1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{epsfig}
3:
4: \shortauthors{Xue \& Cui} \shorttitle{X-ray Flares from Mrk~501}
5:
6: \begin{document}
7:
8: \title{X-RAY FLARES FROM MARKARIAN~501}
9:
10: \author{Yongquan Xue and Wei Cui}
11: \affil{Department of Physics, Purdue University, West Lafayette, IN 47907;
12: xuey@physics.purdue.edu, cui@physics.purdue.edu}
13:
14: \begin{abstract}
15:
16: Motivated by the recent finding of hierarchical X-ray flaring phenomenon
17: in Mrk 421, we conducted a systematic search for X-ray flares from Mrk 501,
18: another well-known TeV blazar, by making use of the rich {\em RXTE}
19: archival database. We detected flares over a wide range of timescales,
20: from months down to minutes, as in the case of Mrk 421. However, the flares
21: do not seem to occur nearly as frequently in Mrk 501 as in Mrk 421 on any
22: of the timescales. The flaring hierarchy also seems apparent in Mrk 501,
23: suggesting that it might be common among TeV blazars. The results seem to
24: imply a scale-invariant physical origin of the flares (large or small).
25: The X-ray spectrum of the source shows a general trend of hardening toward
26: the peak of long-duration flares, with indication of spectral hysteresis,
27: which is often seen in TeV blazars. However, the data are not of sufficient
28: quality to allow us to draw definitive conclusions about spectral variability
29: associated with more rapid but weaker flares. We critically examine a
30: reported sub-hour X-ray flare from Mrk 501, in light of intense background
31: flaring activity at the time of the observation, and concluded that the
32: flare is likely an artifact. On the other hand, we did identify a rapid
33: X-ray flare that appears to be real. It lasted only for about 15 minutes,
34: during which the flux of the source varied by about 30\%. Sub-structures
35: are apparent in its profile, implying variabilities on even shorter
36: timescales. Such rapid variabilities of Mrk 501 place severe constraints
37: on the physical properties of the flaring region in the jet, which have
38: serious implications on the emission models proposed for TeV blazars.
39:
40: \end{abstract}
41:
42: \keywords{BL Lacertae objects: individual (Markarian 501) ---
43: galaxies: active --- radiation mechanisms: non-thermal --- X-rays: galaxies}
44:
45: \section{Introduction}
46:
47: Mrk 501 is among the brightest and closest ($z=0.034$) extragalactic X-ray
48: sources in the sky. It is classified as a BL Lacertae object and is located
49: in the elliptical galaxy UGC 10599 (Stickel et al. 1993). BL Lac objects
50: belong to a more general class of radio-loud active galactic nuclei known
51: as blazars, which are characterized by rapid variability and non-thermal
52: emission at nearly all wavelengths. The emission from a blazar is generally
53: thought to be dominated by radiation from a relativistic jet that is
54: directed roughly along the line of sight (Urry \& Padovani 1995). The
55: spectral energy distribution (SED) of blazars invariably shows two
56: characteristic ``humps'' in the $\nu F_{\nu}$ representation, with one
57: located at optical--X-ray energies and the other at GeV--TeV energies
58: (Fossati et al. 1998). Mrk 501 is among a small number of blazars that
59: have been detected at TeV energies (e.g., Quinn et al. 1996; Bradbury et
60: al. 1997).
61:
62: For TeV blazars, there is a general correlation between fluxes at X-ray
63: and TeV energies (where the SED peaks), although exceptions have recently
64: been noted (Krawczynski et al. 2004; Cui et al. 2004). While the origin of
65: gamma rays is still being debated, there is a general consensus that
66: X-rays originate in the synchrotron radiation from highly relativistic
67: electrons in the jet. Two classes of models have been proposed to explain
68: gamma-ray emission from blazars. In the leptonic models, the TeV emission
69: is attributed to the synchrotron self-Compton (SSC) process (e.g., Marscher
70: \& Gear 1985; Maraschi et al. 1992). The SSC models are attractive for
71: their conceptual simplicity. They can quite naturally account for the
72: observed X-ray--TeV correlation, and have also enjoyed some success in
73: fitting the observed SEDs (e.g., Kataoka et al. 1999; Sambruna et al. 2000;
74: Petry et al. 2000; Krawczynski et al. 2002; Konopelko et al. 2003). However,
75: such models still face challenges, such as the presence of ``orphan TeV
76: flares'' (i.e., those TeV flares with no apparent counterparts in X-rays;
77: Krawczynski et al. 2004; Cui et al. 2004). In the hadronic models, the TeV
78: emission is probably due to synchrotron radiation from relativistic protons
79: in the jet (Aharonian 2000; M\"ucke et al. 2003), although other hadronic
80: processes might also contribute. Although the models are being pushed to
81: the limit by the observed rapid variability of TeV blzars, they are, by no
82: means, ruled out yet. It remains to be seen whether they can account for
83: the X-ray--TeV correlation in a quantitative manner.
84:
85: TeV blazars are known to undergo flaring activities both at X-ray and TeV
86: energies. The flares seem to occur on all timescales but vary greatly in
87: magnitude, as illustrated vividly by the X-ray flaring hierarchy in Mrk 421
88: (Cui 2004). The hierarchy strongly implies the scale-invariant nature of
89: flaring processes in the source, prompting comparison to solar flares and
90: seemingly similar rapid flares in stellar-mass black hole systems. The
91: flares from Mrk 501 have been well observed at X-ray and TeV energies (e.g.,
92: Quinn et al. 1996; Bradbury et al. 1997; Catanese et al. 1997; Hayashida
93: et al. 1998; Kataoka et al. 1999; Sambruna et al. 2000; Petry et al. 2000).
94: In fact, the detection of a very rapid X-ray flare from the source was
95: reported (Catanese \& Sambruna 2000), which represented the first case of
96: sub-hour X-ray flares in any TeV blazar. The short duration of the flare
97: would lead to severe constraints on the properties of the flaring region,
98: as in the case of Mrk 421 (Cui 2004).
99:
100: Motivated by the recent finding of hierarchical X-ray flaring phenomenon
101: in Mrk 421 (Cui 2004), we conducted a systematic search for X-ray flares
102: in Mrk 501 over a broad range of timescales, making use of the rich
103: {\em RXTE} archival database. We describe the data and data reduction
104: procedure in \S~2. The results are presented in \S~3, along with a
105: critical examination of the reported sub-hour X-ray flare. Finally,
106: in \S~4, we discuss the results and their implications on the proposed
107: models.
108:
109:
110: \section{Data and Data Reduction}
111:
112: Mrk 501 has been observed frequently with {\em RXTE}. For this work, we
113: obtained publicly available data from the archival databases,
114: including data both from the All-Sky Monitor (ASM) and the Proportional
115: Counter Array (PCA). We obtained the ASM light curves of Mrk~501 from the
116: MIT archive.\footnote{See http://xte.mit.edu/asmlc/srcs/mkn501.html\#data}
117: We chose to filter out data points with error bars (on the raw count rates
118: in the summed band) greater than 2.0 counts/s, which constitute about 5\%
119: of the data. We then weighted the rates by $1/\sigma^2$ and rebinned
120: them to produce light curves with 21-day time bins. The light curves are
121: available in three energy bands: 1.5--3, 3--5, and 5--12 keV, providing
122: crude spectral information.
123:
124: The PCA is a narrow-field, pointing instrument. It consists of five nearly
125: identical proportional counter units (PCUs) and covers a nominal energy
126: range of 2--60 keV. Due to operational
127: constraints, some of the PCUs were often switched off. Only PCU 0 and PCU 2
128: are nearly always in operation. Data analysis is further complicated by
129: the loss of the front veto layer in PCU 0 (in 2000), because the data from
130: PCU 0 is now more prone to contamination by events caused by low-energy
131: electrons entering the detector. This is particularly relevant to the study
132: of variability of weak sources like Mrk~501. Fortunately, it only affects a
133: about 15\% of the data used in this work. We excluded from timing analyses
134: PCU 0 data collected in 2000. Most PCA observations were made in ``snap-shot''
135: modes, with typical effective exposure times ranging from half a kilosecond
136: to a few kiloseconds. The PCA has numerous data modes and multiple modes
137: are usually employed in an observation. For this work, however, we only
138: used the {\em Standard 2} data, which have a time resolution of 16 s.
139:
140: We followed Cui (2004) closely in reducing and analyzing the PCA data.
141: Briefly, the data were reduced with {\em FTOOLS 5.2}. For an observation,
142: we first filtered data by following the standard procedure for faint
143: sources (see the online {\em RXTE} Cook Book),\footnote{See
144: http://heasarc.gsfc.nasa.gov/docs/xte/recipes/cook\_book.html.} which
145: resulted in a list of good time intervals (GTIs). We then simulated
146: background events for the observation with the latest background model
147: that is appropriate for faint sources (pca\_bkgd\_cmfaintl7\_eMv20031123.mdl).
148: Using the GTIs, we proceeded to extract a light curve from the data (by
149: combining all active PCUs) in each of the following energy bands: 2.0--5.7,
150: 5.7--11, 11--60, and 2.0--60 keV. Note that the boundaries of each band are
151: matched up as closely as possible across different PCA epochs but are only
152: approximate (up to $\pm$0.2 keV). We repeated the steps to construct the
153: corresponding background light curves from the simulated
154: events. Finally, we obtained light curves of the source by subtracting
155: off the background. Following a similar procedure, we constructed an X-ray
156: spectrum and the associated background spectrum for each observation. We
157: note that for spectral analysis we only used data from the first xenon layer
158: of each PCU (which is most accurately calibrated). For the purpose of
159: detailed spectral modeling (see \S~3.4), we included a 1\% systematic
160: uncertainty uniformly across the entire energy range of interest, to
161: take into account residual calibration uncertainties. Note that we did
162: include PCU 0 data from the 2000 observations in the spectral analysis
163: because the background model for the detector seems to work fine in all
164: cases.
165:
166: \section{Results}
167:
168: \subsection{Long X-ray Flares: Months to $>$ Year}
169:
170: Figure 1 shows the ASM light curve of Mrk 501 for the 1.5--12 keV band over
171: roughly an eight-year period. The source was very active in X-rays in 1997.
172: The activity continued, at a much lower level, in 1998 and also the early
173: part of 1999. Mrk 501 has been relatively quiet ever since. At least two
174: major flares are easily identifiable from the light curve,
175: which lasted for months to over a year. The flares show substantial
176: sub-structures, indicating variability on shorter timescales, perhaps in
177: the form of unresolved weaker and shorter flares.
178:
179: Though crude, hardness ratios make it possible to study, in a
180: model-independent manner, spectral variability of a source. Here, we
181: computed the ratio between the ASM count rate in the 3--12 keV band to
182: that in the 1.5--3 keV band to examine spectral evolution of Mrk 501
183: during the 1997 giant flare. The results are shown in Figure 2. For
184: clarity, we have averaged the raw ASM count rates with variable time bins
185: (6-week, 2-week, and 1-week), depending on the brightness of the source.
186: The figure clearly shows a hardening trend of the X-ray spectrum as the
187: source approaches the peak of the flare. This is consistent with
188: findings of previous works (e.g., Pian et al. 1998). Such a
189: hardness-intensity correlation seems to hold in general for X-ray bright
190: blazars (Giommi et al. 1990).
191:
192: To investigate the phenomenon of spectral hysteresis, which is often
193: observed in blazars (e.g., Takahashi et al. 1996, Kataoka et al. 2000,
194: Zhang 2002, Giebels et al. 2002, Falcone et al. 2004, Cui 2004)
195: but may be too subtle to be seen in the ASM data, we made a similar
196: hardness-intensity diagram from the PCA data. There are three observing
197: periods, with two covering the rising portion of the 1997 flare and one the
198: decaying portion, as shown in Figure~3 (top panel). Unfortunately, the
199: coverage of the flare with the PCA is quite sparse, especially during the
200: decaying phase. While there appears to be indication of spectral
201: hysteresis during the rising phase (perhaps associated with smaller
202: flares) and between the rising and decaying phases, it is difficult to
203: draw any definitive conclusions from these results. On the other hand,
204: the trend of spectral hardening with increasing fluxes is also apparent
205: in the PCA data.
206:
207: \subsection{Intermediate X-ray Flares: Days to Weeks}
208:
209: The PCA data span a period of 4 years, which covers weaker flaring activities
210: in 1998, 1999, and 2000, besides the giant flare in 1997 (see Fig.~1). The
211: light curves reveal the presence of X-ray flares with shorter durations, from
212: days to weeks, nearly all the time, as shown in Figure~4. Although the ASM
213: light curve shows that the source seems to be very quiet during the PCA
214: observations in 1999 and 2000, the more sensitive PCA light curves reveal
215: flaring activities during these time periods. Therefore, like Mrk 421, there
216: is no apparent ``quiescent'' state for Mrk 501 either. Unlike Mrk 421,
217: however, the flux of Mrk 501 does not seem to vanish, implying the presence
218: of steady-state emission. The amplitude of these intermediate flares can also
219: be very large, up to 4--5 times the steady-state flux. We attempted to
220: investigate spectral evolution and hysteresis of the source across individual
221: flares but found that the results were ambiguous, mainly due to the lack of
222: quality of the data in this case.
223:
224: \subsection{Rapid X-ray Flares: $<$ Hour}
225:
226: Catanese \& Sambruna (2000) reported the first case of a sub-hour X-ray
227: flare from Mrk 501. Surprisingly, however, we initially failed to find it in
228: the PCA light curves constructed from the same observation. Examining
229: the light curves more carefully, we realized that there was actually a
230: data gap almost exactly at the time of the flare. We suspected that the
231: data had probably been filtered out during the data cleaning process in our
232: case. We then relaxed the data filtering criteria one at a time and in
233: the end found that it was the high ``ELECTRON2'' values that deemed the
234: time interval (of the reported flare) ``bad''. In general, the ``ELECTRON''
235: parameter provides a measure of contamination of the data by events
236: induced by electrons entering the detector. Some of such events (due to
237: low energy electrons) may be registered as good X-ray events because
238: they fail to trigger any of the vetoing logics. Since the PCU 2 is always
239: on, we used ``ELECTRON2'' in the data cleaning procedure (as recommended
240: in the {\em RXTE} Cookbook).
241:
242: To show possible electron contamination of the data, we plotted the X-ray
243: count rates and the values of the ``ELECTRON2'' parameter for the
244: observation in Figure~5 (left panels). The figure clearly shows that there
245: is a strong electron flare nearly at the same time as the reported X-ray
246: flare, which raises the possibility of the latter being an artifact.
247: However, we note that the X-ray flare seems to have started about 200 s
248: before the electron flare and that the profile of the X-ray flare appears
249: asymmetric, unlike that of the electron flare. We cannot think of an
250: obvious explanation for either ``anomaly''. On the other hand, we found
251: another rapid X-ray flare (which has not been reported) from a different
252: observation (obsid: 30249-01-01-01), which also has an
253: asymmetric profile, is accompanied by an electron flare, and leads the
254: electron flare by about the same amount of time ($\sim$200 s), as also
255: shown in Fig.~5 (right panels) for a direct comparison. The presence of
256: two such flares raised serious doubts in our mind about them being
257: physically associated with Mrk 501.
258:
259: We then systematically searched for X-ray flares that were accompanied by
260: electron flares, by visually inspecting all of the light curves, and found
261: a total of 21 cases (although it is entirely possilble than some weaker
262: ones might have escaped our attention), including the two just discussed.
263: Interestingly, in all other 19 cases, we found no apparent timing offsets
264: between X-ray and electron flares. Figure~6 shows some representative
265: examples of those. It is clear, from the figure, that not all electron
266: flares are registered as X-ray signals and that the profile of an electron
267: flare can be complex, e.g., with multiple peaks. Moreover, we found that
268: the spectrum of the X-ray flares varies greatly. In extreme cases, a single
269: X-ray flare is seen in one energy band that coincides with one of the peaks
270: of an electron flare, but another flare in a different energy band that
271: coincides with a different peak of the same electron flare. Therefore, the
272: phenomenology is complex.
273:
274: Finally, we derived the latitude and longitude of the {\em RXTE} satellite
275: at the time of the peak of each of the 21 electron flares. Figure~7 shows
276: a map that summarizes the results. The two cases shown in Fig.~5 are
277: highlighted with different symbols. It should be noted that the peaks of
278: these two electrons flares are separated by almost exactly one satellite
279: orbit ($\approx$1.6 hours). In general, the electron flares seem to cluster
280: in the northeast Pacific region, which confirms what is generally known
281: about the occurrence of such events (K.~Jahoda and C.~Markwardt of the PCA
282: team, private communication). However, particle fluxes in the region are
283: usually over two orders of magnitude smaller than those in the well-known
284: South Atlantic Anomaly (SAA) region, which is why the region is not
285: screened in the same manner as the SAA. Putting
286: together all the evidences, we think that the reported rapid flare (and
287: also the one found in this work) is most likely an artifact caused by
288: intense flaring in soft electron fluxes. Admittedly, outstanding issues
289: remain regarding the observed timing offsets or asymmetric profiles (see
290: Fig.~5). Neither can we rule out the possibility that real flares from
291: Mrk 501 might sometimes include a very soft second delayed component,
292: which increases the ``ELECTRON'' parameter through extra counts absorbed
293: in the propane layer, although the scenario seems a bit contrived.
294:
295: In our search, we discovered yet another sub-hour X-ray flare. It occurred
296: during an observation (obsid: 30249-01-01-02) when there was hardly any
297: enhancement in the soft electron fluxes. Figure~8 shows the flare, along
298: with the ``ELECTRON2'' parameter for the same time period. The flare is
299: quite weak (compared with the other two), reaching a peak amplitude of
300: about 30\% of the steady-state flux. It lasted
301: for about 800 s and shows sub-structures on even shorter timescales. The
302: profile shown might represent the composite of two flares or a sudden
303: decrease and recovery of X-ray fluxes in the middle of a flare. Remarkably,
304: the rising or decaying time associated with the sub-structures is as short
305: as about 20--30 seconds! The flare occurred roughly one satellite orbit
306: after the reported one, which might be a bit worrisome. However, it would
307: appear about 50\arcdeg\ in longitude further west (as shown in Fig.~7).
308: Moreover, Mrk 501 was monitored very intensely during this time period,
309: with data taken in 18 nearly consecutive satellite orbits. Most of these
310: observations did indeed span periods when the satellite was in the general
311: area of the ``Northeast Pacific Region'', but no X-ray/electron flare
312: (other than the ones already mentioned) was detected, which rules out
313: any systematic effects associated with the region. Finally, if the flare
314: shown in Fig. 8 were to be false, it would
315: have to be caused by a surge in particle-induced events that were not
316: vetoed {\em and} not manifested in the increase of the ``ELECTRON'' factor;
317: we are not aware of any examples of such background flaring activities.
318: Therefore, we believe that we have detected a sub-hour flare from Mrk 501.
319: The rare detection can be attributed, to a large extent, to the much
320: intensified coverage of the source.
321:
322: \subsection{Spectral Modeling}
323:
324: We carried out detailed spectral modeling to investigate, in a more
325: quantitative manner, spectral variability of Mrk 501. Here, we eliminated
326: all observations with effective exposure times less than 200 s, because
327: of the lack of statistics. In subsequent analyses, we fixed the hydrogen
328: column density at $1.8\times 10^{20}\mbox{ }cm^{-2}$ (Dickey \& Lockman
329: 1990). We experimented with the following empirical models: power law,
330: power law with an exponential rollover, and broken power law, to fit the
331: observed X-ray spectra. We found that when the source was relatively weak
332: (the 2-20 keV flux less than
333: $2\times 10^{-10}\mbox{ }ergs\mbox{ }cm^{-2}\mbox{ }s^{-1}$), the power-law
334: model fits the X-ray spectra well. However, the simple model fails
335: statistically (in terms of $\chi^2$ values) at higher fluxes. The data
336: taken when the source was brighter (in 1997 and parts of 1998) clearly
337: favor the broken power-law model. Formally, the resultant fits are
338: satisfactory (with the reduced $\chi^2$ values all around unity).
339:
340: Mrk 501 was most active in 1997, with the 2--20 keV flux varying roughly
341: between $3$--$9.5 \times 10^{-10}\mbox{ }ergs\mbox{ }cm^{-2}\mbox{ }s^{-1}$.
342: The best-fit broken power-law parameters are summarized in Figure 9. The
343: first photon index shows an initial decreasing (or spectral hardening) trend
344: and then levels off toward high fluxes; so does the second photon index,
345: although the trend is less apparent. In general, the SED of Mrk 501 is
346: quite flat in the X-ray band during this time period. Spectral variability
347: seems to be dominated by changes in the break energy (roughly between 5 and
348: 16 keV). In the subsequent years (1998--2000), the source became much weaker
349: in X-rays, with the flux ranging between
350: $0.4$--$4 \times 10^{-10}\mbox{ }ergs\mbox{ }cm^{-2}\mbox{ }s^{-1}$. The
351: broken power-law model is still required for some of the 1998 observations,
352: as already mentioned, and the results are also shown in Fig.~9 for
353: comparison. The photon indices show large variations. In some cases, the
354: X-ray spectrum is significantly steeper than that observed in 1997. On the
355: other hand, the values of the break energy seem comparable to those from
356: the 1997 observations.
357:
358: At lower fluxes
359: ($< 2\times 10^{-10}\mbox{ }ergs\mbox{ }cm^{-2}\mbox{ }s^{-1}$), the data
360: can no longer effectively constrain the broken power-law model, mainly
361: due to the lack of statistics (and also to the fact that the photon indices
362: above and below the break do not differ significantly; see Fig.~9).
363: Figure 10 shows the results from fits with a simple power law. The trend
364: of spectral hardening toward high fluxes is more apparent in this case.
365: The photon index goes roughly from 2.6 down to 1.8, clearly indicating a
366: shift in the synchrotron SED peak toward higher energies at higher fluxes.
367: Also apparent from the figure is spectral hysteresis associated with the
368: 1998 observations. As shown in Fig.~1, these observations sampled two
369: distinct periods: the very end of the decaying phase of the 1997 giant
370: flare and the rising phase of a subsequent major flare (of much lower
371: magnitude). To examine spectral hysteresis in more detail, we marked
372: the data points with different symbols for the two periods in Figures 9
373: and 10. For the most part, the points seem to bifurcate, with those for
374: each period clustered around one of the branches. However, the phenomenon
375: is clearly more complex --- spectral hysteresis also seems to be present
376: within each period, perhaps associated with smaller flares that are not
377: resolved.
378:
379: \section{Discussion}
380:
381: Like Mrk 421, Mrk 501 also produces X-ray flares on a wide range of
382: timescales. Unlike in Mrk 421, however, the flares in Mrk 501 do not seem
383: to occur nearly as frequently, which appears to be the case on all
384: timescales. This would be consistent with the scale invariance of X-ray
385: flaring phenomenon, if it is common among TeV blazars. Fig.~11 illustrates
386: the flaring hierarchy observed in Mrk 501.
387: In other words, there might not be any fundamental differences among long
388: flares, intermediate flares, or rapid flares, because they could all be
389: caused by the same physical process operating on all physical scales.
390: Though not fully understood, the flares in blazars are often thought to
391: be related to internal shocks in the jet (e.g., Rees 1978; Spada et al.
392: 2001), or to major ejection events of new components of relativistic
393: plasma into the jet (e.g., B\"ottcher et al. 1997; Mastichiadis \& Kirk
394: 1997). In these scenarios, the flares of different durations might simply
395: be the observational manifestation of a hierarchy of inhomogeneities in
396: the jet, which are energized to produce flares by the shocks. Alternatively,
397: the flares might be associated with magnetic reconnection events in a
398: magnetically dominated jet (Lyutikov 2003), perhaps similar to solar
399: flares in this regard. The flaring hierarchy might be related to
400: reconnection and subsequent avalanche processes. We should note that the
401: proton-synchrotron model actually requires a magnetically dominated jet
402: to account for the observed TeV emission from blazars (Aharonian 2000;
403: M\"{u}cke et al. 2003).
404:
405: The X-ray spectrum of Mrk 501 evolved significantly during the 1997 and
406: 1998 major flares. It shows a general trend of hardening toward the
407: peak of a flare, which is consistent with results from previous
408: works on X-ray bright blazars (Giommi et al. 1990), perhaps implying a
409: hardening in the spectral distribution of the emitting electrons during
410: the major flares. Moreover, there are indications of hysteresis associated
411: with the spectral evolution of the source during major flares in 1997
412: and 1998. Though common among blazars, spectral hysteresis is still not
413: fully understood. Kirk \& Mastichiadis
414: (1999) showed, using an internal shock model, that complex hysteresis
415: behavior could arise from the interplay of three characteristic timescales
416: associated with synchrotron cooling ($\tau_{syn}$), particle acceleration
417: ($\tau_{acc}$), and intrinsic variability ($\tau_{var}$), respectively.
418: The model can qualitatively describe the observed spectral hysteresis
419: associated with rapid flares, but would have difficulty in explaining the
420: phenomenon associated with long-duration flares in Mrk 501, because
421: $\tau_{var}$ almost certainly dominate over $\tau_{syn}$ and $\tau_{acc}$
422: in the latter cases.
423:
424: We critically examined the reported sub-hour X-ray flare from Mrk 501
425: (Catanese \& Sambruna 2000), in light of intense background flaring
426: activities during the observation. We found that it was likely associated
427: with flaring activities in the flux of soft electrons entering the
428: detectors, as opposed to those of Mrk 501. A similar rapid X-ray flare
429: was detected in this work, which was also accompanied with a strong
430: electron flare. On the other hand, we also found a rapid X-ray flare that
431: occurred during a time that seems to be free of any background
432: flaring activities. The flare lasted only for about 800 s; it might in
433: fact be composed of two flares on even shorter timescales. These
434: variability timescales pose severe
435: constraints on the physical properties of the flaring region. First of
436: all, the (jet-frame) size of the region is constrained by
437: $l \lesssim c t_{flare} \delta/(1+z) = 2.4\times 10^{14} \delta_1\mbox{ }cm$,
438: where $t_{flare}$ is the duration of the flare ($=800$ s), $\delta$ is the
439: Doppler factor of the jet ($\delta = 10\delta_1$), and $z$ is the redshift
440: of Mrk 501 ($= 0.034$). The upper limit is already comparable to the
441: gravitational radius ($r_g \equiv G M/c^2$) for a $1.3\times 10^9 M_{\odot}$
442: black hole, which is believed to exist in Mrk 501 (Barth et al. 2003),
443: if the Doppler factor of the jet is not too much larger than 10 (which
444: is probably the case; see discussion below). Since the peak flux of the
445: flare is a significant fraction of the steady-state flux, the size of
446: the flaring region is probably comparable to the lateral extent of the jet.
447: If the jet originates from accretion flows, as is often thought to be the
448: case, the result would also represent an upper limit on inner boundary of
449: the flows (either in the form of a geometrically thin, optically thick
450: cold disk or of a geometrically thick, optically thin hot torus).
451:
452: Secondly, the observed decaying time of the flare
453: sets a firm upper limit on the synchrotron cooling time of the emitting
454: electrons. The cooling time is given by Rybicki \& Lightman (1979),
455: $\tau_{syn} \approx 6 \pi m_e c/\sigma_T \gamma_p B^2$, where
456: $m_e$ is the electron rest mass, $\sigma_T$ is the Thomson cross
457: section, $B$ is the strength of the magnetic field in the region, and
458: $\gamma_p$ is the characteristic Lorentz factor of those
459: electrons that contribute to the bulk of the observed X-ray emission
460: (at $E_{p}\sim 10$ keV, where the synchrotron peak of the SED lies).
461: From $\tau_{syn}/\delta < t_d$, where $t_d$ is the decaying time of the
462: flare ($\approx 400 s$), we can derive a lower limit on the magnetic field
463: strength, $B > 1.4 \delta_1^{-1/2} \gamma_{p,5}^{-1/2} \mbox{ }G$, where
464: $\gamma_p = 10^5\gamma_{p,5}$. This is already larger than typical values
465: of $B$ inferred from fitting the SED with the SSC models (e.g., Kataoka
466: et al. 1999; Sambruna et al. 2000; Petry et al. 2000; Krawczynski et al.
467: 2002; Konopelko et al. 2003). Finally, for synchrotron photons to reach
468: X-ray energies (at $\sim E_p$), the Lorentz factor of the emitting
469: electrons must be sufficiently high,
470: $E_{p}=\delta h\nu_c \equiv (3 e h/4 \pi m_e c) \delta \gamma_{p}^2 B$
471: (Rybicki \& Lightman 1979).
472: Using the derived lower limit on $B$, we can derive an {\em upper} limit
473: on the Doppler factor, $\delta < 6.4 E_{p,1}^2 (\gamma_{p,5}/3)^{-3}$,
474: which seems quite low compared to typical values inferred from the SSC
475: models.
476:
477: \acknowledgments
478:
479: We thank Keith Jahoda and Craig Markwardt for insightful discussions on
480: the issue of electron contamination and the referee for his/her very
481: detailed and constructive comments. This work was supported in part by
482: grants from the US Department of Energy and National Aeronautics and
483: Space Administration. We made use of data obtained through the High
484: Energy Astrophysics Science Archive Research Center Online Service,
485: provided by the NASA/Goddard Space Flight Center.
486:
487: \begin{references}
488: \reference{}Aharonian, F. 2000, New Astronomy, 5, 377
489: \reference{}Barth,~A.~J., Ho,~L.~C., \& Sargent,~W.~L.~W. 2003, ApJ, 583, 134
490: \reference{}B\"ottcher, M., et al. 1997, A\&A, 324, 395
491: \reference{}Bradbury, S., et al. 1997, A\&A, 320, L5
492: \reference{}Catanese,~M., et al. 1997, ApJ, 487, L143
493: \reference{}Catanese,~M., \& Sambruna,~R.~M. 2000, ApJ, 534, L39
494: \reference{}Cui,~W. 2004, ApJ, 605, 662
495: \reference{}Cui,~W., et al. 2004, Proc. "International Symposium on High
496: Energy Gamma-Ray Astronomy" (Gamma-2004, Heidelberg, Germany),
497: eds. F.~A.~Aharonian and H.~Voelk, AIP Proceedings Series (astro-ph/0410160)
498: \reference{}Dickey \& Lockman, 1990, ARAA, 28, 215
499: \reference{}Falcone,~A.~D., Cui,~W., \& Finley,~J.~P. 2004, ApJ, 601, 165
500: \reference{}Fossati, G., Maraschi, L., Celotti, A., Comastri, A.
501: \& Ghisellini, G. 1998, MNRAS, 299, 433
502: \reference{}Giebels,~B., et al. 2002, ApJ, 571, 763
503: \reference{}Giommi, P., Barr, P., Pollock, A. M. T., Garilli, B.,
504: \& Maccagni, D. 1990, ApJ, 356, 432
505: \reference{}Hayashida,~N., et al. 1998, ApJ, 504, L71
506: \reference{}Kataoka,~J., et al. 1999, ApJ, 514, 138
507: \reference{}Kataoka,~J., et al. 2000, ApJ, 528, 243
508: \reference{}Kirk,~J.~G, \& Mastichiadis,~A. 1999, Astropart. Phys., 11, 45
509: \reference{}Konopelko,~A., Mastichiadis,~A., Kirk,~J., de Jager,~O.~C., \&
510: Stecker,~F.~W. 2003, ApJ, 597, 851
511: \reference{}Krawczynski,~H., Coppi,~P.~S., \& Aharonian,~F. 2002, MNRAS, 336,
512: 721
513: \reference{}Krawczynski,~H., et al. 2004, ApJ, 601, 151
514: \reference{}Lyutikov,~M. 2003, New Astr. Rev. 47, 513
515: \reference{}Maraschi, L., Ghisellini, G., \& Celotti, A., 1992, ApJ, 397, L5
516: \reference{}Marscher, A. P., \& Gear, W. K., 1985, ApJ, 298, 114
517: \reference{}Mastichiadis, A., \& Kirk, J. G. 1997, A\&A, 320, 19
518: \reference{}M\"ucke, A., Protheroe,~R.J., Engel,~R., Rachen,~J.~P.,
519: \& Stanev,~T. 2003, APh, 18, 593
520: \reference{}Petry, D., et al. 2000, ApJ, 536, 742
521: \reference{}Pian,~E., et al. 1998, ApJ, 492, L17
522: \reference{}Quinn, J., et al. 1996, ApJ, 456, L83
523: \reference{}Rees, M. J. 1978, MNRAS, 184, P61
524: \reference{}Rybicki,~G.~B., \& Lightman,~A.~P., Radiative Processes in
525: Astrophysics (New York: John Wiley \& Sons)
526: \reference{}Sambruna,~R.~M., et al. 2000, ApJ, 538, 127
527: \reference{}Spada, M., et al. 2001, MNRAS, 325, 1559
528: \reference{}Stickel,~M., Fried,~J.~W., \& Kuehr,~H. 1993, A\&AS, 98, 393
529: \reference{}Takahashi,~T., et al. 1996, ApJ, 470, L89
530: \reference{}Urry,~C.~M., \& Padovani,~P. 1995, PASP, 107, 803
531: \reference{}Zhang,~Y,~H. 2002, MNRAS, 337, 609
532:
533:
534: \end{references}
535:
536: \clearpage
537:
538: %Fig.1
539: \begin{figure}
540: \psfig{figure=f1.eps,width=5in} \caption{ASM light curve of
541: Mrk~501. Each data point was derived from a weighted average of
542: raw count rates over 21 consecutive days (after filtering out the
543: ``bad data''; see text). The error bars shown represent the
544: standard deviations. The dashed-line roughly shows the average
545: count rate of the source in the ``low state'' (which is an average
546: of the rates between MJD 51400--53150). The periods of monitoring
547: campaigns (with the pointing instruments) are indicated.}
548: \end{figure}
549:
550: %Fig.2
551: \begin{figure}
552: \psfig{figure=f2.eps,width=5in} \caption{ASM hardness-intensity
553: diagram for the 1997 giant flare (MJD 50442--50875). Note a
554: general hardening trend of the X-ray spectrum as the flux
555: increases. }
556: \end{figure}
557:
558: %Fig.3
559: \begin{figure}
560: \psfig{figure=f3.eps,width=5in} \caption{(top) Expanded ASM view
561: of the 1997 giant flare. The count rates are weekly averaged in
562: this case. The horizontal lines at the top indicate periods of the
563: PCA observations. (bottom) PCA hardness-intensity diagram. The
564: count rates were computed for the 2--60 keV band. Data taken
565: during the rising portion of the flare are shown in open diamonds
566: (except for the very first data point, which is indicated by a
567: filled circle), while data during the decaying phase in open
568: circles. The direction of the arrows shows the time progression of
569: the flare. Note the general trend of spectral hardening toward the
570: peak of the flare. }
571: \end{figure}
572:
573: %Fig.4
574: \begin{figure}
575: \psfig{figure=f4.eps,width=5in,angle=90} \caption{PCA light curves
576: of Mrk~501. Unlike in Fig. 3, the results here were derived from
577: the PCU 2 data alone. For clarify, they are shown for each year
578: separately. }
579: \end{figure}
580:
581: %Fig.5
582: \begin{figure}
583: \psfig{figure=f5.eps,width=6in} \caption{(left) Reported sub-hour
584: X-ray flare (Catanese \& Sambruna 2000). It is seen only after the
585: data filtering criterion on the ``ELECTRON2'' parameter is
586: relaxed. Note the 200-second lead of the X-ray flare with respect
587: to the electron flare. (right) A similar X-ray flare, with almost
588: identical 200-second difference between the starting time of the
589: X-ray and electron flares. For reference, the RXTE Mission Elapsed
590: Time (MET) is used, which is defined as the number of seconds
591: since 1994 January 1 00:00:00 (UT). }
592: \end{figure}
593:
594: %Fig.6
595: \begin{figure}
596: \psfig{figure=f6.eps,width=6in} \caption{Sample X-ray flares with
597: associated electron flares. Note that not all electron flares
598: induce X-ray flares and that an electron flare may have a
599: complicated multi-peaked profile. }
600: \end{figure}
601:
602: %Fig.7
603: \begin{figure}
604: \psfig{figure=f7.eps,width=5in} \caption{Satellite positions at
605: the times of the peaks of electron flares. The two flares shown in
606: Fig.~5 are indicated by different symbols, with the one associated
607: with the reported X-ray flare in open diamond and the other in
608: open triangle. Purely for the purpose of comparison, the X-ray
609: flare shown in Fig.~8 is also indicated here (in open square). It
610: should be stressed, however, that it is {\em not} accompanied by
611: any electron flare; in this case, the satellite position was
612: derived from the starting time of the X-ray flare. }
613: \end{figure}
614:
615: %Fig.8
616: \begin{figure}
617: \psfig{figure=f8.eps,width=5in} \caption{Same as Fig.~5 but for a
618: real sub-hour X-ray flare from Mrk 501. Note sub-structures in the
619: profile of the flare. }
620: \end{figure}
621:
622: %Fig.9
623: \begin{figure}
624: \psfig{figure=f9.eps,width=5in} \caption{Flux dependence of the
625: model parameters. The results were obtained with a broken power
626: law and are shown in diamonds for observations taken in 1997, in
627: filled circles for those (taken in 1998) near the end of the
628: decaying phase of the 1997 giant flare, and in crosses for those
629: during the rising phase of the 1998 flare (with the dividing line
630: between the two 1998 period drawn somewhat arbitrarily at MJD
631: 50960.6; see Fig.~1). To guide the eye, the dashed lines are drawn
632: to show best-fits to the data with a model consisting of a
633: constant plus an exponential function. }
634: \end{figure}
635:
636: %Fig.10
637: \begin{figure}
638: \psfig{figure=f10.eps,width=5in} \caption{Same as Fig.~9, but a
639: simple power law was used to fit the data. The results are shown
640: in diamonds for observations taken in 1999 and 2000, in crosses
641: for those during the rising phase of the 1998 flare, and in filled
642: circles for those near the end of the decaying phase of the 1997
643: flare. }
644: \end{figure}
645:
646: %Fig.11
647: \begin{figure}
648: \psfig{figure=f11.eps,width=5in} \caption{X-ray flares from Mrk
649: 501 on a wide range of timescales. Apparent irregularity in the
650: width of histogram bins is a manifestation of irregular data
651: sampling. }
652: \end{figure}
653:
654: \end{document}
655: