Deep optical imaging of the gamma-ray pulsar J1048-5832 with the VLT
Andrey Danilenko, Aida Kirichenko, Jesper Sollerman, Yury Shibanov, Dima Zyuzin
aa r X i v : . [ a s t r o - ph . S R ] F e b Astronomy&Astrophysicsmanuscript no. 1048.v17˙arxiv c (cid:13)
ESO 2018January 27, 2018
Deep optical imaging of the γ -ray pulsar J1048 − ⋆ A. Danilenko , A. Kirichenko , , J. Sollerman , Yu. Shibanov , , and D. Zyuzin , Io ff e Physical Technical Institute, Politekhnicheskaya 26, St. Petersburg, 194021, [email protected] ff e.ru, [email protected] ff e.ru St. Petersburg State Polytechnical Univ., Politekhnicheskaya 29, St. Petersburg, 195251, [email protected], [email protected] The Oskar Klein Centre, Department of Astronomy, Stockholm University, AlbaNova, 106 91 Stockholm, SwedenPreprint online version: January 27, 2018
ABSTRACT
Context.
PSR J1048 − γ -rays with Fermi , and also in X-rays with
Chandra and
XMM-Newton . It powers a compact pulsar wind nebula visible in X-rays and is in many ways similar to the Vela pulsar.
Aims.
We present deep optical observations with the ESO Very Large Telescope to search for optical counterparts of the pulsar andits nebula and to explore their multi-wavelength emission properties.
Methods.
The data were obtained in V and R bands and compared with archival data in other spectral domains. Results.
We do not detect the pulsar in the optical and derive informative upper limits of R & . m V & . m A V ≈ Chandra
X-ray data and compare the dereddened upper limits with the unabsorbed X-ray spectrum of the pulsar. We find thatregarding its optical-X-ray spectral properties this γ -ray pulsar is not distinct from other pulsars detected in both ranges. However,like the Vela pulsar, it is very ine ffi cient in the optical and X-rays. Among a dozen optical sources overlapping with the pulsar X-raynebula we find one with V ≈ . m R ≈ . m
3, whose colour is slightly bluer then that of the field stars and consistent with thepeculiar colours typical for pulsar nebula features. It positionally coincides with a relatively bright feature of the pulsar X-ray nebula,resembling the Crab wisp and locating in ∼ ′′ from the pulsar. We suggest this source as a counterpart candidate to the feature. Conclusions.
Based on the substantial interstellar extinction towards the pulsar and its optical ine ffi ciency, further optical studiesshould be carried out at longer wavelengths. Key words. pulsars: general – SNRs, pulsars, pulsar wind nebulae, individual: PSR J1048 −
1. Introduction
Rotation-powered pulsars are believed to be the most numerousof all γ -ray sources in the Galaxy. Nevertheless, only about tenpulsars were discovered in γ -rays until the launch of the FermiGamma-ray Space Telescope (Thompson 2008). The number of γ -ray pulsars has now become about ten times larger, includ-ing many which have not yet been identified in the radio range .The majority of the early known γ -ray pulsars are also identi-fied in the optical and X-rays. This provides a unique possibilityto compile multi-wavelength spectra and light curves for theseobjects for the study of the not yet clearly understood radiativemechanisms responsible for the pulsar emission. Only six suchmulti-wavelength objects are detected in the optical. Increasingthe number of optically identified γ -ray pulsars is highly desir-able and the Fermi discoveries open up a new window for that.The first
Fermi catalogue (Abdo et al. 2010) provided pooraccuracy (several arc-minutes) of the source spatial localisa-tion. Therefore only those new γ -ray pulsars whose coordi-nates were known with higher accuracy from radio and / or X-ray observations were suitable for optical counterparts searches.Here we present optical followup observations of one of the ⋆ Based on observations made with ESO telescope at the ParanalObservatory under Programs 384.D-0386(A) and 386.D-0585(A). see https: // confluence.slac.stanford.edu / display / GLAMCOG / Public + List + of + LAT-Detected + Gamma-Ray + Pulsars ”radio-selected”
Fermi -pulsars, PSR J1048 − − × erg s − , discov-ered in the radio by Johnston et al. (1992) and later in X-rays byGonzalez et al. (2006). It was mentioned as a low significance γ -ray pulsar in the Third EGRET catalogue (Thompson 2008)but was confirmed with Fermi (Abdo et al. 2010). The disper-sion measure (DM) of 129 pc cm − implies a distance ≈ V and R bands. Mignani et al.(2011) reported their non-detection of an optical counterpartbased on our V -band data only. Here we analyse the data in bothbands and examine colours of optical sources located close to thepulsar and within its PWN extent. We confirm the non-detectionin the V band and derive deep optical flux upper limits for thepulsar in both bands within its 1 σ position uncertainty in X-rays.The observations and data reduction are described in Sect. 2, ourresults are presented in Sect. 3 and discussed in Sect. 4.
1. Danilenko et al.: Deep optical imaging of PSR J1048 − Table 1.
Log of the VLT / FORS2 observations ofPSR J1048 − Date Band Exposure Airmass Seeing[s] [arcsec]2010-01-09 V × V × V × V × V × V × R × R ×
13 1.47 0.5–0.92011-01-04 R × R ×
2. Observations and data reduction
The pulsar field was imaged in the V HIGH and R S PECIAL bands with the FOcal Reducer and low dispersion Spectrograph(FORS2 ) at the VLT / UT1 (ANTU) during several service moderuns in 2010, and 2011 (see Table 1). The observations wereperformed with the Standard Resolution collimator providing apixel size of 0 . ′′
25 (2 × . ′ × . ′
8. Setsof twelve- and ten-minute dithered exposures were obtained inthe V and R bands, respectively. Three short, 15 sec, exposureswere taken in the V band to minimise the number of saturatedsources in the crowded pulsar field and were used for astrome-try. The observing conditions were photometric during the runs,with seeing varying from 0 . ′′ . ′′ IRAF and
MIDAS tools.We then aligned and combined all individual frames in each ofthe bands, using a set of unsaturated stars. The alignment accu-racy was < ∼ . ′′
63 and 0 . ′′
64, and the integration times were 24 and 16.8 ks,for the combined V and R images, respectively. For astrometric referencing the shallow V band (Table 1) im-ages were used. The positions of the astrometric standards fromthe USNO-B1 astrometric catalogue were used as a reference.To minimise uncertainties caused by overlapping stellar profilesin the crowded FOV, we selected only 15 isolated non-saturatedstars. Their pixel coordinates were derived using the IRAF taskimcenter with an accuracy of < ∼ IRAF taskccmap, allowing for the image scaling, shift, and rotation, wasapplied for the astrometric transformation of the image. Formalrms uncertainties of the astrometric fit were ∆ RA < ∼ . ′′
056 and ∆ Dec < ∼ . ′′ < ∼ . ′′
17, consistent withthe nominal catalogue uncertainty of ≈ . ′′
2. The combined deep V and R images were aligned to the short V reference frame withan accuracy of < ∼ . ′′
01. The resulting conservative 1 σ referenc-ing uncertainty for the combined images is < ∼ . ′′ For instrument details see http: // / instruments / fors / see http: // / data / fchpix / tion archival X-ray image of PSR J1048 − Chandra / ACIS . In the exposure-corrected ACIS-S3 chip im-age, where the pulsar is located, we found a dozen of point-likeobjects detected at > ∼ σ significance. We identified them withrelatively bright optical reference objects from the USNO-B1catalogue. Their image positions were defined using the CIAO celldetect tool with an accuracy of 0.5–3.0 of the ACIS pixelsize ( ≈ . ′′ ∆ RA ≈ . ′′
424 and ∆ Dec ≈ . ′′
22 with maximal resid-uals < ∼ . ′′
83 and < ∼ . ′′
54 in RA and Dec, respectively. Combiningthe latter with the catalogue uncertainties, conservative 1 σ X-rayimage astrometric uncertainties are ∆ RA < ∼ . ′′
85 and ∆ Dec < ∼ . ′′
58. The shift between the raw and transformed images wasinsignificant, ∼ . ′′ = = − σ -error ellipse of the source position are 0 . ′′ . ′′ . ◦ . ′′
92 and 0 . ′′
64 in RA and Dec, respectively. Combining themwith the optical referencing uncertainty we obtained the RA andDec radii of the 1 σ -error ellipse of the pulsar X-ray position onthe optical images as 0 . ′′
94 and 0 . ′′
67, respectively.
The photometric calibration was carried out using standard starsfrom the photometric standard fields E3, NGC2298, NGC2437,and PG1525 (Stetson 2000) observed during the same nightsas the target. We fixed the atmospheric extinction coe ffi cientsat their mean values adopted from the VLT home page: k V = . m ± . m
01 and k R = . m ± . m
01. The resulting magnitudezero-points for the combined images were V ZP = . m ± . m R ZP = . m ± . m
02, and colour-term coe ffi cients in respec-tive photometric equations are 0.15 ± − ± ffi cient uncertainties, and marginal variations from night tonight.
3. Results
The ∼ ′ × ′ VLT FOV in the V band (leftpanel of Fig. 1) showsa complicated structure of the pulsar environment with a darkfeature extended over the entire field. This is also seen in the R band and is fully consistent with a large scale structure inthe H α archival image (right panel of Fig. 1). The pulsar isnear the middle of its eastern part. In the Spitzer archival im-ages at 8 and 24 µ m the dark part is filled with bright in-frared emission, likely a signature of a warm dust, which istypical for star forming regions. Another neutron star, RRATJ1047 −
58 is located ∼ ′ from the pulsar. Its distance ≈ − . ◦ Obs. ID 3842, date 2003.10.08, Exp. time 36 ks, PI V. Kaspi. see e.g. “A User’s Guide to Stellar CCD Photometry with IRAF”by P. Massey and L. Davis, http: // iraf.net / irafdocs / Obtained with the CTIO 4-m telescope as a part of ChaMPlane sur-vey (Grindlay et al. 2005) GLIMPSE project, PI S. Majewski.2. Danilenko et al.: Deep optical imaging of PSR J1048 − -6.52e+039.77e+021.01e+031.02e+031.04e+031.05e+031.07e+031.10e+031.13e+031.22e+032.19e+05 - : : . : . : . D ec V : . - : : . : . : . : . H-alpha
J1047-5841
Fig. 1.
The field of PSR J1048 − V band with the VLT (left), and in H α with the CTIO (right). The dark horizontallines in the VLT image are due to a gap between the two FORS2 / CCD chips and dithered exposures. The box in the H α image showsthe VLT FOV. The red and black crosses are the positions of J1048 − − - : : . . . RA D ec R a d i o I n t er f . R a d i o t i m i n g V LT , V
10 98 76 - : : . . . RA D ec V LT , R
10 98 76 − R22232425262728 V Fig. 2.
The ∼ . ′′ × . ′′ fragment of the VLT V (top-left) and R (top-middle) band optical, and Chandra − Chandra image issmoothed with a four pixel Gaussian kernel. Colour-bars show bright-ness scales in 1000 counts in the optical and in counts in X-rays.The cross marks the X-ray position of the pulsar. White-dashed, andblack-dashed ellipses show its 1 σ uncertainties in X-rays and tworadio observations, respectively. Yellow contours in the R -band im-age are overlaid from the X-ray image to indicate the PWN structure.Optical sources overlapping with the error ellipses and X-ray PWNare labelled by numbers. Top-right: The observed colour-magnitudediagram of PSR J1048 − ∼ − − In Fig. 2 we compare images of the pulsar vicinity in the VR bands and X-rays. The X-ray image is corrected for the ACIS exposure map and smoothed with a four pixel Gaussian kernelto better show the PWN shape. The X-ray position of the pul-sar and its 1 σ uncertainty are marked together with the radiointerferometric (Stappers et al. 1999) and timing (Wang et al.2000) ones. The X-ray position is in a good agreement with theinterferometric one. This fact and our accurate X-ray astrometric Instrument ATCA, date 13-05-1997, epoch 50581, RA = = − Instrument Parkes, date 25-02-1993–29-03-1997, epoch 49043–50536, RA = = − − referencing allow us to use the X-ray position as a reliable ref-erence point in searching for the pulsar counterpart. The timingposition is shifted significantly and likely su ff ers from system-atic errors. The contours of the brightest regions and the outerboundary of the X-ray PWN are overlaid on the R image.We do not resolve any reliable optical source within the pul-sar 1 σ X-ray error ellipse, but there are several optical sourcesoverlapping with the PWN and radio ellipses. They are labelledby numbers and can be considered as potential counterparts ofthe pulsar, if it has a high proper motion, or its PWN structures.
Pulsars and PWNe typically have peculiar colours as comparedto stars. To investigate whether the marked sources are associ-ated with the pulsar and / or its PWN we performed Point SpreadFunction (PSF) photometry using the psf and allstar tasksof the IRAF DAOPHOT package (Stetson 1987). We set the psf-radius at ten pixels, where the bright isolated unsaturated starsselected for the PSF-construction merged with the background.The fit-radius and aperture radius for the PSF stars and the pre-liminary aperture photometry of the other stars were 2.5 pix-els, while the annulus / dannulus for local background extractionswere 15 /
10 pixels. We made aperture corrections based on pho-tometry of bright unsaturated isolated field stars. The derived
Table 2.
Magnitudes and fluxes of the optical sources labelledin Fig. 2 and PSR J1048 − σ uncertainties referring to the last significant dig-its quoted. Stars 6 and 7 were labelled by D and C respectivelyin Mignani et al. (2011) Star V (mag) R (mag) log F V ( µ Jy) log F R ( µ Jy)1 26.6(1) 25.22(6) − − − − − − − − − D ) 26.1(1) 24.67(4) − − C ) 23.92(2) 22.77(1) − − − − − > ∼ > ∼ < ∼ − < ∼ − σ detection-limits for a point-like object for a half-arcsecondaperture centred at the pulsar X-ray position are V up ≈ . m R up ≈ . m
1, accounting for the aperture corrections. Magnitudesof the sources marked in Fig. 2 and the above upper limits arecollected in Table 2 where uncertainties include the measure-ment and calibration errors. The magnitudes were transformedinto flux densities using zero-points from Fukugita et al. (1995).Comparing the V magnitudes of stars 6 and 7 to those esti-mated by Mignani et al. (2011; stars D and C in their notations),we find that for star 7 ( C ), overlapping with the pulsar timingposition, their estimate, ∼ . m , is consistent with ours, 23 . m ± D ) is about 0 . m allstar task was also used for photom-etry of field stars and construction of the colour-magnitude dia-gram (Fig. 2). To exclude unresolved blends, partially resolvedgalaxies, saturated stars, and stars incorrectly cross-identified inboth bands, we selected only the stars with the allstar output parameters satisfying the following conditions: χ < ∼ < ∼
1; position di ff erences in V and R bands < ∼ < ∼ . m V − R = ± V − R < ∼ / PWNe optical counterparts, which are usually de-tected as faint blue objects. For instance, V − R is 0.4 forthe Crab pulsar (Percival et al. 1993) and 0.7 for its PWNknot (Sandberg & Sollerman 2009). An association with thepoint-like pulsar is, however, excluded due to the large o ff set,1 . ′′ ± . ′′
7, from the pulsar X-ray position. At d = ± − on the 7 yr time-base between the Chandra and VLT observations. The absence of any significant shift onthe 6 yr time-base between the interferometric and X-ray posi-tions excludes such a motion.
In Fig. 3 we compare the non-smoothed X-ray image of the pul-sar vicinity with that in the R band where background stars havebeen subtracted except for source 5. There is a 4- σ significancecompact X-ray structure within the PWN about 2 . ′′ ∼ . m < ∼ V < ∼ . m
4, consistent withthe source 5 brightness, is ∼ − . The respec-tive confusion probability to find an unrelated point-like opticalsource within the 90% Chandra positional uncertainty ellipse ofthe putative wisp is ∼
2% and it becomes considerably smaller, ∼ V − R ) < ∼ σ R flux enhancement seenwithin the pulsar X-ray error ellipse may indicate the presenceof a faint pulsar counterpart candidate, but is consistent witha background fluctuation and the pulsar upper limits derivedabove. It is useful to compare the pulsar optical upper limits with its X-ray spectral data. Gonzalez et al. (2006) reported a spectral anal-ysis of available
Chandra and
XMM-Newton data but only forthe combined emission of the pulsar + PWN system. To examine
4. Danilenko et al.: Deep optical imaging of PSR J1048 − c oun t s X-rays 0.5 -7 keV psr wisp ? c oun t s / optical R-band psr wisp ?source 5 Fig. 3.
The non-smoothed
Chandra image (top) and star-subtracted VLT image (bottom). Magenta contours are X-ray contours ofthe PWN from Fig. 2. The cross and dashed ellipse are the position of the pulsar and its 1 σ uncertainty. The yellow contour marksa relatively bright X-ray structure of the PWN 2 ′′ southeast of the pulsar, presumably a Crab-like wisp. It spatially coincides withthe optical source 5, as can also be seen from the spatial brightness profiles extracted from a slice along the PWN major axis (awhite-dashed rectangular on both images) presented in the leftpanels, where vertical dotted lines indicate the X-ray positions of thepulsar and the wisp. Error-bars indicate typical brightness uncertainties.the X-ray spectrum of the pulsar itself we performed an inde-pendent analysis. We first reanalysed the pulsar + PWN spectrumusing the data from both instruments and our results are fullyconsistent with the published ones. The spectrum is describedby an absorbed power-law, whereas the blackbody model givesan unrealistically high neutron star temperature.We then focused on the
Chandra / ACIS data where the pul-sar is spatially resolved from the PWN (Fig. 3). We used threeapertures shown in Fig. 4 to extract the spectra of the pulsar, thepulsar + PWN system, and the south-east tail of the PWN. Thenumbers of source counts were 71 ±
9, representing > ∼
80% ofthe emission from the point-like pulsar, 176 ±
5, and 50 ± ′′ circularaperture located ∼ ′′ north-east of the pulsar in a region freefrom any sources. We fitted the absorbed power-law model tothe extracted unbinned spectra in the 0.5–10.0 keV range us-ing the Xspec v.12.7.1 (Arnaud 1996) and C-statistics (Cash1979, Wachter et al. 1979). Along with C values the fit qualitieswere estimated using the goodness task . The best fit param-eters, C values, and energy bin numbers (nbins) are presented inTable 3. The pulsar + PWN fit is in agreement with that obtainedby Gonzalez et al. (2006). The
Xspec goodness task simulates data with given response filesand model. The fit is good when about 50% of the simulated spectrahave the value of C less than that of the data in question.
There is a noticeable increase of the absorbing column den-sity N H when we move from the point-like pulsar to the extendedPWN. This increase is significant for the south-east tail, whichoverlaps with the optically dark region in Fig. 2 and, therefore,is most strongly absorbed also in X-rays. The pulsar and north-western part of the PWN are in a more transparent region and N H is lower for the pulsar and has an intermediate value for theentire system. We therefore believe that the N H derived from the1 . ′′ + PWNspectrum.
Table 3.
The absorbing column density N H , photon index Γ , andnormalisation factor PL of absorbing power-laws describing the Chandra spectra (see Fig. 4 and text). Errors correspond to ∆ C = N H Γ PL norm C (nbins)10 cm − − ph cm − s − keV − pulsar0 . + . − . . + . − . . + . − .
245 (649)pulsar + PWN0 . + . − . . + . − . . + . − .
448 (649)south-east tail of the PWN1 . + . − . . + . − . . + . − .
267 (649) 5. Danilenko et al.: Deep optical imaging of PSR J1048 − : . - : : . . . . Right ascension D ec li n a ti on pulsar 1".5south-east part of the PWNpulsar+PWN ACIS-S
Fig. 4.
The fragment of the
Chandra / ACIS image of thePSR J1048 − Distance [ kpc ] A V DM distance entire galactic extinction N H = 0.3 X 10 cm −2 Fig. 5.
The A V –distance relation towards PSR J1048 − A V resulting from the uncertainty intervalof N H and its best-fit value for the pulsar presented in Table 3,respectively. The entire galactic extinction is indicated by thedash-dotted line. The next step is dereddening the optical data. N H values fromTable 3 and a standard relation between N H and the extinc-tion factor A V (Predehl & Schmitt 1995), yield A V of 3.4 + . − . ,10.6 + . − . , and 1.8 + . − . for the pulsar + PWN, PWN tail, and pul-sar, respectively. The first value is nearly equal to the entireGalactic extinction of 3.6 mag along the pulsar line-of-sight(Schlegel et al. 1998). The second one is significantly larger, inagreement with the absence of any stars in the dark region over-lapping with the tail. The pulsar itself is apparently less red-dened. To verify that, we made independent A V estimates using amethod based on the red-clump stars as standard candles, whichprovides an A V –distance relation for a given position on the sky(see e.g. L´opez-Corredoira et al. 2002).We extracted stars located within 0.3 ◦ of the pulsar positionfrom the 2MASS All-Sky Point Source Catalogue , and createda colour-magnitude diagram, K vs J − K . We found mean J − K See http: // irsa.ipac.caltech.edu / applications / DataTag / , DataTag = ADS / IRSA.Gator / /
14 15 16 17 18 19
Log ν [ Hz ] -4-3-2-10 L og F ν [ µ J y ] Power-law, α = 0.46 +0.34−0.35 A V = 1.8 N H = 0.3 X 10 cm −2 pulsar Fig. 6.
Unabsorbed multi-wavelength spectrum ofPSR J1048 − Chandra and VLT. The X-ray part with its uncertainties(hatched regions) is extrapolated to the optical. Black trianglesare optical flux upper limits.colours of the red-clump branch in several magnitude bins andtransformed them to the A V –distance relation, as has been doneby Danilenko et al. (2012) for another γ -ray pulsar J1357 − A V ≈
2, which is consistent with the X-ray spectral fit(the horizontal solid line) and with a mean foreground A V ≈ A V is obviouslylarger (cf. Povich et al. 2011), but these have a negligible con-tribution to the derived relation dominated by more transparentregions, in one of which the pulsar is located.We thus consider A V ≈ N H ≈ × cm − , and d ≈ − The unabsorbed multi-wavelength spectrum of the pulsar isshown in Fig. 6. Its X-ray part is obtained with N H frozen at thevalue of 3 × cm − , and the optical data are dereddened with A V = = L psrX ≈ . erg s − , and the pulsarto pulsar + PWN X-ray luminosity ratio ≈ ≈ . erg s − and 0.34, respectively, obtained for theVela pulsar, which has a similar age and spin-down luminosity.Comparing this together with our constraints of optical luminos-ity L opt < ∼ . erg s − and the e ffi ciency of transformation of itsspin-down luminosity ˙ E to optical photons L opt / ˙ E < ∼ − . withthe data available for other pulsars detected in the optical and X-rays (Zharikov et al. 2004, Danilenko et al. 2012), we concludethat, as the Vela pulsar, PSR J1048 − ffi cient inthe optical and X-rays (Fig. 7).
6. Danilenko et al.: Deep optical imaging of PSR J1048 − L og L X [ e r g s − ] Log τ (years) L og L O p t [ e r g s − ] CrabJ0540
Vela
B0656 Geminga B1929 B0950J1357J1124B1509 B1055 J0108B1133
J1048 -6-4-2 L og η X Log τ (years) -8-6-4 L og η O p t Fig. 7.
Comparison of X-ray and V -band luminosities and e ffi ciencies for pulsars of di ff erent characteristic age τ detected in bothspectral domains. Di ff erent pulsars are marked by di ff erent symbols and PSR J1048 − ffi ciencies with age demonstrate an e ffi ciency minimum near the Vela age, and the Vela-like J1048 −
4. Discussion and conclusions
We did not detect PSR J1048 − ∼
28. The pulsar islocated in a complicated region linked to the northern edge ofthe Carina complex with high star formation (and supernova)activity (Fig. 1). The region is filled by clumpy clouds where theinterstellar extinction and absorbing column density vary sub-stantially even at a 10 ′′ scale. This complicates the optical iden-tification of the pulsar and its PWN. Nevertheless, the derivedpulsar optical flux upper limits are quite informative.First, our VLT observations and reanalysis of the X-ray datashow (Fig. 6) that the optical fluxes of the pulsar do not exceedthe extrapolation of its power law X-ray spectrum towards theoptical range, which is compatible with multi-wavelength non-thermal spectra for other rotation powered pulsars.Second, our results show that this Vela-like pulsar is veryine ffi cient in the optical and X-rays, as is the Vela pulsar itself(Fig. 7). In combination with a significant interstellar extinctiontowards the pulsar, A V ≈
2, this precluded us to detect it in ourdeep observations at a level consistent with the optical e ffi ciencyof the Vela pulsar.We did detect a faint optical source coinciding with a rela-tively bright compact feature of the pulsar X-ray PWN, presum-ably a wisp, located ∼ ′′ from the pulsar (Fig. 3). The sourcecolour V − R ≈ ff ected by the inter-stellar extinction.Recent progress in the multi-wavelength studies of Vela-likepulsars adds new evidence of their low e ffi ciency in the opticaland X-rays. This forms a puzzling minimum in the optical andX-ray e ffi ciency relations vs pulsar age, while in γ -rays it ap-pears to be absent (Abdo et al. 2010). Further studies will showwhether this is a signature of some interesting changes in a neu- tron star magnetosphere and particle acceleration at the 10 kyrage, or just an incomplete sample e ff ect, which disappear whenmore Vela-like pulsars will be detected in the optical and X-rays. Acknowledgements.
We are grateful to anonymous referee for useful com-ments allowing us to improve the paper. The work was partially supported bythe Russian Foundation for Basic Research (grants 11-02-00253 and 11-02-12082), RF Presidential Program (Grant NSh 4035.2012.2), and the Ministry ofEducation and Science of the Russian Federation (Contract No. 11.G34.31.0001and Agreement No.8409, 2012).
Note added: After submission of this paper also Razzano et al. (2013,MNRAS, 428, 3636) appeared. It is based on our data and their conclu-sion that star 6 cannot be the optical counterpart to the pulsar is consis-tent with our results while their pulsar flux upper limits are less deeper.
References
Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010, ApJS, 187, 460Arnaud, K. A. 1996, in Astronomical Society of the Pacific Conference Series,Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H.Jacoby & J. Barnes, 17Cash, W. 1979, ApJ, 228, 939Danilenko, A., Kirichenko, A., Mennickent, R. E., et al. 2012, A&A, 540, A28Fukugita, M., Shimasaku, K., & Ichikawa, T. 1995, PASP, 107, 945Gonzalez, M. E., Kaspi, V. M., Pivovaro ff , M. J., & Gaensler, B. M. 2006, ApJ,652, 569Grindlay, J. E., Hong, J., Zhao, P., et al. 2005, ApJ, 635, 920Hillier, D. J., Davidson, K., Ishibashi, K., & Gull, T. 2001, ApJ, 553, 837Johnston, S., Lyne, A. G., Manchester, R. N., et al. 1992, MNRAS, 255, 401Keane, E. F., Kramer, M., Lyne, A. G., Stappers, B. W., & McLaughlin, M. A.2011, MNRAS, 415, 3065L´opez-Corredoira, M., Cabrera-Lavers, A., Garz´on, F., & Hammersley, P. L.2002, A&A, 394, 883Mignani, R. P., Shearer, A., de Luca, A., et al. 2011, A&A, 533, A101Percival, J. W., Biggs, J. D., Dolan, J. F., et al. 1993, ApJ, 407, 276Pires, A. M., Motch, C., Turolla, R., et al. 2012, A&A, 544, A17Povich, M. S., Smith, N., Majewski, S. R., et al. 2011, ApJS, 194, 14Predehl, P. & Schmitt, J. H. M. M. 1995, A&A, 293, 889Sandberg, A. & Sollerman, J. 2009, A&A, 504, 525Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525Shibanov, Y. A., Koptsevich, A. B., Sollerman, J., & Lundqvist, P. 2003, A&A,406, 645Shibanov, Y. A., Zharikov, S. V., Komarova, V. N., et al. 2006, A&A, 448, 313Smith, N. 2006, ApJ, 644, 1151