1. Optical spectra of quasars in the context of Eigenvector-1 R. Zamanov Bologna Dec. 5, 2002

2. in collaboration with: P. Marziani (Padova, I) J.W. Sulentic (Alabama, USA) M. Calvani (Padova, I) R. Bachev (Alabama, USA) D. Dultzin-Hacyan (UNAM, Mexico)

3. CONTENT: Eigenvector-1 correlations Optical FeII emission of AGNs Average quasar spectra in the context of Eigenvector- 1 diagram “Blue outliers” among AGNs – objects in which the [OIII] lines are blue shifted relatively to the H  with 300-1000 km/s. White dwarfs with spectra similar to quasars and physical drivers of this similarity

4. The main constuituents of an Active Galactic Nucleus Central massive Black Hole (M BH ~10 6 -10 9 M  ) Geometrically Thin Accretion Disk (d  3 R g  10 5 R g ) Thick molecular torus (d  1 pc) Line emitting gas (clouds?) (d  0.1 pc in low luminosity AGN; d  10 4 R g ) Radio Jet along Disk Axis (from Padovani & Urry 1992) Massive Black Hole Molecular Torus Relativistic Jet Accretion Disk

5. FeII emission Average Quasar Spectrum: Francis et al. 1991

6. However the quasar spectra are not similar! Composite Quasar Spectra from the Sloan Digital Sky Survey (Vanden Berk et al., 2001 AJ 122, 549).

7. Eigenvector-1 correlation space During the last decade, several investigations of AGNs emission lines emphasized the importance of a set of correlations conventionally called “Eigenvector-1”. They are related to the principal component analysis of the spectral properties of the Palomar-Green quasars performed by Boroson & Green (1992). This correlation space provide and optimal discrimination between different type of quasars and could play a role for AGNs similar to H-R diagram in regard to stars (Sulentic, Calvani, Marziani 2001, The Messenger 104, 25). Physical drivers of Eigenvector-1 can be: (i) the source luminosity-to-mass ratio (L/M) convolved with the orientation (Marziani et al. 2001, ApJ 558, 553) or (ii) the fraction of the Edington luminosity at which the source emits (L/L Edd ) and the black hole mass (Boroson 2002, ApJ 565, 78).

8. FWHM(H  ) [LIL kinematics] ratio R(FeII)=EW(FeII)/EW(H  ) [LIL em. Regions physical conditions] soft-X photon index  Soft [SED] 3D parameter space for Boroson & Green 1992 + Marziani et al. 1996 sample from Sulentic, Marziani, Dultzin-Hacyan, 2000, ARA&A 38, 521

9. Interpretation of the Eigenvector 1 correlation space AGN “MAIN SEQUENCE” Outliers Outliers are all BAL QSOs Population A Population B

10. where Q is the number of hydrogen ionizing photons. The reverberation mapping studies (Kaspi et al. 2000) : Ionization parameter (Marziani et. 2001):

11. The FeII-H  (optical Eigenvector-1) diagram. The theoretical lines for the range of masses and L/M ratios expected for low redshift quasars (Zamanov & Marziani 2002, ApJ 571, L77).

12. Our data set : The data set includes CCD spectra of 216 Seyfert 1 galaxies and low-redshift quasars (z  0.8). Spectra were obtained for studies of H  region with 2 meter class telescopes: ESO (1.5m), San Pedro Martir (2.2m), Calar Alto (2.2m), KPNO (2.2m), Asiago (1.82m). The spectra were taken with: - similar instrumental setups yielding resolution FWHM  4-7 A, - similar (rest frame) wavelength coverage ( 4300 - 5100 AA), - typical S/N  20 –50 in the continuum, only spectra with S/N > 12 have been used. The sample has an average absolute B magnitude M B  -23.7  2.0.

13. FeII emission in the optical spectra

14. Before to analyse the H  and [OIII] lines we need to subtract FeII emission. The template based on I Zw1 spectrum (Boroson & Green 1992) allows us to satisfactorily subtract the FeII opt emission to about 98% of the the spectra of the whole sample. In the figure is shown the successful rendering of the FeII opt emission by our template, once scaled and broadened, for three objects with very different like width.

15. Examples of subtraction of FeII complex around H  and [OIII] lines. Left panels represent the continuum subtracted spectra and best FeII fit. Left panels represent fit to the H  broad component. The difficulties of FeII subtraction are coming from S/N ratio, wavelength coverage, presence/absence of HeII4686, HeI 4471, etc. from Marziani, Sulentic, Zamanov, et al., 2003, ApJS, accepted

16. FWHM : FeII opt – Hβ correlation Pop. A (FWHM < 4000 km/s ): There is a tight correlation between the FWHM of Hβ BC and FeII, namely 1:1. This implies that both emissions came from the same BLR.Pop. A. (FWHM < 4000 km/s ): There is a tight correlation between the FWHM of Hβ BC and FeII, namely 1:1. This implies that both emissions came from the same BLR. Pop. B. (FWHM > 4000 km/s): Even in this case the correlation seems real, but FWMH(Hβ BC ) exceeds the FeII opt one. Pop. B. (FWHM > 4000 km/s): Even in this case the correlation seems real, but FWMH(Hβ BC ) exceeds the FeII opt one. R Pearson = 0.693, N =69, P = 8.5e-9 R Pearson = 0.882, N = 43, P = 7.3e-9 from Bongardo, Zamanov, Marziani, Calvani, Sulentic, 2002, astro-ph/0211418

17. A special case: IRAS 07598+6508 This intriguing object shows a FIR excess and its location in the E1 diagram is peculiar. It is interesting to note that FWHM(Hβ BC ) = 5000 ± 400 km s -1 and FWHM(FeIIλ4570) = 2000 ± 1300 km s -1. The good S/N ratio and the strength of the FeII opt emission, along with the large EW(Hβ BC ) make this result especially striking. The strong blueward asymmetry of the BC of Hβ suggest that the broadening is due to Balmer emission associated to (1) the highly blueshifted CIV at 1549 Å emission and (2) a narrower unshifted component associated to low ionization emission. from Bongardo, Zamanov, Marziani, Calvani, Sulentic, 2002, astro- ph/0211418

18. The most straightforward implication is that the FeII opt emission mechanism is probably the same in almost all AGN and that FeII opt is mainly from the zone of the BLR where Hβ is also emitted. In fact the FWHM of FeII opt in Pop. B objects seems to be narrower than that of Hβ. It I possible that FeII opt emission in Population B sources comes from the outer part of Hβ emitting region, where the ionization degree is lower.

19. Average quasar spectra

20. Prediction of unification models on spectral properties of Seyfert 1 and quasars: One point somewhere here

21. We present median AGN spectra for fixed regions of the E1 (optical) parameter space [FWHM(Hβ) vs. equivalent width ratio R Feii =W(Fe II λ4570)/W(Hβ)]. We suggest that an E1-driven approach to median/average spectra emphasizes significant differences between AGNs and offers more insights into AGN physics than a single-population median/average spectrum derived from a large and heterogeneous sample of sources. Optical parameter plane of E1. The lines indicate the adopted binning. from Sulentic, Marziani, Zamanov, et al., 2002, ApJ 566, L71

22. Average quasar spectra along E1 sequence Sulentic, Marziani, Zamanov, et al. 2002 ApJ 566, L71 Optical parameter plane of E1. This is the largest sample yet displayed in an E1 context. The lines indicate the adopted binning.

23. Composite quasar spectra following the spectral beams defined in Eigenvector-1 diagram. Left panel: before FeII subtraction. Right panel: same composite spectra with FeII emission subtracted.

24. Continuum subtracted H  composite line profiles for the different E1 parameter bins. The solid colored lines show the H  BC after subtraction of NC. A Lorenztian fit (red line) is superposed on the NLSy1, A1, and A2 profiles. The individual components of a double Gaussian (green lines) and resultant fit are shown for B1 and B1+. We find that the Hβ broad component line profile changes along the E1 sequence in FWHM, centroid shift, and profile asymmetry. While objects with FWHM(Hβ BC ) 4000 km s -1 are better fitted if two broad-line components are used: a ``classical'' broad-line component and a very broad/redshifted component. from Sulentic, Marziani, Zamanov, et al., 2002, ApJ 566, L71

25. The Lorentz profile is consistent with emission from an extended accretion disk. This reinforces the suggestion that the LIL spectra in population A sources arise from a disk. The situation is less clear for population B sources, where the Eddington ratio may be much lower. There is good evidence that sometimes only one of the two emission components is present in population B sources (a pure BLR or a pure very broad line region [VBLR]). Can the double-Gaussian model that is needed to fit population B (and radio-loud) profiles be physically justified? Several lines of evidence point toward the existence of a VBLR at the inner edge of the BLR (Corbin 1997, ApJS 113, 245 and ApJ 485, 517). Emission from this region may be thought of as a sort of inner large covering factor "boundary layer" where gas begins to become optically thick.

26. [OIII] “blue outliers” among the AGNs

27. Forbidden [OIII] emission arises in the NLR of AGNs. This emission has now been partly resolved in the nearest AGN, where the geometry of the line-emitting gas has been found to be far from spherically symmetric. This suggest that measures of the integrated [OIII] emission may correlate with source orientation to the line of sight. At the same time it is generally believed that radial velocity measures of the narrow emission lines (e.g narrow component of H  and [OIII] 4959, 5007) provide a reliable measure of the systemic, or rest-frame, velocity. Several observations, however, indicate that the NLSy1 prototype I Zw1 shows an blue shift of the [OIII] lines  V  -500 km/s relatively to other rest frame indicators (HI 21cm, molecular CO emission). We measured the radial velocity difference (  V) between the H  and [OIII] 4959, 5007 lines in 187 objects (our sample 215 objects, 7 with no detectable [OIII] emission, 16 with poorly defined H  peak).

28. Histogram showing the distribution of the radial velocity difference between [OIII] 5007 and top of H . As it is visible in most of the objects |  V| < 300 km s -1. However there are some objects, with  V down to -1000 km s -1. The values range from –950 to +280 km/s with average = -30 km/s and sample standard deviation  135 km/s. Typical measurement error is  50 km/s. from Zamanov, Marziani, Sulentic, et al., 2002, ApJ 576, L9

29. H  spectral region of the “blue outliers” after the deredshift and subtraction of the FeII template. Spectra are normalized with respect to the normal continuum and arbitrary constant added. Solid curves correspond to the subtraction of IZw1-based empirical template, and dot-dashed curves to the subtraction of a theoretical template (Sigut & Pradham 2002, astro-ph/0206096). Vertical lines indicate the position of H , [OIII] 4959 and [OIII] 5007. The difference in radial velocities between [OIII] lines and H  is obvious. from Zamanov, Marziani, Sulentic, et al., 2002, ApJ 576, L9

30. Radial velocity difference between H  and [OIII] 5007 versus the FWHM(H  BC). Vertical dotted line marks the boundary of the NLSy1 galaxies. Vertical dashed line separates population A and B sources. In our sample of 215 objects we detected 7 objects with  V  -300 km s -1. from Zamanov, Marziani, Sulentic, et al., 2002, ApJ 576, L9

31. Location of the outliers in the FWHM(H  BC ) versus W(FeII)/W (H  BC ) diagram (the optical E1 diagram). They are not randomly distributed (2D_KS test gives probability 0.990 – 0.999). from Zamanov, Marziani, Sulentic, et al., 2002, ApJ 576, L9

32. [OIII] 5007 shifts with respect to the top of H  ( underlying galaxy systemic velocity ?). As it is visible the NLR kinematics is changing along E1 sequence Low W([OIII] 5007) High EW[OIII] 5007) Low EW([OIII] 5007) From Marziani, Zamanov, Calvani, et al. 2003, Mem SAIt, in press).

33. A sketch representing “blue outlier”. The [OIII] lines originate from the wind, the disk is visible face-on, and the receding part of the wind is obscured from the disk. In calculations we adopted cone half-opening angle 85 0, with the line of sight oriented at 15 0, with respect to the cone axis. The receding part of the flow is assumed to be fully obscured by an optically thick disk.

34. Upper panels: CIV 1549 and [OIII] 5007 profiles of Ton 28. Lower panels: CIV 1549 and [OIII] 5007 outflow model profiles, for optically thin gas moving at approximately the local escape velocity. from Zamanov, Marziani, Sulentic, et al., 2002, ApJ 576, L9

35. For our sample, we calculated the masses using reverberation mapping studies(Kaspi et al. 2000). Our sample covers: magnitude range 20 < M B <27 magnitude range 20 < M B <27 BH mass 7 < log(M/M  ) <10 BH mass 7 < log(M/M  ) <10 a well defined strip of a well defined strip of L/L Edd = 0.02 - 1.00. L/L Edd = 0.02 - 1.00. The blue outliers are located between objects with highest L/L Edd ratio. L/L Edd of “blue outliers”

36. The luminosity-to-mass ratio versus the mass of the BH. If the blue outliers are oriented nearly pole-on the effect of orientation could play a role. It could be as high as  M  0.4. Even in these case the blue outliers remain between objects accreting at higher Eddington ratio. (Bear in mind that a lot of other objects also have to be moved in the same way).

37. The “blue outliers” among AGNs seems to represent a special case of high L/M ratio, face-on view, and very compact NLR. They seems to be radio quiet analog of the core dominated radio loud quasars. (!) Not all radio quiet AGNs visible pole-on are “blue outliers”.

38. From White dwarfs to Quasars

39. Figure: UV – region spectral similarity between CH Cyg and I Zw. The middle spectrum is produced by scaling and broadening of the CH Cyg spectrum to imitate the emission lines widths of IZw1. I Zw 1 – narrow line Seyfert 1 galaxy, widely used as template for all quasars. Mass of the black hole ~10 7 M . CH Cyg – symbiotic with ~1M  WD CH Cyg CH Cyg (broad.) I Zw 1 From Zamanov & Marziani, 2002, ApJ 571, 77

40. Comparison between the optical spectra in the H  - H  region of the interacting binaries CH Cyg, MWC 560 and the low redshift quasar I Zw 1. A clear similarity between the emission lines is visible. Practically every emission feature visible in the spectrum of IZw1 has corresponding emission line in the spectra of CH Cyg and MWC 560.

41. The optical emission line spectra of CH Cyg and MWC 560 are subtracted, broadened and scaled to imitate I Zw 1. This standard procedure is widely used for the emission line measurements of AGN, using I Zw 1 itself as a template (Boroson & Green 1992, Marziani et al. 1996). After this processing, good identity is achieved with the spectrum of I Zw 1. Our best fit corresponds to a width FWHM(FeII)= 970  90 km s -1 (Zamanov & Marziani, 2002, ApJ 571, 77)

42. The HI and FeII lines of AGNs are coming from the so-called broad- line region. This poorly understood region is thought to be within  1 pc from the central (supermassive) black hole. The clear spectral similarity means that in objects like MWC 560 and CH Cyg we are observing a scaled down version of the famous broad line region of quasars.

43. JETS: Jet velocity : ~1000-1500 km s -1 in CH Cyg (Taylor et al. 1986, Crocker et al. 2001) and 1000-6000 km s -1 in MWC 560 (Tomov et al. 1992) Galactic microquasars (accreting stellar mass black holes): 0.26c SS 433 (Margon 1984) 0.5c Cyg X-3 (Marti et al. 2001) 0.9c GRS 1915+105 (Mirabel & Rodriguez 1999) Consistent with an overall picture in which the jet velocity is of the same order of the escape velocity (Livio 2001) : V esc (WD)= 0.02 c.

44. JET ENERGY : MWC 560 and CH Cyg: the jets are probably result of the propeller action of a magnetic white dwarf (Mikolajewski et al. 1996) = extraction of rotational energy from the compact object. Quasars – the jet energy is coming from extraction of energy and angular momentum from a rotating black hole via the Blandford & Znajek (1977) mechanism. Microquasars –black hole - Blandford & Znajek (1977) mechanism neutron star - ??? (The jets of Crab are the most pure case of extraction of rotational energy, even without accretion). The jets of CH Cyg and MWC 560 represent probably a low energy (non- relativistic) analog of the jets of quasars and microquasars, having a similar energy source – the extraction of rotational energy from the central compact object.

45. We propose to name these objects NANOQUASARS. NANOQUASARS : white dwarfs with jets and quasar-like spectra, representing the low energy (non-relativistic) analog of quasars and microquasars. Why “nano” ? Denomination: q uasars  microquasars  nanoquasars  (greek) = nano (ital.) = enana (spanish) = dwarf (engl.)

46. Optical spectra demonstrating the spectral similarity and the changes of the FWHM(H  ). The filled circles refer to NLSy 1 galaxies, which are supposed to have systematically lower black hole masses. The two triangles indicate the nanoquasars (CH Cyg and MWC 560). As it could be expected they are located outside of the AGN population but from the side of NLSy1. from Zamanov & Marziani, 2003, ASP Conf.Ser, in press

47. The FeII-H  (Eigenvector-1) diagram. The lines are plotted (from top to bottom) for M BH =1.10 9 M , M BH =5.10 7 M , and white dwarf mass M WD =1.4 M . The L/M ratio was running in the limits 2.5-4.6 for M BH =1.10 9 M  ; 2.5-5.1 for M BH =5.10 7 M  ; 3.0-3.9 for white dwarf mass M WD =1.4 M . The ratio (L/M) is in solar units with the solar value (L/M)  =1.92 ergs s -1 g -1. The position of nanoquasars on the diagram reinforces the interpretation of Boroson &Green Eigenvector-1 as a mainly result of L/M ratio. The efficiency of accretion along with some other factors could play some minor role. from Zamanov & Marziani, 2002, ApJ 571, L77

48. Similarity of the CIV profile of the nova-like variable RW Sex, with those of broad absorption line quasars.

49. The high mass X-ray binaries Can we observe a scaled down version of the quasar broad line region in wind-fed X-ray binaries ? In the most cases they have additional source of ionization – a hot primary OB star, i.e the ionization conditions are quite different from symbiotics and AGNs. It will be extremely interesting to detect a stellar mass black hole accreting from the wind of red giant (although very difficult from the evolutionary point of view). A black hole accreting from wind of red giant will (probably) represent good imitation of quasar !

50. SUMMARY: 1.We are investigating optical E1 on base of the largest sample yet displayed in an E1 context. 2.The average quasar spectra in E1 emphasize the differences between AGNs and offers more insights into physics than a single population spectrum. 3.The FWHM(FeII) is slightly different from FWHM(H  ) in pop.B, although it is very similar in general. 4.We detected objects in which different system velocity indicators gives inconsistent results. We detected in 3% of our sample that the [OIII] lines are shifted with >300 km/s (“blue outliers”). 5.We found striking similarity between the emission lines of two accreting WDs and quasars. This gives us the unique possibility to consider the optical E1 diagram using objects less massive by a factor of ~10 7. Our results reinforce the interpretation of E1 as driven mainly by the L/M ratio.

51. In future: 1.Extension of our sample to higher redshift quasars (IR data). 2.Searching for orientation indicators. 3.Theoretical lines and distributions of objects onto the other Eigenvector-1 plans. 4.We would like to distinguish the influence of M BH and L/M ratio onto the emission line properties.

52. THE END

53. Both Radio Loud and Radio Quiet (  30% RQ;  70% RL in our sample) Predominantly Radio QuietRadio Loudness LargerLowerW([OIII]) LargerLower W(H  ) More frequent high amplitude variability Possible long term variability and flickering Optical Variability Redward Asymmetric Affected by a Very Broad redshifted component? Douple peaked or highly shifted single peaked H  profiles Lorentzian (see Figure), symmetric save for spectral type A3 where a blueward asymmetry becomes prominent H  line profile shape LowerLargerW(FeII) Mainly “mini-BAL QSOs”? Preferred occurrence of low-z (and presumably higher z) of BAL QSOs BAL QSOs (see poster at this meeting) Large EW of [OIII] 4959,5007 and general radial velocity agreement with H  [OIII] 4959,5007 has low EW and frequent blue shifts with respect to H  Forbidden Narrow Lines (see Figure) Only slightly broader than, or equal to FWHM(H  ). Larger than FWHM(H  ) also by a factor of several HeII 4686 emission FWHM Less than FWHM(H  ) by 20%.Equal to FWHM(H  ).FeII 4570 emission FWHM CIV 1549 unshifted or shifted to the red as H  Flatter,  soft  2, no soft X ray excess Population B [FWHM(H  )  4000 Km s -1 ] Large systematic blueshift of CIV 1549CIV 1549 Steeper,  soft  3 Soft X ray spectra index Population A [FWHM(H  )  < 4000 Km s -1 ]

54. where Q is the number of hydrogen ionizing photons. We will use for typical AGN continuum f=0.39, =1.22  10 16 (Laor et al.1997), and for nanoquasars we will adopt (T eff =8500 K) corresponding to f=1  10 -5, =3.48  10 15. The reverberation mapping studies (Kaspi et al. 2000) : The ionization parameter (Marziani et. 2001):

56. Other analogies: Orientation of the source (FWHM and the orientation) FWHM in quasars is (probably) connected with the orientation. In a flattened configuration (Marziani et al. 2001) : where i is the inclination angle. Face on (MWC 560) FWHM(H  )=110 km/s Edge on (CH Cyg) FWHM(H  )=200 km/s This is in qualitative agreement with the expectations for quasars, that a face-on object must have lower FWHM.

57. Figure. The FeII-H  (optical Eigenvector-1) diagram. First theoretical grid proposed (Marziani et al. 2001, ApJ 558, 553). In this grid is the position depends on the L/M ratio and the orientation, supposing that all quasars have equal masses log (M/M  )=8.

58. Immediate results: Narrow-Line Seyfert 1 Galaxies (NLSy1; FWHM(H  )  2000 km s -1 ) represent an extremum in the parameter space, not a disjoint peculiar class of AGN (and they are NOT Seyfert 2!). Not really FeII strong emitters, NLSy1 are discriminated among radio quiet AGN by their small W(H  ). Steep spectrum radio-loud sources represent the opposite extremum. No ultra- soft excess has been revealed in core-dominated RL AGN, which show  Soft  2.5,  therefore better discriminated in the  Soft - R(FeII) than in the R(FeII)-FWHM(H  ) plane. There is a remarkable continuity and correlation in line parameters between NLSy1 and RQ AGN with broader Balmer lines up to FWHM(H  )  4000 km s -1. Perhaps a more significant distinction is between AGN with FWHM(H  )  4000 km s -1, which we may call “Population A” and AGN for which FWHM(H  ) > 4000 (“Population B”). Population A includes 65% of PG quasars but almost no radio loud AGN. Population B encompass RQ quasars that show optical properties rather similar to RL AGN. If EW(FeII) is used instead of R(FeII), Pop. A and B are separated into two disjoint “clouds” of RQ AGN.

59. Two main (sets of) correlations systematize the spread of observed properties among AGN: (1) The so-called Eigenvector 1 correlations; Originally identified by Boroson & Green (1992) by a Principal Component Analysis (PCA) of  the specral properties of 82 Palomar-Green quasars, dominated by an anticorrelation between the strength of FeII and [OIII] 4959,5007; E1 involves primarily Low Ionization Lines (LIL) HI Balmer lines, MgII 2800, FeII multiplets Does not depend on the quasar luminosity Relationship between optical parameters and soft X-rays: Wang et al (1996)  Optimal 3 dimensional correlation spaceOptimal 3 dimensional correlation space The “Baldwin effect,” Anticorrelation between HIL rest-frame equivalent width and continuum luminosiy Involves primarily High Ionization Lines (LIL) CIV 1549CIV 1549, HeII ( 1640 and 4686), OVI 1034.