Machine Matching

First Serial Rights, (c) 1999, Robert Widas - January 13

Machine matching can best be established when the equipment is first introduced into the work area. This is the time to test and collect the parameters that are sure to vary over the life of the tool. This document contains base-lining and troubleshooting information that will permit systematic matching between systems used in etch or deposition. (c) 1999, Robert Widas - January 13 #38744

1. Chamber Hardware
2. Gap
3. Pressure
4. Gas
5. RF

Record the following information for each machine being baselined.

Chamber

In most semiconductor process equipment the chamber is where the action is. Our goal is to force each chamber to react the same to the wafer, pressure, gas, temperature, and RF used in the process. In order to meet this requirement all hardware used in the reaction chamber must be identical. This does not mean it must be new.

• The hardware should be manufactured by one vendor.
• The hardware should be consistently cleaned.
• The surface integrity of the chamber components should be roughly equivalent.

Gap - Electrode Spacing

The gap spacing between the electrodes, or the powered plate and ground needs to be identical between machines being matched. Gap spacing will effect the amount of deposition that reaches the wafer. It will also effect the amount of material etched from the wafers surface. A window of opportunity will present itself where the deposition rate, etch rate, and uniformity are optimized. This point should be the reference for all gap adjustments. You need to ensure that the spacing is identical on all process machines. This is especially critical at the gaps process setpoint.

Pressure

All process chambers have some type of pressure control. I will illustrate a technique below that can be applied across the board on vacuum systems.

What is the lowest process pressure? (Process 1)
What is the average process pressure? (Process 2)
What is the highest process pressure? (Process 3)
What is the size of the manometer?
(You may not need to do the higher or lower pressure tests depending on the operating pressure of your system).

Calibrate all manometers to as tight a tolerance as possible using the same calibration reference for all machines.
The range of 0 through 1000 mt should be checked using a 1 Torr manometer.
The range of 0 through 10,000 mt should be checked using a 10 Torr manometer.
The range of 0 through 100 Torr should be checked using a 100 Torr manometer.
I like using a common gas to all machines for raising the chamber pressure to the appropriate setpoint. It's a good idea to use a process gas flow for this testing. You might as well get some information regarding your gate, or throttle valve position while your performing the test.

Gas

Some equipment manufacturers have software capabilities for measuring the chamber volume. They do this by knowing the chamber temperature, start pressure, end pressure, time in seconds between the start and finish pressure, and the gas flow rate into the chamber.

Gas flow = (Stop pressure - Start pressure)/

RF

The RF delivery system should be broken down into its subsystems for matching purposes. The subsystems referred to are the: generator, coax delivery cable, phase magnitude detector, coax delivery cable, matching network, and the electrode.
The RF generators used on all like machines should be the same model series. This will ensure that harmonics and other variables associated with the generator are consistent between machines.
Generally there will be a phase / magnitude detector in-line between the generator and the matching network. The purpose of the phase / magnitude detector is to drive the matching network to a position where the load impedance equals 50 ohms. This results in minimal reflected power and maximum forward power delivered to the load.
The delivered RF power to the electrode should be adjusted such that the setpoint power is delivered to the load. This is not the output of the generator. We want to check the power as close to the load as possible, preferably at the end of the RF delivery cable that feeds the matching network. This adjustment will compensate for line loss in the coax cable. The setpoint used should be the average or mean process power. A listing of other values to be recorded are contained in the table that follows.
The load termination for all RF power adjustments should be a 50 ohm dummy load. This load must be capable of handling the power from the generator.
The RF generator will normally feed a matching network. The matching network should be calibrated to the manufacturers specification.

BW Global Tech Corp. Home Page