The main thrust of our research is to advance the science of low-temperature plasma physics, particularly as applied to semiconductor fabrication. Emphasis has been on the plasma sources used for this purpose: how they work and how they can be improved. These sources run on radiofrequency (RF) power, which makes probing them difficult; hence, considerable work was done on diagnostics, especially Langmuir probes. Helicon sources have commanded most of our attention because these convert RF energy to plasma density extraordinarily well and may be the source of the future. Unlike other RF sources, these devices work on the principle of wave generation in a plasma. A helicon wave, which propagates in a cylinder along a DC magnetic field, absorbs the RF energy and deposits it efficiently in the plasma. How it does this and why the process is more efficient than others, is a puzzle which took many years to unravel. Helicon research is like an onion: you peel off one layer, only to find another layer underneath that makes you cry. A summary of these layers will follow shortly. At present the semiconductor industry is not yet ready to adopt helicon sources and is depending on older devices such as the capacitive discharge (or Reactive Ion Etcher, RIE) and the Inductively Coupled Plasma, ICP. Though these reactors are well developed, they are by no means well understood. In particular, the ICP, which requires no DC magnetic field, has an anomalous skin depth. That is, the RF energy penetrates into the plasma and produces a uniform density when it should be stopped within a few centimeters of the surface. Our recent work throws light on this problem. RF sources are a gold mine for interesting physics problems, challenging design problems, and inventive ideas. Physics of helicon discharges Click on thumbnails to see full pictures 1. Basic experiments. Measurements of plasma parameters and wave fields were made in a 160-cm long, 5-cm diam discharge tube in a uniform magnetic field B. Results agreed with helicon theory (for nonuniform density profiles) and with classical diffusion theory, with a few surprises. The density peaked far downstream instead of near the antenna; the wave field had a beat pattern downstream; the m = +1 (RH) azimuthal mode dominates over the m = - 1 mode; there is low-B density peak around 50G, etc., etc. All of these effects have been explained. The main papers of this research are:
2. Theory. Helicon theory was extended to treat nonuniform densities analytically. Finite electron mass was added, resulting in the discovery of helicon coupling to Trivelpiece-Gould (TG) modes, thus explaining the unusually high RF absorption efficiency of helicon discharges as well as the dominance of the m = +1 mode. Finally, the helicon discharge problem was reduced to a simple equation, and the HELIC code was developed to compute wave properties including the effects of radial gradients, neutral and ion collisions, TG modes, and antenna coupling. The figure shows that the density gradient makes a big difference; the dashed curve is for a uniform plasma.
3. Advanced experiments. The effect of capacitive coupling was shown in a 10-cm diam device with Faraday shields and 2D imaging of the discharge:
Azimuthally rotating RF fields with bifilar antennas were found to be more efficient than simple helical antennas.
Dual antennas were found to produce a density minimum, not maximum, at the midplane. This is explained by gas depletion.
Half-wavelength antennas were found to produce more downstream ionization than full-wavelength ones. This is not yet understood.
By developing an energy analyzer with RF frequency response, it was shown that non-Maxwellian electrons were not present in high-density helicon discharges, thus disproving the Landau damping hypothesis.
By developing an RF current probe, direct evidence of the existence of TG modes in helicon discharges was found at low magnetic fields.
Large Area Distributed Helicon Source A source consisting of seven individual helicon sources successfully produced densities of 1012 cm-3 over a 40-cm diam area with ± 3% uniformity.
Diagnostics A large number of diagnostics for use in harsh RF environments have been developed: RF compensated Langmuir probes, radial and axial magnetic probes, local optical probes, CCD camera imaging, RF-resolved gridded analyzer, and RF Rogowski coil current probe.
Inductively Coupled Plasmas Classical skin depth theory was carefully examined and found incapable of explaining the uniform radial density profiles observed. A Monte-Carlo calculation showed that the nonlinear Lorentz force can push ionizing electrons into the center of the plasma.
Drift instabilities in helicon discharges The plasma density in a helicon discharge generally increases with DC magnetic field B, but it saturates at high B. With light gases, it even falls as B increases past a certain point. This was found by Max Light to be caused by the onset of a low frequency instability driven by density gradients and electric field shear. M. Light, F.F. Chen, and P.L. Colestock, Low frequency electrostatic instability in a helicon plasma, Phys. Plasmas 8, 4675 (2001). M. Light, F.F. Chen, and P.L. Colestock, Quiescent and unstable regimes of a
L Right: Max's machine at Los Alamos.
Plasma processing induced damage One of the main problems in using plasma to etch high-speed circuits is that charges tend to build up on the thin gate insulators of the MOSFET transistors. Huge voltages drive currents through the oxide layers to damage them. A widely accepted mechanism is electron shading damage, in which negative charges on the photoresist prevent electrons from penetrating into the trenches being etched, and ions cause positive charge buildup. To test this, we scaled the experiment so that the trenches are of mm dimensions and could be probed. It was necessary to develop a discharge with low density and high electron temperature so that the Debye length scaled accordingly. Modeling of the ion orbits was done iteratively, with each part of the trench charging to a self-consistent potential. Tsitsi Madziwa has found that the fields at the entrance control the ion orbits, which are ballistic inside the trench. More figures... from the laboratory tour following the APS-DPP meeting in November, 2001: Results from the 5-cm diameter machine (PPT file) Results from the 10-cm diameter machine (PPT file) Results from the single- and multi-tube machines (PPT file) Status of the plasma damage calculation and experiment (PPT file) |