Metallic local-moment magnetocalorics as a route to cryogenic refrigeration

The study addresses the need for compact, efficient, and low-maintenance sub-Kelvin refrigeration to support quantum technologies and advanced condensed matter experiments. Conventional wet helium-based systems are complex and costly, while adiabatic demagnetisation refrigeration (ADR) offers compactness and operational simplicity but relies on suitable magnetocaloric materials. Insulating salts and garnets used in commercial ADR suffer from low thermal conductivity at low temperatures, limited entropy density, and handling challenges. Metallic magnetocalorics could overcome these limitations by providing higher thermal conductivity and greater entropy density, enabling miniaturisation and improved performance. The central research question is how to identify metallic rare-earth intermetallics that retain large magnetic entropy to very low temperatures without magnetic ordering or strong Kondo screening. The authors propose and investigate YbNi1−xSn as a metallic magnetocaloric that resides in the extreme local-moment regime with exceptionally small Kondo and RKKY interactions, aiming to demonstrate superior entropy characteristics and practical ADR performance.

The framework is based on the Doniach phase diagram for f-electron systems, balancing Kondo screening (TK) and RKKY interactions (TRKKY). Traditional guidance targets quantum critical points where magnetic order is suppressed, but such systems retain significant correlations and field-insensitive entropy at low temperatures, limiting ADR effectiveness. Prior metallic candidates include YbPt2Sn, YbPd2Sn, YbPd2In, YbPt2In, YbCu2Ni, CePt4Sn25, and the diluted heavy fermion Yb0.81Sc0.19Co2Zn20. These exhibit varying degrees of Kondo physics, magnetic ordering, and frustration; QC systems tend to shift entropy recovery to higher temperatures and remain partially polarised at high fields, reducing magnetocaloric efficiency. Some Yb and Ce intermetallics display near-local-moment behavior down to ~0.3 K and below, implying anomalously small exchange J and placing them on the extreme left of the Doniach diagram. The authors position YbNi1.6Sn within this local-moment class, contrasting it with QC systems like Yb0.81Sc0.19Co2Zn20 and YbCu4.6Au0.4, and with insulating salts (CPA, FAA) that have lower volumetric entropy and poor thermal conductivity.

Material synthesis: Polycrystalline YbNi1−xSn was prepared by a two-stage arc-melting process under ultrapure argon. First, Yb and Sn were melted to a button; then Ni was added and repeatedly remelted with the pre-reacted YbSn mixture (optimal mass ratio 0.4 Ni : 1 YbSn). An 8–14% excess of Yb compensated for evaporation losses (total mass loss ~4 wt%). The method is simple and scalable; phase purity is sufficient for ADR use though not maximised. Structural and chemical characterization: Room-temperature X-ray powder diffraction (STOE Stadip, Cu Kα1) confirmed an fcc Fm-3m structure with lattice parameter a = 6.3645(2) Å and refined occupancies near Yb:Ni:Sn = 1:1.6:1. Minor fcc YbNi4Sn was detected. Differential scanning calorimetry (PerkinElmer DSC 8500) between 300–1500 K and EDX (Philips XL30 SEM) were used. The fcc YbNi1.6Sn phase is metastable below 1300 K but long-term stable below 500 K. Physical property measurements: Electrical resistivity and specific heat were measured from 400 mK to 400 K in a QD PPMS with 3He option; millikelvin specific heat down to ~90 mK was measured by relaxation (3He/4He dilution refrigerator, compensated heat pulse). Magnetisation above 1.8 K was measured using a QD SQUID VSM. Nuclear specific heat contributions were modelled and subtracted to obtain the electronic heat capacity; Cph was neglected in low-T entropy analysis; Celec was integrated to obtain Selec. Entropy landscape: Selec(T,B) was constructed from Celec(T,B), and Stotal(T,B) was estimated including nuclear and electronic contributions (phonons negligible at low T). Maxwell relation (∂S/∂B)T = (∂M/∂T)B is noted as an alternative but not primarily used. ADR prototype testing: A simple ADR module for PPMS was built with ~15 g of pressed YbNi1.6Sn powder in a thin-walled brass can, instrumented with a calibrated RuO2 thermometer and a 200 Ω thin-film heater. Thermal anchoring used NbTi wires in CuNi shields routed and varnish-bonded to minimise heat leaks. Demagnetisation runs were performed from Tbath = 1.8 K at initial fields Bi = 4 T, 9 T, 14 T with controlled ramp rates; base temperatures and hold times were recorded under parasitic and applied heat loads (0–5 μW). Measured field–temperature trajectories were compared with computed isentropes from Stotal(T,B).

• YbNi1.6Sn is a metallic, non-superconducting fcc intermetallic (Fm-3m, a = 6.3645(2) Å) with refined composition ~Yb:Ni:Sn = 1:1.6:1; minor YbNi4Sn phase present.
• Transport and magnetism: Resistivity increases monotonically with T; no Kondo scattering signature. Magnetic analysis indicates stable Yb3+ with extraordinarily weak exchange interactions; inferred TK ≤ 1 K and ultra-weak TRKKY.
• Specific heat and entropy: After subtracting nuclear contributions, Celec/T shows a broadened anomaly at Tm ≈ 140 mK, consistent with extremely weak intersite coupling and likely short-range order; Selec(T, B=0) reaches ~R ln 2, confirming a CEF doublet ground state. Applied fields split the doublet, shifting broad Schottky-like maxima to higher T with relatively strong field dependence favorable for ADR.
• Entropy landscape advantage: Compared with QC systems (Yb0.81Sc0.19Co2Zn20; YbCu4.6Au0.4), YbNi1.6Sn concentrates entropy at sub-Kelvin T with stronger field tunability, enabling larger isothermal ΔS for practical fields.
• Volumetric entropy density change: At 1.4 K, ΔS (8 T − 0 T) per volume is at least 2.5× higher than in common insulating ADR salts CPA and FAA, and remains larger down to ~230 mK.
• Nuclear entropy impact: Nuclear contributions negligibly affect Tbase above ~200 mK but become important below ~200 mK and strongly relevant below ~100 mK, potentially limiting Tbase above values inferred from electronic entropy alone.
• ADR performance (prototype, ~15 g refrigerant, Tbath = 1.8 K):
  • 14 T → 0 T: Tbase = 116 mK, estimated parasitic heat leak Q̇ < 0.4 μW.
  • 9 T → 0 T: Tbase ≈ 145 mK at ~0.05 μW; ≈192 mK at ~2.5 μW heat load.
  • 4 T → 0 T: Tbase ≈ 233–240 mK at ~0.05 μW.
  • Hold times below 350 mK exceed 420 min in a 4 T run, surpassing comparable systems (e.g., YbPt2Sn ≈160 min below 350 mK under similar cryostat conditions).
• Practical advantages: Intrinsically higher thermal conductivity than insulating salts simplifies pill design (no metal meshes), improves miniaturisation, and boosts effective volumetric entropy capacity; material avoids precious Pd/Pt used in other metallic magnetocalorics (e.g., YbPt2Sn).
• Position in Doniach diagram: Data place YbNi1.6Sn deep in the extreme local-moment regime with vanishingly small J, ultra-low TK and TRKKY, and minimal magnetic ordering effects (Tm ~ 0.14 K).

The findings demonstrate that the extreme local-moment character of YbNi1.6Sn yields a favorable S(T,B) landscape for ADR: high low-temperature entropy near R ln 2 at zero field, strong field dependence enabling large ΔS upon magnetisation, and minimal suppression by Kondo screening or RKKY-driven correlations. In contrast, QC systems shift significant entropy to higher temperatures and retain field-insensitive correlated entropy at high fields, limiting ADR performance for realistic fields. The observed broadened low-T anomaly (Tm ≈ 140 mK) suggests only weak short-range correlations, thus imposing a less stringent floor on achievable base temperatures than in related intermetallics with higher Tm. The prototype ADR results validate the thermodynamic expectations with base temperatures down to 116 mK and long hold times even under μW-scale loads, supporting real-world applicability. Discrepancies between measured demagnetisation trajectories and calculated isentropes underline the importance of accounting for nuclear entropy at very low temperatures and potential parasitic thermal masses; they also suggest that assumptions about Celec/T → 0 at the lowest T may require refinement for weakly interacting local-moment systems. Overall, YbNi1.6Sn’s combination of high entropy density, metallic thermal conductivity, and absence of precious metals offers a practical route to compact, efficient cryogenic refrigeration below 200 mK.

YbNi1−xSn emerges as a superior metallic magnetocaloric for sub-Kelvin ADR. It combines high entropy density, strong magnetocaloric response, chemical stability, UHV bakeability, and high thermal conductivity, while avoiding precious metals (Pd/Pt) used in comparable intermetallics. Thermodynamic measurements place it in the extreme local-moment regime with TK and TRKKY well below 1 K and only weak low-T magnetic signatures (Tm ~140 mK). Prototype ADR tests with ~15 g refrigerant achieved Tbase = 116 mK (14 T→0 T) and long hold times below 350 mK under realistic heat loads, outperforming common insulating salts (CPA, FAA) in volumetric ΔS by at least a factor of 2.5 at 1.4 K. Future work could target materials with higher CEF ground-state degeneracy to further increase entropy density, and integrate YbNi1−xSn in multi-stage systems (e.g., with paramagnetic salts or PrNi5 nuclear stages) to reach the 10 mK range.

• Nuclear entropy becomes significant below ~200 mK and strongly limits Tbase below ~100 mK; full Stotal(T,B) must include nuclear contributions for accurate performance predictions.
• Discrepancies between measured ADR trajectories and computed isentropes indicate uncertainties in low-T extrapolations (e.g., assumed Celec/T behavior) and potential parasitic thermal masses.
• The functional fcc YbNi1.6Sn phase is metastable below ~1300 K (though long-term stable below 500 K), and samples contain a minor YbNi4Sn phase; phase purity is limited by the simple, scalable synthesis route.
• Further measurements at temperatures well below the current range are needed to refine the understanding of low-T physics (e.g., exact Celec/T behavior, TK) and nuclear contributions.
• While ADR performance is strong above ~150–200 mK, achieving base temperatures deep below 100 mK may require integration with additional stages (e.g., salts or nuclear demagnetisation).