1: %\documentclass[manuscript]{aastex}
2: %\documentclass{emulateapj}
3: \documentclass[12pt,preprint]{aastex}
4: \usepackage{apjfonts}
5: \newcommand{\vdag}{(v)^\dagger}
6: \newcommand{\myemail}{stella@mporzio.astro.it}
7: %
8: \def\ltsima{$\; \buildrel < \over \sim \;$}
9: \def\simlt{\lower.5ex\hbox{\ltsima}}
10: \def\gtsima{$\; \buildrel > \over \sim \;$}
11: \def\simgt{\lower.5ex\hbox{\gtsima}}
12: %
13:
14: \slugcomment{}
15:
16: \shorttitle{SGR 1806-20 and Variability Limit} \shortauthors{Vietri et al.}
17:
18: \begin{document}
19:
20: \title{SGR1806-20: evidence for a superstrong Magnetic Field from Quasi Periodic Oscillations}
21:
22: \author{Mario Vietri}
23: \affil{Scuola Normale Superiore, 56100 Pisa, Italy }
24: \email{m.vietri@sns.it}
25:
26: \and
27:
28: \author{Luigi Stella and Gian Luca Israel}
29: \affil{INAF - Osservatorio Astronomico di Roma, \\ Via Frascati 33,
30: 00040 Monteporzio Catone, Italy.}
31: \email{stella@mporzio.astro.it} \email{gianluca@mporzio.astro.it}
32:
33:
34: \begin{abstract}
35:
36: Fast Quasi-Periodic Oscillations (QPOs, frequencies of $\sim 20 -
37: 1840$~Hz) have been recently discovered in the ringing tail of giant
38: flares from Soft Gamma Repeaters (SGRs), when the luminosity was of
39: order $10^{41}-10^{41.5}$~erg/s. These oscillations persisted for
40: many tens of seconds, remained coherent for up to hundreds of cycles
41: and were observed over a wide range of rotational phases of the
42: neutron stars believed to host SGRs. Therefore these QPOs must have
43: originated from a compact, virtually non-expanding region inside the
44: star's magnetosphere, emitting with a very moderate degree of
45: beaming (if at all). The fastest QPOs imply a luminosity variation
46: of $\Delta L/\Delta t \simeq 6 \times 10^{43}$~erg~s$^{-2}$, the
47: largest luminosity variation ever observed from a compact source. It
48: exceeds by over an order of magnitude the usual Cavallo-Fabian-Rees
49: (CFR) luminosity variability limit for a matter-to-radiation
50: conversion efficiency of 100\%. We show that such an extreme
51: variability can be reconciled with the CFR limit if the emitting
52: region is immersed in a magnetic field $\gtrsim 10^{15}$~G at the
53: star surface, providing independent evidence for the superstrong
54: magnetic fields of magnetars.
55:
56: \end{abstract}
57:
58:
59: \keywords{stars: magnetic fields --- stars: neutron
60: --- stars: individual(\objectname{SGR 1806-20})}
61:
62:
63: \section{Introduction}
64:
65: Soft Gamma Repeaters are a small class of galactic sources of X and
66: soft gamma radiation. They have spin periods of $\sim 5\div10$~s,
67: display a secular spin-down with timescales of $\sim
68: 10^4\div10^5$~yr and do not possess a companion.
69: Unlike radio pulsars, the rotational energy
70: loss of SGRs is a factor of $10\div100$ too small to explain
71: their persistent emission, typically $\sim 10^{33}\div10^{34}$ erg/s
72: (see e.g. \citet{WooTho06}). Like Anomalous X-ray Pulsar (AXPs,
73: \citet{MeSte95}), with whom they share a number of properties, SGRs
74: are believed to host magnetars, neutron stars the emission of which
75: is powered by the decay of their superstrong (internal)
76: magnetic field ($B> 10^{15}$~G, \citet{DT92,TD93}).
77:
78: The name-defining characteristic of SGRs is that they show periods
79: of activity in which recurrent short bursts are emitted, with peak
80: luminosities of $\sim 10^{38}\div10^{41}$~ergs~s$^{-1}$ and
81: sub-second durations. The characteristics of the three giant flares
82: observed so far in about 30~yr of monitoring are much more extreme.
83: Their initial, tenths-of-seconds-long spike releases enormous
84: amounts of energy, $\sim 10^{44}$~ergs in the 1979 March 5 event
85: from SGR 0526-66 \citep{Maz79} and the 1998 August 27 event from SGR
86: 1900+14 \citep{Hur99,Fer99} and as much as $ 5 \times 10^{46}$~erg
87: in the 2004 December 27 event from SGR1806-20
88: \citep{Hur05,Mere05,Palm05,Teras05}. After the initial spike, giant
89: flares display a very bright ringing tail lasting hundreds of
90: seconds and releasing a total energy of about $\sim 10^{44}$~ergs.
91: The emitted spectrum is roughly thermal, with a blackbody equivalent
92: temperature of $\sim 5$~keV in the case of SGR1806-20 \citet{Hur05}.
93:
94: The highly super-Eddington luminosities of the recurrent bursts of
95: SGRs and especially of their giant flares make models involving
96: accretion energy not viable.
97: %HURLEY KTBB = 5 KEV ,BU CLEAR EXCESS ABOVE 40 KEV, OT T= 22 KEV (270-400S INTV)
98: According to the magnetar model, the emission of SGRs (and AXPs)
99: draws from their extremely high magnetic fields
100: \citep{TD95,TD96,TD01}. Within this model the neutron star interior
101: is characterized by a wound-up, mainly toroidal magnetic field
102: configuration with $B_t>10^{15}$~G. A less intense (mainly poloidal)
103: field emerges out of the star magnetosphere, causing spin-down via
104: rotating dipole losses at the observed rate \citep{TD93,TD01}
105: (dipole B-field strength of $B_d \sim 7.8\times 10^{14}$~G). Energy
106: propagates to the neutron star magnetosphere through Alfv\'{e}n
107: waves driven by local ``crust-quakes" and giving rise to recurrent
108: bursts with a large range of amplitudes. Large-scale rearrangements
109: of the internal magnetic field or catastrophic instabilities in the
110: magnetosphere are invoked to explain the sudden release of very
111: large amounts of energy that occurs in giant flares
112: \citep{TD01,Lyu03}. A fireball of plasma expanding at relativistic
113: speeds breaks out of the star's magnetosphere, causing the initial
114: sub-second spike of giant flares. The ringing tail that follows
115: results most likely from the part of the fireball that remains
116: trapped in the star's magnetosphere. The energy release in the
117: ringing tail yields a limit for the external field of magnetars
118: ($\simgt 10^{14}$~G) in agreement with the values inferred from
119: spin-down dipole losses \citep{TD95,TD01}, while an analysis of the
120: initial time scales gives evidence in favor of the crustal cracking
121: mechanism \citep{Sch05}.
122:
123: To confirm this model it is thus essential to measure the surface
124: magnetic field, which complements the measurement of the dipole
125: component (\citet{Woo02}), to include higher order multipoles. In
126: this paper we present an interpretation of the QPOs observed in SGR
127: 1806-20 which provides a lower limit ($B \gg B_q \approx 4.4\times
128: 10^{13}\; G$) to the {\it surface} field. The argument is based upon
129: a seldom-used constraint (\citep{CaRe78,FaRe79,Fab79}) that puts a
130: very strong upper limit on the time-scale on which significant
131: luminosity variations can take place. In the next section we discuss
132: the relevant observations; in Section 3 we will re-derive the limit
133: and show that it is largely violated by QPOs in SGR 1806-20; after
134: vainly trying to circumvent it, we will show that it can be
135: reconciled with observations only if $B\gg B_q$.
136:
137:
138: \section{Quasi Periodic Oscillations in Giant Flares of SGRs}
139:
140: Recent studies led to the discovery that the X-ray flux of the
141: ringing tail of SGRs' giant flares is characterized by fast Quasi
142: Periodic Oscillations, QPOs \citep{Isr05}. Different QPO modes were
143: detected, some of which were excited simultaneously. The ringing
144: tail of the December 2004 event from SGR~1806-20 displayed clear QPO
145: signals at about 18, 30, 93, 150, 625 and 1840~Hz (\citet{WaSt06}).
146: %Evidence for QPOs at about 720 and 2380 Hz was also found.
147: Similarly, QPOs around frequencies of 28, 54, 84 and 155 Hz were
148: detected during the ringing tail of the 1998 giant flare of
149: SGR~1900+14 \citep{StWa05}, while hints for a signal at $\sim$43 Hz
150: were found in the March 1979 event from SGR 0526-66 \citep{Bar83}.
151: These QPOs show large variations of the amplitude with time and,
152: especially, of the phase of the spin modulation in the giant flare's
153: tail.
154:
155: The similarity in some of the QPO modes and frequencies across
156: different SGRs suggests that the production mechanism is the same. A
157: likely interpretation involves the excitation of neutron star
158: oscillation modes, whose expected eigenfrequencies match some of the
159: observed QPOs peaks (\citep{Dun98,Isr05,Pir05}). Infact, if giant
160: flares result from large scale fracturing of the crust induced by
161: instabilities of the internal magnetic fields, then the excitation
162: of crustal and (possibly) global neutron star modes is to be
163: expected (\citep{Lev06}). Regardless of the exact mechanism driving
164: these oscillations we are concerned here with the extremely large
165: and fast luminosity variations of the QPOs.
166:
167: We concentrate on the signals with the largest luminosity time
168: derivative, i.e. the 625 and 1840~Hz QPOs from SGR 1806-20
169: (\citet{WaSt06}). The power spectrum peaks through which these QPOs
170: are revealed, are a few Hz wide, testifying that their signal
171: remained coherent for hundreds of cycles. The signal shape must be
172: close to sinusoidal, as evidenced by the absence of detectable
173: harmonic signals. The 625 and 1840~Hz QPOs were detected only in a
174: $\sim 50$~s long interval of the ringing tail, about 200~s after the
175: initial spike, and were especially prominent over a $\sim 140$~deg
176: interval in rotational phase. The QPO amplitude reached a maximum in
177: this phase interval over two consecutive rotation cycles: for both
178: signals the rms amplitude was $a_{rms}\sim 18$\%. Approximating the
179: QPOs with sinusoids, we estimate their highest luminosity derivative
180: as $\Delta L/\Delta t = 2^{3/2} \pi L a_{rms} \nu_{QPO}$, with
181: $\nu_{QPO}$ the QPO frequency. Here $L \sim 10^{41}$~ergs~s$^{-1}$
182: is the luminosity in the relevant section of the ringing tail (for
183: the likely source distance of 15 ~kpc). This gives $\Delta L/\Delta
184: t \sim 1\times 10^{44}$ and $ 3\times 10^{44}$~ergs~s$^{-2}$ for the
185: 625 and 1840~Hz QPOs, respectively. The effects of beaming might
186: decrease these values somewhat, but not by a very large factor. In
187: fact, these QPO signals were observed over a large interval of
188: rotational phases (about $\sim 140$~deg), translating into
189: approximately the same azimuthal range of emission angles. It is
190: natural to assume a comparably large angular spread in latitude
191: (unless the neutron star rotation axis is very close to our line of
192: sight, which is unlikely given the large amplitude of the rotation
193: modulation and the size of the emission region, see below). This
194: gives a solid angle of order $\sim \pi$~ster. Adopting this beaming
195: factor the luminosity derivatives above reduce to $\sim 2\times
196: 10^{43}$ and $ 6\times 10^{43}$~ergs~s$^{-2}$; this are the values
197: that we adopt in the following discussion.
198:
199: We stress here, because this is essential to our argument (to be
200: presented shortly), that, together with the QPOs, in the ringing
201: tail a strong modulation at the star's spin period is clearly
202: present, similar in relative amplitude and shape to the modulation
203: observed when the source is in its quiescent state: this (together
204: with the lack of significant amounts of beaming) indicates that the
205: emission in the ringing tail originates from a region that remains
206: stably anchored to the star's magnetosphere, and thus that
207: relativistic bulk motions are not present at this stage of the flare.
208:
209: The blackbody temperature and luminosity in the ringing tail
210: translates into a lower limit on the size of the emitting region of
211: about $\sim 30$~km, {\it i.e.} substantially larger than the neutron
212: star. On the other hand the black body-like spectral shape testifies
213: that the emitting region is optically thick (or at least effectively
214: thick), implying a scattering optical depth $\gg 1$. We remark that
215: the size estimate, $\approx 30\; km$, will play an important role in
216: the following.
217:
218: \section{The Cavallo-Fabian-Rees Variability Limit}
219:
220: There is a well--known limit on the rate of change of the luminosity
221: of any given source, which we briefly summarize here
222: (\citet{CaRe78,FaRe79,Fab79}, see also \citet{Law80,Hos89}). Suppose
223: a source undergoes a large luminosity variation on a time-scale
224: $\Delta t$, and there is a luminosity variation, over this
225: time-scale, $\Delta L$. The total energy released within $\Delta t$
226: is related to the total mass within the source dimension $R$ by
227: \begin{equation}\label{energycontent}
228: \Delta L \Delta t = \eta \frac{4\pi}{3}R^3 n m_p c^2\;,
229: \end{equation}
230: where $n$ is the average baryon density, and $\eta$ is the energy
231: extraction efficiency. The time-scale $\Delta t$ must
232: obviously exceed the time over which the photons manage to diffuse
233: out of the source:
234: \begin{equation}\label{timevar}
235: \Delta t > \frac{R}{c} (1+\tau_T)\;
236: \end{equation}
237: where $\tau_T = \sigma_T n R$ is the Thomson optical depth
238: and $\sigma_T$ the Thomson cross-section. Eliminating $R$
239: from the first two equations, we find
240: \begin{equation}
241: \Delta t > \frac{3}{8\pi}\frac{\sigma_T}{m_p c^4} \frac{\Delta
242: L}{\eta}\frac{(\tau_T+1)^2}{\tau_T}\;.
243: \end{equation}
244: Regarded as a function of $\tau_T$, the above has a minimum for
245: $\tau_T = 1$, which yields the limit
246: \begin{equation}\label{cavallo}
247: \Delta t > \frac{3}{2\pi}\frac{\sigma_T}{m_p c^4} \frac{\Delta
248: L}{\eta}\;.
249: %= 0.073 \;s \frac{1}{\eta}\frac{L}{3\times 10^{41}erg\; s^{-1}}\;.
250: \end{equation}{}
251: The above limit is remarkable in that it is independent of both $R$
252: and $n$, or any combination thereof: only the dependence on $\Delta
253: L$, a directly observable quantity, is left. The
254: Cavallo-Fabian-Rees, CFR, limit thus writes
255: \begin{equation}\label{CFR}
256: \Delta L /\Delta t < \eta\ 2\times 10^{42}erg\; s^{-2}\;.
257: \end{equation}{}
258:
259: The 625 and 1840~Hz QPO signals from SGR1806-20 exceed the CFR limit
260: by about an order of magnitude\footnote{We note that also the slower
261: $\sim 90-150~Hz$ QPOs signals in the giant flares' tail of
262: SGR1806-20 and SGR1900+14, exceed the CFR limit, though by a smaller
263: factor.}: the largest value found in the previous section is $\Delta
264: L/\Delta t = 6\times 10^{43}\; erg\; s^{-2}$, which is a whole
265: factor $30/\eta$ larger than the CFR's limit.
266:
267: In order to appreciate how hard it is to circumvent this limit,
268: notice the {\it in situ} re-acceleration of electrons does not help,
269: because it does not change the energies reached by protons; if
270: protons were to escape, leaving electrons behind to be
271: re-accelerated at will, Coulomb forces would quickly make the escape
272: of protons impossible. Nor will having relativistic protons help, as
273: one might think, since this would imply energies per electron to be
274: radiated $\approx \gamma m_p c^2\gg \eta m_p c^2$, because the
275: energy transfer from protons to electrons is too slow. To show this,
276: let us consider what happens when protons transfer promptly to the
277: electrons their energy gain: protons may then be relativistic, in
278: which case the maximum energy which can be extracted from each of
279: them is $\gamma m_p c^2$, rather than the more sedate $\eta m_p
280: c^2$. This however appears like an unlikely way out, because, even
281: admitting that the electrons radiate very promptly their internal
282: energy, the time-scale on which protons manage to transfer to the
283: electrons their internal energy is much longer than the above limit.
284: To see this, we idealize the situation as one where electrons are
285: cold ({\it i.e.} Newtonian) as a result of their short cooling
286: timescales, while the protons are still relativistic. The energy
287: transfer rate is given by the usual formula
288: \begin{equation}
289: -\frac{dE_p}{dx} = \frac{2\pi e^4 n_e}{m_e v^2}\left( \ln(\frac{2m_e
290: v^2 W}{\hbar^2\omega_p^2})+1-2\frac{v^2}{c^2} \right)\;,
291: \end{equation}
292: where $n_e$ is the electron number density, $\omega_p = \sqrt{4\pi
293: e^2 n_e /m_e}$ the plasma frequency, $v \approx c$ the proton speed,
294: and $W \approx 2E_p^2/(m_e c^2)$ is the maximum energy transfer in
295: the not too extreme limit, $\gamma_p \lesssim m_p/m_e$ (but
296: considering the opposite limit, $\gamma \gtrsim m_p/m_e$ would
297: change the argument very little since $W$ only appears as an
298: argument to a logarithm).
299:
300: The energy transfer timescale is of course $t_e \equiv E_p/(c\
301: dE_p/dx)$, which, for this whole idea to work, must be shorter than
302: $\Delta t$. The condition $t_E < \Delta t$ can also be rewritten as,
303: in the limit $v^2 \approx c^2$:
304: \begin{equation}
305: \frac{\gamma}{\ln(\frac{2m_e v^2 W}{\hbar^2\omega_p^2})-1} <
306: \frac{3\sigma_T}{m_p c^4}\frac{L}{\eta} \frac{e^4 n_e}{m_p m_e
307: c^3}\;.
308: \end{equation}
309: Inserting the numerical values for the luminosity of SGR 1806-20
310: during the ringing tail and for the typical magnetospheric density
311: for a pulsar with a {\it normal} magnetic field we find
312: \begin{equation}
313: \gamma \lesssim 10^{-8} \frac{1}{\eta} \frac{L}{3\times 10^{41}\;
314: erg s^{-1}} \frac{n_e}{10^{10}\; cm^{-3}}\;.
315: \end{equation}
316: In order to reconcile the CFR limit with observations, we need
317: $\gamma \approx 30$, corresponding to a magnetospheric density
318: $n\approx 3\times 10^{19}\; cm^{-3}$. While it is certainly true
319: that plasma outside pulsars need not be charge--separated, still
320: this density would exceed the minimum ({\it i.e.}, charge separated)
321: value by more than $9$ orders of magnitude, making it unlikely that
322: such a plasma may exist.
323:
324:
325: This argument must be modified in the presence of a pair plasma,
326: where pair creation processes can easily lead to a strong increase
327: in $n_e \propto T^3$. Still, we know that astrophysical pair plasmas
328: are thermally regulated \citep{PiKr95}, so that their temperatures
329: always lie around $T \lesssim 1\; MeV$. At these temperatures, $n_e
330: \approx 10^{31}\; cm^{-3}$, but we also know that $E_p \ll m_p c^2$,
331: so that the limit of eq. \ref{cavallo} still holds, provided the
332: total energy density is still dominated by the rest mass of the
333: baryons. When instead the total energy density is dominated by the
334: pairs (in other words, when little or no baryons
335: are admixed), eq. \ref{timevar} % and \ref{optdepth}
336: obviously still holds (with $\tau_T$ now indicating the total
337: optical depth due to the pairs and possibly to photons as well),
338: while eq. \ref{energycontent} must be rewritten as
339: \begin{equation}
340: \Delta L\Delta t = \eta \frac{4\pi}{3} R^3 n m_e c^2
341: \end{equation}
342: where we used the fact that $\gamma\approx 2$ for the electrons. The
343: new limit for the variability is
344: \begin{equation}\label{cavallo2}{}
345: \Delta t > \frac{3}{2\pi}\frac{\sigma_T}{m_e c^4} \frac{\Delta
346: L}{\eta} = 134 \;s \frac{1}{\eta}\frac{\Delta L}{3\times 10^{41}
347: erg\; s^{-1}}\;,
348: \end{equation}{}
349: which is even more stringent than eq. \ref{cavallo}.
350:
351: The CFR limit might fail if coherent phenomena are involved, but we
352: are unaware of any coherent mechanism working in the X--ray region
353: of the spectrum where QPOs are observed, rather than the usual radio
354: band.
355:
356: The classic way to circumvent the CFR limit is by means of
357: relativistic effects. After all, GRBs' light-curves often display
358: millisecond variability, and even when they don't they turn on on
359: timescales of a second or less, reaching $L \approx 10^{50}\ erg \;
360: s^{-1}$: this clearly violates eq. \ref{cavallo} by eight to ten
361: orders of magnitude (for $\eta =1$). This is due to a combination of
362: relativistic aberration, blueshift and time contraction, which can
363: drastically increase the luminosity and decrease the variability
364: timescale in the observer's frame. Also BL Lac objects and Quasars
365: have luminosity derivatives that exceed by orders of magnitude the
366: CFR limit (see e.g. \citep{Bas83}), even after their luminosity is
367: corrected for beaming. This explanation is probably correct also
368: for the giant flare of SGR~1806-20, where values as high as $\Delta
369: L/\Delta t \sim 10^{47}\; ergs\;s^{-2}$, for isotropic
370: ejection\footnote{Beaming in the ejection, which is presently
371: unknown, could decrease this value substantially.} are observed: the
372: observation of a radio halo expanding at relativistic speed
373: testifies that the observed variability in the initial spike of the
374: giant flare is most likely magnified by relativistic effects as
375: well.
376:
377: However, this explanation is not suitable for the QPOs in the
378: ringing tail of the same flare: in fact, the presence of modulations
379: at the spin period, with similar pulse shape and amplitude to those
380: during quiescent periods, makes the existence of relativistic bulk
381: motions very unlikely during this phase. Also, one should remember
382: that the photosphere size is estimated to be $\approx 30\; km$ at
383: all times during the ringing tail, which implies that emission
384: during this phase is due to material stably anchored to the pulsar.
385: Lastly, the observed temperatures ($T \approx 5\; keV$) testify to
386: the gas having thermal speeds much below the escape velocity at that
387: radius. For these three reasons, we deem relativistic effects an
388: unlikely explanation for the violation of eq. \ref{cavallo}.
389:
390: The last way in which the CFR limit can fail is the one we propose:
391: the scattering cross-section may differ from Thomson's because of
392: the presence of a strong magnetic field $B$, exceeding the quantum
393: value $B_q = m^2 c^3/(e \hbar) = 4.4\times 10^{13} G$. In this case,
394: the scattering cross section for the ordinary (O) and extraordinary
395: (E) modes, and for the conversion of photons into the other state,
396: are given by \citet{Mes92}, when the dielectric tensor is dominated
397: by vacuum polarization effects, as:
398: \begin{eqnarray}\label{meszaros}
399: d\sigma_{O\rightarrow O} = \frac{3}{4}\sigma_T \sin^2\theta
400: \sin^2\theta' d\cos\theta' \nonumber\\
401: d\sigma_{O\rightarrow E} = \frac{3}{8}\sigma_T\left(\frac{\epsilon
402: B_q}{m_ec^2 B}\right)^2\cos^2\theta d\cos\theta' \nonumber\\
403: d\sigma_{E\rightarrow O} = \frac{3}{8}\sigma_T\left(\frac{\epsilon
404: B_q}{m_ec^2 B}\right)^2\cos^2\theta' d\cos\theta' \nonumber\\
405: d\sigma_{E\rightarrow E} = \frac{3}{8}\sigma_T\left(\frac{\epsilon
406: B_q}{m_ec^2 B}\right)^2 d\cos\theta'\;,
407: \end{eqnarray}
408: where $\epsilon$ is the photon energy, and $\theta$ and $\theta'$
409: are the angles between the photon momentum before and after the
410: diffusion, respectively, with the direction of the magnetic field.
411:
412: This equation shows immediately that, for photons emitted in the
413: extraordinary mode, the cross section is reduced, with respect to
414: the Thomson value, by a factor $\approx (\epsilon B_q /(m_e c^2
415: B))^2$. Thus, to bring the observed value, $\bigtriangleup
416: L/\bigtriangleup t = 6\times 10^{43}\; erg\; s^{-1}$, in agreement
417: with eq. \ref{cavallo}, we just need to have $(\epsilon B_q/(m_e c^2
418: B))^2 \lesssim \eta/30$; here we take for $\epsilon$ the value
419: $\epsilon \approx 14\; keV$, which is the peak of the Planck
420: distribution for the observed temperature $T = 5\; keV$. So our
421: conclusion is that the QPOs' luminosity variation agrees with the
422: CFR limit provided
423: \begin{equation}{}
424: B \gtrsim 1.5 B_q\left(\frac{0.1}{\eta}\right)^{1/2} \approx
425: 6.6\times 10^{13}\; G\;.
426: \end{equation}
427: A technical comment is in order at this point: the description of
428: radiation transfer in terms of separate modal propagation is not
429: always adequate (\citet{Lai03}), because of mode collapse. However
430: this effect seems to be mostly relevant for even higher fields ($B
431: \gtrsim 7\times 10^{13} \; G$) than those derived here, which means
432: that our naive treatment is probably justified.
433:
434:
435: We now remark that this lower limit applies to the field close to,
436: but not {\it at} the surface of SGR 1806-20, because, as discussed
437: in the previous section, emission from the ringing tail is generated
438: within $30\; km$. At this distance from the star surface, the dipole
439: field inferred from pulsar spin-down ($B = 7.8\times 10^{14}\; G$,
440: \citet{Woo02}), is $2.9\times 10^{13}\; G (R_{ns}/10\; km)^3$, which
441: is smaller than the limit just derived, as expected if higher order
442: multipoles are relevant in the star vicinity.
443:
444:
445: Given the rapid decrease of the dipole field (and {\it a fortiori}
446: of the other multipoles) with distance from the star, the surface
447: magnetic field must certainly satisfy
448: \begin{equation}\label{finallimit}{}
449: B \gtrsim 1.8\times 10^{15}\; G \left(\frac{10\;
450: km}{R_{ns}}\right)^3 \left(\frac{0.1}{\eta}\right)^{1/2}\;,
451: \end{equation}
452: where a dipole--like radial dependence has been assumed between
453: $30\; km$ and the star surface, located at $R_{ns}$; we shall
454: refrain from making more elaborate hypothesis about the structure of
455: the magnetic field within $R = 30\; km$ (\citet{TD01}), because our
456: aim is simply to provide a {\it lower limit} to the surface field,
457: for which this minimum hypothesis (pure dipole) is fully adequate.
458:
459:
460:
461: \section{Discussion}
462:
463:
464: Our final result is the lower limit in eq. \ref{finallimit}. This
465: value is close to that of the dipole field, as inferred from pulsar
466: spin-down, $B = 7.8\times 10^{14}\; G$, but our estimate includes
467: higher order multipoles, at least at $30\; km$. It is thus
468: completely independent of the estimates from pulsar spin-down. We
469: note that the limit on $B$ derived by requiring that the magnetic
470: field close to the star can prevent the escape of the "trapped
471: fireball" includes also the contribution of higher multipole
472: components (\citet{TD01}), but gives a substantially lower magnetic
473: field ($B>10^{14}\; G$) than the limit derived above.
474:
475:
476: There is an easy way to test our interpretation of the failure of
477: the CFR limit. The O- and E-mode photospheres of the trapped
478: fireball are located at different heights in the star's
479: magnetosphere, as a result of the different electron scattering
480: cross sections, which therefore sample regions of different B-field
481: strengths and orientations. The polarization fraction and angle of
482: the emerging X-ray flux should thus be modulated with the phase of
483: the QPO signal as a result of the varying (relative) intensity of
484: the O- and E-mode photon component. This in principle gives a clear
485: test for the correctness of our interpretation, which might perhaps
486: become verifiable in the future.
487:
488: Photons in the O-mode suffer a strong Comptonization (\citet{TD01}).
489: If the atmosphere were due to pure scattering, the ensuing photon
490: distribution would differ from a blackbody at low photon frequency,
491: $E_\nu \propto \nu^3$ instead of $\propto \nu^2$, a result due to
492: photon number conservation. At first sight, one might think that
493: there are many processes which may lead to photon absorption and
494: emission in a strong magnetic field, like photon splitting or pair
495: creation via $\gamma + B$ (resulting then in pair annihilation, and
496: photon energy downgrading via Compton recoil), which make photon
497: number conservation unlikely. At the same time, there are important
498: radiation transfer effects taking place, which obviously tend to
499: favor flatter spectra at low energies; also, all rates for photon
500: emission and absorption depart from their values in the absence of
501: magnetic field. Detailed computations (\citep{Lyu02}) show that the
502: ensuing spectrum is flatter, not steeper, than a blackbody; and we
503: remark here that, despite many calibration uncertainties, fits to
504: the spectrum seem to favor spectra flatter than a blackbody, with a
505: preference for thermal bremsstrahlung of temperature $T = 30\; keV$.
506:
507: We have presented our argument by stressing its independence from
508: the estimates of the dipole field from pulsar spin-down, and from
509: Thompson and Duncan's (\citet{TD93, TD95, TD96}) model, but it
510: should be obvious that the limit in eq. \ref{finallimit} is
511: perfectly consistent with the previous measurements, and the model
512: itself.
513:
514:
515: \acknowledgments
516:
517: This work was partially supported through ASI and MIUR grants.
518:
519: %\clearpage
520:
521: \begin{thebibliography}{}
522: \bibitem[Barat et al.(1983)]{Bar83}Barat, C. et al. 1983, \aap, 126, 400
523: \bibitem[Bassani et al.(1983)]{Bas83} Bassani, L., Dean, A.J. \&
524: Sembay, S. 1983, \aap, 125, 52
525: \bibitem[Cavallo \& Rees(1978)]{CaRe78} Cavallo, G. \& Rees, M. 1978, \mnras,
526: 125, 52
527: \bibitem[Duncan(1998)]{Dun98}Duncan, R.C. 1998 \apjl, 498, L45
528: \bibitem[Duncan \& Thompson(1992)]{DT92}Duncan, R.,C., \& Thompson, C. 1992,
529: \apjl, 392, L9
530: \bibitem[Fabian(1979)]{Fab79}Fabian, A.C. 1979, Proc. R. Soc. London, Ser. A., 336, 449
531: \bibitem[Fabian \& Rees(1979)]{FaRe79}Fabian, A.C. \& Rees, M. 1979, Adv. Sp. Res., 3, 381
532: \bibitem[Feroci et al.(1999)]{Fer99}Feroci, M., et al. 1999, \apjl, 515, L9
533: %\bibitem[Ginzburg(1969)]{Gin69}Ginzburg, ? 1969 ???
534: \bibitem[Hurley et al.(1999)]{Hur99}Hurley, K., et al. 1979, \nat, 397, 41
535: \bibitem[Hurley et al.(2005)]{Hur05} Hurley, K., et al.
536: 2005, \nat, 434, 1098
537: \bibitem[Mazets et al.(1979)]{Maz79}Mazets, E., P.,
538: Golentskii, S., V., Ilinskii, V., N.,
539: Aptekar, R., L., Guryan \& Iu., A. 1979, \nat, 282, 587
540: \bibitem[Hoshi(1989)]{Hos89}Hoshi, R. 1989, \pasj, 41, 217
541: \bibitem[Israel et al.(2005)]{Isr05}Israel, G. L. et al. 2005, \apjl, 628, L53
542: \bibitem[Lai \& Ho (2003)]{Lai03}Lai, D \& Ho, W.C.G., 2003, \apj,
543: 588, 962.
544: \bibitem[Lawrence(1980)]{Law80}Lawrence, A. 1980, \mnras ,192, 83
545:
546: \bibitem[Levin(2006)]{Lev06}Levin, Y. 2006, \mnras, 368, L35
547: \bibitem[Lyubarsky(2002)]{Lyu02} Lyubarsky, Y.E., 2002, \mnras, 332, 199.
548: \bibitem[Lyutikov(2003)]{Lyu03} Lyutikov, M., 2003, \mnras, 346, 540
549: \bibitem[Mereghetti \& Stella(1995)]{MeSte95} Mereghetti, S., \& Stella, L.
550: 1995, \apjl, 442, L17
551: \bibitem[Mereghetti et al.(2005)]{Mere05}Mereghetti, S., Götz, D., von Kienlin,
552: A., Rau, A., Lichti, G., Weidenspointner, G. \& Jean, P. 2005, \apjl, 624, L105
553: \bibitem[Meszaros (1992)]{Mes92} Meszaros, P., 1992, {\it High energy radiation from
554: magnetized neutron stars}, Univ. Chicago Press, Chicago.
555: \bibitem[Palmer et al.(2005)]{Palm05} Palmer, D.M., et al. 2005, \nat, 434, 1107
556: \bibitem[Piran (2005)]{Piran05} Piran, T., 2005, {\it Rev. Mod.
557: Phys.}, 76, 1143
558: \bibitem[Piro(2005)]{Pir05}Piro, A.L. 2005 \apjl, 634, L153
559: \bibitem[Pietrini \& Krolik (1995)]{PiKr95}Pietrini, P. Krolik,
560: J.H., 2005, \apj, 447, 526.
561: \bibitem[Schwartz et al. (2005)]{Sch05}Schwartz, S.J., {\it et al.},
562: 2005, \apjl, 627, L129.
563: \bibitem[Strohmayer \& Watts(2005)]{StWa05}Strohmayer, T.E., \& Watts, A.L.
564: 2005, \apjl, 632, L111.
565: \bibitem[Strohmayer \& Watts(2006a)]{WaSt06}Strohmayer, T.E. \&
566: Watts, A.L., 2006, \apj, 653, 593.
567: \bibitem[Terasawa et al.(2005)]{Teras05} Terasawa, T., et al. 2005, \nat,434, 1110
568: \bibitem[Thompson \& Duncan(1993)]{TD93}Thompson, C., \& Duncan, R.,C. 1993,
569: \apj, 408, 194
570: \bibitem[Thompson \& Duncan(1995)]{TD95}Thompson, C., \& Duncan, R. C. 1995,
571: \mnras, 275, 255
572: \bibitem[Thompson \& Duncan(1996)]{TD96}Thompson, C., \& Duncan, R. C. 1996,
573: \apj, 473, 322
574: \bibitem[Thompson \& Duncan(2001)]{TD01}Thompson, C., \& Duncan, R. C. 2001,
575: \apj, 561, 980
576:
577: \bibitem[Woods et al.(2002)]{Woo02} Woods, P.M. et al. \apjl, 576, 381
578: \bibitem[Woods \& Thompson(2006)]{WooTho06}Woods, P.M., \& Thompson, C.
579: 2006, in "Compact Stellar X-ray Sources"
580: Ed. Walter Lewin, W.H.G., van der Klis, M. (Cambridge University
581: Press), p.547.
582: \end{thebibliography}
583:
584: %\clearpage
585: \end{document}
586:
587: \begin{figure}
588: \plotone{f1.eps}
589: \caption{}
590: \label{}
591: \end{figure}
592:
593:
594: \end{document}
595: