1: %%
2: %% Beginning of file 'sample.tex'
3: %%
4: %% Modified 2005 December 5
5: %%
6: %% This is a sample manuscript marked up using the
7: %% AASTeX v5.x LaTeX 2e macros.
8:
9: %% The first piece of markup in an AASTeX v5.x document
10: %% is the \documentclass command. LaTeX will ignore
11: %% any data that comes before this command.
12:
13: %% The command below calls the preprint style
14: %% which will produce a one-column, single-spaced document.
15: %% Examples of commands for other substyles follow. Use
16: %% whichever is most appropriate for your purposes.
17: %%
18: \documentclass[12pt,preprint]{aastex}
19:
20: %% manuscript produces a one-column, double-spaced document:
21:
22: %%\documentclass[manuscript]{aastex}
23:
24: %% preprint2 produces a double-column, single-spaced document:
25:
26: %\documentclass[preprint2]{aastex}
27:
28: %% Sometimes a paper's abstract is too long to fit on the
29: %% title page in preprint2 mode. When that is the case,
30: %% use the longabstract style option.
31:
32: %% \documentclass[preprint2,longabstract]{aastex}
33:
34: %% If you want to create your own macros, you can do so
35: %% using \newcommand. Your macros should appear before
36: %% the \begin{document} command.
37: %%
38: %% If you are submitting to a journal that translates manuscripts
39: %% into SGML, you need to follow certain guidelines when preparing
40: %% your macros. See the AASTeX v5.x Author Guide
41: %% for information.
42:
43: \newcommand{\vdag}{(v)^\dagger}
44: \newcommand{\myemail}{majello@mpe.mpg.de}
45:
46: \newcommand{\drv}[2]{\frac{\partial #1}{\partial #2} }
47: %% You can insert a short comment on the title page using the command below.
48:
49: %\slugcomment{}
50:
51: %% If you wish, you may supply running head information, although
52: %% this information may be modified by the editorial offices.
53: %% The left head contains a list of authors,
54: %% usually a maximum of three (otherwise use et al.). The right
55: %% head is a modified title of up to roughly 44 characters.
56: %% Running heads will not print in the manuscript style.
57:
58: \shorttitle{X-ray Background with {\em Swift}/BAT}
59: \shortauthors{Ajello et al.}
60:
61: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
62: \begin{document}
63:
64:
65: \title{Cosmic X-ray background and Earth albedo Spectra with {\em Swift}/BAT}
66:
67: \author{M. Ajello\altaffilmark{1},
68: J. Greiner\altaffilmark{1}, G. Sato\altaffilmark{2},
69: D. R. Willis\altaffilmark{1},
70: G. Kanbach\altaffilmark{1}, A. W. Strong\altaffilmark{1},
71: R. Diehl\altaffilmark{1}, G. Hasinger\altaffilmark{1},
72: N. Gehrels\altaffilmark{2},
73: C. B. Markwardt\altaffilmark{2}
74: and J. Tueller\altaffilmark{2}
75: }
76: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77: \altaffiltext{1}{Max-Planck f\"ur Extraterrestrische Physik, Postfach 1312,
78: 85741, Garching, Germany}
79: \altaffiltext{2}{Astrophysics Science Division, Mail Code 661, NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA }
80: \email{majello@mpe.mpg.de}
81:
82:
83: \begin{abstract}
84:
85: We use {\em Swift}/BAT Earth occultation data
86: at different geomagnetic latitudes to derive
87: a sensitive measurement of the Cosmic X-ray background (CXB)
88: and of the Earth albedo emission in the 15--200\,keV band.
89: We compare our CXB spectrum with recent ({\it INTEGRAL}, BeppoSAX) and
90: past results (HEAO-1) and find good agreement. Using an independent
91: measurement of the CXB spectrum we are able to confirm our results.
92: This study shows that the BAT CXB spectrum has a normalization
93: $\sim8\pm3$\,\%
94: larger than the HEAO-1 measurement.
95: The BAT accurate Earth albedo spectrum can be used to predict
96: the level of photon background for satellites
97: in low Earth and mid inclination orbits.
98: \end{abstract}
99:
100: %% Keywords should appear after the \end{abstract} command. The uncommented
101: %% example has been keyed in ApJ style. See the instructions to authors
102: %% for the journal to which you are submitting your paper to determine
103: %% what keyword punctuation is appropriate.
104:
105: \keywords{cosmology: observations -- diffuse radiation -- galaxies: active
106: X-rays: diffuse background -- galaxies -- Earth}
107:
108:
109:
110: \section{Introduction}
111: There is a general consensus that the cosmic X-ray
112: background (CXB), discovered more than 40 years ago \citep{giacconi62},
113: is produced by integrated emission of extra-galactic point sources.
114: The deepest X-ray surveys to date \citep{giacconi02,alexander03,hasinger04}
115: have shown that up to virtually 100\% of the $<$ 2\,keV
116: CXB radiation is accounted for by Active Galactic Nuclei (AGN)
117: hosting accreting super-massive black holes (SMBHs).
118: However, the fraction of CXB emission resolved into AGNs declines
119: with energy being $<50$\% above 6\,keV \citep{worsley05}.
120: The unresolved component may be attributed to
121: the emission of a yet undetected population of highly-absorbed AGN.
122: These AGN should be characterized by having column densities
123: $\sim10^{24}$ H-atoms cm$^{-2}$ and a space density peaking at redshift
124: below 1 \citep{worsley05}.
125: Such a population of Compton-thick AGN is invoked by population
126: synthesis models \citep[e.g.][]{comastri95,treister05,gilli07}
127: to reproduce the peak of the CXB emission at 30\,keV \citep{marshall80}.
128:
129: Thus, an accurate measurement of the CXB spectrum in the 15--200\,keV
130: energy range is important to assess and constrain
131: the number density of Compton-thick AGNs. Such measurements are
132: complicated by the fact that instruments sensitive in this energy
133: range are dominated by internal detector background and are
134: not designed to measure the CXB spectrum directly (excluding HEAO-1 A2).
135: The typical approach is to produce an $ON-OFF$ measurement, where
136: taking the difference between the $ON$ and the $OFF$
137: pointings eliminates the internal background component.
138:
139: There are different methods to obtain a suitable $OFF$ observation;
140: the HEAO-1 measurement of the CXB spectrum in the 13--180\,keV range
141: was obtained by blocking the aperture with a movable CsI crystal.
142: Also the Earth disk can be used to modulate the CXB emission.
143: This approach is the one used in recent CXB intensity measurements
144: performed by {\it INTEGRAL} and BeppoSAX \citep{churazov07,frontera07}.
145:
146:
147:
148: Here we report on two independent measurements of the CXB emission
149: using {\em Swift}/BAT. For the first method, we use the Earth
150: occultation technique similarly to the {\it INTEGRAL} and BeppoSAX analyses while
151: for the second one we make use of the spatial distribution of the BAT
152: background.
153:
154: The structure of the paper is as follows. In $\S$~\ref{sec:obs}
155: we present the details of the observations and describe the BAT background
156: components. We also derive a rate-rigidity
157: relation which is fundamental for suppressing the background variability
158: due to Cosmic Rays (CRs).
159: In $\S$~\ref{sec:analysis} we present the details of the Earth's
160: occultation episodes undergone by BAT and the analysis method for
161: the occultation measurement. In $\S$~\ref{sec:err}, we discuss
162: all sources of uncertainties which affect our occultation measurement
163: which is then presented in $\S$~\ref{sec:results}.
164: The alternative measurement used to verify the results of the occultation
165: analysis is reported in $\S$~\ref{sec:quad}.
166: We discuss the broad band properties of the CXB and Earth spectra
167: in $\S$~\ref{sec:disc}.
168: Finally, the last section summarizes our findings.
169:
170: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
171: % Observations
172: %
173: \section{Observations}\label{sec:obs}
174: The Burst Alert Telescope \citep[BAT;][]{barthelmy05}, aboard the
175: {\it Swift} mission \citep{gehrels04}, launched by NASA in 2004,
176: represents
177: a major improvement in sensitivity for X-ray imaging of the hard X-ray sky.
178: BAT is a coded mask telescope with a wide field of view
179: (FOV, $120^{\circ} \times 90^{\circ}$ partially coded)
180: sensitive in the hard X-ray domain (15-200\,keV).
181: BAT's main purpose is to locate and to study Gamma-Ray Bursts (GRBs).
182: While chasing new GRBs,
183: BAT surveys the hard X-ray sky with an unprecedented sensitivity.
184: Thanks to its wide FOV and its pointing strategy,
185: BAT monitors continuously a large fraction of the sky (up to 80\%) every day.
186:
187:
188: The {\it Swift} satellite constraints require that
189: the pointing direction be at least 30$^{\circ}$ above
190: the Earth's horizon. Nevertheless,
191: due to its extent, it may happen that the
192: Earth disk occults a substantial portion (up to 30\%) of the BAT FOV.
193: Moreover, BAT survey data include episodes of large
194: occultation (up to $\sim$70\%) caused by the Earth when the spacecraft
195: was in 'safe' mode.
196:
197: We use 8 months of BAT data which constitutes a well characterized
198: dataset of BAT survey data \citep[see][for details]{ajello08a,ajello07b}
199: to study the different components of the BAT background.
200: Our first aim is to derive the BAT background spectrum in the infinite-rigidity
201: approximation. We then use all occultation episodes, as described in
202: $\S$ \ref{sec:analysis}, to derive
203: a measurement of the CXB and the Earth atmosphere spectra.
204:
205: %
206: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
207: % The BAT background
208: %
209: \subsection{The BAT background}
210: The BAT background is highly complex and structured;
211: it exhibits variability dependent on both orbital position and
212: pointing direction. BAT employes a graded-Z fringe shield
213: to suppress the in-orbit background. The fringe shield, located
214: around and below the BAT detector plane, reduces the isotropic cosmic
215: diffuse flux and the anisotropic Earth albedo by $\sim$95\%
216: \citep{barthelmy05}. The two main background components
217: are the CXB emission and the cosmic ray induced (prompt and delayed)
218: backgrounds.
219:
220: The CXB spectrum in the 3--400\,keV range is derived from HEAO-1 data.
221: The following analytical approximation was suggested by \cite{gruber99}:
222: \begin{equation}
223: S_{CXB}(E) = \left\{
224: \begin{array}{lr}
225: 7.877 E^{-0.29} e^{-E/41.13} & 3 < E < 60\,{\rm keV} \\
226: %
227: 0.0259 \left(\frac{E}{60}\right)^{-5.5} + 0.504\left( \frac{E}{60}\right)^{-1.58} + 0.0288 \left( \frac{E}{60}\right)^{-1.05}
228: & E > 60\,{\rm keV}
229: \end{array}
230: \right.
231: \label{eq:gruber}
232: \end{equation}
233: %
234: %
235: where $S_{CXB}(E)$ is expressed in units of
236: keV cm$^{-2}$ s$^{-1}$ sr$^{-1}$ keV$^{-1}$. Given the large FOV
237: ($\sim$1.4\,sr half coded), the CXB is the dominant background
238: component in BAT up to $\sim$50--60\,keV.
239:
240:
241: The prompt CR background is due to spallation effects
242: of incident CRs on the material of the spacecraft; since the Earth magnetic
243: field modulates the flux of incident CRs across the orbit,
244: such background component is expected to vary with the cut-off
245: rigidity R$_c$ (i.e., the minimum momentum an incident charged particle must
246: have in order to penetrate into the Earth's magnetosphere).
247: The delayed component is caused by the excitation of the materials from
248: the incident CR flux. This component builds up
249: on times short compared to the relevant decay lifetimes,
250: then varies as the slower of the irradiation or the lifetime.
251: The Earth's magnetic field includes an indentation in the southern
252: hemisphere
253: called the South Atlantic Anomaly (SAA).
254: During each SAA passage, BAT experiences a sharp increase in count rate
255: due to the increase of the incident CR flux and a delayed background
256: due to de-excitation of the spacecraft materials.
257:
258: In order to discriminate the various components of the BAT background,
259: we correlated the BAT whole array
260: rate (in each energy channel and normalized by the number of working detectors)
261: with several orbital parameters.
262: The final goal is to derive a ``steady-state'' BAT background model which is
263: unaffected by orbital variations.
264:
265:
266: %
267: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
268: % Data cuts
269: %
270: \subsubsection{Data Selections} \label{sec:cuts}
271: Our aim is to determine a rate-rigidity relation in order to extrapolate
272: the BAT array rates to the infinite-rigidity case; this allows us to
273: model the background variability due to the prompt and delayed
274: CR components.
275:
276:
277: First we selected the data
278: excluding all observations where sources with signal-to-noise
279: ratio (S/N) greater than 8 are detected. An 8$\sigma$ source produces
280: an increase in rate of less than 0.5\% in a typical 300\,s
281: observation\footnotemark{}; thus
282: all point-like sources below this limit give a negligible contribution to
283: the background level.
284: \footnotetext{BAT survey observation have typically an exposure of 300\,s,
285: although shorter and longer exposures might exist.}
286:
287:
288: The next step is to eliminate all observations whose exposure time is
289: less than 300\,s. As Fig.~\ref{fig:saa} (left panel) shows,
290: for exposures below 300\,s,
291: the rates show a clear anti-correlation with exposure
292: time, with an increase of a factor $\sim$3 in
293: the rates for few seconds of exposures. Exposures below 300\,s
294: are usually the result of a truncated observation
295: because: 1) BAT detects a GRB or
296: 2) BAT enters in the SAA and data acquisition is suspended.
297: For any of these reasons, data of truncated exposures are excluded by the
298: present analysis because they are
299: not representative of the average BAT background.
300: \begin{figure*}[ht!]
301: \begin{center}
302: \begin{tabular}{cc}
303: \includegraphics[scale=0.8]{f1a_low.eps}
304: \includegraphics[scale=0.8]{f1b_low.eps}\\
305: \end{tabular}
306: \end{center}
307: \caption{
308: {\bf Left Panel:}
309: BAT rate at 100\,keV versus exposure time. The ``truncated'' exposures below
310: 300\,s are noisy. Despite the impression most ($\geq$90\,\%) of the observations
311: have exposure larger than (or equal) 300\,s.
312: {\bf Right Panel:}
313: BAT rate at 100\,keV versus time after each SAA passage.
314: Note the sharp increase
315: in rate and decay behaviour when the spacecraft exits the SAA.
316: The second sharp peak at $\sim$4000\,s is due to a subsequent passage within
317: the tail of the SAA.
318: The rate modulation visible after $\sim$6000\,s is due to the
319: Earth's magnetosphere. Negative times are used for exposure taken within
320: the SAA.}
321: \label{fig:saa}
322: \end{figure*}
323:
324:
325: Whenever the spacecraft exits the SAA, BAT experiences a rate
326: decline due to de-excitation of spacecraft materials.
327: As shown in Fig.~\ref{fig:saa} (right panel) the rates reach their normal
328: level after $\sim$5600\,s after each SAA passage. By excluding all
329: observations taken within $\sim$5600\,s of an SAA passage, we thus eliminate
330: short-lived radioactivity effects.
331:
332: The BAT rates also show a correlation with the angle between the Sun
333: and the pointing direction. This correlation becomes visible
334: at angles $>$120$^{\circ}$
335: and decreases with energy, disappearing at $70-80$\,keV.
336: The reason of this rate increase with the angle to the Sun
337: is unclear. However, since the number of these
338: observations is small ($\sim$5\,\%), we decided to exclude them
339: from the present analysis.
340:
341:
342: The BAT effective area declines with energy and
343: at $\sim$ 200 keV reduces to 1/5 of its peak value
344: at 50 keV; moreover, at these energies the fringe shield becomes partially
345: transparent. Thus, it is possible to use the high energy channels as a
346: ``particle'' detector to monitor the background level of the instrument
347: (i.e. these channels do not yield much information on any celestial signal).
348: We found that imposing that the rate of the last energy channel (194\,keV--6.5\,MeV) be
349: in the range 10--20\,ct s$^{-1}$ eliminates roughly 1\% of the observations
350: which are outliers in all the correlations we have studied.
351:
352: After these cuts, we find, as expected,
353: that the rates in each energy channel decrease
354: as a function of the cut-off rigidity R$_C$. This effect
355: is shown for two representative energy channels in Fig.~\ref{fig:rate_rig}.
356: We model this behavior with an exponential and a constant
357: ($Rate = C + Be^{\alpha R_C}$),
358: and fit this model to each
359: energy channel. The fitted constant $C$ provides an estimate
360: of the BAT rate in the infinite-rigidity extrapolation.
361: The distribution of the steepness of the rate increase with rigidity
362: ($\alpha$ values), shown in left panel
363: of Fig.~\ref{fig:err},
364: has a mean of -0.34
365: and a RMS of 0.04, in perfect agreement with previous measurements
366: \citep{imhof76}.
367:
368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 2
369: \begin{figure*}[ht!]
370: \begin{center}
371: \begin{tabular}{cc}
372: \includegraphics[scale=0.43]{f2a_low.eps}
373: \includegraphics[scale=0.43]{f2b_low.eps}\\
374: \end{tabular}
375: \end{center}
376: \caption{
377: {\bf Left Panel:}
378: Correlation of BAT rate with the cut-off rigidity at 100\,keV. The dashed
379: line is the best fit using an exponential plus a constant.
380: Only those observations which passed the
381: selection criteria explained in $\S$~\ref{sec:cuts} were used.
382: {\bf Right Panel:}
383: Correlation of BAT rate with the cut-off rigidity at 170\,keV. The dashed
384: line is a exponential plus a constant fit. The outliers present
385: in both figures around 12\, GV are due to the spacecraft being in the
386: vicinity of the SAA.
387: Only those observations which passed the
388: selection criteria explained in $\S$~\ref{sec:cuts} were used.
389: }
390: \label{fig:rate_rig}
391: \end{figure*}
392:
393:
394: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 3
395: \begin{figure*}[ht!]
396: \begin{center}
397: \begin{tabular}{lc}
398: \includegraphics[scale=0.43]{f3a.eps}
399: \includegraphics[scale=0.43]{f3b.eps} \\
400: \end{tabular}
401: \end{center}
402: \caption{
403: {\bf Left Panel:}
404: Distribution of the $\alpha$ values. The mean of -0.34 is in
405: good agreement with measurements from \cite{imhof76}.
406: {\bf Right Panel:}
407: BAT background spectrum extrapolated in the infinite rigidity case.
408: }
409: \label{fig:err}
410: \end{figure*}
411:
412: The BAT spectrum obtained by extrapolating the rates of each energy channel
413: to infinite rigidity is shown in the right panel of Fig.~\ref{fig:err}.
414: The bumpiness between 60 and 100\,keV is due to the
415: numerous fluorescence emission lines from the fringe shield
416: \citep[see][for details]{willis02}.
417:
418:
419: %
420: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
421: %
422: % Earth Occultation
423: %
424: \section{Earth occultation} \label{sec:analysis}
425: The {\it Swift} orbital constraints require that the BAT pointing direction
426: be always at least 30$^{\circ}$ away from the Earth horizon.
427: This is because the Earth is bright in Optical and X-rays, thus
428: it may damage the UV/Optical telescope (UVOT) and the X-ray telescope (XRT).
429: On May 31, June 12 and July 28, 2005, the {\em Swift} spacecraft entered into
430: 'safe'-mode because of star tracker loss of lock.
431: In safe-mode
432: operations, the XRT and UVOT telescopes are closed, but BAT still takes
433: data. The spacecraft remains in sun reference pointing
434: until commanding
435: from the ground recovers {\em Swift} back to its normal status.
436: In the timespan between the safe-mode and the recovering operation,
437: the satellite uses the magnetometers and the sun sensor to derive
438: its pointing direction.
439: At least in the occasions mentioned
440: above\footnotemark{},
441: \footnotetext{A few episodes of Earth occultation were
442: found in BAT data, but some did not pass the criteria
443: explained in Sec. \ref{sec:cuts}}
444: %
445: the Earth passed through the BAT field of view (FOV). Fig.~\ref{fig:occ_ptg}
446: shows the BAT pointing directions during the deep occultation episodes
447: described here.
448:
449:
450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 4
451: \begin{figure}[ht!]
452: \begin{center}
453: \includegraphics[scale=0.9]{f4.eps}
454: \end{center}
455: \caption{Pointing directions for the occulted observations. Marker sizes
456: vary linearly with the occulted FOV fraction. Occultations vary
457: from 0.1\,\% to $\sim$80\,\%. In order to avoid contamination of the CXB signal
458: from the Galactic Ridge emission, we used only occultation episodes at
459: $|$b$|>$20$^{\circ}$ (data points already excised from this plot).}
460: \label{fig:occ_ptg}
461: \end{figure}
462:
463:
464:
465: %%%
466: %%%
467: As the Earth partially occults the FOV, BAT registers a sharp decrease in rate
468: due to occultation of the CXB emission. This is especially evident below 40\,keV
469: where the CXB radiation dominates above the Earth's atmospheric components.
470: The left panel of Fig.~\ref{fig:occ} clearly shows the drop
471: in rates caused by the occulting Earth; between 18--20\,keV
472: the rates drop by a factor 3.5 when $\sim$ 60\% of the BAT FOV is occulted
473: (as shown in the right panel of \ref{fig:occ}).
474: %
475: %
476: Thus, the Earth occultation can be used to measure the CXB emission by means
477: of the depression caused in the BAT rates.
478: Unfortunately, the Earth is not only a passive occulter,
479: but also an active emitter. The Earth is a powerful
480: source of X- and gamma-rays due to cosmic ray bombardment of its
481: atmosphere \citep[see][]{petry05,sazonov07b}.
482: This radiation is usually referred to
483: as albedo, and it is discussed briefly in the next section.
484:
485: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 5
486: \begin{figure*}[ht!]
487: \begin{center}
488: \begin{tabular}{cc}
489: \includegraphics[scale=0.43]{f5a.eps}
490: \includegraphics[scale=0.40,trim=0 -38 0 0,clip=true]{f5b.ps}\\
491: \end{tabular}
492: \end{center}
493: \caption{
494: {\bf Left Panel:}
495: BAT rate, in the 18--20\,keV range, as a function of the elevation angle
496: above the Earth horizon. For ELV $<$30--60$^{\circ}$, BAT starts to experience
497: Earth occultations. The decrease in rate is expected to vary linearly with
498: the occulted solid angle if Earth emission is negligible. Note that this
499: graph is not a light curve as each data point is a separate (in time)
500: 300 s observation.
501: {\bf Right Panel:}
502: Example of deep Earth occultation of the BAT FOV. The black area is
503: the region of the BAT FOV which is completely occulted by the Earth
504: during the 300 s observation; the lighter gray is the un-occulted part of the
505: FOV, while the region in between the black and the lighter gray is partially
506: occulted due to the spacecraft movement in the 300\,s. The colorbar shows the
507: fractional time a given sky pixel is unocculted.
508: }
509: \label{fig:occ}
510: \end{figure*}
511:
512:
513: %
514: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
515: %
516: % Atmospheric Albedo
517: %
518: \subsection{Atmospheric Albedo Gamma-rays}\label{sec:earth}
519: The atmospheric Albedo flux is produced by cosmic ray interactions
520: in the Earth's atmosphere. Hadronic interaction with
521: atmospheric nuclei of the incident
522: cosmic rays leads to the production of an electromagnetic and nuclear
523: cascade with muons, nuclear fragments, and other
524: hadrons.
525: Gamma-rays above 50 MeV are produced mainly by the decay of mesons, while
526: at X-ray energies the main source can be attributed to bremsstrahlung
527: from secondary electrons.
528:
529:
530: Measurements of the X-ray albedo radiation are reported in \cite{schwartz74}
531: in the 1--100\,keV energy range and by \cite{imhof76} above 40\,keV.
532: The Albedo spectrum measured by \cite{schwartz74} shows a cut-off
533: below 30\,keV, probably due to self-absorption of the radiation emitted
534: from the inner layers of the atmosphere and a progressive flattening around
535: 40\,keV. Above 40\,keV the Albedo emission decreases as
536: a power law with photon index of $\sim$1.4-1.7.
537: This power-law behavior is confirmed by other experiments
538: \cite[e.g.][]{schoenfelder80,gehrels92} and by a recent Monte Carlo
539: simulation of the hard X-ray emission of the
540: Earth's atmosphere \citep{sazonov07b}.
541: However the absolute normalization of the
542: observed Earth spectrum depends on the altitude and the inclination of the
543: satellite's orbit and on the solar cycle.
544:
545:
546:
547:
548:
549: %
550: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
551: %
552: % Method of analysis
553: %
554: \subsection{Method of analysis}
555: The rate $R(E)_i$ measured at energy $E$
556: by BAT in a given observation, during which
557: the Earth is in the FOV, can be described as:
558: \begin{equation}\label{eq:occ1}
559: %
560: R(E)_i = I(E)_i - \bar{\Omega}_i\cdot[ R(E)_{CXB,i} - R(E)_{Earth,i}]
561: \end{equation}
562:
563: where the subscript $i$ refers to the $i-th$ observation, $\bar{\Omega_i}$
564: is the ``effective''
565: solid angle occulted by the Earth, $I(E)$ is the total background,
566: and $R(E)_{CXB}$ and $R(E)_{Earth}$
567: are the CXB and the Earth emission respectively.
568:
569: The observations we are dealing with are generally non-contiguous,
570: and thus all changes in the instrument configuration
571: (e.g. number of working detectors) must be taken into account.
572: We do this by computing the ``effective'' solid angle
573: occulted by the Earth for each observation.
574: This is defined as:
575: \begin{equation}\label{eq:effomega}
576: \bar{\Omega}_i = \sum_{j=0}^{N_p} \omega_j \cdot (1-\Delta T^{Frac}_j) \cdot V^i_j
577: \end{equation}
578: %
579: where $N_p$ is the total number of sky pixels,
580: $\Delta T^{Frac}_j$ is the fractional exposure time\footnotemark{}
581: \footnotetext{The fractional exposure time is the fraction
582: of the exposure time the sky pixel is unocculted. Thus, it
583: varies from 0 to 1 for completely occulted and unocculted pixels respectively.}
584: a sky pixel
585: of solid angle $\omega_j$ is unocculted and $V^i_j$ is the
586: vignetting affecting that sky pixel during the $i$-th observation.
587:
588:
589: Equation \ref{eq:occ1} shows the ``degeneracy'' problem which limits
590: the Earth occultation technique when used to determine the CXB emission.
591: Indeed, the measured depression of the rates with respect to the normal
592: sky intensity level are a measurement of the difference
593: of the CXB intensity and the Earth's atmospheric emission.
594: Following the notation of Equation \ref{eq:occ1}, this can be
595: expressed as $R(E)_i=R(E)_{CXB,i}-R(E)_{Earth,i}$.
596: We adopt here an approach similar to the one
597: of \cite{churazov07} and \cite{frontera07},
598: which consists of deriving the ``difference'' ({\it ON - OFF})
599: spectrum and fitting it with a-priori spectral models.
600:
601:
602: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 6
603: \begin{figure}[ht!]
604: \begin{center}
605: \includegraphics[scale=0.80]{f6.eps}
606: \end{center}
607: \caption{Fit example for the 18-20 keV energy channel.
608: The long dashed line is the best fit to the data. Note that
609: the occultation of the CXB emission produces a very strong signal
610: reducing the background rate by $\sim$75\%.
611: \label{fig:fit_ex}}
612: \end{figure}
613:
614: The difference spectrum is derived fitting Equation~\ref{eq:occ1}
615: to each energy channel (an example is shown in
616: Fig.~\ref{fig:fit_ex}). In all these independent fits
617: the two parameters ($I(E)_i$ and $R(E)$) are left unconstrained.
618: Moreover, in order to avoid contamination
619: by the Galactic Ridge emission we used
620: only occultation episodes at Galactic latitude larger than 20$^{\circ}$.
621: The difference spectrum is shown in Fig.~\ref{fig:diff_raw}.
622: However, before describing the spectral fit we discuss in detail
623: the sources of systematic uncertainties affecting our analysis.
624:
625:
626: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 7
627: \begin{figure}[ht!]
628: \begin{center}
629: \includegraphics[scale=0.80]{f7.eps}
630: \end{center}
631: \caption{The {\it ON-OFF} difference spectrum obtained fitting Equation~\ref{eq:occ1}
632: to each energy channel as shown in Fig.~\ref{fig:fit_ex}. The inset shows
633: the negative part of the spectrum. Above 100\,keV the albedo spectrum
634: dominates the CXB emission.
635: \label{fig:diff_raw}}
636: \end{figure}
637:
638:
639: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
640: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
641: \subsubsection{A Note on the ``Degeneracy'' Problem}\label{sec:deg}
642:
643: The ``degeneracy'' problem (i.e. the fact that the CXB and the
644: albedo emissions leave a similar signature during the
645: occultation of the BAT FOV) might in principle be
646: alleviated modeling the albedo emission of the Earth.
647: This involves modeling the emission as a function of the cut-off
648: rigidity of the visible disk as well as the (reasonably) expected non-uniformity
649: of the albedo emission (i.e. limb or disk brightening effects).
650:
651: Indeed, since the albedo emission is generated at different cut-off rigidities,
652: with respect the local rigidity of the satellite, one might reasonably
653: expect that patches of the disk located at lower cut-off rigidities
654: emit a larger X-ray flux. This information might be used to disentangle
655: the albedo from the CXB signal.
656: Moreover, the Earth is known to be a non-uniform emitter at MeV and GeV
657: energies. Both COMPTEL and EGRET \citep{schoenfelder80,petry05} have shown
658: that the Earth exhibits a bright limb. Thus, also this information might be used
659: to model the expedected emission.
660:
661: However, we note that a few factors limit the application, in this analysis,
662: of the modeling described above. The limitations come from the fact that
663: this analysis is entirely based on survey data. As explained in
664: $\S$~\ref{sec:cuts}, the typical integration time for the survey is 300\,s.
665: Thus, we do not have a time-resolved monitoring (e.g. 1\,s time resolution) of the
666: transit of the Earth across the BAT FOV, but only 300\,s snapshot observations with
667: different level of occultations. Moreover, since the Earth is moving in
668: the FOV within these 300\,s, all the physical quantities (e.g. cut-off rigidity,
669: fraction of the occulted FOV, etc.) are necessarily averaged over this time.
670: Another limitation is due to the fact that the observations used here are not
671: contiguous in time, but separated by weeks or months. Thus, the changing background
672: conditions limit the precision of this analysis (as also shown in
673: $\S$~\ref{sec:ratevar}).
674:
675: These facts limit the usage of a precise modeling of the Earth albedo emission
676: which would allow to disentangle the albedo and the CXB signal without assuming
677: a-priori spectral templates. A dedicated Earth observation
678: with BAT in 'burst' mode (i.e. event-by-event mode) would not only allow
679: to overcome the problems shown above, but also would extend the energy
680: range of the measurement up to 350\,keV (instead of 200\,keV) and would also reduce
681: the systematic uncertainties of the measurement to those related to the
682: instrumental response only (see $\S$~\ref{sec:err}).
683:
684:
685: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
686: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
687: \section{Analysis of the Uncertainties}\label{sec:err}
688:
689: %
690: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
691: %
692: % ERROR Analysis
693: \subsection{Rate Variation}\label{sec:ratevar}
694:
695: The rate-rigidity graphs (examples are shown in Fig.~\ref{fig:rate_rig})
696: show a scatter in the rate around the best fit which is generally larger
697: than the statistical errors. This scatter is
698: due to unknown effects.
699: The pointing directions, the solar cycle, the
700: spacecraft orientation with respect to the
701: Earth and the Sun could be at the origin of
702: this scatter which has an amplitude of less than 10\%.
703: %
704: We modeled the scatters
705: as a Gaussian distribution such that the 1\,$\sigma$ width
706: of this distribution gives
707: for each energy channel an estimate of the total (statistical plus
708: systematic) error of the extrapolated rates. This constitutes
709: the baseline uncertainty of this analysis, and it is propagated throughout
710: all the further steps.
711:
712:
713: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
714: \subsection{Uncertainties connected to imprecise attitude determination}
715: During safe-mode operations, attitude determination relies
716: on the magnetometers and Sun sensor. The derived attitude
717: solution has a precision of the order of $\sim$degree.
718: This is confirmed by the
719: analysis of sources detected during safe-mode pointings, which shows that
720: the attitude differs from the nominal pointing direction by 1--2 degrees.
721: The effective solid angle, computed in Eq.~\ref{eq:effomega}, is a
722: slowly-varying function per degree of occultation.
723: As shown in the right panel of Fig.~\ref{fig:errors},
724: the fractional effective solid angle
725: can be approximated by a straight line
726: with a slope of 0.010\,deg$^{-1}$ in the 0.4--0.8 range of
727: fractional occulted FOV.
728: This means
729: that an error of (at most) 2 degrees in the attitude determination
730: translates into an uncertainty of $\sim$2\% in the determination of
731: the occulted portion of BAT FOV. This additional systematic
732: uncertainty is taken into account in our analysis.
733:
734: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 8
735: \begin{figure*}[ht!]
736: \begin{center}
737: \begin{tabular}{cc}
738: \includegraphics[scale=0.43]{f8a.eps}
739: \includegraphics[scale=0.43]{f8b.eps}
740: \end{tabular}
741: \end{center}
742: \caption{
743: {\bf Left Panel:}
744: Variation of the fractional effective solid angle as a function
745: of elevation (distance of the pointing direction with respect
746: the Earth limb).
747: {\bf Right Panel:}
748: Fractional 1\,$\sigma$ total uncertainty as a function of energy. The fractional
749: uncertainty includes all the error estimates outlined
750: in $\S$~\ref{sec:err}.
751: The total error reaches its minimum of 4--5\,\%
752: at the peak of the CXB emission. The peak around 100\,keV is artificial
753: and corresponds to the change of sign of the difference spectrum.
754: Uncertainties above 100\,keV are $\sim$20\,\% and primarily systematic
755: in origin.
756: \label{fig:errors}}
757: \end{figure*}
758:
759:
760:
761:
762: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
763: \subsection{Uncertainties of the BAT instrumental response}
764: \label{subsec:sysbat}
765: The BAT is a well calibrated instrument. However, given the very large
766: FOV and the uncertainty in the modeling of spacecraft materials, the
767: Crab Nebula is detected with slightly different spectral parameters across
768: the FOV. To cope with this uncertainty, users are encouraged, when performing
769: spectral fitting, to use
770: a vector of energy-dependent
771: systematic errors\footnotemark{} which allows a unique spectral fit
772: to the Crab Nebula wherever in the FOV.
773: \footnotetext{A detailed discussion is reported in http://swift.gsfc.nasa.gov/docs/swift/analysis/bat\_digest.html.}
774: In this analysis, we account for such systematic errors, which fortunately have their minimum ($\sim$4\%) in the 20-80 keV band.
775:
776:
777:
778: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
779: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
780: \section{Instrumental response to a diffuse source}\label{sec:resp}
781: The BAT response was developed by characterizing individual CdZnTe
782: detector pixels, and by modeling the absorption and modulation of
783: the coded aperture mask, then finally verifying by Monte Carlo
784: simulations with radio active sources \citep{sato05}.
785: However, since there remained uncertainty in response to continuum emissions,
786: the response was adjusted to fit the Crab nebula spectrum$^4$.
787: The BAT Crab spectrum can be described as:
788: $dN/dE = 10.40\ E^{-2.14}$ photons cm$^{-2}$ s$^{-1}$ keV$^{-1}$.
789: The values of normalization and photon index are well within
790: those used by most of the X-ray missions in a similar band
791: \citep[see ][ for a review of Crab Nebula spectral parameters]{kirsch05}.
792:
793:
794: However, the analysis of a diffuse source (as the CXB) presents some
795: differences with respect to the study of point-like objects.
796: Indeed, the spectrum of a point-like source is modulated by the coded mask pattern.
797: Thus, the indirect unmodulated component which is scattered by the materials
798: of the BAT instrument and of the satellite can be eliminated.
799: Accordingly, the official response generator, {\it batdrmgen},
800: part of the standard BAT software,
801: produces a response only for the direct component.
802: However, the CXB, the subject of this paper, is seen as a diffuse emission and cannot be modulated by the coded mask pattern.
803: We therefore utilized the Monte Carlo simulator to generate a
804: more accurate response for a diffuse emission
805: taking into account the scattered component\footnotemark{}
806: as well as the effect of isotropic illumination of the BAT instrument.
807: This simulator is the one used to verify the response on the ground,
808: but the same corrections to fit the Crab spectrum are also applied.
809: \footnotetext{The so called 'un-coded' (or scattered) component comprises
810: all those events which scatter in the satellite structure and produce
811: a detectable signal in the BAT array.}
812:
813: %
814: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
815: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
816: % Results
817: %
818: \section{Results of the analysis}\label{sec:results}
819: %%%
820: In this section we present the main results of the analysis:
821: the CXB and the Earth's atmosphere spectra.
822: All the uncertainties described in $\S$~\ref{sec:err}
823: were added in quadrature to form the total uncertainty.
824: The dependence of the total uncertainty
825: with energy is shown in
826: Figure~\ref{fig:errors} (right panel). The total uncertainty reaches its
827: minimum value of 4--5\,\% at the peak of the CXB spectrum.
828:
829:
830: In this section, all quoted errors on spectral parameters
831: are 90\% confidence for one interesting parameter.
832:
833:
834:
835: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
836: \subsection{Spectral fitting}
837: The {\it ON-OFF} difference spectrum is folded in XSPEC \citep{arnaud96}
838: with the proper instrumental
839: response for a diffuse source. The model we used for the fit is the
840: difference between the CXB and the albedo spectra.
841: For the CXB spectrum we employ Equation~\ref{eq:gruber}.
842: For the albedo spectrum we use a jointly smoothed double power-law
843: of the form:
844: \begin{equation}\label{eq:joint}
845: \frac{dN}{dE} = \frac{C}{(E/E_{b})^{\Gamma_1} +(E/E_{b})^{\Gamma_2} }\ \ \ \
846: [\rm{photons\, cm^{-2}\,s^{-1}\,sr^{-1}}]
847: \end{equation}
848: %%%%%%%
849: %%%%%%%
850:
851: where $\Gamma_1$ and $\Gamma_2$ are the two spectral indices and $E_{\rm b}$
852: is the break energy.
853: This functional form reproduces well the atmospheric component
854: with its declinde at low energy, bump around 30--40\,keV
855: and a hard spectral
856: index at higher energies \citep{schwartz74,imhof76,gehrels85,frontera07}.
857: Recent Monte Carlo simulations of the Earth emission \citep{sazonov07b}
858: show that Equation~\ref{eq:joint} is a very good approximation of
859: the Earth emission below 300\,keV.
860: We fix the values of the spectral indices and break energy
861: at those suggested by \cite{sazonov07b}, i.e.
862: $\Gamma_1$=-5 and $\Gamma_2$=1.4 and $E_{\rm b}$=44\,keV.
863: Thus, free parameters of our first fit are only
864: the normalizations of Equations~\ref{eq:gruber} and \ref{eq:joint},
865: respectively. The fit is poor, however, with a $\chi^2$ of $\sim220$ for
866: 75 degrees of freedom.
867: Adding free parameters for the high energy spectral index $\Gamma_2$
868: and the break energy $E_{\rm b}$ improves the fit
869: ($\chi^2_{red}$=121.8/73). The F-test confirms that the improvement
870: is very significant (F-test probability of 4.8$\times 10^{-10}$).
871: Adding another free parameter for
872: the low energy spectral index $\Gamma_1$ of the
873: albedo does not improve the fit. Indeed, below 40\,keV the spectrum
874: is completely dominated by the CXB emission, and thus it is not
875: possible to constrain this parameter.
876: Choosing as free variables the parameters of the
877: Earth emission instead of those of the CXB spectrum is well motivated:
878: there are indications \citep{schwartz69,gehrels85,frontera07}
879: that the high-energy spectral index of the albedo emission
880: might be steeper than the classical value of 1.4.
881: On the other side, the formula shown in Eq.~\ref{eq:gruber} \citep{gruber99}
882: is a good representation of the broad band CXB spectrum.
883:
884:
885: Our best-fit parameters (with 90\,\% CL errors) for the albedo spectrum are:
886: $\Gamma_2$=1.72$\pm0.08$, $E_{\rm b}$=33.7$\pm3.5$\,keV
887: and $C$=1.48$^{+0.6}_{-0.3} \times 10^{-2}$.
888: The normalization, and its 90\,\% CL error,
889: of the CXB as measured by BAT
890: with respect Equation~\ref{eq:gruber} is 1.06$\pm0.08$.
891: This error includes also the change
892: by $\pm1$ in the low-energy spectral index ($\Gamma_1$)
893: of the albedo emission. The CXB intensity in the 20--50\,keV band
894: is 6.43($\pm0.20$)$\times10^{-8}$ \,erg cm$^{-2}$ s$^{-1}$ sr$^{-1}$.
895: Figure~\ref{fig:residuals} shows the best fit
896: and its residuals to the difference spectrum.
897:
898:
899:
900:
901:
902: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 9
903: \begin{figure}[ht!]
904: \begin{center}
905: \includegraphics[scale=0.6,angle=270.]{f9.ps}
906: \end{center}
907: \caption{Best fit to the {\it ON-OFF}
908: difference spectrum. The model used is the
909: difference between Equations ~\ref{eq:gruber} and \ref{eq:joint}.
910: Around 100\,keV, the data become negative (as shown in Fig.~\ref{fig:diff_raw}).
911: \label{fig:residuals}}
912: \end{figure}
913:
914:
915:
916: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
917: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
918: \section{Alternative measurement of the X-ray background spectrum}\label{sec:quad}
919:
920: The BAT in-flight background has a peculiar spatial distribution which
921: shows larger count rates towards the center of the detector array and smaller
922: rates towards its edges. This is clearly shown in Fig.~\ref{fig:quadbkg}
923: (left panel).
924: In the process of forming sky images,
925: the BAT software\footnotemark{} removes this background
926: component by means of an empirical bi-dimensional second-order
927: polynomial function. \footnotetext{For reference see the description
928: of the {\it batclean} tool available at
929: http://heasarc.nasa.gov/lheasoft/ftools/headas/batclean.html.}
930:
931:
932:
933: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 10
934:
935: \begin{figure*}[ht!]
936: \begin{center}
937: \begin{tabular}{cc}
938: \includegraphics[scale=0.41,clip=true,trim=0 33 0 34]{f10a.ps}
939: \includegraphics[scale=0.485,clip=true,trim=0 30 0 34]{f10b.ps}\\
940: \end{tabular}
941: \end{center}
942: \caption{
943: {\bf Left Panel:}BAT detector plane image in the 15--55\,keV energy channel.
944: {\bf Right Panel:}Model of the detectors solid angles distribution.
945: The similarity of the model with the real data (left panel) is apparent.
946: \label{fig:quadbkg}}
947: \end{figure*}
948:
949:
950: This feature has
951: an important physical meaning. Indeed, it
952: is the result of a diffuse isotropic source (namely the CXB),
953: shining through the mask. The peculiar shape of this ``background'' component
954: arises from the fact that detectors at the edge of the array have a smaller
955: solid angle of the sky as seen
956: through the mask (they see the mask under large angles)
957: than those at the very center. Given the extent of the BAT mask and array,
958: this effect is noticeable and significant.
959:
960: We thus built a simple model, assigning to each detector its geometrical
961: solid angle through the transparent mask elements. This model is shown
962: in Fig.~\ref{fig:quadbkg}.
963: We can detect the CXB emission by fitting this model to
964: the spatial distribution of the counts in each energy channel.
965: However, this approach is valid only as long as the graded-Z shield remains
966: opaque to X-ray photons($\sim$50\,keV).
967: Indeed, as soon as the shield becomes partially
968: transparent, the effective detector solid angle increases because of the shield
969: transparency. Thus, our model becomes inadequate above this energy.
970:
971: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
972: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
973: \subsection{Model fitting}\label{sec:modelfitting}
974: Among all BAT observations which satisfied the selection criteria
975: outlined in $\S$~\ref{sec:cuts}, we selected only those ones
976: which were unocculted by the Earth.
977: We then summed all the detector plane histograms\footnotemark{}
978: \footnotetext{BAT survey data are in the form of 80 channels detector plane
979: histograms with a typical exposure time of 300\,s.}
980: (DPHs)
981: into a single DPH with an overall exposure of $\sim$1.8\,Ms. Summing
982: the DPHs of observations with different pointing directions achieves the goal
983: of smearing the contribution of sources which are below the detection
984: threshold (8\,$\sigma$ in this case). To each energy channel we fitted a model
985: which is composed of:
986: \begin{itemize}
987: \item a constant term for the edges of the detector modules which register a
988: higher count rate because of the larger exposed area,
989: \item the solid angle distribution model which takes into account the diffuse
990: flux as seen through the mask,
991: \item a constant term which takes into account all other background
992: components including
993: the CR component which penetrates through the shielding.
994: \end{itemize}
995:
996: For each energy channel, the fit independently determines the intensity
997: of the diffuse model.
998: Since all the energy-dependent effects (e.g. absorption through
999: the mask structure and transparency of the lead tiles) are correctly
1000: taken care of in the instrumental response described in $\S$~\ref{sec:resp},
1001: we normalize our model (dividing by the maximum detector solid angle
1002: $\sim$0.6\,sr) and treat the dispersion of the solid angle
1003: distribution ($\sim$0.034\,sr) as a fluctuation.
1004: In this way, we make our approach insensitive to the exact computation
1005: of the solid-angle for each detector and at the same time, it allows us to use
1006: the same response matrix developed for the occultation measurement.
1007: We remark that this response matrix is based
1008: on extensive Monte Carlo simulations.
1009:
1010:
1011:
1012: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1013: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1014: \subsection{Results}
1015: For spectral fitting we convolved the CXB count rate spectrum with the
1016: BAT response matrix. For each energy channel, we summed in quadrature
1017: statistical uncertainty,
1018: the uncertainty on the mean
1019: solid angle (see $\S$~\ref{sec:modelfitting})
1020: and the uncertainty due to the BAT response (see $\S$~\ref{sec:resp}).
1021:
1022: A fit to the data (shown in Fig.~\ref{fig:quadfit})
1023: allowing only the overall CXB normalization
1024: (Equation~\ref{eq:gruber}) to vary yields a $\chi^2$ of 18.8 for
1025: 17 degree of freedom. The normalization with respect to
1026: the level of the CXB as measured by \cite{gruber99} is
1027: $1.09^{+0.03}_{-0.03}$. This measurement is in very good
1028: agreement with the occultation measurement as Fig.~\ref{fig:cxb_double}
1029: shows.
1030:
1031: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 11
1032:
1033: \begin{figure}[ht!]
1034: \begin{center}
1035: \includegraphics[scale=0.6,angle=270.]{f11.ps}
1036: \end{center}
1037: \caption{Best fit to the second CXB measurement using
1038: Equation~\ref{eq:gruber} with only the overall normalization left
1039: as free parameter.
1040: \label{fig:quadfit}}
1041: \end{figure}
1042:
1043: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 12
1044: \begin{figure}[ht!]
1045: \begin{center}
1046: \includegraphics[scale=0.89,angle=0.]{f12.eps}
1047: \end{center}
1048: \caption{The two independent measurements
1049: of the CXB spectrum performed by BAT.
1050: The occultation measurement (gray datapoints) and the measurement
1051: derived using the solid angle distribution (black datapoints) are
1052: in very good agreement.
1053: \label{fig:cxb_double}}
1054: \end{figure}
1055:
1056: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1057: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1058: \section{Comparison with previous measurements}\label{sec:disc}
1059: In this section, we compare the CXB and the albedo spectra
1060: with previously available measurements in the same or overlapping
1061: energy bands. For reference, the values of CXB and albedo emissions
1062: as derived by BAT are reported in Table~\ref{tab:cxb}.
1063:
1064: \input{tab1}
1065:
1066: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1067: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1068: \subsection{The X-ray Background Spectrum}\label{sec:cxbspec}
1069:
1070: Both measurements of the CXB spectrum presented here produce the same
1071: results (within errors) for the normalization of the CXB
1072: intensity at its peak. Combining both measurements
1073: we determine that the CXB intensity at its peak is 8($\pm3$)\,\%
1074: larger than previously measured by HEAO-1 \citep{gruber99}.
1075: We find that the CXB intensity in the 20--50\,keV band
1076: is 6.50$\pm0.15 \times 10^{-8}$\,erg cm$^{-2}$ s$^{-1}$ sr$^{-1}$.
1077: The observed intensity near the peak of the CXB spectrum (expressed
1078: in $\nu$F$_{\nu}$ units) at 30\,keV
1079: is 46.2 keV$^2$ cm$^{-2}$ s$^{-1}$ keV$^{-1}$ sr$^{-1}$.
1080: Figure~\ref{fig:comp} shows the comparison of the BAT CXB spectrum
1081: with all other measurements available above 20\,keV. All measurements
1082: agree well within 10\,\%.
1083: The detailed comparison is reported in Tab.~\ref{tab:comp}.
1084: It is clear that the scatter in CXB intensities does not depend solely on the
1085: adopted spectra for the Crab Nebula. Some of the measurements showed
1086: in Tab.~\ref{tab:comp} might still be affected by systematic uncertainty
1087: in the instrumental response used.
1088: To our knowledge, BAT is the only instrument for which a dedicated
1089: instrumental response has been derived and tested for the analysis of the
1090: CXB.
1091:
1092:
1093: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 13
1094: \begin{figure}[ht!]
1095: \begin{center}
1096: \includegraphics[scale=0.89,angle=0.]{f13.eps}
1097: \end{center}
1098: \caption{Comparison of CXB measurements above 20\,keV.
1099: The BAT spectrum (in red) is in agreement with the HEAO-1 (gray),
1100: {\it INTEGRAL} (blue) and BeppoSAX (black) observations.
1101: For clarity only the BAT occultation measurement is reported.
1102: \label{fig:comp}}
1103: \end{figure}
1104:
1105:
1106: \input{tab2}
1107:
1108:
1109: %%%%%%%%%%%%%%%%%%%%%
1110: %%%%%%%%%%%%%%%%%%%%%
1111:
1112: Fig.~\ref{fig:xall} shows a compilation of the X- and gamma-ray diffuse
1113: backgrounds from keV to GeV energies. In addition to the work of
1114: \cite{gruber99}, we show SMM (MeV) data \citep{watanabe97}, and
1115: {\it COMPTEL} and {\it EGRET} data in a recent revision \citep{weidenspointner00,strong04}.
1116: In particular, the new analysis of {\it EGRET }data shows that the
1117: validity range of the \cite{gruber99} formula (Eq. \ref{eq:gruber})
1118: is now restricted to 3\,keV$<$ E $<$ 1\,MeV.
1119:
1120: We find that a good description of the available data in the
1121: 2\,keV -- 2\,MeV range is achieved using a smoothly-joined
1122: double power-law of the form:
1123: \begin{equation}
1124: E^2 \cdot \frac{dN}{dE} = E^2\cdot \frac{C}{(E/E_{\rm B})^{\Gamma_1} +(E/E_{\rm B})^{\Gamma_2}}\ \ \ \ [{\rm keV}^2 {\rm photons\ cm}^{-2} {\rm s}^{-1} {\rm sr}^{-1}
1125: {\rm keV}^{-1}]
1126: \end{equation}
1127:
1128: The best fit, shown in Fig.~\ref{fig:xall}, yields values
1129: (and 1\,$\sigma$ errors) of: $C= (10.15\pm0.80)\times 10^{-2} $,
1130: $\Gamma_1=1.32\pm0.018$, $\Gamma_2=2.88\pm0.015$
1131: and $E_{\rm B}=29.99\pm1.1$\,keV.
1132: The reduced $\chi^2$ is acceptable ($\sim$1.2)
1133: considering the number (10) of different datasets fitted. The suggested
1134: formula reproduces well the CXB spectrum over two decades in flux
1135: and five in energy. At a given energy, the systematic uncertainty
1136: produced by the scatter of the measurements used here is of the
1137: of the order of 10\,\%.
1138:
1139:
1140: Note that, there is no astrophysical need
1141: to connect the keV and the GeV
1142: diffuse backgrounds with a single formula \citep[e.g.][]{gruber99}.
1143: It is generally agreed that
1144: the source populations contributing to the two diffuse
1145: backgrounds are probably different. Almost all of the
1146: CXB radiation up to 300\,keV can be explained in terms of
1147: emission-line AGN \citep[e.g.][]{gilli07}. Moreover,
1148: taking into account (the likely, but not yet detected) population
1149: of non-thermal electrons in AGN coronae, \cite{inoue08}
1150: successfully reproduce the CXB emission up to 4\,MeV.
1151: On the other hand, blazars account only for $\leq25$\,\%
1152: of the GeV diffuse background and most likely other source
1153: classes contribute to the diffuse emission \citep{dermer07}.
1154:
1155:
1156:
1157:
1158: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 14
1159: \begin{figure}[ht!]
1160: \begin{center}
1161: \includegraphics[scale=0.89,angle=0.]{f14.eps}
1162: \end{center}
1163: \caption{BAT CXB spectrum compared with previous results. The dashed line
1164: is the best fit to 2\,keV$<$ E $<$ 2000\,keV as reported in the text.
1165: \label{fig:xall}}
1166: \end{figure}
1167:
1168:
1169:
1170: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1171: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1172: \subsection{The Earth albedo Spectrum}\label{sec:albedospec}
1173:
1174: Our Earth albedo spectrum is not compatible with the classical high-energy
1175: photon index of 1.4. The BAT data are
1176: consistent with a steeper high-energy photon index at
1177: 99.989 confidence level.
1178:
1179:
1180:
1181: Using the polar-orbiting satellite {\it 1972-076B},
1182: \cite{imhof76} found that above 40\,keV the photon spectrum is
1183: consistent with a power-law with an index ranging from 1.34 to 1.4,
1184: depending on the latitude range scanned. Their measurement is based
1185: on the difference between pointings towards the atmosphere ({\it Down}) and
1186: pointing towards the sky ({\it Up}). In order to derive the albedo spectrum,
1187: the authors sum the CXB emission and the {\it Down-Up} spectrum
1188: \citep[see Eq.~5 in][for details]{imhof76}.
1189: For the CXB emission, they adopt the measurement of \cite{pal73},
1190: which describes the CXB photon spectrum as $dN/dE = 25E^{-2.1}$.
1191: This representation differs from the HEAO-1 CXB spectrum
1192: in both normalization and photon index
1193: in the 40--200\,keV range.
1194: Thus, we adjusted the \cite{imhof76} albedo spectra,
1195: taking into account the
1196: differences between the \cite{gruber99} and the \cite{pal73} CXB
1197: spectral representations.
1198: This is shown in Fig.~\ref{fig:imhof}.
1199: After the correction, the two (equatorial and polar) Albedo spectra
1200: are consistent with a power-law with photon index $\sim$1.7.
1201: In particular the equatorial measurement is in good agreement
1202: with the BAT spectrum.
1203:
1204:
1205: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 15
1206: \begin{figure}[ht!]
1207: \begin{center}
1208: \includegraphics[scale=0.89,angle=0.]{f15.eps}
1209: \end{center}
1210: \caption{Measurements of the Earth albedo of \cite{imhof76}
1211: at different geomagnetic latitudes. The original measurements
1212: (gray) have been corrected for un-subtracted CXB emission (see the text
1213: for details). After the correction, the albedo spectra (black data points)
1214: become steeper.
1215: \label{fig:imhof}}
1216: \end{figure}
1217:
1218: The {\it BeppoSAX} satellite operated in a Low-Earth Orbit
1219: similar to {\it Swift}, but with different
1220: inclination (4$^{\circ}$). The BAT Earth albedo spectrum is compatible
1221: (within the large uncertainties of the {\it BeppoSAX} analysis)
1222: with the measurements obtained by \cite{frontera07}.
1223: It is worth noting that we derived
1224: the Earth intensity using different orbital positions
1225: (as done for {\it BeppoSAX} and {\it OSO-3}), and thus averaging
1226: over the magnetic latitude sampled by {\em Swift}.
1227: Moreover, as generally the Earth enters the FOV
1228: at large angles, we do not observe the upward albedo, but rather the albedo
1229: emerging at large zenith angles. This also seems confirmed
1230: by the similarity of our spectrum with the {\it downward } gamma-ray
1231: flux measured for a balloon over Palestine, Texas \citep{gehrels85}.
1232:
1233: Fig.~\ref{fig:ea_spec} reports also the prediction of the Earth albedo
1234: emission as observed from the orbit of the {\it INTEGRAL} satellite
1235: \citep{sazonov07b}.
1236: Its normalization has been derived during the measurement
1237: of the CXB intensity \citep{churazov07}. It is evident
1238: that this prediction and the BAT measurement agree well in shape but not
1239: so in normalization. Among many factors, the overall normalization depends
1240: strongly on the geomagnetic latitude and the distance to the Earth.
1241: The agreement of the BAT and {\it INTEGRAL} albedo spectra, respectively, with the
1242: equatorial and polar measurements of \cite{imhof76} seem to confirm this
1243: interpretation.
1244: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --------- Fig 16
1245: \begin{figure}[ht!]
1246: \begin{center}
1247: \includegraphics[scale=0.90,angle=0]{f16.eps}
1248: \end{center}
1249: \caption{BAT Earth spectrum as compared
1250: to past measurements. References are shown in the legend.
1251: The datapoints (triangles) from \cite{gehrels85} are a fit to the
1252: {\it downward} gamma-ray flux at 5\,g cm$^{-2}$ over Palestine, Texas.
1253: The measurements from \cite{imhof76} were corrected to take into account
1254: the correct CXB emission (details are in the text).
1255: The thin solid line shows the prediction of the Earth emission as
1256: observed from the orbit of the {\it INTEGRAL} satellite \citep{sazonov07b,churazov07}.
1257: \label{fig:ea_spec}}
1258: \end{figure}
1259:
1260:
1261: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1262: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1263: \section{Discussion}
1264: We have used Earth occultation episodes to derive with {\it Swift}/BAT
1265: an accurate measurement of the CXB emission in the 15--200\,keV energy
1266: range. Moreover, we have proven by means of an independent technique
1267: the accuracy of the occultation analysis and of our results.
1268: The observed BAT intensity near the peak of the CXB spectrum at 30\,keV
1269: is 46.2\,keV$^2$ cm$^{-2}$ s$^{-1}$ sr$^{-1}$ keV$^{-1}$ and
1270: its uncertainty\footnotemark{}
1271: \footnotetext{Derived combining both measurements of the CXB.}
1272: is $\sim$3\,\% (including all systematics). The normalization of the BAT
1273: CXB spectrum at 30\,keV is $\sim8$\,\% larger than the HEAO-1 \citep{gruber99}
1274: measurement and consistent with the {\it INTEGRAL} one \citep{churazov07}.
1275: Moreover, considering that the precision of the HEAO-1 measurement at the
1276: CXB peak is 10\,\% \citep{marshall80} and that {\it BeppoSAX} data
1277: are compatible with a larger (up to 20\,\%) normalization of the CXB
1278: spectrum shows that all measurements above 10\,keV are consistent
1279: within their systematic uncertainties.
1280: %%%%%
1281:
1282: Such consistency is not observed
1283: at lower energies \citep[e.g. see discussion in ][]{revnivtsev05}.
1284: The origin of this inconsistency
1285: is unclear. However, it seems that neither cosmic variance \citep{barcons00}
1286: nor differences in the flux scale calibration of each individual
1287: instrument \citep{revnivtsev05,frontera07} may account for it.
1288: A likely reason for the discrepancy of CXB measurement in the 2--10\,keV
1289: band might reside in
1290: a systematic error in the response function used for diffuse
1291: sources \citep{frontera07}.
1292: To our knowledge, BAT is the only
1293: instrument (beside HEAO-1 A2 which was designed with the purpose
1294: of measuring the CXB) which makes use of a dedicated instrumental
1295: response developed for this particular analysis.
1296: We also note that a recent measurement of the CXB performed, in the 2--7\,keV,
1297: by {\it Swift}/XRT \citep{moretti08} seems to confirm
1298: the results of XMM-Newton \citep{deluca04}, RXTE \citep{revnivtsev05}
1299: and Chandra \citep{hickox06}. If confirmed, this means
1300: that the CXB spectrum, as most recently measured,
1301: is 25--40\,\% larger (with respect the measurement of
1302: HEAO-1) below 10\,keV while only $\sim$10\,\% larger above 20\,keV.
1303: The functional form we provide in $\S$~\ref{sec:cxbspec} for the broad-band
1304: CXB emission approximates well this scenario.
1305: %%%%
1306: %%%%
1307:
1308: A larger, than previously estimated, CXB emission would in turn
1309: require a larger density of Compton-thick AGN both in the local and in the
1310: more distant Universe.
1311: Recently, {\em Chandra} stacking analyses of
1312: mid-IR selected sources unveiled a large
1313: population of Compton-thick AGN at high redshift \cite[][]{daddi07,fiore08}.
1314: This large fraction of Compton-thick AGN found at
1315: z=$\sim2$ can be accommodated if the emitted (obscured) flux
1316: is very low. Indeed, lowering the assumed scattering efficiency
1317: (ratio of reflected to nuclear flux)
1318: would increase by the same amount the number of Compton-thick AGN at
1319: any redshift. In this framework, the recent discoveries of Compton-thick
1320: AGN with an extremely low scattering efficiency
1321: \citep[][]{ueda07,comastri07} fits well. These
1322: AGN are likely buried in a geometrically thick torus that obscures most
1323: of the nuclear flux. Although there can be many of these hidden AGN,
1324: their individual contribution to the CXB is necessarily small.
1325: This seems to be confirmed by the fact that the
1326: contribution of the mid-IR selected, z=$\sim2$, AGN
1327: is $<3$\% of the CXB intensity in the 10--30\,keV band \citep{daddi07}.
1328: Therefore,
1329: a larger contribution should be provided by Compton-thick AGN
1330: at lower redshift.
1331:
1332:
1333:
1334: %
1335: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1336: %
1337: %
1338: \section{Conclusions}
1339: BAT performed a very sensitive measurement of the CXB emission
1340: in the 15--200\,keV energy range.
1341: This measurement takes advantage of several episodes of CXB flux
1342: modulation due to Earth's passages through the BAT FOV.
1343: We find that the BAT CXB spectrum is in good agreement with the {\it INTEGRAL}
1344: one and that its normalization is $\sim8$\,\% larger than the HEAO-1
1345: measurement at 30\,keV.
1346: Additionally, performing an independent measurement of the CXB in the
1347: 15--50\,keV band, we are able to confirm this result.
1348: Remarkably, our
1349: study also shows that all the available measurements in the $>10$\,keV
1350: range agree within their systematic uncertainties.
1351: The new analyses of {\it COMPTEL} and {\it EGRET} data
1352: \citep{weidenspointner00,strong04} show that the formula
1353: suggested by \cite{gruber99} for the diffuse X- and gamma-ray backgrounds
1354: is only valid below 2\,MeV. We derived a simple functional form which, in
1355: the 2--2000\,keV range, approximate well (to a precision of 10\,\%)
1356: the CXB spectrum.
1357:
1358: Our study also derives the Earth albedo spectrum averaged over the magnetic
1359: latitudes sampled by {\em Swift}.
1360: The BAT spectrum is in agreement with all the previous observations performed
1361: by satellites operating in similar LEO orbits.
1362: This work shows that the Earth albedo spectrum declines at energies $>$40\,keV
1363: according to
1364: a power-law with photon index of $\sim$1.7, and not as 1.4 as previously
1365: thought. A re-analyis of the measurements performed by \cite{imhof76}
1366: is in perfect agreement with the BAT Earth albedo spectrum.
1367: The good agreement among the available measurements
1368: allows to use the BAT Earth albedo spectrum
1369: to predict the background contribution from the Earth for other instruments
1370: operating at similar orbits.
1371:
1372: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1373: \acknowledgments
1374: We are grateful to S. Barthelmy, J. Cummings and H. Krimm for
1375: all the effort spent in keeping the BAT perfectly operating.
1376: MA acknowledges the useful suggestions of R. Mushotzky and
1377: the help of C. Gordon for adapting Xspec to the purposes
1378: of this analysis. The anonymous referee is aknowledged for
1379: his/her helpful comments which improved the manuscript.
1380: MA is grateful to S. Sazonov and A. Zoglauer for interesting
1381: discussions about the Earth emission.
1382: This research has made use of data obtained from the
1383: High Energy Astrophysics Science Archive Research Center (HEASARC) provided
1384: by NASA's Goddard Space Flight Center.
1385: MA acknowledges funding from the DFG Leibniz-Prize to GH (HA 1850/28-1).
1386:
1387:
1388:
1389: \bibliographystyle{apj}
1390: \bibliography{/Users/marcoajello/Work/Papers/BiblioLib/biblio}
1391:
1392: \end{document}
1393:
1394: