inferred IFMR is relatively insensitive to the adopted model below the highest masses; mismatched chronometry and progenitor-lifetime models can introduce spurious structure (Cummings et al., 2015, Cummings et al., 2018). This is one reason recent cluster analyses emphasize homogeneous atmospheres, cooling tracks, reddening treatment, and isochrone families.

3. Empirical shape: nonlinearity, steep segments, and kinks

A major development in IFMR studies has been the move away from a single global linear fit. In the $M_i \sim 3$2–$M_i \sim 3$3 regime, NGC 2099 alone yielded

$M_i \sim 3$4

while the combined NGC 2099 + Hyades + Praesepe sample, totaling 29 white dwarfs, gave

$M_i \sim 3$5

significantly steeper than widely used broader-range linear relations such as the Kalirai et al. (2009) form, $M_i \sim 3$6 (Cummings et al., 2015). This steep segment implies curvature in the global IFMR and is consistent with a turnover near $M_i \sim 3$7 associated with second dredge-up (Cummings et al., 2015).

A larger cluster synthesis spanning $M_i \sim 3$8–$M_i \sim 3$9 formalized this nonlinearity as a three-piece relation. Using the preferred MIST-based calibration, the low-, intermediate-, and high-mass segments were

$4\,M_\odot$0

$4\,M_\odot$1

$4\,M_\odot$2

thereby making explicit a steep intermediate segment bracketed by shallower slopes (Cummings et al., 2018). The same study interpreted the slope change near $4\,M_\odot$3–$4\,M_\odot$4 as a signature of second dredge-up and altered TP-AGB core growth (Cummings et al., 2018).

At lower initial masses, the IFMR may become non-monotonic. Cluster white dwarfs in NGC 752, Ruprecht 147, M67, NGC 6819, NGC 7789, and NGC 6121 were used to identify a kink across $4\,M_\odot$5, with a local maximum $4\,M_\odot$6–$4\,M_\odot$7 at $4\,M_\odot$8–$4\,M_\odot$9 (Marigo et al., 2020). The study argued that a kinked three-piece relation fit the data markedly better than a single weighted linear fit and interpreted the feature as the signature of carbon-star formation near the transition between degenerate and non-degenerate helium ignition (Marigo et al., 2020).

Field-white-dwarf analyses independently support nonlinearity. A Gaia DR2 generative-model fit to 1101 bright DA white dwarfs within 100 pc found that the best-fit IFMR flattened over $M_i \sim 1.65$0, producing a secondary peak in the white-dwarf mass distribution near $M_i \sim 1.65$1 (El-Badry et al., 2018). More recent 40 pc work derived a four-piece segmented relation with a knee near $M_i \sim 1.65$2, although that study imposed monotonicity by construction (Cunningham et al., 2023). By contrast, a recent Gaia DR3 open-cluster compilation expressed the preferred Gaia-based cluster IFMR above $M_i \sim 1.65$3 as

$M_i \sim 1.65$4

again emphasizing a slope change near the second-dredge-up regime (Miller et al., 28 Oct 2025).

4. Alternative empirical routes: wide binaries and field populations

Open clusters are not the only route to the IFMR. Wide double white dwarfs provide a complementary coeval laboratory because both components form together yet evolve independently. Their age constraint is differential rather than absolute: the difference in pre-white-dwarf lifetimes equals the difference in cooling ages,

$M_i \sim 1.65$5

or, more explicitly,

$M_i \sim 1.65$6

This allows one white dwarf to set the system age information for the other without requiring an external cluster age (Andrews et al., 2015).

Early wide-DWD work expanded the SDSS sample to 142 candidate systems and developed the basic chronometric method, showing in the benchmark system PG 0922+162 that a $M_i \sim 1.65$7 white dwarf corresponded to $M_i \sim 1.65$8 (Andrews et al., 2014). A subsequent Bayesian hierarchical analysis used 19 high-fidelity DA+DA wide DWDs and a three-segment piecewise-linear model with pivots at $M_i \sim 1.65$9 and $2.10\,M_\odot$0, concluding that robust constraints existed in the $2.10\,M_\odot$1–$2.10\,M_\odot$2 range and that the resulting IFMR lay below commonly used open-cluster semi-empirical linear relations in that interval (Andrews et al., 2015).

Gaia-era samples greatly enlarged the DWD approach. Spectroscopy of 90 Gaia DWDs, with 52 DA+DA, DA+DC, and DC+DC pairs retained, yielded a monotonic piecewise-linear IFMR constrained for $2.10\,M_\odot$3–$2.10\,M_\odot$4, with the $2.10\,M_\odot$5–$2.10\,M_\odot$6 range mostly constrained to a precision of $2.10\,M_\odot$7 or better (Hollands et al., 2023). The posterior median node values included $2.10\,M_\odot$8 at $2.10\,M_\odot$9, $M_i$0 at $M_i$1, $M_i$2 at $M_i$3, and $M_i$4 at $M_i$5 (Hollands et al., 2023). That study also introduced a Bayesian mixture model for outliers, finding $M_i$6 and an additional outlier cooling-age uncertainty of $M_i$7 Gyr, with several high-mass systems interpreted as merger candidates (Hollands et al., 2023).

Field-white-dwarf population methods form a third route. The 100 pc Gaia DR2 DA sample was modeled through a CMD generative framework with a flexible piecewise-linear IFMR and yielded a flattening between $M_i$8 and $M_i$9 (El-Badry et al., 2018). A later spectroscopically complete 40 pc sample adopted population synthesis with merger removal and inferred a four-piece segmented linear IFMR over $M_f$0–$M_f$1, emphasizing that merger contamination above $M_f$2 must be modeled statistically (Cunningham et al., 2023). These field approaches do not depend on cluster ages, but they do depend on assumptions about the IMF, Galactic age distribution, metallicity distribution, spectral-type systematics, and the treatment of mergers.

5. Physical interpretation: mass loss, dredge-up, overshoot, and rotation

The IFMR is a condensed record of TP-AGB physics. At low masses, the balance between third dredge-up efficiency and winds is especially important. In TP-AGB models, the third dredge-up efficiency is parameterized as

$M_f$3

so that larger $M_f$4 suppresses net core growth and therefore lowers $M_f$5 (Marigo, 2022, Addari et al., 2024). A calibration against the observed solar-metallicity IFMR found that $M_f$6 must remain small, $M_f$7, for $M_f$8, rise steeply to $M_f$9 for $M_i$00, and then decline for $M_i$01 (Marigo, 2022). This mass dependence was proposed to explain both the low-mass kink and a possible second kink near the transition between the most massive carbon stars and hot-bottom-burning TP-AGB stars (Marigo, 2022).

A more detailed PARSEC+COLIBRI investigation showed that reproducing the observed kink at $M_i$02–$M_i$03 requires a mass-dependent envelope-overshooting parameter $M_i$04 and a small fixed PDCZ overshooting efficiency, $M_i$05 (Addari et al., 2024). In that framework, the final white-dwarf mass is identified with the core mass at the end of the TP-AGB, and the core is defined by

$M_i$06

The same work emphasized the role of carbon-star winds and the threshold carbon excess for dust-driven outflows, with $M_i$07, in producing a delayed envelope-loss phase and thus extra core growth in the kink regime (Addari et al., 2024).

At higher masses, convective-core overshoot and rotational mixing become more influential than TP-AGB mass-loss prescriptions. A study focused on reconciling theory with the semi-empirical IFMR argued that non-rotating models underpredict white-dwarf masses by $M_i$08 over $M_i$09–$M_i$10 (Cummings et al., 2019). Varying mass-loss rates by a factor of two and reducing third dredge-up to $M_i$11 had only modest effects at high mass, whereas rotational mixing and stronger convective-core overshoot substantially improved agreement (Cummings et al., 2019). Rotation affects the semi-empirical IFMR in two ways: it increases the true core mass and therefore $M_i$12, and it lengthens the progenitor lifetime, which biases the inferred $M_i$13 downward if non-rotating lifetimes are assumed (Cummings et al., 2019). This suggests that part of the observed scatter may trace unmodeled rotation rather than stochastic mass loss.

Metallicity effects appear weaker over moderate ranges in the white-dwarf IFMR than might be expected. NGC 2099, at approximately solar metallicity, and the metal-rich Hyades and Praesepe showed nearly identical $M_i$14–$M_i$15 slopes, with no meaningful offset (Cummings et al., 2015). The larger 13-cluster synthesis likewise found no detectable metallicity sensitivity across $M_i$16 (Cummings et al., 2018). By contrast, a study of 10 white dwarfs in wide binaries suggested that among progenitors with masses in the range of $M_i$17–$M_i$18, more metal-rich stars lose a larger fraction of their mass, implying a metallicity-linked contribution to scatter at the low-mass end (Zhao et al., 2011). A plausible implication is that metallicity effects may be regime dependent, modest in the well-sampled intermediate-mass cluster domain but more visible among low-mass progenitors where RGB contributions to total mass loss are more important.

6. Systematics, variants, and current points of contention

A persistent issue in IFMR work is whether apparent scatter is astrophysical or methodological. Single-cluster analyses have shown that, once low-S/N spectra, membership contamination, and model inconsistencies are removed, observational errors can account for most of the remaining spread in the $M_i$19–$M_i$20 regime, arguing against strongly stochastic mass loss there (Cummings et al., 2015). Uniform reanalyses of high-mass cluster white dwarfs likewise reduced the inferred scatter by roughly $M_i$21, again suggesting that heterogeneous methods had been a dominant contributor (Cummings et al., 2016).

Several variants of the IFMR remain actively discussed. One is atmospheric-composition dependence. A first cluster-based non-DA IFMR, built from six non-DA white dwarfs, suggested that non-DAs may be $M_i$22 less massive than DAs at fixed initial mass (Barnett et al., 2021). The same study, however, regarded this as provisional and likely attributable to small-number statistics, atmospheric-model systematics, and leverage from Procyon B rather than a secure physical offset (Barnett et al., 2021). Another recent analysis of intermediate-separation binaries showed that the isolated-star IFMR can fail outright in interacting systems: some $M_i$23 AU binaries contain low-mass white dwarfs around $M_i$24 that require binary stripping, while others host “spender” white dwarfs that lost more mass than expected from the canonical isolated-star IFMR (Ironi et al., 28 Jan 2025). These systems underscore that the standard IFMR is fundamentally a single-star relation.

There is also tension among different empirical strategies. The recent Gaia DR3 cluster compilation found significant deviations from IFMRs inferred from Gaia field-white-dwarf populations and from double-WD binaries, with the largest discrepancy near $M_i$25 (Miller et al., 28 Oct 2025). Because the cluster analysis explicitly restricted itself to spectroscopically confirmed DA white dwarfs with strong cluster associations and likely single-star origins, it argued that contamination by mergers, non-DAs, or membership ambiguities may underlie part of the discrepancy in the field and DWD determinations (Miller et al., 28 Oct 2025). This suggests that “the IFMR” is not yet a single settled function across all methods, but rather a relation whose exact form depends on how samples are cleaned and how non-single-star evolution is treated.

Finally, IFMR choice matters for downstream inference. A recent theoretical differential study quantified how semi-empirical white-dwarf IFMR choices propagate into cluster ages derived from white-dwarf luminosity functions (Salaris et al., 12 Jun 2026). The systematic age offset is at most $M_i$26 Gyr for old clusters, $M_i$27 Gyr for intermediate ages, and negligible at young ages; neglecting predicted metallicity dependence in old metal-poor globular clusters can underestimate ages by up to $M_i$28 Gyr (Salaris et al., 12 Jun 2026). This establishes the IFMR not only as a stellar-evolution diagnostic but also as an explicit systematic in white-dwarf chronometry.

7. Broader extensions and future directions

Current work points toward a more segmented and conditional view of the IFMR. In the white-dwarf regime, the most robust empirical picture is that the relation is nonlinear, with a moderate low-mass slope, a low-mass kink near $M_i$29–$M_i$30, a steep segment near $M_i$31–$M_i$32, and a shallower high-mass continuation above that (Cummings et al., 2018, Marigo et al., 2020, Miller et al., 28 Oct 2025). The exact morphology below about $M_i$33 remains one of the least secure parts of the relation because of sample limitations and sensitivity to TP-AGB physics (Miller et al., 28 Oct 2025).

Methodologically, progress is coming from larger homogeneous samples and from probabilistic inference frameworks. Hierarchical Bayesian modeling across star clusters has already shown that borrowing strength across clusters can suppress unphysical cluster-specific fits when the local progenitor-mass leverage is poor (Si et al., 2018). Gaia-selected wide DWD samples have brought the $M_i$34–$M_i$35 range into sharp focus, while explicit outlier models now allow departures from single-star evolution to be accommodated statistically rather than simply discarded (Hollands et al., 2023). Open-cluster work continues to benefit from Gaia astrometry, spectroscopic homogeneity, and updated cooling grids, producing the largest cluster-based IFMRs to date (Miller et al., 28 Oct 2025).

The IFMR concept is also being extended observationally beyond white dwarfs. For neutron stars and black holes, astrometric microlensing has been proposed as a route to directly weigh isolated remnants and thereby constrain the massive-star IFMR (Lu et al., 2019). Roman and extremely large telescopes are expected to identify long-duration, unblended dark-lens events and measure their centroid shifts, enabling first constraints on the present-day mass function, multiplicity, and natal kicks of isolated neutron stars and black holes (Lu et al., 2019). Related simulations indicate that different black-hole IFMR prescriptions alter the long-duration tail of the Einstein crossing-time distribution and the small-parallax tail of microlensing events, implying that future microlensing surveys could distinguish competing massive-star IFMRs if $M_i$36 is measured accurately (Rose et al., 2022).

Taken together, these developments suggest that the IFMR should no longer be treated as a universal straight line. It is instead a structured relation shaped by TP-AGB dredge-up, wind physics, second dredge-up, overshoot, rotation, metallicity, and contamination from binary evolution. The empirical challenge is now less the existence of the IFMR than the precise delineation of its regime changes, intrinsic scatter, and dependence on stellar subtype and evolutionary channel.