Featured Article » Nonfiction

Aluminum: Will O' the Wisp?

Written by Iver P. Cooper

There is no doubt that aluminum is a wonder metal. Pure aluminum has a density only about one-third of iron, it is as reflective as silver, and it is a good conductor of heat and electricity. When exposed to air, it quickly acquires a protective coating of aluminum oxide, which shields it from further corrosion. Alloys of aluminum are extensively used as structural materials in the construction of buildings and vehicles.

Because of the extensive up-time use of aluminum, a substantial amount of aluminum products passed through the Ring of Fire. This aluminum will certainly be recycled, where possible (more on that later). What is much more difficult is producing aluminum, and its alloys, from scratch.

The process which made it possible to produce reasonably pure (over 99%) aluminum at a reasonable cost was the Hall-Heroult smelting process (1886), involving the electrolytic reduction of aluminum oxide (alumina) to the metal. Finding the ore (bauxite) and extracting alumina from it are relatively straightforward. However, the Hall-Heroult process has some additional, potentially ticklish material requirements (large amounts of electricity, the rare mineral cryolite, and highly pure carbon)--requirements that the con men targeting Dr. Phil are hoping he will overlook. Also, the more important uses of aluminum are in alloys, so we need to purify the major alloying elements, too.

Finding Aluminum Ores

Aluminum is the third most abundant element in the earth's crust. While hundreds of minerals contain aluminum, the principal ore is bauxite (rocks containing hydrated aluminum oxide minerals such as gibbsite, boehmite and diaspore).

Bauxite takes its name from the town of Les Baux, in France, so that pinpoints one deposit, albeit one in enemy hands. There, it is found as a reddish rock. (EB11, "Les Baux")

There are many other bauxite locales. The Encyclopedia Americana articles on "aluminum" and "bauxite" reveal that bauxite can be found in Europe (France, Ireland, Greece, Hungary, Yugoslavia, Croatia, Bosnia, Herzegovina), the Americas (Arkansas, Jamaica, Suriname, Guyana, Brazil), Africa (Guinea), and Asia (Indonesia). The 1911 Encyclopedia Britannica (1911EB) adds Styria, Austria, India, Italy, Alabama and Georgia (EB11-A), and the modern EB, Hawaii, Australia, Malaysia, China, the Soviet Union, and Ghana. (EB-IEP, 389) The Irish deposit is at Irish Hill (near Larne), in county Antrim, Ireland. The Arkansas deposits are in Saline and Pulaski counties.

Several atlases likely to be in personal libraries, such as the Hammond Citation World Atlas and the Rand McNally Family World Atlas, show, more precisely, the location of major "Al" deposits.

When granitic rocks are weathered, the feldspar minerals (which are complex silicates) are converted into "clay minerals," such as kaolinite (hydrous aluminum silicate). If the granite has a high aluminum content (e.g., aluminum silicates), bauxite is formed. (Hochleitner, 39) The implication is that if you are looking for bauxite, a good starting point is to find out where clay is mined for porcelain.

The up-timers, of course, are most interested in finding bauxite in Germany. There are several clues: (1) it is in basaltic rocks of the "Westerwald," and these rocks are tertiary basalts interbedded with pisolitic iron ore, like those of Antrim (EB11-B); (2) it is found in Hesse (EB11-A), and (3) it is deposited as a reddish clay, between layers of tertiary basalt, at Vogelsberg in Germany. (EB11, "Laterite")

The term "Westerwald" is pretty nondescript ("western forest"), but there is a forest so named at the modern border of Hesse and the Rhineland Palatinate. Vogelsberg is likewise shown by modern maps as a mountainous area in south central Hesse, west of Fulda.

Unfortunately, the German deposits are not significant enough to be shown in the Hammond atlas.

***

We know, within several miles, where to look, the next question is, would we know what are we looking for?

Both The Audubon Society Field Guide to North American Minerals, and the Eyewitness Handbooks: Rocks and Minerals provide photos and "tests" for bauxite. Bauxite isn't necessarily red; it can be white, yellow or brown. (EB11-B) Red is a sign of the presence of ferric iron oxide, although the ore's color "is no sure criterion of the iron content." (EA)

***

Would-be aluminum barons will no doubt show prospective investors the encyclopedia page which states "by far the greatest quantity of commercially exploited bauxite lies at or near the earth's surface. Consequently, it is mined in open pits requiring only a minimum removal of layers of soil and rock covering the ore." (EB-IEP 389)

The modern EB says that "bauxite beds are blasted loose, dug up with power shovel or dragline, and the ore transported by truck or rail to a processing plant. . . . Refining plants are located near mine sites, if possible, since transportation is the major item in bauxite costs."

In the 1632 universe, we will still be able to use explosives to blast away overburden, if necessary. However, access to power shovels and dump trucks will initially be limited to mining sites within driving distance of Grantville. Hence, if we are mining it further away, we will be collecting the bauxite with pickaxe and shovel, carrying it out of the mine by wheelbarrow, hoist, or mine car, and shipping it to the processing plant by pack mule, wagon, barge or ship.

How much ore do we need? According to EB11-Al, the alumina content of the French, Irish and American bauxites is 33-70%. Assuming an alumina content of 50%, it takes two pounds of bauxite to make one pound of alumina, and two pounds of alumina to make one pound of elemental aluminum. (EA)

Bauxite Confidential

How easy is it to find and mine bauxite? Bauxite was discovered in 1821 by Pierre Berthier. Berthier was an amateur geologist, on holiday in Provence, and his attention was caught by a conspicuous band of red earth in the white face of a limestone ridge near Les Baux. (Raymond, 220-1)

In 1858, the prominent chemist Henri Deville was sent a sample of an "iron ore" which an engineer in Marseilles had been unable to smelt. Deville identified it as high-grade bauxite. The sample came from nearby Les Baux, and was found in a kilometers-long exposed seam—much like the one which Berthier had found previously. (Raymond, 224)

Germany. While Germany is not even listed in Brubaker's 1963 overview of bauxite countries, a German website says that in 1918, Germany was the "world's fourth largest bauxite producer." It adds, "The pit 'Eiserne Hose' near the town of Lich has been in production until 1975. Miocene basalt is weathered to depths up to 100 m and overlain by bauxite." A photograph shows a roadside falloff on which a red soil is exposed. http://mindepos.bg.tu-berlin.de/mk/mkb1.htm

Rest of the world. About 80% of world bauxite production is from surface mines, usually exploiting "blanket deposits." (International Aluminum Institute, IAI) Blanket deposits may be exposed, or covered with some kind of overburden which must be removed by open-pit techniques.

According to IAI, "large blanket deposits are found in West Africa, Australia, South America and India. These deposits occur as flat layers lying near the surface and may extend over an area covering many kilometers. Thickness may vary from a meter or less to 40 meters in exceptional cases although 4-6 meters are average." That sounds promising.

In British Guiana and Dutch Guiana (Suriname), we can find blanket deposits up to forty feet thick, and covered by up to sixty feet of overburden (typically sand and clay, not rock). The ore is 58-63% alumina, 2-5% silica, and 3-6% ferric oxide. The Gold Coast deposits are fairly similar: average thickness is 33 feet, maximum 60 feet; and as much as 64% alumina. (Bateman, 558-9)

Lancashire says that Jamaican bauxite is also near the surface (usually not more than 100 feet underground). Moreover, the overburden is soft, and thus easily removed. A map of bauxite mining areas shows that they cover about half of the central third of the island (the region earmarked by the Hammond Citation World Atlas as being of interest).

According to IAI, in southern Europe and Hungary, bauxite is most often found in pockets. These may need to be reached by tunneling.

Extracting Alumina

The alumina can be isolated from bauxite by the 1888 Bayer process. There is a general description of this process in both the Encyclopedia Americana and the modern Encyclopedia Britannica. The bauxite is washed with a hot sodium hydroxide solution, converting the aluminum oxide to a "green liquor" of saturated sodium aluminate. The bauxite impurities (silica, iron oxides, and titanium dioxide) are less soluble and to some degree are filtered out, using cloth filters, as a "red mud."

Crystal "seeds" are added to the liquor, and the solution is allowed to cool, so that an aluminum hydrate (hydroxide) precipitates out. The aluminum hydroxide is then heated (EA says to 1093.3 deg. C (2000 deg. F) to produce the purified aluminum oxide (alumina) in the form of a sugar-like powder. The 1911 EB provides some additional information, such as the specific gravity of the sodium hydroxide solution (from which a chemist can calculate its concentration).

The modern Encyclopedia Britannica says that alumina, after purification for smelting purposes, usually contains less than 0.1% of other oxides.

Alumina Confidential

Our heroes will be pleasantly surprised to discover that most bauxite will not require a great deal of processing. If there is a lot of clay mixed in, it can be removed by "washing, wet screening, cycloning or hand picking." (IAI) The ore should also be crushed so as to increase the surface area over which the dissolution can take place.

The necessary temperatures are dependent on the nature of the bauxite mineral. Gibbsite requires just 135-150 deg. C; Boehmite, 205-245; and diaspore, even higher temperatures. Sodium hydroxide concentrations may also need to be increased to complete extraction of the more stubborn minerals. (Lancaster)

The real bugbear is silica content, which is 1-32%. Lancaster says that ores with more than 7% silica cannot be economically processed, but of course that depends on the prices and availability of alumina and aluminum. Bateman says the allowable silica is only 4.5%. (554)

Essentially, the problem is that the same sodium hydroxide which dissolves the alumina (aluminum oxide) also dissolves the silica (silicon dioxide). The dissolved silica reacts with sodium aluminate to form sodium hydroaluminosilicate, which is essentially a waste product. "As a result, 0.666 kg NaOH and 0.85 kg Al2O3 per 1 kg of silica are lost irrevocably." (Rayzman) Silica content varies from one deposit to the next, and hence it would be prudent to assay it before beginning mining operations.

There is a trick for processing high-silica ores, and it is mentioned in the modern Encyclopedia Britannica. The infamous red mud is heated with limestone (calcium carbonate) and soda to regenerate sodium aluminate, and the solution is fed back into the Bayer process. The residue, rich in silicate, is called "brown mud."

Another possible problem impurity is ferric oxide (range 1-30%). Bateman says that bauxite ore should not have more than 6.5%. (554)

The amount of "red mud" waste generated for each ton of alumina produced depends on the ore, being just 0.33 tons for Surinamese bauxite, one ton for the Jamaican, and two tons for the Arkansan. Efforts have been made to find uses for it, or alternatively to treat as a low grade iron, titanium or aluminum ore and extract metal from it. (Chandra, Waste Materials Used in Concrete Manufacturing, 292)

From Alumina to Metallic Aluminum: The Hall-Heroult Process

In nature, aluminum exists in an "oxidized" state (combined with other elements, especially oxygen). To obtain the metal, the aluminum must be "reduced," usually in an electrolytic cell.

The cell (pot) is the reverse of a battery; a battery uses a chemical reaction to create an electric current; an electrolytic cell uses a current to force a chemical reaction to occur.

Inside the cell is an electrolyte, a fluid medium in which ions and electrons can move. Like a battery, a cell has two poles. The cathode provides the electrons, and they leave the cell at the anode. Reduction occurs at the cathode and oxidation at the anode.

There is a decent description of the Hall-Heroult process in the Encyclopedia Americana. The electrolyte is a melted (982 deg. C) solution of alumina in cryolite; no water is involved. (Cryolite is needed because the melting point of pure alumina is 2050 deg. C.)

The cryolite is held in a carbon-lined cast-iron shell, whose bottom serves as the cathode. Carbon rods are suspended in the melt; they are the anode. Current is passed from the anode to the cathode, reducing the aluminum oxide to aluminum, and releasing oxygen (which attacks the carbon rods, producing carbon dioxide).

The Hall-Heroult process was developed in 1886, and by 1892, it was routinely producing aluminum which was over 99% pure. (Wallace, 9) If you need material of, say, 99.9% purity, you will need to further refine it.

The principal inputs are: alumina (aluminum oxide), cryolite, electricity, and carbon (for the rods). We have already discussed alumina. What about cryolite?

Cryolite

Natural Cryolite. Cryolite ("frost stone") is a mineral, sodium aluminum fluoride. 1911EB reveals that cryolite can be found "almost exclusively at Ivigtut (sometimes written Evigtok or Ivittuut) . . . on the Arksut Fjord in southwest Greenland." The article on "Greenland" notes that the mines are in the district of Frederickshaab, and provides a lovely map showing the location of Ivigtut, "Arsuk" Fjord, and a prominent landmark, Cape Desolation.

This information will be meaningful to down-time mariners, at least the whalers who frequented Greenlandic waters. Both the Cape, and a fjord of the correct shape, are shown on a map made by William Barents and published in 1598. (Braat)

Once our shivering crew of geologists is disembarked at Ivigtut (latitude 61N), they know that they want to look for a "granitic vein running through gneiss," and that the cryolite is "accompanied by quartz, siderite, galena, blende, [and] chalcopyrite."

Once they locate the correct formation, they can look for the actual mineral. 1911EB sets forth its color, crystal form, cleavage, hardness, specific gravity and "flame test" result. Most distinctively, it is "nearly transparent on immersion in water."

A picture would still be nice, and there we are in luck. There is one in Hochleitner, Minerals: Identifying, Classifying, and Collecting Them (1992), which also mentions that it is found in pegmatites (probably more accurate than "granitic veins"), in association with siderite, fluorite, topaz and quartz. There are more photos in The Audubon Society Field Guide to North American Minerals, and Eyewitness Handbooks: Rocks and Minerals.

Whittaker states that cryolite "was traded as early as the beginning of the 18th Century amongst the native people of the western coast of Greenland." It is possible that the mineral was already known in 1632 to the Inuit Eskimos, in which case they can be paid to guide an expedition to the source.

Still, it doesn't take much imagination to appreciate that mining cryolite in Greenland will be arduous and perhaps dangerous. According to 1911 EB, Deville toyed with the idea of using cryolite as an aluminum ore, but, "finding the yield of metal to be low, receiving a report of the difficulties experienced in mining the ore, and fearing to cripple his new industry by basing it upon the employment of a mineral of such uncertain supply," decided to produce aluminum instead by chemical reduction of aluminum chloride (see below).

Anyone seeking to raise money for a cryolite expedition will have to explain away Deville's objections.

Synthetic cryolite. The uptimers know that cryolite can be synthesized (EA), and they at least know its chemical formula (Na3A1F6).

There are several chemists in Grantville and each will have a personal library of chemistry texts. It is within the realm of possibility that even a general chemistry book will explain how cryolite is made. For example, my own library includes a 1993 introductory college chemistry text which suggests adding sodium hydroxide and hydrofluoric acid to aluminum hydroxide (from the Bayer process). The reaction is 3NaOH + Al(OH)3 + 6HF -> Na3AlF6 + 6H2O. (Ebling, 880)

But I think that a good chemist could probably work it out without this help.

Cryolite Confidential

Consumption. In a newly started pot, the initial charge of cryolite is proportional to the initial load of alumina. The 1911EB says that, except for "mechanical losses," the initial charge of cryolite would last indefinitely. In practice, cryolite is lost as a result of absorption by the carbon lining of the cell, vaporization, and so forth.

There are some tricks (which Grantville must re-invent) for regenerating it once smelting is underway. Even so, there will be a continuing demand for cryolite to replace losses (see Appendix).

More on Natural Cryolite. There are contradictory accounts about early mining at Ivigtut. What I think is most accurate is that there were two separate operations there. Julius Thomsen and George Horwitz obtained a license to mine cryolite in 1854, but didn't commence large-scale mining until 1859. In the meantime, in 1854-55, two other entrepreneurs extracted forty tons of silver-bearing galena (itself lead sulphide) (Greenland BMP). They decided, after six months, that the formation wasn't rich enough to warrant further work (Whittaker).

About 3.7 million tonnes of cryolite ore, were mined in the period 1854-1962; previously mined ore was exported until 1987. The mine closed because it had become uneconomic to compete with synthetic cryolite (see below).

Cryolite was initially wanted for use in a new process for production of soda (sodium carbonate), which Thomsen had patented in 1853. Thomsen decomposed cryolite with calcium hydroxide into calcium fluoride and sodium aluminate. He filtered off the former, and added carbon dioxide to get aluminum hydroxide and sodium carbonate (Hornburg 31).

The Thomsen synthesis was used until the 1890s, when it was superseded by the 1864 Solvay ammonia-soda process (Hornburg) . The handwriting had been on the wall for several years, of course, and Thomsen's company, Oresund Chemiske Fabriker, had been trying to develop other markets: soap factories, manufacture of enamel, insecticide ("cryocide"), abrasives (Hornburg; Grossman).

The advent of the Hall-Heroult process, which used cryolite as a flux, was extremely fortunate for the Danish-owned cryolite operation. In 1904, about 25% of cryolite was sold for this purpose, while, by 1939, it was 82% of their market (Travis, 335).

Cryolite was also used, in the period 1855-64, as an aluminum ore in the Rose-Dick process (see Alternatives to Hall-Heroult Process, below). However, relatively few tons were exported from Greenland for this purpose.

At first, the mined material was hand sorted so the shipped ore was at least 85% cryolite. However, the average over the entire history of the mine was 58% (Whittaker). Keep this in mind when calculating how many shiploads of ore are needed.

How easy was it to mine and export cryolite? According to a University of the Arctic course, "it was mined in an open-pit with easy accessibility and an ice-free harbor." However, while Hurlbut's Minerals and Man (88) admits that the cryolite is "easily extracted," it warns that the harbor was only free of ice for a "brief period," and asserts that "mining is difficult because of the harsh climate."

In modern times, the average temperature is above freezing only May-October.

As for ice, the southwestern Greenland coast is somewhat protected by the warm West Greenland Current. As a result, navigation is possible, in open water, year round. However, ships will encounter some sea ice, mostly January-June, and icebergs, which calve off glaciers of the east coast and swing around the tip, are also a threat, especially April-July.

The effect of the Little Ice Age is unclear. While seventeenth century Europe was definitely colder than the twentieth century, it is debatable whether the Greenlandic climate was substantially worse in the 1630s than in the "cryolite soda" era.

It is certainly promising that in OTL, cryolite was mined for 109 years. The very volume of the industry tells us mining cryolite was quite practical in the late nineteenth and the twentieth centuries:

Year(s)	Production	Sales	Consumption	US Imports
1857-67	14,000A
1859			DK: 1,500M
1860-64	13,400M
1862	4,700M
1865-69	44,700M
1865-74				6,000A/yr
1867-77	70,000A
1867			DK: 1,160M
1868			DK: 800M
1870			DK: 1,230M
1880-1910	20,000M/yr
1897-1903				5,383 -10,115E/yr
1904		2,110M		959E
1905				1,600E
1906				1,644E
1914		5,760M
1918 (circa)	6-12,000A/yr
1925		16,760M
1939		30,050M
1854-1962	3,700,000M

M=metric ton, A=American short ton, E=English long ton, DK=Denmark (Sources: Johnson's, "Greenland"; Travis 335; Kragh 40-3; Kentucky Geological Survey; 1918 World Book; BMP Greenland 15; Whittaker 469)

However we do need to ask whether, in 1859, when the cryolite operation began, Europeans were better equipped to sail in Arctic waters (e.g., better maps and navigational equipment) and to live and work under Arctic conditions (e.g., Burberry garbardine windproofs) than they would be in the 1632 Universe. Indeed, by the end of the nineteenth century, they could have used steel-hulled steam ships. Moreover, our ability to reliably access cryolite will be hindered by both war and piracy, neither of which were serious concerns for Thomsen in 1859.

The fact that cryolite has significant uses, other than as a flux in the Hall-Heroult process, will make it easier to attract investors for a cryolite mining venture; they can turn a profit even if the aluminum industry dies stillborn. However, bear in mind that sodium carbonate, while a very important industrial chemical, can be made not only by the Thomsen cryolite process, but also by both the earlier Le Blanc process (1791) and the later Solvay process, both described in Grantville encyclopedias.

I think natural cryolite will be of more interest to downtimers (such as the French), since they are less likely to figure out how to synthesize it, and hence will want natural cryolite as a flux for alumina, or even as an aluminum source.

More on Synthetic Cryolite. Hall didn't use imported cryolite, he made it himself. Which in turn implies that the process isn't real complicated. After all, Hall was working in the woodshed behind his family home. And he had very little formal training in chemistry. (Oberlin)

Synthetic cryolite was commercially available at least as early as 1900, and, by 1930, was offering "serious competition" to the natural product. (Travis, 335)

Safety. A big disadvantage of synthetic production of cryolite is that hydrogen fluoride is involved. HF is highly corrosive (it attacks glass, and some metals), and extremely toxic (it is not unusual for HF burns to lead to amputation of digits and limbs, and even a 2% body exposure can kill). However, even if natural cryolite is used, it is standard practice to add aluminum fluoride, which is made by reacting aluminum with HF, to the "melt." You can dispense with the aluminum fluoride, but at the price of consuming more cryolite and electricity.

Moreover, processing cryolite itself is not without risks—fluorosis was discovered in the workers at the cryolite-to-soda factory in Copenhagen.

Electricity

It will be obvious to anyone reading the encyclopedias that aluminum smelting is energy intensive, although the sources disagree somewhat as to how much electricity is needed. The 1911EB reports an aluminum yield of one pound per twelve e.h.p. hours (nine kilowatt hours). According to the Encyclopedia Americana, it takes about ten kilowatt hours (kWh) to produce one pound of aluminum. The modern EB reports that while, in 1930, it required 12 kWh to produce one pound of aluminum, this had dropped to 4.5 by the early 1980s. (IEP 391) I'll use the 10 kWh figure for now.

The Grantville power plant is a 200 megawatt steam turbine plant, operating at 58% load at the time of the Ring of Fire (Loren Jones, "Power to the People, Ring of Fire). The plant was expected to fail in eighteen to twenty-four months, but be replaced by a steam engine and generator which will supply ten to fifteen megawatts.

According to DOE, in 2001 the average electrical consumption in the United States was 10,656 kilowatt-hours per household. There were about 1,000 households in the Grantville area transported by the RoF.

The current power plant can produce a maximum of 1,752,000 megawatt hours annually. But the immediate post-RoF load, considering only the remaining residential customers, is just 10,656 megawatt hours—less than one percent of the production capacity.

Once the power plant gears down from 200 megawatts to ten megawatts, it still can output 87,600 megawatt hours/year. And the residential load would absorb just 12% of that. The remainder is still sufficient to smelt over 700,000 pounds/year aluminum—not that we'll be making that much anytime soon!

With most of its customer base left behind, and a payroll to meet, the power plant needs to find new customers, fast. Power companies love smelters because they exert a steady, high demand for electricity.

Electricity Confidential

In 1995, the electricity required by modern plants—not counting generation and transmission losses—was 13 kWh/kg (1 kg = 2.2 pounds). The theoretical minimum is 6 kWh/kg (Choate, 25).

However, it is probably more meaningful to look at early practice; the first commercial cells in Pennsylvania (Hall) and Switzerland (Heroult) were drawing over 40 kWh/kg. (Id.) That's about twice as high as the most conservative encyclopedia value.

Carbon

Carbon is needed, both to line the electrolytic cell, and for the anodes. The lining protects the pot from corrosion by molten aluminum and fluorides.

Carbon Confidential

In Grantville, there is no design information on either the anodes or the linings, so that will all have to be worked out empirically.

Anodes. There are two major designs in modern use. "Prebaked anodes" are made by baking blocks of petroleum coke and coal tar pitch ("paste") at 1,000-1,200 deg. C. The advantage of petroleum coke is that it is low in ash (silica, iron oxide, etc.); as the anode is consumed, any silicon or iron would deposit in the aluminum. Coke can be made from coal, but then it has to be purified to remove the ash.

Soderberg anodes are "continuously self-baking." What that means is that the operators are continuously feeding petroleum coke and coal tar pitch into a casing. These materials are baked by the heat of the pot, forming the carbon anode at the bottom of the casing. A smelter using Soderberg anodes doesn't need a carbon baking facility or associated workers.

The prebaked anodes are about 30% more conductive, but Soderberg anodes are baked using waste heat. Overall, smelters using prebaked anodes are about 3% more energy efficient. (Brubaker) Prebaked anodes are also more environmentally friendly; the hydrocarbon waste gases are collected more readily in a carbon-baking facility than at the pot. Cryolite consumption can be up to 60% less in a prebaked anode plant (Reynolds; Beck; Brubaker, 91-96)

The problem for Grantville is not so much the choice between prebaked and Soderberg anodes, but rather appreciating the advantage of using low-ash carbon. The ash content of bituminous and anthracite coals is 1-10%, while that of a typical oil distillate is 0.5-1.5%. (Scurlock)

Linings. One important design parameter is the thickness of the carbon lining. This must vary so the electrolyte (the cryolite) freezes on the inner walls but not on the bottom. (Kirk-Othmer, 195) The advantage of this "ledge" is that it protects the lining from corrosion. Cryolite also freezes to form a crust at the top of the pot, which helps retain heat (and reduce the heating bill).

Its chemical nature is also important. The bottom lining acts, initially, as the cathode. (Once the process has commenced, a pool of aluminum collects at the bottom, and that becomes the true cathode.) Since the carbon lining must be conductive, it is either natural graphite, or carbon baked to form a "graphitic" structure (carbon atoms mostly connected in closely-spaced layers). Anthracite can be used, as high purity is not as important as in the anodes. (Totten, 2:38)

Grantville Knowledge of the Hall-Heroult Process

The electrical current has two functions: (1) heating the cryolite-alumina mixture so it melts, thus forming the electrolyte solution, and (2) reducing the aluminum.

EB11-Al gives a working temperature of 750-850 deg. C., and of course this must be maintained for the reduction to continue. The voltage applied to each cell was 3-5 volts, and 10-12 such cells were connected in series, so the total voltage drop was 30-60 volts. The current isn't stated.

***

We next turn to the modern Encyclopedia Britannica. This, of course, is describing more recent practice. The reported pot temperature is 950 deg. C (higher than before), and the alumina is apparently preheated at this temperature to drive off moisture. The normal pot voltage is about the same (4-6 volts), but pot lines are now 50-150 cells, requiring a total line voltage of 200-900 volts. The currents are 50,000 to 100,000 amperes.

A figure shows an alumina supply hopper over the cell, with the anodes flanking the hopper outlet. The molten aluminum is periodically siphoned off the bottom of the cell. While not shown, there is reference to the addition of aluminum fluoride to "restore the chemical composition of the bath." (EB-IEP 390-1)

Hall-Heroult Process Confidential

In the cell, aluminum oxide is being reduced to aluminum at the cathode (requiring 2.2 volts), while carbon is oxidized to carbon dioxide at the anode (supplying 1.0 volts), for a net minimum requirement of 1.2 volts direct current (at 960 deg. C.) to drive the reaction.

Higher voltages are used because there are electrical energy losses as a result of the electrical resistance of the cryolite, the anodes, the deposited aluminum, and so forth. Each resistance requires a certain voltage to overcome it, and that voltage drop equals the product of the resistance and the current. The biggest voltage drops are in the anode (0.30-0.42 volts), cathode (0.45-0.68 volts), and electrolyte (1.535-1.75 volts). Also, overvoltages at the anode (0.51) and cathode (0.08) help to drive the reaction. (Prasad; Kirk-Othmer 2:197-8; Choate, 31)

***

Electrical energy is converted by the resistance into heat energy at a rate equal to the voltage times the current. Thus, the energy bill exacted by the resistance of the materials is proportional to the square of the current.

Some of this heat is put to good use; it keeps the electrolyte molten, or, if the anodes are of the Soderberg type, it bakes them. The rest is just waste heat. Heat loss occurs at the top of the bath (through the cryolite crust and the carbon anodes), and at the sides and bottom (through the cryolite "side freeze" and the carbon lining). Some heat loss at the sides is desirable to form that "side freeze," which reduces corrosion of the lining. The aluminum metal protects the carbon at the bottom.

It follows that both the electrical conductivity of the electrolyte, and its melting point, are of concern.

***

The 1911 EB says that cryolite dissolves about 30% of its weight of pure alumina (EB11-Al, 78), implying an alumina content of 23% of the melt. That's way too high for economical production. The cryolite-alumina solution with the lowest melting point (960 deg. C) is one which is 89.5% cryolite, 10.5% alumina. (Kirk-Othmer) The solubility of the alumina increases with temperature. Of course, the higher the temperature, the higher the energy requirements.

Usually, other fluoride salts are added to improve the characteristics of the electrolyte. Reducing melting point means less heating is needed, and also reduces the rate of re-oxidation of the aluminum. Unfortunately, at a lower temperature, alumina solubility is lower, making it harder to maintain proper alumina levels.

Increasing conductivity allows use of lower voltages, and saves energy. Decreasing the density below that of aluminum ensures that the metal sinks, lest it get re-oxidized, or short out the anodes. Increasing the surface tension between the molten aluminum and the bath makes it easier to siphon off the aluminum without waste. Decreasing volatility reduces losses to the atmosphere, which are economically and environmentally undesirable.

A typical electrolyte is 2-8% alumina, 80-85% cryolite (sodium aluminum fluoride), 5-7% excess aluminum fluoride, 5-7% calcium fluoride, and 0-7% lithium fluoride. (Kirk-Othmer 2:193, 11:282) The addition of the other salts reduces the solubility, perhaps to about 6%, but also reducing the melting point to as low as 920 deg. C. (Prasad; Aurbach, Nonaqueous Electrochemistry, 503)

The calcium fluoride usually is not added intentionally, because it reduces conductivity and increases density. It is derived from calcium oxide in the alumina. The level in alumina is perhaps 0.4%, but the calcium accumulates in the melt. Calcium fluoride does offer one benefit; it reduces the melting point.

The effects of aluminum fluoride are similar. However, it's deliberately added for several reasons. First, to compensate for loss of the aluminum component of the cryolite. Molten cryolite dissociates to some degree into NaF and NaAlF4. The latter vaporizes more readily, reducing the aluminum-sodium ratio. Also, the moisture in the air reacts with the cryolite to form sodium fluoride, hydrofluoric acid, and alumina, thus further worsening the ratio. (Kirk-Othmer, 11:280)

Another reason to add aluminum fluoride is to eliminate sodium oxide (Na2O), an alumina impurity (~0.6%). The reaction is 3Na2O + 4AlF3 -> 2Na3AlF6 + Al2O3 (Kirk-Othmer, 2:192) Conveniently, this reaction also helps replace cryolite lost as a result of vaporization, absorption by the pot lining, etc.

Cryolite (Na3AlF6) is composed of sodium fluoride (NaF) and aluminum fluoride (AlF3), in a 3:1 ratio. Often, aluminum fluoride is added in excess of what is needed to replace losses or eliminate sodium oxide, as it reduces the melting point and the re-oxidation rate, increases surface tension, and minimizes the anode effect (see below). On the other hand, having excess sodium fluoride increases conductivity and alumina solubility, and reduces volatility. (Both salts reduce density.)

The conductivity of the electrolyte is improved by adding lithium fluoride, which also reduces both melting point and density. But too much lithium reduces aluminum workability.

***

In the course of the operation, the level of alumina will drop. If it drops below 1-2%, the pot voltage rises, and the cryolite is itself electrolyzed ("anode effect"). The fluorides react with the carbon anodes to form polyfluorocarbons, which are carcinogenic. (Beck) If, to avoid or stop an anode effect, you add too much alumina (concentration over 4%), some alumina is deposited, forming a sludge which reduces the current flow. Lowering the pot temperature below the normal 960 deg. C. saves energy, but narrows the "safe" alumina concentration range (Choate, 35)

The pot must be agitated to keep the alumina solubilized in the cryolite. However, if you agitate too aggressively, you run the risk of short-circuiting the pot (by bringing the aluminum in contact with the carbon rods).

Smelters usually run all day and night so that the cryolite, and the extracted aluminum, are kept molten ("Aluminum Smelting," IAI).

***

There are just two basic ways of increasing production: increasing the current running through the pots, and adding more pots (either to an existing pot line, or to a new one).

The theoretical production rate per cell is proportional to the current; if the current is 180,000 amperes, Faraday's Law predicts that the cell will produce 1,450 kilograms aluminum per day.

Hall's first commercial cell used only 1,750 amperes, and in the Thirties, Wallace thought that the maximum amperage was 30,000. Modern cells can reach as high as 500,000 amperes, but 180,000 is quite respectable.

While higher currents increase production, they also create more heat (leading perhaps to ventilation problems) and stronger magnetic fields (which can disturb the aluminum pool). (Brubaker, 94) If currents are increased, without changing pot size, there is faster corrosion of the cathode, and fluorocarbon formation at the anode.

A small amount of the current is wasted, as a result of side reactions, concentration gradients, transient short circuits, and so forth. (Prasad, Kirk-Othmer, Choate) In the first commercial cells, the current efficiency (the percentage of the current which actually resulted in net production of aluminum) was only 75-78%. (Choate, 25) For modern cells, it is 85-95%. So a 180,000 ampere pot which was 90% efficient would produce 1,305 kilograms daily (capacity of 476 tonnes annually).

Don't confuse current efficiency with energy efficiency (the percentage of the electrical energy entering the smelter which actually is used to reduce aluminum). Because additional voltage must be used to overcome electrical resistance, energy efficiency is only about 26%. (Choate, 31)

The total amperage useable by a smelting operation in Grantville may be limited by the power generation and transmission equipment, or by allowable current densities in the pots. At that point, adding pots is the only way to increase capacity. Modern potlines usually don't exceed 300 pots, but of course you can add potlines.

The total size of the smelter is limited by the local availability of electricity, land, building supplies and labor.

Subsequent Processing

The molten aluminum is transported to a holding furnace. If the intent is to make an aluminum alloy, the alloying elements are added at this point. When the composition is correct, the alloy is poured into a mold, in which it cools to form an ingot. The ingot can then be further manipulated.

Aluminum (and its alloys) can be cast by melting it and then pouring it into a packed sand mold or a permanent iron or steel mold. Aluminum has a relatively low melting point (660 deg. C.), which facilitates casting.

Molten aluminum dissolves iron, so if it is cast in an iron receptacle, the latter needs a protective coating. If silica is present, perhaps as a binder, the aluminum will reduce it, producing elemental silicon.

The molten aluminum can be worked in a variety of ways, depending on the alloy, including rolling (into sheets or foils), forging (hammering), extrusion (pushing through a hole) or drawing (into wires).

Heat-strengthening a suitable aluminum alloy (e.g., aluminum-copper) requires controlled temperatures and uniform heating of the furnace. (Hultgren, 297-8) The modern EB describes the process (which it calls "solution heat treatment") in general terms: (1) heating the metal for 6-24 hours at temperatures of 370-535 deg. C., (2) quenching, in hot (66-100 deg. C.) water for casting alloys, or in water at room temperature for wrought alloys; and (3) "aging" the metal, either at room temperature or, to accelerate the process, at a higher one (the encyclopedia says, "somewhat above the boiling point of water"). (IEP 392, 395)

The optimal time-and-temperature conditions are alloy-specific. More detailed specifications may appear in the personal library of one of Grantville's engineers or machinists—for example, treatments for five common alloys appear in Walsh's Machinists' and Metalworkers' Pocket Reference (9-25). Of course, to make sure the temperature is correct in practice, we will need pyrometers.

Welding aluminum, especially to other metals, can be somewhat tricky but hopefully the welders of Grantville already know something about this.

Quality Control

An important consideration in aluminum production is the minimization of impurities, whether you are engaged in primary production (from ore) or secondary production (recycling of foundry scrap or aluminum articles).

The basic problem is that, of the common metallic elements, only potassium, sodium, calcium and magnesium are more reactive than aluminum. Once a less reactive element is associated with the aluminum, it is "practically impossible" to remove it. (NAP, 53) Hence, impurities—at least those which adversely affect the properties of the desired alloy—must be minimized. That means that the raw materials should be as pure as possible.

The impurities in primary (newly made) aluminum primarily come from the anodes and the alumina. (Kirk-Othmer, 2:196); other sources include the cryolite, fluoride additives, and even the air. The impurities include iron, silicon, calcium, lithium, sodium, hydrogen, oxygen, titanium, vanadium and manganese.

The most common impurity in aluminum is iron. In general, it is considered undesirable (note that the Bayer process deliberately removes iron oxide), and the iron content of commercially pure aluminum is typically 0.08-0.5%. (Belov, 185-6) If iron content is excessive, iron aluminide crystals form, which have an undesirable effect on formability, fatigue resistance and surface finish. (Key to Metals) Iron also interferes with heat hardening. (Hultgren, 302)

Next most common is silicon, which is problematic because it can render the aluminum more brittle. However silicon is deliberately added to some alloys to increase castability.

In recycling "secondary" (recycled) aluminum, elements (e.g., zinc, magnesium) deliberately added when making the original alloy may be undesirable if the desired alloy is different.

***

If the ultimate goal is to manufacture an aluminum alloy, then the quality of the source of the alloying element must also be known, and the melt will be tested and adjusted until the desired alloy composition is achieved.

Modern chemists and metallurgists are accustomed to relying on instrumental methods. After the Ring of Fire, the up-timers have access to two appropriate instruments. The power plant has a "Metallurgist XR," which is a portable X-ray fluorescence spectrophotometer specifically designed for alloy analysis. (Boyes) And, even more surprisingly, the high school has a $300,000 atomic absorption spectrophotometer given to them in October 1997 by LaFarge Corp.

Neither of these instruments is going to remain in working order for very long. However, while they last, we can assay a lot of ores from different suppliers, and also reconstruct the compositions of the many different aluminum alloys which are available in Grantville. Just analyzing the parts of an automobile could allow reconstruction of the composition of representative alloys of all of the major types, including the high strength aluminum-magnesium 7000 series. (www.autoaluminum.org)

After they fail, we'll have to rely on wet chemical techniques and microscopic examination of etched metal sections. Fortunately, the two spectrophotometers allow us to assemble a library of photographs of the microscopic appearance of alloys of known composition.

Uses for Pure Aluminum

The uses for pure aluminum are constrained by its relatively low hardness and tensile strength. Nonetheless, there are significant markets for it.

...