27M MACDONALD AND HINSHELWOOD TBE FORMATION AND CCCLXXX.-The Pormation and Growth of Silver Nuclei in the Decomposition of Silver Oxalute. By JAMES YOUNGER MACDONAD and CYRIL ~ToRM-~N HINSHELWOOD. TRIS paper contains first a description of a sensitive method for making direct measurements of the instantaneous rate of reaction in a chemical change where gas is evolved and secondly an account of its application in the investigation of certain interesting phenomena relating to the thermal decomposition of silver oxalate. The decomposition of solid substances frequently off ers the appearance of being autocatalytic the rate of change increasing as the reaction progresses. This is in many instances due simply to an increase in the surface of the solid during the course of the trans-formation (Phil.Nag. 1920 40 569) but it can also be due to the circumstance that the reaction is facilitated by the presence of nuclei of one of the products. To whichever cause the acceleration is due the relation between the velocity of change and the amount of reaction which has already taken place is a very complicated one. The determination of this relation from measurements of the total change 2 which has taken place a t any time t depends upon the drawing of tangents to curves of x and t a procedure involving a large proportional error. The following method was therefore adopted for obtaining direct readings of the actual velocity a t different times during the reaction. The solid is allowed to decompose in a small bulb kept a t constant temperature in a vapour bath The bulb is connected with GROWTH OF SILVER NUCLEI IN THE DECOMPOSITION ETC.2765 M c M gauge and the gaseous products of the reaction are removed by a continuously running Gaede pump. The experiments must be made at temperatures where the rate of evolution of gas is not too great so that the pump is able to mainbin a fairly high vacuum in the apparatus. At intervals measurements of the rate of reaction are made by turning off the tap leading to the pump and raising the mercury in the M c M gauge for a pressure reading. At the moment when the mercury cuts off the bulb of the gauge from communication with the decomposition vessel a stop-watch is started. The pressure reading is completed and the mercury lowered com-munication with the decomposition vessel being once more estab-lished.At the end of a suitable time a second reading of the McLeod gauge is made the watch being stopped at the moment when the mercury once more cuts off the gauge from the rest of the apparatus. In the actual experiments pressure increases of the order of 0.1 or 0.01 ram. were measured in periods of time varying from 45 seconds to 15 minutes in reactions lasting several hours or days. The readings therefore can be regarded as instantaneous. By the use of gauges of various degrees of sensitiveness in different experiments it is possible either to follow the reaction over its whole course or to make a more detailed study of the initial stages. No investigation has previously been made of the mechanism of the decomposition of silver oxalafe.It is usually assumed that the change takes place in accordance with the simple equation C204Ag,= 2Ag + 2C0,. This was confirmed by an analysis of the residue which was more than 99.8% silver and of the gas, which was pure carbon dioxide containing no trace of carbon monoxide. If the reaction involved the intermediate formation of carbonate or oxide carbon monoxide would necessarily appear in the gaseous products. Various specimens of silver oxalate behaved very differently, falling roughly into two classes. One class decomposed with the most marked acceleration the maximum velocity being about 200 times greater than the initial velocity. The other class showed very much less acceleration and the velocity instead of passing through a sharp maximum remained steady over a large part of the whole course.Specimens of this second class were on the whole about ten times as stable as those of the first class. The difference w a ~ traced to one factor in the method of preparation. All the speci-mens were prepared by precipitation of silver nitrate with sodium oxdate. Those precipitated under such conditions that the sodium oxalate was in excess throughout belonged to the unstable accelerat-ing type whilat those precipitated with silver nitrate always in excess belonged to the stable feebly accelerating type 2766 MACDONALD AND HMSHELWOOD THE FOR.MA!LTON AND The method of preparation was carefully standardised. The precipitant which was not to be in excess was run slowly as a k e continuous stream in the course of about 4 hour into a solution of the other salt which was vigorously agitated by a mechanical stirrer.All the solutions were very dilute-an obvious precaution in view of Richards' well-known investigation of the occlusion difEiculties of Stas-and the precipitation was carried out in black-ened vessels at the ordinary temperature. The silver oxalate was repeatedly washed by decantation and dried in a vacuum desiccator kept in a dark cupboard. As far as we are able to judge all factors were exactly the same in the preparation of the two types of silver oxalate except that one or other precipitant was in excess. The autocatalytic acceleration is not due to carbon dioxide since it takes place in a vacuum. It is not due simply to increase in surface as with potassium pemanganate and some other solids (Eoc.cit.) since it cannot be eliminated by preliminary grinding. It must be due therefore t o nuclei of solid silver formed in the decomposition. Since the initial velocity is very small compared with that finally attained the whole course of the reaction is evidently governed by the formation and growth of these nuclei. The decomposition of solids is essentially a surface phenomenon. More-over the devitrification of glass which depends upon the formation of crystal nuclei is known to start from the surface since the wash-ing of the surface with dilute hydrofluoric acid removes the tendency to further devitrification on heating. It is very probable that in a similar way the silver nuclei are formed at the surface of the oxalate crystals.The formation of nuclei and their subsequent growth are independent phenomena both of which are extremely sensitive to the presence of foreign substances (compare Zsigmondy " Kolloid-chemie," 1922 p. 144). When silver oxalate is precipitated with excess of silver nitrate and sodium oxalate respectively there will probably be Werent ions adsorbed. The different behaviour of the oxalate prepared by these two methods is to be attributed to the effect of the adsorbed ions on the chance of formation of silver nuclei in the solid and on their subsequent growth. As the silver oxalate was always very thoroughly washed during preparation only the definitely adsorbed ions need be taken into account. The decomposition bulb was protected from light by a covering of tinfoil.About fifty experiments were carried out with twelve specimens of oxalate. Some of the results are in the following tables and curves; t is the time in minutes and v the velocity of reaction in arbitrary units GROWTH OF SILVER NUCLEI IN THE DECOMPOSITION ETC. 2767 Temperature 131.8'. (a) Silver oxdata precipitated with excess of d u m oxalate. ( b ) Silver oxalate precipitated with excess of silver nitrate. t . U* t. t'. 5 5.3 39.5 756 12 60 43 628 18 147.5 49 434 21 216 53 340 25 384 60 192 28.5 725 75 154 30.5 958 96 46 32.5 850 140 28.5 35-5 888 263 13-3 This specimen waa precipit,ated with a very large excess of oxalate, and is one of the most unstable. t . U. t. U. 3.5 7 240 48 10 13 270 50 33 32 305 48 43 28 335 31 47 33 360 24 60 31 420 18 105 50 540 13 150 43 600 6.3 180 43 The rise and fall in velocity ie quite real and presumably connected with the fact that the oxdate grains are not quite uniform.Using a more sensitive gauge it is possible to see in more detail the initial variations in rate. The fDllowing results were obtained with a rather more stable specimen of oxalate. Up to 100 minutm, the decomposition amounts to about 1% only. Temperature 131.6". t. V. t. 2'. t . U. t. U. 5 3.4 32.5 45.8 77.5 224 135 477 12-5 3.9 37.5 62.8 86-5 255 149 616 17.5 5.8 47.5 107-2 97-5 310 165 649 22.5 15.6 58.5 152 110.5 374 198.5 680 27.5 26.8 67-5 183 121 407 204-5 618 In this experiment the maximum rate of neitrly 3000 wit8 reached in about 6 hours.It is worth while to give in full the results of one more experiment, made at 110" with a specimen of the accelerating type. Attention may be directed to the long period of steady velocity following the sharp fall from the maximum. t. 5 15 25 40 60 85 135 170 245 313 438 533 0. 1.09 1.22 1.28 1.79 2.70 4-72 12.0 18-6 35.7 50-0 69-4 90-8 1. 759 832 881 936 1,126 1,197 1,331 1,400 1,607 1,737 2,293 2,310 V. 182 214-5 224.5 251 298 276 228 189.5 115 98.0 78.0 82-9 t. 2,405 2,575 2,813 3,127 3,858 4,303 5,180 6,012 6,708 8,873 9,617 13,920 19,700 U. 83.2 83.5 78.7 72.9 51.4 45.9 33.7 25-7 19-3 16-7 14.1 8-7 5.6 The relation between the velocity and the amount decomposed G.N. Lewis investigating the some- is not by any means simple 2768 MACDONALD AND HINSHELWOOD THE FORMATION AND what similar decomposition of silver oxide (2. physikd. Chem., 1905 52 310) thought that the results could be expressed by the simple equation for homogeneous autocatalysis dz/& = kx(u - z). It is difficult to see how the complex process of nucleus formation and jgowth can be so represented. The applicability of the equation to the silver oxalate decomposition can be tested in two ways. (a) &/dt is plotted against t and by square counting a curve of x and t is constructed. It is then readily seen by plotting log [d~/dt/(a - x)j against log x whether the rate per unit amount FIG.I. Influence of the inetlwd of precipitation on thc decomposition of silver ox&&. 20 40 Time in minuterr. Curvea I and II silver oxalate precipitated in preeence of wee88 of aodiurn oxaktd. of oxalate is a linear function of x. It is not. The rate depends upon a power of x which not only varies but is always less than unity. (71) For the very early stages of the reaction as measured with a more sensitive gauge the equation reduces to dxJdt = kx, but to account for the fact that the reaction starts at all this has in any case to be written dxldt = E (z + xo) where xo is a small constant quantity which determines the actual initial rate. From this it follows that log dz/& = h! + constant. This gives a con-venient means of plotting the results without resorting to the laborious process of square counting.The following table contains a set of typical results the figures referring to the initial stages of ~ u r v e s III and I V excerr8 of silver nitrate GROWTH OF SILVER NUCLEI IN THE DECOMPOSITION ETC. 2769 one of the experiments already quoted (second table). dx/& = v. When t = 5 log TJ = 0.53. loge - 0.53 log w - 0.53 t - 5. log w - 0.53. t - 6 t - 5. logv - 0.53. t - 5 12.5 0-23 0.0184 53.5 1-652 0.0309 17.5 0-663 0.0379 62.5 1-713 0-0274 22.5 0.898 0-0399 72.5 1.820 0.025 1 27.5 1.131 0.0412 81-5 1.877 0.0230 32-5 1.268 0.0390 92.5 1.961 0-0212 42-5 1.500 0.0353 105-5 2-043 0.0194 Iqluence of Adsorbed Gases on the Silver NucEei.4ome experi-ments were also made in which the silver oxalate was allowed to decompose in a small bulb connected with a gas burette the rate of reaction being measured simply by the total evolution of carbon dioxide.The results were most unexpected. Sometimes there was no trace of acceleration even with specimens prepared by precipit-ation with excess of sodium oxalate. At other times the acceleration made its appearance after varying intervals of time in an uncon-trollable manner. This was all traced to the infiuence of the air initially present in tjhe bulb. Further investigation showed that oxygen has a pronounced poisoning influence on the reaction. This is evidently due to its being adsorbed on the nuclei of silver aa soon as they are formed on the surface. Their growth is thus hindered *or stopped.Experiments carried out in an atmosphere of carbon dioxide followed almost exactly the same come as those carried out in a vacuum and the influence of nitrogen appears to be but slight. Fig. 2 shows clearly the pronounced poisoning influence of air. The explanation is now obvious of the puzzling results obtained when silver oxalate is allowed to decompose in a bulb initially fUed with air which is gradually replaced by carbon dioxide; there is a sudden acceleration of the reaction when the air is all displaced from the surface of the crystals. In order t o study the reaction under conditions where the poisoning effect of air waa constant, experiments were made in a stream of air the rate of reaction being measured periodically by cutting off the stream and allowing the pressure to rise.The stream was sufKciently slow for there to be no appreciable coohg effect as the air passed through a consider-able length of heated tubing before actually coming into contact with the silver oxalate. In a continuous stream of air or even in the presence of still air the reaction is spread out over a time about ten times as long as in a vacuum or in presence of carbon dioxide. The poisoning effect is relatively more marked with the unstable accelerating specimens precipitated with excess of sodium oxdate than with those precipitated in presence of an excess of silve 2770 THE FOBNATION AND GBOWTH OF SILVER NUCLEI ETC. nitrate since in the latter the acceleration is already to some extent inhibited. The changes in colour which accompany the decomposition a m interesting the white oxalate passing through pale to deep brown and then becoming black.When the reaction is nearly complete some sort of recrystallisation of the silver seems to occur and the whole mass becomes almost white. Addition of the h a 1 product does not cause an increase in the rate of reaction. Evidently coamr particles of silver are not effective in catalping the change in the same way as the minute nuclei formed in the space lattice of n u . 2. Influence of air on the decomposition of silver oxukste. 100 80 20 0 60 120 180 240 Time in minutea. the oxalate crystals themselves. Experiments were made on the influence of previous grinding. This produced some increase in the rate of reaction but did not eliminate the autocatalysis aa with potassium permanganate and some other solids.The influence of temperature was investigated with a few of the specimens. It is normal and more or less uniform throughout all stages of the decomposition. Over the range 100" to 150" the rate increases approximately 2.7 times for 10". Summary. The thermal decomposition of silver oxalate takes place in accord-ance with the simple equation (CO*OAg)2 = 2Ag + 2C0, PURVIS THE INFLTJENCE OF DIFFERENT NUCLEI mc 2771 Its rate is governed entirely by the formation and growth of nuclei of silver in the space lattice of the oxdate crystah. These processes are sensitive to the presence of adsorbed ions, since silver oxalate precipitated in presence of excess of sodium oxalate behaves quite Merently from that precipitated in presence of excess of silver nitrate. This is in some respects analogous to the influence of adsorbed ions on the photo-sensitivity of silver bromide described by Fajans and Frankenburger (2. EEektrockm. 1922 28, 499). In a vacuum and in an atmosphere of carbon dioxide the rate of reaction is the same but the presence of oxygen has a very pro-nounced poisoning effect on the silver nuclei. This is an example of catalytic poisoning of a new kind. The influence of adsorbed ions and of oxygen seems to show that the nuclei must be first formed at the surface of the oxalate crystals rather than at any point in the interior. PHYSICAL CHEmsmy LABORATORY, BALLIOL COLLEGE ~LND TRWITP COLLEGE, OXFORD. [Received July 4th 1925.