Chemistry (Annual Reports - Vol.59-1962)

Annual ReportsOn The Progress Of Chemistry

G. 0. ASPINALL,D.Sc., F.R.S.E., F.R.I.C. WILSON BAKER, M.A., D.Sc., F.R.S. C. H. BsMFoRD,&I.A., Sc.D., F.R.I.C. J. W. BARXETT, Ph.D., A.R.C.S., F.R.I.C. R. P. BELL, N.A., F.R.S. D. M. BROWN,Ph.D. G. M. BURNETT, Ph.D., D.Sc. I. G. M. CAYPBELL,B.Sc., Ph.D. G. E. COATES,>LA., D.Sc., F.R.I.C. T. COTTRELL, D.Sc., B.Sc., F.R.I.C. D. P. CRAIQ, Ph.D., D.Sc., F.R.I.C. D. F. ELLIOTT, Ph.D., A.R.C.S., A.R.I.C. V. GOLD, D.Sc., Ph.D. R. H. HALL, Ph.D., A.R.C.S., F.R.I.C. C. H. HASSALL, M.Sc., Ph.D., F.R.I.C. A. W. JOHNSON, Sc.D., Ph.D., A.R.C.S. C. KEYBALL,N.A., Ph.D., F.R.I.C.

J. G. J. J.

A. KITCHENER,D.Sc., Ph.D. KOHNSTAM, Ph.D. W. LINNETT,M.A., D.Phil., F.R.S. F. W. MCONIE, M.A., D.Phil., D.Sc. R. S. NYHOLY, D.Sc., F.R.I.C., F.R.S. L. E. ORQEL,M.A., D.Phil., F.R.S. L. N. OWEN,Ph.D., DSc., F.R.I.C. R. A. RAPHAEL, D.Sc., F.R.I.C., F.R.s.

J.M.RoBERTsoN,C.B.E.,M.A.,D.SC.,F.R.S.A. G. SHARPE, M.A., Ph.D., F.R.I.C. K. W. SYKES, M.A., D.Phil. J. C. TATLOW,B.Sc., Ph.D., F.R.I.C. H. J. V. TYRRELL,B.Sc., M.A. B. C. L. WEEDON,D.Sc., A.R.C.S., F.R.I.C. D. H. WHIFREN, M.A., D.Phil. M. C. WHITINQ, M.A., Ph.D., A.R.C.S. W. WILD, Ph.D.

E d i t o rR. S. CAHN, M.A., D.Phil.Nat., F.R.I.C.

Deputy Editor L. C . CROSS, Ph.D., A.R.C.S., F.R.I.C.

Assistant EditorsI. J. CANTLON, Ph.D.

N. KEEN, Ph.D.

ContributorsR. C. R. K.

M. ACHESON, M.A., D.Phi1. C. BARKER, B.Sc., Ph.D. F. BARROW,B.Sc., M.A., D.Phi1. W. BENTLEY, K A . , B.Sc., D.Phil.

J. H. KNOX, B.Sc., Ph.D.

M. L. MCGLASHAN, Ph.D., D.Sc., F.R.I.C.S. MARES, M.Sc., D.Phil. J. NEREB, B.A. J. MILLEN, B.Sc., Ph.D. NICHOLLS,B.Sc., Ph.D., A.R.I.C. K. H. OVERTON,Ph.D., 13.S~. C. W. REES, B.Sc., Ph.D. D. H. REID, B.Sc., Ph.D. R. E. RICHARDS,M.A., D.Phil., F.R.S. D. J. SUTOR,M.Sc., Ph.D. (N.Z.), M.A., Ph.D. (Cantab.) M. C. R. SYJIONS, D.Sc., F.R.I.C. T. L. V. ULBRICHT,B.Sc., Ph.D. J. V. WESTWOOD, M.Sc., F.R.I.C. A. M, WHITE, B.Sc., Ph.D. R. F. M. WHITE, B.Sc., Ph.D., A.R.I.C. D. W. WILSON, XSc., F.R.I.C. G. A. D. D.

B. CAPON,B.Sc., Ph.D. P. F. S. CARTWIUQHT, M.Sc., Ph.D.,

F.B.I.C. W. COCHRAN, Ph.D., F.R.S. B. A. CRAIG, M.Sc., D.Phi1. L. CROMBIE,Ph.D., D.Sc., F.R.I.C. G. EQLINTON,B.Sc., Ph.D. D. F. ELLIOTT, Ph.D., B.Sc. C. P. FAWCETT, B.Sc., P1i.u. R . P. H. GASSER,M.A., D.Phi1. V. GOLD, Ph.D., D.Sc. G. W. GRAY,B.Yc., Ph.D. A. K. HOLLIDAY, Ph.D., D.Sc., F.R.I.C. J. HONEYXAN, X.A., B.SC., Ph.D., D.Sc. V. I T. . JAXES B.Sc.. PI1.D.. A.R.I.C.

THE CHEMICAL SOCIETY

Annual ReportsOn The Progress Of ChemistryFOR 1962

Volume LIX

LONDON 1963

THE CHEMICAL SOCIETYPATRONHER MAJESTY THE QUEEN

PresidentJ. M. ROBERTSON, C.B.E., M.A., D.Sc., F.R.S.

H. J. EMEL~US, C.B.E., M.A., D.Sc., F.R.S. SIR CYRIL HINSHELWOOD, O.N., M.A., Sc.D., F.R.S. E.L. HIRST,C.B.E., D.Sc., LL.D., F.R.S.

SIR CHRISTOPEER INQOLD, D.Sc., F.R.I.C., F.R.S. SIRERICRIDEAL,M.B.E.,M.A., D.Sc.,F.R.S. LORDTODD, X.A., D.Sc., F.R.S.

D. H. R. BARTON,D.Sc., Ph.D., F.R.S. E. J. BOWEN, M.A., D.Sc., F.R.S. J . CHATT, M.A., Sc.D., F.R.S.

E. R. H. JOKES, D.Sc., F.R.I.C., F.R.S. B. LYTHQOE, M.A., Ph.D., F.R.S. M. S ~ C E Y Ph.D., , D.Sc., F.R.S.

Honorary SecretariesA. W.JOHNSON, Sc.D., Ph.D., A.R.C.S.

E. W. SYRES, M.A., D.Phil. J, W. LINNETT, M.A., D.Phil., F.R.S.

D.Sc., Ph.D., F.R.I.C. G. 0. ASPINALL, J. BADDILEY, D.Sc., Ph.D., F.R.S. L. J. BELLAMY, B.Sc., Ph.D. W. COCKER,M.A., D.Sc., F.R.I.C., M.R.I.A. D. P. CMG, M.Sc., D.Sc. L. CROXBIE, D.Sc., F.R.I.C. F. S. DAINTOP;,M.A., Sc.D., F.R.S. D. H. EVERETT, M.B.E., M.A., D.Phil. B.Sc., Ph.D. G. W. A. FOWLES, W. GERRARD, D.Sc., Ph.D., F.R.I.C. R. N. HASZELDINE, M.A., Sc.D., F.R.I.C.

I ( . HOLLIDAY, Ph.D., D.Sc., F.R.I.C. HONEYMAN, Ph.D., D.Sc. R. KATRITZKY, M.A., D.Phil., Ph.D. J. KING, B.Sc., M.A., D.Phi1. E. A. MOELWYN-HUQHES, D.Phil., D.Sc. Sc.D. W. J. ORVILLE-THOMAS, Ph.D., D.Sc. H. M. POWELL, B.Sc., M.A., F.R.S. R. A. RAPHAEL, D.Sc., F.R.I.C., F.R.S. J. C. ROBB,DSc., Ph.D. A. I. VOGEL, XSc., D.Sc., F.R.I.C.

A. J. A. T.

Ex OWcioSIR JAMES COOK, DSc., F.R.I.C., F.R.S. (Chairman of the Chemical Council) D. H. HEY, D.Sc., F.R.I.C., F.R.S. (Chairman of the Publication Committee) E. D. HUGHES, D.Sc., F.R.I.C., F.R.S. (Chairman of the Joint Library Committee)

General SecretaryJ. I t . RUCKKEENE, M.B.E., T.D., M.A.

LibrarianR. G. GRIFFIN, F.L.A.

CONTENTSPAQE

GENERAL AND PHYSICAL CHEMISTRY 1. INTRODUCTION. By R. E. Richards . 2. ULTRA-HIGH VACUA. By R. P. H. Gasser . 3. GAS-PHASE OXIDATION. By J. H. Knox . 4. ELECTRON SPIN RESONANCE. By M. C. R. Syrnons . 5. NUCLEAR MAGNETICRESONANCE IN ELECTROLYTE SOLUTIONS. By R. A. Craig . 6. EQUILIBRIUM PROPERTIES OF LIQUID MIXTURES. By M. L. McGlashan . By R. F. Barrow 7. SPECTROSCOPY OF DIATOMIC MOLECULES. and A. J. Merer . INORGANIC CHEMISTRY 1. INTRODUCTION. By A. K. Holliday and D. Nicholls 2. TYPICAL ELEMENTS.By A. K. Holliday . 3. THE TRANSITION ELEMENTS. By D. Nicholls .

7 7 18 45 63 73 99

.

129 130 152

ORGANIC CHEMISTRY 1. INTRODUCTION. By L. Crombie and V. Gold . 2. PHYSICAL PROPERTIES AND ORGANIC STRUCTURE. By D. J. Millen and R. F. M. White . 3. REACTION MECHANISMS. By C. W. Rees and B. Capon 4. GENERAL METHODS. By C. C. Barker and G. W. Gray 5. ALIPHATIC COMPOUNDS, By G. Eglinton . 6. ALICYCLIC COMPOUNDS. By K. H. Overton . 7. AROMATIC COMPOUNDS. By D. H. Reid . 8. HETEROCYCLIC COMPOUNDS. By R. M. Acheson . 9. ALKALOIDS. By K. W. Bentley . 10. CARBOHYDRATES.By J. Honeyman . 11. NUCLEIC ACIDS. By T. L. V. Ulbricht .

187 189 207 254 268 281 307 319 342 359 37 1

BIOLOGICAL CHEMISTRY 1. INTRODUCTION. By D. F. Elliott . 384 2. THE BIOSYNTHESIS OF PORPHYRINS. By G. S. Marks . 385 3. VITAMIN B12. By A. M. White . 400 4 . THE NICOTINAMIDE COENZYMES AND THEIR APOENZYMES. By C. P. Fawcett 413 5 . METABOLISM OF STEROID HORMONES. By V. H. T. James 426

.

ANALYTICAL CHEMISTRY. By D. W. Wilson, J. V. Westwood, and P. F. s. Cartwright 1. INTRODUCTION 436 2. GENERAL 437 3. BASICOPERATIONSAND APPARATUS . 438 4 . QUALITATIVE ANALYSIS 440 5. METHODSOF SEPARATION. 443

.

.

.

vi

CONTENTS

6. GRAVINETRIC AKD TITRIMETRIC ANALYSIS . 7. INSTRUMENTAL END-POINT DETERMINATIONS 8. DETERMINATION OF ELEMENTS IN ORGANIC COMPOUNDS

.

9. SPECTROSCOPIC ANALYSIS 10. ELECTRICAL METHODS 11. THERMAL METHODS

. .

.

. . .. .

PAQE

449 457 473476

. 480499501 506515

CRYSTALLOGRAPHY 1. GENERAL.By W. Cochran . 2. ORGANICSTRUCTURES. By D. J. Sutor INDEX OF AUTHORS INDEX O F SUBJECTS

.

...

.

..

. .

556 566

NOTE ON PRINCIPAL REFERENCES USED

ANNUAL REPORTSON THE

PROGRESS OF CHEMISTRYGENERAL AND PHYSICAL CHEMISTRY1. INTRODUCTION

THE reports this year cover a small number of selected topics, and refer to developments over a'period of more than one year. The extensive and detailed studies of the thermodynamic properties of liquid mixtures have not been reported for some time, so this work is summarised in considerable detail. We also include reports on the kinetics of oxidation reactions and on the uses of ultra-high vacuum techniques. The availability of such low pressures now has a very important bearing on the study of gas-solid surface phenomena. I n recent years impressive progress has been made in the study of energy levels of simple molecules. The accurate and extraordinarily detailed information available from the spectra of these molecules is a challenge to the theoretical chemist, and we report the present state of this field of work. The study of electron-spin resonance spectra has grown in a most remarkable way in recent years; it therefore seemed desirable to include a further report on this subject. A report on nuclear magnetic resonance spectroscopy of organic compounds appeared last year, and this is now supplemented by an account of work which has been done on solutions of electrolytes.

R.E. R.2. ULTRA-HIGH V A C U A As this subject has not been reported on previously, the intention is to give a rather general account of results of interest to chemists. Sufficient references have been included to make it possible for those who wish t o pursue the subject further t o do so. By ultra-high vacuum, a pressure below mm. Hg will be implied. A simple calculation based on the kinetic theory of gases demonstrates the advantage t o be gained by reducing the pressure from the conventional mm.) t o the ultra-high vacuum region. The high-vacuum region (ca. rate of collision of gas molecules with a surface, p, is given by p = P(SnmkT)-4 collisions cm.-2 sec.-l where rn = the mass of a molecule. For oxygen a t loF6mm. and 23" c , u is 5 x 1014 collisions cm.-2 sec.-l. A typical metal surface has ca. 10l5 so that, unless the colliding gas has a very small probability of atoms

8

G E N E R A L A N D PHYSICAL CHEMISTRY

sticking, a surface which was clean originally will become covered by a monolayer of gas within a few seconds. However, a thousand-fold decrease in pressure increases the time for monolayer formation by the same factor and the surface stays clean long enough to be studied. It is also possible t o investigate the interaction of the surface with an experimental gas a t pressures in the loe7 mm. region without significant interference from the residual vapour. It is in the field of gas/metal-surface interactions that the most considerable progress has been made and with which this report will mainly deal. Experimental Methods.-There is evidence that pressures approaching the ultra-high vacuum region were being obtained a t times when conventional ionization gauges were recording pressures in the ordinary highvacuum region. This is now known to be due to the X-ray limit of an ionization gauge, which arises as follows. The impact of the electrons, comprising the grid current, causes the grid metal to emit soft X-rays. These X-rays eject electrons from the positive-ion collector and thus give rise to a current at the collector in the same sense as that which records the pressure. The usefulness of the conventional ionization gauge is limited by this process to the recording of pressures down to the mm. region. A pioneering paper in the ultra-high vacuum field is that of Alpert,l who, recognizing the limitations of the ionization gauge, designed a new gauge with a much lower X-ray limit. He also demonstrated the conditions under which ultra-high vacuum could be obtained repeatedly and without sealing off the apparatus from the pumps. The ionization gauge designed by Bayard and Alpert 2 is widely used in ultra-high vacuum studies. It reduces the X-ray limit to below 10-10 mm. by inverting the arrangement of the electrodes. The cathode is outside the cylindrical wire grid and the positive-ion collector is a fine wire running axially down the grid. This wire collects positive ions produced by electron bombardment in the volume defined by the grid. The positive ion current thus produced is directly proportional to the gas density, and therefore to the pressure. This fine wire collector presents a much smaller area for X-ray bombardment than the usual collector so that the X-ray limit is correspondingly lower. Current modifications to the original gauge include : (i) The use of lanthanum boride-coated filaments. These give off thermionic electrons a t much lower temperatures than tungsten, which is normally used. This is an improvement because one of the problems associated with the use of thermionic gauges is that of the interaction of the hot filament with the gases in the system; the lower the filament temperature, the less important are these undesirable side reactions. (ii) The use of still finer wires to reduce the X-ray limit even further. An important condition for obtaining ultra-high vacua is to avoid completely the use of greased taps. Alpert has designed an all-metal closure valve which is turned off by forcing a metal cone very hard into a metal seating. Movement is allowed by use of a flexible metal diaphragm, attached t o the cone. Although such a valve does not switch off completely, con,

(

a

D. Alpert, J. Appl. Phys., 1953, 24, 860. R. T. Bayard and D. Alpert, Rev. Sci. Instr., 1950, 21, 571.

G A S S E R : U L T R A - H I G H VACUA

9

ductances as low as 10-14 1. sec.-l can be obtained. It has the very important properties that it is robust, and can be heated t o temperatures as high as 500" c without harm. It thus becomes possible to outgas an entire vacuum assembly by heating it as a whole in a bake-out oven. It is common to use temperatures approaching the softening point of glass. Further outgassing of the metal electrode assembly of the gauge is usually necessary, and electron bombardment or radio-frequency heating t o red-heat will accomplish this. Conventional diffusion pumps can be used for ultra-high vacuum work, provided that a very efficient trap is incorporated between the pump and the apparatus, to collect back-streaming vapour. A common and reliable trap is a vessel, cooled with liquid air, so designed that back-streaming molecules have to make many collisions with the cold walls before they reach the apparatus. Non-refrigerated traps have also been used with success; trapping materials include rolled up copper foil and artificial mm. can be obtained zeolite.4 By these means a pressure of ca. 3 x as a matter of routine. On isolation of the system from the pumps, the pressure can be reduced still further by operating the gauge a t a high emission (Le. cathode to grid) current, when it will act as an " ion pump " of maximum pumping speed about 2 1. sec.-l. When operating the gauge as a measuring device, this pumping is reduced as far as possible by using small (10-100 FA) emission currents. As an alternative to diffusion pumps, " getter-ion " pumps can be used. The details of the mode of operation of these pumps are still not clear, but in principle they are considered to transfer gas from the gaseous phase to the walls of the pump by a combination of gettering and ionic pumping. The backing pressure required for them to start up can be obtained by cooling an adsorbent material in liquid air. It is thus possible to have a completely oil-free system. I n spite of this, however, hydrocarbons are produced in the system from impurities in the getter. The Theoretical Approach to Ultra-high Vacua.-The experimental methods described above represent a well-tried but largely empirical technique, and it is useful to consider the factors which limit the attainment of ultra-high vacua. The strength of binding of molecules of the atmospheric gases to the walls of the apparatus plays an important part in determining the rate a t which the apparatus can be pumped out. I f the molecules,are strongly held they cannot be removed, but neither do they evaporate, and thus they make no contribution t o the vapour pressure. Weakly-held molecules evaporate rapidly and are removed by the pump. It is thus the molecules with intermediate binding energies which give rise to the continuous evolution of gas experienced in unbaked apparatus. A quantitative treatment has shown that, a t room temperature, gases with heats of adsorption in the range 15-25 kcal. mole-1 are the most troublesome. However, on raising the temperature to 300" c these gases are rapidly evolved. ThisD. Alpert, Westinghouse Research Lab. Scientific Paper 1744 (East Pittsburgh, U.S.A. 1953). L. L. Levenson and N. Milleron, Trans. Vacuum Syrnp., 1961, 91. J. P. Hobson, Trans. Vacuum Syrnp., 1961, 26.

10

GENERAL AND PHYSICAL CHEMISTRY

result implies that rather less extreme conditions of outgassing than those previously described may be adequate t o achieve ultra-high vacua, a conclusion which can be confirmed by the Reporter who has obtained pressures below mm. after bake-out in the 250-3OO"c range. Besides outgassing from the walls, another factor which may limit the attainment of ultra-high vacua is the diffusion of gas from the atmosphere through the walls of the apparatus.l Glass is a very convenient constructional material and is widely used in vacuum apparatus, so that it is important t o know in what circumstances the inflow of gas becomes significant. There is much information about the permeation of helium, and data are now also available for other gases.6 The relative importance of the permeation of various gases can best be appreciated by considering their build-up in a sealed-off system. Consider a vitreous silica bulb of capacity 330 C.C. with walls 1 mm. thick and surface area 100 sq. em. a t 25" c. Then, starting with a negligible pressure, at the end of one year the pressures would be: loW4 mm. of He, mm. of Ne, and mm. of H,. No other gases would be present. Even after 100 years only a few molecules of oxygen would have penetrated. There is a marked difference in permeability between various types of glass. I n a bulb as specified the time taken for the pressure of helium t o build up t o mm. would be as follows: Silica, 3 days; Pyrex, a month; soda-lime, about 100 years. All gases permeate glass a t a rate directly proportional to their partial pressure, the flow being molecular. Steel walls are impermeable to all gases a t room temperature except hydrogen, which flows as atoms. The lowest pressure which could possibly be obtained would occur, of course, when there were no gaseous molecules present a t all. This situation has been approached through a combination of ultra-high vacuum and cryo-, genic techniques.7 The apparatus was first pumped down to ca. mm. and isolated from the pumps. Then part of it was immersed in a liquid helium bath, when physical adsorption of the residual gas occurred on the cold walls. Extrapolation from measurements a t high coverage indicates that, a t equilibrium, the pressure in the cold part should be in the region of 10-35 mm., i.e. there is no gas molecule within its volume for most of the time. Results.-Any property of a surface which is modified by the adsorption of gas can, in principle, be used to study the gas-surface interaction. I n this section three widely used techniques utilizing ultra-high vacua will be described. FieEd-emission microscopy. This is probably the most important and informative method of studying gas-metal interactions. * The principle of operatioh of the field-emissionmicroscope lies in the modification, by a very large potential gradient, of the potential-energy barrier preventing the escape of electrons from the surface of the metal. Under the influence of a gradient of ca. lo7 v cm.-l, electrons can tunnel through the distortedF. J. Norton, Trans. Vacuum Syrnp., 1961, 8. J. P. Hobson, Trans. Vacuum Symp., 1961, 146. R. Gomer, " Field Emission and Field Ionisation," Oxford University Press,

1961.

GASSER: ULTRA-HIGH V A C U A

11

barrier at a rate which is approximately independent of temperature, in the A can be drawn a t temrange 0-300" K . Currents of the order of loA7 peratures hundreds of degrees below that at which thermionic emission is appreciable. The actual current obtained depends on the work function, #, of the surface in a complex manner. However, the appearance of a term of the form exp(4312)shows that i will vary very markedly with changes in 4. The large voltage gradients required for a field-emission microscope are obtained by making the field-emission source in the form of a very fine point (ca. 1000 A diameter), and mounting it at the centre of a bulb of about 5 em. radius. The inner surface of the bulb is coated with conducting and fluorescent layers and the application of 1-20 kv. to this screen produces the necessary voltage gradient at the tip. The fluorescent screen provides a visual demonstration of differences in work function of the surface, magnified some 105-106 times. The observed regions of light and dark, from a clean tip, are due to the differences in work function of the various crystal planes present at the surface. Adsorption of a gas alters the work function, and therefore the field-emission pattern, of a tip. This has been used to study the physical adsorption of inert gases on tungsten.9 with the perhaps surprising results (i) that the adsorbed gas retains liquid-like properties below the bulk melting point of the adsorbate, and (ii)that adsorption gives rise to a substantial dipole moment (0.1-0.8 D) in the adsorbate. The heats of adsorption, however, are in the expected range, 2-10 kcal. molew1. From the large number of papers dealing with chemisorption on fieldemission tips, it has been possible to build up a fairly detailed picture of the surface processes involving diatomic molecules and refractory metals, particularly tungsten. Hydrogen, oxygen, nitrogen, and carbon monoxide are all adsorbed rapidly, and without activation, even at low temperatures ((70" K). The homonuclear molecules are dissociated into atoms, but carbon monoxide is adsorbed without decomposition. The migration of adsorbed species over the surface has been studied at low temperatures (ca. 20-70' K). To do this, part of the tip is given a multimolecular-layer deposit of gas at liquid helium temperature, where the adsorbed layers are immobile. On warming the tip, migration of the physically adsorbed layers occurs (at 27 O K for oxygen) on to the clean regions of the surface, where they become chemisorbed and immobile. More physically adsorbed gas can now migrate over the new deposit and extend the region of chemisorption. This process produces changes in the fieldemission pattern, and it continues until either all the tip is covered or all the original deposit of gas is used up. If the latter happens, no further change in the field-emissionpattern occurs until the migration of chemisorbed atoms sets in at much higher temperatures (about 500" K for oxygen). The activation energies for the two types of migration are about 1 kcal. mole-1 and 25 lwal. mole-l. At still higher temperatures the gas desorbs and eventually the pattern of the clean tip is regained. From these measurements the heats of desorption are obtained. For many systems, the activation energy for migration is about one-fifth of the binding energy of the atom to the surface.OG. Ehrlich and F. G. Hudda, J . Chem. Phys., 1959, 30, 493.

12

G E N E R A L AND P H Y S I C A L CHEMISTRY

The field-emission microscope has recently been used to study the interaction of simple hydrocarbons with an iridium tip10 It was suggested that, for ethylene, adsorption is dissociative and that heating progressively dehydrogenates the adsorbed species, hydrogen being evolved, until a carbonaceous deposit is left. Related t o the field-emissionmicroscope is the field ion microscope. This is essentially a field-emission microscope operated in reverse with the tip as anode, instead of the screen. For its operation it requires a reasonable pressure of gas (ca. mm.), and either hydrogen or helium is commonly used. When a gas molecule comes into the region of high potential gradient, near the tip, it may be ionized by the tunnelling of one of its electrons on to the tip. The positive ion thus produced is accelerated to the screen. The pattern produced on the screen by many such ions is characteristic of the surface of the tip. The resolution of the field ion microscope is greater than that of the field-emissionmicroscope and, indeed, is so great that it has been possible t o see the effect of a single nitrogen molecule upon the pattern from a tungsten tip.11 This very refined technique is unfortunately limited t o the observation of very strongly-bound species. This is because the potential gradients required are some ten times greater than for the fieldemission microscope, and subject the tips to such great mechanical stress, ca. 1011 dyne cm.-2, that any weakly bound species, including atoms of the tip itself, may be field desorbed. Flash-$lament experiments. This technique was introduced in 1953 and, since then, has been widely used as a method of studying the interaction of simple gases with the very high-melting metals. The principle of the method is as follows. The metal is rapidly heated electrically to about 2000K, i.e. flashed, in an ultra-high vacuum to remove the adsorbed layers of atmospheric gases. The experimental gas is then admitted to the mm. range is established. On cooling apparatus and a pressure in the the filament, gas is adsorbed. After a chosen time the filament is flashed thus desorbing all the gas and causing a rapid rise in the pressure. Provided that the time constant of the pressure-measuring equipment is short compared with the flashing time (usually ca. 1 sec.), the evolution of gas can be recorded accurately on an oscilloscope or fast writing recorder. The kinetic theory of gases gives the number of molecules which have collided with a filament of known dimensions and the desorption curve gives the number which have stuck. The results of these experiments are expressed in terms of the probability that, on striking the surface, a molecule is adsorbed. For nitrogen on tungsten there is general agreement that there is a strong probability of a fruitful collision,12,13, l4 which is independent of coverage over a l6 wide range. Similar results have been obtained for carbon monoxide,15~l o J. l1 G. l 2 J.l3

l4l5 l6

G. J. G. J.

R. Arthur and R . S . Hansen, J. Chem. Phys., 1962, 36, 2063. Ehrlich and F. G. Hudda, J. Chem. Phys., 1962, 36, 3233. A. Becker and C. D. Hartman, J . Phys. Chem., 1953, 57, 153. Ehrlich, J . Chem. Phys., 1961, 34, 29. Eisinger, J . Chem. Phys., 1958, 28, 165. Ehrlich, J. Chem. Phys., 1961, 34, 39. Eisinger, J. Chem. Phys., 1957, 27, 1206.


13

oxygen,17 and hydrogen.18 I n all these cases the probability lay in the range 20-60y0. In marked contrast to these results was the behaviour of oxygen on silicon,lS where the sticking probability started at only lyo,for a clean surface, and declined exponentially with coverage. A detailed study of the shapes of the desorption curves gives further information about surface processes. A striking feature observed for many systems is the evolution of gas in discrete stages.13, 15, 20, 2 1 For example, after adsorption a t 115 K nitrogen is evolved from tungsten in three steps. The lowest arises from a physically-adsorbed molecular species while the other two are due t o chemisorbed atomic states, of which the predominant one has atoms bound with an energy of about 155 kcal. molev1. I n the desorption of carbon monoxide from tungsten 21 one physically adsorbed and three chemically adsorbed species have been identified. Kinetic analysis of the desorption curves is also possible and gives the order, and therefore the mechanism, of the desorption processes. Evaporated metalJilms. The use of evaporated metal films for the study of gas-metal interactions was introduced in 1940 and has given much information. The particular property measured in early experiments was the calorimetric heat of adsorption, and its variation with coverage.22 However, more recent work in this field has cast some doubt on the interpretation of the results.23 It was suggested that two effects may have occurred during the evaporation of these films which led t o the accumulation of gas of unknown composition in the system and therefore to an uncharacterized film. These effects are: (i) replacement by the evaporated metal of gas molecules adsorbed on the substrate surface, and (ii) heating of the inner wall of the vessel, which then desorbs gas. It is, however, possible to prepare an evaporated metal film under ultra-high vacuum conditions 24, 2 5 , 26 by evaporating the metal in short bursts. Measurements on such films include those of sticking probabilities and catalytic activity. I n the latter case, a substantial difference in behaviour between films prepared in the mm. and mm. regions has been found. The ultra-high vacuum films are more reactive, but are much reduced in catalytic activity by small quantities of oxygen. The Nature of Adsorptive Processes.-The first important distinction to be drawn in the discussion of surface processes is that between physical and chemical adsorption. At a low enough temperature any gas will be physically adsorbed by any surface. This completely non-specific character of physical adsorption is the result of the operation of van der Waals forces between gas and surface, these forces being themselves non-specific. On the other hand chemisorption is a very specific process, which is characterizedJ . Chem. Phys., 1959, 30, 412. J. Eisinger, J . Chem. Phys., 1958, 29, 1154. l9 J. Eisinger and J. T. Law, J . Chem. Phys., 1959, 30, 410. 2 o R. A. Pasternak and H. U. D. Wiesendanger, J . Chenz. Plays., 1961, 34, 2062. 2 1 P. A. Redhead, Trans. Paraday SOC., 1961, 57, 641. 2 2 B. M. W. Trapnell, Chemisorption, Butterworths, 1955. 23 T. W. Hickmott and G. Ehrlich, J . Phys. and Chem. Solids, 1958, 5, 47. 2 4 P. Della Porta and S. Origlio, Vucuum, 1961, 1 1 , 26. 2 5 S. Wagener, J . Phys. Chem., 1957, 61, 267. 2 6 R. W. Roberts, Trans. Faraday SOC., 1962, 58, 1159.1 7 J . Eisinger, l8

14

G E N E R A L A N D P H Y S I C A L CHEMISTRY

by the rearrangement of electrons t o give rise to a primary chemical bond. The experimental criterion which is widely used to distinguish between these two possibilities is the change in heat content of the process. An adsorption process occurs spontaneously, and therefore has a negative change in free-energy. Therefore in the equation

AF = AH - TAX AH will be negative, since normally the entropy of gas localized on a surface will be smaller than in the gas phase, making A X negative. Thus adsorptionis exothermic. Physical adsorption is akin to the condensation of a vapour and is expected t o have AH comparable with a latent heat of evaporation, i.e. I AH I < ca. 10 kcal. mole-I. Chemisorption should have the full value of AH associated with the formation of a chemical bond, i.e. I AH > ca. 20 kcal. moleA1. The precise nature of the bond involved in chemisorption has been widely discussed.2' It has been concluded that, although allrali-metal atoms are adsorbed on tungsten as the &positive ions, this is an uncommon mechanism and most simple species are held by predominantly covalent bonds. Evidence in support of the covalent-bond postulate come's from infrared spectroscopic studies of carbon monoxide adsorbed on iron; a peak at about 5 p has been attributed t o the metal-carbon monoxide linear structure.28 Attempts t o give a quantitative theoretical account of the bond involved in chemisorption have met with comparatively little success. This is scarcely surprising in view of the electronic complexity of the system and the failure of theoretical chemistry to give a detailed account of any systems other than very simple ones. Physical adsorption. Comparatively little work on physical adsorption seems to make use of specifically ultra-high vacuum techniques. There is, however, a series of papers dealing with the interaction of helium, nitrogen, and argon with a Pyrex-glass surface a t temperatures in the range 4-90' K.' The results can all be expressed in terms of the Dubinin-Radushkevich equation log10 0 = log10 0 , - D[log,o P/Pol2 where 0 is the number of molecules adsorbed per sq. cm., omis the number in a monolayer, D = A T 2 where A is a constant, P is the pressure above the adsorbed layer, and Po is the vapour pressure of the adsorbate a t the temperature of the experiment. It is the extrapolation of these results to low coverage of the surface which yields the equilibrium pressure in an ultra-high vacuum system a t very low temperatures, referred to earlier. Chemisorption. A new approach to the atomic and molecular processes occurring on metal surfaces, instead of the not very successful electronic treatment, has been suggested by E h r l i ~ h . 30 ~ ~ ,I n this approach the aim is to start with experimentally-determined energy parameters and interpret them in terms of the well-established concepts of chemical thermodynamics and kinetics.

I

2728 28

so

P. G. G. G.

M. Gundry and F. C. Tompkins, Quart. Rev., 1960, 14, 257. Blyholder and L. D. Neff, J . Phys. Chem., 1962, 66, 1464. Ehrlich, J . Chem. Phys., 1959, 31, 1111. Ehrlich, Trans. Vacuum Bymp., 1961, 126.


15

One of the most remarkable Qbservationswas the sharp cut-off of chemisorption from the gaseous phase of nitrogen, hydrogen, and carbon monoxide at the end of the three transition series. Among the transition metals themselves there is a considerable and apparently arbitrary variation in activity towards these gases. Thus, for example, nitrogen is not adsorbed by nickel and platinum, which do however adsorb hydrogen. This behaviour can be rationalized as follows. The energy parameters involved in the formation of a dilute layer of a diatomic gas, which is adsorbed as atoms, are (i) the dissociation energy of the molecule, D , and (ii) the binding energy of the atom to the metal, Only if is somewhat greater than D/2 will appreciable adsorption occur, e.g. for hydrogen, a dilute layer corresponding to about 1% coverage will only be stable at 10-3 mm. and 200" K if is greater than 54 kcal. mole-1 (D= 103 kcal. mole-l). Thus a metal may form reasonably strong bonds to the atoms, but if < D/2 an atomic layer, once produced on the surface, is only stable as long as migration of the ad32 ~~~ atoms is prevented. Such a situation can be realized e ~ p e r i m e n t a l l y . Both copper and mercury are inactive towards gaseous hydrogen but, at liquid air temperatures, chemisorb atomic hydrogen (readily produced from molecular hydrogen by the action of a hot filament). As long as the temperature is kept low enough to prevent migration, the layers are stable, but on warming, collisions of ad-atoms occur and the metal can no longer compete successfully, so that molecular hydrogen is evolved. Nitrogen has a very large dissociation energy so that it is only expected to be chemisorbed on those elements with which it forms a very strong bond; among the transition metals these are particularly titanium, zirconium, niobium, tantalum, molybdenum, and tungsten, which are also elements with which it forms very stable interstitial-type nitrides. Thus, whether or not chemisorption occurs depends on the balance between two large energy parameters. Moreover the dependence is an exponential one so that, just as in the case of the solubility of an ionic compound in water, extreme specificity is expected. Migration of adsorbed species. While moving over the surface of a metal an adsorbed atom experiences changes in potential energy which are characteristic of the surface,'and it can therefore serve as a probe for the investigation of that surface.29t 30 Two extreme types of potential-energy variation over the surface can be imagined. (i) The potential energy is uniform over the whole surface-in this case a migrating atom encounters no barrier and migration requires no activation energy. (ii) The potential energy is a minimum at an adsorption site and rises very rapidly on moving away from this site. If the potential energy curve does not overlap with that of a neighbouring site, an atom can only migrate by f i s t desorbing, and the activation energy for migration is the same as for atomic desorption. I n this context an adsorption site is identified with a surface metal atom. The results of field-emission microscope studies indicate that the situation is intermediate between (i) and (ii). Overlap between the potential energy curves of adjacent sites leads to a periodically-varying potential energy over the surface. The energy difference between the peaks and

x.

x

x

x

3 1 K . B. Blodgett, J . Chern. Phys., 1958, 29, 39. 32 J. Pritchard and F. C. Tompkins, Trans. Faraday SOC., 1960, 56, 540.

16

G E N E R A L A N D PHYSICAL CHEMISTRY

valleys is the activation energy for migration; this is usually about one-fifth of the binding energy of the atom to the surface. One very important conclusion can be drawn from the picture of a surface outlined above, namely that although it is meaningful to talk of adsorption " sites " these do not vary very greatly in energy from the rest of the surface. Even at the least favourable positions, i.e. on the potential energy peaks, the binding energy of the atom is still about 431/5. This emphasizes the difficulty which an attempt a t an electronic description of adsorption necessarily encounters, in that it has t o account for a strong surface-ad-atom interaction over the whole surface. Rates of adsorption and desorption. The above model of a surface can 30 The distance be used t o predict the activation energy for between adjacent potential energy minima is the lattice spacing of the metal which will not, in general, be the same as the internuclear distance in the diatomic molecule t o be adsorbed. The formation of the transition state in the reaction can be described as follows. One atom of the incoming molecule binds on t o a site of minimum potential. This makes it impossible for the other atom also to be a t a minimum since the distances do not match. It will be a t a site of energy higher by about the activation energy for migration. This situation is the transition state. The energy quantities involved in its formation are (i) the dissociation energy of the molecule, (ii) the strength of the gas-metal bond, and (iii)the activation energy for migration, Em,, (m. x / 5 ) . Then for the more strongly-bound atom the heat of adsorption, -AH,, is given by -AH, = x - o/2, and for the less strongly-bound atom

-AH,

=x

- D/2

- Emig..

The total energy taken in is the activation energy E and E = AH, A H , = D - 231 Emig. = AH Emig. where - A H is the change in heat content observed calorimetrically. As long as this has a reasonable value, the gas molecule has a lower energy even in its transition state than in the gaseous phase and no activation energy is required. To take a particular case, consider-hydrogen being adsorbed on tungsten. Here D = 103 kcal. mole-l, x = 74 kcal. mole-l, and Emig. = 16 lwal. mole-1. Thus E is negative and no activation energy is required. So far it has not been possible to give a convincing account of the depend~ the limited number ence of the rate of adsorption on c o n ~ e n t r a t i o n . ~ For of systems investigated, the rate of adsorption on metals is constant with coverage up to a critical adsorbed-atom/surface-atom ratio, and then diminishes rapidly. The critical ratio depends on the gas, temperature, and surface preparation. This constancy of rate is a surprising result, contradicting as it does the prediction of the Langmuir hypothesis of chemisorption. According to this, one would expect the rate of adsorption to depend on the rate a t which gas molecules encounter vacant sites on the substrate, and therefore to decrease with increasing coverage. The desorption of gases may follow either first- or second-order kinetics.

+

+

+

G A S S E R : ULTRA-HIGH VACUA

17

First-order kinetics are associated with the desorption of atoms or of a molecular gas which is adsorbed without decomposition. Thus for carbon monoxide the rate equation is -dn/dt = 3.1013.n.exp (-751O3/RT). Oxygen is alone among the diatomic gases in desorbing as atoms. The activation energy for desorption, 147 kcal. mole-I) is close t o the estimated value of the binding energy, as expected. Nitrogen and hydrogen are evolved from tungsten according to the second-order equation -dn/dt = A.n2.exp ( - E I R T ) . For nitrogen E is 81 kcal. mole-1 and for hydrogen, a t low coverage, it is 31 kcal. mole-1. This activation energy is also the heat of adsorption (2% - D ). The frequency factor in both cases closely approximates to the collision number of a two-dimensional gas. This behaviour can be rationalized as follows. Both gases are adsorbed with a substantial evolution of heat and with zero activation energy. On desorption therefore) the activation energy required will be close to the heat of adsorption. If the heat of an adsorption is small there will be no adsorption of the diatomic gas, though atomic adsorption may occur. The activation energy for desorption of such an atomic layer is also the activation energy for migration. The origin of the difference in the desorption kinetics of oxygen and other diatomic gases lies in the very strong bonds that oxygen forms to metals. For all systems there will be competition between the atomic and molecular desorption processes, though normally one or other will dominate. The ratio of atoms to molecules, for the systems quoted above, is given approximately by

For hydrogen and nitrogen on tungsten < D, and the exponential factor ensures that the molecular process dominates. For oxygen the reverse is true. One of the classical concepts of physical chemistry is the dynamic nature of chemical equilibrium, and it is instructive to examine the conditions for dynamic equilibrium between adsorbed and free gas-molecules. Adsorption will occur until the chemical potential of a molecule in the gas phase is the same as that of an adsorbed molecule, at which time the rates of adsorption and desorption will be equal. The theory of rate processes 33 gives the adsorption isotherm, on equating the two rates, as1- 8

x

e ___ _-__ fads. .exp ( & / k T )kTf,,,

where 6 is the fractional coverage, faas. andfgasare the partition functions of the adsorbed gas and free gas respectively, and E is the heat of adsorption evolved per molecule. If now P is reduced, then 8 must decrease. TheS. Glasstone, K. J. Laidler, and H. Eyring, McGraw-Hill, 1941.

The Theory of Rate Processes,

18


rate-determining step in the re-establishment of equilibrium will be the desorption process, since the adsorption process is non-activated. Separating out the desorption rate (dLv/dt)des. = B exp (-Q/RT) where Q is the heat of adsorption and the pre-exponential factor, B, can be expressed by K x f(O), where f (0) is the fraction of the surface from which desorption can occ&, and K is a constant. Now if the average time spent by a molecule on the surface is z then (cW/dt)des.= Nads./r = Nmono. -f(e>/z -*. K = (Nmono./z) -exp ( Q / R T ) .*. z exp (-Q/RT) = zo (a constant) . : z = zoexp (Q/RT). Thus as the heat of adsorption (Q) increases, so does the average time a molecule spends on the surface. I n the limiting case, where there is no binding energy, z = z , and ro is estimated to be of the order of the period of a vibration of the surface atoms, i.e. ca. 10-13 see. From this, the average lifetime of an adsorbed species for various values of Q can be calculated, e.g. a t room temperature for Q = 15 kcal. moleb1, z sec., while for Q = 30 kcal. mole-l, z 100 years. Thus, if the position of equilibrium is altered by reducing the pressure, the rate of attainment of equilibrium may be very small. From the foregoing survey it can be seen that, by studying the behaviour of very simple systems, some insight into the fundamental behaviour of gas-metal systems has been obtained. It remains, of course, to study the adsorption behaviour of more complex gases and of systems in which chemical reactions are occurring a t a surface.

-

-

R. P. H. G.3. GAS-PHASE OXIDATION

GAS-PHASE oxidation has been covered from time to time in Annual Reports 1 but the most recent reviews have been brief. The subject has now been comprehensively treated by Minkoff and Tipper in their book Chemistry of Combustion Reactions which covers the literature t o the end of 1961. However, a number of projects which were then at intermediate stages have since been materially advanced and merit more detailed treatment. The highlights of recent developments, in the Reporters opinion, are (1) the shock-tube experiments of Kistiakowsky and his co-workers on the oxidation of methane and acetylene, (2) the detailed work of Baldwin and his co-workers on the hydrogen-oxygen reaction, (3) the establishment, in considerable detail, of the mechanism of oxidation of methane and (4)the full confirmation of the thermochemistry of the HO, radical by Foner and Hudson. The oxidation of alkanes, olefins, and oxygenated compounds isAnnual Reports, 1950, 47, 39; 1954, 51, 83; 1955, 52, 13; 1957, 54, 41. G. J. Minkoff and C. F. H. Tipper, Chemistry of Combustion Reactions, Butterworths Scientifh Publns., London 1962.a

KNOX: GAS-PHASE OXIDATION

19

now, at last, yielding to sustained pressure and there is an increasing amount of more fundamental work being carried out/on the rates of elementary reactions of simple radicals such as 0, OH, H, CH,, etc. Experiments on the oxidation of alkyl radicals, produced other than by oxidising alkanes, are still in the early stages. Shock Tubes and Flames.-Somewhat unexpectedly the chemical reactions occurring in shock waves and flames are often simpler than those occurring at lower temperatures although the techniques for studying them are more m c u l t . This is because the reactions are very fast; only bimolecular reactions are important and slow degenerate branching reactions do not occur. Schott and Kinsey,3 and Skinner and Ruherwein 4 carried out the first successful oxidations in shock tubes with hydrogen and methane, respectively. The former authors observed the ultraviolet emission of OH from the shocked gas. and determined the temperature-dependence of the induction period ti. log [O,]ti (mole ~ m . sec.) - ~ = -13.65 17,100/4.576T. Since ti is inversely proportional to the rate constant of the branching 0, = OH 0. This reaction they deduced that this could only be H result has been confirmed by Suzuki and Fujimoto although, for some unexplained reason, their induction periods are some five times greater. Kistiakowsky and his co-workers 6 have extended the technique by using several different methods for following the oxidation of methane and acetylene in a shock tube. Gardiner has determined the induction period for the oxidation of acetylene between 1600" and 2020"E by measurement of the change in absorption of soft X-rays. His induction periods, for ignition of mixtures of 1.35% of acetylene and 2.04% of oxygen in xenon, are in precise agreement with those of Schott and Kinsey 3 for the hydrogen-oxygen reaction. Bradley and Kistiakowsky 8 have determined the mechanism of the reaction between 950" and 1090"K by use of a time-of-flight mass spectrometer attached to the shock tube. The induction periods observed agree with other determinations and these results, combined with those of Gardiner,' suggest very strongly that the branching reaction is again H 0, = OH 0. The observation of diacetylene in the products suggests that the other important reactions are 0 + C,H, = OH + C,H

+

+

+

+

+

C,H OH

+ C,H2 = C,H, + H + C2H2= H,O + C,H.

Kistiakowsky and Richards

have measured the vacuum ultraviolet

G . L. Schott and J. L. Kinsey, J. Chem. Phys., 1958, 29, 1177. G . B. Skinner and R. A. Ruherwein, J. Phys. Chem., 1959, 63, 1736. M. Suzuki and S. Fujimoto, 9th Internat. Symposium on Combustion, 1962, preprints. G. B. Kistiakowsky, Proc. Chern. SOC.,1962, 289. [Centenary Lecture.] W. C. Gardiner, J. Chem. Phys., 1961, 35, 2252. J. N. Bradley and G. B. Kistiakowsky, J. Chem. Phys., 1961, 35, 264. G. B. Kistiakowsky and L. W. Richards, J. Chem. Phys., 1962, 36, 1707.

20


emission from shocked acetylene-oxygen-argon mixtures, containing 8599% of argon, between 1400" and 2500" K. The induction periods again agree with those of Schott and K i n ~ e y . ~ The time constant for the acceleration of the reaction, z, was also determined. For the branching reaction H 0, = OH 0, the rate constant and time constant are related by [O,] x z = l/k. The calculated value of Jc agrees precisely with that deter0, = OH 0, from hydrogenmined by Fenimore and Jones lo for H flame studies. From the ratio of ti/z it is deduced that the reaction increases in rate by a factor of lo4 during the induction period. Hand and Kistiakowsky l1 have studied the ionisation accompanying the combustion, which closely parallels the light emission. They conclude, in the light of Hand's earlier experiments l2 on the light emission from acetylene-oxygen flames, that both light emission and ionisation arise from similar sources, namely reactions of CH with 0: CH + 0 = CHO+ + e-

+

+

+

+

CH

+ 0 = CO" + H = CO + H + hv.

Although CH is not included in Bradley and Kistiakowsky's mechanism 8 it probably arises from C,H 0, = CH CO, or C2H 0 = CH + CO. I t s presence in the detonation of acetylene-oxygen mixtures a t 2080"1 ~ ~ 1 3 and in flash-initiated explosions,14 is well established. The ionisation reaction is supported by Green and Sugden,15 who find that CHOf is by far the most important primary ion in hydrocarbon flames. The shock-tube oxidation of methane has been studied by Bradley and Kistiakowsky,8 Hand and Kistiakowsky l1 and by Asaba et aL16 While the different results are in good agreement, they are not as simply explained as are those from acetylene. The time constant of the acceleration is the same as that for acetylene-oxygen ignitions, but the induction periods are much longer and have a higher activation energy. Kistiakowsky et a E .8* l1 obtained 33 kcal., and Asaba et al. 53 kcal. for rich mixtures, and 20 kcal. for weak mixtures.16 Kistiakowsky et l1 conclude that the reaction in the induction period must be a straight-chain, forming a C, hydrocarbon. When this has accumulated the oxidation is effectively that of acetylene. The first part of this conclusion is confirmed by flame studies but the formation of a C, hydrocarbon does not seem necessary to account for 0, = OH 0. branching by the reaction H Fristom and his co-workers 1 7 have studied the methane-oxygen flame a t l/lOth and 1/20th atmosphere pressure, using mass-spectrometer probing

+

+

+

+

+

P. Fenimore and G. W. Jones, J . Phys. Chem., 1958, 62, 693. W. Hand and G. B. Kistiakowsky, J . Chem. Phys., 1962, 37, 1239. W. Hand, J . Chem. Phys., 1962, 36, 2521. 13 R. K. Lyon and P. H. Kydd, J . Chem. Phys., 1961, 34, 1069. 1 4 R. G. W. Norrish, G. Porter, and B. A. Thrush, Proc. Roy. SOC., 1953, A , 216,10 C. 11C. 1 2 C.1 5 J. A. Green and T. M. Sugden, 9th Internat. Symposium on Combustion, 1962, preprints. 16 T. Asaba, Y. Yoneda, N. Kakihara, and T . Hikita, 9th Internat. Symposium on Combustion, 1962, preprints. 1 7 R. M. Fristom, C. Gruenfelder, and S. Favin, J . Phys. Chem., 1960, 64, 1386; 1961, 65, 587; A. A. Westenberg and R. M. Fristom, J . Phys. Chem., 1960, 64, 1393; 1961, 65, 591.

165.


21

for analysis. They show that all important reactions are bimolecular. The reaction is at first a straight-chain, but later branches through formation of H atoms. These appear to arise from the reaction CO OH = CO, H, whose rate constant is well known.l8 Once H atoms are formed the reaction H 0, = OH 0 can occur. The basic correctness of this mechanism is confirmed by comparison of the calculated value of the rate constant of CO OH = CO, H a t 1950"K with the accepted value. Fristom has also carried out a study with a spherical flame l9 at reduced pressure and confirmed earlier findings. The more recent study indicates, in more detail, the reactions involving H, 0, CH,, and OH. The methane-oxygen-argon flame (5.5% methane) has been studied by Levy et aZ.20 The results agree with those of Fristom.lg I n the presence of 0-36y0of hydrogen bromide the time scale of the reaction up to the complete consumption of methane is doubled, and thereafter hydrogen bromide has no effect. The authors suggest that chain branching occurs by the 0, = CHO OH H and that hydrogen bromide reacts reaction CH, mainly with OH.20 It seems more likely, however, that branching is due to H 0, = OH 0 and that the hydrogen bromide intercepts H atoms HBr = H, Br. The bromine atoms regenerate hydrogen bromby H ide by reacting with methane, but as soon as the methane is exhausted the hydrogen bromide is quickly converted into Br,, and the rate of consumption of the residual carbon monoxide is then the same as in the absence of hydrogen bromide. Fenimore and Jones 21 have shown that the ethane and ethylene flames are similar t o that of methane. Both flames commence with a zone of stra,ight-chain reaction. The main reaction, removing ethylene, is 0 + C,H, = (C,H,O)* = CH, + CHO. This decomposition of the activated ethylene oxide molecule is confirmed by the high yield of methane (up to 20% of ethylene consumed) obtained in the presence of a large excess of hydrogen. The rate constant derived for addition of oxygen atoms to ethylene agrees well with values derived from low-temperature work.,, With ethane, the main attacking species are H and OH but the fate of the ethyl radical is uncertain. The main conclusion from the flame and shock-tube studies is that hydrocarbon flames (apart from acetylene) are essentially hydrogen-carbon monoxide-oxygen flames fed by the decomposition products of a straightchain, oxygen-catalysed, pyrolysis of the hydrocarbons. This conclusion casts doubt on the interpretation advanced by Falconer and van Tiggeln 2 3

+

+

+ +

+

+

+

+

+

+ +

+ +

L. I. Avramenko and R. V. Lorentso, Zhur. $2. Khirn., 1950, 24, 207; C. P. Fenimore and G. W. Jones, J . Pizys. Chem., 1958, 62, 1578. l9 R. M. Fristom, 9th Internat. Symposium on Combustion, 1962, preprints. 2o A. Levy, J. W. Droege, J. J. Tighe, and J. F. Foster, 8th Internat. Symposium on Combustion, Williams and Wilkens, Baltimore, Md., 1962, 524. 21 C. P. Fenimore and G., W. Jones, 9th Internat. Symposium on Combustion, 1962, preprints. 2 2 ( a ) R. J. Cvetanovic, J . Chern. Phys., 1960, 33, 1063; ( b ) L. Elias and H. I. Schiff, Canad. J . Chem., 1960, 38, 1657. 2 3 W. E. Falconer and A. van Tiggeln, 9th Internat. Symposium on Combustion, 1962, preprints.

22


for their flame-speed measurements on higher hydrocarbons. They have calculated overall activation energies of 3 0 4 0 kcal. and mean molecular weights of chain carriers of 23-28, for the flames of the butanes and neopentane. The results are interpreted in terms of mechanisms similar to those proposed for low-temperature slow oxidations. The validity of the extrapolation is certainly doubtful when it is remembered that even at 500" c pyrolysis reactions are becoming dominant in hydrocarbon combustion.24 Hydrogen-Oxygen &action.-Our present understanding of this reaction owes much to the work of Baldwin and his co-w~rkers.~~, 26 It is now well established that in an aged reactor coated with boric acid, which preserves HO, and H202, the main elementary reactions occurring just above the second limit areOH

+ H, = H,O + H H + 0, = OH + 0 0 + H, = O H + H

(1)(2)

(3)(4)

H 3-0, + M = H 0 2 + MHO,

+ HO,+ M'

= =

H20, 20H

+ 0,+ M'

H,O,

(the numbering of equations in this section is that of Baldwin). The sequence (l), (4), (lo), (7) constitutes a straight-chain, and reactions (2) and (3) are responsible for chain branching. Chain termination probably occurs by

+ H,O, = H,O + OH O H + H20, = H,O + HO,H

(14)(15)

since H,02, by virtue of reaction (7), can be regarded as equivalent to two free radicals. This mechanism, with minor modifications, predicts with high precision the kinetics of the slow reaction outside the second limit. The rate of oxidation is only slightly greater than the rate of decomposition of the hydrogen peroxide present. Various rate-constant ratios can be obtained by comparison of predicted and experimental rates. For example, the ratio k15/k1 is found to be 7.1 a t 500" c . A value of 6 has been obtained by Baldwin and Bratten 27 from a study of the decomposition of hydrogen peroxide, in the presence of hydrogen, in a flow system. This study has firmly established that the reaction is unimolecular in its second-order region. The rate constant, with nitrogen as activating gas, is k, = 1.7 x lo1, exp (-46,000/1.9872') mole-1~ 1 1 1 sec.-l .~

2 4 V. Ya. Shtern and N. Ya. Chernyak, Doklady Akd. Nauk S.S.S.R., 1951, 78, 91; J. W. Falconer and J. H. Knox, Proc. R o y . SOC., 1959, A, 250, 493. 26 R . R. Baldwin and L. Mayor, Trans. P a r a d a y SOC., 1960, 56, 80, 103; R. R. Baldwin, P. Doran, and L. Mayor, ibid., p. 93. 26 R. R. Baldwin, P. Doran, and L. Mayor, 8th Internat. Symposium on Combustion, Williams and Wilkins, Baltimore, Md., 1962, p. 103. 2 7 R. R. Baldwin and D. Bratten, 8th Internat. Symposium on Combustion, Williams and Wilkins, Baltimore, Md., 1962, p. 110.

K N O X : GAS-PHASE OXIDATION

23

in full accord with all reliable previous work.2* The study 27 also estnbmust be (14) not (14a), lished that the reaction between H and H202H + H,O, = H, + HO2 (14a) since hydrogen accelerates the decomposition. If reaction (14a) occurs, hydrogen should have no effect on the rate since the sequence

is simply replaced by

which is kinetically equivalent.

If reaction (14) occurs we have

This is now a chain reaction. The work also establishes that reaction (6) is inadmissible since it would requireHO,

+ H 2 0 2= H 2 0 +

0 2

+ OH,

(6)

the decomposition of H202 in inert gases, to be a chain reaction. The earlier work of Baldwin et aZ.25showed that as the second limit is approached the quadratic branching reaction (8) HO, + H = 20H (8) becomes important. The overall reaction should still be entirely homogeneous and the nature of the surface should have no effect on the explosion limit. This is not strictly true and, in newly coated vessels, the limit was somewhat raised, indicating some surface initiation a t the limit. 2 9 I n the slow oxidation the hydrogen peroxide concentration builds up to a maximum a t the maximum rate of reaction. Since the position of the second limit is generally found by the withdrawal method, there is some doubt as to whether the peroxide reaches a stationary concentration before the mixture explodes. Baldwin and Doran 29 have investigated the effect of withdrawal rate, and of arrest of the withdrawal, and find that both have a considerable effect on the limit. They conclude that a t 500"c the stationary concentration of hydrogen peroxide is reached before ignition, but a t lower temperatures the build-up is slow since the limit a t first increases as the withdrawal rate is reduced, although further reduction lowers the limit again. The latter effect is attributed to the effect of accumulated water which, because of its high third-body efficiency, increases the rate of reaction (4) t o the detriment of (2). A more extensive study a t 500" c has confirmed this,30 and predicted that water should be about five times as efficient as hydrogen in reaction (4). This explanation has been substantiated by the direct experiments of28 P. A. Giguere and I. D. Liu, Canad. J . Chem., 1957, 35, 283; D. E. Hoare, J. B. Protheroe, and A. D. Walsh, Trans. Paraday SOC., 1959, 55, 548; C. K. McLane, J . C h m . Phys., 1949, 17, 379; C. N. Sattedeld and T. W. Stein, J . Phys. Chem., 1957, 61, 537. R. R. Baldwin and P. Doran, Trans. Paraday Soc., 1961, 57, 1578. 30 R. R. Baldwin, P. Doran, and L. Mayor, Trans. Paraday SOC., 1962, 58, 2410.

24


Baldwin and Brooks 31 on the inhibiting effect of water a t the second limit in vessels coated with potassium chloride and boric acid. The efficiency relative t o hydrogen is 6.4 and is unaffected by temperature or surface between 460" and 540" C . This value compares well with previous values 5.5 (Nalbandyan 32a), 5.0 (Voevodskii and Talrose 32b), 6-5 (Ashmore and Tyler 3Zc). The first limit has been reinvestigated by Ivanov and Nalbandyan 33 with a potassium tetraborate surface. The results confirm Baldwin's earlier conclusion 34 that the predominant termination reaction is removal of hydrogen atoms a t the walls. They obtain a termination efficiency of 0.5 x which compares with Baldwin's value, in a vessel freshly coated with potassium chloride, of 6.6 x a t 520" c. Inhibition of the hydrogen-oxygen reaction. Baldwin and Simmons 35 carried out a careful investigation of the inhibition of the hydrogen-oxygen reaction by ethane, a t the first and the second limit, in a potassium chloridecoated vessel and showed that the effect was due to removal of H atoms by the reactionH

+ CzH6

=

Hz

+ C2H5.

(16)

The alternative inhibition reaction with OH was unimportant, except a t high oxygen mole-fractions. The work showed that the same proportion of ethyl radicals resulted in chain-termination, whatever the pressure, and they therefore concluded that the final removal of ethyl was by the reactions (17) and (5). C2H.5 0,= C2H4 HO2 (17)

+

+

H 0 2 = Wall destruction

(5)

The main features of the inhibition by ethane were the linear fall in the second limit with ethane mole-fraction, i, and the inverse quadratic-relation at the first limit i = / I- u / P 2 (where P = explosion limit pressure). The results with added methane were strikingly The second limit was only slightly depressed until a critical concentration of methane was reached, when explosion was suppressed completely. Baldwin, Corney, and Walker 36 had to assume that some product of the reaction was responsible for the suppression of explosion. They also concluded that, a t concentrations of methane just insufficient to suppress explosion, the limit was thermal, not isothermal as is usual a t the second limit. A significant observation was the high rate of slow reaction when the inhibitor concentration was just sufficient t o suppress ignition. It seemed most likely that the inhibition arose from formaldehyde. The course of the reaction was visualised as follows. Initially there is virtually no inhibition since the only likely reaction of CH, with oxygen is CH, O2 = CH20 OH, which yields

+

+

R. R. Baldwin and C. T. Brooks, Trans. Furaday SOC.,1962, 58, 1782. 32 (a) A. B. Nalbandyan, Zhur. $2. Khim., 1945, 19, 210; (b) V. V. Voevodski and V. L. Talrose, ibid., 1948, 22, 1192; (c) P.G . Ashmore and B. J. Tyler, J. Catalyszs,31

1962, 1, 39. 33 0 . A. Ivanov and A. B. Nalbandyan, Kinetikai Kutaliz, 1960,1,337 (transl. 311). 3 4 R. R. Baldwin, Trans. Faraday Soc., 1956, 52, 1344. 3 5 R. R. Baldwin and R. F. Simmons, Trans. Furaday XOC., 1955, 51, 680; 1957, 53, 955, 964. 36 R. R. Baldwin, N. S. Corney, and R . W. Walker, Trans. Furaday Xoc., 1960, 56, 802.


25

The concentration of H therefore increases exponentially. At the same time the concentration of CH,O increases and begins to inhibit the reaction by virtue of reactions (18) and (19).

OH rather than HO,.

CH,O CHO

+ H = CHO + H2 + 0, = CO + HO,.

(18) (19)

The final stationary concentration of CH,O may, or may not, be sufficient to suppress ignition completely. The experimental results suggest that the establishment of a critical concentration of formaldehyde, at which the branching coefficient is zero, is not a sufficient condition for suppression of ignition, but rather that the rate of the initial burst of reaction must be below a critical value. The final condition is thus thermal. This is supported by the strong effect of diameter on the critical inhibitor concentration and the slight effect of surface. The more recent work of Baldwin, Booth, and Walker 37 has confirmed this view. With methane concentrations just sufficient to suppress the limit, they observe a whitish glow when the expected explosion limit is crossed during withdrawal of the mixture. The glow is often accompanied by a small pressure pulse due to self-heating. The thermal nature of the limit is finally established by the effect of inert gases. Confirmation of the rBle of formaldehyde has been obtained by Baldwin and Cowe 38 who have studied the inhibition of the first and the second limit. The inhibition picture is similar to that of ethane. It requires the removal of CHO radicals by CHO 0, = CO HO, rather than by CHO = CO H or CHO 0, = CO, OH. The rate constant for reaction (18) is El, = 6 x 10l1T4 exp (-3000/1.987T) mole-1 ~ m sec.-l. . ~ The data also enable k, to be determined. The value when taken with those obtained by Fenimore and Jones l o from flame studies gives E, = 3.3 x 1014 exp (-17,600/1-987T) mole-1 ~ 1 1 1 .sec.-l. ~ Voevodskii has used the inhibition of the hydrogen-oxygen reaction by hydrocarbons to evaluate other rate constants of hydrogen-atom reactions. The values have recently been listed.39 Azatyan, Voevodskii, and Nalbandyan40 have obtained a value of E, by an ingenious and independent method. The first explosion limits of mixtures of carbon monoxide and oxygen, containing 0.7-8y0 of hydrogen, were determined. When a suitable surface such as magnesium oxide is chosen all termination reactions are diffusion-controlled and the mechanism is rather simple, comprising O H + CO = CO, + H (20) H + 0, = O H + 0 (2)

+

+

+

+

+

0

0 H37

+ H, = O H + H + Wall = Destruction + Wall = Destruction

(3) (21) (22)

R. R. Baldwin, D. Booth, and R. W. Walker, Trans. Faraday SOC., 1962, 58, 60. 3 8 R . R. Baldwin and D. W. Cowe, Trans. Faraday SOC.,1962, 58, 1768. 39 V. V. Voevodskii and V. N. Kondratiev, Progress in Reaction Kinetics, ed. G . Porter, Pergamon Press, London and New York, 1961, Vol. I, p. 41. 40 V. V. Azatyan, V. V. Voevodskii, and A. B. Nalbandyan, Kinetika i Kataliz, 1961, 2, 340 (transl. 315).

26

GENERAL AND PHYSICAL C H E M I S T R Y

and the explosion limit is given by[O,him. =

(b1/2k2)(1 ~22/h[Hzl)* Since E,, and k,, can be calculated from diffusion theory, E, and k, can be determined by measurement of the explosion pressure a t different small concentrations of hydrogen. The results give k, = 1.0 x 1014 exp (-15,900/1.9862') mole-1 ~ r n sec.-l .~

+

k, = 6.7 x 1013 exp (-1lY700/1-986T) ), Y, YY The absolute value of E, calculated a t 1400" K is half the value obtained by Fenimore and Jones lo and the value a t 800" K is identical with that of Baldwin and C ~ w e . ~ The * enthalpy change in reaction (2) is +16 kcal. and therefore the back reaction probably has zero activation energy. Oxidation of Carbon Monoxide.-Little advance has been made in this field in the last year or two. Cusin and James have studied the inhibition of the oxidation by cyanogen (C,N,) 41 and by ethane.42 The two systems show similar features. At about 900" c, the ignition of either cyanogen alone or cyanogen-carbon monoxide mixtures takes place only when all the cyanogen has been converted into carbon monoxide. The mechanism of inhibition is probably 0 + C,N2 = CNO + CN i n competition with the branching reactions 0 + co = CO,*CO,*

+ 0 , = co, + 2 0 .

The kinetics of the inhibition suggest that the initiation of the oxidation is a first-order activation process - rather than the bimolecular reaction CO 0, = CO, 0. At lower temperatures (600-700" c) and low concentrations the inhibitor seems to act by adsorption on sites which can normally initiate the oxidation. At higher pressures it also acted homogeneously, as a t high temperatures. Kondratiev and Ptichkin 43 have studied the oxidation of carbon monoxide by ozone between 25" and 160" c. The reaction is chemiluminescent and the authors propose that the initiation is by decomposition of ozone t o give oxygen atoms. They 'deduce from their results, and assumed mechanism, that E ( 0 CO) - E ( 0 0,) = 3-8 kcal. The Slow Oxidation of Formaldehyde and Methane.-The slow oxidation of formaldehyde and methane has received renewed attention from Russian workers. FormaZdehyde. Markevich and Filippova 4 4 have used Kovalski's differential calorimetric method 4 5 to show the profound effect of surface on the

+

+

+

+

41F. Cusin and H. James, J. Chirn. phys., 1961, 58, 162, 730.42

43

F. Cusin and H. James, J . Chim. phys., 1962, 59, 454. V. N. Kondratisv and I. I. Ptichkin, Kinetika i Kataliz, 1961, 2, 492 (transl.

449).44

A. M. Markevich and L. F. Filippova, Zhur. $2. Khim., 1959, 33, 2214 (transl.

4 5 A. A. Kovalski and M. L. Bogoyavlenskaya, Zhur. $2. Khim., 1946, 20, 1325; see Semenov " Some Problems in Chemical Kinetics " (Transl. Bradley), Pergamon Press, London and New York, 1959, Vol. I, 187.

358).


27

oxidation and have thereby confirmed previous work carried out in static 46 and flow systems,*7 and partly explained the irreproducibility of much of the earlier work.48 With Pyrex reaction vessels coated with potassium tetraborate about 70% of the heat of the reaction is liberated close to the walls of the reaction vessel, yet throughout the reaction the temperature rise a t the centre gives an accurate measure of the rate of reaction so that there was no change in the percentage heterogeneity. (See also V a n ~ e e . ~ *I)n this reaction vessel, the only detectable products are carbon monoxide, dioxide, and water, and the rate of reaction is given accurately by Rate = Ic[CH20]2[0,]0. These results combined with those obtained in a flow system 47 give log,, k = 19.1 - 50,000/4*5817 (k in mole-1 ~ m sec.-l) . ~ between 380" and 550" c. With a clean surface the reaction is faster, and in clean silica peroxidic products can be isolated. They appear to arise from hydrogen peroxide initially formed by the reactions CH20 + HO2 = CHO + H 2 0 2 CHO + 0 , = CO + H02. I n the borate surface any hydrogen peroxide would be rapidly d e ~ t r o y e d . ~ ~ Markevich and Filippova 4 4 f h d marked activation-energy differences between the borate-coated surface (50 kcal.) and the clean silica surface (26 kcal.). This observation may have some bearing on the activation energy changes observed by Harding and Norrish 50 in the oxidations of formaldehyde and of ethylene (where formaldehyde is probably the branching intermediate) from 21 (26) kcal. at 350" c t o 39 (42) kcal. at 460" c and (53) kcal. a t 550" c (values in parentheses are for ethylene). The isolation of considerable amounts of hydrogen peroxide from the oxidations of both formaldehyde and methane raises the question of its role as a possible branching agent in these oxidations. The definitive work of Baldwin and Bratten 27 gives the following lifetimes for hydrogen peroxide (~ k, in an inert-gas concentration (assumed nitrogen) of 10-5 mole ~ m . about 0-5 atm. at 340" c).Temp. ("c) Lifetime (sec.) 360 5600 380 400 420 1850 650 230 440 90 460 37 480 500 16 7.2 620 3.4

Markevich and Pecherskaya 51 have found that, in a molybdenum-glass4 6 M. D. Scheer, 5th Internat. Symposium on Combustion, Williams and Willrins, Baltimore, Md., 1955, 435. 4 7 A. A. Anisonyan, S. Ya. Beider, A. M. Markevich, and A. B. Nalbandyan, Zhur. Jiz. Klzim., 1959, 33, 1695 (transl. 115). 4 8 R. Fort and C. N. Hinshelwood, Proc. Roy. Soc., 1930, A, 129, 284; R. Spence, J., 1936, 649; F. F. Snowdon and D. W. G. Style, Trans. Paraday SOC., 1039, 35, 426; D. W. E. Axford and It. G. W. Norrish, Proc. Roy. SOC., 1948, A , 192, 518; M. Vanpee, Bull. SOC. chim. belges, 1953, 62, 285; A. M, Markevich and L. F. Filippova, Zhur. $2. Khim., 1957, 31, 2649. 4 9 D. E. Cheaney, D. A. Davies, A. Davis, D. E. Hoar0, J. Protheroe, and A. D. Walsh, 7th Internat. Symposium on Combustion, Buttemorths Scientific Publns., London 1959, 183. A. J. Harding and R. G. W. Norrish, Nature, 1949, 163, 797. 61 A. M. Markevich and Yu. I. Pecherskaya, Zhur. $2. Xhint., 1961, 35, 1418 (transl. 697).

28

GENERAL A N D PHYSICAL CHEMISTRY

reaction vessel, the oxidation a t 330", 390", and 420" c commences with no pressure change although formaldehyde is consumed, probably by the overall reaction CH,O 0, = CO H,O,, and that the reaction is slightly au tocatalytic, as previously observed by Vanpee.48 The addition of up to 10 mm. of hydrogen peroxide to 15 mm. of formaldehyde 300 mm. of air a t 420" c has no lasting catalytic effect, although it removes the slight initial autocatalysis. The results of Baldwin and Bratten 27 show that the pyrolysis of 10 mm. of hydrogen peroxide a t 420" c should generate hydroxyl radicals at a rate of 5 mm. min.-l. The observed initial rate, in the absence of added peroxide, is 4 mm. min.-l and with 10 mm. of peroxide 12 mm. min.-l. Either the chains are exceedingly short or the decompo. sition of the peroxide is somehow short-circuited and does not give hydroxyl radicals. This problem is general throughout hydrocarbon oxidation where substantial yields of hydrogen peroxide are obtained, but where it has very little catalytic effect. A possible explanation is that activation of the peroxide is short-circuited by the collision of a partially activated molecule with formaldehyde, and the occurrence of the reaction

+

+

+

HO*OH*

+ CH,O

=

2H20

+ CO.

In other words, we are dealing with a non-Maxwellian distribution of energy among the peroxide molecules, particularly in the region of 40 kcal. and above. The above reaction is sterically plausible and highly exothermic, and there is some evidence for it in the oxidation of methane. Methane. An important series of papers on the oxidation of methane has appeared under the names of Karmilova, Enikolopyan, Nalbandyan and Semenov. The reaction was studied in a static system in a 200 ml. silica reaction vessel washed with hydrofluoric acid, and aged for 12 months. The surface was believed to be one of silicic acid and therefore of type (i) in Walsh's nomenclature.49 The reaction was studied a t 423,52a472", 491", and 513" c.52b The reaction is initially autocatalytic and a t a small percentage conversion reaches a maximum rate which is maintained, in spite of consumption of the reactants, until nearly the end of the reaction. The major products are carbon monoxide, dioxide, and water, with traces of hydrogen and methanol. Formaldehyde and hydrogen peroxide are intermediate products formed in small yield. The stoicheiometry of the reaction up to the maximum rate is accurately represented byCH,

+ $0, = CO + 2H20.

Later, carbon monoxide is oxidised t o dioxide and the yield of the former sometimes passes through a maximum. The overall activation energy of the reaction is 41.5 kcal. for the stoicheiometric mixture. This agrees with Egerton, Minkoff, and Salooja's value 53 for the same mixture but the latter also obtained lower values ( 2 3 4 1 kcal.) for richer mixtures. Karmilova, Enikolopyan, and Nalbandyan 5 2 tacitly assume that the activation5 2 L. V. Karmilova, N. S. Enikolopyan, and A. B. Nalbandyan, ( a ) Zhur. $2. Khim., 1957, 31, 851; ( b ) ibid., 1960, 34, 550 (transl. 261). 5 3 Sir A. C. Egerton, G. J. Minkoff, and K. C. Salooja, PTOC. Roy. SOC.,1956, A , 235, 158.


29

energy is independent of the methane : oxygen ratio. The overall order of the reaction is 2.7 and the partial orders 1.62 for methane, 0.96 for oxygen, and 0.10 for total pressure. The low exponent of total pressure contrasts with the value of unity obtained by Norrish and F0ord.5~ However, they used a soda-glassreaction vessel [type (ii)or (iii) surface] and their rates were about 20 times smaller. Diffusion-controlled termination might therefore have occurred in their experiments. Karmilova et ~ 1 . ~ 5 investigating 5 the r61e of formaldehyde, have shown that it appears somewhat before hydrogen peroxide although both reach ma-ximumpressures at the maximum rate of reaction. From Markevichs results 44 on formaldehyde they conclude that the peroxide arises from the oxidation of formaldehyde and is not the degenerate branching intermediate, as had been suggested by Egerton, Minkoff, and Salooja.53 The maximum formaldehyde concentration rises, with an activation energy of 7.8 kcal., while that of hydrogen peroxide falls; it increases linearly with methane pressure and is independent of oxygen pressure, above 50 mm. The induction period of the reaction falls, with an activation energy of 36 kcal. These observations are quantitatively explained by the mechanism advanced by Karmilova, Enikolopyan, Nalbandyan, and Semenov : 5 6

Denoting the rate of the ith reaction by ri it was not difficult to show that, at maximum rate when the chains are long, r2 = T , ~ ; r2 = r,; and r l = r2 r2, = r5 r,. Hence the maximum formaldehyde concentration is given by FmaXa = (kJ~4/k2are likely to be small while E , - E5#should be somewhat less than the difference in bond strength between methane and formaldehyde (about 25 kcal.). The experimental value of E , = 7 . 8 kcal. is obviously reasonable. The mechanism also predicts that the reaction should be of second order in methane and first order in oxygen, in reasonable agreement with experiment. The acceleration constant is (k,rC,/k,)[CH,][O,]. If the reaction accelerates exponentially during the induction period, ti, the temperature dependence of l/ti should be the same as that of 4.

+

+

+

R. G. W. Norrish and S. G . Foord, Proc. Roy. SOC.,1936, A, 157, 503. L. V. Karmilova, N. S . Enikolopyan, and A. B. Nalbandyan, Zhur. $2. Khim., 1960, 34, 990 (transl. 470). 5 6 L. V. Karmilova, N. S. Enikolopyan, A. B. Nalbandyan, and N. N. Semenov, Zhur. $2. Khim., 1960, 34, 1176 (transl. 562).5455

30


The experimental activation energy of the induction period was 36 kcal. and this agrees well with that predicted from the mechanism, since Ed = Em,,. rate - E, = 33 kcal. Furthermore, Ed should be somewhat greater than E,, which is expected to be about 30 kcal. The mechanism gives a very good account of the kinetics of the reaction up to the point of maximum rate and vindicates Norrish's original postulate 57 that formaldehyde is the branching intermediate. Features of the reaction which are not considered are (1) the role of hydrogen peroxide as a possible branching agent, (2) the maintenance of the maximum rate of reaction in spite of consumption of reactants, and (3) the formation of carbon dioxide. I n a fourth paper 58 the effect of added hydrogen peroxide is examined. Even in amounts several times in excess of those normally present a t maximum rate, the induction period is only slightly reduced and the maximum rate unaltered. This has been confirmed by Mari et aL59 It is, however, interesting that a t 472" c, when 1.5 mm. of peroxide is added t o 235 mm. of a stoicheiometric mixture, the yield of formaldehyde initially rises rapidly. Thereafter it rises more slowly to the same maximum as is obtained in the absence of peroxide. Meanwhile, the peroxide concentration falls gradually but is always greater than that normally present a t maximum rate. It appears then, that hydrogen peroxide can act as an initiator in the earliest stages but not once any quantity of formaldehyde has appeared. The constancy of the rate of oxidation, in spite of considerable consumption of the reactants, has been noted both by Karmilova et ~ 1 . and ~ 2 by ~ Mari et aZ.,6o but their explanations are radically different. The former 6 1 suggest that a new chain-branching system appears in the later stages of the reaction which just compensates for the expected decline in rate as methane and oxygen are consumed. The system involves the formation of hydrogen atoms which can cause branching by reaction (9). CO + O H = C 0 2 + HH2 + O H = H 2 0 + H H + 0, = O H + 0 H + RH = H2 + R

This scheme also accounts for the formation of carbon dioxide. Application of the kinetic tracer method 6 2 has confirmed the general correctness of this idea by showing that all the carbon dioxide produced in the later stages arises from carbon monoxide. Additional confirmation comes from the calculation of the relative rates of removal of carbon monoxide and methane by using the known rate constant data for the reactions of OH.lS57

R. G. W. Norrish, Proc. Roy. SOC.,1935, A, 150, 36.

L. V. Karmilova, N. S. Enikolopyan, and A. B. Nalbandyan, Zhur. $2. Khim., 1961, 35, 1043 (transl. 512). 59 R . Mari, M. Letort, M. Dzierzynski, and M. Niclause, J. Chim. phys., 1962, 59, 596. 6 o R. Mari, M. Letort, and M. Niclause, J. Chim. phys., 1962, 59, 324. 61 L. V. Karmilova, N. S. Enikolopyan, A. B. Nalbandyan, and V. T. Il'in, Zhur. $2. Khim., 1961, 35, 1435 (transl. 706). 8 2 L. V. Karmilova, N. S. Enikolopyan, and A. B. Nalbandyan, Zhur. $2. Khim., 1961, 35, 1458 (transl. 717).58


31

Mari, Letort, and Niclause,60 on the other hand, consider that the high rate of reaction is maintained because of modification of the surface of the reaction vessel. Mari et uZ.63 have shown that it is necessary to add some eight times the quantity of formaldehyde normally present a t maximum rate, in order to make the reaction start a t this rate. But the pressure-time curves are then of an unexpected shape, starting at a high rate which then declines and finally increases again. This is thought to be due t o acclimatisation of the reaction-vessel surface during the reaction, and pretreatment of the Pyrex reaction vessel with water vapour for several hours certainly accelerates the subsequent reaction. Although the theory of Karmilova et al. is the more attractive, some account should certainly be taken of surface modification during the course of a reaction. The theory of Karmilova et al. supposes that all methyl peroxy radicals decompose to formaldehyde and hydroxyl. Fisher and Tipper 6 4 have shown that this may not be so. In an aged silica reaction vessel a t 400" c measura.ble yields of methyl hydroperoxide are formed in the acetonephotosensitised oxidation of methane and, subsequently, in the oxidation of methane alone. Pyrolysis of methyl hydroperoxide a t 345" c gives formaldehyde and methanol as the major products, probably mainly by heterogeneous decomposition. The methanol which Karmilova et al. report may have come from this source. The lifetime of the peroxide is about 30 sec. and its stability, as regards homogeneous decomposition, must be considerably greater than that of the higher alkyl hydro peroxide^.^^ It is probable that CH,O, abstracts hydrogen from formaldehyde rather than from methane. The oxidation of methane by traces of oxygen has been studied a t 950-1050" c by Germain and Sueur,66 in a flow system. The reaction is autocatalytic and the maximum rate is held until all the oxygen (initially 0--5%) is consumed. The main products are H,, H,O, CO, CO,, C2H4, and C,H,. The carbon dioxide probably arises by reaction (7), molecular hydrogen from H atoms, ethane from unoxidised methyl, and ethylene from ethane. The mechanism is thus similar to that a t lower temperatures. The high-temperature oxidation of methane has also been investigated by Cabannes and his co-workers 6 T who passed a mixture of methane and oxygen over a platinum plate heated to between 600" and 1300"c. From the profile of gas velocity, temperature, and composition, determined by suitable probes, the degree of heterogeneity can be calculated. On clean platinum the reaction is 20% heterogeneous and on platinum coated with alumina it is entirely homogeneous. Enikolopyan and Bel'gorskii 6 8 have compared the oxidation of methane and methanol in glass reactors, coated with potassium tetraborate and6 3 R. Mari, M. Letort, M. Dzierzynsbi, and M. Niclause, J. Chim.phys., 1962, 59, 589; Compt. rend., 1961, 252, 3241. 6 4 I. P. Fisher and C. F. 3. Tipper, Nature, 1962, 195, 489. 65A. D. Kirk and J. H. Knox, Trans. Faraday SOC.,1960, 56, 1296. 6 6 J. E. Germain and R. Sueur, Bull. SOC. chim. France, 1961, 1008. 67 F. Cabannes and Y . Fukuchi-Thibaut, Bull. SOC. chim. France, 1961, 947; F. Cabannes and P. Valentin, ibid., 1962, 166. 6 8 N. S. Enikolopyan and I. M. Bel'gorskii, Zhur. $2. Khim, 1960, 34, 1571 (transl.

749).

32


silver, using differential calorimetry. On the former surface both reactions are homogeneous and autocatalytic, and depend upon formaldehyde as the branching agent. On silver both are heterogeneous and, while the oxidation of methane produces no formaldehyde, that of methanol produces more formaldehyde than the homogeneous oxidation. Thus the effect of silver must be to destroy free radicals, rather than formaldehyde. Barber and Cuthbert 69 have reported some preliminary work on the oxidation of methane in an electric discharge (200 w, 13,560 kc. sec.-1), analysing the products by mass spectrometry. With the discharge through the methane the products are the same as without oxygen (ethane, hydrogen, traces of acetylene). With the discharge through the oxygen the major product is ethane, with smaller amounts of methane and carbon dioxide. Enikolopyan and Konereva 70 have studied the oxidation of methane, catalysed by nitromethane, with the object of establishing whether nitromethane or nitrogen dioxide is the true catalyst. The results are inconclusive. The fall in the nitromethane concentration is slower in oxidisingmethane than in the absence of oxygen, and kinetic-tracer experiments by Miller et aZ.71 show tihat the nitromethane is not only destroyed in the reaction, but formed from CH, and NO,. A striking feature is the sudden fall in the concentration of nitromethane a t a certain stage in the reaction, coinciding with a peak in the heat evolution. This sudden burst of activity seems to be due, to a partial ignition of the accumulated carbon monoxide. The ignition apparently produces a flood of radicals but is quickly inhibited and cannot be detected by any change in the CO concentration. Alkanes.-The present state of knowledge of the oxidation of alkanes has been reviewed in a non-specialist article by C ~ l l i s . ~ ~ The Zower uZEanes. The most significant recent work is that of Zeelenberg et aZ. who have followed the oxidation of isobutane 7 3 and neopentane 7 4 in a static system, from the early stages of the induction period, by gas chromatography. Gas chromatography is now regarded as the most satisfactory tool for studying complex oxidations. 7 5 Sandler and Beech,76 for example, determined some 30 products from the oxidation of n-pentane, and Wright 7 7 some 38 products from the oxidation of o-xylene. Zeelenberg's work on isobutane shows that all the initial products of the reaction increase exponentially with time. Isobutene is the major initial product (about 85% of isobutane oxidised) a t about 300" C. Other initial products areM. Barber and J. Cuthbert, Nature, 1961, 190, 1001. N. S. Enikolopyan and G. P. Konereva, Izvest. Alcad. Nauk S.X.S.R., Otdel. Ichim. Nauk, 1959, 1100; 1960, 419; 1961, 230 (transl. 210). 71 V. B. Miller, P. L. Levin, G . P. Konereva, M. B. Neiman, and N. S. Enikolopyan, Zhur. $2. Khim., 1960, 34, 1980 (transl. 940). 7 2 C. F. Cullis, Chem. and Ind., 1962, 23. 7 3 A . P. Zeelenberg and A. F. Bickel, J., 1961, 4014. 7 4 A. 9 . Zeelenberg, Rec. Trav. chim., 1962, 81, 720. 7 5 J. W. Falconer and J. H. Knox, Proc. Roy. Xoc., 1959, A , 250,493; R. E. Ferguson and C. R. Yokeley, 7th Internat. Symposium on Combustion, Butterworths Scientific Publns., London 1959, p. 113; G . Kyryacos, H. R. Menapace, and C. E. Boord, Analyt. Chem., 1959, 31, 222; C. F. Cullis, A. Fish, F. R. F. Hardy, and E. A. Warwicker, Chem. and Ind., 1961, 1158. 7 6 S. Sandler and J. A. Beech, Canad. J. Chem., 1960, 38, 1455. 7 7 F. J. Wright, J. Phys. Chem., 1962, 66, 2023.70


33

isobutene oxide (10%) and propionaldehyde (6%), with traces of acetone and propene. All can be derived from the decomposition of the two possible peroxy-radicals. The isobutene is probably formed by an HO, radical chain, which has been suggested for other hydrocarbon^.'^ With neopentane the HO, radical chain cannot operate, since there is no olefin with the neopentane carbon skeleton. The initial products at 260-290" c are isobutyraldehyde (85%), acetone (lo%), and traces of pivalaldehyde, dimethyloxetan, epoxyisobutane, and neopentyl alcohol. Isobutene is the major hydrocarbon produced and becomes the major product when the isobutyraldehyde is being consumed during the later stages of the reaction. The main path of oxidation appears to be(CH,),C*CH, X 0, (CH,),C.CH,

+ +

= =

=

(CH,),C*CH, XH (CH3),C*CH2.O, (CH,),CH*CHO CH,O.

+

+

Seakins 79 has reinvestigated the oxidation of propane, paying particular attention to the yields of peroxides and aldehydes. He has confirmed the existence of the negative temperature coefficient between 320" and 390" c and showed that coating the Pyrex reaction vessel with potassium chloride, while reducing the overall rate of re

Chemistry (Annual Reports - Vol.59-1962)

Documents

butterworths

optical rotatory

de la mare

nuclear magnetic

van der waals

lithium aluminium

complete rs

random mixing