I was thinking that on small scales (sub-meter, especially sub-millimeter) it might be more economical to reduce metals from ores by aqueous electrowinning than by smelting, because maintaining large thermal gradients is very difficult.
If the things being constructed are themselves small, the strength of materials is not very important, because at small scales even very weak materials are strong enough to hold together except at very large accelerations. Metals, however, have some other interesting properties: they can conduct electricity, they have very low vapor pressures and so can withstand exposure to space, and they can be readily shaped by electrochemical machining.
Macroscopically, hardness is very important for abrasion or cutting, but I suspect that these shaping processes, like sliding-contact joints, will not be very usable at small scales because of the rapidity of surface wear and the comparatively large forces involved in surface contact. However, at scales above where this is true, hardness is still important, because it determines what can cut what else.
Casting and molding are also very important shaping processes at the human scale. At submillimeter scales, the same thermal problems that impede XXX
pH, CO2, H2O, O2
pressure
Consider the Ellingham diagram for iron, which shows that smelting iron requires a temperature of at least 700°, more practically 1000° or more. If outside the smelting apparatus the temperature is 25° then we have some 975° of temperature difference. If we have a meter of refractory insulation, that’s 975 K/m. Vermiculite’s insulating value is about 16-17 K m/W, a conductivity of about 0.06 W/m/K, giving about 60 W/m² with that gradient, a heat flux which is, in the steady state, uniform throughout the thickness of the material. Aerogel is about three times as good, insulating firebrick about three times worse, and most other insulating materials are in between. The worst insulator of all, diamond, is about 1000 W/m/K.
Now suppose we scale the apparatus down by a linear factor of 1000, cutting the insulation thickness to 1 mm. Because the thermal gradient has increased by a factor of 1000, we are now losing 60 kW/m². This poses a real difficulty inside the apparatus. Because the surface area covered by insulation has increased by a factor of a million (say from 600 m² to 600 mm²) so we are dissipating only one thousandth as much power as before to maintain the kiln at smelting temperature; but the volume over which that power must be generated has diminished by a factor of a billion (say to 1 milliliter), requiring a million-fold increase in the power density of our heating elements, say to 35 W/ml, which is achievable but problematic. Scaling down further rapidly becomes impossible; at 1 micron thickness we are losing 60 MW/m², which for a 10-micron cube amounts to 36 milliwatts.
It also poses a difficulty outside the apparatus, because removing 60 kW/m² requires either radiation at an uncomfortably high temperature (“60 suns”, as they say for solar concentrators) or a lot of coolant, but this is a less serious problem.
Scaling in the opposite direction, we would reach a point where even shitty insulating materials would thermally insulate adequately.
High-temperature processes are possible in a low-temperature environment at the micron scale if they can be carried out very quickly and intermittently. For example, a cubic micron of material, weighing on the order of 5 picograms, can be heated to 2000° for a short period of time with an energy on the order of 10 nanojoules. It cools off through conduction with on the order of a milliwatt, so several milliwatts is required to reach this temperature, which then cools off on a timescale on the order of a microsecond. Lasers and electron beams are straightforwardly capable of being switched with submicrosecond timescales and delivering such power densities.
Resistance heating is also straightforward. 10 milliwatts at 10 volts is a milliamp and thus 10 kΩ. If our joule heater is amorphous carbon at 6 × 10-4 Ω m, a 1-micron cube of it would give us 600 ohms; we could either increase the current to 4 mA and reduce the voltage to 2.4 volts, or we could increase the aspect ratio of the heating element, but either way it seems clear that we will have no trouble reaching the desired temperatures on the desired timescale with easily constructed circuitry.
Electric arcs and pseudosparks are another candidate method for achieving such temperatures rapidly enough.
By contrast, no metal needs as much as ten volts to reduce it. Common insulators have electrical resistivities of 1011 Ω m and up, the best ones exceeding 1023 Ω m, while conductors are in the neighborhood of 10-9 Ω m. Ten volts across a millimeter of a 1015 Ω m substance like sulfur or dry wood produces a current of about 10 pA/m² and thus 100 pW/m², almost 15 orders of magnitude less than the thermal leakage calculated above through a thermal insulator for the temperature needed to smelt iron. If the linear approximation for conductivity were accurate this far down, a 1-nanometer-thick layer of such an insulator would permit only 0.01 mA/m² (and 0.1 mW/m²) of conduction. In fact breakdown voltage becomes a much more significant concern than energy loss to conduction; fused silica can withstand some 500 volts per micron, but other materials are closer to 10, so they’d need over a micron of insulation. At small scales vacuum becomes the best choice of insulator, since most metals don’t suffer field emission until over a gigavolt per meter, which would be 0.01 microns of insulating vacuum.
A micron-thick wire at 10-9 Ω m has 1.2 kΩ/m of resistance, which is an almost entirely insignificant 1.2Ω/mm. So electrical transmission is not perfectly efficient but it does not pose feasibility problems for micron-scale electrowinning in the way that thermal conductivity does for micron-scale carbothermic reduction.
The eight ancient metals are iron, gold, copper, lead, tin, silver, mercury, and, in India, zinc. Today I think the most important metals are aluminum, iron, copper, zinc, tin, tungsten, nickel, chromium, lead, cobalt, molybdenum, vanadium, magnesium, titanium, platinum, gold, zirconium, and the semimetals carbon and silicon. LME’s “non-ferrous” category includes aluminium, copper, zinc, nickel, lead, tin, aluminium alloy, NASAAC (“North American Special Aluminum Alloy Contract”), “aluminium premiums”, alumina, and aluminium scrap; “precious” is gold, silver, platinum, and palladium; and “EV” (“electric vehicle”) is cobalt, molybdenum, and lithium.
Of course many other metallic elements are widely used, in an oxidized form, such as calcium, sodium, and potassium, and there are niche uses of almost all of the metals. But what I’m mostly concerned with here is reducing metals from their oxidized form.
Aluminum is resistant to corrosion in air, nearly as abundant as iron, and although it is not as strong as steel per volume, it is stronger per weight, much easier to shape, and more conductive per mass than copper. It also has an astoundingly high boiling point, 2470°, and an extremely useful oxide. 65 million tonnes are mined per year, and it costs about US$2/kg.
Unfortunately, there is no known way to electrowin aluminum in an aqueous solution; metallic aluminum has a -2.33-volt standard electrode potential to reduce to hydroxyls, while hydrogen is only -2.23 volts, so aluminum will steal oxygens from hydronium. Instead aluminum is electrowon by dissolving alumina in cryolite Na3AlF6, which requires a temperature around 1000°; neat cryolite melts at 1012°, but the eutectic is only 960°.
Of my list of “important metals” above, magnesium, titanium, and zirconium have the same problem, but the others should all be electrowinnable with low-temperature processes.
Alternative processes for reducing aluminum might include plasma electrolysis, mass spectrometry, electron-beam reduction in vacuo, and simple carbothermic reduction using intermittent heating.
Iron is one of the most abundant and strongest metals, and it can withstand moderate heat (1500° or so without oxygen, much more than aluminum or brass, though not in the same ballpark as sapphire, graphite, tungsten, molybdenum, etc.). It’s the main metal used for construction and machinery, having mostly displaced the more expensive bronze and brass as the humans improved their techniques for shaping the more stubborn iron. A couple billion tonnes of it are mined per year, and I think scrap iron costs about 25¢/kg (US$213/ton in 02020).
Electrolytic iron is commercially used in cases that require especially high purity or small particles, such as cereal fortification, powder metallurgy, or high-coercivity powdered-iron magnetic cores.
US Patent 4,134,800 from 01979, by Prasanna K. Samal and Erhard Klar, describes one process, using a bath of ferrous sulfate (36-40 g/l of iron ion) and ammonium sulfate (24-28 g/l of ammonia ion), with 1.4-1.6 grams of iron per gram of ammonium, a pH of 5.6-6.0, a temperature of 38°-49°, and 18-26 amps per square foot (194-280 A/m²), which they say isn’t critical. Their declared aim was to make the iron more brittle so it could be ground, which they hoped to achieve by iron hydroxide formation. As a “prior art bath” they gave as an example 50 g/l ferrous ions, 13 g/l ammonia ions, pH 5.4, 38°-43°, 22 A/ft² (237 A/m²). They carefully didn’t mention their voltage, electrode spacing, agitation, aeration, electrolytic cell size (1 liter or 1 tonne?), or Faraday efficiency, and they didn’t mention any other additives, which hopefully they didn’t have.
If you had sulfate, you could presumably digest iron ores with it and then follow this process. In fact, you could probably continuously digest iron oxides in the sulfate electrolysis bath.
Samal and Klar cite patents 2,464,168 (Fansteel, 01949), 2,481,079 (Chrysler, 01945), and 2,626,895 (Fansteel, 01944). A little further searching turns up patents 1,782,909 (Pike, 01930), 2,464,889 (Pike and Schoder, Tacoma Powdered Metals, 01949), 2,503,235 (Cain, Sulphide Ore Process Co., 01950), 1,162,150 (Estelle, 01915), 2,538,990 (Trask, Buel Metals, 01951), 3,041,253 (Audubert and Lacheisserie, 01962) and, for nickel, patents 3,414,486 (Nordblom and Bodamer, ESB, 01968) and 483,639 (Strap, 01892).
The Estelle patent is particularly interesting for being over a century old and claiming to make iron pyrite an economic source of iron, which it is not at present (though the name of Cain’s company above suggests it used to be). He was electrolyzing ferrous chloride, formed by digesting the pyrite with muriatic acid, and then recycling the resulting ferric chloride solution into muriatic acid and ferrous chloride by reducing it with sulphuretted hydrogen (produced in the first step), producing sulfur as a byproduct. He says that nickel, cobalt, and zinc can be co-precipitated with the iron, but the zinc is easily enough driven off.
Cain’s patent is especially helpful in telling us that at the time (01946) there were two main processes for electrodeposition of iron, one involving the dissolution of an iron anode and one that doesn’t (because it’s digesting an oxide or something similar); and that usually you use an asbestos anode bag to contain the crap formed on the anode. He says it’s good to keep the pH below 2 with muriatic acid. (You’ll pardon me if I prefer polyethylene or polyester to asbestos.)
Audubert and Lacheisserie (concerned with fine particle size) say you can use most ferrous salts, but sulfate and chloride are best, and that if you’re getting oxidized iron, either you have oxygen dissolved in the bath or you have too much ferric iron, and that they use 0.65 volts.
Anyway, so it seems like it’s slightly tricky, but not nearly as tricky as you’d assume from the negative standard electrode potential of iron. And I guess it would have to be not that tricky for Edison’s nickel-iron battery to be rechargeable.
While iron is crucial for moderate temperatures and strength, the much less abundant copper is crucial for electrical conductivity, low-friction bearing surfaces for iron parts, corrosion resistance in oxygen atmospheres, and high thermal conductivity for heat exchangers. 25 million tonnes are produced per year; it costs US$6.20/kg.
Copper is so easy to electrodeposit (and electro-etch) that it’s easier to enumerate the cases where it won’t work: where you’re trying to form an adherent deposit on an electrode that copper will spontaneously oxidize, such as iron, and when the anions in your electrolyte don’t form a soluble copper salt (among the usual suspects, these are iodide (mostly), cyanide (without enough ammonia), thiocyanate, hydroxide (i.e. bases or just water), oxalate (again, without enough ammonia), and phosphate). The USGS says that there are currently 3 electrolytic refineries for copper in the US and 14 electrowinning facilities.
Zinc is used to add corrosion resistance to iron in oxygen atmospheres (its main industrial use today), in Zamak, as an alloying element for copper to form brass, and in its oxidized form, as a white pigment. It has a remarkably low boiling point, 907°. 12 million tonnes are produced per year; it costs US$2.40/kg.
Despite the name “galvanization”, zinc coating was originally done not as electroplating but as a hot-dip process, which is still the most common way to do it today. But electroplating zinc is also a common thing to do, and there’s lots of historical work on producing zinc powder electrolytically.
“Zamak” is a family of low-temperature zinc-based casting alloys, some of which have strength comparable to steel; Zamak 2 (4% aluminum, 2.7% copper, 0.04% magnesium) has a tensile strength of 330 MPa, a Young’s modulus of 96 GPa, and melts over the range 379-390°. Unfortunately the aluminum is a necessary component, and slight lead impurities will wreck Zamak with zinc pest.
In modern practice, brass (about 20% zinc, US$5.40/kg) has mostly been displaced by steel, which is stronger, harder, stiffer, lighter, and cheaper (more than 20× cheaper by weight), and, in high-carbon cases, can be hardened by heat treatment. But brass still has many small-volume niches.
It is enormously easier than steel to cast or, especially with a bit of lead, to cut.
It’s more corrosion-resistant in oxygen atmospheres and in water, especially salt water; “admiralty brass” is 70% copper, 29% zinc, and 1% tin (see below) and is an especially good formulation for this.
Brass has higher thermal and electrical conductivity than steel, and so in particular it lasts much longer for EDM electrodes.
It has much lower friction on steel than steel does, so it can be used for plain bearings (journals), as a cheaper and less durable alternative to bronze (though babbitt is often better still).
It’s used as a solder to join steel parts (“brazing”), which allows a stronger connection than bolts, with lower temperatures and less distortion than welding, and it can join a wider collection of materials than welding, including tungsten carbide (see below).
Because it’s softer than steel, brass doesn’t produce sparks and doesn’t mar steel surfaces, so in some environments and for some purposes brass hammers and other tools are preferred to steel.
Finally, its yellow color is often used for aesthetic purposes. With just zinc and copper, you can make silver (zinc), red (copper), and yellow (brass).
Galvanized steel, steel coated with zinc, has mostly replaced tinplate as an anti-corrosion coating. Zinc is somewhat toxic in food (the oral rat LD50 of the highly soluble zinc chloride is 350 mg/kg, and it’s also used topically to induce skin necrosis in “black salves”) and produces toxic fumes when heated near its boiling point, so this isn’t done for tin cans or cooking pots, but it’s widespread for things like buildings. As mentioned above, this is usually done as a hot-dip thing, but it can be done through electrodeposition.
Tin is crucial for soldering electronics; alloyed with copper it is bronze; alloyed with copper and antimony it is babbitt; coating steel it prevents corrosion; and it melts at only 232°. The largest of its many uses today is as a nontoxic anti-corrosion coating for steel in “tin” cans. Bronze can withstand both higher temperatures and more stress than brass, while retaining brass’s easy castability. Babbitt, which makes the best plain bearings, is tin with 2.5-5% copper (occasionally as high as 8.5%) and 4-8.5% antimony. Some 0.3 million tonnes of tin are mined per year, and it costs about US$18/kg.
You might think its numerous oxidation states (2+ (stannous) and 4+ (stannic), sometimes + and 3+, as well as neutral and negative states) would make it difficult to electrowin. The sulfate, bromide, chloride, and fluoride, all divalent, are water-soluble; the iodide is mildly so, and the bromide is additionally soluble in donor solvents like DMSO. There are also a tetravalent bromide, chloride, fluoride, iodide, sulfide (sphalerite), and nitrate; the tetravalent chloride is a liquid that mixes with all kinds of nonpolar liquids, and the tetrabromide is also water-soluble. The nitrate is, unusually, unstable in water. The sulfate is preferred when stannic ions are undesired, because there is no stannic sulfate.
Tin electroplating is widely practiced using acid baths (I’m guessing sulfuric), alkaline baths (I’m guessing stannate; you can get sodium stannate by digesting tin with lye), and methylsulphonic acid baths. It’s often codeposited with lead, copper, silver, zinc, and/or bismuth.
Tungsten has the highest melting point of any metal (3422°), almost as high as carbon’s sublimation temperature of 3642° and the melting points of tantalum hafnium carbide (3990°), tantalum carbide (3880°), and hafnium carbide (3928°), though well short of tentative results for hafnium carbonitride (4200°). Tungsten also has the highest boiling point of all elements, an astounding 5930°. It’s an essential ingredient in high-speed steel, though vanadium and molybdenum can replace it to some extent, and tungsten carbide (the main current use of tungsten) has largely replaced high-speed steel in modern steel-cutting practice. It’s also essential to TIG welding and important in vacuum tubes and incandescent lights. Some 84000 tonnes are mined per year, 80% in China, but I don’t know what it costs.
Carbides of vanadium, molybdenum, niobium, and the titanium-group metals are possible substitutes for tungsten carbide.
The current industrial process for smelting tungsten is long and involved, but the main article of commerce is tungsten trioxide, which is then either carbothermally reduced or reduced with hydrogen.
Experiments have been made in electrowinning of tungsten at 1080°, but also US patent 2,384,301 (Harford, 01944) and others describe electrodeposition methods for reducing tungsten. Harford recommends complexing your tungsten with 25% ethylenediamine in water, using 25 A/ft², but he explains that people previously just used cyanide.
I think low-temperature electrowinning of titanium, zirconium, and hafnium is basically a lost cause with current electrochemistry. This is a real shame, because titanium is as strong as iron and much lighter.
Perhaps even more interesting than the metals, though, are the carbides, nitrides, borides, and oxides of this group, which are outstanding materials in many ways: ultra-high temperature ceramics, superhard, transformation-toughened, solid electrolytes, photocatalysts, super-high-kappa dielectrics, resistant to chemical attack, high-conductivity semiconductors, etc. They are often produced from the metals, but for example zirconium diboride can be made from refined zirconia, boria, and metallic magnesium, or from boron and zirconia, or boron carbide and zirconia. Nitrides can be made by reacting the oxides with ammonia or nitrogen, etc.
However, of the oxides, only titania (rutile or anatase) occurs in nature. Zirconia (mixed indiscriminately with hafnia) is obtained from zirconium silicate (zircon or jargoon) by calcining.
In most cases it’s difficult to electrodeposit alloys; metals tend to get separated from each other by the process. Sometimes this is because of differing solubilities; lead sulfates, for example, are insoluble, so lead won’t electrodeposit from a sulfate bath. (Chromium has both soluble and insoluble sulfates, and of course barium and calcium have insoluble sulfates, but they’re too reactive to electrodeposit from water.) But that’s not unique to electrochemistry; that’s just regular heap-leach mining chemistry.
The much more interesting fact is that by setting the voltage low enough, you can generally electrodeposit just a single metal from an electrolyte containing different kinds of cations, because no two metals have exactly the same electrode potential. This is potentially very interesting: it’s a high-throughput, high-efficiency, small, low-temperature way to separate many different ionic species. It won’t work for every case, because of considerations like those mentioned above for iron. But it will work in many cases.
By the same token, it’s often possible to dissolve just one metal out of an alloy anode by setting the voltage at the right level.
In general if a metal can be electrowon it can also be precipitated by a single displacement reaction from a more reactive metal. Standard electrode potentials include:
| solutes | metal | E°/V | electrons |
|---|---|---|---|
| Li+ + e- | Li(s) | -3.0401 | 1 |
| Na+ + e- | Na(s) | -2.71 | 2 |
| Mg2+ + 2e- | Mg(s) | -2.372 | 2 |
| Al3+ + 3e- | Al(s) | -1.662 | 3 |
| Ti2+ + 2e- | Ti(s) | -1.63 | 2 |
| Zr4+ + 4e- | Zr(s) | -1.45 | 4 |
| V2+ + 2e- | V(s) | -1.13 | 2 |
| 2H2O + 2e- | H2(g) + 2OH- | -0.8277 | 2 |
| Zn2+ + 2 e− | Zn(s) | -0.7618 | 2 |
| Ta3+ + 3 e− | Ta(s) | -0.6 | 3 |
| Fe2+ + 2 e− | Fe(s) | -0.44 | 2 |
| Co2+ + 2 e− | Co(s) | -0.28 | 2 |
| Ni2+ + 2 e− | Ni(s) | -0.25 | 2 |
| Sn2+ + 2 e− | Sn(s) | -0.13 | 2 |
| Pb2+ + 2 e− | Pb(s) | -0.126 | 2 |
| 2H+ + 2 e− | H2(g) | 0 | 2 |
| Cu2+ + 2 e− | Cu+ | +0.159 | 1 |
| Cu2+ + 2 e− | Cu(s) | +0.337 | 2 |
| O2(g) + 2H2O + 4e- | 4OH- | +0.401 | 4 |
| Cu+ + e− | Cu(s) | +0.52 | 1 |
| Ag+ + e− | Ag(s) | +0.7996 | 1 |
| Au3+ + 3 e− | Au(s) | +1.52 | 3 |
So, if you have some divalent lead salt such as lead acetate in water, and you put a less noble metal into the water, such as aluminum, titanium, zirconium, vanadium, zinc, tantalum, iron, cobalt, nickel, or even tin, you should expect the lead to precipitate, dissolving the other metal into the water; this is the famed Tree of Saturn of the alchemists, and when instead done with a soluble salt of silver, it is the Tree of Diana. The same thing explains the immersion plating of silver ions onto copper with brief immersion at 50° to 60°, immersion plating of gold onto copper at 80° to 90°, immersion plating of gold onto nickel, and so on.
As I understand it, the difficulty in electrowinning aluminum, magnesium, and the titanium-group metals is precisely that they have a more negative electrode potential than hydrogen, so they form an “immersion plating” of hydrogen, consuming the water and the metal. Normally they are protected from this reaction by an impermeable oxide layer, so they don’t dissolve spontaneously in water the way lithium and sodium do.
So, in theory, you ought to be able to precipitate out any of the nobler metals from solution by starting with a hunk of zinc.