KINETICS OF HYDROXYL RADICAL I N AQUEOUS SOLUTiON
47
Kinetics of the Hydroxyl Radical in Aqueous Solution'
by Fred Sicilio, Roland E. Florin, and Leo A. Wall National Bureau of Standards, Washington, D . C.
(Received Mal/ 26, 1966)
The electron spin resonance spectrum of hydroxyl radical, generated from titanous ion and hydrogen peroxide in a flow system, was studied as a function of flow rate, temperature, and mixture composition. Two peaks occur: a principal peak at, g = 2.0128 and a usually minor peak a t g = 2.0114. The intensity of the minor peak is increased by large concent,rations of titanous chloride or sulfate, increased slightly by large concentrations of titanic salts, and lowered by excess sulfuric acid. The minor peak is thought to be the spectrum of hydroxyl associated with titanic ion. At 20 and 28" the concentration of hydroxyl shows a maximum as flow rate is varied, while the concentration increases monotonically with increasing flow rate a t 58". At low flow rates (time after mixing 0.5 to 4 sec.) there appears to be an approach to second-order disappearance nith k = 4.5 X 105 M - I see.-' at 28" and an activation energy of 11 kcal./mole (46 kjoules/mole).
Introduction When aqueous solutions of titanium trichloride and hydrogen peroxide are mixed in a flow system, a rapid reaction generates hydroxyl radicals in concentrations easily observable by electron spin resonance.2 If a reactive organic substrate is present, it is attacked to form detectable concentrations of characteristic organic radicals. The e.s.r. spectra of a large number of such radicals have now been des~ribed.~-~ A number of preliminary observations of the aqueous hydroxyl radical have been reported.2 It is our purpose here to describe some observations of spectral detail and of the rate of formation and disappearance of the hydroxyl radical. Experimental Section Two kinds of titanium trichloride were used: the technical, nominal 20% solutions containing an inorganic acid a13 oxidation inhibitor and a pure anhydrous product. I n connection with a few studies, a measured quantity of 20% Tic13 was converted to the sulfate 3Ti2(S0&* H2S04.25HzO5 by adding equal volumes of concentrated sulfuric acid cautiously, heating to incipient precipitation, cooling, and washing with 60% sulfuric acid. The product contained perhaps 1% chloride, as judged by tests with silver nitrate. In most experiW in TiCh and 0.1 M in H2S04 ments a solution 0.01 i was prepared in 4- to 40-1. quantities by adding the
requisite quantities of 20% solution and sulfuric acid, 6.2 and 5.55 ml., respectively, per liter of solution, to the distilled water. For studies with the pure anhydrous grade, the dry compound was weighed and dissolved in a few hundred milliliters of water and the requisite quantity of sulfuric acid. These operations were performed in a drybox filled with nitrogen. Some hydrochloric acid was evolved. The concentrated solution was removed and diluted to the desired concentration with distilled water which had been boiled and cooled under nitrogen. A separate 40-1. carboy contained 0.1 M hydrogen peroxide, made by diluting the 307, reagent with distilled water in the ratio 10.8 ml./l. of solution. The hydrogen peroxide used conformed to 1955 A.C.S. specifications, involving less than 0.005% residue and less than 0.003% free acid, and presumably contained no stabilizers. Tap water from the Washington, D. C., supply gave apparently the same results as distilled water but was ~ _ _
(1) Based on research supported by the National Aeronautics and Space Administration. (2) W. T. Dixon and R. 0. C. Norman, J . Chfm. SOC.,3119 (1963); 362.5, 4850, 4857 (1964). (3) (a) W. T. Dixon and R. 0. C. Norman, Il'ature, 196,891. (1962); (b) J. T. Pearson, P. Smith, and T. C. Smith, Can. J. Chem., 42, 2022 (1964). (4) H. Fischer, Z.Naturforsch., 19a, 866 (1964). (6) A. Stahler, Ber.,38, 2620 (1905).
Volume 70,,Vumher 1 Janlrary 1966
48
not used in these studies. The more important contaminants in this supply are indicated by results taken from a representative analysis (U. S. Geological Survey, sample drawn Aug. 25, 1961) : Fe, 0.0 p.p.m.; Mn, 0.0; HCOa-, 106; C1-, 15; F-, 0.9; Nos-; 1.2; hardness a,s CaCOs, 137; dissolved solids (ISO"), 108; and pH 7.9. For qualitative experiments with organic substrates, if is Iikely that any municipal water supply, reasonably low in heavy metals and organic matter. wjli serve satisfactorily after adjustment of pH. The e.s.r flow system consisted of a Varian e.s.r. apparatus with 100-kherta field modulation, a flat quartz flow cell, and a Varian mixer. Equal volumes of the titanous and peroxide solution in twin carboys were drawn separately through coils of glass tubing, 0.8 by 100 cm.. in a 10-1. thermostatic bath, then v i a Tygon plastic tubing to the mixer, flow cell, small exit flask, and a receiver made of a 40-1. glass carboy protected by screening and a galvani~ed iron can. Connections between the bath, flow cell, and exit flask were wrapped with insulation. Pressure difference was provided by a water jet aspirator and needle valve air leak and was measured on a mercury manometer, corrections being applied for difierences of solution level. Initial flow calibration was a t 28" in the range of 1.5 to 70 cm. and 0.3 to 8.6 see.-'. The flow calibration was transferred to other teioperatures by the use of the similarity principle6to obtain the flow rate v. It was convenient to compute the ratio AI" = A€'IJ-~-~,find v' = v/qr from the 28' calibration chart, and compute v = v'qr, where qr is the water viscosity relative to that at 28". Preliminary observations of color change with indicator and alkali showed apparently satisfactory mixing a t flow Fpeeds of 1 cm.a/sec. The "dead" volume between mixer orifices and interior of flat cell was measured as 0.1 cm.3. Flow speed WBB converted to nominal time after mixing by the formula t(sec.) = 0. I (em."/v( em. see. -l), To extend the available time scale, use was also made of dela,y chambers, of volumes 0.523 cm.* (approximately 0.3-cm. diameter by 6-cm. length), and 3.4 cm.a (approximately 0.6 cm. by 12 cm.), which were inserted between the mixer and the flow cell. The e.s.r. observations were ordinarily made at about 10-mw. microwave power and 0.4gauss magnetic field modulatjon, which resulted in noticeable but slight modulation broadening. For calibration a t a given temperature, areas under the absorption curve were computed by graphical double integration. The ultimate standard for areas was the highest-field hyperfine peak of LM aqueous manganous sulfate. Both line The Journal of P h y s h l Chemistry
F. SICILIO,R. E. FLORIN,AND L. A. WALL
width and cavity sensitivity were temperature dependent. For magnetic field intervals and reference points the hyperfine lines of FrBmy's salt, potassium nitroso disulfonate, NO(S03K)2 were used, assuming g = 2.0054 f 0.0004 and a = 13.0 f 0.1 gauss.' This salt was, incidentally, too unstable for area calibration in our hands. The dry preparation, after 2 weeks of storage in a refrigerator, gave off nitrogen oxides. A fresh 2cf solution assayed about 11% of the nominal content of radical ions and decayed at a rate that was initially moderate but increased with time. Concentrations a t a fixed temperature were compared using peak heights. The g value was estimated by taking the spectrum of a mixture of FrBmy's salt and hydroxyl in the following manner. A small amount of FrBmy's salt (ca. 0.5 g./l.) was added to the peroxide solution, while the titanous solution was shut off. A scan was made through the high-field and middle peak of the ionradical. Without interrupting the scan, the titanous solution was turned on immediately, the composite spectrum showing two ion-radical peaks and the hydroxyl peak.
Spectra Results. Spectra of the hydroxyl radical are shown in Figure 1. At sufKciently low modulation, the spectrum consists of two incompletely resolved peaks, the major peak being a t g = 2.0128 and the minor peak a t a field 2.4 gauss higher, i.e., at g = 2.0114. Attempts to match the spectrum by superpositions of two Lorentz peaks met with mediocre success. The minor peak is possibly of more complicated structure. The derivative maximum-to-minimum peak width is 0.7 gauss at 28" (Figure la) and 1.5 gauss a t 58" (Figure lb) in the mixture used in most experiments; it also varies with concentration (Figure Id-f) and with either the origin of the TiCla or the amount of oxygen in the TiClasolution (Figure ICvs. Figure la). At higher Ti3+ concentrations, the minor peak becomes major if the amount of Hi304 is held constant (Figure lf), but addition of extra HzSO~in proportion to the Tic13 suppresses this effect and slightly broadens the peaks. The minor peak occurs also when excess hydrochloric acid is present and when titanous sulfate is used in place of chloride in a chloridefree system. The relative intensity of the minor peak is not, greatly different from that, observed in the usual mixture. (G) G . G. Brown, Ed., "Unit Operatiom," John Wiley and Sons, Ino., New York, N . Y., 1960, pp. 132-141. (7) G. E. Pake, J. Townsend, and 5. L. Weissman, Phys. Rev., 85, 683 (1952).
49
KINETICSOF HYDROXYL RADICAL IN AQUEOUS SOLUTION
Aa
+,+ -
-
Figure 1. The e.s.r. spectra from titanium salts and hydrogen peroxide: (a) 28", Ti*+ 0.01 M , HZOZ 0.1 M; (b) 58", same concentrations as in (a); (c) 28", Tia+from anhydrous salt, oxygen excluded; (d) 28", Ti*+0.003 M , H z O 0.1 ~ M ; (e) 2 8 O , Ti3+0.03 M , HzOz0.1 M; (f) 2 8 O , Ti3+0.09 M , HZOZ 0.1 M . Length of bar indicates 5 gauss. Field increases t o right.
As with the chloride, the relative intensity of this peak is increased by large additions of titanous sulfate and also, but less effectively, by additions of titanium sulfate preoxidized with hydrogen peroxide until colorless. Discussion. It will be tentatively assumed that the major peak is due to the hydroxyl radical. The species giving rise to the minor peak at higher field is unknown. Its independence of chloride concentration suggests that it is not a hydrated chlorine atom. It is apparently not derived from the inhibitor in the technical solution since the same peak is shown when pure anhydrous Tic13 is used. It is probably not the Hot. radical, for, although the g value (2.0128) is not too far from the reported8 g = 2.016 for HO2., the line width is much less than the reported 27 gauss. Possibly some of the .OH is coordinated to the Ti4+ ion with resultant modification of g value. It is well known that Ti4+coordinates strongly with hydrogen peroxide to form a yellow complex16and therefore some association with hydroxyl is perhaps conceivable. The coordination may be to a partially hydrolyzed species, which is not abundant when the H2S04 concentration is high. Another possibility is HO2. associated with Ti4+and differing strongly from the same radical in the presence of Ce4+. The strong dependence upon titanous rather than titanic ion might suggest coordination of .OH to titanous ion. This seems unlikely because of the presumed rapid oxidsr tion, Ti3+(.OH) --+ Ti4+(0H-). More conservatively,
we can consider that there is a strong pair correlation between the -OH radical and the Ti4+ ion formed simultaneously with it and that association with more distant Ti4+ ions occurs detectably, but much less frequently. A different assignment of the two peaks is given by Piette, Bulow, and L~effler,~ on the basis of extensive kinetic evidence. They attribute the minor, highfield peak to the hydroxyl radical and the major peak to the radical HOz., appearing at a different field from the H02. coordinated to cerium reported by Sat0 and Bielski.s The principal objection to this assignment is related to the consumption ratio, which is 2Ti3+/H202,within 575, and to the very small oxygen vield. If HO,. is kineticallv imDortant during most of the reaction, 0 2 should be an important product and a larger proportion of peroxide should be consumed.l0Il1 This objection loses its force if applied to the final few per cent of reaction, which may be all that is observed. If the major peak is HO2., it is possible that the great line width of H02. from Ce4+ and H20zis due, not to proton exchange at intermediate rates, but to very short relaxation time involving the magnetic moment of coordinated Ce*+. The same effect may be responsible for the failure to detect 'OH in mixtures of Fe2+ and HZOZ,~ even when the radical possibly is present. Weak spectra of the radical CH20HCHOHcHCH20H have been detected in mixtures of 2-butene-1,4-diol1 ferrous sulfate, and hydrogen peroxide, when the same mixture lacking the butenediol showed no hydroxyl peak. "
I
u
Kinetics Results. The concentration of radicals is shown as a function of flow rate in Figures 2 4 . The plots show reciprocal peak height us. reciprocal flow rate. Because of varying peak width and sensitivity, the conversions to concentration depend on temperature. Table I shows derived concentrations and secondorder rate constants for the presumed disappearance reaction. In the presence of added magnesium sulfate, 0.8 M final concentration, the ratio of minor to major peak height is decreased, the major peak height is incremed by 50% at high flow rates, and the rate of disappearance of both peaks at low flow rate is increased by a larger factor. The addition of lo-* M (8) E. Saito and B. J. Bielski, J. Am. C h m . SOC.,83, 4467 (1961). (9) G.Bulow, private communication. (10) F. Haber and 5. Weiss, Proc. Roy. SOC.(London), 147, 332
(1934). (11) J. H. Baxendale, M. G. Evans, and G. S. Park, Trans. Furaduy SOC.,42, 155 (1946).
Volume 70,Number 1 January 1966
50
F. SICILIO, R. E. FLORIN, AND L. A. WALL
Table I : Concentrations and Rate Constants for Second-Order Disappearance Temp., OC-.
Highest concn., M k , M-1 set.-' a
20
28
58
68'
1.6 x 3.2 X lo6
1 . 6 X 10" 4.5 x 106
3 . 9 x 10-6 3 . 0 X lo6
3 . 9 x 10-6 1.03 x 107
Traversed volume V = 0.1 cm.*; otherwise, V = 0.623 and 3.5 cm.8.
0.8
t
0.3
0.6
I/h
0.4
I
I
I
I
I
2
4
V/v,sec
0.2
Figure 2. Concentration of hydroxyl radicals as a function of flow rate a t 27.2'. Ordinate: reciprocal peak height; 1 unit = 6.9 X lo6 M-l. Abscissa: computed time, V/u; V = volume traversed after mixing, cm.3; v = flow rate, ~ m . ~ / s e c .@,; V = 0.623 ~ m . 0, ~ ; V = 3.5 cm.'
I ' 1 I
I 0
I
I
2
I
I 4
0
9/r 0.2 V/v,sec
0.4
-
I
V / Y I se c
Figure 3. Concentration of hydroxyl radicals as a function of flow rate a t 20". Units are as in Figure 2. One ordinate unit = 6.2 X lo6 M-l. 0, V = 0.623 cm.*; 0, V = 3.5
manganese sulfate, ten times the 0.001% manganese impurity in the magnesium salt, had essentially no effect on the reaction. The highest concentrations are reached at 58", and a maximum in concentration us. time occurs at the two lower temperatures. Presumably, the rate of generaThe Journal of Physical Chemistry
c
Figure 4. Concentration of hydroxyl radicals as a function of flow rate a t 58". Units are as in Figure 2. One ordinate unit = 1.3 X 108 M-1. @, - - - -, V = 0.10 cm.8; 0, , V = 0.623 cm.a.
I
I
.
tion is more temperature dependent than the rate of disappearance. At very low flow rates, Le., long times, the reliability of data is suspect, not only because of the lower precision of measurement, but also because of the possible approach to streamline flow in the mixer, so that mixing may be incomplete even after the stream reaches the e.s.r. cavity. One must also emphasize the wide disagreement between data taken with and without the delay chamber, illustrated in Figure 4. Similar disagreement occurred at all temperatures, the general effect being that a large delay volume expands the apparent time scale. For example, without a delay chamber, the minimum corresponding to that shown in Figure 2 occurred at 0.05 sec. Discussion. The reduction of flow rate to time, t = V/u, is strictly valid only if turbulent flow and a flat velocity profile are maintained in the whole region from the mixer to the sensitive region of the flow cell and if mixing is complete in a small initial part of this region. Consideration of the dimensions of the ap-
KINETICSOF HYDROXYL RADICAL IN AQUEOUS SOLUTION
51
paratus (mixer hole diameter 0.05 cm., channel diameter mental arrangements have been discussed by Rough0.17 cm.) suggest that flow is certainly turbulent in the tonlaand also in a recent symposium.14 mixer holes at most of the flow rates used, 0.5 to 7 ~ m . ~ / Reasonable reactions of the system are sec., but is not necessarily so in any other region. Ti3+ H202 -+- .OH OHTi4+ (1) In ideal pure laminar flow in a cylinder, material arriving a t the detection zone would have a wide dis*OH -OH +HZ02 (2) tribution of lifetimes since mixing, ranging from .OH Ti3++OHTi4+ (3) 0.5 V/v upward. The character of the plots in Figures 2 to 4, especially .OH HzO2 +HO2. H2O (4) the superposition of data from the two larger delay HO2. HO2. +HzO2 0 2 (5) volumes and the approach to second-order behavior a t long times, suggests that the time conversion t = V/v HO2. *OH-+- H20 0 2 (6) has some approximate validity. H02. Ti3+ H + +H2OZ Ti4+ (7) If the flow is slightly perturbed by local irregularities, a typical particle might travel in different flow regions Reactions 5 and 6 lead to formation of 0 2 , which is at different stages and experience an averaged transit observed in minor amounts only; therefore, possibly time. It is obvious from the discordant curves of (4) is minor in the present system. The combination Figure 4 that the effective volume of the delay chambers of reactions 1, 2, and 3 leads simply to the observed must be very different from the measured volume. consumption ratio Ti3+/H202 = 2 ‘although other The resultant derived kinetic data can thus differ by combinations are also compatible, e.g., reactions an order of magnitude for different geometric arrange1, 3, 4, and 7 . If it is permissible to neglect reactions ments. 4 to 7 , the rate of formation of ‘OH is given by Part of this discrepancy can be removed by assuming that mixing is complete in all cases and that flow is satisfactorily turbulent at all points in the O . l - ~ m . ~ volume but approximates ideal streamline flow in the larger delay volumes. If one assumes that there is and with excess H202 thorough mixing initially, that subsequently material along different flow lines is not intermixed during -d [Ti3+]- -k1[Ti3+][H202] - k3[Ti3+][0Hl (9) transit, and that material leaves the delay chamber dt in streamline fashion, then the classical description of viscous flow in a cylindrical tube12leads to the relation Further analysis would require numerical methods or knowledge of Ti3+concentrations. de At long times, [Ti3+]is low, and the disappearance C(2A) = A C(A) becomes simple second order. A plot of [OH]-l vs. dA t should become linear, with slope IC2. The extent to which this behavior is realized can be judged from Figures 2 to 4. Points at the largest values of V/v are suspected because of mixing di.6culties. Nominal wherein A is the ratio of flow rate to volume, cm.3 values of k2 estimated from the final slopes in these sec.-l ~ m . - ~C(A) , is the measured concentration a t figures are given in Table I. The order of magnitude flow rate/volume = A , and C(2A) is the concentration is rather low compared to the usually assumed values in a volume element of transit time t = 1/28. Although in the range of log to loll M-’ sec.-l, e.g., in the spur the application of derivatives requires highly precise theory calculations of Kupperman, Dyne, and Kennedy, data, it can be seen that a t the maximum, where dC/and others15J6; and it is certainly lower than the range dA = 0, and a t long times, where A approaches zero, the gross effect is to convert nominal times V/v to (12) G. Joos, “Theoretical Physics,” 3rd Ed., Hafner, New York, half their value and, in second-order plots, to multiply N. Y., 1958, p. 214. the rate constant by a factor of 2 . (13) J. J. W. Roughton, “Technique of Organic Chemistry,” Vol. The data with the delay chamber (open circles) are VIII, Interscience Publishers, Inc., New York, N. Y., 1953,p. 669. presumably more reliable since there is more assurance (14) B. Chance, Q.H. Gibson, R. H. Eisenhardt, and K. I(.LonbergHolm, Ed., “Rapid Mixing and Sampling Techniques in Biochemisof complete mixing within a negligible fraction of the try,’’ Academic Press Inc., New York, N. Y., 1964, pp. 1-54, 125distance traversed. Relevant principles and experi154, 353-370.
+
+
+
+
+
+ + +
+
+
+ + + + +
+
Volume 70, Number 1 Januury 1966
F. SICILIO, R. E. FLORIN, AND L. A. WALL
52
of lo7 to 1011 M-l see.-’ found for diffusion-controlled reactions of neutral molecules or ions of opposite charge.17 The apparent activation energy, E = 11 kcal./mole (46 kjoules/mole), is high for a purely diffusion-controlled reaction. If .OH is really coordinated to Ti4+, ionic repulsion could account for some additional activation energy. The increased disappearance rate in the presence of added salts is a reasonable effect of ionic strength on reactions of species with like charge. The higher signal levels at short times are more difficult to understand but may reflect simply a decreased mobility of water molecules in the hydration field of the salt, with consequent reduction of dielectric loss and increase of cavity &. The pre-exponential factor, 4.5 X M-l sec.-l, is somewhat higher than the normal range of loll; a similar example is the value of 1.6 X 10l2 computed from the first item in Benson’s table1’ for I I --c 12. These deviations could be the consequence of a moderate systematic error, +2 kcal., in the activation energy. The effect of repulsion between like charges, according to transition state theory, is usually to lower the preexponential factor by a factor of about lo2 for each unit of ZAZB.17b,18A direct analysis of diffusion in a potential fieldlggives v = f v s , where v is the frequency of collisions of ions with a given ion, vs = 8kTn/3q is the Smoluchowski collision frequency for uncharged particles, and the factor f for identical particles reduces to
+
j‘ =
DkTr
In these formulas n is the number of ions per cubic centimeter, q is the viscosity, kT is the average kinetic energy, E is the unit charge in e.s.u., D is the dielectric constant, and r is the radius of the ion in centimeters. The quotient ZlZ2E2/DkTr has a value near 3.5 for univalent ions of ionic radius 2 8. and like charges in water at 300°K. The temperature coefficient of D is critical in the Debye formulation as well as in transition state theory. For the usual static dielectric constant, b In D / b In T = -1.4, DT decreases with increasing temperature, and the end result is a reduced pre-exponential factor and a small negative contribution to activation energy. However, in the frequency range between l O Qand 10l1hertz, which is also the range for d3usive jumps, both the real and the imaginary parts of D are changing rapidly, and the temperature coefficient changes sign.2o When the algebraic value of b In D / b In T is greater than -1, e.g., zero or positive, there will be an increased pre-exponential factor
The Journal of Physical Chemistry
and a positive contribution to the activation energy. For the artificial special case of b In D / b In T = 0, D = 80, the contribution will amount to about 2 kcal. for univalent and 8 kcal. for bivalent ions, and a small increase, 3.5 and 14 times, respectively, over the normal Smoluchowski pre-exponential factor of about 10l1M-l sec.-l. Possibly the same considerations apply to the differently identified and somewhat higher rate constant of OH. If the present OH radicals PietteQfor OH are actually associated with ions, this is probably also true of the OH radicals produced in Fenton’s reagent, and, by analogy, the reaction of Fe2+ Fe3+(HzO),(OH-),OH may be somewhat slower than the reaction of neutral OH with Fez+. Since some of the values in Ferradini’sZ1table of relative reaction rates of OH depend upon the device of setting equal the rate of Fez+ OH in Fenton’s reagent with the rate in a radiation-chemical experiment, where the OH has a better prospect of being uncharged, these values could be subject to revision downward.
+
+
+
Conclusions The e.s.r. spectra of the .OH radical in aqueous solution indicate some molecular complexity, which specificallymay be coordination to the titanic ion. The disappearance rates of the two species seen are practically equal, however, and have only a minor influence upon kinetics. Apparent rate constants for disappearance are of reasonable magnitude. The slowness of generation of OH at low temperatures may sometimes be a complicating factor in studies of other radicals produced by abstraction and addition reactions.
Acknowledgment. The authors wish to thank D. W. Brown and Monick Dousset, guest worker at the National Bureau of Standards, 1964, for their aid in obtaining the electron spin resonance spectra. ~~~~
(16) A. Kuppermann and G. G. Belford, J. Chem. Phya., 36, 1441 (1962),Figure 24. (16) A. Kuppermann, “Actions Chimiques et Biologiques des Radiations,” Vol. 5, M. Haissinsky, Ed., Masson and Co., Editeurs, Paris, 1961,pp. 85-166, and references therein. (17) S. Benson, “The Foundations of Chemical Kinetics,” McGrawHill Book Co., Inc., New York, N. Y., 1960, (a) pp. 501-503, (b) pp. 537,530. (18) K. J. Laidler, “Chemical Kinetics,” McGraw-Hill Book Co., Inc., New York, N. Y., 1950,p. 134. (19) P.Debye, Trans. Electrochem. Soc., 82, 265 (1942). (20) Landolt-Bornstein Tables, 6th Ed., Vol. 2, Part 6, SpringerVerlag, Berlin, 1959, pp. 743, 766, 767. (21) C. Ferradini, Advan. Inorg. Chem. Radwchem., 3, 184, 186 (1961).
