HST Observations of Escaping LyC Radiation from Galaxies & weak AGN at 2.3<∼z<
∼5:
(How) Did they Reionize the Universe, and what JWST must do next.
Rogier Windhorst (ASU) — JWST Interdisciplinary Scientist
Brent Smith, S. Cohen, R. Jansen, L. Jiang, M. Dijkstra, A. Inoue,
A. Koekemoer, R. Bielby, J. MacKenty, R. O’Connell, & J. Silk
Talk at the Lagrange Institute Conference on “Cosmology and First Light”
Monday December 7, 2015; Institute d’Astrophysique, Paris, France
Outline
• (1) HST WFC3 Data & Spectroscopic Sample Selection
• (2) WFC3 & ACS Lyman Continuum Stacking, Systematics, & Fluxes
• (3) Stacked Lyman-Continuum and UV-Continuum Light-Profiles
• (4) SED-fitting & Dust-distribution AV (z)
• (5) LyC Escape Fractions vs. z for Faint Galaxies & Weak AGN
• (6) What critical aspects will JWST add to LyC Escape studies?
• (7) Summary and ConclusionsSponsored by NASA/HST & JWST
Talk is on: http://www.asu.edu/clas/hst/www/jwst/jwsttalks/paris15_jwstlyc.pdf
In what follows,remember that objectsemitting two-sidedand equally brightrelativistic jets, orescaping LyC radiationmay look different,
depending, e.g. onviewing angle, dust,and scattering proper-ties of the medium.
(1a) Hubble WFC3 Data: The Early Release Science (ERS) field.
10 filters with HST/WFC3 & ACS reaching AB=26.5-27.0 mag (10-σ)over 40 arcmin2 at 0.07–0.15” FWHM from 0.2–1.7µm (UVUBVizYJH).
(JWST adds 0.05–0.2” FWHM imaging to AB≃31.5 mag (1 nJy) at 1–5µm + 0.2–1.2” FWHM at 5–29µm, tracing young+old SEDs & dust).
[LEFT] Composite rest-frame far-UV spectra of:SDSS QSOs at z≃1.3 (van den Berk et al. 2001);LBGs at z≃3 (Shapley et al. 2003);LBGs at z≃2–4 (Bielby et al. 2013, Lyα emitters, & absorbers).
• WFC3/UVIS F225W, F275W, F336W, and ACS/WFC F435W filters cancapture LyC (λ<912A) at z≥2.26, z≥2.47, z≥3.08, and z≥4.35.
• Lower z-bounds: no λ> 912A below filter’s red-edge (≡0.5% of peak).
[RIGHT] Total observed throughput curves, designed to maximize through-put and minimize red-leak, which is <∼0.6% of actual LyC signal.
• Filter red-leak wing (λ>∼ 3648A) is <∼3×10−5 of peak transmission.
[LEFT] Cen & Kimm (2015): PDFs of mean fesc over “Nstack” objects:high-mass (top) & low-mass (bottom) at z=4 (left) & z=6 (right).
• Mean fesc from weighted number of photons mimics SED stacking ofgalaxy LyC data with true mean fesc listed. ERS has Nstack=11–37.
[RIGHT] Inoue+ (2014): IGM transmission models for fesc -calculations:Red is median and grey 68% range, based on MC simulations of TIGM(z).
• Uses updated absorber function+available data on Lyα forest, DampedLyman Alpha (DLA) & Lyman Limit Systems (LLS) mean-free paths.
• We do stack z∼5 samples: (z∼5) AGN LyC ∼1m brighter than galaxies.
(1b) Hubble WFC3 ERS — Spectroscopic Sample Selection
Comparison of redshift reliability (spectrum quality) assessments, from best(0.0) to poorest (2.0), by five co-authors [BS, RAW, SHC, RAJ, and LJ]:
• Measuring LyC escape fractions of fesc ≃6.0% at >∼3σ requires low
interloper fraction (Siana+ 2015; Vanzella+ 2015).
• Mask-out all interlopers from 10-band ERS mosaics to AB<∼27.5 mag.
• Use all VLT, Keck, & HST grism spectra to get most reliable samples:
• “Gold” sample: highest fidelity (grades=0–0.63): zsp’s very likely correct.
• “Silver” sample: next highest fidelity (0.64–1.33), with z’s likely correct.
(1b) Hubble WFC3 ERS — Spectroscopic Sample Selection
23 25 27
mAB
1500A[mag]
0.0
0.1
0.2
0.3
0.4Fra
ction
[a] All Galaxies
−24 −22 −20
MAB
1500A[mag]
N=64
N=131
23 25 27
mAB
1500A[mag]
[b] Galaxies (no AGN)
−24 −22 −20
MAB
1500A[mag]
N=50
N=114
23 25 27
mAB
1500A[mag]
[c] Galaxies with AGN
−24 −22 −20
MAB
1500A[mag]
0.0
0.1
0.2
0.3
0.4N=14
N=17
Apparent and absolute magnitude distributions (restframe 1550A) of the“Gold” (highly reliable z) and “Gold+Silver” (reliable z’s) samples:
• The blue dotted curve indicates the faint-end power-law slope of 0.16dex/mag of the galaxy number counts of Windhorst+ (2011).
• Sample incompleteness for AB>∼24, or MAB (1650)>∼–21 mag.
• LyC AB-fluxes & fesc -values only valid for these selected luminosities.
• Galaxies with weak AGN have same N(MAB) as galaxies without AGN.
(2) WFC3 & ACS Lyman Continuum Stacking, Systematics, & Fluxes
F225W F275W F336WThe first & hardest part was to get the WFC3 astrometry right:
• Pre-flight 2009 ERS geo-distortion had <∼0′′.45 offsets at image borderscompared to GOODS v2.0 (Windhorst et al. 2011 ApJS, 193, 27).
• In-flight 2013 geo-distortion correction yielded excellent registration ofall WFC3/UVIS tiles to the ACS F435W mosaics (Kozhurina et al. 2014).
• Compared to GOODS, all offsets are now <∼0′′.02 ±0.06 (rms) in all LyCfilters (Smith et al. 2015) — this no longer blurs any LyC signal!
• Any LyC signal can now be measured and stacked, including removal of allforeground interlopers (AB<∼27.5), and measurement of LyC light-profiles.
Residual sky-background levels in the drizzled WFC3/UVIS ERS mosaics:
• Black lines: Best fit to the 2009 ERS v0.7 mosaics of Windhorst et al.(2011), which used pre-flight thermal vacuum flat-fields.
• Red lines: Current mosaics (ERS v2.0; Smith et al. 2015), using bestavailable on-orbit calibrations.
• Global residual sky-background levels (in ADU/sec) remaining after driz-zling the ERS mosaics are ∼30.29, 29.99, and 28.15 mag arcsec−2 .
• Removed in 3 stages: globally during drizzling (zodi≃25.5 mag/”2),locally before stacking, and again locally after stacking (to do photometry).
This is absolutely critical for optimal LyC stacking.
• Final 71×71 pix (6′′.39×6′′.39) LyC stacks allow residual local sky-subtraction to <∼32.3 mag arcsec−2 .
(2) WFC3 & ACS Lyman Continuum Stacking, Systematics, & Fluxes
med=1.59e-05avg=1.69e-05mod=-4.87e-05
med=3.64e-06avg=4.47e-06mod=-6.34e-05
med=6.47e-06avg=6.00e-06mod=-6.09e-05
med=1.25e-05avg=1.39e-05mod=-4.19e-05
med=1.24e-05avg=1.60e-05mod=-1.54e-04 med=2.20e-06
avg=3.52e-06mod=-6.90e-05
med=1.17e-05avg=1.26e-05mod=-6.00e-05
med=1.27e-05avg=1.37e-05mod=-5.37e-05
med=4.42e-06avg=4.25e-06mod=-5.51e-05
stack1 avgtot =9.39e-06 medtot =8.91e-06modetot =-5.73e-05 σavg=5.02e-06
med=1.66e-05avg=1.61e-05mod=-6.35e-05
med=1.79e-05avg=1.85e-05mod=-6.08e-05
med=5.17e-06avg=4.54e-06mod=-6.87e-05
med=3.64e-06avg=4.32e-06mod=-8.02e-05
med=2.00e-05avg=2.27e-05mod=-1.65e-04 med=3.91e-06
avg=2.58e-06mod=-8.96e-05
med=2.53e-07avg=-2.95e-07mod=-8.27e-05
med=5.51e-06avg=4.76e-06mod=-7.43e-05
med=-3.68e-06avg=-3.67e-06mod=-8.95e-05
stack2 avgtot =6.00e-06 medtot =6.50e-06modetot =-7.58e-05 σavg=7.15e-06
med=4.57e-05avg=4.59e-05mod=-8.40e-05
med=3.53e-05avg=3.45e-05mod=-9.33e-05
med=4.70e-05avg=4.97e-05mod=-8.60e-05
med=4.11e-05avg=3.70e-05mod=-1.14e-04
med=5.36e-05avg=5.67e-05mod=-2.57e-04 med=4.41e-05
avg=4.66e-05mod=-8.54e-05
med=3.12e-05avg=3.36e-05mod=-1.06e-04
med=3.35e-05avg=3.20e-05mod=-1.12e-04
med=4.23e-05avg=4.53e-05mod=-5.35e-05
stack3 avgtot =4.20e-05 medtot =4.12e-05modetot =-8.69e-05 σavg=6.70e-06
med=1.70e-03avg=1.69e-03mod=3.71e-06
med=1.58e-03avg=1.62e-03mod=-1.89e-04
med=1.51e-03avg=1.50e-03mod=-1.25e-04
med=1.69e-03avg=1.71e-03mod=1.71e-04
med=1.62e-03avg=1.63e-03mod=-1.99e-03 med=1.66e-03
avg=1.60e-03mod=5.36e-05
med=1.52e-03avg=1.51e-03mod=-1.78e-04
med=1.59e-03avg=1.60e-03mod=2.62e-05
med=1.59e-03avg=1.61e-03mod=-1.02e-04
stack4 avgtot =1.61e-03 medtot =1.61e-03modetot =-4.04e-05 σavg=6.97e-05
med=3.19e-03avg=3.15e-03mod=1.08e-03
med=3.22e-03avg=3.18e-03mod=1.06e-03
med=3.26e-03avg=3.33e-03mod=1.65e-03
med=3.23e-03avg=3.21e-03mod=1.31e-03
med=4.44e-03avg=4.60e-03mod=1.81e-03
med=3.02e-03avg=3.05e-03mod=1.01e-03
med=3.13e-03avg=3.19e-03mod=1.06e-03
med=3.33e-03avg=3.30e-03mod=1.34e-03
med=3.02e-03avg=3.00e-03mod=1.10e-03
stack1uv avgtot =3.18e-03 medtot =3.17e-03modetot =1.17e-03 σavg=1.04e-04
med=3.06e-03avg=3.00e-03mod=6.83e-04
med=3.10e-03avg=3.12e-03mod=9.85e-04
med=3.19e-03avg=3.19e-03mod=9.68e-04
med=3.17e-03avg=3.24e-03mod=1.07e-03
med=4.44e-03avg=4.49e-03mod=2.26e-03
med=3.09e-03avg=3.10e-03mod=9.06e-04
med=3.13e-03avg=3.10e-03mod=1.08e-03
med=3.09e-03avg=3.09e-03mod=7.97e-04
med=3.19e-03avg=3.27e-03mod=1.06e-03
stack2uv avgtot =3.16e-03 medtot =3.14e-03modetot =9.47e-04 σavg=8.55e-05
med=3.25e-03avg=3.24e-03mod=1.44e-03
med=3.41e-03avg=3.39e-03mod=1.67e-03
med=3.61e-03avg=3.56e-03mod=2.19e-03
med=3.25e-03avg=3.26e-03mod=1.66e-03
med=4.06e-03avg=4.06e-03mod=2.44e-03
med=3.35e-03avg=3.29e-03mod=1.70e-03
med=3.25e-03avg=3.19e-03mod=1.61e-03
med=3.33e-03avg=3.36e-03mod=1.60e-03
med=3.48e-03avg=3.49e-03mod=1.88e-03
stack3uv avgtot =3.35e-03 medtot =3.35e-03modetot =1.73e-03 σavg=1.18e-04
med=2.85e-03avg=2.84e-03mod=1.44e-03
med=2.85e-03avg=2.81e-03mod=1.15e-03
med=2.91e-03avg=2.87e-03mod=1.27e-03
med=2.69e-03avg=2.70e-03mod=1.58e-03
med=3.04e-03avg=3.01e-03mod=1.38e-03
med=2.70e-03avg=2.67e-03mod=1.03e-03
med=2.87e-03avg=2.87e-03mod=1.52e-03
med=2.69e-03avg=2.73e-03mod=1.26e-03
med=2.46e-03avg=2.47e-03mod=9.84e-04
stack4uv avgtot =2.76e-03 medtot =2.76e-03modetot =1.29e-03 σavg=1.26e-04
“Tic-tac-toe” sky-background analysis of 71×71 pixel (6′′.39×6′′.39) stacks:
LyC [left 4 panels] and UVC [right 4 panels].
• Sky-background subtracted in 3 stages: more globally upon drizzling,locally before stacking, and locally before final photometry.
• Residual UV sky-gradients fainter than ∼32.3 mag arcsec−2 across pho-tometric apertures.
• This is fainter than the LyC SB-signal where this can be measured, andmay impose a (fundamental?) limit to how many images can be stacked.
<~ <~z2.3 6 RAJ
a) F225W
g)
c) d) e) f)F275Wb) F336W F435W
h) i) j) k) l)
LyC
LyCLyC
LyC
Gold
Gold+Silver
N=50
Gold+Silver (smoothed)
Gold (smoothed) N=50
N=114
N=114F225W F275W F336W F435W
Galaxies without AGN,
z=2.26–2.47 z=2.47–3.08 z=3.08–4.35 z=4.35–5.5 WEIGHTED ALL: z=2.26–5.5.
[Top Row ]: All galaxies in combined Gold Galaxy sample: N=50;
[Bottom Row ]: All galaxies in combined Gold+Silver sample: N=114.
[Right 2×2 panels]: Weighted “stack-of-stacks” over all 4 LyC filters: bestvisualizes LyC of galaxies at z≃2.3–5.5. Formal detection S/N -ratios:
>∼7σ (∼√50×1.0σ above sky), >∼13σ (∼√
114×1.2σ above sky).
• Equivalent to 22–228 orbit UV stacks with HST, respectively.
Circles: r=8 (0′′.72), 13 pix (1′′.17), centered on the UVC emission.
(3) Stacked LyC Light-Profiles, & Weighted “Stack-of-Stacks”
Combined LyC fi, 2.3 � z � 5.8
HQ
Sp
ecH
Q+
IQ S
pec
N=50c) Galaxies (no AGN) d) Galaxies (no AGN) smoothed
N=131e) All Galaxies N=17f) Galaxies with AGN N=114g) Galaxies (no AGN) h) Galaxies (no AGN) smoothed
N=14b) Galaxies with AGNN=64a) All Galaxies
All Objects Weak AGN Galaxies w/o AGN Smoothed Galaxies
[Top Row ]: All Gold sample (z=2.3–5): 50 Galaxies + 14 weak AGN;
[Bottom Row ]: All Gold+Silver sample (z=2.3–5): 114 Gxys + 17 AGN.
The faint LyC emission has a very flat SB-distribution with radius:
• Not centrally concentrated, with few clear sight-lines per galaxy.
• On average escapes along few random sight-lines through a porous ISM?
• Likeliest escape paths may be somewhat offset from galaxy center.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Isophotal Semi-Major Axis (arcsec)
22
24
26
28
30
32
Surfac
e Brig
htne
ss (m
agAB arcse
c−2)
F435W PSFF275W PSF
Model (z=2.65)
F225W (<z>=2.38)
F275W (<z>=2.68)
F336W (<z>=3.47)
F435W (<z>=5.02)
[Top Curves]: radial SB-profiles of stacked non-ionizing UVC (solid).
[Bottom Curves]: Radial SB-profiles of stacked LyC signal (dashed):
• All LyC SB-profiles are extended compared to the PSFs (dotted).
• Horizontal black dashed line is the 1σ SB-limit of ∼32 mag arcsec−2 .
Light-blue dot-dash: Dijkstra’s z=2.68 UVC-scattering model with ISMporosity + escaping LyC increasing as: fcov(r)=N exp{−(r/10 kpc)x}.
(4) Spectral Energy Distribution (SED)-fitting & Dust (AV )-distribution
1 2 3 4 5 6 7 8 92 10
0
1
2
3
missing for
most SMGs
[LEFT]: Best-fit AV from 10-band SEDs for all ERS galaxies (black dots).
Circles: galaxies; Asterisks: AGN at: z=2.37, z=2.68, z=3.45, z=5.1.
[RIGHT]: Adopted distributions N(AV ) for total Gold + Silver LyC samples:
Median AV increases from ∼0.2m at z=5.1–3.5 to ∼0.6m at z=2.67–2.37.
Gxy+Agn selected at zsp=3.45–5.1 miss ∼45% of dusty (AV>∼1) objects.
(5) LyC Escape Fractions vs. z for Faint Galaxies & Weak AGN
10−1 100 101 102 103
〈fabsesc 〉 (%)
10−5
10−4
10−3
Probab
ility
Avg Val
Med Val
ML Val
Avg σ
ML σ
F225W (〈z〉=2.38)
F275W (〈z〉=2.68)
F336W (〈z〉=3.47)
F435W (〈z〉=5.02)
PDF of absolute fesc -values (Inoue+ 2014 Monte Carlo), folding LyCfluxes ± 1σ errors through 109 random LOS of IGM transmission.
• Filled triangles indicate the resulting modal, and circles the average fesc
-values in each PDF. Tick-marks show the ±1σ MC-range.
1 2 3 4 5 6 7 8 9
1 + z
0.1
1.0
10.0
100.0
〈fabs
esc〉(%
)
〈fabs
esc〉≃ ∆
2·tanh([(1+z)−(1+z0)]/δ)+f0
≃ F0 ·(1 + z)n
13 11 9 7 5 3 1Age of the Universe (Gyr)
Absolute fesc -z: Published + ERS Gold & Gold+Silver samples.
Single power law: fesc ≃(0.006±0.002)·(1+z)1.5±0.7 does not fit well.
1 2 3 4 5 6 7 8 9
1 + z
0.1
1.0
10.0
100.0
〈fabs
esc〉(%
)
〈fabs
esc〉≃ ∆
2·tanh([(1+z)−(1+z0)]/δ)+f0
≃ F0 ·(1 + z)n
13 11 9 7 5 3 1Age of the Universe (Gyr)
Absolute fesc -z: Published + ERS Gold & Gold+Silver samples.
Single power law: fesc ≃(0.006±0.002)·(1+z)1.5±0.7 does not fit well.
Simple tanh[log(1+z)] captures more sudden fesc -increase at z>∼2.5–3.
1 2 3 4 5 6 7 8 9
1 + z
0.1
1.0
10.0
100.0
〈fabs
esc〉(%
)
〈fabs
esc〉≃ ∆
2·tanh([(1+z)−(1+z0)]/δ)+f0
≃ F0 ·(1 + z)n
13 11 9 7 5 3 1Age of the Universe (Gyr)
Absolute fesc -z: Published + ERS Gold & Gold+Silver samples.
Power-laws: fesc ≃(0.006±0.002)·(1+z)1.5±0.7 do not fit well.
Simple tanh[log(1+z)] captures more sudden fesc -increase at z>∼2.5–3.
• fesc of galaxies just high enough to cause reionization at z>∼3.
• LyC of 17 weak AGN in ERS ∼1.0 mag brighter than for galaxies.
• Weak AGN may dominate and maintain reionization at z<∼3.
(6) What critical aspects will JWST add to HST’s LyC Escape studies?
JWST FGS+NIRCam: R≃150, 0.8–5.0µm grism spectra to AB<∼28–29:
• Larger, fainter SED+zspec -samples of LyC candidates in HST UV fields.
NIRSpec: JWST’s short-wavelength (λ≃1–5.0µm) spectrograph:
• 100’s of simultaneous faint-object spectra of LyC candidates to AB<∼27.5.
Concentrate on the most dusty (far-IR selected) AV>∼1 objects at z>∼2.3!
(7) Summary and Conclusions
(1) HST can measure LyC for galaxies + weak AGN at z≃2.26–5.• WFC3 and ACS filters designed with low-enough redleak to enable this.
• Samples of sufficient size (N=11–114) need to be stacked to see LyC signal, preferably many dozen.
• Deepest 10-band images at HST resolution critical to mask-out all foreground interlopers to AB<
∼27.5 mag.
• Careful spectroscopic redshift selection critical for reliable samples: Must correct for MAB and AV -biases.
(2) LyC signal detected in sub-samples of N=11–37 objects at z=2.26–5.• Detections of AB(LyC) generally better than >
∼3–4σ (AB≃29.5–30.5 mag).
• Weak AGN have ∼1.0 mag brighter AB(LyC), but are 4–10× less numerous than galaxies.
• Stacked LyC SB-profiles are on average much flatter than the UV-continuum Sersic-profile.
• LyC may escape along few random sight-lines, offset from galaxy center: Non-Sersic, ISM-porosity increases with r?
(3) fesc (z) may show rapid “tanh[log(1+z)]-like” increase at z>∼2.5.
• Dust-corrected SED-fits and MC simulations essential to interpret this sudden drop in fesc (z).
• Best-fit 10-band ERS SEDs suggests AV increases from z∼6 to z≃2.3.
• Spectroscopic selection at z=2.37–2.68 follows field galaxy AV , but at z=3.45–5.1 misses ∼45% of dusty objects.
• Accumulating HI+AV (t) may shut down fesc (z<
∼3), explaining a sudden fesc -decrease at z>
∼3:
• fesc (galaxies) just high enough to cause reionization at z>∼3.
• (Galaxies with) weak AGN may dominate & maintain reionization at z<∼3.
• JWST NIRISS + NIRSpec spectra for (dusty) LyC objects to AB<∼28–29.
SPARE CHARTS
References and other sources of material shown:
http://www.jwst.nasa.gov/ & http://www.stsci.edu/jwst/
http://ircamera.as.arizona.edu/nircam/
http://ircamera.as.arizona.edu/MIRI/
http://www.stsci.edu/jwst/instruments/nirspec/
http://www.stsci.edu/jwst/instruments/fgs
Bielby, R., et al. 2013, MNRAS 430, 425
Bridge, C., et al. 2010, ApJ 720, 465
Bouwens, R. et al. 2015, ApJ, 803, 35
Gardner, J. P., et al. 2006, Space Science Reviews, 123, 485–606
Hathi, N. P., et al. 2010, ApJ, 720, 1708
Hathi, N. P., et al. 2013, ApJ, 765, 88
Inoue, A. K., Shimizu, I., Iwata, I., & Tanaka, M. 2014, MNRAS 442, 1805
Shapley, A., et al. 2003, ApJ 588, 65
Siana, B., et al. 2015, ApJ, 804, 15
Smith, B., Windhorst, R. A., Jansen, R. A., Cohen, S. H., Jiang, L., Dijkstra, M., Koekemoer, A. M., Bielby, R., Inoue, A.K., MacKenty, J. W., O’Connell, R. W., & Silk, J. I. 2015, AJ, in prep. Telescope Wide Field Camera 3 Observations of
Escaping Lyman Continuum Radiation in Galaxies and Weak AGN at Redshifts z≃2.3–5.”
Vanden Berk, D. E., et al. 2001, AJ 122, 549
Vanzella, E., et al. 2015, A&A, 576, A116
Windhorst, R., et al. 2008, Advances in Space Research, 41, 1965
Windhorst, R., Cohen, S. H., Hathi, N. P., et al. 2011, ApJS, 193, 27
Yan, H., et al.2010, Res. in Astr. & Astrop., 10, 867
40
12 1813 19 29
1110
3734
9461
5 6 8
8
1
37
12
34
6 (53W002)5
61
113
94
113
=2.390 z=2.386
19
10
13
29
40 60
=2.393 =2.397zz=2.388z
1
11
z
18
60
a b
c
F410M(Ly )
d
e f
F450W(B)
F606W(V) F814W(I)
B/I1/4B - r - PSF
α
EN
CO
CO
CO
CO
(Left): WFPC2 BVI + F410M (Lyα ) on 53W002 + surrounding group of17 z=2.39 Lyα candidates (Pascarelle et al. 1996, Nature, 383, 45).
(Right): HST/PC of radio galaxy 53W002 at z=2.390 (Windhorst et al.
1998, ApJL): stellar r1/4-law + Lyα & blue continuum AGN-cloud.
⇒ May need to measure escaping LyC outside (dusty) LBGs with outflows.
JWST can measure AGN hosts <∼6 mag fainter in restframe UV-opt to z<∼15.
Example of SED fits used for fesc (MC) etc, using λ>∼1216 A and z≡zspec.
RAJ
a) b) c) d)
e) f) g) h)
Stacking Tests, Gold sample
F225W F275W F336W F435W
i) j) k) l)
original
rotation tests
split halves − subset 1
n) o) p)m) split halves − subset 2
blank sky tests r) s) t)q)
z=2.37, z=2.68, z=3.45,z=5.1 Gold stacks;
Same Gold stacks af-ter random 90◦ rotation;
First independent datahalves Gold stacks;
Second independentdata halves Gold stacks;
Random sky-stacksto verify null-signal.
RAJ
a) b) c) d)
e) f) g) h)
Stacking Tests, Gold+Silver sample
F225W F275W F336W F435W
i) j) k) l)
original
rotation tests
split halves − subset 1
n) o) p)m) split halves − subset 2
blank sky tests r) s) t)q)
z=2.37, z=2.68, z=3.45,z=5.1 Silver stacks;
Same Silver stacks afterrandom 90◦ rotation;
First independent datahalves Silver stacks;
Second independentdata halves Silver stacks;
Random sky-stacksto verify null-signal.
a) b)
Detector location of “high-CTE” and “low-CTE” sub-samples: [LEFT]:WFC3/UVIS F225W, F275W, F336W. [RIGHT]: ACS/WFC F435W.
Green regions are closest to parallel read-out amplifier. Red regions arefurthest from amplifiers, and may suffer more from CTE-degradation.
• Filled circles show marginal LyC signal in individual objects:
• These are fairly uniformly distributed across individual CCDs.
Average stacked LyC diff: ∆(Lower-CTE–High-CTE) ≃ 0.5±0.35 mag.
=⇒ Less than four months after WFC3’s launch, CTE-induced systematicsare not yet larger than the random errors in the LyC signal.