9.4 Plasma bremsstrahlung 343
b
max
∼ λ
D
, the Debye length. However as we have seen already, bremsstrahlung
emissivity is not especially sensitive to the choices made for the impact parameter
cut-off.
For frequencies close to ω
p
it is no longer correct to neglect correlations between
electrons. Dawson and Oberman (1962) showed that the correction to the Gaunt
factor in the region ω ω
p
due to Langmuir wave generation was insignificant
for a plasma in thermal equilibrium. However for non-thermal plasmas, emission
in the neighbourhood of the plasma frequency may be many orders greater than
thermal levels. A brief account of one aspect of radiation by Langmuir waves is
given in Section 11.6.
9.4.6 Bremsstrahlung as plasma diagnostic
Bremsstrahlung emissivity through its dependence on electron temperature, plasma
density and atomic number clearly has potential as a plasma diagnostic. In the first
place the exponential dependence in (9.54) means that for
¯
hω ≥ k
B
T
e
the slope
of a log-linear plot of the bremsstrahlung emissivity provides a direct measure
of T
e
. Next, the strong dependence of the emission on the atomic number of
the plasma ions in principle allows the impurity content in a hydrogen plasma
to be determined. Moreover, if the plasma electron temperature is known in-
dependently, the level of bremsstrahlung could be used to estimate the plasma
density.
In practice the picture is less clear. Even for thermal plasmas, for which
bremsstrahlung losses do not result in significant modification of the distribution
function, unfolding the electron temperature from the bremsstrahlung spectrum is
not as straightforward as might first appear. Limited spectral resolution may result
in the true slope being masked by recombination edge effects or suffering distortion
from discrete lines in the spectrum. Bremsstrahlung from a tokamak plasma with
a modest content of high Z impurities such as nickel and molybdenum, will be
affected by contributions from these impurities.
Moreover, the assumption of a Maxwellian or near-Maxwellian electron dis-
tribution may not be justified. Since the temperature is deduced from the X-
ray spectrum in the region
¯
hω/k
B
T
e
> 1, any non-Maxwellian component will
lead to errors in the measurements. Non-Maxwellian electron distributions in
both space and laboratory plasmas are commonplace. It may happen for exam-
ple that a Maxwellian distribution that describes the bulk electrons is modified
by a high-energy tail of suprathermal electrons. Even though the population of
suprathermals is only a fraction of that of the bulk electrons, they may nevertheless
exercise an influence on the overall electron dynamics disproportionate to their
numbers.