A simple optical electron polarimeter

It is pointed out that heavy atoms (i.e. those which have spectroscopically resolvable fine structure) are not required for optical measurements of electron polarisation. A polarimeter which uses helium gas instead of heavy-metal vapour is proposed, and several experimental details are discussed. Optical detection of electron polarisation was first proposed more than a decade ago by Farago and Wykes (1969). Their scheme involved the exchange excitation of Hg to the 63P1 state by polarised electron impact. The resulting 61S0-63P1 resonance radiation has circular polarisation (relative Stokes parameter S / I ) which can be related to the incident electron polarisation P. In a subsequent paper, Wykes (1971) suggested an alternative method in which the n s 2 'So ground state of Hg, Cd or Zn is excited to the ns(n + 1)s 3S1 level and the resulting nsnp 3PJ-ns(n + 1)s 3S1 radiation is monitored. The 3PJ-3S1 multiplet must be resolved in order to observe circular polarisation; this dictates the use of relatively heavy atoms as targets. In the latter scheme, S / I and P have a simple relationship (ignoring hyperfine depolarisation) that holds at all incident electron energies for which cascade contributions to the relevant line radiation are negligible. In addition, the problem of radiation self-absorption by the target is significantly reduced. Eminyan and Lampel (1980) have described an experiment in which they measured the longitudinal polarisation of a beam of electrons in this manner, using a Zn-vapour target. More recently, Wolcke et a1 (1983) reported observations of circular polarisation in the Hg 61S0-63P1 transition along the polarised electron beam axis. They found that a 4P Hgresonance 0.03 eV above the 63P1 threshold caused significant deviation of S / I from the threshold value predicted by Farago and Wykes (1969). Optical polarimeters have several advantages over devices based on Mott scattering. Their analysing power is generally higher; S/I for the 43P0-53S1 transition in Zn equals 0.974P (Wykes 1971) compared with typical Mott asymmetries of -0.4P. Polarisation can be measured along any axis without first requiring a spin rotation. Experimental difficulties associated with the use of Au foils (acceleration of the electrons to high energy, the extrapolation of scattering asymmetries to zero foil thickness) in conventional Mott detectors are eliminated as well. The chief disadvantage of the optical method as it has been proposed and employed is the requirement that a heavy-metal vapour be used as the target. + Present address: Physics Department, University of Missouri-Rolla, Rolla, Missouri 65401, USA. @ 1983 The Institute of Physics L553 L554 Letter to the Editor The purpose of this letter is to suggest that He gas be used in place of Zn, Cd or Hg vapours. We consider exchange excitation of the He 33PJ manifold, and measurement of circular polarisation in the 23S-33P, 388.9 nm multiplet. Immediately following production of the 3P state, only the electron spin coordinate is oriented. (Another way to say this is that immediately following excitation, the multiplet must be resolved in order to observe circular polarisation.) At later times, however, the spin-orbit interaction transfers some of this vector polarisation to the atom's orbital coordinates. The time-averaged value of SII is thus non-zero even though the multiplet is unresolved. The use of a He target results in the following simplifications. (i) The gas is essentially mono-isotopic, and there is no hyperfine depolarisation of the excited state. In addition, He is sufficiently light that the Russell-Saunders coupling holds to an extremely good approximation. Thus a particularly simple relationship between P and SII can be derived. (ii) In contrast with the case of Hg (Wolcke et al 1983), spin flips resulting from spin-orbit coupling in resonant Hestates are negligible, as we shall see. In addition, magnetic interactions between the incident electron and the target nucleus, which could also cause spin flips, are greatly reduced. (iii) Investigations of systematic effects, such as the variation of light polarisation with target atom density, are considerably easier. We now obtain an expression relating SII of the unresolved 388.9 nm multiplet to P, using the formalism developed by Ellis (1973, 1977). Excitation of the target by a beam of electrons whose momenta are along the 2 axis and whose polarisation is parallel to , C is considered. The excitation may be thought of as impulsive, because the collision time (-10-l6 s) is much shorter than the relevant spin-orbit periods (lO-''s<). Thus the initial density matrix is written most simply in the LS direct product space: p (33P, t = 0) =PLOPS where, in the m, basis