Thermite

Lessons Learned

Clear PLA burns out completely when ramped to 500 C at 1.5 degrees C per minute.
Brown dirt is mostly not Iron Oxide, and is basically useless for casting. If your thermite burns for more than ~1-2 seconds, it won't work.
A mix of fine ($$) aluminum and a mix of less fine ($) aluminum works OK in at least a 50/50 mix. Using only cheap 5 micron aluminum seems to impact the yield as the burn time is too long and heat is allowed to escape.
A single sparkler stabbed into the mix works well for ignitition. If a sparkler won't ignite it, it is probably crap thermite, and won't work for casting.
Air trapped in the thermite is bad. Incrementally hard pack as much as possible to decrease atmospheric oxygen losses. Do not use a binder.
Natural clay is OK as a funnel inside a flower pot crucible, in fact, the decomposition probably acts as a lubricant to keep iron globules from sticking and not exiting the mold.
Use a lid on the crucible to contain the splatter. Not only is it safer, it redirects splatter back into the melt and increases yield.
A few layers of aluminum foil or aluminum tape on the bottom of the crucible works well as a final exit gate for the molten iron.
Mold materials matter. Firing of the mold materials beforehand matters. A blowtorch is not a kiln, and will not remove all of the PLA or water from the mold.
Playground sand works OK as a support structure for the mold, but one must take care to avoid getting sand into the mold.
Once the mold cools to ~300C, shock using a water bath or the mold will be nearly impossible to remove. Re-heating with a blow torch before dunking is probably fine.

Problem Statement

I don't really know how to start this one. The problem is of a very personal nature. A man, Larry Campbell, whom shared many of the same passions as I, passed away in January of 2018, from the same flu that also had me confined to bed for several weeks. The 'problem' is that a good friend died. A good friend who often annoyed me. This fact is hard to come to terms with. The last time I saw him I probably said something snappy to him, because I didn't always approve of how he handled shared resources. A bitter pill to swallow, as I had and have enormous respect for the man. My intent with this project is to create a memorial using a small amount of Larry's ashes that is worthy of such an amazing man. I only wish he could have joined in while he was still alive.

To truly honor someone's life, use their death as a reason to grow. Learn new things. Help someone out. Explore something you never would have otherwise. Grief is powerful, use it in a positive way and do something they would have been proud of. Otherwise it will consume you and fill you with regret.

Larry was extreme. He loved metalworking. He loved the C64 in his youth, and he loved all methods of high tech fabrication. His primary frustration was that 3D printers don't easily print metal. There are many ways of doing lost PLA casting to aluminum -- but what Larry wanted was steel.

I want to use thermite mixed with a small amount of Larry's ashes to cast a Japanese style dagger with his C64 BBS named, Banzai, embossed on the handle. I want to 3D print the dagger, create a mold, then fill it with liquid steel from thermite.

Log of Experiments

Plaster and Dextrin (fail)

Initially and naively I created a mold from plaster. I used a 3D printed calibration cat as the test piece, and added three vents by welding 1.75mm PLA filament to the high points of the print. I submerged the print into 50% plaster and 50% playground sand. I used a pen to create the main sprue about 0.5" long. The PLA was removed by placing the cured plaster mold upside down in the oven at 500 degrees F with a ramp rate of 50 degrees every 10 minutes.

50g of thermite using laboratory grade Fe2O3 and laboratory grade purified atomized aluminum was mixed with 2% dextrin and pressed into a 3" flower pot with a pencil sized central vertical cavity. An aluminum foil diaphragm covered the bottom of the terracotta pot. The dextrin was added in an attempt to create a solid 'brick' without air as well as a source of carbon for the creation of steel in stead of raw iron. A small amount of metal poured into the mold, but it was clear that the mold was mostly empty, so a second batch of 100g of the same mixture, but not moistened, was placed into a second flower pot and ignited. The mold was still not full, but was destroyed to asses the yield. A total of 35 grams of iron made it into the mold, and it was clear that the inlet was too small and several droplets of iron sat on top of the main sprue.

The lower half of the calibration cat's shape was apparent, but massive amounts of frothing in the iron were observed as well as an extreme sulfur stench similar to the smell of a used ested model rocket engine tube. The sulfur was a byproduct of the plaster decomposing due to temperature, over twice the decomposition point. It was clear that plaster was not an appropriate mold material.

Thermite Yield Experiments

I decided at this point to attempt to quantify the amount of iron that should be expected in an attempt to efficiently size the charge to fill the mold. Stochiometric analysis shows that Fe2O3 is 70% iron. Therefore, the yield should be 70% & 75% = 52.5% iron. In other words, 100g of thermite contains 52.5 grams of iron, which is clearly a theoretical maximum. The thermite-to-mold yield from the previous experiment, 35 grams, is 150*.525 = 78.75g, 35/78.75 = 44% of the theoretical maximum, which honestly is not too shabby considering the iron particles that remain in the slag and adhered to the sides of the terracotta pot crucible.

Cut Star Thermite

The amount of iron beads covering the crucibles from the previous experiments led me to believe that the iron was pooling on top of unreacted thermite, which was ejecting the puddle as it reacted. Therefore I attempted to create 'channels' for the iron to flow towards the bottom by creating 50g of 1/4" thermite cut 'stars'.

Pressed without dextrin

It became clear to me that the dextrin itself may have been the source of the gas which ejected the iron puddles. Therefore, 50g of thermite was pressed using a 3D printed espresso tamper into the bottom of a flower pot as densely as possible. The results were the best yet.

Frustrations with commercial 'red iron oxide'

As my supply of laboratory grade Fe2O3 dwindled, I purchased commercial grade red iron oxide. It became IMMEDIATELY clear that the purity of the commercial grade iron oxide was nowhere near as good as the lab grade. The commercial grade sample refused to sustain combustion when pressed densely. When fluffed, it combusted extremely slowly, taking over 10 seconds to finish, whereas the lab grade materials burned in a fraction of a second. The iron yield for the longer duration burns was negligible, likely due to both the stochiometric offset from the desired as well as a limited maximum temperature due to the extended duration of the burn. Upon inspection under a microscope, it was clear that the samples contained sand and clays. The color was also much more brown than the lab grade oxide, likely also due to quartzes and clays. Apparently selling iron rich dirt qualifies as pure iron oxide....

DIY Iron Oxide

One can make Fe(II) or Fe(III) effectively in an electrolytic cell. The use of sodium chloride and a voltage high enough to allow bubbles from the cathode only, which can be any material such as aluminum or another piece of iron will create iron hydroxide at the anode which will precipitate out as an orange Fe(III) or green Fe(II) gel. The gel must not reach the cathode or it will begin to plate metallic iron onto it and contaminate the gel with iron bits. Therefore, a tall skinny reactor is preferable to a short wide one. A large flower vase was used. The gel was harvested using a large dropper, reacted with a few mL of peroxide to convert all of the Fe(II) to Fe(III) and filtered using pre-rinsed notebook paper. Notebook paper has incredibly small pore sizes, which allows for efficient capture of all of the gel particles with are minuscule and pass through most filter papers. It is also considerably cheaper than filter paper, though the flow rate is only a few drops per minute. One must wait several hours for the initial capture and rinse. Effective rinsing is important to remove excess sodium chloride salts from the gel.

The gel is then heated to 90 degrees C until the dries, exhibiting an extremely large degree of volume shrinkage, then heated to >200 C and held until the material turns dark in color, driving off the waters of hydration and converting the hydroxide to oxide. The resulting oxide must then be pulverized using a pestle or mill.

Analysis of the purity can be performed by dissolving 100mg of the oxide in hot hydrochloric acid or oxalic acid. Iron oxalate is only weakly soluable in water and must immediately be decanted or the yellow precipitate will obscure any insoluble contaminants.

Needless to say, the process is time and effort consuming, and the yield is dismal.

Hope Appears! Ceramics Quality Red Iron Oxide

Creating iron oxide in the lab is a large effort low yield process when going directly to an Iron Hydroxide Gel. The density of the gel is quite low, 8 oz is only a gram or so. Asking around it seems the local ceramic supply store carries multiple 'colors' of Iron Oxide for $6 a pound. The red stuff was VERY red, which is a great sign. It also extremely 'cakey' which indicates very fine particle sizes, also a good sign. Sure enough, a 50g test yield was about 12g of the possible 26g theoretical which is not as good as the lab grade, but certainly better than using dirt based ores.

A 200g test using the ceramic store red iron oxide and 50/50 lab grade aluminum and internet 5 micron aluminum was conducted. A natural clay from my backyard was used to form a funnel in the bottom of a flower pot. It was not fired, only dried at 200 degrees F overnight. The thermite was packed solid with an espresso tamper and ignited with a sparkler. The kiln wash based mold was filled with 55.6g of iron. 61.2g total was recovered (some spilled outside of the mold). This is 58% of the maximum possible yield. Not as good as the lab grade, but significantly better yield than the ore, and at a price point that makes the difference irrelevant.

Mold Process Progression

Furnace Cement

Furnace cement seemed promising at first, but it seems that even the 2700 degree cement decomposes to gassy and glassy byproducts at thermite temperatures, and plastic deforms like modeling clay at red hot temperatures. It is not a good material for coming into direct contact with the thermite reaction or the molten metal itself. It is a good last layer for structural support during the other parts of the process.

Pure Kiln Wash

Kiln wash is designed to never fire to a hardened state in a ceramics environment. It accomplishes this in a two-fold manner. The major constituent of its composition is Aluminum Hydroxide. When fired to ceramic temperatures this decomposes to Aluminum Oxide, which is quite unreactive as well as extremely high temperature, something like 3000 C. There is no danger of sintering or vitrifying aluminum oxide in a standard ceramic environment.

During the dehydration process, a great deal of water is released between 300 and 500 degrees C. Something like 50% of the molecule is water. This leads to a large amount of shrinkage and subsequent loss of strength associated with the new porous oxide, which is helpful for keeping things from sticking to other things in a ceramics kiln.

Kiln wash is promising because it is so extremely high temperature, once we can solve the shrinkage and strength challenge. The shrinkage can be mitigated by ramping extremely slowly, at around 1.5 degrees C per minute to 750 degrees C, but the mold still lack strength and will flake off at detailed areas.

WARNING: Pouring molten steel or other metals into a kiln wash based mold that has not been fired to >500 C will result in rapid steam formation and ejection of the molten mold contents!! The mold must be chemically dehydrated prior to use!

Kiln wash and Grog 50/50

Mulcoa 200 Grog seemed promising to cut the kiln wash with to keep the shrinkage down. The process works extremely well for aluminum and other lower temperature metals. It almost works for the thermite process, but mullite, the major component, melts around 1800 C. The mix seems to plastic deform at temperature leading to a blobby low pass filtered look on mold details. No 3D print lines are visible in the final distorted product, but also no boiling or glassing is observed either. Progress, though not really good enough.

Kiln wash and Fired Kiln Wash 50/50

Kiln wash cut with 50% pre-fired kiln wash seems to be the next logical step. Pre firing half of the material will keep the shrinkage under control, but the ultimate mold strength may be less. A binder such as sodium silicate might be the answer, but might also act as a flux to lower the melting point. A higher firing temperature may be necessary with an oil based binder to maintain mold integrity during the dehydration temperatures <500 C.