Mean ionizing photon energy of a power-law spectrum (source code)

= Mean ionizing photon energy of a power-law spectrum
{title2=$\langle E\rangle=\frac{s}{s-1}h\nu_0\quad(s>1)$}

Suppose an emitted <spectral luminosity> is $L_\nu=A\nu^{-s}$ for every $\nu\ge\nu_0$, where $A>0$ and $h\nu_0$ is an ionization threshold. The total emitted ionizing <luminosity> and <photon> rate use different weights:
$$
L_{\rm ion}=\int_{\nu_0}^\infty L_\nu\,d\nu
=\frac{A\nu_0^{1-s}}{s-1},\qquad s>1,
$$
$$
\dot N_{\rm ion}=\int_{\nu_0}^\infty\frac{L_\nu}{h\nu}\,d\nu
=\frac{A\nu_0^{-s}}{hs},\qquad s>0.
$$
Thus a finite emitted mean <photon energy> for this unbounded power law requires the stronger condition $s>1$ and equals
$$
\langle E\rangle_{\rm em}=\frac{L_{\rm ion}}{\dot N_{\rm ion}}=\frac{s}{s-1}h\nu_0.
$$
For $s=2$, the mean is twice the threshold <energy>. As $s\downarrow1$, the high-<frequency> tail makes the mean diverge even though the <photon> count stays finite. For $0<s\le1$, substituting into the finite-mean formula would therefore be invalid; for $s\le0$ even the unbounded photon-rate integral diverges.

A finite high-<frequency> cutoff changes both integrals and removes these upper-tail divergences. The mean <energy> deposited in a nebula also need not equal the emitted mean: absorption weights the spectrum with its <photoionization cross section>, and <energy> above threshold can be distributed among heat, excitation and secondary ionizations. The excess $h\nu-h\nu_0$, rather than the whole <photon> <energy>, is the initial kinetic <energy> of the photoelectron in the simplest single-ionization event.