The idea of determining the amount of bromine an oil will take up is not a recent one, but dates back to 1857, to Cailletet, a Frenchman, who describes a method of treating oil with an alcoholic solution of bromine, and titrating the excess bromine with a turpentine solution until the colour is discharged.

In 1881, A. H. Allen proposed a method for brominating oil in which the bromine was liberated in contact with the oil by the decomposition of a water solution of sodium hypobromite with hydrochloric acid. Two years later Mills and Snodgrass proposed the use of carbon bisulphide as a solvent. These various schemes were all intended, principally at least, for the identification of vegetable or animal oils, which gave characteristic bromine absorptions, and for the detection of their adulteration.

The most satisfactory method involves the use of carbon tetrachloride as a solvent. This scheme was first used by McIlhenny (Fourn. Am. Chem. Soc., xvi., 275; ii., 1084), and is recommended principally because of the increased stability of the carbon tetrachloride bromine solution, and of a sharper end-point in titrating. One of the reasons for the increased stability of the carbon tetrachloride over the carbon bisulphide can be inferred from a consideration of their boiling-points,

Carbon bisulphide boils at 43° C.; bromine at 63° C.; carbon tetrachloride at 77° C.

Another is the fact that carbon bisulphide forms various sulphur bromides in solution, and liberates free hydrobromic acid.

The method we have used is an adaptation of McIlhenny's. Twenty-five cc. of a tenth normal solution of bromine in carbon tetrachloride, previously standardised against sodium thiosulphate, is agitated with to grm. of the oil to be tested in a stoppered Erlenmeyer flask. Before mixing, the sample of oil should be dissolved in 25 to 30 cc. of carbon tetrachloride. The mixture is then allowed to stand in the dark for thirty minutes. The carbon tetrachloride solution and the oil to be tested should be previously dried by use of a few sticks of fused calcium chloride.

At the end of thirty minutes about 100 cc. of distilled water is added to the flask, with 2 grms. of potassium iodide, and the unabsorbed bromine titrated against a tenth normal sodium thiosulphate water solution. Starch solution is used as an indicator. The end-point is reached when the bromine colour of the oily layer has disappeared, and the iodine blue of the starch and water solution is just discharged. The oily carbon tetrachloride solution should

The hydrobromic acid formed during the bromination is titrated against sodium thiosulphate by adding half a grm. of sodium iodate to the flask, after the titration is finished, and titrating to the second disappearance of the blue colour. Or the hydrobromic acid may be titrated against N/100 caustic after separating from the oily layer. Better results can be obtained from the iodate method by first separating the oily layer from the water solution.

Both the absence of sunlight and the careful drying of both oil and tetrachloride are most necessary, as light accelerates the bromination of the saturated compounds, giving rise to an undue proportion of substituted bromine; and the presence of the smallest amount of moisture is even more effective in accomplishing the same result. The addition of the bromine solution and subsequent titration with sodium thiosulphate need not, however, be made under artificial light, as by carefully duplicating the conditions of experiment very concordant results may be obtained. The total bromination number is the number of grms. of bromine absorbed per 100 grms. of oil. The bromine substitution number is the amount of bromine found as hydrobromic acid after the bromination is completed, calculated to the same basis. The bromine addition number is cal culated by subtracting twice the bromine substitution number from the total bromination. This calculation is based on the following equations :—

2C2H4 Br2 2C2H4Br or C2H4+ Br2 = C2H4Br2
Ethylene. Ethylene
di-bromide.

Ethylene.

Ethylene bromide.

It is evident that for every substituted bromine compound formed an equal amount of hydrobromic acid is formed. Hence, it is necessary to deduct both the bromine of substitution and the hydrobromic acid bromine from the total absorption to find the amount going to form addition products. Both the olefines and the acetylenes form addition products, so that in a general way it may be assumed that large addition numbers indicate acetylenes, olefines, or ring compounds with unsaturated side chains. Large substitution numbers - benzines and small total brominationparaffins or naphthenes. If paraffins were attacked by slight exposure to daylight, substitution products would be formed.

While it would not be fair to assume that the members of the different series are present in direct proportion to the substitution, addition, or absence of bromination in the case of a single oil, between the various oils the results of bromination would indicate which oil contained more of a particular series.

The above described process was carried out on samples of oils differing largely in character and from widely separated fields. Through the courtesy of the St. Louis, San Antonio, Los Angeles, and Denver Gas Companies, and the Standard Oil Company, we received samples of gas oils of the principal types found in the United States. We also tested samples of Canadian crude. Fractionations, sulphur content, heat values, and bromine absorp. tion are given in Tables I. and II.

Pennsylvania oil, Florence, Col., Boulder, Col., and Texas oil appear to have a predominance of paraffins or naphthalines. Detroit City Gas Company's oil, and Laclede (St. Louis) oil seem to be largely olefines or unsaturated ring compounds, whilst Los Angeles oil is remarkable for the presence of benzines, and Canada oil for the absence of them.

The proportionate large amount of additional bromination in the case of Canadian crude is further evidence of the presence of small amounts of olefines. The apparent presence of olefines or other unsaturated compounds in the Detroit City Gas Company's oil, and in the Laclede, is probably due to the cracking of the original oil. Toluol |

The lack of action in the case of benzine is to be expected in the absence of carrier. It is readily brominated in the presence of certain foreign bodies which promote the action. The failure of benzine to react with bromine in a pure state indicates what an important effect a mixture of oils and the products of bromination may have on the amount of bromine absorbed. To test the effect of mixing oils on bromine absorption, the Detroit and San Antonio oils were mixed in exactly equal proportions by weight with the following results :

It is apparent that the effect of increasing the number of constituents increases the amount of bromination. The total bromination of a mixture may then be assumed to be above figures calculated from bromination numbers of pure constituents. Evidently impurities in oil have the effect of accelerating slightly the bromination of hydrocarbons aside from their own action in absorbing bromine. Hydrogen sulphide is the impurity which probably causes the greatest error in interpreting the results of bromination, as it forms hydrobromic acid and free sulphur. If all the 2:09 per cent of sulphur in San Antonio oil were hydrogen sulphide, the error would be 5.2 grms. of bromine per 100 grms. of oil, or sufficient to account for half the total bromination. However, in oils giving large bromine absorptions the amount of hydrogen sulphide present is nil, or negligible, and the possible error due to its presence is very slight. To test the value of the method in identifying a single type of oil, the following samples were brominated: Grand Rapids and St. Louis gas oils are included in Table III. only because they are all approximately of the same type. Detroit gas oil has the following average bromination numbers :

It varies from the average figure by +9.8 per cent to - 10.7 per cent. Such variation would be expected in view of the range in specific gravity from 33.6° B. to 36.1° B. The Denver Gas Co.'s oil shows a striking similarity to the Pennsylvania gas oil, and, in turn, to the San Antonio gas oil, though the first two are undoubtedly largely paraffins, and the last a polymethylene or naphthaline. An idea of the bromine values of oils which are published can be obtained from Table IV.

The report of Mills and Snodgrass on shale oil in Table IV. shows that the absorption of bromine with shale oil varies inversely as the density of the oil.

Summary.

By treating gas oils with bromine dissolved in carbon tetrachloride, under specified conditions, a characteristic bromine absorption number can be obtained having an

tion.

Ο ΙΟ

(a)

(b)

18.3

33'5

22.24

31.9 20.59

28.0

12.59

26.0

(c)

11.72

45'0

27'57

45'5

30.31

(a) Allen, "Commercial Organic Analysis," ii., 95. (b) Mabery, Am. Chem. Journ., April, 1901, p. 255. (c) Mills and Snodgrass, Journ. Soc. Chem. Ind., Nov. 29, 1883, pp. 336, 435.

MOST of those who have worked on the electrolytic production of calcium have been content to describe their methods and apparatus, with perhaps an analysis of a selected lump of their product, but say nothing of the current efficiency obtained. Muthmann (Zeit. Electrochem., 1904, x., 508), in the discussion of a paper by Rathenau, states that by using an electrolyte consisting of two parts of calcium chloride to one part of the fluoride, and raising the iron cathode as the metal accumulated, he obtained a "good" yield. Goodwin (Fourn. Am. Chem. Soc., 1905, xxvii., 1403), electrolysing pure calcium chloride

Paper presented at the Eighteenth General Meeting of the American Electrochemical Society, in Chicago, October 13-15, 1910. The work described in this paper is a continuation of that published by Frary and Badger (Trans. Am. Electrochem. Soc., 1909, xvi., 185), where a summary and discussion of previous work on the electrolytic production of calcium will be found.

7

in a similar way, obtained a current efficiency of 21.5 per cent to 419 per cent. Woehler (Zeit. Elektrochem., 1905, xi., 612), using an electrolyte containing 100 parts of the chloride to 17 parts of the fluoride, claims an efficiency of over 80 per cent, but states that the electrolyte deteriorated in time, probably owing to the formation of the hydrated oxychloride, and that hydrogen (!) is then liberated at the cathode, and the yield decreases. He used a current of 40 ampères at 33 to 38 volts, but does not say how often or for how long a time this efficiency could be obtained. Tucker and Whitney (Journ. Am. Chem. Soc., 1906, xxviii., 84) improved the apparatus of Goodwin, and claim an efficiency of 60 per cent, using the pure chloride as electrolyte, but give no data to support the claim. As far as we are able to ascertain, no one else has published any results on the efficiency of the various processes which have been proposed.

During previous work in this this laboratory (Frary and Badger, loc. cit.), efficiencies as high as 46 per cent were obtained, but, unfortunately, the only runs in which the necessary data were taken were made under unfavourable conditions. As it was certain that better results could be obtained and had been obtained in some of the runs for which no data were at hand, these were not published at the time, and the work was continued by the present authors.

The apparatus used was the same as that described by Frary and Badger (loc. cit.), and consisted essentially of a large crucible of Acheson graphite having a water-cooled bottom and serving as anode, and a water-cooled iron cathode a little over an inch in diameter. The crucible was enclosed in a protective layer of refractory material, and especial pains were taken with the contact between it and the positive cable. The cathode could be raised or lowered at will by means of a screw mechanism.

From preliminary experiments it was decided that the occasional feeding of the fresh electrolyte and irregular raising of the cathode were two important causes of low efficiency.

Intermittent addition of electrolyte caused irregularities in the height of the melt and its temperature, especially the latter. Too great cooling of the electrolyte caused the deposition of the metal in a spongy form, in which state it was readily lost. Spongy portions were also formed when the stick of metal extended too far insufficient to melt the metal at the point of contact with below the surface of the melt, the current density being the electrolyte, and thus produce a solid stick. (In most

cases a rod of about 1 cm. diameter was obtained, so the

cathode area was about 1 to 1 sq. cm.). This spongy metal was very likely to be dislodged and swept away by the vigorous convection currents which were always present; it also gave trouble by growing toward one side or other of the crucible, causing an increase of current at that point, with consequent increased growth of metal and a higher temperature, until finally the calcium on the edges would melt, and, being connected with the main stick only by thin pieces of metal, would be easily swept away by the electrolyte. When the end of the stick of metal was kept just below the surface of the electrolyte, the spongy metal at first formed melted almost at once, and its surface tension drew it into a compact globule which was almost immediately frozen by the cooling effect of the rest of the stick and the water-cooled cathode. Thus a To get the best results it solid stick was easily obtained. was found necessary for one person to devote all his attention to the raising of the cathode and the addition of the electrolyte, making both operations practically continuous.

The chloride used was some of Merck's C.P. granulated, which had previously been used in desiccators. It was dried in an iron crucible over a four-post Bunsen burner, and preserved in a stoppered bottle.

In starting up a run the cold fused chloride remaining in the crucible from a previous run was heated until dry by directing the flame of a blast-lamp upon it. Then a small portion of it was fused by drawing an arc from the side of

the crucible with the aid of a graphite rod. A small current was used until contact was made with the cathode, when more current was put on and the graphite rod removed. As soon as good contact was assured, the direct current was shut off, and an alternating e.m.f. of about 30 volts applied. The current usually rose rapidly to about 200 ampères, being regulated by raising or lowering the cathode, and the upper part of the crucible was soon filled with molten chloride. A sufficient supply of electrolyte was added at this time to fill up the crucible, and when all was ready, the alternating current was thrown off and the direct current thrown on. Readings of the time, voltage, and amperage were taken at once. The voltmeter and ammeter were read every five minutes during the run, and the average of these readings considered to be the average voltage and amperage for the run. The current varied very little during each run, as it was taken from the 110-volt lighting circuit through a constant resistance. There was no trouble with the socalled "anode effect," except when the electrolysis was started before a sufficient amount of the electrolyte had been melted down by the alternating current.

(NOTE.-A sample of the electrolyte was taken at such a time, when the "anode effect" had been more than usually troublesome, and the per cent of silica present determined. Only o'056 per cent was found, confirming the results obtained in the previous paper, and indicating that current density rather than impurities should be blamed for this "effect").

The formation of the metal under the cathode was watched very carefully, great care being necessary to get a good start. We aimed to withdraw the cathode and metal as rapidly as it was possible to do so without striking an arc. It will be noticed in the table of results that practice in doing this increased the yield obtained. When the electrolysis is going on properly, and the stick of metal is at the right depth, the electrolyte seems to flow rather rapidly across the surface in a fixed direction, and is hottest at the cathode. The operator must be guided by the appearance of the electrolyte at the cathode. There should be a rosette-like spot with radial markings here; if the stick of metal is pulled out too fast, this spot becomes almost white hot, and if the metal is not at once lowered, an arc forms, and part of the metal is melted off and lost. If the stick is not raised fast enough, the spot becomes less noticeable, and spongy metal deposits below the sur face, causing losses. When the electrolysis was well under way, new electrolyte was added in small portions, taking care to make these additions as continuous as possible, and in such a spot that the convection currents carried the cold chloride away from the metal. shows the efficiency obtained.

Since Ruff and Plato (Ber., 1902, xxxv., 3612; 1903, xxxvi., 491; D R.P., 153,171; Chem. Centr., 1904, ii., 802), Muthmann (Zeit. Flektrochem., 1904, x., 508), and Woehler (Zeit. Elektrochem., 1905, xi., 612) have recommended the use of a mixture of the chloride and fluoride as the electrolyte, we decided to try it and see if it offered any advantages. The fluorspar was first partially purified by treatment with concentrated hydrochloric acid to remove iron. It was mixed with the chloride in the proportion of 16.5 parts to 100, as recommended by Ruff and Plato. A new crucible was made for use with this electrolyte, having the same size as the other, but a different arrangement of the water-cooling. The bottom of the crucible was turned out on the lathe so as to fit loosely over the hollow brass cooling cylinder of the Borchers furnace for the electrolysis of fused salts ("Die Elektrische Oefen," 1907 ed., p. 35). The crucible was then turned upside down, this space filled with a strong copper sulphate solution, and copper deposited until the cylinder would not enter the space. The copper was now turned out on the lathe to fit the cylinder, and the whole of the outside of the crucible and cylinder protected from the air by a coat of Portland cement and carborundum fire-sand, held in place, as before, by a tin form. The contact being made through the water-cooling cylinder, was kept cool and did not burn away, as the other crucible did in the course of time. A more complete protection of the outside of the crucible was also possible.

With this crucible and the above-mentioned electrolyte a new series of runs was made. Considerable trouble was found in making smooth solid sticks of metal with this electrolyte. The particular advantage of this mixture is supposed to be its low melting-point (660°). This appears to us to be a disadvantage, as the bath must be worked at a temperature of nearly 750° in order to get a solid stick of metal, and the bath is so fluid at the working temperature that convection currents are more violent and more likely to sweep away the metal from the end of the stick, forming an arc. We had a great deal of trouble from this cause. Reduction of the current did not bring the desired result, as the efficiency decreased markedly (Run No. 6, Table II.). The results obtained are shown in Table II.

Table I.

85.5

78.4 77.8

The sticks of metal were cleaned by hammering off the crust on the outside, and then placing them in absolute alcohol for twenty-four hours to remove the rest of the chloride. When taken out of the alcohol, the sticks were always clean, though covered by a thin coat of oxide. They were dried by burning off the alcohol, and weighed. The current was, of course, not absolutely constant, so there is a possible error of 1 or 2 per cent in the efficiency as calculated, but the results are sufficiently accurate for practical purposes.

During the runs with this electrolyte only the chloride was added, it being assumed that the fluoride would not be decomposed. However, at the close of these experiments a sample from the top of the melt showed only 2.21 per cent of insoluble matter, so, evidently, much of the fluoride had been lost. This loss was probably largely mechanical and due to the violent spattering of the electrolyte; it was noticed that a great deal more of the fine dust from the electrolyte collected on the crucible and its surroundings with this than with the simple chloride electrolyte.

We also took samples from the electrolyte in the lower part of the crucible, which had been kept frozen during the electrolysis by the cooling coil, and so, presumably, retained its original composition, and attempted to isolate the fluochloride of calcium described by Poulenc (Ann. Chim. Phys., 1894, [7], ii., 5) and Defacqs (Ann. Chim. Phys., 1904, 18], i., 337; Comptes Rendus, 1903, cxxxvii., 1251; Journ. Chem. Soc., 1904, lxxxvi., [2], 123), but found none of it.