This invention relates to the formation of plasmas for use in semiconductor processing and in particular to the formation of an inductively-coupled plasma.
Plasmas are widely used in the semiconductor industry for depositing thin films, for etching thin films and the underlying semiconductor material, and for dry-cleaning wafers. The plasma is typically formed in a cylindrical vessel or tube around which a wire is coiled. The reactant gases are introduced into one end of the tube and the atoms or ions that are generated by the plasma exit the other end of the tube and flow towards the wafer. Power is applied to the plasma by means of the coil. The mechanism by which the electrical power is transferred to the plasma can be either capacitive or inductive. Normally, both modes are present but one of the modes is predominant. In many applications the inductive mode is preferred because it produces a plasma having a higher ion density and because the ions in the plasma do not bombard the walls of the tube as much as they do when the plasma operates in the capacitive mode. This reduces wear on the tube and increases its life.
One structure for generating a plasma is described in U.S. Pat. No. 6,007,675 (
In practice, it has been found that this structure has several defects. The plasma operates primarily in the capacitively coupling mode when the power supplied to each tube is less than about 500W. For example, when the plasma is used to strip bulk photoresist, it has been found that the coupling mode changes abruptly from capacitive to inductive when the power supplied to each tube increases beyond about 600-700W. This is shown in
Moreover, in the arrangement shown in
These problems are overcome in an arrangement according to this invention, wherein a single tube is supplied with electrical power through a separate impedance match network. There is no direct electrical connection between the coils which enclose different tubes. Each of the coils is bifilar, i.e., each coil contains at least two windings that have turns that are interlaced or interdigitated with the turns of the other winding.
Coil 32 is supplied by an RF generator 40, which operates at 13.56 MHz and which supplies a signal to coil 32 through an impedance-matching network 50. Impedance-matching network 50 includes a phase angle detector 502 and a control motor 504, which drives a load capacitor 506 and a phase capacitor 508 in an LC circuit 514. Circuit 514 also includes inductances 510 and 512, which are connected in series with phase capacitor 508. Matching network 50 is tuned to the impedance of coil 32 by minimizing the reflected power as seen by phase angle detector 502. The minimal reflected power is achieved through a tuning algorithm in which the positions of capacitors 506 and 508 are controlled by motor 504. To minimize the internal losses in matching network 50, capacitors 506 and 508 are preferably vacuum capacitors.
It should be understood that matching network 50 is only illustrative. Numerous forms of matching networks may be employed in other embodiments of this invention.
Coil 32 includes a first winding 36 (solid lines) and a second winding 38 (dotted lines). Windings 36 and 38 are interlaced or interdigitated, e.g., turn 38y of winding 38 is interposed between adjacent turns 36x and 36z of winding 36. Output line 55 from matching network 50 connects to a first end of winding 36, which is located at a first end of coil 32. The other (second) end of winding 36 is located at a second end of coil 32 and is connected via a return line 39 to a first end of winding 38 that is located at the first end of coil 32. The other (second) end of winding 38 is located at the second end of coil 32 and is connected to ground.
The reactor in which coil 32 is housed typically contains other plasma tubes (not shown) that are connected to impedance-matching networks separate from impedance-matching network 50. This allows independent control of the plasma in each tube, and the plasma in a given tube will not be impacted by the conditions in other tubes.
As shown in
The length of the coil may be set so as to provide a helical resonator, with standing waves in the tube. This requires that the length of the coil be a multiple of the wavelength of the electromagnetic radiation within the tube (λ=c/f).
The pressure within the tube may be not uniform from top to bottom, particularly if the tube has a diameter that is relatively small in comparison to its length. The distribution of the plasma inside the coil can be improved by adjusting the position of the coil.
The plasma may be further moved in the direction of the inductive coupling mode by positioning a Faraday shield between the coil and the tube.
If a more neutral species is required for the particular process (e.g., stripping), a recombination mechanism may be placed between the plasma tube and the reaction chamber. In the reactor 90 shown in
The plasma source of this invention has numerous benefits and advantages. The plasma is easy to strike and sustain. For example, it is possible to strike and sustain a plasma at a power level of only 1W per tube. On the other hand, the power may vary widely, up to 3000W per tube, for example. The reaction rate may greatly increased by using higher power levels. For example, in stripping processes more power may be used to dissociate a higher 02 flow rate. As shown in
The foregoing embodiments are illustrative only and not limiting. Numerous alternative embodiments will be apparent to persons of skill in the art. The broad scope of the invention is limited only by the following claims.
This application claims the priority of U.S. Provisional Application No. 60/391,373, filed Jun. 25, 2002.
Number | Name | Date | Kind |
---|---|---|---|
4849675 | Muller | Jul 1989 | A |
6007675 | Toshima | Dec 1999 | A |
6077384 | Collins et al. | Jun 2000 | A |
6224680 | Toshima | May 2001 | B1 |
6527912 | Chen et al. | Mar 2003 | B2 |
6744213 | Wilcoxson et al. | Jun 2004 | B2 |
Number | Date | Country | |
---|---|---|---|
60391373 | Jun 2002 | US |