1 Introduction

The massive stars in the Milky Way are not randomly distributed, but are concentrated in loose groups called OB associations located in the spiral arms of our galaxy (for a review, see e.g. Brown et al. 1999). About 80% of the O stars are member of an OB association; the kinematical properties of the remaining 20% of the field population suggest that these O stars are runaways, i.e. they were born in an OB association, but at a certain stage they escaped from it (Blaauw 1993). The two most popular scenarios to explain the existence of runaway stars are (i) the dynamical ejection from a young cluster (Poveda et al. 1967) and (ii) the supernova of the companion star in a massive binary (Blaauw 1961). A recent study by Hoogerwerf et al. (2000) based on Hipparcos data demonstrates that both scenarios are at work, probably at a rate of 1:2, respectively.

High-mass X-ray binaries (HMXBs) are the descendants of massive binaries (Van den Heuvel & Heise 1972). A neutron star or a black hole, the compact remnant of the initially most massive star (the primary) in the binary system, produces X-rays due to the accretion of matter from the secondary (an OB supergiant or a Be star); see Kaper (1998) for an overview of the OB-supergiant systems. The binary system remains bound after the supernova, if less than 50% of the total system mass is lost during the (assumed symmetric) explosion (Blaauw 1961; Boersma 1961). The latter can be understood if one considers the phase of mass transfer occuring when the primary becomes larger than its critical Roche lobe (e.g. at the end of core-hydrogen burning when the star expands to become a supergiant) and matter flows from the primary to the secondary. This results in a change of the mass ratio from larger to smaller than one. A kick exerted on the compact object due to the eventual asymmetry of the supernova explosion has also to be taken into account when determining whether the binary breaks up or remains bound after the supernova.

Figure 1: O- and B-type stars selected from the Hipparcos catalogue in the field of HD 153919/4U1700-37 (filled circle). The confirmed members of Sco OB1 are shown as filled diamonds. The plus symbols correspond to OB stars with a parallax larger than 1 mas (i.e. distance smaller than 1 kpc) and were eliminated from the membership analysis. The open circles indicate stars with a (photometric) distance within the range of Sco OB1; some of them, with a proper motion similar to the confirmed association members, are indicated by a circle with central dot (cf. Fig. 2). The latter close to Sco OB1 might be members as well

According to the binary-supernova scenario all HMXBs should be runaways. Gies & Bolton (1986) did not find observational evidence supporting this hypothesis on the basis of radial-velocity measurements, though Van Oijen (1989) found strong indications that HMXBs are high-velocity objects. Based on pre-Hipparcos proper motion measurements, Van Rensbergen et al. (1996) suggested that the HMXB Vela X-1 is a runaway system produced by the supernova scenario, and that it originates in the OB association Vel OB1. The discovery of a wind-bow shock around Vela X-1 showed that this system indeed is running through interstellar space with a supersonic velocity, proving the runaway nature of this HMXB (Kaper et al. 1997). The Hipparcos proper motions of a dozen HMXBs (Chevalier & Ilovaisky 1998; Kaper et al. 1999) finally demonstrated that, as expected, likely all HMXBs are runaways. The most massive systems (those hosting an OB supergiant) have a mean peculiar (i.e. with respect to their standard of rest) tangential velocity of about 40 km s^-1, whereas the Be/X-ray binaries have on average lower velocities (about 15 km s^-1). This difference in velocity is consistent with the predictions of binary evolution (Van den Heuvel et al. 2000).

The identification of the "parent'' OB association of a HMXB is important, because it provides unique constraints on the evolution of high-mass X-ray binaries. When the system's proper motion and parent OB association are known, its kinematical age can be derived. The kinematical age marks the time of the supernova that produced the compact X-ray source. The distance of a HMXB usually is quite uncertain (and required to calculate its space velocity), but the distance to an OB association can be determined with better accuracy. The space velocity relates to the amount of mass lost from the system during the supernova explosion (cf. Nelemans et al. 1999). The age of the parent OB association should be equal to the age of the binary system. Consequently, the turn-off mass at the time of supernova yields the initial mass of the primary. Thus, this relatively straigthforward observation can be used to determine the age of the system, the time of supernova of the primary, the initial mass of the primary, and the amount of mass lost from the system during the supernova. Combining this information allows one to put constraints on the initial orbital parameters of the progenitor of the HMXB and on the evolutionary history of the system.

Figure 2: The Hipparcos proper motions of the OB stars shown in Fig. 1. The filled diamonds represent the confirmed members of Sco OB1; these stars cluster both in location and in proper motion. The plusses and open circles indicate the OB stars we could, or could not exclude on the basis of a distance estimate, respectively. The circles with a central dot are stars with similar proper motion and photometric distance as the confirmed cluster members. Also shown is 4U1700-37 (filled circle) which obviously has a proper motion different from that of Sco OB1

Figure 3: The reconstructed path of the runaway HMXB 4U1700-37 intersects with the location of Sco OB1; the error cone is indicated by the dotted straight lines. The Hipparcos confirmed members are shown as filled diamonds. The dotted line marks the region studied by Perry et al. (1991), including the young open cluster NGC 6231 (box). The proper motion of 4U1700-37 is with respect to the average proper motion of Sco OB1. The corresponding kinematical age of 4U1700-37 is $2 \pm 0.5$ million year. The current angular separation between 4U1700-37 and NGC 6231 (at 2 kpc) corresponds to a distance of about 150 pc

Here we apply this to the system HD 153919/4U1700-37. HD 153919 (m_V=6.6) is the O6.5 Iaf+ companion to 4U1700-37, most likely a neutron star powered by wind accretion (Jones et al. 1973; Haberl et al. 1989), although no X-ray pulsations have been detected (Gottwald et al. 1986). According to Brown et al. (1996) 4U1700-37 is a good candidate for a low-mass black hole. HD 153919 is the hottest OB companion star known in a HMXB; therefore, the progenitor of 4U1700-37 potentially is a very massive star. Chevalier & Ilovaisky (1998) showed that the Hipparcos proper motion of HD 153919 (5 mas yr^-1) corresponds to a peculiar tangential motion of 57 km s^-1 for an adopted distance of 1.7 kpc (Bolton & Herbst 1976), which proves the runaway nature of the system.

In the following we will use the Hipparcos data of OB-type stars in the Sco-Cen region to search for the parent OB association of 4U1700-37. The result will be used to reconstruct the evolutionary history of the system.