Magnesium is a one of the most common and non-toxic elements on earth.
In sea salt magnesium is the second most common metal with about 3.7% mass, wedged between spot one sodium 30,6% mass and and spot three calcium 1.2%. Potassium takes spot four is 1.1% mass.
Beside the rare and poisonous beryllium magnesium is the only earth-alkali metal in which the naturally forming oxide/hydroxide layers are dense and tight enough to prevent further reaction with water and air. Mechanosynthesized oxide/hydroxide layers should perform at least equally well.
Magnesium oxide might be one of the most interesting compounds for advanced atomically precise manufacturing. Common elements, non-toxic, no damage of nature when spilled, high thermal stability, decently hard, simple crystal structure, ... .
MgO aka magnesium oxide aka periclase aka fused magnesia:
- MgO is naturally occurring as the mineral periclase. It rarely seems to form bigger clear crystals and is rarely used as gemstone due to its sensitivity to water and lack of very high hardness (green seems to be a common color from impurities).
- Synthetic colorless dense (and sometimes clear) probably polycristalline pieces of MgO go under the name "fused magnesia". This material seems hard to come by in small quantities for collectors.
- High quality colorless single crystals wavers (probably grown via CVD processes) with controlled crystal orientation are extremely expensive (as of 2018).
Dense MgO is ever so slightly water dissolvable (over the detour of its hydroxides/carbonates).
Sidenote: The analogous compound with calcium CaO (commonly called quicklime in its technical use for concrete) is extremely reactive with water.
(TODO: Find out if there have ever been made solid clear crystal pieces of CaO and if there are images.)
Salts of oxo-acids
- Magnesium silicate: @ mineral enstatite:  (Mohs 5-6 - decently hard) (it's a the magnesium pyroxene end-member, other pyroxebne endmembers are calcium and iron)
- Magnesium carbonate:  @ mineral magnesite:  (Mohs 3.5-4.5 - semi soft)
related is dolomite (calcium magnesium carbonate)  (Mohs 3.5-4.0)
- Magnesium hydroxy silicate: Talc  (Mohs 1 - defining mineral - very soft)
- Magnesium sulfates (soluble salts): Epsomite Mg(SO4).7H2O, Kieserite Mg(SO4).H2O, Langbeinite K2Mg2(SO4)3
Just like calium one can get rid of the volatile oxoacid parts (or the hydroxy parts) by heat treatment. This process is called calcination. Since the heat to drive out the volatile parts is not sufficient to melt the product one ends up with a very fine (and thus reactive) powder. This is magnesia MgO. (Related: Sorel cement .) While such thermodynamic processes are not really relevant for constructive mechanosynthesis this seems relevant for conventional destructive thermal recycling of products created via mechanosynthesis. See: "Diamondoid waste incineration".
Salts of Halogenides
- Magnesium clorides (soluble salts): Bischoffite MgCl2.6H2O, Carnallite KMgCl3.6H2O,
Notes on sealing
All but the most hydrogenated/carbonated magnesium compounds oxidize hydrogenate or carbonate on the surface when exposed to weather on earth.
Unlike with other alkali and earth alkali elements this process stops from a macroscale perspective. But on the nanoscale this is still very destructive. Disrupting the first atomic layers very fast in form of significant crystal structure geometry and volume change that would destroy all nanomachinery. After a while of exposure oxide/hydroxide/carbonate naturally forming passivation layers even become visible by human eye. Meaning the thickness of the layers lie in the wavelength of light. That is 100th of nanometers or eqivalently thousands of atomic layers.
It might be possible to mechanosynthesize passivation layers that are much tighter than the ones that naturally form, such that a few atomic layers suffice a a passivation that stops any further reactions. Unfortunately mangesiums passivation minerals (hydroxides/carbonates) (the ones one wants facing the outside of products) are rather soft minerals. So deep scratches are common. At those scratches natural "thick" oxidation layers will start to from. So it might be more desirable to choose other materials for the outermost surfaces. (Maybe magnesium silicate aka enstatite or another pyroxene? Compatible biodegradability plays a role in the choice.)
Well isolated deeper inside gem-gum-products harder weather sensitive magnesium minerals can be used. Especially MgO Periclase seems like an interesting option.
Notes on Biodegradability
Magnesium minerals are very nontoxic, just ever so slightly water soluble when in bulk form and not to hard, which makes them excellently biodegradable (or better bioerodable?).
Mechanosyntesized products with voids inside that are lighter than water and thus when escaped from recycling may swim on the oceans like plastic long enough to be ingested by animals, will be quickly dissolved by gastric acid.
Since magnesium is stronly electropositive (stronger than common transition element metals like iron) it wins out in thermite reactions (getting the oxygen) and conversley magnesium oxide is less prone to dissolution in molten transition metals (or metal rich silicate melts aka lavas / magmas).
For very speculative applications check here: Warning! you are moving into more speculative areas. deep drilling#Earth core probes
- Magnesium Minerals: