Power matching to pulsed inductively coupled plasmas

Abstract

Matching of power delivery to nonlinear loads in plasma processing is a continuing challenge. Plasma reactors used in microelectronics fabrication are increasingly multi-frequency and/or pulsed, producing a non-linear and, in many cases, non-steady state electrical termination that can complicate efficient power coupling to the plasma. This is particularly the case for pulsed inductively coupled plasmas where the impedance of the plasma can significantly change during the start-up-transient and undergo an E–H (capacitive-to-inductive) transition. In this paper, we discuss the results from a computational investigation of the dynamics of power matching to pulsed inductively coupled plasmas (Ar/Cl2 mixtures of tens of mTorr pressure) using fixed component impedance matching networks and their consequences on plasma properties. In this investigation, we used set-point matching where the components of the matching network provide a best-case impedance match (relative to the characteristic impedance of the power supply) at a chosen time during the pulsed cycle. Matching impedance early during the pulse enables power to feed the E-mode, thereby emphasizing capacitive coupling and large excursions in the plasma potential. This early power coupling enables a more rapid ramp-up in plasma density while being mismatched during the H-mode later in the pulse. The early match also produces more energetic ion bombardment of surfaces. Matching late in the pulse diminishes power dissipated in the E-mode at the cost of also reducing the rate of increase in plasma density.

Science-based, data-driven developments in plasma processing for material synthesis and device-integration technologies

“…64) The dynamics of power matching to pulsed ICPs of Ar/Cl 2 mixtures of tens of mTorr pressure using fixed-component impedance matching networks were investigated. 65) This early power coupling enables a more rapid ramp-up in the plasma density while being mismatched during the inductively coupled (H)-mode later in the pulse. The early match also produces a more energetic ion bombardment of the surfaces.…”

“…Matching late in the pulse diminishes the power dissipated in the capacitively coupled (E)-mode at the cost of reducing the rate of increase in the plasma density. 65) Another example is the measurement of microarcs by electrical detection 66) and a microwave flat cutoff probe. 67) Sensory data can be used to detect abnormalities in the plasma processing apparatus.…”

Estimation of plasma properties using an extended Kalman filter with plasma global models

“…Some studies have compared experimental and computational results to note that while qualitative agreement existed between results, some correction is needed for quantitative agreement, indicating that the physics-based model does not adequately represent the true physics [52]. Others cite power dissipation from well-known sources such as exciting or ionizing neutral particles by collisions with electrons and the transfer of kinetic energy to the wall from electrons and ions [55,56].…”

Self-consistent simulation of the impedance matching network for single frequency capacitively coupled plasma

“…However, all the above models treated the plasma as linear or nonlinear equivalent circuit elements of capacitors and resistors, possibly with a global model to estimate the plasma equivalent circuits elements. Another comprehensive work of coupling the matching network and the plasma load is reported recently by Qu et al [30], in which the impedance of the plasma is calculated and coupled with the circuit model. However, the plasma is simplified as some linear or nonlinear equivalent circuit elements in these works, therefore the time-varying and nonlinear nature are only partially addressed.…”