Improving sputtering cathodes and targets using simulation
PlasmaSolve has executed several projects on simulating target erosion. The customers always fell into one of two categories.
OEMs who build sputtering cathodes either as a product or for their own coating systems
Manufacturers of sputtering targets, especially for precision applications in semicondoctor or optical coatings
The need to understand and improve erosion is often driven either by an emerging application (e.g. coating larger wafers with smaller features) or by the need to reduce OpEx of the process (i.e. increase target utilization for high-value materials)
There can be various motivations for wanting to understand and modify the erosion pattern of a sputtering cathode. So far, the PlasmaSolve team has encountered the following types of challenges:
The thickness uniformity on the sample is not sufficient (typical goal +-5% on a glass or +- 1% wafer)
The target utilization is too low (below 40-50%)
The cathode is suffering from over-erosion in some locations, which shortens the lifetime and increases burn-through risks
The magnetic field geometry in this case study is synthetic to protect sensitive IP of our customers.
Results and Benefits
The solver calculates the erosion groove based on the magnetic field, the operating target voltage or power, target material and the gas pressure.
The benefits arising from an erosion groove simulation often include:
Adjusted layout of magnets leading to more uniform erosion profile and deposition profiles
New design of curved or otherwise profiled targets which have a higher material utilization during lifetime
Understanding how the erosion is influenced by process settings - gas mixture, pressure or sputtering power.
The image on the right illustrates how the cross-corner effect of a fixed magnetic field changes with pressure. The corss-corner effect gets stronger with increasing pressure, which is consistent with experimental observations.
Pressure-dependence of target erosion for the magnetic field shown in one of the boxes below. Note the corss-corner effect apparent for the higher pressure values
Magnet layout inside the cathode which serves as an input to the simulation. The image shows a slice through the computational geometry with blue and red being the magnets, green being the target and grey the actual computational domain where electrons and ions are traced.
Magnitude and field lines of the magnet arrangement above.
The inputs check-list for the target erosion model includes:
Burning voltage (or expected range)
Burning power (or expected range)
Relevant gas or gases
Pressure (or expected range)
The only input needing explanation is probably the magnetic field. In the ideal case, a STEP-file containing the magnets assembly and the target position is provided and PlasmaSolve calculates the magnetic field.
A second-best alternative is a data file (csv or dat) with simulated magnetic field data on a regular grid (recommended grid spacing 2 mm). This allows the customer to share the magnetic field but not the exact magnet arrangement.
If neiter of these is available (e.g. for target manufacturers), a measurement of the magnetic field at the target surface is used as an input and PlasmaSolve reconstructs the position and strength of the magnets by simulating various combinations and fitting them to the measurements.
The target erosion model relies on tracing charged particles in self-consistent electric fields and the external magnetic field using the MatSight Target Erosion App.
The MatSight Target Erosion App supports circular, rectangular and cylindrical targets and even targets with steps or ridges at the surface.
The computation times are reasonable - even simulating a cathode for glass coating (over 3 meters long) takes no more than several hours.
In technical terms, the procedure is as follows:
Step 0: Magnetostatic field is computer, initial estimate of plasma density is made and cathode sheath electric field calculated.
Step 1: Electrons are emitted from the target and traced through the ExB fields.
Step 2: Ionization events are recorded and the ions are traced back to the target surface
Step 3: Electric field is corrected and a new iteration is started (see step 1).
Gas pressure and composition
Burning voltage and power
MatSight Target Erosion App
The term cross-corner effect denotes over-erosion of a sputtering target near the bends of the racetrack. It is a surprisingly non-linear phenomenon, depending on the rate of change of the magnetic field in the racetrack bends, the applied voltage and the operating pressure in the sputtering range.
Cross-corner effect is undesirable because it shortens the lifetime of the target. A 10% variation in erosion rate may seem non-critical but it actually shortens the lifetime of the target and reduces the material utilization by the very same 10%.
There are three main strategies for tackling the cross-corner effect:
Finding a working point and magnetic field where the over-erosion is negligible. However, that typically requires to sputter at lower pressures, which may not always be a good fit.
Designing a magnetic field, where the cross-corner effect appears only on one half of the target. That requires the target to be flipped upside down when it reaches 50% of its lifetime but does not restrict the pressure range so much.
Adjusting the erosion by shaping the anode. This solution is also rather pressure-dependent.
The MatSight Target Erosion App, combined with PlasmaSolve's experience with magnetic field design, can help you find solutions or workarounds for this parasitic phenomenon.
Example of a target erosion map at 1 Pa pressure
Erosion shape in the transverse cut through the target - along the red line
Erosion shape in the longitudinal cut through the target - black line
Standard target and 2 iterations of "profiled targets" achieving higher material utilization.
Profiled Targets Design
For most sputtering cathodes, target utilization is below 40%. This is not a major problem for standard materials such as titanium, aluminum or chromium where the end-of-life target can be easily scrapped and re-cycled.
For more expensive materials, target shaping may be a viable way of increasing the target utilization. Changing the shape of the target inevitably influences its deposition characteristics which makes target erosion simulation a suitable tool for designing various target shape modifications.
Segmented Targets Investigation
One of the cases we have applied the model to more than once are segmented targets. Segmented targets allow deposition of alloys from a single target without having to alloy the target.
One of the reasons why segmented targets are used is related to film composition - the atomic content of individual components can be fine-tuned by changing the size of the segments.
Another reason for using a segmented target is the fact that the alloy of the two (or more) materials may not be stable enough or disintegrates when heated up.
In any case, the simulation helps our customers design the optimum layout and size of the segments, limiting the number of experimental iterations. It can also assess the change in the atomic composition of the deposited coating during the target lifetime which is an effect that emerges especially for combinations of materials with very different sputtering yields. The change can be as much as 10 at.% during the target lifetime and can be mitigated especially by designing segments with minimum circumference which still satisfy the coating uniformity goals.
Illustration of a segmented target at the beginning of the lifetime (top) and later during the process (bottom). The segments often form "steps" which impact the atomic composition of the coating.