Cosmology with Distant Supernovae Where Next Richard Ellis

Cosmology with Distant Supernovae: Where Next? Richard Ellis, Caltech Zwicky SN Workshop, Carnegie Jan 17 2004

“Concordance Cosmology”: triumph or sham? Concordance is worrying: • DM 0. 27 0. 04 • B 0. 044 0. 004 • 0. 73 0. 04 (Bennett et al 2003) All 3 ingredients comparable in magnitude but only one component physically understood! 0: why this value and why acceleration now? 2 d. F

Efstathiou et al (2001) Joint analysis of CMB + 2 d. F data CMB alone CMB + 2 d. F Contrary to popular belief CMB alone does not convincingly indicate spatial flatness if is unknown CMB + 2 d. F confirms spatial flatness and non-zero independent of any supernova data WMAP+2 d. F/SDSS: Same idea, higher precision

Role of SNe: Direct method for verifying cosmic acceleration Remarkable conclusions demand remarkable evidence Where next in cosmological applications? • More of the same (Tonry et al 2003) • Better data (HST z<1, Knop et al 2003) • Higher redshift data (GOODS; Subaru) • Check systematics • Independent methods (e. g. SN II, Hamuy et al)

More of the same: Hi. Z team Tonry et al (2003) empty • 23 new If. A/Hi. Z SNe • but only 9 confirmed as Ia • 0. 34 <z < 1. 09 • 15 with z > 0. 7 (doubling #)

Better z < 1 HST data: SCP team Knop et al Ap J 598, 102 (2003) 11 new HST SNe 0. 36<z<0. 86 higher quality multi-color data enabling E(B-V) measures

Probing to higher z with HST: (HDF: Gilliland et al 1999, Riess et al 2001) SN 1997 ff: z = 1. 7 0. 1 GOODS SN Ia 2002 fw z=1. 3 (Riess et al 2003) ACS grism 15 ksec (-- SN Ia 1981 b) Color discrimination of SNIa/II based on the UV deficit of Ia’s

Bias in finding bluer SNe at high z? Possible systematics in GOODs program locating SNe Ia via ACS 850 LP measuring restframe B-band (and UV) with NICMOS F 110 W filter.

Future HST surveys (GOODS, COSMOS. . ) will only modestly increase z > 1 sample (20 -30 events)

Investigating Systematic Effects • Differential extinction – greater amounts of dust in high z host galaxies: mimics > 0 • SN properties may depend on enviroment e. g. galaxy type or mix (Hamuy et al 1996, 2000) • Evolutionary differences e. g. progenitor composition (Höfflich et al 1999)

Evolution? Residuals from best fit to SN Hubble diagram (SCP 1999) Low z 1 mag High z Constant scatter (allowing for obs. errors) with z provides a (weak) case against evolution which would otherwise have to be well-orchestrated with cosmic time.

Reddening? E(B-V) estimates for low & high z SNe in improved HST sample Knop et al SCP 2003

Morphology? Type-dependent SN Ia light curves B-V Type Hamuy et al AJ 120, 1479 (2000) m 15(B)

HST STIS Snapshot Program (Sullivan + RSE) Cycle 8+10 STIS 50 CCD (unfiltered) snapshot imaging (retrospective) • host galaxy morphology • precise SN location • slit arrangement for diagnostic Keck spectroscopy STIS imaging: 59 targets 5 not observed/failed 2 no host visible 52 classified hosts (P 99 42 + new) Keck ESI: 16 targets E(B-V) for 6 plus 24 low z SNe (Hamuy, Riess) Sullivan et al MN 340, 1057 (2003)

SCP Hubble Diagram by Host Galaxy Type • spheroidal • spiral • late/Irr Type N dispersion (flat) P( >0) Spheroidal 13 (15) 0. 167 0. 60 (0. 59) 97. 9 Spiral 23 (28) 0. 197 0. 58 (0. 58) 98. 6 Late/Irr 23 (26) 0. 265 0. 75 (0. 74) 99. 9 Small offset of high z spheroidals (<0. 01) from adopted SCP fit

Light curve “stretch” distributions at high/low z Unfortunately, the similar range in light curve “stretch” at low and high z means we cannot readily test for all possible systematic effects e. g. decline rate versus type as studied locally by Hamuy. Low z High z

Rest-frame color excess versus type (Sullivan et al) Type E(B-V) = (B-V)obs – (B-V)0, s Little extinction in high z SNe and sensible type-dependent trends MB (rest) = MB(spheroidal) + 0. 07 from Hubble diagram AV from 6 ESI spectra: 0. 06 -1. 0 mag Lack of Irregulars in the SN-selected sample c. f. HST-based z surveys

Progenitor studies • Spectroscopic evolution of selected high z SNe c. f. improved local templates (SN Factory) • Metallicity of progenitor? detailed UV spectra near maximum light (Nugent et al 1999) • Nature of Ia progenitor: rate at as a function of z in field (Pain et al) and in clusters (Gal-Yam)

Spectral Evolution of Distant SNe Ia Q: What is the best diagnostic spectroscopic correlation that should be tested for a modest high z sample (z=0. 5)

Nugent et al (1995): Spectral Sequence of SNe Ia R(Ca II) Synthetic & observed spectral sequence MB L R(Si II) blue/red Synthetic sequence reproduces trend via 7400 < Teff < 11000

R(Si II) versus v 10(Si II) (Hatano et al 2000) Do SNe. Ia form a one parameter sequence: can we verify a sequence at high z?

UV Opacity as Probe of SNIa Metallicity (Nugent et al 1999) Strong UV dependence expected from deflagration models when metallicity is varied in outermost C+O layers (Lenz et al 2000)

UV Trends in Nearby SNe Ia (STIS, Nugent) Can we explore these trends at high z and correlate with Hubble diagram?

CFHT Legacy Survey (2003 -2008) Deep Synoptic Survey Four 1 1 deg fields in ugriz 5 nights/lunation 5 months per accessible field 2000 SNe 0. 3 < z < 1 Megaprime Caltech’s role Spectral follow-up of 0. 4<z<0. 6 SNe Ia Tests on 0. 2<z<0. 4 SNe. II

The Need for Photometric Pre-Classification Nearby search Discovery Reference Difference CFHTLS SNe Ia from Sep 2003 • Hi-z SN spectra are much harder to take due to both their faintness & their separation from their host galaxy is comparable to the seeing. • Avoid wasting Keck time taking spectra of objects too close/wrong sub-type.

Photometric redshifts/typing for distant SNe New code SNphot-z pre-classes type, z & epoch prior to taking spectra: only practical for the CFHTLS multi-filter rolling search • Templates from Gilliland, Nugent & Phillips (1999) updated from Nugent et al. (2002). • Calculate color evolution as a function of epoch, z, type, extinction, stretch (Ia’s) in ugriz for all targets. Spectral templates created by homogenizing IUE and HST observations + some modeling to fill in the gaps. SN Ia template weekly for the first 7 weeks.

SN Photo-Z: Results Based on 3 epochs of photometry with only R & I data.

CFHT Legacy Survey: Progress • 17 SNIa 0. 25<z<0. 55 to correlate spectral dispersion with Hubble diagram residuals (in progress) • 3 SNII 0. 1<z<0. 4 to explore feasibility of EPM/Hamuy methods

Results: I - Extending Environmental Range Ref Unlike previous searches the CFHTLS SN search is finding SNe with very low % increases near the cores of bright galaxies, sampling a much broader range of environments. How do they differ? Disc Sub

Results: II - Correlating Spectral Features The large choice of CFHTLS SNe enables us to target for comparisons at same redshifts & epochs. • Two SNe Ia near peak brightness both with z= 0. 45. • Significant difference in Ca II H&K P-Cygni feature (split in 2003 fh, smooth in 2003 fg) • Significant UV flux differences. • Minor velocity shifts of the intermediate mass material (Si. II and SII).

Results: III - Dispersion in UV properties z 0 STIS z 0. 5 Keck Correlating metallicity/UV opacity with light curves is a major goal

Can Cosmic Acceleration be deduced from SN II? Hamuy & Pinto (2002) propose a new “empirical” correlation (0. 2 mag, 9% in distance) between the expansion velocity at the plateau phase and bolometric luminosity for Type IIs. If vindicated with more data, the Hubble diagram of SNII will provide a completely independent check of the cosmic acceleration using Keck QUEST will locate nearby SNIIs on plateau phase; expansion velocities will come from override time on 200 -inch to test this proposition

Expected Numbers of Supernovae v Type Ia SNE Use Rate from R. Pain, et al. (APJ 577, 120, 2002) v Type II SNE • Typically 2 mags fainter than Ia’s (Hamuy & Pinto APJ 566, L 63, 2002) • About twice as numerous per unit volume as Ia’s (Capellaro, et al. , AA 351, 459, 1999) v Estimate numbers of SNe’s for 1000 square degrees, 15 day time window Type Ia SNe’s Type II SNe’s Up to Z Peak m No/1000 sq deg 0. 05 17. 5 2 19. 5 6 0. 10 19. 0 12 21. 0 24 0. 20 20. 5 100 22. 5 200 0. 30 21. 5 300 23. 5 600 0. 40 22. 0 650 24. 0 1300

Keck example: SN 2001 kf z=0. 21 SNIIp (V=23. 0) Measuring the Fe II expansion is feasible at z 0. 3 in 2 -3 hours 10 -20 SNe. IIp free from systematics would confirm 0 at 3

Conclusions • Distant SN programs are entering new, more detailed phases utilising HST and high s/n spectroscopy to provide increased astrophysical data for each event global constraints on evolution & progenitor details. (exciting outcome whether acceleration supported or not) • First enhanced datasets tend to support the SCP conclusions (SN in field spheroidals confirm 0. 7 ) • CFHTLS will extend these SN Ia studies via spectral sequences based on metallicities/environment • Palomar/QUEST 2 will verify the utility of SNe II as cosmic probes: Keck may verify the acceleration! • SNAP/JDEM represents the logical endpoint of the program

Optical ( 36 CCD’s) = 0. 34 sq. deg. SNAP/JDEM – combines SNe Ia and weak lensing as a unique probe of dark energy 4 filters on each 10. 5 m pixel CCD IR (36 Hg. Cd. Te’s) = 0. 34 sq. deg. 1 filter on each 18 m pixel Hg. Cd. Te It should be called the Zwicky telescope! http: //snap. lbl. gov

Conclusions not significantly affected by stretch corrections

Distant SNe. Ia have similar spectra to local counterparts at same epoch

More SNe. IIp…

The current situation – all literature data Tonry et al (2003)

Reddening? SCP (1999): Intrinsic reddening determined from multicolor light curves: • insufficient precision for use on individual SN by SN basis, • zero point uncertain Provides case against overall relative reddening of high c. f. low z sample

Grey dust? E(B-R)/B Grain size ( m) Aguirre Ap J 525, 583 (2000): Grey dust requires larger grains with high metal content and may conflict with far IR background

Keck ESI Spectroscopic Program Keck II Echellette Spectroscopic Imager: R 25000 0. 3 -1 m long slit • emission line properties of host galaxy (correlation with HST morphology) • reddening estimate from H /H • variance in above from longslit data in good seeing

Simulated Results from SNAP

Host Galaxy Types Classification of P 99 sample of 42 into 3 broad types spheroidal/ intermediate/ late R-I from: • ESI (+LRIS) spectrum • HST STIS image • R-I color z

No dependence on projected radial distance

Type versus stretch Stretch versus radius

Detection efficiencies Computed adding fake SN (stars) on real images (galaxies) Set A Set C Set B Set D SN/galaxy relative brightness

Program so far… 17 Type Ia’s at 0. 25 < z < 0. 55 with an average exposure time 4 -5 * longer than what is normally taken during a high-z search program for a given supernova.

Determining High Redshift SN Rate 1 SNu = 1 SN per century per 1010 LB (sun) To estimate rate we require: • SN detection efficiency, i. e. control time t (z, L, ) • Volume and stellar luminosity probed at search limit • Large number of SNe Pain, Sullivan, RSE et al (2002) - old SCP search data • 38 SNe from SCP: 0. 25<z<0. 85 from 12 deg 2 <z> 0. 55 rate is 0. 58 0. 09 ( 0. 09) SNu 1. 53 0. 25 ( 0. 32) 10 -4 h 3 Mpc-3 yr-1

SN rates as a function of redshift (Sullivan et al 2000) SN II rate Various SF histories (Madau et al 1999) SN Ia rate =0. 3 Gyr SCP (Pain et al 2001) =3 Gyr z Must seek higher redshift SNe

Origin of SNe Ia in single degenerate C-O WD systems (Nomoto et al 1999) AGB with C+O core RG+He core WD + MS in common envelope WD + red giant Wind reduces rate Short time delay Significant time delay

Why is a non-zero cosmological constant worrying?

SN Photo-Z: Results - II Best fit z = 0. 96+/-0. 07: Observed z = 0. 979 Success rate is ~95% to 0. 1 in z - helpful in separating Ia & II targets.