Roland A. - PVD for microelectronics (779636), страница 15
Текст из файла (страница 15)
Since the ion's radius is too large to be relevant in sputtering tools, the electron's cyclotron (angular) frequency can beeasily found to beto =eBm(3.13)In Hertz units (1/sec), the oscillation frequency isf = 2.8 x 106B(3.14)where B is the magnetic field in gauss.Example: The cyclotron frequency at a magnetic field of 5 Gaussis 13.5 MHz; at 875 gauss, f = 2.45 gHz.The cyclotron frequency does not depend on the energy of the oscillating electron, only the magnetic field.
However, the radius scales with thesquare of the kinetic energy.There are three general configurations for imposing a magnetic field ona plasma. The first is simply to impose a magnetic field that is perpendicular to the cathode surface (and the sample surface). This forces the secondary electrons emitted from the plasma to move in helical paths and increases the plasma density. This approach is rarely used because (a) it ishard to get a uniform magnetic field (it is usually peaked on the systemaxis), and (b) there are better alternatives.The second two magnetic-enhancement techniques require the use of amagnetic field that is parallel to the cathode surface. The secondary electrons in this case undergo a more complex motion and end up beingtrapped in the plane of the magnetic field, fairly close to the cathode.
Thetrapping process uses an effect known as an E x B drift, which is anothercross product, and moves the electrons on a trajectory perpendicular toboth the electric field (vertical) and the imposed magnetic field (horizontal). This drift is sketched out in Fig. 3.11.70R. POWELL AND S.
M. ROSSNAGELFIG. 3.11S c h e m a t i c o f an E x B drift effect.The E x B drift effect results in a pile-up of electrons at one side of theelectrode system: sort of a plasma-phase Hall Effect. While this would notbe desirable, the second type of magnetic-enhancement tool uses this effect combined with a moving magnetic field. The field is set up by oscillating electromagnets (Fig. 3.12), which cause the net magnetic field to rotate around the vertical axis of the tool.
The moving magnetic field spreadsthe E x B-trapped electrons around the cathode surface and can lead togood uniformity. This effect is the basis of the AME5000 tool [3.8].The third magnetic-enhancement system also uses the E x B-trappingeffect, but instead of moving the magnetic field, it uses a magnetic fieldthat is not uniform around the system. As shown in Fig.
3.13, the magneticfield in this case is configured radially and parallel to a round, planar cathode surface. The result is a closed-loop E x B drift path for the secondaryelectrons. The secondary electrons are trapped in this ring close to thecathode and can lead to very high levels of ionization of the backgroundgas. This geometric design is known as a magnetron.The formation of the closed-loop path for the E x B-drifting secondaryelectrons is what defines a magnetron. The simplest geometric design isthat shown in Fig.
3.12, which is called generically a circular planar magnetron. As a measure of the trapping efficiency of this class of cathode,PLASMA SYSTEMS71FIG. 3.12 (a) Schematic of an RF diode with lateral magnetic field set up by a coil, (b) schematicof a moving magnetic field E • B-trapped cathode.measurements were made of the relative amount of circulating current andwere compared to the net discharge current [3.9]. An example of that workfor a small, 15-cm-diameter cathode is shown in Fig. 3.14, where the circulating current was inferred by measuring the magnetic field it induced.72R.
POWELL AND S. M. ROSSNAGELFIG. 3.13Top and side view schematic of a circular planar magnetron cathode.The slope of that graph indicates that the electrons can be considered to goaround the E x B loop path from 3 to 8 times before they are lost to thewalls of the system. This amount of trapping is indicative of an electronmobility consistent with Bohm diffusion, rather than classical electron diffusion in a magnetic field [3.9].The circular planar magnetron is the most widely used example of themagnetron cathode.
However, since the overall requirement is simply thatPLASMA SYSTEMS73FIG. 3.14 Circulating E x B drift current in a 150-mm-diameter circular planar magnetron deviceas a function of discharge current: (a) as a function of c h a m b e r pressure, (b) as a function of workinggas 13.91.the E x B drift path form a closed path, there are many other geometricalpermutations. A common example is the rectangular planar magnetron,shown in Fig.
3.15. This geometry is similar to the circular planar one, butis simply stretched in one direction, forming an E x B drift path somewhatlike a racetrack oval. The magnetic fields in a rectangular magnetron areusually 10-20% stronger at the ends to compensate for curvature-drivenloss processes for the electrons. Dimensionally, there is no real limit to the74R. POWELL AND S. M. ROSSNAGELFIG.
3.15Rectangular planar magnetron.length of the rectangle, and sources have been constructed several metersin length. Typically this class of magnetron is used in a linear system,where samples pass by the long sides of the cathode. Since this type ofsource design scales up so easily, this is usually the kind of magnetron usedto coat very large surfaces, such as architectural glass or large rolls of plastic sheeting.A three-dimensional variant on the circular planar magnetron uses abeveled, cone-shaped cathode surface, as shown in Fig. 3.16.
This class ofmagnetrons, known as the S-Gun class, was first developed by PeterClarke [3.10]. The magnetic field is oriented such that it is parallel to thebeveled cathode surfaces, and the E x B drift path is adjacent to the cathode surface. Larger versions of this class of sources use two nested loops,independently powered, for better deposition uniformity control.The final geometrical variant of the magnetron uses a cylindrical geometry in which the magnetic field is uniform and the cathode surface isarranged in the form of a cylinder with the magnetic field aligned with theaxis.
This comes in two variations, one with the plasma on the externalface of the cathode cylinder and the second with the plasma on the internal face (Fig. 3.17). The external-plasma version is known as the cylindri-PLASMA SYSTEMSFIG. 3.1675S-Gun class of magnetron geometries.cal post magnetron, and the internal version is known as the hollow cathode magnetron. The essential part of these designs is, again, a closed-looppath for the secondary electrons, which is now a cylindrical band in eachgeometry. The external-plasma version of the cylindrical magnetron isoften used to coat large areas, whereas the internal version is usually usedto deposit metal on fibers passing along the axis of the cathode.There are additional variants of all of these design classes of magnetrons.
For example, wedge-shaped cathodes have been designed thathave a serpentine E x B drift path around the surface of the wedge to provide uniform erosion of the cathode. Another geometrical variation is therotating cylindrical magnetron, which uses an E x B drift path that lookssomewhat like a Hawaiian lei wrapped around a cylinder, which is thenmotor-driven to provide uniformity [3. l l ].
A widely used variation of thecircular planar magnetron has a small heart-shaped E x B drift path, whichis motor-driven behind the cathode to circle the cathode face and provideuniform erosion. This design is the basis for most current-day semiconductor wafer deposition systems and will be discussed at greater depth inChapter 4.76R. POWELL AND S. M. ROSSNAGELFIG. 3.17(a) Cylindrical post magnetron, (b) hollow cathode magnetron.3.9 Reactive Sputter DepositionReactive gas species, such as oxygen or nitrogen, are routinely added tothe inert gases used during sputtering for the purpose of depositing oxideor nitride films. In the case of the DC diode, this was found (as discussedabove) to strongly alter the cathode surface, resulting (particularly for thecase of oxygen) in a much-reduced discharge current.
In the case of magnetron sputtering, due to the very high levels of metal sputtered from thecathode, an interesting behavior can be observed.A typical reactive deposition system is shown in Fig. 3.18. The systemcontains a sputtering source (usually a magnetron), gas inlets for inert andPLASMA SYSTEMSFIG. 3.1877Reactive sputter deposition system.reactive species, a pumping system, and various surfaces such as the wallsor the sample on which films are deposited. As a starting place for this explanation, consider the operation of the magnetron at a fixed DC powerwith an AI target in a background of inert gas without any reactive speciespresent.
At this point, the films deposited around the chamber are metallicand the deposition rate is relatively high.If a small amount of a reactive gas species is added m oxygen, for example m very little change is observed. This is because the oxygen israpidly adsorbed by the freshly sputtered AI films on the walls and thesample. Chemical analysis of these films would show a mostly metallicfilm with a low level of oxygen contamination. If the flow of the oxygenis increased (Fig. 3.19), very little change is observed in the system: Thedeposition rate is similar, the discharge voltage on the cathode is unchanged, and there is no change in the chamber pressure.
The only realchange is that the films are becoming increasingly more oxygen-rich.If the oxygen flow is increased farther, eventually the deposited filmswill reach a terminal oxidation level and will not be able to absorb any additional gas. For AI, this occurs at the formation of A1203. Any additionalflow of oxygen cannot be picked up by the already-saturated films, and78R. POWELL AND S. M.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
