RF Langmuir probes, revisited

Though probes have been used for $n_{i}$ and $T_{e}$ measurements in rf plasmas for many years and commercial systems are available for automatic scans of $I-V$ curves, accurate results are by no means guaranteed. Strange $I-V$ curves obtained in the past two years have led us to re-examine the problem of Langmuir probes in low-density plasmas in severe rf environments. There are five main problems. First, the Bernstein-Laframboise theory used for small $\lambda _{D}$/$R_{p}$ (Debye length/probe radius) have been shown to be inaccurate$^{1}$. We therefore use thin probes and the OML (orbital-motion-limited) theory for large $\lambda _{D}$/$R_{p} $, but at low $n$ the electron curve is not exponential. Second, we find that the plasma potential drifts with probe voltage $V_{p}$ on a msec or longer timescale. This can be explained, but the effect is larger than expected. Third, the drift can be overcome by rapid sweeping of the $I-V$ curve, but too fast a $V_{p}$ scan would be distorted by the inductance of the choke chain. Fourth, rf compensation is done with a chokes and a large compensation electrode$^{2}$ (CE). We find that CE cannot do most of the job; the chokes have to have large impedance at their self-resonant frequencies. But tuning the choke chain is not easy. Finally, cleaning the probe tip is difficult at $n \quad \approx $ 10$^{9}$ cm$^{-3}$ because not enough current can be drawn to the probe. This is seen in very slow drifts of the ion current, in the order of seconds. A good rf probe requires careful construction of the compensation elements and judicious choice of the scan speed. \newline \newline $^{1}$F.F. Chen, Phys. Plasmas \textbf{8}, 3029 (2001). \newline $^{2}$I.D. Sudit and F.F. Chen, Plasma Sources Sci. Technol. \textbf{3}, 162 (1994).