Sandia III-V Wafer Transfer Process
This process was observed by Will Doyle at CINT 1/7/25 - 1/10-25 for the purpose of bringing it back to MRC to replicate.
The process is designed to transfer III-V MBE grown epilayers onto a transparent substrate (sapphire) for subsequent NLO testing. This has to be done because some of the response we are interested in observing is absorbed/obscured by the presence of the substrate.
The outline of the process is as follows
Cleave III-V and sapphire pieces to squares
Glue them together in the correct orientation
Lap away almost all of the III-V
Finish with alternating acid dips to work through the etch-release and etch-stop layers until only the epi grown layers remain glued to the sapphire.
More details of the process can be found below. Please note that this process has been transcribed as exactly as possible from Sandia - LASErs may subsequently modify it to accommodate alternate tools or supplies that we have available to us. A similar process can be found in the wiki page on laser thinning.
1. Sample Cleaving
Supplies required: III-V epi material, sapphire 10mm squares, tweezers, acetone, methanol, IPA, N2 gun, scribe, sample boxes, cleanroom wipes
Before beginning, obtain 10mm C-cut sapphire squares. These can be custom ordered, diced, or cleaved, although cleaving sapphire is tricky. You will likely want to scribe along the whole wafer and use cleaving pliers.
At a suitably ventilated bench, cleave the III-V samples into 11 or 12 mm squares. This sizing is fairly important - you want them to be a bit bigger than the 10mm sapphire squares.
Feel free to mark the backside of the III-V pieces with a small scratch - it will get lapped away anyway, and makes your life easier in case you flip the sample over or something by accident.
Clean the III-V pieces with acetone, methanol, IPA and blow dry.
Securely store the samples until ready to proceed with the next step.
2. Epoxy
The epoxy used at Sandia had the following part number: 353ND/4GM/BIPAK. A cursory look online seems to show it available at Digikey as of January 2025. This is an epoxy that is specially designed for optical use, so if you choose to replace it, perhaps look for an epoxy that is similarly designed.
Supplies required: III-V squares, sapphire 10mm squares, SOFT TIPPED tweezers, acetone, methanol, IPA, N2 gun, sample boxes, hot plate, timer, scissors, small plastic epoxy dish, glass slides, very fine tipped epoxy spreader (toothpick sharpness), cleanroom wipes
Clean everything again with acetone, methanol, IPA (both the III-V squares and the sapphire squares), as well as your tweezers. Blow everything dry. Pay special attention that there are NO particles on the sapphire or sample so that the bond will be smooth. The entire epoxy step should either be performed in a cleanroom or a well ventilated hood. If you are using a wooden toothpick for your epoxy transfer implement that won’t be allowed in the cleanroom, so work in a LASE hood or something.
Start heating the hot plate to 85C (actually confirm it gets to 85C, don’t just trust the dial!)
Place n glass slides onto the hot plate, where n is the number of samples you are planning to transfer.
Mix the epoxy - when I observed this, the epoxy was split into two parts in a single pouch. By applying pressure from a sharpie, we were able to break the “inner seal” and mix the two solvents to activate the epoxy. Mix considerably inside the pouch. The epoxy should not be very viscous, and should be well mixed before cutting open the pouch fully.
Cut open the epoxy pouch and pour into the plastic epoxy dish.
Put your first sample epi-up on a cleanroom wipe. Dip your toothpick (or equivalent) into the plastic dish to get a drop of epoxy on it, and transfer that drop to the center of the III-V epi square. A picture can be seen below for the relative amount - I was told that this was borderline too much epoxy. If you have too much epoxy, it will overflow the side of the sample after you make the “sandwich”, and glue it to the glass slide.
Using your soft-tipped tweezers, place a sapphire square on top. Gently press down using your tweezers, and slide it around a bit to get full coverage of epoxy between the III-V and sapphire.
Transfer the sample onto one of the glass slides that has been heating up on the hot plate. The heat will aid in the following epoxy spreading steps.
Continue applying pressure, sliding the sapphire around to spread out the epoxy. You should start to see interference fringes as the epoxy things - this amount of epoxy should thin out to about a 1 micron layer.
Continue spreading and applying pressure until you see 4 or fewer interference fringes - the fewer the better, since that means the coating is more uniform.
If you did put too much epoxy and some drips down onto the glass slide, you will want to lift up the sample every 20-30 minutes during the epoxy cure to try and keep it from fully gluing itself to the glass slide.
Repeat all of these steps for your other samples.
Once every “sandwich” is made and on a glass slide on a hot plate, leave for at least 30 minutes. Increase temperature to 120C and leave for at least 30 minutes. Increase temperature to 150C and leave for at least 30 minutes.
After the final curing step, remove the samples, allow to cool, and securely store until ready to proceed with the next step. When I observed this process at Sandia, this is where we stopped after day 1.
3. Lapping
Supplies required: Samples, lapping tool, lapping pads, drop micrometer, wax, hot plate, lapping chuck, sample boxes, tweezers, Opticlear, acetone, methanol, IPA.
Prior to starting, check your stack and the thickness of your samples (sapphire thickness and III-V thickness, ask the grower and/or confirm the thickness of the wafers that were ordered). Once you have those thicknesses, you will want to check the thickness of your entire stack to see how much thickness the adhesive has added.
Set hot plate to 80C and put lapping chuck on there
Pick up a small piece of wax and wipe it along the hot chuck to melt a bit. Using tweezers, mount the sample III-V up on the chuck, and press down to improve adhesion and promote levelness. If the sample is not level, the lap will not be level either. Remove chuck and allow to cool.
Check height of the whole stack using the drop micrometer(if you know your III-V thickness and sapphire thickness, this should give you your epoxy + wax thickness).
Now, calculate the height you will have to lap to in order to only have 15µm of substrate left. For us at Sandia, the entire stack was 1140µm tall (500µm sapphire, 15µm adhesive, 625µm III-V), and we were aiming to lap down to 530µm (500µm sapphire, 15µm adhesive, 15µm III-V)
Load the lapping tool with a 15µm grit lapping pad. Set to 25RPM, start water flow, and start lapping down. Stop frequently to check how much has been lapped, want to get down to 50µm of III-V remaining. When checking height with the drop micrometer, check in multiple places on the samples to confirm it is level and being lapped evenly.
Once there is around 50µm of substrate left, swap in a 9µm lap wheel. Set to 20RPM, and lap until there are 15µm left. This is the most sensitive part of the lapping process, stop and check the lap amount often so that you do not do too much.
Once only 15µm remain, set in a 3µm lap wheel, 15RPM, 3 minutes exactly. Check thickness.
Finish with 1µm lap for polishing, 15RPM, 55 seconds. Check thickness - you should hopefully be around 520 or so, with just 5µm substrate remaining after the polish.
You are now done. Set the stage back on the hot plate to melt was again, remove the sample, put in opticlear for 1 minute before cleaning with acetone, methanol, IPA and blowing dry. Store samples securely until ready for the final step.
Note, these steps should probably be performed in a well ventilated hood, and waste dust should be handled as III-V waste.
4. Acid Etch
This is the final step in the process. For convenience, an example stack is shown below to help explain the purpose of this step.
As you can see, there are four layers marked for etch release. At this point, there are only ~5µm of substrate left beneath the stack. Recall that the substrate is actually the “top” side of the sample at this point, the epi layers are epoxied to the sapphire. We will finish the process by acid etching the remaining substrate, GaAs etch release layers, and one the thick AlGaAs layer.
Supplies required: Samples, 2x stir plates, , 2x 400mL glass beakers, 2x stirrers, assorted graduated cylinders/measurement beakers, cleanroom scale, citric acid, phosphoric acid, hydrogen peroxide, sample dippers
The two acid mixes are as follows
Citric Acid Mix: 100g Citric : 100mL H20 : 32.5mL H2O2. Stir at 160 RPM
Phosphoric Acid Mix: 20mL H3PO4 : 200mL H20 : 4mL H2O2. Stir at 200 RPM.
The acid mixes can be kept for 5-6 weeks. Note that this is not typically done at MRC, but if you do, handle with care, label and date all containers, make sure they have tight lids. While these are not as dangerous as HCL or HF, it is still acid.
After obtaining all of your plates, glassware, and acid, set up the stir plates in an acid hood. Pour your water, then acid, then peroxide into the beakers. Leave to mix for 15 minutes.
In the mean time, you can load your samples into sample dippers. Sandia used ones that held the squares vertically, unsure how important this is, whether a sample dipper basket would suffice. Individual dippers do allow you to remember/distinguish between separate samples, which aids in processing multiple samples at a time.
After the 15 minutes, dip sample(s) into citric mix. After 10 minutes, you should observe the III-V side being roughened as the citric acid etches away the GaAs. After ~1 hour, the surface should be shiny again, and it may start appearing red/transparent. This means we have reached the AlGaAs etch stop layer. Allow to etch 10 additional minutes after reaching this point. Note: The AlGaAs stop layer is not a perfect stop - you can’t forget about the sample and leave it here for hours.
After the 10 minutes have elapsed, remove the sample(s) and double rinse in DI water, carefully blow dry.
Dip in phosphoric for 1 minute, 40 seconds. Remove and double rinse in DI water, carefully blow dry.
Some notes are in order here - the phosphoric etch of the AlGaAs layer is the most finnicky part of this whole process. The etch rate is approximately 1.2nm/second, but this can fluctuate. Some calibration may be necessary. If the AlGaAs layer is thicker than 120nm, adjust dip time accordingly. If the ratio is not 0.55 Al, the etch rate may also vary.
Finally, we will etch the last 100nm GaAs etch layer up to the second AlGaAs etch stop layer. Dip sample(s) in citric for 30s, remove and double rinse, blow dry.
At this point, the sample is done and ready for characterization. You can tell which side is the sapphire side because you will see the epoxy fringes when looking at it from that side. So if you can see epoxy fringes, flip it over, and now it is III-V up.
Reach out to Will Doyle with questions - it is likely that this process will require some adjustments for UT.