Asgard/NOTT: L-band nulling interferometry at the VLTI -- III. The mid-infrared integrated optics beam combiner for NOTT

TL;DR

This work demonstrates the first broadband L$^{\prime}$-band deep null achieved with a four-telescope integrated optics beam combiner designed for NOTT. Using Ultrafast Laser Inscription in GLS glass, the authors fabricate a 4T-nuller based on the double-Bracewell scheme and validate its performance in a lab with a four-beam, broadband setup, achieving a raw null of $8.13\times 10^{-3}$ and a self-calibrated null of $1.14\times 10^{-3}$ over a 200 nm band around $3.8~\mu$m. They confirm single-mode operation in the 3.65–3.85 $\mu$m range, demonstrate an achromatic central 50/50 directional coupler with 7.5 mm interaction length, and quantify throughput (37% with room for AR-coated improvements) and polarization effects (low differential birefringence but non-negligible impact at certain polarization angles). The results show the feasibility of broad-null performance via the Angel & Woolf double-Bracewell approach and indicate the necessary steps—cryogenic testing, post-processing calibration, and polarization control—to meet the stringent nulling requirements for NOTT and exoplanet spectroscopy at the water-ice line.

Abstract

The NOTT visitor instrument at the VLTI will characterize hot exozodiacal dust and young Jupiter-like planets at the water snowline via L' band nulling interferometry. The beam combination will be achieved by a four-telescope integrated optics beam combiner (IOBC) that fulfills specific requirements. Our goal was to manufacture the mid-infrared IOBC for NOTT based on the double-Bracewell architecture and run a detailed laboratory characterization in the L' band. We focus on the achievable raw and self-calibrated nulling ratios. We use a double Michelson interferometer to produce four broadband coherent beams simulating the four telescopes of the VLTI and perform broadband nulling at room temperature. We analyze the modal, chromatic, and polarization behavior of the IOBC, and measure its total throughput. We were able to manufacture a single-mode four-telescope double-Bracewell IOBC in GLS mid-infrared transparent glass using Ultrafast Laser Inscription. We show that the directional couplers forming the four-telescope IOBC (4T-nuller) have an achromatic splitting ratio across the band 3.65-3.85 um with a 40/60 and 50/50 splitting for the side couplers and the central coupler, respectively. We report a total throughput of 37%, including the Fresnel losses that will be mitigated with anti-reflection coatings, and quantify differential birefringence. Operating at room temperature, with 200 nm bandwidth centered at 3.8 um and without polarization control, we measure an average raw null of 8.13+/-0.03x10-3 and a self-calibrated null of 1.14+/-0.01x10-3. Finally, we show that a theta^6 broad null can be experimentally reproduced in these conditions. This is, to our knowledge, the first measurement of a broadband L' deep null obtained with a four-telescope integrated optics beam combiner. The next step foresees testing the 4T-nuller in cryogenic conditions.

Figures (12)

• Figure 1: Functional phasor diagram of the double-Bracewell nuller with complex amplitudes of inputs 1 (green), 2 (yellow), 3 (red), and 4 (blue). DC1, DC2, and DC3 are three directional couplers with 50/50 splitting ratio in intensity. DC1 and DC2 produce constructive outputs at #1 and #4, while the two nulled outputs are feeding DC3. The nulled outputs #2 and #3 feature two transmission maps, which are centro-symmetric to one another w.r.t. the central star.
• Figure 2: Top left: NOTT's 4T-nuller photonic chip with the network of waveguides manufactured by ULI (see Sect. \ref{['sec:manufacturing_uli']} for details). Top right: cross-section of the input/output facet of a triplet waveguide; Bottom left: schematic of the 4T-nuller showing the four inputs, the side-step S-bends, and the three cascaded directional couplers; Bottom right: visualization of the different outputs of the 4T-nuller (photometry and interferometry) observed on the testbench camera (Sect. \ref{['sec:results_extinction_ratios']}). The photometric signals from each input are extracted by the Y-junctions and routed to photometric outputs $P_1$, $P_2$, $P_3$, and $P_4$. The remaining signals pass through the interferometric section consisting of three directional couplers, producing the four circled interferometric outputs $I_1$, $I_2$, $I_3$, and $I_4$. In the configuration of Fig. \ref{['fig:phasor']}, $I_1$ and $I_4$ correspond to constructive outputs while $I_2$ and $I_3$ yield isolated nulled signals. $l$ is the interaction length of the coupler, $s$ is the center-to-center separation between the waveguides or pitch, $A$ is the amplitude of the lateral offset and $r$ is the radius of curvature.
• Figure 3: Layout of the mid-infrared characterization testbench and its components. SCS: supercontinuum source; BBS: blackbody source; AC: achromat; AP: aperture; L$^\prime$: broadband filter; PH: pinhole; BS: beam splitter; M1: fixed mirror; M2, M3, M4: movable mirrors with delay lines; PL: mid-infrared linear polarizer; HeNe: metrology laser to calibrate the delay lines; Photonic Chip contains the 4T-nullers. Top inset: spectral profile of the SCS measured before the photonic chip.
• Figure 4: Measurement of the splitting ratio of the three directional couplers with 7.5 mm interaction length forming the 4T-nuller (three plots from the left) and for the Y-junction photometric tap. The error bars correspond to three times the standard deviation obtained from four independent measurements. In the last plot, the gray curves depict one of the four Y-junctions in the characterized 4T-nuller, while the green curves show the properties of the targeted 80/20 splitter, which will be implemented in a future version.
• Figure 5: Relative throughput in percent for the reference straight waveguide and S-bends of 1.1 mm amplitude, along with the 4T-nuller. Error bars denote standard deviations from multiple identical straight and S-bend waveguides on the same photonic chip, while the standard deviation for the 4T-nuller is based on the throughputs of its four individual inputs.