ADDITION AND ISOMERIZATION
IN PHOTOCHEMICAL
REACTIONS
987
ysis a t 200-300"C. can be obtained by the rotating sector method (W. L. Haden, Jr., and 0. K. Rice: J. Chem. Phys. 10,445 (1942); V. E. Lucas and 0. K. Rite: J. Chem. Phys., in press). The fraction of the radicals striking the wall which is and the activation energy removed (accommodation coefficient) is about appears from the later work to be low. The accommodation coefficient apparently does not depend on the concentration of the radicals being removed, so if the reaction is a recombination of these radicals the surface must be fully covered. If it is a combination of the chain-carrying radical with some other radical, complete coverage might not be necessary, but the fraction of the surface covered must not depend on the concentration of the radicals being removed.
PHOTOCHEMICAL REACTIONS BETWEEN BROMINE AND 1,2-DICHLOROETHYLENE
THERELATIONBETWEEN ADDITIONAND
ISOMERIZATION'*a
J . .4. A. KETELAAR, P. F. VAN VELDEN, G. H . J. BROERS,
ANO
H . R . GERSMANN
Laboratory f o r General and Inorganic Chemistry, University of Amsterdam, Amsterdam, Netherlands Received August 10, 1950 INTRODUCTION
The kinetics of photochemical reactions between halogens and olefins have been studied frequently, in the gas and liquid phases. While addition of the halogen to the unsaturated compound usually takes place, several cases are known in which tie-trans isomerization occurs at the same time (cinnamic acid and derivatives, l,Z-dihaloethylenes, etc.). Many authors focussed their attention on the addition reactions (2, 3, 6, 8, 19, 17, 27; see also the survey in reference 22), while others studied the isomerizations (12, 22). However, we know of no attempt to study both processes simultaneously in such a way that the two reactions can be followed in one and the same experiment. The present investigation was undertaken in order to search for a relation between the kinetics of addition and isomerization. The existence of such a relation is suggested by the fact that in both cases the kinetic behavior can be explained satisfactorily by assuming the occurrence of free radicals of the type A-X(whete A designates the olefin and X a halogen atom) as intermediates. For a survey, we refer for instance t o reference 22. The rather easily Presented before the Symposium on Anomalies in Reaction Kinetics which waa held under the auspices of the Division of Physical and Inorganic Chemistry and the Minneapolis Section of the American Chemical Society at the University of Minnesota, June 19-21,1950. More complete details of this work are t o be found in a thesis by the second author, Amsterdam, 1950 (25).
'
988
KETEL.4AR, VAN VELDEN, BROERS AND QERSMANN
accessible compound 1,2-dichloroethylene and the halogen bromine seemed to be the most suitable for our purpose. The rate of isomerization can be followed easily by dielectric constant measurements, as the two isomers differ considerably in this property (trans, 2.14; cis, 9.28 a t 25OC.). The rate of the addition reaction was determined by measuring the change of light absorption, caused by the bromine, as a function of time. All experiments were carried out in the liquid phase a t 25"C., using the green mercury line (5461 A.) for illumination. Special emphasis must be laid upon the fact that no solvents were used. Reactions between free radicals and solvent molecules, obscuring the true picture of the proper reaction studied, are by no means uncommon. In our case, the use of solvents appeared to be unnecessary, provided the concentration of the bromine was kept low enough to allow us to follow the reactions (the bromine concentrations did not exceed 0.05 mole/ liter). As pointed out many years ago by Chavanne and coworkers (9, 26) and confirmed by later investigators in this field, the presei.:; even of very small quantities of oxygen influences considerably the course of both isomerization and addition. After some of our preliminary experiments indicated the same conclusion, the reproducibility of addition and isomerization reactions separately being very poor even though nitrogen was used in all operations, further experiments were carried out so as to exclude oxygen as completely as possible. Comparison of the two series of experiments led to clarification of the nature of the reaction mechanism, which will be dealt with in a later section.
EXPERIMENTAL^ 1. Materials Cis- and trans-1 ,2-dichloroethylenes were prepared and purified in the manner described previously (15, 16, 25), following the methods also used by Bull (7), Wood and Dickinson (30), and others. Dielectric constant measurements were applied in controlling the purity of the samples. Several physical properties of the pure isomers were determined in this Laboratory (15, 16, 25). A careful comparison with values given in the literature has convinced us that the purity of our samples was equalled only by that obtained by Bull (7). We list only the following physical constants for our material: CB",
Boiling point (760 mm. of mercury) . . . i Density (25°C.). . . . . . . . . . . . . . . . . . . . Dielectric const. (25°C.) . . . . . . . Refractive index (ne).. . . . . . . . . . . . . . .
'
,I
47.7"C. 1.2463 2.145 1 ,44255
I
Ii
cis
60.5"C. 1 ,2757 9.28' 1.44615
* T h i s value differs somewhat from that formerly given ( e = 9.31) (15), which is only the result of a better evaluation of calibration measurements (25).
Bromine (Merck's c.P.) was treated with potassium bromide, distilled, dried over phosphorus pentoxide, and distilled once more in a dry glass apparatus. a Only a brief description is given here; for full details the reader is referred to reference 25.
::a.t.;; i I
'
.
*
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
989
1. Apparatus
The reaction vessel (figure 1) consisted of a cylindrical glass absorption cell
C (length, 29 mm.; diameter, 22 mm.) in which two parallel rectangular platinum sheets t, measuring 25 x 11 mm. and held 8 mm. apart by small glass rods (see front view, figure I), served as condenser plates for measuring the capacitance of the cell when filled with reaction mixture or standard liquid. Platinum wires connected the sheets to copper leads p, ending in the sockets q. With the exception of the plane parallel glass windows r through which the light beam could pass, C was painted black on the outside. A known volume of liquid from the calibrated vessel F could be brought into the cell without the possibility of leakage of air or moisture into the apparatus, dry nitrogen being ap-
FIG.1. Reaction vessel plied during all operations (a description of the process of filling and emptying, using a rather complicated stopcock system connected to R, P, and T, for direction of the nitrogen through the different glms tubes, is given in reference 25). When emptying the reaction cell, the liquid was caught a t Q, this outlet being closed otherwise. The apparatus was cleaned by rinsing with pure sodiumdried petroleum ether and subsequent drying with a stream of dry nitrogen. During the experiments made with rigorous exclusion of oxygen, the ground connection E was replaced by an all-glass seal. As it was impossible to install a stirrer in the cell C , the reaction mixture was kept in continuous movement during an experiment by alternately applying a slightly increased or reduced pressure to the gas in tubes A and B, using a pumping system connected to the above-mentioned stopcock system. This set-up proved to be satisfactory in 6.. keeping the mixture homogeneous (25). b 5.
'.
990
KETELAAR, VAN VELDEN, BROERS AND GERSMANN
The reaction cell a3d its immediate connections were thermostated a t a temperature of 25°C. f 0.05” by means of an electrically heated and regulated bath of colorless paraffin oil. Glass windows in the metal walls of the thermostat permitted the light beam to pass through the reaction cell and impinge on the photocell used for the absorption measurements. Dielectric constant measurements were made by means of the apparatus used previously (15) and described in detail by Cohen Henriquez (10). For calibrations, carefully purified and dried liquids of known dielectric constants were used, standard values a t 25°C. being taken from the work of Dekker ( l l ) , i.e., benzene (e = 2.278), monochlorobenzene ( c = 5.641)) and 1,2-dichloroethane (e = 10.433). Evaluation of the calibration data was carried out in a more satisfactory manner than before, correcting empirically for deviations from a linear relation between the dielectric constant of the medium and the capacitance of the cell (25). The light source used was a “Heraeus” mercury-vapor arc (120-v. D.c.). Isolation of the green mercury line (5461 b.)was achieved by means of a type-77 Wratten filter, combined with a 2-cm. layer of 0.4 M aqueous cupric chloride solution to remove the red light to an extent sufficient for our purpose. An optical system provided a sufficiently homogeneous and parallel light beam, the intensity of which could be varied by means of a rotating sector disk or by gray filters of known transmission properties. For absorption measurements, two “Eel” selenium barrier photocells were used. One of these (I) was placed directly behind the reaction cell in a special recess of the thermostat and served for measuring the light absorption by the bromine in the reaction mixture. The second (11) was used in controlling the intensity of the light emitted by the mercury arc during the experiment. Both cells were connected to a wiring and switching system in which a high-sensitivity galvanometer‘ was included. The current put out by either cell could be measured without altering the external load resistance (320 ohms) of the other cell, since it is known that illumination of this type of cell under “open circuit” conditions may not give reproducible results. The load resistance of 320 ohms proved to be sufficiently low to secure a linear relation between intensity of illumination and current output, minor deviations being corrected easily. The light intensity never surpassed 1.4 x lo-‘ watt/cm.* The sensitivity of the galvanometer was 2.5 X lo-’’ amp. (1 mm. deflection a t 1 m. distance). All wiring was made with lead-shielded and grounded cable. Thermocurrents were too small to be troublesome. Cell I was calibrated by comparison with a calibrated thermoelectric pile‘ and was found to give a current of 3 x lo-* amp. when illuminated by green mercury light of intensity lov6natt/cm.’ under an external load resistance of 320 ohms. (For further details see reference 25.) The nitrogen used in the preliminary experiments was taken from commercial tanks and passed through an alkaline pyrogallol solution, after which it was Type Kc “double coil”, Kipp and Zonen, Delft, Yetherlands. Our thanks are due to M r . F. A . Rodrigo of the Physical Laboratory, University of Utreeht, for carrying out this calibration. 4
ADDITION AND ISOMERIZATION I N PHOTOCHEMICAL REACTIONS
991
carefully dried. After it was considered necessary to remove oxygen as thoroughly as possible, a gas lift pump was constructed similar to that described by Uhrig, Roberts, and Levin (24), in which the gas, mixed with ammoniacal ammonium chloride solution, was led over copper ribbon. Ammonia was then removed by leading the gas through a fine fritted-glass filter into a solution of sulfuric acid in water (l:l), and it was dried with concentrated sulfuric acid and phosphorus pentoxide. The dried pure nitrogen was carried by glass tubes to all parts of the equipment where it was needed. Rubber connections were avoided as much as possible ; when absolutely necessary to prevent breakage, specially prepared vacuum rubber tubing was used. All connections were vacuum tested. A 0.5 per cent solution of indigo carmine in water, reduced by zinc as described by TABLE 1 The relation between dielectric constant ( e ) and composition for mizlures of cis- and trans-1 .d-dichloroethulenes at 86OC. CiS-lSOXER
I
CiS-lSOMEE
I
~
mole pn cenl
mole pn
~
Lcnl ~
0.0 4.9 6.0 11.7 12.3 17.35 19.3 20.4 22.15 25.4 32.6 38.4 39.4 44.7 46.4 52.4
, ~
I I,
i 1
I ~
~
i
2.14s 2.42 2.48 2.78 2.82 3.115 3.23 3.28 3.39 3.60 1.085 4.45 4.50 4.89 5.00 5.43
57.1 57.9 61.1 62.4 64.8 65.7 69.2 71.2 75.4 79.0 70.6 85.5 90.8 93.2 96.05
100.0
I I ~
i ~
1
5.80 5.87 6.08 6.19 6.34 6.43 6.75 6.89 7.21 7.50 7.55 8.01 8.48 8.68 8.93 9.28
___-___
992
KETELAAR, VAN VELDEN, BROERS AND OERSMANN
in determining the content of cis-isomer from the dielectric constant of the reaction mixture. A complete experiment was carried out in the following way (only the technique used with careful elimination of oxygen is described here; for complete details see reference 25) : A sufficient portion of freshly distilled dichloroethylene was freed from air by passing a current of “oxygen-free” nitrogen through the liquid and boiling. The all-glass apparatus provided for distillation of the “oxygen-free” liquid into a glass container, from which a known volume was subsequently brought into the calibrated vessel F (figure 1) of the reaction apparatus; during all these operations, the liquid was in contact only with the dry “oxygen-free” nitrogen. After the reaction cell C had been filled with the liquid, its dielectric constant was measured, the content of cis-isomer thus being known from the calibration curve (cf. table 1). At the same time, the liquid was illuminated and the value of J O (intensity of light falling upon photocell I with zero bromine concentration) was checked. Meanwhile, a second portion of the same liquid was expelled from the container into a special mixing bottle, made from dark red glass and filled with nitrogen. This mixing bottle already cont,ained the bromine to be used, enclosed in a glass “bromine bulb” with very thin walls. A special technique had been developed for filling these bulbs with oxygen-free bromine (25), of known weight. After weighing the dichloroethylene in the mixing bottle, the bromine bulb was broken in the liquid, without danger of contamination by air or moisture. From this moment, the room was kept dark or at most dark red light of low intensity was used. After thorough mixing by shaking the bottle, the reaction mixture of known composition was forced by nitrogen into the reaction apparatus (F in figure l ) , which meanwhile had been emptied, cleaned, and dried. After transfer of the liquid into the reaction cell C in the usual way, reaction rate measurements could be started.6 Immediately after illumination of the reaction mixture was started, readings of the galvanometer deflection caused by the current output of photocell I and measurements of the capacitance of the reaction cell were made a t regular intervals. At certain intervals, the intensity of the mercury arc was controlled by connecting photocell I1 to the galvanometer. Corrections for deviations in J o were small in most experiments and could be easily accounted for. Preliminary experiments, in which oxygen had not been rigorously excluded, had a rather long duration (up to 4 hr.) while the later experiments were ended after about 20-30 min. in most cases, the rates of reaction sometimes even being too fast for measurement. An experiment was stopped as soon as the velocities of the addition and isomerization reactions became negligibly small. The reaction mirture was then illuminated for some time by unfiltered mercury light, the last traces of bromine disappearing and isomerization equilibrium being reached in most cases. The equilibrium mixture of dichloroethylene was found to contain 81.6 f 0.2 per cent of the &-isomer, by averaging the results‘of a large number of experiments a t 25°C. (see reference 25). 8
For more details as to the construction and use of the mixing bottle, see reference 25.
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
993
EVALUATION O F EXPERIMEXTAL DATA
From the measurements of light absorption in the reaction mixture, the bromine concentration was calculated using the Lambert-Beer relation in the following form:
J
=
Jo x
Here J = the intensity of transmitted light, K = the molar extinction coefficient, d = the length of the light path in the reaction mixture, and C = the bromine concentration in moles per liter. As we were only interested in changes of the bromine concentration with time, K and d were taken together into one constant K and the product KC was used in all calculations and graphical representations:
KC
=
log ( J o / J )
Measurements of extinction coefficients in a Beckman spectrophotometer proved Beer’s law to be valid for the concentration range used; moreover, value: of K were substantially the same for both isomers (log ( K ) = 1.42 at 5461 A.). From the dielectric constant measurements, &:trans ratios ip the reaction mixture were calculated by means of the calibration curve (cf. table 1). I t might be argued that the presence of bromine and/or the addition product 1,2-dichloro-l,2-dibromoethane in the mixture makes this procedure invalid. However, it can be shown (25) that the influence of the very small quantities of these compounds (0.05 mole/liter or less) involved in our measurements can be neglected within the limits of accuracy obtained. The Onsager-Bottcher equation for liquid mixtures, relating the dielectric constant to the polar properties of the compounds involved, was used in calculating the corrections to be spplied, which never were larger than about 0.2 per cent of the content of cisisomer. Quantum yields of reactions were calculated in the usual way, using the data on sensitivities of photocell I and of the galvanometer. RESULTS AND DISCUSSION
In figure 2a the dependence of bromine concentration (expressed in K C ) on time is given for two typical experiments of the “old” series (not “oxygenfree”). In figure 2b the corresponding isomerization reactions are presented :
(.
=
cis
+ trans
The occurrence of “induction periods” as well as the poor reproducibility for experiments starting with almost the same initial concentrations suggested that impurities such as oxygen in the reaction mixtures played a part in the reaction mechanism, a result which is in agreement with the findings of former workers for reactions of this kind.
994
IiETEL4AR, VAN VELDEN, BROERS AND GERSMANN
The effect of oxygen shows itself clearly in figure 3, where addition curves are given from the “old” series (not “oxygen-free”) (experiments 14/7 and 10/8) as well as from the “new” (“oxygen-free”) series (F and P). Induction periods still exist. However, lowering of the oxygen content evidently produces considerable acceleration of the process. This behavior is the more pronounced as for both F and P the intensity of illumination was reduced for some time to 1.0.
0.8.
T
o*6’
8 0.4
0.2
1
75 100 126 MINUTES 4 FIG.2a. The course of addition for two experiments of the “preliminary” series (oxygen not rigorously excluded). KC = measure of bromine concentration (see section on evaluation of data); t , = starting point of illumination; a = induction period; b = point of largest reaction velocity (end of induction period). Experiment No. 27/5: (Bn), = 0.0125 mole/ liter; z, = 0.018. Experiment S o . 2 8 / 5 : (Br?), = 0.0127 mole/liter; z, = 0 013. 25
50
one-fourth of the initial value (see breaks in curves), the latter being the same for all measurements (1.4 X watt/cm.’). The isomerization curves, not drawn here, are similar. In fact, isomerization and addition curves for a single experiment were always found to show the same peculiarities. Moreover, another effect is revealed. Comparing experiment 14/7 with experiments F and P in figure 3, we see that the addition rates during the first parts of the reactions are not strikingly different. However, as soon as the reaction 14/7, which was started with a mixture chiefly consisting of trans-
ADDITION AND ISOMERIZATlON I N PHOTOCHEMICAL REACTIONS
995
isomer, has reached a point where the cis-isomer begins t o predominate (owing to the simultaneous isomerization reaction), a strong retardation occurs. Likewise,, the reaction 10/8, started with a mixture of high content of cis-isomer, proceeds very slowly. It thus appears that the retarding effect of oxygen for some reason is the more pronounced the higher the cis content of the reaction mixture. This effect, which has been mentioned also by Chavanne et al. (9, 26),
0.4
t
0.3
x 0.2
0.1
25
50
76
MINUTES
100
+
125
FIG.2b. The course of isomerization for two experiments of the “preliminary” series (oxygen not rigorously excluded). KC = measure of bromine concentration (see section on evaluation of data); t i = starting point of illumination; a = induction period; b = point of largest reaction velocity (end of induction period). Experiment No. 27/5: (Brr)i = 0.0125molebiter; z i = 0.018. Experiment N o . % / 5 : (Br3)i 0.0127 mole/liter; z i = 0.013.
-
can be further illustrated by the data in table 2. Here quantum yields, Q, of the addition reaction are listed for various experiments and for times a t which both intensity of illumination and bromine concentration (KC = 1.00), and thus the total quantity of absorbed light quanta, were the same. The very high values of Q (up to IO’) clearly indicate that we have to do with a chain reaction. For reactions in liquid systems such quantum yields are unusually high indeed. As is generally the case in dealing with chain reactions, good agreement
996
KETELAAR, VAN VELDEN, BROERS AND GERBMANN
between like sets of experiments could not be reached, even after the most complete removal of oxygen. However, some relation between isomerization and addition seemed to exist for each experiment (compare the curves in figures c ti
VCiS
1.8 1.6
I .4 1.2
7 Y
1.0
0.8 0.6
0.4
0.; 10
20
30
40
50 MINUTES
60
70
80
90
4
FIG.3. Rates of addition as a function of oxygen content and content of cis-isomer. startingpoint of illumination. Experiment KO.14/7: zi = 0.02,K C i = 1.83,appreciable
ti =
-
oxygen content. Experiment KO.10/8: xi 0.82, K C i = 1.54, appreciable oxygen. Experiment No. P : z, = 0.85, K C i = 1.53, utmost exclusion of oxygen. Experiment No. F : zi = 0.08, K C , = 1.70,utmost exclusion of oxygen.
2a and 2b), peculiarities in kinetic behavior always being reflected in both processes. Now, considerations concerning the types of intermediate reactions most likely to occur led to a rather simple equation expressing the rate of isomerization as a function of the rate of addition or vice versa. This equation was tested and was shown to hold for each individual experiment, independent of the coltditions, such as oxygen content, under which it wos carried out.
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
997
We write: Br2
+ hv + 2Br
+ C + ABr ABr Br + T Br
Scheme A: ABr
+C
+ Br
ABr+T+Br ABr
+ Brz + ABrz + Br
TABLE 2 Quantum yields of addition reactions under comparable conditions AT
L. . . . . . . . . . . 3/8.. . . . . . . . . . . . . w.................... H. . . . . . . . . . . . . . . . F......................
++,
65 62 44 39 26 23
KC
-Q
1.00
O X Y G I N MSTENT.
2 x 10‘ 3 x 104 6 X 10’ 1.5 X 10‘ 1.5 X lo6 1.5 X lo* 4 x 104 2 x 10‘ 3 x 10’ 3 x 104 1 x 106 3.5 x 104 1 x 104 2 x 104
+,
* “old” series, “high” oxygen content; -, “new” series, “oxygen-free”; “new” series, oxygen added by shaking in contact with air before starting the experiment. Here C represents the cis-isomer, T the trans-isomer, and ABr is written for the radical C2HzC12Br--. Reaction 4 essentially represents the addition process, while isomerization is described by reactions 2’ and 3‘ in connection with reactions 2 and 3. Some of the above reactions had been used already by others (e.g., Berthoud et al. (3)). I t must be remarked that we have not yet included chain-breaking reactions in this scheme. These will be dealt with further on. The relation between addition and isomerization already follows from reactions 1to 4.However, it can be stated that in our caseundiluted 1,2-dichloroethylene being used-the chain-breaking reactions most likely will not involve bromine atoms. Every bromine atom produced according to reaction 1 is surrounded by unsaturated molecules with one of which it can react immediately, forming a radical ABr. The actual concentration of free bromine atoms therefore will be exceedingly small, even in comparison with the ABr concentration.
998
KETELAAR, VAN VELDEN, BROERB AND Q E R B U N N
From reactions 2, 3, 2', and 3' it follows that:
+
in which z = (C)/[(C) (T)]. If we assume that the rate constants kZ and ka have the same value (a justification for which will he dealt with in the following), the second term in equation I disappears. Combination with the condition for isomerization equilibrium, dz/dt = 0 for 5 = z., then leads to:
On the other hand,, the addition process can be expressed by
-
dt
=
iJ.
+ kr(ABr)(Br2)
which equation follows from reactions 1and 4.Here J. = Jo - J is the absorbed quantity of light a t time t , while the factor i is equal to 1, if every absorbed light quantum results in dissociation of a bromine molecule and J,, (Br2), and (Br) are expressed in corresponding units. I n view of the high quantum yields (see table 2) the term iJ. in equation 111 may be neglected in comparison with kr(ABr)(Brz). If we then eliminate (ABr) from equations I1 and I11 we obtain: dr
This relation between isomerization and addition was tested for each experiment in the following manner: Putting [(C) (T)] = constant (which is true as a first approximation, pure cis-isomer having a concentration of 13.2 moles/ liter and pure trans-isomer a concentration of 12.9 moles/liter a t 25°C.) and integrating we get:
+
- x,) = H (z, - z) = H
In (z In
+ P1 In (Br2) + Pp In (Br2)
where
(z (z
> 5,) < 2,)
(Va) (Vb)
+ 1) + (TI1
H = kV(K. k4[(C)
and PI and PZ are cpnstants determined by the initial concentrations of the components in the reaction mixture (z = xi; (Brz) = (Brz),). For experiments starting with trans-isomer equation Vb is used; for experiments starting with cis-isomer, equation Va. According to these equations, plots of log 1 x - xI I against log (Brz) should
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
999
result in straight lines. In figures 4a, 4b, and 4c the experimental confirmation is to be found. For each experiment, corresponding values of 5 and of the bromine concentration (expressed in terms of K C ) were taken from the 2-t and KC-t curves, examples of which were shown for instance in figure 2 or 3. For zb the value 0.816, resulting from the measurements of isomerization equilibrium, was used. Preliminary experiments in the presence of oxygen are shown in figure 4c, while in figures 4a and 4b the “oxygen-free” experiments are assembled. For comparison, experiment 11/8 is reproduced in figure 4a as well as in figure 4c. Changes in intensity of illumination are indicated by vertical arrows. Numbers along the curves give the time in minutes after the illumination of the reaction mixture was started. The curves in figure 4 very strikingly illustrate that for all experiments, regardless of the oxygen content of the reaction mixtures, the same relation between isomerization and addition holds over the whole range of time of the experiment (apart from the small bending in the curves which occasionally occurs just after the starting points), although the “old” experiments with oxygen present took considerably longer time than those of the “new” series. Comparing the slopes of the straight lines (indicated by the number a t the end of each line) we see that these tend to increase somewhat with increasing content of cis-isomer in the starting mixture. However, no particular significance need be attached to this behavior, as the slopes, particularly in the case of the experiments with the cis-isomer, are rather sensitive to the choice of the value of re.For instance, if we take 5. = 0.814 instead of 2, = 0.816 (the difference being within the limits of accuracy), the mean value of the slopes of the “trans” curves (E, F, G, H, W) changes from 0.67 to 0.69, while that of the slopes of the “cis” curves (A, B, C, D) turns out to be 0.68 instead of 0.74. The comparatively low values of the slopes of the “trans” curves in figure 4c seem to be related to the increased oxygen content in these experiments; however, in the case of the experiments with the cis-isomer (compare experiment 11/8 with curves A, B, C, and D), no such tendency is found. So far, no satisfactory explanation can be offered for this fact, although it is probable that, in these cases, side reactions due to the oxygen are responsible. In the same way, the small curvatures exhibited by the lines G, F, W, L, and 27/5 for the very first minutes of the reactions might be caused by similar side reactions entering into the scheme together with reactions 2 and 3, the result being the same as if k2 # k3 for a small time during which the reacting impurities are gradually consumed. As we shall see later, deviations from linearity are to be expected theoretically, if kz # k3. The condition kz = ka, used in deriving equation I1 on the basis of scheme A , can be shown to be necessary as well as sufficient to account for the experimental results. From reactions 1 t o 4 it follows that
loo0
KETELMR, VAN YELDEN, BROERS AND QERSMANN
I
I I
I I
.-m
U
$?
?
t-
z+
IX-aXI
301
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
1001
1002
KETELAAR, VAN VELDEN, BROERB AND GERSMANN
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
1003
Omitting again the term with J . (high quantum yields) and putting d(Br)/dt 0 for the stationary state, we obtain:
=
For the very small bromine concentrations used in our experiments, we may consider the term kl(Br2) to be negligible compared with ( k z , k3,) (for a detailed discussion of this point see reference 25). Taking this into account, combining equation VI1 with equation I, and using again the condition dz/dt = 0 for z = xe,we get:
+
with a = ks/k2 and K. = ak2,/k3,. This expression changes into equation I1 for k2 = ka, i.e., a = 1. Eliminating (ABr) from equations VI11 and I11 (omitting the term iJJ, we obtain: dz
According to this equation we should expect the slopes of the curves log I z - re 1 us. log (Brz), as drawn in figure 4, to change continuously with the factor l/[s a(1 - z)]. However, we have seen that the slopes turned out to be essentially constant. This would indicate that a = 1. The fact that the best agreement between experiment and theory is reached for a value a = 1 can be further illustrated as follows: Integration of equation IX leads to:
+
log (z
- 2,) - A(z - z,)
log
- x)
=
H’ log (Br2)
+ Q1(zd
z)
(Xb)
and (5.
+ A(%,
- z) = H‘ log (Brz)
where
Q1and Q2are constants determined by the starting conditions. If we now plot respectively log (z - 2,) - X(z - x.) or log (2. - z) A(z, - z) against log ( K C ) , which is a measure of (Br2), as before, we expect a straight line when using the ‘‘true” value of a. I n figure 5, some plots for different values of a (and subsequently of A) are shown for a typical experiment (F). For a = 1 (A = 0) a linear dependence is found (actually the same straight line as in figure 4b), while for a = 2 and for a = 0.5 irrefutable deviations from linearity appear. The same conclusions were reached for all other experiments. I t is not possible to decide by this method whether a is exactly equal to 1 or not, but
+
1004
KETELAAR, VAN VELDEN, BROEFS AND GERSMANN
it can be concluded with certainty that for all values of a larger than 1.5 or smaller than 0.75, no agreement between experiment and theory can be reached. Thus, the rate constants of reactions 2 and 3 must be equal or must at least have practically the same values. This conclusion is in agreement with the fact that for radical reactions of this kind activation energies have been found which are practically zero (see, for instance, Stewart and Weidenbaum (23), Muller and Schumacher (19), and Williams (28)), especially as collision numbers for reactions 2 and 3 must be practically the same. We have thus found that the experimental data can be satisfactorily explained with the aid of reaction scheme A, provided we assume kl and ka to be practically equal. However, the form of reactions 2’ and 3’ is not entirely unobjectionable.
T N
2.21 2.0
i-1.8-
P
>:
-X 1.6-
4
+
1.4-
X
I
w
2s
1.2-
‘
0.8
1.4 1.6 1.8 2.0 2.2 LOG(KC) +2 --f FIG.5. Graphical proof that a = 1 (kl = k3) in scheme A (compare equations IX and X )
0:6
0.8
1.0
1.2
A rough calculation using Pauling’s (21) values of bond energies shows that an activation energy of a t least 13 kcal./mole is needed for the endothermic decomposition of the radical ABr. A unimolecular decomposition of the kind shown in equations 2‘ and 3’ therefore is rather improbable. However, in the reaction mixture, collisions continuously occur, which suggests the following form of reactions, rather than 2’ and 3‘:
+ M C + Br + M ABr + M + T + Br + M ABr
--*
(2’4 (3’4
in which M = solvent molecule. On the other hand, we have already remarked that in our case-undiluted 1,2-dichloroethylene-free bromine atoms are not
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
1005
likely to persist as such. I n view of this consideration, we write the following modification of reaction scheme A:
Br2
+ hv -+2Br
(1)
Br+C+ABr
(2)
Br+T+ABr
(3)
scheme B: ABr
+ T-+ABr + C
(2")
+ C-+ABr + T ABr + Br2 ABr2 + Br ABr
(3") (4)
---f
in which the isomerization process is now described by reactions 2" and 3". It must be kept in mind that this scheme applies only to the reaction in pure dichloroethylene. As soon as solvents are used, reactions like 2'a and 3'a are to be included. Clearly, reactions of the kind ABr C -+ ABr C or ABr T -+ ABr T need not be included as long as one considers the radicals ABr to exist only in one form. If, however, one admits with Dickinson and coworkers (12, 20) (see also Berthoud and Urech (4)) the existence of two different types, CBr and TBr, which determine the isomerization by mutual transition, this reasoning no longer applies and the number of intermediate reactions in both schemes has to be enlarged considerably. This would make the elaborations of the kind described before rather complicated. As the explanation of the experimental facts in our case did not require such an extension, we prefer the simpler schemes given here. It can be shown easily (25) that scheme B, using again the condition k2 = kr, leads to the follawing relation analogous to equation V:
+
+
+
+
+ constant (XI) with H" = ( k 3 " / k 4 ) ( K o+ 1) and K . = (k2,,/k3,t).As we see, H" differs from H in equation V only with respect to the factor [(C) + (T)], which is no longer In 11: - ze I
=
H" In (Br2)
included. This was to be expected, as both k 3 ~and k, belong to bimolecular reactions, while k3, is a unimolecular reaction constant. I t is obvious that the linear relation between log I x - x. I and log ( K C ) as shown in figure 4 is expected from equation XI as well as from equation V. In view of the foregoing considerations, however, the reaction scheme B seems to be the most appropriate in this case. What information can be obtained from the values of the slopes in figure 4? Using formula XI we may calculate ( k 3 . , / k 4 )from the values of H" and K , ( = 4.44). Taking the mean value of H" to be 0.7 (figures 4a and 4b) we find ( k 8 , , / k 4 ) = 0.13, or, consequently, (k2"/k4) = 0.58. The rate constants are related to the energies of activation by the well-known formula k = Ze-"IRT. The collision numbers Z2" and Z4 may be roughly calculated by applying the
1006
KETELAAR, VAN VELDEN, BROERS AND QERSMANN
classical formulas (see for instance Moelwyn-Hughes (18)), assuming the molecular diameters of the radical ABr and the transdichloroethylene to be equal, the latter being estimated from the molar volume. In this way we (25) calculated ( Z Y , / Z , ) = 1.29, from which, by combination with (k2*’/k4) = 0.58, it follows that (E*,,- Ed) = 0.5 kcal./mole. Measurements of the temperature dependence of the reactions are needed to confirm this result. However, a large temperature effect can not be anticipated, in our opinion, as the activation energies of radical reactions of the kind encountered here are likely to be very small indeed, and consequently also their differences. If we assume for a moment that Ev, and E4 are practically equal, we calculate that the value of (z2,,/&) obtained by the classical methods used above differs from the “true” value only by a factor (1.29/0.58) = 2.2. This difference can not be considered important, knowing that even a factor of 10 is regarded as a “normal” one in considerations along classical lines (18). In the foregoing, we have compared k2”, rather than k3,,,with k4, because an activation energy of 0 might be possible for reaction 2”, while reaction 3” should require at least an activation energy equal to the difference in heat .content between the cis- and the trans-isomers (the cis-isomer having the lower heat content by an amount of about 0.5 kcal./mole; see references 25 and 31). So far, chain-breaking reactions have not yet been considered in setting up the reaction schemes A and B, as this was not necessary in discussing the relation between isomerization and addition. However, it is obvious that the course of both addition and isomerization separately will depend a great deal on the way in which chains are broken. As already said, it is highly improbable that bromine atoms enter into the terminating reactions. As both isomerization and addition velocities are determined by the concentration of ABr (equations I1 and 111), it is evident that inconsistencies between like sets of experiments are caused by the action of impurities (oxygen) on the ABr radicals in some way. Especially also, the large differences in reaction times, found between the “oxygen-free” and the preliminary experiments, point to this conclusion. In discussing the possible kinds of terminating reactions, we may confine ourselves to either the addition process or the isomerization process, as they have been shown to be related closely. In choosing the addition reactions for this purpose, we first call attention to the fact that the reaction rates have been found to vary with J, rather than with 6. This can be seen in table 3. In evaluating the tangents a t the breaks in the (KC)-t curves, caused by changing the intensity of illumination by some known factor (using the rotating sector disk or gray filters), no great accuracy can be expected, as it was often difficult to be certain to draw the correct curves through the experimental poinis near the breaks. However, the mean value of the power of J. certainly is closer to 1 than to 0.5. Also, no systematic dependence on either cis-isomer content, oxygen content, or bromine concentration was found (compare table 3). The dependence on the first power of J . suggests the following kind of chainbreaking reactions:
ABr
+X-Y
(S)
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
1007
in which X and Y are as yet undefined products. Combination of reaction (s) with reaction scheme A or B, using the condition for stationary state (d(ABr)/dt) = 0, leads to the expression:
(ABr) =
2iJ
-0
1 -
(XI11
k, (X)
TABLE 3 The variations of addition rates with intensity of illumination ~
EXPERIMENT NO.
c
I
...............
D ............... F ............ F. . . . . . . . . . . . G . . . . . . . . . . G . . . . . . . . . . . H... .......... H . . . . . . . . . .
1 ~
,
P P
w w 16/7 16/7 2017 20/7 3/8 3/8 6/8 6/8 11/8. . . . . . . . . . . . . . . 11/8.. . . . . . . . . . . . .
I
~
KC
TRANOE IN INTENSITY.
--
0 . 5 -+ 1 0.5 1 1 0.25 0.25- 1 (i)1-+0.25 0.25 -+ 1 1-0.25 0.25 -+ 1 (9) 1 + 0.295 (9) 0.29s -+ 1 1 + 0.25 0.25 1 1 -+ 0.25 0.25 -+ 1 1 -+ 0.25 0.25 -+ 1 1 -+ 0.125 0.125 -+ 1 1 -+ 0.25 0.25 1 (i) 1 -+ 0.25 (i) 0.25 1 (1)
0.204 0.088
0.340 0.166 0.638 0.238 0.720 0.320 0.118 0.061 0.612 0.173 0.472 0.268 0.544 0.300 0.697 0.470 0.444 0.408 0.940 0.900
-
-
I
1.1 0.95 1.2 1.1 0.6 0.9 1.0 1.25 1.1 1.2 0.7 0.8 1.2 0.8 1.1 0.8 1 .o 0.8 1.0 0.95 1.1 0.8
0.928 0.836 0.542 0.650 0.258 0.514 0.481 0.624 0.822 0.819 0.537 0.705 0.360 0.478 0.412 0.534 0.695 0.698 0.780 0, 782 0.960 0.954
OXYOLN CON’IENTt
-
+ +-
+ + -* + ++ ++ ++ ++ ++ ++ ++ ++ ++
’
* Changes in J . marked by ( a ) were caw by aray*fiIters; in all other cases the sector disk was-used. Changes marked by (i) were- made during the induction period of the reaction. t Oxygen content: -, “oxygen-free” series; “old” series, relatively high oxygen content; “oxygen-free” series with addition of oxygen by shaking i n contact with air.
+,
++,
Omitting the term iJ. from equation 111 and combining this equation with equation XII, we obtain:
(XIII) From equation XI1 it is obvious that the course of both isomerization and addition will depend upon the concentration of X and its change with time. Formula XI11 suggests a linear relation between -(d(ln (Rr,j)/dt) and J , if the concentration of X were so large as to remain practically constant during the reaction. In figure 6 some plots of this kind are shown, illustrating quite a different behavior. Experiments A, L, and H all belong to the “oxygen-free”
1008
KETELAAR, VAN VELDEN, BROERS AND GERSMANN
series (cf. figures 4a and 4b). Again, bromine concentrations are expressed in terms of KC, while (1 - ( J / J o ) ) = J,,/Jo. The larger values of this latter quantity correspond to the first periods of the reactions, of course. Heavy lines indicate experimental curves drawn so as to fit the experimental points as well as possible. Dashed lines were calculated theoretically and will be elucidated in the following.
T
I
I
a25
0.5
(I-%)
4
0.75
ID
FIQ.6 . Experimental comparison with equations XI11 and XVI. See also the corresponding text.
First, it appears that all curves have negative slopes from the starting points of the reactions until a more or less pronounced maximum in - (d(1n (Br2))/dt) is reached. Obviously, the concentration of X in the reaction mixture rapidly decreases. This behavior is especially pronounced in the case of “trans” experiments (see, e.g., H, where only two points are given), while for mixtures with relatively high content of &isomer (L, A) the decrease of (X) occurs more slowly.
ADDITION AND ISOMERIZATION IN PHOTOCHEMICAL REACTIONS
1009
It is clear that the compound X is consumed by reaction (s). Reaction (9) thus most probably also accounts for the induction periods observed at the beginning of the experiments. In any case these iriduction periods can not be ascribed to a cumulation of ABr radicals, which would be the case if the formation of these radicals by reactions 2 and 3 were to exceed their consumption by consecutive reactions in the beginning of the experiments. No such cumulation appearp, as was proved experimentally by interception of illumination during experiments for some time. In figure 7 the results are shown for two addition reactions (experiments P and 9/8). While no reaction at all takes place during darkness, the rate of the reactions a t the point where illumination is started again is exactly the same as it would have been without interception of illumination. We may now try to express (X) as a function of time. If we assume 1.2'
T
\-\
V r
0.8
'
0.6 .
*-
MINUTES
MINUTES
+
- -- - -_
-
FIG.7. Behavior of addition reactions on interruption of the illumination
that every absorbed light quantum is used for dissociation of a bromine molecule into two bromine atoms (reaction 1) (i = l ) , we get: d(X) dt
=
-k,(ABr)(X)
=
-2J.
(XIV)
owing to the fact that for every absorbed light quantum two reaction chains are started, both of which are broken by a terminating reaction of the type of reaction (s). If the initial concentration of X at time t = t , (beginning of illumination) is designated by (X)i, we may write: (X) = (X)i
-2
Combination with equation XI11 results in:
/' J. dt
1010
KETELAAR, VAN VELDEN, BROERS AND QERSMANN
From equation XVI it follows that a maximum is to be expected in the curves of figure 6 as soon as the following condition is fulfilled:
Thus one might think that with proper choice of the values of the ‘!parametera” 2kr/k, and (X)i/Jo the experimental curves could be explained theoretically by formulas of the kind of equation XVI. Actually, however, it was found that for certain k e d values of these “parameters’’ only parts of the curves could be explained on the basis of equation XVI. For instance, the dashed curves for the experiment A in figure 6 were calculated in this way. The values (kJk.) w 0.6 and (X)i/Jo w 15 best fitted the Erst part of the reaction (negative slope), while the last part could be best represented by taking for the “parameters” 4.5 and 30, respectively. The integral
i:
(1
-
(J/Jo))dt was calculated by graphical integration, from plots of 1
-
J / J o against time t . The value (X),/Jo w,15 can be shown easily (25) to correspond to a concentration of impurity X a t the starting point of the experiment of about lo-’ mole/liter, which amounts to an impurity content of only lo-‘ per cent in the dichloroethylene used. Analogous conclusions are reached for the other experiments of the “oxygen-free” series. The fact that the values of kr/k, to be used in representation of the last parts of the curves of figure 6 by formula XVI appeared to be considerably larger than those computed for the first parts of the same curves (in the case of experiment A by a factor 7 . 5 ) might point to the conclusion that, apart from the terminating reaction (s), a t least one other chain-breaking reaction takes place (s’), its rate constant being smaller than k,. This second reaction, however, should influence the course of the overall reaction only after the induction period is ended, i.e., from about the maximum in the curves of figure 6. It might be possible that this second terminating reaction has to be interpreted as a reaction between ABr radicals and the product Y which results from reaction (s), and the concentration of which increases during the induction period of the addition and isomerization processes. I n this case we should write: Y+ABr--tZ
(s‘)
For the last parts of the reactions, the compound Y then would take the place of X in formulas XIV, XV, and XVI. We have not yet considered the possible nature of the compounds X, Y, and Z, occurring in the terminating reactions (8) and (s’). I n view of the fact that oxygen has a great influence upon the course of both addition and isomerization, one may be inclined to identify X with some product formed by the oxygen present in the reaction mixture. The compound Y then could be some sort of radical containing oxygen (ABr02 might result if X is identified with molecular oxygen, but the presence of the latter in unsaturated systems of the kind con-
.\DDITlOS .\SD ISO~IERIZATIOZI IN PHOTOCHEMICAL REACTIONS
1011
sidered is highly improbable). This radical would be rather stable, and would react again with ABr in the way indicated by reaction (s’), forming a peroxide. Somr formulation of this kind has already been suggested by Daniels et al. (6,27). Support, for this interpretation can be found in the work of Bockemuller and Pfeuffer (.?I), who were able to isolate bis(dibromoisopropy1) peroxide from the products obtained by the bromination of allyl bromide in benzene with the addition of oxygen. On the whole it seems to be rather difficult to determine the true nature of the chain-breaking reactions. This usually is the case in dealing Ivith chain react ions, because the impurities which are responsible are often present only in very small quantities, as indeed we have shown to be the case in our experiments. A few words may be added as to the possible explanation of the experimental fact that t,he retarding influence of oxygen is considerably greater when the reaction mixture contains a high content of cis-isomer than when the transisomer predominates. It is highly probable that oxygen does not enter into reaction (s) in molecular form, but that some binding with the unsaturated dichloroethylene molecules occurs as soon as t.he liquids take up oxygen, forming some sort of peroxides. In this case, reaction (5) possibly should be replaced by two reactions: ABr
+ (T-On)
ABr
+ (C-0,)
Y
(ST)
+ Y’
(sc)
-+
and where (T-Oa) and (C-On) designate these peroxides of different and as yet unknown structures formed from the trans- and the cis-isomer, respectively. If now k: were much smaller than k:, chains would be broken much more effect>ivelyin the case of mixtures with high content of cis-isomer. It is evident that much more experimental work is needed to test this hypothesis. Finally, we must emphasize that the considerations as to the nature of the terminating reactions, here given, are all based upon the observed dependence of the reaction rates on the first power of illumination intensity. Several workers mentioned a dependence on fi in studying reactions of this kind (3, 14, 17). Most investigations, however, were carried out in dilute solutions, and also in many cases the influence of oxygen was totally disregarded. In solvents which themselves do not readily react with bromine atoms, these atoms are likely t o occur in the chain-breaking reactions, for instance in the following way: Br Brz or Br ABr .+ ABr2. All these reactions result in a dependence Br upon fi,as can easily be shown mathematically (25) (compare also Bauer and Daniels (2) and Rollefson and Burton (22)).
+
--j
+
S UMDIARY
The reactions between bromine and cis- and trans-1 ,2-dichloroethylen~s have been studied in the liquid phase a t 25’C., using light of wave length 5461 A. No solvents have been used. It has been shown that a close relation exists be-
1012
KETELAAR, VAN VELDEN, BROERJ AND QERJMANN
tween the two processes which occur-namely, isomerization and additionand that this relation is independent of the oxygen content of the reaction mixture, even though both isomerization and addition separately are strongly influenced by small quantities of this impurity. Reaction schemes have been suggested which account for the observed relation. Possible explanations of the influence of oxygen have been based on the experimentally observed dependence of the reaction rates on the first power of the light intensity, and the nature of the chain-breaking reactions likely to occur. REFERENCES
(1) BAXDERET, A,, A N D RANBY,B.: Helv. Chim. Acta 30, 1190 (1947). F.: J. Am. Chem. SOC.66,378,2014 (1934). (2) BAUER,W. H.,A N D DANIELS, A. (and former references): Helv. Chim. Acta 13, 385 (1930). (3) BERTHOUD, BERTHOUD, A., A N D MOSSET,M.: J. chim. phys. 33, 272 (1936). BERTHOUD, A., A N D PORRET,D . : Helv. Chim. Acta 17, 1548 (1934). (4) BERTHOUD, A., A N D URECH,c.: J. chim. phys. 27, 291 (1930). (5) BOCKEMULLER, W.,A N D PFEUFFER, L.: Ann. 537,178 (1939). (6) BROWN, R. F., A N D DANIELS, F . : J . Am. Chem. SOC.62,2820 (1940). (7) BULL, R.: Thesis, Wurzburg, 1930. (8) CARRICO, J. L., A N D DICKINSON, R. G.: J. Am. Chem. SOC.67, 1343 (1935). G.:Bull. soc. chim. Belg. 28, 234 (1914). (9) CHAVASNE, (10) COHENHENRIQUEZ, P . : Thesis, Delft, 1935;Rec. trav. chim. 64, 327 (1935). (11) DEKKER,A . J.: Thesis, Amsterdam, 1945;Physica 10, 768 (1943). R.G., WALLIS,R. F., A N D WOOD,R. E.: J . Am. Chem. SDC. 71,1238 (1949). (12) DICKINSON, This paper gives references to earlier work. (13) FLATT, R., A N D HEM, W.:Helv. Chim. Acta 21, 525 (1938). S. K . , AXD BHATTACHARYYA, S. C.: Z. physik. Chem. (14) GHOSH,J. C., BHATTACHARYYA, 32B, 145 (1936). (15) KETELAAR, J. A. A., V A N VELDSN, P . F., A N D ZALM,P.: Rec. trav. chim. 66,721 (1947). J. A . A , , DE \'RES, L., V A N VELDEN,P. F., A N D KOOY,J. S.: Rec. trav. (16) KETELAAR, chim. 66, 733 (1947). (17) LEERMAKERS, J. A., A N D DICKINSON, R. G.: J. Am. Chem. SOC. 64, 4648 (1932). (18) MOELWYN-HUGHES, E. A.: The Kinetics of Reactions i n Solution. Clarendon Press, Oxford (1947). (19) MULLER,K. L., A N D SCHUMACHER, H . J.: Z. physik. Chem. 36B, 285 (1937);42B. 327 (1939). (20)XOYES,R. M., DICKINSON, R . G., A N D SCHOMAKER, V.: J. Am. Chem. SOC. 67, 1319 (1945). (21) PAULIXG, L.: The Nature of the Chemical Bond. Cornell University Press, Ithaca, New York (1945). (22) ROLLEFSON, G. K.,A N D BURTON, M.: Photochemistry. Prentice-Hall, Inc., New York (1939). (23) STEWART, T. D., A N D WEIDENBAUM, B.: J. Am. Chem. SOC.67, 2036 (1935). (24) UHRIG,K., ROBERTS,F. M., A N D LEVIN,H . : Ind. Eng. Chem., Anal. Ed. 17,31 (1945). P . F.: Ph.D. Thesis, Amsterdam, 1950. (25) VAN VELDEN, (26) VERHOOGEN, D.: Bull. soc. chim. Belg. 34, 434 (1925). (27) WILLARD, J., A N D DANIELS, F.: J. Am. Chem. SOC. 67, 2240 (1935). G.:Trans. Faraday SOC.37,749 (1941). (28) WILLIAMS, (29) WISSLOW,E . H., A N D LIEBHAFSKY, H . A , : Ind. Eng. Cliem., Anal. Ed. 18,565 (1946). (30) WOOD,R. E., A N D DICKINSON, R. G.: J . Am. Chem. SOC. 61, 3259 (1939). (31)WOOD,R. E.,A N D STEVENSON, D . P . : J. Am. Chem. SOC.63,1650 (1941).
