Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea.iki.rssi.ru/integral06/papers/B33_Schanne02-12Jan07_15:38:00_color.pdf Äàòà èçìåíåíèÿ: Fri Jan 19 20:22:08 2007 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 23:30:41 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: vela satellite

EVIDENCE FOR 1809 keV GAMMA-RAY EMISSION FROM 26Al DECAYS IN THE VELA REGION WITH INTEGRAL/SPI
StÈphane Schanne(1,*), Patrick Sizun(1), David Maurin(1), Bertrand Cordier(1), Andreas von Kienlin(2), Clarisse Hamadache
(1) (2) (1)

CEA-Saclay, DAPNIA/Service d'Astrophysique, F-91191 Gif sur Yvette, France Max-Planck-Institut fÝr extraterrestrische Physik, D-85740 Garching, Germany (*) corresponding author: schanne{at}hep.saclay.cea.fr

ABSTRACT The Vela region is a promising target for the detection of 1.8 MeV -rays emitted by the decays of radioactive 26 Al isotopes produced in hydrostatic or explosive stellar nucleosynthesis processes. COMPTEL has claimed 1.8 MeV -ray detection from Vela at a 3 level with a flux of 3.6â10-5 ph/cm2/s. In this paper, we present first results of our search for 1.8 MeV -rays from Vela with the spectrometer SPI aboard INTEGRAL. Using the data set acquired during 1.7 Ms at the end of 2005 in the frame of our AO-3 open-time observation, we determine a flux of (6.5 ± 1.9(stat) ± 2.4(syst))â10-5 ph/cm2/s from 26Al decays in the Vela region. 1. VELA REGION NUCLEOSYNTHESIS

in the 1.8 MeV domain) with good sensitivity thanks to its Ge-detectors. 2. NEW INTEGRAL VELA OBSERVATIONS

We use for our analysis the SPI data acquired in November-December 2005 in the frame of our accepted AO-3 open-time observation. Data were taken from INTEGRAL revolution 373 to 383, in 473 different satellite pointings, corresponding to 1.7 Ms of on-time. At variance with the previous deep Vela observations [3], during the AO-3 Vela data taking no strong solar flare polluted the data set (see Fig. 1). The SPI cryostat remained at a very stable temperature, without any cryocooler activity, hence the standard SPI energy calibration per revolution is sufficient for a line analysis with a few keV resolution at 1.8 MeV.

The Vela region is a key target for detecting gammarays emitted by the decays of radioactive isotopes produced in hydrostatic and explosive stellar nucleosynthesis processes, such as 26Al (gamma-ray energy: 1809 keV), 60Fe (1173 and 1332 keV) and the associated positrons, detectable through e+-eannihilation (511 keV). 26Al nuclei in particular, with a half-life of 0.74 Myr, can escape their production environment, before their decay to 26Mg associated with the emission of observable 1809 keV -rays. The importance of the Vela region as a nucleosynthesis laboratory is underlined by the presence of potential nucleosynthesis sites (young Supernova Remnants (SNR), Wolf-Rayet (WR) stars) at distances as close as a few hundred parsec. Candidates for 26Al production in the Vela region are for instance: the Vela SNR (core collapse SN, age~104 yr, distance~250 pc) and the historical SNR Vela Jr (680 yr, ~250 pc), the WR star 2-Vel (~260 pc) and distant galactic OB associations (few kpc). COMPTEL observations have revealed 1.8 MeV -ray emission from the Vela region [1] at a 3 level, with a flux of (3.6±1.2)â10-5 ph/cm2/s, yet the source of the emission has so far not been identified. In this paper, we present results of our search for 26 Al gamma-ray emission in the Vela region with the spectrometer SPI aboard INTEGRAL [2], capable of resolving high-energy -ray lines (and particularly well

ACS

IREM e

T cryo

Fig. 1. The count-rate profiles of the shield (ACS) and radiation monitor (IR data taking are smooth, and show entry/exits. Without cryomachine activit the Ge cryostat increase only slowly.

SPI anti-coincidence EM electrons) during only radiation belt y, the temperatures of

3.

METHOD OF DATA ANALYSIS

We adopt a light-bucket approach, in which the counts of all active 17 SPI detectors are added together, ignoring the mask imaging properties. Background tracers, i.e. available dead-time-corrected count rates (such as saturating Ge rates, PSAC, IREM, and line

_____________________________________________________________________ Proc. 6th INTEGRAL Workshop `The Obscured Universe', Moscow, 3-7 July 2006 (ESA SP-622)


dead-time corrected count rate [c/s]

rates), are used to build background models. Those are fit to an OFF dataset from which no 26Al signal is expected.. The background model is then applied to the ON database, using the tracers from the ON dataset, to predict and subtract the background present in the 1.8 MeV region. For the OFF database we use all high-latitude (|B|>20°) SPI data since July 2004, from the moment on where only 17 working SPI detectors were left. For the ON database we use all public data since July 2004 and our private Vela AO-3 data. All data are cleaned in order to remove pointings of bad quality (containing radiation belt passages, or covering solar flares), following a procedure described in [4]. In particular cuts are applied to the cryogenic temperatures, PSAC, IREM proton (11-30 MeV) and electron (>.55 MeV) count-rates, before removal of spikes in the 12C, 16O, D lines and ACS count rates (using a boxcar algorithm). After cleaning, the OFF database contains 1697 pointings and the ON database 3560. For each pointing we extract a list of 259 relevant numbers (mostly deadtime corrected count rates, such as Ge saturating counts or ACS counts, or from selected regions in the Ge spectra, including e.g. background lines). 4. REFERENCE 26Al ANALYSIS

eigenvalues sorted from the most to the least significant. The new orthogonal set of base vectors forms the "principal components" (Fig. 3).
eigenvalues

t ra ce rs

components

ts onen comp

Fig. 3. Eigenvalue of each principal component and matrix of the "principal components" as linear combination of the original (centered/normalized) tracers.
OFF bkg model fit (yellow): reduced 2=0.995, degrees of freedom 1691
0.5 0.4 0.3 0.2 0.1 0 1700 1800 1900 2000 data model const pca0 pca1 pca2 pca3 pca4 2100 days since 1/1/2000 data model const pca0 pca1 pca2 pca3 pca4 1700 1800 1900 2000 2100 days since 1/1/2000

ON predicted bkg (green)
0.5 0.4 0.3 0.2 0.1 0

The spectrum of the SPI background in the 1.8 MeV region is shown in Fig. 2. The 26Al signal is searched in the single event spectra in a 10 keV wide bin centered at 1809 keV (which implies a narrow-line hypothesis, see Fig. 2). For the following reference 26Al analysis, we define also spectral regions (labeled tracers 3,4,5,6) covering the 205Bi and 28Al peaks (well correlating to the background in the signal region), and control bands (lo,hi) around the 26Al signal region. The background in the signal region is a combination of a continuum background (which can be deduced from the control bands) and a background peak from excited 26Na and 56 Mn. We define also as tracers 1,2,7 the Ge-saturating counter, the 69mZn spectral band, and the PSAC.
SPI Single Events spectrum
26Al

Fig. 4. OFF/ON (top/bottom) count rates in the signal region vs time, superimposed with a linear background model fitted on the OFF data, and applied to the ON data. Each component used is shown with its weight ci.
OFF residuals frequency

-0.06

-0.04

-0.02

0

0.02

0.04

0.06 [c/s]

205Bi

frequency

28Al

26Al signal region

sig region: 1803-1815 keV => hypothesis: narrow signal line around 1809 keV 26Al bkg tracer count rates: 1 - GeSat 2 - 69mZn 436-441 keV 3 - 205Bi 1760-1768 28Al 41774-1783 5 - LoBand 1790-1802 6 - HiBand 1815-1827 7 - PSAC

ON residuals

-0.06

-0.04

-0.02

0

0.02

0.04

0.06 [c/s]

Fig. 5. Histogram of the (dead-time corrected) residual count rate in the signal region after background subtraction. While the OFF histogram (top) is centered, the ON histogram (bottom) is shifted towards positive values, with a significantly non zero mean value, showing the detection of an overall Galactic plane 26Al emission.

34

5

6

Fig 2. SPI background in the 26Al signal region.

All 7 tracer vectors (for the OFF and ON together) are then centered (in order to have zero mean values) and normalized, their covariance matrix diagonalized, their

Using either all tracers or all components for the background model is equivalent. However by using only the first components it is possible to remove the noise accumulated in the components with the smallest eigenvalues. In the following we will apply the socalled Kaiser-cut, and use only the 5 first components


for background modeling (note that in some cases even the 1st component alone gives already good results). A linear background model is built using the OFF data pointings p by fitting the coefficients ci to reproduce the (background) count rate Bp in the signal region using the principal components Ti,p as tracers: B p = c0 + ci Ti , p
i =1.. Ncomponent s

This formula (with the ci found) is then used to predict, in the ON data, the background rate Bp under the measured rate Rp with the ON data tracers Ti,p (shown in Fig. 4). The signal rate is then simply the residual Sp=Rp-Bp (Fig. 5 and Fig. 6).
OFF (|B|>20° residual rate [c/s] vs L )
0.015 0.010 0.005 0 -0.005 -0.010 -0.015

point source on axis is 32.8 cm2 at 1.8 MeV. Assuming a uniform diffuse source with an extent of -30° to 30° in latitude, a reduction by 0.47 applies (SPI off axis efficiency integrated over this region). Hence we deduce a flux of (2.4±0.35(stat))â10-4 ph/cm2/s originating from 26Al in the Galactic central radian, in agreement with the COMPTEL value of (2.8±0.15)â10-4 ph/cm2/s [5]. The Vela 1809 keV rate detected is 2.7±0.6(stat) counts/ks from SPI single events, which translates to a flux of (8±2(stat))â10-5 ph/cm2/s, in the case of a point-like source (extent typically < 10°), only 2 above the COMPTEL value. 5. SYSTEMATIC
26

Al STUDIES

210°
0.015 0.010 0.005 0 -0.005 -0.010 -0.015

260°

310°

0°

50°

100°

150° L

ON (|B|<20° residual rate [c/s] vs L )

Vela 26Al is detected at ~3
210° 260° 310°

The inner central radian 26Al emission is detected at 7.6
0° 50° 100° 150° L

ON Residual SIGMA IMAGE (B vs L)
B 20° 10° 0° -10° -20° -30°
3 2 1 0 -1 -2

210°
0.8 0.6 0.4 0.2 0

260°

310°

0°

50°

100°

150° L

EXPOSURE TIME [Ms] vs L ON OFF

In order to assess its systematic error, the previous analysis is repeated for all combinations of the following parameter variations: (i) three different separations between ON/OFF data sets are chosen (|B|<10, 20, 30°); (ii) three background model types are built (LNP=linear, with original tracers used for the background model; LP5=linear using the 5 first principal components as tracers, like presented in §4; BP5=Bayesian model, using the 5 first principal components ­ it replaces the linear model by the probabilistic neural network PNNV code given in [6], as described in [4], and illustrated in Fig. 7.); (iii) five different combinations of tracers are used (the 7 tracers presented in §4, a list of 69 tracers from rate meters and single event lines, a list of 52 tracers reduced to strong lines, a list of 79 tracers with single and double event lines, and a list of 22 tracers containing only rate meters and no spectral tracers). In this systematic test, all presented OFF background models have good fits: their 2 values range from 0.9 for Bayesian models (an effect known as neural network "over-learning" from statistical data fluctuations) to 1.04 for the "worst" linear model. Therefore the 2 value can not reasonably be used to discriminate the best models.
L IN E A R B A C K G R O U N D M O D E L (7 tr)
dead-time corrected OFF data count rate [c/s]
0 .4 8 0 .4 6 0 .4 4 0 .4 2 0 .4 0 0 .3 8 0 .3 6 0 .3 4 -3 0 .4 8 0 .4 6 0 .4 4 0 .4 2 0 .4 0 0 .3 8 0 .3 6 0 .3 4 -3 -2 -1 0 1 2 pca0 -2 -1 0 1 2 pca0

210°

260°

310°

0°

50°

100°

150° L

d a ta

Fig. 6. As a function of the Galactic longitude (L in °), are shown the OFF residual rate (fluctuating around 0), the ON residual rate (non zero indicates 26Al flux), the -significance map (latitude vs longitude) of the ON residuals, and the exposure (mostly Vela region and Galactic central radian).

m odel

B A Y E S IA N B A C K G R O U N D M O D E L (2 2 tr)

The background model fit on the OFF data is quite good, with a reduced 2 of 0.995 for 1651 degrees of freedom. The results (i.e. the residual 26Al -ray fluxes) are presented in Fig. 6 as a function of the Galactic longitude. 26Al emission is significantly detected in the Galactic central radian (at a 7.6 level with the dataset used), and in the Vela region (with 4.4 ). The Galactic central radian 1809 keV -ray rate detected in this analysis is 3.8±0.5(stat) counts/ks from SPI single events, for which the efficient area for a

Fig. 7. Normalized OFF data rate (blue) and model (yellow) vs first principal component rate for a linear and Bayesian model (which obviously allows for non linearity).

The results of all combinations of systematic tests are shown in Fig. 8, presenting for each background model the 1809 keV flux obtained, which is computed using


26

A l F l u x ( ph /c m 2 /s )
2. E - 04

O N |B |< 1 0 °
LP 5 BP 5 LN P

O N |B |< 2 0 °
LP5 B P5 LN P

O N |B |< 3 0 °
LP 5 BP5

LNP

2. E - 04

1. E - 04

5. E - 05

0.E + 00

- 5. E - 05
2 .E -0 4 2 .E -0 4 1 .E -0 4 5 .E -0 5 0 .E + 0 0 -5 .E -0 5 -1 .E -0 4 -2 .E -0 4 -2 .E -0 4 1

V e l a s i g n a l s y s te m a ti c s
1 3 5 7 9 11 13 15 17 19 21 23

7 , 6 9 ,5 2 , 7 9 , 2 2 , m e a n
25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67

C r a b s y s t e m a ti c s ( c r o s s - c h e c k )

3

5

7

9

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47

49

51

53

55

57

59

61

63

65

67

Fig. 8. The 1809 keV -ray flux detected in Vela (top) and Crab (bottom) versus a sequential trial number, ordered in 3 groups (of ON regions |B|<10, 20, 30°) with 3 types (of background models LNP, LP5, BP5) using 5 different tracer lists (with 7, 69, 52, 79, 22 tracers). The error bars shown are 1 -statistical. Each "-" marks the mean flux value for each group of 5 tracer lists.

the measured rate and a SPI single event efficient area of 32.8 cm2, i.e. the FOV central value is used. This is a correct assumption for a source with an extension<10° FWHM (FOV FWHM~24°). For an ON with |B|<30°, too few OFF data are left over to get a stable background model, hence the larger spread. For an ON with |B|<10°, there might be still signal in the OFF data (FOV HWHM~12°). Therefore we consider the ON with |B|<20° the best compromise. It gives indeed a lower systematic error. Notice also that in the case of the Crab, no signal is found and systematic errors are similar. The Bayesian background models show slightly lower systematic errors. From this analysis, we obtain a detection of 26Al in the Vela region at 3.4 (statistical) with a flux, computed as the mean value for ON with |B|<20°, of 6.5 ± 1.9(stat) in units of 10-5 ph/cm2/s, the systematic error in the same units being ±2.4 (standard deviation of all models) or +4.9/-3.2 (their min-max values). 6. CONCLUSIONS

Our 26Al gamma-ray flux value from the Vela region is compatible with the one from COMPTEL. With our light-bucket approach we still can not disentangle the 26Al source, and an improved method, capable of using the mask imaging properties of SPI will have to be applied. From the fact that we have an evidence for 26Al gamma-ray emission at the same level as COMPTEL, but in a 10 keV narrow energy bin, we can speculate that the 1.8 MeV line in Vela could be a narrow line. Further analysis is needed in order to measure the 1809 keV line width. If it turns out to be narrow, as the line observed from the Galactic central radian [7], this could then be a hint that the Vela region 26Al source might not be one of the recent and nearby supernova remnants present in the region, but rather be part of the overall Galactic 26Al emission. The localization of this source can be determined, with more statistics, using model fitting methods or SPIROS associated with a precise background model. 7. REFERENCES

From the analysis of 1.7 Ms of SPI data, corresponding to the first part of our Vela deep observation in the frame of INTEGRAL AO-3, taken during NovemberDecember 2005 (1.3 Ms more expected in 2006), we detect a 1809 keV -ray flux of (6.5±1.9(stat) ±2.4(syst))â10-5 ph/cm2/s originating from 26Al decays in the Vela region. We considered all SPI pointings falling inside a L=10° and B=40° wide region centered at L=265° and B=0°, using SPI single events only, and assuming an efficient area of 32.8 cm2 at 1.8 MeV (corresponding to the value on the SPI optical axis). This assumption is correct, only if the emission region is small in size compared to the SPI field of view. The systematic error is still large and the method has still to be carefully evaluated.

1. Diehl, R., et al., 1995, A&A 298, L25, 1.809 MeV gammarays from the VELA region. 2. Schanne, S., 2006, IOP Journ. Phys. 41, 46, and references herein, Observing with a space-borne gamma-ray telescope: selected results from INTEGRAL 3. Maurin D., et al., 2004, ESA SP-552, 107, Status of the Search for 1809 keV -rays in Vela with INTEGRAL/SPI 4. Sizun P., et al., 2007 (to be submitted) Background model studies for INTEGRAL/SPI 5. Diehl, R., et al., 1995, A&A 298, 445, COMPTEL

observations of Galactic 26Al emission
6 n 7 a . Masters, T. 1993, Academic Press, textbook: Practical eural recipes in C++ 26 . Diehl, R., et al., 2006, Nature 439, 45, Radioactive Al nd massive stars in the Galaxy