arXiv:astro-ph/0303142v1 6 Mar 2003 rvdsa neetn ehns htcnepanbt po- both explain can model counterrotating (2000) and that Yu ring mechanism lar & interesting Tremaine the an counte Thus, provides a form stream. and disk, disk the ste rotating in the p retrograde prograde, halo become to zero could the through orbits if increase to that, equ were pr Note two speed contain model tern will streams. The way disk this counterrotating past in orbits. formed mass, outward polar ring polar into sweeps a t levitated that slowly of dicts are resonance speed they the pattern preces- , the As of the matches rate orbit the halo. tilted when infal triaxial slightly resonance by a matter a of dark at triaxial sion acquires trapped a halo are of the sl stars plane as that disk (e.g., symmetry speed zero pattern a retrograde to in initially tends lies an accreti with no halo a requires dark If that formation merger. ring or polar for model tive polar rela- dissipative small a with in model formed disk different velocity. mass is tive a comparable galaxy two suggested ring of has polar merger a (1998) which Bekki g in central the after formed. 1984) capture Boksenberg has gas to & The formation Ulrick, ring Schechter, attributes 1983; galaxies. Rubin Wh of (Schweizer, & formation more, mergers They ring polar on 1996). for information model Arnaboldi conventional & provide Combes Schweizer also 1994; & may al. McElroy, et (Whitmore, sh Sackett halos three-dimensional 1987; po- matter the dark of reveal their kinematics ultimately of the could understanding evolution rings and example, lar formation For the into bec insight galaxies. interesting unique are a objects offer these they population, galaxy the of oa igpoieamast iciiaebtenteefor these between discriminate kinematics to stellar means the a most direction, provide would ring one stars polar in ring rotating the be models stellar likely merger levita- counterrotating in resonance contains whereas the ring ulations, in polar Because the model 1994). tion Collett & Evans also rpittpstuigL using typeset Preprint T HE eety rmie&Y 20)hv fee nalterna- an offered have (2000) Yu & Tremaine Recently, lhuhplrrn aaiscnttt nyasalfracti small a only constitute galaxies ring polar Although A STROPHYSICAL u egrsestems ieyepaainfrtestruct the for headings: We explanation Subject likely t hole. most in central the predicted seems a kinemati merger is lacks The A as Yu. disk streams, ring. polar polar a the stellar that in counterrotating suggests than S0 rather the disk of a in pr resides directi curve, rotation material same flat the approximately in an rotating suggest gas kinematics and stars the with rotation, epeettefis esrmn fteselrknmtc in kinematics stellar the of measurement first the present We eateto ersra ants,Crei Institutio Carnegie Magnetism, Terrestrial of Department A J T eateto hsc n srnm,JhsHpisUnivers Hopkins Johns Astronomy, and Physics of Department OURNAL E tl mltajv 11/12/01 v. emulateapj style X 1. , INTRODUCTION RF VERSION DRAFT eecp cec nttt,30 a atnD. Baltimor Dr., Martin San 3700 Institute, Science Telescope disk aais niiul(G 60)—glxe:knmtc an kinematics galaxies: — 4650A) (NGC individual galaxies: TLA OIN NTEPLRRN AAYNC4650A NGC GALAXY RING POLAR THE IN MOTIONS STELLAR aaiswtotamre (see merger a without galaxies D ECEMBER h srpyia ora,datvrinDcme 4 2018 24, December version draft Journal, Astrophysical The 4 2018 24, R alaxy V owly [email protected] pop- ause OB fa of ape ABSTRACT llar ERA at- on on he fWsigo,54 ra rnhR W ahntnD 20015 DC Washington NW, Rd Branch Broad 5241 Washington, of n it- l), of e- al r- - .S A. AND .R C. 1 ersnnelvtto oe rpsdb rmie& Tremaine by proposed model levitation resonance he WATERS n n ihsmlrapiue h aeu n stellar and gaseous The amplitude. similar with and on, t,30 .CalsSr,Blioe D228 n Space and 21218, MD Baltimore, Str., Charles N. 3400 ity, vdn ute upr o h yohssta h polar the that hypothesis the for support further oviding UBIN ftewr ftern,asnl ltpsto antcvrbo cover cannot position slit single a ring, the of warp the of G8 of range end the observed. and in were types beginning K4III spectral the to with At stars template night taken. a each were and exposure calibration wavelength sky argon offset an exposure science each pix Å f3 p.A hsdsac,1 distance, this At Mpc. 38 of te naplrrn sNC45A tcnit facnrlS0 central V a adopte of of an velocity With consists centric ring. It Group polar Local 4650A. prominent NGC a is by surrounded ring galaxy, polar a in ities scenarios. mation oiCDa h a apnsBae(.-)tlsoe For telescope. 1 (6.5-m) was width Baade slit Campanas the Mar- observations, Las all the the and at spectrograph CCD Chivens coni and Boller the with 4650A otnet ali ig oaodcnuinwt h 0disk S0 the with confusion avoid to ring, b a disk, it a call as the to ring are continue the of these identification that the believe support We servations 4650A. of NGC measurement of first ring the and S0 l osato 0k s km 70 of constant ble nraetesga-onierto h aawr indon binned 0 were 4 data by the ratio, signal-to-noise the increase g a 60,wt WMrslto f15k s km 135 of resolution FWHM a with 1600Å, was age t2-m(raod ta.19)rvae ptal extend spatially a revealed 1997) al. H et (Arnaboldi Observ 21-cm therein). (see references at and studied 2002 extensively al. et been Gallagher has e.g., 4650A NGC proximity, tive ta.20;Idc ta.2002). al. et (Gallag disk Iodice polar 2002; the al. of et proje structure Heritage warped the complex, of a part reveal as take Telescope Space observations Hubble disk Recent the polar with 1993). a Wakamatsu called also appropriately (see more galaxy is 4650A NGC that gest r n ieaiso G 4650A. NGC of kinematics and ure I h hieo ltpsto eisseilmnin Because mention. special merits position slit of choice The rm addt o h esrmn fteselrveloc- stellar the of measurement the for candidate prime A ehv bandsetao h 0ado h igo NGC of ring the of and S0 the of spectra obtained have We nti etr erpr tla n a eoiisi oht both in velocities gas and stellar report we Letter, this In h oa igo G 60.Teei eldefined well is There 4650A. NGC of ring polar the so h msinln a tadna h center the near and at gas line emission the of cs ikcicdn ihtern,laigteeatost sug- to authors these leading ring, the with coincident disk aentdtce vdnefrto qa mass, equal two, for evidence detected not have − ,M 11;[email protected] 21218; MD e, . 1 25 .Osrainldtisaepoie nTbe1 After 1. Table in provided are details Observational ). 2. ′′ iesi h pta ieto.Teseta cover- spectral The direction. spatial the in pixels BEVTOSADDT REDUCTION DATA AND OBSERVATIONS stellar dynamics d − 1 Mpc eoiisi oa ig u ob- Our ring. polar a in velocities − ′′ 1 14p.Bcueo t rela- its of Because pc. =184 G 60 sa distance a at is 4650A NGC , LG ′′ 23 ms km =2635 n h ltlnt 72 length slit the and ; − 1 n Hub- a and , − 1 ations twe ut (0.78 ′′ -chip To . her . III he ed th ct n n d 2 Swaters & Rubin

′ ′ FIG. 1.— Image of NGC 4650A from the Heritage program, with the slit positions overlayed. The image size is 2.7 by 1.3 . North is 110◦ and east is 200◦ counterclockwise from the top.

within 10′′ of the S0 major axis for the ring slit positions). At TABLE 1 ′′ other locations, the data were binned in 4 intervals. The cal- JOURNAL OF OBSERVATIONS culated errors of the derived velocities are the quadratic sum of the formal errors on the cross correlation and the dispersion re- Date Spectral Integration Position Region region time angle sulting from using different template stars. Example CCPs are shown in the top four panels of Fig. 2. The typical width of 2001 May 29 Hα 1800s 155 N ring the cross-correlation peaks is about 230 km s−1. The reduction α 2001 May 30 H 1800s 150 S ring process was not optimized to measure the stellar dispersions. 2002 Apr 6 Mg 12 x 1800s 335 N ring The wavelength range of these observations also contains the 2002 Apr 7 Mg 7 x 1800s 63 S0 major β 2002 Apr 7 Mg 4 x 1800s 335 S ring [OIII] 5007, 4959, and H emission lines. Emission line ve- locities were measured by Rubin using the customized DTM measuring program in the software package VISTA (Rubin, the center of the S0 galaxy and the brightest regions of the N Hunter, and Ford 1990). The argon frames for each slit po- and S part of the ring. To maximize the signal from the po- sition were used for wavelength calibration. Only the [OIII] lar ring, we chose to align the slits along the brightest parts of 5007 line was measured because the spectrum degraded rapidly the ring. As a result, the ring slit positions are each displaced ′′ toward the blue and the Hβ emission was located in a fairly about 4 from the nucleus of the S0 galaxy. For the major axis ◦ deep absorption trough. The 2001 Hα spectra were measured spectrum of the S0 galaxy, a position angle of 63 was adopted by Rubin using the same procedures as for the [OIII] line, ex- (Whitmore et al. 1990). The positions and orientations of the cept that the night sky emission lines on each frame were used slit for each of these pointings are indicated in Fig. 1. for the wavelength calibration. During an earlier observing run with the Baade 6.5m tele- scope, we obtained Hα emission line spectra of the N and S 3. RESULTS parts of the ring (Table 1). The slit positions for these spec- tra were close to those for the 2002 observations indicated in In Fig. 3 we plot the LOS velocities of the polar ring stars Fig. 1. as a function of their distances from the major axis of the S0, The frames were bias subtracted, flat fielded, cleaned of cos- dS0. Note that dS0 = 0 corresponds to the location where the slit crosses the major axis, rather than the distance from the nu- mic ray events, wavelength calibrated and combined in IRAF. ′′ Sky subtraction was done by combining all offset sky images cleus. With the exception of the region within |dS0|∼ 10 of for the appropriate slit position and subtracting it, after scaling, the S0 major axis, the stellar velocities along the ring resemble from the corresponding science frame. those expected for a normal, late-type disk galaxy. A notable feature in Fig. 3 is the apparently double valued velocities in the central regions. We attribute these to the com- 2.1. Derivation of the velocities bination of the strong contribution of light from the S0 that To determine the stellar kinematics, the underlying stellar overwhelms the contribution from the ring stars, and the slit continuum emission was first subtracted from both the galaxy positions that are offset by 4′′ from the S0 center. The radial and the template frames. Next, these continuum subtracted velocities at these positions along the S0 major axis (see Fig. 5) frames were cross correlated by Swaters to obtain the stellar are indicated in Fig. 3 by the arrows, and match well the veloc- line-of-sight (LOS) velocities. The absorption velocities were ities observed in the ring spectra. determined by fitting Gaussian profiles to the cross-correlation The observations reveal no evidence for equally populated peaks (CCP), every 1′′ where the signal was strong (along al- counterrotating stellar streams in the polar ring. With a veloc- most the entire length of the S0 major axis slit position, and ity difference close to 200 km s−1 between the approaching and Stellar motions in NGC 4650A 3

FIG. 2.— Example cross-correlation peaks used to determine the FIG. 3.— Line-of-sight stellar absorption line velocities for the north- stellar radial velocities. The positions of the spectra used for these ern part of the ring (filled circles) and the southern part (open circles). cross-correlations are given in the upper right corner of each panel Each arrow indicates the stellar velocity in the S0 galaxy (Fig. 5) at (see also Figs. 3 and 5). the point where the slit crosses the major axis of the S0. receding sides of the ring, the spectral resolution is sufficient to note that the asymmetric drift correction is largest just outside detect two equal mass components. This can be seen in Fig. 2e, the galaxy center, and gets smaller with radius. A final correc- which shows the CCP for a model equal mass, counterrotat- tion is needed is because the spectra are integrated along the ing stream, constructed by coadding the N and S spectra. This − LOS, but these corrections are small compared to the correc- model CCP, with its width of 350 km s 1, is significantly wider tions for asymmetric drift (Sackett & Sparke 1990). Because than the observed CCPs. Note also that the model CCP peaks the corrections are complex or uncertain, we have not derived near the systemic velocity, as expected for two equal, counter- the stellar rotation curve for the polar ring. However, given that rotation streams. Thus, merely the fact that we observe signifi- the corrections are small, or get smaller with larger |dS0|, the cant rotation in the ring rules against the existence of two equal flatness of the rotation curve is a secure result. mass counterrotating streams. However, we cannot rule out the The agreement between the stellar and emission line kine- existence of a weaker, counterrotating component, whose ab- matics seems remarkable given that gas and stars are expected sorption features would be lost in our instrumental resolution to have different corrections for asymmetric drift and LOS in- and sensitivity. From analysis of models constructed by adding tegration. However, after convolution with the spectral reso- the N spectrum to a scaled down S spectrum, we estimate that lution, any narrow spectral features are washed out, resulting at most about 15% of the stars can be in counterrotation. in observed stellar and emission line profiles that have similar A comparison of the measured LOS stellar and [OIII] ion- shapes and derived LOS velocities. ized gas velocities in the ring shows that the velocities of the LOS stellar velocities along the equatorial plane of the S0 stars and the gas in the polar ring agree closely (see Fig. 4). (Fig. 5), show a smooth rotation curve (r < 20′′ = 3.6 kpc) Note how well the gas velocities bisect the double valued stel- with LOS velocities approaching 100 km s−1. Emission lines of lar velocities, indicating that the emission line gas is associated [OIII] and Hβ are observed within ±15′′ of the nucleus. Their with the ring, not the S0. We attribute the double valued gas ′′ ′′ LOS velocities, close to the NGC 4650A systemic velocity, in- velocities, 8 < r < 15 , to the different regions sampled by dependent of radius, suggest that we are probably observing the the N and S ring spectra in the region north of the S0 major ionized gas component of the polar ring. axis (Fig. 1). The measured Hα velocities are virtually indis- tinguishable from the [OIII] velocities (Fig. 4); differences are probably due to minor differences in the slit positions. 4. DISCUSSIONANDCONCLUSIONS Calculating the stellar and emission line rotation curves from In this Letter we report the first detection of stellar velocities the observed radial velocities requires corrections for inclina- in the ring of polar ring galaxy NGC 4650A. To our knowledge, tion, slit position, asymmetric drift, and LOS integration ef- these are the first stellar velocities obtained for a polar ring. The fects. The inclination correction is complex because of the kinematics reveal well-defined rotation, both in gas and in stars. ′′ warped nature of the ring (Arnaboldi et al. 1997; Gallagher et The rotation curve appears to reach a flat part for |dS0| > 15 . ′′ al. 2002). However, outside of |dS0|∼ 6 the main part of the The original impetus for this work was to look for evidence polar ring appears close to edge-on (Gallagher et al. 2002), and for two equal mass, counterrotating streams to test the intrigu- there the inclination corrections are negligible. The correction ing polar ring formation model proposed by the Tremaine & ′′ for the off-center position of the slit is 7% at dS0 = 10 , and Yu (2000). However, the fact that we detect well-defined rota- less for larger dS0. The correction for asymmetric drift de- tion, and the structure of the CCPs, rule out a polar ring that pends on the velocity dispersion and the shape of the velocity consists of two equal mass counterrotating streams. Nonethe- ellipsoid, and their dependence on radius (Binney & Tremaine less, given the sensitivity and resolution of our observations, < 1987, p. 198). In principle these can be derived from the data we cannot rule out that a small fraction ( ∼ 15%) of the stars is presented here, but this is beyond the scope of this Letter. We counterrotating. 4 Swaters & Rubin

FIG. 4.— Line-of-sight [OIII] emission line velocities for the N part FIG. 5.— Line-of-sight stellar absorption line velocities (gray points) of the ring (filled circles) and the S part (open circles) overlayed on and emission line velocities (black points) along the major axis of the the absorption line velocities in gray, with the same coding of the S0 galaxy. We attribute the emission to the polar ring gas component symbols. Velocities of the stars and the gas in the polar ring are strik- ingly similar, except in the region 8′′ < d < 15′′, where the two slits S0 polar ring is offered by the emission line kinematics along the sample different regions in the ring (see Fig. 1). The insert shows α major axis of the S0. These velocities, constant at the systemic a comparison between the H (filled circles) and the [OIII] (open velocity, most likely arise from polar ring gas. With the inner circles) velocities. ◦ ring warped with an inclination of 63 (Gallagher et al. 2002), Thus, it appears that the Tremaine & Yu (2000) resonance the emission lines seen at the center indicate that ring gas ex- levitation mechanism is not at work in NGC 4650A. However, tends into the center. The emission likely arises in the chaotic as Tremaine & Yu (2000) point out, the self-gravity of the ring dust lanes which are oriented at random and extend into the S0 (likely for NGC 4650A given its luminous mass is similar to nucleus (Gallagher et al. 2002). that of the S0, Iodice et al. 2002) makes it energetically favor- Iodice et al. (2002) argue that the luminous mass in the ring able for all stars to be a single stream, rather than two counter- is similar to or even higher than that of the S0. Its large lumi- rotating streams. Another complication is the uncertain effect nous mass, combined with its disklike morphology, suggest the of the presence of dust and gas in the polar ring of NGC 4650A polar ring in NGC 4650A is a disk galaxy that is the victim of on resonance capture. Furthermore, because the resonance lev- a dissipational polar merger of disks of approximately similar itation model predicts that the polar material does not extend masses (Bekki 1998). Further support for a merger scenario to the galaxy center, the possibility that NGC 4650A has a po- comes from the location of NGC 4650A in a chain of galaxies lar disk that does (Iodice et al. 2002; see also below) makes in the cluster (Sersic 1968). It seems unlikely that it less likely it has formed through resonance levitation. Per- a single accretion event could transfer the required large mass, haps galaxies with narrow polar rings that have little or no gas or that extended accretion over a period of time could produce and dust, or disk galaxies with counterrotatingstreams are more sufficient coherence to form a ring. likely to have formed through resonance levitation. Although we did not detect counterrotating star streams in Absent evidence for a resonance levitation mechanism for the polar ring of NGC 4650A, which would have been more the formation of the polar ring in NGC 4650A, we briefly exciting, we are continuing the study of this fascinating object. consider the merger or acquisition alternatives. The observed We postpone a detailed kinematic model until we have more flatness of the rotation velocities in the polar gas and polar stars extensive observations. offers strong kinematic evidence that this feature is a spatially extended disk rather than a narrow ring, as suggested earlier We thank Dr. Maxine Singer, who, as President of the from gas kinematics (Arnaboldi et al. 1997) and from morphol- Carnegie Institution of Washington, insured that the two- ogy (Gallagher et al. 2002). Insight into the radial extent of the Magellan telescope project proceeded, and the astronomers and technical staff who insured that it succeeded.

REFERENCES

Arnaboldi, M., Oosterloo, T., Combes, F., Freeman, K. C., & Koribalski, B. Sackett, P. D. & Sparke, L. S. 1990, ApJ, 361, 408 1997, AJ, 113, 585 Sackett, P. D., Rix, H., Jarvis, B. J., & Freeman, K. C. 1994, ApJ, 436, 629 Bekki, K. 1998, ApJ, 499, 635 Schweizer, F., Whitmore, B. C., & Rubin, V. C. 1983, AJ, 88, 909 Binney, J. & Tremaine, S. 1987, Galactic Dynamics, Princeton, NJ, Princeton Schechter, P. L., Boksenberg, A., & Ulrich, M.-H. 1984, ApJ, 277, 526 University Press Sersic, J. L. 1968, Atlas de Galaxes Australes, Universidad Nacional de Combes, F. & Arnaboldi, M. 1996, A&A, 305, 763 Cordoba, Argentina, p. 66, Evans, N. W. & Collett, J. L. 1994, ApJ, 420, L67 Tremaine, S. & Yu, Q. 2000, MNRAS, 319, 1 Gallagher, J. S., Sparke, L. S., Matthews, L. D., Frattare, L. M., English, J., Wakamatsu, K. 1993, AJ, 105, 1745 Kinney, A. L., Iodice, E., & Arnaboldi, M. 2002, ApJ, 568, 199 Whitmore, B. C., McElroy, D. B., & Schweizer, F. 1987, ApJ, 314, 439 Iodice, E., Arnaboldi, M., De Lucia, G., Gallagher, J. S., Sparke, L. S., & Whitmore, B. C., Lucas, R. A., McElroy, D. B., Steiman-Cameron, T. Y., Freeman, K. C. 2002, AJ, 123, 195 Sackett, P. D., & Olling, R. P. 1990, AJ, 100, 1489 Rubin, V. C., Hunter, D. A., & Ford, W. K. J. 1990, ApJ, 365, 86