MULTIWAVELENGTH STUDY OF THE BRIGHT X-RAY SOURCE POPULATION IN THE INTERACTING GALAXIES NGC 5774/NGC 5775

Chandra Advanced CCD Imaging Spectrometer (ACIS) observations of the NGC 5774/5775 system were carried out on 2002 April 5 for 58.2 ks (ObsID 2940, PI: Philip Maloney). A level 2 event list for this data set was retrieved from the Chandra archive. NGC 5774 is located on the front-illuminated CCD S2 and NGC 5775 is on the back-illuminated CCD S3. We use the locally developed software tool LEXTRCT (Tennant 2006) for data analysis, which is described in detail in Swartz et al. (2003). Source detections in LEXTRCT were performed using a circular Gaussian approximation to the point-spread function (PSF), which gives higher weight to sources with a central concentration of events. Point-source counts and spectra were extracted from within the 95% encircled-energy aperture of the model PSF. Background regions were typically chosen from annular regions surrounding the source regions except in crowded regions of the field where background regions adjacent to the source were used. The background-subtracted counts within the source regions were scaled to obtain the aperture-corrected count values. The background-subtracted point source detection limit is 14 counts for the 2.8 minimum signal-to-noise ratio (S/N) threshold and a minimum 5-σ above background. We define the total field as the 66 × 58 rectangle, in R.A. and decl., that inscribes both D₂₅ ellipses. Forty-nine sources were detected in this field. Their properties are listed in Table 1. The Chandra image of the NGC 5774/5775 field is shown in Figure 2 with the distribution of X-ray sources detected in the field. The ellipses shown delineate the D₂₅ isophotes of these two galaxies.

Zoom In Zoom Out Reset image size

Source and background spectra and light curves were used for spectral and timing analysis, respectively. X-ray light curves were binned into 1000 s bins. We computed the Kolmogorov–Smirnov (KS) statistic to determine the time variability of sources by comparing the cumulative event arrival times, binned at the 3.24 s frame time, to that expected for a steady source. The resulting KS probabilities are listed in Column 10 of Table 1. We also performed a χ² test against a constant flux hypothesis for each light curve and the results are presented in Column 11 of Table 1.

X-ray spectra of all the sources and the corresponding backgrounds were binned so that there were at least 10 counts per fitting bin. Spectral fit were made to sources with more than 40 detected source counts. Spectral redistribution matrices and ancillary response files were generated using Chandra X-ray Center software CIAO version 3.4. XSPEC version 11.3.2t was used to fit the 0.5–8.0 keV energy spectra with an absorbed (phabs) power-law (powerlaw) model. Best-fitting model parameters and fit statistics are listed in Columns 5–7 of Table 1. These spectra have also been fitted with other models as described in the sections on individual sources (Sections 3 and 4). The observed spectra, with best-fitting power-law models and fit residuals, are shown in Figure 3. Intrinsic luminosities of these sources based on these spectral fits are listed in Column 8 of Table 1. Intrinsic luminosities for sources with fewer than 40 source counts were estimated by allowing only the model normalization to vary in the fitting procedure (using appropriate response matrices and ancillary response files) and freezing the power-law photon index to 1.8 and the absorption column to the Galactic N_H value in the direction of NGC 5774/5775 (3.47 × 10²⁰ cm⁻²) in WebPIMMS.⁶ The photon index value of 1.8 is very close to the average value of the power-law index, for sources with statistically acceptable powerlaw fits measured in our catalog of ULXs (Swartz et al. 2004).

Zoom In Zoom Out Reset image size

A 47 ks XMM-Newton observation of NGC 5774/5775 was carried out on July 2003 (ObsID 0150350101). The data was processed using the XMM-Newton Science Analysis System (version 6.5.0). This observation suffered from background flares. After filtering these flares, only 22.6 ks of useful data remained. The XMM-Newton/MOS 1 image of NGC 5774/5775 is shown in Figure 4. Seven sources were detected in the entire field. Their X-ray properties, using the EPIC MOS and PN data, are listed in Table 2. Six of these were previously detected with Chandra (and they are numbered in Table 2 using the numbering scheme of Table 1). The one source detected only in the XMM-Newton data is listed as source #49 in Table 1. We did not perform variability studies on the XMM-Newton data due to the strong background flare activity. Absorbed powerlaw model fit parameters to the spectra of these sources are listed in Table 2. Sources were detected down to a detection limit of ∼7 × 10³⁸ erg s⁻¹ for 14 counts in a source above the background in the 0.2–10.0 keV band, for the EPIC MOS and 2.0 × 10³⁸ erg s⁻¹ in PN observations, respectively.

Zoom In Zoom Out Reset image size

For completeness, we analyzed archival ROSAT observations of NGC 5774/5775. A 6 ks PSPC observation was taken in 1993 July/August and a 35 ks HRI observation was taken in 1997 July. No sources were detected in the PSPC data and only two sources (#5 and #26 of Table 1) were detected in the HRI data. Count rates of these two sources were used to estimate their intrinsic fluxes and are listed in Table 2.

Photometric data of NGC 5774/5775 observed with the Hubble Space Telescope (HST), the European Southern Observatory's 3.5 m New Technology Telescope (NTT), and the Sloan Digital Sky Survey (SDSS) program were obtained from their respective archives.

A 640 s Hubble/WFPC2 observation of NGC 5774 was carried out on 2001 May 20 using the F814W filter. This observation covered only X-ray source s01 in Figure 2. Hubble astrometry was done using the USNO-B1.0 star 0935-0243054. Based on the absolute positional uncertainties of the USNO stars and the X-ray sources, we estimate that the astrometric accuracy was better than 070. No counterpart to source #1 was detected in the Hubble data.

Two Hubble/ACS–WFC observations of NGC 5775 were carried out on 2005 August 21 with F625W and F658N filters. Astrometry between Hubble/ACS–WFC and Chandra images were carried out using the source #26. Considering the centroiding uncertainties of these images, the estimated astrometric accuracies were in the range of 02–03.

NGC 5774/5775 was observed with the ESO/NTT in Hα and R-band filters on 1991 May 7. Multiple short observations were taken in each filter. The total integration time for the Hα filter (full-width at half-maximum = 75 Å) was 60 m, which reached to the rms noise level of 7 × 10⁻¹⁸ erg cm⁻² s⁻¹ arcsec⁻² (Lee et al. 2001). Photometric analysis of these images were carried out using IRAF⁷ software. Astrometry between the ESO/NTT and Chandra images was done using two sources (#5 and #26), which are bright in both the images.

SDSS photometric (u, g, r, i, and z bands) images of the field and spectroscopic observations of the nuclei of NGC 5774 and NGC 5775 were carried out on 2001 May 16. Astrometry was performed following the same procedure used for the ESO/NTT images. Astrometrically corrected SDSS images (25''× 25''), centered on the positions of the Chandra sources, are shown in Figure 5. The X-ray (0.5–8.0 keV) to optical (V-band) flux ratios (F_X/F_O) are given in Column 9 of Table 1. When no candidate optical counterpart to an X-ray source is detected, then the value of the V band has been assumed to be 25 mag and the corresponding lower limit values of F_X/F_O are listed in Table 1. The SDSS detection limit in the V band is 25 mag. For sources in confused region of NGC 5774/5775, no F_X/F_O is given.

Zoom In Zoom Out Reset image size

There are 12 X-ray sources with potential counterparts in the SDSS images. Their properties are listed in Table 3. Where objects are identified in the SDSS catalog, these magnitudes are quoted. For sources #9 and #26, magnitudes were estimated by fitting normalized Gaussian profiles to the spatial distribution of counts in each filter image. Most of these objects lie outside the D₂₅ isophotes of the galaxy pair. Three of the X-ray sources, #5, #23, and #26, are ULX candidates based on their X-ray flux and assuming they are at the distance of NGC 5774/5775. An optical color–color diagram including these 12 sources is displayed in Figure 6. Usually, the nominal photometric accuracy of SDSS images is around 0.03 mag for sources not limited by photon statistics (Ivezić et al. 2004). However, some of the 12 sources are relatively faint and we estimated that the photometric accuracies in (u − g) and (g − r) colors will be approximately between 0.10 and 1.0 mag.⁸ Five sources (sources #5, #6, #9, #15, and #26) are located in a small region near the center in Figure 6. As shown in Section 2.3, optical spectra of sources #5 and #26 show they are background QSOs. The (u − g) and (g − r) colors of these sources are consistent with being z < 3.0 SDSS-QSOs (Richards et al. 2001). We conclude all five sources are likely background QSOs with redshifts less than 3 but only have strong spectroscopic evidence for this classification for sources #5 and #26.

Zoom In Zoom Out Reset image size

Three objects located in the lower right hand region of Figure 6 could be globular clusters (GCs) based on their colors and absolute magnitudes (assuming they are at the distance of NGC 5774/5775) compared to those of Galactic and extragalactic GCs (e.g., Harris 1996; Miller et al. 1997; Lotz 2004). Source #23 is in the halo of NGC 5775 and source #11 in the halo of NGC 5774 so this identification is plausible. Source #11 is located close to a background galaxy on the (u − g) versus (g − r) plot. However, source #11 is a point source. Thus, we believe that it is a GC associated with NGC 5774. Source #2 is about 2 D₂₅ radii from NGC 5774 and it may instead be an unrelated background source.

Sources #25 and #39 are peculiar in that their positions in the upper left of Figure 6 are due to weak emission in the g band relative to the u and r bands.

We reanalyzed the Sloan images of source #39 and found that the u band emission is significant only at about the 1−σ level. Therefore we feel that the u band flux listed in the Sloan catalog is overestimated and that the tabulated colors are unreliable for this source.

Source #25 is located at the southern tip of what is clearly an elongated background galaxy, SDSS J145355.82 + 033431.8 (Figure 5). Our interpretation of the Sloan image is that the X-ray source is coincident with a blue color region different from the rest of this galaxy. The SDSS color of the astrometric corrected X-ray position is located at the upper left in Figure 6. We were unable to improve on these color estimates but feel that they are not reliable because of the spatial overlap. We note that the colors of the background galaxy center (labeled "galaxy" in Figure 6) alone is very close to the colors of source #11, which is reasonable in that the colors of the old stellar population of a galactic bulge at low redshift should match the colors of a globular cluster.

The fact that source #25 is not at the center of the background galaxy suggests that it could be a very luminous ULX in that galaxy. The Sloan spectrum of this galaxy gives a redshift of 0.0953 ± 0.0002, which corresponds to a cosmology corrected luminosity distance of 424 Mpc. The Galactic absorption corrected flux for the source #25 in the 0.2–10 keV band is 3.1^+2.0_−1.1 × 10⁻¹⁵ erg cm⁻² s⁻¹and the corresponding luminosity is ∼6.7 × 10⁴⁰ erg s⁻¹. The immediate environment of this source in the host galaxy is blue, which can be seen from Figure 6. No HST data are available for this galaxy. To search for the optical counterpart of source #25, we added all the available R-band images obtained from ESO/NTT. Astrometric and photometric studies were performed with the final image. No counterpart within the error circle at the astrometric corrected Chandra position of source #25 was detected in this image, at the limit of 27.5 mag. This suggests that the X-ray to optical flux ratio (F_X/F_O) is more than 17. The absence of radio detections and such high value of F_X/F_O indicates that the source #25 is not a background AGN and most likely a bright ULX in SDSS J145355.82+033431.8.

Source #13 is weak and consequently has poorly measured optical colors. Formally, this source has colors similar to a star forming region but it is located far from any active star forming region so this interpretation is doubtful.

The remaining source #17 has not been cataloged in the SDSS. Our measurements are listed in Table 3 and shows that it is an extended object with extremely blue colors. This is a blue compact dwarf galaxy.

In summary, of 12 optical sources spatially coincident with X-ray sources two could be globular clusters (s11 and s23), one bright ULX (s25) in a background galaxy, one blue compact dwarf galaxy (s17), six are likely background QSOs (s02, s05, s06, s09, s15, and s26), and two are too faint or too confusing to identify with confidence (s13 and s39).

Comparing Tables 1 and 3, we find three of the 15 ULX candidates (source #5, #23, and #26) have potential optical counterparts. Sources #5 and #26 are brighter than 21 mag and we obtained optical spectra of these two objects.

The optical spectrum of source #5 was obtained on January 2004 with the William Herschel Telescope (WHT)⁹ using the red arm of the ISIS spectrograph and the R158R grism during an 1800 s exposure. The slit width was between 120 and 145. Cu–Ar and Cu–Ne lamps were used for wavelength calibration. This provides a sampling of 1.62 Å pixel⁻¹ and an effective resolution of 8–10 Å (depending on the slit width used and on atmospheric seeing). The spectrum was analyzed following a standard procedure using IRAF software that comprises bias subtraction, extraction of the spectra, and wavelength calibration. We used the standard spectroscopic stars Feige 67 (Oke 1990), and BD+26 2606 (Oke & Gunn 1983) to correct approximately for the response of the configuration to different wavelengths. Given the prohibitive time needed to obtain flat-field images (especially in the blue part of the spectrum), we did not correct for that effect. However, we have checked that this correction would have been very small (⩽1 %) and would not have affected any of the identifications of the main spectral features reported here. The spectrum of source #5 is shown in the top panel of Figure 7. A strong Mg ii λ2798 Å emission line and a weak emission line of O ii λ3727 Å are present in this spectrum. From these emission lines we determine that this object is a background quasar at a redshift of 1.365.

Zoom In Zoom Out Reset image size

The optical spectrum of source #26 was obtained with the 3.6 m Telescopio Nazionale Galileo (TNG) telescope on June 2005. A single exposure of 1800 s with a slit of 2'' using the DOLORES spectrograph and the MR-B grism was taken. For wavelength calibration we used He and Ar lamps. The spectrum was reduced according to standard procedure within IRAF as mentioned above. The spectrum is displayed in the bottom panel of Figure 7. It is a background quasar at a redshift of 0.661 based on the Mg ii λ2798 Å emission line.

A second spectrum of source #26 was obtained at the ESO 3.6 m telescope using the ESO Faint Object Spectrograph (EFOSC2) on June 6th, 2005. The grism Gr#13 was used, providing a dispersion of ∼2.77 Å pixel⁻¹ with a wavelength coverage between 3720 and 9330 Å. Two spectra of this source were obtained with an exposure time of 1800 s each and a S/N of about 5–10. They were reduced using standard IRAF tasks, shifted to the heliocentric rest frame and flux-calibrated with a spectrophotometric standard star. The spectral resolution is 440 at the central wavelength (6530 Å  5.4 pixel element). Some fringing is present at the level of ∼5%–10% at wavelength long ward of 8000 Å. The Mg iiλ2798 Å as well as Balmer lines Hβ λ4861 Å and Hγ λ4342 Å are seen, as are forbidden lines of O and Ne. While Mg ii is clearly broad, with a FWHM ≈ 5300 km s⁻¹, Hβ is narrow, with a FWHM of only 400 km s⁻¹, similar to the forbidden line widths. However, most of the narrow lines are either weak or affected by systematic effects such as fringing, and the typical error in the measurement of their width is about 40%. This spectrum is shown in Figure 8.

Zoom In Zoom Out Reset image size

The SDSS obtained optical spectra of the nuclear regions (circle of 3'' radius corresponding to the size of the SDSS fiber) of NGC 5774 and NGC 5775. The only strong emission lines present in the nuclear spectrum of NGC 5774 are Hα, [N ii] λ6583Å, and [S ii] λ6725Å. The intensity ratios of these lines suggest that the emission is from normal H ii regions (Ho et al. 1993). The nuclear optical spectrum of NGC 5775 is much redder than that of NGC 5774, as anticipated by the high inclination of NGC 5775, but the spectrum is otherwise similar to that of NGC 5774 with relatively weaker emission lines. No nuclear point source is detected in the Chandra image of either galaxy to a limiting luminosity of 3 × 10³⁷ erg s ⁻¹. Thus, we conclude that these galaxies do not host active nuclei.

The only counterpart to an X-ray source detected in the infrared Two Micron All Sky Survey (2MASS) data is source #5. Its J, H, and K magnitudes are 17.0 ± 1.4 mag, 16.7 ± 1.6 mag, and 15.8 ± 1.8 mag, respectively. These colors are consistent with the quasar interpretation for this source (Cutri et al. 2002).

The NRAO VLA–Faint Images of the Radio Sky at Twenty-centimeters (VLA–FIRST, Becker et al. 1995) image of NGC 5774/5775 field is shown in Figure 9. Only source #47 has a potential counterpart in this image. Its radio peak flux-density is (2.94 ± 0.15) mJy beam⁻¹. We also queried the positions of ULX candidates in the NRAO VLA Sky Survey (NVSS and FIRST were operated at 1.4 GHz, Condon et al. 1998) and no source was detected. This is expected, because the limiting peak source brightness of NVSS was about 2.5 mJy beam⁻¹ (Condon et al. 1998).

Zoom In Zoom Out Reset image size

Radio observations of the NGC 5774/5775 galaxy pair were carried out using the Australia Telescope Compact Array (ATCA) in 2005 January. The continuum observations (over a time-slot of 7 hr) were made at all four frequencies 1.4, 2.3, 4.8 and 8.4 GHz by alternately observing at 1.4/2.3 GHz and 4.8/8.4 GHz frequency pair using turret rotation. The array used for the observations was 1.5D. The synthesized beams are elongated in declination as a consequence of the northern declination of the source and hence only the results for the higher frequency pair have been considered. The data at the two frequencies were reduced using standard procedure within the MIRIAD package (Sault et al. 1995; Sault & Killeen 1998). The quoted fluxes are from the cleaned and primary beam corrected images at the two frequencies. The rms noise achieved at 4.8 GHz and 8.4 GHz was respectively 0.21 and 0.06 mJy and the synthesized beams were 24 × 35'' and 2'' × 17''. We have detected radio emissions from sources #19, #47, and #49, with peak flux–densities of 1.58, 0.68, and 1.12 mJy beam⁻¹, respectively, at 4.8 GHz with a typical rms of 0.40 mJy. These results can be compared with results of the VLA–FIRST. However, it must be mentioned that the ATCA and the VLA–FIRST measurements were made using two different instruments with different calibration, imaging and analysis methods. In addition, 1.4 and 4.8 GHz frequencies were used for the VLA–FIRST and the ATCA observations of NGC 5774/5775  respectively. Thus, it is important that we compare the peak flux densities at a common frequency, say at 1.4 GHz. To convert the ATCA peak flux densities (S_ν) from 4.8 GHz to 1.4 GHz, we need to know the value of radio spectral index. Results of radio observations of ULXs suggest that the values of radio spectral indices (α) are in the range of 0.4 to 1.2, where S_ν ∝ ν^−α (Neff et al. 2003; Kaaret et al. 2003; Kording et al. 2005; Miller et al. 2005; Soria et al. 2006; Lang et al. 2007). We have used the conservative-value of α as 0.5 and the estimated 1.4 GHz peak flux densities for sources #19, #47 and #49 are 2.95 ± 0.56, 1.24 ± 0.56, and 2.08 ± 0.56 mJy, respectively. Comparison of these results with that of the VLA–FIRST results suggest that the 1.4 GHz peak flux densities of these three sources varied by (2.95–0.15) ± 0.58, (2.94–1.24) ± 0.58, and (2.08–0.15) ± 0.58 mJy, respectively, where 0.15 mJy is the typical rms of the VLA–FIRST observations. Considering the large uncertainties, which were computed through the propagation of errors, we find that sources #19 and #49 were variable at or above the 3-σ level. Even though these two sources appear variable, but these results should be taken cautiously because of the differences between the VLA–FIRST and the ATCA observations and analysis procedures. However, it should be emphasized that whether sources #19, #47, and #49 are radio variable or not, but they are radio emitters.

There are eight (excluding the AGN, s05) X-ray sources in NGC 5774. Grimm et al. (2003a) proposed a universal cumulative X-ray luminosity function (XLF) of high-mass X-ray binaries (HMXRBs), suggesting a normalization factor (NF) proportional to the SFR of a galaxy. Thus, the formalism of Grimm et al. (2003a) would predict only one HMXRB in NGC 5774 with 2–10 keV luminosity ∼1.78 ± 0.3) × 10³⁸ erg s⁻¹. This is extremely small compared to the observed total X-ray luminosity of the eight point sources of NGC 5774, which is (6.85 ± 0.39) × 10³⁹ erg s⁻¹ (Table 1). Such huge discrepancy may be due to the nonlinear relation between the total luminosity of HMXRBs and the low-SFR of NGC 5774 (Grimm et al. 2003a). NGC 5774 has a low SFR of 0.2 M_☉ yr⁻¹, but the presence of three ULXs and five other X-ray binaries is remarkable. Using the mass and the SFR of NGC 5774 and Equation (4) of Colbert & Ptak (2004) we can determine the total point-source X-ray luminosity of this galaxy, which is (4.15 ± 0.5) × 10³⁹ erg s⁻¹. This is comparable to the observed value and suggests that the X-ray source population of NGC 5774 consists of the older stellar population (including low mass X-ray binaries) and the younger stellar population (including HMXBs), which are directly related to the mass and SFR of NGC 5774 (Colbert & Ptak 2004; Grimm et al. 2003b). Three ULXs of NGC 5774 are located in active star forming regions on the spiral arms of this galaxy. This indicates that the formation of ULXs strongly depends on its immediate local environment but not on the global properties of the host galaxy.

Twenty-four X-ray sources (excluding the AGN, s26) are detected in NGC 5775. Observed counts of these 24 sources are plotted against the cumulative number of sources in Figure 10. It can be seen from this figure that the XLF of the starburst galaxy NGC 5775 is best represented as a single power law (slope = −0.71) whereas the XLF of all the sources studied requires a broken power law with a steeper slope at high flux. For NGC 5774, both a single and a broken power law provide acceptable fits (there are too few sources to differentiate) but the slope of the single power law is −0.82, steeper than that of NGC 5775. The value of the slope of the XLF of NGC 5775 is consistent with the mean values of −0.65 ± 0.16 for the interacting/merging and −0.79 ± 0.24 for spirals (Colbert & Ptak 2004). The XLFs of X-ray sources in spiral galaxies with moderate SFRs are relatively flat (e.g., Kilgard et al. 2002; Colbert & Ptak 2004), which means that the most luminous individual sources dominate the total X-ray luminosity (L_X) from these galaxies (Grimm et al. 2003b). Comparison of XLFs between NGC 5775 and interacting/merging galaxies shows that their slopes are also around −0.70 with a dispersion of 0.2, even though they have different SFRs [M82 (Kaaret 2001), NGC 253 (Strickland et al. 2000), NGC 3256 (Lira et al. 2002), NGC 3395/3396 (Brassington et al. 2005), NGC 4485/4490 (Roberts et al. 2002), The Mice (Read 2003), The Antennae (Fabbiano et al. 2001), The Cartwheel Ring Galaxy (Gao et al. 2003)]. However, it may be noted that the NFs of respective XLFs are different for these galaxies. Here we would like to mention that the number of sources detected above 2 × 10³⁸ erg s⁻¹ in NGC 5775, are consistent with the number of HMXRBs predicted by the universal XLF of Grimm et al. (2003a). Thus, the XLFs with same slopes but different NFs indicate for evolutionary stages of galaxies. For example, comparison of XLFs between NGC 5775 and The Antennae or NGC 4485/4490 shows that these galaxies have similar spectral slopes but the NF values are large for The Antennae or NGC 4485/4490. These results suggest that The Antennae or NGC 4485/4490 are more evolved than that of NGC 5775. This is independently supported by the radio results, which suggest that the star formation in NGC 5775 was triggered by an encounter between the two galaxies some 100 Myr ago (Irwin 1994).

Zoom In Zoom Out Reset image size

SDSS obtained optical spectra of the nuclear regions (circle of 3'' radius corresponding to the size of the SDSS fiber) of NGC 5774 on 2001 May 16. The only strong emission lines present in the nuclear spectrum of NGC 5774 are Hα, [N ii] λ6583 Å, and [S ii] λ6725 Å. The intensity ratios of [N ii] λ6583/Hα is less than 0.6. This suggests that the emission-lines are formed in the nuclear H ii regions (Veilleux & Osterbrock 1987; Ho et al. 1993, 1997). No nuclear point source is detected in the Chandra image of this galaxy to a limiting luminosity of 3 × 10³⁷ erg s ⁻¹. Source # 3 (s03) is close to the optical center, but the astrometric corrected Chandra position of s03 is ∼100 away from the optical nucleus. Thus, we conclude that this galaxy does not host active nucleus to a limiting luminosity of 3 × 10³⁷ erg s ⁻¹. SDSS spectrum of the nuclear region of NGC 5775 is much redder than that of NGC 5774, as anticipated by the high inclination of NGC 5775, but the spectrum is otherwise similar to that of NGC 5774 with relatively weaker emission lines. No nuclear point source is detected in the Chandra image of this galaxy to a limiting luminosity of 3 × 10³⁷ erg s⁻¹. Thus, again we conclude that this galaxy also does not host active nucleus.

We extracted the X-ray spectrum of the diffuse emission from the D₂₅ region, excluding the point sources. The spectrum was binned to obtain at least 20 counts per fitting bin. A corresponding background spectrum was extracted from an ellipse of twice the D₂₅ area from a source-free region of the same CCD. Absorbed single power-law model does not fit well with this spectrum (N_H consistent with the Galactic value and Γ = 3.2^+0.83_−0.37 for χ² = 91.2 for 67 degrees of freedom (dof)). Similarly, absorbed single component thermal models like blackbody, mekal, bremstrahlung, disk blackbody, etc., fit poorly with this spectrum. Finally, we find that an absorbed mekal plus power-law model fits well with the diffuse gas spectrum of NGC 5775 (χ² = 65.2 for 65 dof). The spectrum of diffuse gas, model and residuals are shown in Figure 11. In this model N_H was required to freeze with the Galactic value. The mekal temperature and abundance values are ∼0.31 ± 0.04 keV and 0.05^+0.20_−0.03, respectively. Photon index of the power-law component is 1.75^+0.68_−0.51. From the X-ray spectral fits of the diffuse gas, we have found that the fluxes from the mekal and the power-law components are 2.6 ± 0.7 and 5.4 ± 1.5 × 10⁻¹⁴ erg s⁻¹ in the 0.5–8.0 keV band, respectively. Parameters of the mekal component can be used to compute different parameters (emission measure, particle number density, total gas mass, thermal energy, cooling time, etc.) to determine the physical properties of the diffuse gas. However, large uncertainties of the volume-filling factor will introduce huge errors in the values of these parameters (Strickland & Stevens 2000). Instead, we will use the SFR to determine the origin of the diffuse gas. The temperature of the diffuse gas is slightly lower than the expected diffuse gas temperature of ∼0.5 ± 0.07 keV detected in interacting galaxies (Read & Ponman 1998). This indicates that NGC 5775 is in early phase of interaction. The value of the abundance from the mekal model suggest that the metallicity of the diffuse gas is subsolar. Kewley et al. (2006) have shown that the metallicity of gas in the central regions of starburst galaxies in pairs like NGC 5774/5775 tends to be subsolar. They attribute this to channeling of metal-poor fuel from the outskirts of the galaxies toward the central star forming regions. Most likely, the NW and SE bridges are transporting outer materials into NGC 5775 and thus we expect the diffuse gas of this galaxy to be composed of subsolar materials.

Zoom In Zoom Out Reset image size

We have detected outflow from the central region of NGC 5775, which is shown in Figure 12. The observed outflow may be the starburst-driven superwinds from the starburst galaxy, NGC 5775. This indicates that NGC 5775 is in the superwind phase, which can be used as tool to study its evolutionary process. Comparison of the observed results with the post-starburst evolutionary model results of Taniguchi et al. (2000) suggest that NGC 5775 have gone through both the starburst phase and the early-B star phase and entered into the superwind phase. This means that the widespread star formation took place around 2 × 10⁷ yr ago (Taniguchi et al. 2000). This is consistent with the estimated age of the interaction from dynamical considerations (Irwin 1994). In addition, these results are also consistent with the independent results derived from the studies of XLF and the X-ray spectrum of the diffuse gas of NGC 5775.

Zoom In Zoom Out Reset image size

The fact that star formation is occurring throughout the disk of NGC 5775 yet there is very little visible distribution of the disk/spiral structure of the galaxy pair indicating that the starburst in NGC 5775 is triggered by tidal forces redistributing the gas within the galaxy disk rather than by gas transfer between the galaxies. Thus, the hot diffuse gas was produced during the widespread star formation episodes in NGC 5775 and is proceeding at a rate of 5.4 M_☉ yr⁻¹. Diffuse gas has two origins: thermal and nonthermal. Thermal component is due to supernova remnants (SNRs) and stellar winds. Contributions from SNRs can be computed using a relationship between the far-infrared luminosity and the supernova rate of starburst galaxies, which is as follows (Mattila & Meikle 2001):

$$\begin{equation} r_{SN}=2.7\times 10^{-12} L_{\rm FIR}/L_{\odot }\;{\rm yr}^{-1}, \end{equation} \tag{ 1 }$$

The far-infrared luminosity (L_FIR) of NGC 5775 is 3.14 × 10¹⁰ L_☉ (Devereux & Eales 1989), which corresponds to 0.084 SN per year. Thus, 1.7 × 10⁶ SNe exploded during the interval of 2 × 10⁷ yr. If we assume that a massive star evolves to explode into a SN in 10⁷ yr and 10⁵¹ erg is released from each SN explosion, then total thermal energy associated with the diffuse gas of NGC 5775 will be 1.7 × 10⁵⁷ erg. In addition, there will be some more contribution to the thermal energy from stellar winds. Of course, part of the total thermal energy will be lost due to radiative processes, gravitational potential energy and kinetic energy of the galactic wind/outflow. A large fraction of the thermal energy will be dissipated during the superwind phase of NGC 5775, which may last ∼8 × 10⁷ yr (Taniguchi et al. 2000).

The results of the spectral fitting show that the power-law component contributes more than double to the diffuse gas compared to the mekal component. Power-law sources are most likely unresolved X-ray binaries plus x-ray emitting stars. In addition, we find that the total diffuse gas contributes less than ∼5.0% to the total X-ray luminosity of NGC 5775. These results along with the nondetection of the nuclear source plus detection of the galactic wind suggest that NGC 5775 is in the beginning of the evolutionary process similar to NGC 3395/3396 (Brassington et al. 2005). This is further supported by the evolutionary model of Taniguchi et al. (2000), which suggests that, presently, NGC 5775 is in the superwind phase and after spending ∼8 × 10⁷ yr this galaxy will enter into the LINER phase (Taniguchi et al. 2000). However, it may be mentioned that the evolutionary stage of NGC 5775 is determined based on the evolutionary models of Taniguchi et al. (2000), which have been constructed to explain the optical narrow emission-lines. These models are yet to be examined with the multiwavelength properties of starburst galaxies. Thus, presently, we are not certain about the evolutionary tracks of these models and the evolutionary results of NGC 5775, present here, should be taken cautiously.

From the comparison of Figures 1 and 2, we find that sources #14, #16, and #25 are located on the NW-bridge region. We know that source #25 is associated with a background galaxy. Sources #14 and #16 have no optical counterparts and are most likely not background AGNs. Most likely, they are HMXRBs. It appears that source #16 is probably associated with the halo of NGC 5775. Thus, we conclude that source #14 is an HMXRB located on the NW-bridge. Similarly, we find that sources #6, #10, #15, #18, and #23 are located on the SE-bridge. Based on the optical colors, sources #6 and #15 are possible background AGNs. Source #23 is a potential globular cluster (Figure 6 and Section 2.2). No objects brighter than 25 mag in the V band are present within the error circles at the astrometric corrected Chandra positions of sources #10 and #18. Both the sources #18 and #23 are located very close to the D₂₅ isophote of the edge-on galaxy, NGC 5775. Thus, we suggest that they are most likely the halo-objects of NGC 5775. Finally, we are left with two sources, s10 and s14, which are around 2 × 10³⁷ erg s⁻¹ and are probably HMXRBs formed on the SE- and NW-bridges, respectively. The presence of these objects provides evidence for recent star formation on the bridges that might have been initiated by the interaction between the galaxy pair (Smith et al. 2007). Sources #18 and #23 are ULX candidates. Chandra spectra of these two ULX candidates can be described by the power-law model (Table 1). The X-ray to optical flux ratio (F_X/F_O) is larger than 13 for source #18 and source #23 is hosted in a globular cluster.

Sources #2, #13, #17, #19, #29, #39, and #48 are within the field but they are located neither on the bridges nor within the D₂₅ isophotes of two galaxies. Properties of sources #2, #13, and #39 are discussed in Section 2.2. Source #17 is associated with a blue compact dwarf galaxy (Figure 5), which contains large clusters of young, hot and massive stars. Radio emission was detected from source #19 and most likely this is a BL Lac object. We could not detect the optical counterparts of source #29 and #48 and they are close to D₂₅ isophote of NGC 5775 and they are most likely HMXRBs in the halo of this galaxy. However, it should be mentioned that hydrodynamical simulations of interacting galaxies suggest that the gas in the connecting bridges is shocked or compressed due to interaction, which triggers widespread star formation episode and some gas may fall onto the outer disk (Struck & Smith 2003). If this is true, then we would expect many X-ray sources to form at the interface of the disk and bridges. Thus, the halo-objects of NGC 5775 (s16, s18, and s23), which are also located at the interface of the disk and bridges, might have formed due to the interaction between the outer disk and the falling gas from the bridges.

In this section, we describe interesting individual sources in some detail, particularly the most luminous X-ray sources which are candidate ULXs. In total, 49 sources were detected in two X-ray observations within the 66 × 58 field (Figures 1 and 2) above a detection limit of ∼10³⁸ erg s⁻¹. Nine X-ray sources were detected near or within the D₂₅ isophote of NGC 5774. There are 25 X-ray sources within the D₂₅ isophote of NGC 5775. Rest 15 sources were also detected within the field but they were out side the D₂₅ isophotes of these galaxies. Properties of these 15 objects have already been discussed in Sections 2.2 and 3.3. Here we will discuss about the candidate ULXs detected within the D₂₅ isophotes of NGC 5774 and NGC 5775. Due to the large amounts of unresolved optical emission from the disk of NGC 5775 and its nearly edge-on orientation, identification of potential optical counterparts to the detected X-ray sources cannot be made for all the sources with high confidence. Thus, we concentrate mainly on the X-ray properties of the sources in most cases.

In 1996, a bright supernova was detected in NGC 5775, SN 1996ae (Vagnozzi et al. 1996). We used the Chandra and XMM-Newton data to search for X-ray emissions from SN 1996ae. No source was detected at the location of this supernova. A 2-σ upper limit of 2 × 10³⁷ erg s⁻¹ was determined.

Before we switch over to candidate ULXs, we have to determine the possible foreground/background objects, which mimic as ULXs. Since the value of F_X/F_O for foreground stars will be less than 0.1 and that for AGNs is of order unity (e.g., Green et al. 2004), the brightest X-ray sources with optically bright counterparts are very likely either Galactic stars or high-redshift AGNs. The flux from a ULX (L_X > 10³⁹ erg s⁻¹) at the distance of NGC 5774/5775 is 8.4 × 10⁻¹⁵ erg cm⁻² s⁻¹ in the 2–10 keV bandpass assuming a Γ = 1.8 power-law spectrum and an absorption column density equal to the Galactic value. We expect three background sources at or above this flux level within the 66 × 58 field analyzed here, according to the log(N)–log(S) relation deduced from the Chandra Deep Field South survey (Rosati et al. 2002). Two ULX candidates with bright optical counterparts lying at or just beyond the D₂₅ isophotes of the NGC 5774/5775 galaxy pair have spectroscopically confirmed redshifts higher than that to the galaxy pair. Discovery of high redshifts to ULX candidates outside the optical extent of target galaxies is now a common occurence (e.g., Arp et al. 2004; Gutiérrez & López-Corredoira 2005; Gutiérrez 2006) and agrees with the statistical evidence that the radial distribution of ULXs in galaxies asymptotically approaches the expected background level at roughly the radius of the D₂₅ isophote (Swartz et al. 2004; Swartz 2006). Based on the SDSS colors, we have identified that s06, s09, s13, and s15 are likely background AGNs. The two ULX candidates, s18 and s23, have already been discussed in Section 3.3 and will not be repeated here. Finally, we are left with three and seven candidate ULXs in NGC 5774 and NGC 5775, respectively. These 10 objects will be described in the following section. SDSS images of individual sources, their Chandra spectra and spectral results are presented in Figure 5, Figure 3 and Table 1, respectively.

3.4.1. Source #1

The Chandra light curve of source #1 displayed significant variability. Its X-ray light curve is shown in Figure 13. Assuming a constant count rate results in a value of χ² of 85.7 for 58 dof and a KS probability < 10⁻⁵. The Chandra spectrum of this ULX candidate can be described as an absorbed power law. This source was not detected during the XMM-Newton observations. We queried at the astrometric corrected Chandra position of this source and the derived value of L_X is ∼7.7 × 10³⁸ erg s⁻¹ in the 0.5–8 keV band. Comparison of this value with that of Chandra, listed in Table 1, shows that it varied by a factor of 3 between these two observations separated by around 15 months.

Zoom In Zoom Out Reset image size

HST/WFPC2 image in F814W filter was used to search for the optical counterpart of source #1. The astrometry between the Chandra and the HST images were performed using the USNO-B1.0 star 0935-0243054 and the estimated astrometric error is ∼070. No optical counterpart was detected within the error circle at the astrometrically corrected Chandra position of the X-ray source, to the limiting magnitude of ∼26 mag. The source is located on the spiral arm and also within a star forming complex. The Hubble image shows many bright knots near this ULX candidate that are also apparent in the SDSS composite image.

Despite the lack of an optical counterpart, this source is similar in the X-ray temporal and spectral behavior to the transient ULX in Cen A that has been attributed to a Be/X-ray binary (Ghosh et al. 2006). At the distance to NGC 5774, a late O or early B companion to source #1 would not be detectable even with Hubble.

3.4.2. Source #4

3.4.3. Source #12

3.4.4. Source #24

Source #24 was bright (L_X ∼7.78 ± 0.32 × 10³⁹ erg s⁻¹ in the 0.5–8 keV band) during the Chandra observation but not detected in the XMM-Newton data with a 2-σ detection upper limit of 5.6 × 10³⁸ erg s⁻¹ in the 0.2-10 keV band. This shows that it varied by more than a factor of 10 between the two observations i.e., over a period of 15 months. However, it was steady during the interval of Chandra observation. A Chandra spectrum of this source fitted with an absorbed power-law model is shown in Figure 3 and the spectral parameters are listed in Table 1. Residuals between the data and the model are shown in the lower panel of Figure 3. It can be seen from the residual plot that an emission feature around 3 keV is present in the spectrum of this absorbed and flat power-law object. We added a Gaussian emission-line component to the power-law model to fit the Chandra spectrum. From the fit parameters, we find that Δχ² = 7.2 for 18 dof and the F-test statistic and probability values are 5.92 and 0.0054, respectively, which suggest that the detection of the line at 2.99 keV (width = 0.16 keV and equivalent width is 541 eV) is significant at ∼99.5% level. It is important to mention that it is a flat power-law object (Γ = 1.26^+0.60_−0.30). Figure 14 shows the HST/ACS/WFC Hα (F658N) image of the region surrounding this ULX candidate. A circle of 02 radius, which is the size of the error circle, is also shown. It may be seen that this source is located within the diffuse Hα emission region and close (∼39 pc) to an object of absolute magnitude around −12.2 mag in the F625W filter. This object may be a young star cluster, because most of the H ii regions are much brighter than this object (Lee et al. 2001). In fact, the error circle covers part of this star cluster, suggesting that this ULX candidate is associated with this star cluster.

Zoom In Zoom Out Reset image size

Flat power laws have been detected in the low/hard state spectra of Galactic black hole X-ray binaries (McClintock & Remillard 2006). In this state, the accretion rate is sub-Eddington and compact radio jets may be present. In addition, radio and X-ray intensities are correlated (McClintock & Remillard 2006; Remillard & McClintock 2006). Thus, it is believed that X-ray beaming is absent and X-ray emission-lines can be detected in this low/hard state (Remillard & McClintock 2006). Flat powerlaw and the presence of a broad (∼0.16 keV) emission line at 2.99 keV (Ca XX) in the Chandra spectrum of this ULX suggest that beaming was not present in source #24 and it was in low/hard state during the Chandra observations. If we assume that its observed X-ray luminosity was due to isotropic emission at the sub-Eddington (∼0.02 × L_Edd) rate, then the mass of the accretor is ∼3000 M_☉. It entered into further lower flux level during the XMM-Newton observations. Such dramatic variation indicates that the accretion disk temperature at the outer edge was below the hydrogen ionization temperature. This can happen when the mass transfer rate falls below a critical value and the mass of the accretor will be much higher than those of stellar-mass black holes, if the companion is a high-mass star (King 1996). Source #24 being located in the star forming region and most likely hosted in a young star cluster, it is expected that its companion will be a high-mass star. Thus, source #24 is a prospective IMBH candidate.

3.4.5. Source #28

3.4.6. Source #40

It is the brightest ULX candidate (7.6 ± 0.7 × 10⁴⁰ erg s⁻¹ in the 0.5–8 keV band) among 82 nearby galaxies in the survey of ULXs described in Swartz et al. (2004). It was not detected by XMM-Newton with a 2-σ detection upper limit of 3.5 × 10³⁸ erg s⁻¹. The 2–10 keV flux did not vary, but the 0.2–2 keV light curve, shown in Figure 15, indicates for possible variability during the Chandra observation (χ²/dof ∼75/58). This type of variability was not detected in any of the sources of Table 1 that had at least 100 counts in total, compared to 1354 counts in source #40. We have searched for pulsations in the soft, hard and total light curves of this source, but no dominant pulsation was detected. Close inspection of the soft light curve suggest that this source can be a periodic variable, which is not clearly visible due to the presence of flares. We have fitted this light curve with a periodic variable component (∼6.2 hr period) plus flares, which could be from the accretion disk (Figure 15).

Zoom In Zoom Out Reset image size

Both absorbed power-law and disk blackbody models fit equally well with the Chandra spectrum of source #40. The power-law spectral parameters are listed in Table 1. The disk blackbody model parameters are T_in = 1.82^+0.36_−0.20 keV, Normalization = 0.00304 ± 0.0015 and χ²/dof = 109/110. This model fits slightly better than the powerlaw model (Δχ² = 3.9). The parameters of the disk blackbody model suggest that the mass of the accretor of source #40 is 30 ± 9.0 M_☉  assuming that the inclination angle ranges between 0° and 45°. An HST/ACS/WFC Hα (F658N) image of this ULX candidate field is shown in Figure 16. The astrometric-corrected Chandra position of this ULX is near to a starforming region (Lee et al. 2001), indicating that this source is not hosted in any star cluster.

Zoom In Zoom Out Reset image size

Source #40 in NGC 5775 is the most luminous ULX in our catalog (Swartz et al. 2004). It displayed remarkable variability by more than a factor of 500 between Chandra and XMM-Newton observations. In addition, its energy dependent light curve displays variations with a possible period of 6.2 hr and the X-ray spectrum suggest that the mass of the accretor is in the range of 20–40 M_☉. Based on these results and using the formalism described in Ghosh et al. (2006), we suggest that this system may be similar to that of Cygnus X3 (Paerels et al. 2000). The X-ray modulation in Cygnus X-3 is explained as variable scattering in the dense photoionized wind from a Wolf-Rayet companion (Paerels et al. 2000). Energy dependent light curve of source #40 (Figure 15) also indicates that the possible modulation may be either due to high absorbing column or due to occultation of the X-ray source by the mass transfer stream (White & Swank 1982). If the former is true, then this ULX is most likely a wind-fed system. Otherwise, the mass transfer stream will collide with the accretion disk and that will generate strong turbulence. This will cause the disk to swell in the vicinity of the confluence and this bulge or thickened region of the accretion disk may cause modulation. In this scenario, the dips will have variable depths. Long-term changes and instabilities in the outer structure of the disk will result in phase jitter and anomalous dips (White & Holt 1982). The presence of flare-like features in the light curve of source #40 may be due to instabilities on the disk. These can be tested from future observations over a longer period.

Intrinsic X-ray luminosity of source #40 is ∼10⁴¹ erg s⁻¹ in the 0.2–10 keV energy band, which suggest that the mass of the accretor has to be at least 770 M_☉ even if it is accreting at the Eddington rate. However, from the HST/ACS data we found that this ULX is not hosted in a star cluster, which can host the IMBH. In addition, we have determined that the mass of the accretor of this ULX system will be in the range of 20–40 M_☉ and its the huge X-ray flux variability between the Chandra and XMM-Newton observations could be due to either geometrical or relativistic beaming along our line of sight. High-mass accretion rates lead to thick accretion disks causing geometrically beamed emission (Fabrika & Mescheryakov 2001; King et al. 2001; Fabrika 2004; Poutanen et al. 2007). Geometrical models suggest that the X-rays will be beamed through the geometrical funneling and will be absorbed by the H- and He-like ions of abundant heavy elements present in the photosphere of the funnel's inner wall. These blue-shifted absorption lines will be superimposed on the emergent multicolor-continuum spectrum from the funnel (Fabrika & Mescheryakov 2001; Fabrika 2004; Poutanen et al. 2007). However, we could not detect any such line in the Chandra spectrum of this ULX, though the 0.5–8 keV continuum fits well with the MCD black body model. Thus, it is very likely that this ULX is beaming relativistically. Relativistic beaming could be either due to inverse-Compton or synchrotron mechanism. In the inverse-Compton scenario, if the optical photons from the companion star are up-scattered by the jet, then the X-ray emission can power the ULX. X-rays produced by this process will have a broken power-law shape with a break-energy at around 1 keV (Georganopoulos et al. 2002). However, we do not find the signature of broken power law in the Chandra spectrum of this ULX. Relativistically beamed synchrotron jet will produce spectral curvature (Γ/dE is >0), which was suggested based on the EXOSAT spectra of X-ray selected BL Lac objects (Ghosh & Soundararajaperumal 1995). Recently, this has been confirmed with high S/N XMM-Newton spectra of these objects. (Perlman et al. 2005). Thus, to search for the spectral curvature we divided the Chandra spectrum of source #40 in the 0.5–2.0, 2.0–4.0, and 4.0–8.0 keV bands, and fitted these spectra with the absorbed single power-law model. Figure 17 shows a plot of Γ versus energy. A constant model was fitted to this plot that resulted with Γ = 1.6 and χ²_r = 1.64 for 2 dof. However, a powerlaw model with a spectral slope of 0.75 fits this plot with χ²_r = 0.038 for 1 dof. Using these results we find that the F-test statistic and probability values are 85.1 and 0.0687, respectively. This means that the powerlaw model is significant at 93.1% level with respect to the constant model. These results are not robust, which will firmly establish that dΓ/dE is > 0. Thus, we consider these results as a possible signature for the spectral curvature. Spectral curvature can also be measured using a logarithimicparabola model (Perlman et al. 2005 and references therein):

$$\begin{equation} dN/dE = ke^{-\sigma (E)N_{H, {\rm Gal}}} e^{-\sigma (E)N_{H, {\rm int}}(1+z)} E^{(-\Gamma +\beta {\rm Log}(E))} \end{equation} \tag{ 2 }$$

where N_H,Gal, N_H,int are the Galactic and intrinsic absorbing columns and β is the curvature parameter. We fitted the Chandra spectrum of source #40 using this model and determined the value of β as 0.4 ± 0.2. Again, this is not a highly significant result, but indicates that this spectrum may be curved. If the observed flux variations of source #40 are due to the presence and absence of jet emission, then we can assume that beaming has boosted X-ray emissions at least by a factor of 500. This can be equated with $\delta ^{3+2\alpha _{x}}$, where δ is the Doppler boosting factor for the approaching jet with X-ray spectral index α_x. This gives the value of δ as 3.5, which is smaller than the value of δ determined for the brightest ULX of NGC 5408 (Kaaret et al. 2003). Thus, we suggest that, most likely, the luminous X-rays from the source #40 is due to relativistically beamed synchrotron jet (Kording et al. 2002).

Zoom In Zoom Out Reset image size

3.4.7. Source #41

This ULX candidate was detected in both Chandra and XMM-Newton observations. The X-ray spectra fits well with the powerlaw model. The photon index and the hydrogen absorbing column density were almost steady, within the error limits, between the two observations although the 0.5–8.0 keV flux varied by almost a factor of 2. It is interesting to note that both the Chandra (Γ = 1.48^+0.32_−0.24) and XMM-Newton (Γ = 1.40^+0.17_−0.14) spectral indices of s41 suggest that it is a flat X-ray spectrum source. If flat spectral indices are the signatures of the sub-Eddington accretion rate (∼0.02 × L_Edd) at the low hard state of this ULX, then its observed XMM-Newton luminosity will correspond to ∼6.3 × 10⁴¹ erg s⁻¹ as the intrinsic luminosity in the 0.2–10 keV energy band (McClintock & Remillard 2006; Remillard & McClintock 2006). This suggests that the mass of the accretor has to be at least 4846 M_☉ even if it is accreting at the Eddington rate and emitting X-rays isotropically.

HST/ACS/WFC Hα (F658N) image of this ULX candidate field is shown in Figure 18. It can be seen from this figure that the source #41 is located within a bright Hα (F658N) complex. With the available data, we cannot determine whether this source is associated with a H ii region or a star cluster.

Zoom In Zoom Out Reset image size

3.4.8. Sources #46 & 47

These two ULXs lie within 4'' of each other and cannot be resolved by XMM-Newton. The sum of the luminosities of the two sources in the Chandra observation is within errors equal to the luminosity of the combined object in the XMM-Newton observation. Thus, it is likely that both the sources did not vary between the Chandra and XMM-Newton observations. Source #46 is a relatively absorbed steep X-ray spectrum ULX. HST/ACS/WFC Hα (F658N) image shows that source #46 is embedded in diffuse emission (Figure 19). Radio emissions were detected in the VLA-FIRST and in our ATCA observations from the region but the angular resolution is not sufficient to determine which source is the likely radio emitter. However, astrometric results between the VLA-FIRST and the Chandra images of NGC 5775 suggest that the VLA-FIRST position is more consistent with s47. We have already discussed in Section 2.4 that s47 did not vary between the VLA-FIRST and the ATCA observations but is a strong radio emitter. No optical counterparts were detected within the Chandra error circles in the HST/ACS/WFC Hα (F658N) image around s47 but this source is located ∼20 away from a bright-blue object (Figure 20). Photometric colors and absolute magnitudes, assuming that this object is in NGC 5775, suggest that it is a large star-forming region. Both the Chandra and the XMM-Newton results clearly show that s47 is a flat X-ray spectrum ULX. Thus, the radio, optical and X-ray properties of this ULX is interesting. For example, the lack of any optical point source at the astrometric-corrected Chandra position of s47 suggests that it is not a background object because its value of F_X/F_O will be much larger than the upper limits of B L Lac object (REF). On the other hand, its X-ray results clearly show that this ULX was in the low hard state and the observed high X-ray luminosity indicates for an accretor of mass more than, if accreting at the Eddington rate and emitting isotropically (McClintock & Remillard 2006; Remillard & McClintock 2006). This has been discussed in section. Again, the scenario of a high-mass accretor is further supported by the radio/X-ray results, even though these observations were not simultaneous. A "fundamental plane" relating black hole mass (M), X-ray (L_X) and radio luminosities (L_R) in unbeamed sources has been discovered (Corbel et al. 2000, 2003; Gallo et al. 2003; Falcke et al. 2004; Merloni et al. 2003, 2005). Using this relation,

$$\begin{equation} \log L_{\rm R}=\big(0.60^{+0.11}_{-0.11}\big) \log L_{\rm X} + \big(0.78^{+0.11}_{-0.09}\big) \log M + 7.33^{+4.05}_{-4.07}, \end{equation} \tag{ 3 }$$

and the observed radio/X-ray fluxes, we find that the mass of the accretor will be very high. However, large vertical spread (σ_R = 0.88) of this relation suggest that the errors will also be large. This puts a limit on the mass the accretor between several hundreds to several thousands of solar mass. Thus, based on the nonsimultaneous radio/X-ray results we can not firmly claim that s47 is an IMBH system but it is a very promising candidate.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.4.9. Source #49

This source was detected only in the XMM-Newton data. Its flux remained steady during this observation. The X-ray spectrum of this ULX candidate fits well with the powerlaw model. No source was detected within the error circle at the astrometric corrected Chandra position of this source in the HST/ACS/WFC F625W image. However, it appears from the F658N image (Figure 21) that this ULX is located either within a spherical nebula or within a dissolved star cluster (Pellerin et al. 2007). The UVW2 image obtained from the XMM-Newton optical monitor's telescope shows that the ULX is embedded in the diffuse UV emission, which is the typical characteristic of dissolved star clusters (Pellerin et al. 2007). We have discussed in Section 2.4 that s49 is most likely a radio variable ULX, which was detected during our ATCA observations (1.12 ± 0.1 mJy beam⁻¹ at 4.8 GHz). Using the "fundamental plane" relation, Equation (3), and the observed radio and X-ray fluxes we estimate that the mass of the accretor as 8.3 × 10⁴ M_☉. Due to the large vertical spread (σ_R = 0.88) of this relation we find that the mass of the accretor could as low as a few hundred solar-mass. In addition, the radio and the X-ray observations were not simultaneous. Thus, we should consider these results cautiously. However, it should be mentioned that s49 is a potential IMBH candidate and future simultaneous radio/X-ray observations are highly essential.

Zoom In Zoom Out Reset image size

Ten ULXs are detected within the D₂₅ isophotes of NGC 5774 and NGC 5775. Sources #1, #12, #28, and #40 are all located within bright star forming complexes and no optical counterparts have been detected within the astrometric corrected Chandra positions of these sources. Based on their observed properties we suggest that these are stellar-mass black hole systems.

X-ray spectra of sources #4 and #46 are steep and they are embedded in diffuse emission, which may be ionized nebulae. Ionized nebulae with bubble-like morphology are detected around some ULXs (Roberts et al. 2003; Pakull & Mirioni 2003; Pakull et al. 2006). These bubbles are several hundred parsec in diameter with expansion velocities around 50–80 km s⁻¹ (Rosado et al. 1981; Rosado et al. 1982; Valdez-Gutiérrez et al. 2001). It has been suggested that the expansion could be due to the combined action of energetic supernova explosions and stellar winds, or to continuous inflation by jets (Miller 1995; Valdez-Gutiérrez et al. 2001; Wang 2002; Pakull & Mirioni 2003; Pakull et al. 2006). Some nebulae show barrel-type shapes or enhanced emission along opposite directions that could be interpreted as excitation from a beamed source (Roberts et al. 2003; Pakull & Mirioni 2003). However, some nebulae have displayed spur-shape, which requires an isotropic flux of energetic photons to explain the observed He ii flux (Kaaret et al. 2004; Pakull et al. 2006). Nebulae formed by the combined action of SNRs and stellar winds are not supposed to have He ii emission unless it contains very massive stars. Thus, detection of an nonspherical He ii nebula at the position of the ULX will support the beaming model of these two ULXs. Nebulae formed around IMBHs will be much smaller in size compared to the nebulae formed by SNe or jets. Thus, the observed sizes of diffuse emission around s04 and s46 suggest that they are not IMBH systems.

Another mechanism that is prevalent in interacting galaxies and may lead to the formation of luminous X-ray sources is the formation of super star clusters (SSCs). There is a general correlation between the formation of SSCs and galaxy mergers (Whitmore 2000). Keel & Borne (2003) have shown that this trend applies also to systems like NGC 5774/5775 and does not require direct contact to occur. Theory suggests that the most massive black holes may come from mergers of massive main sequence (Portegies Zwart et al. 2004) or protostars (Soria 2006) within these young massive SSCs. If these IMBHs with masses ≫ 20 M_☉ can form in SSCs and can capture a stellar companion (Baumgardt et al. 2006) then X-ray luminosities near the Eddington limit, 1.3 × 10³⁸ erg s⁻¹, are likely to produce ultraluminous X-ray sources. The ULX X-1 in M82 is believed to be of this type (Strohmayer & Mushotzky 2003; Agrawal & Misra 2006; Dewangan et al. 2006; Mucciarelli et al. 2006). Luminous sources #24, #41, #47, and #49 are all associated with either young star clusters or bright starforming complexes. These ULXs are potential IMBH systems, which have been discussed in Sections 3.4.4, 3.4.7, 3.4.8, and 3.4.9, respectively.

From the comparison of X-ray luminosities of ULXs detected in NGC 5775 during the Chandra and XMM-Newton observations, we find that s41 and s47 varied at the most by less than a factor of 2. On the other hand, s24, s28, s40, and s49 displayed dramatic variations. Both these two types of ULXs may form in starforming environments that have subsolar metallicities. We have already seen that the diffuse gas in the starburst galaxy NGC 5775 is composed of subsolar materials. As a result of this the newly formed massive stars in the disk of NGC 5775 may leave more massive compact remnants than is typical for solar-metallicity stars. The combination of more massive black holes and more massive companions can lead to more luminous and longer-lived X-ray emission once mass transfer begins via a stellar wind or through Roche lobe-filling. Wind-fed binaries will steadily emit X-rays over long time. However, Roche lobe-fill systems will display dramatic variations both in fluxes and spectra.

From our Chandra survey of ULXs, we found a correlation between the observed number of ULXs per galaxy (N_ULX) and the far infrared luminosity (L_FIR). In addition we have also found that the observed number of ULXs per galaxy and the nearest neighbor distance are anticorrelated (Swartz et al. 2004). This means that the observed number of ULXs in interacting/merging galaxies strongly depends on L_FIR. However, Brassington et al. (2005) found a weak correlation between N_ULX and L_FIR of five interacting/merging galaxies. Instead, they found that L_ULX, determined following the prescriptions of Colbert & Ptak (2004), is relatively better correlated with L_FIR of the same five galaxies. It may be mentioned that such a correlation is obviously expected because the value of L_ULX strongly depends on L_FIR (Colbert & Ptak 2004). Thus, we decided to study the relation between N_ULX and L_ULX using The Cartwheel (AM 0035-335), M82 (NGC 3034), NGC 3256, Arp 270 (NGC 3395/3396), Arp 299 (NGC 3690/IC 694), Antennae NGC 4038/4039), NGC 4485/4490, The Mice (Arp 242, NGC 4676A/B), NGC 5194/5195, NGC 5774/5775, Arp 220 (IRAS 15327+2340), and NGC 7252, which are interacting/merging galaxies with elevated number of ULXs. To maintain uniformity, we have analyzed all the archival data from Chandra and XMM-Newton observations and detected X-ray sources within the D₂₅ isophotes of these galaxies. When the intrinsic luminosities of the X-ray sources are more than 10³⁹ erg s⁻¹ in the 0.5–8.0 keV band, then those objects have been identified as ULX candidates. Next, we computed the number of possible background AGNs within the D₂₅ isophotes using the log(N)–log(S) relation deduced from the Chandra Deep Field South survey (Rosati et al. 2002). Finally, we determined total ULXs per galaxy, which will be considered as N_ULX, after subtracting the background AGNs. L_ULX was computed using Equation (4) of Colbert & Ptak (2004). The values of L_K were derived using the K-band apparent magnitude from 2MASS and following the expression given in Seigar (2005). We have used GALEX data or the XMM-Newton/OM data in the UV1 and UV2 bands to determine the values of L_UV and that for FIR were derived using the 60 and 100 μ IRAS flux densities (Devereux & Eales 1989). Figure 22 displays a plot of N_ULX versus log (L_ULX) and fitting gives a slope of 1.03 with a correlation coefficient of 0.34. However, if we exclude The Cartwheel and Arp 220 from fitting, then the slope changes to 1.38 with a correlation coefficient of 0.83. This fit is shown with the straight line in Figure 22 and indicates that the number of ULXs per galaxy depends on L_ULX, which is a function of FIR, near-IR, and UV luminosities. This is an interesting result and suggests that the value of N_ULX depends not only on the SFR but also on the mass of the host galaxies (Colbert & Ptak 2004). It can also be seen from this figure that the large number of ULXs detected in The Cartwheel may not be the true ULX population of this galaxy system. At least four to five sources are in excess compared to the expected value. Detection of these four to five sources may be the result of source confusion. On the other hand, a large number of ULXs are missing in Arp 220. These missing ULXs may be highly variable, which were below the detection level during the Chandra observations of this galaxy. It is also evident from Figure 22 that at least two to three ULXs are missing in NGC 5775. This indicates that the two ULX candidates (s18 and s23) marked as possible halo-objects of NGC 5775, may really belong to this galaxy. Finally, we conclude that the interacting/merging galaxies host a comparable number of ULXs based on their mass and SFR.

Zoom In Zoom Out Reset image size