Phonon-Blocked Junction Microcalorimeter

• The paper demonstrates that junctions can serve as phonon transducers, thermal isolators, and on-chip refrigerators in advanced cryogenic applications.
• Phonon-blocked junction microcalorimeters are cryogenic detectors that use engineered phonon suppression to achieve high sensitivity and precise energy resolution.
• Experimental architectures reveal design trade-offs, balancing electron cooling efficiency with phonon isolation to optimize detector performance.

A phonon-blocked junction microcalorimeter is a cryogenic calorimetric architecture in which thermal isolation is obtained by suppressing phonon transport across engineered junctions, while the same or adjacent junctions provide refrigeration, thermometry, or quasiparticle-sensitive readout. In the literature summarized here, the concept appears in several closely related forms: superconducting tunnel-junction phonon transducers for spectrally resolved ballistic phonons, suspended semiconductor–superconductor junctions with strong phonon boundary resistance, semiconductor–superconductor junction refrigerators that use the junction as both electronic cooler and phonon bottleneck, and a dedicated cold-electron microcalorimeter framework based on phonon-blocked NIS junctions (Otelaja et al., 2013, Mykkänen et al., 2018, Mykkänen et al., 2022, Geng et al., 7 Aug 2025).

1. Historical formation of the concept

A first essential lineage is the use of superconducting tunnel junctions as phonon transducers. A microfabricated phonon spectrometer based on Al–Al$_x$O$_y$–Al STJs demonstrated emission and detection of tunable, non-thermal, and spectrally resolved acoustic phonons in silicon microstructures, with frequencies ranging from $\sim 100$ to $\sim 870\ \text{GHz}$ and a spectral resolution of $\sim 15\text{–}20\ \text{GHz}$ (Otelaja et al., 2013). That work established that a junction can function not only as an electrical element but as a spectrally selective phonon emitter and detector, which is foundational for later calorimetric uses of junctions in phonon engineering.

A second lineage is the emergence of junctions as thermal weak links with intrinsic phonon isolation. In suspended degenerately doped Si–Al semiconductor–superconductor devices, the Sm–S interface was shown to combine efficient thermionic electron cooling with phonon transfer blocking arising from the transmission bottleneck at the junction. These devices suspended different size degenerately doped silicon chips directly from the junctions and cooled them by biasing the junctions, with a measured total thermal resistance $R(T)=aT^{-3}$ and explicit decomposition into interfacial thermal resistance $R_{\mathrm{PTB}}$ and lead resistance $R_{\text{lead}}$ (Mykkänen et al., 2018).

A third lineage generalized the same idea from cooling to broader cryogenic thermal engineering. Junction refrigerators based on phonon-blocked semiconductor–superconductor junctions were proposed as solid-state cooling platforms for quantum devices, with the junction simultaneously acting as an electronic cooler for semiconductor electrons and a phonon bottleneck between semiconductor and superconductor (Mykkänen et al., 2022). This directly anticipates calorimetric architectures in which the junction defines the dominant thermal conductance.

The term itself is made explicit in the dedicated cold-electron formulation. The paper “Phonon-blocked junction calorimeter” introduces a cold-electron microcalorimeter based on phonon-blocked junctions, integrating on-chip electron cooling and boundary phonon isolation, and presents a general theoretical framework with approximate analytical expressions for cooling efficiency, thermal time constant, and energy resolution (Geng et al., 7 Aug 2025). In parallel, low-gap Josephson junction development has supplied an additional sensor-side ingredient: Hf–HfO$_x$–Hf junctions were explicitly framed as enabling low-gap, quasiparticle-sensitive elements for meV-scale phonon and THz-photon detection architectures, although that work did not fabricate a full phonon-blocked calorimeter (Balaji et al., 29 Oct 2025).

2. Physical basis of phonon blocking

The most direct phonon-blocking mechanism is interfacial thermal boundary resistance. In the cold-electron NIS formulation, phonon heat flow is written as

$P_{ph} = \frac{1}{4a_{eITR}}(T_B^4-T_N^4),$

where $_y$0 is the effective interfacial thermal resistance coefficient, $_y$1 is the bath temperature, and $_y$2 is the absorber temperature (Geng et al., 7 Aug 2025). In the suspended Sm–S experiments, the total phonon thermal resistance is modeled as

$_y$3

with the measured aggregate behavior

$_y$4

so that the corresponding heat flow becomes

$_y$5

This $_y$6 heat-flow form is the low-temperature signature of boundary-limited phonon transport across the junction-plus-lead system (Mykkänen et al., 2018).

A second blocking mechanism is quasi-1D phonon constriction. In phonon-blocked Sm–S junction refrigerators, the total phonon thermal resistance is written as

$_y$7

and the phonon heat leak as

$_y$8

with $_y$9 for a planar boundary and 3D leads, $\sim 100$0 for diffusive quasi-1D nanowire constrictions, and $\sim 100$1 in the ballistic 1D thermal-conductance-quantum limit. The upper bound is set by the phonon thermal conductance quantum

$\sim 100$2

This framework is directly relevant to microcalorimeters because it provides a route from interfacial blockade to mode-count blockade (Mykkänen et al., 2022).

A third mechanism is geometric scattering and resonance in junctioned nanostructures. In a 1D silicon nanophononic metamaterial with cross junctions, phonon transport reduction was separated into particle and wave contributions by a combined Monte Carlo plus atomic Green’s function method. The wave contribution fraction is

$\sim 100$3

and the particle effect can be as high as $\sim 100$4 for a 4-legs-junction NCJ when the cross-sectional area is $\sim 100$5 (Ma et al., 2018). This result is significant because low-frequency phonons are strongly affected by wave effects through resonance hybridization, whereas high-frequency phonons remain strongly suppressed by particle scattering even when junction legs are reduced to one atomic layer. A related 3D silicon-nanowire-cage study found $\sim 100$6 at 300 K for the smallest reported cage bars and showed that local resonance and hybridization at nano-cross-junctions can yield strong localization without requiring strict periodicity; a periodic structure with $\sim 100$7 and a random 3D-NCJ structure with $\sim 100$8 differed by only about $\sim 100$9 (Ma et al., 2015). In calorimetric language, these results mean that junctions can block phonons through both spectral and geometric channels.

3. Junction functions: refrigeration, thermometry, and quasiparticle sensing

In cold-electron implementations, the junction is simultaneously a thermal link, refrigerator, and thermometer. The absorber is a large normal metal electrode with heat capacity

$\sim 870\ \text{GHz}$0

connected to two identical NIS tunnel junctions. The absorber temperature is read out through the tunnel current, and the same junctions extract hot electrons when biased near the optimal cooling point (Geng et al., 7 Aug 2025). In the Sm–S suspended devices, the simplified optimal-bias cooling expression is written as

$\sim 870\ \text{GHz}$1

with

$\sim 870\ \text{GHz}$2

so the achievable base temperature depends directly on the competition between ideal cooling and sub-gap leakage (Mykkänen et al., 2018).

A distinct readout modality is thermoelectric. In the superconductor–ferromagnet tunnel-junction thermoelectric detector, the junction is unbiased and the signal arises from the giant thermoelectric effect in a spin-split superconductor with spin filtering. The small-signal current is

$\sim 870\ \text{GHz}$3

and the loop-gain analogue is

$\sim 870\ \text{GHz}$4

This formulation is valuable beyond SFTEDs because it makes explicit how junction transduction, thermal isolation, and readout inductance generate coupled electrothermal dynamics (Geng et al., 2022).

A further sensor-side development is the use of low-gap Josephson junctions as quasiparticle traps and readout elements in phonon-detection architectures. For Hf–HfO$\sim 870\ \text{GHz}$5–Hf SIS junctions with area $\sim 870\ \text{GHz}$6, the measured parameters include $\sim 870\ \text{GHz}$7, $\sim 870\ \text{GHz}$8, and a junction gap $\sim 870\ \text{GHz}$9 with $\sim 15\text{–}20\ \text{GHz}$0 (Balaji et al., 29 Oct 2025). The same study states explicitly that it does not fabricate a full phonon-blocked calorimeter, but instead provides the low-gap, quasiparticle-sensitive junction element for such architectures. This distinction is important: phonon blocking is not identical to low-gap sensing, but low-gap sensing becomes especially effective once phonon escape is reduced.

4. Thermal dynamics, pulse formation, and performance metrics

The defining thermal equation of the dedicated cold-electron phonon-blocked junction calorimeter is

$\sim 15\text{–}20\ \text{GHz}$1

where $\sim 15\text{–}20\ \text{GHz}$2 is absorbed signal power and $\sim 15\text{–}20\ \text{GHz}$3 parametrizes heat return from the superconducting side (Geng et al., 7 Aug 2025). This equation expresses the two central design tensions: the calorimeter must have weak enough thermal coupling to the bath to produce a large pulse, but not so weak that recovery is impractically slow; and the junctions must cool efficiently without being compromised by subgap tunneling or heat backflow.

Within the thermoelectric analytical model, pulse shape is governed by the competition of $\sim 15\text{–}20\ \text{GHz}$4, $\sim 15\text{–}20\ \text{GHz}$5, and $\sim 15\text{–}20\ \text{GHz}$6. The two-eigenvalue case without an RC shunt yields rise and decay constants

$\sim 15\text{–}20\ \text{GHz}$7

Critical damping occurs at $\sim 15\text{–}20\ \text{GHz}$8, while $\sim 15\text{–}20\ \text{GHz}$9 gives an underdamped oscillatory response (Geng et al., 2022). This is directly relevant to junction calorimeters with SQUID readout because the junction and the input coil generically form a coupled electrothermal-electrical system.

Two phonon-mediated comparison points show why phonon collection and phonon blocking are inseparable in practice. In a 1-gram silicon kinetic-inductance phonon-mediated detector, the baseline resolution on the energy absorbed by the phonon sensor was $R(T)=aT^{-3}$0, but the resolution on energy deposited in the substrate was limited to $R(T)=aT^{-3}$1 because the phonon collection efficiency was sub-percent (Temples et al., 2024). In a contact-less Al KID on a 30-gram silicon crystal, the RMS energy resolution was about $R(T)=aT^{-3}$2, and the result was mainly limited by a conversion efficiency of about $R(T)=aT^{-3}$3 from deposited energy to quasiparticles (Goupy et al., 2019). Although these are not junction calorimeters, they quantify a central implication for phonon-blocked junction devices: once intrinsic junction sensitivity becomes good, performance is often limited by uncontrolled phonon escape rather than by the readout element itself. This suggests that boundary phonon isolation and phonon collection efficiency are coequal design variables.

5. Implemented architectures and experimentally relevant motifs

A direct junction-isolated platform was demonstrated in suspended, degenerately doped Si–Al Sm–S devices. The sub-chip thickness was approximately $R(T)=aT^{-3}$4, each sample had 24 Sm–S junctions, and the best device showed a relative temperature reduction of $R(T)=aT^{-3}$5 at $R(T)=aT^{-3}$6 and an absolute temperature drop of $R(T)=aT^{-3}$7 at $R(T)=aT^{-3}$8 (Mykkänen et al., 2018). These experiments established that macroscopic Si islands can be suspended and cooled directly from the junctions, with phonon isolation arising from the junction contact itself rather than from a separate membrane weak link.

A second experimentally important motif is junction-side heat isolation inside a microcalorimeter. In a magnetic microcalorimeter with integrated dc-SQUID readout, the first variant was limited by Joule power dissipation of the Josephson junction shunt resistors, athermal phonon loss, and slew-rate limitations, and achieved $R(T)=aT^{-3}$9. A second variant introduced a tetrapod absorber geometry and a membrane-technique for protecting the temperature sensors against the power dissipation of the shunt resistors, achieving $R_{\mathrm{PTB}}$0 (Krantz et al., 2023). This was not a phonon-blocked junction calorimeter in the narrow NIS sense, but it demonstrates a closely related principle: junction-associated dissipation must be geometrically and thermally isolated from the absorber if one wishes to preserve calorimetric resolution.

A third motif is nanoscale support-leg phonon blocking by junctioned branches. The silicon nanowire-cage and nanowire cross-junction studies show that junctions can act as local resonators and scattering centers that strongly reduce thermal conductivity without requiring long-range periodicity (Ma et al., 2015). In the microcalorimeter context, these results are often interpreted as support-leg analogues of phonon-blocked junctions: the main beam is the thermal leg, and side branches or cross junctions provide wave-based and particle-based suppression of heat transport (Ma et al., 2018). This suggests that a full device may combine two distinct junction functions: a tunnel or Josephson junction for sensing, and phononic junctions in the support geometry for passive thermal isolation.

6. Design trade-offs, misconceptions, and open problems

One common misconception is that nanoscale phonon blocking in junctioned structures is predominantly a coherent wave effect. In the silicon nanophononic metamaterial study, the particle effect remained significant and could contribute as much as $R_{\mathrm{PTB}}$1 to the total thermal conductivity reduction; even at the smallest cross section, the particle fraction remained about $R_{\mathrm{PTB}}$2–$R_{\mathrm{PTB}}$3 depending on leg number (Ma et al., 2018). Another misconception is that strong phonon blocking requires strict periodicity. In silicon-nanowire cages, a periodic structure with $R_{\mathrm{PTB}}$4 and a random 3D-NCJ structure with $R_{\mathrm{PTB}}$5 differed by only about $R_{\mathrm{PTB}}$6, which suggests that local resonance and hybridization can survive substantial geometrical disorder (Ma et al., 2015).

A second open issue is that low-temperature thermal models cannot always assume the standard $R_{\mathrm{PTB}}$7 electron–phonon law. In a 50 nm Au film on a 300 nm SiO$R_{\mathrm{PTB}}$8 platform, micromachining and suspension changed the fitted exponent from $R_{\mathrm{PTB}}$9 on bulk to $R_{\text{lead}}$0 on the released membrane, and the change was attributed to a modified phonon spectrum (Saira et al., 2019). For phonon-blocked junction calorimeters on membranes or thin platforms, this means that extracting $R_{\text{lead}}$1, $R_{\text{lead}}$2, and $R_{\text{lead}}$3 from bulk-material formulas can be quantitatively misleading.

A third difficulty is that practical junctions are never ideal. The dedicated cold-electron phonon-blocked calorimeter framework explicitly includes non-ideal effects such as subgap tunneling and heat backflow through the $R_{\text{lead}}$4 term (Geng et al., 7 Aug 2025). In the Hf low-gap Josephson junction study, the discrepancy between $R_{\text{lead}}$5 and $R_{\text{lead}}$6 was attributed to subgap states, with a Dynes parameter estimate $R_{\text{lead}}$7 (Balaji et al., 29 Oct 2025). In Sm–S junction refrigerators, the practical design window is limited by the competition among tunnel transparency, the gap-leakage parameter $R_{\text{lead}}$8, and Andreev reflection (Mykkänen et al., 2022). A plausible implication is that the decisive engineering problem is not merely maximizing phonon blocking, but matching phonon blocking to a junction technology whose subgap leakage, backflow, and readout impedance remain compatible with stable calorimetric operation.

The unresolved program is therefore twofold. First, junction microcalorimeters must combine boundary phonon isolation, low heat capacity, and a junction transduction scheme that remains linear and well damped. Second, the thermal model must include modified phonon spectra, interfacial phonon resistances, and non-ideal quasiparticle return. The available literature shows that each ingredient has already been demonstrated separately: spectrally selective STJ phonon transduction, suspended Sm–S phonon bottlenecks, low-gap Josephson junction sensors, membrane-isolated junction dissipation, and nanophononic support-leg blocking. The remaining task is their full co-design within a single calorimetric architecture (Otelaja et al., 2013, Mykkänen et al., 2018, Krantz et al., 2023, Geng et al., 7 Aug 2025).