Fig. 4: Octave-wide operation of dichroics.

a–d First dichroic, whose purpose is to separate 2 μm light from shorter wavelengths. a Simulations at 1050, 1550, and 2050 nm showing extraction of the 2 μm band into the cross port. b Measured broadband experimental spectra at the bar and cross ports. The input is the microcomb shown in the inset. c Measured (symbols) and simulated (solid lines) octave-wide transfer function. At the cross or 2 μm port, extinction ratios of (21.4 ± 1.1) and (19.9 ± 0.8) dB are measured in the 1 and 1.55 μm bands, respectively. At the bar port, an extinction ratio of (18.1 ± 2.9) dB is measured in the 2 μm band. d Magnified individual spectral bands. e–h Second dichroic, whose purpose is to separate 1.55 μm light from shorter wavelengths. e Simulations at 1050 and 1550 nm, showing extraction of the 1.55 μm band into the cross port. f Measured broadband experimental spectra at the bar and cross ports. The input is the microcomb shown in the inset of (b). g Measured (symbols) and simulated (solid lines) octave-wide transfer function. At the cross or 1.55 μm port, an extinction ratio of (20.1 ± 1.0) dB is measured in the 1 μm band, and at the bar port, an extinction ratio of (18.6 ± 3.3) dB is measured in the 1.55 μm band. h Magnified individual spectral bands. The performance of the dichroic in the spectral region shaded in (f, g) is relatively unimportant, as this region is filtered out by the first dichroic in the full interposer chip. In (a, e), cross-sections of |E(x,y,z)|2 are plotted with z set to half the height of the dichroics. The measured transfer functions shown in (c, g) are extracted from the corresponding transmission of the comb teeth in (b, f). The corresponding uncertainties reported in (c, g) correspond to line-to-line fluctuations in the measured comb spectra and include variations in coupling and are one standard deviation values