PROPERTIES OF FREE HYDROGEN ATOMS1

K. F. Bonhoeffer, Berlin.

I. Methods of Formation of Free Hydrogen Atoms

Under ordinary conditions hydrogen is a gas consisting of diatomic molecules. Free hydrogen atoms have become accessible to experiment only recently. Certain facts from the field of chemistry had already earlier indicated the possibility of the independent existence and special properties of hydrogen “at the moment of liberation” (in statu nascendi).

a) Hydrogen “at the moment of liberation.” If hydrogen is formed in the course of some chemical reaction, it possesses increased chemical activity. This was associated with the formation of hydrogen atoms. For the most part, this concerned hydrogen formed during the dissolution of metals in acids or alkalis, or liberated at the cathode during the electrolysis of aqueous solutions2. The hydrogen atoms formed in this way can be called free only with great reservation, since they are adsorbed by the metal on which they are liberated; their activity may thus depend also on the adsorbing substance.

Refinement of our ideas about the course of chemical reactions has made it possible to assume the intermediate appearance of free hydrogen atoms also in those reactions where it cannot be directly proved, for example in the formation of hydrogen bromide from bromine and hydrogen in the gas phase1, in the combination of chlorine with hydrogen under the action of light2, or in an induced reaction3. (There are still some doubts concerning the mechanism of this reaction.) Recently indications have been obtained of the formation of atomic hydrogen in the reaction between gaseous \(H_2SO_4\) and solid \(Na\)4. However, for studying the properties of free hydrogen atoms these systems are for the most part too complex and unsuitable.

b) Free atoms in thermal equilibrium. According to Nernst’s theorem, one can calculate, with sufficient approximation, the equilibrium between free atoms and molecules for a given temperature and pressure. The heat of formation of a hydrogen molecule from atoms is known. According to spectroscopic data it is equal to \(4.34\,V = 100.1\) gram-calories at absolute zero. The variation of the specific heats is also known sufficiently well. In the equation

\[ \ln K_p = -\frac{U_0}{RT} - \frac{1}{R} \int_0^T \frac{C_p(H_2)}{T}\,dT + \frac{2}{R} \int_0^T \frac{C_p(H)}{T}\,dT' - i_{H_2} + 2i_H, \]

where \(K_p\) is the dissociation constant (measured in partial pressures), \(U_0\) is the heat effect at absolute zero, \(C_p\) is the heat capacity at constant pressure, \(i\) is the chemical constant, \(R\) is the gas constant, \(T\) is the absolute temperature, a small numerical uncertainty is introduced only by the chemical constant \(i\). This is explained by the fact that the statistical weight of the molecule and the atom entering into it still

PROPERTIES OF FREE HYDROGEN ATOMS

has not been established. The experiment of Stern–Wrede (Stern, Wrede)1 permits one to regard the statistical weight of the atom as equal to 2; nevertheless, in the calculation there remains an uncertainty due to the residual moment of the protons, both for the atom and for the molecule. For the latter, experimental investigations of the elasticity of the vapor2 and of the gaseous equilibrium3 have not yet yielded final results.

The experimental approach to the question of the thermal formation of free atoms is carried out by two different methods, which give results consistent with one another and with the requirements of Nernst’s theorem.

The temperatures at which dissociation becomes noticeable are obtained in explosions of gas mixtures. In the presence of gas molecules, dissociation manifests itself in an anomalous increase of the specific heats and pressure with temperature. A closer investigation shows that the thermodynamic equilibrium of dissociation has time to become established, despite the short duration of the reaction. On the basis of his experiments Wohl (Wohl)4 gives the equilibrium constant in a form determined by the choice of the statistical weights of H and H₂, which he takes to be equal to unity. If for H one puts it equal to 2, as indicated above, then the Witmer heat of dissociation is obtained directly, within the errors of experiment.5

When the pressure is lowered, dissociation occurs at lower temperatures. Langmuir (Langmuir) (see the bibliography, Nos. 9–21), in a series of classic works, studied dissociation on incandescent tungsten filaments. On the basis of his results he represents the dissociation constant in the following form (in accordance with the form given by Nernst’s theorem):

\[ \log_{10} K_p = - \frac{21200}{T} + 1.765 \log_{10} T - 9.85 \cdot 10^{-5} T - 0.256 . \]

The following table gives some values calculated in this way (see the bibliography, No. 18).

Table 1.

Langmuir’s method made it possible to elucidate many properties of free atoms. In what follows we shall often return to these works.

If hydrogen is dissolved in another medium, then the equilibrium constant depends on the nature of this medium. In many metals, for example Pt, Cu, complete dissociation is already attained at low temperatures.\(^1\)

Under these conditions one must also allow for an increased dissociation of hydrogen adsorbed on the surface. The increased reactivity of adsorbed hydrogen in catalytic reactions can then be explained by its atomic state. This conception

\(^1\) Sieverts und Jurisch, Ber. d. deutsch. chem. Ges. 45 (1912) 221. ZS. f. phys. Chem. 77 (1911) 591. See also: Borelius, Ann. d. Phys. 83 (1927) 121; references therein.

was substantiated by Polanyi1. We shall not, however, use them further, since these atoms cannot be regarded as free, owing to the adsorptive forces binding them.

c) Free atoms in gas discharges. In discharges, hydrogen can be decomposed into atoms: this is evident from the emission of the Balmer spectrum. However, by this route it is impossible to find out how many atoms have thereby been formed. Wood (see bibliography Nos. 35—38) gave a very effective method for obtaining hydrogen atoms by a discharge at low pressure. Below we shall return to this method again. From the amount of heat liberated upon the recombination of atoms, as well as from chemical reactions, one may conclude that concentrations considerably exceeding 20% are readily obtained (see bibliography No. 2). Experiments carried out by Wrede (bibliography No. 39) and by Taylor and Fipson (No. 29) on an atomic beam also show that the concentration of atoms here is rather high.

The mechanism of decomposition of molecules may be different. In all probability, the predominant process is one in which the molecule is split by an electronic impact into an excited and an unexcited atom; the excess energy may be converted into the energy of motion of the atoms flying apart and may be detected by the resulting Doppler effect2.

Hydrogen atoms also appear in a voltaic arc and, as Langmuir showed (Nos. 18, 22), at high concentration (see p. 71).

d) Photochemical production of free atoms. There are several photochemical methods for obtaining hydrogen atoms. Thus, for example, they appear when molecular hydrogen is illuminated by light whose wavelength

which is less than 850 Å. This assertion is based on the form of the absorption spectrum, which, beginning at 849.4 Å, is continuous and corresponds to the dissociation of the molecule into an excited (two-quantum) and an unexcited atom1.

Another route is a sensitized photochemical reaction. A mixture of Hg and H₂ is illuminated with the mercury line 2536.7 Å (reference no. 6). The excited Hg collides with H₂, and some portion of these collisions leads to dissociation, either directly or indirectly through the formation of an HgH molecule and an H atom2. The atoms thereby obtained can be detected chemically. Instead of mercury in the \(2^{3}P_{1}\) state, mercury in the \(2^{3}P_{0}\) state may act (see nos. 26, 27). Finally, the formation of hydrogen atoms can be proved in the photochemical decomposition of hydrogen iodide3. This was predicted by Warburg’s theory.

II. Methods for Obtaining Free Hydrogen Atoms.

Not all of the indicated methods of forming atomic hydrogen can be used for obtaining it in more or less appreciable quantities. Here we shall describe the methods that have been employed up to the present.

a) Obtaining atoms by thermal dissociation.

The properties of free atoms were first studied by Langmuir (reference no. 14), who developed this method. The principle of his apparatus is very simple. It consists of a pump producing a high vacuum, a McLeod manometer, and a glass bulb containing a tungsten wire heated by an electric current (a lamp with a tungsten filament). Between the manometer and the bulb there is a \(U\)-shaped tube cooled by liquid air,

for the removal of mercury vapor. The bulb (with the wire) is first heated and evacuated to a very low pressure. If hydrogen is then introduced into the apparatus, then when the tungsten wire is heated to incandescence it dissociates into atoms to a greater or lesser degree, depending on the pressure.

This dissociation manifests itself above all in the abnormally high thermal conductivity (Nos. 9, 10, 13) of the gas. Table 2 gives the dependence of the Langmuir “vessel-shape factor” on temperature for nitrogen and hydrogen. This factor, defined by the equation

\[ W = s \int_{T_1}^{T_2} K\,dT, \]

where \(W\) is the loss of energy in watts per second, \(T_2\) is the temperature of the filament, \(T_1\) is room temperature, and \(K\) is the thermal conductivity as a function of temperature, depends only on the nature of the gas and on the geometrical proportions.

Table 2.

Whereas for nitrogen \(s\) remains practically constant at all temperatures, for hydrogen it begins to increase anomalously, starting from \(2000^\circ\).

Already qualitatively one can establish that dissociation is occurring here, since when the pressure is lowered (which promotes dissociation) the effect becomes clearer. At reduced pressure the heat transfer, with increasing temperature, becomes greater than at high pressures (see Table 3).

This shows that the effect is not caused by the formation of an endothermic compound such as \(H_3\).

From this representation one can calculate, with sufficient approximation, the heat of dissociation, independently of special assumptions concerning the mechanism of the phenomenon (see p. 77).

Further, dissociation manifests itself at low pressures \((10^{-2}—10^{-3}\ \mathrm{mm})\) in the so-called “clean-up” effect: at temperatures of the tungsten filament from \(1300^\circ T\) abs. to \(2500^\circ T\) abs., the pressure of hydrogen in a closed space gradually decreases, which is explained by the adsorption of free hydrogen atoms on the glass walls (see adsorption of atoms). Finally, at low pressures it is possible to demonstrate and (within limited bounds) study the chemical activity of H-atoms.

Table 3.

In both of the latter cases reproducible results are obtained only with a perfectly definite character of the vessel walls and in the absence of greased joints and stopcocks. In Langmuir’s experiments the gas comes into contact only with glass and mercury and—in micro-gas analysis—with platinum.

In another apparatus, Compton1 and Duffendack2 obtained hydrogen atoms by a thermal method. A tungsten furnace, which can be used for dissociation, is described in Journ. Opt. Soc. 8 (1922) 910.

b) Production of atoms by a silent discharge. The foundations of this method were laid by Wood (literature reference No. 35). The path which he first entered upon proved the most convenient for studying the properties of H atoms. We shall describe here the apparatus (Fig. 1), which justified itself in subsequent investigations (Nos. 1, 4).

In this apparatus, the hydrogen atoms arising in the discharge tube must be led as far as possible from the place of discharge. They are studied in the space \(R\), before they can disappear as a result of recombination.

The criterion of whether those conditions have been attained under which many H atoms are obtained is, above all, the intensity of the Balmer spectrum in the discharge space. The concentration of H atoms is determined to a considerable degree by the character of the glass walls, on which, under certain conditions, strong catalytic recombination of H atoms takes place (see p. 82). Complications caused by ions appearing in the discharge are insignificant in view of the small concentration of these ions. This was proved by special experiments (literature reference No. 2). We did not take into account the possibility of the formation of \(H_3\), since the existence of such molecules proved unrealistic3.

c) Production of atoms in a sensitized light reaction. The above-described method of Cario and Franck (literature reference No. 6) is often used for obtaining hydrogen atoms. The original apparatus is very simple. A quartz tube containing several drops of mercury is evacuated by a diffusion pump. Hydrogen is introduced through a heated palladium tube, and the pressure is measured by a McLeod gauge. If now illu-

...to illuminate the quartz tube with the light of a mercury lamp with the unreversed line \(2536.7\ \text{Å}\) (cooling, magnetic field), then in the closed volume of hydrogen a lowering of the pressure is observed as a result of the adsorption of hydrogen atoms, as in Langmuir’s experiment. Since the absorption line is very narrow, the effect disappears immediately if the central part of the line is absent from the lamp radiation owing to self-reversal. Cario and Franck worked at a comparatively low pressure (not

Fig. 1. The hydrogen freed from oxygen and entering through tap \(H\) may be wet or dry. In the latter case oxygen is added to it at \(O\), and the water vapor formed is frozen out in \(U_1\). The strength of the discharge current between the electrodes \(E_1\) and \(E_2\) fluctuates between 100 and 500 mA. The gas emerging from the discharge tube at a rate of several meters per second and at a pressure of 0.5 mm can be destroyed by cooling with liquid air in \(U_2\). The chemical reactions taking place in \(R\) may give gaseous products, which are frozen out in \(A\) and then analyzed. \(E_3\) is an auxiliary electrode by means of which discharges are directed into the apparatus for its purification.

In the figure: “Hydrogen purifier”; “To the pump”; “To the McLeod gauge”; “To the freezing vessel”; \(5000\ \Omega\); \(5000\) volts.

above several mm). For certain purposes it is convenient to use considerably higher pressures, up to atmospheric. The absorption line \(2537\ \text{Å}\) is then greatly broadened, and a more complete utilization of the energy is obtained. Such an arrangement is already suitable for investigating the reactivity of hydrogen atoms in gas mixtures. Fig. 2 shows the apparatus used by Marshall and Taylor (Nos. 24, 25, 33, 34).

This method gives unambiguous data in gas mixtures for the course of the reaction, since for the most part it remains unclear which of the reaction components is excited by collision with the mercury atom in the \(2^3P\) state. However, the good agreement of the results obtained by this method with those given by other methods shows that, up to now, such complications have not in practice been encountered.

d) Hydrogen atoms in the voltaic arc. Finally, one must mention one more method, which has already found technical application (bibliographic reference No. 22). Langmuir (No. 18) connected the above-mentioned great lifetime of H-atoms, formed in a glow discharge, with the circumstance that, in order to maintain an arc in hydrogen at atmospheric pressure, an extraordinarily large potential drop is necessary.

The correspondingly large outflow of energy in the stationary state cannot be effected either by thermal conduction, or by convection, or by radiation, but only by diffusion of free hydrogen atoms.

Experience shows that a twenty-ampere arc between tungsten electrodes of \(6\ \mathrm{mm}\) cross-section, inserted obliquely into a tube of alundum (cross-section \(10\ \mathrm{cm}\)), through which hydrogen flows, melts within several seconds 2–3-millimeter iron wires at a distance of \(3\text{—}5\ \mathrm{cm}\) from the arc. The voltage at the electrodes varies from 300 to 800 volts,

Fig. 2. Mercury lamp with water-cooled refrigerator \(AB\); the evacuated annular space \(D\) allows one to keep the temperature in the light filter \(F\) (filled with a mixture of Cl and Br) and in the reaction space \(F\) independent of the temperature of the water (Marshall, Journ. of Phys. Chem. 30 (1926) 37).

the distance between the electrodes being up to \(2\ \mathrm{cm}\). The rapid melting of the wires can partly be explained by the catalytic combination of hydrogen atoms into molecules on the ...

surface of the metal. In favor of this view speaks the circumstance that a molybdenum rod (with a melting point of \(2800^\circ\) abs.), owing to its catalytic activity, melts sooner than a quartz rod (softening point \(1900^\circ\) abs.). If the explanation is correct, then from Langmuir’s data on the mean free path of free H atoms (not less than 7 mm) it follows that, at atmospheric pressure, equilibrium does not have time to become fully established in \(1/10\) second. This is rather surprising and stands in some contradiction to data obtained by the explosion method.

Fig. 3. Arc for a flame of atomic hydrogen.

The technical significance of Langmuir’s discovery lies not so much in the high temperature of the flame of H atoms (Langmuir estimates it at \(4000^\circ\) abs. by analogy with the estimate of the temperature of ordinary flames—from the heat of reaction, the change of heat capacity with temperature, and the degree of dissociation), as in the complete absence of carbon, nitrogen, and oxygen.

Fig. 3 shows an apparatus designed for welding in an arc with atomic hydrogen. Between two tungsten electrodes a rapid jet of \(\mathrm{H}_2\) is blown. From small openings arranged in the form of a sieve there simultaneously emerges a slow stream of hydrogen, washing over the place being welded and thus protecting it from oxidation.

Special grades of steel and other metals, which are difficult to weld by other methods, are welded by this method quite reliably and retain their mechanical properties.

III. PHYSICAL PROPERTIES OF THE HYDROGEN ATOM.

a) Absorption of light and dispersion. The absorption spectrum of the hydrogen atom in the normal state (to which we shall confine ourselves here), under definite conditions, has not yet been studied. Recently Lyman¹), in connection with other investigations, obtained the first four members of the Lyman series in the absorption spectrum.

Langer²) attempted to determine experimentally the dispersion in atomic hydrogen. He obtained the gas in a discharge tube. A substantial uncertainty is introduced by the fact that during the measurements the concentration of H atoms remains unknown. A direct determination is difficult. A preliminary estimate, made on the basis of other work (cited literature No. 2) and suffering from considerable uncertainty, gave him for \(n^2 - 1\) the value \(1.36 \cdot 10^{-4}\) instead of \(2.29 \cdot 10^{-4}\), obtained theoretically (on the basis of the new quantum theory) from the magnitude of the quadratic Stark effect³). Here \(n\) denotes the refractive index for infinitely long waves.

The discrepancy is not so great that it could be regarded as real⁴).

b) Excitation and ionization potential. The excitation potentials of atomic hydrogen were measured by Olmsted and Compton⁵). At a pressure of several hundredths of a mm, hydrogen was heated to \(2800^\circ\) in a tungsten furnace and in this process almost completely dissociated. By the photoelectric method it was observed at what electron velocities the luminescence of the atoms was excited. The first six excitation potentials—

¹) Lyman. Nature, 118 (1926) 156.
²) Langer. Proc. of the Nat. Acad. of Sciences (USA) 12 (1926) 639, 644.
³) Van Vleck. Proc. of the Nat. Acad. of Sciences (USA) 12 (1926) 662.
⁴) Langer himself derived theoretically the value \(3.34 \cdot 10^{-4}\) and cautiously expresses the opinion that the discrepancy between theory and experiment is real.
⁵) Loc. cit.

excitation gave values of 10.15; 12.05; 12.70; 13.00; 13.17; 13.27; 13.54 volts, in complete agreement with Bohr’s theory.

The accuracy of the measurements is estimated by the authors as 0.05 volt.

The ionization potential of the hydrogen atom was found by Duffendack1, and here too complete agreement with the theory was found. He observed the curve of the dependence of the current on the voltage for a low-voltage arc in an atmosphere of atomic hydrogen obtained by thermal dissociation. At 13.5 V the arc formed and broke.

c) Magnetic moment of free atoms. The magnetic moment of the free atom was determined simultaneously and independently by Wrede (No. 39) at Stern’s institute and by Phipps and Taylor (No. 29), using Stern and Gerlach’s atomic-beam method. Both found a value coinciding with Bohr’s magneton; according to the new theory this moment is explained by the “angular momentum” of the rotating electron. The principal experiments in both works were carried out with hydrogen atoms obtained in a silent discharge. In both cases very similar methods were used. The atomic beam emerges through a glass slit \(Sp_1\) (in Phipps and Taylor, \(0.075 \times 3\) mm; in Wrede, \(0.05 \times 3\) mm), obtained by sealing a metal sheet into glass and subsequently dissolving it in acid. A metal slit is unsuitable here owing to its catalytic action on the recombination of atoms. Details of Wrede’s apparatus are shown in Fig. 4.

The presence of atoms was proved by their reducing action on a mixture of molybdenum oxide with silver nitrate. Fig. 5 shows one of Wrede’s photographs. The simple strip indicates the place where the H-atoms fall in the absence of a magnetic field. (The plate \(Pl\) was then rotated through a certain angle.) From the distance between the two lines of the double strip one determines the splitting of the beam in an inhomogeneous field, corresponding to one magneton and agreeing with the requirements of the theory for an \(S\)-term doublet.

Fips and Taylor also qualitatively established the magnetic moment of atoms obtained by Langmuir’s method.

![Figure 4]

Fig. 4. The atomic beam \(R\) is cut out by the slits \(Sp_1\) and \(Sp_2\); fast-acting pumps are connected to \(P_1\) and \(P_2\). In the field of the magnets \(Pm\) the beam \(R\) is split and falls on the plate \(Pl\). The spaces into which the atomic beams, \(St\), enter, and the space into which they fall, \(A\), are connected only through \(Sp_2\). The parts \(Tr\), \(Kr\), and \(Mr\) serve for mounting and adjusting the system of slits. After Brede.

Quantitative determinations here are hampered by the different thermal velocities of the atoms.

d) Effective diameter of the free atom. Fraser1 has recently carried out an experiment to determine the effective cross-section (Wirkungsquerschnitt) of the free hydrogen atom as a function of its orientation in space. A visual representation of the Bohr model of the atom leads one to expect that, for a section in the plane of the planetary orbit of the electron, the diameter of the atom is considerably greater than for a section perpendicular to it. Earlier experiments2 had already shown that such a pred—

![Figure 5]

Fig. 5. After Brede.

representation imposes too great demands on the model of the atom. Fraser proved experimentally that the diameter of the atom does not depend on its orientation. In order to obtain a definite orientation of H-atoms, he made use only of the quantization of directions in a magnetic field described above.

Fig. 6. — \(AB\) discharge tube. \(B\) and \(C\) serve as cathode and are grounded. \(K\), \(E\), and \(F\) are connected to two pumps. \(A\) and \(K\) — gas inlets. \(C\) and \(S\) — magnetic shielding; solenoid \(L\), which creates the magnetic field for quantizing the directions in the beam. \(G\) — diaphragm. \(I\) — thermopile for removing ions. \(H\) — potassium beam. \(D\) — shutter. After Fraser.

The atomic beam was obtained from canal rays, the charged part of which was removed by an electric field. By means of a solenoid, quantization of directions was effected. When, in this process, the planes of the electronic orbits are oriented perpendicular to the direction of the beam, which increases the effective diameter and the number of collisions with the molecules of the remaining gas, the number of neutral atoms disappearing as a result of ionization in these collisions should have increased. It turned out, however, that the number of collisions and, consequently, the effective diameter, within the accuracy of the measurements (2–3%), do not depend on the orientation.

Schrödinger’s theory gives this a clear explanation: the characteristic function of the differential equation has, for the normal state, a spherically symmetric form.

Thus, in electrical respects the H-atom is spherically symmetric.

IV. Combination of H-atoms into \(H_2\) molecules.

a) Energy of combination.

To determine the heat of dissociation of the hydrogen molecule into atoms there exist several methods, completely independent of one another, re-

results of which are in complete agreement with one another. First of all, one must mention the spectroscopic method of Witmer and Dieke–Hopfield, the most direct and precise one. From the theoretical analysis of band spectra, carried out by Witmer1, the vibrations of the nuclei in the molecule can be found for the normal state almost up to the dissociation of the molecule. If one carries out a slight extrapolation to the dissociation itself, then, assuming that the separating atoms are in the normal state, one obtains the heat of dissociation at absolute zero equal to 4.34 volts = 100.1 kg-cal (upper limit 105.2 kg-cal, lower limit 94.6 kg-cal). This assumption seems quite natural and has also proved justified in other cases.

Further, Dieke and Hopfield2, considering the H₂ molecule with an excited electron, showed that there exist two systems of bands whose terms, for strong vibrations of the nuclei, pass into systems corresponding to one excited and one normal atom. Since the energy of the excited (two-quantum) H atom is known, the heat of dissociation is also obtained from this. It is equal to \(4.34 \pm 0.1\,V\). If, however, one starts from the beginning of the continuous spectrum, \(4.38\,V\) is obtained.

Results practically coinciding with these are obtained by applying the explosion method, the principle of which was indicated above. Here one obtains 95 kg-cal or, if the statistical weight for the hydrogen atom is taken to be two, a somewhat higher value, coinciding with Witmer’s. Since many variables enter into this calculation, it would not be surprising if the agreement were even somewhat worse.

Langmuir (Nos. 13, 19–21) was the first to begin determining the heat of dissociation. From his experiments he at first derived 130 kg-cal, then 84 and finally 97 kg-cal; Isnardi, using a very simplified calculation, found 95 kg-cal.

The order of magnitude here, too, turned out to agree with other data.

Bodenstein and Jung1 calculated from the kinetics of the formation of hydrogen bromide \(107 \pm 3\) g-cal. The calculation is based on the assumption that the rate of the process is determined by the reaction \(\mathrm{Br}+\mathrm{H}_2=\mathrm{HBr}+\mathrm{H}\), and that the observed temperature dependence may be interpreted as a heat of activation exactly equal to the thermal effect.

Finally, the heat of dissociation can be estimated from the chemiluminescence spectra (cited literature no. 3) produced by atomic hydrogen. Thus, it turns out that it can still excite the \(\mathrm{OH}\) group, whose excitation energy is equal to 92 g-cal, whereas the mercury line \(2537\ \text{Å}=112\) g-cal cannot be excited. Hence, with some caution, one may conclude that the dissociation energy is greater than 92 and less than 112 g-cal.

Unfortunately, it has not proved possible to convert measurements of ionization potentials by the electron-impact method into a method for determining the heat of dissociation, as was at first hoped. The primary process here always consists in the formation of a molecular ion. The appearance of atomic ions is explained by complications caused by a secondary transfer of energy to the molecular ion2.

b) Kinetics of the combination of H atoms. — a) Homogeneous combination of atoms. Until quite recently there existed no ideas whatever as to the path by which a chemical reaction such as the combination of two hydrogen atoms with one another proceeds. It was generally assumed that such reactions proceed very rapidly. This view contributed to the fact that the question of studying free hydrogen atoms was regarded as very difficult. Only one, perhaps analogous, association reaction was known, \((2\mathrm{NO}_2=\mathrm{N}_2\mathrm{O}_4)\), for whose course experimental data were available, and this reaction did indeed proceed at a very high rate. Grüneisen and Goens3 established, by Nernst’s method, measur—

PROPERTIES OF FREE HYDROGEN ATOMS

...to measurements of the speed of sound in dissociating gases, which established that the attainment of equilibrium continues for less than \(\frac{1}{10000}\) second.

When, in considerations concerning the kinetics of reactions, it was necessary to know the rate of reactions of the type \(2A = A_2\), it was usually assumed that the reaction is successfully completed at every collision (if one disregards the insignificant “steric” factor).

This assumption was called into question by one observation of Wood (ref. no. 35), who noted that the hydrogen atoms formed diffuse a measurable distance from the place of formation. When it subsequently became possible, by a chemical method, to follow quantitatively the disappearance of hydrogen atoms combining into molecules, the lifetime of free atoms at a pressure of several tenths of a mm proved to be equal to at least a third of a second. The measurement was made from the formation of hydrogen sulfide in the reaction between solid sulfur and hydrogen atoms (ref. no. 2). Further observations in the formation of hydrogen bromide from bromine vapor gave a lifetime of the H atom exceeding one second. This means that, in the extreme case, every millionth collision between two free atoms leads to combination. To explain this fact the hypothesis was proposed that, in an ordinary collision between two molecules, combination does not occur at all, and that it becomes possible only in the presence of a third molecule (no. 2).

This conception is compatible with the order of magnitude of the lifetime of H atoms at these pressures and explains the repeated failures of experiments on the activation of hydrogen at high pressures, since under such conditions recombination is accelerated and no accumulation of free atoms can occur. According to the new conceptions, the following picture may be constructed. Following the path chosen by Witmer and Dieke–Hopfield for determining the heat of dissociation from the vibrational terms of the molecule, one may conclude that, when two normal atoms combine into a normal molecule, the electronic configuration does not change. Therefore,

if radiation were associated with the combination of two normal atoms, then it would be caused not by an electron jump, but only by such transitions as lead from a state unquantized with respect to rotational and vibrational quantum numbers to the states of an already formed molecule. However, in nonpolar molecules the intensity of this radiation vanishes in the first approximation (the absence of a rotational-vibrational spectrum), so that, in the absence of external fields, we may disregard it. In other words, the dissociation state of the molecule H$_2$ is only a limiting case of the vibrational or rotational motion of a nonpolar molecule, and likewise metastable; therefore it does not pass into the normal state with the emission of energy. If, however, one denies the possibility of radiation and assumes smooth association in the collision of two atoms, then, together with Born and Franck$^{1}$, we arrive at a contradiction with the laws of conservation of energy and momentum. Indeed, the internal energy of the molecule formed on the basis of the conservation laws will not, in general, coincide with the molecular states permitted by the quantum laws. The angular momentum before combination will likewise not be quantized. If we disregard the very rare cases where this nevertheless occurs (within the limits of the indefiniteness of quantum states$^{2}$), then combination by this route cannot take place at all.

The situation is different in a triple collision. Here the third molecule may either, by its presence, remove the prohibition on radiation, or itself take care of the realization of the quantum state, taking upon itself the excess energy and momentum.

Independently of these arguments, based on quantum theory, the idea of a triple collision had already been introduced earlier on the basis of considerations of classical

$^{1}$ Born und Franck, ZS. f. Phys., 31 (1925) 411.

$^{2}$ See Polanyi and Wigner (Polanyi und Wiegner). ZS. f. Phys., 33 (1925) 429; Franck and Jordan (Franck und Jordan). Anregung der Quantensprünge durch Stösse, Berlin, J. Springer, last chapter.

PROPERTIES OF FREE HYDROGEN ATOMS

theory. Owing to lack of space, we cannot dwell on them here. This was done chiefly by Boltzmann, Jeans, and Herzfeld.

If the conception that we have adopted as a basis is correct—that for a more frequent combination into molecules what is chiefly lacking are partners which, by taking upon themselves the excess energy in a collision, would hinder the atoms from flying apart—then one should expect that also in a double collision not leading to combination, the minimum of the potential energy of the atoms is determined by the dissociation energy of the molecule. Hence it follows that this determines the Sutherland constant (the order of its magnitude should be \(10^4\) instead of the usual \(10^2\)) and that the temperature dependence of the internal friction of atomic hydrogen should be abnormally large. This could be tested experimentally without any special difficulties.

With the idea of a triple collision the following observation was associated (ref. lit. No. 4). If vapors of Na are admixed to a stream of atomic hydrogen, the \(D\)-line is emitted. It is natural to explain this by saying that, upon recombination of two hydrogen atoms, a sodium atom is excited in a triple collision. This would be an exact converse of the experiment of Cario and Franck described above. Here, however, the formation of molecules in a triple collision is just as unproven as the dissociation of a molecule at the moment of collision there.

Both here and there there exists the possibility that the process proceeds through the mediation of a chemical reaction (in our case—the formation of the compound NaH). It is noteworthy that in this case no emission of the higher members of the principal series of Na is observed, which might have been expected on the basis of the heat of dissociation of the molecule \(\mathrm{H}_2\). Likewise, the blue doublet does not appear in cesium vapors, as was recently confirmed by Moler (No. 28)\(^{1}\). In this selectivity one should not see an argument against the conception of a triple collisio—

\(^{1}\) Further negative results of Moler with Mg, Th, Zn agree with unpublished investigations of the author.

... The Cd line is excited by this route (No. 28), while the mercury line 2537 Å is not excited1.

In the ultraviolet region there is a band which is always excited by atomic hydrogen obtained in Wood’s tube. This is the so-called ultraviolet band of water vapor, whose carrier is the hydroxyl group. OH is formed in the discharge process and, at these pressures, evidently possesses such stability that it is pumped off together with the free hydrogen atoms. Its presence is caused by the presence of slight traces of water vapor or oxygen, which is deliberately added to the gas. As was indicated above, 92 kcal are required for the excitation of hydroxyl; these are probably released when H-atoms recombine into a molecule.

β) Combination on solid surfaces. The difficulties in experimenting with H-atoms are due not so much to their rapid recombination in the gas space as to their recombination on solid surfaces. The rate of disappearance of hydrogen atoms in ordinary apparatus is determined precisely by this process. As we shall see below, it is necessary here to assume the adsorbability of hydrogen atoms.

Already a priori one may assume a strong adsorbability of hydrogen atoms, proceeding from their “unsaturated character.” Indeed, there exists a certain relation between the adsorption potential and the “Sutherland constant” \(C\)2. It depends, as was indicated in the last section, on the heat of dissociation. If, as was said there, \(C\) for the hydrogen atom is 100 times greater than for argon, then the adsorption potential is approximately 10 times greater; i.e., hydrogen atoms are adsorbed in the same way as argon atoms at an absolute temperature 10 times lower.

The fact that the extremely high adsorbability of H-atoms has been experimentally demonstrated shows that it is inadmissible to restrict the formation of molecules to the action of van der Waals—

…of these forces only in the case of molecules with a small dissociation energy1.

Quantitatively, very little is known about adsorption. Equilibria, in view of experimental difficulties, have not been studied at all. Langmuir showed that when hydrogen atoms strike a glass surface at room temperature they are partially retained there. This is manifested in the well-known clean-up effect. However, even under the most favorable conditions no more than \(1/7\) of all atoms are adsorbed (calculated from the heat loss of the heated filament); the remaining atoms turn into molecules. At room temperature only an insignificant part of the glass surface is covered with atoms; further atoms falling on this surface evaporate again before recombination occurs. When the surface of the glass tube is in such a “state of saturation,” free hydrogen atoms can pass through it. At the temperature of liquid air the surface is completely covered with hydrogen atoms. Further H-atoms recombine with the adsorbed ones, so that atomic hydrogen does not pass through a glass tube cooled with liquid air. On the contrary, walls of ice at the temperature of liquid air adsorb not so strongly. This is evident from the experiments of Hansen2.

On metals adsorption is very strong; this is understandable, since already in equilibrium with molecular hydrogen at ordinary pressure many metals are saturated with hydrogen in the atomic state. This adsorption has not been measured directly, but from the kinetics of the decomposition of \(\mathrm{H}_2\) on incandescent tungsten filaments one may conclude that even at \(1500^\circ\) a noticeable part of the surface is covered with atoms. At higher temperatures the surface is freed from them (cited lit. No. 21).

The strong catalytic action on the combination of H-atoms speaks most clearly in favor of the adsorption capacity of metals with respect to hydrogen atoms.

Directly from the experiments, without resorting to considerations of a general chemical nature, one may conclude that in catalysis the adsorption of atoms is a necessary condition. Indeed, the activity of the walls is so great (this can easily be verified even by an approximate calculation), or, in other words, so many atoms are converted into molecules per unit time, that mere random collisions between two molecules from the gaseous space alone could not explain the result, since they occur too rarely. It is therefore necessary to suppose that a free atom collides with one already adsorbed and that both leave the surface in the form of a molecule, as was indicated above. The catalytic activity proved to depend both on the material of which the vessel walls were made and on the cleanliness of the surface. Different activity is observed spectroscopically, by introducing the substance under investigation into the space in which H-atoms are formed during discharge, and by examining the ratio of the intensities of the Balmer and many-line spectrum (Wood). Another method of observation is to compare the temperatures to which the investigated catalytic bodies are heated, other conditions being equal, in a stream of atomic hydrogen. In this case the substances may be applied directly to the bulb of a thermometer. The temperature differences are very considerable, for example, for

\[ \mathrm{Pd}\ 340^\circ,\quad \mathrm{Ag}\ 278^\circ,\quad \mathrm{Cu}\ 258^\circ,\quad \mathrm{Pb}\ 142^\circ . \]

By these and analogous methods it was found that the catalytic ability of metals decreases in the following series (cited lit. No. 2):

\[ \mathrm{Pt},\ \mathrm{Pd},\ \mathrm{W},\ \mathrm{Fe},\ \mathrm{Cr},\ \mathrm{Ag},\ \mathrm{Cu},\ \mathrm{Pb}. \]

It is easy to notice that in this series the metals are arranged in the same order as is observed with respect to the cathodic overvoltage in electrolysis. In the electrolytic production of hydrogen on metallic

at cathodes, to maintain gas formation, a generally higher voltage is required than that which corresponds to the reversible process. This overvoltage has long been connected with the difficulties which the primarily formed atoms encounter when combining into molecules. In this sense, metals with low overvoltage should catalyze well, metals with high overvoltage poorly, and this is indeed confirmed by experiment. This is a strong argument in favor of the given conception of overvoltage.

Further, the oxides of alkaline-earth metals and of the trivalent elements Al, Cr, Fe possess catalytic properties. Unglazed porcelain and glass powders also catalyze the reaction \(2H = H_2\). No substantial differences are observed between quartz and ordinary glass. Even with such small quantities as \(10^{-8}\) g of silver on the surface of glass in \(1\ \text{m}^2\), their catalytic activity is detected by thermal means. The strong influence exerted by oxygen on the emission spectrum of a discharge in hydrogen has also been connected with catalytic actions. The addition of oxygen to hydrogen in an amount of several percent significantly intensifies the atomic spectrum; in hydrogen completely deprived of oxygen, the atomic spectrum almost does not appear at all (in Wood’s apparatus). According to Langmuir, this may be explained by the fact that oxygen poisons the surface of the walls and thereby slows the recombination of ions. The addition of oxygen does indeed increase the concentration of atoms. That this is not a purely spectroscopic effect is seen most clearly from determinations of the concentration by chemical means (cited literature, No. 1).

Table 4.

To the hydrogen atoms obtained in the discharge, bromine vapors were added. The quantities of cm³ have been recalculated to room temperature and atmospheric pressure. Oxygen was admixed with the stream of hydrogen before the discharge.

The poisoning action of oxygen on catalysts is demonstrated by the following experiment. An apparatus similar to a radiometer is used, in which temperature differences are detected by the rotational moment imparted to a vane (No. 2). A cover glass silvered in this way—for example, on the front right and rear left—and suspended on a thread is a very sensitive device for detecting free atoms and has little inertia. With such an instrument it is easy to show the suppressing action of oxygen on the catalytic ability of metals.

If, under constant conditions, the deflection of the “radiometer” is measured during short current closures (by the ballistic method) at intervals of 5 minutes, the same deflection is always obtained. But if shorter intervals of time (1/2 minute) pass between individual readings, the values continually increase. The same thing is observed if the current is closed not for several seconds, but for a longer time. In other words, the radiometer behaves as though, from treatment with hydrogen atoms, it became more sensitive, and this sensitivity slowly disappeared. When the current is passed for one minute, the increased sensitivity remains for an hour. If oxygen or air is brought into contact with the surface, the sensitivity disappears. A surface that has been in contact with air for a long time is, at the first moment, catalytically completely inactive.

Senftleben (Nos. 30, 31) found the same effect of preliminary treatment in an entirely different apparatus for hydrogen atoms obtained by the method of Cario and Franck.

Langmuir (Nos. 12, 15, 17) established the “poisoning” action of oxygen in the dissociation of molecules, proving that in the presence of oxygen the dissociation of hydro-

THE PROPERTIES OF FREE HYDROGEN ATOMS

hydrogen on incandescent tungsten filaments proceeds not at all.

It has apparently not yet been established whether the action of oxygen can be reduced quantitatively to poisoning of the walls, or whether other factors may also operate here.

If a high concentration of free atoms can be obtained, as in the case of producing them electrically, then upon their contact with solid surfaces the emission of light is observed, just as, upon combination in the gaseous space, the glow of excited admixed gases is observed. The phenomenon becomes especially beautiful if phosphors with zinc sulfide are used as the contact substance (No. 4). The spectrum emitted in this case differs only slightly from that excited by light. It can be shown, however, that direct contact is required for excitation here and, consequently, that the luminescence is not excited indirectly by ultraviolet rays.

V. Chemical Reactions.

a) Reactions with metals. The behavior of metals with respect to free hydrogen atoms is not exhausted by catalytic activity. The possibility of chemical reactions must be taken into account. Indeed, according to Newman1, a Na–K alloy gives with hydrogen atoms (?) a white solid compound resembling hydrides; however, in Newman’s experiments it was not established whether the atoms are in fact the carriers of the activity.

It may be more surprising that mercury forms with hydrogen atoms a volatile hydride, which can be detected spectroscopically in the gaseous space (No. 4). When atomic hydrogen flows over the surface of mercury, an intense dark-blue glow is noticeable above it. Spectral decomposition shows that the glow is caused by the so-called HgH bands. It may be considered experimentally and theoretically established that the carri—

the carrier of the spectrum is a compound of mercury with hydrogen (No. 1). It is very probable that these are diatomic HgH molecules. In addition to the hydride bands, a continuous spectrum is visible, attributed by Franck and Grotrian to the Hg₂ molecule, and from the atomic spectrum only the line 2537. The line 2537 appears in the same parts of the tube as the hydride bands, i.e., near the mercury surface. Evidently, HgH is formed only as the result of a heterogeneous reaction with liquid mercury. The appearance of the line 2537 Å, which apparently cannot be excited by recombination of H atoms (112.6 cal), at first seems surprising. On closer consideration, however, it becomes understandable if one assumes that the excitation process is determined by the equation: HgH (excited) + H = Hg \((2^{3}P_{1})\) + H₂. Indeed, the presence of HgH (excited) has been proved spectroscopically.

It is possible that analogous hydrides are also formed with other metals; however, the experiments that were carried out gave a negative result¹). When a silver wire is heated in an atmosphere of atomic hydrogen to 600°, “spraying” is observed, which can hardly be explained by simple evaporation. It may proceed through an intermediate stage of hydride formation (as, for example, with As; see below).

b) Reactions with metalloids. Among reactions with metalloids, the interaction of H atoms with N₂, As, P, Sb, O₂, S, Br₂, Cl₂, and J₂ was investigated. The following may be said about these reactions.

The experiments showed that free hydrogen atoms are so chemically active that they react not only with most of the indicated substances, but also with their hydrogen compounds. In this case it is possible that, as a result of the latter reaction, the initial substance is partially obtained again. In such cases not a complete transformation is attained, but rather a certain stationary state, the position of which is determined not only by the rate constants of both reactions competing for the H atom, but also by the rate of removal from the reacti-

¹) Experiments of Möller (cited lit. No. 29) with Cd and unpublished experiments of the author with Cu and Ag.

...ing mixture of one or another component (for example, by freezing out). As we shall see, however (for example with HCl), it is not necessary that in the reaction of H-atoms with a hydrogen compound the original metalloid should again be obtained. The reaction of a hydrogen atom may here be only the first link in a chain of reactions, the final result of which is only the recombination of hydrogen atoms. We shall explain this in more detail below.

We shall consider the reactions in the order of the series written above.

Most of the investigations were carried out by the method shown in Fig. 1, according to which H-atoms are obtained in a silent discharge. Unless specifically mentioned, this is the method in question. In many cases there are control data on H-atoms obtained by other methods. The agreement is always very good.

Nitrogen. Nitrogen is completely indifferent. Under the action of hydrogen atoms on it, ammonia is not formed.

Langmuir (No. 14) found the same indifference of nitrogen with respect to atoms formed on heated tungsten filaments (Kario and Frank1 and Marshall and Taylor), to atoms obtained on contact with excited Hg; Tiede and Schleede2 showed that hydrogen atoms deposited on the cathode are incapable of reducing nitrogen.

This indifference of nitrogen with respect to H-atoms is in agreement with Haber’s conception, according to which in the synthesis of ammonia the nitrogen molecule undergoes substantial changes while being adsorbed by the catalyst; this, in his opinion, is indicated by the nature of the catalysts employed3.

Phosphorus, arsenic, antimony. P, As and Sb give the corresponding hydrogen compounds. For P and As the loss in weight over a definite time was determined quantitatively. Phosphorus te—

for 5 minutes: P—12 mg, As—10 mg. The quantities that disappear are not converted entirely into hydrogen compounds. Part of the phosphorus or arsenic is deposited as the element on the glass wall, getting there not in the form of vapor, but forming secondarily—as a result of the decomposition of the hydrogen compounds by active hydrogen. This was directly demonstrated for AsH₃ (see below: action on hydrogen compounds).

Oxygen. Oxygen reacts with hydrogen atoms (cited lit. No. 3); but the reaction is not as energetic as with the halides (see below). Two reaction products are possible: water and hydrogen peroxide. By analogy with what is known to us from laboratory practice, the second should be expected: in the interaction of hydrogen in statu nascendi with oxygen, H₂O₂ is constantly obtained1.

The amount of hydrogen peroxide formed is so large that it is possible to determine its percentage content (with respect to the method see No. 1). In this way yields of up to 76% by weight are obtained. From this it may be concluded that only hydrogen peroxide is formed primarily, and that water is obtained from it secondarily.

It was natural to test whether hydrogen atoms obtained by the method of Cario and Frank likewise give hydrogen peroxide, especially since Dickinson (No. 8) showed that hydrogen atoms obtained in this way react with oxygen. The formation of H₂O₂ was subsequently proved by two investigations (see Bates, Marschall and Taylor and cited lit. No. 5).

Sulfur. Sulfur reacts very rapidly with active hydrogen, giving hydrogen sulfide. The loss in weight of a piece of sulfur in 5 minutes is 6 mg. The above-mentioned measurements of the lifetime of free atoms were carried out with this reaction. The reaction does not proceed to completion, as with P and As; sulfur acids form on the walls of the vessel, of the same origin as in the case of P and As.

Chlorine, bromine, iodine. All three halides form the corresponding hydrogen compounds. When iodine comes into contact with atom-

gen, no chemiluminescence is observed (emission of the iodine fluorescence spectrum). By analogy with what was said on p. 81, one might have expected luminescence. But, in all probability, the reaction \(J_2 + H = JH + J\) makes excitation in the sense described above impossible. The formation of \(HJ\) was qualitatively established.

If bromine or chlorine vapors are brought into contact with a stream of atomic hydrogen, the point of their contact becomes very strongly heated. In quantitative experiments this point was cooled externally with water in order to eliminate the possibility of a thermal reaction between molecular hydrogen and the haloid. With equal amounts of atomic hydrogen, chlorine reacts more than bromine (Nos. 23, 1). This fact is of interest from the standpoint of the Nernst chain mechanism, from which it follows directly. The chain to whose formation one might try to reduce the abnormally large yield of \(Cl_2\) is in any case very short. However, Marshall\(^1\) showed that the length of the reaction chains in the photochemical formation of hydrogen chloride becomes ever smaller with falling pressure, and connected this with the small chain length observed in the present case. The interpretation of this decrease in chain length is still unreliable. We shall consider this question in connection with the behavior of halogen-hydrogen compounds, which, as has already been indicated, are not indifferent with respect to hydrogen atoms.

c) Reactions with hydrogen-containing compounds.
The following were investigated: \(CH_4\), \(NH_3\), \(AsH_3\), \(H_2O\), \(H_2S\), \(HCl\), \(HBr\).

Methane and ammonia behave indifferently or at least are very little active. On bringing them into contact with atomic hydrogen, no thermal effect was observed\(^2\).

Water vapor. Water vapor likewise gives neither a chemical reaction (formation of \(H_2O_2\)) nor a thermal effect. This excludes the possibility that hydrogen peroxide, of which—

\(^1\) A. L. Marshall, Journ. of Phys. Chem., 29 (1925) 1453.
\(^2\) See \(H_2S\), \(HCl\), \(HBr\), \(CH_3Cl\).

...which was discussed on p. 90, is formed as a result of the reaction \(H_2O + H = OH + H_2;\ OH + OH = H_2O_2\).

Hydrogen sulfide and arsine. These gases react at once with hydrogen atoms. In this reaction the metalloids are deposited on the walls of the vessel; the activity of the hydrogen disappears. At the same time the place of contact of the gases becomes heated. The reactions must proceed according to the following schemes: \(H_2S + H = HS + H_2;\ HS + HS = S + H_2S;\ AsH_3 + H = As + 2H_2\). The reactivity of \(AsH_3\) can also explain the behavior of \(As_2O_3\) toward H-atoms; here an arsenic mirror is deposited from the gas phase onto the glass, appearing as the result of the reaction of H-atoms with the hydrogen compounds formed as intermediates.

Hydrogen chloride and hydrogen bromide. When these gases come into contact with hydrogen atoms, strong heating occurs. The hydrogen atoms disappear instantaneously. It might have been expected that hydrogen bromide would not behave indifferently toward hydrogen atoms: the reaction \(H + HBr = Br + H_2\) is strongly exothermic; the equilibrium is strongly displaced to the right1. New data on the heat of dissociation of hydrogen chloride likewise make it possible to consider a rapid reaction in the direction \(H + HCl = H_2 + Cl\) possible. If, however, one starts from Nernst’s chain mechanism, the opposite might have been expected.

The question is what becomes of the free halogen atoms appearing in this reaction. From the thermochemical data it follows that a bromine atom can only recombine with another bromine atom or with an H-atom. In both cases a reaction product is formed (\(HBr\) or \(Br_2\)) capable of reacting further with hydrogen atoms. Thus hydrogen atoms may disappear, and, depending on the rate constants, different amounts of bromine are formed. With chlorine atoms the matter is the same, but here there is one further complicating possibility: the reaction \(Cl + H_2 = HCl + H\), since here, in all probability, the two oppositely proceeding reactions possess the same—

... rates (the equilibrium constant is approximately equal to 1). As experiment shows, the total effect in both hydrohalic compounds is that the hydrogen atoms disappear. In the case of hydrogen bromide, the simultaneous presence of free bromine can be proved (Ref. 1). This, of course, is reflected in the course of the above-mentioned reactions between active hydrogen and the halogens. Bromine was studied especially thoroughly in this respect. It turned out that only with a large excess of bromine is a constant amount of bromine converted into hydrogen bromide obtained; under ordinary conditions, as the amount of added bromine is increased, the amount of HBr formed also increases, despite the constant amount of H-atoms (see Table 5).

Table 5.

It might have been thought that the different amounts of hydrohalides which, as indicated, are formed from \(\mathrm{Cl}_2\) and \(\mathrm{Br}_2\) are connected with the fact that the hydrogen bromide arising in the reaction destroys the hydrogen atoms more rapidly than hydrogen chloride does. However, with a large excess of halide a constant amount is obtained, and in this case one must consider the influence of competing reactions of the hydrohalides with H-atoms to be excluded1. Under this assumption it is natural to suppose that the higher yield for chlorine is explained by a chain mechanism and that, with a large excess of bromine, the number of HBr molecules formed is equal to the number of free hydrogen atoms. At the place of mixing it is approximately equal to 5%.

d) Reactions with oxides, sulfides, and halide compounds of metals. The reactions of these compounds were studied only insofar as their occurrence is indicated by a change in color. Table 6 serves for orientation; the sign \(+\) means reduction to the metal; the sign \(-\): within 10 minutes no noticeable reduction occurred.

Table 6.

In addition, it is necessary to recall the transition of tungsten trioxide into the blue \(W_2O_5\), which served Langmuir as an indication of the presence of hydrogen atoms, the reduction of various nitrates and sulfates \((Cu, Pb)\) to metals, and of \(RaSO_4\) to \(BaS\).

e) Carbon compounds. Finally, let us mention some reactions with carbon compounds, although in this field only an insignificant fraction of reactions promising success has been studied. CO and \(CO_2\) seem to be indifferent to the hydrogen atoms obtained in the discharge. In both cases the formation of small amounts of formalde-

hydride. The reactivity of these gases is lower than that of all those mentioned above.

For the reduction of carbon monoxide, apparently, the experimental conditions used by Marshall (No. 25), who obtained significant quantities of formaldehyde, are more favorable. Marshall explains this by the dependence of the reaction on concentration, which he established.

With $\mathrm{CH}_4$ no heating, such as occurs with other hydrogen compounds, was observed, nor were there any other signs of a chemical reaction. It is apparently indifferent.

$\mathrm{CH}_3\mathrm{Cl}$ (methyl chloride), on the contrary, reacts very vigorously. The reaction products have not yet been studied.

Hydrogen atoms are capable of adding to double bonds and hydrogenating them. Oleic acid, after several minutes, solidifies into a mass resembling lard. Determination of the iodine number indicates strong hydrogenation. The carboxyl group is not affected. The reaction product appears to be stearic acid.¹

Taylor and Marshall (No. 34) showed, using hydrogen atoms obtained by the method of Kario and Frank, that they are capable of hydrogenating double bonds. They studied the formation of ethane from ethylene.

Finally, we shall mention the experiments of Kopo, Perepo, and Okara (No. 7), who dealt with the reduction of organic dyes by free hydrogen atoms.

VI. Conclusion.

If we now look back at the results obtained in the study of free hydrogen atoms, a very coherent picture emerges. On all essential points there is complete agreement, which is especially valuable in view of the differences in the methods used, including the methods for obtaining H-atoms. Despite several experiments that admit of different interpretations, the agreement of the results makes it possible to consi—

¹ According to the experiments of Waterman and Bertram (Waterman and Bertram), what occurs here is not the addition of hydrogen, but polymerization.

K. F. BONHOEFFER

to say that, in general, we are on the right path. A large part of the data concerning the physical properties of free atoms, their combination into molecules, adsorbability, and chemical activity may be regarded as firmly established. True, a considerable part of the material is qualitative in character. This is partly due to the fact that the question of the exact determination of the concentration of hydrogen atoms obtained by discharge has not been finally resolved. The development of an exact and reliable method from this point of view is very desirable. Another shortcoming, which is not so easy to remove, is the difficulty of obtaining large quantities of substance, connected with the necessity of producing discharges at low pressures. This considerably narrows the prospects of a chemical-preparative character. Nevertheless, in modern apparatus one can obtain a mole of free atoms at an expenditure of 14 kWh \((1.2 \cdot 10^4\) large calories), and no special efforts have been made in this direction to improve the results. True, in one apparatus this takes 70 hours. For laboratory experiments this method may still be of importance even now. It is possible that, for this purpose, the production of hydrogen atoms from molecules with the aid of excited mercury atoms will present certain advantages. Despite a number of shortcomings—for example, the complicated and uncertain course of possible reactions, the necessity of introducing mercury into the system, etc.—this method has the advantage that it permits work at high pressures. The main question here is a light source that would give the mercury line 2537 Å without great broadening and with high relative and absolute intensity.

LITERATURE

1) Boehm und Bonhoeffer, ZS. f. phys. Chem. 119 (1926) 385.
2) Bonhoeffer, ibid. 113 (1924) 199.
3) ” ZS. f. Elektrochem. 31 (1925) 521.
4) ” ZS. f. phys. Chem. 116 (1925) 391.
5) Bonhoeffer und S. Loeb, ibid. 119 (1926) 474.
6) Cario und J. Frank, ZS. f. Phys. 11 (1922) 161.

Literature

7) Coraux, Perepot and Hocart, Bull. de la Soc. chim. de France, (4) 37 (1925) 141.
8) Dickinson, Proc. of the Nat. Acad. of Sciences (USA) 10 (1924) 409.
9) I. Langmuir, Phys. Rev. 34 (1912) 401.
10) —, Transact. of the Electrochem. Soc. 20 (1911) 225.
11) —, ibid. 29 (1916) 294.
12) —, ibid. 29 (1916) 261.
13) —, Journ. of the Amer. Chem. Soc. 34 (1912) 860.
14) —, ibid. 34 (1913) 1310.
15) —, ibid. 38 (1916) 227.
16) —, Gen. Electr. Rev. 16 (1913) 962.
17) —, ibid. 25 (1922) 445.
18) —, ibid. 29 (1926) 153.
19) —, and Mackay, Journ. of the Amer. Chem. Soc. 36 (1914) 1708.
20) —, ibid. 37 (1915) 417.
21) —, ibid. 38 (1916) 1145.
22) —, and Weinmann, Gen. Electr. Rev. 29 (1926) 159.
23) A. L. Marshall, Journ. of Phys. Chem. 29 (1925) 842.
24) —, ibid. 30 (1926) 34.
25) —, ibid. 30 (1926) 1634.
26) E. Meyer, ZS. f. Phys. 37 (1926) 639.
27) A. C. G. Mitchel, Proc. of the Nat. Acad. of Sciences (USA) 11 (1925) 458.
28) Mohler, Phys. Rev. 29 (1927) 419.
29) Phipps and Taylor, ibid. 29 (1927) 309.
30) H. Senftleben, ZS. f. Phys. 32 (1925) 922.
31) —, ibid. 33 (1925) 871.
32) H. S. Taylor, Journ. of the Amer. Chem. Soc. 48 (1926) 2840.
33) —, and Marshall, Transact. of the Faraday Soc. 1925, Oxford-Band d. Zeitschr. f. phys. Chem. 120 (1926).
34) —, Journ. of Phys. Chem. 29 (1925) 1140.
35) R. W. Wood, Phil. Mag. (6) 42 (1921) 729.
36) —, ibid. (6) 44 (1922) 528.
37) —, Proc. of the Roy. Soc. of Amsterdam, 27 (1921) 455.
38) —, ibid. 102 (1923) 1.
39) Wrede, ZS. f. Phys. 41 (1927) 569.

1. According to Bodenstein and Lotkemeier (loc. cit.), the probability of the reaction \(\mathrm{H} + \mathrm{HBr} = \mathrm{H}_2 + \mathrm{Br}\) is approximately 10 times smaller than that of the reaction \(\mathrm{H} + \mathrm{Br}_2 = \mathrm{HBr} + \mathrm{Br}\). ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. Tiede und Schleede, ZS. f. Elektrochem. 27 (1921) 112. ↩↩↩↩↩↩↩↩↩↩↩↩

3. G. Kistyakovsky and Yu. S. Taldor came to this same conclusion on the basis of special experiments. ↩↩↩↩↩↩

4. Wohl, ZS. f. Elektrochem. 30 (1924) 49. ↩↩

5. Simon, ZS. f. phys. Chem. 123 (1926) 383. ↩