Introduction In a solar flare, charged particles (for example fast electrons) are accelerated along magnetic field lines in the solar atmosphere, travelling through the ambient coronal plasma. As they pass through the plasma, collisions generate a positive slope in the beam velocity space, which leads to the generation of Langmuir waves. These in turn interact further to produce Type III radio bursts (the wavelength of which is directly related to the plasma frequency and hence the ambient density and the velocity of the particles along the magnetic field). The growth rate of Langmuir waves is a function of the wavenumber and the beam electron distribution function. We study the development of an electron beam under the influence of Coulomb collisions, and investigate conditions for reverse-drift Type III radio emission. Modelling the Velocity Evolution of an Electron Beam with Stochastic Simulations JA Pollock, L Fletcher Physics & Astronomy Dept. University of Glasgow, Glasgow, G12 8QQ Stochastic Simulations The general distribution function of electrons in a beam can be described by [1] : (3) Boltzman Equation which is equivalent to the Boltzman Equation if. When electromagnetic forces are introduced, this can be written: (4) Vlasov Equation which is the Vlasov Equation. This equation, when collisions are introduced, provides the basis of the evolution of the electron distribution function, which can be described thus (e.g. Kovalev & Korolev, 1981): (5) Fokker-Plank Equation where μ=cosθ, the angle of particle injection, and ( Λ ≈20 is the Coloumb logarithm and n ≈10 9 cm -3 is the coronal number density). We assume a constant magnetic field. This is a Fokker-Plank Equation, which has been shown (in the normalised situation) to be equivalent to the set of stochastic differential equations [2] : (6) where r(t) is a Gaussian random noise process of mean 0 and variance 2. We have used these stochastic differential equations to calculate v, z and µ incrementally in a Java program, and from this produced histograms of the resultant velocities of particles in the z direction (along the field lines). We have assumed straight field lines. [1] “Physics of Space Plasmas” GK Parks [2] “Stochastic simulation of fast particle diffusive transport.” AL McKinnon & IJD Craig, Astron. & Astrophys. 251, 1991 [3] “Instabilities In Space & Laboratory Plasmas” DB Melrose [4] “On the generation of loop-top impulsive hard X-ray sources.” L Fletcher, Astron.& Astrophys. 303, 1995 [5] “Collisional interaction of a beam of charged particles with a hydrogen target of arbitrary ionization level.” AG Emslie, Astrophysical Journal, 224, 1978 [6] “Plasma Astrophysics” 2 nd Edition, AO Benz, ASSL / Kluwer 2002 [7] “Physics of the Solar Corona - An Introduction” M Aschwanden, Springer / Praxis 2004 Acknowledgements: Iain Hannah, Ross Galloway, Dr Norman Grey & Dr Graeme Stewart. Further Work We intend to consider further the effects of different initial velocity and injection angle distributions with the existing simulations, and later to include further scattering terms, curved and varying magnetic fields and relativistic particles. We also hope to be able to incorporate different atmospheric models rather than assume constant density as we have here. Type III Radio Bursts The electrons in the atmosphere are primarily thermal, and therefore have a thermal distribution function, however the accelerated component has higher energy and a different distribution – we shall consider a power law distribution. Coulomb collisions affect the low energy part of the distribution (see Eq. 6) and this leads to a ‘bump-in-tail’ instability in the distribution function of the electrons. The growth rate of this instability depends on the rate of change of the distribution function with velocity along the field line according to the following equation [3] : (1) where the sign of is the same as the sign of hence a positive slope in a histogram of v z, the component of the velocity along the field line, results in the growth of waves. Langmuir waves Type III radio emission These are Langmuir waves which, through a series of further interactions, result in Type III radio emission. The wavelength of the resultant emission is related to both v z and n, the coronal number density, by the following equation (in cgs units): (2) Hence for Type III radio emission of minimum wavelength λ =10cm, the minimum velocity at which the slope must appear positive is v z =2.389×10 9 cms -1. Fig. 1: Reverse-drift Type III radio bursts. [Aschwanden & Benz, ApJ 480, 1997] Results We have simulated the evolution of the distribution function of an electron beam passing through the ambient coronal plasma using Eqs. 6 in different situations. The velocities (energies) predicted by the simulation at given distances for given initial energies agree with those predicted by the equation for energy as a function of column depth and initial angle in Emslie, 1978 [5], thus verifying the simulation. We have assumed non-relativistic electrons, and consider only Coulomb collisions as a source of scattering. It should be straightforward in the future to add further forms of scattering to the simulation. We assume Type III radio emission of wavelength greater than 10cm, which corresponds to a minimum velocity (energy) in the direction of the field (via Eq. 8) of 2.839×10 9 cms -1. These graphs allow us to see where a positive slope in velocity space develops in relation to the required minimum velocity for Type III radio emission and hence at what distance along the field line a wave resulting in this sort of emission is created. Thus we can postulate where Type III emission originates. The y-axis is number density (the data were binned into histograms) and represents the electron distribution function. Figure 2 shows that for electrons of injection angle 60° and initial energies chosen randomly from a power law distribution between 7.997keV and keV, a positive slope at the velocity threshold first appears after a distance of approximately 1×10 7 cm (~100km) for a coronal density of 10 9 cm -3. Varying the initial energies very soon results in all positive gradients appearing before or after the minimum threshold velocity, in which cases there is the possibility of Type III emission at no or all distances along the field line respectively. The same is true of varying the initial injection angle. Figure 3 shows the results of a simulation with the initial injection angles chosen randomly from a uniform distribution between 0° and 90°, and the initial energies chosen randomly from a power law distribution between keV and keV. In this situation, increasing the initial energies results in much less change in the evolution of the distribution function over time compared to the situation presented in Fig. 2. In fact, even at these near relativistic energies, the positive slopes in velocity space do not reach the minimum velocity required for Type III emission of λ > 10cm. The results so far would suggest that Type III emission only results from beams, since for initial injection angles chosen randomly from a uniform distribution, positives gradients in velocity space do not appear at the threshold minimum velocity even for near-relativistic particles. Further work into relativistic particles will investigate this further. Fig. 3: Log-log plot of electron number density (≡ distribution function) against component of velocity along the field ( v z ) line at different distances along the field line ( z ). Initial angle of injection chosen from uniform distribution between 0° and 90°, initial velocities randomly chosen from power law distribution in the range ×10 10 cms -1. Fig. 2: Log-log plot of electron number density (≡ distribution function) against component of velocity along the field ( v z ) line at different distances along the field line ( z ). Initial angle of injection constant (60°), initial velocities randomly chosen from power law distribution in the range ×10 9 cms -1.