The CMBR fluctuations from HI perturbations prior to reionization

Short version: this paper patches two omissions in Loeb & Zaldarriaga's 21 cm absorption proposal, and both patches are real. The gas-temperature response g(z) and the peculiar-velocity term in the optical depth are derived here rather than quoted, and the second one matters — it produces a Kaiser-like (1+mu²) that raises sqrt(Cl) by more than 50% at the redshifts of interest. That is the sentence to remember from the abstract [abstract:s5], and it holds up in Eq. 14 and Fig. 4. What is good: the chain from Eq. 1 through Eq. 17 is clean linear perturbation theory, the collision rates are properly cited to Allison & Dalgarno, and the paper states its assumptions instead of hiding them. The g(z) discussion is the more interesting of the two additions — the spin-temperature response changes sign around z~90, and dropping gas-temperature fluctuations flips the sign of the effect at z>100. That is a qualitative correction to the earlier treatment, not a percentage tweak. Soft spots, in order of how much they matter. The Gaussian cutoff at k=14 h/Mpc is hand-chosen and never varied; it sets the absolute mK amplitudes in Fig. 4, so treat those numbers as indicative. The 50% velocity enhancement is a ratio at fixed cutoff and survives that choice. Second, the growth assumption Delta_H proportional to a is stated explicitly and is fine — at z=30 the universe is matter-dominated to a part in 10⁴, so I would not lose sleep over it, contrary to what a first read might suggest. Third, Eq. 13 has a sign error: v = -i H a k Delta/k² should carry a plus sign, since Eq. 14's (1+mu²) is the correct growing-mode answer. It is a typo, not a flaw, but it will confuse a student working through the algebra. Who it is for: anyone building on the dark-ages 21 cm absorption signal, and anyone who wants the correct linear response function before quoting Loeb & Zaldarriaga's numbers. It is a calibration paper, not a measurement, and its value will track whether the signal is ever detected. Recommendation: send it out. It is short, correct where it counts, and the two additions are worth having in the literature.

The paper's central claim, that peculiar velocities enhance the brightness-temperature power spectrum by more than 50 percent, rests on the standard linear redshift-space distortion factor (1 + mu^2) in Eq. 14, which follows from Eq. 11 once the growing-mode velocity is inserted. The reader's concern about the linear growth assumption for g(z) and s(z) is valid in principle but weak in practice: at 30 < z < 200 the universe is matter-dominated (Omega_m(z) is within 10^-4 of unity at z = 30 for the stated cosmology), so Delta_H proportional to a(z) is an excellent approximation and the growth-rate term in Eq. 5 is well determined. The one concrete inconsistency I find is between Eq. 13, which states v(k,z) = -i H(z) a(z) k Delta(k,z)/k^2, and Eq. 14, which uses (1 + mu^2). Combining Eq. 11 with the sign in Eq. 13 would give (1 - mu^2), not (1 + mu^2). The correct growing-mode relation is v = +i H a Delta k/k^2, so Eq. 14 is right and Eq. 13 has a sign typo. Because the final result, Eq. 16, and Fig. 4 all use the correct (1 + mu^2), the conclusions are unaffected; this is a presentation error, not a load-bearing flaw. The reader's CONDITIONAL verdict remains appropriate given the acknowledged limitations on small scales and the arbitrary cutoff, but the specific concern about linear growth is weaker than the reader suggests.

The paper is a careful linear-theory extension of Loeb & Zaldarriaga (2003). The two new ingredients, gas-temperature fluctuations and peculiar-velocity redshift-space distortion, are derived from standard perturbation theory, and the algebra from Eq. 1 through Eq. 17 is internally consistent once the sign of Eq. 13 is corrected. The headline claim, that velocities raise the predicted power spectrum amplitude by more than 50 percent, is a ratio of two computed quantities and is therefore insensitive to the arbitrary Gaussian cutoff at k = 14 h Mpc^-1; the cutoff affects only the absolute normalization, which the paper does not claim to fix precisely. The reader's weakest-assumption concern about linear growth is legitimate but not strongly load-bearing: at the redshifts of interest the growth is matter-dominated and Delta proportional to a is an excellent approximation, so g(z) and s(z) are well determined. The more concrete issue is the sign inconsistency between Eq. 13 and Eq. 14, which is a typo rather than a substantive error because the final result uses the correct (1 + mu^2). The paper is honest about its assumptions, cites the collision rates properly, and does not exhibit circularity. The reader's CONDITIONAL verdict is reasonable given the acknowledged small-scale nonlinearity and the arbitrary cutoff, but the specific concern about the growth history is weaker than the reader suggests, and no adjustment to the verdict is warranted on the basis of the sign typo alone.

The two effects are physically real and were omitted by Loeb & Zaldarriaga (2003); the paper's contribution is to quantify them for the pre-reionization absorption signal. The derivation is transparent and self-contained given the cited collision rates (Allison & Dalgarno 1969) and the stated cosmology, and the separation of the optical-depth and spin-temperature responses in Fig. 3 is a useful diagnostic. The results are falsifiable predictions for future 21-cm experiments, and the paper is candid about the arbitrary k-cutoff and the Jeans-scale restriction. The main strengths are the clean isolation of the RSD term and the explicit demonstration that gas-temperature fluctuations matter at the tens-of-percent level.

Before the first stars form, hydrogen gas fills the universe and, over a stretch of cosmic time corresponding roughly to redshifts $30 < z < 200$, it absorbs a thin slice of the cosmic microwave background at the 21-cm wavelength. This paper asks what the pattern of that absorption would actually look like: how a small ripple in hydrogen density becomes a ripple in the observed brightness temperature, and therefore what a radio telescope would really be measuring. The authors work through both channels by which density ripples imprint themselves, the spin temperature that sets how strongly the gas absorbs and the optical depth that sets how much gas a line of sight crosses, and they add two effects earlier treatments left out. One is that density ripples also heat or cool the gas, which feeds back into the spin temperature. The other is that the gas is moving, and its bulk motion slides absorption features up and down in frequency exactly as peculiar-velocity distortions do in galaxy surveys. That second effect alone changes the predicted angular power spectrum by more than 50 percent, which matters because the entire promise of this signal is a precise map of the primordial power spectrum.

What carries the argument. The argument is carried by two response functions, $g(z) = partial Delta_g / partial Delta_H$ and $s(z) = partial Delta_s / partial Delta_H$, which record how much a fractional density perturbation turns into a fractional gas-temperature and spin-temperature perturbation. They are obtained by linearising the gas-temperature evolution equation, which balances adiabatic cooling against heat pumped in by residual electrons scattering CMB photons, and the spin-temperature evolution equation, which balances collisional spin flipping against radiative transitions. Both are integrated forward assuming the density perturbation grows in proportion to the scale factor. The second piece of machinery is the $(1+mu^2)$ angular factor inherited from the velocity term, which in the spherical-harmonic expansion replaces the ordinary spherical Bessel function $j_ell(kr)$ with a fixed combination of $j_{ell-2}$, $j_ell$ and $j_{ell+2}$ given in the paper as $J_ell(kr)$.

Core claim. The paper's central claim is that the relation between hydrogen density perturbations and 21-cm brightness-temperature fluctuations is not the one used in the earlier proposal of Loeb & Zaldarriaga (2003): two additional terms change it materially. First, density perturbations perturb the gas temperature, not only the collision rate, and because the spin temperature is coupled to the gas temperature this propagates into the brightness temperature; the authors find the spin-temperature response changes sign near $z = 90$ and that ignoring the gas-temperature channel makes $z<100$ power about 10 percent too high and $z>100$ power about 30 percent too low. Second, peculiar velocities shift the absorption features in frequency, so the optical depth picks up a term proportional to the line-of-sight derivative of the velocity, the same structure as linear redshift-space distortion. In Fourier space this turns the density response into $(1 - T_0/T_s)(1+mu^2) + (T_0/T_s) s$, with $mu$ the cosine between the wave vector and the line of sight, and the authors report that including it raises $sqrt{C_ell}$ by more than 50 percent. The combined response peaks near $z approx 55$, lower than the earlier estimate, and the angular power spectrum of the signal peaks at $z approx 50$.

The premise it rests on. The two response functions are integrated assuming the hydrogen density perturbation grows in proportion to the scale factor, and the authors themselves note that a perturbation which did not evolve in time would produce no gas-temperature fluctuation at all, so any departure from linear growth between $z=200$ and $z=30$ shifts the derived responses and the percentages quoted with them.

Any attempt to invert a measured 21-cm absorption angular power spectrum into a primordial power spectrum must include the peculiar-velocity term, or it will misestimate amplitude by of order 50 percent.

The signal is predicted to be anisotropic in a specific way, with power depending on the angle between the line of sight and the wave vector through $(1+mu^2)$, so the anisotropy itself becomes a signature rather than a nuisance.

The redshift of peak response moves down to $z approx 55$, which sets where a survey should concentrate its frequency coverage.

Gas-temperature fluctuations must be tracked separately from collision-rate fluctuations, since they push the spin-temperature response in opposite directions above and below $z approx 90$.

The clean mapping onto the dark matter power spectrum is limited to scales larger than the Jeans scale, about $k_J approx 500\,h\,\mathrm{Mpc}^{-1}$ over the redshifts of interest.

The $(1+mu^2)$ anisotropy gives a way to separate the velocity contribution from the density contribution empirically: a survey comparing angular power along and across the line of sight could confirm or rule out the 50 percent enhancement without needing an absolute calibration of the signal.

Because the predicted enhancement depends on the growth history assumed for the density perturbation, the same measurement could in principle be turned around and used as a probe of growth between $z=200$ and $z=30$, a period otherwise poorly constrained.

The quoted percentages depend on the Gaussian suppression the authors impose beyond an arbitrarily chosen $k = 14\,h\,\mathrm{Mpc}^{-1}$, so a reader should treat the 50 percent figure as tied to that choice rather than as a cutoff-independent number.

The authors note that near and below the Jeans scale the density field is only mildly nonlinear and the Zel'dovich approximation may describe it better; that regime is where the linear treatment used here would be expected to fail first.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review