Proximity-Junction NanoSQUIDs

• Proximity-junction nanoSQUIDs are nanoscale superconducting quantum interference devices that use the proximity effect to create Josephson weak links with high critical current density and low noise.
• They employ normal metals or semiconductors in contact with superconductors to induce superconducting correlations, forming versatile and phase-sensitive weak links without traditional tunnel barriers.
• Advanced implementations, including scanning SQUID-on-tip probes, offer enhanced flux transfer functions and minimal noise, enabling precise magnetic imaging and multimodal sensing.

Proximity-junction nanoSQUIDs are nanoscale superconducting quantum interference devices in which the Josephson junctions are formed by weak links created through the superconducting proximity effect rather than by a tunnel barrier or a lithographically defined constriction. In this class, a normal metal or semiconducting conductor in good contact with superconducting electrodes acquires superconducting correlations, so that a proximized wire, nanobridge, or nanowire acts as a Josephson junction without a conventional insulating barrier. Within the broader nanoSQUID taxonomy, proximity-junction devices sit alongside SIS-junction devices, SNS-junction devices, constriction-junction devices, and high-$T_c$ grain-boundary devices, and they are attractive because the proximity effect can produce very small, coherent weak links with high critical current density and, in many implementations, non-hysteretic current-voltage characteristics suitable for ultrasensitive flux detection (Martínez-Pérez et al., 2016).

1. Physical basis and defining characteristics

The underlying mechanism is the superconducting proximity effect. A normal metal in good contact with superconducting electrodes acquires superconducting correlations; a mini-gap opens in the density of states of the normal metal, and Andreev pairs can propagate through it while retaining information about the superconducting phase. The junction behavior depends on whether the system is in the long- or short-junction regime, determined by the relation between the Thouless energy of the normal metal and the superconducting gap. In practical terms, this produces a phase-sensitive weak link that can be substantially smaller than a conventional junction (Martínez-Pérez et al., 2016).

The device remains a SQUID in the standard sense. Its operation relies on the Josephson relations

$$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$

together with phase quantization around the loop. For a dc SQUID, the phase differences in the two junctions satisfy

$$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$

In the small-inductance limit $\beta_L\ll 1$, the critical current becomes

$$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$

This flux-dependent interference is the basis of flux transduction: a small change in applied flux shifts the critical current and, in voltage-biased operation, generates a measurable voltage response (Martínez-Pérez et al., 2016).

The key figures of merit are the transfer function $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$, the equivalent flux noise $S_\Phi=S_V/V_\Phi^2$, and, for magnetic-particle measurements, the spin sensitivity $\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$, where $\phi_\mu\equiv \Phi/\mu$ is the coupling factor. For thermal white noise in the common $\beta_C\ll 1$, $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$0 regime, the review gives

$$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$1

This directly shows why nanoSQUID miniaturization is central: shrinking the loop lowers $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$2, improves flux noise, and strengthens magnetic coupling to localized moments (Martínez-Pérez et al., 2016).

2. Device architectures and material platforms

The review literature places proximity-junction nanoSQUIDs in a heterogeneous family of proximized interferometers. Examples explicitly identified include a dc SQUID based on a carbon nanotube intersecting an Al ring, a graphene-proximized SQUID, and proximized nanowire SQUIDs such as highly doped InAs nanowire devices enclosed in a V ring. The Superconducting Quantum Interference Proximity Transistor, or SQUIPT, is conceptually related but is read out through a tunnel probe connected to a proximized normal-metal island inside a superconducting loop rather than as a conventional dc SQUID (Martínez-Pérez et al., 2016).

A particularly clear realization is the hybrid InAs nanowire–vanadium proximity SQUID. In that device, the superconducting loop is patterned from Ti/V and interrupted at two places by weak links formed on a single highly doped InAs nanowire. The nanowires have charge density

$$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$3

diameter $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$4 nm, and length about $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$5; the contact metal is Ti/V with thickness $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$6 nm, and the contact transparency is tuned with NH$$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$7S$$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$8 passivation of the nanowire surface. Below the vanadium transition temperature, $$I_{\rm s}(t)=I_0\sin\delta(t),\qquad U(t)=\frac{\Phi_0}{2\pi}\dot{\delta},$$9, superconducting correlations are induced in the nanowire regions under the electrodes, and Josephson coupling appears through nanowire segments of length $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$0 (Spathis et al., 2010).

In this nanowire platform, the diffusion constant is $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$1, so the Thouless energy $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$2 is about one order of magnitude larger than the V gap $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$3. The junctions are therefore not in the long-junction limit but in the intermediate regime between long and short diffusive SNS behavior. Experimentally, the devices show critical currents $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$4 of $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$5, normal-state resistance $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$6 of $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$7, and $$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$8, which supports the intermediate-length diffusive interpretation and suggests influence from a residual Schottky barrier or an oxide layer at the nanowire/V interfaces (Spathis et al., 2010).

The broader comparative picture is equally important. Compared with SIS tunnel junctions, proximized weak links can reach much higher critical current densities and do not necessarily require external shunt resistors to avoid hysteresis. Compared with constriction junctions, their current-phase relation is often more ideal and the effective weak-link region can be better controlled. The review also notes limitations: proximity-junction devices may have lower characteristic voltages, can be constrained by fabrication complexity or narrower temperature windows, and are not yet as mature or as widely deployed as Nb/Al-AlO$$\delta_1-\delta_2+2\pi n=\frac{2\pi}{\Phi_0}(\Phi+LJ).$$9/Nb SIS nanoSQUIDs (Martínez-Pérez et al., 2016).

3. Interference, symmetry, and readout performance

The InAs nanowire–vanadium device shows standard SQUID interference with period $\beta_L\ll 1$0. The critical current follows

$\beta_L\ll 1$1

which is the standard relation for sinusoidal current-phase relations and negligible loop inductance. For one device, the fit yields $\beta_L\ll 1$2 and $\beta_L\ll 1$3, so $\beta_L\ll 1$4. Because both weak links are made from two nearby and homogeneous sections of the same nanowire, the junctions are nearly identical, and the measured modulation depth

$\beta_L\ll 1$5

is exceptionally large (Spathis et al., 2010).

The same device illustrates the inductance requirements for strong interference. The magnetic-field period is $\beta_L\ll 1$6 Oe, corresponding to an effective area $\beta_L\ll 1$7, consistent with the geometrical area $\beta_L\ll 1$8. The geometric self-inductance is $\beta_L\ll 1$9, and

$$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$0

so self-inductance is negligible. The device functions as a flux-to-voltage transducer above $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$1, with maximum transfer function up to $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$2 at optimum bias current $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$3. The corresponding power dissipation is a few tens of pW, and Josephson coupling is observed up to about $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$4, while transfer-function oscillations are observed up to $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$5 in the reported data range (Spathis et al., 2010).

A more recent proximity implementation changes the practical readout regime. In tapping-mode SQUID-on-tip microscopy, the nanoSQUID is an Nb–Cu–Nb SNS device whose weak link is a $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$6 nm long Cu nanobridge exposed by focused ion beam milling. The paper emphasizes a large transfer function of a few mV/$$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$7, at least an order of magnitude larger than typical nanoSQUIDs, and states that the large voltage output allows readout with a simple four-wire electronic measurement in current-bias mode using only a lock-in amplifier and without cryogenic amplification. The same work reports resolution of nanoscale currents as small as 100 nA, with successful imaging of 100 nA currents after integrating 35 scans (Rog et al., 29 Aug 2025).

4. Scanning implementations and multimodal microscopy

Scanning nanoSQUID microscopy has been shaped by both proximity-junction devices and closely related weak-link architectures. The early self-aligned SQUID-on-tip is not a proximity-junction nanoSQUID in the usual sense: its two Josephson weak links are self-formed geometric weak links in an evaporated Al apex ring rather than explicit SNS junctions. Even so, it established the extreme tip-based geometry that later proximity-junction devices adopted. The device was fabricated on the apex of a sharp quartz tip with effective diameters down to 100 nm, showed an effective area of $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$8, flux sensitivity $$I_{\rm c}=2I_0\left|\cos\left(\frac{\pi\Phi}{\Phi_0}\right)\right|.$$9, projected spin sensitivity $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$0, and operation in fields as high as $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$1 (Finkler et al., 2010).

Planar scanning implementations based on weak-link-like junctions provide a complementary route. The 3D nano-bridge-based SQUID susceptometer employs a 2-junction SQUID built from 3D Nb nano-bridges. In the nanoSQUID context it is directly relevant to proximity-junction and nano-bridge-based SQUIDs because it uses weak-link-like 3D nano-bridges rather than conventional tunnel junctions. Its gradiometric architecture uses two counter-wound pickup loops and one-turn field coils for susceptibility measurements; the smallest pickup loop was $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$2 in diameter, the flux noise was around $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$3 at 100 Hz and 4.2 K under zero applied field with FLL readout, and the device remained functional up to $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$4 perpendicular magnetic field. The same platform demonstrated scanning magnetometry, susceptometry, and current magnetometry (Pan et al., 2019).

The tapping-mode proximity SQUID-on-tip extends scanning capability beyond magnetic imaging alone. The probe is fabricated on a commercial Akiyama tuning-fork AFM probe, is self-sensing and self-actuating, and places the SQUID plane at a $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$5 angle relative to the sample surface, giving sensitivity to both out-of-plane and in-plane magnetic field components. By integrating the nanoSQUID with tapping-mode AFM, the probe minimizes nanoSQUID-sample distance and operates without lasers. Frequency multiplexing allows simultaneous imaging of current, magnetism, dissipation, and topography; in one demonstrated scan of a Cu/Co heterostructure, these channels were recorded concurrently, with the a.c. and d.c. signals differing by more than two orders of magnitude yet remaining nearly artifact-free because of frequency separation (Rog et al., 29 Aug 2025).

5. Noise, microstructure, and weak-link engineering

Low-noise operation is a central design criterion for proximity-junction nanoSQUIDs because both magnetic sensitivity and spin sensitivity depend directly on $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$6. The review emphasizes that reducing loop inductance is one route to improved white noise, while maximizing the coupling factor $V_\Phi=(\partial V/\partial\Phi)_{\rm max}$7 requires placing the magnetic object extremely close to a constriction or narrow region of the loop. This is one reason proximized or nanowire-based junctions are attractive: they can be integrated into compact loops and fine structures. The same review also cautions that proximity-junction devices, while competitive, are not always the absolute lowest-noise class and do not yet match the record spin sensitivity of the best SQUID-on-tip devices (Martínez-Pérez et al., 2016).

A related but distinct lesson comes from high-$V_\Phi=(\partial V/\partial\Phi)_{\rm max}$8 grain-boundary nanoSQUIDs. The YBa$V_\Phi=(\partial V/\partial\Phi)_{\rm max}$9Cu$S_\Phi=S_V/V_\Phi^2$0O$S_\Phi=S_V/V_\Phi^2$1/SrTiO$S_\Phi=S_V/V_\Phi^2$2 superlattice devices are not proximity-junction nanoSQUIDs; they use bicrystal grain-boundary Josephson junctions rather than true proximity weak links. Their relevance lies in weak-link microstructure and $S_\Phi=S_V/V_\Phi^2$3 noise suppression. In these devices, low-frequency noise is dominated by critical-current fluctuations $S_\Phi=S_V/V_\Phi^2$4 in the grain-boundary junctions, and bias reversal at $S_\Phi=S_V/V_\Phi^2$5 kHz strongly suppresses this contribution. Measured at 1 Hz under dc bias, the flux noise is $S_\Phi=S_V/V_\Phi^2$6 for SQ-14 and $S_\Phi=S_V/V_\Phi^2$7 for SQ-15, compared with $S_\Phi=S_V/V_\Phi^2$8 for reference single-layer devices; under bias reversal, $S_\Phi=S_V/V_\Phi^2$9 for SQ-15. The authors attribute the improvement to an improved microstructure at the grain boundaries, but explicitly note that they do not yet have direct microstructural evidence, so the explanation remains inferential (Lin et al., 2020).

For proximity-junction nanoSQUIDs, this suggests that weak-link microstructure is as important as nominal circuit topology. The YBCO study is not evidence about SNS transport itself, but it does show that improving weak-link microstructure can suppress critical-current noise dramatically without sacrificing white-noise performance, which is directly relevant to any nanoSQUID based on nanoscale Josephson weak links (Lin et al., 2020).

6. Mesoscopic limits, phase control, and theoretical extensions

Proximity-junction nanoSQUIDs are constrained not only by standard SQUID interference but also by mesoscopic pair breaking. In a superconducting thin ring closed by a Josephson junction, superconductivity can be destroyed by inverse proximity effects at the junction, so that the loop has a minimum radius

$\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$0

below which the only stable state is normal. The mechanism combines bulk depairing by superflow with local junction-induced depairing, described by the Ginzburg–Landau equation

$\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$1

with boundary conditions

$\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$2

and current relation

$\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$3

The threshold radius depends on phase difference, magnetic flux, Josephson coupling, and interfacial pair breaking (Barash, 2016).

The phase dependence is explicit: $\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$4 At $\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$5, the limiting radii are

$\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$6

For a 0 junction ($\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$7), the maximum $\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$8 occurs at $\sqrt{S_\mu}=\sqrt{S_\Phi}/\phi_\mu$9; for a $\phi_\mu\equiv \Phi/\mu$0 junction ($\phi_\mu\equiv \Phi/\mu$1), the maximum occurs at $\phi_\mu\equiv \Phi/\mu$2. Near the threshold size, the current-phase relation is strongly distorted, the critical current vanishes as $\phi_\mu\equiv \Phi/\mu$3, and the critical temperature is reduced below the bare value $\phi_\mu\equiv \Phi/\mu$4 (Barash, 2016).

A separate theoretical extension addresses a different limitation of nanoSQUID operation: the periodic occurrence of high sensitivity only near specific flux values. The multi-terminal, multi-junction dc SQUID introduces extra terminals and control currents so that the critical-current interference pattern can be shifted continuously with respect to applied flux. In the 4-terminal, 4-junction model, each junction obeys

$\phi_\mu\equiv \Phi/\mu$5

and the normalized fluxoid condition is

$\phi_\mu\equiv \Phi/\mu$6

The practical result is that operation at maximum sensitivity can be obtained at any value of the magnetic field by applying control current to the extra terminals; the paper states that the 3-terminal device shifts the interference pattern by about half a period, while the 4-terminal device extends the accessible shift to more than one full flux period when both control currents are used. The same framework is also presented as a direct method to measure the current-phase relations of individual weak links, and the authors emphasize that the analysis can be generalized from $\phi_\mu\equiv \Phi/\mu$7 to a generic $\phi_\mu\equiv \Phi/\mu$8, which is particularly relevant for proximity-coupled nanostructures (Meltzer et al., 2016).

7. Scope, adjacent junction families, and persistent misconceptions

A recurrent source of confusion is the boundary between proximity-junction nanoSQUIDs and other weak-link nanoSQUIDs. Not every nanoscale weak-link SQUID is a proximity-junction device. The self-aligned Al SQUID-on-tip employs geometric weak links rather than explicit SNS junctions (Finkler et al., 2010). The 3D Nb susceptometer uses 3D nano-bridges and is directly relevant to weak-link nanoSQUIDs, but it is not described as an SNS proximity device (Pan et al., 2019). The YBCO/STO nanoSQUIDs use bicrystal grain-boundary Josephson junctions rather than superconductor-normal-superconductor weak links (Lin et al., 2020). These distinctions matter because the microscopic sources of hysteresis, $\phi_\mu\equiv \Phi/\mu$9 noise, field tolerance, and current-phase nonideality are not identical across junction types.

A second misconception is that proximity-junction nanoSQUIDs form a single mature technology class. The available evidence points instead to a broad design space. Some platforms emphasize nearly symmetric interferometers and low dissipation, as in nanowire–vanadium SQUIDs (Spathis et al., 2010). Others emphasize large transfer function and simplified readout, as in Nb–Cu–Nb tapping-mode SQUID-on-tip probes (Rog et al., 29 Aug 2025). The review explicitly notes that some early concepts, such as the carbon-nanotube dc SQUID, promised very strong coupling to nearby magnetic moments, but that a full proof-of-principle magnetometer based on this concept is still missing (Martínez-Pérez et al., 2016).

The application domain is correspondingly broad. In the review, nanoSQUIDs are presented as sensors for magnetic flux, current, magnetization, magnetic field, and position, with strong relevance to scanning SQUID microscopy and magnetic-particle characterization (Martínez-Pérez et al., 2016). In explicit scanning implementations, closely related weak-link devices have demonstrated local flux detection, susceptibility mapping, and current reconstruction by FFT inversion using the pickup-loop point-spread function (Pan et al., 2019), while proximity-junction SQUID-on-tip probes have demonstrated simultaneous imaging of currents, magnetism, dissipation, and topography without external radiation or cryogenic preamplifiers (Rog et al., 29 Aug 2025). Taken together, these results define proximity-junction nanoSQUIDs not as a single circuit topology but as a technically diverse class of nanoscale interferometers in which proximitized weak links are used to combine compact geometry, phase-sensitive transport, and experimentally useful readout characteristics.