Fast Solar Polarimeter

Fast Solar Polarimeter

A. Feller, F. Iglesias, K. Nagaraju, S. K. SolankiMax Planck Institute for Solar System Research

and colleagues from the Max Planck semiconductor lab

A. Feller FSP IAUS 305 1 / 15

Overview

Fast Solar Polarimeter (FSP) in a nutshell

Novel ground-based solar imaging polarimeter developed by MPS incollaboration with the MPG semiconductor lab (HLL) and PNSensorBased on

fast low-noise pnCCD sensor andferro-electric liquid crystals for polarization modulation

Polarimetry of small and dynamic solar structuresat increased polarimetric sensitivity (< 10−3) orat high temporal cadence

in particular also in the chromosphereDevelopment in 2 phases:

2012-2014 Proof of concept with small pnCCD prototype(264x264 pixels2), single-beam

2014-2016 Development of full-scale, science-ready version with1kx1k pnCCD, dual-beam

Funded by MPG and European Commission (SOLARNET)


Why FSP?

Photon budget and solar evolution

Tradeoff between solar evolutionand noise:

Maximum integration time ∆tallowed by solar evolution:

∆te = 2 ∆x/v

Minimum integration time toreach a given required rmsnoise level σ:

∆ts = (Fσ2∆x2)−1

∆x : spatial samplingv : evolution speedF : Flux [phot / (s · arcsec2)]

Δx

Δt

Δts

Δte

optimum (Δx, Δt) ~ F-1/3 σ-2/3


Why FSP?

Photon budget and solar evolution

107 108 109 1010

Flux [phot / (s · arcsec2)]

10−2

10−3

10−4

RM

S no

ise

0.010", 1.0s

0.017", 0.2s

0.020", 2.1s

0.034", 0.4s

0.040", 4.1s

0.068", 0.8s0.080", 8

.3s

0.137", 1.7s

0.160",16.6s

0.274", 3.3s0.320",3

3.1s

0.547", 6.6s

Fe I 525.0 nm

CaII 393.3 nmSr I 460.7 nm

1m telescope:(Δx, Δt) = (0.12", 12.5s)


Why FSP?

Why fast modulation?

Slow dual-beam modulation is not sufficient for . . .high accuracy in the presence of strong polarization signalshigh spatial resolution

The demodulated imagesstill suffer from crosstalk between Q, U, V ...... which is not reduced by AO (see poster by Nagaraju)

Only corrective:Keep modulation cycle as close as possible to seeing time scale(∼ 10 ms)→ 100 Hz modulation!


Why FSP?

FSP is beneficial for 2 dedicated observing regimes

High-precision polarimetry (σ < 10−3)Fast modulation suppresses systematic errorsImage reconstruction and statistical techniques like

Feature-based spatial averaging (image segmentation)Feature tracking in time

conserve small-scale spatial information

Low-precision, high-cadence polarimetry

High duty cycle (95%)→ S/N in shortest possible ∆t1 reconstructed Stokes image set per s possible, due to

high frame rate (400 fps)short mod. cycle (4 states)


How does FSP work?

Main specifications

FSP I FSP IISensor size 264 px x 264 px 1024 px x 1024 pxMax. frame rate 800 fps 400 fpsPixel pitch 48 µm 36 µmQE > 90% 500 nm - 870 nm 350 nm - 500 nmDuty cycle 97% 95%RMS readout noise 3 - 4 e−

Sensitive subst. depth 450 µmReadout ASICS x number CAMEX x 4 VERITAS-1 x 16Max. data rate 0.78 Gb/s 6.7 Gb/s


How does FSP work?

pnCCD camera

Key conceptsFast split frame transferColumn-parallel readoutNo shutter→ numerical frametransfer correction(Iglesias et al. 2015)Multi-correlateddouble-sampling to reducenoiseCustom coating to optimize QE

Thick substrate→ no internalfringing

Sensor layout scheme

From Ordavo et al. 2011


Does FSP work as expected?

VTT test campaigns

Campaigns

Jun 2013 SpectrographNov 2013 TESOSJun 2014 TESOS

Setups (for data shown later)

VTT aperture 0.7 m

Spectrograph, 422.7 nm

Sampling 0.8" x 17 mÅ

FOV 72" x 3.7 Å

Efficiency 6 · 10−4

TESOS, 630.2 nm

Sampling 0.08" x 0.08"

FOV 20" x 20"

Spec. bandwidth 25 mÅ

Efficiency 1 · 10−2



Ca I 4227 Å, Scattering polarization

I

422.50 422.60 422.70 422.80

0

20

40

60

arcsec

Q/I

422.50 422.60 422.70 422.80

nm

0

20

40

60

arcsec

I

422.50 422.60 422.70 422.80

020

40

60

80

100

120

140

e-/(fram

e*pixel)

Q/I

422.50 422.60 422.70 422.80nm

0.0

0.5

1.0

1.5

2.0

2.5

%

Figure: Black: FSP obs. at µ ∼ 0.15; Blue line: atlas of the Second SolarSpectrum (Gandorfer 2000)



Figure: Time series of 19 MFBD reconstructed line scans (1.6s / spectralposition)



Fe I 6302 Å, Quiet Sun

Figure: Top: Simple averaging; Bottom: MFBD reconstructed



Fe I 6302 Å, noise behaviour (modulator off)

Figure: RMS noise vs. number of averaged frames


What’s next?

DEPFET/Infinipix - on-sensor charge caching

In a nutshell . . .Decoupling of frame rate and modulation frequencyPeriodic on-sensor charge caching, in phase with pol. modulationNo covered sensor areas, no charge transfer, 100% fill factorSwitching time ∼ 100 nsEssential FSP sensor properties(e.g. QE, frame rate, noise char., . . .) are conservedHeritage from particle physics and X-ray astronomy (BELLE-II,MIXS, ATHENA, . . .)EC "Horizon 2020" proposal submitted: polarimetry tests with32x32 4-DEPFET prototype sensor (2016-2018)


Summary

Summary

For high-precision polarimetry the light gathering capability of alarge-aperture telescope is more important than pushingdiffraction-limited resolution!FSP combines high duty cycle and fast modulation, which isessential for polarimetry at increased spatial resolutionThe FSP I prototype has successfully demonstrated the potentialof this novel polarimetry conceptWith future large-aperture solar telescopes at the horizon we willtry to improve solar polarimetry, based on pnCCD (and potentiallyDEPFET) sensor technology


Appendix Why fast modulation?


modulator

senso

r

S

pol. beamsplitterIu(t)

Id(t)

seeing, jitter, ...

u

d

Dual-beam modulation

with 2nd beam-exchange measurement

Iu(t1) =12

g (I + δI1) +12

4∑i=2

Si + δSi,1

Id (t1) =12

(g + δg)(I + δI1)− 12

4∑i=2

Si + δSi,1




modulator

senso

r

S


Id(t)

seeing, jitter, ...

u

d

Dual-beam modulation with 2nd beam-exchange measurement

Iu(t2) =12

g (I + δI2)− 12

4∑i=2

Si + δSi,2

Id (t2) =12

(g + δg)(I + δI2) +12

4∑i=2

Si + δSi,2




modulator

senso

r

S


Id(t)

seeing, jitter, ...

u

d

Modulated intensities after dual beam + beam exchange(neglecting higher-order errors)

I1 = Iu(t1)− Iu(t2)− Id (t1) + Id (t2) ≈(

g +δg2

) 4∑i=2

m1,i

(Si +

δSi,1 + δSi,2

2

)Same for I2 and I3 . . . (S1,2,3: Stokes Q, U, V; g: gain table; m: mod. matrix)




Slow dual-beam modulation is not sufficient for . . .high accuracy in the presence of strong polarization signalshigh spatial resolution

The demodulated imagesstill suffer from crosstalk between Q, U, V ...... which is not reduced by AO (see poster by Nagaraju)

Only corrective:Keep modulation cycle as close as possible to seeing time scale(∼ 10 ms)→ 100 Hz modulation!


Appendix How does FSP work?

FSP setup at VTT/TESOS



FSP setup at VTT/TESOS



Modulator

SOLIS/ZIMPOL design: 2 static retarders + 2 FLCsTemp. controlled (±0.1 K)Broadband efficiency optimization following Gisler 2006



Modulator

Polarimetric efficiencies

wavelength [nm]

modulation frequency [Hz]


Appendix Does FSP work as expected?

Fe I 6302 Å, Active region

Figure: 33s averages of MFBD reconstructed framesA. Feller FSP IAUS 305 5 / 9


Hα 6563 Å, Active region

Figure: Line scan, 55s average / spectral position



Expected performance at a 2m telescope

Fe I 6302 Å, active region

VTT test meas. 2m telescopeAperture 0.7 m 2 mEfficiency 1% 2% (dual beam)Duty cycle 50% 90%Spatial sampling 0.08" 0.03" (diff. lim.)1 spec. scan cycle (5 pos.) 15s 3.3s (solar evol.)No. of cycles 1 1Obs. time 15s 3.3sS/N 4.7 · 102 4.5 · 102



Expected performance at a 2m telescope

Example: Fe I 6302 Å, quiet Sun

VTT test meas. 2m telescopeAperture 0.7 m 2 mEfficiency 1% 2% (dual beam)Duty cycle 50% 90%Spatial sampling 0.08" 0.06"1 spec. scan cycle (2 pos.) 6.6s 6.4s (solar evol.)No. of cycles 35 2Obs. time 230s 12.8sS/N 3 · 103 3 · 103


Appendix What’s next?


A few words on the workingprinciple . . .

Based on the combineddetector-amplifier structureDEPFET (Treis et al. 2004)On-pixel, non-destructive chargesampling via FET conductivitymeasurementSuperpixel with 4 DEPFET cellsfor charge storing and readout

Shield electrodes induce periodicphoto-charge drifting into each ofthe 4 DEPFETs


Appendix What’s next?


Status and next steps . . .Prel. design of 4-DEPFET Infinipix sensorEC "Horizon 2020" proposal submitted: expected funding period2016-2019First conceptual study in terms of numerical simulationsTest of a small prototype sensor (32 x 32 superpixels) to assesspotential for polarimetry


Documents

Fast Solar Polarimeter