CALDER

Measuring tiny amounts of light at cryogenic temperatures is complicated and, in experiments like CUPID, sets stringent requirements on the features of the desired light detector:

• baseline resolution better than 50 eV;
• response time faster than 1 ms;
• active surface of up to 5x5cm² for an efficient coupling with commonly used cryogenic detectors;
• operation with high performance in a rather wide temperature range (5-20 mK) without need for fine-tuning;
• ease in fabrication and reproducibility;
• ease in operating hundreds of detectors with low heat-load for the cryogenic system;

None of the technologies for light detection fulfilled all these requirements. The CALDER collaboration proposed to realize a new light detector by exploiting devices which have already been succesfully used in astrophysics: Kinetic Inductance Detectors (KIDs).

When a superconductor is biased with an AC current, Cooper pairs oscillate and acquire kinetic inductance L_K. The superconductor can be inserted in an LC circuit with high merit factor, that acts as a resonator (blue plot). Interactions with energy larger than the binding energy of Cooper pairs (Δ₀) can break them into quasiparticles, pruducing a variation of the kinetic inductance. The change in L_K modifies the frequency and shape of the resonance (red plot). Thus, we can infer the energy of the interaction by monitoring the shape of the resonance. The binding energy of Cooper pairs is much lower than 1 meV, thus even small energy deposits can produce a signal.

Pros and Cons

The most important features of KIDs are:

• the intrinsic energy resolution, which can be as low as few eV;
• the natural aptitude to a frequency multiplexed read-out. Each detector can be designed to resonate at a slightly different frequency, thus hundreds of KIDs can be read-out with a single RF line and a cryogenic amplifier.
• They are operated well below the superconductor critical temperature, where the quasiparticles lifetime and the internal quality factor saturare, leading to a stable detector behavior over a wide temperature range.
• Most of the electronics is located at room-temperature, so the installation of KIDs array requires only minor modifications to the existing facilities.

The only limit of these devices resides in their rather poor active area: future light detectors need a surface of several cm², while the largest KIDs barely achieve few mm². To overcome this limit, the CALDER collaboration developed light detectors based on the phonon-mediated approach.

An effective coupling with commonly used bolometers, that have surfaces of several cm², requires large light detectors. Realizing a 5x5 cm² detector with hundreds of KIDs of a few mm² is not practicable, not only because of the too large number of devices (expecially for experiments that need hundreds of light detectors), but also because the emitted photons would barely interact in such a small device.

Nevertheless, KIDs can be used as sensors to probe a substrate that acts as mediator between the photon and the Cooper pairs. The photon interacts inside the (insulating) substrate producing phonons, that travel until they are absorbed by a KID or lost through the substrate supports/surfaces. Using phonons instead of the original photons decreases the efficiency in the collection of the impinging energy but, on the other hand, allows to exploit much larger active surfaces.

The goal of CALDER was to demostrate that this approach guarantees the realization of sensitive cryogenic light detectors.

For the first prototypes, we decided to exploit a commonly used material for KIDs: Aluminum. This superconductor is not the most sensitive (its transition temperature is of about 1.2K), but it is well known and it is rather simple to handle for KIDs applications. We started by depositing a 40 nm thick single pixel on a 2x2 cm² Silicon substrate. With this prototype, that features an active surface of 2.0x1.4 mm², we achieved a baseline resolution of about 130 eV and an average efficiency of about 5% (using a non-collimated X-rays source).

We reproduced the same design in a 4-pixel array, proving that the efficiency can be increased up to 18% without loss in energy resolution.

Afterwards, we optimized the geometry of the resonator, by enlarging the active area of almost a factor of 2 and by increasing the quality factor of the resonator up to 150x10³. These improvements allowed to reach a baseline resolution of about 80 eV RMS using a single resonator. The detector demonstrated to be insensitive to temperature changes up to 200 mK.

Then we moved the superconductor to trilayers of aluminum titanium and aluminum and reached an energy resolution of 26 eV and a response time of 0.02 ms, and finally we enlarged the active area to 5x5 cm² reaching a resolution of 34 eV and response time of 0.1 ms.

As spin-off, we started fruitful collaborations to investigate the impact of radioactivity and cosmic rays on superconducting qu-bits, triggered by the analogies of this technology with KIDs.

Check out our publications for an extensive description of the steps and results of the project!