---
title: "How a clump of plasma delays and bends a radio burst"
description: "Derive the dispersive delay of a free-electron column, the bending it causes and the lens strength that decides whether one burst splits into copies."
canonical_url: "https://sandboxphysics.com/stories/fast-radio-burst-echoes/plasma-delay-and-bending/"
language: "en"
page_type: "story"
alternate_languages:
  zh-CN: "https://sandboxphysics.com/zh/stories/fast-radio-burst-echoes/plasma-delay-and-bending/"
last_modified: "2026-11-14"
---

[Fast radio bursts: repeats, echoes and plasma lenses](https://sandboxphysics.com/stories/fast-radio-burst-echoes/)

# How a clump of plasma delays and bends a radio burst

Sandbox Physics · 2026-11-14

Free electrons make radio waves travel slower than light by an amount that grows as the square of the wavelength. A column of electrons therefore delays a burst, and a column that changes across the sky also bends it. For a Gaussian clump, one dimensionless number, the lens strength $\alpha$, collects the electron column, the clump size, the distance and the frequency. When $\alpha$ is larger than about 2.24, one burst can reach us along three paths.

Before this chapter, read [is a repeating radio flash the source repeating, or its path?](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repeat-or-echo/) You need the small-angle approximation, the derivative of $e^{-x^2}$ and the idea that a wave's phase advances by $2\pi\nu t$ in a time $t$.

## The refractive index of a plasma

In a cold, unmagnetized plasma of electron density $n_e$, waves below the plasma frequency $\nu_p$ cannot propagate. Above it the refractive index is

$$
n=\sqrt{1-\frac{\nu_p^2}{\nu^2}}\approx1-\frac{\nu_p^2}{2\nu^2},\qquad \nu_p^2=\frac{n_e r_e c^2}{\pi},
$$

where $r_e=2.818\times10^{-15}\,\mathrm{m}$ is the classical electron radius. For interstellar gas $\nu_p$ is of order a kilohertz, so at radio frequencies $\nu_p\ll\nu$ and the first-order form is enough.

A burst travels at the group velocity $v_g=cn$. Integrating $1/v_g-1/c\approx\nu_p^2/(2c\nu^2)$ along the path gives the extra arrival time

$$
t_{\rm DM}=\frac{c\,r_e}{2\pi\nu^2}\int n_e\,\mathrm dl=\frac{c\,r_e\,\mathrm{DM}}{2\pi\nu^2}=4.149\,\mathrm{ms}\;\frac{\mathrm{DM}}{\nu_{\rm GHz}^2},
$$

with the dispersion measure $\mathrm{DM}=\int n_e\,\mathrm dl$ in $\mathrm{pc\,cm^{-3}}$. The coefficient follows from SI constants: our independent check gives 4.1488 ms. Cordes and Chatterjee's review quotes 4.15 ms as the leading term of the arrival time, with smaller terms in $\nu^{-3}$ and $\nu^{-4}$ that we neglect. [Fast Radio Bursts: An Extragalactic Enigma, Eq. 1](https://arxiv.org/abs/1906.05878v2)

## Phase runs ahead, so rays bend away

The phase velocity $c/n$ is larger than $c$. Over the same path the phase therefore runs ahead of vacuum by

$$
\Delta\phi=\frac{2\pi\nu}{c}\int(n-1)\,\mathrm dl=-\lambda\,r_e\,N_e,
$$

where $N_e$ is the electron column in $\mathrm{m^{-2}}$. A wavefront is a surface of constant phase. Where the column is larger the phase is further ahead, so the wavefront tilts and the ray turns toward smaller columns. The deflection angle is

$$
\theta(x)=\frac{\lambda}{2\pi}\frac{\mathrm d\Delta\phi}{\mathrm dx}=-\frac{\lambda^2 r_e}{2\pi}\frac{\mathrm dN_e}{\mathrm dx}.
$$

A clump denser than its surroundings has $\mathrm dN_e/\mathrm dx<0$ on its far side, so rays are pushed outward: an overdense plasma clump is a diverging lens. Because $\theta\propto\lambda^2$, low frequencies bend more.

## A Gaussian clump and its lens equation

Follow Clegg, Fey and Lazio and take a one-dimensional clump with a Gaussian column, $N_e(x)=N_\ell\,e^{-x^2/a^2}$, of peak dispersion measure $\mathrm{DM}_\ell$ and width $a$. [The Gaussian plasma lens in astrophysics](https://arxiv.org/abs/astro-ph/9709249v1) Its deflection is

$$
\theta(x)=\frac{\lambda^2 r_e N_\ell}{\pi a^2}\,x\,e^{-x^2/a^2}.
$$

Put the lens a distance $D$ in front of the source, with the observer much farther away. A ray that leaves the source at transverse position $x_s$, crosses the lens plane at $x$ and then heads for the observer must have been deflected by $\theta(x)=(x_s-x)/D$. Measure lengths in units of $a$: $u=x/a$ for the crossing point and $u'=x_s/a$ for the source offset from the lens centre. Then

$$
u\left(1+\alpha\,e^{-u^2}\right)=u',\qquad \alpha=\frac{\lambda^2 r_e\,\mathrm{DM}_\ell\,D}{\pi a^2}=3430\,\frac{\mathrm{DM}_\ell\,D_{\rm kpc}}{(\nu_{\rm GHz}\,a_{\rm AU})^2}.
$$

This is Eq. 2 of Cordes and colleagues, with their Eq. 3 for $\alpha$; for a lens far from both ends, $D$ becomes $d_{\rm sl}d_{\rm lo}/d_{\rm so}$. [Lensing of fast radio bursts by plasma structures in host galaxies, Eqs. 1 to 3](https://arxiv.org/abs/1703.06580v1) Our independent calculation of the coefficient from SI constants gives 3429.8.

Each solution $u$ is a separate path, an image of the source. The left side is a straight line bent by a bump. When the bump is weak, every $u'$ has one solution. When it is strong, the curve turns back on itself and some offsets have three. The next chapter shows the turn-back needs $\alpha>e^{3/2}/2\approx2.24$.

## Two delays for each path

Each path is late for two reasons. It is longer than the straight line, by $(x-x_s)^2/2D$ in the small-angle limit, and it crosses part of the electron column. In units of $a$,

$$
t(u)=t_{g0}\,(u-u')^2+t_{\rm DM0}\,e^{-u^2},\qquad t_{g0}=\frac{a^2}{2cD},\qquad t_{\rm DM0}=4.149\,\mathrm{ms}\;\frac{\mathrm{DM}_\ell}{\nu_{\rm GHz}^2}.
$$

These are Eqs. 12 to 14 of the 2017 paper. A useful identity ties the scales together: $t_{g0}\,\alpha=t_{\rm DM0}$. For $a=1\,\mathrm{AU}$ and $D=1\,\mathrm{kpc}$, $t_{g0}=1.21\,\mu\mathrm s$.

**Worked example.** Take the Lab default, $\mathrm{DM}_\ell=0.6\,\mathrm{pc\,cm^{-3}}$, $a=30\,\mathrm{AU}$, $D=1\,\mathrm{kpc}$, at $600\,\mathrm{MHz}$. Then $\alpha=3430\times0.6/(0.6\times30)^2=6.35$, $t_{\rm DM0}=4.149\times0.6/0.36=6.91\,\mathrm{ms}$ and $t_{g0}=1.21\,\mu\mathrm s\times900=1.09\,\mathrm{ms}$. Their ratio is 6.35, as the identity says. A source exactly behind the core, $u'=0$, has one image at $u=0$: it pays the full dispersive delay of 6.91 ms and no geometric delay.

## Lab tasks

1. In [the plasma-lens echo Lab](https://sandboxphysics.com/labs/frb-plasma-lens-echoes/), choose the preset that puts the source behind the lens. Predict the delay of the single copy from $t_{\rm DM0}$, then read it from the data row.
2. Halve the lens column with the first slider. Predict how $\alpha$ and $t_{\rm DM0}$ change, then check the secondary readouts.
3. Move the observing frequency from 600 to 400 MHz. Explain from the formula for $\alpha$ why the scene changes.

## Exercises

1. Compute $t_{\rm DM0}$ for $\mathrm{DM}_\ell=0.6\,\mathrm{pc\,cm^{-3}}$ at 400 MHz and at 800 MHz.
2. Show that $t_{g0}\,\alpha=t_{\rm DM0}$ from the definitions.
3. A clump has $\mathrm{DM}_\ell=2\,\mathrm{pc\,cm^{-3}}$, $a=10\,\mathrm{AU}$ and sits $0.1\,\mathrm{kpc}$ from the source. Find $\alpha$, $t_{\rm DM0}$ and $t_{g0}$ at 1 GHz.
4. Where across the Gaussian clump is the deflection largest, and how large is it for the Lab default?

Answers

1. $t_{\rm DM0}=4.149\times0.6/\nu_{\rm GHz}^2$ gives $15.6\,\mathrm{ms}$ at 400 MHz and $3.89\,\mathrm{ms}$ at 800 MHz. Halving the frequency multiplies the delay by four.
2. $t_{g0}\,\alpha=\dfrac{a^2}{2cD}\cdot\dfrac{c^2 r_e\,\mathrm{DM}_\ell\,D}{\pi\nu^2a^2}=\dfrac{c\,r_e\,\mathrm{DM}_\ell}{2\pi\nu^2}=t_{\rm DM0}$, using $\lambda=c/\nu$.
3. $\alpha=3430\times2\times0.1/(1\times10)^2=6.86$, $t_{\rm DM0}=4.149\times2=8.30\,\mathrm{ms}$ and $t_{g0}=1.21\,\mu\mathrm s\times100/0.1=1.21\,\mathrm{ms}$. The lens is strong enough to make three images.
4. $\theta\propto u\,e^{-u^2}$ peaks where its derivative $(1-2u^2)e^{-u^2}$ vanishes, at $u=1/\sqrt2$. There $\theta_{\max}=\alpha\,(a/D)\,e^{-1/2}/\sqrt2=0.429\,\alpha\,a/D$. With $\alpha=6.35$, $a=30\,\mathrm{AU}$ and $1\,\mathrm{kpc}=2.06\times10^8\,\mathrm{AU}$, $\theta_{\max}=4.0\times10^{-7}\,\mathrm{rad}$, about 0.08 arcseconds.

Next: [images, gains and delays of a Gaussian plasma lens](https://sandboxphysics.com/stories/fast-radio-burst-echoes/gaussian-plasma-lens-images/).

## Try it in the Lab

[FRB echoes through a plasma lens](https://sandboxphysics.com/labs/frb-plasma-lens-echoes/): Put the source behind the lens core and read the delay of the single copy.

## Sources

1. [Fast Radio Bursts: An Extragalactic Enigma, Eq. 1](https://arxiv.org/abs/1906.05878v2)
2. [The Gaussian plasma lens in astrophysics](https://arxiv.org/abs/astro-ph/9709249v1)
3. [Lensing of fast radio bursts by plasma structures in host galaxies, Eqs. 1 to 3](https://arxiv.org/abs/1703.06580v1)

## Keep reading

Fast radio bursts: repeats, echoes and plasma lenses · Reading path

### From one burst to source-or-path tests

1. [Article **Is a repeating radio flash the source repeating, or its path?**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repeat-or-echo/)
2. [You are here **How a clump of plasma delays and bends a radio burst**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/plasma-delay-and-bending/)
3. [Article **Images, gains and delays of a Gaussian plasma lens**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/gaussian-plasma-lens-images/)
4. [Article **Telling source repetition from path effects with timing statistics**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repetition-statistics-source-or-path/)
5. [Article **Search one CHIME burst for a lens echo and test R3's repeats**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/fit-a-chime-burst-and-r3-repeats/)
6. [Article **How to read the CHIME paper on the 16-day FRB cycle**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/reading-the-chime-r3-periodicity-paper/)
