CATALYSIS¹

Eric Rideal, Cambridge.

The extensive literature and frequent discussions on catalysis testify to the enormous scientific and technical significance of catalytic processes. The science of catalysis has expanded so much that it is no longer separate and distinct. Even chemical reactions and processes in living tissues may, in a certain sense, be called catalytic. It would be quite impossible, in the course of a short report, to give even a simple outline of the various ramifications into which ideas about catalytic processes extend; therefore it will be much better to dwell on the various views that have been expressed concerning the mechanism of the simplest reactions. True, even in this limited field there are many grounds for discussion, but I think that an introductory address at a broad meeting such as ours is precisely meant to give impetus to such discussion.

Contact Catalysis

If we turn our attention even to the simplest case, it will become apparent that we are only at the initial stage of a real understanding of molecular reactions. In considering the processes of hydrogenation and dehydrogenation

¹ Address delivered at the opening of the session on catalysis at the First Institute of Chemistry of the American Chemical Society in Pennsylvania. Printed in Chemical Reviews, 5, 67 (1928). Eric K. Rideal. Cambridge.

of such substances as ethylene or acetaldehyde and their derivatives, we see that, over wide ranges of pressures, the dehydrogenation process is a reaction of zero order¹), whereas in the case of hydrogenation, for partial pressures of the reacting substances there exists a definite optimum. For many dehydrogenation reactions the critical activation energy proves to be constant, provided only that the surface remains unchanged. I shall cite as an example the value \(\varepsilon = 22\,000\) cal/g-mole for the dehydrogenation of a series of alcohols on copper, or the values: \(\varepsilon = 9\,800\)—on nickel, \(\varepsilon = 15\,700\) on palladium, \(\varepsilon = 19\,000\) on platinized carbon, and \(\varepsilon = 18\,500\) for platinum on carbon for many dehydrogenation processes such as the decomposition of hexahydrobenzene, decahydronaphthalene, and piperidine. If we also take the decomposition of formic acid, then we find \(\varepsilon = 22\,000\) for platinum black, \(\varepsilon = 31\,000\) for silver, \(\varepsilon = 25\,000\) for gold, and the exceptional value \(\varepsilon = 39\,000\) for palladium. For dehydrogenation processes a larger number of metals is suitable, the limits apparently being determined only by the magnitude of the atomic radius, which must lie between 1.23 and 1.39 Ų). We may add to this that the hydrides of these metals are unstable and more similar to LiH than to HCl. Further, we know that the chemical changes occurring at the surface of a catalyst, at temperatures not too high, are in general confined to certain isolated areas, but that at high temperatures the entire surface, or the greater part of it, is active. These regions active at low temperatures are connected with an unstable state of the surface, large changes in energy, and a long lifetime of the adsorbed substances. All this

¹) A reaction of zero order is a reaction whose rate does not depend on the concentration of the reacting substance.

²) It is interesting to note the following order in the data of Remy and Schäfer (Remy und Schäfer), ZS. f. anorg. allgem. Chem. 136, 1924 [for the acceleration of the combination of hydrogen with oxygen in the case of metals: a) treated with hydrogen:
\(\mathrm{Os} > \mathrm{Pt} > \mathrm{Pd} > \mathrm{Rh}\) (increasing value of \(\sigma\)), and
b) treated with oxygen: \(\mathrm{Ir} > \mathrm{Pd} > \mathrm{Pt}\) (decreasing value of \(\sigma\))].

is confirmed by experimental facts, and any theoretical interpretation must encompass them.

It seems to me not without interest to give a brief outline of those views that have been put forward to explain catalytic reactions, and to ask to what extent one or another of these conceptions is in agreement with the data established by experiment. In Faraday’s original views there were two assumptions: 1) a polymolecular adsorbed film and 2) a weakening of the bonds which, in the normal state, hinder chemical action. The existence of polymolecular films—if one does not speak of substances having the structure of a gel, where there can scarcely be any question of the formation of a film—is now known to be observed only in very rare cases, and we may be certain that reactions, with the possible exception of combustion reactions and chain processes, take place on the very surface of the catalyst. The bond of the adsorbed substance with the adsorbent, owing to the fact that the molecules become more complex, must automatically lead to an increase in the number of collisions that bring about activation, as has recently been shown by Christiansen, Hinshelwood, Fowler, and myself. This bond, according to earlier conceptions of De la Rive and to the more modern views of Sabatier, is identical with the bond that is obtained in the formation of definite chemical compounds. This view was also expressed in a somewhat different form, namely, it was asserted that we are dealing with the formation of a large number of compounds, hydrides of various types, each of which is capable of hydrogenating various substances, for example the benzene nucleus, substances with double bonds, or nitrobenzene. Some of the proposed formulas, based on determining the average composition, are rather unusual. I do not think that such an interpretation of activation as a consequence of collisions of the second kind between complex molecules formed by the combination of the adsorbent substance with molecules of the adsorbed substance, or of some other of the reacting substances, could

be regarded as corresponding to reality; for it does not by itself provide an explanation either for the existence of active and inactive parts of surfaces, or for the change in the critical activation energy in the transition from one substance to another.

The simplest acceptable hypothesis, if one does not consider the definition of the type of compound of the adsorbent with the adsorbed substance, will be the admission of the existence of adsorption of the molecule and of the fact that, on the corresponding part of the catalyst, the molecule is stretched and, as a result of such stretching, undergoes activation. We shall return to this view later. If such stretching reaches a limit, then we are dealing with the dissociation of the molecule into radicals or atoms. Some investigators adhere to the extreme view, namely, that activation occurs only when this limit has been reached. However, little evidence can be adduced in favor of such a hypothesis. Despite numerous attempts, atomic hydrogen has not been detected in hydrogenous palladium, bearing in mind the ease with which the cold metal adsorbs hydrogen atoms, and the favorable conditions for their formation at surfaces at high temperature; I think that the detection of hydrogen atoms at the electrodes of thermionic tubes cannot serve as definitive proof of the presence of dissociation by a cold metallic plate. It is true that hydrogen atoms may be detected in many surface processes, but it is difficult to prove that they are not secondary products of a chemical action that has taken place between excited molecules.

Adsorption under the action of electric forces.

Many valuable assumptions have been put forward according to which the adsorption compounds that are formed possess a definite electrical structure. It proves not difficult to calculate the heat of adsorption for a gas when it consists of dipolar or linear quadrupolar molecules. This can be done by assuming that adsorption takes place

thanks to attraction to the surface caused by the image of the molecular electric moment in a medium with a definite dielectric constant. However, we must note that this representation does not give good values for the ratios between the heats of adsorption on substances with different dielectric constants¹) and that the absolute values calculated in this way are small and constant, thereby giving no explanation for the observed changes in the heat of adsorption as a function of the form of the surface.

We further find indications in the literature that adsorption is a surface phenomenon which occurs owing to the existence of positive and negative ions already present on the surface of the adsorbent. For simplicity one may suppose that some atom on the surface of a metal loses an electron and in

¹) For example, the heats of adsorption of argon, nitrogen, and methane on mica and homogeneous charcoal, in ergs per molecule, will be

If the heats of adsorption are wholly due to a loss of potential energy associated with the transition of the dipole into the position of equilibrium with respect to its mirror image, then the heats of adsorption on charcoal and on mica must be in the following ratio:

$$ \frac{Q_m}{Q_c}= \frac{\dfrac{K_m-1}{K_m+1}}{\dfrac{K_c-1}{K_c+1}}, $$

where the subscript $m$ refers to mica, and $c$ to charcoal. If $K_c$ is taken sufficiently large and for $K_m$ the usual value 6 is taken, then for the ratio we obtain

$$ \frac{Q_m}{Q_c}=\frac{5}{7}, $$

which contradicts experiment.

upon adsorption of a hydrogen molecule, a molecular hydrogen ion is formed. This idea may be expressed in the following form:

\[ \mathrm{M}^{+}+\mathrm{H}_{2}\longrightarrow \mathrm{MH}_{2}^{+};\quad \mathrm{M}' + \mathrm{C}_{2}\mathrm{H}_{4}\longrightarrow \mathrm{MC}_{2}'\mathrm{H}_{4}. \]

Many arguments may be advanced in favor of the view that ions are formed upon adsorption on active parts, or, at high temperatures, also on inactive ones. We can point to the enormous chemical activity of charged ions in the gaseous phase. Thus \(\mathrm{N}^{+}\) and \(\mathrm{N}_{2}^{+}\) readily react with hydrogen, forming ammonia. \(\mathrm{CO}^{+}\) decomposes into \(\mathrm{CO}_{2}\) and carbon. Many such reactions have been investigated by means of collisions with electrons and \(\alpha\)-particles (Lind). Further, in many reactions on the surface, especially in oxidation reactions of incandescent filaments, we can observe an abundant emission of positive and negative ions, which may be thought of as products that have broken away from the adsorption layer without entering into combination, of course—on the assumption that the ionic form is the normal state in the layer. Finally, it has been established that it is possible to construct an ethylene–hydrogen element with palladium electrodes, and that by this method, with permissible accuracy, one can measure the free energy of hydrogenation; moreover, the reactions may be written as follows:

\[ \begin{matrix} \mathrm{H}_{2} \rightleftarrows 2\mathrm{H} \rightleftarrows 2\mathrm{H}^{+} \\ \mathrm{C}_{2}\mathrm{H}_{4} \rightleftarrows \mathrm{C}_{2}\mathrm{H}_{4} \end{matrix} \left\} \,\mathrm{C}_{2}\mathrm{H}_{6}. \right. \]

The close parallelism between thermionic emission and a chemical reaction on a surface was first noted by Thomson and was often later commented upon by Langmuir. There are also experiments indicating the heterogeneous character of thermionic emission by surfaces, as well as the reduced magnitude of the work of detachment associated with amorphous surfaces or with the presence of a layer of amorphous material on a crystalline surface. Schottky’s electron emission is also represented as being limited to spots on the surface; these

regions with a small value of the work of detachment may be regarded as identical with active regions at low temperature in the case of catalysis. Despite these arguments, I do not think that such an ionic representation can be maintained in this simple form. This can be shown by studying reactions between incandescent filaments and oxygen or halides with the formation of volatile compounds. Langmuir found that the rate of oxidation of tungsten follows a simple monomolecular law. He admits the possibility of the existence of a large number of free electrons in the metal, some of which combine with oxygen molecules at the surface. If we modify this view so that we do not admit the existence of a multitude of free electrons in the metal, then we shall see that the critical activation energy must come close to the difference between the work of detaching an electron from the metal and the electron affinity of the reacting gas.

Langmuir showed that the critical activation energy for the surface oxidation of tungsten is equal to 0.6 volts; in our laboratory the value for the analogous process on platinum was found to be 2.3 volts.

The work of detaching electrons from these two metals is equal to:

If the critical activation energy \(\varepsilon\) is the difference between the work of detaching an electron from the metal and the affinity to the electron of molecular oxygen \(\varepsilon_r\), then we have: \(\varepsilon=\varphi-\varepsilon_r\), whence

\[ \begin{aligned} \text{for platinum}\ .\ .\ .\ .\ .\quad &2.3=5.5-\varepsilon_r,\quad \text{i.e. } \varepsilon_r=3.2\ \text{volts},\\ \text{for tungsten}\ .\ .\ .\ .\quad &0.6=4.25-\varepsilon_r,\quad \text{i.e. } \varepsilon_r=3.65\ " \end{aligned} \]

For oxygen there is a series of proofs in favor of the existence of \(O_2^{-}\), and it appears quite probable that, according to Mackay’s data, the normal molecule has an electron affinity of about 3.5 volts; this value is in good agreement with that given above.

It is obvious that not all the reacting gas is present on the surface in the form of ions, but that molecules are converted into ions only after a certain quantity of energy has been imparted to them, measured by the critical activation energy. Further, although Thomson showed long ago that the reaction between hydrogen and oxygen on a platinum filament becomes intense at temperatures at which thermionic emission in a vacuum becomes appreciable, rapid combination can occur on the surface of many metals even at very low temperatures. Finally, the rate of combination of hydrogen with oxygen considerably exceeds the rate of formation of the metal oxide, which indicates that oxygen does not need to pass through the oxide form in order to enter into reaction with hydrogen.

We shall dwell, finally, on one more view, which deserves some attention and which also admits the existence or appearance of induced positive and negative charges on the surface of the adsorbed substance—more precisely, the polarization of molecules during adsorption and the entry of adsorbed molecules into the interior of the double layer and onto the surface of the adsorbent. These polarized molecules are taken to be chemically active.

It seems very probable that the ease with which a gas is adsorbed depends on the magnitude of its electric moment, but that the mechanism of attraction on the metal is not

As regards the halogens, as is seen from Langmuir’s work, they first thermally dissociate, after which the atoms react with the metal. The total energies required for dissociation are: \(\mathrm{Cl}_2\)—5.2 volts at \(1000^\circ\mathrm{C}\), \(\mathrm{Br}_2\)—2.47 volts; \(\mathrm{I}_2\)—1.51 V. The electron affinities are: 4.1, 3.8, and 3.5. This gives the values \(E\) on platinum for the reaction \(\mathrm{X}_2 \to \mathrm{X} \to \mathrm{PtX}\): \(\mathrm{Cl}_2\)—6.6, \(\mathrm{Br}_2\)—3.72, and \(\mathrm{I}_2\)—3.51, and on tungsten: \(\mathrm{Cl}_2\)—5.35 and \(\mathrm{I}_2\)—2.26 volts. The values of \(\varepsilon\) for the halogens are much smaller, namely:

so that the reaction proceeds even at low temperatures.

It is interesting in this connection to note that Davy (Phil. Trans. 1807) showed that negatively charged zinc cannot be oxidized and that silver becomes oxidizable when it receives a positive charge.

may be explained by a simple conception of the mirror image. It might seem that on catalytic surfaces the adsorbed substance must enter into the field present at the surface,¹ and we must have the presence of natural or induced dipoles which are characteristic of the metal or other catalyst and which depend on the structure as well as on the nature of the material. Some arguments in favor of this view may be obtained from other sources, for example—from the properties, investigated by Wood, Langmuir, and Frenkel, of molecular beams of metals falling on a cold surface. There is sufficiently convincing evidence that twins or small aggregates of atoms on a plane surface not only evaporate less readily than free atoms, but also produce a noticeable local field. If Frenkel’s conception of a solid metal as a system in which the valence electrons are never free and are in constant motion from atom to atom is correct, then we obtain a simple mechanism for the formation of dipoles possessing both electrostatic and electromagnetic moments in the form of twin atoms situated on the surface.

Thus the surface of a catalyst, where adsorption takes place, according to this view must consist of a fluctuating system of electric moments of various magnitudes, which depends on the closeness of the grouping of the atoms and the degree of their removal from the mean surface of the catalyst, and, finally, on the presence of true atoms—promoters.²

¹ If the free surface energy of a substance, \(\sigma\), is a manifestation of electrostatic energy, the energy of surface dipoles, then it can be shown that

\[ \sigma=\frac{9V^{2}}{80\pi r}, \]

where \(V\) is the potential for photoelectric emission. For mercury \(V=4.2\) volts, \(r=1.4\ \text{Å}\), so that \(\sigma=472\ \text{dyn}/\text{cm}\) (observed \(\sigma=465\)).

² Promoters are substances whose presence intensifies the action of a catalyst.

Deformability of Molecules

In adsorption, deformation of the adsorbed molecule takes place. This may also be an actual stretching, but a deformation that leads to an increase in the electric moment \(P\) need not lead to an increase in the magnitude of the mechanical moment \(I\), since there is no necessity to suppose that, when a dipole molecule with small values of \(P\) is excited, any change occurs in the spatial arrangement of the centers of the atoms. From the examples given below it is evident that in some cases \(I\) decreases as \(P\) increases.

Here \(S\) denotes the number of symmetries in the expression for the chemical constant, \(m\) the mass of the molecule, \(I\) its mean moment of inertia, and \(R\) the Boltzmann gas constant1.

Further, in the case of sodium chloride, Reis observed that, if one assumes that the equilibrium condition of the salt molecule both in the gaseous phase and in the lattice can be expressed as

\[ \frac{d}{dr}\left(\frac{e^2}{r} - \frac{B}{r^9}\right)=0, \]

then the magnitude \(r\) in the case of gaseous NaCl will be smaller than in the case of the lattice, where the deformation is greater.

The conditions of activation during adsorption thus require the existence of a dipole with a sufficiently large electrostatic or electromagnetic moment on the surface,

presence of an electrostatic moment in the adsorbed substance1 and the ability of its molecule to undergo deformation. We may indicate that the deformability of a molecule was defined by Born by the expression \(a = \dfrac{P}{E}\), where \(P\) is the electric moment of the dipole induced in an electric field of intensity \(E\). The magnitude \(a\) for the reacting substance must be a factor on which the ease of activation depends. The classical method of calculating it consists in determining the refractive power

\[ a = \frac{3}{4}\,\frac{n^{2}-1}{n^{2}+2}\cdot\frac{M}{\pi N d}, \]

and the values obtained in this way can in many cases be compared with values determined from the conditions of equilibrium of crystals, from electrostriction, or from the temperature coefficients of the dielectric constant.

For example, we find the following values of \(a\) for two oxides of carbon:

Obviously, carbon dioxide has a large electric moment and is, consequently, more readily adsorbed, but its deformability and, consequently, the ease of its excitation will in general be somewhat less than for carbon monoxide.

Various mathematical problems connected with the question of the dielectric constant and temperature, similar to the expression first formulated by Langevin and Debye for the relation between magnetic susceptibility and temperature, can apparently be more easily developed with the aid of the methods of de Broglie–Schrödinger wave mechanics.

The final rupture of the dipole naturally leads to the formation of ions, not atoms, so that the ions observed in surface reactions, especially at high temperatures, may be not only by-products of the chemical reaction, as Haber assumes, and not only, as Brewer believes, those ions which in small numbers have succeeded in escaping from the circular process by virtue of which the action of surfaces takes place. They may also be formed from certain, not very numerous molecules which fly off, owing to thermal motion, after adsorption on the corresponding parts of the surface, where the energy gained during adsorption exceeds the lower limit of the energy at which the surface process of dissociation can already occur. It is possible that, in connection with this, the character of the surface action changes when the temperature is sufficiently elevated for these ions to fly off, and we obtain a transition from a surface reaction to a chain reaction—an effect that is easily observed, for example, in the case of the hydrogenation of oxygen and, still better, of bromine on a platinum surface1.

In connection with what has been said here about electric moments, I would like to point out one further matter. Considering

changes of the potential jump at the water–air boundary when a monomolecular film of the given substance is formed at this boundary, Guyot and then Frumkin, who used Guyot’s method, as well as Smythe, from determination of the refractive index, came to the conclusion that the electrical moments must be assigned to a certain number of so-called polar groups—OH, COOH, etc. It now appears that the interesting phenomenon of residual charge, as well as the loss of energy in dielectrics under the influence of an alternating field, are connected with substances possessing electrical moments or capable of acquiring them when small displacements take place. This energy is not released at once, but only gradually.

Observations indicate that the process of adsorption proceeds at first isothermally, and that the adsorbed molecule still possesses for some time the potential energy lost upon falling onto the surface. Such adsorbed molecules are thus potentially active until they lose their potential energy with the liberation of heat, i.e., the heat of adsorption. When the surface is only sparsely covered, this liberation of heat occurs only slowly. Hence one may conclude that the paths of transformation of potential energy into kinetic energy proceed either through collision with adsorbed neighboring molecules, as Garner imagines it, or through isochronous induction in neighboring molecules, as is indicated by Perrin’s work on fluorescence. It may be noted, however, that in the adsorption of gases by substances such as coal, the molecules undergo stretching, so that it would be necessary to investigate the accompanying changes in energy before drawing any conclusions from measurements of the heat liberated.

Reactions in Solutions

One of the most interesting features of reactions in homogeneous systems was the regularity with which reactions

are represented as conforming to the ideas of Bjerrum and Brønsted (Bjerrum, Brönsted) concerning the mechanism of such actions. You know that, with the collapse of the simple formulation of the reaction rate in solution as a function of the concentration of the reacting substances, i.e.

\[ \frac{dx}{dt}=K(A)(B)^1), \]

Mc. C. Lewis and Scatchard (Mc. C. Lewis, Scatchard) introduced the Lewis (G. N. Lewis) concept of activity in the following form:

\[ \frac{dx}{dt}=K f_A(A) f_B(B) \]

where \(f_A\) and \(f_B\) are the activity coefficients of the reacting substances. A more thorough investigation showed that, whereas this formulation is correct for certain reactions, for example for the decomposition of hydrogen peroxide with the aid of hydrochloric acid, it is, generally speaking, not correct. Brønsted and Bjerrum assumed that the reaction mechanism includes the formation of “associated complexes” or “quasi-compounds.” Such a complex either may break down into the original reacting substances, with which it is in equilibrium according to the law of mass action, or else may react further at a rate proportional to its concentration. We can thus formulate our reaction mechanism in the form

\[ \frac{dx}{dt}=K(AB). \]

Since, according to the law of mass action, the equilibrium constant is expressed by the formula

\[ \frac{f_A(A)f_B(B)}{f_{AB}(AB)}=K, \]

\(^1\) \((A)\) and \((B)\) denote the concentrations of substances \(A\) and \(B\).

then we have

\[ \frac{dx}{dt}=K\frac{f_A(A)f_B(B)}{Kf_{AB}}, \]

an expression different from that which was proposed by Mac-Lewis (W. C. Mc. C. Lewis), but closely approaching it if the activity coefficient of the complex \(f_{AB}\) is the activity coefficient of the uncharged complex, for, as Brønsted experimentally showed and as the works of Milner, Debye, and Hückel indicate, the magnitudes of the activity coefficient depend mainly on the charge of each of the kinds of particles.

It may be shown that the recent investigations of Dowson correspond to these ideas, but it must be added that, according to these investigations, in reactions catalyzed by oxonium and hydroxyl ions, specific catalytic activities may be ascribed to all “donors” and “acceptors” of oxonium ions if the nomenclature of Wieland is adopted.1 Thus, in the case of catalysis by a dilute aqueous solution of acetic acid, the specific catalytic activities may be ascribed to the “donors” of oxonium ions: \(H_2O\), \(CH_3COOH\), \(H_3O^+\), and to the “acceptors” of oxonium ions: \(OH'\), \(CH_3COO'\).

I do not think it necessary to suppose that, in the process of hydrolysis of an ester by a dilute acid, the oxonium ion is “given off” at the same moment at which another ion, like it, “is added”; but in such a reaction there are interesting features, as is seen from the following series of transformations depicting the process of hydrolysis of methyl acetate into acid and alcohol under the action of hydrogen ions.

$$ \begin{gathered} \text{methyl acetate} \\[2mm] \ce{CH3-C(=O)-OCH3 <=>[H2O] CH3-C(OH)(OCH3)-OH2+} \\[3mm] \ce{<=> CH3-C(OH)(OCH3)-OH2+ -> CH3-C(OH)=O + H3O+ + CH3OH} \end{gathered} $$

(acetic acid)

(methyl alcohol).

It must be noted that one of the stages of the reaction consists in the addition of one hydrogen ion (1) and the splitting off of another (2), which occurs owing to the transfer of one electron from (2) to (1). We must remember that complex (A) is nothing other than a critical complex or quasi-compound and can, consequently, also decompose back into the ester and the oxonium ion. While in this way we may regard the action of hydrolysis as a reaction caused by the displacement of an electron, or even as a kind of electrolysis—which, however, may perhaps already be a somewhat too sweeping interpretation of Prof. Armstrong’s views—we still do not know what the conditions are for the forward or reverse process, and whether hydrolysis occurs at the moment of some crisis in the molecular motions, and whether the communication of activation energy affects not the concentration of the quasi-compound but its stability. It is clear, however, that if any one of these views is correct, then the quasi-compound (A) must have an objective existence, even if it is present in quantities not detectable by analysis, and the two stages of the process—the addition of the oxonium ion to (1) and its splitting off from (2)—cannot be simultaneous.

Molecular induction.

In the initial stage of his “radiation” theory of chemical processes, Perrin believed that excited molecules must lose their excitation energy not only

in collisions of the second kind, as is assumed by Klein and Rosseland, but also in “resonance excitation.” Recently attempts have been made, by various methods, to calculate the energy obtained when a molecule enters either the field of another molecule with a definite electric moment, or the field of an ion. These investigations were carried out chiefly by J. J. Thomson, Merritt, and Taylor, with the aim of clarifying what influence ions and polar molecules may have on the rate of chemical action. J. J. Thomson showed that the collision rate between reacting molecules may increase considerably when one of these substances is replaced by an ion, for example \(A^+\), or by a compound of the type \(bA\), where \(bA\) is a primary complex between a polar molecule \(b\) and the reacting substance \(A\). Not all such collisions, as we know, actually cause chemical interaction; therefore it is very important to pose the question: can the critical activation energy of these reactions, as well as the collision frequency, be changed when the field produced by a molecule with a definite electric moment or by an ion is changed? It is assumed here that, when the dipole molecule \(B\) enters the electric field of a polar molecule \(b\) or of an ion \(A^+\), it acquires potential energy determined by the orientation of its axis with respect to the lines of force; moreover, the moment of inertia \(I\), as well as the electric moment of molecule \(B\), changes, and a certain amount of energy \(E\) may either be released or absorbed in this process. Merritt assumes that this energy supplies part of the activation energy. The following simple calculation shows the relative magnitudes of the changes in energy obtained in these two cases.

Assuming that the molecule enters the field at a distance of about 4 molecular radii, we may set the field strength equal to \(\dfrac{\mu}{(4\sigma)^3}\), so that, in the case of water, where \(\mu = 1.10^{-18}\) electrostatic units and \(\sigma = 2.3\ \mathring{\mathrm A}\), the field strength will be \(1.3 \cdot 10^3\) electrostatic units, or \(3.85 \cdot 10^5\ \mathrm{V/cm}\). The maximum potential-

of the molecule, having a moment \(\mu\), in this field will be \(\mu \times 1.3 \cdot 10^3\) electrostatic units. For a molecule similar to hydrogen chloride, where \(\mu_1 = 3 \cdot 10^{-18}\), we obtain \(= 3.84 \cdot 10^{-15}\) erg, or only about \(55\ \mathrm{cal}/\mathrm{g\text{-}mol}\). The insignificance of this quantity clearly refutes Cathala’s view of the catalytic influence of water in the reaction of combination of hydrogen chloride; but at the same time it is evident that intense fields may be obtained if the water molecule is more closely connected with the reacting substance than at a distance of \(4\sigma\), as assumed in this calculation. At a distance of one radius \(\varepsilon\) will already be increased to \(3000\ \mathrm{cal}/\mathrm{g\text{-}mol}\).

If the polar molecule of water is replaced by an ion, we shall have around the ion an accumulation of molecules, as was assumed by Langevin, J. J. Thomson, Erikson, and Loeb, and the mean potential energy of a molecule in the first layer at a distance of \(2.3\ \text{Å}\) will be about \(12\,000\ \mathrm{cal}/\mathrm{g\text{-}mol}\).

It follows from this that the possible gain in the potential energy of a molecule entering the field of an ion is very considerable and, consequently, in the case of a reaction in which ions play a role, appreciable catalytic effects may be expected.

Oxidation Processes.

At the present time much attention is being paid to the mechanism of slow and rapid combustion. In view of the technical importance of explosive reactions, on the one hand, and of processes of autoxidation and slow combustion, on the other, the interest in this field, which has again intensified in recent years, should be welcomed.

It appears highly probable that many oxidation processes in solution take place on surfaces; we may mention the oxidation of oxalic and other acids on charcoal, and of benzaldehyde and turpentine on glass, pumice, and similar surfaces. By means of the method of selective poisoning it can be shown that only a small part of the total surface—for example, in the case of powdered glass—is

catalytically active. Many substances exhibit what Moureu calls an antioxidizing action owing to selective adsorption on these surfaces; this applies to iodine, diphenylamine, and even to organic acids. These oxidation reactions include, in many cases, the formation of a definite and relatively stable peroxide, which in the case of aldehyde peroxide reacts further in the homogeneous liquid phase, passing into an acid. It seems to me almost certain that, before the formation of the stable peroxide, there must exist another, more active form of peroxide, which can serve as the starting point for the formation of chains in the sense assumed by Christiansen and Kramers1. In any case, it is very probable that typical reactions of this kind occur on surfaces: thus, the combustion of an aldehyde molecule on the surface of glass is conditioned by the disappearance of many molecules on the surface as a result of the propagation of the combustion process. On carbons one can establish at least five different types of surfaces. There exists a certain small surface which is auto-oxidizing. Ward recently showed that these auto-oxidizing parts of the surface are identical with those surfaces on which Blench and Garner observed particularly large values for the heats of adsorption of oxygen. Further, there exist catalytically active parts of the surface on which, as Warburg and Miss Wright have shown, various oxidative processes can proceed. It is highly interesting that the rates of oxidation on this surface for various substances, at their corresponding optimum concentrations, run parallel with their ability to lower the overvoltage of oxygen on a platinum anode.

There exist, finally, two more “kinds of sites” on surfaces containing promoters, of which some are connected with complexes containing carbon and iron, while others—extremely active—are connected with complexes containing iron, carbon, and nitrogen. These curious reactions compel one to pay attention to the possible connection with Prof. Armstrong’s ideas about reactions as a special kind of electrolysis. They may be represented visually by the following equations, illustrating the process of oxidation of oxalic acid to carbonic acid on the surface of platinum and carbon:

\[ \mathrm{Pt} \begin{array}{c} \diagup \mathrm{OH}\!:\!\mathrm{H\,OH}\!:\!\mathrm{HOOC}\\[2pt] \diagdown \mathrm{OH}\!:\!\mathrm{H\,OH}\!:\!\mathrm{HOOC} \end{array} \;\longrightarrow\; \mathrm{Pt} \begin{array}{c} :\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2}\\[2pt] :\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2} \end{array} \]

\[ \begin{array}{rcl} \text{(2a)}\quad \begin{array}{c} \mathrm{C}\!\begin{array}{l}\diagup \mathrm{OH}\\[-2pt]\diagdown \mathrm{H}\end{array}\\[-2pt] \begin{array}{c}\mathrm{OH}\end{array}\\[-2pt] \mathrm{C}\!\begin{array}{l}\diagup \mathrm{H}\\[-2pt]\diagdown\end{array} \end{array} +\mathrm{O_2} &\longrightarrow& \begin{array}{c} \mathrm{C}\!\begin{array}{l}\diagup \mathrm{OH}\\[-2pt]\diagdown \mathrm{OH}\end{array}\\[6pt] \mathrm{C}\!\begin{array}{l}\diagup \mathrm{OH}\\[-2pt]\diagdown \mathrm{OH}\end{array} \end{array} \quad \text{(2b)} \end{array} \]

\[ \begin{array}{c} \mathrm{C}\!\begin{array}{l} \diagup \mathrm{OH\,H\,OH\,HOOC}\\[-2pt] \diagdown \mathrm{OH\,H\,OH\,HOOC} \end{array}\\[6pt] \mathrm{C}\!\begin{array}{l} \diagup \mathrm{OH\,H\,OH\,HOOC}\\[-2pt] \diagdown \mathrm{OH\,H\,OH\,HOOC} \end{array} \end{array} \;\longrightarrow\; \begin{array}{c} \mathrm{C}\!:\!\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2}\\ \phantom{\mathrm{C}}\!:\!\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2}\\ \mathrm{C}\!:\!\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2}\\ \phantom{\mathrm{C}}\!:\!\mathrm{OH_2}\!:\!\mathrm{OH_2}\!:\!\mathrm{CO_2} \end{array} \]

However, it must not be forgotten that the existence of a “perhydrol” of carbon,

\[ \mathrm{C}\begin{array}{l} \diagup \mathrm{OH}\\[-2pt] \diagdown \mathrm{OH} \end{array}, \]

similar to that which is assumed for platinum, is not sufficiently experimentally substantiated.

In explosive oxidations the situation is, of course, much more complicated. Further work, I think, is necessary in order to prove the existence of chains formed owing to collisions of the second kind between excited molecules of the reaction products and the reacting substances, as postulated by Christiansen and Kramers. There is no doubt, as was especially emphasized by Egerton, that peroxides play a significant role in explosive reactions; the same view—that they are formed after the primary dissociation of saturated hydrocarbons into unsatu-

enriched in hydrogen, does not rest on such a firm foundation, and it is also probable that peroxide formation may occur in the reaction between excited molecules of saturated hydrocarbons and oxygen. Egerton believes that the conditions determining the formation of peroxides of oxidizing substances also determine the conditions of detonation. This view is not in agreement with that derived from consideration of the emission spectra in explosions of gaseous mixtures. Whereas it had long been suggested that a zone of electrons or radiation spreads ahead of the explosive wave and acts as an exciting agent for the remaining gas, Garner now concludes that the ionization produced during a gas reaction is no greater than the normal thermal ionization which must occur at the high temperatures present during an explosion. The situation appears to be otherwise with respect to the quantity and nature of the infrared radiation emitted. In dry gas mixtures—for example, in the case of carbon monoxide—the emission of such radiation is much greater than in moist mixtures or in mixtures in which water is formed during combustion. The fine structure of the band spectrum indicates that excited water molecules, containing energy of both rotational and vibrational motion, are the sources of the radiation. Garner regards water as an “energothermal” catalyst: it absorbs radiation which otherwise would have been emitted, and thereby facilitates the distribution of energy in the system. We may represent the sequence of reactions taking place as follows:

\[ \begin{aligned} \text{First stage}\quad & C_nH_{2n} + x \longrightarrow CO + H_2 + \text{other products}.\\ \text{Second stage}\quad & CO + O_2 \longrightarrow CO_2\\ & H_2 + O_2 \longrightarrow H_2O. \end{aligned} \]

It is assumed that antiknock agents1 lower the rate of formation of water, and that, owing to this, the dissipation of energy in the form of radiation is increased.

1. Substances added to the fuel in internal-combustion engines in order to raise the flash temperature. ↩↩↩↩↩↩