---
title: Intrabeam Scattering (IBS) in Accelerators
url: https://www.emergentmind.com/topics/intrabeam-scattering-ibs
type: topic
---

# Intrabeam Scattering (IBS) in Accelerators

Searching arXiv for recent and foundational accelerator papers on intrabeam scattering.
Intrabeam scattering (IBS) is the cumulative effect of Coulomb interactions between charged particles within the same bunch in an accelerator. Its principal beam-dynamical consequence is a change in transverse and longitudinal emittances, with corresponding changes in beam size, bunch length, momentum spread, brightness, and luminosity. In the conventional description, IBS is treated as the net outcome of many weak Rutherford scatterings that redistribute momentum among the degrees of freedom rather than directly removing particles from the beam [2602.11952]. Depending on machine regime, IBS may produce slow emittance dilution over hours, establish a synchrotron-radiation-limited equilibrium, drive diffusion across RF acceptance, or significantly increase slice energy spread in dense low-energy electron beams [1009.1562] [1108.1644] [2510.02058].

## 1. Definition and physical content

Intrabeam scattering refers to the effects of the Coulomb force acting between particles in a bunch. In a discrete-particle picture, particles scatter through many weak encounters, transferring momentum between transverse and longitudinal motion. The process is internal to the bunch: no external material is involved, and the dominant consequence is evolution of the phase-space distribution rather than direct particle loss [2602.11952].

A standard accelerator-physics interpretation is that IBS is a diffusion-like process. It broadens the beam by increasing rms momentum spread and emittances; in hadron colliders and storage rings this increases beam sizes and can reduce luminosity, while in low-emittance electron machines it can raise equilibrium emittance and energy spread above the values set by synchrotron radiation alone [1806.07317] [1308.0035]. In heavy-ion operation, IBS can also indirectly cause particle loss by diffusing particles in longitudinal phase space until they cross the RF-bucket boundary and debunch, a mechanism identified as especially important in RHIC without stochastic cooling [1009.1562].

The sign and qualitative interpretation of IBS depend on the longitudinal dynamics. One synthesis states that below transition the beam tends toward bounded redistribution and equipartition-like behavior, while above transition the transverse and longitudinal emittances may grow over time without limit [2602.11952]. For below-transition operation, the conserved combination is written as
$$
\sqrt{k_x}\varepsilon_x + \sqrt{k_y}\varepsilon_y + \sqrt{k_z}\varepsilon_z = \textrm{constant},
$$
with equilibrium condition
$$
\sqrt{k_x}\varepsilon_x = \sqrt{k_y}\varepsilon_y = \sqrt{k_z}\varepsilon_z.
$$
Above transition, the corresponding conserved combination contains a minus sign in the longitudinal term,
$$
\sqrt{k_x}\varepsilon_x + \sqrt{k_y}\varepsilon_y - \sqrt{|k_z|}\varepsilon_z = \textrm{constant},
$$
which explains why all three emittances can increase together in that idealized picture [2602.11952].

IBS is often contrasted with Touschek scattering. One formulation states that IBS mainly redistributes momentum and energy among the particles’ degrees of freedom, whereas Touschek scattering can kick particles out of the beam [2602.11952]. This distinction is operationally important because the measurable effects of IBS are often emittance growth and bunch-profile evolution rather than immediate intensity decay.

## 2. Analytical frameworks and growth-rate formalisms

The standard analytical treatments of IBS are based on Coulomb scattering theory and kinetic descriptions of weak collisions. A recent overview identifies three main theoretical levels: a simple ideal-gas model for physical insight, Piwinski’s formulae for quantitative growth rates, and Bjorken–Mtingwa formulae together with later approximations for faster computation [2602.11952].

In Piwinski’s formulation, the emittance growth rates are defined by
$$
\frac{d\varepsilon_i}{dt} = \frac{2\varepsilon_i}{\tau_i},
$$
with $i=x,y,z$ [2602.11952]. The growth rates are expressed through a lattice average of a function $f(a,b,q)$, with a prefactor
$$
A = \frac{\pi r_0^2 c N_b}{8\gamma_0 \Gamma},
$$
and beam-geometry variables
$$
a = \frac{\sigma_h}{\gamma_0}\sqrt{\frac{\beta_x}{\varepsilon_x}}, \qquad
b = \frac{\sigma_h}{\gamma_0}\sqrt{\frac{\beta_y}{\varepsilon_y}},
$$
$$
\frac{1}{\sigma_h^2} = \frac{1}{\sigma_\delta^2} + \frac{\eta_x^2}{\beta_x\varepsilon_x} + \frac{\eta_y^2}{\beta_y\varepsilon_y}.
$$
The dependence on $a$ and $b$ is strong, whereas the dependence on the impact-parameter cutoff parameter $q$ is relatively weak [2602.11952].

The Bjorken–Mtingwa formalism gives the same physics in a matrix form involving an auxiliary matrix $L$ and a Coulomb logarithm,
$$
(\log)_\mathrm{BM} = \ln\!\left(\frac{b_\mathrm{max}}{b_\mathrm{min}}\right),
$$
with typical choices $b_\mathrm{max}=\min(\sigma_x,\sigma_y)$ and $b_\mathrm{min}=r_0$ [2602.11952]. In practical accelerator calculations, Nagaitsev’s reformulation is widely used because it replaces slow quadratures by complete symmetric elliptic integrals of the second kind,
$$
R_D(x,y,z)=\frac{3}{2}\int_{0}^{\infty} \frac{dt}{\sqrt{(t+x)(t+y)(t+z)^3}},
$$
yielding efficient expressions for growth rates and diffusion coefficients suitable for tracking [2310.03504] [2308.02196].

Several papers emphasize that the Coulomb logarithm is not merely a numerical detail. In comparisons between SIRE, Piwinski, Bane, and Bjorken–Mtingwa-based models, differences in predicted growth rates are traced in part to different parameterizations of the impact-parameter range [1806.07317]. A separate CesrTA study showed that, in strongly damped lepton rings, agreement with measurements requires a tail-cut procedure that excludes rare large-angle events from the growth rate used to describe the measurable Gaussian core [1308.0035]. There, the maximum and minimum impact parameters are taken as
$$
b_{max}=\min\left(n^{-1/3},\sigma_x,\sigma_y,\gamma\sigma_z\right),
$$
$$
b_{min}=\sqrt{\frac{1}{n\pi\tau_b\nu}},
$$
leading to
$$
C_{\Lambda}=\log\frac{b_{max}}{b_{min}}.
$$
At $1.6\times10^{10}$ particles per bunch, the reported average Coulomb logarithm is approximately $9.4$ with tail-cut and approximately $17.6$ without it [1308.0035].

A plausible implication is that the apparently technical choice of cutoffs is tied to what observable is being modeled: total scattering activity, the Gaussian beam core, or the full non-Gaussian distribution.

## 3. Stochastic, kinetic, and collective descriptions

Beyond closed-form growth-rate formulae, IBS is implemented numerically in several distinct ways. In multi-particle tracking for heavy-ion colliders, IBS can be modeled as a random momentum kick given to each particle every turn. One RHIC/LHC study uses a modified Piwinski model, with kick amplitude modulated by the local longitudinal density $\rho_t(t)$ so that non-Gaussian bunch profiles can be handled [1009.1562]. A representative kick formula is
$$
\Delta p_u=r\sigma_{pu}\sqrt{2^{-1} T_\mathrm{rev} \sigma_t \sqrt{\pi}\rho_t(t)},
$$
with $u=x,y,z$ and $r$ a unit Gaussian random number [1009.1562].

A related effective-kick method is used in the CLIC damping-ring studies, where the kick is applied turn by turn as
$$
\Delta p_u=\sigma_{p_u}\sqrt{2\,T_{\text{IBS},u}^{-1}\,\sigma_z\sqrt{\pi}\,\rho(z)}\,R,
$$
with $R$ a Gaussian random variate [2308.02196]. In the CERN ion-injector chain, a Langevin form is used,
$$
P_i(t+\Delta t)=P_i(t)-F_i P_i(t)\rho(z)\Delta t +\sigma_{p_i}\sqrt{\Delta t\, G_{i}\rho(z)} \varsigma_{i},
$$
which incorporates plane-dependent friction and diffusion coefficients and modulates the kick by longitudinal line density [2310.03504]. These formulations treat IBS statistically rather than as explicit pairwise binary collisions in the lab frame.

A second class of methods uses Monte Carlo binary-collision models. In the HALF storage-ring study, an IBS module is developed for IMPACT within a PIC/MCC framework: PIC handles external-field tracking and MCC handles scattering events inside a bunch, with Piwinski’s binary-collision model used in the center-of-mass frame [2311.15454]. In a photoinjector context, a dedicated Monte Carlo model in REPTIL uses Nanbu’s cumulative binary collision method with spatial cells, random pairing, and momentum rotations in the center-of-mass frame; the local scattering strength is set by a dynamically computed Coulomb logarithm [2510.02058].

A third class consists of kinetic-theory formulations based on Fokker–Planck equations. In the SPS/LEIR work, IBS is introduced through friction and diffusion terms,
```latex
\frac{\partial \Phi}{\partial t}=-\frac{\partial}{\partial t} (F_m\Phi)+\frac{1}{2}\frac{\partial^2}{\partial P_m \partial P_{m'}(D_{m,m'}\Phi).
```
The corresponding momentum variables are
```latex
\vec{r} = \begin{pmatrix} z-z_s \\ x \\ y \end{pmatrix},\quad\quad
\vec{P} = \begin{pmatrix} \frac{1}{\gamma}\frac{\Delta p}{p} \\ x' \\ y' \end{pmatrix},
```
and the coefficients are derived from a Bjorken–Mtingwa-type auxiliary matrix [2310.03504].

A more radical revision of the standard picture appears in the work on long-range correlations in relativistic beams. That analysis argues that the binary-collision description is only partially valid because a particle is simultaneously influenced by fluctuating electric fields generated by many neighbors. In that approach, the energy-diffusion coefficient is written as an autocorrelation integral of the longitudinal self-field,
$$
{\cal D}_\eta = \left(\frac{e}{\gamma mc^2} \right)^2 \int_{-\infty}^0 ds\, \langle \delta\tilde E_z(0)\delta\tilde E_z(s) \rangle,
$$
and the conventional Coulomb-log result is recovered only after restricting attention to short-range fluctuations [2504.01867]. For finite energy spread, the numerical result is summarized as
$$
J = 33.3 + \frac{\pi^{3/2}}{8} \log(K_\mathrm{max} \sqrt{\epsilon \beta}),
$$
which the authors interpret as a long-range-correlation correction to the usual logarithmic behavior [2504.01867]. This suggests that conventional IBS theory may underestimate the effects of long-range correlations in very cold relativistic beams.

## 4. Interplay with synchrotron radiation, space charge, and beam distributions

IBS is rarely the only collective mechanism shaping beam evolution. In hadron colliders, storage rings, and damping rings, its operational significance is usually determined by competition with synchrotron radiation damping, space charge, beam-beam interactions, lattice nonlinearities, and the evolution of the distribution shape itself.

The HE-LHC study presents an unusual hadron-collider regime in which synchrotron radiation damping is strong enough to dominate IBS early in the store [1108.1644]. For a beam energy of $16.5$ TeV, the reported SR parameters are an energy loss per turn $U_0 = 206.3\ \mathrm{keV}$, SR power $67\ \mathrm{kW}$, transverse damping times $\tau_x=\tau_y=1.93\ \mathrm{h}$, and longitudinal damping time $\tau_E=0.96\ \mathrm{h}$ [1108.1644]. Using the smooth optics approximation of V. Lebedev, the initial IBS growth times are quoted as $82$ h horizontally and $72$ h longitudinally, far longer than the SR damping time [1108.1644]. The beam therefore evolves toward an SR–IBS balance,
$$
\left(\frac{d\varepsilon}{dt}\right)_{\mathrm{SR}} + \left(\frac{d\varepsilon}{dt}\right)_{\mathrm{IBS}} = 0,
$$
with the resulting beam emittance approximately half the initial value [1108.1644].

Heavy-ion studies at RHIC and the LHC show a different pattern. In RHIC without stochastic cooling, IBS-driven longitudinal diffusion can populate neighboring buckets and cause debunching losses that a Gaussian ODE model does not reproduce accurately, whereas a particle-based 6D tracking model does [1009.1562]. In the LHC at collision energy, by contrast, radiation damping is strong enough that IBS is largely balanced by damping and debunching losses are negligible in the scenarios studied [1009.1562].

The role of non-Gaussian beam profiles is central in the LHC-focused SIRE study. There, IBS and synchrotron radiation are treated as mechanisms that reshape the full 3D distribution rather than merely changing rms sizes [1806.07317]. The luminosity of two head-on bunches is generalized from the Gaussian expression
$$
{\cal L}^G=\frac{N_1N_2N_b f_{\mathrm{rev}}}{4\pi\sigma_x^G\sigma_y^G}
$$
to a q-Gaussian form,
$$
{\cal L}^{qG}=\frac{N_1N_2N_b f_{\mathrm{rev}}}{4\pi\sigma_x^{qG}\sigma_y^{qG}\, {\cal I}_x^{qG}{\cal I}_y^{qG}},
$$
with the tail-weight parameter $q$ determining the correction factors [1806.07317]. For fixed rms beam size, a $10\%$ change in tail weight can yield about a $5\%$ change in luminosity estimate [1806.07317]. This is important because IBS alters the population of the core and tails, especially in the horizontal and longitudinal planes.

Space charge can strongly amplify IBS effects. In LEIR and SPS, stand-alone IBS or stand-alone space-charge simulations are reported to be insufficient to explain measured degradation; the interplay of both is needed [2310.03504]. Near resonances, space charge broadens the tune footprint and IBS enhances diffusion into resonance-driven loss channels [2310.03504]. A similar conclusion is reached for the CLIC damping rings, where IBS, space charge, and synchrotron radiation compete near an excited skew resonance $3Q_y=31$ [2308.02196]. There, synchrotron radiation damping is reported to mitigate the combined IBS+SC degradation, reducing residual vertical emittance growth to about $5\%$ maximum and losses to below $1\%$ in the resonance-sensitive cases [2308.02196].

A plausible implication is that IBS should not be regarded as a single growth-rate correction superposed on otherwise independent dynamics. In many contemporary machines it acts as part of a coupled collective-effects system whose behavior depends on damping, optics, resonance structure, and distribution shape.

## 5. Machine-dependent manifestations

The operational meaning of IBS differs markedly across accelerator classes.

In hadron and ion colliders, IBS is commonly associated with luminosity degradation through emittance growth and, in some cases, debunching. For RHIC and the LHC, one study treats IBS as one of the key collective mechanisms driving the time evolution of luminosity and bunch intensities [1009.1562]. In the HE-LHC scenario, however, the initial SR damping time of about $2$ h is much shorter than the initial IBS growth time of about $70$–$80$ h, so IBS does not prevent emittance shrinkage and only becomes important as the beam approaches equilibrium [1108.1644].

In flat-beam ion operation, IBS becomes strongly asymmetric between planes. RHIC measurements with gold beams of transverse emittance ratio about $11{:}1$ show that IBS growth can be modeled successfully with the Lebedev–Nagaitsev formalism both without coupling and with weak coupling [2606.13745]. At $31$ GeV/nucleon, the measured average growth times over a five-minute observation period are effectively no growth horizontally, $25.1$ min vertically, and effectively no growth in bunch length; the model gives $-167$ h, $25.7$ min, and $31.6$ h, respectively [2606.13745]. At $100$ GeV/nucleon, the measured horizontal, vertical, and bunch-length growth times are $4.9$ h, $11.1$ h, and $4.8$ h, compared with modeled values of $4.3$ h, $13.2$ h, and $4.2$ h [2606.13745]. These results are directly tied to EIC design, where the target hadron-beam emittance ratio is about $10{:}1$ [2606.13745].

In low-emittance electron and positron storage rings, IBS is often an equilibrium-setting mechanism because synchrotron-radiation damping acts on similar or shorter timescales. The CesrTA program provides two complementary examples. A dedicated IBS study reports good agreement between measurements and theory, provided a tail-cut procedure is applied [1308.0035]. In low-current minimum-emittance positron operation, the horizontal emittance increases from $3.8$ nm-rad at low current to $10.4$ nm-rad at $1.3\times10^{11}$ particles per bunch, a substantial IBS effect attributed to the ring’s large horizontal dispersion [1308.0035]. A separate CesrTA study on horizontal crabbing shows that once RF-cavity-induced $xz$ tilt is corrected or eliminated, the remaining RF-voltage dependence of the horizontal beam size is attributed entirely to IBS [1311.1763].

For diffraction-limited storage rings, IBS can be severe enough to require dedicated mitigation. In the HALF storage ring, with natural emittance around $86.3$ pm·rad and full-coupling zero-current horizontal emittance about $50.2$ pm·rad, IBS is reported to nearly double the equilibrium emittance at nominal bunch charge if no mitigation is used [2311.15454]. Bunch lengthening by factors of $3$ and $5$, together with damping wigglers, reduces the equilibrium emittance to $64.5$ pm·rad and $59.3$ pm·rad, respectively, for a bunch charge of $0.875$ nC [2311.15454].

In fourth-generation light sources, the balance between IBS, synchrotron radiation, quantum excitation, betatron coupling, and vertical dispersion becomes a design problem. A self-consistent ODE framework developed for BESSY III evolves projected emittances as
$$
\frac{d \varepsilon_x}{dt} = -\alpha^C (\varepsilon_x - \varepsilon_y) - 2\alpha_x^{SR}(\varepsilon_x - \varepsilon_{x,0}) + 2\alpha_x^{IBS}\varepsilon_x,
$$
$$
\frac{d \varepsilon_y}{dt} = -\alpha^C (\varepsilon_y - \varepsilon_x) - 2\alpha_y^{SR}(\varepsilon_y - \varepsilon_{y,0}) + 2\alpha_y^{IBS}\varepsilon_y,
$$
$$
\frac{d \varepsilon_z}{dt} = -2\alpha_z^{SR}(\varepsilon_z - \varepsilon_{z,0}) + 2\alpha_z^{IBS}\varepsilon_z.
$$
The BESSY III case study finds that coupling can significantly reduce horizontal emittance at large vertical-emittance ratios, but vertical excitation is likely preferable for a modest target vertical emittance of about $10$ pm·rad because it avoids optics distortion and tune-space constraints [2604.01892].

In RF photoinjectors and XFEL injectors, IBS manifests differently again. The SwissFEL study identifies IBS as a previously neglected source of slice energy spread (SES) growth in dense low-energy electron bunches [2510.02058]. At the SwissFEL baseline charge, the highest measured SES reaches roughly $5.8$–$6$ keV at $192$ pC, while conventional space-charge-only modeling underestimates this by about an order of magnitude [2510.02058]. The paper concludes that 5D brightness is largely conserved but 6D brightness decreases noticeably because IBS increases SES [2510.02058].

## 6. Coupling, dispersion, and present directions

A recurring theme in recent IBS research is that transverse coupling and dispersion cannot always be treated as secondary corrections. For arbitrary $x$–$y$ coupling, an analytical method based on the Landau collision integral and the extended Mais–Ripken parametrization derives closed-form average growth rates for the transverse eigen-emittances and momentum spread [1812.09275]. In the coupled formalism, the growth rates depend on the eigen-emittance geometry rather than uncoupled $x$ and $y$ separately, and the bunched-beam case is obtained from the coasting-beam formulae by replacing the circumference factor with $2\sqrt{2\pi}\sigma_s$ and halving the longitudinal kinetic growth rate [1812.09275].

This eigenmode viewpoint reappears in the RHIC flat-beam work, where global linear coupling is described by the coefficient
$$
C^{-} = \frac{1}{2\pi} \oint \sqrt{\beta_x \beta_y \left[ k_{1s} + k_s \left(\frac{\alpha_x}{\beta_x} - \frac{\alpha_y}{\beta_y}\right) - i k_s\left(\frac{1}{\beta_x}+\frac{1}{\beta_y}\right) \right] e^{i(\Psi_x-\Psi_y)} dl },
$$
and EIC operation is said to require $|C^-|$ below about $0.002$ to preserve an $11{:}1$ emittance ratio [2606.13745]. In the BESSY III framework, coupling also redistributes damping partitions, changing the synchrotron-radiation damping rates and therefore the SR–IBS equilibrium [2604.01892].

Vertical dispersion can mitigate IBS by increasing vertical emittance and lowering phase-space density, but it also changes optics and radiation integrals. The BESSY III analysis compares three routes to generating vertical emittance—betatron coupling, vertical dispersion, and transverse-feedback-driven excitation—and concludes that each has different trade-offs in horizontal emittance reduction, dynamic aperture, and operational flexibility [2604.01892]. This suggests that IBS mitigation cannot be discussed independently of the mechanism used to produce the target transverse aspect ratio.

Two misconceptions are explicitly challenged by the recent literature. The first is that IBS is adequately described by Gaussian rms growth rates in all situations. The LHC non-Gaussian study and the RHIC heavy-ion tracking study both show that distribution shape and RF-bucket dynamics can be decisive [1806.07317] [1009.1562]. The second is that IBS is merely a binary-collision process with a weak logarithmic sensitivity to long-range interactions. The collective-field treatment argues that long-range self-field correlations can contribute a finite offset beyond the usual Coulomb-log term and may become especially relevant in beams with very small energy spread [2504.01867].

Taken together, these developments place IBS at the intersection of kinetic theory, nonlinear beam dynamics, and accelerator optimization. In conventional proton and ion machines it remains a principal source of emittance growth and, in some regimes, debunching loss. In electron storage rings it often defines a self-consistent equilibrium with synchrotron radiation. In photoinjectors it can degrade longitudinal phase-space quality without strongly affecting projected transverse metrics. The common thread is that IBS is fundamentally a density-driven internal Coulomb process whose practical consequences depend on the balance of optics, damping, collective interactions, and the full phase-space distribution [2602.11952] [2510.02058].

Source: https://www.emergentmind.com/topics/intrabeam-scattering-ibs