David Reid writes (via BoingBoing):

Research is nearing completion on a system that will allow the melting and casting of bronze, silver, gold, and even cast iron, using an unmodified domestic microwave oven as the energy source.
A potential foundry in every kitchen !!

The first part of this Foundry Note describes a technique for using a domestic microwave oven to melt and cast, to accurate shape, small quantities (up to a quarter of a kilo) of bronze, silver, white metal or iron. The technique has been used to cast pieces from ceramic shell moulds up to about 18cm high, and is an accessible alternative to other small-scale melting set-ups, for example the flask casting of jewellery.

The second part of the note describes thoughts and tests which led to the procedure. It offers guidance and some warnings, to anyone making investigations into metal heating by microwave.

Some pictures of the process

Background
The microwave work was triggered by a short reference to the refining of rare earth metals, at Illawara Technology Centre, which was mentioned by a visitor to the Central Saint Martins foundry, Dennis Glaser. Since these metals melt at temperatures above 800 degrees Celcius, it seemed possible that the method could be adapted to melt and cast small objects in the workshop or studio. If this could be done a domestic microwave would, effectively, become a cheap and accessible furnace.

Trials were begun which simply aimed to melt metals such as silver and bronze in open crucibles. However, it soon became obvious that casting to shape could also be accomplished by adapting the Reid Technique (RT) - a simplified ceramic-shell procedure for the casting of non-ferrous metals, patented in 1990. RT was first developed to avoid the problem of heat loss, which makes the the pouring of small melts very difficult - these difficulties arise however the metal is heated, and while the microwave technique set out here can be used for heating small amounts of metal in open crucibles, its greatest potential lies in its use as a flameless furnace in processes such as the Reid Technique. The crucial discovery, made during extended tests with various susceptors - materials which heat up when exposed to microwaves - was that two substances, graphite and magnetite, working together were required to achieve the kind of heating we were looking for.
The Method in Brief
A wax object is prepared, attached to a hemispherical wax cup.

A second blank wax cup is prepared.

Both waxes are coated with a patent ceramic shell slurry containing some graphite.

These are then stuccoed with a magnetite sand.

Further layers of normal ceramic shell slurry are applied and stuccoed with molochite grain to build up the shells.

Both shells are dewaxed, by rapid heating in a flame.

The shell cup, containing enough metal to fill the mould cavity, and a lump of carbon, is glued with ceramic paste to the mould.

An insulating, but microwave transparent, ceramic fibre block is placed around the cup area of the mould.

The assembly is placed in the oven chamber, and the timer set for a specified time. The firing time depends on the type and mass of metal to be melted e.g. 330g of sterling silver would need 17 minutes in an 850 watt microwave.

When the beeper sounds, the mould together with the insulation, is removed from the chamber and inverted, allowing the metal to run into the casting cavity without loss of temperature.

When the mould has cooled, the shell is removed to reveal the casting.

Equipment and Materials
The Microwave Oven
Domestic microwave appliances are based on the magnatron; an electronic device which converts electrical energy to microwave energy, which is fed via a waveguide to the cooking chamber. Since the conversion is somewhat less than perfectly efficient, the magnatron has to be cooled by a stream of air from a fan. This air is then led to the oven to help remove steam produced during cooking. Once in the chamber, the microwaves are reflected by the metal walls until they are absorbed, (usually by water-containing food), their energy being converted to heat. Should absorption not take place - if, for example, the oven is activated when empty, some energy will re-enter the waveguide and cause over-heating of the magnetron. Usually a safety switch turns the machine off when this happens. Note that the reflecting walls and the constant frequency of the microwaves set up standing waves in the chamber. This results in some areas being much more active than others and is the reason why food must be rotated through the varying field to cook evenly.

To be of use for metal casting, a domestic microwave oven rated D or E (850W or 1000W) needs two slight modifications: the rotating glass plate must be removed and the holes which admit air to the cooking chamber must be taped over (masking tape works reasonably well). The air from the magnetron cooling will then be re-directed to the exterior. No other modifications should be made. Microwaves are potentially dangerous and the uninitiated should treat the oven with respect.

Insulation
This is critical. Microwave energy transformed into heat within the shell must be contained if the temperature is to rise to the point at which it will melt metal. Insulation also protects the walls of the oven. Ceramic fibre wool, in various forms, proved to be a very useful insulating material.
Working Towards a Method
It must be stated that, at the out-set of these experiments, the researcher was completely ignorant of microwave technology. Much was learnt along the way, mainly by cautious empirical investigations which took a serious view of possible dangers. A very early purchase was a microwave leakage detector. As anyone who has left a fork in a microwave oven knows, metal with sharp projections placed directly in a microwave field will cause arcing. This spectacular abuse will burn the interior walls and over-heat the magnetron. Although most modern ovens are protected by heat sensitive cut-outs, arcing will eventually ruin the oven. However, if the microwaves are absorbed and their energy is converted to heat before they meet the metal, no such damage should occur.

When food, (containing water, a very efficient absorber), is placed in a microwave field having a frequency of 2.4 5GHz, virtually all the microwave energy is converted to heat. So, the problem was to find a substance which, when put in a microwave field and in contact with a refractory container, would absorb heat (be a good `susceptor’) and raise its temperature to about 1200 degrees Celcius, thus allowing alloys within it to melt and become castable.

Early experiments using carbon as a susceptor were discouraging. The uninsulated crucibles barely attained red-heat and after running for 5 or so minutes the machine shut off. Insulation around the crucible helped, as did the realisation that the cooling air from the fan could be redirected. But it was obvious that a more efficient absorber had to be found. The literature on absorbers mentioned both silicon carbide and ferrites as susceptors, so an SiC paste was mixed with clay and applied to a thickness of about 8 mm and dried on the inside of a ceramic shell crucible, which we knew from earlier tests to be non-absorbing. After drying, it still didn’t show red heat after 10 minutes in the microwave field. Powdered ferrites proved very difficult to obtain until it was realised that they were just modified iron oxides. A quick visit to the Ceramics Department gave samples of red, yellow, black and granular iron oxides. Similar sized samples of these were similtaneously put in the oven and fired for a couple of minutes. They all showed warming, but the granular substance (magnetite) was hot enough to burn the finger.

Another crucible was prepared, lined in the same way, but this time with a clay-ferrite paste. 50 g of sterling silver was added, and the crucible was capped with a carbon lined shell. Although it became very hot the silver had not melted after 10 minutes of firing. The carbon-lined cap was replaced by one lined with magnetite/clay and the test was re-run. After 15 minutes firing the silver was found to have melted. A very exciting moment! Although the methods were a bit crude, a temperature of 900 degrees Celcius had been reached. Using a similar crucible, with a small mould attached, the first castings were made soon after. However, the process was not efficient enough to be really useful. Some simple calorimetric calculations showed that very little of the energy entering the chamber was actually getting to the metal. Much of it was being absorbed by the crucible, and the walls of the chamber were getting quite hot, showing that less than perfect absorption was taking place around the metal. Closer examination of the absorbancy of various materials followed, and attempts were made to formulate more efficient ferrite susceptors.

How thin could an absorbing layer of magnetite be made in order to reduce its thermal capacity but retain its heating qualities? The granularity of the magnetite suggested it could be applied as a stucco in a ceramic shell build-up. How many layers would be necessary? Two were tried for a start, then one. Both crucibles heated well when placed close to the wave-guide port, but the heating was by no means even. Small areas of the shell would ‘light-up’, getting hot enough to melt the refractory. Test shells, after cooling, that were replaced in the oven, sometimes fired up and sometimes did not. Again, much testing was undertaken to try and solve this problem.

Parallel experiments going on in the foundry involved heating glasses of various compositions in a microwave oven. It was found that although these were transparent to the microwaves when cold, they would absorb microwave energy when just below red heat. Maybe magnetite had to be above a critical temperature to absorb well. The carbon coated shells, we already knew, would always heat from a cold start. A double susceptor - a carbon (graphite) loaded primary coat stuccoed with magnetite sand - was tried with some confidence. The crucible, with a mould attached was fired and the temperature climbed steadily from cold, to something sufficient to melt the silver which was then cast. Further experiments led to temperatures high enough to melt small amounts of cast iron. This proved to be a limit - any increase above it caused the magnetite to flux and destroyed the shell. The hunt is on for susceptors that can take the temperature of the crucible beyond the magnetite limit. It may then be possible to cast even higher melting point metals, such as steel, by the microwave method.