1. The First Constraint on the Reionization from GRBs: the Case of GRB 050904 Tomonori Totani (Kyoto) Nobuyuki Kawai, George Kosugi, Kentaro Aoki, Toru Yamada, Masanori Iye, Kouji Ohta, Takashi Hattori To Appear in PASJ (astro-ph/0512154) STScI workshnop “ The End of the Dark Ages ” 2006, March 13-15

2. Why GRBs? a Unique Reionization Probe Brightness Brightness detectable to very high redshift of z>~10 detectable to very high redshift of z>~10 much brighter than LAEs/LBGs and even than QSOs if observed quickly enough much brighter than LAEs/LBGs and even than QSOs if observed quickly enough less biased, probing more normal IGM less biased, probing more normal IGM most likely, simply tracing SFR most likely, simply tracing SFR QSOs and bright galaxies normally in strongly biased, clustered regions QSOs and bright galaxies normally in strongly biased, clustered regions host galaxy luminosity irrelevant for GRB detection host galaxy luminosity irrelevant for GRB detection No proximity effect that is a major problem for quasars No proximity effect that is a major problem for quasars Clean spectrum Clean spectrum almost single power-low SED, in contrast to QSOs or LAEs almost single power-low SED, in contrast to QSOs or LAEs GRBs are the best astronomical source for searching the red damping wing of HI absorption in IGM

3. Probing the Reionization by the Red Damping wing GP trough in QSOs only gives a lower limit of x HI >~10 -3 GP trough in QSOs only gives a lower limit of x HI >~10 -3 The red damping wing, if detected, would give a precise measurement of x HI The red damping wing, if detected, would give a precise measurement of x HI since optically thin, it is insensitive to any clumpiness of IGM since optically thin, it is insensitive to any clumpiness of IGM However, this method is problematic in QSOs, since: However, this method is problematic in QSOs, since: proximity effect proximity effect complicated intrinsic spectra complicated intrinsic spectra GRBs are an ideal object to search this red damping wing! (Milarda-Escude 1998) GRBs are an ideal object to search this red damping wing! (Milarda-Escude 1998) wavelength flux 1215 (1+z) Å red damping wing GP trough

4. GRB 050904 @ z=6.3 Swift Detection Swift Detection Cusumano et al. ’ 05 Cusumano et al. ’ 05 photometric colors of optical afterglows indicating z~6 photometric colors of optical afterglows indicating z~6 Haislip et al. ’ 05, Price et al. ’ 05, Tagliaferri et al. ’ 05 Haislip et al. ’ 05, Price et al. ’ 05, Tagliaferri et al. ’ 05 optical spectroscopic observation by Subaru optical spectroscopic observation by Subaru 3.4 days after the burst, z ’ = 23.7 3.4 days after the burst, z ’ = 23.7 confirmed z=6.3 by: confirmed z=6.3 by: clear Lyman break clear Lyman break metal absroption line systems at z=6.295 ± 0.002 metal absroption line systems at z=6.295 ± 0.002 Signature of the Damping Wing! Signature of the Damping Wing! Haislip et al. ‘ 05 GRB 050904, 3.4 days (Kawai et al. ’ 06)

5. Analysis of the Red Damping Wing to Constrain the Reionization two possibilities for the origin of the damping wing: two possibilities for the origin of the damping wing: A damped Lyα system (DLA) associated with the host galaxy A damped Lyα system (DLA) associated with the host galaxy DLAs often found in GRB afterglows DLAs often found in GRB afterglows log N HI (cm -2 ) ~ 21-22 log N HI (cm -2 ) ~ 21-22 IGM neutral hydrogen (damping wing of GP trough) IGM neutral hydrogen (damping wing of GP trough) If detected, it would give a crucial information on x HI If detected, it would give a crucial information on x HI GRB 030323 at z=3.372 (Vreeswijk et al. 2004)

6. Model Fitting Intrinsic GRB afterglow spectrum Intrinsic GRB afterglow spectrum β= - 1.25±0.25 power-law (F ν ∝ ν － β ) β= - 1.25±0.25 power-law (F ν ∝ ν － β ) ~0.5 day observations by Haislip et al., Tagriaferri et al. ~0.5 day observations by Haislip et al., Tagriaferri et al. spectral change by cooling frequency break passage is unlikely spectral change by cooling frequency break passage is unlikely no evidence for extinction  assume A V =0 at host (check later) no evidence for extinction  assume A V =0 at host (check later) Galactic Extinction: E(B-V) = 0.066 mag Galactic Extinction: E(B-V) = 0.066 mag Absorptions by DLA and IGM Absorptions by DLA and IGM 4 model parameters: z DLA, N HI, z IGM,u, x HI 4 model parameters: z DLA, N HI, z IGM,u, x HI DLA: τ= N HI σ[(1+z)ν obs ] DLA: τ= N HI σ[(1+z)ν obs ] damping width >> reasonable velocity dispersion ( > reasonable velocity dispersion (<~200 km/s) IGM: Formula of Miralda-Escude 1998 IGM: Formula of Miralda-Escude 1998 uniform IGM distributed from z IGM,l to z IGM,u uniform IGM distributed from z IGM,l to z IGM,u z IGM,l = 6.0 assumed in the baseline model z IGM,l = 6.0 assumed in the baseline model

7. redshift range of IGM contributing the damping wing at z~6.3 more than 80% (half) of IGM optical depth is by IGM at z > 6 (6.2) more than 80% (half) of IGM optical depth is by IGM at z > 6 (6.2) (z IGM,u = 6.3)

8. z DLA versus z metal z metal = 6.295 ± 0.002 (no other redshift around z~6.3) z metal = 6.295 ± 0.002 (no other redshift around z~6.3) fine-structure Si* detected  ~pc scale from the burster fine-structure Si* detected  ~pc scale from the burster z DLA vs. z metal ? z DLA vs. z metal ? z DLA = z metal : straightforward, the baseline model z DLA = z metal : straightforward, the baseline model [S/H] ~ -1.3, [Si/H] ~ -2.9 for log N HI ~ 21.6 [S/H] ~ -1.3, [Si/H] ~ -2.9 for log N HI ~ 21.6 (large Si depletion is not rare in GRB afterglows) (large Si depletion is not rare in GRB afterglows) z DLA = z host, but different from z metal z DLA = z host, but different from z metal host galaxy must be very low metallicity of [Si/H] < -3... likely? host galaxy must be very low metallicity of [Si/H] < -3... likely? | z DLA – z host | < ~ 0.005 (~200 km/s) | z DLA – z host | < ~ 0.005 (~200 km/s) z metal – z host < ~ 0.07 (~3,000 km/s, by accelerated metal absorbing shell around the GRB) z metal – z host < ~ 0.07 (~3,000 km/s, by accelerated metal absorbing shell around the GRB) -0.005 < z DLA – z metal < +0.07 -0.005 < z DLA – z metal < +0.07 Mirabal+ ‘ 03 Vreeswijk et al. ‘ 04

9. z IGM,u versus z metal z IGM,u versus z metal z IGM,u vs. z metal ? z IGM,u vs. z metal ? z IGM,u vs. z host z IGM,u vs. z host ionized bubble size: z IGM,u – z host ~ 0.02 (LAEs), 0.003 (theory for typical GRBs at z~6; Barkana & Loeb 2004) ionized bubble size: z IGM,u – z host ~ 0.02 (LAEs), 0.003 (theory for typical GRBs at z~6; Barkana & Loeb 2004) IGM gas infall to GRB host galaxies: negligible (Δz < 0.001) IGM gas infall to GRB host galaxies: negligible (Δz < 0.001) -0.02 < z IGM,u – z metal < +0.07 -0.02 < z IGM,u – z metal < +0.07 +0.07 comes from the possible accelerated shell for z metal +0.07 comes from the possible accelerated shell for z metal

10. Fitting Results: DLA? IGM? (1) Totani et al. 2005 DLA model with z=6.295, log N HI = 21.62

11. Fitting Results: DLA? IGM? (2) Both the DLA and IGM models can explain the damping wing (degeneracy!) Both the DLA and IGM models can explain the damping wing (degeneracy!) marginally z DLA =z metal allowed for DLA marginally z DLA =z metal allowed for DLA If IGM, z metal must be blueshifted by about 3,000 km/s If IGM, z metal must be blueshifted by about 3,000 km/s possible in GRBs! possible in GRBs! First detection of almost neutral IGM!? z metal = 6.295±0.002

12. Discrimination by LyβFeature The degeneracy can be broken by Lyβ profile! The degeneracy can be broken by Lyβ profile! The IGM model inconsistent with the data The IGM model inconsistent with the data upper limit z IGM,u < 6.314 upper limit z IGM,u < 6.314 DLA is dominant under this constraint DLA is dominant under this constraint the plausible model: z DLA =z meteal = 6.295 the plausible model: z DLA =z meteal = 6.295 x HI = 0 consistent with the data x HI = 0 consistent with the data but x HI =1 with z IGM,u =6.295 would affect the damping wing but x HI =1 with z IGM,u =6.295 would affect the damping wing DLA (z DLA =6.295) IGM (z IGM,u =6.36)

13. Constraint on x HI ? What x HI is preferred, when N HI is treated as a free parameter without any prior? What x HI is preferred, when N HI is treated as a free parameter without any prior? z IGM,u = z DLA = 6.295 best fit x HI = 0.00 ± 0.17 best fit x HI = 0.00 ± 0.17 x HI < 0.60 (95% C.L.) x HI < 0.60 (95% C.L.) The first upper limit on x HI at z>6! The first upper limit on x HI at z>6! zero level: best fit with x HI =0 solid: best fit with x HI =1 The data obey to Gaussian well

14. Uncertainty Check check done for spectral index, dust in host galaxy, redshift parameters, weak absorption lines, and time variability check done for spectral index, dust in host galaxy, redshift parameters, weak absorption lines, and time variability Totani et al. 2006, PASJ in press

15. Comparison with other constraints The constraint of x HI <0.6 at z=6.3 is consistent with those derived from QSO (HII region size etc.) and LAE statistics The constraint of x HI <0.6 at z=6.3 is consistent with those derived from QSO (HII region size etc.) and LAE statistics The strength of the GRB constraint : The strength of the GRB constraint : derived directly from absorption optical depth (like GP trough) derived directly from absorption optical depth (like GP trough) model uncertainty is very small model uncertainty is very small insensitive to clumpiness, giving mass-averaged x HI along the sight line insensitive to clumpiness, giving mass-averaged x HI along the sight line Fan et al. 2006 LAE statistics Malhotra & Rhoads ’ 05 Stern et al. ’ 05 Haiman & Cen ‘ 05 GRB 050904

16. Prospects for future observations low N HI GRBs? low N HI GRBs? only weak constraint on x HI when log N HI > 21.5 only weak constraint on x HI when log N HI > 21.5 However, there are GRBs with log N HI <~ 20 However, there are GRBs with log N HI <~ 20 promising chance for a better constraint on x HI promising chance for a better constraint on x HI Vreeswijk et al. ‘ 04 solid: DLA long-dashed: IGM+DLA(20.0)

17. Prospects for future observations (2) How often do we expect events like GRB 050904? How often do we expect events like GRB 050904? spectrum of GRB 050904 was 3.4 days. It would be x10 brighter if 0.5 day spectrum of GRB 050904 was 3.4 days. It would be x10 brighter if 0.5 day If GRB 050904 occurred at z=1, R=17.9, F~0.2 mJy at 1 day  one of the brightest optical afterglows If GRB 050904 occurred at z=1, R=17.9, F~0.2 mJy at 1 day  one of the brightest optical afterglows NIR flash of GRB 050904 was as bright as the opt. flash of GRB 990123 （ Boer et al. 05)  rare object? NIR flash of GRB 050904 was as bright as the opt. flash of GRB 990123 （ Boer et al. 05)  rare object? NIR spectroscopy necessary at z>7  lower sensitivity than optical NIR spectroscopy necessary at z>7  lower sensitivity than optical Fynbo et al. 01 Panaitescu & Kumar 01 GRB050904

18. Conclusions GRB 050904 gave us the first constraint on reionization from GRBs GRB 050904 gave us the first constraint on reionization from GRBs Opened a new era of GRB cosmology Opened a new era of GRB cosmology degeneracy: the red damping wing can be explained by DLA or IGM degeneracy: the red damping wing can be explained by DLA or IGM Lyβ feature can be used to break this degeneracy Lyβ feature can be used to break this degeneracy In the case of GRB 050904, the DLA is dominant In the case of GRB 050904, the DLA is dominant Lower x HI at z~6-6.3 is preferred from our data Lower x HI at z~6-6.3 is preferred from our data x HI = 0.00 ± 0.17 ， x HI < 0.60 (95% C.L.) x HI = 0.00 ± 0.17 ， x HI < 0.60 (95% C.L.) The IGM was largely reionized at z=6.3 The IGM was largely reionized at z=6.3 The first quantitative upper limit on x HI at z>6 by a direct method The first quantitative upper limit on x HI at z>6 by a direct method

19. Identified Absorption Lines Kawai et al. 2006

20. Prospects for future observations low N HI GRBs? low N HI GRBs? only weak constraint on x HI when log N HI > 21.5 only weak constraint on x HI when log N HI > 21.5 However, there are GRBs with log N HI <~ 20 However, there are GRBs with log N HI <~ 20 promising chance for a better constraint on x HI promising chance for a better constraint on x HI Lyα line emission from host galaxies? Lyα line emission from host galaxies? line emissivity affected by x HI, giving a probe for reionization line emissivity affected by x HI, giving a probe for reionization GRB host galaxies may have strong Lyα emission (in EW) (Jakobsson et al. 2005) GRB host galaxies may have strong Lyα emission (in EW) (Jakobsson et al. 2005) Lyα emission search for GRB host galaxies may also be interesting Lyα emission search for GRB host galaxies may also be interesting For GRB 050904, we set upper limit on Lyα corresponding to SFR < 0.8 M sun /yr For GRB 050904, we set upper limit on Lyα corresponding to SFR < 0.8 M sun /yr Vreeswijk et al. ‘ 04 solid: DLA long-dashed: IGM+DLA(20.0)