To receive quarterly process tips directly to your inbox, subscribe to our Ask a Process Geek newsletter. To view previous tips and answers to your process questions, check our process tips archive.
Q: What is the recommended process for curing my specific polyimide?
A: The process is easy with the right equipment. First lets define a polyimide... Polyimides are high temperature engineering polymers utilized by semiconductor manufacturers for their excellent mechanical, thermal and electrical properties. Polyimides are usually applied in liquid form and then soft-baked to set the film. After the polyimide has been imaged and developed, it is ready to be thermally cured. The cure cycle activates the desired properties of the polyimide. While temperatures may change depending on the particular polyimide, the techniques developed to achieve an ideal film are universal. Now, let me go into detail on what the curing process for polyimide is doing. First, the polyimide needs to be cured in an oxygen-free environment. During the cure process oxygen will interfere with the cross- linking of the polyimide and result in a weak film. Our YES-450PB Series of tools first purge oxygen from the system through a series of vacuum pumpdowns and nitrogen backfills. Then a controlled temperature ramp commences, which provides an ideal process for curing almost any polyimide. I prefer slightly lower ramp rates when running processes which decreases stress in the film of polyimide. Many cure recipes include "Holding" a process at a temperature around 200°C for an indefinite amount of time allows the solvent in the film to evaporate. Since the YES-450PB tools cure at a reduced pressure, the solvent is being drawn out of the film from the beginning of the process. Therefore the hold step isn't necessary, especially if the ramp is slowed down slightly.
After the solvent has been removed from the film, the next step is to fully imidize the polyimide. This is the conversion of the film from polyamic acid to polyimide. The ramp continues, usually until the film reaches the glass transition temperature. The temperature will be held at 300-400°C for 30-60 minutes depending on the polyimide. The product is then cooled to allow for safe removal from the system. The cooling rate can be much faster than the heating rate because the characteristics of the film are set.
This is a standard application for the YES-450PB. The system can be programmed to follow the manufacturer's recommended cure cycle exactly or however it is appropriate for the application. For instance, when YES ran a bond pad re-routing using multiple layers of polyimide as the dielectric, after the first layer, we could not exceed the glass transition temperature of the polyimide on subsequent layers. Consequently, Layer 1 was cured to 400°C, Layer 2 was cured to 360°C and Layer 3 was cured to 350°C.
Yield Engineering Systems can work with your engineers to determine which tool within the YES- 450PB Series will work best for what you are trying to achieve.
We invite you to contact YES to troubleshoot your specific process.
Is it possible to control the sidewall angle in the image reversal process? I worry that I am getting too much undercut. I made 2.5 micron wide lines with image reversal, and the structures were entirely undercut and released in the middle following the image reversal process.
The wall angle can be controlled using an ammonia process to neutralize the exposed resist followed by a flood exposure. I have experienced angles from +22 degrees to vertical to -22 degrees. The trick is balancing the process.
Using X as the original exposure to define a 1:1 image, the profile can be tailored by varying the amount of exposure. Underexposure leads to an undercut profile, while overexposure can lead to a rounded profile.
To troubleshoot a process, I recommend doing an exposure matrix. Vary the exposure from 50% to 200% of your normal imaging exposure. The next step to try is lengthening the ammonia exposure. This is a diffusion reaction with the exposed resist and cannot be overdone. A typical process is 45 minutes, so try 60 - 90 minutes. The flood exposure also cannot be overdone, so be sure to give the resist a 2X - 3X exposure time. Then develop. Looking at undercut images under a low power microscope, you should be able to see dark lines at the edge of the features (view image).
7/24/2008
Regarding a mask for ICP (dry etching), is it possible to do a quick bake above 100 ºC then cool it down with N2 gun and immediately spin coat HMDS?
The trick is to stop any chance of moisture before HMDS is applied. The easiest way to achieve this is using vacuum/hot nitrogen dehydration, followed by an HMDS vapor while the product is still in a vacuum dehydration chamber.
7/24/2008
What is an effective way to do lift-off with resist AZ5214E?
Using image reversal to create a negative slope resist line allows production control down to 0.1 micron metal lines.
7/24/2008
What is the process for manufacturing semiconductors?
The manufacturing process requires the following steps:
Producing silicon wafers from very pure silicon ingots (boules)
Slicing the boules into wafers and polishing
Using multiple oxidation, photoresist etching doping steps to fabricate bipolar integrated circuits (ICs) onto the wafers (front-end processing)
Metal Oxide Semiconductor (MOS) devices only use one doping step, i.e. N channel or P channel MOS devices
Complementary Metal Oxide Semiconductors (CMOS) have N channel and P channel on the same chip
Assembling all ICs onto the wafer in a finished product
Testing and back-end processing the finished products
7/24/2008
How can I improve adhesion between SU-8 2015 and a glass wafer? I am making a mold for electroplating Cu. Up until the developing process, it looks there is no adhesion problem, but when I electroplate Cu, the SU-8 mold layer peels off.
As SU-8 is an epoxy based resist, it is probable that an epoxy based silane adhesion promoter would work better. Also, to promote adhesion, try vacuum dehydration before deposition.
7/24/2008
What are copper pillars?
In a flip chip, it's common practice to use solder balls. These have a very high lead component and less conductivity than the conductive metals copper, silver, etc. If copper pillars are substituted for solder balls the metal is more conductive and the amount of lead drops to less than 1%. The 1% is caused by solder tinning the tops of the copper pillars to improve adhesion of the copper pillar to the chip.
7/24/2008
What is copper capping?
Copper capping is the practice of inserting a barrier between the copper and the surrounding dielectric. When the copper is annealed some of the copper molecules tend to diffuse into the dielectric reducing the insulation. If a cap is placed in between them the copper can not diffuse into the dielectric.
7/24/2008
How does HMDS bond to a wafer surface?
Hexamethyldisilizane (HMDS) {NH-2Si(CH3)3} removes the H on the end of a Hydroxyl ion, removing 2 Hs breaks down the NH bond of the HMDS. This liberates NH3 (ammonia), and leaves Silicon tri-methyl attached to the {2} old hydroxyl bonds.
7/24/2008
What is photoresist?
Photoresist is a liquid that, when exposed to light, changes its structure. Positive resist becomes mildly acidic and a basic solution, TetraMethylAmmoniumHydroxide (TMAH), dissolves away the exposed resist. The remaining resist is baked and becomes resistant to acid and/or plasma etching steps. A negative resist crosslinks and/or becomes harder when light hits it. The unexposed resist is washed away using a solvent that the resist material is carried in.
7/24/2008
I see water used for some silane depositions and not others. What is the role of water in silane coatings and what is best?
The YES-1224P can do different "water injection" processes depending on the chemicals and process being used. I will tell you the different uses and how each one works.
1. There are two chemical structures we are concerned with. The first is hydroxyl ions (-OH groups) and the second is surface bound water (H2O). The hydroxyl ions are necessary for the reaction of the silane and will form a strong bond on the surface. Surface bound water is weakly bound to the surface and if the silane bonds to it, will form a weak coating on the substrate. When we do a dehydration step in the process, we are trying to remove surface bound water only. Surface bound water evaporates at about 150°C (at 1 atmosphere of pressure, it evaporates at a lower temperature with lower pressure). Hydroxyl ions are bound to the surface up to about 450°C. The big difference in energy makes it easy to keep hydroxyl ions and remove water.
When we perform a plasma surface clean, we may remove both water and hydroxyl ions. When we do a "water injection" step, we are trying to restore the hydroxyl ions. So one way to use the YES-1224P is to complete a three-step process:
1) Plasma clean (using oxygen or argon plasma).
2) Pump down and inject 0.1ml H2O. Pump out the water vapor.
3) Inject silane.
If the silane functional group can react with water (such as an amine or epoxy silane) it is important to remove the water from the system before injecting the silane.
2. If the silane is non-reactive (such as a perfluoro silane or a simple alkyl silane) then the following process can be used:
1) Plasma clean (using oxygen or argon plasma).
2) Pump down and inject silane. Delay 5-10 minutes. Inject 0.1ml H2O.
With this process, the water reacts with the trichloro - or trimethoxy end of the silane and helps the reaction of the silane with the substrate.
The YES-1224P allows the engineer to choose the correct sequence of steps to give the BEST results. We can work with your engineers to determine the best process for your substrates and coatings.
