The youngest X-ray binaries

X-ray binaries (XRBs) usually consist of an accreting neutron star (NS) or black hole (BH) and an optical companion/donor star (Nagase 1989). According to the masses M_C of the companion stars, XRBs can be mainly divided into high-mass X-ray binaries (HMXBs) where M_C ≳ 8 M_⊙ and low-mass X-ray binaries (LMXBs) where M_C ≲ 1 M_⊙. HMXBs can be further classified into supergiant X-ray binaries (SGXRBs), Wolf-Rayet X-ray binaries (WRXRBs), and Be/X-ray binaries (BeXRBs) (see Reig 2011, for a review). Except for a few close binaries where mass transfer occurs via Roche-lobe overflow (RLOF) of the companion star, in most HMXBs the compact star accretes by capturing the wind material from the massive companion star. In the case of SGXRBs and WRXRBs, the stellar winds are traditionally thought to be radially expanding (probably in the form of clumpy winds), while in BeXRBs the donor stars are fast-rotating B-type stars and the winds are in the form of a circumstellar decretion disk.

Based on the main-sequence (MS) and post-main-sequence (PMS) lifetimes of the companion star, LMXBs and HMXBs have typical ages of ∼ 10⁹ – 10¹⁰ yr and ∼ 10⁶ – 10⁷ yr, respectively. These objects provide unique astrophysical laboratories for studies of stellar evolution, accretion processes and astrophysics of compact stars. HMXBs are especially at the key transitional stage between young massive binaries and double degenerate binaries, which are the important sources of gravitational wave radiation.

Due to the strong interaction between the compact stars and the accreting material, the properties of compact stars such as mass, spin, and magnetic field may have been significantly changed, and thus it is difficult to trace their origins from their current status. Most NSs and BHs are produced by supernovae (SNe) of massive stars. Because supernova remnants (SNRs) are short-lived (with ages ≲ 10⁵ yr), studies of XRBs associated with SNRs provide valuable insights into the earliest stages of XRB evolution and accretion history. Fortunately there have been an increasing number of XRBs discovered to be associated with SNRs. In this paper we review the observational characteristics of these youngest XRBs and discuss their possible implications on the formation and evolution of compact stars.

Currently there are six XRBs that are (possibly) associated SNRs, including one BH XRB and five NS XRBs. Three of them are in the Milky Way (MW), two in the Small Magellanic Cloud (SMC) and one in the Large Magellanic Cloud (LMC). Table 1 summarizes the observational and derived parameters of these XRBs.

2.1. SS433

2.2. Cir X-1

2.3. SXP 1062

2.4. SXP 1323

2.5. MCSNR J0513–6724

2.6. CXOU J053600.0–673507

The discovery of NS XRB-SNR associations provides interesting information about the initial spin period and magnetic field of the NSs, as well as the interaction between young NSs and their companion stars. When the NS captures the envelope material from the companion star by gravity, the motion of the accreting material is also strongly influenced by the NS magnetic field and the particle wind. At the stopping radius the ram pressure of the accreting material is balanced by the magnetic/radiatioin pressure. If the magnetic pressure dominates, the stopping radius is called the magnetospheric radius, which is given by

$$\begin{eqnarray}&&{R}_{{\rm{m}}}=\xi {\left(\displaystyle \frac{{\mu }^{2}}{\dot{M}\sqrt{2GM}}\right)}^{2/7},\end{eqnarray} \tag{ 1 }$$

where μ = BR³, M, and $\dot{M}$ are the magnetic moment, mass, and mass accretion rate of the NS, respectively. Here G is the gravitational constant, B the surface magnetic field, R the radius of the NS, and ξ a dimensionless factor of order unity. A necessary condition for steady accretion onto the NS is that the magnetospheric radius is smaller than the corotation radius where the local Keplerian angular velocity of the accreting material equals the angular velocity Ω of the NS, i.e.,

$$\begin{eqnarray}&&{R}_{{\rm{co}}}\equiv {\left(\displaystyle \frac{GM}{{\Omega }^{2}}\right)}^{1/3}.\end{eqnarray} \tag{ 2 }$$

Accordingly the spin period P of the NS should be longer than the equilibrium period

$$\begin{eqnarray}&&{P}_{{\rm{eq}}}={2}^{11/14}\pi {\xi }^{3/2}{\mu }^{6/7}{\dot{M}}^{-3/7}{(GM)}^{-5/7},\end{eqnarray} \tag{ 3 }$$

or

$$\begin{eqnarray}&&B\le 8.5\times {10}^{10}({\xi }_{0.5}^{-9/7}{M}_{1.4}^{1/3}{R}_{6}^{-5/2}){P}^{7/6}{L}_{35}^{1/2}{\rm{G}},\end{eqnarray} \tag{ 4 }$$

where ξ_0.5 = ξ/0.5, M_1.4 = M/1.4 M_⊙, R₆ = R/10⁶ cm, and L₃₅ = L_X/10³⁵ erg s⁻¹. For MCSNR J0513–6724, assuming a correction factor of two for the bolometric luminosity and adopting the spin period of 4.4 s, one can derive the upper limit of the NS surface magnetic field strength to be (2 – 5) × 10¹¹ G (see also Maitra et al. 2019). In the case of Cir X-1, the presence of type I X-ray bursts (Tennant et al. 1986; Linares et al. 2010) indicates that the NS possesses a relatively low magnetic field strength (< 10¹² G) (Joss & Li 1980; Fujimoto et al. 1981). At the observed luminosity of Cir X-1, the emission radius of the blackbody is small enough to be associated with the accretion hot spot as the X-ray emitting region, which also points to a field strength below 10¹² G (Schulz et al. 2020). Thus these two objects could belong to the group of so-called "anti-magnetars" (young NSs born with a weak dipole field), originally discovered in the compact central objects (CCOs) in SNRs, like 1E 1207.4–5209 in G296.5+10.0 (Zavlin et al. 2000; Gotthelf & Halpern 2007), CXOU J185238.6+004020 in Kes 79 (Haberl 2007), and RX J0822–4300 in Puppis A (Gotthelf & Halpern 2009). An interesting scenario for these objects suggests that rapid fallback accretion after the SN explosion(Chevalier 1989) may bury the magnetic field of the NS and decrease its surface strengths (Bernal et al. 2013; Torres-Forné et al. 2016). The buried field then gradually reemerges through Hall drift and Ohmic dissipation when accretion stops (Geppert et al. 1999; Ho 2011; Pons et al. 2013; Viganò et al. 2013; Gourgouliatos & Cumming 2015). The typical timescale of the reemergence is ≳ 10³ yr (Ho 2011; Gourgouliatos & Cumming 2015), depending on the burial depth.

A newborn NS is believed to rapidly rotating. The relativistic winds from the rotation-powered NS can prevent the companion's matter from penetrating the light cylinderical radius or the Bondi gravitational radius, so the NS usually first appears as a radio pulsar and spins down by magnetic dipole radiation. Once the plasma enters the light cylindrical radius, the pulsar mechanism switches off and the incoming matter forms a quasi-static atmosphere surrounding the NS (Davies et al. 1979). Since R_m > R_co in this situation, the infalling material is stopped at the magnetosphere by the centrifugal barrier, which prevents material from accreting onto the NS. The ejected material carries away the angular momentum of the NS and decelerate its spin by the propeller mechanism (Illarionov & Sunyaev 1975). The system finally switches on as an accreting XRB when the NS spin period has reached P_eq, where R_m = R_co. Thus, to reach the current spin period the NS should have already experienced the ejector and propeller phases. Most X-ray pulsars in HMXBs have a spin period of tens or hundreds of seconds with an age in excess of 10⁶ yr. In contrast, the spin periods of SXP 1062 and SXP 1323 are longer (∼ 10³ s), but their ages are much shorter (∼ 10⁴ yr). So it is interesting to ask how the NSs can spin down to such a long period within such a short time.

Haberl et al. (2012) argued that, if an NS has a normal magnetic field (B ∼ 10¹² – 10¹³ G), it is difficult to spin-down to a period ∼ 1000 s within a few 10⁴ years. Assuming that SXP 1062 has spun down with a constant rate $\dot{P}=100$ s yr⁻¹ as observed over its whole lifetime, they derived the lower limit of the initial spin period of the NS to be 0.5 s. Assuming that SXP 1062 is spinning at the equilibrium period, Popov & Turolla (2012) derived the current NS magnetic field to be ≲ 10¹³ G. To reconcile the long spin period and short age of SXP 1062, they suggested that the NS magnetic field must be in excess of 10¹⁴ G in the past and then decayed to its present, normal value. Thus SXP 1062 may be a unique example of evolved magnetars in HMXBs. Fu & Li (2012) explored the possible evolutionary tracks of SXP 1062, taking account of various initial parameters and spin-down mechanisms of the NS. They showed that the current magnetic field of the NS may be ≳ 10¹⁴ G. As such, SXP 1062 would be an accreting magnetar. González-Galán et al. (2018) estimated the magnetic field for SXP 1062 adopting the quasi-spherical accretion model of Shakura et al. (2012), and found no need for an extremely high magnetic field at present day for SXP 1062. However, it is noted that in the model of Shakura et al. (2012) the estimation of the NS B-field is sensitively dependent on a few parameters whose magnitudes are uncertain. Additionally, the Shakura et al. (2012) model was constructed for NSs accreting from spherically expanding winds in HMXBs, while in the case of SXP 1062 the winds from the Be star is in the form of a decretion disk, and it is unknown whether the model applies in that case. For example, the timing analysis by Serim et al. (2017) shows that SXP 1062 has a long-term secular steady spin-down that could be a result of a steady disk accretion. Using the standard accretion theory, Serim et al. (2017) obtained a magnetar-like surface magnetic field estimation that is consistent with the estimations by Fu & Li (2012)

If SXP 1062 is or has been a magnetar, so is SXP 1323. Thus, we see the diversity of youngest NSs in the small sample of youngest XRBs ² Considering the fact that young isolated NSs can manifest themselves as magnetars, high-B pulsars, and CCOs, some of NS sub-populations could be connected with similar evolutionary paths but with different initial field configurations in terms of the magnetothermal evolution model (Kaspi 2010; Popov et al. 2010; Pons et al. 2013; Viganò et al. 2013; Gullón et al. 2015). While the magnetothermal evolutionary model is mainly applicable for NSs with high magnetic fields (≳ 10¹³ G), to unify NSs with both low and high surface fields one may need to invoke the field burial and reemergence model (Viganò & Pons 2012; Heinz et al. 2015; Popov et al. 2015; Rogers & Safi-Harb 2016; Liu & Li 2019).

The mass transfer in XRBs occurs through two ways: capture of the wind from the companion star and RLOF. In the former case, to ensure efficient mass accretion onto the compact star requires both high wind mass loss rate and short orbital separation. This is why most HMXBs possess a supergiant donor star in close orbit or a Be star in long and eccentric orbit. While MCSNR J0513–6724, SXP 1062, and SXP 1323 readily fit the BeXRB picture, the luminous X-ray emission and jet activity in SS433 and Cir X-1 clearly indicate rapid mass transfer caused by RLOF. It is then interesting to ask how the companion star can fill its RL shortly after the SN for the latter systems.

Although there have been extensive investigations on the formation and evolution of various types of XRBs, to our knowledge, Han & Li (2020) and Xing (2020) are the only theoretical works that investigate the formation of RLOF XRBs within the SNR's lifetime, which focus on SS433 and Cir X-1, respectively.

Assuming that the compact object in SS433 is a BH, Han & Li (2020) employed a binary population synthesis (BPS) method ³ to study the formation of SS433. After evolving a large number of primordial binaries, they searched the BH binaries in the BPS results whose XRB phase starts within 10⁵ yr since the birth of the BH. Other selection criteria include: (1) the mean mass transfer rate > 10⁻⁸ M_⊙ yr⁻¹, (2) the companion mass is initially less than 50 M_⊙, and (3) the orbital period at the beginning of RLOF is less than 1 yr. It is found that the properties of the selected binaries span a wide range and can be divided into three groups according to the evolutionary stage of the companion star: (1) BH binaries with an MS companion, (2) BH binaries with a Hertzsprung gap (HG) companion, and (3) BH binaries with a core helium burning (CHeB) star or helium main-sequence (HeMS) star companion. The overall birthrate of the BH binaries is a few times 10⁻⁶ yr⁻¹. If one sets more stringent constraints for SS433 such as the time the companion spent to fill its RL after the BHs birth is less than 10⁴ yr, the orbital period is shorter than 20 days at the end of the RLOF, and the duration of the RLOF phase is longer than 10³ yr, then only BH/HG binaries are left, and the birthrate decreases to ∼ 10⁻⁷ yr⁻¹. Detailed mass transfer calculation shows that at the moment of SN, both the companion mass and the orbital period were larger than the current values. The companion star evolved rapidly within the lifetime of the SNR and transferred mass to the BH, which causes significant mass loss and orbital shrink. Considering the fact that the birthrate of BH binaries in the Galaxy is a few 10⁻⁵ – 10⁻⁴ yr⁻¹ (Shao & Li 2018) and there is only one BH XRB (i.e., Cygnus X-1) discovered in the Galaxy, the very low birthrate of SS433 suggests that either we are extremely lucky to observe such a rare object or the actual age of SS433 is much longer than 10⁵ yr.

The case of Cir X-1 is probably more complicated because the spectral type of the optical counterpart is highly uncertain. Xing (2020) also investigated its formation with a BPS method, taking into account asymmetric SN kick imparted on the newborn NS. Due to mass loss and the SN kick, the binary orbits are likely to be eccentric after the SN. By selecting binaries where RLOF initiates within 10⁵ yr at periastron after the SN, he found that the birthrate of NS XRBs with an MS donor star is ∼ (5 – 10) × 10⁻⁶ yr⁻¹. Thus, if taking the upper limit of the SNR ages to be 10⁵ yr, there could be at most one such source in the MW. The birthrate for a supergiant donor is smaller, ∼ (1 – 4) × 10⁻⁶ yr⁻¹, because of the much shorter lifetime of supergiants. If one considers the donor star to be a Be star and assumes that RLOF takes place at periastron when the Be star disk is larger than its RL, then the birthrate of these binaries can be obtained to be ∼ (5 – 10) × 10⁻⁵ yr⁻¹. It clearly demonstrates that the BeXRB model is most likely to account for the characteristics of Cir X-1 (see also Schulz et al. 2020).

Figure 1 illustrates the possible evolutionary tracks for an SGXRB (left) and a BeXRB (right) that starts emitting X-rays within the lifetime of the SNR. Here we assume that the compact star is an NS. Before it accretes from the companion star and becomes an X-ray source, the NS experiences a phase of radio pulsar because the stellar wind was originally relatively weak and/or the NS was initially rapidly rotating.

Zoom In Zoom Out Reset image size

BHs and NSs are born in core-collapse of massive stars, and are naturally expected to be found inside or nearby SNRs. While there are more than 120 isolated pulsars thought to be associated with Galactic SNRs (Ferrand & Safi-Harb 2012) ⁴ , we have only five or six cases for accreting compact binaries. Their very low birthrate implies that the total number in the MW and MCs is limited. Nevertheless, this small group of youngest XRBs can provide valuable clues to the initial condition of newborn compact stars, their interaction with environments, and the formation channels of compact star binaries. However, there are quite a few critical issues needed to be addressed, such as the mass of the compact star and the nature of surrounding bubble for SS433, the evolutionary state of the companion star of Cir X-1, the magnetic field strengths of SXP 1062 and SXP 1323, etc. More careful multi-wavelength observations are needed to set stringent constraints on the stellar and binary parameters and differentiate between models for their formation and evolution.

This work was supported by the National Key Research and Development Program of China (2016YFA0400803), the National Natural Science Foundation of China under grant No. 11773015, and Project U1838201 supported by NSFC and CAS.