Ask any chemical engineer what’s the most important process variable, and they are likely to reply, ‘temperature’. When we plasma etch a wafer, or deposit a layer by PECVD, is the temperature just as critical? In plasma processes, most of the chemistry is driven not by the surface temperature, but by the electron temperature in the plasma. The electrons behave like a separate gas, co-existing with the neutral gases flowing into the chamber, but only weakly coupled to that gas. Because the electrons are charged particles, they pick up energy from the applied RF voltage, which they shed in collisions with neutral particles.
These collisions drive chemical reactions by splitting up molecules into highly reactive radicals. The electron temperature is in the range 10,000 – 40,000K, which is so far above the wafer temperature that the wafer temperature is less important. Energy is also supplied to the surface by ion bombardment, where the impact energy lies between 20eV – 1000eV, far above the thermal energy of atoms, even if the wafer is at 1000K.
But that’s not the whole story. Surface processes include:
• Adsorption of gases and radicals
• Surface diffusion
• Reaction
• Desorption of reaction products
The reaction process is dominated by the plasma energies, and desorption can be stimulated by ion bombardment. But the surface temperature strongly drives adsorption, diffusion and desorption, especially on sidewall surfaces which are not strongly bombarded by ions.
While measuring and even controlling the wafer temperature may be the holy grail, the industry currently works only on the table temperature under the wafer. The wafer temperature then depends on the heat flux (either heating or cooling the wafer) and the degree of thermal coupling between the wafer and the table. Radiation coupling is weaker at lower temperatures (especially with silicon, which is fairly transparent to infrared below 500K). At process pressures below 1 Torr, there is little conduction through the gas between the table and the wafer, so OIPT offers ‘helium backside cooling’. In this technique, the wafer is clamped to the table (either by electrostatic clamping or by a mechanical clamp), and 5 -20 Torr of helium gas is maintained behind the wafer, with pressure control and flow monitoring. This pins the wafer temperature close to the table surface, even in the presence of high heat fluxes from the plasma.
In PECVD applications, the process pressure is high enough to deliver similar temperature differences between table and temperature, without needing the extra heat transfer gas feed. We have shown that it is very beneficial to raise the pressure after loading a wafer, to improve heat transfer and decrease stabilisation time. This is especially true if a carrier plate is used, because of its higher thermal capacity. Without this step, we have shown that the plate temperature deviates from the table temperature by more than 10ºC. With a 2 Torr stabilization step, the plate temperature settles within 30ºC of the table temperature.
Summary
Wafer temperature does matter, even in a plasma process. A reproducible thermal stabilization history is necessary for a reproducible substrate temperature. A well-characterised heat transfer environment is essential.