1. Hard X-ray spectra of AGN observed with Suzaku
--- Time behaviour of AGN ---
I would like to report the time behaviour of hard X-rays from AGN observed with Suzaku.
Nagoya University
Hideyo Kunieda
T. Nakamura, Y. Haba
and Suzaku team

2. Suzaku mission High throughput X-ray telescopes 4 CCD Cameras
(3 FI and 1 BI)
The US-Japan collaborative X-ray mission Suzaku was launched in July last year.
Highthrouput X-ray telescopes are placed at the top of the space craft and CCD cameras are mounted at their foci.
A hard X-ray detector is also looking at the same direction.
These five detection systems are working well and producing many new results, which can be found in papers presented in this general assembly.
HXD
(PIN + GSO)
ISAS/JAXA
Launched in July 2005

3. Effective area of Suzaku detecthion systems
GSO
XIS
(XRT x CCD)
PIN
Effective area of Suzaku detection systems is summarized here.
Area of X-ray imaging spectrometer XIS is plotted in the left from 0.2 to 10 keV. Two detectors of HXD have the area are shown by the curves for PIN and GSO.
With these detectors, the broadband coverage from 0.2 to 300 keV makes Suzaku the most favorable system for AGN study.
In this talk, I would like to present time behaviour of AGN by connecting XIS and HXD light curves, especially from PIN detector.
50keV
10keV

4. Suzaku HXD Sensitivity(BKGD limited)
This is the plot of sensitivity of various missions in orbit.
HXD is about three times better than previous missions.
In order to achieve this limit, we need to reproduce the BKGD as accurate as possible.

5. BKGD reproduction Night Earth data Fukazawa et al.
In this figure, black points are the PIN light curve when SC was looking at the night earth. Red points are the BKGD reproduced by the equation created by HXD after long effort. The residual shown in the lower panel is about 5%. Now we have high sensitivity data up to 50 keV.
Fukazawa et al.

6. Suzaku AGN observations
Spectral analysis
Broad band spectra ( keV)
Determination continuum components
Emission line and edge structure
and then broad iron line profile
For AGN analysis, Suzaku provides us with broad band spectra with better accuracy and allow us to determine accurate continuum spectra and then we can get much robust results of emission lines and edge structure, especially for the broad iron line

7. Suzaku AGN observations
Spectral analysis
Broad band spectra ( keV)
Determination continuum components
Emission line and edge structure
and then broad iron line profile
For the timing analysis, difference spectra between high and low states is made to examine the spectra of variable components. In my talk, I will show some approaches of timing analysis to distinguish various components proposed based on static analysis.
Timing analysis
Difference spectrum = variable component
“Difference of time behaviour
of various components”

8. Mkn 3
For the first target, we picked up Mkn 3, which is known with small variability in shorter time scale than weeks.

9. PIN Light Curves of Mkn 3 1 day T. N.
Let’s take a look of PIN light curves. Top panel is the raw counting rate with time binning of 1000 sec.
The second is the BKGD thus estimated, which seems 2/3 of raw count.
The bottom is the net PIN light curve.
In order to confirm the stability of PIN light curve, we have compared the observed variance with statistical error of each data point.
T. N.

10. Time behavior of PIN Variance vs statistical error
In this plot green data points show the statistical error of each bin for different binning time scale from 256 sec to 100 ksec.
Red points are varinace of the measured light curves binned with the time on the horizonta axis.
σ2= Σ(xi - μ) 2 1 N
T. N.

11. Time behavior of PIN Variance vs statistical error
Of course, if there is no variability, all data points should come on the straight line of the square root T. Green points are there, and red points are also very close to. In this case, no extra variability is detected as we expected.
T. N.

12. MCG
The next example is MCG known with variability and enough counts in HXD band.

13. X-ray Spectrum of MCG-5-23-16
This is the energy spectrum of MCG-5 fitted with a single power law model. You can find a soft excess component below 1 keV and excess hump above 10 keV, may be due to the stable reflection component.
Reeves

14. X-ray Spectrum of MCG-5-23-16
XIS
PIN
Variable power law
Scattered
Reflection
This is a trial model spectrum with the variable power law component, stable cold reflection, and the stable scattered component.
We selected three energy bands to charcterize the time behavior of three components.
Haba

15. XIS Light Curves of MCG-5-23-16
Scattered + power law
Variable power law
Let’s take a look of XIS LC.
The first line is the LC of below 1 keV.
The second and third panels show LC of the main variable power law component.
T. N.

16. PIN Light Curves of MCG-5-23-16
How about the PIN light curve?
In this case the fraction of the BKGD shown in the second panel is about half of the raw counts shown above.
After the subtraction, the net PIN LC shown below is really varying.
Let’s compare the variance with statistical errors for different bin size.
Reflection + Power law
T. N.

17. Time behavior of PIN Variance vs statistical error
Now the measured variance shown with red points are much above the statistics from 1000 sec and longer.
T. N.

18. Time behavior Excess Variance of XIS & PIN
Excess variance (fractional)
σINT2= Σ{(xi - μ) 2 -σi2} 1
Nμ2
In order to measure the excess variability above the statistical error, we have calculated excess variance defined by this equation, according to Edelson, Nandra and Papadocuis. μ is the average counting rate. The first term is the measured deviations and the second term is the statistical errors.

19. Time behavior Excess Variance of XIS & PIN
This is the plot of excess variance of XIS and PIN light curves with different time binning.
From the left, the time bin is 256 sec. We have about 1000 data points during the 2.5 days observation.
Red points are from three XIS systems, while blue diamonds show that of PIN
Since the excess variance does not change up until 4000 sec, there is not much power of variation in higher frequency, I would say.
The decrease of the excess variance means the power of variation in these time scales of sec are being smeared out by the binning.
σINT2= Σ{(xi - μ) 2 -σi2} 1
Nμ2

20. Time behavior Excess Variance of XIS & PIN
The variance is related to the power density spectrum (PDS) as ν1
σ2INT(ν1) = ∫PDS(ν）dν ν0
PDS(ν1, ν2) = {σ2INT(ν2) - σ2INT(ν1)}／(ν2- ν1)
Therefore, PDS is given by measured excess variance.
The variance is related to the power density spectrum with this equation.
Then PDS of the frequency band between ν1 and ν2 should be given by difference of the variance of different bin size.

21. Time behavior Power Density Spectra of XIS & PIN
Smaller
scale?
Dillution?
The power density spectra thus obtained from the excess variance is shown for XIS with red and PIN with green. The major power of variability is in the range of 10 ksec or longer.
10 k light sec corresponds to the size of 10**15 cm.
if you assume the mass of BH of 10**8 Solar masses,
Power law component with such time scale could be created at about 10 Rs or outer.
On the other hand PIN hard X-ray light curve is diluted by the stable reflection component for example. It pushes the PDS curve downward.

22. Time behavior Flux correlation of XIS & PIN
Another simple approach to the time behaviour of different energy band is the correlation plot of counts in two energy bands. Here the PIN counting rate is plotted against the CCD counts in 2-5 keV.
T. N.

23. X-ray Spectrum of MCG-5-23-16
XIS
PIN
Reflection + Power law
Variable power law
If you remember the spectral model, PIN band flux is evenly dominated by the varaible power law and the stable reflecction components, while the 2-5 keV band is essentially variable power law component.
Haba

24. Time behavior Flux correlation of XIS & PIN
If the reflection component is one half of the observed PIN flux, data points should come to the black line.
T. N.

25. X-ray Spectrum of MCG-5-23-16
Variable power law
Reflection + Power law
If the stable component is less as is suggested by Reeves with blue dashed line,
Reeves et al.

26. Time behavior Flux correlation of XIS & PIN
Data points may align on the blue line.
Unfortunately, the amplitude of the variation of MCG-5 is not wide enough to distinguish two models. We are going to apply the same method to other sources with larger flux variation.
T. N.

27. Samples AGN Spectra Watanabe et al
These are the counting rate plots of various AGN observed with Suzaku so far. We have at least 5 Sy I and one Sy II in our hand with the same or higher PIN flux level than Mkn 3. May be more soon.
Two timing analysis methods I showed here may not be new and very primitive way. But different time behavior of different energy band examined by these methods should be explained by the correct spectral models.
The detail study of AGN in PIN energy band is also important to examine the contribution of AGN to the CXB as Guenther mentioned.
One year after the launch, Suzaku can contribute to explore the origin of the AGN radiation from detail analysis of broad band spectra with much better sensitivity now available after the establishment of HXD BKGD.

28. Summary Mkn 3 : No detection of variability
in both XIS and PIN light curves
MCG
Excess variance is clearly found.
σ(XIS) > σ(PIN) at most frequencies
PDS can be derived
Correlation may constrain
the combination of multiple components
Apply the same method to other sources.
Error estimation is necessary to establish this approach