The stability of a cylindrical keyhole is investigated using the energy and pressure balance. Non-equilibrium evaporation from the keyhole surface, surface tension, hydrostatic and hydrodynamic pressure in the melt as well as heat conduction into the workplace are considered. In contrast to former investigations, the temperature at the keyhole wall and the radius of the keyhole adjust themselves self-consistently. A threshold for the laser power per thickness of the workpiece is found above which the formation of a stable keyhole commences. For iron, this threshold is 790 W mm-1. The temperature at the keyhole wall exceeds evaporation temperature by approximately 100 K. The keyhole radius exceeds the radius of the laser beam and is at least 1.7 times the laser radius.

The conjecture which explains the humping phenomenon in terms of Marangoni convection is discussed and rejected. Instead, Rayleigh's theory of the instability of a free liquid cylinder due to surface tension is applied. The width-to-length ratio of the weld pool has to exceed 1/2 pi to avoid humping. The growth time of a disturbance is found to be approximately the same as the growth time of a hump. The analysis of a bounded cylinder provides a new stability criterion which allows the introduction of a bounding function to distinguish between arc and laser welding. The weld pool dimensions are estimated in terms of a simple heat conduction model. The threshold value predicted theoretically for the travel speed above which humping commences agrees well with the experimental value. It decreases with increasing power, which is in qualitative agreement with experimental results.

The dynamic behaviour of a keyhole in laser welding is studied theoretically. Starting from the stationary state, where the recoil pressure from ablating particles is in equilibrium with the surface tension at the keyhole wall, the collapse time due to a sudden laser shut-down is calculated. The characteristic time constant (r0 3 rho / gamma )1/2 of the system (r0 is the initial keyhole radius, rho is the density of the melt, gamma is the gamma coefficient of surface tension) which is approximately 0.1 ms for Al, Fe and Cu turns out to be a lower limit of the keyhole closing time. Linear stability analysis of the stationary state reveals that under conditions relevant in practice, the keyhole is expected to perform oscillations with frequencies of several hundred Hertz. The results of this investigation are particularly important for pulsed laser applications.

The heat conduction model of a cylinder-type source in laser keyhole welding is extended to a time-modulated laser beam with a prescribed energy flux density instead of a prescribed temperature at the wall of the cylinder. A new non-dimensional parameter Omega = omega r02/ kappa ( omega denotes the frequency of the time-modulated laser beam, r0 the focus radius and kappa the thermal diffusivity of the metal) is found, characterizing the behaviour of the system. The resulting temperature oscillations mainly affect the weld seam, whereas the heat-affected zone is not influenced. The time behaviour of the surface temperature of the keyhole is calculated. The study is also applied to a laser beam modulated as a pulse train.

In scanning tunneling microscope experiments with high voltage pulses hiBocks are formed on the substrate. The conjecture which explains these nanostructures in terms of Beld ion evaporation is discussed and rejected. Another model for the Au-Au system is proposed on the basis of local melting of the tip by the Nottingham eKect. The resistive Joule heating is shown to be negligible.