Important Disclaimer: Chip decapsulation involves using hot concentrated acids. It should only be done by persons experienced at working with acids safely. This article is for entertainment/education please do not try this at home.
I have had readers ask to show the process I use for decapsulation, so here goes. There are only a few places on the web where amateurs are doing chip decapsulation and imaging chips. It is not tremendously difficult, but it does involve using hot acids. Something normally done in a laboratory setting within a fume hood/wet bench. Doing this safely in a home environment takes some special care.
Equipment & Safety
Here is my set-upSafety is paramount. I do this for fun and don’t want my hobby to turn into a dangerous activity. So a few comments about my set-up regarding safety
- I always use gloves and eye protection. A bit of acid spilt on your skin will give you a burn, but you only have one pair of eyes!
- It is done under a range hood with a strong fan that vents outside. A range hood that re-circulates the air is not suitable. This is critical as the hot acids will give off some nasty fumes.
- Only use small amounts of acid are used. Most of my chips are small so I find using 10 ml beakers half full works fine for nearly all the chips I look at (Less than 8 ml of acid).
- A pipette is used to transfer acid to the beaker so I am not pouring large bottles of acid.
- There is a double containment. I built a glass tray (From an old oven door) and also have the hot plate in a metal tray. Thus if I do spill a beaker of acid it is contained and easily cleaned up.
Most silicon chips in consumer products are bonded to a metal lead frame and then potted in a plastic compound. To decap them you have to dissolve the plastic away using acids. There are three acids I know that can be used for decapsulation of silicon chips, Sulphuric acid, Nitric Acid and Red Fuming Nitric Acid. Fuming Nitric Acid is the most aggressive acid and can even decap most plastic packages without needing heat. However it is very expensive, very difficult to obtain. And you really do not want to be using even small amounts of this outside of a proper lab web bench/fume hood containment.
Nitric vs Sulphuric
I have tried both Nitric and Sulphuric acids, both have their pluses and minuses
- Nitric acid can decap at lower temperatures ~110°C
- It is more aggressive etching plastics and can result in cleaner decaps. – However
- Nitric acid tends to attack metals very strongly so the bond pad area and metal leading off the bond pad gets etched away.
- It is expensive and harder to obtain
- Sulphuric acid does not attack the metals so aggressively
- It is significantly cheaper. – However
- You have to heat it to higher temperatures ~260°C.
- (Because it does not attack the bond pads as much) you are usually left with ultra-thin gold bond wires still attached that you have to pluck off the die.
I have taken to using Sulphuric in recent months. I bought the cheapest grade (Technical grade.) You can see the acid is pretty brown, it should be clear. It was like that when I bought it, the discolouration is apparantly caused by a small amount ~1% of organic contamination (Perhaps from being in a plastic container too long). For what I need it does not matter, the decapsulation process is very dirty.
I use a cover glass over the beaker to contain fumes (though it is intentionally not air tight so gases can escape though the beaker lip). The pyrex jug is for collecting the waste acid (A viscous black sludge) that I dispose of as hazardous waste. In almost a year and nearly 200 decaps I only have about 1L collected.
Here is my target chip. I usually cut the leads off. A small chip like this goes straight in; a larger die I may cut, or sand off some of the plastic to reduce the amount of work the acid needs to do.
I heat the acid up for ~5 mins until it reaches ~250°C (Not quite there yet in this shot). When you lift the cover glass, there are fumes clearly coming off.
Into the acid it goes.
After about 30 secs the acid is starting to turn black as the plastic is dissolved.
A minute in and the acid is now black.
I leave the chip in the acid at ~255°C for 10 mins. It looks like each chip needs a slightly different time for perfect results, but normally 10 mins is good.
Decant and Slide Preparation
After 10 mins, I take it off the heat and allow to cool for ~3 mins, then I drain off the liquid. and rinse in acetone (Note: adding acetone to acid is extremely dangerous, you have to ensure the beaker is cool and only a tiny amount of liquid sludge remains) and I then tip the contents out onto kitchen paper.
What you see here is small metal fragments (the lead frame and bond wire attachments) and a small die, can you see the die? If not you are not alone,this is a very small die (0.5 mm x 1.5 mm) it literally took me two to three minutes to find it amongst the lead frame and other small pieces of metal. I can barely see it without a magnifying glass.
I use a magnifying glass with light. At this stage (If I used Sulphuric acid) I usually need to pluck bond wires off of the die, using tweezers. This is a difficult process and a stereo microscope would be better, but I don’t have one of those 🙁 I use teflon plastic tweezers to avoid/reduce damage to the die.This is a process that I still need to improve at, as I frequently scratch the die trying to pull off these really thin (Typically ~25 μm diameter) wires that are difficult to grab with the tweezers. (I can barely see them in the magnifier).After this I clean the die up in a beaker of acetone sitting in an ultrasonic bath of warm water. A cheap ultrasonic jewelry cleaner for a few minutes does the job.The die is then mounted onto a microscope slide. It is important the die is perfectly clean on the underside and sits flat on the slide. To hold it on the slide I use a strip of double sided tape.
Here is the small die ready for the microscope.
It’s a metallurgical microscope (Meaning the light hits the specimen from above through the objective.) Manufactured by Olympus in the mid 1980’s. The rate of progress of optical microscopes has slowed down, with very little improvements in recent time. This 30 year old Olympus BH2 microscope likely still has better optics than any consumer level microscopes you can buy new today (Such as OMAX brand etc.). I made one change to the microscope. The lamp housing had broken off so I removed it (with its 100W tungsten filament bulb) and retrofitted an LED light source. I designed and built the LED light and it has been working very well.
The microscope has a trinocular head with a Canon DSLR camera (EOS 1300D) to capture the images. I use Canon’s EOS Utility live view software to display the image on my PC. The camera is set-up so that it is parfocal with the eyepiece objective. That means when it is in focus in the eyepieces it is in focus on the camera. Which is key to making it easy to use. There are four Olympus NeoDplan objectives 5x (NA=0.1), 10x (NA=0.25), 20x (NA=0.4) and 80x. The 80x objective has an NA=0.9 but only a 0.18mm working distance. For imaging bare silicon die is ok as they are pretty flat surfaces. The NA (Numerical aperture) is important because it is the NA that defines what you can resolve not the magnification. You can zoom an image taken with 5x objective lens to the same magnification as one taken with 20x or 80x objective, but you will not resolve the same features because the NA is lower.
My die photos are typically imaged with the 10x or 20x objectives. They are a mosaic of images that are stitched together. I take between 10 and 120 images depending on the size of the die and objective used. Between images the stage is moved manually. I would like to acquire an automatic x-y stage that has servo motors moving the slides more efficiently.
To stitch the images together I use Hugin which does a good job but is not so easy to use. Ken Sherriff published a very useful tutorial for how to use Hugin to stitch die photos. Once assembled I then use another good but complex software tool, Raw Therapee to adjust contrast and tone. It can also do a Richardson-Lucy deconvolution which I have found to work quite well at sharpening my silicon chip microscope images taken with the lower NA objectives.