next up previous contents index
Next: Large and Small Up: StarsStellar Populations, Previous: Fundamental Properties of

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

GEMINGA as Seen by the Hubble Space Telescope

Patrizia A. Caraveo and Roberto Mignani
Istituto di Fisica Cosmica del CNR, Via Bassini, 15, 20133 Milano, ITALY



Cycle 4 observations of the optical counterpart of the X and -ray source GEMINGA yielded the measure of the source parallactic displacement as well as a more precise knowledge of its color distribution. This has definitely improved our understanding of this isolated neutron star, the first discovered and studied without the help of radio astronomy.

Keywords: Neutron Star


Geminga has been a High Priority target for the Hubble Space Telescope since the very beginning of the mission. FOC and WFPC observations G", suspected to be the optical counterpart of the peculiar X-ray source 1E0630+178, in turn presumed to be the counterpart of -ray source Geminga, were approved for Cycle 1. At the time, the outstading problem was to say a final word on the correctness of the optical identification through a study of the colors of G", with the hope to find a clear signature of its neutron star nature. Owing to the faintness of the target () the observations had to be deferred and, while waiting for the refurbishing mission, the multi-wavelength observational panorama on Geminga changed significantly.

First came the discovery of the 237 msec periodicity in the X and -ray data (Halpern & Holt, 1992, Bertsch et al. 1992, Bignami & Caraveo 1992). This secured the identification of 1E0630+178 with Geminga and provided, through timing, a clear idea of the energetics of the source, whose distance upper limit turned out to be 340 pc (Bertsch et al. 1992, Bignami & Caraveo 1992). This limit, in agreement with the distance inferred from the absorption of the soft X-ray data (Bignami, Caraveo, & Lamb 1983, Halpern & Ruderman 1993), was confirmed by the discovery of the proper motion of the suspected optical counterpart.

Bignami, Caraveo & Mereghetti (1993) comparing data acquired over a 8 year time-span, found G" to be moving at 170 marcsec/y. G" was indeed a low-luminosity, high proper motion object inside the small error box of an X--ray emitting, radio-quiet, neutron star. Thus, the proper motion was taken as evidence of G" being the optical counterpart of Geminga. When coupled with the average velocity of neutron stars, the measured proper motion pointed again to a nearby object, in the 100--200 pc range.

Distances of this order can be measured directly exploiting the source annual parallactic displacements, amounting to 0.01 for a distance of 100 pc. Obviously, to be able to assess this tiny effect, the source position must be known with milliarcsec precision. Ground-based measurements, while of sufficient precision and duration to fix the proper motion, cannot reliably determine such a milli-arcsec annual parallax, especially for a very faint object like G", and the refurbished Hubble Space Telescope was required.

When it came the time to write proposals for Cycle 4 , we wanted to use HST to address the parallax of Geminga together with its UV colors. Both requests were granted and successfully carried out. In the following we shall report on how the Space Telescope changed our views on Geminga.

The Parallactic Displacement

Owing to its near-equinoxial position, close to the ecliptic plane, the expected parallactic factor of G" is maximum close to the equinoxes, and it is almost all in right ascension. Of course, any annual parallactic displacement would appear, in the plane of the sky, as superimposed to the observed proper motion of 170masyr in the NE direction. Thus, the source will follow a wiggling path: the value eventually found for the parallax would determine the amplitude of the oscillation, while its phase would be fixed by the time of the year. Assuming an hypothetical distance of 100 pc, i.e., a parallactic displacement of 0.01, we can compute the expected trajectory: this is shown in Fig. 1. Note, in particular, that the RA component of the composite motion is ``asymmetric'' in time, in the sense that the March-to-September RA displacement is bigger than the September-to-March one. This is due to the RA component of the parallax (in fact twice its actual value) being added to the proper motion in the first case and subtracted in the second. Such asymmetry would represent a signature in the observational data. It also results in a difference, again mainly in the RA values, between the apparent total displacement of the object in the sky in the two semesters.

Figure: Combination of proper motion and parallax for a source at the Geminga coordinates for a distance of 100 pc. The small triangles mark the dates of maximum parallatic displacements.

In order to measure the predicted displacement, very precise relative astrometry is crucial. This requires both good angular resolution and a field of view large enough to contain a reasonable number of stars. The Faint Object Camera has excellent angular resolution, but its field of view is so tiny that no reference object is present in an exposure centered on G". The Planetary Camera (PC) on the WFPC2, with a pixel size of 0.0455 and a FOV of 3535, appears as the ideal instrument, allowing for the relative astrometry with the required accuracy. G" has been observed three times by the Planetary Camera on March 19, 1994, on September 23, 1994 and on March 18, 1995. All observations were performed with the filter F555W, roughly equivalent to the classical V filter, where ground-based observations had shown the source to be brightest (Bignami et al. 1987, Halpern & Tytler 1988, Bignami, Caraveo, & Paul 1988). After careful alignment of the images (see Caraveo et al. 1996 for a detailed description of the procedure), we compared the positional residuals of the centroids of all the objects detected in the three images.

While all field stars show residuals comfortably close to zero, the behavior of G" is quite different. The three relative PC positions, plotted in Fig. 2 in the plane of the sky, show the presence of both proper motion and parallactic displacement. The March 94--March 95 displacement is a direct measure of the proper motion, since it gives the displacement totalled in exactly one year, between two positions with identical parallactic factors. This turns out to be masyr (with a RA component of masyr, and a DEC one of masyr) i.e., fully consistent, but much more accurate, than the ground based-one. On the other hand, the September point is obviously displaced from the ``zero parallax'' line joining the two March points. As expected from Fig. 1, it is displaced in the sense of increasing right ascension. Moreover, the March 94--Sept 94 displacement is definitively bigger than the Sept 94--March 95 one.

When all corrections are taken into account (see Caraveo et al. 1996 for a complete discussion) there remain a difference of 0.6 0.1 pixel between the two segments of RA trajectory. Multiplied by the cosine of the declination, this is four times the parallax of Geminga. Using the actual numerical values of the time differences and the correct parallax factors, the end result is:

Figure: Total motion of Geminga as measured by the HST/WFPC2. The straight line corresponds to the pure proper motion, and the ondulating one to the combination of proper motion and parallax for a distance of 157 pc.

The Colors of Geminga

Our knowledge on the color of G" had not improved significantly since 1988, when the B color of the source was measured by Bignami, Caraveo & Paul (1988). The source R,V and B colors were found not to be compatible with a planckian distribution and well above the Rayleigh-Jeans extrapolation of the black-body best fitting the Einstein X-ray data. Although ROSAT data had lowered the best fitting temperature of the source (Halpern & Ruderman 1993), the perculiarity of the optical distribution remain unchanged.

Figure: Fluxes of Geminga as measured from ground-based (I,R,B,V) and HST (F675W,F555W,F342W) observations compared with the low-energy BB extrapolation of the soft X-ray spectrum. The corresponding and temperature combinations are also shown for different source radii and a distance of 157 pc.

In order to improve our knowledge on the source spectral shape, Geminga was observed with the FOC on Sept. 23, 94 through filter F342, roughly equivalent to U, for a total exposure time of 5000 sec. The source was clearly detected yielding a magnitude of 24.9 0.2. This is an important new piece of information to be combined with the three WFPC2 observations used for the parallax, and one more WFPC2 shorter observation taken through the F675 filter, equivalent to R.

Obviously, these data have to be compared with the flux expected on the basis of the extrapolation of higher energy spectral data. Apart from ROSAT, Geminga has also been detected with EUVE and, taking advantage of the distance measurement, one can use the data collected by the EUVE satellite to obtain an independent evaluation of the effective temperature of the NS surface (Bignami et al.,1996). There is little doubt that the temperature range suggested by the EUVE data is 2--, definetely lower than the best fitting value of of Halpern & Ruderman (1993). Although significantly different in the far UV/soft X-ray range, the Rayleigh-Jeans parts of the different planckians do substantially agree in the optical domain and can be compared with the colors measured for G". This was done in Fig. 3 (Bignami et al. 1996) where we have plotted (1) an upper limit in the ``I'' filter (6,000Å to 8,000Å) obtained in Feb. 1995 from the ESO NTT, (2) the R and V points, also ground based, obtained by now several times and fully compatible with those of Halpern & Tytler (1988), of (3) the difficult B point of Bignami et al. (1988), also from ESO, and (4) the three HST colors described above. The I-to-FOC spectral region, with the colors now available, does carry a lot of information on Geminga.

First, taken as a whole, it is apparent that the cluster of ground-based and HST observations are largely compatible with the Rayleigh-Jeans side of the planckians fitting the EUVE/ROSAT data. One important characteristic is however apparent in the I-to-FOC colors: they most definitely cannot be all fit by the planckians shown in Fig. 3. While the FOC and B points and the important I upper limit are well in agreement with the Rayleigh-Jeans slope and flux values, the R, V and WFPC2 points are not. In fact, it is quite apparent that no single monotonic law can fit all the Geminga colors.

The simplest interpretation, proposed in Bignami et al. (1996), is that of the presence of a spectral feature on the thermal continuum. This could either be an emission peak (centred around the V color) or an absorption trough, with a minimum around B. While reasonable (and interesting) interpretations of such a feature in terms of a magnetized neutron star are possible, the need for confirmation of the reality of this feature must be pointed out. Especially for the B point the error () reflects a measurement at the limit of the instrumental capability. On the other hand, the HST magnitude determinations are quite accurate, and the R and V points are the result of repeated ground based measurements.

More FOC observing time has been granted in Cycle 6 to confirm the B magnitude and to obtain a further UV color to better constrain the underlying planckian.


Whatever the nature of the Geminga spectral feature, it is interesting to note that we own to it the possibility to identify and study the optical emission of this radio-silent neutron star. No proper motion and no parallax measurements would have been possible if the optical flux were to be on the extrapolation of the planckian seen in soft X-rays.

However, Geminga is not going to be a unique case. PSR 0656+14, a radio pulsar with parameters very similar to those of Geminga, is showing a very similar behavior. As noted by Bignami et al. (1996) and confirmed by Pavlov et al. (1996), the combination of the FOC data with the ground based detection (Caraveo et al. 1994) reveal a maximum in the optical domain.


Bertsch, D.L., Brazier, K.T.S., Fichtel, C.E., Hartman, R.C., Hunter, S.D., et al. 1992, Nature, 357, 306

Bignami, G.F., Caraveo, P.A., & Lamb, R.C. 1983, ApJ, 272, L9

Bignami, G.F., Caraveo, P.A., Paul, J.A., Salotti, L., & Vigroux, L. 1987, ApJ, 319, 358

Bignami, G.F., Caraveo, P.A., & Paul, J.A. 1988 A&A, 202, L1

Bignami, G.F., & Caraveo, P.A. 1992, Nature, 357, 287

Bignami, G.F., Caraveo, P.A., & Mereghetti, S. 1993, Nature, 361, 704

Bignami, G.F., Caraveo, P.A., Mignani, R., Edelstein, J., & Bowyer, S. 1996, ApJL, 456, L111

Caraveo, P.A., Bignami, G.F., Mignani, R., & Taff, L.G. 1996, ApJL in press

Caraveo, P.A., Bignami, G.F. & Mereghetti, S. 1996, ApJL, 422, L87

Halpern, J.P., & Tytler, D. 1988, ApJ, 330, 201

Halpern, J.P., & Holt, S.S. 1992, Nature, 357, 222

Halpern, J.P., & Ruderman, M. 1993, ApJ, 415, 286

Pavlov, G., Stringfellow, G.S. & Cordova, F.A. 1996, ApJ in press

next up previous contents index
Next: Large and Small Up: StarsStellar Populations, Previous: Fundamental Properties of