C. BAWN, A. PENNINGTON AND C. TIPPER 291 THE REACTIONS OF LIQUID ETHYL BENZENE WITH OXYGEN IN GLASS VESSELS BY G. M. HENDERSON Received 1st February, 1951 Pure ethyl benzene, in clean glass systems, reacts with oxygen at 130OC first t o give a very slow radical reaction. This is replaced by a faster wall- catalyzed reaction. Both these reactions produce only the hydroperoxide of the hydrocarbon. The wall-catalyst apparently consists of stabilized free radicals derived from ethyl benzene. The amount of the wall-catalyst laid down from any one sample of ethylbenzene at any one temperature is a function of the original volume. The stabilized free radicals show a remarkable degree of stability. General experience with industrial liquid-phase air oxidations suggested that an experimental investigation of the oxidation of an isolated methylene group at atmospheric pressure, in glass at 1 3 0 O C both with and without catalysts, would aid in understanding what happens in the induction period of the more complex catalyzed technical oxidations which are usually necessarily carried out under pressure in metal equipment, and on materials presenting several places of attack per molecule.Ethyl benzene was chosen as it presented an isolated methylene linkage, open to oxidation, between the relatively inert phenyl and methyl groups. In the preliminary investigation, with which we are alone concerned here, no catalysts were used and the attempt was made to exclude all292 REACTIONS O F ETHYL BENZENE foreign ions known to be oxidation catalysts. Carefully purified ethyl benzene was used throughout and, although different specimens showed the familiar scatter of results of this field under any fixed experimental conditions, large bulked samples gave reproducible results.Consequently in making comparative experiments samples were withdrawn from the same stock sample as far as possible at the same time (Fig. 3). After the experimental work discussed in this paper was finished, a recent patent disclosed that commercial isopropyl benzene, even after careful fractionation, contains traces of styrene and its derivatives. Unless these are removed by special chemical means, isopropyl benzene will react only sluggishly with oxygen to form hydroperoxides. As the purification employed in our experiments involved only the sequence fractionation, washes with sodium bisulphite and sodium bicarbonate solutions, water washing and refractionation in a column of 30 theoretical plates, this fact must be borne in mind.Jacketted oxidizers in Pyrex glass were constructed whose inner cylindrical reaction tube was 5 cm. wide by 30 cm. deep. Sintered glass discs 2-5 cm. in diameter were set on the end of flared-out narrow tubes forming a spoon. The porosity could be varied from G1 (coarse) to G, (fine) and were adapted from standard Pyrex Gooch crucibles. These discs were set with their faces upwards, at 45" to the vertical. The shallow spoon on the end of the air line under these circumstances was completely self-draining on commencing air flow. Various jacketting liquids or vapours could be used to adjust the reaction temperature.The reactor was " blacked out " for dark reactions (as in the bulk of the present experi- ments) by swathing in layers of aluminium foil. Porosity of the disc and air flow rate were shown to have little effect provided the system was saturated with oxygen and a flow of 20 l./hr. of air was fixed as a standard. Gas analysis on oxidations using air showed that the exit gas contained 20 yo oxygen on average. The ascend- ing stream of bubbles from the sintered disc at the foot of the reactor gave excellent agitation by an air lift effect, bubbles being swirled below the level of the injection area. Cooled portions of the liquid contents were withdrawn at any. time for analysis by means of a built-in water jacketted pipette with a capillary stem dipping to the foot of the reactor.Analytical samples were then repipetted from this cooled zone and the residue re- turned to the reactor. The most important experimental feature is that the various Quickfit joints must be dry or unlubricated. The whole apparatus before each experiment must be thoroughly cleaned with a strong metal-free oxidizing agent ; concentrated hot nitric acid or mixtures of nitric and hydrochloric acid were used with equal success. The acid was then removed with dis- tilled water and the apparatus finally baked at 130-150" C in an electric oven for at least I hr. If this cleaning was not carried out properly, a series of repeat experiments showed a steady decay in rate each time. A few experiments suggested hot concentrated sulphuric acid was also effective, but it was much less convenient and more suspect of metallic contamination.One or two experiments were carried out in spherical reactors of different volumes so as to vary the ratio, wetted wall area/volume. It proved difficult to fill the reactors with glass rods or tubes without disturbing the even distribution of gas bubbles in the system and creating stagnant enclosures of liquid. The results obtained over several years have not fallen easily into the current theories and a very tentative hypothesis is now advanced which seems to cope with some of the experimental facts. The rate of reaction in all cases was followed by titrating samples of the reactor contents for hydroperoxide by the Kokatur and Jelling a method and the hydroperoxide content was expressed as ml.of 0.1 N sodium Kokatur and Jelling, J . Amer. Chem. Soc., 1941, 63, 1432. 1 Distillers Company Ltd., Brit Pat., 630,286.G. M. HENDERSON 29 3 thiosulphate finally consumed by I ml. of reactor contents. The sum- marized findings are : (I) The specific rate of hydroperoxide formation at 130’ C in the pure thermal reactions depends primarily on the volume of the charge being oxidized from any one batch of purified ethyl benzene (Fig. 4). ( 2 ) Dilution of ethyl benzene with up to 33 yo of pure diphenyl before oxidation does not alter this rate dependence on the original volume of the solution (Fig. 6). ( 3 ) The rate of the thermal reaction, once established, is independent of rate of air flow, bubble size or even the partial pressure of oxygen. The solutions in our experimental range were always fully saturated with oxygen. (4) Once the steady rate for a given volume of ethyl benzene has been established, exposure to u.-v.light leads to a rise in rate, which is mostly retained on removal of the exciting source of radiation (Fig. 10). ( 5 ) Oncc the steady rate for a given volume has been established (0) addition of fresh ethyl benzene, say by doubling the volume, does not alter the rate (Fig. 8) ; ( b ) removal of a major portion of the charge, say by reducing the volume to one-half, greatly enhances the specific rate to somewhat over twice the original figure (Fig. 9). (6) If a volume of ethyl benzene is oxidized at its steady rate, and the reactor contents are wholly removed from the reactor, two effects may be subsequently observed : ( a ) If the reactor is cleaned with methanol, water, and then baked at 130’ C an attempted repeat experiment with the previous volume of ethyl benzene or larger volume of ethyl benzene, only reproduces the steady rate of the previous experiment.Some factor of the original rate has been imposed on the walls as a “reaction memory ”. The rate shows a slight but variable decay according to the severity of the solvent washing. If this subsequent experi- ment takes the form of using a smaller volume than the original experiment, higher than normal rates may be experienced. (b) If the walls are cleaned with nitric acid following the standard procedure, the now clean walls are without effect and the specific rates observed initially depend once more only on the initial volume of the ethyl benzene charged to the oxidizer; the other two variables (the temperature and provenance of the samples) being kept constant.(7) If a volume of ethyl benzene is oxidized at its steady rate and is then poured into a new clean reactor, the reaction proceeds at the same steady rate in the new reactor, in spite of depositing rate influencing sur- faces on the walls of both reactors (Fig. 7). (8) Very pure and fresh samples of ethyl benzene show an induction period during which there is a slow but steady rate of oxidation which is then replaced by a faster and steady rate. This induction period is fleeting and can only be observed under favourable conditions. This suggests that the reaction commonly observed is catalyzed in some way, even with attempted exclusion of catalyst ions (Fig.3). (9) In all these reactions, hydroperoxide is the only detectable reaction product, the formation of water characteristic of the metal-catalyzed reaction has never been observed without the deliberate addition of traces of metal catalyst such as cobalt naphthenate. (10) I t is noticeable that the light-catalyzed reaction shows signs of sensitivity to oxygen partial pressure, apparently being poisoned by high partial pressures of 0, at temperatures of IOOO C. From these facts i t seems that the reaction generates a limited quantity of material on the walls of the glass reactor which acts as a catalyst. This294 REACTIONS OF ETHYL BENZENE is presumably organic in nature.This bears some relation to the original volume of the system and is one factor in influencing the specific rate. Furthermore, this catalyst is able to short circuit the original catalyst- forming reaction, which it is reasonable to suppose to be a homogeneous chain reaction by analogy with other hydrocarbon systems. Since the system is at all times saturated with oxygen, the two reactions cannot compete over the system : hydrocarbon + dissolved oxygen. We are forced to assume that there is a slow equilibrium absorption of some of the oxygen into a more closely bound form slow RH2 + 0, [RH,-++O,] the oxygen here being held as an activated complex with the methylene bridge group under attack. We then suppose that this may decompose in at least two ways, back to dissolved oxygen or most infrequently into a hydrocarbon free radical.We may now set down the slow chain reaction of the induction period : very slow CHAIN INITIATION. [RH, c --f O,] - RH' + H02 CHAIN PROPAGATION. RH' + 0, + RHO2' RHO,'+ RH, -+ RHOOH + RH'. CHAIN TERMINATION (at least in part). RH' + wall -+ RH- wall+. I t is now supposed that at least part of the captured radicals on the walls are stabilized by withdrawing electrons from the glass and are then sufficiently stable to form long-lived complexes. It is suggested that a wall charged with such stabilized free radicals will lead to a facile re- arrangement on its surface of the ethyl benzene-oxygen complex leading to the peroxide and a regenerated ion or radical with only a movement of charges.One further supposes that this surface reaction is much faster than the original chain-initiating step ; it thus greatly reduces the stationary concentration of the active complex and hence practically suppresses the original classical chain reaction in solution which led to the setting-up of this wall process. The wall reaction may. then be regarded as a chain reaction carried out on the surface of the wall by a fixed quota of stabil- ized free radicals, or as a wall reaction catalyzed by a highly efficient organic ion catalyst supported on glass. The wall process resists simple organic solvents, water and dry baking for brief periods with only slight decay but is completely destroyed by rezgents such as hot concentrated nitric acid, We can now fit the hypothesis to the experimental facts if we suppose that the totaI charge of wall catalyst is a function of the original volume Vo.Total catalyst oc fV,. We now suppose the system circulates over the surface, the catalyst coming in contact with succeeding units of volume. The system may be best treated as the pouring of a volume V , of ethyl benzene through a catalyst bed containing a quantity of catalyst fV, in unit time, the amount of catalyst alone controlling the efficiency of the reaction. If we double the volume, we suppose we exactly double the efficiency of the catalyst bed and again we have a volume 2 V 0 treated as pouring through a bed of catalyst 2fV0 in unit time, the catalyst bed now being of twice the efficiency. This gives us a rate dependence as found on varying the initial volume ; furthermore it matches the movement of specific rates &-hen the volume is subsequently doubled or halved, The invigoration This is illustrated in Fig.I and 2 . The wall coating is quite invisible.G. M. HENDERSON 29 5 of the reaction by u.-v. irradiation, leading to a permanent enhancement of the reaction rate is explicable by supposing the superimposed photo- initiated chain reaction is able to generate free radicals anew by an in- dependent route which generates a further quantity of catalyst. FIG. RELATIVELY RH- CHAIN TERUINATION t Slow chain reaction in brief induction period. PHOOH WROPL ROXlDE ph phase I. RH, + Q, 1"'"" CHAIN REACTION COATED WALLS RHOOH HYDROPEROX1 DE FIG. 2.-Phase I J . Final fast wall reaction governed by stabilized radicals.XPIJ. 0 AND 8) E.B. PURIFIED E.B I 2 3 4 5 FIG. 3.-Effect of provenance on specific rate of oxidation of ethyl benzene. Conditions : 130' C, 150 g. ethyl benzene.296 REACTIONS OF ETHYL BENZENE The experimental data are appended in the form of graphs showing rates of reaction under different conditions. The rates are expressed as the build-up in time of the peroxide titre of I ml. of the reactor content in ml. 0.1 N sodium thiosulphate. t5 2 3 4 5 I '- I I - I I L FIG. +-Specific rates of peroxide formation in relation to volume of charge. Conditions : varying E.B. charges, 130' C. t5 5 2 T I- 2 3 4 5 FIG. 5.-Effect of temperature on thermal reaction a t 100' C and 130' C on same bulked sample ethyl benzene, Conditions : 150 g.charges of E.B. 150 9. ETHW- BENZENE COMPARED WITH EQUAL VOLUME SOLUTION OF DIPHENYL IN ETHYL BENZWE AND 13OoC. (SAME STOCK) AT too0c. FIG. 6.-Comparison of pure ethyl benzene with a mixture of diphenyl (I mole) and ethyl benzene (2 moles), equal volumes of solution being compared.G. M. HENDERSON 297 It is suggested that this type of oxidation probably only appears under favourable circumstances with particular hydrocarbons, but nevertheless may explain the variable results of past workers in this field. 4 4 w " @= DISC IMMERSED 8.5 Cm. IN NARROW CYLINDRICAL REACTOR. A= DISC IMMERSED 443Cm. IN SPHERICAL REACTOR n r 4 vl L E -3 TIME SCALE ADJUSTED. REACTOR, FIG. 7.-Effect of transferring charge of oxidizing ethyl benzene to new clean reactor of different shape.Conditions : 150 g. ethyl benzene, 130° C. HOURS 3 3 4 3 FIG. 8.-Effect oi doubling ethyl benzene charge during oxidation (temperature 1 3 0 O C). FIG. g.-Effect of removing half the charge of oxidizing ethyl benzene during an oxidation. Conditions : 500 ml. ethyl benzene in I litre round flask in oil bath at 130' C. It may be best to visualise this reaction as an extension of the original homogeneous chain reaction which has colonized the reactor walls. To draw another analogy, it is the developed and fixed latent image on the walls of the stationary state in the earlier homogeneous reaction.298 OXIDATION OF LUBRICATING OILS We had at one time supposed that this type of wall-process might also take place on the molecules of oil soluble cobalt catalysts used in catalyzed air oxidations of hydrocarbons, thus giving the striking colour charge (blue +green) of cobaltous salts to cobaltic complexes when the rapid oxidation commences after an induction period. The recent ex- periments of Bawn and co-workers 8 demonstrate that hydroperoxides alone can account for this phenomenon, ... w t5 5 I FIG. 10.-Effect of period of exposure t o u.-v. lamp on a reaction. Conditions : I 50 g. ethyl benzene, I 30° C. However, for metal-catalyzed oxidations, the slow growth of hydro- peroxide during the induction period may well arise by such a wall-process which will then in turn be suppressed by the hydroperoxide activated metal ions, whose reactivity is such, that the concentration of dissolved oxygen in many catalyzed reactions is nearly reduced to zero once the rapid catalyzed reaction is running. Imfierial Chemical Industries Limited, Research Laboratories, Hexagon House, Blackley, Manchester, 9. Private communication.