Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 1/22

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Name

Prepared by:

A. Chapon J.C. Cl´emens A. Secroun

Date

Signature

30/11/2012

Approved by:

xx/xx/2012

Authorized by:

xx/xx/2012

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 2/22

Modifications

Issue

Date

Modifications

1.0

xx/xx/2012

Document creation

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 3/22

Contents I

Flux varying measurements

1 (Non-)Linearity 1.1 Experimentation . . . . . . . 1.2 Analysis . . . . . . . . . . . . 1.2.1 Standard non-linearity 1.2.2 Reciprocity failure . . 1.3 Output . . . . . . . . . . . .

6 . . . . .

6 7 7 7 7 8

. . . . . . . .

8 8 9 9 10 11 12 12 12

3 Persistence 3.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12 13 13 13

4 Well Capacity 4.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13 13 13 13

II

14

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

2 Conversion Gain and Interpixel Capacitance 2.1 Experimentation . . . . . . . . . . . . . . . . 2.2 Analysis . . . . . . . . . . . . . . . . . . . . . 2.2.1 Conversion gain . . . . . . . . . . . . . 2.2.2 IPC and photon transfer method . . . 2.2.3 IPNL auto-correlation method . . . . 2.2.4 Pixel uniformity . . . . . . . . . . . . 2.2.5 Pixel operability . . . . . . . . . . . . 2.3 Output . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

Noise measurements

5 Total Noise 5.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14 14 14 14

6 Dark Current 6.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15 15 15 16

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 4/22

7 Reset Anomaly 7.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16 16 17 17

III

17

Wavelength varying measurements

8 Quantum Efficiency 8.1 Experimentation . . . . . . . . . . 8.1.1 Preliminary work . . . . . . 8.1.2 Illumination flatness . . . . 8.1.3 Photodiode cold calibration 8.2 Analysis . . . . . . . . . . . . . . . 8.3 Output . . . . . . . . . . . . . . .

IV

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

Other measurements

18 18 18 19 19 20 20

21

9 Cosmic Rays 9.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21 21 21 21

10 Intrapixel Sensitivity 10.1 Experimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21 21 22 22

V

22

Summary

11 Test plan or schedule

22

12 Output

22

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 5/22

Abbreviations ADU CDS CPPM DQE EMC EQE FET H2RG HAWAII IPC IPNL IQE JWST LED LN2 LNE MCT MTF NIR NISP NIST PC PSF PV QE QTH ROIC RT UTR

Arbitrary Detector Unit Correlated Double Sampling Centre de Physique des Particules de Marseille Detector Quantum Efficiency Electromagnetic Compatibility External Quantum Efficiency Field Effect Transistor HAWAII 2k×2k RG detector HgCdTe Astronomical Wide Area Infrared Imager Interpixel Capacitance Institut de Physique Nuclaire de Lyon Internal Quantum Efficiency James Web Space Telescope Light Emitting Diode Liquid Nitrogen Laboratoire National de M´etrologie et d’Essais (France) Mercury Cadmium Telluride or HgCdTe or “MerCadTel” Modulated Transfer Function Near InfraRed Near Infrared Spectrometer National Institute of Standards and Technology (USA) Photoconductor Point Spread Function Photovoltaic Quantum Efficiency Quartz Tungsten Halogen Read Out Integrated Circuit Room Temperature Up The Ramp

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 6/22

This document aims at specifying the characterization plan that should be followed to validate the performances of the H2RG detectors that will be used on the NISP instrument onboard EUCLID. This plan implies two modes of characterization: • a “noise mode” where the parameters are measured in electrons. This mode includes read noise and dark current measurements. • a “flux mode” where the measurements are flux dependent and given in percentage of flux. Here, the uniformity of the illumination source is critical and should be controlled both temporally and spatially. Thus outputs will be given as maps for each pixel and include: (non-)linearity, interpixel capacitance and conversion gain, persistence, well capacity and quantum efficiency. All measurements should be done with the same read mode as much as possible. Flux and wavelength shall be varied. Four sets of measurements may follow: 1. flux varying measurements (with given temperature and wavelength modes) from which can be derived (non-)linearity, reciprocity failure, conversion gain and IPC, persistence, well depth, pixel operability, pixel-to-pixel (non-)uniformity; 2. noise measurements for dark current, reset anomaly and total / read noise; 3. wavelength varying measurements for quantum efficiency, associated to cutoff and wavelength range; 4. intrapixel sensitivity which requires a specific setup. Here are some noteworthy comments on this characterization plan: • If most measurements are planned to be done at CPPM, some (in particular intrapixel sensitivity) could be obtained from our partners (for instance IPNL or University of Michigan). • It should be planned some time to check each setup before any detector acquisition. • It should also be planned a setup that allows to test two detectors at a time, in order in particular to cross-check and cross-calibrate their performances. Of course, it would also allow to make some EMC testing. • Another important issue that should be looked at: are there any correlations between the measured parameters, in particular between QE, dark, intrapixel and read noise which are all related to the conversion gain and interpixel nodal capacitance ?

Part I

Flux varying measurements Those measurements could be driven with the same setup and same set of acquired frames, from which the data would be analytically derived. This setup is based on a uniform (flat-field) illumination for which the flux is varied so that: • at low fluxes: more frames should be acquired • at high fluxes: reach saturation

1

(Non-)Linearity

See Unlinearity Internal Note. Should be explored: • standard non-linearity: including saturation effects at high fluxes, The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument •

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 7/22

reciprocity failure: systematic count rate dependent non-linearity.[?]

Note: reset anomaly (or pixel self-heating) is often considered as a non-linearity all the more than it can be accounted for with a four-parameter linearity correction. However, it is actually due to local (at the pixel cell level) temperature changes that occur when the clocking cadence is interrupted [?]. Thus it will be treated in the dark current section.

1.1

Experimentation

Measurement strategy Dependencies Linearity depends on • temperature • count rate • wavelength • readout cadence or mode Acquisition Linearity Reciprocity failure

1.2 1.2.1

non destructive UTR read under illumination vary flux while keeping integration time constant. Vary temperature. Flux variation can be obtained through the varying illumination of a reference LED. UTR readout mode may be used. At least four fluxes up to saturation.

Analysis Standard non-linearity

Define saturation or full well as deviation of say 10% from linearity (linear fit) at high fluxes. Fourparameter fit allows to take into account both reset anomaly and saturation 1.2.2

Reciprocity failure

The reciprocity law states that the detector response will be determined by the total exposure, defined as intensity × time. Therefore, the same response should result from reducing duration and increasing light intensity, and vice versa. Reciprocity failure is a count-rate dependent non-linearity in the detector. It is distinctly different from the well-known total count dependent non-linearity due to saturation as the well is filled. Reciprocity failure • might be wavelength dependent • exhibits a power law behavior with pixels with high count rates detecting slightly more, and pixels with low count rates detecting slightly less flux than expected for a linear system

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument 1.3

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 8/22

Output Output data

10 after correction non-linearity coefficient maps vs fluence, reset fluctuations map

Acquisition time Storage needs Note : one frame = 2048px × 2048px × 16 bits = 67 Mb to which some space for headers and setup information should be added. During one acquisition, frames should be stored as raw data, one frame each 17 s. So if the acquisition lasts 10 minutes, then we would need to store 10 × (60s/17s) × 70Mb = 2.5 Gb. Considering the characterization plan could last three weeks in continuous, it would add to some 6,5 Tb for one detector and some 130 Tb for 20 detectors.

2

Conversion Gain and Interpixel Capacitance

Conversion gain links electrons (created by incoming photons) to ADUs (digital unit given by the ROIC). It is the factor used to convert from ADU measured by the readout electronics to electrons collected at the detector node. It has units of e-/ADU. The conversion gain provides the absolute calibration for read noise, dark current, and quantum efficiency measurements and therefore its accurate knowledge is critical. Conversion gain is also directly linked to interpixel capacitance. Two phenomena are responsible for the interpixel capacitance: the lateral diffusion of charges and the coupling capacity. Lateral diffusion is a random process that does not create signal correlation between pixels. It takes place within the active layer before the accumulation of charges. Coupling capacity is a deterministic phenomenon that creates signal correlation between pixels. It takes place after the accumulation of charges. Approximately half of the coupling occurs in the multiplexer, and it can be asymmetric depending on the multiplexer geometry. This component is minimized with careful multiplexer design. The remaining coupling occurs between the depletion regions of neighboring pixels, and might be unavoidable. Small amounts of interpixel capacitance can cause large errors in the measurement of poissonian noise versus signal, and all subsequently derived measurements such as nodal capacitance and quantum efficiency. Crosstalk and MTF can also be significantly influenced by interpixel capacitance.

2.1

Experimentation

Measurement strategy Conversion gain is generally obtained through the photon transfer method (see below). This method is based on the statistical properties of electrons generated by a uniform illumination of the detector. The gain is then derived from the inverse slope of the variance as a function of the mean signal. Two steps must be considered. 1. Flat field illumination Precisely, the mean of the pixel distribution is easy to estimate, but the variance can be difficult to measure, due to the presence of correlated noise. There are two ways of measuring the variance: • Temporal sampling estimates the variance on a pixel by pixel basis, by taking a large number of exposures (> 104 for an error less than 1%) and calculating the standard deviation for each pixel. Temporal sampling takes many hours and requires a stable illumination source.

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 9/22

Spatial sampling calculates the variance from the difference of two frames. Spatial sampling assumes all the pixels are identical; differencing two frames corrects pixel to pixel variations, making this assumption valid. The biggest issue when using a spatial statistic is the impact of correlations in the pixel outputs on the variance estimate. Two variations of this method: ”full-frame method” and ”gray-scale method” (Cf. JWST independent testing) 2. Correction to take pixel correlations into account There are many ways of measuring the interpixel capacitance that correlates neighboring pixels and thus obtain a proper evaluation of the conversion gain: • Capacity comparison: an external capacity is used as a reference to measure the nodal capacitance.[?] 55 Fe source: only possible with substrate removed devices, which is the case. Gives • Use of a precise knowledge of the number of created electrons with an X-ray source.[?] • Autocorellation method or so-called photon transfer method: the knowledge of intercorrelations allows correction of conversion gain. Smadja et al. has developed a thorougher analysis strategy taking into account neighboring pixels, also based on uniform illumination. This strategy will be applied here.[?] • Point spread function uses the windowed reading of a pixel to know its PSF and correct the conversion gain. •

Dependencies Gain and interpixel capacitance depend on • pixel (histogram) • temperature • biases • version of multiplexer Acquisition δG ≤ 1%) G Varying fluxes (10) or integration time up to well below the full well to avoid non-linear or saturation effects CDS or Fowler-1 acquisition mode for photon transfer method or Multiaccum for IPNL method (Differential mode to minimize the effect of pixel to pixel variations) @100K @1λ Error estimated with 20 acquisitions (Desired precision:

2.2

Analysis

Figure 1 shows the schematics of a pixel: the incoming photon is converted to electrons in the MCT p-n junction, then thanks to the indium bump, the electron signal is transferred to the ROIC. 2.2.1

Conversion gain

Conversion gain G links electrons (created by incoming photons) to ADUs (digital unit given by the ROIC). It is thus defined as: Ne G= A The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 10/22

incoming photon

AR coating MCT p-n junction indium bump Vdd

ROIC Si multiplexer

Vreset

bus toward readout amplifier

Figure 1: Schematics of a pixel and its ROIC

where Ne is the number of electrons accumulated within the pixel (e-), and A is the ADU response (ADU). The ADC (Analog to Digital Converter) responds to a voltage change between two frames according to: A=k·V where V is the pixel voltage and k is the conversion factor defined as the product of the ADC conversion factor by the on chip (MOSFET) source follower gain. For a single pixel, V varies linearly with the number of accumulated electrons: Ne V =q C0 where q is the charge of the electron and C0 is the capacitance of one pixel (≈ 40 fF according to Teledyne). It comes out that: Ne Ne Ne C0 C0 G= = = = A k·V kqNe kq 2.2.2

IPC and photon transfer method

This method[?, ?] is based on the statistical properties of electrons generated by a uniform illumination of the detector. It compares the mean signal to the variance. For a given pixel, the ADU response is: A=k·V =

kq Ne C0

and its corresponding variance is: 2 σA =



kq C0

2

2 σN e

The electron signal may be considered as Poisson distributed (shot noise) so that: 2 σN = Ne e

It comes out that: 2 σA

 =

kq C0

2

2 σN e

 =

kq C0

2

 Ne =

kq C0

2 A

C0 kq 1 = A= A kq C0 G

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 11/22

For a Poisson process, the mean and variance must be equal; the conversion gain is thus the factor that multiplies the pixel distribution (in ADU) to make this condition true. The conversion gain is thus the inverse slope in the relation between the variance σA of the stochastic contribution to the noise (in ADU) in a given time and the charge A accumulated during the same time (also in ADU). The capacitance C0 of the pixel can be obtained once the conversion factor k is known. Inter-pixel capacitance shares charge between neighboring pixels and leads to an underestimation of the variance (overestimation of the conversion gain) when using a standard variance estimator. A new variance estimator, which accounts for nearest neighbor correlations, must be considered, as given by Moore [?]:  1 X 2 2 σA = Di [k, l] (1) i 2N k,l X X + Di [k, l]Di [k, l + 1] + Di [k, l]Di [k, l − 1] (2) k,l

k,l

X

X

 +

Di [k, l]Di [k + 1, l] +

k,l

Di [k, l]Di [k − 1, l]

(3)

k,l

where N is the total number of pixels, Di [k, l] the signal measured in pixel [k, l] in the Fowler-1 difference frame (equivalent to CDS (correlated double sampling)). Again here, the gain is obtained as the inverse slope of the variance σA versus the incoming flux A, where the variance is evaluated with the Moore formula taking into account the nearest neighbors. 2.2.3

IPNL auto-correlation method

Generalized cross-correlation method using more neighboring pixels. See references for details on the method.[?, ?] This method is based on a redundant determination of the interpixel capacitance using groups of pixels and a temporal (rather than spatial) analysis of the signal. For a group of 5 pixels in the following configuration, 2 3

0

1

4 the electrostatic influence matrix is C = C0 M with 

1 x  M = x x x

x 1 0 0 0

x 0 1 0 0

x 0 0 1 0

 x 0  0  0 1

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 12/22

Cij , with C0 is the capacitance of the central pixel and Cij is interpixel capacitance or C0 otherwise an electrostatic influence coefficient.   1 −x −x −x −x −x 1 − 3x2 x2 x2 x2    1 −x x2 1 − 3x2 x2 x2  M −1 =   2 1 − 4x  −x x2 x2 1 − 3x2 x2  −x x2 x2 x2 1 − 3x2

where x =

2.2.4

Pixel uniformity

Non-uniformity may be evaluated from the average and standard deviation values of gains. Nonuniformity is evaluated on a pixel-to-pixel basis as well as over large scales of say 100×100 pixel regions. 2.2.5

Pixel operability

Bad or non operational pixels include: • dead pixels: charge is not integrated. • hot/bright pixels : looking for very high pixel signal slopes dI/dt, non-linear response with time (similar to capacity discharge), looking for pixels that have a value greater than 75% of the full A/D range in the zeroth read • open pixels: value strongly falling down compared to neighboring pixels for high illumination. These pixels are mis-connected and have a large impedance. JWST has used different criteria: • deviating by more than 30% from the mean DQE • total noise higher than 12 e- rms Note: all analyses should correct for mis-connected and/or hot pixels that often contaminate the images (specially in the case of engineering-grade detectors) thanks to masks applied to remove outlying pixels.

2.3

Output Output data Acquisition time Storage needs

IPC map and conversion gain map vs fluence 36 h per detector 540 Gb per detector

Derived from these measurements are • pixel-to-pixel non uniformity (over adjacent pixels) • pixels non uniformity over large scales (100×100 pixel regions) • SCS figure of merit map for pixel operability

3

Persistence

After an intense illumination above and below saturation and after reseting the detector, persistence remains for a long time. This persistence can also be observed when changing the level of reference biases. The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument 3.1

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 13/22

Experimentation

Measurement strategy The detector is exposed to a large flux for a certain time, followed by many ”row-by-row resets ”, and some ”darks”. JWST: mean image signal following a 12 s exposure to full well and 3 resets. Dependencies Persistence depends on • time and flux: depending on above or below saturation • temperature • detraping time, global reset should be tested Acquisition

3.2

Analysis

3.3

Output Output data Acquisition time Storage needs

4

Well Capacity

Full well is defined as the level where non-linearity is greater than a certain value, typically 5 or 10%. May be derived from non-linearity measurements.

4.1

Experimentation

high flux of 50e-/px/sec read every second without reset and accumulate up to 1500 s The average well depth is measured by illuminating an infrared LED and taking a series of very short non-destructive readout at regular interval during a long exposure. The number of ADU on average for each pixel determines the well depth in electron (with the measured gain).

4.2

Analysis

Check for saturation and deviation from linear response. Criterion: d2 Q >C d2 ADU

4.3

Output Output data Acquisition time Storage needs

well capacity maps 30 minutes 7.5 Gb per detector

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 14/22

Part II

Noise measurements 5

Total Noise

Total noise includes dark current (treated hereafter) and read noise. Read noise includes: shot noise in FETs, Johnson noise, 1/f noise, shot noise on the dark current, drifts in the bias voltages. It may originate in the readout electronics, the multiplexer and the detector material.

5.1

Experimentation

Measurement strategy Noise requires • typical EUCLID acquisition mode • UTR-150 for spectroscopy (expected total noise < 9.0 e- for integration time = 500 s) • Fowler-16 for photometry (expected total noise < 7.3 e- for integration time = 100 s) • other acquisition modes ? Dependencies Noise depends on • integration time • readout mode: UTR, Fowler, CDS • sampling method: spatial or temporal • statistics chosen to define the noise Acquisition Noise is usually measured through the variance of the photon signal (here for n acquisitions, typically n > 50): v u n u 1 X σi,total = t (si,j − < si >)2 n−1 j=1

5.2

Analysis

Rauscher [?] shows that total noise in a multi-n×m acquisition mode can be described as: 2 σtotal =

12(n − 1) 2 6(n2 + 1 2(2m + 1)(n − 1) σread + (n − 1)tgroup f − (m − 1)tf rame f m · n(n + 1) 5n(n + 1) m · n(n + 1)

where f is the flux (e-/s/px) including photonics current and dark current, tgroup is the time between two groups of m frames, and tf rame is the time between two frames. UTR mode corresponds to m = 1 and Fowler modes correspond to n = 2. Under ultra-low flux (f ≈ 0) and for CDS acquisition mode (n=2, m=1), this equation comes down to the familiar: √ σCDS = 2σread

5.3

Output Output data Acquisition time Storage needs

noise maps, noise histogram 10 h 150 Gb per detector

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument 6

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 15/22

Dark Current

Dark current comes from the thermal generation of electrons. It is measured in e-/px/s. Underlying is the measure of the mean and dark current and its associated noise. Study of dark current is similar to that of total noise, except that measurements are done without illumination and the signal is generated from thermal effect.

6.1

Experimentation

Measurement strategy Dark requires • very stable source over 24h • varying temperature between 85-90K and 140K To prevent contamination of the measurement, dark current measurements are performed with the detector inside a light-tight enclosure mounted to the cold stage at a temperature of 100K. The dark current measurements are always performed after the detector is powered on continuously for about 24 hours and thermal equilibrium is established. Reference biases might derive and must be taken care of: • thermal-induced bias drifts: temperature must be controlled down to 1mK • electricaly-induced bias drifts: biases must be very stable Reference pixels allow for compensation. Each output amplifier must be treated individually and a linear ramp subtracted from the digitized video signal. This ramp is determined from the mean values of the first 4 and last 4 pixels of a line. Need to check the validity of subtracting reference pixels in Fourier space. Dependencies Dark current depends on • temperature • reverse bias (typically 0.25V) • distribution in bands and pixels Acquisition ”Long dark exposure and taking a series of non destructive CDS read at regular interval also known as Up the Ramp exposures.” Statistical measures are required through either • spatial averaging through the derivation of pixel to pixel variance in a single dark frame, Fowler-1 • temporal averaging through the measurement of the fluctuation in a given pixel when an exposure is repeated (at least 50 frames). Care must be taken of electronic and thermal drifts. Reference pixels can be used to correct for DC shifts.

6.2

Analysis

Dark current is defined as the accumulation of charge generated by thermal excitation in the bulk material rather than incident photons, and thus depends sensitively on the temperature as follows: Eg

Idark ∝ kT 3/2 e 2kT where k is Boltzman constant, T is the temperature (K), and Eg is the energy bandgap (eV). The bandgap Eg of Hg1−x Cdx Te is a function of the alloy composition ratio x of CdTe to HgTe and the temperature of the material. It’s given by[?]: Eg = −0.302 + 1.93x − 0.81x2 + 0.832x3 + 5.3510−4 T (1 − 2x) The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 16/22

For 2.3 µm cutoff HgCdTe detectors, Eg =

hc = 0.54 eV λco

where λco is the cutoff wavelength, h is Planck constant (6.62 10−34 m2 .kg/s), and c is the speed of light (3 108 m/s). The bandgap temperature dependency is expressed as: Eg (T ) = Eg (0) −

αT 2 T +β

where Eg (0) = 0.814 eV, α = 4.10−4 eV/K and β = 182 K. Measurements of the dark current require a photon free environment and an understanding of bias and temperature drifts that cause changes in the output signal. Measurements of the dark current are subject to contamination due to read glow, a small amount of charge injected into each pixel when the detector is read out. When a pixel is addressed and read out, current flows through the source follower FET in the multiplexer, which can stimulate the emission of infrared photons. This appears as an increased dark current when a fast readout cadence is employed. The glow is expected to be temperature dependent. The read glow and dark current contribute to the total detector noise. The common assumption is that the noise is Poisson distributed and scales as the square root of the signal, however measurements by Figer et al. indicate that this may not be true.[?] An analysis similar to that total noise may be applied to derive the dark current, using the statistics of generated electrons.

6.3

Output Output data Acquisition time Storage needs

7

dark maps vs temperature (4), mean dark current 100h 1,5Tb per detector

Reset Anomaly

Reset anomaly (or pixel self heating) was initially thought of as a non-linearity consequent to the reseting applied before any acquisition. Brown’s results indicate that this anomaly is actually due to local (at the pixel cell level) temperature changes that occur when the clocking cadence is interrupted.[?] When a pixel is read out, current flows through the pixel source follower FET and power is dissipated in the pixel. Slowing the readout cadence leads to cooling of the output FET and a positive voltage change on the output. Increasing the readout cadence leads to local heating within the pixel, which appears as a negative signal at the pixel output. Continuously clocking (both before and during the exposure) and reading the detector should eliminate this effect.

7.1

Experimentation

Local heating is observed as a negative signal in short exposure time dark current ramps following a reset. Temperature drifts due to local heating lead to systematic offsets in the pixel output level and add noise to Fowler sampled data. Reset anomaly should be seeked on dark current frames as the small magnitude of the dark signal is easily contaminated by temperature and bias drifts. The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 17/22

Figure 2: The impact of pixel self heating on the detector output for Fowler sampled data with a 300 s exposure delay.[?]

7.2

Analysis

The reset anomaly can introduce systematic errors into dark current measurements if it is not correctly accounted for: systematic bias tends to overestimate the dark current for a straight line model.[?] The reset anomaly has been observed to have time constants ranging from seconds to hours before the pixels reach the asymptotic portion of the ramp. Reset anomaly is highly repeatable for a given pixel. It contributes almost no additional noise (degrades the total noise by only a few percent). It calibrates out during matching dark or sky subtraction. The reset anomaly may be accounted for in dark current measurements by fitting a four-parameter function to up-the-ramp sampled pixels. Pixels showing reset anomaly can be well-modeled by a four-parameter function of the form: si (t) = ai + bi t + ci edi t Thus the non-linearity correction with a four-parameter polynomial description should be sufficient to correct the reset anomaly.

7.3

Output Output data Acquisition time Storage needs

reset anomaly maps analysis of dark current frames NA

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 18/22

Part III

Wavelength varying measurements 8

Quantum Efficiency

Quantum efficiency is defined as the ratio of the flux of collected electrons to the flux of incident photons. EQE (External Quantum Efficiency) should be differentiated from IQE (Internal Quantum Efficiency) for which only incoming photons that are absorbed are taken into account. Here we only consider EQE.

8.1

Experimentation

Measurement strategy requires flat field better than a few percents QE calibration of reference photodiode at LN2 requires a precise knowledge of the amplification of the electronics chain requires a monochromatic flat-field ! meaning extremely low fluxes to be expected on the detector irradiance or flux mode ? flux level ? choice of photodiode ?

Dependencies QE depends on • wavelength • temperature: MBE detectors do not depend on temperature, unlike LPE, however cutoff varies with temperature • position: large scale variations Acquisition measure every 50 nm between 900 nm and 2400 nm (provided the detector cut-off is 2.3 µm), thus a total of 30 measurements typical flux of 2 e-/px/s 8.1.1 •

•

• •

Preliminary work choice of photodiode according to – flux level will be relatively low, cold dark level should be as low as possible – photodiode QE should not be much worse than that expected of the H2RG – it must have a cut-off wavelength outside of the range to be characterized preparatory calibrations – photodiode: QE vs wavelength, cold and warm calibration for reference – monochromator: wavelength – lamp: flux stability – filters: transmission homogeneity check flatness of flat field check optical alignment

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument 8.1.2 •

•

8.1.3

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 19/22

Illumination flatness CPPM: 1 photodiode (not necessarily calibrated) moved in X and Y in place of the detector, moved thanks to under vacuum micro-displacements, say following an array of 5x5 positions to which should be added the locations of the fixed photodiodes planned for follow up. IPNL: 16 photodiodes cross-calibrated placed as an array in the focal plane, therefore there are no displacements, but cross-calibration could be tricky. Photodiode cold calibration

option to have the calibration done at either LNE (Laboratoire National de M´etrologie et d’Essais), France or NIST (National Institute of Standards and Technology), USA

Which material for the photodiode ? (Hamamatsu InGaAs or Judson MCT) Basically there are few materials in the range of NIR wavelengths. The table hereafter summarizes existing photodiode types. Of course, there are other parameters to consider: the sensitive area, the type of diode (photovolta¨ıc or photoconductor). Major vendors are Hamamatsu and Judson (Teledyne) but others could be considered ??? material RT cutoff LN2 cutoff dark linearity comment InAs 3.6µm 3.1µm to consider InSb 7µm 5.5µm fairly high to consider even @77K PbS 2.9µm non linear ”old” PbSe 4.8µm non linear ”old” MCT 2.8µm large @RT good PV better than PC, requires good amplifier, more universal than InSb InGaAs 1.7µm 1.4µm cutoff too short Ext. InGaAs 2.6µm excess noise requires cooling Which diameter of photodiode ? Choice is typically between 1mm2 or 3mm2 , small diameter allows only overfill or irradiance mode, whereas larger diameter might bring in spatial discrepancies. Pulsed or continuous mode ? The sensitivities of the photodiode and the detector may not be compatible, namely the photodiode dark current is much higher than that of the detector and at low fluxes, the signal seen by the detector might not be seen by the photodiode. Here are some ways to by-pass this issue, though not completely satisfying: • work in pulse mode on the detector and integrate the whole signal on the photodiode, • use a larger area of photodiode, • use a photodiode with a lower dark current (as might be the case of the MCT photodiode). Should the calibration been done in underfil or overfil mode ? (refer to the question between irradiance or flux illuminating mode) • underfil mode: incoming light is completely received by the photodiode sensitive area, thus giving directly access to QE as the ratio of the number of measured electrons to that of incoming photons The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 20/22

overfill mode: as the beam diameter of incoming light is larger than that of the photodiode, the number of incoming photons is harder to evaluate precisely. Is the ratio of beam to photodiode areas precise enough ? Should there be a correction to take the borders into account ? Could we check the validity of our calculation by moving the photodiode within the beam (for instance two measurements with one centered and one on the side) ? Reversely, considering that H2RG detectors will be in overfill mode, it might be more appropriate to make QE measurements in overfill or irradiance mode for both the photodiode and the H2RG, thus avoiding the issue of area correction. •

Cold and warm calibration at IPNL IPNL is looking into obtaining the cold calibration of a photodiode from its warm calibration. The idea is to use the same setup for both measures with and without cooling the photodiode. The light source would be LEDs maintained in both cases at a constant temperature with a Pelletier cooler. Cold and warm calibration at LNE Discussions are ongoing to have our photodiodes calibrated at LNE They can apparently provide us with warm QE calibration, but they are not equipped for cold calibration. We are trying to find a solution.

8.2

Analysis

Quantum efficiency is derived from the spectral response of the detector. It is defined as the ratio of the flux of electrons collected by the detector to the flux of incident photons. QE(λ) =

Rλ hc . λ q

where Rλ is the responsivity (A/W), λ is the wavelength (nm), h is Planck constant, c is the speed of light and q is the elementary charge. When working in flux (or underfil) mode, its absolute value may be obtained as the ratio of the incoming photon flux to the received electron flux. However, when dealing with overfill mode, it may be obtained as the diode current (in Amps) as a function of irradiance (W/cm2 ). Is it convenient to work with this expression of QE ? note: it might interesting to consider the DQE (detective quantum efficiency), also called noise factor, which also takes into account the spatial frequency of the incoming signal. DQE(f ) =

8.3

2 (f ) SN Rout 2 (f ) SN Rin

Output Output data

30 QE maps, QE vs wavelength (5), cutoff, QE max gradient, operating wavelength range, mean QE

RAW data Acquisition time Storage needs The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 21/22

Part IV

Other measurements 9

Cosmic Rays

Proton effects and long term performance degradation.

9.1

Experimentation

9.2

Analysis

9.3

Output

10

Intrapixel Sensitivity

[?, ?, ?]

10.1

Experimentation

Specific setup with submicron exploration capability (not available at CPPM)

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.

Characterization Plan and Performances of the H2RG Detectors of the NISP Instrument 10.2

Analysis

10.3

Output

Ref.: EDMS ref.: Issue: Date: Page:

EUCL-CPPM-ALT I-xxxxxx 1.0 xx xxx 2012 22/22

Part V

Summary 11

Test plan or schedule

12

Output

Parameter

Output maps

Other outputs

QE

QE vs λ

IPC / Gain

IPC, conversion gain vs fluence

Noise Dark Well capacity Non-linearity

noise dark vs temp. well capacity non linearity coefficients, reset fluctuations

maximum cutoff wavelength, QE maximum gradient, operating wavelength range, mean QE pixel to pixel non uniformity, pixels non uniformity over large scales histogram mean dark current

Acq. time

Prep. time

Data storage*

36 h

540 Gb

10 h

150 Gb

30 min

7.5 Gb

Intrapixel For 1 detector For 20 detectors *Storage capacity should be doubled if aiming at secured data.

note : comment valuer l’uniformit d’un dtecteur au voisin ?

The presented document is Proprietary information of the Euclid Consortium. This document shall be used and disclosed by the receiving Party and its related entities (e.g. contractors and subcontractors) only for the purposes of fulfilling the receiving Party’s responsibilities under the Euclid Project and that the identified and marked technical data shall not be disclosed or retransferred to any other entity without prior written permission of the document preparer.