The paper presented below was written by J. R. German (Society of Vacuum Coaters - 38th Annual Technical Conference Proceedings, pg. 414). It details the use of magnetic "shunts" to alter and enhance the performance of traditional magnetron systems. Using computerized magnetic FEA analysis, Soleras has incorporated this modification into many existing coating systems. Customers that use the modified design enjoy a stable plasma impedance over target lifetime and an increase in target utilization because of decreased trenching on the target surface. Call for more information about upgrading your own coating system.

also see: Soleras Magnetron Enhancement Page

A Simple Low Cost Method for Increased
Performance of Planar Magnetron Sputtering Targets

INTRODUCTION
An important cost concern in thin film sputtering is the performance of the targets. The production engineer would like to increase the life of each target and reduce variations in process parameters related to the target. It is an added bonus if these improvements can be developed and implemented at little cost. Discussed in this article is a method for improving the performance of planar magnetron sputtering targets by use of a ferromagnetic shim which is placed between the magnet assembly and the target. The shim acts as a magnetic shunt which favorably modifies the magnetic field at the surface of the target leading to the advertised improvements in target performance. The purpose of a magnetic field in a sputtering plasma is to increase the efficiency of ionization by capturing electrons emitted from the target to increase the rate of collisions between electrons and neutral gas atoms. The lack of uniformity of the magnetic field produces a non-uniform plasma density, hence differential sputtering rates across the surface of the target. It is obvious that increasing uniformity of the magnetic field will improve the uniformity of erosion of the target.

  FIGURE 1

A - UNSHUNTED DESIGN
Standard magnetron and target configuration. Effect of the magnetic field is maximized where field lines are most parallel to target surface.

B - SHUNTED DESIGN
Steel shim shunts normal path of field lines causing the field to become more uniform near target surface resulting in improved sputtering performance.

EXPERIMENTAL DETAILS AND RESULTS
One way of increasing uniformity in a planar magnetron system is by use of a ferromagnetic shim which partially shunts the magnetic circuit in the region of the target surface. The effect of the shim is to reduce the curvature of the field at the target surface as shown in figure 1. Part A shows the shape of the original field and part B shows the field modified by placing the shunt between the target and backing plate. The shunts where originally made of cold rolled steel and sandwiched between the target and the cathode. They were later made from low carbon iron, for better uniformity, and bonded into recesses cut into the cathodes. The dashed lines indicate erosion patterns for the given magnetic fields. The field lines are from computer generated models which were verified by physical measurements. The erosion patterns are from mechanical traces of used targets which where run for equal kilowatt-hours. Note the direction of the magnetic field lines relative to the erosion contour in each case. Since the effect of the magnetic field is greatest where it is most parallel to the target surface, it is apparent that the shimmed target has greater potential for extended usefulness than the unmodified target since it is not as deeply eroded and the curve of the field more closely follows the contour of the target surface. The depth of the trench for the shimmed target is 80% of that of the un-shimmed target and there was about 6% increase in utilization of the targets original material. This seemingly modest change at least doubled the usefulness of each target. With the original assembly, targets had to be changed twice weekly, during system cleans, to prevent target burn-through and expensive unscheduled shutdowns. The modifications allowed the targets to last for at least two system cleans instead of just one. For some applications, the greatest benefit may be an increase of machine utilization due to less frequent shutdowns for target changes. An additional benefit with the shunted target is that power parameters are more constant over the life of the target. This is illustrated in figure 2 which shows Voltage levels of four targets as a function of run number. The left half of the graph is of four unmodified targets. The right side is of two modified and two unmodified targets run together in the same chamber at the same power settings. The discontinuity represents a change of targets. The more consistent voltage level indicates a more stable process. As targets erode, trenching ensues. This trenching confines the most active plasma to an increasingly narrow region and sputter yield decreases. Power must then be increased to maintain desired yield. In the voltage plot, an increase in power is seen as an increase of voltage for the modified targets, but a continued decrease is seen for the unmodified targets. Voltage levels decay over the life of the shimmed target as well, but since they do so more slowly, power settings need be altered less frequently. Thus, a more stable process and a more consistent product is realized. In addition, with reduced trenching there may be a reduction in spitting or sputter splatter. There is some thought that the splatter effect is due to high intensity sputtering in the trench. However, there are no statistics to confirm or deny this speculation.

  FIGURE 2

 

 OTHER ADAPTATIONS
This technique of modifying magnetrons with a shunt was developed for use in a large in-line sputtering system with rectangular targets. It could be easily adapted to other planar targets that use static magnetic fields produced by permanent magnets or DC electromagnets. Although the design work would be best done with the aid of computer modeling it is possible to design a useful shunt without the computer. One can cut a shim to a best guess shape and size, put it in place, and measure the field with a gauss meter. Plotting the component of the field which is parallel to the target surface as a function of position, one can see whether the physical model has the desired structure. In fact, this procedure was used to confirm the results of the computer model before using a modified target in production. For most targets, especially those with axial symmetry, the computer modeling can be done with relatively inexpensive two dimensional software. More complicated geometries may require the much more expensive three dimensional software. There are several commercially available packages which vary in cost and user friendliness. Although the basics of the software may be easily mastered, it may take some time to develop the intuition needed to design and optimize magnetic circuits. Better results may likely be obtained in different target assemblies. It is likely that the example shown here may still be improved. Note that the field lines are parallel to the sputtering surface across a wide area about midway through the modified target in figure IB. This suggests that a thinner target may have a better erosion pattern yet and allow equal or greater target utility from lower cost targets. It also suggests that target assemblies with different relative positions of magnets, shim, and target may allow for even better results in many cases.

1. J.R. German, “Magnetron Shunt for Enhanced Performance of Sputter Targets", IBM Technical Disclosure Bulletin, 36(11), (1993)

1. Thanks to Dieter Meyer of IBM in Mainz, Germany for providing useful data.
2. Thanks to Alan Plaisted and Soleras Ltd. for encouragement and assistance in preparing this paper.