2896 HINSHELWOOD AND BURK : CCCCI1.-The Relation of Homogeneous to Cutulysed The Catalytic Decomposition of Hydro- Reactions. gen Iodide on the Xurface of Platinum. By CYRIL NORMAN HINSHELWOOD and ROBERT EMMETT BURB. THE results of several previous investigations (this vol. pp. 327, 1105 1552; Proc. Rmj. Soc. 1925 A 108 211) havc rcndere probable the conclusion that simple gaseous decompositions wh.i~h as homogeneous reactions are bimolecular become llnimolecuh when they take place in contact with the surface of a solid catalyst. Thus for example the thermal decomposition of nitrous oxide, 2N,O = ZN2 + O, aa an uncatalysed reaction depends upon the collision of two molecules but the reaction which fakes place upon the surface of gold appears to be simply N,O = N + 0 followed by the combination of atomic oxygen fo the molecular form.The decompositions of nitrous oxide on the surface of platinum and of ammonia on the surface of platinum have ako been shown to be nnimoleculaz. It is not always possible to decide from measurements of reaction rate whether a catalytic reaction is unimolecular or bimoleculm. If the active surface of the catalyst is completely covered with adsorbed molecules-'saturated'-then the reaction rate is inde-pendent of the pressure of the gas. The reaction appears to be of 'zero order.' When on the other hand the adsorption is small, the chance of a group of n molecules occupying positions on the surface near enough for interaction to be possible is proportional to the nth power of the total number adsorbed and this in turn to the nth power of the pressure of the gas.The ' order ' of the reaction obtained from kinetic measurements gives under these conditions the number of molecules actually participating in the decomposition. An intermediate condition is possible when the adsorption is neither very large nor quite small where a birnole-cular reaction might simulate a unimolecular reaction over a small range of pressure. But if the reaction order appeared to be unity by a coincidence of this kind it would vary very markedly both with pressure and with temperature as may easily be seen. This state of affairs would readily be detected and the results discounted, so that in general we may say that unless the reaction order is zero it gives the number of molecules participating in the change.The decomposition of ammonia on tungsten and of hydrogen iodide on gold are both approximately of order zero; consequently we cannot conclude how many molecules are involved. The homogeneous decomposition of hydrogen iodide is one of the best known bimolecular reactions. We were anxious therefore to find a catalyst at the surface of which the true order of the hetero-geneous reaction could be found. Platinum fulfils these con-ditions. The reaction takes place in the simple unimolecular manner HI = H +I followed by the combination of atomic hydrogen and iodine to form molecules. This illustrates once more the f undameiital importance in hetero 2898 HINSHELWOOD AND B U X : geneow crtfalysis of the rtfEinity which metal surfaces possess for free atoms.Method of Ezperimnt.-The hydrogen iodide ww prepared by the action of phosphoric acid on potassium iodide and purified by fractional distillation from liquid air the middle fraction being collected in evacuated blackened glass holders. The decomposition was allowed to take place at the surface of a heated platinum wire and measured exactly 8.5 described for the corresponding experiments with a gold wire (this vol. p. 1552). The reaction vessel was kept in ice. Since the iodine condensed, the reaction 2HI = H + I is attended by a decrease in pressure which allows the rate to be measured by simple manometric means. That the action of the platinum wire was 'catalytic,' and that the wire w&s not attacked by the iodine was shown by the fact that the theoretical change in pressure was observed after complete decomposition and by the fact that the resistance of the wire remained absolutely unchanged during the whole series of experiments.Its catalytic activity moreover remained steady. The mercury in the manometer was protected from the action of the hydrogen iodide by a buffer of hydrogen in the capillary tube leading to the decomposition bulb. In$uence of the Pressure of the Hydrogen Iodi&.-Three typical experiments showing the course of the reaction a t different tem-peratures will first be recorded t is the time in seconds x the percentage change and k the unimolecular velocity coefficient. 670". 200 Mm. HI. 100 Mm. H,. 100 22 24.9 200 41 26.4 300 56 27-4 400 64 25.6 500 '72 25-5 600 76 24.0 t .x. k x 104. 563O. 200 Mm. HI. 100 Mm. H,. t. x. k x lo4. 120 12 10.6 300 27 10.3 600 47 10.6 900 61 10.5 1200 71 10.4 1600 78 10.1 439". 200 Mm. €€I. 100 Mm. H,. t . x. k x lo' 600 13 2-32 1200 26 2-50 1800 38 2.65 2400 46 2-49 3300 55 2.42 4200 63.5 2.40 6000 74.5 2.28 The influence of the pressure of the hydrogen iodide is shown 100 Mm. of hydrogen were by the following values found a t 563". present in each experiment. Press. of HI (mm.) ......... 100 200 300 k ................................. 0.001 18 0.00 105 0.00095 The values of k are sufficiently independent of pressure to ahow that the reaction is unimolecular. The slight dropping off was traced to the circumstance that hydrogen has a small but defmite retarding effect on the reaction.When the initial pressure o hydrogen iodide is great the pressure of hydrogen a t every stage of the reaction is greater than when the initial pressure of hydrogen iodide is small. The slight falling off in rate due to the adsorption of hydrogen by the wire is shown by the following table. 200 Mm. of hydrogen iodide were used in each experiment. Preai. of H (mm.) ......... 0 100 200 k (average) .................. 040126 040105 0.00092 The Influence of Temperature.-This is shown in the following The values of k are those corresponding to a hydrogen table. pressure of 100 mm. T (abe.) ........................ 943O 836O 712O k 0-00105 0400244 ................................. 0.00268 E (cals.) 13,700 13,850 ........................lKec7mni.m of the Reaction.-We suggest the simplest possible interpretation of the resulh namely that the reaction at the surface of the platinum is the simple change HI = H + I. The apparently unimolecular course could be explained in another way by assuming that there existed one complete layer of hydrogen iodide molecules on the surface of the catalyst and that reaction took place when a molecule from the gas phase struck one of the molecules in this layer. There are however several vital objec-tions to this. First there is abundant evidence to show that where two molecules interact in heterogeneous reactions they must in general both be actually adsorbed. This is shown by the fact that in reactions where A and B interact excess of either A or B can actually retard the reaction by displacing the other from the surface.Secondly if there is a complete layer there is no reamn why the molecules in it should not react among them-selves instead of waiting for impacts from the gas phase. Thirdly, if the heterogeneous reaction depended upon impacts from the gas phase the rate could never become independent of pressure as it does when it takes place at the surface of gold. Attention may be directed to one further point of interest which emerges from these experiments. The retarding effect of hydrogen on the reaction is but slight. When the catalytic decomposition of ammonia was investigated on the same wireand actually at a higher temperature-it was found to be retarded in a most pro-nounced manner by hydrogen. It cannot be argued that hydrogen iodide is too strongly adsorbed to be displaced by hydrogen whiM ammonia is easily displaceable because the come of a d o l e -cular reaction can be revealed only when the adsorption is small. Hence we must conclude once more that different reactions are provoked at different points (‘active centres ’) on the catalytic surface Summary. In continuation of a general investigation of the influence of catalytic surfaces on reactions which in the homogeneous gas phase are bimolecular it has been shown that the thermal decom-position of hydrogen iodide in contact with platinum is kinetically unimolecular. Reasons are given for believing that this proves the decomposition to take place in the simple manner HI = H + I. PHYSICAL CHEMISTRY LABORATORY, BALLIOL COLLEGE AND TRINITY COLLEGE, OXFORD. [Received November a h 1925.