Kinetic Inductance Phonon-Mediated Detector

• Kinetic Inductance Phonon-Mediated Detectors are superconducting calorimeters that use kinetic inductance resonators to sense athermal phonons generated by particle interactions in substrates.
• They convert energy deposited in a bulk crystal into phonon signals that break Cooper pairs, resulting in measurable shifts in resonator frequency and quality factor.
• Recent advances focus on improving phonon collection efficiency, reducing noise sources, and lowering superconducting gaps to achieve sub-eV energy resolution.

A kinetic inductance phonon-mediated detector is a superconducting detector architecture in which a kinetic-inductance resonator senses energy that reaches it through phonons rather than only by direct electromagnetic absorption. In the most specific usage, a Kinetic Inductance Phonon-Mediated (KIPM) detector is a calorimeter that “uses kinetic inductance detectors to read out phonon signals from the device substrate,” so that a particle interaction in a bulk crystal creates athermal phonons, those phonons break Cooper pairs in a superconducting resonator, and the resulting quasiparticle population shifts the resonator response (Temples et al., 29 Sep 2025). Closely related literature uses the same underlying transduction chain for wide-area cryogenic light detectors on silicon or germanium substrates, for phonon-engineered direct-absorption MKIDs, and for thermal kinetic inductance detectors on suspended membranes, where the signal is mediated by thermal phonons in a bolometric thermal link rather than by direct athermal phonon capture (Temples et al., 2024, Jabbari et al., 13 Mar 2026).

1. Conceptual scope and historical development

The modern literature contains three adjacent but non-identical meanings of the topic. The canonical KIPM definition is the substrate-coupled calorimeter used for rare-event detection: energy is deposited in a crystalline target, converted into athermal phonons, and read out by MKIDs on the substrate surface (Temples et al., 29 Sep 2025). A second family consists of phonon-mediated light detectors, especially in the CALDER program, where optical or X-ray energy is absorbed in a several-$\mathrm{cm}^2$ silicon or germanium substrate and only a fraction of the resulting phonons is intercepted by small superconducting resonators (Cardani et al., 2015, D'Addabbo et al., 2017, Delicato et al., 2024). A third family is the thermal kinetic inductance detector, in which a suspended membrane stores deposited power as heat and the kinetic-inductance resonator functions as a bolometric thermometer; here the relevant mediator is the thermal phonon conductance of the membrane and support legs rather than direct pair-breaking by ballistic substrate phonons (Jabbari et al., 13 Mar 2026).

The field emerged first as a demonstration of fast phonon sensing with multiplexed superconducting resonators. An early silicon-wafer LEKID array provided fully synchronous readout with a per-resonator bandwidth of $1.2\ \mathrm{MHz}$, enabling sub-$\mu\mathrm{s}$ phonon imaging across the wafer (Swenson et al., 2010). Position- and energy-resolved phonon-mediated particle detection was then demonstrated in silicon with $< 1\ \mathrm{mm}$ position precision at $30\ \mathrm{keV}$ and $\sigma_E = 0.55\ \mathrm{keV}$ at $30\ \mathrm{keV}$ after position correction (Moore et al., 2012). In parallel, CALDER established the large-area light-detector variant on $2\times2\ \mathrm{cm}^2$ Si substrates, progressing from a four-KID aluminum prototype with $154 \pm 7\ \mathrm{eV}$ RMS baseline to a single-KID design at $82 \pm 4\ \mathrm{eV}$, and later to an Al/Ti/Al trilayer device at $1.2\ \mathrm{MHz}$0 RMS (Cardani et al., 2015, Bellini et al., 2016, Cardani et al., 2018).

More recent work has shifted the center of gravity toward dark-matter and neutrino applications. A 1 g silicon KIPM prototype operated at the NEXUS Cryogenic Facility reached a baseline resolution on energy absorbed by the phonon sensor of $1.2\ \mathrm{MHz}$1, while the corresponding resolution on energy deposited in the substrate remained $1.2\ \mathrm{MHz}$2 because the phonon collection efficiency was only $1.2\ \mathrm{MHz}$3 (Temples et al., 2024). The consortium view codifies the current program as one of improving phonon collection efficiency $1.2\ \mathrm{MHz}$4, suppressing noise, and moving to lower-$1.2\ \mathrm{MHz}$5 superconductors, with a long-term goal of sub-eV threshold on energy deposited in the substrate (Temples et al., 29 Sep 2025).

2. Microscopic transduction and energy-conversion physics

The core transduction chain is the same across the athermal-phonon implementations. Deposited energy in a substrate launches athermal phonons; phonons entering the superconducting film with energy above $1.2\ \mathrm{MHz}$6 break Cooper pairs; the excess quasiparticles alter the complex conductivity; and the resulting change in kinetic inductance and dissipation shifts the resonator frequency and quality factor. In the NEXUS KIPM formulation, the small-signal response is written as

$1.2\ \mathrm{MHz}$7

with the corresponding resonator variables

$1.2\ \mathrm{MHz}$8

For absorbed energy $1.2\ \mathrm{MHz}$9, the frequency responsivity is

$\mu\mathrm{s}$0

while for deposited substrate energy the effective responsivity is reduced by the phonon collection efficiency $\mu\mathrm{s}$1 (Temples et al., 2024).

This detector physics is inseparable from nonequilibrium quasiparticle–phonon dynamics. A detailed Chang–Scalapino treatment for THz KIDs showed that even sub-gap microwave readout drives the quasiparticle distribution $\mu\mathrm{s}$2 far from equilibrium at $\mu\mathrm{s}$3, while pair-breaking signal photons create additional structure through phonon-mediated downconversion and recombination (Goldie et al., 2014). In that framework, the useful quasiparticle yield is governed by competition between phonon pair-breaking time $\mu\mathrm{s}$4 and phonon loss time $\mu\mathrm{s}$5. The paper’s simple cascade argument gives the pair-breaking probability

$\mu\mathrm{s}$6

and for the modeled thin-film Al case this leads to a source detection efficiency of order $\mu\mathrm{s}$7, illustrating that finite phonon escape can remove roughly half the useful quasiparticle yield before readout (Goldie et al., 2014).

The same issue appears empirically in optical MKIDs. For TiN resonators, the downconversion efficiency was parameterized as

$\mu\mathrm{s}$8

and the maximum resolving power as

$\mu\mathrm{s}$9

with $< 1\ \mathrm{mm}$0. Using thermal quasiparticle calibration and optical pulses, the measured mean result for three TiN pixels was

$< 1\ \mathrm{mm}$1

substantially below the frequently assumed bulk value $< 1\ \mathrm{mm}$2, while the inferred $< 1\ \mathrm{mm}$3 remained linearly entangled with the uncertainty in the single-spin density of states $< 1\ \mathrm{mm}$4 (Hernandez et al., 2020). This is directly relevant to phonon-mediated KIDs because the same downconversion, phonon escape, and recombination losses determine how much deposited energy survives to the gap scale.

Phonon loss can also be engineered. A membrane-less Hf/In optical MKID improved resolving power from 11 to 20 at $< 1\ \mathrm{mm}$5, reducing the inferred phonon-loss term from $< 1\ \mathrm{mm}$6 to $< 1\ \mathrm{mm}$7; the dominant mechanism was identified not as simple acoustic mismatch but as the inability of high-energy phonons to enter the low-Debye-temperature indium layer because of the lack of available phonon states (Zobrist et al., 2022). Although that device was not a canonical substrate-phonon KIPM detector, it established that interface spectral engineering can materially alter phonon escape and hence quasiparticle yield.

3. Architectural families

The literature now supports several stable architectural patterns.

Architecture	Mediating excitations	Representative papers
Substrate-coupled athermal-phonon MKID / KIPM	Athermal phonons in bulk Si or Ge	(Moore et al., 2012, Temples et al., 2024, Delicato et al., 2024, Temples et al., 29 Sep 2025)
Wide-area cryogenic light detector	Optical/X-ray energy $< 1\ \mathrm{mm}$8 substrate phonons $< 1\ \mathrm{mm}$9 KID	(Cardani et al., 2015, Bellini et al., 2016, Cardani et al., 2018)
Sensor–collector-separated phonon KID	Athermal phonons collected in high-gap metal, quasiparticles trapped in lower-gap KID	(Pesce et al., 13 Jan 2026)
Thermal kinetic inductance detector	Thermal phonons in suspended membrane and support legs	(Jabbari et al., 13 Mar 2026)

The earliest particle-detector geometry used aluminum LEKIDs patterned directly on 1 mm thick high-resistivity silicon, with 20 resonators centered around $30\ \mathrm{keV}$0. Athermal phonons generated in the wafer were sensed by the resonator inductors themselves, and the relative pulse amplitudes and delays across the array provided both energy and position information (Moore et al., 2012). The NEXUS KIPM platform retained the same basic idea at larger scale: a 1 g silicon target, 11 resonators in the 3.9–4.4 GHz band, and one primary aluminum phonon absorber whose $30\ \mathrm{keV}$1 Al inductor occupied only $30\ \mathrm{keV}$2 on a $30\ \mathrm{keV}$3, 1 mm thick substrate (Temples et al., 2024). The consortium summary explicitly treats this as the current reference architecture and organizes its improvement program around more active area, less dead superconducting area, and lower-gap sensing elements (Temples et al., 29 Sep 2025).

The wide-area light-detector family was developed because direct-absorption KIDs have active areas of only a few $30\ \mathrm{keV}$4, whereas bolometric light detectors need several $30\ \mathrm{keV}$5. CALDER therefore placed one or more aluminum resonators on $30\ \mathrm{keV}$6 silicon substrates $30\ \mathrm{keV}$7–$30\ \mathrm{keV}$8 thick, using the substrate as the optical absorber and the KIDs as phonon sensors (Cardani et al., 2015, Bellini et al., 2016). This design space includes both silicon and germanium absorbers. The first phonon-mediated germanium array used a $30\ \mathrm{keV}$9, $\sigma_E = 0.55\ \mathrm{keV}$0 thick Ge tile with three $\sigma_E = 0.55\ \mathrm{keV}$1 Al resonators at $\sigma_E = 0.55\ \mathrm{keV}$2, $\sigma_E = 0.55\ \mathrm{keV}$3, and $\sigma_E = 0.55\ \mathrm{keV}$4, demonstrating that high-$\sigma_E = 0.55\ \mathrm{keV}$5 KIDs can be fabricated on Ge if native oxide removal is sufficiently aggressive (Delicato et al., 2024).

A more explicit decoupling of absorption and sensing appears in the FunKID architecture. Here the resonator is a lower-gap Al/Ti/Al trilayer meander, while the phonon collectors are separate $\sigma_E = 0.55\ \mathrm{keV}$6 aluminum funnels. Athermal phonons are absorbed in the higher-gap Al collectors, the generated quasiparticles diffuse into the lower-gap trilayer, and gap engineering traps them in the KID. The paper states that the phonon collection volume is $\sigma_E = 0.55\ \mathrm{keV}$7, the active sensor volume is $\sigma_E = 0.55\ \mathrm{keV}$8, and the resulting responsivity enhancement is about a factor of five relative to a standard phonon-mediated KID fabricated on the same silicon tile (Pesce et al., 13 Jan 2026).

The thermal-bolometric branch is architecturally different but still belongs within the broad topic. In the MgB$\sigma_E = 0.55\ \mathrm{keV}$9 TKID, a lumped-element resonator is placed on a free-standing SiN$30\ \mathrm{keV}$0 membrane with four narrow support legs, and an on-membrane Au heater is used for calibration. Deposited power first raises the membrane temperature, and only then changes $30\ \mathrm{keV}$1, $30\ \mathrm{keV}$2, and $30\ \mathrm{keV}$3. The resonator is therefore a thermometer on a bolometer island rather than a direct athermal phonon absorber (Jabbari et al., 13 Mar 2026).

4. Readout, calibration, and pulse analysis

Across the field, readout is performed through the complex forward transmission $30\ \mathrm{keV}$4 of a microwave feedline, with resonator parameters extracted from notch-like or circle-fit models. The KIPM particle-detector literature commonly uses a full complex fit

$30\ \mathrm{keV}$5

while the simpler canonical form

$30\ \mathrm{keV}$6

appears in several phonon-mediated light-detector implementations (Temples et al., 2024, Delicato et al., 2024). Operationally, signals are usually projected into phase/frequency and amplitude/dissipation quadratures, and then processed with matched or optimal filters.

CALDER established a particularly influential two-channel analysis. The single-KID $30\ \mathrm{keV}$7 detector used $30\ \mathrm{keV}$8 windows sampled at $30\ \mathrm{keV}$9, converted $2\times2\ \mathrm{cm}^2$0 to $2\times2\ \mathrm{cm}^2$1 and $2\times2\ \mathrm{cm}^2$2, and then applied a 2D matched filter

$2\times2\ \mathrm{cm}^2$3

which accounts for the noise covariance between amplitude and phase (Bellini et al., 2016). This was important because phase pulses were about ten times larger than amplitude pulses, yet amplitude noise was much lower and closer to the amplifier limit. The practical lesson, repeated across CALDER papers, is that the quadrature with the larger pulse is not necessarily the quadrature with the better final energy estimator (D'Addabbo et al., 2017).

Absolute calibration has been achieved in three main ways. First, thermal quasiparticle calibration uses the equilibrium expression

$2\times2\ \mathrm{cm}^2$4

together with Mattis–Bardeen fits of resonant-frequency shift versus temperature; optical pulse amplitudes can then be converted to quasiparticle number and hence to downconversion efficiency $2\times2\ \mathrm{cm}^2$5 (Hernandez et al., 2020). Second, LED photon shot noise provides a direct absolute calibration. In the NEXUS KIPM detector, pulsed $2\times2\ \mathrm{cm}^2$6 light was used so that the variance of the pulse-amplitude distribution obeyed

$2\times2\ \mathrm{cm}^2$7

allowing the intercept to determine baseline noise and the slope to determine the responsivity-per-photon $2\times2\ \mathrm{cm}^2$8 (Temples et al., 2024). A closely related method was used for the germanium array with $2\times2\ \mathrm{cm}^2$9 photons, where

$154 \pm 7\ \mathrm{eV}$0

yielded both responsivity and baseline resolution (Delicato et al., 2024). Third, the TKID branch uses electrical substitution through an integrated heater and the membrane thermometry relation

$154 \pm 7\ \mathrm{eV}$1

to extract thermal conductance, responsivity, and NEP without an optical setup (Jabbari et al., 13 Mar 2026).

Pulse morphology itself carries physical information. In substrate-phonon devices, rise times are commonly governed by athermal phonon propagation and decay times by quasiparticle recombination or more complex nonequilibrium processes. CALDER reported $154 \pm 7\ \mathrm{eV}$2–$154 \pm 7\ \mathrm{eV}$3 rise times and $154 \pm 7\ \mathrm{eV}$4–$154 \pm 7\ \mathrm{eV}$5 decay times in its four-pixel array, while the NEXUS KIPM detector required a two-component empirical template with prompt and delayed channels to fit millisecond-scale pulses over a wide temperature range (D'Addabbo et al., 2017, Temples et al., 2024). In the thermal TKID regime, the pulse is instead a single-pole bolometric response with

$154 \pm 7\ \mathrm{eV}$6

so the signal bandwidth is set by membrane heat capacity and leg conductance rather than by quasiparticle recombination (Jabbari et al., 13 Mar 2026).

5. Performance landscape and application domains

The particle-detection branch has already demonstrated simultaneous energy and position sensitivity. In silicon, phonon-mediated MKIDs achieved $154 \pm 7\ \mathrm{eV}$7 position resolution at $154 \pm 7\ \mathrm{eV}$8, a baseline resolution $154 \pm 7\ \mathrm{eV}$9, and $82 \pm 4\ \mathrm{eV}$0 at $82 \pm 4\ \mathrm{eV}$1 after position correction (Moore et al., 2012). The NEXUS KIPM prototype later separated intrinsic sensor performance from substrate-level performance: $82 \pm 4\ \mathrm{eV}$2 for energy absorbed by the phonon sensor, $82 \pm 4\ \mathrm{eV}$3 for energy deposited in the substrate, and $82 \pm 4\ \mathrm{eV}$4 in the February 2023 calibration (Temples et al., 2024). The consortium summary uses this result as the present record for sensor-absorbed-energy resolution and frames future dark-matter and low-energy neutrino searches around improving $82 \pm 4\ \mathrm{eV}$5 while preserving the strong MKID sensor physics (Temples et al., 29 Sep 2025).

Wide-area phonon-mediated light detectors define a second performance axis. The first CALDER array, consisting of four $82 \pm 4\ \mathrm{eV}$6 Al resonators on a $82 \pm 4\ \mathrm{eV}$7, $82 \pm 4\ \mathrm{eV}$8 silicon chip, achieved a baseline $82 \pm 4\ \mathrm{eV}$9 and total efficiency $1.2\ \mathrm{MHz}$00 (Cardani et al., 2015). An optimized single Al KID on the same substrate size improved the baseline to $1.2\ \mathrm{MHz}$01, and to $1.2\ \mathrm{MHz}$02 when the source was directly below the resonator (Bellini et al., 2016). Replacing the $1.2\ \mathrm{MHz}$03 Al film with an Al/Ti/Al trilayer exploiting the superconducting proximity effect lowered $1.2\ \mathrm{MHz}$04 to $1.2\ \mathrm{MHz}$05, increased $1.2\ \mathrm{MHz}$06 to $1.2\ \mathrm{MHz}$07, and reduced the best baseline to $1.2\ \mathrm{MHz}$08 RMS, close to the $1.2\ \mathrm{MHz}$09 target for Cherenkov light discrimination in bolometric $1.2\ \mathrm{MHz}$10 experiments (Cardani et al., 2018).

The same substrate-coupled approach has now been extended beyond silicon. The first germanium-target phonon-mediated KID array reported loaded $1.2\ \mathrm{MHz}$11 values of $1.2\ \mathrm{MHz}$12–$1.2\ \mathrm{MHz}$13, internal $1.2\ \mathrm{MHz}$14 values of $1.2\ \mathrm{MHz}$15–$1.2\ \mathrm{MHz}$16, phase responsivities of $1.2\ \mathrm{MHz}$17–$1.2\ \mathrm{MHz}$18, baseline resolutions of $1.2\ \mathrm{MHz}$19–$1.2\ \mathrm{MHz}$20, and energy conversion efficiencies of $1.2\ \mathrm{MHz}$21–$1.2\ \mathrm{MHz}$22 (Delicato et al., 2024). The paper argues that the lower efficiency relative to CALDER-17 is dominated by geometry and surface condition, notably the smaller active/total metallized area ratio and the use of single-side-polished Ge, rather than by an intrinsic disadvantage of germanium as a phonon absorber.

Noise engineering has also begun to change the achievable operating point. Introducing a wideband KI-TWPA to a KIPM detector chain spanning a 70 MHz band near $1.2\ \mathrm{MHz}$23 produced a $1.2\ \mathrm{MHz}$24 improvement in the inferred detector energy resolution in the best sensor, while making explicit that passive insertion loss and TLS noise, rather than HEMT noise alone, are now the obstacles to approaching the standard quantum limit (Ramanathan et al., 2024). This result is important because KIPM detectors often favor large superconducting absorber volumes and high readout powers, a regime in which the first-stage amplifier has historically dominated the error budget.

The thermal-bolometric branch occupies a different application space. The MgB$1.2\ \mathrm{MHz}$25 TKID operated from below $1.2\ \mathrm{MHz}$26 to $1.2\ \mathrm{MHz}$27, was explicitly phonon-noise limited from $1.2\ \mathrm{MHz}$28 to $1.2\ \mathrm{MHz}$29, and at $1.2\ \mathrm{MHz}$30 achieved a measured NEP of $1.2\ \mathrm{MHz}$31, matching the expected phonon noise of $1.2\ \mathrm{MHz}$32 (Jabbari et al., 13 Mar 2026). This regime is relevant less to sub-keV particle spectroscopy than to scalable membrane-supported bolometers operating at elevated cryogenic temperatures.

6. Limitations, ambiguities, and current design directions

A persistent conceptual ambiguity is terminological: not every KID whose performance is controlled by phonons is a canonical substrate-phonon KIPM detector. Membrane-supported direct-absorption MKIDs in the mid-IR are explicitly described as direct absorbers whose energy resolution is nonetheless phonon-loss limited because recombination and downconversion phonons escape to the substrate unless a membrane suppresses that channel (Ras-Vinke et al., 26 Feb 2026). Likewise, the membrane-less Hf/In optical MKID is a direct absorber with phonon-blocking engineering rather than a remote-absorber phonon detector (Zobrist et al., 2022). By contrast, the TKID is a bolometer in which the signal is mediated by thermal phonons in a membrane thermal circuit, not by ballistic athermal phonons (Jabbari et al., 13 Mar 2026). The canonical KIPM definition remains the substrate calorimeter of the rare-event community (Temples et al., 29 Sep 2025).

For the canonical athermal-phonon architecture, the dominant present limitation is phonon collection efficiency. The most dramatic example remains the NEXUS prototype, where $1.2\ \mathrm{MHz}$33 was sub-percent, so an excellent $1.2\ \mathrm{MHz}$34 translated into a mediocre substrate-level $1.2\ \mathrm{MHz}$35 (Temples et al., 2024). The consortium summary makes the same point quantitatively: the best recent device had $1.2\ \mathrm{MHz}$36 but only $1.2\ \mathrm{MHz}$37, yielding about $1.2\ \mathrm{MHz}$38 on deposited energy (Temples et al., 29 Sep 2025). Wide-area light detectors face an analogous problem at a different scale: CALDER’s four-resonator prototype recovered only $1.2\ \mathrm{MHz}$39 of the energy deposited in the substrate, and the germanium array $1.2\ \mathrm{MHz}$40–$1.2\ \mathrm{MHz}$41, because phonons are lost to supports, inactive metal, or downconversion before they reach active inductors (Cardani et al., 2015, Delicato et al., 2024).

A second limitation is that material parameters inferred from resonator response are not always robust. In TiN optical MKIDs, the inferred downconversion efficiency scales linearly with the uncertain $1.2\ \mathrm{MHz}$42, and the difference between Gao’s estimate $1.2\ \mathrm{MHz}$43 and the smaller Leduc/Dridi estimate $1.2\ \mathrm{MHz}$44 propagates directly into $1.2\ \mathrm{MHz}$45 and hence into the predicted intrinsic resolving power (Hernandez et al., 2020). Disordered TiN also shows anomalous electrodynamics: a 1550 nm TiN MKID had different decay times in the frequency and dissipation quadratures at low temperature, low-temperature frequency shifts inconsistent with simple Mattis–Bardeen theory, and behavior interpreted in terms of quasiparticle traps or subgap states (Gao et al., 2012). For phonon-mediated TiN devices, this implies that pulse decay cannot always be identified with a single recombination lifetime and that quadrature choice may change the physical interpretation of the same event.

A third limitation is the transition from amplifier-limited to microphysical-noise-limited readout. The KI-TWPA study showed that once the HEMT contribution is reduced, TLS noise in the frequency-like quadrature and lossy passive components around the parametric amplifier become the next bottlenecks (Ramanathan et al., 2024). The consortium summary is explicit that the best current architecture is TLS-limited in frequency readout and that dissipation readout, together with a KI-TWPA, is now a serious candidate for the next step (Temples et al., 29 Sep 2025).

The design trajectory is correspondingly clear. One branch enlarges active phonon collection while reducing dead superconducting area and mounting loss. The consortium projects that, with 33 KIDs at 2% surface coverage, Nb interdigitated capacitors, negligible mount losses, and $1.2\ \mathrm{MHz}$46, a 1 g Si target could reach $1.2\ \mathrm{MHz}$47; for a 27 g Si substrate with 50 resonators and similar coverage, the projection is $1.2\ \mathrm{MHz}$48 (Temples et al., 29 Sep 2025). A second branch lowers the superconducting gap, using Hf, Ir, or AlMn to improve quasiparticle yield and to support quasiparticle trapping. A third branch, the phonon-absorber-assisted KIPM (PAA-KIPM), attempts to decouple $1.2\ \mathrm{MHz}$49 from $1.2\ \mathrm{MHz}$50 by combining large Al phonon absorbers with small low-$1.2\ \mathrm{MHz}$51 trapping segments; the consortium cites projected single-PAA-KID resolution of $1.2\ \mathrm{MHz}$52 and deposited-energy resolution of $1.2\ \mathrm{MHz}$53 at $1.2\ \mathrm{MHz}$54 and 4% area coverage (Temples et al., 29 Sep 2025). The FunKID result, with its measured factor-of-five responsivity enhancement from integrated collectors, is the clearest current proof that separating phonon absorption from microwave sensing is experimentally viable (Pesce et al., 13 Jan 2026).

Taken together, these results establish the kinetic inductance phonon-mediated detector as a detector class defined less by a single geometry than by a common transduction problem: preserving deposited energy as useful pair-breaking excitation long enough, and in a favorable enough volume, for a microwave resonator to measure it. The central unresolved issue is no longer whether MKIDs can sense phonons; it is how efficiently phonon energy can be routed, trapped, and interpreted before readout noise, dead metallization, interface loss, or anomalous superconducting electrodynamics erase the available information.