AlMn Alloy Superconducting Films

• AlMn alloy superconducting films are dilute manganese-doped aluminum films that offer tunable critical temperatures and sharp transitions ideal for TES applications.
• They are fabricated using sputtering or ion implantation methods, with process parameters like film thickness, annealing, and geometry critically affecting performance.
• Their optimized superconducting properties underpin advanced detectors in CMB, 0νββ, and X-ray microcalorimetry, highlighting the interplay of materials engineering and device design.

Searching arXiv for recent and foundational papers on AlMn alloy superconducting films and TES applications. AlMn alloy superconducting films are manganese-doped aluminum films whose superconducting transition can be tuned into the millikelvin regime required for transition-edge sensors (TESs). Across TES platforms for cosmic microwave background (CMB) instrumentation, neutrinoless double-beta decay $(0\nu\beta\beta)$ detectors, and X-ray microcalorimetry, AlMn is used because it offers tunable critical temperature $T_c$, sharp resistive transitions, and fabrication routes ranging from alloy-target sputtering to post-deposition Mn ion implantation (Lv et al., 2020, Vavagiakis et al., 2019, Liu et al., 1 Oct 2025, Zhang et al., 1 Oct 2025, Xie et al., 12 Feb 2026). A central theme in the literature is that the superconducting properties are not set by composition alone: film thickness, geometry, implantation damage, added layers, annealing history, and magnetic-field orientation all materially affect $T_c$, transition width, normal resistance, and detector-level performance.

1. Material definition and functional role

AlMn films are dilute manganese-doped aluminum superconducting films used as the absorber/thermometer material in TESs, and AlMn alloy superconducting films are described as the main components of TES devices in both bolometric and spectroscopic detector programs (Lv et al., 2020, Liu et al., 1 Oct 2025). In the TES context, the relevant operating principle is the sharp change in resistance near $T_c$, which converts a small thermal perturbation into a measurable electrical response.

Several application domains motivate the development of AlMn. For CMB-S4, AlMn$_{2000\,\mathrm{ppm}}$ is pursued because it offers simple film manufacturing and good uniformity over large areas, with target device parameters of approximately $T_C \sim 150{-}200$ mK and $R_N \sim 10{-}20$ m$\Omega$ (Vavagiakis et al., 2019). For $0\nu\beta\beta$ searches, the target $T_c$ is roughly $T_c$0, because TES detectors are expected to provide faster response and better energy resolution than neutron-transmutation-doped Ge thermistors in experiments such as CUORE (Liu et al., 1 Oct 2025). For X-ray microcalorimetry, AlMn has been presented as an alternative to bilayer TESs such as Mo/Au, Ti/Au, or Mo/Cu, with the stated advantages of a simpler fabrication process and annealing-based $T_c$1 tuning (Xie et al., 12 Feb 2026).

The literature repeatedly identifies the tunability of AlMn as the defining materials property. Manganese suppresses superconductivity in aluminum, allowing $T_c$2 to be adjusted from the kelvin scale of undoped Al toward the tens-to-hundreds-of-mK regime relevant for TES operation (Lv et al., 2020, Liu et al., 1 Oct 2025). In detector practice, this tunability is coupled to the requirement of a narrow superconducting transition: one $T_c$3 study explicitly notes that the transition width $T_c$4 is tied to temperature sensitivity, with smaller $T_c$5 giving larger responsivity (Liu et al., 1 Oct 2025).

2. Fabrication routes and process-dependent control

Two distinct fabrication paradigms are emphasized: sputtering from AlMn alloy targets and Mn ion implantation into pre-deposited Al films. In sputtering-based work, films were prepared by DC magnetron sputtering using AlMn alloy targets with Mn concentrations of 1800 ppm and 2000 ppm, with a base vacuum of about $T_c$6 Torr and a target-substrate distance of 153 mm (Liu et al., 1 Oct 2025). The sputtering rate was calibrated as

$T_c$7

where $T_c$8 is the sputtering rate and $T_c$9 is the sputtering power; increasing Ar pressure from 2 mTorr to 10 mTorr reduced the rate by only about 20%, so sputtering power was the dominant control parameter for deposition rate (Liu et al., 1 Oct 2025).

For ion-implanted films, an Al film is first grown by magnetron sputtering on $T_c$0, after which Mn ions are implanted at multiple energies to create a nearly uniform Mn distribution through the film thickness (Lv et al., 2020). The reported motivation is that single-energy implantation produces a Gaussian Mn profile in depth, so three implantation energies were chosen so the Gaussian peaks span the film thickness. For 100-nm films, typical implantation energies are 30, 65, and 100 keV; for 300-nm films, 100, 200, and 300 keV were used. SRIM/TRIM simulations and SIMS measurements confirmed that the resulting Mn profile is much more homogeneous than a single-energy implant (Lv et al., 2020). The same work states that, compared with sputtering from Mn-doped targets, ion implantation offers more precise and reliable control of Mn concentration and avoids the practical difficulty of tuning sputter targets (Lv et al., 2020).

Process stacks and interfaces also matter. Argonne’s CMB-oriented TES development used a stack of 2000 ppm AlMn on SiO$T_c$1 with a Ti 15 nm / Au 15 nm top layer, selected because it showed good critical-temperature repeatability and was derived from the Ti/Au SPT-3G fabrication flow (Vavagiakis et al., 2019). The gold cap was used to prevent oxidation and keep the top conductor electrically useful for wiring. The same study notes that lift-off patterning can produce tapered edges and promote interaction between Al and Au, and that direct Al/Au contact can form intermetallic compounds that modify superconducting behavior (Vavagiakis et al., 2019).

In X-ray TES development, annular AlMn structures introduced an additional geometric degree of freedom. One fabrication flow used a 180 nm AlMn film sputtered at 100 W DC and 5 mTorr Ar from a 2000 ppm target, followed by a $T_c$2C, 10 min anneal, wet etching into a 100 $T_c$3m diameter circle, deposition of 220 nm RF-sputtered SiO$T_c$4, 240 nm Nb electrodes, a 10 nm Ti / 200 nm Au seed layer, Au electroplating to form absorbers, and final membrane release by DRIE/Bosch etching (Zhang et al., 1 Oct 2025). A later annular design deposited Nb first by DC magnetron sputtering, patterned it by dry etching, then deposited a 300 nm AlMn film from a 2000 ppm target and annealed at $T_c$5C for 10 minutes to tune $T_c$6 near 100 mK (Xie et al., 12 Feb 2026).

3. Superconducting transition tuning and transition sharpness

The experimentally accessible range of $T_c$7 in AlMn films spans from undoped Al behavior near 1.2 K down to the $T_c$8 mK range. In ion-implanted films, undoped Al showed $T_c$9 K, with $T_c$0, and for a 300-nm film as high as 7.6, indicating good film quality (Lv et al., 2020). Once Mn was implanted, all samples showed dramatic suppression of superconductivity: $T_c$1 dropped below 100 mK while the resistive transition remained sharp, with widths less than 20 mK (Lv et al., 2020).

Annealing is a major tuning knob in sputtered AlMn. In $T_c$2-motivated films, a 150 nm, 2000 ppm AlMn film annealed at $T_c$3 reached $T_c$4, directly placing it in the target TES window (Liu et al., 1 Oct 2025). In the interval $T_c$5 to $T_c$6, the $T_c$7-annealing relation for 150 nm and 180 nm, 2000 ppm films was approximately linear, with slopes around 2.43 and 2.55 (Liu et al., 1 Oct 2025). For 1800 ppm films, the explicit linear fits were reported as

$T_c$8

for 100 nm films, and

$T_c$9

for 200 nm films (Liu et al., 1 Oct 2025). The same study concludes that increasing Mn content makes $_{2000\,\mathrm{ppm}}$0 less sensitive to annealing temperature, whereas decreasing Mn content raises $_{2000\,\mathrm{ppm}}$1 and increases the slope of the $_{2000\,\mathrm{ppm}}$2-annealing relation (Liu et al., 1 Oct 2025).

Transition width is treated as a fabrication constraint as much as a materials property. In the $_{2000\,\mathrm{ppm}}$3 films, $_{2000\,\mathrm{ppm}}$4 was about 2 mK at first and then increased as annealing temperature rose; above about $_{2000\,\mathrm{ppm}}$5, the transition width grew noticeably (Liu et al., 1 Oct 2025). An annealing window around $_{2000\,\mathrm{ppm}}$6 was therefore identified as especially useful because it preserves a small $_{2000\,\mathrm{ppm}}$7 while remaining compatible with standard fabrication constraints (Liu et al., 1 Oct 2025).

Argonne’s TES development for CMB-S4 further shows that thickness, geometry, and bake conditions alter both $_{2000\,\mathrm{ppm}}$8 and $_{2000\,\mathrm{ppm}}$9 across a broad range (Vavagiakis et al., 2019). Examples reported for AlMn$T_C \sim 150{-}200$0 100 nm / Ti 5 nm / Au 20 nm include $T_C \sim 150{-}200$1 mK at 0 °C bake, $T_C \sim 150{-}200$2 mK after $T_C \sim 150{-}200$3C for 10 min, and $T_C \sim 150{-}200$4 mK after a sequence of $T_C \sim 150{-}200$5C, $T_C \sim 150{-}200$6C, and $T_C \sim 150{-}200$7C bakes (Vavagiakis et al., 2019). For AlMn$T_C \sim 150{-}200$8 550 nm / Ti 15 nm / Au 15 nm, reported values include $T_C \sim 150{-}200$9 mK and $R_N \sim 10{-}20$0 mK at 0 °C, $R_N \sim 10{-}20$1 mK and $R_N \sim 10{-}20$2 mK after $R_N \sim 10{-}20$3C anneals, and $R_N \sim 10{-}20$4 mK or $R_N \sim 10{-}20$5 mK for sequences that combined short and long bakes at $R_N \sim 10{-}20$6C (Vavagiakis et al., 2019). These data explicitly show that the same nominal material system can be driven across widely separated TES operating points by thermal processing.

4. Magnetic-impurity physics, pair breaking, and defect engineering

The microscopic interpretation of Mn in Al is a central point of the literature. Ion-implanted AlMn films were analyzed with both the Abrikosov–Gor'kov (AG) magnetic-impurity model and the Kaiser non-magnetic-impurity model, using the implicit transition-temperature relation

$R_N \sim 10{-}20$7

where $R_N \sim 10{-}20$8 is the digamma function, $R_N \sim 10{-}20$9 is the transition temperature of the undoped superconductor, and $\Omega$0 is the pair-breaking parameter (Lv et al., 2020). For magnetic impurities in the AG picture, $\Omega$1 and

$\Omega$2

with $\Omega$3 corresponding to ferromagnetic coupling and $\Omega$4 to antiferromagnetic coupling (Lv et al., 2020).

The fitted values were described as inconsistent with the Kaiser picture: $\Omega$5 for 100 nm and $\Omega$6 for 300 nm, which the authors called unrealistic because the model would normally predict suppression of electron-phonon coupling rather than enhancement (Lv et al., 2020). By contrast, the AG fits were interpreted as evidence that Mn dopants behave as magnetic impurities in ion-implanted Al, and that the 300-nm films have a smaller antiferromagnetic coupling parameter than the 100-nm films (Lv et al., 2020). The reported AG critical concentrations were $\Omega$7 ppm for 100 nm and $\Omega$8 ppm for 300 nm, while the unannealed 300-nm sample had $\Omega$9 (Lv et al., 2020).

The physical mechanism proposed is pair breaking by localized Mn moments, modified by Ruderman–Kittel–Kasuya–Yosida (RKKY) exchange among Mn impurities. The relevant interaction in the implanted films was interpreted as antiferromagnetic and therefore partially counteracting the destruction of superconductivity by individual magnetic impurities (Lv et al., 2020). That conclusion was explicitly contrasted with earlier sputtered AlMn work that inferred non-magnetic behavior from a sharp BCS-like tunneling density of states (Lv et al., 2020). The resulting controversy is not that one interpretation is simply incorrect, but that the dopant magnetic character depends strongly on how the film is prepared. A plausible implication is that local disorder alters both Mn moments and Mn–Mn coupling.

Defect engineering enters through implantation damage and annealing. The displacements per atom were defined as

$0\nu\beta\beta$0

where $0\nu\beta\beta$1 is the atomic displacement cross-section and $0\nu\beta\beta$2 is the radiation flux spectrum (Lv et al., 2020). TRIM simulations showed that 300-nm films experience more severe radiation damage than 100-nm films, and the authors argued that this greater defect density suppresses RKKY exchange among Mn impurities (Lv et al., 2020). Post-implantation annealing at $0\nu\beta\beta$3C and $0\nu\beta\beta$4C increased the fitted magnetic coupling parameter from $0\nu\beta\beta$5 in the unannealed state to $0\nu\beta\beta$6 and $0\nu\beta\beta$7, respectively, which was interpreted as defect healing that restores exchange coupling between Mn moments (Lv et al., 2020).

A related but experimentally distinct annealing mechanism was reported in sputtered $0\nu\beta\beta$8 films. TOF-SIMS showed strong Mn accumulation near the film/silicon nitride interface, and with increasing annealing temperature the Mn intensity near the surface increased, the Mn concentration in the middle depth decreased, and the overall depth profile became more homogeneous (Liu et al., 1 Oct 2025). This was suggested as a possible mechanism for $0\nu\beta\beta$9 tuning, and the authors connected it to earlier hypotheses that annealing may restore exchange coupling among Mn impurities or influence Mn clustering (Liu et al., 1 Oct 2025).

5. Device architectures and geometry-dependent behavior

The dominant AlMn TES geometries in the cited work are rectangular patterned films for bolometric arrays and annular films for X-ray microcalorimeters. In the CMB-S4 context, geometry was used as a tuning parameter for resistance and transition temperature. For AlMn$T_c$0 200 nm / Ti 15 nm / Au 15 nm TES films, the following geometries were reported: $T_c$1m with $T_c$2 mK and $T_c$3, $T_c$4m with $T_c$5 mK and $T_c$6, $T_c$7m with $T_c$8 mK and $T_c$9, and $T_c$00m with $T_c$01 mK and $T_c$02 (Vavagiakis et al., 2019). A representative TES island geometry was identified as AlMn 200 nm / Ti 15 nm / Au 15 nm, 15 $T_c$03m long by 80 $T_c$04m wide, with Pd for heat-capacity stabilization, Nb leads, and a load resistor on the island (Vavagiakis et al., 2019).

Annular AlMn TESs were introduced to decouple normal resistance from thermal conductance more flexibly than conventional rectangular designs (Zhang et al., 1 Oct 2025). In the first reported implementation, the effective TES region was a ring with inner radius $T_c$05 and outer radius $T_c$06, giving the normal-resistance relation

$T_c$07

where $T_c$08 is the resistivity and $T_c$09 is the AlMn thickness (Zhang et al., 1 Oct 2025). Using the measured resistivity $T_c$10, the estimated normal resistance was $T_c$11 (Zhang et al., 1 Oct 2025). The membrane thermal conductance was described by

$T_c$12

with $T_c$13 related to the outer perimeter times the membrane thickness, so increasing the outer perimeter increases thermal conductance (Zhang et al., 1 Oct 2025).

A later annular architecture used a 300 nm AlMn ring with inner radius 28 $T_c$14m and outer radius 45 $T_c$15m, with an inner Nb electrode of width 10 $T_c$16m connected through a 40° notch in the annular TES (Xie et al., 12 Feb 2026). The gold absorber measured $T_c$17, had thickness 1.7 $T_c$18m, and was suspended above the TES by five gold pillars (Xie et al., 12 Feb 2026). That work notes that the curved geometry can broaden the superconducting transition due to nonuniform current density (Xie et al., 12 Feb 2026). This suggests that the annular geometry is not merely a layout change; it modifies the electrodynamic operating conditions of the TES.

6. Electrical, thermal, and magnetic-field characteristics

The standard transport definitions are largely common across the papers. In the $T_c$19 study, $T_c$20 was defined as the normal-state resistance, $T_c$21 as the temperature at which $T_c$22, and $T_c$23 as the interval between $T_c$24 and $T_c$25 (Liu et al., 1 Oct 2025). In the Argonne study, $T_c$26 was defined at 50% of $T_c$27, while $T_c$28 was taken 2 mK above the last superconducting data point on the $T_c$29 curve (Vavagiakis et al., 2019). Four-terminal sensing was used in both film and device characterization to remove lead and contact resistance (Vavagiakis et al., 2019, Liu et al., 1 Oct 2025).

A Ginzburg–Landau-type critical-current relation was reported for AlMn films in the $T_c$30 work: $T_c$31 with

$T_c$32

(Liu et al., 1 Oct 2025). This was presented as confirmation that the films follow expected superconducting phenomenology.

Magnetic-field response shows a notable anisotropy. In low-$T_c$33 AlMn films intended for $T_c$34, the out-of-plane magnetic field significantly suppressed $T_c$35, with an experimentally observed slope

$T_c$36

A change of about $T_c$37 lowered $T_c$38 from $T_c$39 to $T_c$40, while the in-plane magnetic field had no significant effect within the measured range (Liu et al., 1 Oct 2025). The same study noted that room-temperature magnetization at 1.1 T, 4 T, and 7 T for 10 minutes had little measurable effect on $T_c$41, implying that the relevant concern is the magnetic field present during low-temperature operation (Liu et al., 1 Oct 2025).

This should be read alongside later X-ray TES work, which still described AlMn as less sensitive to magnetic fields than bilayer TESs, while simultaneously devoting substantial attention to shielding (Xie et al., 12 Feb 2026). In the 2026 annular X-ray detector, a composite shield consisting of a 1.5 mm Cryoperm 10 cover plate, a 2 mm Nb bottom plate, and an aluminum enclosure around the SQUID was modeled with COMSOL to reduce the magnetic field near the TES to $T_c$42T and the field at the SQUID location to $T_c$43T (Xie et al., 12 Feb 2026). A plausible implication is that “lower magnetic-field sensitivity” is relative rather than absolute, and that very low-$T_c$44 operation still requires careful field control.

Thermal characterization in full TES devices used the standard power-law form

$T_c$45

or

$T_c$46

with thermal conductance extracted from

$T_c$47

For Argonne devices with 865 $T_c$48m and 446 $T_c$49m legs, best-fit values were $T_c$50 mK, $T_c$51 pW/K, $T_c$52 pW/K$T_c$53, $T_c$54 m$T_c$55, $T_c$56, and $T_c$57 mK, $T_c$58 pW/K, $T_c$59 pW/K$T_c$60, $T_c$61 m$T_c$62, $T_c$63, respectively (Vavagiakis et al., 2019). In annular X-ray TESs, the measured thermal exponents were around 3.2–3.3 in one study and 3.68 in another (Zhang et al., 1 Oct 2025, Xie et al., 12 Feb 2026).

7. Detector implementations and measured performance

The detector-level record of AlMn films spans CMB bolometers, $T_c$64 light detectors, and X-ray TES microcalorimeters. For CMB-S4-oriented devices at Argonne, time-division SQUID multiplexing was used to determine $T_c$65, $T_c$66, $T_c$67, $T_c$68, and $T_c$69, with the extracted $T_c$70 m$T_c$71 close to the design goal of $T_c$72 m$T_c$73 (Vavagiakis et al., 2019). Bias-step measurements showed that the time constant decreases substantially as the detectors are biased lower on the superconducting transition, consistent with previous AlMn TES work (Vavagiakis et al., 2019).

For $T_c$74 detectors, the key materials result is that sputtered and annealed AlMn can be tuned into the required $T_c$75 mK range while maintaining small $T_c$76 over a useful fabrication window (Liu et al., 1 Oct 2025). The same work explicitly identifies strong sensitivity to out-of-plane field as a systems-level constraint requiring magnetic shielding (Liu et al., 1 Oct 2025).

X-ray TES demonstrations provide the most direct evidence that AlMn films support high-resolution microcalorimetry. The 2025 annular AlMn TES study evaluated three devices with absorber sizes $T_c$77, $T_c$78, and $T_c$79 (Zhang et al., 1 Oct 2025). Their $T_c$80 values were 115.0, 115.2, and 114.1 mK, and the tested FWHM at 5.9 keV was $T_c$81 eV, $T_c$82 eV, and $T_c$83 eV, respectively (Zhang et al., 1 Oct 2025). The I–V curves were reported as highly consistent across the three detectors, with measured normal resistance about 27 m$T_c$84 for all three, higher than the design estimate of 19.7 m$T_c$85 (Zhang et al., 1 Oct 2025). The discrepancy was attributed to film-thickness deviation and possible contribution from the AlMn layer beneath the Nb electrodes (Zhang et al., 1 Oct 2025).

The same study also reported a theoretical best resolution of 3.5 eV for the smallest-absorber detector, versus a measured 11.0 eV at 5.9 keV, and attributed the difference mainly to readout electronics noise, with average current noise below 20 kHz of about $T_c$86 (Zhang et al., 1 Oct 2025). It additionally noted that the heat capacity was 3–4 times larger than the expected value from reference models, with the cause not yet explained (Zhang et al., 1 Oct 2025).

The 2026 annular AlMn TES advanced this line by reporting intrinsic FWHM energy resolutions of $T_c$87 eV at 5.9 keV, $T_c$88 eV at 8.0 keV, and $T_c$89 eV at 17.48 keV (Xie et al., 12 Feb 2026). The $T_c$90 keV result corresponds to $T_c$91 energy resolution, and was explicitly described as the first AlMn TES X-ray detector to surpass one-thousandth relative energy resolution (Xie et al., 12 Feb 2026). That device had $T_c$92 m$T_c$93, $T_c$94 mK, $T_c$95 pW/K at 98.4 mK, $T_c$96, $T_c$97, loop gain $T_c$98, measured decay constant $T_c$99 ms, intrinsic decay time $T_c$00 ms, and total heat capacity 0.6 pJ/K (Xie et al., 12 Feb 2026). The estimated fundamental limit from the noise budget was 3.4 eV, so excess noise remained dominant (Xie et al., 12 Feb 2026).

Taken together, these results establish AlMn alloy superconducting films as a tunable TES material platform whose practical performance depends on the interplay of alloy concentration, annealing, geometry, interfaces, defects, and magnetic environment. The literature supports two simultaneous conclusions: AlMn is operationally attractive because it enables simple and scalable fabrication, and its superconducting properties remain highly process-sensitive, so detector optimization is inseparable from detailed materials engineering (Vavagiakis et al., 2019, Lv et al., 2020, Liu et al., 1 Oct 2025, Zhang et al., 1 Oct 2025, Xie et al., 12 Feb 2026).