Orbital-Selective ARPES Techniques

• Orbital-selective ARPES is a spectroscopic technique that uses polarization, incidence geometry, and photon energy to control and resolve specific orbital contributions in Bloch states.
• It employs a matrix element formulation that couples orbital coefficients with final-state control, enabling precise mapping of interference effects and orbital textures.
• This approach has been successfully applied to Fe-based superconductors, WSe₂, and nickelates to uncover orbital differentiation and complex many-body interactions.

Orbital-selective angle-resolved photoemission spectroscopy is the use of polarization, incidence geometry, photon energy, and final-state control in ARPES to enhance or suppress contributions from specific orbital characters within a Bloch state. In this regime, ARPES is not merely a band-mapping technique: the measured photocurrent depends on a matrix element that is strongly momentum-, geometry-, and polarization-dependent, so the observable intensity reflects a coherent product of spectral-function physics and selection rules, structure factors, and interference among orbital and sublattice emission channels (Yen et al., 2024).

1. Conceptual basis and formal structure

The standard ARPES intensity is written, within the sudden and three-step approximations commonly used to interpret ARPES, as

$$I(\mathbf{k},\omega)\propto |M_{fi}(\mathbf{k})|^2\,A(\mathbf{k},\omega)\,f(\omega),$$

where $A(\mathbf{k},\omega)$ is the single-particle spectral function and $f(\omega)$ is the Fermi–Dirac distribution. The spectral function encodes many-body physics, whereas the dipole matrix element $M_{fi}$ encodes orbital sensitivity, symmetry filtering, and geometry dependence (Zhang et al., 2022).

A more explicit formulation recasts the photocurrent through Fermi’s Golden Rule as

$$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$

with

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$

where $\psi_{\mathbf{k}\alpha}$ is the initial Bloch state and $\chi_{\mathbf{k},E}$ the final photoelectron state. In a Wannier tight-binding representation, the initial state is expanded in localized orbitals,

$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$

and the matrix element becomes a coherent sum of orbital-resolved terms,

$$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$

The intensity therefore separates into diagonal orbital terms and cross terms,

$A(\mathbf{k},\omega)$0

so orbital selectivity is intrinsically a coherent phenomenon rather than a direct readout of orbital weight alone (Yen et al., 2024).

This structure has two immediate consequences. First, low intensity does not imply low orbital content, because matrix elements can strongly suppress a visible channel. Second, momentum-space asymmetries and dichroic textures are often signatures of inter-orbital or inter-sublattice interference rather than simple occupancy differences. This is why orbital-selective ARPES is most informative when intensity is analyzed together with symmetry, polarization, and photon-energy dependence rather than as a static band image.

2. Symmetry filtering, polarization, and parity selection

The most widely used form of orbital selectivity exploits mirror-plane symmetry. When the scattering plane coincides with a crystal mirror plane, the parity of the dipole operator fixes which initial states can contribute. For p-polarized light, whose electric field lies in the mirror plane, the dipole operator is even under reflection and predominantly couples to even-parity initial states. For s-polarized light, whose electric field is perpendicular to that plane, the operator is odd and predominantly couples to odd-parity initial states (Wang et al., 2012).

In Fe-based superconductors this logic is often implemented by choosing the emission plane along a high-symmetry cut. Along $A(\mathbf{k},\omega)$1–M with an approximately $A(\mathbf{k},\omega)$2 emission plane, $A(\mathbf{k},\omega)$3, $A(\mathbf{k},\omega)$4, and $A(\mathbf{k},\omega)$5 are even, whereas $A(\mathbf{k},\omega)$6 and $A(\mathbf{k},\omega)$7 are odd; p-polarized light therefore enhances the former set and s-polarized light the latter. Along rotated cuts such as $A(\mathbf{k},\omega)$8–X, the parity of $A(\mathbf{k},\omega)$9 and $f(\omega)$0 interchanges with azimuth, so angle-dependent combinations

$f(\omega)$1

become the natural even and odd channels with respect to the chosen measurement plane (Wang et al., 2012).

Soft-x-ray ARPES in layered nickelates provides a closely analogous example. In Eu$f(\omega)$2Sr$f(\omega)$3NiO$f(\omega)$4, with the scattering plane coinciding with the $f(\omega)$5 mirror plane, s polarization enhances the odd $f(\omega)$6-derived band and suppresses the even $f(\omega)$7 band, whereas p polarization enhances $f(\omega)$8 and suppresses $f(\omega)$9. At soft-x-ray photon energies this parity filtering is combined with improved $M_{fi}$0 resolution and increased bulk sensitivity, making orbital-selective mapping of three-dimensional Fermi surfaces possible (Uchida et al., 2011).

These rules are exact only within the adopted symmetry frame. Small misalignments alter the effective mirror plane and can locally change even–odd assignments. Moreover, once orbitals are strongly hybridized, the measured contrast is no longer binary: the component-wise matrix elements $M_{fi}$1, $M_{fi}$2, and $M_{fi}$3, their relative phases, and the azimuthal dependence of the Bloch coefficients determine how much of a mixed state survives a given geometry (Wang et al., 2012).

3. Light–matter gauge, final states, and the ubiquity of interference

The light–matter coupling in ARPES starts from minimal coupling,

$M_{fi}$4

which yields two common interaction forms. In the velocity gauge,

$M_{fi}$5

and in the length gauge,

$M_{fi}$6

Gauge invariance requires exact initial and final states of the same Hamiltonian with a complete basis and consistent approximations. In practice, tight-binding or Wannier truncations, atom-centered approximations, and approximate final states inevitably break gauge invariance, so gauge comparison becomes a diagnostic rather than a mere formal choice (Yen et al., 2024).

Final-state modeling is correspondingly central to orbital selectivity. In the plane-wave approximation, the photoelectron is taken as $M_{fi}$7, which captures geometry-dependent structure-factor phases such as $M_{fi}$8 and often gives useful intuition. However, plane-wave final states have known deficiencies: in velocity gauge they vanish for $M_{fi}$9, and with purely real $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$0 they suppress circular dichroism. Dipole gauge with the atom-centered approximation alleviates some of these shortcomings, because it permits $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$1 coupling and allows circular dichroism to depend on orbital character (Yen et al., 2024).

More realistic descriptions expand the final state in partial waves. In the locally distorted wave approximation, the photoelectron near an atom is written in spherical harmonics with radial functions determined by a local Schrödinger equation and outgoing boundary conditions. The local Coulomb-wave approximation replaces the atomic potential by $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$2, yielding analytic Coulomb radial functions and scattering phases $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$3. These phase shifts and energy-dependent radial amplitudes are the microscopic origin of many photon-energy-dependent dichroic patterns and channel crossovers. In the WSe$$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$4 benchmark, dipole gauge plus LDWA captures photon-energy trends and CDAD sign reversals, whereas dipole gauge plus PWA reproduces only the gross texture and fails in the energy dependence (Yen et al., 2024).

A complementary physical interpretation is the Huygens picture of ARPES, in which each contributing orbital acts as a source of dipole-allowed outgoing Coulomb wavelets. In the Fraunhofer or plane-wave limit, ARPES becomes an orbital Fourier-transform problem. Once Coulomb distortion is restored, the two allowed dipole channels $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$5 acquire kinetic-energy-dependent amplitudes and relative phases, and linear dichroism emerges from their interference. This framework explains why plane-wave treatments often reproduce gross momentum structure but miss kinetic-energy-dependent final-state effects (Moser, 2022).

Recent first-principles work reformulates the final-state problem through a Lippmann–Schwinger equation that enforces the time-reversed LEED boundary condition directly inside standard DFT frameworks. In that approach, the realistic final state includes open Laue channels and evanescent components, and the matrix element is evaluated in the velocity gauge as $$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$6. The resulting calculations reproduce photon-energy- and polarization-dependent ARPES in graphene and WSe$$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$7, including the dark corridor and circular dichroism tied to hidden orbital polarization (Ryoo et al., 1 Aug 2025).

4. Dichroic and interferometric observables

Circular dichroism in ARPES is conventionally defined as

$$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$8

It can probe intrinsic orbital angular momentum when interference is weak or symmetry-filtered, when a single orbital channel dominates, and when the geometry preserves the relevant mirror symmetry. In general, however, CD also contains extrinsic geometric and interference contributions, including both intra-atomic and inter-atomic terms. This intrinsic-versus-extrinsic distinction is a persistent interpretive issue in orbital-selective ARPES rather than a marginal correction (Yen et al., 2024).

Time-reversal dichroism in photoelectron angular distributions extends this logic by replacing polarization switching with a crystal reorientation that maps one valley to its time-reversal partner. In bulk 2H-WSe$$I(\mathbf{k},E)=\sum_{\alpha} f_\alpha(\mathbf{k})\, |M_{\alpha}(\mathbf{k},E)|^2\, \delta(\varepsilon_\alpha(\mathbf{k})+\omega - E),$$9, a $$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$0 azimuthal rotation maps K to K′, and one defines

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$1

followed by

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$2

By construction, TRDAD removes geometry-invariant contributions and isolates the component that reverses under effective time reversal. In the minimal three-orbital description of WSe$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$3, the photoemission intensity contains a large interference term

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$4

and TRDAD images the sign structure of this interference, thereby revealing the hidden orbital pseudospin texture (Beaulieu et al., 2020).

Polarization-modulated ARPES introduces a related but distinct interferometric observable. In that method the linear polarization axis is continuously rotated, so that

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$5

which can be recast as

$$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$6

The Fourier harmonic $$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$7 separates the ordinary linear dichroism, $$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$8, from an interference quantity $$M_{\alpha}(\mathbf{k},E)=\langle \chi_{\mathbf{k},E}| \hat{\Delta} | \psi_{\mathbf{k}\alpha} \rangle,$$9 called Fourier dichroism in PAD. Together with $\psi_{\mathbf{k}\alpha}$0 and $\psi_{\mathbf{k}\alpha}$1, this permits reconstruction of the magnitude of CDAD without circular photons and, in a two-orbital subspace, inversion for the orbital pseudospin components $\psi_{\mathbf{k}\alpha}$2, $\psi_{\mathbf{k}\alpha}$3, and $\psi_{\mathbf{k}\alpha}$4 of the Bloch wavefunction (Schüler et al., 2021).

Resonant photoemission adds another orbital filter by tuning the photon energy to a core-level absorption edge. In angle-resolved resonant photoemission at the Fe $\psi_{\mathbf{k}\alpha}$5 edge, the total amplitude is a coherent sum of direct valence photoemission and a second-order resonant channel,

$\psi_{\mathbf{k}\alpha}$6

with Fano interference between direct and resonant contributions. Because the intermediate 2p core hole has strong spin–orbit coupling and a controllable hole polarization $\psi_{\mathbf{k}\alpha}$7, AR-RPES selectively enhances spin- and orbital-flip channels and has been used in bcc Fe to image pure spin-flip and entangled spin-flip–orbital-flip excitations in the diffraction pattern of the emitted electron (Pieve, 2015).

5. Material realizations and orbital fingerprints

Monolayer and bulk WSe$\psi_{\mathbf{k}\alpha}$8 have become a benchmark system because the top valence manifold near K and K′ combines strong Berry curvature, intrinsic orbital angular momentum, and nontrivial orbital texture. Wannier-based simulations show that the top valence band is dominated by W 5d $\psi_{\mathbf{k}\alpha}$9 orbitals with Se 4p admixture at finite binding energy, and that simulated CD-ARPES exhibits a polar texture around K with nodal lines and a pronounced photon-energy dependence from 10 to 120 eV. Decomposition into intra-atomic W-only and inter-atomic W–Se contributions shows that local OAM fingerprints are strongly modified by both intra-atomic d-channel interference and inter-atomic phase factors $\chi_{\mathbf{k},E}$0, whose photon-energy dependence drives sign changes and suppression or enhancement of CDAD (Yen et al., 2024).

The same material has also been used to establish TRDAD and PM-ARPES. TRDAD on bulk 2H-WSe$\chi_{\mathbf{k},E}$1 shows strong alternating positive and negative contrast around K and K′, while the hexagonal $\chi_{\mathbf{k},E}$2-centered band with mostly $\chi_{\mathbf{k},E}$3 character vanishes after antisymmetrization, consistent with a single-orbital sector. PM-ARPES on 2H-WSe$\chi_{\mathbf{k},E}$4 recovers valley-dependent modulation phases, reconstructs the absolute value of CDAD without circular photons, and yields an orbital pseudospin texture whose $\chi_{\mathbf{k},E}$5 component is dominated by $\chi_{\mathbf{k},E}$6 near the valley center and decreases anisotropically away from it (Beaulieu et al., 2020); (Schüler et al., 2021).

Fe-based superconductors provided some of the earliest systematic implementations of orbital-selective ARPES. In Ba$\chi_{\mathbf{k},E}$7K$\chi_{\mathbf{k},E}$8Fe$\chi_{\mathbf{k},E}$9As$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$0 and FeSe$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$1Te$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$2, polarization- and geometry-dependent simulations reproduce strongly asymmetric Fermi-surface intensity patterns and support specific orbital assignments: in Ba$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$3K$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$4Fe$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$5As$$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$6, $$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$7 is predominantly $$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$8, $$\psi_i(\mathbf{k},\mathbf{r}) = \sum_{\alpha} c_{\alpha}(\mathbf{k})\, \Phi_{\alpha}(\mathbf{k},\mathbf{r}),$$9 predominantly $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$0, and $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$1 predominantly $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$2 at $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$3, while the electron ellipses at M ունեն tip regions with $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$4 character and inner segments with $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$5 character (Wang et al., 2012). In LiFeAs, polarization-resolved ARPES at $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$6 eV isolates $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$7 as predominantly $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$8, $$M_{fi}(\mathbf{k}) \approx \sum_{\alpha} c_{\alpha}(\mathbf{k})\, m_{\alpha}(\mathbf{k})\, e^{i\mathbf{k}\cdot \boldsymbol{\tau}_\alpha}.$$9, $A(\mathbf{k},\omega)$00, and $A(\mathbf{k},\omega)$01 as predominantly $A(\mathbf{k},\omega)$02, and $A(\mathbf{k},\omega)$03 as mixed $A(\mathbf{k},\omega)$04 and $A(\mathbf{k},\omega)$05. The $A(\mathbf{k},\omega)$06 band shows the largest renormalization, a band renormalization factor of approximately 4 and an effective mass of about $A(\mathbf{k},\omega)$07, together with a much stronger temperature-induced loss of coherence than the $A(\mathbf{k},\omega)$08-derived bands, supporting a Hund’s-coupling-driven orbital-differentiated coherence–incoherence crossover rather than an orbital-selective Mott transition (Miao et al., 2016).

In NdFeAs$A(\mathbf{k},\omega)$09P$A(\mathbf{k},\omega)$10O$A(\mathbf{k},\omega)$11F$A(\mathbf{k},\omega)$12, orbital selectivity resolves a different phenomenon: P-polarized light detects $A(\mathbf{k},\omega)$13 $A(\mathbf{k},\omega)$14 and $A(\mathbf{k},\omega)$15 $A(\mathbf{k},\omega)$16, while S-polarized light detects $A(\mathbf{k},\omega)$17 $A(\mathbf{k},\omega)$18 and $A(\mathbf{k},\omega)$19 $A(\mathbf{k},\omega)$20. The $A(\mathbf{k},\omega)$21 band shifts from about $A(\mathbf{k},\omega)$22 meV at $A(\mathbf{k},\omega)$23 to about $A(\mathbf{k},\omega)$24 meV at $A(\mathbf{k},\omega)$25, and a weak $A(\mathbf{k},\omega)$26 component appears at the binding energy of the $A(\mathbf{k},\omega)$27 band top, indicating reconstruction of the originally degenerate $A(\mathbf{k},\omega)$28 doublet by the unoccupied $A(\mathbf{k},\omega)$29 band. The reported correlation between the evolution of this incipient $A(\mathbf{k},\omega)$30 component and the composition dependence of $A(\mathbf{k},\omega)$31 suggests that near-$A(\mathbf{k},\omega)$32 $A(\mathbf{k},\omega)$33 weight is an important ingredient for superconductivity in this 1111 family (Tin et al., 2022).

Soft-x-ray orbital-selective ARPES in Eu$A(\mathbf{k},\omega)$34Sr$A(\mathbf{k},\omega)$35NiO$A(\mathbf{k},\omega)$36 resolves a quasi-two-dimensional $A(\mathbf{k},\omega)$37 hole cylinder together with a $A(\mathbf{k},\omega)$38-centered small $A(\mathbf{k},\omega)$39 electron pocket that is present at $A(\mathbf{k},\omega)$40 and absent at $A(\mathbf{k},\omega)$41. The coexistence of these two $A(\mathbf{k},\omega)$42 sectors near $A(\mathbf{k},\omega)$43, together with substantial off-nodal $A(\mathbf{k},\omega)$44 admixture to the large hole surface, provides a concrete multiband contrast to the more nearly single-band near-nodal cuprate phenomenology (Uchida et al., 2011).

Angle-resolved resonant photoemission in bcc Fe shows that orbital selectivity can also operate in the excitation manifold rather than only in the direct valence-band matrix element. At the Fe $A(\mathbf{k},\omega)$45 edge, parallel geometry with circular polarization reveals a near-$A(\mathbf{k},\omega)$46 majority $A(\mathbf{k},\omega)$47 peak and a spin-down channel containing an exchange-induced spin-flip contribution from spin-up valence states. The resonant diffraction patterns exhibit helicity-dependent $A(\mathbf{k},\omega)$48 twists and vortex-like wavefronts associated with different $A(\mathbf{k},\omega)$49 continuum components, linking orbital symmetry, spin flip, and local correlation physics in an itinerant ferromagnet (Pieve, 2015).

6. Computational practice, interpretive limits, and persistent misconceptions

Orbital-selective ARPES is unusually sensitive to theoretical approximations. High-quality Wannier functions, symmetry-consistent orbital centers and phases, and tight-binding bands that reproduce DFT are necessary before any matrix-element decomposition can be trusted. Within that framework, a common practical recommendation is to prefer dipole gauge with atom-centered approximations for transparency, compute velocity-gauge counterparts as a robustness check, and use the local gauge constraint

$A(\mathbf{k},\omega)$50

to tune LCWA effective charges and diagnose gauge artifacts (Yen et al., 2024).

Several caveats recur across the literature. Intensity is not orbital weight; matrix elements can strongly suppress or enhance a band, so missing intensity is not a proof of missing orbital character. Symmetrization can erase left–right asymmetries, quadrant selectivity, and dichroic contrasts that are themselves the orbital diagnostic. Circular dichroism is not automatically a direct measure of intrinsic OAM, because it may contain large geometric and interference contributions. Final-state effects, surface scattering, and photon-energy-dependent cross-sections can qualitatively alter visibility outside simple parity arguments, so systematic scans in polarization, azimuth, and $A(\mathbf{k},\omega)$51 are required (Wang et al., 2012); (Zhang et al., 2022).

Computationally, the field spans a hierarchy from compact matrix-element engines to fully first-principles one-step solvers. The software package chinook evaluates ARPES matrix elements in the length gauge with localized initial states, a plane-wave final state expanded in spherical harmonics, attenuation $A(\mathbf{k},\omega)$52, and explicit site-resolved structure-factor phases $A(\mathbf{k},\omega)$53. It is designed for arbitrary photon energy, polarization, spin projection, and slab or bulk models, and is particularly useful for rapid experiment design and interpretation of orbital, spin, and interference effects (Day et al., 2019). At the more rigorous end, Lippmann–Schwinger-based final states provide TLEED boundary conditions directly within standard DFT workflows, improving the realism of polarization- and photon-energy-dependent matrix elements without requiring specialized KKR infrastructure (Ryoo et al., 1 Aug 2025).

A stable interpretive strategy therefore combines symmetry-aware geometry, polarization switching, and photon-energy scans with explicit matrix-element simulations. When those ingredients are used together, orbital-selective ARPES can separate intrinsic orbital texture from extrinsic interference effects, map hidden orbital polarization, identify orbital differentiation in correlated metals, and expose multiorbital band reconstruction that would be inaccessible in symmetry-averaged spectra (Yen et al., 2024).