The Nernst Lamp
Electrical Conductivity in Non-Metallic Materials
Retired from Dept. of Physics, University of Leicester, U.K.
The use of carbon and tungsten filaments mounted in vacuum as early incandescent lamps is well-known, but it is less often realised that an early competitor was Nernst’s ceramic lamp. This made use of the property that , although good insulators at room temperature, many oxides become partially conductive when strongly heated. This is due to the migration of ions rather than electrons, and was commonly considered a deleterious property. Nernst realised however that it might permit the construction of a sturdy incandescent lamp that did not require a vacuum. His early lamps embodied a thin rod of magnesium oxide, initially heated with a coil of platinum wire. Later, he used a modified composition based on zirconia. However, these ‘ceramic’ lamps were unable to compete with a new generation of ‘coiled-coil’ tungsten lamps with filaments mounted in low-pressure argon.
Keywords: Nernst, Incandescent lamps, Ceramic lamps
Thomas Edison introduced the first commercially-successful electric incandescent lamp in the early 1880s, employing a high-resistance filament of carbonised bamboo enclosed in a high vacuum within a glass envelope. These lamps lasted over 1200 hours. Later versions using filaments of carbonised viscose lasted longer, but their luminous efficiency was still < 5%. All these carbon-based filaments must have contained a percentage of graphitized carbon in order to achieve their electrical conductivity, which was therefore due to the migration of electrons – just as in metals. Therefore, like metals, these filaments showed a positive resistance/temperature characteristic: their resistance increased with temperature. This made it possible to choose a length and diameter that stabilized at a maximum desirable temperature when supplied with electricity at a given voltage. (Thus, a modern nominally 60W tungsten ‘coiled coil’ lamp showed a resistance of 111 ohms when cold, rising to 1065 ohms and passing 0.23A when operating from a supply at 245 V, i.e 56W).
The high vacuum was essential to prevent burning of the carbon by reaction with oxygen, and thin filaments of considerable length were mandatory to give a high resistance permitting
operation of lamps in parallel. This resulted in a susceptibility to breakage, particularly when relatively thinner spots were developed on ageing. It was therefore inevitable that Edison’s competitors would seek some stronger refractory material that possessed sufficient electrical conductivity to be heated to brilliant incandescence by the passage of an electric current, yet was sufficiently inert not to require enclosure in vacuum.
Walther Hermann Nernst (1864-1941) was a brilliant physical chemist working at Göttingen University (Ref. 1). He was destined to win a Nobel prize in 1920, but well before then, in the early years of the century, he had become widely known for his invention of a ceramic lamp.
Nernst knew that an alternative form of electrical conductivity existed in electrolytes, with the current being carried by charged ions. The electrolysis of water is very familiar in this context, but Nernst’s broad knowledge led him to wonder if a refractory solid electrolyte – with a sufficient mobility of ions at less than a very high melting point – might replace carbon as a filament to give a rugged incandescent lamp. An oxide-based ceramic (i.e. already fully oxidized) should allow operation in air, obviating the requirement for vacuum.
The Nernst lamp
Many researchers of the period knew of refractories that were good insulators at ambient temperatures, but tended to become more conductive as their temperature increased. This was generally regarded as a deleterious property, but Nernst realised that it might form the basis of a ceramic-based lamp (Refs. 2, 3). Possessing more commercial acumen than most academics, he patented the idea (Ref. 4).
Initial research pointed to magnesium oxide (magnesia, MgO) as a possible candidate, but it was necessary to heat it externally above a red heat before it would conduct sufficiently by the migration of ions to maintain itself at a high temperature. This he accomplished in a practical lamp by surrounding a thin rod of the oxide with a platinum heating spiral (Fig. 1). Once sufficient current was flowing in the magnesia rod, it tripped a magnetic relay that disconnected the heating spiral.
Now, this non-metallic oxide conducted electricity by the migration of ions (O—), and like aqueous electrolytes (and gas discharges) possessed a negative resistance/temperature characteristic. This meant that as it heated-up more current flowed – so it got even hotter – so still more current flowed until (hopefully) a fuse ruptured before something else failed. It was therefore necessary to include a metallic ‘ballast’ resistor
in series in the circuit, with oppositely matching characteristics. This is shown in Fig. 2 as the uppermost tubular component, and consists of an iron filament within a hydrogen-filled glass bulb. It operated at a dull red heat to limit the current flowing through the magnesium oxide rod.
The Nernst lamp was sold as a complete assembly (Fig. 3), and was a bulky and expensive object. An obvious improvement was to make the magnesia rod – which tended to fail first – as a replaceable unit along with its associated heater (Fig. 4). Nernst probably also introduced trace elements into the magnesia to increase its luminous efficiency and spectrum: he would have known for example that cerium in the thoria-based Welsbach gas mantle exerted a beneficial effect on its emission. Then, in collaboration with Westinghouse, he improved it still further by using yttria-stabilised zirconia for the filament (Refs. 5, 6). At around 8% the final design was more efficient that the carbon filament lamp (Ref. 7), but for domestic and commercial lighting it could not really compete with the much simpler tungsten filament lamp filled with argon that was introduced in 1904. In particular, the latter did not suffer from the annoying half-minute delay after switching on – a defect shared with the modern coiled fluorescent lamps that we are now expected to use.
The ‘Nernst glower’ hung on for a while as a source of infra-red radiation in IR spectrometers and similar apparatus, but the need for preheating was always a nuisance. Silicon carbide8 then began to be made with sufficient conductivity at room temperature not to need preheating, and soon replaced other materials in IR sources and high temperature furnaces.(The ‘carborundum’ of commerce is a dark blue-grey, but this is due primarily to iron as an impurity. Pure SiC is a colourless, transparent, highly refractive and very hard crystalline solid, that may be cut and polished into diamond-like jewellery (Ref. 8)).
The Nernst lamp is now recognized as employing the first commercially-produced solid electrolyte, and as such is the progenitor of modern developments in solid oxide fuel cells (Ref. 9). However, the non-aqueous electrolysis of alumina dissolved in a molten eutectic of synthetic cryolite (Na3AlF6) and calcium fluoride (CaF2) has long been of great economic importance for the production of aluminium.
The conductivity of hot glass
The zirconia-based ceramic used in the later Nernst lamps is not readily available in small quantities, but the principle may be demonstrated with ordinary soda glass. At ambient temperatures this is a very viscous supercooled liquid, but when heated to a red heat the viscosity decreases greatly and the contained sodium ions (Na+) become sufficiently mobile for the matrix to display substantial electrical conductivity (Ref. 10). (Borosilicate glass – ‘Pyrex’ –is not suitable.) A fragment of soda glass (e.g. from a broken bottle) is held between two 4 mm loops of nichrome wire spaced about 5 mm apart upon rigid conductors of copper wire insulated with glass fibre. (Brass clamps removed from plastic connecting blocks serve very well for these connections.) The assembly is held between two pieces of kaolin-based heatproof board, and the glass melted with a butane torch (Fig. 5).
When cool, the glass bead mounted on its electrodes is inserted in the circuit diagrammed in Fig. 6, the 60 watt incandescent lamp acting as a suitable ballast resistor.
Wearing insulating rubber gloves and a Perspex/Plexiglass face shield, the glass bead is carefully heated once again. It will begin to conduct at around a red heat, and the tungsten lamp will light up. The hotter it gets, the more current will flow, and the lamp will become brighter. The butane torch may then be removed, the heat generated by resistive heating in the glass bead being sufficient to keep it molten. The yellow radiation characteristic of sodium is emitted (Fig. 7), and peculiar twinkling points may be seen around the electrode wires. My set-up, employing a 60 W incandescent lamp as the ballast resistor, operated with 64 V across the bead and a current of 0.2 A flowing. This is equivalent to 13 W dissipated in the hot glass, with a resistance of 320 ohms. This ‘solid state sodium lamp’ operated for > 10 minutes, but eventually the hot glass sagged and broke the circuit.
It will be obvious that this demonstration should only be conducted by experienced laboratory personnel with adequate safety precautions.
- K. Mendelssohn, The World of Walther Nernst: The Rise and Fall of German Science, Macmillan, 1973. A chronology of Nernst’s life may be found here: Chronology
- H. Monmouth Smith, ‘The Nernst lamp’, Science, 11 (1898) 689-690.
- Anon., ‘The Nernst light’, Scientific American, 80 (1899) 150.
- W. Nernst, Verfahren zur Erzeugung von elektrischen Glülicht, German patent DRP 104872, filed 1897.
- W. Nernst, Material for electric-lamp glowers, American patent USP 685730, filed 1899.
- Yttria-stabilized zirconia. Wikipedia.
- L.R. Ingersoll, ‘On the radiant efficiency of the Nernst lamp’, Phys. Rev., 17 (1903), 371-7.
- Silicon carbide. Wikipedia.
- H.-H. Möbius, ‘On the history of solid electrolyte fuel cells’, J. Solid State Electrochem., 1 (1997), 2-16.
- W.Thomson, ‘Electrolytic conduction in solids – first example, hot glass’, Proc. Roy. Soc., 23 (1874-5)