ON THE SPECTRUM OF BORON.* By Sir WILLIAM CROOKES, O.M., For.Sec.R.S.

In 1902 Exner and Haschek again took measurements of the lines in the ultra-violet spectrum of boron from a sample of boric acid obtained from Merck, and gave the wave-lengths of the three lines as 3451 49, 2497'79, and 2496.87.

In 1904 A. Hagenbach and H. Konen (1905, Jena, G. Fischer; London, W. Wesley) published photographs of the boron spectrum in which the wave-lengths of the dominant lines are given as 3451, 2498, and 2497. The spark was taken between carbon poles with boric acid. Many other lines are shown, but the authors ascribe these to the electrodes.

* A Paper read before the Royal Society, November 9, 1911. + The lines so marked are the dominant lines.

In March, 1901, I was examining the silicon spectrum, using fused silicon electrodes, and a pair of lines were seen at about 2497 and 2498 which could not be found in any silicon spectrum photographed by other observers. I soon identified these strangers as lines given by a boron impurity in the fused silicon. A pure sample of fused silicon supplied by Johnson and Matthey did not show these lines.

Properties of Boron.

Amorphous boron when pure is a fine dark brown powde Its conductivity for electricity is very poor, and it is difficult to manipulate so as to get an induction spark through, as it is readily blown away. I could not fuse it in the arc.

Quite recently, Dr. Weintraub, of the West Lynn Research Laboratory, General Electric Co., U.S.A,, has prepared Hitherto the boron in a solid state and chemically pure. physical properties of this element were unknown, notwithstanding the efforts of many chemists who had worked on the subject. Moissan, who came nearest to obtaining the pure element, only succeeded in getting it in the form of an amorphous powder. He said it was not possible to melt or volatilise boron in a carbon crucible or arc as it was changed into carbon boride, and concluded that boron passed from the solid to the gaseous state without becoming liquid.

Dr. Weintraub has not only obtained boron in a state of purity but has prepared it in a fused homogeneous form. His process consists in running one or more alternating current arcs, fed by a high potential transformer, between water-cooled or air-cooled copper electrodes, in a mixture of boron chloride with a large excess of hydrogen in a glass or copper vessel. The boron is thrown off in fine powder on to the walls of the vessel or agglomerates on the ends of the electrodes, where it grows in form of small rods. After a while the arc runs between two boron electrodes, and if the current is of proper value the rods melt down to boron beads which eventually fall off, whereupon the same process repeats itself.

The first specimens I received from Dr. Weintraub were deposited from a vaporous state from boron chloride Subseand hydrogen in the manner already described. quently I received some lumps of fused boron which had been prepared from magnesium boride obtained in the reaction between boric anhydride and excess of magnesium. The magnesium boride dissociates at a relatively low temperature (1200°), especially in vacuo, and with rapidity at 1500. The fusion is effected between copper electrodes. Under the conditions of the experiment no disintegration takes place, and according to my informant, the affinity of copper for boron is so slight that it can be directly fused on to the electrode without being contaminated with copper. Another way of fusing boron is

in what Dr. Weintraub terms a mercury arc furnace, based on the fact that most refractory bodies, such as tungsten, tantalum, boron, &c., have no affinity whatever for mercury.

The fused lumps of boron obtained by the reaction between boron chloride and hydrogen were found by Dr. Weintraub to be quite pure, analysing from 99.8 to 100.2 per cent. The difference being due to errors of experiment, and perhaps to a trace of silica abraded from the agate mortar during pulverisation.

The boron prepared by fusing that obtained by the gas process is not quite so pure. The sample sent me contains about 97 per cent boron, the main impurities being nitrogen in the form of boron nitride, magnesium in the form of magnesium boride, a trace of iron, and a trace of

carbon.

The most interesting property of solid boron is its extraordinary rise in electric conductivity with slight increase in temperature. A piece of fused boron measured by Dr. Weintraub, which at the room temperature (27° C.) had a resistance of 5,620,000 ohms, dropped to 5 ohms at a dull red heat. Another noteworthy property of both the melted and agglomerated boron is extreme hardness. It comes next to the diamond in hardness, a splinter easily scratching corundum. Its fracture is conchoidal, and when melted no decided crystalline structure is seen. The agglomerated boron, deposited in the arc from boron chloride and hydrogen, is, on the contrary, highly crystalline. A rod of boron heated to whiteness before the blowpipe shows no sign of fusion, but the flame is coloured green, and when the cold rod is microscopically examined it is seen to be covered with minute globules of fused boric anhydride.

The first samples of boron under experiment were of the agglomerated variety, in the form of thin flakes deposited from the vaporous condition in the reaction already described. Several fragments were clamped together to form the electrodes.

It is not easy to get the spark to pass between cold boron poles owing to its high electrical resistance. When the two pole pieces are held in brass clips in front of the spectroscope the spark passes across from one clip to another; it is only when the boron poles get heated by the spark that they begin to conduct sufficiently to let the current pass. The light of the spark is somewhat feeble, and exposures from one to two hours are required to bring out with intensity the three dominant lines. Generally, when the boron electrodes are well heated the current assumes the form of a minute arc, starting from a luminous point at one edge of the pole, which soon becomes red-hot. Occasionally an intense yellow-green flame shoots from a corner which immediately fuses, but this only lasts a few seconds. After the current has passed for forty or fifty minutes an accumulation of boric anhydride causes a resist ance to the current, and it then again begins to pass across between the brass clips.

The melted boron is easier to manipulate in the spectro graph. In the form of solid blocks, the electrodes do not get so hot, and very little oxidation takes place. The spark is of a faint apple-green colour, and when the poles have become sufficiently hot to carry the current it passes quietly for hours without change.

The spectrum of boron consists essentially of three lines, the wave lengths of which, according to careful measurements made in the manner described in 1903, are 3451'50, 2497 83, and 2496-89 ("The Ultra-violet Spectrum of Radium," Proc. Roy. Soc., 1903, vol. lxxii., p. 295) For easy comparison I give in a tabular form the wave-lengths of these lines measured by different observers :Hartley (1883). 3450 3 2497

Rowland (1893)

Eder and Valenta (1893) Exner and Haschek (1897)

..

2496.2 2497 821 2496-867 2497 7 2496.8 2497.8 2496.88 2496.87 2498 2497 3451 50 2497 83 2496.89

34513

3451 4 3451 49 2497'79

3451

Besides these three lines two others of wave-lengths about 3274 and 3248 were seen on the photographs of each sample of boron, the agglomerated and the melted, the lines in the melted being the stronger. The fact that no other observer had noticed these lines, and their being of different intensities in the two samples, proved that they were due to some other element accidentally present. A run over my photographed spectra of the elements soon showed these lines to be the dominant copper lines, 3274 096 and 3247-688. A short time ago Sir Walter Noel Hartley (Roy. Soc. Proc., vol. lxxxv., p. 271) traced the occurrence of these two copper lines in the spectrum from pure cadmium electrodes to the proximity of his laboratory to the overhead conductors of the Tram-lines. The condensation of the vapour of copper following the sparking at the rubbing contact yielded a dust of extreme tenuity, such as could not arise from mechanical abrasion of the solid metal. Sir Walter Noel Hartley found that the amount of copper passing between the sparking terminals sufficient to produce an impression of the copper lines on the photographic plate need not be more than from o'o01 to 0.0014 mgrm. This explanation could not apply to my own case, for no tram-lines are near the laboratory and nothing was going on that could give rise to a dust of copper. The atmosphere of my laboratory was perfectly free from any trace of copper, but the clips that hold the electrodes in front of the slit of the spectrograph were made of brass. When the spark first passes the boron is cold, and in that condition is a very bad conductor; consequently, some discharge may take place between the brass clips. After a little time, however, the boron gets hot enough to conduct the whole current.

To guard against such an accidental contamination, I made a pair of clips of pure gold for boron and other electrodes, and repeated the spectrographic tests with each sample of boron.

My reasons for selecting gold as the material for the clips were that it shows no lines near those in question of copper or aluminium; that it is a soft metal well adapted for clips holding hard bodies; that I had convenient blocks of it in a state of purity; and that all the lines of gold have been measured and mapped with accuracy.

The poles of agglomerated boron after sparking for two hours-using gold clips-gave a spectrum in which no trace of copper lines could be seen, whereas a similar experiment in which melted boron was used in the gold clips showed decided lines of copper and traces of magnesium.

Another pair of lines occurred in the neighbourhood of wave-lengths 3082 and 3093 in the spectrum given by the melted boron, but not in that of the agglomerated plates. This pointed to other impurities in the melted boron. On comparing these lines with spectra of other elements it was soon discovered that they were two of the dominant lines of aluminium, 3082-275 and 3092-818.

These

Experiments were now made to see if by greatly prolonged exposures in the spectrograph other boron lines could be brought out. Melted boron, sparked for seven hours in gold clips, gave a photograph showing many additional lines, which, however, were for the most part ill-defined air lines; four, however, occurring between wave-lengths 3930 and 3970, were definite and sharp although faint. four lines might be due to boron, or to impurities, or they might be air lines. To ascertain if they were air lines a control experiment was tried by photographing the spectrum of pure tungsten sparked for seven hours. Tungsten was selected as the metal to put in comparison with boron because in its spectrum there is a blank space where the four lines brought out by long exposure of boron occurred. Had any of these four lines been brought out by long exposure in the tungsten spectrum it would show that they were common to both and not peculiar to boron. Close examination of the tungsten spectrum showed no trace of the four lines. The two photographs were enlarged, and most of the lines in common were found to be air lines.

It having been proved that the four lines in question were not air lines, the remaining alternatives were

CHEMICAL NEWS, Jan. 5, 1912

spectra.

The result of my work on boron is to show that its photographed spectrum consists of three lines; that the fourteen other lines given by J. M. Eder and E: Valenta, and the five other lines given by F. Exner and E. Haschek failed to record themselves on my photographs, notwithstanding the excessively long exposures I gave in the attempts to bring out additional boron lines.

A RE-DETERMINATION OF THE DENSITY AND COEFFICIENT of LINEAR EXPANSION OF ALUMINIUM.*

By F. J. BRISLEE, D.Sc., &c., Chief Chemist, British
Insulated and Helsby Cables, Ltd.

THE increase in the use of aluminium for many technical purposes has rendered a re-determination of some of its physical constants necessary. The metal, as now placed upon the market, is in a high state of purity; but many of the earlier determinations of its physical constants were made upon specimens of aluminium of doubtful purity and unknown composition, which probably contained consider. able quantities of iron, silicon, carbon, and oxide of aluminium, the oxide being mixed through the metal as an emulsion. Aluminium absorbs certain gases, when molten; hence the difficulty of obtaining sound castings; and blowholes would cause too low a value for the density to be obtained. The aluminium used in the determinations of the density and coefficient of linear expansion, described below, was obtained from (1) the Société Electrométallurgique Française, La Praz, Savoy, and (2) from the British Aluminium Company. A few specimens of remelted scrap metal were used in one or two of the density determinations. All the metal used was carefully analysed, and the iron, silicon, the chief impurities present, were estimated. The iron was estimated by dissolving 10 grms. of the metal in caustic soda. The solution was allowed to stand, and the clear liquid syphoned off, water was then added, the preeipitate again allowed to settle, and the clear solution removed by a syphon. The precipitate was then filtered off, thoroughly washed with water, dissolved in hydrochloric acid, precipitated with ammonium hydrate after oxidation, and weighed as oxide, or the hydrochloric acid solution was reduced and titrated with a N/10 solution of potassium dichromate. Repeated determinations, made upon the same sample, showed a good agreement between the two methods. The silicon was determined by the method of Otis-Handy. (Described in Lunge's" ChemischTechnische Untersuchungsmethoden," ii., p. 789). The analyses of the aluminium gave, as a mean of many determinations:

* A Paper read before the Faraday Society, December 6, 1911.

Silicon .. Iron Aluminium

S.E.M.F. Per cent.

3

British Aluminium Co.

Per cent.

0.26

O'25

0.26

0.23

99'48

99'52

The specific gravity was determined upon the cast metal, soft annealed rod § in. diameter. The determinations were as received, upon hard-drawn rod in. diameter, and upon made by both the ordinary method of weighing in air and water, and by employing the displacement method, using wire. The cast aluminium was cut into rectangular blocks, a 100 cc. flask, the latter method being employed for fine weighing from 20 to 30 grms. each; the round rod was cut into cylinders of about the same weight. The fine wire was cut into small pieces about an inch long; the weight used for a single determination was from 10 to 30 grms. The aluminium was carefully cleaned with petroleum ether, in order to remove grease, &c., which might have been taken up by the metal during working. The temperature of the water was taken at the time of weighing. The results obtained are given in Table I.

The density of re-melted aluminium is lower than the above, due probably to the fact that a certain proportion of iron and silicon are taken up during melting, and also to the fact that gases are absorbed while molten. The value found for the density of re-melted aluminium was The most probable 168 16=2.687, or 16 8 4=2·6821. value for the density of aluminium is 2.708 for cast metal, and metal upon which a large amount of work has not been done. The worked metal reaches 2'72, in some of the above instances, but this value was obtained in only three determinations, and was not found for hard-drawn wire.

The above values differ from the values recorded in the literature, e.g., in Landolt and Börnstein (" PhysikalischChemische Tabellen "), and in a paper by Wilson (Fourn. Inst. Elect. Eng., 1901, xxxi., 332) in which a large amount of data are collected.

The Coefficient of Linear Expansion of Aluminium.

The coefficient of linear expansion was measured directly by determining the increase in length of a rod of the metal, about 1 metre long, when it was heated from 10° C. to 100° C. The apparatus is shown in the figure. It consists of a stout mild steel tube A, made of hydraulic tubing, which carries two heavy bronze clamps, B1, B2, which can be moved at will, and rigidly fixed in any position by means of clamping screws. BI carries a mild steel stop D, against which the end of the aluminium bar G rests when the determination is being made; the other (B2) carries a micrometer head c for measuring the change in length. The tube A is supported by the bronze supports M, N, which in turn were screwed on to the wooden blocks and base WI, W2, W3. A steady stream of water was kept flowing through A, the temperature of which was measured by the thermometer T4, placed in the glass vessel F. The temperature of the water did not vary more than 0.2° C. during the measurement, and the slight error caused thereby can be neglected. The aluminium bar was placed inside the water jacket K, and supported by a half tube, i.e., a brass tube cut out so as to form a shallow tray underneath the aluminium rod, the tube being complete only at the ends which carry the stop D and the stuffing-box H. The temperature of the bar was measured by the three thermometers T1, T2, T3. The change in length was measured by the micrometer C. Two micrometers were used, made by Messrs. Brown and Sharpe, and measured to o'oor mm. and o'0001 inch respectively. The increase in length of the bar G pushed the piston P through the stuffing-box H, and the change in length was then measured by the micrometer head c, the temperature remaining constant during

The aluminium rods were cut so as to be approximately I metre in length. The ends were then turned up in a lathe, exactly at right angles to the long axis. The exact length was determined by comparison with a Whitworth standard steel bar I metre long, the difference being measured by a micrometer. The rod was then placed in position in the water-jacket K, which was well lagged with asbestos, the stop D, and the piston and the stuffing-box, P and H, fixed in position. Cold water was allowed to flow through the apparatus until the thermometers showed that the temperature was constant to within o'1° C. The thermometers were calibrated from time to time so as to check the constancy of their indications. The thermometers showed that the temperature was practically constant during the actual determinations. The micrometer was then screwed down until the piston P pressed upon the end of the rod, and a reading of the position of the micrometer was then taken. A series of readings were taken of the thermometers and the micrometers, over a period of half-an-hour, so as to ensure that the temperature and

was then run out, and a steady current of steam was maintained, until the thermometers showed that a constant temperature was reached. The micrometer c was again screwed down on to P, and the reading taken. A series of readings were taken of both micrometer and thermometers. The bar was kept at the temperature of the steam for half-an-hour, and comparative tests showed that temperature equilibrium between the steam and metal was established in ten to fifteen minutes. At the completion of the measurement the steam was shut off, and cold water circulated through the apparatus. When the initial temperature was reached, the first readings of the micrometer were checked. The readings at the higher temperature were repeated by again heating to the temperature of steam. It was found that successive determinations made with the same bar agreed to within o 2 to 0.3 per cent. During the whole of the measurements the temperature of the tube A was kept constant by a steady current of water, and the bronze clamps B1 and B2 were shielded from radiant heat by thick asbestos screens. The temperature of the bar, as indicated by the thermometers T1, T2, T3,