The eclipsing millisecond pulsar PSR J1740 5340 and its red straggler companion

We present a high-resolution echelle spectrum taken with the Very Large Telescope and analyse archival Hubble Space Telescope photometry of the recently identified companion of the eclipsing millisecond radio pulsar PSR J1740-5340 in the globular cluster NGC 6397. From the spectrum, we show that the companion is metal poor, as expected for a member of NGC 6397. Using synthetic photometry, and assuming a true distance modulus of 12.13 +/- 0.15 mag and a colour excess of EB-V = 0.179, we derive a radius of 1.60 +/- 0.17 R-. and an effective temperature of 5410 +/- 50 K, implying a luminosity of 2.0 +/- 0.4 L-. (1sigma errors). These properties make it similar to the so-called "sub-subgiants" in the old open cluster M67 and "red stragglers" in 47 Tuc, which have luminosities comparable to those of turn-off stars, but cooler temperatures and larger radii. The light curve of the companion is well described by ellipsoidal variations, and despite the incomplete (similar to 60%) phase coverage, we are able to derive good constraints on a number of the system parameters. In particular, for the inclination we find a 2-sigma lower limit of i > 48degrees (i > 46degrees at 3sigma). Assuming a pulsar mass of 1.2 < M-MSP < 2.4 M-., this implies a companion mass in the range 0.14 <M-comp < 0.38 M-.. Combined with the photometric constraint, we find a best fit for i similar or equal to 50degrees and M-comp similar or equal to 0.3 M-.. We infer a Roche lobe filling factor by radius of similar to97%. Surprisingly, we find no evidence whatsoever for irradiation of the companion, despite the high inferred rotational energy loss of the pulsar ((E) over dot ' 1.4 x 10(35) erg s(-1)). We discuss possible reasons, but find most lacking. We hypothesise that the system is a triple, and that the acceleration due to a third body in a wide orbit around the binary led to an overestimate of the intrinsic spin-down rate and hence the spin-down luminosity. This can be tested by further timing observations. We also discuss two other puzzles, viz., the system's location far outside the cluster core and the companion's large radius and luminosity. We suggest that the system was formed in a binary-binary encounter in the core, due to which the system acquired a substantial velocity, and the companion - which must have been a somewhat evolved turn-off star - lost much of its envelope. We suggest other "red stragglers" and "sub-subgiants" might have formed by similarly drastic encounters.