1. XIS Calibration Report 2006 Feb 22 (revised 2006 Feb23)
K. Hayashida (Osaka Univ.) and the SUZAKU XIS team

2. Menu Minimum Notes on XIS data processing Gain/CTI QE degradation
Charge Trail Correction
CTI Correction /Charge Injection
Ex-PHA Relation / Response
QE degradation
RXJ1856,E0102,Cyg Loop, Earth Atmosphere
Modeling
Background Study and Data Base
COR, PINUD dependence
Data Base and Subtraction method
Data Selection Criteria
Bad Columns / Hot Pixels

3. XIS Data Reduction Frame Data /8sec Event data XIS cleaned event list
PHAS(6)
PHAS(7)
XIS Response depends on the reduction procedure
PHAS(0)
Frame Data /8sec
Dark-level Subtraction
Event Pickup (PHAS(0)>Event Threshold)
5x5 mode, 3x3 mode or 2x2 mode
Event data
Charge Trail Correction
CTI correction
Grading / PHA-reproduction for PHAS(i)>Split Threshold
PHA-dependent Split Threshold for BI
EHK screening
Bad Columns Filter
xisclean
XIS cleaned event list
Onboard DE
On the ground

4. Darklevel Estimation and Subtraction
Dark-level of each Pixel “Dark” is determined onboard after every SAA passage in DarkInit / DarkUpdate DE-mode.
DE also calculate the Darklevel shift with time for each frame (8sec) for each (e.g. 64x64pixels) partition. We call it “LightLeak” value.
One LightLeak value for one partition
Pixel-Level “PHAS[]” = Raw-Pixel-Level – “Dark” – “LightLeak”(of the Previous Frame)
“LightLeak ” value shifts slowly in time.
Only when bright Earth come into the FOV, “LightLeak ” increases dramatically. It makes error in pixel levels, and thus in PHA and in event number.
*) D LLL(=increment of the LL from previous frame to the current frame) can be an indicator of such effect.
*) “LightLeak written in the XIS data products is in 1/8 ADU unit.

5. Event Grades Grades 02346 are used as X-ray events. grade0 grade1
Pixel level is maximum among 3x3 area and larger than Event threshold
Grades are used as X-ray events.
Pixel level is larger than Split threshold and added to the PHA
Pixel level is larger than Split threshold but NOT added to the PHA
grade0
grade1
grade2
grade3
grade4
grade5
grade6
grade7

6. Charge Trailing ACTY Some charge is deposited in trailing pixels
Grade0
Grade2
events
ACTY ０
１０００
５００
２００
Vertical
Some charge is deposited in trailing pixels
transfer
Trailing charge
Grade0
Grade2 PH

7. Charge Trail ＰＨ（２）= preceding pixel ，ＰＨ（７）=trailing pixel
BI1 5.9keV X-ray incidence
PH [ADU]
RAWY a b
Near readout node d c b
Far from readout node a c d

8. PHAS(2),PHAS(7) Distribution before/after Charge Trail Correction
PHAS [ADU]
PH [ADU]

9. Charge Trail Correction
Grade02346 event distribution
３００　　　　　　　　　　　　０
３００　　　　　　　　　　　　０
couns
counts
not uniform!
After Correction
Berore Correction ０
５００
１０００ ０
５００
１０００
ACTY
ACTY
Save grade7 events
Correction saves 10—20% of event at high energy

10. Calibration Source 55Fe Doors cannot be closed again Door Close
Door Open
Doors cannot be closed again

12. Q’ = Q(1-CTI)N → CTI = (Q- Q’) / Q / N
Determination CTI parameters included in rev0.6 data (H. Yamaguchi)
Q’ = Q(1-CTI)N
→ CTI = (Q- Q’) / Q / N
Q = PHA(ACTY=0)
Q : Initial charge
Q’ : Readout charge
N : Number of P transfer
PHA(Y=0) – PHA (Y=896, T)
PHA(Y=0)×896
Whole Area cal src data (8/11)
CTI =
CTI(Seg1) = CTI(Seg 2)
= [CTI(Seg0) + CTI(Seg3)] / 2
Corner cal src data
(8/15～11/20)
CTI = CTI_CONST
+ CTI_NORM×(PHAS)CTI_POW
In rev0.6…
CTI_POW = -0.5
CTI_CONST = 0 T
Q’(T) = PHA(ACTY=896, T)

13. CTI correction in rev0.6 Determine CTI_NORM for each quadrant
Column by Column difference is not taken into account.
Fit CTI_NORM as a function of time with a straight line.
Make a CALDB table which has columns, time, CTI_NORM, CTI_CNST etc. One row per a week.
Fill (modeled) CTI_NORM.
Correction is done with a critical ftool “xispi” to PHAS[], after Charge trail correction but in prior to event grading.
Corrected PHAS[] is not written in XIS event file.
In principle, gain shift by CTI is corrected in xispi (data reduction) not in response.
Line broadening is inevitable even with CTI correction. (Charge Injection data helps the broadening partially). The effect must be taken into account the response but not yet.
Response builder for that has almost prepared.

14. Results of CTI correction
Mn-Kα
5.84keV (rev0.3)
→ 5.88keV (rev0.6)
Cal src (rev0.6)
by Nakajima
@2006/01
OVIII line
0.654keV
CTI parameters should be improved.
Cygnus Loop (rev0.6)
by Katsuda
(CI data were obtained on 2006/01.)

15. Residual error in CTI Correction of rev0.6
Simplified Assumptions:
CTI (ACTY=0,t)=CTI(ACTY=0,t0=2005Aug)
There must be degradation in CTI Imaging to Frame Store Area Transfer
CTI is proportional to Ex-0.5
Need to be checked its accuracy with orbital data.
Mn-Ka peak ch history was made with G02346
CTI correction is for each pixel PHAS[], G0 data should be employed.
We need to check the energy scale in rev0.6 data with various t, Ex, ACTY, (ACTX)
Feedback from SWG members will be acknowledged. Expected Energy is sometimes uncertain. Need detailed knowledge for each source, too.
(for XIS team) Ni-K line, Al-K etc in BGD might be useful, though its not certain they are imaging area event.
NOTE: If Charge-Trail correction parameters mismatch, not only energy scale but also (effective) QE is affected.

16. Revision of Ex-PHA Relation: XIS2 Segment 0
Old: Single linear function fit: 2 parameter model (caldb=ae_detgain_ ext)
New: 2 linear functions for E<Esik and E>Esik: 4 parameter model (caldb=ae_xi2_makepi_ fits)
In this case, the systematic offset of ~4ch (~15eV) in the soft X-ray region is largely reduced.
Si K edge (E=1839 eV)

17. PKS2155 (rev0.3) residuals
Structure around Si-Kedge is mainly due to broken line approximation of the Ex-PHA relation.
Should update Ex-PHA relation (PHA-PI conversion) in xispi

18. RXJ1856.5-3754 Discovered with ROSAT
Nearby (D~120pc) Isolated Neutron Star
X-ray spectrum is fitted with a simple blackbody ( against NS atmosphere model).
R~4-5km Quark Star ?

20. Suzaku Obs 2005-10-24~10-26 RMF 20051210 a-d for XIS1
Rev0.3 data -10eV offset
Suzaku Obs ~10-26 RMF a-d for XIS1
a: Based Cal on the Ground
b: a x excess0.15mmC
c: Dead Layer =Design Value
d: c x excess0.15mmC

21. Calibration Task Share
Components
Location
X-ray Source
QE reference
Chip level
CSR/MIT
Fluorescent X-rays (C,O,F,Al,Si,P,Ti,Mn,Cu)
ACIS chips calibrated at BESSY
Camera without OBF
+FM AE
Osaka
Grating Spectrometer
keV
Polypro-window Gas PC & XIS-EU
Kyoto
Fluorescent X-rays (Al,Cl,Ti,Mn,Fe,Zn,Se)
Window-less SSD
OBF
Synchrotron Facility
Synchrotron X-rays + monochrometer
(Transmission measurement with PIN diode)
Camera onboard the satellite
ISAS/JAXA
55Fe

22. QE(PC)->QE(EU)->XIS BI1 QE
-1deg offset slant-PC is assumed
Best Fit Estimates
HfO2
0.005mm fixed Ag
0.001mm fixed
SiO2
0.000±0.0005mm
Si depletion
43.6±0.7mm
Constant Factor
0.961±0.003
Some data points exceeds 1
Difficult to reproduce 0.28keV QE with simple absorption models.
At low energy side, artificial cutoff in the form of erf(Ex) is introduced.
This model QE is used as a new reference.

23. Rev0.6 data (with no gain shift) Wabs x BBodyrad x Varabs
O/C <10%

26. Suzaku/XIS Contamination Measurements with E0102
E0102: SNR in SMC, bright in soft X-ray lines
excellent calibrator for low-E gain, QE changes
contamination degrading low-E eff. area of all XIS’s
model
thermal bremss + 24 Gaussian emission lines
Galactic + SMC absorption
pure C absorption from contaminant (varabs)
gain shift -5 eV ~ -15 ev
r2 ~ 1.6 (FIs) to 2.5 (BI)
OVIII
NeIX
NeX
OVII
MgXI

27. XIS Contamination Rate
empirical correction for observers
contamination rate turnover (?)
SMC NH uncertainty  systematic error ±0.02 m independent of epoch
change in effective C column:
chip slope intercept
(1016 cm-2/day)
XIS0 1.6 ± ±4.0
XIS1 2.7 ± ±15
XIS2 3.1 ± ±14
XIS3 4.1 ± ±50.

28. Count Rate History

29. Count Rate History

30. Count Rate History

31. Atmospheric N-K line Map XIS1(BI)
From Anabuki et al.’s poster
Color code is adjusted for each map
N-K line
Day Earth
0 < DYE_ELV < 5
5 < DYE_ELV < 10
10 < DYE_ELV < 15
15 < DYE_ELV < 20
20 < DYE_ELV < 25

32. Atmospheric O-K line Map XIS1(BI)
0 < DYE_ELV < 5
Day Earth
5 < DYE_ELV < 10
10 < DYE_ELV < 15
15 < DYE_ELV < 20
20 < DYE_ELV < 25

33. Day Earth Radial Profile (vignetting corrected,normalized by center region)
N-K line
O-K line
SN1006_NE_BGD
Mrk 3
N-K line
O-K line
A2811_offset
NGC 4388
MBM12_off Cloud

34. Center 6mm radius / Other area
N-K line
O-K line
Mean Free Path in C(2.2g/cc)
0.182mm for N-K line
0.375mm for O-K line
Spatial Difference in Carbon contamination thickness can be modeled with Atmospheric N-K, O-K data.
Thickness at the center is evaluated by E0102 and RXJ1856 obs.
Thickness (t,detx,dety) will be modeled/introduced in arfbuilder (or rmfbuilder).

35. N-K line
O-K line
[Central 6mm radius count rate] / [Outer area count rate]
c.r.(center) - c.r.(outer)=0.9x10-3 mm/day
c.r.(center)~ 2.5x10-3 mm/day
c.r.(outer)=1.6x10-3 mm/day
Time(sec)

36. Bad (CTE) Columns Bad CTE
X-ray image (number of events /pixel)
Bad CTE
Typically long trail in each event.
Sometimes flickering pixel is observed.
Rows near the readout node can be used.
Identification logic without accumulating 10^7events was developed.
EU= 21 bad columns/chip
XIS0=14, XIS1=50, XIS2=17,XIS3=24
How should we do for adjacent columns ?

37. Calibration Status and Plan
QE degradation are going to be modeled as a function of time and distance from the FOV center.E0102, RXJ1856, DarkEarth. => arf-builder or emfbuilder.
QE at C-K line on the ground is ½ of the orbital QE?
CTI correction was introduced in rev0.6 processing. Need to be checked in various sources. (Energy, time, position dependence) Feedback from SWG is expected.
PHA->PI conversion (xispi) should be upgraded.
Charge Trail Correction (introduced from rev0.3) parameters should be checked with 2006 data.
Resolution decrease (inevitable with CTI correction) should be included in the response. rmf-builder can do it. CI measurement and CTI correction by each column might be used to save the degradation
Background data base is under construction. Bad Columns data base should be updated.
Data-Selection Criteria will be revised from (ELV>5 && DYE_ELV>20) to (ELV>5).