Dec. 5, 1932 PHOTOCHEMICAL DECOMPOSITION OF HYDROGEN PEROXIDE 3999

[CONTRIBUTION FROM THE GEORGE HERBERT JONES LABORATORIES, CSIVERSITY OF CHICAGO]

The Photochemical Decomposition of Hydrogen Peroxide. and Fractionation Effects

BY JOHN P. HUNT AND HENRY TAUBE RECEIVED MAY 3, 1952

Quantum Yields, Tracer

The quantum yield of the photochemical decomposition of hydrogen peroxide at relatively high light inteiisity is independ- ent of the concentration of hydrogen peroxide, of acidity and of the presence in the solution of Br-, C1-, S H 4 + or Mn++. At 25", the limiting quantun yield a t X 2537 A. io 0.98 =t 0.05 and at 0" it is 0.76 fk 0.05. The primary efficiencies are taken as '/z of the limiting quantum yields. Tracer experiments show that the oxygen formed in the photodecomposition origi- nates entirely in the hydrogen peroxide. The exchange between 0 2 and H20s during thc photodecomposition is a t most very slight. The fractionation effects associated with the non-chain process for decomposition have been determined. They do not appear to be compatible with hydroxyl radicals as the sole net products of the primary act.

The original purpose of the work described in this paper was to measure the isotope fractionation effects in the reaction

HO + H~01---+ Ha0 + E102 1 1 0 2 + HpOz + HO + HzO + 0.

using light acting on hydrogen peroxide to generate the radicals. The fractionation factors for these reactions are of interest in comparison with the factors which have been measured for a number of catalysts acting on hydrogen peroxide. In order to determine the fractionation factors in a simple way from measurement on the photodecomposition, i t is necessary to choose conditions so that the total decomposition produced by the chain process be large compared to that by the separate path consisting of the chain-initiating and -terminating steps, i.e., the chains must be long. The work of others has demonstrated, 2 - 4 and our experience has confirmed, the great difficulty that arises in purifying the solvent and the reagents and other- wise conducting the experiments so that the chain decomposition is restricted to some intrinsic path for the systein such as is represented above. The original goal has been set aside at least for the present and, in the work reported here, we have limited our studies to conditions under which the chain decomposition is eliminated. This is achieved by using light of sufficiently high inten- sity5 and results in a system which is much less sensitive to impurities.

Our results on the kinetics of the photodecom- position in general confirm those obtained by Lea,6 and considerably amplify his. In addition to experiments on the quantum yield of the reaction, results obtained in tracer experiments and in the study of fractionation effects are presented and dis- cussed with reference to mechanisnis for the pri- mary act and subsequent processes.

Experimental The light sotirce was :I G.E. -!-watt gcrrnicitl:rl lnnip opcr-

ating from it Sola constant voltagc transformer. The light cinittctl i i i thc ~vave length region affcctiiig thc uranyl oxal- ,ttc actinoinrtcr is almost entirely 25.37 A . Shortrr wixvc lengths are efliciently removed by the glass walls of the lamp. When a Pyrex glass filter was interposed, the effect

(1) A. E. Cahill and H. Taube, THIS JOURNAL, 74, 2312 (195%). (2) F. 0 Rice and 11. I.. Kilpatrick, J. Phys . C h e m . , 81, 1607

( 1827). 13) A. J. Allmand and D. W. G. Style, J . Chcm. Soc., 5!lG (lY30). (4) L. J . Heidt, THIS JOURNAL, 64, 2840 (1932). ( 3 ) D. E. T,ea. T w n s . Fiirririay Soc., 46, 81 (1949).

on thc uranyl oxalate actinometer mas observed to be only 370 of the effect when it was omitted, thus showing that mer- cury lines of longer wave lengths contribute little to the light energy. The lamp was placed in a fused quartz tube closed : I t oiie end, and this end was lowered into the neck of the reaction flask. Provision was made for sealing the quartz tube to the mouth of the flask. The flask was fitted with an entry tube for adding solutions, and an exit tube, which could be used to collect evolved gases.

In most of the experiments, the light was used without collimation, the lamp being held several centimeters above the surface of the solution. Thelight flus for each particular geometry was determined by replacing the hydrogen per- oxide solution with an equal volume of actinometer solution .6 The geometry described required correction of the results for incomplete absorption of light when the hydrogen per- oxide was a t low concentration. I n the worst case this cor- rection was estimated to be 22%. Since the correction can- not be made accurately, a series of quantum yield determina- tions were made using a collimated light beam which was confined to the central portion of the solution. For this geometry the maximum correction amounted to only 6% and the results agreed with those obtained using full illu- mination. The solutions were stirred by rocking the reaction cell assembly, or by inserting :I glass stirring blade. No difference in results was noted for the two methods.

Except where the influence of tap distilled water was studied, redistilled water was used. The hydrogen peroxide was Merck and Co., Inc., 30%, inhibitor free. The results a t high light intensity were found to be independent of lot iiuxnber of the hydrogen peroxide, of the perchloric acid and of the redistilled water, and as the results will show :;e re- markably independent of the concentration of certain cata- lysts."

Analyses for hydrogen peroxide were made by titration with standardized ceric sulfate solution. This was added in excess and the excess determined by titration with stand- ardized ferrous sulfate solution using the iron orthophenan- throline indicator.

Results Table I contains a summary of the experiments devoted

to a study of the kinetics of the photoreaction. No special effort was made to vary the intensity of the

light. However, during the course of the work on changing the geometry, it varied ca. 2-fold, but no effect on 4 was observed. The observation that the quantum yield is sub- stantially independent of (HzOZ) further supports the con- clusion that it is also independent of the rate of absorption of light since the number of quanta absorbed per unit volume changes. The rate of absorption of light was < 5 X 10'' quanta 1.-1 set.-' for all the experiments.

In some experiments, the effect of passing gases through the hydrogen peroxide solution was tested. When COZ, A or 02 was swept through, the rate of decomposition was U I I - affected. The tests were made wit: H z O ~ at 8 X lo-* M , with acid at M and a t 1 M a t 0 . Under the same con- ditions, H1 does affect the rate, enhancing it when the gas is passed at a high rate. The gases were scrubbed by a solution of potassium iodide after leaving the reaction vessel. Only with 0 2 was a noticeable amount of oxidizing agent carried

(6) W. G. Leighton and G. S. Forbes, THIS JOURNAL, 62, 3134 i l030) .

6000 JOHN P. RUNT AND E ~ E N R Y TAUBE h i . 74

T A B L E I T H E QUAXTUM YIELD OF THE PHOTODECOMPOSITIOV OF

Concentrations in moles litrr; time of insolation, 4 to 6 hr. HYDROGEN PEROXIDE AT HIGH LIGHT INTEN5ITY

No. T , "C. (HCIOII Orhcr suhs.

1 2 5 & I 1 . 0 . . . . 2 23 & 1 1 . 0 . . . , .

3 25 & 1 1.0 . . , . . . . , . 4 25 + I 1 . 0 , . . . . . . .i (J 1 . 0 , , . . . . , . t i ( 1 1 . 0 . , . . . . . . 7 0 1.0 , , . . , . . . . 8" 0 1 . 0 . . . . , . . . . !I 0 1.0 0 . 0 2 .\I sac1

I O o i . ( ~ 2 x 10-.3 AI SH,CI I 1 ( 1 1.0 2 X l W 3 .if SaBr I2 0 1 .ti 0 , O t 5 .lI Mn( C l 0 , ) ~ 1 3 0 1 . 0 0 .01 .lf I~e(C104),j 1.1 0 10-1 . . I 3 " 0 1 . @ . . . , . . . . . . l i i h o 1 .0 , . . . . . . . . 1 T h 0 1 .0 . . . . . , . . IS5 0 10-4 . . . . . . . . . .

(HzO?) 0 017 ,048 .1b6 186 030

,020 ,047 047 0 4 7 04 7

.047

.047 ,048 ,021 ,0078 ,020 ,054 022

6 0.99 1.06 1.29 1 .:30 0.77

.:ti

.84

.92 ,236 . 80 .XI) . S.5 .49 .8U

.T9

.83

.84

-- . i i

Tap distilled, rather than redistilled water \\-as used. h Collimated light beam.

over. The amount, which represented only a small fraction of the hydrogen peroxide decomposed was very nearly the same even when hydrogen peroxide is omitted from the sys- tem, and is attributed to the formation of 0 3 in the oxygen.

The effect of Ce(CIOl)a was also studied, as part of the scries in Table 11. The presence of Ce+++ a t 0.011 -11 caused a decrease in 4 of c a ~ 10%.

A tracer experiment showed that in the photodeconiposi- tion under the conditions we have employed, the oxygeii comes solely from the hydrogen peroxide. Therefore, the intermediates which yield 0 2 cannot undergo appreciable exchange with the solvent, nor can the hydrogen peroxide undergo appreciable exchange with water during the reac- tion. The experiment was conducted by dissolving hydro- gen peroxide of normal isotopic composition in water %fold enriched in 0l8, without added acid. The mole fraction of 0 ' 8 in the initial hydrogen peroxide (N,O) was 2.123 X 10-8, in the oxygen liberated after decomposition of 0.4 of the hydrogen peroxide, (N,) 2.117 X and in the re- sidual hydrogen peroxide (iVp) was 2.148 X The deviations of N8 and Np from ~vp" are real. Within experi- niental error they are the same as those observed when the environment has the same isotopic composition as the per- oxide, and are attributed to isotope fractionation.

The second type of tracer experiment was perforrnctl i n which 0 2 of normal isotopic composition was passed through :i solution of hydrogen peroxide fa. %fold enriched i t i OL8 it: water of normal isotopic composition. After a consitlcrahlc fraction of the hydrogen peroxide had decomposed photo- chemically, its isotopic composition was determined. 1 1 1 an experiment with the concentration of H202 initially a t 0.037 M and no acid added to solution, analysis of the oxy- gen released from the residual hydrogen peroxide (17%) on oxidation with Ce(1V) showed it to be enriched in O'* by 1.8%. This increase is about what would be expected of normal isotope fractionation in the photodecomposition ( v i d e infra) and is opposite in direction from changes ex- pected for exchange. We conclude that exchange betweeii oxygen and H202, direct and catalyzed by the photoproc- esses is very slight, at any rate not great enough to invalidate the measurements of fractionation fa.ctors which are rc- ported. A slight exchange effect (1% change in Np pro- duced by thephotolysis) would however have escaped notice.

The method of determining fractionation factors for the isotope effects in the decomposition of hydrogen peroxide, and the significance of data of this type, were outlined in an earlier paper.' Three fractionation factors are defined'

(7 ) The definition of the fractionation factors differs from tha t i i i

Although t h e present form is more correct, tlic \.dtics reference 1. yielded are identical a t low enrichment levels.

fo =

fr =

It = dOgls + dow'' (OPl6) dOg16 + dOWL6 (OPl8)

The subscripts g, p and w refer to oxygen, hydrogen per- oxide and water, respectively. fo and fi describe the frac- tionation for the production of oxygen and water, respec- tively, andf', the fractionation on formation of products, ir- respective of their identity. With sufficient accuracy for our purposes

f o +f* ft = - 2

f,, is obtained by comparing the isotopic composition of the residual hydrogen peroxide after a measured fractional de- composition ( x ) with that of the hydrogen peroxide initially

The ratiof,/f, is given by X

- .q- 1 - 1Vg,2N,0 - xN, - 2Np(1 - X )

The sum and ratio of fo and fr being known, each factor can be calculated. The highest precision in measuringf, can be obtained by comparing No after very small fractional de- composition with N p o . I n the limit as x -+ 0, f,, = N , / N P o . This procedure was adopted in a number of the experiments, the value being corrected for the slight change in Np during decomposition.

The results obtained in a study of the fractionation effectr are presented in Table Il. Measurements of the quantum yield were made for some of the experiments, more as a check on the process being studied than as quantum yield determinations. I t is of interest to note that a t 25' with- out added acid, substituting tap distilled water for the re- distilled water introduces a chain decomposition. At 0" with added acid, the corresponding substitution caused only a slight increase in 4 (cf. experiments 7 and 8, Table I ) .

Discussion Lea6 has shown that a t sufficiently high light

intensity (> 2.4 X 10'' quanta 1.-' set.-' when (H202) is cu. 0.01 M ) the quantum yield for the decomposition of hydrogen peroxide approaches a limit which is independent of acidity and of the concentration of hydrogen peroxide. Our observa- tions made in the range of high light intensity confirm these conclusions. However, the limiting value of 4 (0.9s f 0.05) a t 2.5" we have measured is lower than that obtained by Lea, 1.30 rt 0.11. \Ye are unable to account for the diff erencc.

The observation that 4 decreases as markedly as i t does when the teinperature falls proves that the limiting value of Q - 1.0 a t 25" cannot be attributed to a primary efficiency of unity. Two mechanisms which are consistent with the require- ment on the primary efficiency, with the kinetics of the change and which otherwise appear reason- able are

I

I1 HzOn + k~ --f H20 + 0 0 + Hz02 = HO + HO:

(1') and ( 2 ) followed by ( 3 ) and (4) as above. A reaction scheme in which the products of the

primary act are H and HOz has been considered but rejected because changing the concentration

(1') ( 2 )

Dec. 5, 19.52

(HiOz)a

0.1 .1 .1 .1 .1 -05 .025 .1 .05 .05 .05 .05 .05

(HC104) added

None None None None None None None None 1.00 1 .oo 1.00 1.00 1.00

PHOTOCHEMICAL DECOMPOSITION OF HYDROGEN PEROXIDE

TABLE I1

Concentrations in moles /liter ISOTOPE FRACTIONATION EFFECTS IN THE PHOTODECOMPOSITON OF f1202 AT = 2537 A.

Temp., "C.

25 f 1 25 f 1 25 f 1

0 . 5 f 0 . 5 25 f 1 2 5 f 1 2 5 f 1 25 f 1 25 f 1 25 f 1 25dz 1 25 f 1 25 f 1

X

0.378 .097 .204 .162 .05 .985 .436 .815 .679 . 697 .740 .67 ,687

fo

0,990 .991 .990 .991 .989 .989

,988 .989 .989

. . .

. . .

. . .

.981

of oxygen over a aide range has no effect on the rate. Oxygen competes effectively against Hz02 for hydrogen atoms,8 and would therefore cause a decrease in 4 if hydrogen atoms were present.

Both mechanisms I and I1 are consistent with the tracer experiments. By each mechanism a limiting value of 4 is reached which is twice the primary efficiency. The primary process can be considered in terms of the scheme

H202 + hv + H202*

H2O2* --f HsOz, k d €LO2* -+ dissociation, k,

ZIzOz* represents the activated entity, whether an electronically excited species or radicals trapped in a solvent cage. Using the limits of 4 as establish- ing the primary efficiencies as 0.49 and 0.38 a t 25 and O", respectively, kr/kd at the two tempera- tures is 0.96 and 0.61. If the temperature coeffi- cient of k d is assumed to be unity, the activation energy corresponding to dissociation to effective products is 2.9 kcal. This value seems reasonable in comparison with cu. 7 kcal. for the corresponding process for bromine and the value for chlorine (wave length dependent) of cu. 3 kcal.9 In all cases, the values represent lower limits for the activation energies of the dissociation steps.

Even if hydroxyl radicals are assumed to be formed as primary products of the absorption of light (mechanism I), i t is not unreasonable to suppose that they can undergo reaction in the solvent cage to form the products HzO + 0 (mechanism 11). "Escape from the solvent cage" on this interpretation means formation of these products in competition with reunion and deactiva- tion. Good evidence that the reaction of hydroxyls to form HzO + 0 takes place has been presented by Hardwick.lo

The fact that substanccs such as Br- and C1- which arc known to react with HOII do not affect the quantum yield does not prove that HO is absent.

HO + Br- --f OH- + Br

The effect of the reaction

( 8 ) H. Pricke, J . Chcm. Phys., I), 349 (1934). (9) A. C. Rutenberg and H. Taube, THIS JOURNAL, 73, 4426

(10) T. J. Hardwick, Can. J . Chcm., SO, 23 (1952). (11) H. Taube and W. C. Bray, TXIS JOURNAL, 69, 3357 (1940).

(1951).

600 1

f r f t Remarks @

0.976 0.983 .. . . . . . . . . . . . . . . . 0.99 . . . . . . . . . . . . . . . 1.00

. . . . . . . . .

... . . . . . . . . . . . . . .

. . . . . .

.978 ,984

. . . .9sg

.950 ,969 ,995 .992 .995 .992 . . . * 993 . . . .986 .951 .966

. . . . . . . . . . .

. . . . . . . . . 1.19

. . . . . . . . . . . Tap distilled HzO 3.8

. . . . . . . . . 1.00

. . . . . . . . . 0.99

. . . . . . . . . .96 6 X M F e + + + .91

M Fe+++ .88

for example, may merely be to replace reaction 3

Br + HpOz + H + + Br- + HOz Similarly with Mn++, the formation of Mn+++ would be followed by

Mn+++ + H202 --+ Mn++ + H0z + H + without necessarily producing a change in quantum yield. Evidence that Fe+++ does participate, although affecting $I only slightly a t low concentra- tion is that f,, and fr change when Fe+++ is added. The changes in the fractionation factors prove that new substances are reacting with H z 0 ~ when Fe+++ is present. When tap distilled water replaces rc- distilled water, a chain decomposition of hydrogen peroxide is initiated. The values off are different from those observed when the limiting quantum yield is observed, and show that new intermediates are involved. It is very doubtful that these are HO and HOz, and they are presumably formed from some impurity, dissolved or solid, in the tap dis- tilled water. The experiments with Fe+++ added are difficult to evaluate quantitatively. At 10-2M Fe+++ only 3% of the light is absorbed by HzOz; however, the quantum yield is decreased by only 40%. A question of interest which will require more data to be settled is whether light absorbed by Fe+++ is also photochemically effective. If Fe+++ acts as an efficient internal filter, i t must act also to increase the chain length since i t produces a rela- tively slight change in 4 when present in sufficient amount to absorb nearly all the light. The de- crease in 4 produced by Ce+++ may be due to an internal filter action.

It is of interest to consider the measured values of f, andf, in relation to mechanism I and 11. For mechanism I, fo = fs.f4 and fr = fi. fi describes the discrimination between isotopic forms of HzO2 in rcaction (l), f a in reaction (3) and f4 describes the isotope discrimination when HOz forms 0 2 . On the basis of mechanism I , the fractionation of isotopes in forming water takes place only in the primary act (HO once formed yields water with no further iso- tope discrimination). The observations on fi. are inconsistent with this mechanism since it provides no way of explaining the change off, with acidity. Further, the magnitude of the fractionation effect in low acid cfr = 0.977) seems too great to be attri-

by

buted to a primary act of etliciency nearly 0.3. llechanisni I1 as written similarly givesj; = .II’ and meets the same difficulty. However, if it is sup- posed that at low acid, reaction 2 takes place as

O* -+ OOH- = 0’0 J- oir -

the necessary degree of ireedom to account for the change in fr is introduced. -In analogous change in mechanism does not appear as reasonable with HO as the reactant. The fractionation data teiid to disqualify mechanism I, and favor T I a i a iiicans of explaining all the observations.

The transition from the limiting deconiposition to the chain decomposition has been discussed by Lea.5 The slight trend in I#I with concentration of hydrogen peroxide (Table I , experiments 1-4 and 5-7) is presumably due to a residual chain decom- position, which diminishes as the peroxide cuncen- tration decreases. I t is by n u means certain that the chain decomposition which sets in is carried b>- HO and HOz. The difficulty that has been exper- ienced in obtaining reproducible data when the

chain lengths are long makes i t questionable to ;is- sume that the chain carriers are these radicals i n an!. particular case. The fractionation factors

.fo andJ should be useful in characterizing the inter- mediates in future work. They have the advan- tage over rate measurements that they are inde- pendent of chain length, and are affected only by changing the identity of the intermediates. Thus dn accidental inhibitor that acts only by breaking chains will affect the quantum yield but not the \.alues of i o aiid J , . I t IS assumed in these remarks that the chains are long enough so that the princi- pal path lor deconiposition of HLO? is by the chain mechanism I

Acknowledgment.- -This work was supported by the Oftice of Kava1 Research under contract NB- ori-020%. The funds for the purchase of the mass spectrometer used in the research were supplied by the -1toiiiic. Energy Commission under contract - \ t i l l -1)-!12. CHICA(>O 37, 1I,I,I’i<JIS

Kinetics of the Gas Phase Reaction of Olefins with Ozone Jlu I<ICH.IRD TI. C,\DLE AND COKR.\D SCKIDT

I<l?CEIVEn .kPRII, >!I, 1

A I-apid gas phase rcactiim occurs lietu-ecii ozoiic aiid various olcliiis. T h c ovcr-;tll reactions arc quitc coinplicatctl, more By extrapolating rate5

The initial rate was mostly homogeneous Certaiti

oletin disappears than ozoirc oii a mole I):&, :tiid the stoichioinetry varies with rc:tct:iiit pressures. I ~ c k to initial conditions, ii relativcly simple swoiid-order rate law was observed. atid independcnt of foreign gas pressure. xsixcts of a mechanism are discussed.

1Yithiti cspcritrirtital error the ratcs arr also independent of tcmperature.

’The reaction of ozone and olefins has been studied in the liquid phase, especially with rcspect to rcaction products.‘

’The present paper reports a study of the kinetics o l the gas phase reaction of olefins with ozone. ,\nother group in this Laboratory has been in- vesticating the products of the gas phase reactioii between olefins and ozone,? and this study will lie reported separately. This work was part of a broad program to inirestigate reactions which iri;ty occur in contaminated atmospheres, m d thus particular emphasis has been pIaced on low coil- centrations.

Experimental The olcfiris investigated n w c 1-hesenc, ethylene, cyclo-

hexene arid (very briefly) 1-decene, 1-octene, 1-heptciic, 1- ])cntene and propylene. They werc obtained from coin- niercial sources. The liquid hydrocarbons were purified by distillation. The ethylene was condensed in a trap cooled with liquid nitrogen. Gaseous ethylene was produced as riccdcd by allowing the trap to warm slightly. Ozonc was prcparcd from cylinder osygcri by it corona discharge ozo-

tilethod of Crribirer :iiid t ; c i~ ip .~ This method had the atl- vltiitagc that vcry low coiicciitratious of mtct:mts could bc crn],loyetl.

The other technique ernploycd the cell of a recording it i - frarcd spectroineter :IS the reaction vessel. This ccll M ~ S 1 0 cin. long and :3 ern. in imide diameter. Thc cheniicd re- action was followctl by measuring absorption at a fised imvc lcugth. The gas handling system ~ r a s similar to that uscd by hlills and Johnston for reactioiis at intermediate prcs- surcs.

The reaction cell, mixing cell and double stopcock were iiiiinersed iii a tlierinostated oil-bath. Each of thc liiicq 1c:rding to the gas-handling system included a coil irniiicrscd in the oil-bath which aided in bringing the gases to bath tcinpcrature. A run wts started by evacuating the reac- tion cell, closing the stopcock between cell and pump, and admitting reactants through the double stopcock. This technique required higher concentrations of reactants than the chamber technique, but temperature and total pressure werc inore easily controlled.

Thc ratcs of decomposition of ozone in the two types of equipitient were investigated. Thc ratc of disappearancc

)tic in the chamber was about 15% per hour. S o ob- ble dccompositioti of ozoiic occurred iti the ccll of thc

ilxctroirieter duriiig tivo iiiiiiutcs which wits longer t1i:iu 1llObt or the ruiis.

Results I - H ~ ~ ~ ~ ~ . - T ~ ~ gas phase reaction of l-hexene

with Ozone was studied entirely with the chamber technique. Two sets of runs were made to deter-

nizcr. Two techniques were used t o folloiv the course of the re-

actions. One involved mising the reactants in the air of cl nalvanized iron chamber of 10 m.3 volume. Samples of the air in the chamber were withdrawn periodically and ana- lvzed chemically for oxidant using .t modification of t h c ____ ~ ._ - - iiiine the ortfcr and stoichiometry of the reaction.

(1) R. Criegee, “The Oaonolysis of Olefins and -4cetylenes,” paper