Chemical Analysis by Means of the Spectroscope

Kirchhoff and Bunsen

Many substances are well known to have the power when burned in a flame of causing certain bright lines to appear in the spectrum. It is possible to base on these lines a method of qualitative analysis that immensely widens the field of chemical analysis and solves questions hitherto unanswered. We shall here limit ourselves merely to the application of this method to the detection of the metals in the alkalies and alkali earth and to a series of illustrations as to the value of their methods.

Such lines appear plainer the higher the temperature and the less the natural light of the flame in which the substance is heated. The Bunsen burner described by one of us has a flame of very high temperature, but very small luminosity, and is therefore specially suited to experiments on the substances that produce bright lines in the spectrum. [p.390]

[p.391] In figure one the spectra are illustrated which the flames mentioned produced when the salts, as pure as possible, of potassium, sodium, lithium, strontium, calcium, and barium, are heated to a vapor in it. The solar spectrum is also reproduced in order to render a comparison easy.

Figure two represents the apparatus that we have usually used in observing the spectrum. The box A, resting on three feet, has the form of a trapezium and is blackened on the inside. The two inclined sides of the box make an angle with each other of about fifty-eight degrees, each carrying a small telescope, B or C. The eye-glass of the first is removed and in its place is put a plate, in which there is made a slit formed by two brass bands at the focus. The lamp D is so placed in front of the slit that the mantle of the light would be cut by the axis of the telescope B. Just under the point where the axis would cut the mantle the end of a very fine platinum wire passes into it. This wire is bent into a hook and is carried by the holder E. On this hook is melted a drop of the chloride previously dried on it. Between the object glasses of the two telescopes B and C there is placed a hollow prism F, filled with carbon disulphide. This prism rests upon a brass plate that can be moved round on a vertical axis. This axis has at its lower end the mirror G, and above this the arm H, which is used as a handle to rotate the mirror and the prism. Another telescope is adjusted in front of the mirror. This gives an image of a horizontal scale located a short distance away. By turning the prism we can make the entire spectrum of the flame pass before the vertical thread of the telescope C and can make every part of the spectrum correspond with this thread. A particular point of the spectrum corresponds to every reading on the scale. When the spectrum is very weak the cross hair of the telescope C may be illuminated by a lens throwing part of the rays from a lamp through a small opening made laterally in the eye-glass of the telescope C.

The spectra in figure one, produced by the pure chloride already mentioned, have been compared by us with those obtained when we have introduced the bromides, iodides, hydrated oxides, sulphates, and carbonates of the various metals into the flame of sulphur, the flame of carbon disulphide, the flame of aqueous alcohol, the non-luminous flame of coal gas, the flame of carbonic oxide, the flame of hydrogen and the oxyhydrogen flame.

From these inclusive and detailed investigations, the particulars of which we may omit, it becomes evident that the difference in the compound in which the metals were used, the great variety of the chemical reactions in the various flames, and the immense differences of temperature in such flames have absolutely no effect on the position of the spectral lines characteristic of the different metals. The same metal compound seems to give in any flame a more intense spectrum as the temperature is higher. The compounds of these metals that have the greatest volatility give the most intense flame.

For another proof that each of these metals always gives the same bright lines in its spectrum we have compared its spectra with the spectra of an electric spark passing between electrodes made of these metals.

Bits of potassium, sodium, lithium, strontium, and calcium were fastened on a fine platinum wire and so melted, two at a time, within glass tubes, the wires piercing the sides of the tubes so that they were distant from one to two millimeters from each other. Each tube was adjusted before the slit of the spectroscope. We made electric sparks pass between these metal pieces by using a Ruhmkorff induction coil and compared the spectrum thus given with the spectrum of the chloride of the same metal, when brought into the gas flame, placed behind the glass tube. When the Ruhmkorff apparatus was alternately thrown in and out of action one was easily convinced without need of measuring accurately that the bright lines of the spectrum from the flame remained present without being displaced in the brilliant spectrum of the spark. In addition to these, however, in the spark spectrum there appeared other bright lines, some of which must be laid to the presence of foreign metals in the electrodes, others to nitrogen which remained in the tubes after the oxygen had combined with the electrode.

It is therefore unquestionable that the bright lines described in the spectra may be regarded as absolute proof of the presence of the metal in consideration. They can serve as tests by means of which this material may be more certainly and readily and minutely detected than by any other analytical method.

The spectra in a figure are produced when the slit is wide enough so that only the most noticeable of the dark lines of the solar spectrum appear. The observing telescope magnifying only about fourfold and the light not being intense. These conditions appear the most suitable for carrying out a required chemical analysis by spectral experiments. The looks of the spectrum may vary considerably under other circumstances. If the purity of the spectrum is increased (by greater magnification), a number of the lines that seemed single are resolved into several, the sodium line, for instance, into two; if the intensity of the light is increased, in many of the spectra shown new lines make their appearance and the old ones change their relative brightness. In general darker lines increase in brightness with the greater intensity of the light more rapidly than a brighter one, but not enough to surpass it in brightness. The two lithium lines constitute a good example of this fact. We have noted only one exception to the rule. The line Ba e, which with the light not intense is scarcely visible, while Ba g shows very distinctly, becomes much brighter than the latter when the light is more intense. This fact seems to be important and we shall make a further study of it.

We shall now enter more minutely into the characteristics of the various spectra, a knowledge of which is important from a practical point of view, and illustrate the advantage offered by a chemical analytic method founded upon the spectrum.

Of all the spectral reaction the most sensitive is that of sodium. The yellow line Na a, the only one appearing in the sodium spectrum, coincides in position with Fraunhofer’s line D in the solar spectrum and is marked by its particularly sharp boundary and its unusual brilliancy. If the temperature of the flame is very high and the amount used large, indications of a continuous spectrum appear adjacent to the line. Lines of other elements, in themselves not strong, coming near it, appear still weaker, and are therefore often visible only after the sodium reaction begins to fade away. The reaction is strongest in the oxygen, chlorine, iodine and bromine compounds in sulphuric and carbonic acids, but it is also evident in the silicates, borates, phosphates and other non-volatile salts.

Swan has already noted the minute quantity of common salt sufficient to produce a clear sodium line.

The experiment that follows emphasizes the fact that no reaction in chemistry compares even remotely in sensitiveness with this analytic spectral detection of sodium. In one corner of the experiment room, which contained, say sixty cubic meters of air, we detonized, as far away from our apparatus as possible, three milligrams of chlorate of sodium with milk sugar while the non-luminous flame was under observation in front of the slit. After some moments the flame gradually turned pale yellow and gave a strong sodium line, which again vanished in about ten minutes. Now it is easy to estimate, noting the weight of the detonized salt and the amount of air in a room, that in a unit weight of air not one twenty-millionth part of sodium smoke could have been suspended. As the reaction can easily be seen the first second, and as in this time, in accordance with the flow and composition of the gases in the flame, only about fifty ccm. or .0647 grams of air containing less than 1–20,000,000 of sodium salt burn in the flame, we conclude that the eye can detect less than 1–3,000,000 of a milligram of sodium salt as the greatest distinctness. With so sensitive a reaction it is evident that there would rarely be a sodium reaction not detectable in glowing atmospheric air. The earth over more than two-thirds of its surface is covered with a solution of chloride of sodium. As the waves break in the foam, this is continually changed into spray and the particles of sea water which thus enter the atmosphere, evaporate and leave behind them traces of salt varying in size, but rarely absent from the air, perhaps being of use in supplying small organisms with salt, the same as the ground supplies it to the larger plants and animals. The presence of salt in the air, easily detected by spectrum analysis, is of interest from still another point of view. If, as we can hardly yet doubt, there are catalytic influences that aid the spread of disease, it is possible that an antiseptic substance, such as salt, may not be, though even in the smallest quantities, without noticeable influence on such reactions in the air. Now it would be easy to discover from continued daily observation of the spectrum whether the variation in strength of the spectral line Na a, representing the sodium reaction in the air, is in any way concomitant with the rise and spread of endemic diseases. The wonderfully sensitive reaction of sodium may also explain why all bodies when heated in the flame show the sodium if left exposed to the air, and why one can eliminate the last trace of the sodium line Na a in only a few compounds, and after crystallizing them ten times or more from water that has come in contact with platinum vessels only. A platinum hair wire that had been freed by heat from every trace of sodium gives this reaction very vividly if it is exposed some hours in the air. Dust settling in the room from the air shows it as plainly, so that, for example, the slapping of a dusty book is enough to give brilliant flashes of the Na a line at a distance of several paces.

The incandescent vapors of the compound of lithium produce two well defined lines, one a pale yellow Li b, the other a bright red line [p.395] Li a. This reaction is more certain and delicate than any hitherto known in analytical chemistry. It is not as delicate as the sodium reaction, perhaps because the eye is more sensitive for yellow than for red. By means of experiments the unlooked-for result is reached that lithium is one of the substances most widely distributed throughout nature. In the production of compounds of lithium on a large commercial scale spectrum analysis is an invaluable method of selecting the raw material used and deciding on the most efficient process of manufacture. For example, it is necessary to evaporate only a drop of the various mother-liquors in the flame and note the result through the telescope in order to discover at once that a rich and heretofore unnoticed supply of lithium exists in many saline residues. And in the course of preparation we can trace any waste of lithium in the by-products by means of spectrum analysis, and hence at once seek more efficient processes of manufacture than those being used.

The unstable compounds of potassimu produce a wide continuous spectrum in the flame, showing only two characteristic lines: the first, Ka, in the edge of the red and bordering on the ultra red rays, coincides exactly with the dark line A of the solar spectrum; the second, Kb, far in the violet at the other end of the spectrum, also is identical in position with a Fraunhofer line. There is also a very weak line corresponding to Fraunhofer’s line B, but it is only visible with an intense flame, and so less characteristic. The violet line is not strong, but is about as well adapted for discovering the presence of potassium as the red line. The position of both lines, near the last visible to the eye, makes this reaction less sensitive than those already mentioned.

The spectra given by the alkali earths are not as simple as those from the alkalies. Strontium is peculiar for the absence of green bands. Of its spectrum eight lines are remarkable, six of them red, one orange, and one blue. The orange line Sr a appearing close to the sodium line near the red; the two red lines Sr b, Sr g, and finally the blue line Sr d are the most important in position and brightness. We can according to experiment estimate that 6–100,000 of a milligram of strontium is discoverable to the eye. The reaction of potassium and sodium is not interfered with by the presence of strontium. The lithium reaction is distinctly visible along with these three, if the amount of lithium be not too small in comparison with that of [p.396] strontium. In this case the lithium line shows itself as a narrow red line, very intense and highly defined, against the weaker red background of the broad strontium line Sr b.

The spectrum of calcium can be distinguished at first glance from the four others already described by the perfectly characteristic marked green line Ca b. A second equally remarkable distinction is the very bright orange line Ca a, situated considerably nearer the red end of the spectrum than the sodium line Na a, or the orange line of strontium Sr a. By burning a mixture of calcium chloride, chlorate of potassium, and milk sugar we produced a smoke the reaction from which has about the same sensitiveness as that from the chloride of strontium fumed under the same conditions. From an experiment made in this way it resulted that 6–100,000 milligrams could be easily and definitely detected. Only the calcium compounds that volatilize in the flame give this reaction, and the more volatile the compound the plainer the reaction. Chloride of calcium, iodide of calcium, and bromide of calcium are the best. Sulphate of calcium produces a spectrum only when it has become basic, then a brilliant and lasting one. In the same way the reaction of the carbonate becomes visible after the acid is driven away.

Barium has the most complicated spectrum of the alkalies and the alkaline earths. It is at once distinguishable from those already considered by the green lines Ba a and Ba b. These surpass all others in brilliancy, and appear first and fade last in weak reactions. Ba g is not so distinct, but is nevertheless to be noted as a characteristic line. The relatively wide extension of the spectrum explains why the barium reaction is considerably less sensitive than those hitherto examined. Three-hundredths of a gram of chlorate of barium, burned in our room along with milk sugar, gave in air, mixed by moving an open umbrella, the line Ba a for a long time very distinctly. We may therefore conclude by a calculation as in the case of sodium that the reaction will show 1–1,000 of a milligram with absolute distinctness.

For those that are familiar through repeated observation with the various spectra, no accurate measurement of the separate lines is necessary, as their color, position, peculiar degree of definiteness and shade, and difference in brilliancy are differentia sufficient for even the inexperienced to clearly recognize them. These individualities may be compared with the distinguishing marks used as a reaction test borne by the different precipitates in outward appearance. As the precipitate shows gelatinous, pulverized, flocculent, granular, or crystalline, so the spectral lines are characteristic in the definiteness of their edges, in their regular or irregular shading away on one or both sides, and in their breadth or narrowness, as the case may be. As also we use only those precipitates for analysis that are producible from the greatest possible dilution, so in spectrum analysis we use only those lines requiring only the smallest amount of substance and only a moderate temperature to produce them. In such ways the two methods are quite comparable. But in the color phenomena of the reaction, spectrum analysis has a quality that gives it infinite advantage over any other method of analysis. The precipitates that are used in chemical analysis are most of them white, and but few colored. Even of the colored the tint is not very constant and differs considerably, in accordance with the greater or less condensation of the precipitate. The smallest mixture of some foreign matter is often enough to change entirely a characteristic color. Hence small differences in the color of precipitate cannot be used as a sure chemical test. But in spectrum analysis, on the other hand, the colored bands remain unaffected by such foreign influences or the presence of other bodies. The positions in which they lie in the spectrum is a definite characteristic that is as unchanging and fundamental as the atomic weight of the substance, and gives us power to detect the substance with almost astronomical exactness. What, moreover, is peculiarly important in the case of spectrum analysis, is that its power extends almost infinitely beyond the limit to which the chemical analysis of matter has hitherto reached. It promises invaluable results on the distribution and ordering of geological substances in their original formation. The few experiments included in the present treatise already point to the unlooked-for result that not only potassium and sodium, but also lithium and strontium must be numbered among the substances most widely scattered throughout the earth, though in only minute quantities.

Spectrum analysis will have fully as important a share in the discovery of unknown elements. For if there are substances scattered in such minute quantities throughout nature that the methods of analysis hitherto used in observation have failed to detect them, we may expect to discover and differentiate many of them that would be hidden to the usual method of chemical analysis, merely by the investigation of the spectra inflamed. That such elements, previously unknown, do really exist, we have already been able to show. We also believe that we shall yet be able to declare absolutely, on the strength of unquestionable conclusions from spectral analysis, that besides potassium, sodium and lithium there remains a fourth metal of the alkali group. This will give quite as characteristic a spectrum as lithium; it shows with our spectrum method only two lines, a weak blue one nearly opposite the strontium line Sr d, and a second blue line lying a little farther toward the violet, and almost as bright and clearly marked as the lithium line.

Spectrum analysis gives us, as we believe we have already shown, a method of marvelous simplicity for discovering the slightest traces of many elements throughout the earth on the one hand, and on the other, it throws open to chemical investigation a field hitherto entirely inaccessible, extending far beyond the limit of the earth and the solar system itself. Since it is enough for this method of analysis merely to see the gas in an incandescent state to analyze it, it is evident that this same method would apply to the atmosphere of the sun and the brighter fixed stars.

Right here there must be introduced a modification in regard to the light the nuclei of these heavenly bodies radiate. In a treatise "On the Relation Between the Emission and Absorption of Heat and Light" one of us has shown theoretically that the bright lines of an incandescent gas are transformed into dark ones when a sufficiently bright source of light giving a continuous spectrum is placed behind it, that is, the spectrum of the gas is reversed. From this it results that the sun spectrum with its dark lines is nothing else than the reversal of the spectrum which the (outer) atmosphere of the sun would show. Hence, to analyze chemically the sun’s atmosphere (not the sun itself) requires only the investigation of the substances that produce in the flame bright lines corresponding with the dark ones of the sun’s spectrum. In the memoir mentioned are given the following examples as proofs by experiment of the theoretical law referred to:

The bright red line in the spectrum from a bead of a chloride of lithium in a flame is changed into a black line when we allow full sunlight to pass through the flame.

Substituting a bead of sodium chloride for the lithium, the dark double line D, corresponding to the bright sodium line, shows itself in the sun’s spectrum with remarkable clearness.

The dark double line D is given in the spectrum of the Drummond light if its rays are passed through the flames of aqueous alcohol, into which chloride of sodium has been introduced.

A still further confirmation of this remarkable theoretical law may not be uninteresting. This may be reached by the experiment described below:

We heated a thick platinum wire in a flame until it was incandescent and by means of an electric current nearly brought to the melting point the wire (like other solids at white heat) gave a bright spectrum with no traces either of specially brilliant or dark lines. If a flame of alcohol much diluted with water, containing common salt in solution, was introduced between the platinum wire and the slit of the apparatus the dark line D appeared very distinctly.

We can obtain the dark line D in the spectrum of a platinum wire incandescent in a flame by merely holding before it a test tube, into which some sodium amalgam has been put, and heated to the boiling point. This experiment is important because it shows that far below the incandescent point of sodium vapor it absorbs exactly the same parts of the spectrum as when at the highest temperature we can produce, or at that which the sun’s atmosphere is kept.

We have been able to change bright lines into dark in the spectra of potassium, strontium, calcium, and barium by using sunlight and mixtures of the chlorates of these metals with milk sugar. The mixture was put into a small iron trough before the slit of the apparatus, the full sunlight was let to pass along the trough of the slit and the mixture set on fire at one side by an incandescent wire. The telescope was set with the intersection of its cross hairs, which were placed at an acute angle with each other, on the bright line in the flame spectrum, the reversal of which was to be looked for. The observer put all his attention on this point in order to be sure whether at the instant of the flash a dark line appeared passing through the intersection of the cross hairs. It was not hard by this method, the mixture to be burnt being properly proportioned, to note the reversal of the lines Ba a and Ba b and the line Kb. The last of these corresponded with one of the most distinct lines of the solar spectrum; although it is not noted by Fraunhofer, this line showed itself most distinctly at the moment of ignition of the potash salt.

To observe the reversal of the bright lines of the strontium spectrum by this method, the chlorate of strontium must be very carefully dried, for a very slight moisture weakens the sun’s rays and produces the positive spectrum of strontium, because the flame becomes filled with salt particles that have been scattered about by the combustion.

We have limited ourselves in this record to investigate in the spectra of the alkali metals and earths, and these only as far as necessary for analyzing substances here on earth. We reserve for ourselves the further extension of such investigations suitable to analyzing terrestrial substances, and the analysis of the atmospheres of the stars.

Heidelberg, April, 1860.