The reactions of "conjugated double bonds" had not lent themselves to an experimental decision between the two theories. The reduction of ketones to pinacones had, even on the basis of the Thiele theory, necessitated the assumpion that it was the oxygen which reacted first. But if these reactions had not given any indication of free affinity at the carbon atom of a carbonyl-group, they also had not disproved it. It is in the aromatic series that a crucial test has now been possible, and the verdict is against an unsaturated carbonyl-carbon atom as postulated by Thiele's theory of partial valencies, and therefore against that theory itself. GAS ANALYSES BY FRACTIONAL DISTILLATION AT LOW TEMPERATURES.* By G. A. BURRELL and F. M. SEIBERT. THIS paper describes experiments that resulted in the separation of a natural-gas sample into the individual paraffin hydrocarbons present. This had not been accomplished hitherto. Natural gases may contain only methane as the combustible constituent or may be mixtures that contain large quantities of the higher gaseous paraffins. In some samples the latter predominate. In addition, there may be vapours of the liquid paraffin hydrocarbons present, sometimes enough to warrant the installation of a plant for their extraction. The natural gas used in Pittsburgh, Pa., is a complex mixture, and is typical of gas that is supplied to many cities to the extent of billions of cubic feet per year. The exact composition of this gas is of importance to the Paper presented before the Spring Meeting of the American Chemical Society, April 7-10, 1914, by permission of the Director of the Bureau of Mines, U.S.A. Bureau of Mines, because it is used in testing explosives, safety lamps, electrical mining machinery, and other mining appliances. By the scheme shown herein it is also possible to determine more closely the quantity of the vapours of the liquid paraffins in a natural gas mixture than has been possible heretofore. Gases that contain enough of these vapours are compressed and cooled at many plants and the condensate sold as gasoline. It is generally known that ordinary combustion gas analyses give but little indication regarding the individual hydrocarbons present in a natural gas mixture. Only the two predominating paraffins are shown. In the experiments reported herein, natural gas was first liquefied by means of liquid air, and the different paraffin hydrocarbons separated by properly adjusting temperatures and removing the various fractions with a mercury pump. These fractions were analysed by the ordinary slow combustion methods. Advantage was taken of the work of P. Lebeau and A. Damiens (Comptes Rendus, 1913, clvi., 325), who prepared various mixtures of the gaseous paraffins, liquefied them, and partially separated them. This work is an advance over their work in that the separation was made into single constituents. The important part of this paper, however, is the application of the work to the determination of the constituents of natural gas. Such a separation is possible because in the liquid condition the boiling points of the gaseous paraffins are rather widely separated. These boiling points follow:Methane, 160 C.; ethane, - 93° C. ; propane, - 45° C.; N-butane, +1° C., and isobutane - 10 C. The two butanes were not separated. In order to finish the work with fractions large enough for accurate analyses the experiment given herein was started with about 1 litres of gas (1531 cc.). Other experiments were performed with various natural gases in which smaller quantities were used. The sample as analysed by ordinary slow combustion methods contained the following constituents : There is also about o'03 per cent of carbon dioxide in the gas mixture. Carbon monoxide, hydrogen, and olefine hydrocarbons are not present. The experimental procedure follows: Fig. 1 shows the general arrangement of the apparatus. The Topler pump is on the left of the photograph. (A) is a small glass vessel used for holding the liquefied gases. It could be enclosed in the Dewar flask (B). Surrounding this Dewar flask is shown another and larger one. This arrangement was adopted in order to provide better insulation than was afforded by only one flask. The gas sample prior to liquefaction was measured in the glass vessel (c), then transferred to the gas burette (D), and from there passed into the liquefying bulb (A). At (E) is shown a mercury manometer for registering pressures in the pump. At the base of the Topler pump are shown the glass vessels for trapping the different gas fractions over mercury as they were removed. air or liquid oxygen. The entire sample was first liquefied by means of liquid With the gas in the liquid con. dition, connection was made between it and the mercury pump, and as much of the gas removed with the pump as possible. This process divided the original quantity into two portions; first, a gaseous portion; and second, a liquid residue. In other words, the vapour pressure of liquid ethane (boiling-point - 93° C.) is so small at the temperature of liquid air that none could be detected in the distillate within the experimental error of making the analysis. It was found that when liquid air was used that had stood for some time so that its boiling-point had risen to a point FIG. 1.-APPARATUS FOR FRACTIONAL DISTILLATION OF GASES AT LOW TEMPERATURES. methane and nitrogen were removed from the original again liquefied at the temperature of liquid air. Connear to the boiling-point of oxygen (- 183° C.), that the | fractionation was allowed to volatilise, measured, and mixture more quickly than when newly made liquid air was nection was again made to the pump, and more methane used. This is to be expected. The residue from this first removed. In other words, although the residue from the first fraction was treated in exactly the same manner as the original sample more methane was obtained. Upon volatilising the entire residue, however, and again liquefying a re-arrangement of the solution occurred, and chance for faster evaporation of this last methane portion was afforded. In no case could the last minute portions be recovered, so the attempts at complete recovery were stopped when it was found that only such a small propor tion was being left behind as did not sensibly affect the results. To this point the first series of fractionations had reached a stage where the larger portion of the methane had been removed by the first fractionation and where the first residue had been volatilised, re-liquefied, and pumped to obtain another small portion of methane. The residue from the second liquefaction was treated again in the same identical manner, and more methane obtained. A further identical treatment resulted in no additional recovery of methane. No indication of methane was found in the ethane portion within the error of making the analysis. The distillate obtained by the above scheme undoubtedly contained a trace of ethane, but so small that it could not be detected by analysis. The analysis of a portion of the total methane and nitrogen fraction follows: TABLE I.-Analysis of a Portion of the Total Methane and Nitrogen Fraction. Sample taken O2 added.. Total volume Cc. Cc. Volume after combustion (a) Methane from contraction (a) Methane from CO2.. Per cent methane from contraction Average per cent methane .. 70'10 59'00 59'40 36.80 40'50 29'50 29.60 29.44 29.64 29'59 29.67 98.2 98.3 98.2 98.1 98.2 97.8 (a) Corrected for the molecular volume of CO2. Second Series of Fractionations. The next step in the process involved the separation of the ethane from the methane free residue. This necessi tated the employment of a temperature such that practically all of the ethane could be separated from the still higher paraffins, propane, the butane, &c. The temperature used could not be too low, else the ethane itself could not be separated, nor so high as to also remove all the propane. A natural gas condensate obtained from a natural gas gasoline plant by subjecting natural gas (casing head gas) from an oil well to a pressure of 250 lbs. per square inch, and then cooling it to ordinary temperature, proved excellent when cooled by liquid air for obtaining temperatures higher than the temperature of liquid air. This condensate is known in the natural gas gasoline trade as "wild" gasoline. It contains large quantities of liquid propane and the butanes (especially the latter), as well as some of the ordinary gasoline constituents, the pentanes, hexanes, &c. Other substances tried for obtaining low temperatures, such as alcohol, ether, methyl, and ethyl chloride, &c., jellied so much at low temperatures that they could not be used satisfactorily. The mass did not remain of uniform temperature from top to bottom. In order to obtain a temperature of 145° C., for instance, the condensate was placed in a Dewar flask and stirred with a test-tube into which liquid air was run until - 145° C. was reached. Upon removal of the liquid air the condensate warmed up very slowly, about 5° to 10° C. per hour, thereby affording sufficient time for the withdrawal of the vapour from the liquefaction bulb. In separating ethane from the methane free residue the latter was first cooled to a temperature of - 145° C., and pumping started and continued until the temperature had risen to - 125° C. By this process there was obtained a distillate consisting of ethane and propane. In other words, some propane (boiling-point - 45° C.) is removed at - 125° C. as well as the ethane (boiling point - 93° C.). The residue was then treated twice in the same manner, the final separation of the ethane being made at a temperature of -155° to -140°C. The temperature was purposely lowered to - 125° C. to obtain practically all of the ethane as well as some propane, because it was found easier to separate the ethane from that part of the propane that came over than to attempt to pull off all of the ethane from the original residue. All the ethane was obtained, there being only small quantities of propane remaining as a residue. The analysis of a portion of the total ethane fraction follows :— According to the equation C2H6 + 3'5O2 = 2CO2 + 3H2O, the contraction should be equal to the CO2 × 125. In No. I analysis the contraction then becomes 511 CC. X 125 = 63.87 cc. This corresponds well with the contraction actually observed, 63 80 cc. In No. 2 analysis the contraction is equal to 40 50 cc. X 1'25 = 50-€2. The contraction actually observed is 50 50 cc. In calcu lating the ethane from the carbon dioxide and contraction use was made of equations that correct for the deviations of carbon dioxide and ethane from the ideal conditions as follows("Errors in Gas Analysis Due to Assuming_that the Molecular Volumes of all Gases are Alike," by G. A. Burrell and F. M. Seibert, Technical Paper 54, U.S. Bureau of Mines; CHEM. NEws, cix., 188, 196): No. 1 Analyses. 0.990 C2H6+3'5O2=1·992 CO2 + 3H2O. cc. ethane = 0.396 x contraction = 25.26. cc. ethane = 0.497 × CO2 = 25'39. No. 2 Analyses. 0.990 C2H6+3'5O2 = 1·994 CO2+3H2O. cc. ethane = 0.397 × contraction = 20:05. cc. ethane 0'496 × CO2 = 20.09. Third Series of Fractionations. The final residue from the second series of fractionations then contained propane and higher paraffins. The ethane free residue was next liquefied and pumped at a temperature that started at 125° C. and ended at - 110° C., the object being to remove practically all of the propane (boiling point -45°), and also some of the butanes (boiling-points + 1° C. and -10° C). The residue from this operation was again treated in the same manner to obtain any propane that still remained behind. It was thought that if a temperature was used that would permit the distillation of an appreciable quantity of butane, practically all of the propane should come over. The total distillate obtained in this manner was then liquefied, and pumped at a temperature ranging from 135° C. to 120°. There resulted a distillate that consisted of propane only. In other words, propane can be separated from the butanes at a temperature between - 135° and 120° C. The analysis of a portion of the total propane fraction follows: According to the equation C3H8 +502=CO2+4H2O, the contraction - CO2 =0'0. In No. I analysis 38.8 cc. -39'1 cc. -03 cc., and in No. 2 analysis 38.1 cc. -37'9 CC. = 0.2 CC. The propane was calculated from the following corrected equation : 0986 C3H8 +502 = 2·991 CO2+4H2O. Then according to No. 1 analysis the propane when calculated from the contraction is 0.329 × 38 80 = 12.76 cc., and when calculated from the CO2 is 0.329 x 39°10=12.86 cc. According to No. 2 analysis the propane when calculated from the contraction is 0.329 × 40 ̊5 = 13·32 cc., and when calculated from the CO2 is 0.329 x 40'3 - 13.26 cc. In the case of both analyses the cc. of propane as calculated from the CO2 and the contraction agree closely. The value o 986 or the molecular volume of propane at 0° C. and 760 mm. of mercury was calculated from Van der Waals' equation- or a (P + ~ ) (V+ b) = RT M do 2 · (1 +a) (1 − b) = R. This value, as far as the authors are aware, has never been verified experimentally as in the case of oxygen, methane, ethane, and carbon dioxide. The same procedure was followed in the case of the propane separation as in the case of the ethane and methane separations. Distillates and residues were lique fied and pumped until no propane could be obtained. The analysis of a portion of the final residue follows. This should consist of butane only, providing no vapours of the liquid paraffins are present. Analysis of a Portion of the Total Butane Fraction. from, the equation C4H10+502-4CO2+5H2O, the CO2. From the above analysis, contraction X I'14 = 330X11437.62 cc. 37·62 -37.00 -0.62 cc. difference. was calculated to butane only, and was not the pressure of methane at that temperature, because some nitrogen was present. The pressure persisted throughout the pumping around this value until near the end, when it suddenly dropped to o'o mm. Then the pumping was stopped, and the residue allowed to volatilise, and again liquefied and pumping continued until no more distillate was obtained. Three liquefactions of the residues were usually necessary for the removal of all the methane. When the methane- and nitrogen-free residue was liquefied for the removal of the ethane the vapour pressure of the mixture was about 2 mm. at 155°, and about 4 mm. at 145°. When nearly all the ethane was removed, the pressure dropped off suddenly to o'o mm. when the pumping was stopped. The residue was again liquefied, and treated in the same manner until all the ethane was removed. Three successive treatments of the residue were usually sufficient. After the removal of the nitrogen, methane, and ethane from the mixture had been accomplished, it became necessary to separate the propane from the butanes, &c. This was accomplished at -130° C. to -120° C. The vapour pressure at 130° C. was about 0.5 mm., and about 1 mm. at 125° C. The pressure dropped suddenly to o'o mm. after nearly all the propane had been removed. The residues were then treated as previously described until all the propane had been removed. The final residue consisted of the butanes and any vapours of the liquid hydrocarbons that were present. (To be continued). THE SEPARATION OF TITANIUM FROM IRON, ALUMINIUM, AND PHOSPHORIC ACID WITH THE AID OF THE AMMONIUM SALT OF NITROSOPHENYLHYDROXYLAMINE ("CUPFERRON "). By WILLIAM M. THORNTON, Jun. IT has been shown by the author (Am. Journ. Sci., 1914, xxxvii., 173) that titanium can be quantitatively precipitated in solutions moderately acidified with sulphuric acid and containing also tartaric acid by the ammonium salt of nitrosophenylhydroxylamine ("cupferron "). Making use of this fact, an indirect separation of titanium from iron has been accomplished. In this method, after precipitating the iron as ferrous sulphide by ammonium sulphide in the presence of ammonium tartrate and filtering from the aforesaid ferrous sulphide, the iron free filtrate is acidified with sulphuric acid, the hydrogen sulphide boiled out, and the titanium_precipitated in the cold by the "cupferron" reagent. The yellow precipitate thus produced is collected and ignited to titanic oxide. At the close of the article it was stated that it would seem that there should be no great difficulty attending the separation of titanium from iron, aluminium, and phosphoric acid. Experiments with a view to accomplishing these separations have been fraught with very interesting and very gratifying results. Although Schröder (Zeit. Anorg. Chem., 1911, lxxii., 89) has made the statement that titanium and zirconium could be quantitatively precipitated by the "cupferron reagent, no experimental data have appeared on the subject until Bellucci and Grassi (Gazzetta Chimica Italiana, 1913, xliii. [I.], 570) showed that titanium could be quantitatively thrown down by the reagent and that a brought about by a single precipitation in acid solution. The above analysis appears to be almost entirely this gas, but undoubtedly a very small proportion of the vapours of the liquid paraffins clean separation of titanium from aluminium could be were contained in the mixture. Vapour pressures obtained by means of a manometer Their results leave little to be desired for accuracy. But attached to the pump furnished evidence which indicated it is indeed surprising that these writers do not give in any when a separation had been accomplished. For instance, but the most indefinite terms the acid concentration or the at the temperature of liquid air the vapour pressure of the absolute volume of the solutions in which the titanium hydrocarbon mixture was 22 mm. This vapour pressure was precipitated. They say that the solution should be notably but not excessively acid with either sulphuric or hydrochloric acid. Again, under the separation of titanium from aluminium they say that the same acidity was used for the separation as for the precipitation of titanium alone, and that the precipitate was washed first by decantation and then on the filter with very dilute hydrochloric acid. Just how large a quantity of either of the above two acids can be employed and the precipitation of the titanium still be complete has not yet been determined. But the author's experiments have clearly demonstrated that if the concentration of sulphuric acid be too small, the aluminium is carried down with the titanium in considerable measure. the analysis, viz., the separation of titanium from aluminium. Since the conditions in experiments (3) and (4) are the same as those in experiments (1) and (2) respectively, excepting the presence of tartaric acid, it is obvious that if this acid be present in the solution, the concentration of free sulphuric acid necessary to bring about a clean separation can be greatly diminished. In experiments (5) and (6) it will be noticed that with small increments in acidity the error is very slightly lowered. If the original solution for the analysis occupy a volume of 100 cc. the author has found that in his experience the filtrate and washings from the ferrous sulphide will occupy a volume of about 350 cc. An absolute volume of 400 cc. was therefore chosen as a convenient one in which to make the precipitation of titanium. Accordingly experi ments (7) and (8) were performed with the object of learning the quantities of sulphuric acid and of tartaric acid necessary to effect a good separation in the above given absolute volume. In these two experiments the concentration of sulphuric acid was the same as in expre-periments (4) and the tartaric acid increased somewhat over the amount previously used. Since the results of experiments (7) and (8) were satisfactory it was not thought necessary to carry the series further. For these experiments two standard solutions of titanic sulphate were employed. These were made by acting on specially prepared potassium fluotitanate with concentrated sulphuric acid until all the hydrofluoric acid had been displaced, diluting and making up to known volume. The first solution was standardised by taking two weighed portions of 25 cc. each and precipitating the titanium by hydrolysis of the acetate upon boiling in the manner viously described by the author (loc. cit., 174). Duplicate determinations gave the following results:- 01427 grm. 0.5226 per cent Since these two results agreed so closely, the value obtained in (b) was arbitrarily taken as correct. The second solution was standardised by taking weighed portions of 25 cc. and 24 cc. respectively and determining the titanium in one (a) by the acetate method and in the other (b) by the "cupferron" method of Bellucci and Grassi (loc. cit.). Two determinations gave the following results : Titanic oxide. Titanic sulphate solution. (a) 25 cc. = 27.814 grms. o'1066 grm. 0.3822 per cent (b) 24 cc. 26.667 grms. O'1022 grm. 0.3822 per cent Since these two determinations agreed exactly, the value here obtained was taken as correct. The first series of experiments were carried out with a view to obtaining favourable conditions for the separation of titanium from aluminium. Known quantities of aluminium were added by weighing off dry ammonium aluminium sulphate, which had been purified by two re-crystallisations. To the solution containing the titanium and aluminium and a weighed amount of tartaric acid (except in the case of experiments (1) and (2) which were made in the absence of tartaric acid), re-distilled ammonium hydroxide was added until the solution became neutral to litmus paper. A definite volume of sulphuric acid (made by diluting one volume of acid of sp. gr. = 1.84 with an equal volume of water) was then added and the solution made up to the volume designated in the last column of Table I. Twenty cc. of a 6 per cent "cupferron " solution was then added and the beaker set aside for the precipitate to settle. The time of standing does not appear to matter a great deal. Two hours was found to be satisfactory; but no doubt a shorter time would suffice, and in a few cases the precipitate was allowed to stand about twelve hours with no harmful consequences. The precipitate was filtered on paper with the aid of very gentle suction, washed twenty times with water containing 20 cc. of hydrochloric acid of sp. gr. = 120 per litre, placed in a tared platinum crucible, and dried at 110° C., ignited at first very carefully with the cover on, then with the crucible inclined, and open until all the carbon had been consumed, and finally brought to constant weight over the Meker burner. From experiments (1) and (2) it is evident that if the concentration of free sulphuric acid be too small, very serious errors are made in the direction of gain of weight on the titanium. Since, after the separation of the iron, the filtrate from the ferrous sulphide contains already tartaric acid, it was considered advisable at this point to earn the effect of tartaric acid on the next operation in 2. 0.1066 Grm. Grm. Error. Tar- Vol. of 01094 +0.0113 5 200 0.0590 + 0.0018 5 I'2 200 ΙΟ 01127 O'1127 00577 I'2 200 O'1127 0.0579 I'2 200 01127 00575 1.2 200 O'1127 01072 20 I'5 400 01127 01070 I'5 400 The second series of experiments was begun with the object of determining how great a concentration of sulphuric acid could be employed in the presence of tartaric acid without any resultant loss of titanium. The mode of procedure was the same as in the separation of titanium from aluminium except that in experiment (9) the precipitate was washed with water and in experiments (11) and (12) with hydrochloric acid (made by diluting 100 cc. of acid of sp. gr. 120 to one litre). Since in experiment (12) the concentration of sulphuric acid was twice as great and the quantity of tartaric acid greater by half a grm. also than were found necessary to ensure a clean separation of titanium from aluminium, and since the error on titanium was one of gain, it appeared that there was little to be profited by carrying the series further. TABLE II. Fresenius (Zeit. Anal. Chem., 1911, 1., 35) has succeeded in obtaining a quantitative separation of iron from phosphoric acid in a solution acid with hydrochloric acid by means of the "cupferron" reagent. Mindful of this, it occurred to the author that in a similar way it might be possible to separate titanium from phosphoric acid. The third series of experiments was therefore undertaken with this object in view. Approximately known quantities of phosphoric acid were added by weighing off dry portions of Baker's analysed disodium hydrogen phosphate. The conditions of experimentation were like those described above for the titanium aluminium separation. Table III. |