COMBUSTION ANALYSIS.*

By JAMES WALKER, F.R.S., and THOMAS BLACKADDER,

B.Sc., University College, Dundee.

THE process of Dennstedt for the elementary analysis of organic compounds by combustion in oxygen with the help of platinised quartz was tried in this laboratory, and in expert hands was found to be both rapid and accurate. The average student, however, experienced great difficulty in the conduct of the combustion, and it occurred to us that the advantages of the apparatus of Dennstedt, so carefully worked out by him in detail, might be applied to the ordinary method of combustion by means of copper oxide.

If one inspects a tube in which a copper oxide com. bustion is being conducted, it is found that the oxide actually reduced to metallic copper after the combustion of the volatile matter is completed, rarely extends for more than an inch or two along the tube, unless the process has been accidentally "rushed." It seemed, therefore, possible to reduce the dimensions of the combustion-tube to such an extent as to secure the advantage of the Dennstedt furnace, which from its lightness of construction admits of rapid heating and cooling. This we discovered to be the case, and in reality there is no greater

5

while still hot to a tube A, Fig. 2, which is then closed tube to protect the copper oxide from atmospheric moisture. with a stopper, through which passes a calcium chloride If the substance to be analysed is a solid, it is weighed off in a small stoppered bottle B, Fig. 2, the neck of which fits into the constriction of A, so that the substance may be mixed in B with a quantity of copper oxide from A, with as little exposure to the atmosphere as possible. The end of the combustion-tube fits into the wider part of A, so that it also may be conveniently charged with copper oxide from the ignited supply. After the combustion - tube has received its charge of copper oxide, the mixture of substance and copper oxide is transferred from B into the combustion-tube, the neck of the stoppered bottle being of such a diameter as to fit inside the end of the latter. The bottle is then "washed out" once or twice with copper oxide, received as before from A, and emptied into the combustion-tube. The absorption-tubes and oxygen supply are next attached and the combustion begun. The two burners at the front (absorption end) of the tube are lit and tiles are placed over them, the heat being as far as possible confined to the portion of the tube covered by the tiles by means of screens of asbestos paper. When the copper oxide has attained dull redness, the third Bunsen is lit at the other extremity of the tube without a tile. The heat from this Bunsen gradually volatilises the substance,

difficulty in performing a copper oxide combustion in a shortened. Dennstedt furnace, heated by three Bunsen burners, than in a furnace of the customary type heated by thirty. The saving of initial outlay on the furnace, and on the current consumption of gas is, of course, comparatively great, but a still more considerable advantage is that the combustion may be done on the worker's bench without inconvenience to himself or to his neighbours.

The furnace employed by us, together with the burners, absorption-tubes, and the purifying apparatus for the supply of oxygen, are practically all as described by Dennstedt, and shown to scale in Fig. 1. The chief modification is that the furnace is cut down to 60 cm. in length. The combustion-tube is of Jena glass, 66 cm. long, and need not be more than 8 mm. internal diameter. The total volume of such a tube is only about 30 cc., and the charge of copper oxide, including spirals, weighs only 35 grms. Two of the burners are supplied with attachments for spreading the flame into a flat sheet; the third is used for local heating, but towards the end of the combustion, when the whole tube is heated, it also may be provided with a flame-spreader.

The copper oxide employed is coarsely powdered and sifted free from fine dust, which during the combustion might clog and stop the tube. The combustion is carried out in a current of oxygen, and the calcium chloride and soda lime absorption-tubes are always weighed filled with the same gas.

The method adopted for mixing the substance with copper oxide, and transferring it to the combustion-tube, is a slight modification of that used by Professor Thiele, of Munich. The copper oxide after ignition is transferred

* From the Proceedings of the Royal Society of Edinburgh, xxviii., Part ix., No. 44. We are indebted to the Royal Society of Edinburgh for the blocks illustrating this Paper.

A

B

FIG. 2.

with or without decomposition, the volatile products being for the most part burned in the moderately rapid stream of oxygen (two to three bubbles per second) which is all the time passing through the apparatus. The Bunsen is gradually moved forward as the combustion proceeds, tiles being placed behind it to keep the tube still hot. Under ordinary conditions there is no visible reduction of copper oxide to metallic copper, although towards the end of the combustion the oxide usually glows immediately behind the tiles at the absorption end of the tube. To burn any carbon that may be left on the tube by decomposition, all the tiles are placed in position, and the three burners adjusted so that the tube is heated as uniformly as possible to dull redness. The carbon at this temperature is not graphitised and burns off readily. When the oxygen comes freely through the indicator bottle at the end of the apparatus, the combustion is finished, and very little sweeping out is necessary, owing to the small volume of the apparatus. Immediately after the combustion, the copper oxide is returned to the tube A, and is ready for the next analysis.

The time occupied between attaching and removing the absorption-tubes need not exceed half an hour. The

0'1890 0'1733

Percentage carbon. Percentage hydrogen.

40.61

40'44

40'48

40.56

5:28

5'20

5:18

5'20

(The theoretical percentages for C4H6O4 are carbon= 40.68, hydrogen 5'09).

If the substance contains nitrogen, a metallic copper spiral 7 cm. long is, as usual, placed at the front of the tube, and the current of oxygen is passed at a slower rate at the beginning of the combustion. For acetanilide the following percentages were obtained: - Carbon, 7103, 7124; hydrogen, 672, 6.89, the numbers for the formula C&H NO being 71'11 and 6.67 respectively.

When the substance is a liquid it is weighed off in a small bulb with a sealed capillary, the tip of the capillary being broken off before its introduction into the tube. The best results are got when the capillary is comparatively wide and about 8 cm. long. The open end of the capillary, surrounded by the copper oxide, faces the stream of oxygen. The liquid, when the copper oxide in the front of the tube has reached dull redness, is slowly distilled out of the bulb into the copper oxide at the cool end of the tube by means of a small Bunsen flame applied directly to the upper side of the tube over the bulb. The combustion is then continued as for a solid.

When a volatile liquid, such as benzene, is burned, it is introduced into the tube in a small sealed bulb terminating in a capillary at either end. The longer capillary (8 cm.) is plugged with fusible metal (compare Hempel, "Gas Analysis," p. 341), and the shorter (3 cm.) is sealed off in the flame after the bulb has received its charge of benzene.

In this case it is advisable not to surround the bulb and capillaries with copper oxide, but to leave the whole clear at the back of the tube. The little plug of fusible metal at the end facing the current of oxygen is melted by the application of the flame above the combustion-tube, and the benzene is gradually vaporised by very gentle heating. The following result was obtained by this method :Carbon = 91 94 per cent; hydrogen 7.83 per cent. Calculated for C6H6-Carbon = 92:31 per cent; hydrogen 7.69 per cent.

A combustion-tube of the same dimensions may be used for estimating nitrogen by the direct method. In this case a current of carbon dioxide is substituted for the current of oxygen, the nitrogen being collected over concentrated potash solution. The reduced copper spiral necessary in this case need not exceed 7 cm. in length. The percentage of nitrogen in aniline found by this method was in two experiments 15:09 and 15:25, the percentage required for the formula C6H,N being 15.05. Equally satisfactory results were obtained with acetanilide and hippuric acid.

To show that the method gives satisfactory results in the hands of a beginner, the following consecutive analyses may be quoted from the notebook of a student who had no previous experience of organic combustions :

Succinic acid. Found I.

Carbon.

Hydrogen.

It is apparent, then, that the method as here described is well adapted to ordinary laboratory work, and compares very favourably with the copper oxide method as usually employed in respect of economy in apparatus, gas, and time, and of the greatly increased comfort of the operator.

ON THE DETERMINATION OF HALOGENS IN
ORGANIC COMPOUNDS.*
By C. W. BACON.

1. Introductory.

HALOGENS are at present determined in organic compounds almost exclusively by the method of Carius. While generally reliable, the method is quite laborious. First, it requires some hours of actual work, including as it does a gravimetric determination of silver halide. Besides, eight or nine hours heating with fuming nitric acid involves a delay in getting the final result, which is at times extremely inconvenient.

About two years ago Stepanoff (Ber., 1906, xxxix., 405) published a method for the quantitative determination of halogens in organic compounds based on the reducing He dissolves a weighed action of nascent hydrogen. amount of the halogen compound in 20 to 40 cubic centinected with a reflux condenser, places the flask on a watermetres of 98 per cent alcohol in an Erlenmeyer flask conbath, and adds gradually through the condenser twenty-five times the amount of sodium corresponding to the reaction R.Hlg+ C2H5.OH+2Na Na. Hlg+C2H5.ONa + R.H. When the reaction is over he adds 20 to 40 cubic centimetres of water, distils off the alcohol, acidifies strongly with nitric acid, and determines the halogen in the resulting solution by Volhard's method.

=

Stepanoff's method was tried in these Laboratories, exactly as described by the author, in connection with a study of the esterification of some refractory aromatic acids containing halogens. The results, however, were entirely unsatisfactory. The figures yielded by consecutive analyses of one and the same substance differed greatly from one another, and were many per cent removed from the truth.

At the request of Professor Rosanoff I then undertook to re-investigate the subject with a view to ascertaining under what conditions, if at all, nascent hydrogen can really be relied upon to reduce organic halogens quantitatively. The analytical conditions and procedure recommended below are the result of a large number of painstaking trials. If the simple directions are followed, success may be confidently expected in every case. A single determination requires about two hours; but if three or four determinations are carried on simultaneously a considerable number of analyses can be conveniently performed in the course of a working day.

2. Analytical Directions.

Introduce about o 2 grm. of the halogen compound into a dry pear-shaped (Kjeldahl) flask.

If w is the number of grms. of the compound taken, add 156 cubic centimetres of alcohol (at least 98 per cent) if the compound contains chlorine, or 68w cubic centimetres if the compound contains bromine, or 44w cubic centimetres it the compound contains iodine. Connect the flask with a reflux condenser, clamp it over a square of wire gauze covered with a thin sheet of asbestos, and warm with a Bunsen burner until the substance is dissolved. Introduce very gradually, through the condenser, a total of 19.5 grms. of sodium if the compound contains chlorine, or 8.5 grms. of sodium if the compound contains bromine, or 5'5 grms. of sodium if the compound contains iodine. (These quantities represent about fifteen times the amount

* Contributions from the Chemical Laboratories of Clark University.

CHEMICAL NEWS,} Jan. 1, 1909

7

Taking into account the fact that the four substances used are among the most stable organic halogen compounds, these results may be considered as sufficient proof of the general applicability and accuracy of the method. The investigation was carried out under the guidance of the Director of these Laboratories, Professor M. A. Rosanoff, whom the writer wishes to thank again for his friendly interest and assistance in the work. Acknowledgment is also due to Dr. W. L. Prager for his assistance in the early part of the work. Clark University, Worcester, Mass., June, 1908.

A few additional remarks may not be surperfluous. The statement that the liquid should be boiled for an hour after the solution of the sodium is complete may seem surprising; for sodium ethylate is believed by some to have no action upon halogen compounds of the aromatic series. While the additional boiling is probably unnecessary in the case of aliphatic compounds, the following experiment will show that the operation is not useless in the case of refractory aromatic compounds. An alcoholic solution of hexachloro-benzene was boiled for one hour with the amount of sodium ethylate required by the above directions. The solution, diluted with water and acidified, was found to contain no less than 25 per cent of the amount of sodium chloride corresponding to the hexachloro-benzene used. While in similar experiments I was unable to transform monobromo-benzene into phenetol, there is no doubt but that the alcoholates of sodium attack halogens in a benzene ring quite vigorous'y. In fact, if the action were not somewhat too slow for analytical purposes, it might be employed, independently of any other method, for the quantitative estimation of halogens in organic compounds.

Special experiments showed that distilling off the alcohol (as Stepanoff does) before titrating with sulphocyanate is unnecessary. In the solution copiously diluted with water the alcohol has no effect upon the end-point.

Finally, Kahlbaum's sodium was found to be sufficiently pure for use in connection with the method here proposed. 3. Some Test Results.

The following results were obtained by the method proposed :

1. A sample of 1-2-4-6-trichlorobenzoic acid prepared in this laboratory was found to contain 47 29 per cent of chlorine. The amount corresponding to the formula C7H3O2C13 is 47 19 per cent. 2. Two samples of ethyl 1-2-4-6-tribromobenzoate, containing theoretically 61.99 per cent of bromine, gave 61.32 per cent and 61.25 per cent respectively. The substance was probably not quite dry, but the agreement of the two results is satisfactory.

3. A sample of pure benzene hexachloride was found to contain 73 23 per cent of chlorine. The theoretical percentage for C6H6C16 is 73°15.

4. Ten consecutive analyses of carefully dried hexachlorobenzene (Kahlbaum's), with theoretically 74 71 per cent chlorine, gave the following results :Percentage of chlorine found.

Error.

METHOD TO THE ESTIMATION OF POTASSIUM IN SOILS.

By W. A. DRUSHEL.

In a previous paper from the Kent Chemical Laboratory of Yale University it was shown that potassium may be estimated with a fair degree of accuracy by precipitating it as potassium sodium cobalti-nitrite in a solution acidified with acetic acid and oxidising the precipitate with standard potassium permanganate (Am. Journ. Sci., 1907, xxiv., 433; CHEMICAL NEWS, xcvii., 124). In the same paper the applicability of the method to the estimation of potassium in commercial fertilisers was shown by a series of experiments.

In the method as previously worked out an excess of concentrated sodium cobalti-nitrite solution acidified with acetic acid is added to a neutral solution of a potassium salt, and the mixture is evaporated to a pasty condition on the steam-bath. After cooling, the residue is stirred up with sufficient cold water to dissolve the excess of sodium cobalti-nitrite. The precipitate, consisting of K2NaCo(NO2)6. H2O, is filtered on a rather close asbestos felt in a perforated crucible and well washed with cold water, or preferably with a half saturated sodium chloride solution. The precipitate and felt are transferred to an excess of standard N10 or N/5 potassium permanganate which has been diluted to about ten times its volume and heated nearly to boiling. If particles of the precipitate stick persistently to the walls of the crucible and cannot be removed with a spray of water, the crucible is put into the permanganate solution. After stirring for a few minutes the solution is gradually acidified with 5 cc. to 20 cc. of dilute sulphuric acid, and the oxidation is allowed to go to completion, a process which seldom requires more If no particles of the yellow precipitate than five minutes. settle out on standing a minute, the oxidation may be considered complete. The hot solution is then bleached by running in a measured amount of standard oxalic acid, containing 50 cc. of concentrated sulphuric acid per litre. The solution after bleaching is titrated to colour with standard permanganate in the usual manner.

In this process the cobalt in the molecule is reduced from the trivalent to the bivalent condition and not re-oxidised, consequently from the molecule of the potassium sodium cobalti nitrite we find o 000857 grm. K2O équivalent to 1 cc. of strictly N/10 potassium permanganate. This factor, of course, must be corrected for any variation in the normality of the permanganate solution used.

For the extraction of the alkalis 10 grms. of dry soil are placed in an Erlenmeyer flask with 25 cc. to 35 cc. of about 20 per cent hydrochloric acid. The flask is thoroughly shaken and a small funnel is hung in its neck to avoid too great a loss of acid by evaporation. The contents of the flask are digested on the steam-bath for twenty-four hours.

From this point several methods for the final preparation of the sample were tried with satisfactory results, given in the table. Since duplicate estimations were to be made by the gravimetric chlorplatinate method, for which it was

necessary to remove the iron, aluminium, calcium, magnesium, phosphoric acid, and ammonium salts, if present, from the soil extract, the following general procedure was found to be most expeditious. The soil extract was filtered through paper into an evaporating dish and the residue was washed with boiling water until the filtrate gave no reaction for chlorine with silver nitrate. The filtrate was evaporated almost to dryness to remove the excess of hydrochloric acid as far as possible. The residue was dissolved in about 200 cc. of water, and after heating to boiling, a little ammonium hydroxide and ammonium oxalate were added. The mixture was boiled a minute, settled, filtered, and the precipitate was washed with hot water until a drop of the filtrate gave no chlorine reaction. The filtrate was concentrated, transferred to a 200 cc. flask, cooled, and made up to the mark. After thoroughly shaking, 50 cc. aliquots were pipetted off for the gravimetric and volumetric estimations. The aliquots were evaporated to dryness in platinum dishes, and gently ignited to remove the ammonium chloride. After cooling, the residue was moistened with dilute sulphuric acid and again ignited, gently at first and finally at the full heat of the Bunsen flame, to remove the last trace of ammonium present as the sulphate, and to destroy any organic matter which night be present.

The residue for the gravimetric estimation was dissolved in a little water and a few drops of hydrochloric acid over the steam-bath, and the estimation of the potassium was made according to the modified Lindo-Gladding method.

To dissolve the residues for the volumetric estimations a little water and a few drops of acetic acid instead of hydrochloric acid were used. In the volumetric work approximately N/5 potassium permanganate was used, 26'08 cc. of permanganate being equivalent to 50 cc. of exactly N/io oxalic acid. From this ratio the factor for K2O was found to be 0.001642. In each case the potassium was precipitated as the cobalti-nitrite by evaporating off with io cc. of sodium cobalti-nitrite prepared according to the method of Adie and Wood (Journ. Chem. Soc., lxxvii., 1076).

were removed in the separate aliquots by ammonium hydroxide and ammonium oxalate. In VII. the aliquots were made directly from the hydrochloric acid extract. That of VII. (2) was treated in the usual manner for the gravimetric estimation of potassium. The other aliquots of VII. were evaporated to dryness and gently ignited to remove any ammonium chloride present and to char the organic matter. The residues were extracted with hot water and a little acetic acid, filtered, and evaporated with sodium cobalti-nitrite in the usual manner.

In this work the results are based on a small amount of soil (2-5 grms.) in each case because but a limited amount of each sample was available. A higher degree of accuracy may be secured by using 10 grms. of soil for each estimation instead of 2.5 gims.

Summary.

A weighed amount of dry soil is extracted with an excess of hydrochloric acid over the steam-bath. The excess of acid is removed from the extract by evaporation. The bases which might interfere with the process are removed with sodium carbonate or ammonium hydroxide and ammonium oxalate. Ammonium salts and organic matter are removed by ignition. The residue is dissolved in a little water and a few drops of acetic acid, and the mixture evaporated with an excess of sodium cobalti-nitrite to a pasty condition, stirred up with cold water, and filtered upon asbestos in a perforated crucible. The precipitated potassium sodium cobalti-nitrite is washed with halfsaturated solution of chloride, and treated with an excess of permanganate in hot dilute solution. The colour of the permanganate is destroyed by an excess of standard acidulated oxalic acid, and the excess of oxalic acid titrated to colour with permanganate.-American Journal of Science,

xxvi., 329.

THE CORROSION OF IRON.*

By ALLERTON S. CUSHMAN, Assistant Director, Office of Public Roads, Dept. of Agriculture, U.S.A,

Introduction.

IN 1905 the Department of Agriculture published Farmers' Bulletin No. 239 on the subject of corrosion of fence wire. A large number of complaints had reached the Department from farmers in regard to the rapid corrosion of steel of modern manufacture used for various purposes, and especially for wire fencing. As the same difficulty has been constantly met in the use of corrugated iron and steel culverts for road drainage, it was decided to begin an investigation of the entire subject. The intention has been not only to get at the facts in the case, but also to aid as much as possible in bringing about an improvement in the rust-resisting quality of iron and steel used for the special purposes mentioned above.

Coincidently with the beginning of the investigation inaugurated by the Office of Public Roads much attention began to be given to the same subject by a number of prominent metallurgists and chemists, both in this country and abroad. The American Society of Mining Engineers, at their Washington meeting, held in May, 1905, engaged in a discussion of the relative corrosion of iron and steel, and a vigorous debate followed the presentation of several papers on the same subject at the meeting of the American Society for Testing Materials, held at Atlantic City in June, 1906. The subject was considered so important that the latter Society appointed a Special Committee of ten members to make an investigation of it.

(NOTE.-Prof. W. H. Walker, of the Massachusetts Institute of Technology, a member of this Committee, has for some time past been making an independent investigation of the cause of the corrosion of iron. The writer

*Bulletin No. 30, U.S. Department of Agriculture, Office of Public Roads.