Page images

effects, and to determine if possible to what definite compound each particular effect was due.

The effect of introducing a bead of cupric chloride into a Bunsen-flame has been carefully described by Lecoq de Boisbaudran (Spectres Lumineux, p. 156). As already stated, he distinguished four cases, but of these we need only take the first, in which a large quantity of salt is used. The salt first melts, and then in a few moments is seen to be surrounded by a brilliant patch of yellow, like a piece of ordinary candleflame; the exterior parts of this patch are reddish (again resembling carbon luminosity). Outside the yellow a bright blue colour appears, and outside this the flame is coloured green. The yellow luminosity is of short duration, the blue lasts longer, but soon the only tint remaining is the green.

These three effects were so local and distinct that it seemed possible to collect the substances to which they were due, and this was done by holding in the flame glass or porcelain basins filled with water. The deposit obtained in this way from the yellow part of the flame was red in colour, it transmitted greenish light, it could be burnished, it dissolved in nitric acid with evolution of red fumes, and in fact answered in every respect to metallic copper. The yellow luminosity observed with a large quantity of cupric chloride in a Bunsenflame must, therefore, be attributed to the liberation and incandescence of minute particles of solid copper.

The deposit obtained from the blue part of the flame was of a very pale yellow colour when freshly formed. On standing or by breathing upon it, the film absorbed moisture and became quite white: it answered in all its properties to cuprous chloride containing a little of the cupric salt.

The green part of the flame produced a deposit which was almost black and corresponded in appearance and chemical properties to cupric oxide. As the film was very thin, the possibility of it having been originally cuprous oxide and having subsequently oxidized was not excluded.

From these experiments it is obvious that the three distinct colour-effects noticeable when cupric chloride is introduced into a Bunsen-flame correspond to three different substances, viz., metallic copper, cuprous chloride, and an oxide of copper.

It is highly improbable from a chemical point of view that cupric chloride when introduced into a flame should afford a spectrum. The easy decomposability of this salt and the stability of cuprous chloride (which is volatilizable without change) would lead one to anticipate the decomposition of CuCl into Cu,Cl, and Cl2 long before there could be any question of incandescence. That this is the case can be

easily seen by holding a bead of cupric chloride well above the tip of a small Bunsen-flame, and supporting above it a porcelain basin filled with cold water. Though no flame is seen the salt melts and volatilizes sufficiently to give a considerable film on the basin, which, on examination, is seen to be almost wholly cuprous chloride.

Taking it as established that the blue part of the flame is due to cuprous chloride, the question arises, How does this salt become converted into oxide of copper to which the green part is due? This could be easily accounted for by the action of steam upon the chloride. That this is the correct explanation is established with something like certainty by experiments with flames of combustibles containing no hydrogen. In the flame of carbon monoxide or carbon bisulphide, cupric chloride produces almost exclusively the blue colour; whilst in flames. of hydrogen in which, according to the hypothesis, the existence of cuprous chloride should be precluded by the presence of steam, the green colour greatly predominates. Again, a small jet of hydrogen or steam impinging on a carbon-monoxide flame coloured intensely blue by cuprous chloride produces the green at the point of contact.

The remaining point of interest is to ascertain which of the oxides of copper produces the green colour. When carefully purified CuO is dusted on to a Bunsen-flame, a green tint is at once produced; and the deposit obtained on porcelain from the green part of a copper-chloride flame appears to be black at the instant of deposition. These facts point to cupric oxide as the substance which produces the green colour. On the other hand, cupric oxide is a substance which loses oxygen at a high temperature. The dissociation of cupric oxide was studied by Debray and Joannis (Compt. Rend. xcix. p. 583, 1884). Heated in vacuo they found it to yield oxygen at 350° C. At the melting-point of silver the tension of this oxygen amounted to 56 millim., and a little above the meltingpoint of gold to 1000 millim. It would appear to be impossible for cupric oxide to exist at the average temperature of a Bunsen-flame, for this would demand an oxygen-tension vastly greater than exists within or around such a flame. It seems therefore necessary to ascribe the green glow to cuprous oxide or some lower oxide of copper, but I have been unable after several attempts to volatilize cuprous oxide by external heating.

It has already been stated that a cuprous-chloride flame containing plenty of hydrochloric acid is surrounded by a dull red margin. This appears to be due to cupric chloride. When a super-aërated flame is obtained in the separator and

is supplied with a spray of copper chloride, it is tinged wholly green and is surrounded by a green halo. Round such a flame the existence of cuprous chloride is impossible, and when a piece of asbestos, soaked in hydrochloric acid, is introduced into the halo we do not get as a matter of fact any blue colour, but the ruddy margin immediately appears, and it seems impossible to attribute this to anything but the formation of cupric chloride, which, in the presence of oxygen and hydrochloric acid, remains as such and gives the feeble red glow.

The behaviour of cupric chloride in the separator admits of easy explanation. In the inner cone the average temperature is extremely high, and the products there generated consist largely of carbon monoxide and free hydrogen. The cupric chloride will therefore not only lose its chlorine but the cuprous chloride, if we suppose it to be formed for a moment, will be immediately attacked by the reducing gases and lose the remainder of its chlorine. We have therefore metallic copper, and the average temperature is not sufficient to produce its characteristic spectrum. The copper and the hydrochloric acid resulting from the decomposition of the cupric chloride pass upwards. Some of the former is deposited as a thin metallic film on the inner walls of the outer tube. The rest passes to the outer cone where, at the lower average temperature and in contact with atmospheric oxygen, some cuprous oxide is formed and gives the green tint and oxide. spectrum. At the same time some cuprous chloride is formed by the hydrochloric acid, and gives the faint traces of the chloride spectrum.

It is important to observe that, according to the above explanation, cupric chloride is decomposed in the inner cone without evincing any spectrum at all. The fact that the salt is easily reduced to cuprous chloride at a comparatively low temperature, and that cuprous chloride can itself be easily deprived of its chlorine by heating in a current of gases of the same composition as those that are passed into the separator, would lead one to the conclusion that the salt undergoes these changes before actually entering the inner cone. In this case we might at first expect to see some coloration below it, and, indeed, Gouy (loc. cit. p. 29) describes such an appearance, stating that a line of coloration is distinctly visible parallel to and within the cone. I have repeatedly tried to verify this observation, but could not succeed in doing so either by the eye or the spectroscope. The matter is, however, of no great consequence.

Phil. Mag. S. 5. Vol. 39. No. 236. Jan. 1895. K

Flame-Spectrum of Gold Trichloride.

Owing to the easy decomposability of gold trichloride and the desirability of ascertaining more accurately the source of the spectrum commonly attributed to it, I have examined it with the separator.

When gold trichloride is introduced on the end of a platinum wire into a Bunsen-flame, a bright green flash of light is produced accompanied by some bright sparks. The coloration only lasts a very short time, and a considerable residue of metallic gold is left on the wire. These facts are noted and a map of the spectrum is given by Lecoq de Boisbaudran (Spectres Lumineux).

A moderately dilute solution of gold trichloride, introduced into the separated cones by means of the apparatus used for copper salts (p. 123), does not colour either of them; but when a piece of asbestos moistened with strong hydrochloric acid is held in the upper cone, a brilliant green colour is produced and the green lines are seen in the spectroscope. When the coal-gas is previously passed through the saturator containing asbestos moistened with strong hydrochloric acid, the outer cone becomes slightly green, especially at the edge. If, however, the saturator is jacketed with steam, so as to increase the volatilization of hydrochloric acid, the upper cone becomes intensely green. The same effect is obtained by using chloroform in the saturator.

If when, in either of the above ways, the upper cone has acquired a bright green colour, the outer tube of the separator is slid down over the inner one, the colour diminishes in intensity, and when ultimately the orifices of the two tubes are level, it disappears almost entirely except at the extreme edge. The only alteration which the outer cone can suffer by this approximation to the inner one is that its average tempe rature must increase, whence it appears that the coloration by the gold salt depends upon two circumstances-abundance of hydrochloric acid and a low average temperature.

This conclusion is easily confirmed. If, instead of bringing the cones nearer, the upper one be surrounded by oxygen instead of air, it becomes smaller and of higher average temperature, and its green colour disappears except at the extreme edges.

The fact that a chloride of gold can under any circumstances produce its individual spectrum is at first thought remarkable when we consider the extreme ease with which the salt is decomposed, and though the lower chloride or chlorides might be formed, they are likewise easily decomposed, leaving

only metallic gold. There can be no doubt, indeed, that gold trichloride is decomposed completely at a temperature far below the average one of the flame in which it yields a spectrum. It has, however, been observed by Debray (Compt. Rend. Ixix. p. 984, 1869) that gold chloride may be obtained at 300° by passing a current of chlorine through a tube containing the metal, from which it is evident that the salt may exist at abnormally high temperatures provided an excess of chlorine be present. To test this question further, I performed the following experiment.

By means of an electric current a pure gold wire was raised to bright redness in a tube partially exhausted of air. Chlorine was then allowed to enter the tube in considerable quantity. When this was done, a slight sublimate was immediately formed on the sides of the tube opposite to the glowing wire, and this sublimate on examination proved to be gold trichloride. From this it is apparent that in presence of abundance of chlorine, gold chloride may be formed at a red heat, and so emit its characteristic spectrum. The case is not an exception, but rather an example of the generalization that dissociable bodies become stable in presence of excess of one of the products of dissociation.

It is now easy to understand why gold chloride introduced into a flame on a platinum wire gives a spectrum. The great bulk of the salt is decomposed so as to give an atmosphere of chlorine in which a small portion of the salt volatilizes without decomposition. As has been already noted, much metallic gold remains on the wire.

I have shown that in using the separator the gold-chloride spectrum may be maintained by introducing abundance of hydrochloric acid or chloroform (the interconal gases contained no free chlorine), and that though the coloration may be for the most part quenched by making the flame hotter, it is most persistent at the edges. From these facts it appears that, so far as the formation and stability of gold chloride are concerned, an atmosphere of hydrochloric acid and oxygen is potentially one of chlorine, the stable arrangement of the system at a high temperature changing as follows*

302, 12HCl, 4Au becomes 6H2O, 4AuCl ̧.

It is clear, therefore, that provided we have a sufficient quantity of hydrochloric acid or free chlorine, gold chloride may exist in a flame at a temperature sufficiently high to produce its spectrum.

It is stated in Gmelin's 'Handbook,' vi. p. 215, on the authority of Proust, that gold is soluble in hot hydrochloric acid in presence of air.

« PreviousContinue »