The size distribution of interstellar dust particles as determined from polarization: Infinite cylinders
S. H. Kim, P. G. Martin;
ApJ, 1994, 431, 783
ABSTRACT:To extract the size distribution of polarizing dust grains from the wavelength dependence of interstellar linear
polarization as objectively as possible, we have used the maximum-entropy method
(MEM), as in an earlier study of size distributions based on extinction (Kim,
Martin, & Hendry).
There are additional complications using polarization
data since polarization depends on shape and alignment.
In this first
investigation, we adopted infinite cylinders with perfect spinning alignment.
To
constrain a wide range of sizes it is necessary to use data from the infrared to the
far-ultraviolet.
Much of our analysis is based on bare silicate grains.
The modified
Serkowski law represents interstellar polarization quite well for the
wavelength range 0.3 to 2 micrometers using one parameter, lambda(sub max), the
wavelength at which the polarization is maximum.
Recent ultraviolet
polarimetric observations of eight stars of differing lambda(sub max) indicate
that extrapolation of the modified Serkowski curve into the ultraviolet
produces a reasonable approximation for larger lambda(sub max) (greater than
or approximately 0.55 micrometer), but for smaller lambda(sub max)
there is an excess of polarization observed.
Therefore, we have
investigated how the size distribution of the polarizing grains changes with
lambda(sub max) simply by fitting the modified Serkowski curve evaluated for
lambda(sub max) = 0.55, 0.615, and 0.68 micrometers.
But for HD 25443 (lambda(sub
max) = 0.49 micrometer) which shows super-Serkowski behavi or, and for HD
197770 (lambda(sub max) = 0.51 micrometers) which might exhibit a 2175 A
polarization bump, we combined the modified Serkowski curve in the infrared and
optical with the actual far-ultraviolet data.
The size distributions found
for silicates bear little resemblance to a power law.
Instead, when
expressed as contributions to the total mass, they peak roughly at 0.14
micrometer and are skewed, with the relative rate of decrease to larger and smaller
sizes depending on lambda(sub max).
For the particles larger than 0.1
micrometer, the size distribution does not change much with lambda(sub max)
because the infrared polarization is fairly invariant; on the other hand there
is a remarkable change for the smaller particles--the drop-off gets
faster as lambda(sub max) increases, explaining the decreasing ultraviolet
polarization.
In terms of the evolution of the size distribution, there is nothing to
signal the transition from Serkowski to super-Serkowski behavior; this
distinction is probably artificial.
We have explored size distributions and
polarization fitting using homogeneous grains of organic refractory material and
other more highly absorbing materials like iron, magnetite, and an
artificial one introduced by Chlewicki & Greenberg.
Among three organic
refractories studied, two give a fairly good fit to the average polarization curve;
but in detail, one of these gives excess polarization in the infrared and
the other predicts unobserved spectral structure in the
ultraviolet.
The third more processed org anic refractory, thought to be most
characteristic of interstellar grains, does not have features in the ultraviolet;
however it produces a very flat polarization curve in the far ultraviolet,
unlike that observed.
On the whole, these organic refractories appear less
satisfactory than silicates.
Iron and magnetite have features in the visible and
near-infrared which do not agree with the polarization data.
KEYWORDS: cosmic dust, infrared radiation, interstellar extinction, linear polarization, particle size distribution, polarized radiation, cosmology, maximum entropy method, refractivity, silicates
PERSOKEY:polarisation, dust, nir, ,
CODE: kim94