We present the discovery of one or two extremely faint z~6 quasars in 6.5 deg^2 utilizing a unique capability of the wide-field imaging of the Subaru/Suprime-Cam. The quasar selection was made in (i'-z_B) and (z_B-z_R) colors, where z_B and z_R are bandpasses with central wavelengths of 8842A and 9841A, respectively. The color selection can effectively isolate quasars at z~6 from M/L/T dwarfs without the J-band photometry down to z_R<24.0, which is 3.5 mag. deeper than SDSS. We have selected 17 promising quasar candidates. The follow-up spectroscopy for seven targets identified one apparent quasar at z=6.156 with M_1450=-23.10. We also identified one possible quasar at z=6.041 with a faint continuum of M_1450=-22.58 and a narrow Lyman-alpha emission with HWHM=427 km/s, which cannot be distinguished from Lyman-alpha emitters. We derive the quasar luminosity function at z~6 by combining our faint quasar sample with the bright quasar samples by SDSS and CFHQS. Including our data points invokes a higher number density in the faintest bin of the quasar luminosity function than the previous estimate employed. This suggests a steeper faint-end slope than lower-z, though it is yet uncertain based on a small number of spectroscopically identified faint quasars and several quasar candidates are still remain to be diagnosed. The steepening of the quasar luminosity function at the faint-end does increase the expected emission rate of the ionizing photon, however, it only changes by a factor of ~2-6. This was found to be still insufficient for the required photon budget of reionization at z~6.

We have identified a very interesting Ly-alpha emitter, whose Ly-alpha emission line has an extremely large observed equivalent width of EW_0=436^{+422}_{-149}A, which corresponds to an extraordinarily large intrinsic rest-frame equivalent width of EW_0^{int}=872^{+844}_{-298}A after the average intergalactic absorption correction. The object was spectroscopically confirmed to be a real Ly-alpha emitter by its apparent asymmetric Ly-alpha line profile detected at z=6.538. The continuum emission of the object was definitely detected in our deep z'-band image; thus, its EW_0 was reliably determined. Follow-up deep near-infrared spectroscopy revealed emission lines of neither He II lambda1640 as an apparent signature of Population III, nor C IV lambda1549 as a proof of active nucleus. No detection of short-lived He II lambda1640 line is not necessarily inconsistent with the interpretation that the underlying stellar population of the object is dominated by Population III. We found that the observed extremely large EW_0 of the Ly-alpha emission and the upper limit on the EW_0 of the He II lambda1640 emission can be explained by population synthesis models favoring a very young age less than 2-4Myr and massive metal-poor (Z<10^{-5}) or even metal-free stars. The observed large EW_0 of Ly-alpha is hardly explained by Population I/II synthesis models with Z>10^{-3}. However, we cannot conclusively rule out the possibility that this object is composed of a normal stellar population with a clumpy dust distribution, which could enhance the Ly-alpha EW_0, though its significance is still unclear.

We report the discovery of a protocluster at z~6 containing at least eight cluster member galaxies with spectroscopic confirmations in the wide-field image of the Subaru Deep Field (SDF). The overdensity of the protocluster is significant at the 6 sigma level, based on the surface number density of i'-dropout galaxies. The overdense region covers ~36 sq. arcmin, and includes 30 i'-dropout galaxies. Follow-up spectroscopy revealed that 15 of these are real z~6 galaxies (5.7 < z < 6.3). Eight of the 15 are clustering in a narrow redshift range centered at z=6.01, corresponding to a seven-fold increase in number density over the average in redshift space. We found no significant difference in the observed properties, such as Ly-alpha luminosities and UV continuum magnitudes, between the eight protocluster members and the seven non-members. The velocity dispersion of the eight protocluster members is 647 km/s, which is about three times higher than that predicted by the standard cold dark matter model. This discrepancy could be attributed to the distinguishing three-dimensional distribution of the eight protocluster members. We discuss two possible explanations for this discrepancy: either the protocluster is already mature, with old galaxies at the center, or it is still immature and composed of three subgroups merging to become a larger cluster. In either case, this concentration of z=6.01 galaxies in the SDF may be one of the first sites of formation of a galaxy cluster in the universe.

We carried out extended spectroscopic confirmations of Ly-alpha emitters (LAEs) at z=6.5 and 5.7 in the Subaru Deep Field. Now, the total number of spectroscopically confirmed LAEs is 45 and 54 at z=6.5 and 5.7, respectively, and at least 81% (70%) of our photometric candidates at z=6.5 (5.7) have been spectroscopically identified as real LAEs. We made careful measurements of the Ly-alpha luminosity, both photometrically and spectroscopically, to accurately determine the Ly-alpha and rest-UV luminosity functions (LFs). The substantially improved evaluation of the Ly-alpha LF at z=6.5 shows an apparent deficit from z=5.7 at least at the bright end, and a possible decline even at the faint end, though small uncertainties remain. The rest-UV LFs at z=6.5 and 5.7 are in good agreement, at least at the bright end, in clear contrast to the differences seen in the Ly-alpha LF. These results imply an increase in the neutral fraction of the intergalactic medium from z=5.7 to 6.5. The rest-frame equivalent width (EW_0) distribution at z=6.5 seems to be systematically smaller than z=5.7, and it shows an extended tail toward larger EW_0. The bright end of the rest-UV LF can be reproduced from the observed Ly-alpha LF and a reasonable EW_0-UV luminosity relation. Integrating this rest-UV LF provides the first measurement of the contribution of LAEs to the photon budget required for reionization. The derived UV LF suggests that the fractional contribution of LAEs to the photon budget among Lyman break galaxies significantly increases towards faint magnitudes. Low-luminosity LAEs could dominate the ionizing photon budget, though this inference depends strongly on the uncertain faint-end slope of the Ly-alpha LF.