Electric lamp and discharge devices: systems – Discharge device load with fluent material supply to the... – Plasma generating
Reexamination Certificate
1996-11-04
2001-06-26
Bettendorf, Justin P. (Department: 2817)
Electric lamp and discharge devices: systems
Discharge device load with fluent material supply to the...
Plasma generating
C333S017300, C118S7230IR, C156S345420
Reexamination Certificate
active
06252354
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
The invention concerns tuning an RF signal generator to a plasma-loaded RF signal applicator of a plasma reactor, such as an antenna or an electrode, by serving the RF frequency so as to optimize the value of a selected parameter, such as, for example, delivered current or voltage or power, using a fixed tuning circuit. The fixed tuning circuit may be switched to a selected one of several transformers, depending upon the tuning range anticipated.
2. Background Art
RF plasma reactors of the type employed in processing semiconductor wafers require a large amount of RF power to maintain a plasma within a vacuum chamber, typically on the order of a thousand watts at RF frequencies on the order of several megaHertz. To maintain a high density plasma, the RF power is best inductively coupled into the chamber via an overlying coil antenna, while plasma ion energy can be controlled by controlling the voltage on the semiconductor wafer being processed. Typically, the RF signal source is an RF generator having an output impedance of 50 &OHgr; with substantially no reactance. The input impedance presented by the plasma-loaded coil antenna is typically not 50 &OHgr; and has substantial reactance, so that there is a substantial impedance mismatch. For example, the coil antenna typically has a length much less than a quarter wavelength of the RF signal, so that the coil antenna presents an impedance having a real part much less than that of the RF generator (which is typically 50 &OHgr;) and having a very high inductive reactance. Such a mismatch causes RF power to be wasted by reflection back to the RF generator rather than being delivered to the plasma, so that it is difficult to control the amount of RF power delivered to the plasma. As a result, process control is compromised. One solution to this problem is to provide a fixed RF match circuit having lumped reactive elements so as to maintain a zero phase angle between RF voltage and current. Moreover, optionally a transformer can be employed to provide a match between the output and input impedance magnitudes.
The problem with such a fixed match circuit is that the input impedance of the plasma-loaded coil antenna changes as process conditions inside the reactor chamber change. Thus, as changes in plasma conditions change the plasma-loaded antenna impedance, the match circuit no longer can perform its function, and RF power delivered to the plasma falls off. Such a reduction in delivered RF power typically distorts the plasma processing of the wafer and in many cases is unacceptable. Therefore, the best solution in the art is to provide an RF impedance matching apparatus that adjusts the impedance match in response to changes in the plasma-loaded impedance of the antenna.
A conventional plasma reactor having such a variable RF impedance match circuit is depicted in FIG.
1
A. The plasma reactor includes a reactor chamber
100
evacuated by a pump
105
, a wafer support pedestal
110
on which a wafer
115
may be placed, an overhead coil antenna
120
, and a gas inlet
125
into the chamber coupled to a process gas supply
130
. An RF plasma source signal generator
140
is connected through an RF impedance match box
150
while an RF bias signal generator
160
is connected through another RF impedance match box
170
to the wafer pedestal
110
. The power applied by the plasma source signal generator
140
controls plasma ion density in the chamber
100
while the power applied by the bias signal generator
160
controls plasma ion energy near the wafer
115
. In some cases, both ends of the coil
120
may be connected to ground through respective capacitors shown in dashed line in FIG.
1
A.
The RF impedance match boxes
150
and
170
are generally the same and will be described with reference to the RF impedance match box
150
. The impedance match is provided by a conventional “pi-network” consisting of a pair of parallel capacitors
180
,
185
(which are really capacitor circuits) on either side of a series inductor
190
. Each of the capacitor circuits
180
,
185
is controlled by an impedance match controller
200
. The controller
200
monitors the forward voltage, reverse voltage and current/voltage phase angle via a conventional directional coupler
210
at the RF input
150
a
and computes from these three parameters a correction to the capacitance of each variable capacitor circuit
180
,
185
, using a network model
220
. The controller
200
issues control signals at its control outputs
200
a
,
200
b
to the variable capacitors
180
,
185
to effect the needed corrections in their capacitance values. Each of the variable capacitors
180
,
185
can be a mechanically variable capacitor or an electrically variable capacitor circuit as illustrated in the drawing, the latter choice being preferable.
FIG. 1A
illustrates one example of the latter case, in which each variable capacitor circuit
180
,
185
consists of an electrically variable inductor
230
connected in parallel with a fixed capacitor
240
. The variable inductor
230
is a saturable reactor consisting of a primary winding
232
, a magnetically permeable core
234
and a smaller control winding
236
connected to a variable current source
238
. A respective one of the control outputs
200
a
,
200
b
is connected to an input of the current source
238
. The controller
200
can decrease the capacitance of the variable capacitor
180
by increasing the D.C. or low frequency current through the control winding
236
. This in turn reduces the permeability of the core
230
(by inhibiting the magnetic domains in the core from following the field fluctuations in the primary winding
232
) and hence reduces the inductance presented by the primary winding
232
, thereby decreasing the predominance of the capacitive reactance presented by the fixed capacitor
180
over the inductive reactance. Such a change represents an effective decrease in capacitance of the variable capacitor
180
. The reverse process produces an increase in capacitance.
One disadvantage of such a device is that it requires a measurement of the forward and reflected voltages and the phase therebetween, or a measurement of the current and voltage and the phase therebetween. Another disadvantage is that it is bulky and costly. Another more important disadvantage is that there are hysteresis losses in each core
234
that vary as the load impedance varies. Referring to
FIG. 1B
, as the applied magnetic field H (from the control winding
236
) increases and then decreases, the induced magnetic field B that fixes the polarization of the core magnetic domains changes at different rates so that there is a net loss of energy with each cycle of the induced field. Referring to
FIG. 1C
, the complex impedance plane of the match network includes a tuning space
300
within which the impedance match controller provides a theoretically exact solution to the impedance matching problem. Assuming the signal generator output impedance is a purely resistive 50 &OHgr;, if the controller
200
commands a higher control current through the control winding
236
, it moves the impedance presented by the pi-network
180
,
185
,
190
to a lower impedance in the region
310
of the control space. In this case the magnetic core
234
fluctuates near the origin of the hysteresis loop of FIG.
1
B and the losses are slight. If on the other hand the controller
200
commands a lower control current, the resulting impedance can be found in the region
320
of higher resistance in the control plane, and the fluctuations in the core
234
may reach the outer extreme of the hysteresis loop of
FIG. 1B
, in which case the losses in the core
234
are very high. Thus, it is seen that the delivered power to the coil antenna
120
will necessarily vary as the plasma-loaded impedance of the coil antenna
120
varies, a significant disadvantage.
Various other impedance match techniques are known in addition to the foregoing. For example, to avoid the problem of load-dependent
Buchberger Douglas
Collins Kenneth
Roderick Craig
Shel Viktor
Trow John
Applied Materials Inc.
Bettendorf Justin P.
Michaelson and Wallace
Wallace Robert
LandOfFree
RF tuning method for an RF plasma reactor using frequency... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with RF tuning method for an RF plasma reactor using frequency..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and RF tuning method for an RF plasma reactor using frequency... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2493966