69 VII.-On the Decomposition of Silver Fulmina,te by Hydrochloric Acid. By EDWARD DIVERS, M.D., and MICHITADA KAWAKITA, M.E. THE Society has receired from us a note of the fact that silver fulmi- nate differs from mercury fulminate in yielding much less than the full amount of hydroxyammoninm chloride, and in yielding ammo- nium chloride. The present communication contains the results of our further examination of the action of hydrochloric acid on silver fulminate, and also those of the examination of the action of dilute hydrochloric acid on mercury fulminate, and on fulminurates, in relation to the production of hydroxyammonium chloride, formic acid, and ammonia, and of Steiner's production of oxalic acid from mercui-y fulminate. The silver fulminate was prepared in small quantities at a time, and was dried, with certain precautions, in an oven at or near 100".I n two preparations we determined the silver as chloride, and found 71.79 and 71.76 per cent., in place of 72.00 as calculated. Unlike the mercury salt, silver fulminate is energetically attacked by concentrated hydrochloric acid. As mentioned in our paper on mercury fulminate, it is hardly necessary to use Steiner's precau- tion of working with that salt only when moist; and although the reaction between silver fulminate and the acid is of a kind sugges- tive of danger, we have used it in the dry state without accident. The concentrated acid, as i t comes in contact with the bulky salt', causes i t to shrivel up and decompose with a loud hissing noise and development of heat. If the heating is not checked, the silver chlo- ride produced is a t first stained orange, but rapidly loses this colour, imparting it to the acid mother-liquor. The heating may be pre- vented or reduced, both by taking only a small quantity of the fulmi- nate at a time, and by dropping it into a considerable excess of the acid artificially cooled.The colour of the solution disappears during subsequent evaporation. Resides the colouring matter, there is a t first among the products of the reaction an unstable, colourless substance, which gives an intense wine-red colour with ferric chloride, both in the strongest acid solution or in one t,hat has been neutralised. Long standing in the cold or a few minutes' hating, deprives the solution of the power of changing in colour with ferric chloride.This substance is also formed from mercury fulminate, and, as we have already pointed out in our preliminary note, is probably identical with that met with by Gay- Lussnc and Liebig in their examination of silver fulminate, and70 DIVERS AND KAWAKITA: DECOMPOSITION O F SILVER described by them as a chlorinated acid containing nitrogen, and not precipitable by silver nitrate. Thi8 acid, however, is stated by them to have given the red colour only when the solution had been previously neutralised with alkali. In all cases when silver fulminate is treated with hydrochloric acid, the odour of hydrogen cyanide is perceptible, but more markedly when dilnte acid is taken. Two determinations were made of the amount of hydrogen cyanide produced when concentrated hydro- chloric acid is used ; in one experiment this was found t o be 0.29, and in another 0.38 per cent.The determinations were made with cai*e, but the results are to be taken as only approximately accurate, and as serving to show the smallness of the quantities of hydrogen cyanide produced. The procedure was first to let hydrochloric acid through a tap-funnel iiito a retort containing the dry, weighed fulminate, and connected air-t'ight with two U-tubes in series containing solution of potassium hydroxide, and, next, slowly to aspirate air through the apparatus for some time, the retort being kept warm. The contents of the receivers were then treated with enough silver nitrate to cause a precipitate, and were filtered. Lastly, the silver in solution in the alkaline liquid was determined and taken, in calculating the hydrogen cyanide, as having existed in the solution as silver potassium cyanide.Liebig's vo1umeti.i~ method could not be employed, because of the presence in the solution along with the cyanide of a minute quantity of some substance sharply reducing silver to the metallic state. The process of estimating the hydrogen cyanide indicated, there- fore, the possibility of the simnltaneous generation of a very small quantity of a highly volatile and oxidisable matter. Besides showing this reducing action on silver nitrate, the alkaline solution acquired in some experiments a slight brown colonr during the aspiration. Our endeavours to isolate or identify any such substance, 01% to get uniform indications, were, however? unsuccessful , and we believe now that traces of the contents of the retort must have been carried over mechanically, although unnoticed, by being first thrown up into the head of the retort by the violent action between the acid and the fulminate when they met, and then washed down into the receiver by condensing vapours.The hissing noise of the decomposition of silver fulminate by hydrochloric acid is connected with a not inconsiderable escape of carbonic anhydride. We have collected this escaping gas by decom- posing the fulminate with acid in a tube exhausted of air by the Sprengel pump, and have obtained in this way 4.4 per cent., which is the equivalent of & of the total carbon. This quantity is largely in excess of what can possibly be liberated when the fulminate is dropped suddenly into abundance of acid, and its production is due toFULMINATE BY HYDROCHLORIC ACID.71 the fact that, when working in a vacuum, the acid has to be added to the fulminate, and then meets it in the form of fine streams or jets- a condition of things allowing of great local beating and thus favour- i n g the generation of carbonic anhydride. Working in a retort, filled with air as usual, and adding the whole of the acid to be used suddenly to the fulminate, we have collected the escaping carbonic anhydride in barium hydroxide solution, and thus obtained evidently much less than when working in a vacuum, although as yet we have made no exact measurement of the quantity formed in this way.During the evaporation of the acid solution, a little carbonic anhydride continues to escape from it with the vapour, even after the precaution has been taken to expose it for some time in au open vessel on a water-bath before distilling. In the decomposition of silver fulminate by hydrochloric acid, there is liberated no gas but carbonic anhydride and hydrogen cyanide, neither nitrogen, nitrous oxide, nitric oxide, nor carbon monoxide, The yield of hydroxyammonium chloride depends on the strength of the hydrochloric acid used. In decomposing the fulminate, care was taken to miriimise the spontaneous heating-up of the mixture. The fulminate was added to moderate excess of the fuming acid in some cases, and in others to large excess. The product was warmed, diluted, well shaken to collect together the silver chloride, and filtered.The hydroxyammonium chloride was titrated with iodine solution. The quantities obtained in separate experiments were thus found to be:- 29.8, 29.1, 32.1, 29.8, 28.9, 30.4 37.4, and 31.6 per cent. These results show considerable uniformity, wit,h some exceptions, and represent very nearly two-thirds of the hydroxyammonium chloride equivalent to the whole of the nitrogen, that is, 46.33 per cent., two-thirds of which is 30.9. We have no reason to doubt the accuracy of any of these numbers, and attribute the irregularities observable to variations in the proportions of hydrochloric acid taken, although we are not able now to establish this to have been the case in these particular analyses.But, as we shall show that we can make the yield vary to some extent by varying the conditions of the ex- periment, we do not consider that any special significance is to be attached to the fact that about two-thirds of the nitrogen appears as hydroxyammonium chloride when concentrated acid is used. Uni- form procedure has given us uniform results, and by breaking through this we can make the results vary, and can increase the yield of this salt. For the estimation of formic acid, distillation was necessary before titration could be made, because of the presence of ammonium chlo-72 DIVERS AND KAWAKITA ON THE DECOMPOSITION OF ride. The distillation was finished in a current of air, and the heat so managed as t o leave the hydroxyammonium chloride as far as practicable undecomposed." The formic acid was measured by de- ducting the quantity of alkali equivalent to the hydrochloric acid found by titration with silver nitrate from the alkali required by the total acid, and calculating the difference as formic acid.In the ex- periment which gave us the fifth of the enumerated quantities of hydroxyammonium chloride, that is, 28.9 per cent., we obtained 19.8 per cent. of formic acid. The full quantity of formic acid is 30.67 per cent., and two-thirds of this is 20.44, so that we obtained almost two-thirds of the total formic acid possible, just as we obtained nearly two-thirds of the hydroxyammoninm chloride. We have attempted to estimate the formic acid? by weighing the * We have fewer determinations of formic acid than of hydroxyammonium chlo- ride, because of several failures in measuring it.Some of these were due apparently to the passing over of spray into the receiver in consequence of distilling in small retorts too full, to which practice we were tempted by the difficulty, when the retort was large, of getting over all the formic acid without overheating the hydroxy- ammonium chloride. This source of error was afterwards avoided by the obvious expedient of distilling with the retort not so full. Other failures were experienced through the presence of a very large excess of hydrochloric acid along with the formic acid, auch an excess having been taken in order to keep down by its mass the temperature of the reaction as much as possible, and lessen the production of hydro- gen cyanide.Sodium phosphate added to the contents of the retort, to fix some of the hydrocliloric acid, rendered the distillation difficult and unsatisfactory. After- wards, by taking a much smaller proportion of hydrochloric acid, this cause of failure was removed. It is, besides, dificult to see in what way silver fulminate could be decomposed in which the formic acid would not be equivalent to the hydroxyammonium chloride, unless the liberation of much hydrogen cyanide took place. f I n our paper on n.ercury fulminate we described an experiment in which, by titration of the formic acid from 2.6665 grams of mercury fulminate, and working on thirds, we determined its amount to be 3 per cent. less than the calculated quan- tity--31.3 instead of 32.4-surely a result having no great pretension to close accu- racy, considering that our mercury and hydroxyaminonium chloride determinations were in almost perfee: accordance with theory.As mercury fulminate has been known since the beginning of this century, as it is 60 years since it was examined by Liebig, and as no analysis of this salt had been published in all that time, except the partial one by Liebig, in which he got 56.9 per cent. mercury instead of 70.4, we thought it of interest to use our results for the purpose of showing for the first time the composition of mercury fulminate by analysis. This having been done by us, Dr. Schotten, in abstracting our paper for the Berichte der deutsch. chem. #es., has endeavoured to raise doubts in the minds of his readers as to the accuracy of our results.There was and is no better method of determination open to us; a finding by titration of only 97 per cent. of the calculated amount of formic acid was cer- tainly no refiiiement of accuracy beyond the capability of the method ; and a devia- tion of even double that extent from the truth would still have left our result suffi- cient to prove all we attempted to make it do. In the text of this paper, we gim After all, sufficient determinations for the purpose are given.SILVER FULXINATE BY HYDROCHLORIC ACID. 73 calomel produced by it from mercuric chloride, but not with success. H. Rose lias proved that in the presence of alkali chlorides, the pre- cipitation of mercurous chloride is incomplete, but this fact was no obstacle to us.We had before us a mixture of formic and hydro- chloric acids, and this we proceeded to neutralise with washed pre- cipitated mercuric oxide. To test the method, we operated on lead formate and found if work quite successfully. The lead salt was treated with enough hydrochloric acid to decompose it, and mercuric oxide was then added in excess. Digestion on a water-bath, aided by frequent stirring, and then a second digestion with dilute hydrochloric acid, gave the theoretical quantity of mercurous chloride. In one experiment we got 30.80, in another 30.83 per cent. of formic acid, whilst pure lea,d formate should yield 30.97 per cent. The method has the great recommendation of depending on the weighing a compound more than ten times heavier than the substance estimated.But un- fortunately, as we have said, we found that it would not serve our purpose. The large proportion of hydrochloric acid present proved a fatal obstacle. This acid formed an oxychloride during the long digestion necessary to insure the decomposition of all the formic acid, which it was generally impossible to dissolve out thoroughly from the mercurous chloride by fresh hydrochloric acid, without at the same time deconiposing some of it into metal and mercuric chloride. After many trials we gave up our attempt to use this method. Whatever inay be the yellow or orange matter which is formed when the silver fulminate is allowed to grow hot by its reaction with hydrochloric acid, the following experiments serve to show t h a t it is formed at the expense of that portion of the fulminate which would otherwise become hydroxylamine and formic acid in contact with con- centrated hydrochloric acid.That it is not formed from the other third of the fulminate, or a t least from that third alone, is also to be seen in the fact that, with cooling, only a trace of it is produced, although no portion of this third is gained as formic acid and hydroxy- ammonium chloride. I n two separate portions, in order to moderate the rise in tempera- ture, 2.6330 grams were drenched with fuming hjdrochloric acid, without cooling. The resulting solution, which had a distiiict yellow colour, gave, by direct titration with iodine, hydroxyammonium chloride 27.65 per cent., and by titration with alkali, after distillation, other determinations of formic acid by this method, the degree of accuracy of which may safely be measured approximately by the hydroxyammonium determinations, and as we have suppressed nothing, we offer these to show what the method is capable of.We should add that in all proba,bilit.y the hydroxgammonium is not an exact measure of the formic acid, but measures this and the very small quantity of hydrogen cyanide together.74 DIVERS AND KAWAKITA ON TIE DECOMPOSITION OF formic acid 18.67 per cent. These quantities are about one-twelfth less than two-thirds of the whole. Next, in order to lead to a still greater rise in temperature, 1.9830 grams were treated in one lot in the same way. Much heat was developed, and a t first the silver chloride was very strongly coloured, but soon gave up its colouring matter to the solution. The hydroxyammonium chloride was in this case found to be only 21.55 per cent., or less than half the f u l l quantity.The change in the results obtained by using dilute hydrochloric acid instead of the concentrated, is considerable, and is important as serving t o tone down the contrast between mercury fulminate and silver fulminate. Although the use of the acid in the dilute state seems to favour the production of hydrogen cyanide, both hydroxy- ammonium chloride and formic acid are increased in quantity much above two-thirds. A small quantity of silver fulminate, 0.3175 gram, being put uuder water, concentrated hydrochloric acid solution was added to the extent of about half the volume. The hydroxFammonium chloride obtained in this case was 43.52 per cent., which is not very far short of the full quantity, 46-33 per cent., whilst the yield with undiluted acid was only 30 per cent., as already fully recorded.A modification of the preceding experiment was made by using hot liquids. 0.36975 gram of fulminate was placed in hot water, and hot, diluted hydrochloric acid added. Iu this case, 41.60 per cent. of hydroxyammonium chloride was produced, a lower yield, and illus- trating the injurious effect of heat. Once more, 4.28225 grams of fulminate, in two quantities, were treated with cold water and hydrochloric acid, and the resulting solu- tion, after the usual heating on the water-bath, divided into portions in which both hydroxyammonium chloride and formic acid were deter- mined.We found :- Per cent. 42.15 evaporation ................................ 42.17 with iodine.. ................................ 42.88 28.35 Hydroxyammonium chloride, by direct titration .... 3 7 ,, weighed as residue on Hydroxyammonium chloride, by tit'ration of residue Formic acid, by titration with alkali.. ............ From these results, it appears that about nine-tenths of the silver fulminate had become formic acid and hy droxyammoniu m chloride. When silver fulminate has been treated with concentrated hydro- chloric acid, we always find ammonium chloride present in quantity among the products. I n this respect, silver fulminate is in marked contrast with the mercury salts. But when dilute acid is used, thisSILVER FULMINATE BY HYDROCHLORIC ACID.7 5 difference between the salts is no longer observable. Along with the increasing yield of hydroxyammonium chloride and formic acid that attends dilution of the hydrochloric acid, there goes a lessening of the quantity of ammonium chloride down even to nothing. The results last given show that in the experiment with dilute acid, no ammonium chloride was formed, or practically none, for the difference 0.3 between 42.17 and 41.88 may safely be set down to the presence of water and hydrochloric acid as impurities. Qualitative testing gave only doubtful evidence of any ammonia being present. I n contrast with the result in which dilute hydrochloric acid was employed, may be presented results which were obtained with con- centrated hydrochloric acid :- Hydroxyammonium chloride, by direct Mixed chlorides as residue on evapora- Hydroxyammonium chloride, by titra- I.11. 111. IV. titration. ....................... 29.80 29.08 32.13 30.42 tion .......................... 34.90 35.03 37.39 35.37 tion of residue .................. 27.60 - 30.07 - Ammonium chloride, by difference .... 7.3 - 7.3 - Although among these experiments T I and I V were not followed out, the two numbers given enable the others to be inferred. Ammonia was tested for qualitatively, after destruction of the hgdroxyam- moriium chloride with iodine, and found abundantly. Ammonium chloride 7.3 per cent. IS equivalent to more than a fifth of the total nitrogen of the fulminate. Ammonia, so far, having been determined only by difference in weight between the evaporation-residue and the hydroxyammonium chloride measured by titration, an attempt was made to estimate i t directly.Some silver fulminate having been treated with con- centrated acid in the usual way, a portion of the solution was treated with copper sulphate and potassiiim hydroxide in a closed flask, in order to destroy hydroxylamine. The resulting alkaline solution and precipitate were then distilled for ammonia. B u t the process proved unsatisfactory, inasmilch as free ebullition was impracticable in con- sequence of the mixture bumping; and therefore only part of the liquid could safely be distilled over, so that some ammonia must have been retained by the copper hydroxide. We succeeded, however, in getting in this way three-fourths of the quantity of the ammonia indicated by the indirect method.Instead of making further trials to determine the ammonia directly, we made two chlorine determinations in the evaporated residue of chlorides, as well as hydroxyammonium chloride determi-76 ON THE DECOMPOSITION OF SILVER FULMINATE. nations by iodine titration. We found in one case, as difference between total residue and hydroxyammonium chloride, ammonium chloride 4.92 per cent. of the fulminate, and, as the equivalent of the difference between the total chlorine found and that calculated from ihe hydroxyammonium chloride, 4.33 per cent. ammonium chloride. I n the other case we found, as difference by weight, 7-65, and as difference by the chlorine method, 7.20 per cent. ammonium chloride. These determinations of the chlorine served not only to estimate the ammonia, but also to confirm the view that the oxidisable part of the residue was hydroxyammonium chloride and nothing else. It was in the experiments just detailed that we also determined the hydrogen cyanide, and as the only nitrogenous products we have observed have been hydroxylamine, ammonia, and hydrogen cyanide, the quantities obtained of these three substances should contain all the nitrogen.This they do not quite do, as the following statement shows :- I. 11. Nitrogen as hydroxyammonium.. .. 7-34 6-36 97 ammonia 1.13 1.88 7, hydrogen cyanide 0.15 0.20 9 7 unaccounted f o r . . 0.51 0.89 7 7 silver fulminate 9.33 9.33 ............ .... .... -- - ...... It is tolerably certain that some ammonia escapes, if not duying the evaporation, at least during the drying of the residue in the dish.Menxyy fulnzinnte, when treated with dilute hydrochloric acid, yields no ammonia. By using one measure of concentrated acid to three measures of water, mercury fulminate can be thoroughly decomposed a t the heat of a water-bath. The yield of hydroxy- ammonium chloride is somewhat diminished, and that, of hydrogen cyanide increased. Hydroxyammonium chloride, 43.51 per cent., was thus obtained, the calculated quantlity being 48.94 per cent. There no longer exists, therefore, any difference between the two salts as fulminates, what difference they show being caused by the metallic radicals. The same hydrolysis occurs with both fulminates ; but in the case of the silver salt, in consequence probably of the sharp separation of the silver chloride in the solid state, the decom- position generates heat SO rapidly that the formic acid and hydroxy- ammonium chloride change into carbonic acid and ammonium chloride :- CzAg2N202 + 4HC1 + 40H2 = 2CH302 + 2NH40C1 + 2AgC1 and CH2Oz + NH40C1 = CH203 + NH,Cl.DIVERS ON THE COX3TITUTION OF FULMINATES.77 Hydrochloric Acid aml Fulviinurates. We have only made a special examination of the behaviour of hydrochloric acid with fulminurates, to see whether these salts were capable of generating hydroxyammonium chloride and formic acid. As already known, fulminurates yield some ammonium chloride when heated with aqueous hydrochloric acid. We have experimented with potassium fulminurate and silver f u lminurate, and have obtained neither hydroxyammonium chloride nor formic acid. Hydrochloric acid does not seem to have much action on fulminurates. Non-productiosi, of Oxalic Acid from Fzhainates. We have already published an account of our failure t o find even a trace of oxalic acid among the products of decomposition of mercury fulminate by sulphuric acid, or by aqueous hydrogen sulphide. But as it is stated that Steiner got oxalic acid by decomposing mercury fulminate with hydrogen sulphide, using ether instead of water as a menstruum, and even to have obtained it in crystals (see Watts’s Dictioi~n~y), we have followed his process, using as little ether as practicable, so as to avoid as far as possible the accidental presence of water. We have repeated the experiment, and on evaporating the ethereal solution, after saturating it with hydrogen sulphide, as directed, could find no trace of oxalic acid in the residue, and for our own part we believe t,hat oxalic acid cannot be produced from fulmi- nates by any method yet published.