SPECTROGRAPHIC ANALYSIS OF MINERALS, 61 JV.-A Simplified Method foy the Spectroyruphic Analysis of Minerals. By WALTER NOEL HARTLEY, F.R.S. and HUGH RAMAGE, F.I.C., A.R. C. Sc.1. PRIOR to the close of the eighteenth century the composition of minerals was never stated quantitatively, and even now in all cases where the analysis of a mineral is of importance, it is necessary first to know the nature of its constituents before determining their pro- portions. Testing by the blow-pipe mas first systematically employed by J. Gahn, and very extensively applied to mineral analysis by him and by Bergman (Thomson’s “Annals of Philosophy,” 1818, 40, 40-47. See Thornson’s “History of Chemistry,” 2, 199, edition of 1831) ; it was afterwards in many respects perfected by Wollaston and Plattner.Its application is limited when unaided by other methods, thus in minerals of a complex composition it is difficult t o detect small quantities of beryllia and some of the other rare earths when associated with alumina and magnesia. The method of dissolving minerals by the action of special solvents and separating their constituents by the addition of suitable precipit- ants was first systematised by Klaproth, and largely practised by Vauquelin, who not only improved analytical processes, but reduced the a r t of analysis to a greater degree of simplicity and precision. The work was advanced to a greater measure of perfection by Berzelius, Heinrich Rose, and many others down to the present day. This is the method pay excellence both for qualitative and quantitative purposes, but it is tedious when carried out in minute detail, and requires to be applied with considerable judgment, tact, and skill.It is frequently necessary to supplement it by other methods, as for instance in the detection of small quantities of beryllium, yttrium, cEsium, or rubidium. Spectrum analysis of minerals found a pioneer in Swan (Trans. Roy. SOC. Edin., 1853, 20, 335), the inventor of the collimator, and Bunsen, who with Kirchhoff devised the first serviceable spectroscope for the chemist’s use, and applied it practically to mineral analysis, It is hardly necessary to refer to the discoveries of gallium, indium, thallium, rubidium, and ciesium. Spectrum analysis is the only absolute method of diagnosing the chemical composition of a mineral, or of a substance separated by precipitation in the course of a chemical analysis. As no two substances can give the same spectrum, it follows that the spectrum of a substance is peculiar to itself, and, provided that light sufficient in amount and in purity can enter the instrument, i t is immaterial how far from a self-luminous object the62 HARTLEY AND RAMAGE: A SINPLIFIED METHOD FOR observer is distant.Moreover, it is possible to determine the com- position of materials too minute in quantity to be handled, and which cannot therefore be submitted to chemical analysis. llTe are also able to determine the composition of a substance the constituents of which are not amenable to any known process of chemical separation. Notwithstanding these advantages, this method in some directions does not completely satisfy the requirements of the chemist. It is desirable that a marked distinction should be drawn between chemical analysis, spectrum analysis, and merely chemical testing.The first implies the actual separation of the constituents of a sub- stance by taking advantage of their differences in volatility, solubility, or other chemical properties ; spectrum analysis is the separation in the order of wave-length of the rays proceeding from any material either self-luminous or not, and the identification of these rays with the presence of a substance whether element or compound to which these rays belong. This physical method when suitably modified in detail can very advantageously be employed as an aid to, and with certain limitations even instead of, chemical analysis.The Method of Spectrographic Anulysis. Oxyhydrogen flame spectra both of elements and compounds have been closely investigated and described by one of us and shown to be capable of very useful applications (Hartley, Phil. T~ans., 1894, 185, A, 161, 1029). Photographs of spectra are produced with extreme ease either from metals, oxides, or other compounds in the solid or liquid state. The alkali metals were proved to be volatilised from refractory silicates by reason of the high temperature em- ployed. A general application of this fact led to a method of examination being devised whereby the spectra of the alkalis can be separated from those of the alkaline earths even in a Bunsen flame (Hartley, Trans., 1893, 63, 138).We have also shown by a study of manganese slags and siliceous minerals containing this element that it can be volatilised in the oxyhydrogen flame by simply heating the silicate. Owing to the high :temperature to which the substances are sub- jected, it has already been shown that platinum wire supports are of no use, and fragments of a highly crystalline silicate of alumina in the form of the mineral kyanite have been substituted (Hartby, “Flame Spectra a t High Temperatures,” Phil. Trans., 1894, 185, A, 16s). Later it was found that little rods of pure alumina, also chips off the bowls of clay tobacco pipes, could serve the purpose of supports, and that solutions as well as solid salts, minerals, or metals could conveniently be examined therewith.THE SPECTROGRAPHfC ANALYSIS OF MINERALS.63 We propose now to give an account of a simplification of the method of obtaining these spectra, together with some examples of its appli- cation to the chemical analysis of very minute quantities of mineral substances, If the substance to be examined is a metal it should be in the form of filings, turnings, powder, or metallic sponge. If a mineral, it should be finely powdered. I n either case, the powder in quantity up to half a gram is rolled up in one-half of an (‘ ashless ” filter paper about 5 inches in diameter, the powder is spread equally over the paper with exception of a strip a t the edge, and is kept as near to the centre of the roll as possible, that it may be in the midst of the reducing gases given off by the paper when charred and in contact with the carbon.I n this manner a t so high n temperature, some oxides which are non-volatile or are volatilised with great difficulty, as they do not undergo dissociation in the flame, are actually reduced and a spectrum of the metal is thus obtained. F o r example, we find that certain lines in the spectrum of zinc and even that of aluminium may be photographed when the oxides are burnt in this way. AS regards the thermochemistry of the substance, this fact is of con- siderable interest because it informs us of the possibility of reducing aluminium oxide in presence of carbon a t the temperature of the oxyhydrogen flame. I n the electric arc the product is aluminium carbide. The lines referred to have approximately the wave-lengths 3962 and 3944, and correspond with two of the strongest lines of aluminium common to the arc and spark spectrum, namely, those with wave-lengths 3961.68 and 3944.26 (Kayser and Runge). I n front of the spectrograph is a quartz lens 3 inches in dia- meter, which projects on to the slit of the instrument the eimage of the flame from an oxyhydrogen blow-pipe.It is advisable to make the focal length of the lens 4 or 5 inches. That used by us has a focus of 3 inches, and molten particles which adhere to the quartz are liable to be projected on t o it, necessitating the repolishing of the lens. The flame may take a vertical or horizontal direction, or a direction inclined towards the optic axis of the instrument. The operator having placed the dark slide containing the sensitive plate in position and exposed it, protects his eyes by wearing very dark neutral tint or black glasses and, seated near the flame, introduces the point of the paper pencil containing the ore or mineral into the lowest part of the flame.As i t burns rapidly it is pushed farther in until it becomes necessary to hold the last two or three inches with platinum pointed forceps. Some minerals burn with sparks, others form fused globules which drop off, and care should be taken with these that they do not fall into and choke the oxygen jet of the blow- pipe. One pencil suffices, as a rule, to yield a strong spectrum. The64 HARTLEY AND RAMAGE: A SlMPLlFlED METHOD FOR time occupied in burning is about two minutes, and five spectra can very conveniently be photographed on one plate.A short spark spectrum is generally photographed on the middle of the flame spec- trum, by turning the slit of the spectrograph towards the spark emitted by an alloy giving well-known and clearly defined lines, from the measurements of which wave-lengths along the whole spectrum may be deduced by means of a curve. Of course it is advisable first of all to ascertain definitely the substances to be found in the filter papers used ; Sclileicher and Schull’s ashless filters generally gave weak lines of sodium, and weaker lines of potassium, calcium, and iron. These elements are present in the ash of the filter paper and the dust of the air. Precipitates may be collected on ashless filters and the paper burnt. If the precipitate be small, as, for instzznce, a few milligrams in weight, the paper is cut into strips and so burnt.The photographs are taken on Edwards’ ‘‘ Snap Shot ” isochrornatic plates or Cadett’s spectrum plates, and are developed with quinol, the development occupying from two t o three minutes, When burnt in the manner described, the lines in the spectra are not obscured by the emission of white light, but the presence of a large quantity of sodium salts is decidedly disadvantageous, inasmuch as it overpowers or suppresses weak lines of other elements. The spectra observed for the most part have already been described (Phil. Trans., 1894, 185, A, 168, 1029). In one or two cases we have considered it advisable to revise these spectra. The -water-vapour lines from the flame, the lines of iron, calcium, and sodium, may be very commonly observed.Precipitates show the spectrum of potassium in a manner which leads to no other conclusion than that this element remains in combination with insoluble hydr- oxides much more frequently than is generally supposed. The lines in the spectra are measured directly by applying an ivory scale to the photograph and clamping it to the glass as described else- where in the examination of absorption spectra (Hartley, Phil. Trans., 1885, 176, 471). We have used divided scales 4 inches long with 200 divisions to an inch. A very little practice with a sufficiently powerful magnifier or low-power microscope enables each division to be subdivided by judgment into fifths. With a curve constructed from the principal lines of iron and either Kayser and Runge’s wave-lengths or Rowland’s solar lines, we may determine the wave-lengths of any line so accurately that as a rule for the identification of lines of known wave-lengths nothing more is necessary.But occasionally it becomes desirable that wave-lengths be determined with still greater accuracy, and micrometer measure- ments are resorted to. As an example of what may be accomplishedTHE BPECTROORAPHIC ANALYSIS OF MINERALS. 65 in this manner, we give in a tabulated statement the lines observed in "blast furnace metal" from Middlesbrough, side by side with the lines with which they have been identified in the solar spectrum or in the arc spectrum of iron. It will be seen that the agreement between the wave-length numbers is very close.We proved the identity of the lines due to foreign metals by carefully executed chemical analyses aided by measurements from the photographs of the spectra of the various precipitated substances which were separated in the process, and not solely from the spectrum of the crude metallic iron. Lines may be identified in many spectra by simultaneous coinci- dences with groups of lines on photographed spectra of the elements, without precisely measuring their wave-lengths. In this case, the one plate is placed film to film against the other. Spectrogyaphic unulysis of the crude metal with which the converters are The iron wus heuted in the oxyhydrogen churged at Miclcllesbrough. same on suppwts of cyunite. Wave-leiigt hs of lines.(Row land's scale.) 5735'2 5622'2 5582'4 5371'7 5328'6 5270-2 4481 -7 61'5 27 *3 15'7 05.1 75 -8 25'5 07.9 4253.9 i 4 . 6 71-5 54'4 26'8 16.9 02 '4 4171'6 43'6 31 -4 4071% 63 '4 4383.1 Lines in solar spectiuiii for comparison. ( Rowland.) 5371 '686 5328 '696 5 328 *7 47 52iO-533 4482.336 61.81 8 27'482 15'293 04.927 4383*i20 76.1 07 25 939 L s s * I l y l 4269 853 i4.958 i l . 9 3 4 54 505 ~d 904 15.i2 01 98 417 2 '21 1 44.038 31'235 4071 '908 63'755 Remarks and references showing by figures the intensity of' the lines, (1) being the lowest. Narrow band Edge of band more refrangible. Edge of band more refrangible. Fe, C!r (7) Fe (2) Fe (2) Fe (4) Fe Fe (4) Fe (5) Fe ( 6 ) Ft! (10) Fe (15) Fe (6) Fe (8) Fe (6) Cr (5) Cr (7 d) Fe ( 6 ) Cr (8) ('a (20 d 1) lit), Rowland, also 4216'351 Fe (3 d P) KI), Itowland, also 4202'198 Fe (8) 3a (1) Fe (15) Pe (10) Fe (15) Fe ( Y O ) VOL.LXXlX. F66 HARTLEY AND RAMAGE: A SIMPLIFIED METHOD FOR Spectrographic cwlalysh, &c. (continued). Wave-lengths of lines. (Rowland's scale.) 4047.4 45 '8 44-0 34.5 33.2 31 '0 05 -3 3969.1 305 28 '0 22.8 20.0 06.4 3899-9 95.9 86.2 78.9 72.9 65.6 60.1 56.6 49.6 40 '2 34 '2 27.3 25-9 24 *5 20.6 15.6 1.2'6 3799'4 98'5 94'9 87'9 67'2 63.8 58'4 49 -3 48'2 45.8 37-4 35 '1 33 *3 27.9 22 '5 19'9 09.3 07'9 05.7 3687.1 83'6 79.9 47 '8 Lines in solar spectium for comparison. (Rowland.) 4047.338 45.975 44.294 34.641 33,224 30'914 05.408 3969 -413 30.450 28 *075 23.054 20.410 06.628 3 89 9 -8 50 95-803 86.421 78'720 72.639 65'674 60.055 56-524 50,118 40'580 34'364 27'973 26.027 24.591 20.566 15.987 13-100 3799.693 98-662 95-147 88'046 67'341 63'945 58.379 49'631 48-408 45'717 37'231 35'014 33.469 27.778 22'692 20.086 09.397 08'068 05-711 3687 -607 83.202 80-064 47'995 Remarks and references showing by figures the intensity of the lines, (1) being the lowest.TRE SPECTROGRAPHIC ANALYSIS OF MINERALS.Spectrographic analysis, &c. (continued). 67 Wave-lengths of lines. ( Rowland's scale.) 3631 '2 18 '7 09 '0 05-3 3593'7 87.6 85 *5 81'5 78.8 70.1 65.5 58'6 25'9 25 *O 21 -2 13.7 3497'6 90.7 76 '6 75'4 71.9 66'0 43.8 40'7 3301.9 01 '4 3273'6 47 -0 Lines in solar spectrum for comparison. (Row land. ) 3031.619 18'924 09.015 05.483 3593 -636 87'130 85,479 81 '344 78.832 70.225 65'528 58-670 45.986 24.677 21 '404 13.947 3497 '991 90.721 76.831 75.594 71.499 ? 65'991 44-032 40'759 3303.1 07 02.501 3274 -092 47.680 Remarks and references showing by figures the intensity of the lines, (1) being the lowest.Fe (20) Fe (20) Fe (15) Cr (4) Cr (9) Pe (8) Fe (7) Fe (40) Cr (10) Fe (20) Fe (12) Fe ( 8 ) Fe (7) Fe (6) Fe (7) Fe (10) Fe (8) Fe (10) Fe (3). Identity with 3471.499 doubt- ful. I t is more probably 3472'06 Fe, a reversed line in the arc spectrim (Kayser and Rnnge). E "0) Fe (6) Na (5) Na (6) c u (6) cu (9) ;: {Z) The Spectrographic Analysis of Silicates. Silicates are sometimes of a very refractory nature and do not yield spectra of the bases present other than the alkalis unless these are in some way liberated and converted either into oxides or salts. A means universally applicable for removing the silica had to be devised.Several methods were tried, but only one which has been in use for thirteen years past in the laboratory of this college appeared in all cases to be satisfactory; that is, to decompose the silicate with a mixture of pure ammonium fluoride and strong sulphuric acid warmed in a platinum crucible which is covered by a lid. It is necessary that the ammonium fluoride and the acid should be pure. To purify the ammonium fluoride from silicon fluoride and from all fixed bases, it is necessary to distil it in a platinum retort, an operation which is not F 2(1) Corundum ... (2) Corundum ... (3) Spinel ......... (6) Rutile ......... (6) Basalt ......... (7) Serpentine Co. Donegal (8) Hornblende granulite ...(9) Felsite ......... (Orthoclase). (12) Plumose mica (14) Beryl, (1) Ruby ......... (10) Felspar (11) Muscovite ... (1 3) Lepidolite ... Glencullen, Co. Dublin (16) Schiat con- taining large garnets ...... Garnets and mica slate, Airolo, Can- ton Ticino. (16) Garnets ...... (17) Mica slate ... Minerals from a coar8e granite from Co. Dublin. (18) Felspar ...... (19) Mica ......... (20) Quartz ......... (21) Clay, Shank. hill Quarry, Co. Dublin.. (22) Residue, in. soluble i n HCI, from Cleveland clay iron. stone ......... (23) Shale from ironstone beds ......... (24) Bauxite from Wales ...... - .r( $ e 's - Li Li Li Li Li Li Li - $ $ m - Na Na Na Na Nfl Na Na Na Na Na Na Na Na Na Na Na Na Na Na Na Na Na Na A'a - - ii & .r( W a3 4J - K K K K K K K K K K K K K K K K K K K K K K K R - Examples of the spectrographic analysis oj refi.actory minerals.- .rl El 2 P - Rb Rb Rb Rb Rb Rb Rb Rb Rb Rb Rb Rb Rb F; I% e u - c u c u c u c u c u c u c u c u cu c u c u c u c u Cn c u c u c u c u c u cu c u c u cu - - E * S W .2 3 d o - Ca CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO CaO Ca - 3 :z ; a Sr Sr Br Sr Sr Sr sr Sr Sr Sr - .r( !i c( R s - Ga Ga Ga Ga Ga Ga Ga Ga Gtl Ga Ga Ga Ga Ga Gs Ga Ga Ga Ga Ga Ga - d 2 Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe * - Fe Fe Fc Ft? Fe Fe Fe Fe Fe Fe Fe - M 4 iz - Ni Ni Ni Ni Ni Ni Wi Ni Ni Ni - - 2 % Fi i - Mn Mn Mn Mn MU Mn Mn Mn Mn Mn Mn Mn Mn Mn Mn hl n Mn MH Mn Mn - $ 'ii 2 d 0 - Cr Cr Cr Cr Cr Cr Cr Cr - W $ 4 - Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pt- Pi Remarks.iVhere the symbol of the element is printed in italics, it means that there werr traces only. Jpinel contains more Cr and Ga than ruby. Ruby contains Mn and Cu, and more Fe than spinel. rhe symbol Ca indicate8 the calcium line 4228. When the quantity of substance is small this very strong line alone is photographed. rhe formula CaO shows that the calcium oxide bands in addition are conspicuous. MgO indicates bands of magnesium oxide. Gallium doubtful herr. The mica slate contains also TiOa as rutile. Practically all the Mn, Cs, as, and Ag cr!s- tallic~e in the mica. The felrpar contains alarger proportion of Na, Ca, and Pb, but smaller pro- portions of Rb and Fe. It is a little doubtful whether chromium is present here.70 SPECTROCRAPTi[IC ANALYSIS OF MINfiRALS.difficult, since the salt distils freely and does not solidify in the neck of the retort. The purity of the sulphuric acid is of course easily ascertained by ignition in a platinum dish, when it should leave no fixed residue. The proportion of fluoride found convenient will of course differ with the nature and composition of the silicate to be examined. If we take the substance which is perhaps the most refractory, namely, cyanite, because it contains about 96 per cent. of aluminium silicate, we may reckon that for every gram of the mineral there will be required at least two and a half times its weight of ammonium fluoride and five times its weight of oil of vitriol. I n practice we find three times its weight of ammonium fluoride and seven times its weight of oil of vitriol are sufficient.These materials may be intimately mixed with the very finely powdered mineral and kept for some time a t a temperature of about 50°. When silicon fluoride ceases to be evolved, the temperature may be gradually raised until the sulphuric acid and ammonium sulphate are completely ex- pelled. The solid may be then converted into mixed bases by re- peatedly heating with ammonium carbonate and igniting gently. Lead is found sometimes in ordinary concentrated oil of vitriol by the appearance of one lead line in the spectrum, but when it is desirable to get rid of this impurity, the acid made from sulphur trioxide may be used. The only impurities observed in the spectra, including those yielded by the paper and the sulphuric acid, were traces of sodium, potassium, calcium, and the merest trace of iron.Otto Vogel has described the use of the oxy-coal gas flame for spec- trum analysis, using pieces of retort carbon from gas-works as a sup- port for the substance (Zeit. anorg. Chem., 1894,5, 42-62). The spectra were not photographed. Such observations were made by one of us using iridium wires as a support as far back as 1885, and the methodof examin- ing such minerals as heavy-spar, fluorspar, felspar, and mica in such a flame was taught in the laboratory of the Royal College of Science, Dublin, from that time until 1889 and 1890, when the oxyhydrogen flame and cyanite supports were used. Silicates such as felspar, mica, &c., we now examine in the following manner. The residue from the treatment of 1 gram with ammonium fluoride, &c., after ignition to expel the ammonium sulphate, is boiled with water and a slight excess of ammonia.The precipitate which contains the alumina, &c., is collected by filtration, dried, and ignited in a roll of filter paper. The residue from the filtrate, after ignition, is collected in a small quantity of water, and the mixture poured on to a filter paper, which is dried and burnt. Two spectra are thus obtained from each mineral, the one of bases precipitable, the other of those which are not precipitable by ammonia, Silicates such as basalt, and those present in siderolites, &c., may contain iron (ferrous or ferric oxide),HYOSCYAMUS MUTICUS AND DBTURA STRAMONIUM. 71 calcium oxide, and magnesium oxide, besides the alumina and alkalis. The residue is dissolved in hydrochloric acid, and before precipitating with ammonia theiron must beoxidised. The calcium in thefiltrate is pre- pitated by ammonium carbonate, and the magnesium in the filtrate from this by ammonium phosphate. The filtrate from the magnesium phos- phate is then examined as in the preceding case for the alkalis, The reaction for caesium and rubidium is more delicate if these bases are first separated from the bulk of the potassium and the sodium by precipitating with platinic chloride and boiling the precipitate with water. Potassium and sodium yield strong continuous spectra which mask weak lines of other elements, Lithium is only detected in the photographed spectra by its blue line and one in the ultra-violet when it is present in appreciable quantity, but traces may be easily detected by eye observation of the red line. On pp. 68 and 69 a tabulated statement is given of the substances detected by spectrographic analysis in a number of very refractory minerals which were analysed by the method here described.