Feedback stabilization of instabilities was demonstrated in inductively coupled electronegative plasmas by modulating the radio frequency power in response to a feedback control signal. Two modes were observed and stabilized: a fast bursting mode (designated 'B' mode) and a slower oscillatory mode (designated 'O' mode). 'B' mode is a low duty-cycle non-linear phenomenon characterized by short (<1 µs) spikes on optical signals, at a pulse repetition rate of 10-40 kHz. 'O' mode is characterized by continuous oscillations in ion saturation current at lower frequency (200 Hz-1 kHz). Some parameter ranges exhibit both modes. Fluctuations in similar processing plasmas and having similar characteristics have previously been observed and identified with instability due to electron attachment to the electronegative neutral gas species.Because 'O' mode exhibits fluctuations in ion density, it may be of more concern than 'B' mode for repeatable plasma processing. Stabilization of 'O' mode may be accomplished using the ion saturation current signal collected by a Langmuir probe as a measure of the plasma density. The floating potential obtained from the capacitively coupled plasma bias electrode is similar in shape to the ion saturation current, and can also be used as the feedback control signal.The bandwidth required for stabilizing the 'O' mode is in the kHz range, whereas the bandwidth required to stabilize 'B' mode may be closer to a MHz. In practice, it is far easier to stabilize the 'O' mode than the 'B' mode, but both may be effectively suppressed. However, since a purely 'B' mode does not seem to measurably modulate the ion density, and unless it dramatically changes the average electron temperature it may not be necessary to stabilize it in order to achieve repeatable plasma processing.

Multiply charged ions are studied in the Constance B quadrupole mirror experiment [Phys. Rev. Lett. 59, 1821 (1987)] in order to better understand the ion physics of electron cyclotron resonance (ECR) high charge-state ion sources. By measuring the ion densities and end loss fluxes, the parallel confinement times for the first five charge states of oxygen plasmas are determined. The parallel ion confinement times increase with charge state and peak on axis, both indications of an ion-confining potential dip created by the hot electrons. The radial profile of ion end loss is normally hollow, with the peak fluxes occurring at the edge of the ECR zone. An attempt is made to increase the end loss flux of a selected ion species by decreasing its parallel confinement time using minority ion cyclotron resonance heating (ICRH). In addition, an ion model is developed to predict the ion densities, end loss fluxes, and confinement times using the ion particle balance equations, the quasineutrality condition, and theoretical confinement time formulas. The model generally agrees with the experimental data to within experimental error.

Detailed experimental measurements of multiply-chargered ion densities, endloss currents,and temperatures have been made in a single cell quadrupole mirror. The purpose of these experiments is to determine the confinement physics of ions over a range of charge states in an ECR ion source. The parallel ion confinement times agree well with a theoretical model including Pastukhov and flow confinement for a potential dip of ϕi/Ti ~ 1. The ion endloss currents were found to be up to 50% hollow. Radial transport of ions'was found to be important in limiting the extraction of high charge state ions. In addition, by heating O4+ with ion cyclotron resonance heating we were able to increase its extracted current in the plasma center by 25%