Process Information

Common Mistakes

Sticking to the standard format does not guarantee the approval of your process flow. After checking the format, you should take a closer look at whether your wafer cleanliness level complies with the Process Verification Scheme. In addition to this, you have to consider whether your processes will damage any of the equipment that you intend to use. This page lists some of the common mistakes that users make.

1. Oversimplification: Although such vital processes as resist coating, striping, development, hardbake, softbake, descum, HMDS priming and wafer cleaning are too obvious to be omitted, people still manage to forget them. Contamination due to the omission of one or more of these processes accounts for more than half of the contamination incidents in the NFF (CWB).

Can you spot the mistakes in the following example?

Hint: How is the photoresist coated on the wafer?

2. Wrong initial wafer status: You cannot declare the initial wafer cleanliness of, say glass, as "CLEAN", because it has been defined as Non-standard in Section B of the Process Verification Scheme. However, a lot of people start off on the wrong foot by making this mistake. If only they had read this page earlier, their flows would not have been rejected immediately after the first line was read.

Do you know how to correct the following process flow?

Process Flow of LED

Substrate: Ga Nitride Wafers, Dimensions: 2 inches in diameter, xxx thick

Hint: Once the initial cleanliness has changed, every thing else has to be changed accordingly.

3. Incompatible equipment cleanliness level:Section C of the Process Verification Scheme clearly tells you how to evaluate the wafer cleanliness level. Having read this section, you should have no difficulty in figuring out which machine can be used in the next step and which one is incompatible with the current step. A lot of rejected process flows reveal that many users are ignorant of this point.

What's wrong with the following process flow?

Hint: Changing one item may mean changing all subsequent items.

4. Wrong photoresist stripping process: In the NFF (CWB) there is more than one way to strip photoresist. Which process should be used depends on the cleanliness of your wafers. If your wafers are Clean, no doubt you can use sulfuric acid to strip the photoresist. Yet if your wafers are Semi-Clean, the process to use will depend on the type of materials on your wafers. Different materials require different treatments. MILC wafers and wafers without metal or silicides are fine with sulfuric acid. But wafers with standard metals and silicide must be processed with MS2001 rather than sulfuric acid, as acid will attack the metals, extract the metallic ions which will in turn contaminate clean wafers.

What change should be made to the following process flow?

Photoresists that have undergone high-temperature treatments such as implantation, long-time plasma etching, and high-temp hardbake cannot be stripped clean by wet processing alone. In addition, an implanted photoresist has heavy cross-links with ions. So it is strongly suggested that users perform an O₂ ashing step right before wet processing.

5. Misuse of decontamination: Decontamination is not a panacea. Not every sample can be decontaminated, as clearly stated in Section E of the Process Verification Scheme. If your wafers are non-standard, don't even think about decontaminating them. Only certain semi-clean wafers can be decontaminated. To determine whether your wafers are "decontaminable", try answering the following question (we call this the decontamination test):

1.Are your wafers classified as unclean simply because they have undergone CMP and post-CMP cleaning?

If you answered "yes" to the question, congratulations! You can decontaminate your wafers.

Do you know how to correct the following process flow?

Hint: Simplify the process flow into two steps instead of three.

6. Misuse of low stress nitride as oxidation masks:The physical properties of low stress nitride are different from those of normal nitride. In terms of thickness, normal nitride is no match for low stress nitride, but one of the drawbacks of low stress nitride is that it is likely to crack as a result of thermal expansion. Serious contamination could occur if low stress nitride is used as oxidation masks.

7. Double-side photoresist coating: The SVG coater track and developer track cannot perform double-side coating, as the photoresist on the other side will contaminate the chucks. So if you insist on coating both sides, you must use manual coaters. But once you have done so, the wafers must never be processed with Clean equipment. Also you are not allowed to use certain Semi-Clean equipment like the CF-3000 implanter to treat double-side coated wafers. There was an incident where a wafer got stuck on the chuck of the implanter and could not be unloaded because the photoresist on back of the wafer had glued wafer and chuck together. Finally if you intend to perform double-side coating, please state it clearly in your user application as it is not immediately apparent from the process flow.

8. Wafer edge coating: : If you decide to paste photoresist around the wafer edge to prevent it from being etched, that piece of wafer must never be treated with Semi-Clean or Clean equipment again. That is because any wafer with its edge covered in photoresist is prone to contamination.

9. Omitting the backside rinsing step before implantation and dry etch:When you use photoresist as the implantation/dry etch mask, make sure you perform a backside rinse before implantation/dry etch to remove the photoresist from the ream and backside of the wafers. Otherwise, the photoresist will be left in the handler of the implanter/dry etcher which will lead to contamination.