One of the most enigmatic features of the "second solar spectrum" (the linearly polarized spectrum that is due to coherent scattering processes on the Sun) has been the spectral structure across the Na I D1 5895.94 and D2 5889.97 Å lines. Although the general shape of the polarized profile with its remarkable signature of quantum-mechanical interference between the two scattering transitions was uncovered already in 1978 in observations at the National Solar Observatory (NSO), both at NSO/Sac Peak and NSO/Kitt Peak (Stenflo et al. 1980, 1983), it was only with the much higher sensitivity of the ZIMPOL system (Z urich Im aging Pol arimeter, cf. Povel 1995) that the polarization peaks in the Doppler cores of the two lines could be recorded with convincing precision (Stenflo & Keller 1996, 1997). The core peak in the D1 line has been the main mystery, since according to standard quantum mechanics such a line that represents a scattering transition should be intrinsically unpolarizable.
An elegant explanation of the core peaks was recently proposed by Landi Degl'Innocenti (1998, 1999) in terms of a combination of hyperfine structure splitting and optical pumping. Due to the nuclear spin of for sodium the lower and upper states get split into hyperfine states with different quantum numbers, which are polarizable in principle. To obtain a net polarization for the emitted radiation it is however necessary that the initial, lower state of the scattering transition is polarized (aligned). This may be achieved by an optical pumping process. While alignment in the excited state is induced by the anisotropic excitation, the spontaneous emission process transfers some of that alignment to the lower state. With many such transitions a statistical equilibrium becomes established that leaves the lower level in a polarized state. This ground-state polarization can be imprinted on the polarization of the scattered light, making the D1 line emission polarized. With this theoretical framework Landi Degl'Innocenti (1999) could model the polarized line profiles for different center-to-limb distances.
This explanation however requires that the alignment of the lower state survives destruction by magnetic fields (Hanle depolarization) and collisions during its life time, until the next radiative excitation. Since the life time of the lower state is longer by approximately two orders of magnitudes as compared with the excited state, it is correspondingly more vulnerable to such destruction. Landi Degl'Innocenti (1998) therefore concluded that either the field strength (of a small-scale field with random orientations of the field vectors) would need to be smaller than about 10 mG, or the field orientation would have to be very close to vertical, to make the Hanle depolarization effects sufficiently small. However, Zeeman-effect observations have shown that the magnetic field in the lower chromosphere (where the cores of the Na I D1 and D2 lines are formed) is highly inclined with a canopy-like structure (Giovanelli 1980; Jones & Giovanelli 1983). Hanle effect observations (Bianda et al. 1998a,b, 1999) also rule out that the magnetic field is everywhere vertical, and show that fields as weak as 10 mG do not exist (at least not with sufficient filling factor), but that the solar atmosphere is instead permeated by fields with a strength on the order of 10 G, which may be turbulent lower down and canopy-like higher up.
We are therefore stuck in a paradoxical situation. There is not even a consensus on whether its resolution is to be found within atomic physics or within astrophysics, although we are convinced (because of the mentioned Hanle-effect evidence from other spectral lines) that the problem is basically one of atomic or quantum physics. Further theoretical progress and modelling need to be guided by qualitatively new observations that can better constrain the possible physics involved. The objective of the present paper is to provide such new constraints.
© European Southern Observatory (ESO) 2000
Online publication: March 9, 2000