The Contributions by Volta, Children and Pepys

The story effectively starts with the letter, dated 20th March 1800, of Alessandro Volta (1745–1827) describing his famous pile (4), which led to a flurry of interest in things electrical and, in 1809 (5) and 1813 (6), to the development of two large batteries by John George Children (1777–1852). These were enormous: the 1815 development of the 1813 model, for example, had 63 plates of zinc and copper each measuring 6 ft by 2 ft 8 in dipping into 945 gallons of a 3:1 mixture of fuming nitric and sulphuric acids diluted with water, and the flow of electric current was stopped by raising the plates out of the acid by a system of ropes and pulleys. This battery was located at his father’s house, Ferox Hall, near Tonbridge, Kent (7), and was also used by several other Fellows of the Royal Society, including Sir Humphry Davy, for electrical experimentation. For his work with his batteries Children was awarded a Royal Society Medal in 1828. He established, among other things, that the heating powers of metals varied with their specific resistance; but the most interesting observation from our viewpoint was that of William Hasledine Pepys (1775–1856) who was interested in the controversy then raging on whether diamond could convert soft iron into steel on heating (8), or whether the necessary carbon came from an outside source (9,10,11). Use of Children’s battery ensured that there would be no source of carbon other than diamond and accordingly (6):

So was born the first electrically heated wire resistance furnace, although the heating element was not platinum.

The Lost Years, 1815–1893

Why did nearly a century elapse between the above experiment and the invention of the platinum-wound tube furnace?

The answer is not to be found in lack of technical advance, since continuous current could be obtained from reliable compact batteries before 1840 (12, 13) and the electrical generator had been developed from the rudimentary form of Hippolyte Pixii in 1832 to the commercial model of Zénole Theophile Gramme (1826–1901) in 1870 (14,15). Filament lamps, which employ a somewhat analogous principle for another purpose, developed over the period 1820 to 1878 from early attempts to use platinum wires in a vacuum to usable glass-enclosed lamps (15,16). Moreover, in 1841, James Prescott Joule (1818–1889) quantified the amount of heat liberated when an electric current is passed through a resistance (17).

However, the reference by Powell and Schroeder (15) and Jarvis (16) to the 1820 work of Warren de la Rue must be treated with caution as de la Rue was then only 5 years old. Comments by Grove (Phil. Mag., 1840, 16, 338–339) suggest the reference might be to de la Rive, and presumably to Gaspard (1770–1834), since the first scientific paper of Auguste (1801–1873) appears to have been published in 1822.

Lack of interest is likewise not a tenable answer as two reviews which appeared about the turn of the century reveal a continuous interest in certain types of electric furnace. Thus, Street (1) in 1895 considered such furnaces could be divided into two main classes: “les uns … basés sur l’incandescence d’une fraction résistante de circuit et les autres sur l’emploi de l’arc électrique”, the first referring only to systems where the resistance of the contents was the source of heat. Minet (2) in 1905 lists three stages of development, that in the period 1808 to 1886 being attributed to “laboratory furnaces”, but again refers only to the types recognised by Street. In his final complex classification of furnaces, a type not discussed in the text is given as a “solid resistance not in contact with the material” and under this he has “iron wire round the reaction chamber” and “platinum wire” associated with the names “Charpy (1893)” and “Nernst (1900), Howe (1900)” respectively but with no references. In general, therefore, it would appear that electric furnaces were developed with essentially technical and industrial applications in mind—and Street’s two classes would fit these requirements.

Another factor in the delay was undoubtedly the use of coal gas. With the introduction of the Bunsen burner in 1853 (18) gas became the normal source of heat in laboratories and small gas furnaces and muffles (19,20) were adequate for most laboratory purposes. Electricity, in contrast, was not readily available until into the twentieth century and even then the position varied from laboratory to laboratory, as illustrated by the use of gas heating for thermogravimetry as late as 1925 (21).

Analysis of these facts suggests that the main causes of delay in the introduction of the wire-wound furnace were the ready availability of laboratory gas furnaces, the familiarity of scientists with gas and possibly the fact that highly accurate temperature control was not then considered essential for many of the operations performed.

Le Chatelier, Osmond, Moissan and Charpy

In the late 1880s, however, interest in metallurgy grew, and convenient temperature control became a reality in 1886 when Henry Louis Le Chatelier developed the thermocouple as an accurate temperature-measuring device (22), thereby confounding earlier prognostications (23). At about this period too, Floris Osmond (1849–1912) carried out his extensive metallurgical investigations (24) and Ferdinand-Frédéric-Henri Moissan (1852–1907) was involved in related studies. It was in this climate that Augustin Georges Albert Charpy entered l’École Polytechnique in 1887 (25,26). Becoming an Assistant in the Chemistry Course in 1889, he gained his doctorate in 1892 and then, not surprisingly as Henri Moissan had presented his papers to the Academy of Sciences from 1891 on (27), he commenced metallurgical studies at the Laboratoire Central de l’Artillerie de Marine. His first paper on this subject, published in 1893 and dealing with the effect of annealing on the physical properties of brass, contains the comment (28):

This is the first record in the literature of a wire-wound electric resistance furnace—and the winding was platinum, not the “iron wire” mentioned by Minet (2), whose notes, regrettably, are not always accurate. Berthélemy was thus quite correct in stating in 1947 (26) that “il a construit le premier four à résistance électrique, à éléments de platine, qui nous paraît très simple, mais qui fut et qui reste un outil extrêmement précieux pour les mesures”. For a study of the allotropic forms of iron, in 1894, Charpy used a similar furnace of which he gives more details (29). The platinum wire was wound around a refractory tube 20 cm diameter by 60 cm long, giving a zone 20 cm long in which the temperature, as measured by a thermocouple, remained constant for several hours over the range 500 to 1300°C. The furnace tube was insulated with asbestos, the current was regulated by a rheostat and the furnace was pivoted so that it could be quickly tilted vertically to allow the iron bar being heated to fall into the tempering bath without its temperature falling significantly in transit.

On 25th November 1892 the Council of La Société d’Encouragement pour l’Industrie Nationale offered a grant of 3000 francs for studies on the tempering and annealing of steel to Floris Osmond, who at first accepted but later withdrew as his research had taken a different direction. Thereupon the Council approached Georges Charpy who, with the approval of his superiors at the Laboratoire de la Marine, entered into an agreement with the Society in June 1894. After visiting his laboratory, a Committee of Council, who were most impressed by his research, handed over to Charpy the 3000 francs previously set aside for Osmond (30). On 25th January 1895, Charpy described his work to a meeting of the Society and this was published in their Bulletin for June 1895 (31). This remarkable piece of work for such a short period deals not only with the tempering of steel in detail but also describes several platinum-wound electric resistance furnaces and was very highly praised by the Society’s assessors, who refer to “la profondeur de ses recherches, la vigeur et la précision de ses methodes et l’originalité de ses procédés d’investigation” (30). Apparently Charpy also exhibited one of his furnaces at the meeting on 25th January; as this excited great interest (32), one wonders why it is not referred to in Street’s review (1) and how Minet (2) got his facts wrong.

The first illustration of such a furnace and its associated temperature measuring and recording arrangement is reproduced herewith; Charpy describes its construction as follows:

A rather more complex furnace was used by Charpy in his studies on the tempering of steel. It consisted of a platinum-wound refractory tube 60 cm long by 2.5 cm diameter insulated with asbestos inside a brass tube sealed with end-pieces. The refractory tube was extended at each end by cooled bronze tubes forming axles set in bearings fixed to the outer frame: one axle had a pulley attached and the other the electrical contacts. This arrangement was rotated by an electric motor attached to the outer frame and again the whole system was pivoted so that it could be rapidly brought to the vertical position, allowing the metal bar sample to drop into the tempering bath. Temperature profiles, obtained using a thermocouple, showed that the sensibly “constant-temperature” region tended to become smaller as the temperature increased up to 1000°C, although, even at this temperature, the variation was less than 10°C over 15 cm in the centre of the tube. Temperature control was by manual adjustment of a rheostat. The winding used consisted of four platinum wires 7 m long and 0.15 mm diameter arranged in parallel; two wires were wound directly on the refractory tube and were covered with a sheet of asbestos on which were wound the other two. In an appendix to the paper, Charpy gives instructions for constructing a similar furnace 40 to 50 cm long but without the rotating arrangement or water-cooled ends, indicating that this can be used when great temperature precision is not required. In this furnace, constructed for Charpy by M. Jobin, precision instrument maker, Paris:

Thus, not only did Charpy construct the first platinum-wound furnace in 1893, but in the next year he arranged it on a pivot so that it could be turned rapidly from the horizontal position to the vertical, and in 1895 he produced a rotating, pivoted version with water-cooled ends that seems complex even today. He also used part of the winding as an inbuilt platinum resistance thermometer for less precise work. Clearly he was an engineer of resource and ingenuity and it is little wonder that, later in 1895, he was awarded a gold medal of the Société d’Encouragement pour l’Industrie Nationale for his efforts (33).

Developments in the Period to 1900

An interesting electric resistance furnace with a spherical cavity, heated by coils of iron wire, rotating on axles and fitted with an inner iron sheath was constructed by William Chandler Roberts-Austen (1843–1902) in 1897 and was used for calibrating thermocouples against an air thermometer (34). Iron wire could not give the temperature stability or reproducibility of platinum—although Le Chatelier, by packing the space around an iron winding with charcoal, claimed long life when used up to 1000°C (35)—but Robert-Austen’s furnace was as carefully constructed as Charpy’s and was perfectly adequate for the purpose for which it was made. In 1899, Holburn and Day constructed a tube furnace, also for calibrating thermocouples against an air thermometer, using a winding of nickel wire (36). Base metal windings were thus introduced at an early stage; obviously these were cheaper than platinum, but a contributory factor may have been the difficulty of obtaining pure platinum wire. Many of the papers at that period comment on the fact that specially purified metal must be used for windings and claim that the iridium, which was apparently then a normal impurity, tended to evaporate during heating, contaminating the furnace. Although this seems most unlikely, furnace tubes certainly tended to become conducting after a period of use and poor quality materials gave rise to much trouble. For example, in 1898 a batch of commercial rhodium for the construction of thermocouple wires was found to contain 30 per cent iridium (37) and, up to at least 1908, the quartz present in British porcelain refractory tubes (but not in Berlin porcelain) gave rise to spurious thermal analysis effects (38).

Early Commercial Manufacture

Platinum-wound laboratory resistance furnaces, both tube and muffle, were first produced commercially by the firm of W.C. Heraeus, Hanau, Germany, around 1900. Repeated heating and cooling caused the thin platinum wire spiral on early models to expand and separate in places from the ceramic tube with resultant non-uniformity of temperature distribution and development of local hot-spots where the wire eventually fractured. To overcome this, Heraeus in 1902 developed a furnace using a very thin platinum ribbon (0.007 mm thick) in place of the wire (39). This ribbon required only about one-sixth of the amount of platinum, that is 6 g for a furnace 20 cm long (40). As the coils were laid 1 mm apart there was virtually a complete platinum cylinder around the ceramic tube. Because of their advantages of uniformity of heat supply and long winding life, these furnaces were highly regarded and were extensively used during the early part of this century. An example is illustrated herewith (41).

Heraeus claimed that furnaces with tubes 2.5 cm in diameter could be heated to 1400°C in about 5 minutes and could attain 1700°C without fusion of the winding (39). The refractory material available for the tube, however, became electrically conducting above about 1500°C. Haagn (42) confirmed that, for 6 to 8 hours’ operation, the practicable maximum was 1500°C, although the furnace could be heated to 1700°C for a short period without permanent damage.

Later Developments

Despite the ready availability of these furnaces, many laboratories still preferred to make their own, sometimes for financial reasons but sometimes because the sizes available were inappropriate for the work to be performed. For example, in 1902 the Physics Division at the National Physical Laboratory, Teddington, constructed two large electric furnaces for a McLeod, gas thermometer reading up to 1200°C, two large furnaces for thermocouple comparison and two smaller furnaces for miscellaneous work. By 1904 the Thermometry Department of the Physics Division had 5 wire-resistance furnaces. The winding material is not known with certainty but it would appear to have been platinum or nickel; manganin or eureka wire then being used for low-temperature studies.

Graded windings, to lengthen the uniform-temperature zone, appear to have been first introduced with nickel wire in 1903 (43). In 1904, at the U.S. Geophysical Laboratory, where much early development took place, platinum-iridium windings were placed inside the furnace tube to obtain temperatures 200°C higher than those obtainable with a conventional system (44); in 1908 this arrangement was further improved by using graded windings along with supplementary end-coils to give temperatures varying by not more than 1°C over the central 20 cm of a tube 30 cm long by 6 cm internal diameter (45). Within fifteen years from its introduction, therefore, the wire-wound resistance furnace had developed into a precision tool and its technology was highly advanced.

The usual sources of power for these furnaces were high-capacity accumulators of 70 to 100 volts output, or occasionally dynamos, as these provided a more reliable output than mains electricity, where this was available (43,44). By 1912, however, local mains supplies were apparently sufficiently stable in some places to be useful (46). Power consumption is only very rarely stated, but seems to have varied enormously—for example, that of the Heraeus range is quoted as 400 to 2000 watts (42), but 3000 and even 4000 watts are mentioned elsewhere (44,47).

Temperature control was normally effected as described above by Charpy (31), that is using visual observation of the galvanometer deflections and manual adjustment of a rheostat. Temperature indicators for use with thermocouples were introduced at an early stage and in the 1900s control at a fixed temperature was effected by using the needle of the indicator as a relay contact. The first automatic control system for continuously increasing temperature seems to have been introduced in 1912 (46); despite the mechanical nature of this, the linearity of the temperature/time curve is impressive.

Replacement of gas furnaces by electric ones was generally slow. Evidence in the literature suggests that by 1906 gas and electricity were fairly equally utilised, whereas by 1909 electricity had probably become more common. Yet the preference for gas was very strong in some countries and gas furnaces remained in use for purposes where electricity would have been more convenient well into the 1920s (21). Moreover, visual temperature observation with manual control persisted until the 1950s (48).

Perhaps it is because of this gradual transition that the history of the platinum-wound electric resistance furnace has been so neglected.