P . BONET-MAURY 79 MECHANISM OF DECOMPOSITION OF WATER BY IONIZING RADIATIONS BY AUGUSTINE 0. ALLEN Chemistry Department, Brookhaven National Laboratory, Upton, Long Island, New York, U S A . Received 14th January, 1952 Irradiated water undergoes a decomposition to molecular H2 and H202, simultaneously with the decomposition to free radicals, H and OH. With gamma rays or hard X-rays, the yield of the niolecular decomposition is about 0.6 Hz molecules formed per 100 eV absorbed, and the yield of free radicals appears to be about 3-5 radical pairs per 100 eV. As the ionization density of the radiation is increased, the molecular yield increases and the free radical yield falls. The decomposition of pure water reverses itself because the free radicals initiate a back reaction between the decomposition products.The rate of the back reaction increases with increasing concentration of dissolved hydrogen but de- creases with increasing concentration of dissolved oxygen or hydrogen peroxide. This80 DECOMPOSITION OF WATER unusual type of kinetics leads to some peculiar phenomena in water radiolysis. The molecular decomposition is ascribed to reactions occurring in the very small regions of high energy density along the charged particle track (hot spots) which correspond to the " ion clusters " formed by fast particles in a gas. Determination of what occurs when high energy radiation acts on water or simple aqueous solutions constitutes a difficult experimental problem. The first requisite for understanding a reaction is establishment of a complete material balance.Only by a knowledge of amounts of all products formed and all sub- stances disappearing can one have reasonable confidence of knowing what is going on. Since hydrogen gas is generally an important reaction product, this demands setting up a refined system for determining small concentrations of gas dissolved in small samples of water, as well as sensitive and precise methods for determination of other dissolved materials. A second necessity is care in purifying the material studied and in establishing reproducibility of results, since the behaviour of most solutions seems to be greatly affected by trace contamin- ations. Another highly desirable condition is that radiation sources should be available which allow reactions to be studied over a wide range of radiation in- tensity as well as radiation quality.During the last five years, researches have been carried out at several laboratories in the United States and Canada which have attempted to meet these difficult requirements. Results of these studies are now beginning to appear in the scientific journals. These researches have greatly enhanced our previous knowledge of radiation decomposition of water, and this paper will attempt to summarize the present status of the subject. DosImTRY.-Radiation reactions are discussed in terms of the yield of molecules reacting per unit energy input. In practice, laboratories working with X-rays or gamma rays generally obtain their yields by comparison with the yield of oxidation of ferrous sulphate in 0-8 N sulphuric acid solution.Unfortunately disagreement exists as to the absolute value of the ferrous sulphate oxidation yield. By comparison with ionization in air, Fricke and co-workers obtained a value of 18.2 molecules oxidized per 100 eV ; 1 Miller later obtained a value of 20.6.2 Calorimetric determination of the heat produced when solutions are irradiated would appear to provide a more direct method. Hochanadel has carried out such a determination with a cobalt gamma ray source at Oak Ridge and finds for the ferrous sulphate solution a yield of only 15.5 molecules oxidized per 100 eV.3 The determinations appear equally sound and it remains for further work to effect a choice between the values. from recent work is that in addition to the free radicals formed from water, a direct decomposition of water occurs to yield gaseous hydrogen independent of what material may be dissolved in the water.I had postulated the molecular reaction in a previous paper,4 largely on the basis of preliminary results obtained by J. A. Ghormley and myself on decomposition of water by X-rays and cathode rays. The molecular reaction has since been verified in numerous systems. Solutions of HCI, CuSO4, KBr and others at various concentrations, placed in the Oak Ridge nuclear reactor, were found to give the same initial yield of hydrogen gas within experimental error.5 Subsequent more accurate work by Dr. Everett Johnson and myself at Brookhaven,6 using hard X-rays from a 2 MeV electrostatic generator with gold target, has shown that a constant yield of hydrogen is obtained from irradiation of a variety of solutions, including KBr, KI, HBr, H202, oxygen-saturated ferrous sulphate, air-saturated water and ceric sulphate solutions.The only difference found was that from solutions containing 0.8 N H2SO4 the yield of hydrogen was consistently 18 % lower than from dilute solutions in pure water. The results were most striking with H202 solutions. Here the H202 rapidly decomposes to oxygen. Throughout the entire reaction, however, the hydrogen still keeps being generated at a constant rate (provided the initial con- centration of peroxide is high enough) which is exactly equal to the rate at which hydrogen is generated in KBr and other unrelated solutions. Changes in the details of preparing the solutions and intentional addition of foreign substances, which had a large effect on the peroxide decomposition yield, did not affect the hydrogen yield.7 THE MOLECULAR YIELD OF WATER DECOMPOSITION.-The main fact which emergesAUGUSTINE 0.ALLEN 81 The hydrogen is accompanied by an equivalent quantity of oxidizing material which appears to consist mostly, if not entirely, of H202. In acid bromide solutions the actual peroxide found in the solution is equivalent to the hydrogen formed. In neutral bromide or iodide solutions the oxidizing agent appears mostly as oxygen rather than peroxide,39 8 but this may be accounted for by the rapid decomposition of peroxide to oxygen under these conditions.7 The production of a constant yield of hydrogen gas was shown by Fricke and Hart for solutions of a number of simple reducing agents, viz.iodide, bromide, nitrite, arsenite, selenite and ferrocyanide.8 In all these cases the yield of hydrogen was the same, and was independent of concentration of dissolved material and of pH. With the above reducing agents other than bromide and iodide the oxidized form of the material appeared in equivalent amount to the hydrogen formed. In these cases the HzOz apparently oxidized the reducing agent quantitatively, this action being perhaps catalyzed by the free radicals present. As the concentration of dissolved materials is reduced to a low level, the hydrogen still initially appears with its characteristic rate but the rate soon slackens and the hydrogen concentration levels off at some definite value.59 6 When no solute is added, the hydrogen concentration produced by X-rays levels of€ at a very low value, generally close to the limits of detection, which is very sensitive to the presence of trace impurities and is practically impossible to reproduce.Evidently water decomposes directly to molecular hydrogen as well as forming free radicals that may react with solutes. In the absence of a sufficiently high concentration of solutes, the hydrogen reacts with radicals and is converted back to water ; but almost any solute capable of oxidation or reduction will destroy the radicals and allow the molecular hydrogen to appear. In a redox system, the radicals may produce equivalent amounts of reduction and oxidation and hence give no net effect, so that the sole overall reaction occurring is the production of molecular hydrogen and an equivalent amount of H202 or reaction products of H202.In HzO2 solutions, the radicals do produce a net reaction, decomposition of the peroxide to oxygen, but as long as peroxide and oxygen are present in sufficient amounts they use up all the radicals and the hydrogen keeps coming out at its characteristic rate. In ceric sulphate solution the overall yield of re- duction of cerium is greater than one would expect from the amount of molecular H202 formed along with the hydrogen gas. The interplay of radicals here is such as to result in some net reduction. Deaerated solutions of ferrous sulphate give an unusually high oxidation yield, the interplay of radicals resulting in some net oxidation.The mechanisms of these reactions are not yet entirely understood. The molecular hydrogen yield was demonstrated by Hart 9 for solutions of formic acid and oxygen. Deaerated formic acid solutions give a high yield of hydrogen, which arises mostly from the acid, as shown by the presence of deuterium in the hydrogen pro- duced by irradiation of solutions of DCOOH in ordinary water. When oxygen is added, the acid is oxidized instead of decomposing and the hydrogen yield falls to the value characteristic of the molecular decomposition of water. The hydrogen from DCOOH + oxygen solutions contains practically no deuterium. The molecular yield of hydrogen from hard X-rays or gamma rays, expressed as a fraction of the yield of ferrous sulphate oxidation, is 0.030 according to Hochanadel,3 0.036 according to Fricke and Hart,8 0.022 according to Hart9 and 0-039 according to Johnson and Allen in pure water, 0.031 in 0-8 N H2S04.6 The results of various laboratories appear to agree within experimental error, except perhaps for Hart’s low value.To convert these ratios into yields of hydrogen molecules formed from water per 1OOV energy input, the above fractions are to be multiplied by either 20.6 or 15.5, depending upon the value accepted for the absolute yield of the ferrous sulphate actinometer. The hydrogen yield thus lies between 0.45 and 0-7 molecules per 100 eV. of hydrogen from decomposition of water by natural alpha rays has been studied by many investigators and is equal to about 1.9 - 0.1 molecules of hydrogen per 100 eV.103 1 1 , 1 2 This number presumably represents the molecular yield of hydrogen from water by this type of radiation since the initial yield does not drop as the hydrogen accumulates, and solutions of HI were found to give nearly the same value as pure water.]] For radiations of ionization density intermediate between natural alpha rays and hard X-rays the avail- able information is only qualitative but strongly suggests that the molecular yield increases continuously with the ionization density.EFFECT OF IONIZATION DENSITY OF RADIATION ON THE MOLECULAR YIELD.-The yield82 DECOMPOSITION OF WATER Results with cyclotron deuterons, protons and alphas have been published by Toulis.13 He found that as the energy of the protons reaching water was increased from 2 to 12 MeV, the peroxide yield per proton instead of increasing linearly with the proton energy, showed a curve which bent over, reached a maximum at 7 MeV and decreased at higher proton energies.Thus the part of the proton beam lying above 7 MeV actually caused more destruction of H202 than formation, while the peroxide formation occurred chiefly in the lower energy parts of the proton tracks. Similar results were obtained with cyclo- tron deuterons and alphas over a comparable range of ionization densities. The resuits of such an experiment must depend on how the HzO2 is mixed into the body of solution, since the local concentrations of peroxide may vary with depth in the water as the mean ionization density of the rays varies. We cannot agree with Toulis in the detailed interpretation that he offers of his results ; but the results themselves are extremely suggestive.Observations have been made of the steady-state level in the decomposition of water with a variety of types of radiation which tend toward the same conclusion. Such observations are not quantitative since the steady-state level is greatly affected by trace impurities and is generally not well reproducible. However, in the Oak Ridge reactor,5 the effect of changing radiation quality on the steady-state level of decomposition of water was demonstrated in a vessel which was kept sealed, thus keeping the chemical con- ditions as constant as possible. When the mean ionization density of the reactor radi- ation was decreased by surrounding the vessel with paraffin, the pressure of gaseous decomposition products of water within the bulb showed a precipitate drop ; when the ionization density was increased by surrounding the vessel with lead, the pressure rose.The change of steady-state decomposition pressure with radiation quality is a good indication that the molecular yield increases with increasing radiation density. In the decomposition of water by the soft beta rays from tritium (average energy 5700 eV) steady- state decomposition levels were obtained which were much higher than those expected from similar experiments with hard X-rays.14 Hart has irradiated solutions of formic acid 9 and of ferrous sulphate,ls in the presence and absence of oxygen, with tritium beta rays as well as gamma rays. An exact com- parison of the molecular yields from the two types of radiation is difficult to make, because of uncertainties in the absolute dosimetry and in the reaction mechanisms, but a higher value for the tritium rays is indicated, and the difference probably amounts to about 20 %.The indications are then that the molecular yield increases with increasing ionization density but more quantitative information on this point is badly needed. decomposition is always the resultant of the rate of decomposition of water (re- ferred to as forward reaction) and the rate of re-formation of water from the decomposition products (back reaction). The forward reaction ordinarily arises entirely as a result of the molecular decomposition of water ; the free radicals formed by water decomposition usually disappear by reaction with the molecular reaction products and are responsible for the back reaction.The kinetics of water decomposition are to be explained by a study of the back reaction, and the most effective way of studying radiation decomposition in water is to irradiate solutions containing H2, 0 2 and H202 at various proportions and concentration levels. Extensive experiments with the mixed fast neutron and gamma radiation in the Oak Ridge reactor5 and later experiments with gamma radiation alone3 have shown that the back reaction is accelerated by increasing the hydrogen concentration, but is decelerated by increasing the concentration of either 0 2 or W202. This leads to some curious phenomena in the decomposition of water.If the water is irradiated in a closed vessel full of pure water, a steady state is soon reached at a low level of product concentrations. If, however, a gas space exists over the water the hydrogen gas produced escapes into the gas phase leaving an excess of H 2 0 2 dissolved in the liquid phase. The excess H 2 0 2 inhibits back reaction and the decomposition proceeds to a much greater extent than in a full vessel. If hydrogen is added to water prior to irradiation the decomposition is completely repressed, but if oxygen or H 2 0 2 is added initially the decomposition proceeds to high levels, with much hydrogen being produced on long irradiations. REACTION KINETICS IN WATER DECOMPOSITION.-The observed yield Of WaterAUGUSTINE 0 . ALLEN 83 A trace of organic impurity in the water will decompose initially with the forma- tion of hydrogen which acts to repress the water decomposition so that formation of oxygen or peroxide is not observed.Traces of certain inorganic impurities, however, may inhibit the back reaction almost entirely so that large amounts of decomposition are seen. These facts explain apparent anomalies in some of the older work on the effect of X-rays on pure water. Risse 16 and Fricke and co- workers 17 irradiated water in filled vessels and observed very little decomposition. Guenther 18 irradiated water in the presence of a very large available gas volume and observed continuing production of hydrogen. These apparently contra- dictory results are exactly what is expected on the present picture.The kinetics of the back reaction in the presence of excess peroxide or oxygen are unfortunately irreproducible.5 Irreproducibility is also characteristic of the kinetics of decomposition of peroxide to oxygen in irradiated peroxide solutions. In systems containing dissolved hydrogen and oxygen or peroxide, with hydrogen not present in excess, conversion of peroxide to oxygen ‘or vice versa always occurs and the kinetics are erratic, However, a solution of H202 and hydrogen with the hydrogen in excess gener- ates little or no oxygen on irradiation, the sole reaction being the disappearance of equimolecular quantities of H2 and H202 with formation of water. Under these conditions reproducible reaction kinetics were found both with nuclear reactor radiation 5 and with gamma radiation.3 The yield of disappearance of hydrogen or peroxide, after correction for the amounts of these materials pro- duced by molecular decomposition of water, is the3 strictly proportional to the hydrogen concentration, inversely proportional to the peroxide concentration and independent of the radiation intensity : where Gr; is the yield of molecular decomposition of water with the radiation used, as evaluated by rate of hydrogen evolution from other solutions such as bromide or HCl.The molecular decomposition equation is 2H20 -- W? 4- H202. Eqn. (1) suggests that a chain reaction is occurring leading to reaction of hydrogen, with peroxide acting to break the chains as well as to carry them on, so that peroxide acts as an inhibitor. The most likely scheme for the chain reaction is (F) H20 r= H -1 OH H + H2Oz -- OH f H20 OH + H2 = H + 1320 The chain-breaking reaction involving peroxide is probably OH + M201= HO2 + H20 which is universally postulated to account for the photochemical or radiation chemical decomposition of peroxide to oxygen.In the presence of excess hydrogen, all the HO;! radicals formed must be reduced back to €3202 and none oxidized to oxygen, since no oxygen is formed in the reaction. This probably occurs because with excess H2 the concentration of hydrogen atoms is kept high by means of reaction (3), so that all HO2 radicals disappear by the reaction (4) Reaction ( 5 ) is not a rate-determining step, since under the conditions postuIated every H02 formed disappears by (5). Furthermore, as long as a reasonable con- centration of HzOz is present all hydrogen atoms not consumed in (5) react by84 DECOMPOSITION OF WATER (2), so that reactions of hydrogen atoms are not rate-determining either.The only rate-determining steps then are reaction (R), which gives the rate at which chains are started, and reactions (3) and (4), the ratio of which gives the proba- bility that the chain will either be carried on by reaction of OH with H2 or ter- minated as a result of reaction of OH with H202. The observed reaction kinetics (eqn. (1)) follow immediately from these considerations, since the rate of any chain reaction equals the rate of initiation times the ratio of the rate of continu- ation to the rate of termination. The constant K is equal to the product G~k3/k4, where GR is the yield of reaction (R).In the papers referred t0,395 the rate expression is derived by the standard steady-state method of homogeneous reaction kinetics, assuming that the bulk concentrations of the different free radicals may be taken as uniform. Objection has been taken to this method of treating radiation chemical reactions, since the concentration of radicals is not in fact uniform but varies as the radicals diffuse away from the regions where they are initially formed. In the above case such considerations make no difference, since the rate-determining step is merely the competition of hydrogen and peroxide molecules for the OH radicals, which depends only on the concentrations of these molecules and not on that of the radicals.The effects of " track overlap " on radiation chemical kinetics, on which considerable speculation has appeared in print, would appear to be important only in cases where the rate of reaction between radicals actually determines the rate of the overall observed reaction. This may be true for radiation-induced polymerization of such solutes as acrylonitrile.19 The course of most radiation- chemical reactions, however, seems to be determined more by competition of dissolved substances for reaction with radicals rather than by reaction between radicals, and in such cases the overlapping of tracks has no special significance. THE FREE RADICAL YIELD IN WATER DECOMPOSITION.-The chemical effect Of any given radiation on water can be characterized by two numbers, GF and GR, one giving the yield of molecular products, the other the yield of free radicals.GF can be readily obtained by measuring the yield of hydrogen produced on irradiation of a variety of solutions. The value of GR is much more difficult to estimate. The difficulty is that whatever action is produced by one radical of the H-OH pair can, in general, be re- versed by the other radical. This is most obvious in reversible redox systems, but even with organic compounds such possibilities exist. Thus with any organic compound we may have the sequence RH + OH = H20 + R; R + H = RH. The yield of conversion of dissolved materials will, in general, give only a lower limit to CR. An attempted way out of the difficulty has been to study a solution containing two solutes, one of which was supposed to react readily with hydrogen atoms and remove them from the system, the other to react with the OH radicals.Then the total chemical change should be a measure of the total free radical yield, providing no subsequent reactions occur to reverse the changes. Possiblity of such reversal still appears, however, to provide an essential difficulty. Attempts to evaluate GR in this way have been made using results obtained on solu- tions of the mixtures formic acid + oxygen, hydrogen + oxygen and ferrous sulphate + oxygen. The formic acid t- oxygen mixture on irradiation yields the reaction 9 HCOOH + 0 2 = H202 -I- C02. The assumed mechanism was H + 0 2 = HO2 OH t HCOOH = HCOO + H20 H02 + HCOO == H202 + C02. According to this mechanism the yield of conversion of formic acid and oxygen should be equal to GR.The difficulty is that if HO2 radicals after being formed react with them- selves or with OH or H, the value of GR will have been underestimated. The radical yield estimated by Hart from this reaction was 2-78 radical pairs per 100 eV for cobalt gamma rays if the dosimetric standard is 15.5, 3.6 if the standard is 20-6, and we may assume that at least this many free radicals are produced.AUGUSTINE 0. ALLEN 85 The formation of H202 in solutions of hydrogen and oxygen was treated by Hochanadel in a similar way.3 The assumed reactions were H + 0 2 = H02 OH 3- H2 = H20 + H 2H02 = H202 + 0 2 and the peroxide yield was taken as a direct measure of the number of free radicals produced. The yield obtained was very close to that found by Hart with formic acid but here again the same difficulty occurs with the reaction of HO2.For the ferrous sulphate + oxygen system, Krenz and Dewhurst 20 and also Hart 15 have used the mechanism H + 0 2 = H02 OH $- Fe2+ = Fe3+ + OH- H+ + HO2 + Fe2+ = Fe3+ + H202 H202 + Fez+ = Fe3+ + OH- + OH. This gives a yield of 4 or 5 radical pairs per 100eV (depending upon the dosimetric standard), which is significantly higher than the free radical yield values found for the formic acid and hydrogen cases. Evidently the mechanisms assumed are not entirely correct in at least one of the above cases, and further work is being carried out on all these reactions. In any case it would appear that the results on any one such system can, in general, give only a lower limit to the free radical yield, and authentic values will be on hand only when the numbers deduced from several different independent reactions are found to show precise agreement.PHYSICAL PICTURE OF THE ACTION OF RADIATIONS ON WATER.-Much has been written about the processes occurring when radiations traverse water, mostly from the viewpoint that water should behave much like a highly compressed perfect gas. Such an assumption is so far from reality that we cannot expect it to furnish a very good picture of the real process, much less to provide quantitative predictions of experiments. However, some sort of physical model seems required by our minds as a framework for our thinking about the observed chemical phenomena, and models of this sort do no harm if not taken too seriously.The rate of energy loss by charged particles passing through matter is nearly independent of the state of chemical combination of the atoms present, but the nature of the excited states formed certainly depends strongly on the state of combination and state of aggregation of the material. We may therefore assume that the distribution of energy loss is the same in liquid water as in a vapour. But the distinction between ionization and excitation, so prominent in the dilute gas, may be largely lost in the liquid state. The upper excitation bands should be broadened in a liquid and merge with the ionized state, while on the other hand a slow electron ejected from a molecule in a liquid by ionization may well recoil, return and recombine with the positive ion thus giving in effect an excited molecule from a process which would be an ordinary ionization in the vapour state.Both ionization and excitation processes, however, result in the dissociation of water molecules to form free radicals, probably mostly H and OH. The best procedure in drawing our picture is therefore to ignore as far as possible the classification of processes in terms of excitation and ionization and to consider rather the distribution of free radicals in terms of distribution of energy loss within the liquid. The distribution of energy loss itself may legitimately be derived from experiments and theories relating to the gas phase. The loss of energy by fast electrons passing through matter occurs not uniformly but in bursts of varying size, leading in gases to the formation of ions in clusters along the track.The size of the clusters can be determined roughly from well- resolved cloud chamber pictures, and counts of the distribution of ions in the clusters have been published by Wilson21 and by Beekman.22 They find that the relative number of clusters of any given size decreases approximately exponen- tially with size of the cluster. The mean size of the clusters is about 2.5 to 386 DhCOMPOSITlON OF WATER ion pairs.23 Thus the “ mean cluster ” corresponds to an energy loss of about 100 eV. Considering a fast electron, say 100 keV, passing through liquid water, we have the picture of some isolated occurrences involving only enough energy loss to dissociate one water molecule into radicals, while other occurrences will involve dissociation of several water molecules located very close together.Assuming that a total of six water molecules are dissociated per 100 eV, a typical “ radical cluster ” will consist of a group of about six H and six OH radicals. Since only a third of the total energy is required to dissociate six water molecules, much energy will appear immediately as heat and the region is perhaps best denoted by the term “ hot spot ”. A typical hot spot arises when an electron of 100 eV energy is ejected from a water molecule. According to Lea’s tables,24 such an electron should have a total range in water of about 30A. This electron will suffer so many collisions that its path will resemble a random difflision path and its course will lie in a region having a diameter equal to about one-third of its total range, or lOA.The six OH radicals formed by it will lie closely along its path, but some question exists as to the location of the six H atoms. Some of these are formed by ejection of an electron from a water molecule and subsequent capture of this electron at a distance. Lea assumed that a free electron in water would diffuse for some time at thermal velocities before being captured and that the mean separation of H and OH was about 150 A.24 There is some evidence that this distance is too large and that the electrons are rapidly captured by liquid water. Bradbury and Tatel25 found that water vapour has a high cross-section for capture of low- energy electrons when the pressure is near the saturation point, but capture ceases when the pressure is low.If water vapour below saturation captures electrons by virtue of incipient molecular association, the liquid itself should capture them immediately. The hydrogen atoms may be no farther apart than the OH radicals. The most that can be said about size of the hot spots is that their radii should lie between 7 A and 150 A, probably closer to the smaller value. Considerable immediate recombination of radicals must occur in such a small region, and when like radicals combine we get the molecular H2 and H202 Nhich is actually observed. Thus the existence of the molecular yield in water decom- position is ascribed to the existence of the hot spots. A calculation of the ratio of the initial rate of recombination to the initial rate of diffusion out of a cluster is readily made if the initial radical distribution is assumed to be Gaussian in distance from the centre.The ratio is equal to 0.054 cc No/Db, where NO is the initial number of radicals, CL and D the reconibina- tion and diffusion coefficients, and b the mean radius. If we take NO = 12 and cc/D = 5 x 10-7 cm, the ratio becomes unity if 0 = 32A, and will be larger if b is smaller. For larger radical clusters (higher No) the ratio will be proportionally greater. As the ionization density of the primary radiation increases, the isolated radical pairs must approach closer and closer to the clusters and eventually will begin to be swallowed up by the clusters. This will result in an increase of the molec- ular yield and decrease of the free radical yield in the irradiation of water.Such a change should not become very prominent until the mean distance between primary events becomes comparable with the mean cluster radius. A quanti- tative theory for the change in molecular yield with ionization density would not be difficult to formulate on the basis of the above considerations, but is perhaps better postponed until detailed experimental data are available. The above formulation has little, if anything, to do with the frequently dis- cussed problem of the diffusion of radicals out of the tracks of fast electrons. The formation of a column of radicals around the track of an electron can hardly be said to occur until the radicals have diffused a distance from the track equal to their mean separation along the track.For a 100 keV electron this distanceAUGUSTINE 0. ALLEN 87 will be of the order of 1000 A, and the process of initial recombination within the radical clusters and loss of the cluster identity by diffusion will be complete by the time this separation has occurred. The assumption of six radical pairs formed per 100 eV is based on the idea that about four pairs appear as free radicals in gamma or X-irradiations, one pair appears as H2 and H202 and one other pair must have disappeared within the radical cluster by recombination between unlike radicals to re-form water. With natural alpha rays the molecular yield is only two H2 molecules per 100 eV and the free radical yield appears to be negligible,26 so that we must conclude that only one-third of the radicals escape from the alpha-ray column as H2 or Hz02. At least one-half of the radicals in the column should, however, combine as like pairs to give molecular products, and if there is segregation within the column as proposed by Lea 24 the fraction will be greater than one-half.We must assume that some H2 and Hz02 molecules formed by initial recombination are destroyed by back reaction to water, occurring as a result of free radical action within the cluster. We then expect that the sum of the molecular yield and free radical yield should increase as the ionization density decreases. The above picture is certainly a grossly over-simplified explanation of the complicated changes that must be occurring within a hot spot. The real processes are probably much less clear-cut than the mere dissociation of molecules to radicals and subsequent recombination of these radicals in pairs. Nevertheless the existence of the hot spots can hardly be doubted, and the connection between these hot spots and the molecular decomposition of water seems highly probable. Work on this paper was performed under the auspices of the United States Atomic Energy Commission. 1 Fricke and Morse, Phil. Mag., 1929, 7, 129. 2 Miller, J . Cheni. Physics, 1950, 18, 79. 3 Hochanadel, J . Physic. Chem., 1952, 56 (in press). 4 Allen, J . Physic. Chem., 1948, 52, 479. 5 Allen, Hochanadel, Ghormley and Davis, 1951, Report AECU-1413 ; J . Physic. 6 Johnson and Allen, paper submitted to the J . Atner. Chern. Soc. 7 Johnson, J. Chem. Physics, 1951, 19, 1204. 8 Fricke and Hart, J . Chem. Physics, 1935, 3, 596. 9 Hart, 1951, Reports AECU-1250 and ANL-4636 ; J . Physic. Chem., 1952,56 (in press). 10 Nurnberger, J . Physic. Chem., 1934, 38, 47. 11 Lanning and Lind, J. Physic. Chem., 1938, 42, 1229. 12 Duane and Scheuer, Le radium, 1913, 10, 33. 1 3 Toulis, 1950, Report UCRL-563. 14 Ghormley and Allen, 1948, Report ORNL-128. 15 Hart, J . Amer. Chem. SOC., 1951, 73, 1891. 16 Risse, Z. physik. Chem. A , 1929, 140, 133. 17 Fricke, Hart and Smith, J . Chem. Physics, 1938, 6, 229. 18 Guenther and Holzapfel, 2. physik. Chem. B, 1939, 44, 374. 19 Dainton and Collinson, Ann. Rev. Physic. Chem., 1951, 2, 99. 20 Krenz and Dewhurst, J . Chem. Physics, 1949, 17, 1337. 21 Wilson, Proc. Roy. Soc. A, 1923, 104, I, 192. 22 Beekman, Physica, 1949, 15, 327. 23 Kara-Michailova and Lea, Proc. Camb. Phil. SOC., 1940, 36, 101. 24 Lea, Actions of Radiations on Living Cells (Cambridge University Press, 1946). 25 Bradbury and Tatel, J. Cliem. Physics, 1934, 2, 835. 26 Dale, Gray and Meredith, Phil. Trans. Roy. SOC. A, 1949, 242, 33. Chrm., 1952, 56 (in press).