The plasma reactor (see Fig. 4.3) is constructed from a pyrex half-nipple adapter (~ 4″ OD x 4″ long) having an o-ring face sealing groove from Larson Scientific
Glassware with a gas inlet and pumping port added. The reactor is sealed on one end by an aluminum plate which houses a pyrex protection disk and 3″ OD x 0.060″
molybdenum (to avoid excessive sputtering) bias plate directly exposed to the plasma. On the opposing end, an O-ring on the reactor OD seals against an 8″ Conflat flange that houses another pyrex disk containing the extraction aperture (304 stainless steel). This aperture (see next section) is composed of two pieces which key together through a 3/4″
hole in the pyrex plate (electrically isolating the aperture) with a snap ring on the high vacuum side. The high vacuum side of this flange also serves as the mounting surface for the puller, buncher, and front beamline acceleration electrodes using alumina standoffs.
Gases are introduced through separate mass flow controllers to a mixing station before entering the reactor through Teflon tubing (to maintain electrical isolation). As well, the pumping arm of the reactor is excessively long to keep the plasma volume electrically isolated from the grounded turbo-pump and pressure readout. The whole reactor setup is pumped by a Balzers 210 L/s drag turbo and mechanical backing pump (both were Fomblin oil prepped for oxygen and corrosive gas service). A heated 50 mTorr Baratron functions for pressure readout and the operating pressure (0.2-10 mTorr) is set by adjusting the gas inlet flowrate.
The plasma is excited by a two-turn, solenoid-type RF antenna (1/8″ copper tube cooled with a glycol bath) wound around the glass reactor with a grounded Faraday shield sandwiched in-between. The shield was made of copper sheet after Johnson, (1993) with several 1 mm wide slits lengthwise to aid in striking the discharge in
capacitive mode initially. Several layers of Kapton sheeting prevent electric breakdown between the antenna and Faraday shield. Capacitive coupling between the antenna and plasma is minimized because the induced eddy-currents in the shield cancel out the E- field of the antenna. It is imperative that capacitive coupling be minimized to keep the plasma potential low (< 20 eV) and ion energy distribution narrow (Leiberman and Lichtenburg, 1994). The entire reactor is forced air cooled with two high volume fans because of the high RF levels used to power the discharge.
The electrical circuit used to excite and bias the plasma is shown in Fig. 4.4. The RF antenna is driven through a Π-network match box from an ENI 1250W RF power supply operating at 13.56 MHz. Bird power meters are used to monitor the forward and reflected power from the antenna. The plasma bias is supplied by a Spellman +3 kV, DC switching power supply (25 kHz) which had to be protected from the high frequency electron oscillation in the plasma potential (5 V peak-to-peak @ 13.56 MHz). A 4-stage low-pass LC filter was necessary to reduce the 13.56 MHz noise to < 0.1V peak-to-peak on the bias line to allow the feedback regulator in the DC supply to function properly.
In the inductive excitation scheme, plasma electrons behave like an inductor (Le)
in series with a ″plasma resistance″ (Rp). A ″transformer-like″ equivalent circuit for the
The oscillating B-field from the antenna, when properly matched, can induce a standing wave in the plasma ″tank″ circuit, thereby transferring power. Maximum coupling efficiency occurs when the plasma inductance is large (high electron density) and impedance matched to the antenna circuit. As one might expect, this coupling through magnetic induction depends directly on the ″tank″ circuit of the plasma. If the effective inductance
(
due to the plasma electrons is small (low plasma density), then inductive coupling does not occur, and power can not be transferred through the antenna B-field. This behavior manifests itself in the lab as a weak, capacitively coupled plasma formed initially that will suddenly jump to inductive mode once a threshold electron density is reached. The initial discharge couples to the E-field of antenna and as power input increases, more and more electrons are produced through ionization. A point is reached when the plasma electron inductance becomes large enough and the tank circuit of the plasma begins to ″ring″ with B-field of the antenna like the secondary winding of a transformer. When this occurs, the electron density (ion density too) jumps up abruptly by several orders of magnitude. The transition from capacitive to inductive mode depends strongly on the operating pressure, working gas, and power input (Leiberman and Lichtenberg, 1994). At low pressures or when working with highly electronegative gases, inductive coupling will just not occur without a strong magnetic field around the plasma to keep the electron density high.)
e L
For our reactor, operating pressures were routinely 1-5 mTorr with input powers of 300-700W (depending on the working gas) needed for reasonable inductive coupling. One important point that should not be overlooked is the extremely high resonance Q- factor. Proper impedance matching of the antenna to the plasma to within ± 2 pF on the match box was critical to maintain inductive operation. Gases that were hard to break down and drive ″inductive″ were routinely mixed with argon to provide enough electrons for good coupling. Therefore, the need for magnetic confinement was avoided. Mixing gases was not deemed a concern because the ion beam was mass filtered downstream.