The following is the surprisingly successful implementation of a bad idea. I cobbled together a thermal evaporator to deposit aluminum onto a silicon wafer, with the goal being to form ohmic contacts.
The vacuum pump and thermistor-type vacuum gauge are salvage/surplus. The transformer was made from the core and primary of a transformer scrapped from a dead microwave oven. The chamber was made from an assortment of commodity vacuum fittings. The most expensive component, by far, is the length of black rubber tubing connecting the pump to the chamber.
The heater is a sliver of tungsten cut from an evaporation boat. A small strip of aluminum foil, cleaned of residue and folded in half, is hung over the middle. With the input power properly managed, once the middle of the strip has melted onto the tungsten, the rest of the strip is drawn into the melt by surface tension.
One somewhat novel feature is that the chamber doesn't have to be big enough to accommodate the entire target object. The wafer (or in this case, a glass slide used as a test substrate) simply sits against a seal on top of the chamber, and atmospheric pressure seems to provide enough down-force to maintain a vacuum-tight seal (which I've supplemented with a bit of silicone grease for good measure). It is necessary to aggressively and meticulously clean the surface before attempting to coat; otherwise the adhesion is very poor.
Coming eventually: Photos and data from the resulting aluminum-on-silicon contacts.