3 On the evolutionary history of HD 153919/4U1700-37

Our analysis shows that HD 153919/4U1700-37 originates in the OB association Sco OB1, from which it escaped about 2 Myr ago due to the supernova of 4U1700-37's progenitor. At the time of the (assumed symmetric) supernova explosion less than half of the total system mass was lost from the system, as the system remained bound. The amount of mass lost during the supernova explosion ( $\triangle M$) can be estimated from the current space velocity $v_{\rm sys}$ of the system. For a circular pre-supernova orbit and a symmetric supernova explosion, Nelemans et al. (1999) derive the following relation between $\triangle M$ and $v_{\rm sys}$:

$$\left( \frac{\triangle M}{M_{\odot}} \right) = \left( \frac{v... ...ac{1}{3}} \left( \frac{M+m}{M_{\odot}} \right)^{\frac{5}{3}},$$

where M is the present mass of HD 153919, m the mass of 4U1700-37, and $P_{\rm cir}$ the orbital period after re-circularization of the orbit due to tidal dissipation. The current orbital period of the system is 3.41 day and there is no indication that the orbit is non-circular. As the X-ray source is not pulsating, only the radial-velocity orbit of the O supergiant can be measured, so that the masses of both stars are not uniquely determined. Heap & Corcoran (1992) propose $M= 52 \pm 2$ $M_{\odot}$ (i.e. a mass corresponding to its spectral type) and $m = 1.8 \pm 0.4$ $M_{\odot}$; Rubin et al. (1996) argue that M= 30⁺¹¹_-7 $M_{\odot}$and m = 2.6^+2.3_-1.4 $M_{\odot}$. For a space velocity of 75 km s^-1, $\triangle M$ becomes 8 $M_{\odot}$ or 6 $M_{\odot}$ for the solution of Heap & Corcoran and Rubin et al., respectively. Therefore, the mass of the star that exploded was about 9 $M_{\odot}$. This is significantly higher than model calculations by e.g. Wellstein & Langer (1999) predict.

What can be said about the initial mass of 4U1700-37's progenitor? Given its origin in Sco OB1, the system should have the same age as the association. As discussed in Sect. 2.4, there likely is some spread in age within the association, but the observations indicate that at the moment of the supernova Sco OB1 was not older than $6 \pm 2$ Myr. The corresponding turn-off mass is $\geq 30^{+30}_{-10}$ $M_{\odot}$ (Schaller et al. 1992). Following Iben & Tutukov (1985) (and case B mass transfer), the initial mass of a star that will explode as a 9 $M_{\odot}$ star is 25 $M_{\odot}$. Although Iben & Tutukov do not take into account the mass lost by the helium star, this result is consistent with our estimate of the progenitor mass based on the age of Sco OB1. If we take the initial mass of the primary to be 30 $M_{\odot}$, the initial mass of the secondary must have been less, and thus the present mass of HD 153919 cannot be higher than about 60 $M_{\odot}$(i.e. conservative mass transfer).

It is difficult to reconstruct the evolution of the massive binary before the supernova explosion. The main problem is the current short orbital period of the system. Applying Eqs. (4) and (5) in Nelemans et al. (1999), the orbital period before the supernova was a bit longer, about 4 days. In such a close binary it might well be that the primary starts transfering mass when it is still on the main sequence (case A mass transfer), but then one would predict a relatively large increase of the orbital period in case of conservative mass transfer (cf. Wellstein & Langer 1999). It might be that the evolution has been highly non-conservative due to strong stellar-wind mass loss and/or non-conservative Roche-lobe overflow. Case B mass transfer followed by a contact phase could produce systems like the Wolf-Rayet binary CQ Cep/HD 214419 (e.g. Marchenko et al. 1995) with an orbital period of 1.64 day. The latter system shows that in principle a short-period system like 4U1700-37's conjectured pre-supernova configuration can be produced (cf. Van den Heuvel 1973). A non-conservative evolutionary scenario for 4U1700-37 is also suggested by Wellstein & Langer (1999).