I shouldn''t have been surprised that when I went to actually smelt the magnetic powder we recovered, the iron we recovered was far too brittle, and even repeated melt cycles with added flux didn''t fix the problem.  Twenty-two days of troubleshooting and problem solving got me a few answers.  I had previously run the powder past the electromagnetic separator a few times to get everything out, but if I instead only melted down the first pass''s powder, the iron was relatively fine.  When I melted down the extra pass material separately, the result was a moderate amount of incredibly brittle iron that was full of impurities, and a particularly dense slag material.


    As part of troubleshooting, I realized that despite how finely ground the powder was, it was actually comprised of different sized particles, and that those particles were likely comprised themselves of multiple materials.  I came to this conclusion after attempting to do a vibrational test on a container of the powder, to separate it into different density strata.  If the particles had been roughly the same size, and each particle was homogenous, I''d have expected clear layers to form.  Instead, I got three vague layers.


    It''s clear from the layers I found that there probably are at least two, but likely three, separate clusters of density within the material comprising the rock that have significant difference from each other.  However, the individual particles are probably still too large, and are made of different materials, meaning that any particle could have varying density based on it''s own composition.  Additionally, larger particles in a shaken bed can act like more dense particles, obscuring the result even further.


    My guess, which could be wrong, is that the magnetic particles are also dragging with them other bits of material, and the very pure particles experience a stronger net attraction, yielding a much purer iron.  If that theory holds true, then finer grinding would increase the total yield of good quality iron from the stone, as each particle is likely to be of a higher purity.


    The issue with testing that is that making even smaller steel balls than I''ve already made would require a lot more work, and an entirely different setup.  So, I''m left with a choice.  Either we leave it be, and end up with a volumetric yield of about 0.5% for iron, or I do some experiments with making smaller steel balls and a stirred ball mill, and see if we can increase our yields.  Since I don''t have any other pressing things to work on, this seems like a decent project.  Even if we don''t get a higher yield from the rock, having access to even finer grinding methods could give us better access to other chemicals down the road.


    <hr>


    I tried a lot of really dumb ideas for making small steel balls over the course of twenty days before I realized we''ve already pretty much developed everything we need to make it work.  I was trying things like grinding and casting, when what I should have realized would be the best option for us is actually just cold pressing steel cut from a wire into the shape of a ball, then grinding that down in a standardized way.  We already pull hot steel through a die to make wire for our cabling, so we''re already part of the way to what we need.


    I took two more days making a few test apparatuses, using a large lever arm to compress a cutting of wire down to a ball shape, then ground it down to make a nice and neat ball that was only a few millimeters in size.  After some heat treatment it became quite tough and difficult to polish, and while it didn''t have a mirror finish the abrasiveness of the particles of rock should smooth them out while in use in a ball mill.  I''ll have to work out a lot of details before we''re at the point where I can actually test that, however.


    With balls this small, I''ll definitely need a stirred mill, and the particles coming in will need to be pre-processed.  If I want standardized steel balls of different sizes, all we would need to do is up or downscale the production. For anything smaller than an inch or two in diameter this method should work.  That means we will probably need to use rock crushers to break rocks small enough to the point where we can run them through a regular mill once, filled with whatever our largest size steel balls are, then into the fine mill before we magnetically separate it.


    <hr>


    Before I try mechanizing anything, I wanted to make sure that the process would actually be worth upscaling.  If it turned out that it wasn''t worth it now, I could just pocket the project to continue later when it was actually needed.  As a result, I hired on three goblins to use my rudimentary press to make hundreds of very small steel balls a day while I assembled a small stirred ball mill and a stirling engine to hook into it to get it up to high enough speeds.


    The principle is pretty straight forward.  In a regular ball mill, as the drum rotates it brings material up the edge until it falls, and causes collisions, breaking up the material.  The limiting factors there are the kinetic energy achieved by whatever size ball we''re using, the contact surface area of those balls, and the gravitational force exerted by our planet.  What that sums up to is that you need to accelerate small balls manually to achieve even finer grinding. By essentially spinning the whole mill with stirring rods at high speeds, we can take advantage of centripetal force to increase the kinetic energy of small balls, and thus achieve that finer ground product.Stolen content warning: this content belongs on Royal Road. Report any occurrences.


    In practice, there were a lot of problems to solve to make a stable stirred mill.  I turned it vertically, and made the whole frame and drum stationary, but even then I had to stoneshape it''s frame into the ground.  The reason being that as I ran small test runs, it was clear that the medium would occasionally get off balance internally, and that would impart a decent amount of force on the sides of the drum of the mill, wanting to spin or flip the whole thing.  Other than that, the design is pretty straight forward, with a handful of steel rods rotating about a central axis to impart the energy to mill the materials.


    So, after thirteen days and a few unstable designs, I had one that I was comfortable trying at full capacity.  That capacity was actually fairly small, at only five cubic feet, but it was enough for basic testing.  So, I loaded the powdered material in from the previous milling alongside over a thousand small steel balls, and then ran the mill for a day.


    If the material going in was powder, then this was a fine dust.  When the lid was opened to the mill, the air disturbance alone threw a small amount of the dust up into the air.  I backed away quickly and exhaled as I waited for the dust to settle.  That dust is almost certainly a health hazard, given how fine it is.  In the other mill, the amount of dust it put in the air at the end was a lot less than this, so I was less concerned about long term exposure being a problem.


    In either case, after the dust settled, came another difficult step, separating the steel balls from the dust itself.  Which made me realize I should probably just fill the mill drum with water before it''s opened to both prevent dust from getting in the air, but also to let me easily separate the dust from the steel balls.  If we pour a slurry of the dust and balls through a screen to catch the balls, what comes out the other end should be just dust filled water.  That could then be dried before going through the magnetic separation process.  Speaking of that process, it will also need to be upgraded to help handle such fine dust.


    The upgrades should be fairly straight forward though.  We''ll just be installing some glass between where observations happen and the actual electromagnet does the separations.  It was already quite dusty in there, and it needed to be cleaned regularly, but now it''ll get so dusty that it''ll probably need to be cleaned daily.  A large amount of the dust is so fine that it practically floats on the air, so the glass separator should provide the necessary physical barrier to protect operators.  Though I''ll also need to make a device to allow dusting off the glass from the other side, but that should be pretty straightforward.


    <hr>


    After another five days of preparations, and working on details to make sure things were relatively safe, I went ahead with trying magnetic separations for the new dust.  What I ended up with was a much higher yield than I expected.  Nearly 15% of the dust was separated on the first go, and a negligible amount was separated on a second pass.


    However, when I melted that dust, what I got was once again a very brittle iron.  Intrigued, I went through the trouble over three days of making another batch to test.  This time, despite the potential risk to my health due to dust inhalation, I slowly moved the chute further and further from the electromagnet as the dust was poured out.  At a certain point, the dust started to separate into three batches: dust that went all the way to the electromagnet drum before falling off, dust that didn''t react at all to the electromagnet, and dust that clearly was deflected but didn''t actually make it to the drum.


    If I moved the chute too far away, the strongly attracted dust would also start missing the chute well before the weaker attracted dust would stop deflecting, meaning I had a fairly small sweet spot to try to collect all three different lines.  So, another three days, and a bit of rigging to try to collect all three different kinds, and I''d gotten a sample of the three different kinds of dust.


    The strongly magnetically attracted dust ended up making a good quality iron, which is about what I expected.  The lesser magnetically attracted dust mostly made slag with a small amount of brittle iron, and while some of the non-magnetic dust melted into a slag, some of it remained solid throughout, which might end up being useful for separating it in the future.  As for final yields, the good quality iron made up about 9% of the initial dust, the brittle low quality stuff was 6%, and the remaining 85% was non-magnetic.


    If we hadn''t ever mined actual iron ores before, 9% by mass yield wouldn''t seem that unreasonable as plenty of ores have lower yields than that on earth.  However, iron ores are usually only quite high yield at 40% or more.  With our limited resources however, a 9% yield over a large amount of rock on the island actually does seem quite good.  The next step for me is scaling everything up, and industrializing it.  If we can automate a large amount of this process, then we can actually get a bunch of our demons back to work breaking rocks, with those rocks just ending up ground down for metal, so they won''t actually need to be adjusted with stoneshaping at all.