Spin Sorting: Methods and Applications

• Spin sorting is a family of operations that selectively separates spin-labeled degrees of freedom based on explicit rules across diverse systems.
• In mesoscopic transport, devices such as Y-shaped spin separators use Zeeman splitting and quantum point contacts to achieve near-perfect spin routing.
• In optical and astrophysical contexts, spin sorting manifests as chirality discrimination and inferential ranking, demonstrating its broad practical utility.

Spin sorting denotes a family of operations in which degrees of freedom labeled by spin, magnetic-moment projection, optical spin, or spin-ranked components are separated, routed, filtered, or re-parameterized according to an explicit rule. In mesoscopic spintronics, it can mean a three-terminal Y-shaped separator that routes opposite electron spins into different output leads (Wojcik et al., 2014). In relativistic magnetic reconnection, it denotes Stern–Gerlach branch separation across a current sheet by the sign of the field-aligned magnetic-moment projection $\mu_\parallel$ (Nykyri, 15 May 2026). In nanophotonics, it may refer either to chirality-sorting by transverse optical spin in an evanescent field (Hayat et al., 2014) or to spatial routing of opposite optical spins in a subwavelength nanoparticle waveguide (Hinamoto et al., 2021). In gravitational-wave population inference, it means ranking the two black holes in a binary by spin magnitude rather than by mass (Szemraj et al., 31 Jul 2025). The term is therefore domain-dependent: some usages describe literal spatial separation, some describe transmission or reflection filtering, and some describe a change of variables for inference.

1. Semantic scope and domain-specific meanings

Across the literature surveyed here, “spin sorting” is not a single standardized procedure but a cluster of related operations. The shared structure is selective discrimination of states carrying a spin-related label, followed by a measurable difference in output channel, force direction, inferred population component, or state classification. The label being sorted, however, varies substantially.

In semiconductor nanostructures, the sorted quantity is usually electron spin projection, and the outcome is transport into distinct terminals or transmission windows (Wojcik et al., 2014). In strong-field plasma physics, the sorted quantity is not simply raw spin projection but the sign of the field-aligned magnetic-moment projection, so the relevant branches are $\mu_\parallel>0$ and $\mu_\parallel<0$ rather than “electron up” and “electron down” in isolation (Nykyri, 15 May 2026). In optical settings, the situation bifurcates further: in one case optical spin angular momentum in the field sorts particles of opposite chirality (Hayat et al., 2014), while in another the optical spin itself is routed into opposite propagation directions or branches through spin-momentum locking (Hinamoto et al., 2021). In astrophysical inference, spin sorting is not a transport mechanism at all but a relabeling of binary components into the “more-spinning” and “less-spinning” objects, $\chi_A$ and $\chi_B$, with $\chi_A \ge \chi_B$ (Szemraj et al., 31 Jul 2025).

This semantic spread matters because closely neighboring terms are not equivalent. A spin filter may suppress one spin channel without generating two spatially separated outputs. A spin detector may infer spin from output current without autonomously sorting an initially mixed beam. A rank-ordering of inferred component spins is not a beamline separator, even though it is explicitly called spin sorting in the population-inference literature (Szemraj et al., 31 Jul 2025). A related foundational point is that for higher-spin statistical assemblies, classification by one magnetic-substate probability vector $p_m$ is complete only for oriented systems; more general assemblies require the full $SU(n)$ description and multiple empirically determinable axes (Ramachandran, 2019).

2. Mesoscopic transport: separators, filters, and serial discrimination

A concrete spin-sorting device is the three-terminal spin separator proposed in a Y-shaped two-dimensional electron gas with a quantum point contact in one output branch (Wojcik et al., 2014). Its effective-mass Hamiltonian is

$$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$

with Landau gauge $\mathbf A=(yB,0,0)$ and total scalar potential $\mu_\parallel>0$0. The device accepts an initially unpolarized current through lead 1, places the QPC near contact 2, and exploits a joint effect of Zeeman splitting and magnetic-field-induced edge-state transport. At the operating point, the QPC is tuned so that only the spin-up subband remains below $\mu_\parallel>0$1 in the constriction. Spin-up electrons pass through the QPC to lead 2, while spin-down electrons are reflected from the QPC and, because transport is carried by edge states, are diverted to lead 3 rather than back to the injector. The paper’s conductance-based indicators are

$\mu_\parallel>0$2

For the simulated $\mu_\parallel>0$3 device with $\mu_\parallel>0$4, $\mu_\parallel>0$5, $\mu_\parallel>0$6, $\mu_\parallel>0$7, $\mu_\parallel>0$8, and $\mu_\parallel>0$9, the clearest separation occurs at $\mu_\parallel<0$0 and around $\mu_\parallel<0$1, where the structure acts as an “(almost) perfect spin splitter.” The calculations use a tight-binding discretization with $\mu_\parallel<0$2 and the Kwant package. Disorder averaging over $\mu_\parallel<0$3 runs, with realistic mean free paths consistent with $\mu_\parallel<0$4, shows that impurity scattering does not destroy the effect in the non-ballistic regime.

Other transport-based realizations sort by transmission channel rather than by spatial branch in a Y-junction. In scattering from a two-dimensional crystal of aligned point magnets, the spin degree of freedom is the projection along the common magnetization axis, and sorting occurs through different transmission and reflection probabilities for the two spin channels (Ghazaryan et al., 2020). The spin-dependent scattering lengths satisfy

$\mu_\parallel<0$5

and the periodic layer maps at low energy to an effective one-dimensional sheet with

$\mu_\parallel<0$6

The corresponding single-layer transmission and reflection coefficients are

$\mu_\parallel<0$7

with transmitted-beam polarization

$\mu_\parallel<0$8

Near the mirror condition $\mu_\parallel<0$9, one spin channel becomes strongly reflected. A two-layer “quantum cavity” can then restore high transmission at resonant energies while retaining strong spin selectivity.

A different transmission-based construction is the one-dimensional magnetic chain for particles of arbitrary spin $\chi_A$0 (Pal et al., 2016). There the moving spin-$\chi_A$1 particle is recast as a spinless particle on an effective $\chi_A$2-strand ladder, one strand per spin projection $\chi_A$3. Substrate engineering controls the strand on-site energies and couplings, thereby opening spin-resolved spectral gaps. The central analytic relation is

$\chi_A$4

so gap opening requires

$\chi_A$5

At a chosen incident energy or Fermi energy, only selected spin projections have propagating states, and the device therefore functions as a spectral spin sorter.

A useful contrast is the “Single-Spin CCD,” which demonstrates initialization, transport, preservation, and sequential single-shot readout of individual electron spins in a triple quantum-dot array but does not autonomously route $\chi_A$6 and $\chi_A$7 electrons into distinct destinations (Baart et al., 2015). It achieves single-shot read-out of three spins with fidelities of $\chi_A$8 on average and shows negligible influence on spin projection after hundreds of shuttling events over a cumulative distance of $\chi_A$9. The architecture is thus a spin-preserving shift register with serial discrimination, not a literal spin sorter.

3. Stern–Gerlach branch separation in relativistic magnetic reconnection

In relativistic magnetic reconnection, spin sorting is defined as Stern–Gerlach-driven spatial separation inside a reconnecting current sheet according to the sign of the field-aligned magnetic-moment projection

$\chi_B$0

with

$\chi_B$1

The setting is a Harris sheet in which $\chi_B$2 varies across the sheet-normal direction $\chi_B$3. Under the assumption that Bargmann–Michel–Telegdi precession around the inflow field is fast during the transit, transverse spin components average away, and the sheet-normal Stern–Gerlach force reduces to a branch law of the form

$\chi_B$4

For $\chi_B$5, the sign rule is

$\chi_B$6

The effect is explicitly described as moment-tagged rather than species-tagged: in a pair plasma, particles with the same spin projection may have opposite $\chi_B$7, while particles with the same $\chi_B$8 may have opposite spins (Nykyri, 15 May 2026).

The importance of the sorting is quantified by comparing the SG-induced cross-sheet displacement accumulated during one diffusion-region transit to the relativistic Larmor radius. For a fully projected branch,

$\chi_B$9

and for an effective ensemble,

$\chi_A \ge \chi_B$0

The regime interpretation is explicit: $\chi_A \ge \chi_B$1 means negligible SG transport, $\chi_A \ge \chi_B$2 means transitional behavior, and $\chi_A \ge \chi_B$3 means strong sorting. Electron–positron 2.5D spin-kinetic PIC simulations with isotropic initial spins and zero net ensemble polarization show the predicted branch separation: the north side becomes enriched in $\chi_A \ge \chi_B$4 particles and the south side in $\chi_A \ge \chi_B$5 particles. In the SG-active run, the time-averaged branch centroid separation is $\chi_A \ge \chi_B$6, with peak $\chi_A \ge \chi_B$7, while the reconnected flux histories overlap to within $\chi_A \ge \chi_B$8. The astrophysical hierarchy obtained from $\chi_A \ge \chi_B$9 identifies the magnetotail, solar corona, AGN/blazar jets, and pulsar-wind nebulae as quantitative negative controls, magnetar current sheets as transitional, and magnetar surfaces as extremely strong-field targets.

4. Optical spin, chirality sorting, and photonic routing

One optical meaning of spin sorting is actually chirality sorting mediated by optical spin angular momentum (Hayat et al., 2014). In an evanescent wave produced by total internal reflection, the field propagates along $p_m$0, decays along $p_m$1, and carries a transverse spin angular momentum along $p_m$2. A small chiral particle with induced dipoles

$p_m$3

can convert that transverse spin structure into a real lateral force. The remarkable feature is that this force acts along $p_m$4, a direction in which there is neither wave propagation nor intensity gradient. Its sign depends on the chiral polarizability $p_m$5, so particles of opposite handedness are pushed in opposite lateral directions. The sorted objects are therefore chiral particles, not particles being separated by their own internal spin state. The paper emphasizes that the net chirality-dependent lateral effect comes from the dipole–dipole interaction term and that the final spin-related term in the force expression produces linear momentum transfer on chiral particles in opposite directions for opposite signs of $p_m$6.

A second optical meaning is direct photonic routing of opposite circular polarizations in a subwavelength dielectric waveguide (Hinamoto et al., 2021). The proposed structure is an asymmetric zigzag chain of high-index silicon nanospheres with basic parameters

$p_m$7

periodicity $p_m$8, and $p_m$9 large spheres. A circular electric dipole source

$SU(n)$0

is placed at the center of the middle large sphere. The key mechanism is engineered transverse spin density: in the asymmetric chain, one spin parity is concentrated inside the large sphere and the opposite parity in the gap region near the side sphere. Because spin-momentum locking ties local spin to propagation direction, a source inside the sphere couples preferentially in one direction, while a source in the gap couples with opposite directionality. In the first transmission band, $SU(n)$1–$SU(n)$2, about $SU(n)$3 of the coupled light reaches the preferred side and less than $SU(n)$4 reaches the opposite side; near $SU(n)$5, after averaging over finite-size effects, $SU(n)$6 of the transmitted light reaches the right. The asymmetry is strongest around $SU(n)$7, or $SU(n)$8. A star-shaped junction built from mirror-opposite asymmetric branches then performs explicit spin sorting: at $SU(n)$9, $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$0 light is routed into the upper branch, while reversing the source rotation routes light into the lower branch. Directional transport is also preserved in a bent chain with bending radius $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$1, although with stronger decay than in the straight structure.

5. Statistical, inferential, and algorithmic reformulations

In binary-black-hole population inference, spin sorting is introduced as a reframing of component-spin analysis (Szemraj et al., 31 Jul 2025). Standard analyses mass-sort the components into $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$2 and $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$3, the spins of the more massive and less massive black holes. Spin sorting instead defines

$$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$4

where $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$5 is the larger spin magnitude and $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$6 the smaller. Under an IID component-spin distribution $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$7 with cumulative distribution $$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$8, the spin-sorted distributions are post-processed by order statistics:

$$\hat{H} =\left [ \frac{ (-i\hbar \nabla +e\mathbf{A}) ^2}{2 m_e} + U(\mathbf{r}) \right ] \mathbf{1} + \frac{1}{2}g\mu_B B \sigma _z ,$$9

$\mathbf A=(yB,0,0)$0

This change of variables is especially informative for near-equal-mass binaries and one-spinning-component populations. Simulated populations show that even with the Default LVK spin-magnitude model, which cannot place a literal delta-function at zero spin, the spin-sorted outputs distinguish fully nonspinning, singly-spinning, and mixed populations. The paper concludes that current observations are inconsistent with a fully nonspinning binary-black-hole population, but could be explained by a population with only one spinning black hole per binary or by a population with up to $\mathbf A=(yB,0,0)$1 nonspinning sources.

A more foundational statistical usage arises in the theory of assemblies of particles with spin (Ramachandran, 2019). For spin-$\mathbf A=(yB,0,0)$2, sorting by a single projection axis is complete: after choosing the axis along the polarization vector, the density matrix can be diagonalized and represented by probabilities $\mathbf A=(yB,0,0)$3 and $\mathbf A=(yB,0,0)$4. For higher spins $\mathbf A=(yB,0,0)$5, that simple picture describes only oriented systems. A general spin-$\mathbf A=(yB,0,0)$6 density matrix requires the generators of $\mathbf A=(yB,0,0)$7, with $\mathbf A=(yB,0,0)$8, and the diagonal probability domain is identified with the interior of a regular polyhedron in $\mathbf A=(yB,0,0)$9. The center corresponds to an unpolarized assembly, the vertices to pure states, and interior points to mixed states. The paper further states that a higher-spin system contains

$\mu_\parallel>0$00

independent empirically determinable axes. Only when all these axes are collinear does the oriented-system description in terms of probabilities $\mu_\parallel>0$01 apply. In this sense, higher-spin sorting may require full multipole classification rather than mere projection counting.

A terminological outlier is the algorithmic literature on “Spin-the-bottle Sort” and Annealing Sort (Goodrich, 2010). Here the word “spin” has no quantum-mechanical content. Spin-the-bottle Sort is a randomized round-robin comparison algorithm with expected running time $\mu_\parallel>0$02 on a worst-case input and $\mu_\parallel>0$03 time with very high probability on any input, while Annealing Sort uses a temperature-restricted comparison schedule and admits an $\mu_\parallel>0$04 implementation with very high probability. The usage is lexically similar but conceptually unrelated to physical spin sorting.

6. Comparative distinctions and adjacent concepts

Several recurring distinctions prevent conceptual slippage. First, not every spin sorter performs literal spatial beam splitting. The Y-shaped separator routes spin-up and spin-down electrons into leads 2 and 3 (Wojcik et al., 2014), but the aligned-point-magnet crystal separates spins mainly into transmitted and reflected channels (Ghazaryan et al., 2020), and the arbitrary-spin magnetic chain sorts through energy-selective transmission windows (Pal et al., 2016). Second, not every “spin sorting” operation sorts raw spin projection. In reconnection, the relevant label is the sign of $\mu_\parallel>0$05, the field-aligned magnetic-moment projection (Nykyri, 15 May 2026). In evanescent-field nanophotonics, optical spin sorts material chirality rather than the particle’s own spin state (Hayat et al., 2014). In gravitational-wave population inference, the operation is purely inferential: the two black holes are reordered by spin magnitude (Szemraj et al., 31 Jul 2025).

Third, several adjacent systems are explicitly not literal spin sorters. The Single-Spin CCD performs high-fidelity serial spin-state discrimination after transport to a common readout point, but it does not route opposite spins to distinct branches (Baart et al., 2015). Active spinner materials display effective spin-like ordering, chirality degeneracies, and chirality-controlled edge currents, yet the study concerns spatiotemporal order in a monodisperse same-chirality ensemble rather than separation of mixed spin or chirality populations (Zuiden et al., 2016).

A plausible commonality is that successful sorting appears when the spin-related label is embedded in a transport or classification mechanism that remains dynamically legible. In the Y-shaped 2DEG device, that mechanism is the combination of Zeeman splitting and edge-state transport (Wojcik et al., 2014). In relativistic reconnection, it is the accumulation of SG displacement relative to the Larmor radius (Nykyri, 15 May 2026). In the nanoparticle chain, it is the engineered overlap between local spin density and guided-mode propagation direction (Hinamoto et al., 2021). In binary-black-hole inference, it is the order-statistic map from IID component spins to $\mu_\parallel>0$06 and $\mu_\parallel>0$07 (Szemraj et al., 31 Jul 2025). This suggests that “spin sorting” functions less as a single technical doctrine than as a domain-specific label for selective separation, routing, filtering, or re-expression of spin-bearing degrees of freedom under well-defined dynamical or statistical rules.