2014ApJ...794..133S

We present a method to detect small atmospheric signals in Kepler's planet candidate light curves by averaging light curves for multiple candidates with similar orbital and physical characteristics. Our statistical method allows us to measure unbiased physical properties of Kepler's planet candidates, even for candidates whose individual signal-to-noise precludes the detection of their secondary eclipse. We detect a secondary eclipse depth of 3.83_–1.11^+1.10 ppm for a group of 31 sub-Saturn (R < 6 R_⊕) planet candidates with the greatest potential for a reflected light signature ((R_p/a)² > 10 ppm). Including Kepler-10b in this group increases the depth to 5.08_–0.72^+0.71 ppm. For a control group with (R_p/a)² < 1 ppm, we find a depth of 0.36±0.37 ppm, consistent with no detection. We also analyze the light curve of Kepler-10b and find an eclipse depth of 7.08±1.06 ppm. If the eclipses are due solely to reflected light, this corresponds to a geometric albedo of 0.22±0.06 for our group of close-in sub-Saturns, 0.37±0.05 if including Kepler-10b in the group, and 0.60±0.09 for Kepler-10b alone. Including a thermal emission model does not change the geometric albedo appreciably, assuming A_B= (3/2)*A_g. Our result for Kepler-10b is consistent with previous works. Our result for close-in sub-Saturns shows that Kepler-10b is unusually reflective, but our analysis is consistent with the results of Demory for super-Earths. Our results also indicate that hot Neptunes are typically more reflective than hot Jupiters.