Products of Molten Silicate Electrolysis
We have identified several products of silicate electrolysis, partly on the basis of our experiments and partly on the basis of phase diagrams for equilibrium processes. The main products are oxygen produced at the anode and a suite of metals and metal alloys produced at the cathode and consisting of iron, silicon, or iron-silicon alloys containing 0.2-1 percent titanium and chromium. The metal compositions vary as a function of imposed potential and magma composition. (Theoretically, but not yet observed in our experiments, aluminum, magnesium, and calcium could be reduced at increasingly negative potentials and at higher melt temperatures.)
The mineral spinel precipitates from the residua! melt at sufficiently low temperature or with sufficient removal of silicon and oxygen. This material varies in composition from an iron- and chromium-rich spinel to a magnesium- and aluminum-rich spinel, depending on the composition of the magma and the extent of electrolysis.
The remaining molten silicate will be an important byproduct; it can be cast into bars, beams, and sheets, or its CaO- and MgO- enriched composition may make it suitable for use in cements. The so-called "waste heat" carried off with the products or radiated by the cell is another potentially useful byproduct.
Electrode and Container Materials
The primary disadvantage of the molten silicate electrolysis process is that these high-temperature silicate melts are very corrosive, and suitable materials for containers and electrodes are yet to be tested. We discuss here four general types of possible electrode or container materials.
One type of material is simply inert to the silicate and its products. As an example, platinum has been used extensively in experimental petrology as an inert container for silicate melts at temperatures as high as 1650°C. However, although platinum does not react with silicate melt, it does combine with silicon to form an alloy that melts below 1000°C. Because silicon is expected to form at the cathode and accumulate in the container, platinum is unsatisfactory as a cathode or container. Platinum does appear to be a suitable anode material (Haskin et al. 1990).
The second type of material involves a steady-state equilibrium. An "iron skull" container or cathode could be formed by balancing the heat generated by the electrolysis with heat lost to the surroundings to form a solid skin of product or feedstock enclosing the silicate melt and metal product.
The third type of material would be in thermodynamic equilibrium with the silicate melt and electrolysis products and would therefore not react with them. Because the product is Si-Fe metal, Si-Fe alloys might serve as the cathode material (Haskin et al. 1990). Similarly, the presence of sl;!inel (MgAI204) on the liquidus of the residual silicate of the electrolysis process suggests the use of spinel as the containing material.
The fourth type of material WOUf~i be destroyed by the process, but slowly. This option detracts from one of the intended advantages of unfluxed silicate electrolysis-the absence of any need to resupply reagents or other materials from Earth or to recover them from the products. Nevertheless, such an option may prove to be the most cost-effective.
Problems and Work Yet To Be Done
As discussed above, our first steps in the study of molten si1icate electrolysis have been to investigate the fundamental chemistry of the electrolysfs and to address some specific questions such as the product comoosition. We have not addressed certain complexities of the process such as (1) problems that might arise in scaling up from our small experiments to a factory- size process, (2) the problem of designing a system that will effectively transfer the corrosive electrolysis products from the cell, (3) the problem of maintaining the cell at high temperature so that the silicate melt does not freeze and destroy it, (4) the problem of getting ore from the lunar surface to the cell, and (5) the still untested state of our proposed container and electrode materials. Our purpose thus far has been not to demonstrate that molten silicate electrolysis in its current state of development is the best process, but to determine whether, theoretically and experimentally,lt might be the best process if certain technological hurdles can be overcome. As do all the proposed methods for extracting oxygen from lunar materials, the silicate melt electrolysis method requires considerable work before an operational factory can be built.
Criteria for Comparing Processes
Several other promising processes for extracting oxygen from lunar materials have been proposed and are being studied, including reduction of ilmenite by hydrogen gas (see the preceding paper by Knudsen and Gibson), reduction by carbon monoxide gas (see the paper by Rosenberg et al. preceding Knudsen and Gibson's), extraction by processing with hydrofluoric acid or fluorine (e.g., Waldron 1985; Burt 1989, 1990), and electrolysis using a fluxing agent to reduce the melting temperature and increase the electrical conductivity of the melt (e.g., Keller 1989). Understanding which of these processes is the most convenient, reliable, and economical is one of the goals of current research efforts. At present, we are the primary investigators of the molten silicate electrolysis method (also called the "magma electrolysis" method) for extracting oxygen from lunar materials, and our work to date has increased our confidence in its promise. Here, we compare our method, as we now assess it, with other proposed technologies. We recognize the fine line between advocacy and objectivity (Johnson 1980), and we realize that only the test of time and adequate experimentation can determine which technology is the most appropriate.
Informed speculation and preliminary studies of these and other extraction processes have proceeded for over three decades, but slowly for the following reasons: (1) The exact characteristics of nonterrestrial resources are, in most cases, only poorly known, although our knowledge of lunar resources is at least based on experience on the Moon's surface and materials collected there. (2) Conditions on the Moon (vacuum, intermediate gravity, extreme temperatures, and nontraditional ores) are foreign to Earth experience in mining and materials extraction. (3) Lunar conditions (with which we have little experience) and the uncertain future demand for lunar materials make significant investment of time and money in the development of specific processes seem risky. Thus, many studies have been of the paper, rather than the laboratory, variety.
Given the nontraditional ores and conditions on the Moon, it can be argued that nontraditional extraction processes may prove more practical there than transplanted terrestrial technologies (e.g., Haskin 1985). We form certain general criteria for judging the various processes as discussed below (and elsewhere in this volume).
The successful lunar process must rely on proven resources, preferably abundant ones. Especially in the short term, the cost of searching for specialized or superior ore bodies (which mayor may not exist) could overwhelm the cost of extracting the desired material from less specialized ore. The use of a common material also requires that a process should accommodate a substantial range of feedstock compositions and thus be relatively insensitive to the selection of a Moon base site. The process should use a feedstock that is easily mined and requires minimal processing. It should operate automatically or by teleoperation from Earth. Particularly in the early days of Moon base development, the process and accompanying mining, beneficiation, and other operations should not require a large fraction of the astronauts' time or of the available power.
The process should not be compromised by, and if possible should take advantage of, lunar conditions such as 2-week day~ (with dependable sunlight-there are no cloudy days), large temperature swings between day and night, vacuum, intermediate gravity, abundant unconsolidated lunar "soils," clinging dust, and the absence of traditional processing agents such as air, water, coal, and limestone.
The process should be simple, with few steps and few moving parts. It must be easy to install and robust against physical jarring during transport and installation. Initially, all operations on the lunar surface will be awkward and expensive. Thus, the simplest technologies that can produce crucial products will presumably be the first technologies developed (Haskin 1985). Keeping the process simple will make it easier to automate, will require fewer replaceable parts, and should reduce operational problems, resulting in less downtime and fewer people needed to operate the plant. Simplicity can also decrease development time and cost. However, simplicity must be balanced against flexibility to yield more specialized products later on in the development of the Moon base. For example, some compromise should be reached between the ability to extract a single product from lunar soil and the ability to extract several valuable products by a more complex process or processes.
The process must require little or no continuing supply of reagents from Earth (such as fluorine, hydrogen, nitrogen, or carbon). One of the principal costs of setting up and maintaining a lunar factory will be the need to bring supplies from Earth (see Simon's paper in volume 2). If the process uses reagents that need to be recovered, their use raises power and mass requirements (in contrast to on Earth, where cheap reagents often need not be recovered) and increases the comp1exity of the process (since additional steps are required to recover them).
In addition to app1ying these first- order criteria, we can make rough comparisons of various processes for extracting oxygen by asking the following questions. How much power is required to produce a given amount of oxygen? What fraction of the feedstock is converted into products? What are the products of the process? What technology must be developed before the process is viable? What plant mass is required to produce oxygen at a given rate? What must yet be learned about the theory of the process before any or all of the questions above can be answered? Exact answers to these questions cannot be obtained until much more research is done, but in the next section we describe a possible magma electrolysis operation, pointing out its advantages as judged by these criteria. Similar comparisons between various methods for extracting oxygen were made earlier by Eagle Engineering (1988).