18 NILLS AND DONALD ON THE IV.-On the Action of Oxides on Salts. .Part IV.* Potassic Chlwate and Ferric Oxide. By EDMUND J. MILLS D.Sc. F.R.S. and GEORGE DONALD. 33. THE work accomplished in previous parts naturally led us to con-sider the decomposition of potassic chlorate by oxides a reaction which during the 50 years since Dohereinert first ohserved it has been and still continnes a problem in theoretical chemistry. There are several oxides which are known to facilitate the evolution * For Part I11 see Journal 1881 1 533. + Ann. Pharm. 1 236. Dobereiner thus states the problem :-" Welche Rolle spielt nun dcr Braunstein in diesem Processe ? wirkt er blos als guter Wfirmeleiter oder als Electromotor ? oder ist es endlich noch ein Minimum yon inharirendem Wasser welches die totale Zerzetzung des Salzes bedingt und die Bildung des oxy-chlorsauren Kslis verhindert ? BCTION OF OXIDES ON SALTS.19 of oxygen from potassic clilorate and to prevent perhaps entirely the formation of perchlorate. Of these manganic dioxide is t'he most strik-ing example. One of us acc )rdingly in association with Mr. John Stevenson M.A. made iiumerous experiments with this substance ; but trhe investigation was eventually abdndoned partly because we were unable to prepare* a srable anhydrous dioxide and partly on account of the irregulai.ity and comparative violence of the reaction. We shall again refer to tbese results in a supplementary note. 34. The chlorate employed in our experiments consisted of a fair commercial sample which was purified by repeated crystal lisation and filtration.J t was powdered sified dried a t loo" and kept in dry air. The ferric oxide was prepared lrorn a sample of piire ferrous sulphate by oxidation with hydric ni hate precipitation hy ammonia complete washing solution in hydric chloride and reprecipitation with ammonia. Washing was again proceeded with until the washings were free from chloride or sulphate. The hydrate was then dried over the water-bath and ignited irl successive small quantities in a porcelain crucible. I n order to preclude thP gas flame which surrounded the crucible from attacking the oxide the crucible was cemented into a circular apertme cut centrally i n a plate of iron about 2 feet square. All the ignitions were as far as possible carefully executed in the same manner.The residual preparations of oxide were all mixed together powdered and sifted thr0uF.h fine muslin. Renewed absorp-tion of water was prevented by heating to loo" alid preservation in B desiccator. 35. The apparatus which was used for the reaction consisted of a shallow iron pot inside which a low stage of porous tile was con-structed ; upon this stage lay the horse-shoe-shaped horizontal bulb of a Schlosing's rqydator and within the bulb was placed the porcelain crucible containing the mixture of oxide and chlorate. The crucible was covered with a lid ;md the bulb of a mercurial thermometer was placed very near to it. The iron cover of the pot protected the whole arrangement. As a source of heat we employed a Fletcher's burner, and the entire a,pparatus was as far as could conveniently be managed, screened on all sides from draught.36. The actual course of an experiment will siiggest itself to the reader. All we need remark is (1) that the air-bath was heated for half-an-hour before each operation; (2) that the insertion of the crucible was an almost momentary act. Thus the time consumed in heating up the mixture of chlorate and oxide must hare bezn very short ; and the disturbance introduced into our work by this condition -especially when it is remembered that the duration of heating was always four hours-we regard as inappreciable. At the end of our * Comp. Picbering Chem. fieus 43 226. c 20 MILLS AXD DONALD ON THE experiment the crucible was instantly removed placed in a desiccator, and weighed the next morning.37. Our trial series of experiments was performed with pure sub-stances prepared by Mr. Pratt (Part II) in 1878. The constant weight of potassic chlorate was 5 grams ; the weight of oxide ranged from 0.5 to 3.0 grams and the meantemperature" was 189.8". Under these conditions the loss of oxygen amounted to 0.0083 gram to 0.0311 gram. We found however on repeating some of our work, that we could not obtain constant results neither were these of suffi-cient magnitude to serve as a basis for accurate determinations of a " factor of chemical effect." This factor had values extending from 0.11200 to 0.35742 and amounted on the average to 0.20750. It was evidently necessary t o obtain a larger evolution of oxygen. We therefore (now taking substances of our own preparation and the same constant weight of chlorate) raised the temperature to about 19.5".39. I n the course of our calculations we have had occasion to use the following numbers viz. :-KC103 = 122.47 Fe,O = 159.92 0 = 15.96. Actual weights appropriately divided by these figures, map be considered as expressed in " chemical units." The temperatures reported below are very good means of quarter-hourly observations. If the action of ferric oxide upon potassic chlorate be similar to that of an ordinary oxide on an ordinary salt the numerical results should admit of represefltation under some form of the general equation-In this E is the chemical effect on oxygen expelled a and y are respectively the masses of oxide and chlorate ; xr yo are the residues of these masses after action ; and a is a " factor of chemical effect," depending upon the particular conditions of chemical change.In our calculations we have taken a = zr since the oxide remains constant in each experiment ; and chemical units have been employed throughout. The necessary details will be found in the following Table -* All our temperatures are corrected for zero error and exposure. For the expo-sitre correction the ordinary formula and the factor 0.00013 were employed Experiment . -. I I1 111 IV v ,, PI VII VIII IX x XI XI1 XI11 XIV xv * . . Oxide taken . grams . 0 - l G 0.15 0 *20 0 *25 0 '50 0 -75 1 -00 2 -00 3 -00 5 -00 8.00 8 '00 8 *oo 10 *00 10 .00 .-ACTION OF OXIDES ON SALTS . TABLE XI . X . -.-0 *00062532 0 *00093797 0 *0012506 0 4015632 0 * 0031 266 0 *0046898 0 *0068532 0 *0125064 0 *0187596 0 -0312660 0 *050026 0 *OS'i532 9 ) 3 ) Oxygen expelled . gram . 0 -0213 0 -0312 0 *0230 0.0833 0 *0435 0 *0591 0 -0629 0 '0993 0 -1246 U -1563 0 *1900 0 *1103 0.1158 0 -1320 0 * 1228 E . 0 '0013346 0.0019559 0 9014411 0 *0014599 0.0027250 0 * 0037230 0 *0039411 0.0062218 0 *0078070 0 -0097932 0 *011915 0 *0069110 0 '0072556 0 -0082707 0 -0076942 a . 2 *1437 2 *0998 1 '1769 0 *95862 0.91900 0.65643 0 -70656 0 *6246S 0 *68086 0 '52806 0 50688 0 -29964 0 -31416 0 -32591 0 *30377 21 .!emp . C . 195* 6 O 195 * 1 . 195 *1 194 *7 194 *7 195 -6 194.9 194 -7 194 *7 194 -7 195 * 1 195 '2 194 *7 195 '1 194 *7 39 . I t is evident from the above results that the chemical effect of the oxide increases rapidly a t first. and afterwards a t a diminishing rate ; and the numbers suggest that the values of a are inversely pro-portional to the ralues of o . In order to test this hypothesis. we took all the values of a excepting IX and X (the mean of the value8 a t 8 grams* and 10 grams oxide respectively being assumed. instead of the single quantities there). and calculated t. he following equation :-( a - 0.27420) (Z + 0*00095942) = 0.0029056. A comparison between theory and experiment is given in the Table below -TABLE XI1 .Experiment . I I1 . . . . . . . . . . . . . . I11 IV v VI VI I VIII IX x x r-XI 1 I XIV. xv a calculated . 210770 1.80560 1.58890 1.42600 0.98532 0.78854 0.67706 0.48998 0.364367 0.331191 0-319965 0.42 1 5 *5 a found . 2.14370 2.09980 1.1 7690 0.95862 0.9 1900 0.85643 0-70656 0.62468 0.58086 0.52806 0.31484 0.37356 * I n Experiment X it is probable that the temperature accidentally rose WIIC t. he thermometer was not under observation. but as we are not certain of t. Ilia explanation we have retained the result 23 MILLS Ah9 DONALD Ulu’ THE Probable error of a single comparison 0.15215 ; of 12 comparisons, -043921. Sum 02 the errors 0.0231 6:;. The hypothesis therefore, with regard to a may be accepted as fairly correct’.40. Some of the relations between the acting quantities are worthy of special consideiqation. The factor a of chemical effect is the number of chemical units of oxygen expelled per unit of oxide. Now since-it follows that when o is very small a = 3.3028. Hence within the Limits of experimental error FeLO expels 0.:.3 or 3Fe20 expels Ole, when the weight of oxide is very small ; and a unit of the oxide then acts on rather more than a unit of the dhlorste. On the other hand, when o is very large a = 0.27240. Heuce 0.2 7240ay E = _____ x + y r ’ an expression which on the understanding stated reduces to-E = 0*27240 y = 27240 x 0.040826 = 0.01112 unit = 0.1775 gram. Assuming then 1 he qiicznt,i ty of oxide to be indefinitely great the 5 grams of xchlorate would not under the conditions of our experiments, have lost more than 0.1775 gram oxygen.This amount has not in fact been exceeded except in Experiment IX which we believe to be of an accidental character. Another inference which may be drawn from our work is very obvious. As the mass of oxide increases so does its efficiency decrease; it stands so to speak in ihs own way. Large quantities of oxide present little real advantage over medium quantities such for instance, as an equal weight. There can be no doubt if we compare our first with our second series of experiments that bhe value of 01 rises with the temperature. The constant X seems to depend on the physical condition of the oxide. 41. In Part ZI (23) of these researches i t was shown that the action of ferric oxide on -5 grams potassic carbonate leads to the expulsion of carbonic dioxide in three stages and that in the first of these (ie., from 2.4 grams oxide downwards) the factor of chemical effect is inversely proportional to the mass of the oxide.Thus the entire course of the action of ferric oxide upon potassic chlorate is strictly analogous to the first stage of the action of the same oxide on potassi ACTIOS OF OXlDES OX SALTS. 23 carbonate ; and both actions are particular ca8ses of the general effect of an oxide on a salt represented by the equation-any :c'r j- yr* E = -It is obvious then that the case of chemical change which me have had under consideration presents nothing abnormal or peculiar in its features.From the carbonate an oxide of carbon is the matter expelled ; from the chlorate an oxide of oxygen. The law of action is the same in both instances. The name catalysis therefore which was applied by Berzelius to this case of chemical change ceases to have any reason for its existence. SUPPLEMENTARY NOTE. Jlanganic Dioxide and Potassic Chlomte. By EDXUSD J. MILLS D.Sc. F.E.S. and JOHN STEVENSOF M.A. 42. As stated in (33) Ke performed a number of experiments with manganic dioxide. The particular mode of preparation which we eventually selected was that of Beilstein and Jawein (Ber. 1879 p. IS%) which leads t o a body really consisting of manganic dioxide in a hydrated condition. Our hydrate contained 331 per cent. of water. With this preparation we uniformly obtained a very violelit action at 210'.As previous trials with this and another sample of hydrous dioxide had convinced us that evolution of oxygen begins at about 160", we selected an intermediate temperature (180°) the time being four hours and the weight of chlorate 5 grams. With quantit,ies of oxide less than 0.05 gram we obtained only very irregular results ; hut with that quantir;y and higher f act0 ry . [MnO. = 86-92.] -weights the values of z were fairly satis-TABLE XIII. Anhydrous oxide. 0 '0483 0 -0967 0.4835 0 -9669 1.9338 4 *8345 9 . a 9 0 X . -1 2 10 20 40 100 200 The above calculation Oxygen expelled. 0 *4361 0 *6087 0'7731 0.7643 0 '7798 0.7416 0 '6-188 -u found.38 838 24.567 6.451 3 937 2.517 1 .ti57 1 *239 a calculated. 39 '1'75 24 '034 6.486 3 %9 2 -386 1.528 1 '239 has been simplified b j taking proportiona 24 O’SULLIVAN x - AND P-AMYLAN. values for x. The ‘( anhydrous oxide” is calculated from a determina-tion of water lost at 280-300° and the ‘‘ oxygen expelled” by deduct-ing the loss of water by the hydrate at 180” from t’he total loss. The equation is-(Z - 0.94800) (X + 0.32577) = 58.310. Probable error of a single comparison of theory with experiment, 0.18123 ; of seven comparisons 0*068499. 43. The relation which we have indicated between a and x rests upon a certain supposition. The percentage of water present in the hydrate was 3.31 a t the beginning and 0.144 a t the close of the reaction ; and this amount might very possibly have been sufficient, in accordance with Dobereiner’s snggestion to furnish an incentive for the powerful effects in the 0.0483-9.6690 gram experiments. On the other hand it might have produced no such effect but may have had influence merely in retarding the evolution of oxygen in the <O.O4SS gram experiment,s as seems to have been the case. What we can my with considerable confidence is that in either event the effect upon the form of the relation 01 to x of the presence of this quantity of water in the oxide is not traceable ; the action of manganic dioxide upon potassic chlorate resembles the ordinary behaviour of an oxide towards a salt