In Section 4, Paragraph 1, Line 14 of our paper [1], there is an error in the estimated overall detection efficiencies η_s, η_i for signal and idler photons, respectively, in our photon-pair coincidence measurement. The estimates of η_s ≃ 1.8% and η_i ≃ 2.8% in [1] did not include the waveguide-to-fiber coupling loss, which was ~50% for each channel. The correct efficiencies should therefore be η_s ≃ 0.9% and η_i ≃ 1.4%.

The efficiency corrections do not affect any of the experimental results and conclusions, including the pair generation rate of 2.0 × 10⁷ pairs/s/mW which was calculated correctly in [1]. We thank Don Schmadel from the University of Maryland, College Park for pointing out the error.