1:
2: \documentclass[12pt,preprint]{aastex}
3: %\usepackage{graphicx}
4: %\usepackage{amsmath}
5: %\usepackage{lscape}
6:
7: \newcommand{\aox}{$\alpha_{\rm ox}$}
8: \newcommand{\daox}{$\Delta \alpha_{\rm ox}$}
9: \newcommand{\aoxcorr}{$\alpha_{\rm ox}{\rm (corr)}$}
10: \newcommand{\daoxcorr}{$\Delta \alpha_{\rm ox}(corr)$}
11: \newcommand{\aoxl}{$\alpha_{\rm ox}(l_{2500})$}
12: \newcommand{\CIV}{\ion{C}{4}}
13: \newcommand{\SiIV}{\ion{Si}{4}}
14: \newcommand{\MgII}{\ion{Mg}{2}}
15: \newcommand{\NV}{\ion{N}{5}}
16: \newcommand{\vmax}{$v_{\rm max}$}
17: \newcommand{\nh}{\mbox{${N}_{\rm H}$}}
18: \newcommand{\cmsq}{\mbox{\,cm$^{-2}$}}
19: \newcommand{\flux}{\mbox{\,erg\,cm$^{-2}$\,s$^{-1}$}}
20: \newcommand{\fnu}{\mbox{\,erg\,cm$^{-2}$\,s$^{-1}$\,Hz$^{-1}$}}
21: \newcommand{\flambda}{\mbox{\,erg\,cm$^{-2}$\,s$^{-1}$\,\AA$^{-1}$}}
22: \newcommand{\lumin}{\mbox{\,erg~s$^{-1}$}}
23: \newcommand{\kms}{\mbox{\,km\,s$^{-1}$}}
24: \newcommand{\xmm}{\emph{XMM-Newton}}
25: \newcommand{\chandra}{\emph{Chandra}}
26: \newcommand{\xspec}{\emph{XSPEC}}
27: \newcommand{\rosat}{\emph{ROSAT}}
28:
29: \begin{document}
30: \title{The Correlation between X-ray and UV Properties of BAL QSOs}
31: \author{LuLu Fan,\altaffilmark{1,3}
32: HuiYuan Wang,\altaffilmark{1,2} Tinggui Wang,\altaffilmark{1,2}
33: Junxian Wang,\altaffilmark{1,2} Xiaobo Dong,\altaffilmark{1,2}
34: Kai Zhang,\altaffilmark{1,2} Fuzhen Cheng,\altaffilmark{1,2}}
35: \date{}
36:
37: %\email{whywang@mail.ustc.edu.cn}
38: \altaffiltext{1}{Center for Astrophysics, University of Science and
39: Technology of China, Hefei, 230026, China;} \altaffiltext{2}{Joint
40: Institute of Galaxies and Cosmology, USTC and SHAO, CAS }
41: \altaffiltext{3}{SISSA/ISAS,Via Beirut 2-4,I-34014 Trieste,Italy}
42: \begin{abstract}
43: We compile a large sample of broad absorption lines (BAL) quasars
44: with X-ray observations from the \xmm\ archive data and Sloan
45: Digital Sky Survey Data Release 5. The sample consists of 41 BAL
46: QSOs. Among 26 BAL quasars detected in X-ray, spectral analysis is
47: possible for twelve objects. X-ray absorption is detected in all of
48: them. Complementary to that of \citet{gall06} (thereafter G06), our
49: sample spans wide ranges of both BALnicity Index (BI) and maximum
50: outflow velocity (\vmax\ ). Combining our sample with G06's, we find
51: very significant correlations between the intrinsic X-ray weakness
52: with both BALnicity Index (BI) and the maximum velocity of
53: absorption trough. We do not confirm the previous claimed
54: correlation between absorption column density and broad absorption
55: line parameters. We tentatively interpret this as that X-ray
56: absorption is necessary to the production of the BAL outflow, but
57: the properties of the outflow are largely determined by intrinsic
58: SED of the quasars.
59: \end{abstract}
60:
61: \keywords{quasars: absorption lines - X-rays: general}
62:
63: \section{INTRODUCTION}
64:
65: About 10\%-30\% of optically selected QSOs show broad absorption
66: lines (BAL) in their UV spectra, indicative of outflows with
67: velocities up to 0.1c\citep{hewett03,reichard03}. The similarity in
68: the UV continuum and emission lines between BAL and non-BAL QSOs
69: suggests that BAL QSOs are otherwise normal QSOs viewed in the
70: direction covered by the outflow \citep[e.g.][]{weymann91}. One
71: exception to these similarities is that BAL QSOs are soft X-ray
72: faint compared to non-BAL QSOs \citep[e.g.][]{green95,brinkmann99}.
73: The weakness in X-rays is interpreted as due to strong absorption
74: rather than intrinsic difference. Evidence for this has been
75: accumulated now from detailed studies of X-ray spectra of a few
76: bright BAL quasars, which display X-ray absorption with column
77: densities from $10^{22}$ to $\geq10^{24}$\cmsq\
78: \citep{wang99,gall99, gall02}.
79:
80: Giving the ubiquity of X-ray absorption in BAL quasars, it is
81: natural to ask whether and how the X-ray absorbing gas is
82: connected to the UV BAL phenomenon. It has been known for quite long
83: time that BAL gas should be either confined into small clumps or
84: shielded from the intense soft X-rays in order to match the observed
85: profile. \citet{murray95} proposed that the highly ionized gas at
86: the base of disk wind (shielding gas) can naturally filter the soft
87: X-ray radiation to prevent the gas to be over-ionized so that the
88: radiative acceleration is effective \citep[See also][]{proga00}. As
89: both UV and X-ray absorbers are part of the continuous outflow, the
90: column densities of the two are expected to be correlated. Indeed,
91: \citet{brandt00} identified a correlation between the equivalent
92: width of \CIV\ absorption line and the soft X-ray weakness in a
93: sample of bright quasars, including half a dozen BAL QSOs.
94:
95: \citet{wang05,wang07} found that electron scattering of the
96: shielding gas can explain the distribution of continuum polarization
97: in quasars, and the resonant scattering of BAL outflow can explain
98: the observed polarized spectrum of BAL. They further noted that
99: certain special features should appear in the polarized spectrum if
100: the size of the shielding gas is comparable with that of the BAL
101: outflows. As these features are only found in several low-ionization
102: BAL (LoBAL) QSOs \citep[see their paper for details and
103: also][]{ogle99}, the shielding gas is likely well inside the BAL
104: outflow except in LoBAL QSOs. A similar conclusion has been reached
105: by studying the X-ray spectra of BAL QSOs
106: \citep[e.g.][]{gall04,gall06}. On the other hand, as
107: pointed by \citet{wang00}, in order to keep sufficient opacity in
108: the soft X-ray between $0.2-0.3$~keV, the absorber must have large
109: column density also of Li-like ions because those ions are responsible
110: for both the soft X-ray absorption between $0.2-0.3$~keV and the high
111: ionization UV BALs. Though this band is notoriously difficult to be studied,
112: they argued that at least in three bright low redshift BAL QSOs, the
113: X-ray absorption opacity around $0.2-0.3$ ~keV is large, suggesting
114: very large column density of Li-like ions. However, a relatively
115: small fraction of X-ray absorbing gas at moderate ionization level
116: will be sufficient to suppress the soft X-ray flux.
117:
118: If the X-ray shielding is critical to the ionization balance in the
119: BAL outflow, which in turn affects the radiative accelerating force
120: on the outflow, one would expect that kinematic properties and
121: column density of BAL outflow will somehow correlate with the
122: properties of the X-ray absorber. In a sample of BAL quasars
123: observed by \chandra\ , G06 found a weak correlation between the
124: maximum outflow velocity(\vmax\ ) of BAL and the indicator (\daox\ )
125: of X-ray absorption. Their finding agrees with the qualitative
126: analysis that the strong soft X-ray absorption leads to more Li-like
127: ions, thus more efficiently radiative acceleration by UV photons.
128: However just as G06 pointed out that they had only four sources at
129: the low \vmax, and their sample of BAL QSOs is biased towards
130: strongly absorbed sources comparing to the BI distribution of SDSS
131: EDR BAL QSOs \citep{reichard03}. Giving the importance of this
132: question, more study based on a uniform sample is clearly required.
133:
134: X-ray absorption is not the only factor that affects the ionization
135: equilibrium of BAL gas. \citet{steffen06} showed that the X-ray
136: luminosity of non-BAL QSOs has a large scatter for a given optical
137: luminosity. According to the current popular scenario that BAL and
138: non-BAL are only a matter of whether our line of sight passes
139: through BAL region or not, the intrinsic spectral energy
140: distribution (SED) between UV and X-ray of BAL QSOs should be also
141: diverse. Therefore, it would be interesting to study how the wind properties
142: depend on the intrinsic SED because ionization equilibrium is also
143: closely related to the intrinsic SED. If such a relation does exist,
144: it may offer insight into the driver of the outflows.
145:
146: In this paper, we present a study of BAL QSOs from SDSS Data Release
147: 5 (DR5) \citep{adelman07} that observed by \xmm\ satellite in X-ray
148: in order to explore the relations between UV and X-ray absorbers as
149: well as the relations between BAL properties and the intrinsic UV to
150: X-ray spectra. In \S2 we describe the selection of our \CIV\ BAL
151: QSOs sample and the data analysis in \S3. We show our results and
152: discuss the underlying physics in \S4. Finally, we summarize our
153: results in \S5. Throughout the paper, we assume the cosmological
154: parameters $H_0$ = 70 km s$^{-1}$ Mpc$^{-1}$, $\Omega_\mathrm{M}$ =
155: 0.3 and $\Omega_{\Lambda}$ = 0.7.
156:
157: \section{THE BAL QUASARS OBSERVED BY \xmm\ }\label{ref ss}
158:
159: Starting from the spectroscopic quasar sample in the SDSS DR5
160: \citep{adelman07}, we compiled a sample of definitive \CIV\ BAL
161: quasars that have been observed by \xmm\ either serendipitously or
162: as a target. We restricted the redshift $1.5<z<4.0$ in order to make
163: sure that the \CIV\ $\lambda 1549$ is shifted into the SDSS
164: wavelength regime ($3800-9200$\AA). We matched these quasars with
165: \xmm\ pointing, and resulted in 225 quasars in the FOV of \xmm\
166: observation to date of April,2007.
167:
168: We measured the BALnicity Index \citep[BI,][]{weymann91} and the
169: maximum outflow velocity for these 225 quasars using our own fitting
170: code (see \S3.2 for detail). We adopted the conventional definition
171: for the BAL QSO: the equivalent width (in \kms\ ) of any contiguous
172: absorption (at least $10\%$ below the continuum) exceeds 2000 \kms\
173: that falls between $3000-25000$ \kms\ blueshifted from the
174: systematic redshift \citep{weymann91}. All quasars with non-zero BI
175: were checked by eye, and ambiguous sources were removed.
176:
177: Our final sample consists of 41 \CIV\ BAL quasars, including 5
178: LoBAL quasars, 22 HiBAL quasars, and 14 BAL quasars with unknown
179: BAL subtype because \MgII\ is not within the SDSS spctral coverage.
180: Most (25) of them have been included in the large BAL QSOs catalog
181: from SDSS DR3\citep{trump06}. In comparison with G06, BAL quasars
182: in our sample cover somewhat larger ranges of redshift ($1.579-3.776$)
183: and UV luminosity ($log(l_{2500})$:$30.212-32.230$), and are fainter
184: on average (31.055). Our sample has more uniform distributions in BI
185: ($3-4610$ \kms\ with an average $1101$ \kms\ ) and in \vmax\
186: ($5306-25000$ \kms\ with an average $14482$ \kms\ ) while G06's sample
187: consists mainly of BAL QSOs with the large BI (with an average $3437$
188: \kms\ ).
189:
190: We notice that there are only four \xmm\ targeted objects: SDSS
191: J091127.61+055054.1, SDSS J111816.95+074558.1, SDSS J152553.89+513649.1
192: and SDSS J154359.44+535903.2 (Table 1). The first two are the lensed
193: BAL QSOs \citep{bade97}, which will be excluded from the following
194: correlation analysis. The third quasar was observed because of its
195: high optical polarization \citep{shemmer05}. The fourth object was
196: observed because of its X-ray detection by previous missions, thus
197: may bias towards X-ray bright sources \citep{grupe03}.
198:
199: Some radio loud BAL QSOs show anomalous X-ray properties in
200: comparison with radio quiet counterparts
201: \citep{brotherton06,wang08}. In order to mark such sources, we
202: calculate the radio-to-optical flux ratios,
203: $R_{i}=log(S_{1.4GHz}/S_{i})$, following the definition of
204: \citet{ivezi02}. The flux densities at 1.4~GHz, $S_{1.4GHz}$, are
205: taken from the Faint Images of the Radio Sky at Twenty centimeters
206: survey \citep[FIRST;][]{white97}. We estimate the $R_i$ upper limits
207: by taking the $2\sigma$ errors as the upper limit of radio flux
208: density. Only, three sources (SDSS J092345.19+512710.0,SDSS
209: J133004.72+472301.0 and SDSS J133553.61+514744.1) have radio
210: counterparts with the measured flux density of 1.72~mJy ,1.18~mJy
211: and 2.92~mJy, which give $R_i$=1.42, 1.18 and 1.14, respectively.
212:
213: We listed our sample in Table 1 including the SDSS ID, the redshift,
214: $i$ band fiber magnitude of SDSS, the flux density at rest-frame
215: 2500{\AA}($f_{2500}$),Galactic \nh\ from \citet{dickey90},the BAL
216: subtype and the radio-to-optical flux ratios $R_{i}$. Also we listed
217: the BI and \vmax\ in Table 1(see \S3.2). The values of $f_{2500}$
218: are calculated either by averaging the flux densities in the
219: rest-frame range of $2500\pm20$\AA\ or by extrapolating from the
220: continuum given by our fitting code.
221:
222: \section{DATA ANALYSIS}
223:
224: \subsection{X-ray Data Analysis}\label{sec xray}
225:
226: The X-ray data were retrieved from \xmm\ Science Archive (XSA) and
227: prepared using $SAS\ 7.0.0$ with the most recent calibration files.
228: We extracted the background lightcurve above 10~keV, and the light
229: curve was used to filter the data obtained during the flaring
230: background periods using a threshold of 1.0 count~s$^{-1}$ for PN
231: and 0.5 count~s$^{-1}$ for MOS.
232: A sliding box cell detection algorithm ($eboxdetect$) was applied to
233: the images obtained by PN-CCD detector and two MOS-CCD detectors in
234: the soft ($0.3-2.0$~keV), hard ($2.0-10.0$~keV) and full
235: ($0.3-10.0$~keV) bands to search for X-ray sources. We selected $L =
236: -ln(P)= 10$ as the minimum detection likelihood value, which in turn
237: corresponded to a probability of Poissonian random fluctuations of
238: the counts of $P = 4.5\times10^{-5}$. Among 41 BAL quasars, 26 were
239: detected in the full band and 25 (13) were detected in the soft
240: (hard) bands at least on one EPIC instrument. The spectra were
241: accumulated from a circle region with a 30$\farcs$\ radius except
242: for two sources locating close to the edge of the CCD whereas a
243: circle with 20$\farcs$\ radius was adopted. The backgrounds were
244: extracted from a source-free annulus surrounding each target on the
245: two MOS-CCD detectors and from a source-free circle along the
246: read-out direction on the pn-CCD detector. The photon counts were
247: extracted from the circular source regions centered on the SDSS
248: optical positions with the mentioned radius and the aperture
249: corrections were performed (See Table 2).
250: For non-detections the upper limits of counts are the $90\%$
251: confidence limits from Bayesian statistics\citep{kraft91}.
252: The redistribution matrix file ({\em rmf}) and
253: auxiliary response file ({\em arf}) were
254: generated using the tasks {\em rmfgen} and {\em arfgen}
255: respectively.
256:
257: X-ray spectral modelling are performed using the package \xspec\
258: \citep{arnaud96}. We fit all the spectra with a uniform model, an
259: absorbed power law. It is found that the photon index $\Gamma$ of
260: the power-law is around 2.0 with a small scatter for radio-quiet
261: quasars \citep{george00,reeves00}. The broad band X-ray spectra of
262: BAL QSOs are quite similar to those of radio quiet non-BAL QSOs,
263: \citep{gall02,chart02,chartas03,aldcroft03,grupe03,page05},
264: therefore, in following analysis, $\Gamma$ is fixed to 2.0. Both the
265: Galactic neutral HI absorption and an intrinsic absorption are
266: included in the model. The Galactic neutral HI column density is
267: fixed at the value derived from Galactic HI maps\citep{dickey90}
268: (See Table 1). Due to limited count rates and the large
269: uncertainties, we do not consider more complex X-ray absorption
270: models (e.g., a partially covering or ionized absorber), and just
271: adopt a simple neutral absorption with a solar chemical composition
272: at the source rest frame ({\em zwabs} in \xspec\ )
273: \citep{morrison83} for the intrinsic absorption.
274:
275: To deal with very different X-ray counts available, two different
276: methods are used to estimate the intrinsic absorption column
277: density. The X-ray spectrum is fitted directly with an absorbed
278: power-law if the source is detected in the both soft and hard band.
279: Only the intrinsic absorption and the power-law normalization are
280: free parameters. Other parameters, including redshift, the Galactic
281: neutral absorption and the photon index $\Gamma$, are fixed to the
282: proper values. Either $\chi^2$-statistic or C-statistic
283: \citep{cash79} is taken as the merit of the fit depending on the net
284: source counts. If the net source counts are greater than 100, the
285: spectrum is re-binned with at least 15 counts per bin, and the fit
286: is performed by minimizing $\chi^2$. Otherwise, the spectrum is not
287: binned and the fit is performed by minimizing the C-statistics. In
288: order to test the validity of the fixed $\Gamma$ power-law model, we
289: make $\Gamma$ free to fit eight sources with greater than 100 counts
290: .The average value of $\Gamma$ of six sources is about 1.90, which
291: is very close to 2.0.The other two (SDSS J091127.61+055054.1 and
292: SDSS J100728.69+534326.7) show a rather flat spectra with $\Gamma
293: \sim 1.2$. \citet{page05} mentioned that SDSS J091127.61+055054.1
294: should be better modelled by a broken power law with the values of
295: $\Gamma$ 0.92 and 1.96,respectively.
296: The fitted intrinsic absorptions are very similar to those
297: listed in Table 3 except the two flat spectral sources. A flat spectrum
298: can be caused by complex absorption, strong reflection component or an
299: intrinsic flat power-law. Because all
300: other six sources show normal X-ray spectra and there is no evidence
301: for strong Fe\,K$\alpha$ in the X-ray spectrum, we believe that the
302: flat spectra in these two objects are caused by complex absorptions.
303: However, due to limit counts available, we will not try more
304: complicated models.
305:
306: For those sources detected only in hard or soft band, the
307: upper/lower limit of the column density of intrinsic absorption is
308: estimated from the hardness ratios, defined as ${\rm HR} =
309: (h-s)/(h+s)$, where {\em h} and {\em s} are referred to the hard and
310: soft band counts, respectively. First for each source, we calculate
311: hardness ratios for a grid absorbed power-law models with column
312: densities in the range of $10^{20}-10^{24}$\cmsq\ using the {\em
313: arf} and {\em rmf} at the source position. The observed hardness
314: ratio is then compared to the models and then the upper/lower limit
315: of the intrinsic absorption could be derived.
316:
317: In order to compare our sample with that of G06, we also calculate
318: \aox\ , \daox\ and \aoxcorr\ ,\daoxcorr\ , defined in the G06
319: \footnote{Our definition of the soft($s:0.3-2.0$~keV) and hard
320: ($h:2.0-10.0$~keV) bands is slightly different from
321: theirs($s:0.5-2.0$~keV;$h:2.0-8.0$~keV).}, as follows. First, the
322: above hardness ratios for all BAL QSOs in the sample were estimated.
323: Hardness ratios for a grid of power-law models only absorbed by the
324: Galactic column density in that direction were then calculated using
325: the \xspec. The observed HR was compared to the model HRs to
326: estimate the photon index $\Gamma_{\rm HR}$. With the best fitted
327: photon-index, the normalization at 1~keV was derived from the
328: count-rates. The Galactic absorption corrected 2~keV flux is
329: determined from the model. We defined the UV to X-ray broad band
330: spectral index as \aox\ $=0.384log(f_{2keV}/f_{2500})$
331: \citep{tana79}. It is found that \aox\ is correlated with the
332: optical luminosity of quasars \citep{yuan98,avni86,wilkes94,green95},
333: though its reality has been questioned by \citet{yuan98,tang07}.
334: Following G06, we introduce a
335: quantity of \daox\ = \aox\ $ -$\aoxl\ to characterize the weakness
336: of the X-ray emission of the quasar relative to the average quasars
337: at that UV luminosity, where \aoxl\ was the expected \aox\ based on
338: the 2500 \AA\ monochromatic luminosity, $l_{2500}$\citep{stra05}.For
339: sources not detected in both band, an upper limit to the X-ray flux
340: was derived by assuming $\Gamma=1.0$. Finally, \aoxcorr\ and
341: \daoxcorr\ were calculated from an ``absorption-corrected" value for
342: the 2~keV flux density estimated by a fixed $\Gamma = 2.0$ power-law
343: model normalized by the counts rate in the observed-frame 2-10~keV
344: bandpass. Note that for two lensed sources, SDSS J091127.61+055054.1
345: and SDSS J111816.95+074558.1, we do not calculate \daox\ or
346: \daoxcorr\ since their intrinsic $l_{2500}$ are unknown.
347:
348:
349: \subsection{Ultraviolet Spectral Analysis And \CIV\ Absorption-Line
350: Parameters}
351:
352: Following the procedures described in \citet{zhou06}, we calculate
353: both the BI of \CIV\ and \MgII\ absorption lines and their maximum
354: outflow velocity, \vmax. Briefly, we use the SDSS composite quasar
355: spectrum \citet{vanden01} as the template for continuum and emission
356: line spectrum. The template is reddened and scaled to match the
357: observed quasar spectrum in the absorption line free windows. The BI
358: of \CIV\ and \MgII\ are calculated following the definition given by
359: \citet{weymann91} and \citet{reichard03}, respectively, as follows,
360: \begin{equation}
361: BI=\int^{25000}_{0\ or\ 3000}dv[1-\frac{F^{obs}(v)}{0.9F^{fit}(v)}]C(v)
362: \end{equation}
363: where $F^{obs}(v)$ and $F^{fit}(v)$ are the observed and fitted
364: fluxes, respectively, as a function of velocity in \kms\ from the
365: systematic redshift within the range of each absorption trough and
366: \begin{equation}
367: C(v)=\left\{
368: \begin{array}{ll}
369: 1.0,&\quad \mbox{if $[1-\frac{F^{obs}(v)}{0.9F^{fit}(v)}]>0
370: \ over\ a\ continuous\ interval\ of\ \gtrsim W\ \kms\ $} \\
371: 0,&\quad \mbox{otherwise}
372: \end{array}
373: \right.
374: \end{equation}
375: The integral in equation (1) starts from $v=3000$ \kms\ for \CIV\
376: and from $v=0$ \kms\ for \MgII. The threshold interval in equation
377: (2) is $W=2000$ \kms\ for \CIV\ and from $W=1000$ \kms\ for \MgII.
378: Five sources show the non-zero BI of \MgII\ consistent with $\sim
379: 10\%$ fraction of low ionization BAL QSOs. The maximum outflow
380: velocity are calculated, simultaneously. The BI and \vmax\ of \CIV\
381: absorption lines are listed in Table 1.
382:
383: Finally, we also remeasure the BI and \vmax\ for G06's sample on the
384: LBQS spectra
385: \citep{foltz87,foltz89,hewett91,chaffee91,morris91,hewett95} using
386: our method because we will combine G06's sample with ours in the
387: statistical analysis between the properties of X-ray and UV. We find
388: that our measurements of either BI or \vmax\ are well correlated
389: with those of G06 although there is considerable scatter. For six
390: objects both in this sample, the SDSS BAL QSO sample of
391: \citet{trump06} and in G06, our measurements appear in between
392: theirs. Note that the results of correlation analysis by using G06's
393: BI and \vmax\ are very similar to those obtained by using ours.
394:
395: \section{RESULTS AND DISCUSSION}\label{sec_res}
396:
397: \subsection{X-ray Properties Of BAL QSOs}\label{sec xrayprop}
398:
399: To investigate the general X-ray properties of our SDSS/\xmm\ BAL
400: sample (this paper), we try to measure the intrinsic absorption
401: adopting a simple neutral absorption. However, only 12 of 41 sources
402: can be fitted directly to give the intrinsic absorption column
403: densities \nh\ in the range $\sim 4\times10^{21}$ to $\sim
404: 2\times10^{23}$\cmsq. For 14 sources, upper/lower limits can be
405: placed by the hardness ratios HR. The final sample spans a wide
406: range of intrinsic absorption column densities from $<10^{20}$\cmsq\
407: to $\sim 10^{24}$\cmsq\ (Table 3). The lowest limit is obtained for
408: the LoBAL QSO, SDSS J092238.43+512121.2. We have rechecked the
409: optical spectrum, the identification of this quasar as LoBAL might
410: be questionable because of the presence of narrow absorption lines.
411: Notably, five LoBAL QSOs do not show stronger absorption than HiBAL
412: QSOs.
413:
414: Following G06,we measure \aox\ or place upper/lower limits on it,
415: which ranges from $-1.36$ to $-2.26$ with an average $-1.86$ (Table
416: 3). Similar to G06, we show the \daox, to account for the luminosity
417: dependence of \aox(Table 3,see also the dot-dashed line in Fig.
418: \ref{fig_xdis}). The average value of \daox\ is $-0.25$, suggesting
419: that 2~keV X-ray luminosities (at rest frame) of our SDSS/\xmm\ BAL
420: sample are roughly three times fainter than the SDSS/\rosat\ non-BAL
421: sample\citep{stra05}. And in Table 3 we also present the \aoxcorr\
422: and \daoxcorr\ as a surrogate of the intrinsic X-ray properties of
423: BAL QSOs. \aoxcorr\ is calculated by assuming $\Gamma=2.0$ and using
424: the hard-band counts rate to normalize the X-ray continuum and
425: \daoxcorr\ =\aoxcorr\ $-$ \aoxl\ (see G06 or \S3.1 for the
426: definition). We find \daoxcorr\ in the range from $-0.36$ to $0.29$
427: with an average value of 0.11, which indicates our SDSS/\xmm\ BAL
428: sample is slightly X-ray brighter, relative to the average quasars
429: at that UV luminosity, than the SDSS/\rosat\ non-BAL
430: sample\citep[see Fig. \ref{fig_xdis}]{stra05}.The X-ray brighter of
431: our sample may be due to the relative shallower detection threshold
432: of \xmm\ relative to \chandra . Comparing the \daoxcorr\
433: distribution of the LBQS/\chandra\ BAL sample(G06) with the
434: SDSS/\rosat\ non-BAL sample\citep[figure 2 of G06]{stra05} one can
435: find the LBQS/\chandra\ BAL sample(G06) is slightly intrinsic X-ray
436: weaker with a median \daoxcorr$=-0.14$ than the normal QSOs.
437: Alternately, even hard X-rays are absorbed in the LBQS/\chandra\ BAL
438: sample(G06) so that the simple assumption is broken down (See G06 or
439: \S3.1 for detail). We have carried out simulations to test this
440: effect. Using \xspec, we simulate the dependence of the \daoxcorr\
441: on the varying neutral hydrogen column density \nh. We assume that
442: \daoxcorr\ equals to zero for a single $\Gamma=2$ power-law with
443: \nh$=10^{20}$\cmsq\ at the redshift 2. We find that an absorption
444: column density of $3\times10^{23}$\cmsq\ is required in order to
445: account for the mean offset, about $-0.14$, of \daoxcorr\ of the
446: LBQS/\chandra\ BAL sample(G06) relative to the SDSS/\rosat\ non-BAL
447: sample\citep{stra05}. If this is the main cause, most of X-ray weak
448: sources in the LBQS/\chandra\ BAL sample(G06) will have a column
449: density at least order of this. Future X-ray observation is
450: certainly needed to assess this.
451:
452: %It means
453: %the method to correct the absorption is only unsuitable for small
454: %fraction of QSOs.
455:
456: Either intrinsic X-ray weak or large column density of absorber may
457: indicate that the LBQS/\chandra\ BAL sample(G06) is biased in X-ray
458: properties. Their sample obviously has larger values of BI and
459: \vmax\ than the SDSS BAL QSOs\citep{reichard03} and is not uniform
460: on UV properties too. Note that our SDSS/\xmm\ BAL sample is more
461: uniform, especially on UV properties, and can be used as a
462: complement to the LBQS/\chandra\ BAL sample(G06). For direct
463: comparison, we show the distribution of \daox\ for the SDSS/\rosat\
464: non-BAL sample\citep{stra05} and \daoxcorr\ for our SDSS/\xmm\ BAL
465: sample(this paper) and the LBQS/\chandra\ BAL QSO sample(G06) in
466: Fig. \ref{fig_xdis}. In the following we used the combined sample of
467: ours 41 and G06's 35 sources to study the relations between X-ray
468: and UV properties. Note again, the UV properties of the
469: LBQS/\chandra\ BAL QSO sample(G06) used in this paper are obtained
470: by using our procedures so that we can use the consistent definition
471: of BI and \vmax. We also use the hardness ratios presented in G06 to
472: calculate \nh\ of these G06 QSOs following the same approach as we
473: have done for \xmm\ sources.
474:
475: \subsection{X-ray And UV Absorptions}\label{sec_ab}
476:
477: One of the purposes of this paper is to study the relationship between
478: the UV and X-ray absorbers. It is generally believed that the
479: X-ray absorber shields the disk winds from soft X-rays and
480: makes line driving more efficient. A naive deduction is
481: that the properties of UV and X-ray absorptions are correlated.
482: Basing on this idea, G06 presented a correlation analysis between
483: the X-ray absorption using \daox\ as an indicator and the UV
484: absorption properties such as BI, DI, \vmax\ and $f_{\rm deep}$.
485: They found only a weak correlation between \daox\ and \vmax. We
486: will carry out a similar analysis using a larger sample covering
487: more uniformly the whole BI range. In the following analysis, we
488: will use Kendall-$\tau$ test to quantify the significance of a
489: correlation.
490:
491: First, we check whether \daox\ is a good indicator of X-ray
492: absorption. In the left panel of the Fig. \ref{fig_nhaox}, we show
493: \nh\ versus \daox\ for the combined subsample of 51 BAL QSOs that
494: \nh\ is obtained either from spectral fit or from HR analysis.
495: Similar to G06, we find a clear correlation between the two
496: quantities. The probability for null hypothesis is less than 0.01\%~
497: using non-parametric Kendall $\tau$-test (See Table 4). Then we
498: compare the redshift distributions of sources with \daox\ $> -0.2$
499: and \daox\ $< -0.2$.The result is their distributions are very
500: similar,which indicates that the correlation between \nh\ and \daox\
501: is not from selection effect of redshift.These suggest that \daox\
502: can be used as a measure for X-ray absorption indeed.
503:
504: Next, we examine the correlations between BAL properties and the
505: X-ray absorption column density \nh\ (Fig. \ref{fig_nhuv})
506: measured through X-ray spectral fit or HR analysis. We do not find
507: any correlation with a probability of null hypothesis less than
508: 1\% (Table 4). However, a weak correlation between BI and \nh\
509: cannot be rejected because the large uncertainty in the \nh\
510: measurement may reduce the significance of a weak correlation to
511: the measured level (2\%).
512:
513: We show \daox\ versus BI and \daox\ versus \vmax\ in Fig.
514: \ref{fig_daouv}. Two lensed BAL QSOs are excluded from following
515: analysis because their UV and X-ray light may have been differntly
516: amplified. There appears a correlation between \daox\ and BI with
517: the probability for null hypothesis of only 0.05\% (Table 4). The
518: correlation appears not linear, rather there is an upper envelope.
519: Since LoBAL QSOs may be different from the HiBAL QSOs
520: \citep{boroson92,wang07}, we also make Kendall test for 45 HiBAL
521: QSOs only. The correlation is marginally significant with a null
522: probability of 1\%. The decrease in significance is caused by
523: reducing the sample size. However, we do not find any significant
524: correlation between \vmax\ and \daox, which was seen in G06, in
525: neither the whole sample nor in the HiBAL subsample with a null
526: probability of 5\% and 68\%, respectively (Table 4). Comparison with
527: the LBQS/\chandra\ BAL sample(G06), our sample has a handful BAL
528: QSOs on the upper right of the figure. These BAL QSOs destroy the
529: weak correlation trend of \vmax\ vs \daox\ in the LBQS/\chandra\ BAL
530: sample(G06). These correlations are more or less similar to the
531: correlations using \nh\ with an exception of higher significance. It
532: is worthwhile to note that \daox\ reflects a combination of the
533: X-ray absorption and the intrinsic deviation to the average quasar
534: SED. Therefore, one must be careful as using it as an indicator of
535: X-ray absorption. We will discuss below the implication of these
536: results.
537:
538:
539: \subsection{Intrinsic X-ray Properties And The Outflow}
540:
541: Previous studies have shown that UV properties, such as the
542: blueshift and the equivalent width of \CIV\ emission line, are
543: correlated with X-ray to optical flux ratio for non-BAL QSOs (e.g.
544: \citet{baskin04,richards06}). It would be interesting to explore
545: whether the BAL properties are correlated with the intrinsic \aox.
546: Unlike the correlation with X-ray absorption, such correlation
547: should give information for the primary driver of the outflow. Here
548: we use a corrected \aox\, i.e. \aoxcorr\ to represent the intrinsic
549: \aox\, and investigate its relation with the UV absorption line
550: properties. We also study the correlations between UV properties and
551: \daoxcorr\ so that we can compare the results with those in previous
552: section and G06. We note that the range of UV luminosity,
553: $l_{2500}$, for the combined sample is only 2 dex, which introduce a
554: scatter in \aoxcorr\, through \aox$\sim l_{2500}$ correlation, of
555: less than 0.27, a factor of about 2.5 smaller than
556: the dynamic range of \aoxcorr.
557: Therefore, the difference between \daoxcorr\ and \aoxcorr\ should
558: be small in correlation analysis. This is verified below
559: that the relationships between \daoxcorr\ and UV properties have a
560: very similar behavior as those between \aoxcorr\ and UV properties.
561:
562: Before exploring UV and X-ray connection, we first check whether
563: \aoxcorr\ and \daoxcorr\ are affected by the X-ray absorption or
564: not. We plot \nh\ versus \daoxcorr\ on the right panel of Fig.
565: \ref{fig_nhaox}. There is no apparent correlation between the two
566: quantities. Kendall test gives a probability of chance coincidence
567: of 36\% (Table 4). Therefore, we can conclude that there is no
568: evidence that \aoxcorr\ and \daoxcorr\ are significantly affected by
569: X-ray absorption.
570:
571: We then explore the correlations between \aoxcorr\ and BI or
572: \vmax\ with Kendall and Spearman tests. We find that \aoxcorr\ is
573: significantly correlated with both BI and \vmax\ with a Null
574: probability of less than 0.1\% for either test(See Fig 5; also
575: Table 4). The correlation is still very significant ($P<0.1\%$)
576: for HiBAL QSO subsample (45 QSOs). For clarity we show only QSOs
577: detected in the hard X-ray band in Fig.\ref{fig_aocuvd}.
578: Furthermore, these QSOs are more important for the Kendall and
579: Spearman tests than the rest objects, and can give us a clear
580: trend about these correlations. We note that one LoBAL QSOs, SDSS
581: J133553.61+514744.1, which appears largely discrepant with the
582: main sample in Fig. \ref{fig_aocuv} and \ref{fig_aocuvd}. This
583: quasar has very steep X-ray photon index $\Gamma\sim2.56$ so that
584: our `absorption-correction' underestimate the intrinsic X-ray
585: luminosity. If we set \aox$=-1.82$ as the lower limit of \aoxcorr
586: (it is reasonable since \aoxcorr\ is larger than \aox), this
587: quasar would move rightward and is consistent with other quasars.
588: The correlations between \daoxcorr\ and UV properties are very
589: similar to above correlations using \aoxcorr\ except the latter
590: appear slightly more significant (Table 4). This may indicate
591: that the dependence of UV properties on \aoxcorr\ is more
592: fundamental than on \daoxcorr.
593:
594: Is it possible that these correlations are introduced by some
595: selection effect in the sample? If it is the case this effect would
596: tend to miss the objects which occupy the bottom-left (X-ray weak
597: and high UV absorption) and top-right (X-ray strong and low UV
598: absorption) corners of Fig. \ref{fig_aocuv} and \ref{fig_aocuvd}.
599: Note that the sample selections of ours and G06's are both based on
600: the optical luminosity and have nothing to do with the X-ray
601: properties. If the relative X-ray luminosity is uncorrelated with
602: BAL properties it is hard to understand why G06 and we select the
603: objects at the bottom-right(top-left) corner but miss those at the
604: bottom-left(top-right) corner. Since objects at top-right
605: (bottom-left) corners, if they really exist, should have the same
606: optical properties as these at top-left (bottom-right). All of these
607: analysis indicate the correlations between the intrinsic \aox\ and
608: the properties of UV absorber are real. We discuss the implications
609: of the correlations in details in the next subsection.
610:
611:
612: \subsection{Discussion}\label{sec_dis}
613:
614: Using a larger and more uniform sample, we reexamine the
615: correlations between BAL properties and X-ray absorption presented
616: in G06. Two indicators of X-ray absorption, \nh\ and \daox, are used
617: in the work. We identify the correlation between \daox\ and $BI$ as
618: the only significant one. In particular, we do not find the
619: correlation between \daox\ and \vmax\ claimed in G06, and any
620: correlation between \nh\ and the UV absorption properties. Although
621: we can not rule out a weak correlation between \nh\ and the UV
622: absorption line properties due to relative large error bar of \nh\,
623: our results clearly suggest that X-ray absorption is {\em not} the
624: major factor that determines UV absorption properties. Given the
625: fact that almost all BAL QSOs show strong absorption in X-ray, it
626: seems that X-ray absorption is a necessary condition for launching
627: of the BAL winds, but the properties of the wind depend on other
628: factors. As shown above, the observed correlation between \daox\ and
629: BI may be the secondary effect of the correlation between BI and
630: \aoxcorr\ or \daoxcorr, as \daox\ is composed of the contributions
631: of absorption and of \daoxcorr.
632:
633: In passing, we note that lack of correlations between the X-ray
634: absorption column density and UV properties does not necessarily
635: contradict with the scenario of radiatively accelerated wind as
636: naively thought. For locally optically thin material, the ratio of
637: the radiation force to the gravitational force is a function of
638: Eddington ratio and the cross-section ratio of effective absorption
639: to Thomson scattering. If resonant scattering is responsible for the
640: absorption opacity, the cross-section will be determined by the
641: fraction Li-like ions. According to the equatorial wind
642: model\citep{murray95}, a clump of highly ionized gas (shielding
643: gas), which accounts for most X-ray opacity, blocks the soft X-ray
644: interior to the wind. The transmitted flux of soft X-rays that
645: ionize Li-like ions in the wind depends strongly on the X-ray column
646: density, \nh. If \nh\ along the direction is very small,
647: high-velocity wind cannot be launched because of the reduction of
648: the radiation force caused by the over-ionization. On the other
649: hand, if \nh\ is very large, the wind will end up with a turbulent
650: flow due to the blocking of thick very-low-ionized gas
651: behind\citep[see the figure 4 of ][]{proga00}. Thus, high-velocity
652: wind can only be launched when the radial column density \nh\ is
653: moderate as the fraction of Li-like ions, such as \CIV\ and \NV, is
654: large enough. As far as the X-ray absorption column density is in
655: the right range, the fraction of Li-like ions should be the dominant
656: species. The flow properties are then determined self-consistently
657: by the launching radius, the gas density at the launching radius and
658: the radiation intensity. If X-ray absorber is well separated from
659: the UV absorber, then we would not expect any correlation between
660: the X-ray absorption column density and the flow properties for BAL
661: QSOs apart.
662:
663: On the other hand, the X-ray absorber may
664: be the 'hitchhiking' gas just located at
665: the inner edge of the wind, and its properties may have a close
666: connection with the boundary conditions of the disk wind\citep{murray95}.
667: In that case, we should consider globally the structure of gas along
668: a line of sight. The wind starts at a radius where the radiation force
669: is substantially larger than the gravitational force. As far as the gas
670: density is high enough, such a region can certainly exist. \citet{murray95}
671: has worked out a consistent line acceleration model, and they found
672: that gas column density and final velocity are correlated for a constant
673: Eddington ratio and at a given launch radius. However, if the launching
674: radius is not exactly scaled with luminosity as $L^{1/2}$ and there are a
675: range of Eddington ratio, as they assumed, the correlation can be smeared
676: out.
677:
678: More interesting results of our work are the strong correlations
679: between the parameters of outflow and intrinsic \aox. We argue that
680: these correlations are essential rather than due to some selection
681: effect or the secondary effect of other correlations (see previous
682: subsection for details). In fact our results are consistent with
683: \citet{richards06} who found that the QSOs with large blueshifts of
684: \CIV\ emission line, i.e. the parent population of BAL QSOs as
685: suggested by \citet{richards06}, tend to have lower X-ray luminosity
686: for given optical luminosity (their figure 5). It is also upheld by
687: \citet{laor02} who presented significant correlation between the
688: equivalent width of \CIV\ absorption and \aox\ in a sample of
689: non-BAL QSOs. This correlation is actually predicted by
690: \citet{murray95}, in which they found that quasars with a large
691: X-ray to UV ratio can only produce weak low velocity winds while
692: quasars with a small X-ray to UV ratio can produce strong and large
693: velocity winds. This is exactly what we have found here. As we
694: discussed above, their model also predicted a correlation between
695: the X-ray absorption column density and the maximum velocity of the
696: flow, which is not observed in this sample. Lack of such correlation
697: may be due to two important factors that (1) variation in the
698: Eddington ratio and launching radius; (2) the large uncertainties in
699: the measurement of absorption column density.
700:
701: We do not fully understand why the variation in the Eddington
702: ratio and launching does not completely smeared out the correlation
703: with \aoxcorr. There seems one reason for this.
704: \citet{wang04} found that the 2-10kev luminosities
705: to bolometric luminosities ratio tightly anti-correlated with
706: Eddington ratio for a sample of broad-line and narrow-line Seyfert 1
707: AGNs. If this correlation holds up for BAL QSOs, one would expect
708: that quasars with high Eddington ratio would have larger radiative
709: acceleration force, or large terminal velocity, and at the same time
710: X-ray weaker. \citet{ganguly07} find that \vmax\ as a function of
711: Eddington ratio has an upper envelope in the SDSS3 BAL catalog,
712: exactly as expected. Since there is no clear correlation of BI with
713: UV luminosities\citep[cf.][]{laor02}, Eddington rate and black hole
714: mass\citep{ganguly07},it is very likely that the BAL properties
715: are more likely determined by the SED of quasars rather than Eddington ratio.
716:
717:
718: \section{SUMMARY}\label{sec_sum}
719:
720: We compile a large \CIV\ BAL QSOs sample from the \xmm\ archive data
721: and SDSS DR5. The sample consists of 41 BAL QSOs, among which 26
722: QSOs are detected in the X-ray band. Our sample spans wide and
723: homogeneous ranges of both BI and \vmax\ and can be used to
724: complement the LBQS/\chandra\ BAL sample(G06). In addition, the
725: combined sample of ours and G06s show a more homogeneous
726: distribution of intrinsic X-ray properties than G06s. Using this
727: combined sample, we investigate the correlations between X-ray and
728: UV properties of BAL QSOs.
729:
730: We briefly summarize our conclusions below:
731: \begin{enumerate}
732: \item{We confirm the previous results that BAL QSOs are generally
733: soft X-ray weak, which is mainly due to the intrinsic X-ray
734: absorption. We also find the X-ray luminosities of BAL QSOs with
735: given optical luminosity have large scatter. The scatter is
736: caused by both the various column densities of X-ray absorber and
737: the scatter of intrinsic X-ray emission at given optical luminosity.}
738: \item{We do not find any evidence for the claimed correlation between the BAL
739: properties and soft X-ray absorption, with an exception of the
740: correlation between BI and \daox. The correlation between BI and
741: \daox\ can be induced by the correlation between BI and the
742: intrinsic \aox. The X-ray absorber is important for launching the
743: high-velocity wind but do not directly determine the BAL
744: properties.}
745: \item{There are significant correlations between intrinsic X-ray strength,
746: \aoxcorr, and BI and \vmax\ in the combined sample. These
747: correlations are essential rather than due to any artificiality. We
748: preliminarily interpret that the BAL properties are influenced by
749: the intrinsic SED of QSOs, which is consistent with the prediction
750: of a radiatively accelerated disk wind model \citep{murray95}.}
751: \end{enumerate}
752:
753:
754: \acknowledgements We thank the anonymous referee for helpful comments.We thank
755: Paul Hewett for providing electronic data of LBQS spectra.We also acknowledge
756: the Sloan Digital Sky Survey(http://www.sdss.org).
757: This work was supported by the Knowledge
758: Innovation Program of the Chinese Academy of Sciences, Grant No.
759: KJCX2-YW-T05 and the National Basic Research Program of China (973 Program)
760: under Grant No. 2007CB815400.
761:
762: \begin{thebibliography}{}
763: \bibitem[Adelman-McCarthy et al.(2008)]{adelman07}
764: Adelman-McCarthy, J.~K., et al.\ 2008, \apjs, 175, 297
765: \bibitem[Aldcroft \& Green (2003)]{aldcroft03}Aldcroft, T. L. \&
766: Green, P. J., 2003,\apj ,592,710
767: \bibitem[Arnaud(1996)]{arnaud96} Arnaud, K. A., 1996, in ASP Conf. Ser. 101:
768: Astronomical Data Analysis Software and Systems V,eds. G. Jacoby \&
769: J. Barnes vol. 5,17
770: \bibitem[Avni \& Tananbaum(1986)]{avni86} Avni, Y., \& Tananbaum, H.\ 1986, \apj, 305, 83
771: \bibitem[Bade et al.(1997)]{bade97} Bade, N., Siebert, J., Lopez, S., Voges, W., \& Reimers, D.\ 1997, \aap, 317, L13
772: \bibitem[Baskin \& Laor(2004)]{baskin04} Baskin, A., \& Laor, A.\ 2004, \mnras, 350, L31
773: \bibitem[Boroson \& Meyers(1992)]{boroson92} Boroson, T.~A., \& Meyers, K.~A.\ 1992, \apj, 397, 442
774: \bibitem[Brandt,Laor \& Wills(2000)]{brandt00} Brandt, W.~N., Laor, A.,
775: \& Wills, B.~J.\ 2000, \apj, 528, 637
776: \bibitem[Brinkmann et al.(1999)]{brinkmann99} Brinkmann, W., Wang,
777: T., Matsuoka, M., \& Yuan, W.\ 1999, \aap, 345, 43
778: \bibitem[Brotherton et al.(2006)]{brotherton06} Brotherton, M.~S.,
779: De Breuck, C., \& Schaefer, J.~J.\ 2006, \mnras, 372, L58
780: \bibitem[Cash(1979)]{cash79}Cash,W., 1979, \apj, 228,939
781: \bibitem[Chaffee et al.(1991)]{chaffee91} Chaffee, F.~H., Foltz,
782: C.~B., Hewett, P.~C., Francis, P.~A., Weymann, R.~J., Morris, S.~L.,
783: Anderson, S.~F., \& MacAlpine, G.~M.\ 1991, \aj, 102, 461
784: \bibitem[Chartas et al.(2003)]{chartas03}Chartas, G.,Brandt,
785: W. N. \& Gallagher, S. C., 2003, \apj, 595,85
786: \bibitem[Chartas et al.(2002)]{chart02}Chartas, G.,Brandt, W. N.,Gallagher, S. C. \&
787: Garmire, G. P., 2002, \apj, 579 , 169
788: \bibitem[Dickey \& Lockman(1990)]{dickey90} Dickey,J. M. \& Lockman, F. J., 1990, \araa, 28,385
789: \bibitem[Foltz et al.(1987)]{foltz87} Foltz, C.~B., Chaffee,
790: F.~H., Jr., Hewett, P.~C., MacAlpine, G.~M., Turnshek, D.~A., Weymann,
791: R.~J., \& Anderson, S.~F.\ 1987, \aj, 94, 1423
792: \bibitem[Foltz et al.(1989)]{foltz89} Foltz, C.~B., Chaffee,
793: F.~H., Hewett, P.~C., Weymann, R.~J., Anderson, S.~F., \& MacAlpine, G.~M.\
794: 1989, \aj, 98, 1959
795: \bibitem[Gallagher et al.(1999)]{gall99} Gallagher, S.~C.,
796: Brandt, W.~N., Sambruna, R.~M., Mathur, S., \& Yamasaki, N.\ 1999,
797: \apj, 519, 549
798: \bibitem[Gallagher et al.(2002)]{gall02} Gallagher, S. C., Brandt,
799: W. N., Chartas, G. \& Garmire, G. P., 2002, \apj, 567 , 37
800: \bibitem[Gallagher et al.(2006)]{gall06} Gallagher, S.~C.,
801: Brandt, W.~N., Chartas, G., Priddey, R., Garmire, G.~P., \&
802: Sambruna, R.~M.\ 2006, \apj, 644, 709
803: \bibitem[Gallagher et al.(2004)]{gall04} Gallagher, S.~C.,
804: Brandt, W.~N., Wills, B.~J., Charlton, J.~C., Chartas, G., \&
805: Laor, A.\ 2004, \apj, 603, 425
806: \bibitem[Ganguly et al.(2007)]{ganguly07} Ganguly, R.,
807: Brotherton, M.~S., Cales, S., Scoggins, B., Shang, Z., \&
808: Vestergaard, M.\ 2007, \apj, 665, 990
809: \bibitem[Gehrels (1986)]{gehr86}Gehrels,N.,1986,\apj, 303,336
810: \bibitem[George et al.(2000)]{george00} George, I. M., Turner, T.
811: J., Yaqoob, T., Netzer, H., Laor, A., Mushotzky, R. F., Nandra, K.,
812: \& Takahashi, Y., 2000, \apj, 531,52
813: \bibitem[Green et al.(1995)]{green95} Green, P.~J., et al.\
814: 1995, \apj, 450, 51
815: \bibitem[Grupe et al.(2003)]{grupe03}Grupe, D., Mathur, S., \&
816: Elvis, M., 2003,\aj,126,1159
817: \bibitem[Hewett \& Foltz(2003)]{hewett03} Hewett, P.~C., \&
818: Foltz, C.~B.\ 2003, \aj, 125, 1784
819: \bibitem[Hewett et al.(1995)]{hewett95} Hewett, P.~C., Foltz,
820: C.~B., \& Chaffee, F.~H.\ 1995, \aj, 109, 1498
821: \bibitem[Hewett et al.(1991)]{hewett91} Hewett, P.~C., Foltz,
822: C.~B., Chaffee, F.~H., Francis, P.~J., Weymann, R.~J., Morris, S.~L.,
823: Anderson, S.~F., \& MacAlpine, G.~M.\ 1991, \aj, 101, 1121
824: \bibitem[Ivezi{\'c} et al.(2002)]{ivezi02} Ivezi{\'c}, {\v Z}.,
825: et al.\ 2002, \aj, 124, 2364
826: \bibitem[Kraft et al.(1991)]{kraft91}Kraft,R.P.,Burrows,D.N.,\&
827: Nousek,J.A.,1991,\apj,374,344
828: \bibitem[Laor \& Brandt(2002)]{laor02} Laor, A., \& Brandt,
829: W.~N.\ 2002, \apj, 569, 641
830: \bibitem[Morris et al.(1991)]{morris91} Morris, S.~L., Weymann,
831: R.~J., Anderson, S.~F., Hewett, P.~C., Francis, P.~J., Foltz, C.~B.,
832: Chaffee, F.~H., \& MacAlpine, G.~M.\ 1991, \aj, 102, 1627
833: \bibitem[Morrison \& McCammon(1983)]{morrison83} Morrison, R. \&
834: McCammon, D., 1983, \apj , 270 ,119
835: \bibitem[Murray et al.(1995)]{murray95} Murray, N., Chiang, J.,
836: Grossman, S.~A., \& Voit, G.~M.\ 1995, \apj, 451, 498
837: \bibitem[Ogle et al.(1999)]{ogle99} Ogle, P.~M., Cohen, M.~H.,
838: Miller, J.~S., Tran, H.~D., Goodrich, R.~W., \& Martel, A.~R.\ 1999,
839: \apjs, 125, 1
840: \bibitem[Page et al.(2005)]{page05}Page, K. L., Reeves, J.
841: N.,O\'Brien, P. T.,\& Turner, M. J. L.,2005,\mnras,898
842: \bibitem[Park et al.(2006)]{park06} Park, T., Kashyap, V.~L.,
843: Siemiginowska, A., van Dyk, D.~A., Zezas, A., Heinke, C., \&
844: Wargelin, B.~J.\ 2006, \apj, 652, 610
845: \bibitem[Proga et al.(2000)]{proga00} Proga, D., Stone, J.~M.,
846: \& Kallman, T.~R.\ 2000, \apj, 543, 686
847: \bibitem[Reeves \& Turner(2000)]{reeves00} Reeves, J. N. \& Turner,
848: M. J. L., 2000,\mnras, 316,234
849: \bibitem[Reichard et al.(2003)]{reichard03} Reichard, T.~A., et
850: al.\ 2003, \aj, 125, 1711
851: \bibitem[Richards(2006)]{richards06} Richards, G.~T.\ 2006, ArXiv
852: Astrophysics e-prints, arXiv:astro-ph/0603827
853: \bibitem[Shemmer et al.(2005)]{shemmer05} Shemmer, O., Brandt,
854: W.~N., Gallagher, S.~C., Vignali, C., Boller, T., Chartas, G.,
855: \& Comastri, A.\ 2005, \aj, 130, 2522
856: \bibitem[Steffen et al.(2006)]{steffen06} Steffen, A.~T.,
857: Strateva, I., Brandt, W.~N., Alexander, D.~M., Koekemoer, A.~M.,
858: Lehmer, B.~D., Schneider, D.~P., \& Vignali, C.\ 2006, \aj, 131,
859: 2826
860: \bibitem[Strateva et al.(2005)]{stra05}Strateva,I.V.,Brandt,
861: W.N.,Schneider,D.P.,Vanden Berk,D.G.,\& Vignali,C. 2005,\aj,130,387
862: \bibitem[Tananbaum et al. (1979)]{tana79}Tananbaum,H.,et al. 1979,\apj,234,L9
863: \bibitem[Tang et al.(2007)]{tang07} Tang, S.~M., Zhang, S.~N.,
864: \& Hopkins, P.~F.\ 2007, \mnras, 377, 1113
865: \bibitem[Trump et al.(2006)]{trump06} Trump, J.~R., et al.\
866: 2006, \apjs, 165, 1
867: \bibitem[Vanden Berk et al.(2001)]{vanden01} Vanden Berk, D.~E.,
868: et al.\ 2001, \aj, 122, 549
869: \bibitem[Wang et al.(2008)]{wang08} Wang, J., Jiang, P., Zhou,
870: H., Wang, T., Dong, X., \& Wang, H.\ 2008, \apjl, 676, L97
871: \bibitem[Wang et al.(2004)]{wang04} Wang, J.-M., Watarai,
872: K.-Y., \& Mineshige, S.\ 2004, \apjl, 607, L107
873: \bibitem[Wang et al.(2000)]{wang00} Wang, T.~G., Brinkmann,
874: W., Yuan, W., Wang, J.~X., \& Zhou, Y.~Y.\ 2000, \apj, 545, 77
875: \bibitem[Wang et al.(1999)]{wang99} Wang, T.~G., Wang, J.~X.,
876: Brinkmann, W., \& Matsuoka, M.\ 1999, \apjl, 519, L35
877: \bibitem[Wang et al.(2005)]{wang05} Wang, H.-Y., Wang, T.-G.,
878: \& Wang, J.-X.\ 2005, \apj, 634, 149
879: \bibitem[Wang et al.(2007)]{wang07} Wang, H.-Y., Wang, T.-G.,
880: \& Wang, J.-X.\ 2007, \apjs, 168, 195
881: \bibitem[Weymann et al.(1991)]{weymann91}Weymann,R.J.,
882: Morris,S.L.,Foltz,C.B.,\& Hewett,P,C., 1991,\apj,373,23
883: \bibitem[White et al.(1997)]{white97} White, R.~L., Becker,
884: R.~H., Helfand, D.~J., \& Gregg, M.~D.\ 1997, \apj, 475, 479
885: \bibitem[Wilkes et al.(1994)]{wilkes94} Wilkes, B.~J.,
886: Tananbaum, H., Worrall, D.~M., Avni, Y., Oey, M.~S., \& Flanagan,
887: J.\ 1994, \apjs, 92, 53
888: \bibitem[Yuan et al.(1998)]{yuan98} Yuan, W., Siebert, J., \& Brinkmann, W.\ 1998,\aap, 334, 498
889: \bibitem[Zhou et al.(2006)]{zhou06} Zhou, H., Wang, T., Wang,
890: H., Wang, J., Yuan, W., \& Lu, Y.\ 2006, \apj, 639, 716
891:
892:
893: \end{thebibliography}
894:
895:
896: \clearpage
897: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
898: \input{tab1}
899: \clearpage
900: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
901: \input{tab2}
902: \clearpage
903: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
904: \input{tab3}
905: \clearpage
906: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
907: \clearpage
908: \input{tab4}
909: \clearpage
910: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
911:
912:
913: \begin{figure}[t]
914: \plotone{f1.eps}\caption{The top panel shows the distribution of
915: observed \daox\ = \aox\ $-$ \aoxl\ for the SDSS/\rosat\ non-BAL
916: sample\citep{stra05}.The three lower panels show the distributions
917: of \daoxcorr\ = \aoxcorr\ $-$ \aoxl\ for the LBQS/\chandra\ BAL
918: sample(G06),the SDSS/\xmm\ BAL sample and the combined sample(this
919: paper),respectively.For all four panels,solid lines indicate the
920: full samples and dashed lines only show upper limits.Dot-dashed line
921: in the three lower panels represents the distribution of \daox.The
922: arrow in the top panel shows the direction of the X-ray weak objects
923: for our convention of \aox. }\label{fig_xdis}
924: \end{figure}
925:
926: \begin{figure}[t]
927: \plottwo{f2a.eps}{f2b.eps}\caption{The left panel shows the \nh\ vs.
928: \daox\ = \aox\ $-$ \aoxl\ and the right panel shows the \nh\ vs.
929: \daoxcorr\ = \aoxcorr\ $-$ \aoxl\ for the detected BAL QSOs in the
930: combined sample.SDSS BAL QSOs in our sample are shown with squares
931: and LBQS BAL QSOs in G06's sample are shown with diamonds.The
932: open,filled and half-filled symbols indicate HiBALs,LoBALs and BAL
933: QSOs of unknown type,respectively. }\label{fig_nhaox}
934: \end{figure}
935:
936: \begin{figure}[t]
937: \plottwo{f3a.eps}{f3b.eps}\caption{Plots of \daox\ = \aox\ $-$
938: \aoxl\ vs. \CIV\ absorption-line parameters for the combined sample
939: : BALnicity index(BI , left panel) and maximum outflow velocity of
940: absorption,(\vmax\ , right panel).Symbols are the same as in Fig.
941: \ref{fig_nhaox} .}\label{fig_daouv}
942: \end{figure}
943:
944: \begin{figure}[t]
945: \plottwo{f4a.eps}{f4b.eps}\caption{Plots of \nh\ vs. \CIV\
946: absorption-line parameters for the detected BAL QSOs in the combined
947: sample : BALnicity index(BI , left panel) and maximum outflow
948: velocity of absorption,(\vmax\ , right panel).Symbols are the same
949: as in Fig. \ref{fig_nhaox} .}\label{fig_nhuv}
950: \end{figure}
951:
952:
953: \begin{figure}[t]
954: \plottwo{f5a.eps}{f5b.eps} \caption{Plots of \aoxcorr\ vs. \CIV\
955: absorption-line parameters for the combined sample : BALnicity
956: index(BI , left panel) and maximum outflow velocity of
957: absorption,(\vmax\ , right panel).Symbols are the same as in Fig.
958: \ref{fig_nhaox} .}\label{fig_aocuv}
959: \end{figure}
960:
961: \begin{figure}[t]
962: \plottwo{f6a.eps}{f6b.eps}\caption{Plots of \aoxcorr\ vs. \CIV\
963: absorption-line parameters for the BAL QSOs detected in the hard
964: X-ray band in the combined sample : BALnicity index(BI , left panel)
965: and maximum outflow velocity of absorption,(\vmax\ , right
966: panel).Symbols are the same as in Fig. \ref{fig_nhaox}
967: .}\label{fig_aocuvd}
968: \end{figure}
969:
970: \end{document}
971: