KID arrays. One of the fundamental assets of KIDs is that they can be easily multiplexed, reading out hundreds of pixels with a single feedline. While this is true on paper, actually achieving large format arrays, with thousands of good working pixels, remains a challenging task. The GIS collaboration has been working in this direction constantly over more than a decade, passing from the tens or hundreds of pixels per array of the NIKA camera, to ~1000 pixels per array in NIKA2 and ~2000 in CONCERTO. This has been possible thanks to the experience gained developing such cameras, coupled to the effort done to get an always better understanding of the key aspects that play a role in the final quality of an array. These include in particular mastering the various fabrication steps that are needed to produce an array, simulating the devices and choosing the designs that can accommodate the largest number of pixels while minimizing the risks of frequency collisions, and optimizing the architecture of the focal plane, to assure an efficient optical coupling, the good thermalization of the arrays and a solid mounting. Today, we are working on the next generation of large format arrays, called 8KID: one single 4 inches wafer containing more than 8000 pixels, optimized for the mm-wave bands. We have already fabricated and tested the first working samples. Being able to consistently produce high quality 8KID arrays will be a crucial step towards the next generation of experiments, with focal planes requiring tens of thousands of detectors.

Silicon lenses. In the recent years our collaboration acquired a lot of experience in the design and fabrication of plastic lenses (High Density Polyethylene and Polypropylene) currently used in instruments such as NIKA2, KISS and CONCERTO. However, for the next generation instruments we need to develop higher Field-of-View and consequently larger lenses up to a diameter of the order of 300-400 mm. For such large optical elements, plastic is no longer a suitable material because of its low refractive index. On the contrary, silicon is an excellent choice due to its low loss at millimeter and sub-millimeter wavelengths and to its high index of refraction leading to high-throughput and large field of view optical designs ideal for a large sky survey. However the high index of refraction of the silicon lenses also means that a significant fraction of the incident light is reflected. This not only causes less light to reach the detectors, causing a decrease in the overall sensitivity, but it can also lead to other instrument systematics due to multiple reflections occurring in the instrument. The undesirable consequences of the high index means that the lenses must have Anti-Reflection (AR) coatings in order to deliver state-of-the-art measurements of the sky. To solve this problem we develop meta-material AR coatings. Metamaterial AR coating consists of sub-wavelength features machined with a milling machine into an optical surface which act as effective dielectric layers. The shape and dimensions of the sub-wavelength features of the metamaterial coating can be tuned to result in sub-percent reflections across the identified bandwidth. This R&D work is possible through the use of a specific milling machining which is provided thanks to a LABEX ENIGMASS funding. The first prototypes of Silicon lenses are supposed to be produced in 2023.

On-chip spectroscopy. The development of KID based spectroscopy aims to provide different technological options to study millimeter waves at different energy resolutions R=Δν/ν ranging from 10 to 10,000. Among other astrophysical challenges, two are of greater importance to us. The first is to separate the so called B modes of the Cosmic Microwave Background polarization from those of the foreground emission of galactic dust and synchrotron. The second is to map the intensity fluctuations of the Carbon II line in early galaxies to understand better galaxy evolution and star formation at the epoch of reionization. For low resolution (R~10), we have already coupled KIDs to a Martin-Puplett interferometer on KISS and CONCERTO. We are investigating another solution based KID arrays on wedge-shaped substrates acting as Fabry-Perot cavity. For R~100, we are investigating on-chip spectroscopy with superconducting resonant filters. For even higher resolutions R of 1000 to 10,000, sub-gap modes of KID made out of particular superconducting materials are explored as a technological option.

Readout electronics. The readout electronics, designed to be hosted in rugged micro-TCA crates, can probe up to 400 KID per readout card over a large bandwidth (950 MHz). A microTCA can host up to 12 of these boards and each board uses less than 40W. Each readout board heavily relies on the use of a Field Programmable Gate Array (FPGA) and fast digital to analog and digital to analog converters (2 Gigasamples per second). The radiofrequency front-end is in charge of performing the up and down conversion of the signal between the converters and the cryostat (back and forth). This electronics is used for radioastronomy and in-house developments. The firmware of the FPGA can be adapted to photometry or spectroscopy observations.