Measurements of plasma parameters in a semiconductor processing reactor*

Applied Materials (AMAT) and Princeton Plasma Physics Laboratory have collaborated to investigate plasma parameters in an AMAT processing reactor equipped with optical and electrical diagnostics, as well as radiofrequency (RF) and pulsed-voltage (PV) power sources. In the AMAT reactor, ionization and dissociation are controlled by the RF power delivered to the first electrode embedded in the substrate-supporting cathode assembly, whereas the ion energy towards the substrate is generated by establishing a PV waveform at the second embedded electrode separated from the substrate by a thin dielectric layer. Both RF (source) and PV (bias) powers are applied as synchronized trains of bursts each containing 3 level-states of either power. By selecting zero source power for state-2 (S2) and varying the PV level, different afterglows can be studied. During the afterglow, plasma is no longer produced by the source, whereas ions are lost towards both the biased substrate and the grounded top electrode. A non-zero PV during S2 (afterglow) not only affects ion dynamics and residence time, but also allows to infer the current to the substrate (due to ions and secondary electrons) from the pulser current measured with the current transducer at the pulser output. In this study, ions in the discharge afterglow were used as a proxy for neutral radicals to understand the species residence time in the reactor volume. Substrate-current decay times were compared with N2+ decay times measured by laser induced fluorescence (LIF) at the center of the discharge. Trends of the ion density decay time measured by LIF at different S2 voltages and duty cycles were found to be consistent with those of substrate current, whereas absolute decay times from the two techniques showed some differences. In turn, the photomultiplier tube based broadband optical emission spectroscopy with nanosecond time resolution was used to infer the electron temperature decay time in the beginning of the afterglow. Results were found to be consistent with background emission decay times observed in LIF experiments. Also reported are measurements of the sheath thickness using optical imaging for a range of S1 PV levels and zero S2 levels. Results were found to be consistent with estimates based on substrate current and voltage measurements.