EXCHANGE vr/VR A Spectrographic Study by Means of a Grating (Replica) Spectroscope and The Determination of The Wave Lengths of The Arc Spectrum of Tantalum JO XLJSH3AINQ 2H1 PRESENTED TO THE FACULTY OF VANDERBILT UNIVERSITY AS A THESIS, FOR THE DEGREE OF DOCTOR OF SCIENCE, BY ALLAN F. ODELL, M. S. A Spectrographic Study by Means of a Grating (Replica) Spectro- scope and The Determina- tion of The Wave Lengths of The Arc Spectrum of Tantalum. PRESENTED TO THE FACULTY OF VANDERBILT UNIVERSITY AS A THESIS FOR THE DEGREE OF DOCTOR OF SCIENCE BY ALLAN F. ODELL, M. S. CONTENTS. Page Acknowledgment 3 Dedication 4 Introduction 5 Residues Containing Doubtful Elements 6 The Arc 7 The Spectrograph 8-9 Purification of Tantalum 9-10 Methods of Measurement 10 The Comparator 10-11 The Comparator Method 11-12 The Projection Method 12 The Arc Spectrum of Tantalum 14 The Spectrum of a Clay Residue 19 Summary 20 ACKNOWLEDGMENT. I wish to express my sincere appreciation and thanks to Dr. William L. Dudley, of this University, at whose suggestion this research was commenced, and by whom it has been directed. 4698^2 As a token of esteem and affection to my old class-mate ERNEST WILLIAM GOODPASTURE INTRODUCTION. The original object of this research was to investigate the contents of rare elements in some of the clays of the South ; and for this purpose, samples were collected from as many different points as possible. The manner in which the investigation was to be carried out, was to separate the clay into its various constituents, and exam- ine these in the spectroscope, and, if unusual elements were found, to estimate the quantity. To give an example, a clay was treated thus : It was first qualitatively analyzed for the ordinary elements, as iron, aluminum, manganese, calcium, magnesium, potassium, sodium, etc. The precipitates obtained for these metals were dried, powdered, and put in the arc for photographing the spec- trum. The resulting spectographs were examined, and the lines carefully measured. In this manner, the spectra of the rare ele- ments would identify their presence without making the tedious qualitative examination for them in the first instance necessary. RESIDUES CONTAINING DOUBTFUL ELEMENTS. In the first set of clays examined, three specimens from Spruce Pine, Alabama, a residue was obtained which gave lines of Ti, and another element, which we could not identify from its spec- trum. The method by which this residue was obtained is as follows: The clay is finely pulverized and heated with a mixture of eight parts cone, sulphuric and two part cone, nitric acid, un- til solution has taken place and only a white siliceous residue is left. By this method the majority of the elements are taken into the solution. It is diluted with water and the siliceous matter filtered off. The Fe and Mn and some rare elements are precipitated from the filtrate by adding an excess of KOH. The solution is again filtered and the precipitate washed on the filter. The precipitate is then dissolved in HC1, and the resulting solution treated with NrLjOH, until precipitation occurs, (NH4) 2Sx is added until the hydroxide which was first precipitated, is changed into sulphide. H2SO3 is then added until the solution clears, leaving a white precipitate. Ti, Cb, Ta, etc., remain in the precipi- tate, which is filtered off and washed. The Fe, Mn, Cr, etc., will be in the filtrate. The precipitate is dried, and removed from the paper, and fused with Na2CO3 in a platinum crucible. The fusion is digested in cold water and filtered. The precipitate is treated with HC1, and to the resulting solution Na2S2O3 added. A precipitate is gotten which is designated as precipitate A. This precipitate was examined under the spectroscope, and will be referred to later. If the filtrate from precipitate A is treated with KOH, another precipitate is gotten, which is precipi- tate B, and which has not yet been investigated, as the quantity obtained was tpo small. The filtrate from the Na2CX>3 fusion is made acid with HC1, and the CO2 expelled by boiling, NH-4OH is then added, and the excess expelled by boiling. Another pre- cipitate is obtained. This, also, has not been examined. Precipitate A gave lines for Ti, and was contaminated by traces of Fe; but there were also other strong lines present, which could not be identified in the spectra of either of these elements. As tantalum occurs widely distributed, and associated with titanium, some lines of the former were suspected of being pres- ent in the spectrum of precipitate A. And as no arc spectrum of tantalum has been made heretofore, and as the continued recur- rence of this element would present serious difficulty in the future work on clays, the arc spectrum of tantalum has been made be- tween ^3200 to ^6300. The apparatus available would only permit of the measure- ment of the lines to the fourth place, which is sufficiently accurate for all practical work. THE ARC USED. The first arc used was a closed lantern with carbon poles, but as so many bands occurred in the carbon spectrum, they obscured the fainter lines, and the carbon poles with the closed lantern were discarded and replaced by a hand feed lantern using copper poles. As copper gives a sharp spectrum with lines wide- ly distributed, it served perfectly. The impurity in it is chiefly Fe, and this assists in finding standards for measurement. The poles are 5-16 of an inch in diameter, and are turned down a half inch from the end to about 3-16 of an inch. The lower pole is cupped out for receiving the substance. The cur- rent used was about 104 volts and four to six am- peres. The bottom pole containing the substance, was used as the positive pole. (Fig. I.) An attempt was made to secure spark spectra of some of the clay residues, but as these were usually -i- in the form of a fine powder, the spark, on turning on the current, would brush the substance off the pole immediately. In taking the spectra from the arc, it was found that by shifting the upper pole slightly, the arc could be kept constantly on the fused substance on the lower pole, thereby diminishing the copper cpectrum and intensifying the spectrum with the substance to be examined. Figure 1 THE SPECTROGRAPH. The spectograph used is one of special construction, and de- signed by Dr. Dudley, and made so that the entire spectrum be- tween ^3000 and ^6500 can be taken on one plate. Its construction will be easily understood from the diagram. (Fig. 2.) A is the slit, regulated by the micrometer screw B which adjusts the slit in fortieths of a millimeter. C is the screw adjustment which focuses the light from A on to the lens D, which parallels the rays upon the grating G. From thence they are dispersed upon a vertical slit at H, behind which is arranged the photographic plate. K is the screw which raises and lowers the table L. E is a hinge, regulated by F, which raises and lowers the portion of the spectrograph at M so that the extreme red or violet may be taken by raising or lowering it, respectively. I is an arrangement for moving the photographic plate before the slit H. Upon this plate about twelve spectra can be taken. This spectrograph has a focus of 2 1-2 feet, and uses a replica grating with 20150 lines to the inch. The size of the plate used is 3^ by 4 inches. The instrument is capable of transmitting wave lengths to about 2600. PURIFICATION OF TANTALUM. The tantalum, in the form of Ta2O5, was prepared from tan- talite obtained from Pilbarra Dist. W., Australia. The method used in the separation and purification of the tantalum is a mod- ification of the original method of Wolcott Gibbs for the prep- aration of Columbium from Columbite,* and is as follows: The finely powdered tantalite is fused with twice its weight of KHF2 in a platinum dish, until the tantalite is dissolved, and the excess of HF is driven off. The fusion is dissolved in water, and the gangue filtered off. The filtrate from this is evaporated a little and allowed to stand to crystallize. If crystals do not form on cooling, it is evaporated more and stood aside again. In this manner successive small crops of tantalum potassium fluoride crystals are prepared. These are kept separate, and a specimen of Ta2O5 is prepared from the first crop, by the method explained later, and is examined in the spectroscope. This first crop of crystals of the double fluoride is again dissolved in water. And fractionated further, a specimen of Ta2O5 is also prepared from *Am. J. Sc. & Arts 37, 357. 10 the next first crop and examined under the spectroscope similarly. This fractionation and recrystallization of the double fluoride is continued until the spectrum of the specimen of Ta2Os from the last crops shows no loss of lines, indicating that the impurities have vanished through repeated recrystallization. This final crop of double fluoride crystals is dissolved in water and treated with H2S as a precaution to insure the absence of Sn, W, etc. The precipitate, if any, is filtered off, and the filtrate treated with concentrate H2SO4, and evaporated to dryness. The residue is treated with water and Rochelle salt. All metals will go into solution except the tantalum, which is left as Ta2Os, and which, on washing well, will be found in a sufficiently pure condition for spectroscopic work. THE METHODS OF MEASUREMENT. The measurement of the lines is made in two different man- ners. First the measurement with the comparator, by which means a relative number is given to each line. A curve is then plotted upon cross section paper from which this number can be converted into wave lengths. The error of this method is about 0.5 Angstrom unit. The second manner is by means of the projection method. The plate upon which is the spectrum to be measured is put in an ordinary projection lantern and the image thrown upon a screen from twelve to sixteen feet removed. The lines are meas- ured by means of a centimeter scale, which will be more fully explained later. THE COMPARATOR. The comparator mentioned above was made to order by Zeiss & Co. It consists of two moderately high power stationary microscopes mounted six inches apart. Beneath these is a movable stage, half of which has a scale graduated in tenths of millimeters on its surface, which slides back and forth under, and is read through the right-hand microscope. The other half of the stage is constructed so that it will hold a plate. It has in its middle a slit three-eighths of an inch wide, which traverses the entire length of the surface that is intended to hold the plate. When the stage slides back and forth, this slit is beneath the left-hand microscope, and is free to receive the light reflected from 11 a mirror placed beneath the stage, directly under the microscope. The microscopes are about twenty diameters in power, and have the usual focusing adjustment. Each has beneath the eye- piece a stationary pointer, and two parallel wires moving hori- zontally and governed by a micrometer screw on the outside. The pointer of the left-hand microscope is placed directly over the lines to be measured, and the width of the line may be deter- mind by the parallel wires and micrometer screw attached to this microscope. The pointer of the right-hand microscope has the parallel wires placed directly across it, at which point the microm- eter screw reads zero. The parallel wires are then moved across the nearest scale number, and, then reading of the fractions taken on the micrometer. In this manner, measurement are made to the one-thousandth of a millimeter. The spectrum to be meas- ured is put upon the slit under the left-hand microscope. The stage is then adjusted so that some arbitrary number is under the right-hand microscope, for example, 40.000. The plate is then adjusted under the microscope so that some known line, which is used as a standard, comes between the parallel wires. The stage can be drawn back now so that the far edge of the spectrum is under the microscope, and as each line comes between the cross wires, the scale is read at the right-hand microscope. For exam- ple, we find lines in the copper spectrum on numbers 39.025, 41.000, 51.060, etc., when the line ^5179 is set on 40.000. THE COMPARATOR METHOD. Now to convert these numbers into wave lengths, a chart must be drawn upon cross section paper. Copper has in its spec- trum several prominent lines which are easily recognized. There are three lines in the red, ^5218, ^5153, and ^5105, which read on the comparator 39.025, 40.645, and 41.800 respectively; two lines in the green, ^4651 and '-4275, which read on the comparator 52.900 and 62.340; two lines in the violet, ^4062 and ^4022, on the comparator 67.720 and 68.710. There occur also, several iron lines with the copper in the spectrum, as the copper was not pure; and these assist in plotting the curve. A piece of cross section paper is taken which has along one co- ordinate, the wave length numbers, and along the other the com- 12 parator scale numbers. (Fig. 3.) As the measurement of the plate commenced in the red, the comparator numbers read up, while the wave lengths read down. A place is now found on the chart along the wave length co-ordinate for ^5218, and along the comparator scale co-ordinate 39.025. If the two points found are extended, along the perpendicular and horizontal lines of the cross section paper, they intersect in a point, which is the starting point of the curve. The other lines, of which we know the wave lengths and the comparator reading, are placed in a similar man- ner. When all the points have been found, they are connected by a line which forms the curve. In this case, where the diffrac- tion grating was used, the curve is so slight as to approximate a straight line. Now the method by which the wave lengths are gotten, is to take the comparator reading, find its place along the proper co-ordinate, extend it out until it strikes the curve, and drop a perpendicular from this point to the wave length co-ordi- nate, and the wave length is read off this. THE PROJECTION METHOD. The projection method has the double advantage of being much simpler and more accurate, and the relative intensities of the lines may be determined with greater facility, but at the same time, the faint lines cannot be seen and measured so easily. For this purpose it is only necessary to place the plate in the lantern, and project the image on a screen twelve to sixteen feet away. The lantern is shifted back and forth until the distance between ^5218 and ^5105 is, we will say, thirty centimeters. The wave length difference between these two lines, as can be seen, is one hundred thirteen Angstrom units, the linear differ- ence is thirty centimeters, hence thirty cm. equals 113 A. U., one m. m. equals 113-300 A. U. Now to measure any line between the two standards, one must get the linear distance from either of the standards to this line, and convert the distance into A. U. as shown above, and add it to or subtract it from the standard, according to whether the measurement is made in the direction of the red or violet end. The dispersion not being precisely equal for all colors, new values of A. U. in m. m. must be taken between new standards, as the measurement progresses along the spectrum. Seed's Panchromatic plates were used at first, and then Cra- mer's "Pan Iso Plates" were adopted, the latter being very sensi- tive to the red rays. o o Figure 3 14 About twelve dozen plates Were developed with from eight from twelve spectra on each plate, making a total of about eleven hundred spectra examined. THE ARC SPECTRUM OF TANTALUM. Following, the wave lengths of the tantalum spectrum will be given from ^3 200 to ^6300. The intensities range from i, for the lines easily visible, to 10, for the strongest lines, in the copper spectrum, o is used for lines just visible. An n denotes nebulous, a line not sharply denned ; s denotes sharp ; b denotes band ; and is inclosed in brackets, the wave lengths of the two edges being given. A D following a character letter (as sD) means double. After some of the lines will be seen the symbol of an element followed by a question mark, indicating that the probability is that the line belongs to the element mentioned, Which might have occurred in the spectrum as an accidental impurity. Wave Length. Intensity. Character. Wave Length. Intensity. Character. 6268 6264 6252 0248 6203 6195 6049 6041 6034 6010 6000 5996 5990 5987 5978 5974 5970 5963 5938 5934 5931 5923 5907 5887 5883 5873 sD s s s s n s s s s s n n n n n 5870 5863 5859 5856 5848 5847 5842 5837 5830 5820 5810 5808 5805 5792 5787 5782 5696 5692 5687 5682 5642 5635 5631 5619 5614 5611 3 s I s I s I s I s I s 2 s 2 s I n I n s s o s 4 s i s i s i s 6 s 4 s 2 n 4 s 3 n 2 n iSn? n i s i s i s 15 Wave Length. Intensity. < Character. Wave Length. Intensity. C :hara 5606 i s . 5348 s 5605 o n 5340 s 5602 n 5336 s 5595 i n 5334 s 5564 o s 5330 s 556o o s 5328 s 5530 I n 5327 s 55i8 2 n 5324 s 5497 I n 5322 2 s 5494 I n 5317 2 s 5492 I n 5313 i n 5488 2 s 5307 I s 5482 I s 5304 i s 5478 I s 5302 I s 5473 I s 5298 3Cb? n 5467 I s 5294 8 s 5463 I n .5289 3 s 5459 2 s 5062 8 s 5456 I s 5055 2 s 5450 I s 5052 2 n 5449 I s 5050 I s 5447 I s 5047 I s 5444 I n 5045 2 n 5440 I s 5044 I s 5436 1 s 5040 2 s 5433 I s 5034 3 s 5431 I sD 5031 I s 5420 I s 5022 n 54i8 I s 5016 n 54U I s 5012 o n 51 T 1 s 5009 n 5407 I s 5003 I nD 5405 I s 5000 oTi-La? n 5400 5 s 4996 o n 5397 iFe? s 4993 i n 5393 s 4989 i n 5391 s 4988 i V 5388 s 4986 i n 5382 s 4980 2 s 5378 s 4976 2 s 5372 s 4972 I s 5369 s 4870 I n 5366 s 4965 I n 536o s 4951 3 s 5356 n 4949 i s 5354 s 4940 i s 5352 s 4938 I s 16 Wave Length. Intensity. Character. Wave Length. Intensity. Character. 4927 \ 4865) J2Fe.Ti. (2Fe.Ti. Cb? nb Cb?nb 4118 4113 3 3 s s 4850 2 n 4109 2 s 4847 I n 4105 2 s 4837 2 . s 4095 2 s 4834 2 s 4091 2 s 4829 3 n 4074 2 s 4826 3 n 4070 2 s 4823 2Mn? s 3996 2 n 4821 I s 3989 2 n 4819 I s 3986 2 n 4815 o s 3983 2 s 4812 2 s 3979 2 Cb? n 4810 3 s 3975 2 s 4807 i n 3971 2 s 4803 2 s 3969 2 s 4800 n 3961 4 s 4769 s 3956 2 FeCa s 4764 s 3952 2 s 4754 s 3947 2 s 4752 s 3944 2 s 4686 2 s 3926 2 s 4656 2 n 3921 I n 4637 2 s 3914 4 s 4582 s 3910 4 s 4577 s 3905 s 4458 s 3901 s 4432 s 3900 s 4430 s 3898 s 4427 s 3896 s 4424 s 3888 s 4421 s 3881 s 4397 s 3874 3 s 4387 s 3851 i s 4376 Fe? s 3787 3 s 4372 s 3785 3 s 4362 s 378i 2 n 4357 o s 3778 4 s 4267 s 3774 2 s 4240 s 3770 2 s 4214 s 3768 I s 4211 s 3764 2 s 4207 s 3691 3 s 4205 s 3686 3 s 4204 s 3682 4 s 4198 s 3678 7 s 4188 s 3672 3 s 17 Wave Length. Intensity. Character. Wave Length. Intensity. Character 3670 5 s 3462 i s 3669 3 s 3456 i s 3667 i s 3452 i s 3665 3 s 3448 o Cb? s 3659 2 s 3446 o s 3651 2 s 3440 5 n 3649 3 s 3430 o s 3640 3 Fe ? s 3423 5 n 3639 2 Fe? s 3414 o 3 3635 4 s 3412 o s 3631 2 Fe? s 3405 o s 3620 4 s 3397 o s 3612 2 Fe? n 3395 o n 3590 4 s 3393 o n 3586 2 s 3390 o n 3584 2 s 3387 o n 3576 2 s 3385 o n 3572 2 s 3383 o n 3567 3 s 3380 o n 3554 5 "D 3378 o s 3547 5 n 3375 o s 3545 i s 3371 o s 3542 5 nD 3368 o s 3538 4 s 3366 o s 3532 2 s 3361 o s 3520 2 s 3347 o Cb? s 3507 3 n 3345 o s 3503 2 s 3339 o s 3499 2 s 3333 s 3496 2 s 3306 o Fe? s 3493 i s ,3300 ^ o s 3491 5 Cb? n 3297 o s 3489 i s 3295 o s 3487 i s 3292 o s 3485 5 n 3287 o s 3482 i s 3283 o s 3480 i s 3280 o s 3475 2 n 3262 o s 3471 2 s 3257 i s 3470 i n 3251 i s 3466 i n 3246 i s In the spectrum of columbium recently published by Hilde- brand* there are several lines to be found, which correspond *J. Am. Ch. S. 30, 1677. 18 both in wave length and in intensity, to lines found in the spec- trum of tantalum. There is every reason to suppose that the Cb2O5 used by Hildebrand was as pure as could be prepared, With the present knowledge of columbium and its allied elements. And since We have the same assurance with regard to the Tb2C>5 used in this work, there is either an unknown element associated with tantalum and columbium or a common impurity inseparable by the methods used in purification ; for if the impurity in the specimen of Tb2C>5 used by us was columbium, the lines should appear weaker in the tantalum spectrum, and vice versa. Following, the coincident lines are given. To the right are the tantalum lines with intensity, to the left are Hildebrand 's lines for columbium. Wave Length. Intensity. Wave Length. Intensity. 5366.0 I 5366 I 5340.1 I 5340 I 5321.9 2 5322 2 5317.1 2 5317 2 5302.4 I 5302 I f 5054.9 2 5055 2 4992.6 I 4993 I 4764-0 I 4764 I 4357-0 i 4357 o 4203.6 I 4204 I 3764.2 2 3764 2 3667.1 I 3667 I 3496.2 2 3496 2 3387-1 i 3387 o 3361.0 i 3361 o 3262.0 i 3262 o The following lines, which appear in the tantalum spectrum, do not appear in Exner and Hascheck's Spectrum of Columbium, but are attributed to this latter element by Hildebrand* and inserted in his table of wave lengths. To the right are the tanta- lum lines, to the left are Hildebrajid's lines: Wave Length. Intensity. Wave Length. Intensity. 3888.6 I 3888 I 3287.1 i 3287 o 3257.2 i 3257 o *Loc. cit. 19 The following strong lines are unidentified in Rowland's Solar Spectrum,* and are probably due to tantalum : Wave Length. Intensity. Wave Length. Intensity. 5682.427 2 5682 4 5012.335 i 5012 o 5031.199 3 5031 i 3910.469 3QIO 4 3910.670 2j 3682.310 2 3682 4 3554-593 2 3554 5 3538.643 3538.399 SPECTRUM OF A CLAY RESIDUE. Below are given the Wave lengths, which occur in the spec- trum of precipitate A, and which have not been positively identi- fied. The lines corresponding to Ti, Ta, Fe, Na, etc., have been removed from the table : 5890 2 Na? s 4940 4 s 5885 2 s 4275 4 Cr? s 5883 2 s 4110 3 . n 5880 2 s 4100 3 Cb? n 5875 3 s 4094 3 n 5866 2 n 3995 3 s 5860 3 n 3990 6 s 5853 3 Ba? s 3982 6 s 5850 3 s 3973 7 Ca? s 5844 8 s 3951 8 n 5834 8 Cb? s 3942 9 n 5749 10 s 3931 6 n 5167 4 Mg? s 3919 5 Cb? s 4946 2 Cb? s 3909 5 Cb? s The following lines are unidentified in Rowland's Solar Spec- trum, and occur in the spectrum of precipitate A. Wave Length. Intensity. Wave Length. Intensity. 4094.573 2n 4094 3n 3995.352 2 3960 6 3990.248 o 3942 9 3942.510 2 3931 6 3942.380 o 2995 2 3931.030 o *See Rowland's Solar Spectrum. 2(1 SUMMARY. 1 . The arc spectrum of tantalum has been measured between ^3200 and ^6300. 2. The method of using H2SO4 and HNC>3 for obtaining clays in solution, is satisfactory. 3. The spectrographic method for examining precipitates, for rare elements and traces, is pre-eminently superior to ordinary spectroscopy. . 4. There is probably an unknown element associated with columbium and tantalum, or a common impurity, inseparable from either element, by ordinary means of purification. 5. There is a doubtful element or group of elements in the clay from Spruce Pine, Alabama. 6. Tantalum has been identified as one of the metals in the sun from lines which Were marked unknown by Rowland. 4698o - i c UNIVERSITY OF CALIFORNIA LIBRARY