⬇ Download Markdown  ·  ⬇ Download LaTeX  ·  ⬇ original  ·  view original

Intrinsic Spin Filter Effect in a d-wave altermagnet KV 2 Se 2 O with Open Fermi Surface Bin Liu, 1 Pei-Hao Fu, 1, 2, # Yu-Xuan Sun, 1 Xiao-Lin Zhang, 3 Si-Cong Zhu, 4 Xiang-Long Yu, 1, * Hua

Wu, 5, † Yuan-Zhi Shao 1, 6, ‡ 1 2 3 School of Science, Sun Yat-sen University, Shenzhen 518107, China Department of Physics and Astronomy, University of Florence, I-50019 Sesto Fiorentino, Italy Key Laboratory of Artificial Micro- and Nano-structures of Ministry of Education, School of Physics

and Technology, Wuhan University, Wuhan 430072, China 4 The State Key Laboratory for Refractories and Metallurgy, Wuhan University of Science and

Technology, Wuhan 430081, China 5 Laboratory for Computational Physical Sciences (MOE), State Key Laboratory of Surface Physics,

and Department of Physics, Fudan University, Shanghai 200433, China 6 Guangdong Provincial Key Laboratory of Magnetoelectric Physics and Devices, Sun Yat-sen

University, Guangzhou 510275, China # * Corresponding author: [email protected] Corresponding author: [email protected]

† ‡ Corresponding author: [email protected] Corresponding author: [email protected] ABSTRACT Altermagnets offer a unique pathway to functional spintronics by combining vanishing

magnetization with large spin splitting. Here, we demonstrate that the canonical d-wave

altermagnet KV 2 Se 2 O can deliver giant tunneling magnetoresistance through orientation-

dependent spin filtering. By analyzing the crystallographic spin segregation, we show that

transport along specific crystallographic axes is nearly fully spin-polarized within the

symmetry-protected ballistic channels. We implement this mechanism in a lattice-matched

KV 2 Se 2 O/Bi 2 O 2 Se/KV 2 Se 2 O magnetic tunnel junction, which achieves a robust half-metallic

transport regime. The symmetry-protected spectral gap in the parallel/anti-parallel

configuration ensures a high tunneling magnetoresistance ratio, resulting in substantial

tunneling magnetoresistance, robust thermally driven spin filtering, and spin Seebeck effect at

room temperature. These findings provide a path of altermagnetic heterostructures as a high-

performance platform for scalable, field-free, and thermally stable spin logic.

1 / 24

1. INTRODUCTION

Spin-polarized charge currents constitute the fundamental information carrier in spintronic

technologies, underpinning key functionalities such as nonvolatile memory, logic operations,

and spin-torque-driven switching [1] . Achieving robust spin polarization without sacrificing

device scalability and stability remains a central challenge. Ferromagnets (FM) naturally

generate longitudinal spin-polarized currents but suffer from stray magnetic fields that limit

device density and introduce cross-talk [2-5] . Antiferromagnets (AFM) eliminate stray fields and

exhibit ultrafast spin dynamics, yet their compensated spin symmetry typically enforces spin-

degenerate electronic bands, resulting in negligible longitudinal spin polarization and severely

restricting their applicability in spin-charge conversion and transport [6-9] . Altermagnetism (AM)

has recently emerged as a distinct magnetic order that bridges this long-standing dichotomy by

unifying key attributes of ferromagnets and antiferromagnets [10-12] . Altermagnets host large,

exchange-driven spin splitting comparable to ferromagnets and unconventional spin Hall

responses enabled by magnetic crystal symmetries, while maintaining a strictly vanishing net

magnetization enforced by combined crystal rotation and time-reversal symmetries [13-17] .

Owing to the absence of net magnetization, AM avoids the stray fields that challenge FM-based

devices. At the same time, the presence of spin splitting enables spin-dependent transport, in

contrast to AFM-based systems in which the spin information is largely hidden [18-23] .

However, the symmetry protection that stabilizes altermagnetism also imposes a fundamental

constraint: spin textures related by combined crystalline and time-reversal symmetries, exhibit

an orientation-dependent compensation that leads to an exact cancellation of spin polarization

when integrated over the Brillouin-zone, thereby prohibiting the emergence of sizable

macroscopic spin-polarized charge currents in equilibrium. In this sense, the defining advantage

of altermagnetism simultaneously constitutes its central obstacle for spintronic functionality [24-

27] . From a symmetry perspective, both strain- and electric-field-based approaches rely on

explicit symmetry breaking to enable spin polarization in altermagnets. Uniaxial strain induces

fully spin-polarized transport by breaking mirror and rotational symmetries, analogous to

electric-field-based schemes that rely on strong gate fields to lift symmetry constraints.

Crucially, both mechanisms yield momentum-dependent and valley-confined spin responses [28,

2 / 24 29] . Such strong external perturbations raise concerns regarding robustness, scalability, and

experimental feasibility. In terms of physical mechanisms, these strategies rely on explicit

symmetry breaking to lift the symmetry-enforced momentum-space cancellation of spin

polarization. The resulting spin-polarized transport can be interpreted as arising from an

effective Zeeman-like field that induces a finite net magnetization [30-34] . While this route enables

longitudinal spin polarization, it effectively drives the system back to a

ferromagnetic/ferrimagnetic-like regime, undermining the symmetry-protected nature of

altermagnetism and potentially reintroducing familiar limitations such as stray-field coupling

and reduced device robustness. These considerations underscore a critical open challenge in

altermagnetic spintronics: achieving intrinsic longitudinal spin-polarized charge transport

without resorting to external symmetry-breaking fields or net magnetization.

In this work, we propose an intrinsic spin-filter effect in AM driven by open Fermi-surface

geometry, enabling highly spin-polarized charge transport without net magnetization. Focusing

on KV 2 Se 2 O as a representative material platform, we first establish the microscopic

mechanism underlying the intrinsic spin-filter effect using a simple low-energy effective model.

We show that the closed-to-open Fermi surface transition selectively suppresses one spin

channel in a direction-selective manner, resulting in ideal spin polarization in longitudinal

transport. We then provide a proof of concept by investigating KV 2 Se 2 O—a material exhibiting

strong crystalline and magnetic anisotropy that enables pronounced directional modulation of

spin-dependent charge transport. Owing to its quasi-two-dimensional lattice, strong in-plane

anisotropy, and intrinsic magnetic order, KV 2 Se 2 O supports symmetry-constrained, direction-

dependent spin textures, enabling field-free spin-polarized transport within a realistic materials

platform [14, 35] . By combining first-principles electronic-structure calculations with device-level

transport modeling, we elucidate the symmetry origin of the anisotropic spin splitting, track the

evolution of momentum-resolved spin textures under different transport geometries, and

identify specific crystallographic directions along which nearly approaching fully longitudinal

spin polarization emerges despite the absence of net magnetization. Guided by the Density

Functional Theory (DFT) results, we further propose lattice-matched tunneling heterostructures

based on KV 2 Se 2 O. The sharp contrast between the conducting parallel-configuration (PC) state

3 / 24 and the insulating antiparallel-configuration (APC) state yields a giant tunneling

magnetoresistance (TMR) ratio of approximately 2.24 × 10 3 % in the absence of any external

magnetic field. Moreover, we predict a colossal thermal tunneling magnetoresistance exceeding

750% at low temperatures, which remains substantial even at room temperature. Our findings

demonstrate that altermagnets with open Fermi surfaces enable a distinct class of spintronic

functionalities, including nearly fully spin filters and giant tunneling magnetoresistance,

without requiring magnetic fields, spin--orbit coupling, or ferromagnetic components.

2. RESULTS

2.1 Low-Energy Effective Model for Fermi Surface Controlled Spin Transport In this section, we use a low-energy effective model for a two-dimensional d-wave altermagnet

to elucidate the mechanism of the geometry of the Fermi surface enabling theoretically perfect

spin-polarized charge transport. The purpose of this effective model is to capture, in a minimal

but analytically transparent manner, how the closed-to-open Fermi-surface transition in

altermagnets controls (i) the density of states (DOS) and (ii) the direction-dependent spin-

resolved carrier velocity, which together dictate the spin polarization of the longitudinal

conductance.

The effective Hamiltonian for two-dimensional 𝑑 𝑥 2 −𝑦 2 -wave altermagnet is [10, 14]

𝐻 ^ A M 𝜎 ( 𝑘𝑥 , 𝑘𝑦 ) = 𝑡𝑥𝜎 𝑎2 𝑘𝑥2 + 𝑡𝑦𝜎 𝑎2 𝑘𝑦2 + 𝐸𝐹 , (1)

where 𝑎 is the lattice constant and 𝐸 𝐹 denotes the Fermi level, 𝜎 = +/- for spin-up/down

states. The d-wave altermagnetic anisotropic spin splitting is encoded in the spin- and direction-

dependent hopping parameters 𝑡𝑥 (𝑦) 𝜎 = 𝑡0 ± 𝜎𝑡𝐴 𝑀 , with 𝑡0 representing the isotropic

nearest-neighbor hopping amplitude and 𝑡 𝐴𝑀 the anisotropic spin-dependent hopping that

captures the d-wave altermagnetic spin splitting. This anisotropic spin splitting is directly

manifested in the spin-resolved equal-energy contours and dispersions of Eq. (1), shown in Fig.

  1. For 𝑡 𝐴𝑀 /𝑡 0 < 1 [Fig. 1(a)], the dispersion consists of two upward-opening parabolic bands

with spin- and direction-dependent curvature, leading to two spin-dependent closed elliptical

Fermi surfaces whose principal axes are mutually orthogonal [see the insert of Fig. 1(a)]. For

4 / 24 𝑡 𝐴𝑀 /𝑡 0 > 1 [Fig. 1(b)], the dispersion consists of two parabolic bands with opposite opening

directions (along one momentum axis) and spin-dependent curvature. This yields open Fermi

surfaces: within a finite momentum window, only a single spin species is present at the Fermi

level as exhibited in the insert of Fig. 1(b). Moreover, the band structure hosts a saddle point at

the spin-degenerate momentum (𝑘 𝑥 , 𝑘 𝑦 ) = (0, 0) [36] , where the curvature is positive along one

direction and negative along the other. Notably, at the critical point 𝑡 𝐴𝑀 /𝑡 0 = 1 [Fig. S1], a

flat band emerges: along one direction, only one spin sub-band remains dispersive, while the

5 / 24 other is dispersionless. FIG. 1. Mechanism of open-Fermi-surface-driven perfectly spin-polarized transport. (a), and (b) Energy dispersion of the low-energy effective d-wave altermagnetic model [Eq. (1)] along 𝒌𝒙 with 𝒌𝒚 = 𝟎, together with the corresponding density of states [Eq. (2)] as a function of energy. Blue and red represent different spins, and the direction of the arrows indicates the direction of the speed. Insets illustrate the Fermi-surface geometry determined by the altermagnetic parameter 𝒕𝐀 𝐌/ 𝒕𝟎 : closed Fermi surfaces with positive curvature for 𝒕𝐀 𝐌/ 𝒕𝟎 < 𝟏, open Fermi surfaces with negative curvature for 𝒕𝐀 𝐌/ 𝒕𝟎 > 𝟏. (c), and (d) Schematic illustration of the spin-resolved transport corresponding to the different Fermi-surface geometries. For Fermi surfaces with nonpositive curvature (𝒕𝐀 𝐌/ 𝒕𝟎 ≥ 1), direction-dependent and nearly perfectly spin-polarized charge transport emerges.

In the altermagnet with the open Fermi surface, the occurrence of the flat band and saddle

point gives rise to a van Hove singularity in the DOS near the corresponding energy [36] . A

straightforward calculation yields the spin-resolved DOS [37]

𝐷𝜎 (𝐸) = 1 8𝜋 1 √|𝑡A M 2 − 𝑡02 | { 2Θ(𝐸 − 𝐸𝐹 ), 𝑡A M 𝑡0 < 1, l n ( 𝐸c ut |𝐸−𝐸𝐹 |) , 𝑡A M 𝑡0 ≥ 1, (2)

where 𝐸 cut = 8𝑡 0 is a bandwidth cutoff, Θ(𝑥) is the Heaviside step function. Importantly,

𝐷𝜎 (𝐸) = 𝐷− 𝜎 (𝐸), so the spin density 𝐷↑ (𝐸) − 𝐷↓ (𝐸) = 0, consistent with the absence of net magnetization. The total DOS is then 𝐷(𝐸) = 𝐷↑ (𝐸) + 𝐷↓ (𝐸) = 2𝐷𝜎 (𝐸) . Figs. 1(a-b)

summarize the energy dependence of 𝐷(𝐸) for representative 𝑡 AM /𝑡 0 . For 𝑡 AM /𝑡 0 < 1 ,

𝐷(𝐸) is essentially the constant DOS of a two-dimensional parabolic band [37] , with

modification induced by the altermagnetism via 𝑡 AM , but no singular behavior. By contrast,

when 𝑡 AM /𝑡 0 ≥ 1, the flat band and saddle point generate a logarithmic van Hove singularity

near 𝐸 ≃ 𝐸 𝐹 . Notably, in the vicinity of the critical point 𝑡 AM /𝑡 0 = 1, 𝐷(𝐸) → ∞ within the

present effective description, implying a strong enhancement of the transport response. This

sharp contrast between the DOS of closed and open Fermi surfaces is a direct signature of the

distinct Fermi-surface geometries.

The impact of Fermi-surface geometry is also encoded in the carrier velocity. For Eq. (1), the

velocity expectation value along 𝛼-direction (𝛼 = 𝑥, 𝑦) reads 𝑣𝛼 𝜎 = 1 ℏ ⟨𝜕𝑘 𝛼 𝐻 ^ A M 𝜎 ⟩ = 2𝑡𝛼𝜎 𝑎2 𝑘𝛼 ℏ , (3)

where 𝑣 𝛼𝜎 denotes the quantum-mechanical expectation value in the spin-𝜎 eigenstate. The dwave altermagnetic symmetry implies 𝑣 𝑥𝜎 = 𝑣 𝑦−𝜎 , inherited from Eq. (1). As illustrated in Fig.

1(c), for 𝑡 AM < 𝑡 0 (closed Fermi surfaces), both spin channels have finite velocities along both

directions, so spin-up and spin-down electrons contribute comparably to transport. In contrast,

for 𝑡 AM ≳ 𝑡 0 [Figs. 1(d)], one spin channel becomes nearly blocked along a given direction while the other channel remains highly conductive. In particular, 𝑣 𝑥↑ = 𝑣 𝑦↓ ∝ 𝑡 0 + 𝑡 AM ,whereas

𝑣𝑥 ↓ = 𝑣𝑦 ↑ ∝ 𝑡0 − 𝑡A M ≃ 0 , so transport along 𝑥(𝑦) is dominated by spin- (↑↓) carriers,

providing the kinematic basis for perfectly spin-polarized currents.

The effect of the Fermi-surface geometry on the DOS [Eq. (2)] and the velocity [Eq. (3)] is

6 / 24 then manifested in the longitudinal conductance. To demonstrate how the Fermi-surface

geometry controls transport, we evaluate the longitudinal conductance within semiclassical

Boltzmann theory using the relaxation-time approximation. In the linear-response regime,

𝐼 𝛼 = 𝐺 𝛼𝛽 𝑉 𝛽 , and we focus on the longitudinal component 𝛼 = 𝛽 . At zero temperature, the

conductance can be written as

𝐺𝛼 𝛼 (𝐸) = 𝜏𝑒 2 ∑ 𝑣𝛼 2 𝐤 𝐷(𝐸), (4)

where 𝜏 is the momentum relaxation time. For the 𝑥direction, substituting Eq. (3) into Eq. (4)

yields the scaling form

𝐺𝑥 𝑥𝜎 (𝐸) ∝ (𝑣𝑥 𝜎 )2 𝐷𝜎 (𝐸) ∼ 𝑡02 (1+𝜎 𝑡𝐴 𝑀 𝑡0 ) 2 𝐷𝜎 (𝐸), (5)

which explicitly contains 𝑡 AM /𝑡 0 through the spin- and direction-dependent velocity. For ∣

𝑡A M/ 𝑡0 ∣< 1 , both 𝐺𝑥 𝑥↑ and 𝐺𝑥 𝑥↓

AM /𝑡 0 ∣< 1 , both 𝐺 𝑥𝑥 and 𝐺 𝑥𝑥 are comparable, and the net conductance contains

↑ contributions from both spins. For ∣ 𝑡

AM /𝑡 0 ∣≳ 1 , however, 𝐺 𝑥𝑥 is strongly enhanced

↓ (particularly near the van Hove singularity at 𝐸 ≃ 𝐸

𝐹 ), whereas 𝐺 𝑥𝑥 becomes negligible,

yielding a nearly perfect spin polarization

𝑃𝑥 𝑥 = 𝐺𝑥 𝑥↑ − 𝐺𝑥 𝑥↓ 𝐺𝑥 𝑥↑ + 𝐺𝑥 𝑥↓ ≃ 1, (6)

Repeating the same analysis for transport along 𝑦gives the opposite polarization, 𝑃 𝑦𝑦 ≃ −1 =

−𝑃𝑥 𝑥, consistent with the 𝑑-wave altermagnetic symmetry and the relation 𝐺𝑦 𝑦𝜎 (𝐸) = 𝐺𝑥 𝑥−𝜎 (𝐸),

as illustrated in Fig. 1.

We emphasize that Eq. (1) is a minimal effective description. In realistic materials such as

RuO 2[38-41] , La 2 O 3 Mn 2 Se 2 , and Ba 2 CaOsO 6[42] , 𝑑 -wave altermagnetism often involves

additional orbital and sublattice degrees of freedom (e.g., two-sublattice tetragonal [38-40] or

Lieb-lattice models [42] ). Here we isolate the universal role of Fermi-surface geometry, for which

Eq. (1) already captures the key distinction between closed and open Fermi surfaces. We also

note that the divergence in Eq. (2) is regularized in real systems by disorder and finite

quasiparticle lifetimes [43] . A standard phenomenological broadening amounts to the

replacement ∣ 𝐸 − 𝐸𝐹 ∣→ √(𝐸 − 𝐸𝐹 )2 + Γ2 , with 𝛤 ∼ ℏ/(2𝜏′ ) set by the elastic lifetime

𝜏 ′ due to impurity scattering [44-46] . Finally, the Boltzmann treatment in Eq. (4) assumes a 7 / 24 constant (isotropic) relaxation time 𝜏, weak disorder, zero temperature, and a small bias 𝑉 𝛽 .

Despite these approximations, the analysis demonstrates the central point: open Fermi surfaces

naturally suppress one spin channel in a direction-dependent manner, enabling giant and nearly

perfectly spin-polarized longitudinal conductance when the chemical potential is tuned close to

the flat-band or saddle-point energies. In the next section, we substantiate this mechanism using

material-based calculations.

2.2 Nearly Perfectly Directional Spin Polarization in KV 2 Se 2 O To validate the mechanism proposed in Eq. (1), in this section and the following we perform

DFT calculations for the altermagnetic candidate material KV 2 Se 2 O. The calculations were

performed using the nonequilibrium Green’s function formalism combined with density

functional theory (DFT+NEGF), as implemented in QuantumATK (version T-2022.03,

Synopsys QuantumATK) [47-49] , A Hubbard 𝑈 value of 1 eV was applied to the V 3𝑑 orbitals of

one pair of V atoms with opposite spin orientations [50-52] , thereby explicitly treating spin

polarization. The electrically-driven spin transport performance can be quantitatively analyzed from the Landauer-Buttiker equation [53] :

I(𝑉𝑏 ) = 2𝑒 2 ℎ ∫ 𝑇(𝐸, 𝑉𝑏 )[𝑓𝐿 (𝐸 − 𝜇𝐿 ) − 𝑓𝑅 (𝐸 − 𝜇𝑅 )]𝑑𝐸 +∞ −∞ (7)

where 𝑓 𝐿(𝑅) (𝐸, 𝜇)is the equilibrium Fermi distribution of the left (right) electrode, and 𝜇 is

the electrochemical potential of the electron. Fig. 2(a) shows the crystal structure of KV 2 Se 2 O,

where the magnetic V atoms form a square lattice separated by Se and O ligands. The

corresponding spin-resolved Fermi surfaces are shown in Fig. 2(b). Two bands (labeled as

Bands 35 and 36) cross the Fermi energy and exhibit pronounced anisotropy: one spin sub band

is dispersive along one direction (e.g. spin-up is dispersive along Γ-X direction, spin-down is

dispersive along Γ-Y direction), it becomes flat band without dispersion along the orthogonal

direction. As a result, the Fermi surfaces are open, extending across the Brillouin zone rather

than forming closed contours. As what is expected in other altermagnetic candidate materials

such as MnTe, Nb 2 FeB 2 , and V 2 Se 2 O [54-56] . This geometry closely resembles the open-Fermi8 / 24

surface regime of the effective model Eq. (1) for 𝑡A M ≳ 𝑡0 . FIG. 2. Realization of crystallographic spin segregation in the d-wave altermagnet KV2 S e2 O . (a) Crystal and magnetic structure of KV2 S e2 O . Vanadium (V) atoms form two antiferromagnetically aligned magnetic sublattices (VM ↑ and VM ↓ ), which are related by a crystallographic rotation symmetry. (b) Spin-resolved Fermi surface in the kz = 0 plane. The Fermi surface exhibits complete spin separation: electronic states along the Γ–X direction is exclusively spin-up (blue), whereas those along the orthogonal Γ–Y direction is entirely spin-down (red). (c) V-atom projected spin-resolved band structure. The marker size is proportional to the spectral weight from the VM ↑ (open circles) and VM ↓ (solid circles) sublattices, with blue and red denoting spin-up and spin-down states, respectively. Transport channels along Γ–X are dominated by VM ↑ spin-up states, while those along Γ–Y are dominated by VM ↓ spin-down states, demonstrating sublattice-selective transport. (d) Projected density of states (PDOS) of the V 3d orbitals. The spin-symmetric total spectral weight reflects complete magnetization compensation, while the pronounced local exchange splitting ensures robust metallic transport channels.

The origin of this open Fermi surface can be traced back to the orbital character and bonding

geometry of KV 2 Se 2 O. As a results, in Fig. 2(c), the band structure near the Fermi level consists

of two parabolic bands with opposite opening directions (along one momentum axis) and spin-

dependent curvature, as shown in Fig. 1(b) via the effective model [Eq. (1)]. Using a

dimensionless parametrization with 𝑡 0 = 1, the band evolution shown in Fig. 1 corresponds to

𝑡 am = 0.5, 1.0, 1.5 , representing the closed-Fermi-surface, critical, and open-Fermi-surface

9 / 24 regimes, respectively, and the DFT bands of KV 2 Se 2 O are consistent with the 𝑡 am ≳ 𝑡 0 . In

particular, one spin channel remains dispersive along one in-plane direction while becoming

nearly flat along the orthogonal direction. In particular, the presence of Se atoms strongly

suppresses the effective V-V hopping along one in-plane direction, whereas hopping along the

perpendicular direction remains sizable. This anisotropic hybridization leads to an almost

dispersionless band along one momentum axis, giving rise to a quasi-flat band and a saddle-

point-like structure near the Fermi energy. This band anisotropy is further reflected in the spin-

resolved DOS, shown in Fig. 2(d). While the total DOS remains spin compensated, consistent

with the absence of net magnetization, a pronounced enhancement of the DOS appears near E F ,

originating from the flat-band and saddle-point features. This behavior is qualitatively

consistent with the logarithmic van Hove singularity predicted by the effective model in Eq. (2)

for the open-Fermi-surface regime. Taken together, Figs. 2(a-d) demonstrate that KV 2 Se 2 O

realizes, at the material level, the key ingredients of the low-energy d-wave altermagnetic model:

direction-dependent hopping, open spin-resolved Fermi surfaces, and a strong DOS

enhancement near the Fermi level. These features provide the microscopic foundation for the

giant and nearly perfectly spin-polarized transport discussed in the following sections.

2.3 Tunneling Magnetoresistance in Lattice-Matched Heterostructures In the following, we investigate two fundamental spintronic phenomena: tunneling

magnetoresistance and spin-dependent thermoelectric transport, that are strongly enhanced by

the nearly perfect spin polarization realized in altermagnetic spintronic devices constructed

from KV 2 Se 2 O -based heterostructures.

To transform the crystal spin filtering of KV 2 Se 2 O into a functional device, we construct a

fully epitaxial magnetic tunnel junction (MTJ) in which a Bi 2 O 2 Se barrier is sandwiched

between two KV 2 Se 2 O electrodes in Fig. S2. Owing to the small lattice mismatch between

Bi 2 O 2 Se and KV 2 Se 2 O, Bi 2 O 2 Se is chosen as the tunnel barrier, which preserves the intrinsic d-

wave spin texture at the interface without introducing significant momentum scattering. Here,

the parallel (PC) and antiparallel (APC) configurations refer to the relative orientation of the 10 / 24 alternating magnetic sublattice spin textures projected along the transport direction, rather than

a net magnetization alignment. In both configurations, the net magnetization remains zero. The

detailed device geometry and the local spin-resolved electron density are shown in Fig. S3. Fig.

3 presents the spin-dependent transport characteristics by directly linking the microscopic

energy-resolved state, described by the projected local density of states (PLDOS) [Figs. 3(a)-

(d)], to the macroscopic momentum-resolved transmission [(Figs. 3(e)-(h)]. In the PC

configuration, the junction exhibits a half-metallic-like transport behavior. For the spin-down

channel, the PLDOS [Fig. 3(b)] reveals a continuous electronic-state bridge connecting the

source and drain across the Bi 2 O 2 Se barrier. This transportation channel connectivity gives rise

to a robust transmission peak near the Fermi energy, as indicated by the red curve, which

directly corresponds to a high-transparency window (bright region) in momentum space [Fig.

3(f)]. In contrast, the spin-up channel [Fig. 3(a)] is strictly blocked due to a symmetry-enforced

spectral gap along the transport direction, resulting in a vanishing transmission coefficient [T(E)]

and a dark momentum-resolved map [Fig. 3(e)]. As a consequence, the tunnel current becomes

highly spin polarized, with a spin polarization reaching approximately 92.5% in the absence of

any external field. This performance significantly surpasses that of the benchmark altermagnet

RuO 2[22] , and rivals the efficiency of conventional ferromagnetic spin injectors [57-59] . In the APC

configuration, the crystal spin filter of the drain electrode is opposite to the spin-polarized

current injected from the source. As shown in Figs. 3(c) and 3(d), although transport states may

exist in one electrode, they lack matching counterparts in the other electrode, leading to rapid

decay of the wave function at the interface. As a result, the total transmission for both spin

channels is strongly suppressed, with the red curves remaining close to zero, and the

momentum-resolved spectra [Figs. 3(g) and (h)] show no propagating channels. The sharp

contrast between the conducting PC state and the insulating APC state gives rise to a giant

tunneling magnetoresistance (TMR) ratio of approximately 2.24 × 10 3 % in the absence of any

external magnetic field, which is more than an order of magnitude larger than the previously

reported TMR of about 93% in RuO 2 -based devices. [22] . These results establish the KV 2 Se 2 O

/Bi 2 O 2 Se heterostructure as a promising platform for field-free spintronic logic applications.

11 / 24

FIG. 3. Ballistic spin transport and tunneling magnetoresistance effect in a KV2 S e2 O /Bi2 O 2 S e/KV2 S e2 O junction. (a–d) Energy-resolved transport mechanism. The colormaps display the Projected Local Density of States (PLDOS) along the transport direction (kz ) , while the overlaid red curves represent the energy-dependent transmission coefficient T(E). The dashed grey line indicates the Fermi level (EF ) . (a, b) Parallel Configuration (PC): An effective half-metallic state is realized. The spin-down channel (b) shows a continuous conducting pathway and a high transmission peak at EF , whereas the spin- up channel (a) is strictly gapped with vanishing transmission. (c, d) Anti-Parallel Configuration (APC): Transport is blocked for both spins. The mismatch between the source and drain band structures leads to interface-localized states and negligible transmission coefficients (flat red curves), explaining the high resistance state. (e–h) Momentum-resolved transmission spectra at EF corresponding to the cases in (a–d). The bright yellow region in (f) confirms the highly transparent spin-down channel in the PC state, while the dark regions in (e, g, h) indicate the effective spin blockade driven by spin segregation.

2.4 Giant Spin-Dependent Thermoelectricity Benefiting from the high-temperature altermagnetic nature of KV₂Se₂O, which stabilizes spin

order under thermal gradients, we can explore thermally generated spin currents in

KV 2 Se 2 O/Bi 2 O 2 Se/KV 2 Se 2 O heterostructures. Here, we explore the potential of

KV 2 Se 2 O/Bi 2 O 2 Se heterostructures for spin caloritronic by investigating thermally generated

spin currents. The thermal transport properties were evaluated under two distinct regimes: (i)

fixed temperature difference (ΔT) with varying base left electrode temperature (T L ), and (ii)

fixed base left electrode temperature near room temperature with varying ΔT. Fig. 4(a)

illustrates the thermally induced currents under a fixed ΔT. In the parallel configuration (PC),

the device operates as a highly efficient thermal spin generator. The spin-down current (red)

rises significantly with temperature, driven by the Seebeck effect in the metallic minority

12 / 24 channel. Conversely, the spin-up current (blue) remains suppressed which blocks thermally

excited carriers just as effectively as electrically injected ones. This asymmetry results in a

colossal thermal tunneling magnetoresistance, exceeding 750% at low temperatures and

remaining substantial even at 300 K [Fig. 4(b)]. The robustness of this effect for ambient

applications is demonstrated in Figs. 4(c) and 4(d), where the T L is fixed near room temperature

(T L in [290, 300, 310] K). A nearly linear dependence of the spin-polarized current on ΔT is

observed in the PC state, confirming the diffusive nature of the thermal transport. Most notably,

the spin filtration efficiency remains remarkably high (>80%) across the entire temperature

range tested [Fig. 4(d)], which indicates that the crystallographic spin segregation mechanism

is energetically robust against thermal broadening, positioning KV 2 Se 2 O as a promising

13 / 24 candidate for zero-bias spin logic and waste-heat spin harvesting devices. FIG. 4. Thermally driven spin currents and magnetoresistance. Spin-dependent thermoelectric transport properties of the KV2 S e2 O /Bi2 O 2 S e/KV2 S e2 O junction. (a, b) Fixed Temperature Difference (ΔT). (a) Thermally generated current as a function of the left electrode temperature (TL ) for ΔT = 10, 20, 30 K. In the PC state, a pure spin-down current (red) is generated, while the spin-up current (blue) is blocked. (b) The corresponding spin filtration efficiency (SFE) and thermal TMR ratio. The TMR remains robust up to room temperature. (c, d) Fixed left electrode temperature (TL ) . (c) Thermal current as a function of ΔT with TL fixed near room temperature (290–310 K). (d) The device maintains a near-unity SFE (>80%) and a rising TMR with increasing thermal gradient, confirming that the crystal orientation filtering mechanism is stable against thermal fluctuations at room temperature.

3 DISCUSSION

Our investigation establishes a unified framework for realizing high-performance spintronics

in compensated magnetic systems, governed by the principle of crystallographic spin

segregation. By bridging the symmetry-resolved toy model (Fig. 1) with specific material

realizations (Figs. 2-4), we demonstrate that robust spin filtering can be rigorously enforced by

magnetic crystal symmetries that allow a d-wave spin splitting, providing a deterministic route

to generate pure spin currents without net magnetization.

Our approach is conceptually based on the distinction between direction-selective spin

transport and symmetry-compensated spin splitting. While previous studies have classified

altermagnets primarily based on global Fermi surface topology, our work highlights that

functional spin selectivity is fundamentally dictated by local momentum-space connectivity. In

KV 2 Se 2 O, the characteristic d-wave symmetry imposes a strict orthogonal segregation of spin

channels along the principal axes. Consequently, ballistic transport along the Γ-X (or Γ-Y)

direction is populated exclusively by a single spin species. This crystallographic locking

effectively utilizes the anisotropic band structure to "switch on" high spin polarization simply

by aligning the device geometry with the symmetry-broken crystal axes. The translation of this

intrinsic quantization into macroscopic device performance is validated by the giant tunneling

magnetoresistance observed in the all-epitaxial KV 2 Se 2 O/Bi 2 O 2 Se heterostructure. The

calculated TMR ratio of 2.24 × 10 3 % represents an order-of-magnitude improvement over

previously reported values in RuO 2 -based devices.

Crucially, our results extend the utility of altermagnets into the domain of spin caloritronic.

The persistence of a high spin filtration efficiency (SFE> 80%) and substantial thermal TMR

at room temperature indicates that the energy scale of the symmetry-protected spin splitting

significantly exceeds the thermal broadening. This resilience is a direct consequence of the

large spectral gaps identified in the PLDOS analysis, which effectively block thermally excited

carriers in the forbidden channel. The ability to harvest pure spin currents from waste heat

without external magnetic fields positions KV 2 Se 2 O as a versatile platform for energy-efficient,

zero-bias spin logic.

14 / 24 We emphasize that the orientation-dependent filtering mechanism identified here is generic

to the d-wave symmetry class. By exploiting the synergy between altermagnetic ordering and

epitaxial engineering, our work paves the way for a new generation of scalable, field-free, and

room-temperature spintronic devices that combine the ultrafast dynamics of antiferromagnets

with the high readout fidelity of ferromagnets.

DATA AVAILABILITY

The data supporting the findings of this study are available from the corresponding authors

upon reasonable request.

ACKNOWLEDGEMENTS

This work is supported by the Shenzhen Science and Technology Program (Grant No.

JCYJ20250604174400001), the Guangdong Basic and Applied Basic Research Foundation

(Grant No. 2023A1515011852), and the Guangdong Science and Technology Innovation

Strategy Special Fund (Grant No. pdjh2025bk012). P.-H. F. acknowledges the financial support

from the Fondazione Cariplo under the grant 2023-2594. The authors would like to thank

National Supercomputer Center in Guangzhou for providing high performance computational

resources.

REFERENCE [1] [2] [3] [4] [5] Song Cheng, Bai Hua, Zhou Zhiyuan, et al. Altermagnets as a new class of functional materials [J]. Nature Reviews Materials, 2025, 10(6): 473-85. 10.1038/s41578-025- 00779-1 Wolf S. A., Awschalom D. D., Buhrman R. A., et al. Spintronics: A Spin-Based Electronics Vision for the Future [J]. Science, 2001, 294(5546): 1488-95. 10.1126/science.1065389 Žutić Igor, Fabian Jaroslav, Das Sarma S. Spintronics: Fundamentals and applications [J]. Reviews of Modern Physics, 2004, 76(2): 323-410. 10.1103/RevModPhys.76.323 Chappert Claude, Fert Albert, Van Dau Frédéric Nguyen. The emergence of spin electronics in data storage [J]. Nature Materials, 2007, 6(11): 813-23. 10.1038/nmat2024 Bader S.D., Parkin S.S.P. Spintronics [J]. 2010, 1(Volume 1, 2010): 71-88.

15 / 24 [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] https://doi.org/10.1146/annurev-conmatphys-070909-104123 Jungwirth T., Marti X., Wadley P., et al. Antiferromagnetic spintronics [J]. Nature Nanotechnology, 2016, 11(3): 231-41. 10.1038/nnano.2016.18 Baltz V., Manchon A., Tsoi M., et al. Antiferromagnetic spintronics [J]. Reviews of Modern Physics, 2018, 90(1): 015005. 10.1103/RevModPhys.90.015005 Šmejkal Libor, Mokrousov Yuriy, Yan Binghai, et al. Topological antiferromagnetic spintronics [J]. Nature Physics, 2018, 14(3): 242-51. 10.1038/s41567-018-0064-5 Chen Hongyu, Liu Li, Zhou Xiaorong, et al. Emerging Antiferromagnets for Spintronics [J]. Advanced Materials, 2024, 36(14): 2310379. https://doi.org/10.1002/adma.202310379 Šmejkal Libor, Sinova Jairo, Jungwirth Tomas. Emerging Research Landscape of Altermagnetism [J]. Physical Review X, 2022, 12(4): 040501. 10.1103/PhysRevX.12.040501 Fukaya Yuri, Lu Bo, Yada Keiji, et al. Superconducting phenomena in systems with unconventional magnets [J]. Journal of Physics: Condensed Matter, 2025, 37(31): 313003. 10.1088/1361-648X/adf1cf Ashcroft Neil W., Mermin N. David. Solid State Physics [M]. New York, NY: Holt, Rinehart and Winston, 1976. Šmejkal Libor, Hellenes Anna Birk, González-Hernández Rafael, et al. Giant and Tunneling Magnetoresistance in Unconventional Collinear Antiferromagnets with Nonrelativistic Spin-Momentum Coupling [J]. Physical Review X, 2022, 12(1): 011028. 10.1103/PhysRevX.12.011028 Šmejkal Libor, Sinova Jairo, Jungwirth Tomas. Beyond Conventional Ferromagnetism and Antiferromagnetism: A Phase with Nonrelativistic Spin and Crystal Rotation Symmetry [J]. Physical Review X, 2022, 12(3): 031042. 10.1103/PhysRevX.12.031042 Chen Xiaobing, Liu Yuntian, Liu Pengfei, et al. Unconventional magnons in collinear magnets dictated by spin space groups [J]. Nature, 2025, 640(8058): 349-54. 10.1038/s41586-025-08715-7 Bai H., Han L., Feng X. Y, et al. Observation of Spin Splitting Torque in a Collinear Antiferromagnet RuO2 [J]. Physical Review Letters, 2022, 128(19): 197202. 10.1103/PhysRevLett.128.197202 Karube Shutaro, Tanaka Takahiro, Sugawara Daichi, et al. Observation of Spin-Splitter Torque in Collinear Antiferromagnetic RuO2 [J]. Physical Review Letters, 2022, 129(13): 137201. 10.1103/PhysRevLett.129.137201 Lai Junwen, Yu Tianye, Liu Peitao, et al. d-Wave Flat Fermi Surface in Altermagnets Enables Maximum Charge-to-Spin Conversion [J]. Physical Review Letters, 2025, 135(25): 256702. 10.1103/bf1n-sxdl Zhu Haiyuan, Li Jiayu, Chen Xiaobing, et al. Magnetic geometry induced quantum geometry and nonlinear transports [J]. Nature Communications, 2025, 16(1): 4882. 10.1038/s41467-025-60128-2 Peng Rui, Fang Shibo, Ho Pin, et al. Ferroelastic altermagnetism [J]. npj Quantum Materials, 2025. 10.1038/s41535-025-00835-7 16 / 24 [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] Yang Zongmeng, Yang Xingyue, Wang Jianhua, et al. Unconventional thickness scaling of coherent tunnel magnetoresistance in altermagnets [J]. Physical Review B, 2025, 112(20): 205202. 10.1103/2thy-fzzj Shao Ding-Fu, Jiang Yuan-Yuan, Ding Jun, et al. N'eel Spin Currents in Antiferromagnets [J]. Physical Review Letters, 2023, 130(21): 216702. 10.1103/PhysRevLett.130.216702 Shao Ding-Fu, Tsymbal Evgeny Y. Antiferromagnetic tunnel junctions for spintronics [J]. npj Spintronics, 2024, 2(1): 13. 10.1038/s44306-024-00014-7 Fu Pei-Hao, Lv Qianqian, Xu Yong, et al. All-electrically controlled spintronics in altermagnetic heterostructures [J]. npj Quantum Materials, 2025, 10(1): 111. 10.1038/s41535-025-00827-7 Gonzalez Betancourt R. D, Zubáč J., Gonzalez-Hernandez R., et al. Spontaneous Anomalous Hall Effect Arising from an Unconventional Compensated Magnetic Phase in a Semiconductor [J]. Physical Review Letters, 2023, 130(3): 036702. 10.1103/PhysRevLett.130.036702 Chi Boyuan, Jiang Leina, Zhu Yu, et al. Crystal-facet-oriented altermagnets for detecting ferromagnetic and antiferromagnetic states by giant tunneling magnetoresistance [J]. Physical Review Applied, 2024, 21(3): 034038. 10.1103/PhysRevApplied.21.034038 Yan Zhi, Yang Ruixia, Fang Cheng, et al. Giant electrode effect on tunneling magnetoresistance and electroresistance in van der Waals intrinsic multiferroic tunnel junctions using VS2 [J]. Physical Review B, 2024, 109(20): 205409. 10.1103/PhysRevB.109.205409 Wu Yanzhao, Deng Li, Yin Xiang, et al. Valley-Related Multipiezo Effect and Noncollinear Spin Current in an Altermagnet Fe2Se2O Monolayer [J]. Nano Letters, 2024, 24(34): 10534-9. 10.1021/acs.nanolett.4c02554 Zhang Run-Wu, Cui Chaoxi, Li Runze, et al. Predictable Gate-Field Control of Spin in Altermagnets with Spin-Layer Coupling [J]. Physical Review Letters, 2024, 133(5): 056401. 10.1103/PhysRevLett.133.056401 Fernandes Rafael M., De Carvalho Vanuildo S., Birol Turan, et al. Topological transition from nodal to nodeless Zeeman splitting in altermagnets [J]. Physical Review B, 2024, 109(2): 024404. 10.1103/PhysRevB.109.024404 Zhai Yahui, Yu Longju, Lv Jian, et al. Zeeman-type spin splittings in strained $d$-wave altermagnets [J]. Physical Review B, 2025, 112(17): 174411. 10.1103/7q2d-jcqg Sumita Shuntaro, Naka Makoto, Seo Hitoshi. Phase-modulated superconductivity via altermagnetism [J]. Physical Review B, 2025, 112(14): 144510. 10.1103/3k12-2467 Fu Pei-Hao, Mondal Sayan, Liu Jun-Feng, et al. Light-induced Floquet spin-triplet Cooper pairs in unconventional magnets [J]. 2025. Fu Pei-Hao, Mondal Sayan, Liu Jun-Feng, et al. Floquet engineering spin triplet states in unconventional magnets [J]. Physical Review Letters, 2026. 10.1103/lkf9-jgv6 Yang Xin-Ren, Fang Shibo, Yang Zongmeng, et al. Altermagnetic Flatband-Driven Fermi Surface Geometry for Giant Tunneling Magnetoresistance, F, 2025 [C]. Ziletti A., Huang S. M., Coker D. F., et al. Van Hove singularity and ferromagnetic 17 / 24 [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] instability in phosphorene [J]. Physical Review B, 2015, 92(8): 085423. 10.1103/PhysRevB.92.085423 Datta Supriyo. Electronic Transport in Mesoscopic Systems [M]. Cambridge: Cambridge University Press, 1995. Liu Jiayu, Zhan Jie, Li Tongrui, et al. Absence of Altermagnetic Spin Splitting Character in Rutile Oxide RuO2 [J]. Physical Review Letters, 2024, 133(17): 176401. 10.1103/PhysRevLett.133.176401 Feng Zexin, Zhou Xiaorong, Šmejkal Libor, et al. An anomalous Hall effect in altermagnetic ruthenium dioxide [J]. Nature Electronics, 2022, 5(11): 735-43. 10.1038/s41928-022-00866-z Tschirner Teresa, Keßler Philipp, Gonzalez Betancourt Ruben Dario, et al. Saturation of the anomalous Hall effect at high magnetic fields in altermagnetic RuO2 [J]. APL Materials, 2023, 11(10): 101103. 10.1063/5.0160335 Li Yu-Xin, Chen Yiyuan, Pan Liqing, et al. Exploration of Altermagnetism in RuO2 [J]. 2025. Jaeschke-Ubiergo Rodrigo, Bharadwaj Venkata-Krishna, Campos Warlley, et al. Atomic altermagnetism [J]. 2025. https://doi.org/10.48550/arXiv.2503.10797 Von N. W. Ashcroft Und N. D. Mermin; Holt Rinehart and Winston. Solid State Physics. New York 1976, [J]. Physik in unserer Zeit, 1978, 9(1): 33-. https://doi.org/10.1002/piuz.19780090109 Borda László, Fritz Lars, Andrei Natan, et al. Theory of inelastic scattering from quantum impurities [J]. Physical Review B, 2007, 75(23): 235112. 10.1103/PhysRevB.75.235112 Zaránd Gergely, Borda László, Von Delft Jan, et al. Theory of Inelastic Scattering from Magnetic Impurities [J]. Physical Review Letters, 2004, 93(10): 107204. 10.1103/PhysRevLett.93.107204 Quang Doan Nhat, Tung Nguyen Huyen, Tuoc Vu Ngoc, et al. Quantum and transport lifetimes due to roughness-induced scattering of a two-dimensional electron gas in wurtzite group-III-nitride heterostructures [J]. Physical Review B, 2006, 74(20): 205312. 10.1103/PhysRevB.74.205312 Smidstrup Søren, Markussen Troels, Vancraeyveld Pieter, et al. QuantumATK: an integrated platform of electronic and atomic-scale modelling tools [J]. Journal of Physics: Condensed Matter, 2020, 32(1): 015901. 10.1088/1361-648X/ab4007 Smidstrup Søren, Stradi Daniele, Wellendorff Jess, et al. First-principles Green'sfunction method for surface calculations: A pseudopotential localized basis set approach [J]. Physical Review B, 2017, 96(19): 195309. 10.1103/PhysRevB.96.195309 Brandbyge Mads, Mozos José-Luis, Ordejón Pablo, et al. Density-functional method for nonequilibrium electron transport [J]. Physical Review B, 2002, 65(16): 165401. 10.1103/PhysRevB.65.165401 Jiang Bei, Hu Mingzhe, Bai Jianli, et al. A metallic room-temperature d-wave altermagnet [J]. Nature Physics, 2025, 21(5): 754-9. 10.1038/s41567-025-02822-y Anisimov Vladimir I., Zaanen Jan, Andersen Ole K. Band theory and Mott insulators:

18 / 24 [52] [53] [54] [55] [56] [57] [58] [59] Hubbard U instead of Stoner I [J]. Physical Review B, 1991, 44(3): 943-54. 10.1103/PhysRevB.44.943 Dudarev S. L., Botton G. A., Savrasov S. Y., et al. Electron-energy-loss spectra and the structural stability of nickel oxide: An LSDA+U study [J]. Physical Review B, 1998, 57(3): 1505-9. 10.1103/PhysRevB.57.1505 Büttiker M., Imry Y., Landauer R., et al. Generalized many-channel conductance formula with application to small rings [J]. Physical Review B, 1985, 31(10): 6207-15. 10.1103/PhysRevB.31.6207 Li Shuyi, Zhang Yu, Bahri Adrian, et al. Altermagnetism and strain induced altermagnetic transition in Cairo pentagonal monolayer [J]. npj Quantum Materials, 2025, 10(1): 83. 10.1038/s41535-025-00793-0 Ma Hai-Yang, Hu Mengli, Li Nana, et al. Multifunctional antiferromagnetic materials with giant piezomagnetism and noncollinear spin current [J]. Nature Communications, 2021, 12(1): 2846. 10.1038/s41467-021-23127-7 Zhu Yu-Peng, Chen Xiaobing, Liu Xiang-Rui, et al. Observation of plaid-like spin splitting in a noncoplanar antiferromagnet [J]. Nature, 2024, 626(7999): 523-8. 10.1038/s41586-024-07023-w Moodera Jagadeesh S., Mathon George. Spin polarized tunneling in ferromagnetic junctions [J]. Journal of Magnetism and Magnetic Materials, 1999, 200(1): 248-73. https://doi.org/10.1016/S0304-8853(99)00515-6 Manno M., Frakie R., Bolon B., et al. Spin-dependent intergranular transport in highly spin-polarized Co1−xFexS2 thin films [J]. Applied Physics Letters, 2009, 95(18): 182510. 10.1063/1.3262958 Yang Zhi, Qin Shen-Ao, Ruan Bao-Fu, et al. Van der Waals multiferroic tunnel junction with giant tunneling electroresistance and magnetoresistance toward bidirectional photoresponse and photoassisted memory [J]. Physical Review B, 2025, 112(11): 115312. 10.1103/9m1x-q7dn 19 / 24 Supplemental Material of “Intrinsic Spin Filter Effect in a d-wave altermagnet KV 2 Se 2 O

with Open Fermi Surface” Bin Liu, 1 Pei-Hao Fu, 1, 2, # Yu-Xuan Sun, 1 Xiao-Lin Zhang, 3 Si-Cong Zhu, 4 Xiang-Long Yu, 1, * Hua

Wu, 5, † Yuan-Zhi Shao 1, 6, ‡ 1 2 3 School of Science, Sun Yat-sen University, Shenzhen 518107, China Department of Physics and Astronomy, University of Florence, I-50019 Sesto Fiorentino, Italy Key Laboratory of Artificial Micro- and Nano-structures of Ministry of Education, School of Physics

and Technology, Wuhan University, Wuhan 430072, China 4 The State Key Laboratory for Refractories and Metallurgy, Wuhan University of Science and

Technology, Wuhan 430081, China 5 Laboratory for Computational Physical Sciences (MOE), State Key Laboratory of Surface Physics,

and Department of Physics, Fudan University, Shanghai 200433, China 6 Guangdong Provincial Key Laboratory of Magnetoelectric Physics and Devices, Sun Yat-sen

University, Guangzhou 510275, China # * Corresponding author: [email protected] Corresponding author: [email protected]

† ‡ Corresponding author: [email protected] Corresponding author: [email protected] 20 / 24

Fig. S1. Energy dispersion of the low-energy effective d-wave altermagnetic model [Eq. (1)] along 𝒌𝒙 with 𝒌𝒚 = 𝟎, together with the corresponding density of states [Eq. (2)] as a function of energy, Insets illustrate the Fermi-surface geometry determined by the altermagnetic parameter 𝒕𝐀 𝐌/ 𝒕𝟎 : a critical (flat) Fermi surface with zero curvature at 𝒕𝐀 𝐌/ 𝒕𝟎 = 𝟏

21 / 24

FIG. S2. Atomic geometry of the all-epitaxial magnetic tunnel junction. Schematic illustration of the KV2 S e2 O /Bi2 O 2 S e/KV2 S e2 O heterostructure used for quantum transport calculations. (a) Top view: The device consists of a central Bi2 O 2 S e tunnel barrier (scattering region, blue background) sandwiched between two semi-infinite KV2 S e2 O electrodes (beige background). Transport is defined along the z-axis. (b) Front view: Illustrating the periodic arrangement perpendicular to the transport direction. (c) Side view: The atomic stacking viewed along the transport axis, highlighting the epitaxial registry at the interface.

Fig. S2 displays the spin-resolved local density of states (LDOS) projected onto the scattering region of the magnetic tunnel junction (MTJ). The heterostructure is constructed using Bi 2 O 2 Se as the tunnel barrier, selected for its excellent lattice matching and chemical compatibility with the KV 2 Se 2 O electrodes, which minimizes interface scattering and preserves the intrinsic dwave spin texture. In the parallel configuration (PC), a striking dichotomy in wavefunction penetration is observed. The spin-down channel [Fig. S2(b)] exhibits continuous, delocalized states extending from the source electrode, through the Bi 2 O 2 Se barrier, and into the drain, confirming the high-transparency state predicted by the orientation-dependent spin filter model. In contrast, the spin-up channel [Fig. S2(a)] shows strong localization within the electrodes with negligible spectral weight inside the barrier region, indicative of the symmetry-forbidden transport gap along the stacking direction. The blocking mechanism is further visualized in the anti-parallel configuration (APC). Here, the mismatch between the spin-projected pass-bands of the two electrodes leads to a suppression of wavefunction overlap. As shown in

22 / 24 Supplementary Figs. S2(c, d), electronic states are confined to the injection side of the junction, decaying rapidly within the barrier without coupling to the collector. This spatial segregation of states in the APC limit, contrasted with the robust tunneling channel in the PC spin-down limit, visually corroborates the giant magnetoresistance effect arising from the crystallographic spin filtering.

23 / 24

FIG. S3. Real-space projection of spin-dependent transport states. Spatial distribution of the spin- resolved Local Density of States (LDOS) at the Fermi energy for a KV2 S e2 O /Bi2 O 2 S e/KV2 S e2 O magnetic tunnel junction. (a, b) Parallel Configuration (PC): The spin-up channel (green) is strongly attenuated due to the absence of bulk transport states, whereas the spin-down channel (blue) demonstrates significant wavefunction penetration across the Bi2 O 2 S e barrier, corresponding to the high-conductance state. (c, d) Anti-Parallel Configuration (APC): Both spin- up (c) and spin-down (d) states exhibit spin-selective mismatch between the source and drain electrodes.

24 / 24