Skip to main content
Sandbox Physics
Fast radio bursts and the cosmic baryon census

Adding up electrons across an expanding universe

Sandbox Physics ·

The average cosmic dispersion measure out to redshift zz is the mean free-electron density today times the Hubble distance times one integral over redshift. With the Planck 2015 baryon density and 85 percent of the baryons in diffuse ionised gas, it is 84 pc cm−384\,\mathrm{pc\,cm^{-3}} at z=0.1z=0.1, 449449 at z=0.5z=0.5 and 918918 at z=1z=1. This chapter builds that curve from the baryon density, checks it against two universes where the integral has a closed form, and shows what a measurement of the curve tells you: the product of the diffuse fraction and Ωbh70\Omega_{\mathrm b}h_{70}.

The previous chapter, How free electrons delay a radio burst, defined the dispersion measure. Here you also need the Hubble parameter H(z)H(z) and the comoving distance.

How many free electrons there are per cubic metre

Today's mean baryon density is ρb,0=Ωbρc,0\rho_{\mathrm b,0}=\Omega_{\mathrm b}\rho_{\mathrm c,0}, with ρc,0=3H02/(8πG)\rho_{\mathrm c,0}=3H_0^2/(8\pi G). Ionised hydrogen gives one electron per proton mass. Helium, with mass fraction YY, gives two electrons per four proton masses. The free-electron density of the diffuse gas is therefore

nˉe,0=fd ρb,0mp(1−Y2),\bar n_{\mathrm e,0}=f_{\mathrm d}\,\frac{\rho_{\mathrm b,0}}{m_{\mathrm p}}\left(1-\frac{Y}{2}\right),

where fdf_{\mathrm d} is the share of baryons in diffuse ionised gas rather than in stars or cold neutral gas. Macquart et al., 2020, eq. 2

Worked example: the electron density of the universe today

With H0=67.74 km s−1 Mpc−1H_0=67.74\,\mathrm{km\,s^{-1}\,Mpc^{-1}}, ρc,0=8.619×10−27 kg m−3\rho_{\mathrm c,0}=8.619\times10^{-27}\,\mathrm{kg\,m^{-3}}. Multiplying by Ωb=0.0486\Omega_{\mathrm b}=0.0486 gives ρb,0=4.19×10−28 kg m−3\rho_{\mathrm b,0}=4.19\times10^{-28}\,\mathrm{kg\,m^{-3}}, a quarter of a proton per cubic metre. With Y=0.25Y=0.25 the factor 1−Y/21-Y/2 is 0.8750.875, so nˉe,0=0.219 m−3=2.19×10−7 cm−3\bar n_{\mathrm e,0}=0.219\,\mathrm{m^{-3}}=2.19\times10^{-7}\,\mathrm{cm^{-3}} if every baryon were diffuse. With fd=0.85f_{\mathrm d}=0.85 it is 1.86×10−7 cm−31.86\times10^{-7}\,\mathrm{cm^{-3}}.

Each slice of the path, carried to today

Take a slice of the path at redshift zz with proper length dl\mathrm dl. Its electron density is higher than today by (1+z)3(1+z)^3, since the same electrons filled a smaller universe. Its contribution to the observed dispersion measure carries the factor 1/(1+z)1/(1+z) derived in the previous chapter. The proper length of the slice is dl=c dz/[(1+z)H(z)]\mathrm dl=c\,\mathrm dz/[(1+z)H(z)]. Putting the three together,

d⟨DM⟩=nˉe,0(1+z)31+z c dz(1+z)H(z)=nˉe,0 cH0 (1+z) dzE(z),\mathrm d\langle\mathrm{DM}\rangle=\frac{\bar n_{\mathrm e,0}(1+z)^3}{1+z}\,\frac{c\,\mathrm dz}{(1+z)H(z)}=\frac{\bar n_{\mathrm e,0}\,c}{H_0}\,\frac{(1+z)\,\mathrm dz}{E(z)},

with E(z)=Ωm(1+z)3+ΩΛE(z)=\sqrt{\Omega_{\mathrm m}(1+z)^3+\Omega_\Lambda} for a flat universe of matter and a cosmological constant. Integrating,

⟨DMcosmic⟩(z)=K∫0z1+z′E(z′) dz′,K=nˉe,0 cH0.\langle\mathrm{DM}_{\mathrm{cosmic}}\rangle(z)=K\int_0^z\frac{1+z'}{E(z')}\,\mathrm dz',\qquad K=\frac{\bar n_{\mathrm e,0}\,c}{H_0}.

For Planck 2015 and fd=0.85f_{\mathrm d}=0.85, the Hubble distance c/H0c/H_0 is 4426 Mpc4426\,\mathrm{Mpc} and K=824.3 pc cm−3K=824.3\,\mathrm{pc\,cm^{-3}}. This is the mean relation drawn as the solid line in Fig. 2 of Macquart et al., 2020, which keeps a slowly varying fd(z)f_{\mathrm d}(z) where this chapter holds it constant.

Table 3.1 Mean cosmic dispersion and comoving distance for Planck 2015 with fd=0.85f_{\mathrm d}=0.85.
RedshiftComoving distance (Mpc)Mean cosmic DM (pc cm−3)(\mathrm{pc\,cm^{-3}})
0.1432.184.48
0.251040.1217.4
0.51945.6448.8
0.752725.4684.4
1.03396.0918.2

Two universes that check the integral

The integral has a closed form in two limits. In a universe of matter only, E=(1+z)3/2E=(1+z)^{3/2} and the integrand is (1+z)−1/2(1+z)^{-1/2}. In a universe of vacuum energy only, E=1E=1 and the integrand is 1+z1+z:

∫0zdz′1+z′=2(1+z−1),∫0z(1+z′) dz′=z+z22.\int_0^z\frac{\mathrm dz'}{\sqrt{1+z'}}=2\left(\sqrt{1+z}-1\right),\qquad \int_0^z(1+z')\,\mathrm dz'=z+\frac{z^2}{2}.

At z=0.5z=0.5 these give 0.44950.4495 and 0.62500.6250; at z=1z=1, 0.82840.8284 and 1.50001.5000. Our universe, with Ωm=0.309\Omega_{\mathrm m}=0.309, gives 0.54450.5445 and 1.11391.1139, between the two as it must. Matter slows expansion, so a matter-only universe reaches a given redshift over a shorter path and collects fewer electrons.

At low redshift, expanding the integrand gives ⟨DM⟩≈Kz [1+(1−32Ωm) z/2]=Kz (1+0.268 z)\langle\mathrm{DM}\rangle\approx Kz\,[1+(1-\tfrac32\Omega_{\mathrm m})\,z/2]=Kz\,(1+0.268\,z). At z=0.1z=0.1 this gives 84.684.6, within 0.2 %0.2\,\% of the full integral. The saved materials evaluate the integral by Simpson's rule and by a separately written Gauss–Legendre rule; the two agree to one part in 101510^{15}.

What a measured curve tells you

Since nˉe,0∝ΩbH02\bar n_{\mathrm e,0}\propto\Omega_{\mathrm b}H_0^2, the constant KK is proportional to fd ΩbH0f_{\mathrm d}\,\Omega_{\mathrm b}H_0. H0H_0 does not appear in E(z)E(z), so localised bursts measure the product fd Ωbh70f_{\mathrm d}\,\Omega_{\mathrm b}h_{70}, with h70=H0/(70 km s−1 Mpc−1)h_{70}=H_0/(70\,\mathrm{km\,s^{-1}\,Mpc^{-1}}). For Planck 2015, Ωbh70=0.04703\Omega_{\mathrm b}h_{70}=0.04703. Fixing fdf_{\mathrm d} from an independent census of stars and cold gas turns the curve into a measurement of the baryon density. Macquart et al., 2020, main text after eq. 2

Two shorthand forms circulate. Petroff et al., 2022, Section 4.1 quotes about 1000 z1000\,z for the intergalactic part with all baryons in the intergalactic medium. Our low-redshift slope with fd=1f_{\mathrm d}=1 is 969.8969.8, 3 percent lower. Connor et al., 2025, eq. 2 quotes 1085 z fd (Ωbh70/0.04703)1085\,z\,f_{\mathrm d}\,(\Omega_{\mathrm b}h_{70}/0.04703) for z≲1z\lesssim1. That is a chord, not a tangent: at z=1z=1 with fd=0.85f_{\mathrm d}=0.85 it gives 922.3922.3, within 0.44 %0.44\,\% of the integral, but at z=0.1z=0.1 it gives 92.292.2, 9 percent high.

Worked example: one real burst against the curve

FRB 190711 sits at z=0.522z=0.522 with DM=593.1 pc cm−3\mathrm{DM}=593.1\,\mathrm{pc\,cm^{-3}}. Removing 80 pc cm−380\,\mathrm{pc\,cm^{-3}} for the Milky Way leaves 513.1513.1. The mean cosmic value at that redshift is 469.5469.5. The remaining 43.643.6 is what the host must supply, which corresponds to 43.6×1.522=66.4 pc cm−343.6\times1.522=66.4\,\mathrm{pc\,cm^{-3}} in the host's own frame. The burst agrees with the curve if its host is modest and its sightline is typical.

Lab task

Lower the diffuse fraction in the Lab from 0.850.85 to about 0.40.4, the Half the baryons missing preset. The mean curve in the dispersion–redshift chart drops below the real ASKAP bursts. Return to 0.850.85 and the bursts sit on or above it, with the host term filling the gap.

Exercises

  1. Show that 1−Y/21-Y/2 is the number of free electrons per proton mass in fully ionised gas of hydrogen and helium.
  2. By how much does the mean cosmic dispersion at z=0.3z=0.3 fall if fdf_{\mathrm d} drops from 0.850.85 to 0.60.6?
  3. Use the low-redshift form to estimate the mean at z=0.25z=0.25 and compare with Table 3.1.
  4. A colleague uses 1085 z1085\,z with fd=1f_{\mathrm d}=1 to predict a burst at z=0.1z=0.1. How large is the error compared with the integral?
Answers
  1. Per unit mass, hydrogen contributes (1−Y)/mp(1-Y)/m_{\mathrm p} electrons and helium 2×Y/(4mp)=Y/(2mp)2\times Y/(4m_{\mathrm p})=Y/(2m_{\mathrm p}). The sum is (1−Y/2)/mp(1-Y/2)/m_{\mathrm p}.
  2. The mean is proportional to fdf_{\mathrm d}: 263.0×0.6/0.85=185.6 pc cm−3263.0\times0.6/0.85=185.6\,\mathrm{pc\,cm^{-3}}, down by 77.477.4.
  3. 824.3×0.25×(1+0.268×0.25)=219.9824.3\times0.25\times(1+0.268\times0.25)=219.9, against 217.4217.4 from the full integral: 1.1 %1.1\,\% high, because the next term in zz is negative.
  4. The integral with fd=1f_{\mathrm d}=1 at z=0.1z=0.1 is 84.48/0.85=99.484.48/0.85=99.4. The shorthand gives 108.5108.5, 9 %9\,\% too high. Over a large sample this would bias the inferred host contribution low.

Sources

  1. Macquart et al., 2020, eq. 2
  2. Petroff et al., 2022, Section 4.1
  3. Connor et al., 2025, eq. 2

Keep reading

What this page links to directly

See this series on the content map →

Fast radio bursts and the cosmic baryon census · Reading path

From one delayed pulse to the missing baryons

  1. ArticleHow do fast radio bursts count the ordinary matter between galaxies?
  2. ArticleHow free electrons delay a radio burst
  3. You are hereAdding up electrons across an expanding universe
  4. ArticleWhy every sightline gives a different dispersion
  5. ArticleRecompute the localized-burst baryon census
  6. ArticleReading the FRB baryon census papers
Something broken, confusing, or missing on this page?Tell us
What kind of feedback?

Sent with this page’s address, its Lab ID, the site language and your device type. Please leave out personal details. Feedback is kept for up to 12 months.Privacy policy