2014MNRAS.445..726C

Intrinsic alignments of galaxies have been shown to contaminate weak gravitational lensing observables on linear scales, r > 10 h-1 Mpc, but studies of alignments in the non-linear regime have thus far been inconclusive. We present an estimator for extracting the intrinsic alignment signal of galaxies around stacked clusters of galaxies from multiband imaging data. Our estimator removes the contamination caused by galaxies that are gravitationally lensed by the clusters and scattered in redshift space due to photometric redshift uncertainties. It uses posterior probability distributions for the redshifts of the galaxies in the sample and it is easily extended to obtain the weak gravitational lensing signal while removing the intrinsic alignment contamination. We apply this algorithm to groups and clusters of galaxies identified in the Sloan Digital Sky Survey `Stripe 82' co-added imaging data over ∼ 150 deg2. We find that the intrinsic alignment signal around stacked clusters in the redshift range 0.1 < z < 0.4 is consistent with zero. In terms of the tidal alignment model of Catelan et al., we set joint constraints on the strength of the alignment and the bias of the lensing groups and clusters on scales between 0.1 and 10h-1 Mpc, b_LC_1ρ_crit = -2_-14^+14 ×10^-4. This constrains the contamination fraction of alignment to lensing signal to the range between [ - 18, 23] per cent below scales of 1 h-1 Mpc at 95 per cent confidence level, and this result depends on our photometric redshift quality and selection criteria used to identify background galaxies. Our results are robust to the choice of photometric band in which the shapes are measured (i and r) and to centring on the Brightest Cluster Galaxy or on the geometrical centre of the clusters.