Skip to main content

Sputtering targets: purity, erosion and the running cost nobody budgets

How to read a target certificate rather than the headline purity, why a bonded target fails when you raise the power, how much of a target you actually get to use, and what reactive sputtering control really depends on.

Consumables budgets for a deposition tool usually cover argon, wafers and pump oil. The line that gets left out is targets, and for a group working in platinum, gold or iridium the targets can cost more over five years than the tool did.

Worse, the target is the component that quietly sets film quality. A tool that is behaving badly is often a tool with a tired, contaminated or wrongly specified target in it, and no amount of adjusting the process recipe will fix that.

Purity, and what the extra nine buys you

Targets are sold by nominal purity: three nines, four nines, five nines. The number is a summary and it hides the thing you usually care about.

Bonded or monolithic

A target has to conduct heat out of its face into the cooled backing, or it cracks. How that thermal path is made is a specification point that gets skipped, and it decides how much power the tool can actually run.

Magnetic materials are their own problem. Nickel, cobalt and iron short the magnetron's field, and a standard target thickness in a standard gun will simply not strike a useful plasma. The fix is a thin target, a magnetron designed for magnetic material, or both. Suppliers quote a pass-through flux figure for magnetic targets, and it needs to be checked against the gun before the order goes in rather than after.

The racetrack, and how much target you actually get

A planar magnetron confines the plasma into a closed loop over the target face, and that loop is where erosion happens. The target does not wear down evenly. It develops a groove, the racetrack, while the centre and the rim stay close to full thickness.

The consequence is that a large fraction of an expensive disc is never sputtered. How large depends on the magnet design, on the target thickness, and on whether the magnet assembly moves. Rotating or scanned magnet designs spread the erosion and improve utilisation, and they are worth paying for when the target material is precious.

Flaking is the other consumable question nobody plans for. Material builds up on shields and on the chamber wall, and eventually it delaminates and lands on a substrate. Sputter-down geometries suffer more than sputter-up. Removable shields that can be cleaned or replaced on a schedule cost less than the wafers lost to a single flake event.

Reactive sputtering and the hysteresis problem

Reactive sputtering, running a metal target in a mixture of argon with oxygen or nitrogen, is how most useful nitride and oxide films are made. It is also the process most likely to be blamed on the tool when the real issue is control.

The target has two stable states. In metallic mode the surface is clean metal, the sputter yield is high and the rate is high, but there may not be enough reactive gas to fully form the compound. In poisoned mode a compound layer has formed on the target face, the yield collapses and the rate falls, though the film is fully reacted. Between them is an unstable region where a small change in reactive gas flow flips the target from one state to the other, and the transition does not happen at the same flow going up as going down. That loop is the hysteresis.

A reactive recipe that is not under closed loop control is not a recipe, it is a coincidence that happened to work on the day it was written.: Common experience on shared deposition tools

Budgeting for targets in Bangladesh

The running cost of a sputter tool is dominated by three things: what the target material costs, how much of it you use rather than throw away, and how often you vent.

Back to all Insights