---
title: Momentum-Space Nonsymmorphic Symmetry
url: https://www.emergentmind.com/topics/momentum-space-nonsymmorphic-symmetry
type: topic
---

# Momentum-Space Nonsymmorphic Symmetry

Momentum-space nonsymmorphic symmetry denotes the momentum-space action of symmetry operations that combine a point-group part with a fractional translation, so that Bloch eigenstates acquire momentum-dependent phases and, in the broader projective setting, crystal momentum may transform affinely as $k \mapsto Rk+\kappa_R$. Its characteristic consequences are projective little-group representations, momentum-dependent symmetry eigenvalues, band sticking on Brillouin-zone boundaries, and symmetry-enforced band connectivity such as nodal lines, Dirac points, hourglass dispersions, and Möbius-like band structures [2202.06927][2306.14410][2512.21844].

## 1. Definition and algebraic structure

In Seitz notation, a space-group element is written as $g=\{R|t\}$, with $R$ a point operation and $t$ a translation that may be non-primitive. On Bloch states or Bloch eigenfields, the defining momentum-space feature is the phase from the fractional translation. In the electronic notation used for PtPb$_4$,
$$
U_g(k)\psi_k(r)=e^{-ik\cdot t}\,u_m\,\psi_{mk}(r),
$$
while in the photonic formulation,
$$
gF_k(r)=e^{-ik\cdot t}U(R)F_{Rk}(r).
$$
The same structure underlies glide mirrors, screw rotations, and related nonsymmorphic operations in electronic, photonic, and acoustic settings [2202.06927][2505.19452].

The algebra of these operators depends on both momentum and the spin character of the representation. For a glide or screw with fractional translation $t$, squaring gives
$$
g^2=e^{-i2k\cdot t}\quad\text{for spinless electrons},
\qquad
g^2=-e^{-i2k\cdot t}\quad\text{for spinful electrons},
$$
and for a spinful glide one therefore obtains eigenvalues of the form
$$
\lambda_\pm(k)=\pm i\,e^{-ik\cdot t}.
$$
At Brillouin-zone boundaries where $e^{-i2k\cdot t}=-1$, these relations force special eigenvalue structures and, in many cases, band sticking [2202.06927][2207.00020].

A broader formulation replaces purely linear momentum-space actions by affine ones. In the momentum-space crystallographic-group framework, point-group elements act as
$$
\rho_F(R)k=Rk+\kappa_R,
$$
and the resulting group is the twisted semidirect product
$$
\Gamma_F=L_F\rtimes_{(c_F,\omega_F)}G.
$$
Within this language, momentum-space nonsymmorphism is encoded by the fractional reciprocal shifts $\kappa_R$ and the corresponding cocycle $\omega_F$; in flux-based constructions, these affine actions arise from projective representations of real-space space groups in the presence of symmetric gauge flux [2512.21844][2604.26575].

## 2. Little groups, band sticking, and enforced connectivity

The immediate k-space consequence of a nonsymmorphic operation is that irreducible representations of the little group depend explicitly on momentum through the phase $e^{-ik\cdot t}$. Along symmetry-invariant lines or planes, this modifies compatibility relations and can force connectivity that point-group symmetry alone does not require. In the simplest case, glide eigenvalues differ by a sign and a $k$-dependent phase, and at zone boundaries bands carrying opposite glide eigenvalues must connect, producing band sticking and hourglass-like dispersions [2505.19452][2207.00020].

When several nonsymmorphic operations coexist, their algebra on the boundary little group can enforce degeneracy more strongly. In PtPb$_4$, the spinless no-SOC representation on X–M satisfies $\{g_x,g_z\}=0$, so every eigenstate is paired with another of opposite glide eigenvalue and the entire boundary line is twofold degenerate. More generally, orthogonal nonsymmorphic operations may anticommute on boundary manifolds, whereas in spinful double-group representations the same operators may instead commute, allowing a symmetry-allowed mass term and removing the line degeneracy away from special points [2202.06927].

Several distinct connectivity mechanisms follow from this algebra. Order-two nonsymmorphic symmetries generate momentum-dependent twists of the form $e^{-ik_1}$ in Bloch-space representations, and these twists underlie hourglass and Möbius connectivities on symmetry-invariant lines and surfaces. In two-band systems, the momentum dependence can impose global constraints on the Bloch Hamiltonian, such as the relation $q(k)e^{ik}=q^*(k)$, forcing unavoidable crossings and yielding a global $\mathbb{Z}_2$ charge that differs from the local topological charge of Weyl points. Off-centered nonsymmorphic rotations and mirrors provide a related mechanism: on appropriate invariant lines or planes they force Kramers doublets to carry fixed symmetry eigenvalues, and crossings between opposite-eigenvalue doublets then produce Dirac point or line nodes even with strong spin-orbit coupling [1606.03698][1604.00843][2207.00020].

## 3. Electronic realizations in crystalline solids

PtPb$_4$ provides a concrete three-dimensional realization in which the square-net motif generates two orthogonal glide mirrors and a screw axis in tetragonal $P4/nbm$ (No. 125). In the absence of SOC, the orthogonal glides anticommute along X–M and R–A and enforce essential Dirac nodal lines along the Brillouin-zone boundaries. ARPES places the main nodal feature at a binding energy of approximately $-0.6$ eV relative to $E_F$, and constant-energy maps at $-0.6$ eV show an intense grid-like feature precisely on the BZ boundaries. With SOC, however, the double-group algebra changes: $g_x$ and $g_z$ commute, the boundary lines become generically nondegenerate apart from Kramers pairing, and fourfold degeneracies remain only at X, M, R, and A. Experimentally, the nodal-line bandwidth is less than $0.2$ eV in ARPES, substantially smaller than in DFT, and DMFT-based renormalization reduces but does not eliminate this discrepancy, leading to the description of PtPb$_4$ as a system with narrow crossings approximately protected by nonsymmorphic crystalline symmetries [2202.06927].

A complementary two-dimensional realization is $\alpha$-bismuthene, a centrosymmetric nonmagnetic layer-group-$pman$ material whose key nonsymmorphic element is the glide mirror $\widetilde M_z=\{m_z|\tfrac12,\tfrac12\}$. In the spinful representation,
$$
(\widetilde M_z)^2=-e^{-i(k_x+k_y)},
\qquad
g_z=\pm i\,e^{-i(k_x+k_y)/2}.
$$
At $\bar X_1=(\pi,0)$ and $\bar X_2=(0,\pi)$, the glide and inversion anticommute, $\{\widetilde M_z,P\}=0$, and together with $T$ this enforces fourfold Dirac degeneracies. $\mu$-ARPES observes Dirac-like dispersions at $\bar X_1$ and $\bar X_2$, with DP1 at approximately $0.7$ eV below the Fermi level and DP2 at approximately $0.4$ eV below the Fermi level. Without SOC, the system has a nodal line along X$_1$–M–X$_2$; with SOC, that nodal line is gapped everywhere except at X$_1$ and X$_2$, which remain symmetry-enforced Dirac points [1906.08456].

The same general theme reappears in $P4/nmm$ systems. At the tetragonal M point, the nonsymmorphic little group admits a single four-dimensional spinful irrep, so all bands there are fourfold degenerate. In one direction, this feeds directly into magnetic exchange: the momentum-dependent electronic structure near M produces Rashba-type spin-momentum locking and diversified RKKY interactions, including Heisenberg-type, Dzyaloshinskii–Moriya-type, Ising-type, and anisotropic terms, with strong dependence on whether the relevant carriers are near $\Gamma$ or M. In another direction, the same nonsymmorphic setting supports spin–sublattice–momentum locking that favors the inter-sublattice odd-parity triplet $A_{2u}$ pairing channel, making spin-triplet superconductivity the leading instability even with weak SOC when the Fermi surface lies near the BZ boundary [2210.08439][2208.09409].

## 4. Artificial crystals, wave systems, and numerical implementations

Artificial platforms have made momentum-space nonsymmorphic symmetry directly engineerable. In a two-dimensional acoustic crystal with a chess-board pattern of $\pi$ and $0$ fluxes, real-space mirror reflections acquire projective minus signs with translations,
$$
\{M_x,L_y\}=\{M_y,L_x\}=0,
$$
and become momentum-space glide reflections,
$$
G_x:(k_x,k_y)\mapsto(-k_x,k_y+\pi),\qquad
G_y:(k_x,k_y)\mapsto(k_x+\pi,-k_y).
$$
These k-space glides act freely on the Brillouin torus, reduce the momentum-space manifold to a Klein bottle, and replace the ordinary Chern classification by a $\mathbb{Z}_2$ invariant adapted to the non-orientable manifold. Acoustic experiments observed the glide-reflected iso-frequency contours, the associated boundary identifications, and symmetry-protected in-gap edge states predicted by this momentum-space group structure [2305.07174].

A closely related acoustic realization shows that a pair of anticommutative momentum-space glide reflections can protect a higher-order topological insulator. In that checkerboard-$\pi$-flux system, the symmetry operators satisfy
$$
\{M_x,M_y\}=0,\qquad
M_x^2=e^{ik_y},\qquad
M_y^2=e^{ik_x},
$$
and enforce half-translation pairing of edge bands on a cylinder geometry. The measured edge spectra satisfy
$$
E_{\text{top}}(k_x)=E_{\text{bottom}}(k_x+\pi),
$$
and momentum-resolved probability densities confirm the corresponding shift by half a reciprocal period. The same platform exhibits both an intrinsic HOTI transition, signaled by a bulk-gap closure when couplings along the open boundary are varied, and an extrinsic HOTI transition, signaled by an edge-gap closure when couplings along the closed boundary are varied, together with quadrupole corner modes detected by transmission and field-distribution measurements [2409.08196].

In photonics, nonsymmorphic symmetry has been turned into a computational principle. The nonsymmorphic symmetry adapted finite element method implements the boundary constraint
$$
F_k(r_b)=e^{-ik\cdot t}U(R)\,F_k(R^{-1}(r_b-t)),
$$
decomposes the Maxwell eigenproblem into irrep sectors of the little group, and reduces the FEM system via $x=Px'$, leading to $P^TAPx'=\lambda P^TBPx'$. The method was demonstrated for a rod group, a plane group, and a space group, including a metacrystal with a photonic nodal line between bands 2 and 3 at approximately $16$ GHz at U and Y. The metacrystal example reduced the number of degrees of freedom from $1{,}036{,}485$ to $248{,}369$ and the total run time from $23{,}366$ s to $4{,}851$ s, corresponding to approximately $4.8\times$ speedup [2505.19452].

Bosonic systems also realize symmetry-enforced Weyl nodes through nonsymmorphic little-group algebra. In several chiral nonsymmorphic space groups, boundary high-symmetry points admit only two-dimensional irreps, so every enforced doublet is a Weyl phonon. For K$_2$Sn$_2$O$_3$ in space group 199, first-principles phonons show twofold degeneracies at P that are monopole Weyl phonons and a spin-1 Weyl phonon at H near $17.5$ THz; the large separation between P and H produces long, clearly visible surface arcs [1911.07461].

## 5. Topological classification and cohomological formulations

The systematic classification of nonsymmorphic band topology first appeared in twisted equivariant $K$-theory for order-two nonsymmorphic space groups. In that setting, the Bloch-space symmetry relations take the form
$$
U(\sigma k)U(k)=\epsilon_\sigma e^{-ik_1},
$$
and the fractional translation is encoded as a $2$-cocycle twist in $K_G^\tau(T^d)$. This formalism yields a modulo-$2$ periodicity in the number of flipped coordinates, allows $\mathbb{Z}_2$ phases even without time-reversal or particle-hole symmetry, and produces novel $\mathbb{Z}_4$ phases when nonsymmorphic symmetry coexists with $T$ and/or $C$. The associated boundary manifestations include Möbius and hourglass surface states, with the $\mathbb{Z}_4$ hourglass phase realized in three-dimensional class AII glide systems [1511.01463].

A more algebraic reformulation uses momentum-space crystallographic groups $\Gamma_F$, defined by the affine action
$$
\rho_F(R)k=Rk+\kappa_R,
\qquad
\Gamma_F=L_F\rtimes_{(c_F,\omega_F)}G.
$$
In this framework, the determinant phases of symmetry operators define $U(1)$-valued $1$-cocycles, and there is a canonical isomorphism
$$
H^1(\Gamma_F,\mathcal F(\mathbb R_F^d,U(1)))\cong H^2(\Gamma_F,\mathbb Z).
$$
Accordingly, $H^2(\Gamma_F,\mathbb Z)$ classifies Abelian crystalline topological insulators, while
$$
H^3(\Gamma_F,\mathbb Z)\cong H^2(\Gamma_F,\mathcal F(\mathbb R_F^d,U(1)))
$$
classifies twistings of point-group actions on the Brillouin torus relevant to twisted equivariant $K$-theory. The two-dimensional momentum-space glide group Pg provides an explicit example, with
$$
H^2(Pg,\mathbb Z)\cong \mathbb Z_2,\qquad H^3(Pg,\mathbb Z)\cong 1,
$$
so the Abelian classification carries a $\mathbb Z_2$ invariant and all twistings are trivial in that case [2512.21844].

The general theory of momentum-space nonsymmorphic space groups shows that every k-NSG can be realized from a real-space symmorphic space group via a projective representation with canonical multiplier
$$
\nu[(t_1,R_1),(t_2,R_2)]=\exp\big(-i\,\kappa_{R_1}\cdot R_1t_2\big),
$$
provided the fractional reciprocal shifts satisfy
$$
\kappa_{R_1}+R_1\kappa_{R_2}-\kappa_{R_1R_2}\in\widehat{\mathcal L}.
$$
Gauge-flux constructions furnish explicit realizations of these projective multipliers. The flux-mediated correspondence sharpens this picture further: in the presence of symmetric flux,
$$
[\overline{\omega_F}]=[\overline{\Phi\cdot\omega}]+[\overline{\Lambda_\Phi}]
$$
in $H^{2,\tilde c_F}(G,Z_F/L_F)$, so real-space nonsymmorphicity can enforce momentum-space nonsymmorphicity, and in three exceptional arithmetic classes symmetric flux requires nonsymmorphicity in both real and momentum space [2306.14410][2604.26575].

## 6. Boundary manifestations, probes, and conceptual distinctions

Momentum-space nonsymmorphic symmetry is not only a bulk classification principle; it also leaves direct spectroscopic and boundary signatures. In quasiparticle interference, internal nonsymmorphic symmetries lock momentum components to internal symmetry eigenvalues, and this produces selection rules for replicated QPI peaks. If the STM tip satisfies $\hat m\hat R=\omega^\beta \hat R\hat m$, then the QPI channel for scattering between symmetry flavors $\nu$ and $\nu'$ obeys
$$
N_{\nu'-\nu}(\mathbf q)=0
\quad\text{if}\quad
\omega^{(\nu'-\nu)+(\beta-\alpha)}\neq 1,
$$
with $\alpha=(\mathbf q\cdot \mathbf a)/(2\pi)\bmod n$. The universal information is therefore encoded in missing QPI replicas or systematic intensity alternation across them, rather than only in peak positions [1712.09316].

Boundary geometry can also determine which manifestation of the same bulk phase is visible. In the two-dimensional nonsymmorphic antiferromagnetic Dirac model with screws
$$
\tilde{\mathcal S}_x(\mathbf k)=i\,e^{-ik_x/2}\tau_x\sigma_x,
\qquad
\tilde{\mathcal S}_y(\mathbf k)=i\,e^{-ik_y/2}\tau_x\sigma_y,
$$
screw-compatible edges retain gapless first-order edge states because the boundary mass is forbidden by the nonsymmorphic screw symmetry. A $45^\circ$ diamond termination breaks the screw at the boundary, gaps the edges, but still preserves the antiunitary magnetic mirrors $\mathcal M_x\mathcal T$ and $\mathcal M_y\mathcal T$, which enforce an alternating edge-mass pattern and bind corner zero modes. The second-order phase is characterized by a quantized quadrupole moment, and the corner states remain pinned to zero energy by the chiral symmetry $\mathcal C=\tau_y$ [2605.07206].

Several conceptual distinctions follow. First, nonsymmorphic protection is not synonymous with strict gaplessness: PtPb$_4$ shows that SOC can be symmetry-allowed, can change the little-group algebra from anticommuting to commuting glides, and can gap a nodal line everywhere except isolated high-symmetry points or protected accidental crossings [2202.06927]. Second, momentum-space nonsymmorphism is not restricted to the reciprocal shadow of real-space nonsymmorphic space groups: projective and flux-mediated constructions produce genuinely affine k-space actions, modified Brillouin-zone manifolds, and topological invariants that are absent in the conventional symmorphic reciprocal-space framework [2305.07174][2306.14410]. Third, the relevant topological data are often global rather than local: Möbius and hourglass connectivities, Klein-bottle reductions, and the global $\mathbb Z_2$ charges of some two-band semimetals all express constraints on how bands are sewn across the entire Brillouin zone rather than only at isolated nodes [1606.03698][2207.00020].

Momentum-space nonsymmorphic symmetry has therefore become a unifying language for a broad class of phenomena: symmetry-enforced band sticking on zone boundaries, SOC-sensitive deprotection and re-protection of crossings, artificial k-space glides generated by gauge flux, algebraic classifications via cohomology and twisted $K$-theory, and boundary responses ranging from hourglass fermions and Möbius manifolds to quadrupole corner modes and QPI selection rules. Across these settings, the central mechanism remains the same: fractional translations, whether in real space or reciprocal space, make symmetry representations projective in momentum space and thereby reshape the topology and connectivity of Bloch bands.

Source: https://www.emergentmind.com/topics/momentum-space-nonsymmorphic-symmetry