05.12.01

columbia

Calibration of the ReflectionGrating Spectrometers

n Physical modeln Scientific performancen Current SAS implementationn Conclusions

• SRON: J.W. den Herder, A. Brinkman, A.J. den Boggende, J.Kaastra, T. Tamura, C. de Vries

• Columbia: S. Kahn, J. Cottam, F. Paerels, J. Peterson, A.Rasmussen, D. Reynolds

• MSSL: G. Branduardi-Raymont• PSI: M.Audard, M.Güdel• ESA/SSD: C. Erd• + others

05.12.01

columbia

Physical model

• Optical design• Mirror response• Grating response• Camera/CCD response

05.12.01

columbia

Optical design (1)

CCD

Cal. source

6700

.0

561.0

X

Z

X

Y EPIC

γ=2.275068o

α+βM=4.804 o

RFC

6698.8

n=7.333o

α=1.576191o

mirrors

gratings

24.8

05.12.01

columbia

Optical design (2)

γ

δ

α

200

100.4

97.5

X-ray

β

Y

Z image storage

F Al F Al

X

Z

CCD bench 4 cal sources

image: 1024 x 384

storage: 1024 x 384

spectral image

pre-amp2 pre-amp1

Y

Z

dead space ~ 0.5 mm

CCD: 9 m=-1: 5 Å m=-2: 2.5 Å

CCD: 1 m=-1: 38 Å m=-2: 19 Å

05.12.01

columbia

Optical design (3)

β, outgoing angle from gratings

β, outgoing angle from gratings

C

CD

PH

cro

ss d

ispe

rsio

n

05.12.01

columbia

Mirror response

• Empirical fit to raytraceresults (consistent with toflight data PKS 0312-770),projected on dispersionaxis

• mirror effective area

• empirical fit to crossdispersion data (Mkr 421)on RGS detector (Row-land circle is not opticalmirror focus)

CCD X (dispersion)

CC

D Y

(cro

ss d

ispe

rsio

n)

CCD X (dispersion)

Cou

nts/

bin

05.12.01

columbia

Grating response

• Calculation of efficiencyusing Maxwell equations

• Fit blaze angle to variousdatasets (Bessy, αdependence, orders)

• incoherent scattering(scalar theory) with twodistributions (small andlarge angle) in thedispersion direction

• large angle scattering incross dispersion direction

• Add alignment information

Bessy data and model

05.12.01

columbia

CCD response: redistribution

• CCD energy responseused to separate orders(less critical)

• CCD redistributionfunction includes partialevent tail and CCE atthe backside

• CCD redistributionfunction verified in orbitas for a given positionon the detector theenergy is ‘monochro-matic

Partial eventsDaresbury + thresholds

Partial events:CCE

Ephoto

05.12.01

columbia

CCD response: QE• QE based on CCD

thickness (verified by IR)and thickness of variouslayers (optical Al filter of45, 68 and 75 nm andMgF2 insulation layer)

• QE verified during groundcalibrations

• CTI/gain correctionsdetermined in orbit usingonboard calibrationssources (Al and F) aswell as astrophysicalobjects

QE

05.12.01

columbia

Scientific performance

• Line spread function(LSF)• Wavelength scale• Effective area• Background• In-orbit monitoring

05.12.01

columbia

Line spread function

• Expected LSF isconvolution of

- mirror response

- grating response

- and CCD response

- (+ electronics)mirror

Grating scattering:

small angle

Large angle

CCD response

05.12.01

columbia

In-orbit verification of the LSF

• Verified on number ofstrong unblended lines

• Model (raytrace) in goodagreement with data(some uncertainty inbackground/continuum)

• Calculated response(FWHM) in goodagreement with largerset of data

Wavelength [Å]

Res

olut

ion

(λ/∆

λ, F

WH

M)

Continuum/backgrounduncertainty

RGS1

RGS2

05.12.01

columbia

λ scale verification

• Ground alignment doesnot give ultimateaccuracy

• Comparison of stronglines (Lyα) withlaboratory wavelengthsfor a number of pointings

• Result: ± 8mÅ (1 σ)• RGS1 + 1.5 mÅ• RGS2 – 1.6 mÅ

8 mÅ

05.12.01

columbia

Effective area

• Based on physicalmodel

• Validated during longbeam tests (which is notfully representative)

• Verified in-orbit on BLLac (PKS2155) resultingin number of changes

05.12.01

columbia

Effective area 2

Optimization of scatter parameters to describe the flight data(red = model, black = data)

λ = 7 Å λ = 15 Å λ = 21 ÅX

= 0

.5 a

rcm

in

1

.5 a

rcm

in

5.

5 ar

cmin

05.12.01

columbia

Effective area (3)

• Apply correction factorsfor:

– Difference RGS1 andRGS2 (β dependent)

– O-edge

– Second orderspectra

05.12.01

columbia

Effective area: result

• Predicted effective areafor RGS1 and RGS2

• includes hot columnsand failing CCD chain(only shown for RGS2)

• No sharp features due toscattering wings

• Edges due to O, F, Mgand Al + finite thicknessof Si (~ 30 µm)

RGS2

RGS1

05.12.01

columbia

Effective area: verification

• Result for PKS2155 consistent with single power law forboth instruments (within 5% except for < 7 Å)

RGS1 RGS2

05.12.01

columbia

Background

• Different backgroundcomponents:

– Minimum ionizingparticles

– Electrons– Onboard calibration

source– Fluorescence lines– Read-out noise– Soft protons entering

the telescope (highlyvariable)

Increase due to PH filter

05.12.01

columbia

In orbit monitoring (1)

• Dark current and opticalload (• uniform, less than 1 e-

contribution to read-outnoise of 5 e-,

• few defects• Particle background

• CTI + gain (on chipamplifier and electronics)• Effect less than 1%

(normalized on orbit 165)• Slope change, presum-

ably since electronics isoff during perigee

Off during perigee

Bias voltage change

Z (dispersion) [pixels]

X-d

isp

ersi

on

[p

ixel

s]

05.12.01

columbia

In-orbit monitoring (2)

• Hot and flickering pixels(small fraction << 1%with intensity whichdepends on time)

• Increase following solarevents and gradualrepair of defects (at–80oC)

time

CC

D P

HA

CC

D P

HA

05.12.01

columbia

In-orbit monitoring (3)• Contamination verified using

intensity of the onboardcalibration sources

– Al and F continuouslyilluminate parts of CCD

– No significant difference inslope (but also not consistentwith T1/2 of Cm source)

– No contamination on detector

– Marginal contamination onsource stopping a particles

Al, slope = -0.00063(10)

F, slope = -0.00081(10)

Cm T1/2

Cm T1/2

05.12.01

columbia

SAS products

• SAS version 5.2 (and related CCFs) give reasonableresults

• Not included are:- correction to force RGS1 to RGS2 (β dependent)- Correction for second orders- Proper vignetting function of the gratings- Proper description of O-edge including fine structure

• Without these corrections LSF, λ scale are correct, Aeff lessaccurate

• After these corrections response can be normalized toreference source

05.12.01

columbia

SAS products (2)

• Not all parts of calibration in public version of SAS (yet)

RGS1 RGS2

05.12.01

columbia

Conclusions

• Quality of calibrations in good agreement with pre-flightpredictions based on a physical model of the instrument

• Improvements are feasible in λ scale and response model (butrequire significant further work)

• Normalization of response to PKS2155 reduces uncertainties

• In-orbit performance as expected

• Current public SAS reasonable but various improvementsidentified to improve Aeff