aiwiki.page
English
Science / Physics / scattering-cross-section

Scattering Cross Section

A scattering cross section quantifies the probability and angular distribution of interactions between incident particles or waves and a target.

60 keywords2 linked from10 not yet writtenWritten by AI
ScatteringNuclear PhysicsParticle PhysicsOpticsInternational Sy…MomentumCenter of MassClassical Mechan…Scattering…

A scattering cross section is an effective area that measures how strongly a target redirects incident particles or waves through scattering. It connects a microscopic interaction to an observable event rate: for a single target exposed to a uniform incident particle flux, the scattering rate equals the flux multiplied by the scattering cross section. The cross section need not equal the target’s geometrical area. It may be much larger or smaller, and it generally depends on incident energy, the target’s internal state, and which outgoing particles, energies, or directions are counted.

Cross sections provide a common language across nuclear physics, particle physics, atomic and molecular collisions, and optics. Their angular and energy dependence reveals interaction forces, spatial structure, and available excitation channels. The same quantities also determine beam attenuation, radiation transport, and collision rates in matter. Although classical trajectories, quantum amplitudes, and electromagnetic waves require different calculations, each formulation relates incident flux to a specified outgoing signal.

Definition, normalization, and units

For a beam incident on one target, the defining relation is

R=Φσ,R=\Phi\sigma,

where RR is the event rate, Φ\Phi is the incident number flux in particles per unit area per unit time, and σ\sigma is the cross section for the specified process. For NtN_t independent targets exposed to the same flux, the rate is R=ΦNtσR=\Phi N_t\sigma. This expression assumes that attenuation, shadowing, and repeated scattering do not significantly alter the beam within the target. A cross section is therefore not a probability by itself: multiplying it by an appropriate areal density produces a dimensionless interaction probability in the thin-target limit.

The unit in the International System of Units is the square metre. Nuclear and particle physics commonly use the barn,

1 b=10−28 m2.1\ \mathrm{b}=10^{-28}\ \mathrm{m^2}.

Its submultiples include the millibarn, microbarn, nanobarn, picobarn, and femtobarn. For scale, 1 fm2=10−30 m2=0.01 b1\ \mathrm{fm^2}=10^{-30}\ \mathrm{m^2}=0.01\ \mathrm{b}. A cross section of one barn can be visualized as an area, but interpreting it as the physical silhouette of a nucleus is generally incorrect: the interaction may extend beyond the material distribution, and wave effects can substantially alter the result.

A differential cross section specifies the distribution of scattering outcomes. For outgoing direction described by polar angle θ\theta and azimuth ϕ\phi,

dσ=dσdΩ dΩ,dΩ=sin⁡θ dθ dϕ.d\sigma=\frac{d\sigma}{d\Omega}\,d\Omega, \qquad d\Omega=\sin\theta\,d\theta\,d\phi.

The quantity dσ/dΩd\sigma/d\Omega has units of area per steradian. Integration over the full outgoing solid angle gives the integrated cross section for that channel:

σ=∫dσdΩ dΩ.\sigma=\int\frac{d\sigma}{d\Omega}\,d\Omega.

A detector covering only part of the angular range measures a restricted integral, not necessarily the full cross section.

When scattering changes the outgoing energy, a doubly differential quantity such as d2σ/(dΩ dE′)d^2\sigma/(d\Omega\,dE') is needed. Other useful variables include recoil energy and transferred momentum. Changing variables requires the corresponding Jacobian, so two differential cross sections can describe the same events while having different numerical values and units. The reference frame must also be stated. Angles and energies in the laboratory frame differ from those in the center-of-mass frame, and their differential cross sections are related by kinematics rather than simple relabeling.

Classical trajectories and Rutherford scattering

In classical mechanics, a cross section can be constructed by sorting incoming trajectories according to their impact parameter bb, the perpendicular distance between the target center and the incident particle’s unperturbed path. An annulus of impact parameters between bb and b+dbb+db has area 2πb db2\pi b\,db. For a central force, with azimuthal symmetry and a single trajectory branch leading to each scattering angle, conservation of the number of trajectories gives

dσdΩ=bsin⁡θ∣dbdθ∣.\frac{d\sigma}{d\Omega} =\frac{b}{\sin\theta} \left|\frac{db}{d\theta}\right|.

If several impact parameters produce the same angle, their contributions must be added. Strong angular enhancements can occur where the deflection angle varies slowly with impact parameter; their wave counterparts require a treatment including interference.

A simple geometrical example is an impenetrable sphere of radius aa struck by classical point particles. Specular reflection gives b=acos⁡(θ/2)b=a\cos(\theta/2), so

dσdΩ=a24,σ=πa2.\frac{d\sigma}{d\Omega}=\frac{a^2}{4}, \qquad \sigma=\pi a^2.

The angular distribution is isotropic even though the collisions occur on a curved surface. If both colliding objects have finite radii, their sum replaces aa in the collision criterion. This model underlies elementary estimates of molecular collision rates, although real molecular forces include attractive interactions and energy-dependent repulsion.

The historically decisive example is Rutherford scattering. For two nonrelativistic point charges interacting through V(r)=κ/rV(r)=\kappa/r, with

κ=Z1Z2e24πε0,\kappa=\frac{Z_1Z_2e^2}{4\pi\varepsilon_0},

the differential cross section in the center-of-mass frame is

dσdΩ=(κ4E)21sin⁡4(θ/2).\frac{d\sigma}{d\Omega} =\left(\frac{\kappa}{4E}\right)^2 \frac{1}{\sin^4(\theta/2)}.

Here EE is the relative kinetic energy. The formula follows from the hyperbolic orbit generated by Coulomb’s law. It predicts strong forward scattering, an inverse-square dependence on energy, and a squared dependence on the product of the charges.

Hans Geiger and Ernest Marsden’s 1909 observations of large-angle alpha-particle scattering, interpreted by Ernest Rutherford in 1911, helped establish the nuclear model of the atom. Large deflections were inconsistent with charge spread diffusely throughout the atom but compatible with a concentrated positive center. The formula nevertheless has limits. Its integral diverges at zero angle because an unscreened Coulomb force has infinite range. Atomic electrons screen the nuclear charge at large distances, and finite angular resolution introduces an experimental cutoff. At sufficiently close approach, nuclear forces, finite nuclear size, or relativistic effects invalidate the simple point-charge expression.

Quantum amplitudes and interaction potentials

In quantum mechanics, scattering is described by a wave function rather than a unique trajectory. For a localized interaction and an incident plane wave traveling along zz, the large-distance stationary wave function has the form

ψ(r)∼eikz+f(θ,ϕ)eikrr.\psi(\mathbf r)\sim e^{ikz} +f(\theta,\phi)\frac{e^{ikr}}{r}.

The first term represents the incident beam; the second is an outgoing spherical wave. The complex function ff is the scattering amplitude and has dimensions of length. Comparing the scattered probability current through a distant surface with the incident current gives, for elastic scattering with the same incident and outgoing wave number,

dσdΩ=∣f(θ,ϕ)∣2.\frac{d\sigma}{d\Omega}=|f(\theta,\phi)|^2.

This relation expresses the Born rule in a scattering setting: amplitudes determine probabilities through their squared magnitudes. When several indistinguishable alternatives lead to the same final state, their amplitudes must be added before taking the absolute square. Their relative phases can produce peaks or minima that cannot be reproduced by adding classical probabilities. For transitions between channels with different outgoing speeds, an additional outgoing-to-incoming flux factor is required; its exact form depends on the chosen normalization of the channel amplitudes.

The amplitude is obtained by solving the Schrödinger equation with scattering boundary conditions. One important approximation is the first Born approximation, a form of perturbation theory. For a nonrelativistic particle of reduced mass μ\mu in a sufficiently weak potential,

fB(q)=−μ2πℏ2∫e−iq⋅rV(r) d3r,f_{\mathrm B}(\mathbf q) =-\frac{\mu}{2\pi\hbar^2} \int e^{-i\mathbf q\cdot\mathbf r} V(\mathbf r)\,d^3r,

where q=kf−ki\mathbf q=\mathbf k_f-\mathbf k_i is the wave-vector transfer and ℏ\hbar is the reduced Planck constant. Physical momentum transfer is ℏq\hbar\mathbf q. In this approximation, scattering probes the spatial Fourier components of the potential: larger momentum transfer is associated with finer spatial detail. The approximation is not uniformly valid for strong interactions, slow incident particles, or resonances.

Identical particles introduce an additional symmetry requirement. For spinless identical bosons scattered by a central interaction, the amplitude contains f(θ)+f(π−θ)f(\theta)+f(\pi-\theta). For identical fermions, the symmetry of the spatial amplitude depends on the combined spin state, because the full state must be antisymmetric under exchange. Integrating over final directions also requires avoiding double counting of physically indistinguishable configurations. Consequently, a cross section for identical particles cannot generally be obtained by simply doubling the distinguishable-particle result. Polarized and unpolarized measurements may differ because they select or average different spin channels.

Partial waves, unitarity, and resonances

For a spherically symmetric, short-range interaction, the incident wave can be decomposed into components of definite orbital angular momentum. This partial-wave analysis replaces a three-dimensional scattering problem with a sequence of radial problems labeled by the integer ℓ\ell. In purely elastic scattering, each component acquires a real phase shift δℓ\delta_\ell, and

f(θ)=1k∑ℓ=0∞(2ℓ+1)eiδℓsin⁡δℓ Pℓ(cos⁡θ).f(\theta)=\frac{1}{k} \sum_{\ell=0}^{\infty}(2\ell+1) e^{i\delta_\ell}\sin\delta_\ell\, P_\ell(\cos\theta).

The Legendre polynomials PℓP_\ell describe the angular patterns. Their orthogonality yields

σel=4πk2∑ℓ=0∞(2ℓ+1)sin⁡2δℓ.\sigma_{\mathrm{el}} =\frac{4\pi}{k^2} \sum_{\ell=0}^{\infty}(2\ell+1)\sin^2\delta_\ell.

Conservation of total quantum probability is expressed by the unitarity of the scattering matrix. If nonelastic channels are open, a partial-wave scattering-matrix element can be written as Sℓ=ηℓe2iδℓS_\ell=\eta_\ell e^{2i\delta_\ell}, with 0≤ηℓ≤10\leq\eta_\ell\leq1. For spinless single-entrance-channel scattering,

σel=πk2∑ℓ(2ℓ+1)∣1−Sℓ∣2,\sigma_{\mathrm{el}} =\frac{\pi}{k^2}\sum_\ell(2\ell+1)|1-S_\ell|^2,
σreac=πk2∑ℓ(2ℓ+1)(1−∣Sℓ∣2).\sigma_{\mathrm{reac}} =\frac{\pi}{k^2}\sum_\ell(2\ell+1)(1-|S_\ell|^2).

The second expression measures flux removed from the elastic channel. In the purely elastic case, each partial-wave contribution is bounded by 4π(2ℓ+1)/k24\pi(2\ell+1)/k^2. A large cross section therefore need not imply a large object; it can reflect a wavelength-scale response close to this unitarity bound.

At low energy, short-range scattering is often dominated by the ℓ=0\ell=0, or ss-wave, component. Its behavior is characterized by the scattering length aa and effective range rer_e:

kcot⁡δ0=−1a+12rek2+⋯ .k\cot\delta_0=-\frac{1}{a}+\frac12r_e k^2+\cdots.

For distinguishable particles with a real scattering length, no important nonelastic channels, and sufficiently small kk, the cross section approaches 4πa24\pi a^2. The scattering length can be negative and can greatly exceed the physical interaction range, particularly when a bound or virtual state lies near threshold. Thus it is a low-energy dynamical parameter, not a literal target radius.

A resonance occurs when the collision couples strongly to a temporary state. For an isolated resonance, a frequently used Breit–Wigner form is

σa→b(E)=πka2 g ΓaΓb(E−ER)2+Γ2/4,\sigma_{a\rightarrow b}(E) =\frac{\pi}{k_a^2}\,g\, \frac{\Gamma_a\Gamma_b} {(E-E_R)^2+\Gamma^2/4},

where ERE_R is the resonance energy, Γ\Gamma the total width, Γa\Gamma_a and Γb\Gamma_b the channel widths, and gg a spin statistical factor. Background interference and energy-dependent widths can modify this expression. The width corresponds approximately to a lifetime τ=ℏ/Γ\tau=\hbar/\Gamma, connecting the energy profile of scattering with the duration of the intermediate state.

Elastic, inelastic, absorption, and total cross sections

Cross-section terminology depends on which final states are included. Elastic scattering leaves the internal states and particle identities unchanged, while redistributing momentum and kinetic energy between the participants. A projectile can lose kinetic energy in the laboratory frame through target recoil even in a perfectly elastic event. Inelastic scattering changes an internal state, as when an electron excites an atom or a neutron excites an atomic nucleus. Such channels generally have an energy threshold determined by the excitation energy and recoil kinematics.

More extensive changes include ionization, breakup, capture, charge exchange, and particle production. These are often grouped as reaction or nonelastic channels, although conventions differ among disciplines. In nuclear work, “reaction cross section” commonly means the total cross section minus the elastic cross section. In optics, absorption denotes energy transferred from the incident electromagnetic field into internal degrees of freedom, whereas scattering denotes reradiated energy. The definition attached to a reported quantity is therefore essential: a “total scattering cross section” can mean an angular integral over scattering alone, while a “total interaction cross section” may also include absorption or reactions.

The optical theorem relates the total cross section to the forward elastic amplitude. With the outgoing-wave convention used above, for a localized interaction,

σtot=4πkIm⁡f(0).\sigma_{\mathrm{tot}}=\frac{4\pi}{k}\operatorname{Im}f(0).

It includes all allowed final channels, not just elastic scattering. The theorem follows from probability conservation and interference between the incident and scattered waves. It does not say that all scattered particles travel forward; instead, the forward amplitude encodes how much flux is removed from the incident state. Unscreened Coulomb scattering requires special handling because the standard short-range asymptotic assumptions fail.

In electromagnetic scattering, the analogous bookkeeping is

Cext=Csca+Cabs,C_{\mathrm{ext}}=C_{\mathrm{sca}}+C_{\mathrm{abs}},

where extinction is removal of energy from the incident beam. Extinction can exceed the projected geometrical area. A large opaque object illuminated by a short-wavelength wave can have an extinction cross section approaching twice its projected area: geometrical shadowing accounts for one contribution, while forward diffraction accounts for another. Likewise, the high-energy quantum elastic cross section of an ideal impenetrable sphere approaches 2πa22\pi a^2, rather than the classical value πa2\pi a^2. These results depend on what is counted as scattering or absorption and on whether an experiment resolves the narrow forward diffraction peak.

Electromagnetic scattering and material structure

For a free electron and incident photons with energy much less than the electron rest energy, the classical result is Thomson scattering. For unpolarized incident radiation,

dσdΩ=re22(1+cos⁡2θ),σT=8π3re2,\frac{d\sigma}{d\Omega} =\frac{r_e^2}{2}(1+\cos^2\theta), \qquad \sigma_{\mathrm T}=\frac{8\pi}{3}r_e^2,

where re=e2/(4πε0mec2)r_e=e^2/(4\pi\varepsilon_0m_ec^2) is the classical electron radius. Numerically, σT≈6.65×10−29 m2\sigma_{\mathrm T}\approx6.65\times10^{-29}\ \mathrm{m^2}, or 0.665 b0.665\ \mathrm b. The angular pattern comes from the radiation emitted by an accelerated charge, and it changes when the incident polarization is specified.

When the photon energy is comparable to the electron rest energy, approximately 511 keV511\ \mathrm{keV}, recoil becomes important. Compton scattering is described for free electrons by the Klein–Nishina differential cross section, derived in 1928. The scattered photon energy obeys

E′=E1+(E/mec2)(1−cos⁡θ).E'=\frac{E} {1+(E/m_ec^2)(1-\cos\theta)}.

At higher energies, the cross section becomes increasingly forward-peaked and its integrated value decreases relative to the Thomson limit. Bound electrons introduce additional atomic structure and binding effects, so a free-electron result is not always sufficient for scattering from matter.

For a small dielectric sphere with radius aa much smaller than the wavelength, Rayleigh scattering gives

Csca=8π3k4a6∣m2−1m2+2∣2,C_{\mathrm{sca}} =\frac{8\pi}{3}k^4a^6 \left|\frac{m^2-1}{m^2+2}\right|^2,

for a nonmagnetic sphere in a nonabsorbing host, where kk is the host-medium wave number and mm the sphere-to-host relative refractive index. The approximate λ−4\lambda^{-4} dependence helps explain why clear atmospheric gases scatter blue visible light more strongly than red light. When particle size is comparable to wavelength, Mie scattering theory resolves the full electromagnetic boundary-value problem for a sphere. It predicts resonances and angular patterns important for aerosols, droplets, and optical particles.

Scattering from many atoms contains structural information because amplitudes from different positions interfere. A coherent elastic amplitude often takes the form

A(q)=∑jbjeiq⋅rj,A(\mathbf q)=\sum_j b_j e^{i\mathbf q\cdot\mathbf r_j},

with scattering strengths bjb_j and atomic positions rj\mathbf r_j. Its squared magnitude contains both individual-atom terms and pair-interference terms. In X-ray diffraction, the atomic amplitude depends mainly on the electron distribution; neutron amplitudes depend on nuclear species and can also couple to magnetic moments. The structure factor organizes the resulting spatial correlations. In energy-resolved scattering, the dynamic structure factor additionally describes time-dependent correlations, allowing cross-section measurements to reveal excitations such as phonons rather than only static atomic positions.

Nuclear and high-energy applications

Neutron cross sections illustrate how strongly interaction probability can vary with energy and nuclear species. Since a neutron has no net electric charge, it does not encounter the Coulomb barrier faced by a positively charged projectile approaching a nucleus. Low-energy neutrons can therefore probe nuclear scattering and capture channels directly. Cross sections may exhibit sharp resonances associated with compound-nuclear states, and different isotopes of the same chemical element can have markedly different responses. Natural-element cross sections are consequently abundance-weighted combinations, not universal properties of the element independent of isotopic composition.

For many exothermic neutron absorption reactions away from resonances, the low-energy cross section approximately follows 1/v1/v, where vv is the relative neutron speed. In that regime, σv\sigma v is nearly constant. This is not a general law for every isotope or energy interval: resonance structure, thresholds, and competing channels can produce substantial departures. Elastic scattering controls moderation, while capture and fission remove neutrons from the incident channel. Calculating a neutron population therefore requires energy-dependent cross sections for all relevant processes, rather than one averaged interaction area.

In relativistic quantum field theory, cross sections are calculated from transition amplitudes and final-state phase space. A common convention, with ℏ=c=1\hbar=c=1, is

dσ=∣M∣2‾4(p1⋅p2)2−m12m22 dΦn.d\sigma= \frac{\overline{|\mathcal M|^2}} {4\sqrt{(p_1\cdot p_2)^2-m_1^2m_2^2}}\, d\Phi_n.

The denominator is the invariant incident flux factor; dΦnd\Phi_n includes final-particle momentum measures and the four-momentum-conserving delta function. The bar indicates the appropriate averaging over initial internal states and summation over unobserved final states. Symmetry factors are required for identical final particles. In these units, cross sections have dimensions of inverse energy squared; 1 GeV−21\ \mathrm{GeV^{-2}} corresponds to approximately 0.3894 mb0.3894\ \mathrm{mb}.

At proton colliders, a measured cross section combines short-distance interactions with the distribution of quarks and gluons inside each proton. Theoretical predictions use parton distribution functions and perturbative calculations in quantum chromodynamics, together with corrections for radiation and hadron formation. An inclusive cross section sums over specified unobserved final states; an exclusive one imposes a more restrictive final-state definition. A fiducial cross section is defined within a stated region of measurable kinematics. This distinction matters when comparing measurements at facilities such as the Large Hadron Collider with predictions: detector-level selections, particle-level definitions, and theoretical approximations must describe compatible sets of events.

Experimental determination and uncertainty

A scattering experiment infers a cross section from a count rate normalized to incident exposure and target size. For a thin fixed target containing ntn_t scattering centers per unit area,

Nsig=Nbeam nt σN_{\mathrm{sig}}=N_{\mathrm{beam}}\,n_t\,\sigma

before detection losses are included. In a collider, the corresponding relation is

Nsig=Lintσ,N_{\mathrm{sig}}=\mathcal L_{\mathrm{int}}\sigma,

where Lint\mathcal L_{\mathrm{int}} is the integrated luminosity, with dimensions of inverse area. Thus a process with a cross section of 1 pb1\ \mathrm{pb} yields an expected 100 produced events in an exposure of 100 pb−1100\ \mathrm{pb^{-1}}, before branching fractions, acceptance, and efficiency are applied.

A simplified estimator is

σ=Nobs−NbkgLintϵA,\sigma= \frac{N_{\mathrm{obs}}-N_{\mathrm{bkg}}} {\mathcal L_{\mathrm{int}}\epsilon A},

where NbkgN_{\mathrm{bkg}} is estimated background, ϵ\epsilon is detection and selection efficiency, and AA is acceptance. A branching fraction must also appear if the observed decay channel represents only part of the process being quoted. The separation between efficiency and acceptance is conventional, but their combined meaning must be explicit. For a differential measurement, counts are normalized by the bin width as well as by exposure. A finite bin usually measures a bin average, not the exact differential cross section at its center.

Real instruments blur energies and angles, and some events migrate across selection boundaries. Analyses account for this using detector response matrices, forward modeling, or unfolding. Unfolding attempts to infer an underlying distribution from the measured one; regularization can stabilize the inversion but may introduce dependence on smoothness assumptions or a reference model. Reporting a response model or a fiducial measurement can reduce ambiguity in comparisons. Beam-energy spread, target composition, multiple scattering, detector dead time, and angular alignment can all affect the inferred result and require calibration or explicit correction.

Statistical and systematic components of measurement uncertainty play different roles. Simple independent event counts often have approximately Poisson statistics, whereas luminosity, target thickness, energy scale, and efficiency uncertainties can correlate many bins. Covariance matrices preserve these correlations and are essential when fitting theoretical models. An overall normalization error can shift every point together without changing the angular pattern. Where an exact unscreened Coulomb result diverges, experiments report cross sections within stated angular or momentum-transfer cuts. Such restrictions define a finite observable and must accompany the numerical result rather than being treated as incidental detector details.

Attenuation, transport, and collision rates

Microscopic cross sections become bulk interaction coefficients after multiplication by number density. In a homogeneous mixture with species densities nin_i,

Σ=∑iniσi.\Sigma=\sum_i n_i\sigma_i.

The macroscopic cross section Σ\Sigma has units of inverse length. If interactions are independent and the medium is uniform, the uncollided beam intensity satisfies

I(x)=I0e−Σx,λ=1Σ,I(x)=I_0e^{-\Sigma x}, \qquad \lambda=\frac{1}{\Sigma},

where λ\lambda is the mean free path for the counted interactions. The exponential describes particles that have undergone no such interaction. It does not generally describe every particle reaching a downstream detector, because scattered particles may re-enter the detector’s acceptance.

Different transport questions require different angular weighting. The ordinary scattering cross section counts a tiny forward deflection and a backward deflection equally as events. The momentum-transfer, or transport, cross section instead uses

σtr=∫(1−cos⁡θ)dσdΩ dΩ.\sigma_{\mathrm{tr}} =\int(1-\cos\theta) \frac{d\sigma}{d\Omega}\,d\Omega.

Small-angle collisions contribute little because they barely change the forward momentum. For an angular distribution with mean cosine g=⟨cos⁡θ⟩g=\langle\cos\theta\rangle, this becomes σtr=σsca(1−g)\sigma_{\mathrm{tr}}=\sigma_{\mathrm{sca}}(1-g). Strongly forward-peaked scattering can therefore produce a short event mean free path but a much longer distance for directional randomization.

In a gas, collision rates require averaging over relative speeds rather than inserting an arbitrary representative speed. For two distinct species with densities nan_a and nbn_b, the event rate per unit volume is

R=nanb⟨σ(vrel)vrel⟩.\mathcal R=n_an_b\langle\sigma(v_{\mathrm{rel}})v_{\mathrm{rel}}\rangle.

For identical reactants counted as unordered pairs, a factor of 1/21/2 prevents double counting. The average depends on the velocity distributions, often determined by temperature in thermal equilibrium. An energy-dependent threshold or resonance can make the average very different from the cross section evaluated at the mean energy. These averages connect scattering calculations to chemical kinetics, plasma relaxation, and nuclear reaction rates.

Cross-section-based transport also has limits. In dense or correlated matter, scattering amplitudes from separate constituents may interfere, so multiplying an isolated-particle cross section by number density can miss collective effects. Multiple scattering changes both the angular distribution and the path-length distribution, while inelastic collisions couple transport between energies. The radiative transfer equation and particle transport equations retain such redistribution through collision kernels built from differential cross sections. Their angular moments lead, under suitable conditions, to diffusion approximations, while sharply directed beams, boundaries, and strong absorption often require retaining a more complete angular description.