Mitigating CTE losses: Charge Injection and Pre/Post Flash

November 21, 2002 Curing CTE degradation Mauro Giavalisco 1

Mitigating CTE losses: Charge Injection and Pre/Post Flash


Traps and CTE losses

CTE loss is caused by traps formedin lattice by cosmic radiation.

During charge transfer operations, charge from transiting packets is captured and retained by “traps”, thus lost to the packet.

Traps retain charge for some timeand then releases it.

Charge released by traps can add tonearby following packets.

Process is stochastic in nature:packets loose, but also gain, charge in a random way:

• Space-dependent photometric bias

• Increased scatter• Decreased S/N• Net charge is lost• Fluctuations increase


Measuring CTE Losses

Effects of CTE loss can be reproduced in the LAB on CCDs subject to radiation damage of controlled magnitude.

Sources of known flux provided by radioactive isotopes, such as 55Fe. For example, an X-ray line from 55Fe promotes 1620 e- in the CCD detector

Equivalent to a flat f-l source with V=25.84 in a 1,800 sec exposure.

Photometry of CR hits made with Sextractor. Used circular aperture (5 pix d.) and isophotal aperture

Used CCD43-152 irradiated to•0 year•2.5 year•5 yearworth of damage


CTE degradation: 2.5 and 5 year damage

Comparison of photometry of CR hits:

• new detector• 2.5 year irradiated

detector • 5-year irradiated CCD

Effects of CTE losses:• degradation of

photometric uniformity

• loss of sensitivity1. charge is lost2. additional noise

introduced


Mitigating CTE Losses

Degradation of CTE mitigated by filling traps with charge.

Filled traps become passive and do not subtract additional charge from transiting packets.

Two methods to dispense charge:1. Charge injection,

• Discrete• continuous

2. Post or pre-flash with light

Shown here is charge injection of ~104 e- every 200 lines in 2.5yr CCD

Unfortunately, traps release charge after some time, becoming active again.


Release of Charge

Release of charge by trapsdiminishes effectiveness of of added charge to mitigate CTE losses.

Spacing between injectedlines is key parameter for Discrete Charge Injection .

Spacing must be such that traps are not allowed to “dry up” without charge and becomeactive.


Discrete Charge Injection: every 25 lines

By injecting charge more frequently, one can mitigate the charge release problem.

Shown here is charge injection of 104 e- every 25 lines in 5yr CCD.

Note that CTE losses, released charge from injected lines is much less than in the previous case.


Pre Flash and Continuous Charge Injection

Filling traps with charge can be done by either:

•Post/Pre Flash (injection by light)

•Charge Injection (electronically)•Discrete Charge Injection•Continuous Charge Injection

Shows here is the pattern of C.C.I. with ~10,000 e-/pix


C.C.I. Residual Map

Noise: s = 15 e- rms

C.C.I. repeatable and “calibratable”.


Pre Flash at 5 year - 1

Shown here are the curvesrelative to pre-flash with

100 and 200 electrons.

Also shown are the curves for the undamaged CCD and for the 5 year CCD

Improvement in the photometric uniformity is modest and overall similar to D.C.I. at 25 lines.

Photometric scatter is better than D.C.I. (probably due to filling all traps)

However, note the lower S/N ratios.


Pre Flash at 5 year - 2

Shown here are the curves

relative to pre-flash with 500,

1000 and 2000 electrons.


Improvement in the photometric uniformity is good and overall similar to D.C.I. at 25 lines for the 2.5 year CCD.

Photometric scatter is significantly better than D.C.I (probably due to filling all traps).

However, note the lower S/N ratios.


Continuous Charge Injection vs. Pre Flash at 5 yr

Continuous Charge Injection (CCI)very promising: •Same remedial effects as P.F.•In principle, much less noise

Curves relative to Pre Flash with

2000 electrons, and C.C.I. With

10,000 electrons.


Improvement in the photometric uniformity is very similar

C.C.I. has very high photometric scatter and lower S/N ratios.


Effects of isophotes’ size

Isophotal aperturesVs.Fixed Apertures (5 pix)


Faint Source Limit

• Case of faint sources not empirically tested for WFC3 CCD.• Studies with WFPC2 CCD (e.g. Whitmore et al. 2002) showed that fainter

sources proportionally more affected by CTE losses than brighter ones:– Flux no P.F 25 e- 250 e- 1700 e-– 20-50 DN 37.7 +/- 4.7 11.8 +/- 2.4 3.9 +/- 5.3 not enough stars– 50-200 DN 23.3 +/- 2.1 8.3 +/- 1.4 3.3 +/- 2.0 5.8 +/- 3.4 %– 200-500 DN 16.6 +/- 4.0 8.7 +/- 1.8 5.5 +/- 2.2 -1.8 +/- 2.8 %– 500-2000 DN 8.8 +/- 5.6 10.2 +/- 4.3 -1.2 +/- 4.2 2.3 +/- 1.6 %– Dm ~ 0.0 0.4 0.9 1.5

• Low-level pre-flash effective in mitigating faint/bright difference.• However, CTE mitigation with low-level pre-flash not terribly effective for

~1600 e- source (e.g. still 10% losses with 100 e-) in WFC3 CCD. • WFPC2 case suggests similar losses at faint levels, at best.• High-level P.F. has devastating effects of Poisson noise.• This suggests that C.C.I. is still optimal solution: increased noise from 5

to 15 e- corresponds to Dm~0.3 in V-band for a V~28 (220 e-) point source.


Conclusions

• CTE losses significantly degrade CCD performance:– Space-dependent photometric bias, photometric scatter– Decreased sensitivity (S/N), e.g.: Dm~0.8 loss in limiting

flux at 5 years – Decreased photometric accuracy (increased scatter)

• D.C.I. (25 lines) and P.F. (2000 e-) provide comparable mitigation to CTE loss

• C.C.I. superior to both P.F. and D.C.I. at 5 year with relatively good noise performance (15 e-)

• SOC recommended to implement C.C.I. capability• Work ongoing to further reduce C.C.I. noise.

Mitigating CTE losses: Charge Injection and Pre/Post Flash

Documents

Transcript of Mitigating CTE losses: Charge Injection and Pre/Post Flash