Active Nitrogen

G. O. Kneser, Marburg

I. Historical survey. II. Preparation. III. Chemical properties. 1. Compounds. 2. Catalytic phenomena. 3. Formation of surface layers. IV. Presence of atoms and measurement of concentration. V. Non-luminous modification. VI. Influence of the electric field, pressure, and temperature. VII. Volume and surface deactivation. Influence of the properties of the walls. VIII. Decay of the afterglow. 1. Decay function. 2. Measurement of decay. 3. Triple collisions. IX. Spectrum after the glow. X. Energy of active nitrogen. 1. Energy of formation. 2. Deactivation energy measured by the temperature effect. 3. From dissociation. 4. From ionization. 5. From excitation of spectral lines. 6. From excitation of band spectra. 7. Energy of the non-luminous modification. XI. Summary of experimental results. XII. Hypotheses on the nature of active nitrogen. 1. Metastable molecules. 2. Atomic nitrogen, triple collisions. 3. Combined hypothesis. 4. Index of the literature.

I. The Concept of Active Nitrogen. Historical Survey

Gaseous nitrogen, subjected to the action of an electric discharge, under certain circumstances acquires the ability to produce an “afterglow,” i.e. it continues to glow after the direct action of the discharge has ceased. This phenomenon is observed either in the same gas, the tubes filled with nitrogen often glowing over the course of an entire meter, or in a quiet gas after the current has been switched off; in such cases this glow is sometimes observed for a quarter of an hour. We call the state of nitrogen in which, among other things, it possesses the ability to give an afterglow, active; remark-

The same is true, however, of the fact that in certain cases, to be discussed later, afterglow is regarded as an insufficient indication of the presence of an active modification of nitrogen. It differs from the afterglow of a mixture of nitrogen and oxygen by its characteristic yellow color, whose spectral composition has been precisely studied. In this review we shall be concerned only with the phenomena observed for nitrogen. Their first systematic study was undertaken in 1900 by E. Warburg and P. Leube. Later (1911–1918) R. Strutt (Lord Rayleigh) published a whole series of papers in which highly important material for acquaintance with the properties of active nitrogen was presented. The study of active nitrogen entered a new stage from the time when Franck’s ideas on the transfer of energy by collisions of the second kind found wide application. As to the very nature of active nitrogen, we still cannot say the final word; therefore it is hardly worthwhile to refine the definition of the active state of nitrogen given above. A review article on this question must confine itself to a critical comparison of the experimental material and to an attempt to unite it on the basis of a single general hypothesis.

II. Preparation of Active Nitrogen

Active nitrogen is obtained by a discharge in nitrogen*, in most cases at low pressures. The usual arrangement of the inductor, Leyden jars, and spark gap is shown in Fig. 1a. For other purposes, for example for demonstrations, it is preferable to use the method of an electrodeless ring discharge (Fig. 1b). As it turned out, it is most convenient for this purpose to use tubes 10–20 cm in diameter.

At a pressure of 2 mm of mercury the largest amount of active nitrogen is usually obtained, but it depends strongly on the kind and intensity of the discharge. Correspondingly

* To obtain pure nitrogen one usually uses either the method of Kautsky and Thiele (Kautsky und Thiele, Z. anorg. Chem., 102, 342, 1925), or the method of Tiede (Tiede, Ber. Deutsch. Chem. Ges., 49, 1724, 1916).

ACTIVE NITROGEN

thereby the spectrum of the discharge that produces the afterglow changes. Apparently, only the presence of the second and fourth positive groups of bands of the nitrogen molecule^67 and of separate red and infrared arc lines of the atom^14 is a necessary condition for the occurrence of the afterglow. A discharge in which only the atomic spectrum is emitted, according to Fowler and Strutt, gives no afterglow at all. This, however, may be ascribed to the influence of some secondary effects^37.

Fig. 1a. Fig. 1b. Usual arrangements for obtaining active nitrogen.

Fig. 2. Intensity of the afterglow \((J)\) as a function of the place of origin in the discharge at 4.5 and 1.4 mm Hg, according to Strutt^66.
\(I\)—Crookes dark space; \(II\)—negative glow; \(III\)—Faraday dark space; \(IV\)—positive column.

The place where the afterglow arises, according to Strutt, should be sought near the cathode. Strutt photometrically measured the intensity \(J\) of the afterglow as a function of the position in the tube (the discharge was produced by a constant current), from which the gas was pumped out. The results are presented in Fig. 2. In a similar way he found that \(J\) depends little on the strength of the electric field, but increases with increasing current density. Strutt’s experiments are not entirely unambiguous, since in the positive glow there occurs not only activation but also deactivation. He showed this by causing a gas glowing with afterglow to pass through a second, considerably weaker discharge, whereupon the after-

the glow ceased. Wrede^86, by superimposing a weak discharge upon one already present, was able to show that the concentration of atoms decreases very appreciably (cf. Section IV).

To study the properties of the active gas it must be removed from the region of the direct action of the discharge. This is achieved either by causing the nitrogen to flow continuously through the discharge tube (as in A in Fig. 1a), whence it emerges already activated, or by examining the gas at rest only when the current is switched off. For optical studies in this case it is convenient to use a rotating sector, which admits light when the current circuit is open.

III. Chemical Properties

1. Combination reactions. Active nitrogen, in contrast to ordinary nitrogen, possesses to a high degree the ability to enter into compounds, which is what prompted its name. The products of the reactions have been little studied, as in the case of active hydrogen.

With the metals Na, K, Mg, Ca, Zn, Cd, Hg, Al, Tl, Sn, Pb, Fe, active nitrogen forms nitrides, which have been detected partly chemically and partly spectroscopically^64, 17, 18, 72, 73. Reactions are possible only with metal vapors, with the exception of Ca and Hg, which form nitrogenous compounds in the solid and liquid states.

The formation of nitrides also occurs with B, As, and S^64 (in the last case \(S_4N_4\) is obtained), whereas phosphorus only passes from the white modification into the red^60. With the halogens^81 no reaction products have likewise been obtained.

The formation of oxides in the presence of oxygen or ozone has been proved only spectroscopically, by the appearance of the so-called \(\beta\)- and \(\gamma\)-bands belonging to NO*. Ammonia

* On the formation of nitrogen oxides upon activation of oxygen, see A. König^23; in the arc—see F. Fischer und E. Hene, Chem. Ber. 46, 603, 1913.

is not formed61, 69, 80. H₂O, CO₂, and CO are not changed; NH₃ rapidly decomposes60, 80; HJ and HBr also decompose, but HCl does not decompose61, 80; on the contrary, in a mixture of haloids with hydrogen, under the action of active nitrogen HCl is formed, while HJ and HBr are not obtained61, 80. Nitrogen oxides, in the presence of active nitrogen, give a characteristic green luminescence with a continuous spectrum; in the case of NO the reaction product is N₂O₃87, 61. The reaction with CS₂ gives the blue sulfide NS and CS. With S₂Cl₂ and H₂S, as also with S, N₄S₄ is obtained60, 64. SnCl₂ and TiCl₄ give solid metallic nitrides, scarcely studied64. A whole series of organic substances gives hydrogen cyanide with active nitrogen; among these are acetylene, benzene, pentane, heptane, methyl bromide, ethyl chloride and iodide, chloroform, bromoform, ethylene and ethylidene dichloride, and ethyl ether64, 65. With methane, on the contrary, HCN is not formed80*. For acetylene, Strett65 assumes the following reaction: C₂H₂ + 2N = 2HCN, and an analogous one for benzene, although there are indications of the formation of cyanobenzene (C₂H₅—CN).

Glycerin does not react with active nitrogen; indigo, dissolved in strong sulfuric acid, is decolorized65. The glow that accompanies the majority of the reactions known to us has been separately and thoroughly studied and described17, 18, 60, 61. Since spectra are emitted in this process, the structure of which is known, we shall consider this question in Section X, §§ 5 and 6.

A very large number of solid bodies of more complex chemical structure are excited by active nitrogen and give luminescence, without reacting with it noticeably***. Since these

* The formation of NH₃ apparently occurs in the presence of free atoms. Caross and Rideal, Proc. Roy. Soc., London, A, 115, 684, 1927.

** Cf. also A. König u. E. Elöd, Chem. Ber. 14, 165, 1914. A series of hydrocarbons behaves in exactly the same way with respect to active hydrogen. Methane is likewise indifferent. See K. Bonhoeffer und P. Harteck, Z. physikal. Ch. A, Haber-Band, S. 64, 1928.

*** Very substantial material on this question was collected in the dissertation of Tannenberger72, p. 46 ff.; cf. further Lewis49.

phenomena are of much greater interest in the sense of studying the properties of the substances excited, and not of active nitrogen; therefore here we shall cite only the following two remarks from the summary of Tiede and Schlede^78:

1) Almost all strongly luminescing substances contain oxygen bound in their lattice, or, in general, elements with a small atomic number.

2) Substances known as having the ability to luminesce strongly—for example, the sulfides and oxides of the second group of the periodic system—are excited comparatively little, or are not excited at all.

1. Catalytic phenomena. Some substances absorb the afterglow, and no signs have been observed of any chemical reaction between them and active nitrogen; their action may therefore be called catalytic. In this respect the oxides of metals are typical, especially CuO, in whose presence the afterglow suddenly goes out, although neither luminescence nor a change in weight has been detected.^80 In the same way act (arranged in decreasing order of catalytic effect) Cu, Fe, Zn, Ag, Pt, W, and Mo^80; ^81; some of them do so only within a definite temperature interval. Bîley^81 explains this by saying that the actual catalysts in these cases are nitrides formed on the surface of the metal, stable only at that temperature. The disputed question, long discussed, of whether Hg vapor has a quenching action was settled by Tiede and Tannenberger in the negative.^84

If all known catalysts extinguish the afterglow, then, on the other hand, it has been firmly established that completely pure nitrogen, generally speaking, does not give an afterglow* and becomes capable of giving it only after admixtures of small quantities of catalytically acting gases. Such gases are the following (arranged in decreasing—

* This fact was established by Warburg–Lewis and especially emphasized by Tiede and Domke^72; at first it was disputed by Strutt and gave grounds for doubting whether the afterglow should indeed be attributed to nitrogen, which was later proved by studying the spectrum.

effect): $\mathrm{H_2S}$, $\mathrm{H_2O}$, $\mathrm{CO_2}$, $\mathrm{CO}$, $\mathrm{C_2H_2}$, $\mathrm{C_2H_4}$, $\mathrm{CH_4}$, $\mathrm{O_2}$, $\mathrm{Cl_2}$ and $\mathrm{H_2}$ ^15, 86*.

The quite different chemical structure of the substances acting in the same way makes it very probable to suppose that catalytic processes are involved here; in individual cases, moreover, a chemical reaction apparently also occurs. We shall confine ourselves here to the purely descriptive side of the study of these phenomena; in the section on the deactivating action of vessel walls we shall examine in more detail the action of catalysts.

1. Formation of surface layers. In individual cases a change has been observed in the surface of bodies exposed to the action of active nitrogen: platinum becomes coated with black, and polished surfaces of copper and zinc become matte ^69, 81, probably as a result of the formation of nitrides. Indirectly—from the change in the action on the afterglow—one may infer the formation of a surface layer on mercury. A quiet mercury surface does not harm the glow; when set in motion by shaking, it instantly extinguishes it; if it is covered with sulfuric acid, the effect remains the same ^62. By means of the formation of a heat-conducting layer, C. A. Kenty and Turner ^28 explained the lowering of temperature experienced by an incandescent tungsten wire in a stream of active nitrogen. The thickness of the layer is estimated at one atomic diameter**.

IV. Detection of nitrogen atoms and measurement of their concentration

Only recently has it been possible to prove the presence of atomic nitrogen in the active gas, and moreover by two different methods: 1) Bay and Steiner ^2 obtained, with the aid of

* The data of Pirani and Lachs ^53 cannot be cited here, since they concern the blue afterglow. According to experiments by Terenberg ^14 and Ilyin ^40 it appears doubtful whether oxygen alone can cause nitrogen to give an afterglow.

** Na, K, Ca, Ba and Mg exhibit surface luminescence; Mg always only on a surface that has long been exposed to the action of active nitrogen ^52.

of a weak electrodeless discharge in active nitrogen, the arc and spark lines of the nitrogen atom; in inactive nitrogen they are either entirely absent or very weakly expressed. 2) Wrede[^86] found a pressure difference in front of and behind the wall of a vessel with a very narrow diffusion slit ($Sp$, Fig. 3), on one side of which ($E$) active nitrogen is located. Under a stationary state this pressure difference is caused only by diffusion of the partly dissociated gas (from $E$), while only undissociated gas diffuses back, which is ensured by the corresponding catalysts ($K$).

Fig. 3. Apparatus for determining atomic concentration, according to Wrede[^86]. Control devices are omitted.

Wrede’s method is directly applicable to the study of atomic concentration. Under ordinary conditions it is found to be about 2% (by volume), and with frequent, stirring discharges, 30–40%. Experiments in which the dissociation in an electric discharge was determined by direct measurement of pressure* were probably unsuccessful because the degree of dissociation is too small for this not very accurate method to be applied.

The chemical method for determining the concentration was given by Strett[^61]; he mixed a known amount of activated nitrogen with NO and, after removing the excess NO, determined the weight of the resulting amount of $N_2O_3$. Assuming that the reaction proceeds between N atoms and breaks down into two processes, namely:

\[ 2NO + N = NO_2 + N_2, \]

\[ NO_2 + NO = N_2O_3, \]

the atomic concentration is found to be equal to 2.5% by weight. Since, however, it cannot be proved that every other process is impossible, and since, as always, part of the active nitrogen

* W. H. Crew and Hulbert[^10]; for criticism of their method see Wrede[^86].

catalytically, i.e., without the formation of reaction products, passes into an inactive state, then this method is not free from errors. Willey, for measuring the concentration of atoms, made use of the liberation of heat upon deactivation; moreover, one thermometer \(T_1\) (Fig. 4) was surrounded by a net of oxidized copper wire, while the other—\(T_2\)—was placed in nitrogen activated in \(A\). The first indicated a higher temperature than the second, owing to catalytic deactivation. The temperature difference measures the degree of activation, but since it cannot be assumed that heat is evolved only

Fig. 4. Measurement of concentration and detection of the non-luminous modification according to Willey.

upon recombination of atoms, these investigations, especially in comparison with the first method indicated, permit only relative conclusions about the atomic concentration.

V. Non-luminous modification

If the yellow afterglow is a visible sign of the existence of an anomalous form of nitrogen, this still does not prove the necessity that every chemically active modification should manifest itself by the ability to give an afterglow. In fact, one can observe a whole series of phenomena characteristic of active nitrogen, while at the same time it either seems to have lost its ability to glow, or has been artificially deprived of it. For example, if a tube through which luminous nitrogen is flowing is strongly heated at one point, then at this point the afterglow disappears, but it appears again in the unheated parts of the tube (Strutt \(^{60}\)). Cario and Kal-

G. O. KNEZER

Lang 7, 26 showed that, with sufficiently strong heating, the \(D\)-line of Na appears at this place; consequently the sodium vapors emitted by the glass are excited by a non-luminous gas. By weakening the activating discharge (by reducing the capacity), the afterglow can be eliminated, but an entirely weak auxiliary discharge in the nitrogen obtained in this way gives a different spectrum (the 4th positive group of bands) than nitrogen in the normal state 3. Phosphorus vapors are excited by this active, but non-luminous, nitrogen and give luminescence 34. Further, an experiment of the highest importance is described by Strutt 61. Into a wide tube (Fig. 5), through \(B\) enters luminous nitrogen; through \(C\)—inactive nitrogen mixed with phosphorus vapors. At \(D\) this mixture is discharged.

Fig. 5. Detection of a non-luminous modification, according to Strutt.

The luminescence of the phosphorus vapors is observed not near the aperture \(C\), but chiefly at \(E\), where the afterglow of the nitrogen has already completely died out.*

Willey 82 made an attempt at systematic investigation in this direction. He caused the activated gas (at \(A\), Fig. 4) to flow through a tube; at \(B\) there occurs a weak electric discharge, which destroyed the glow; then he measured the atomic concentration. It turned out that, over wide limits, it does not depend on the intensity of the afterglow.**

The only obvious result of all these experiments is that active nitrogen, losing its abil-

* Strutt, in addition, observed the following phenomenon: if \(C\) is moved along the axis of the tube, the luminous cloud of phosphorus vapors remains in place until the orifice of the tube reaches it; if \(C\) is moved still farther, then it moves along with it, though, as before, the glow decays. The same phenomenon was observed by König and Elöd 38.

** In a certain sense this is contradicted by the aforementioned result of Wrede, who, under the same conditions, observed a decrease in the atomic concentration.

ability to glow, but does not thereby lose the ability to induce luminescence and to exhibit chemical and thermal actions. Bay and Steiner[^3] suppose that active nitrogen in this nonluminous modification consists chiefly of molecules possessing high energy levels.

In what follows we shall throughout note what pertains to the luminous and to the nonluminous modification, especially in cases where conclusions are drawn from the properties of luminous nitrogen.

VI. Influence of the Electric Field, Pressure, and Temperature

When active nitrogen was passed between electrodes between which an electric field had been excited (up to \(3000\ \mathrm{V}/\mathrm{cm}\)), neither quenching of the afterglow, nor deflection of the luminous gas, nor any change whatever was detected[^61],[^68]. The current between the electrodes is proportional to the surface of the electrodes; this current could be only a photocurrent. Thus ions are not contained in active nitrogen in either modification[^29],[^9]. No influence of a magnetic field has been found either[^28].

Fig. 6. Dependence of the afterglow on pressure.

The influence on the afterglow of a change in pressure can be investigated only in a quiescent gas and far from the discharge tube, since otherwise the discharge conditions and the composition of the active gas will not remain constant. Otrett[^62] used the apparatus schematically shown in Fig. 6. Nitrogen activated in \(A\) by an electrodeless discharge diffused rapidly into \(B\) and could be confined there by means of a tube filled with mercury and covered with a layer of sulfuric acid. Thus compressed in \(B\), it glowed more brightly and the gas glowed faster; while traces of luminescence still remained, further compression intensified it. If, however, by lowering \(C\), we bring the gas to the initial pressure, then

glow in \(B\) proves to be much weaker than in \(A\), where no noticeable change in pressure occurred. Since the rate of extinction depends on the pressure, deactivation is evidently not a monomolecular process (cf. Section VIII, 2).

A spontaneous decrease of pressure during the process of decay was observed by Bredig \(^{36}\) and explained by him as a consequence of recombination of atoms. However, in the steady state, i.e. during prolonged formation and destruction of active nitrogen in a closed space, no change whatsoever in pressure could be detected, even at the temperature of liquid air; from this it may be concluded that active nitrogen contains no constituent that liquefies at this temperature.

If the part of the vessel in which the afterglow is observed is strongly cooled, then at the boundary of the coldest parts the gas glows more brightly; on both sides of them the glow dies out almost at once. This may be explained as a consequence of the increased density occurring in the cooled parts and producing greater brightness and a more rapid extinction of the afterglow (see above). If now the entire vessel is immersed in liquid air, then after the current has been switched off an afterglow appears that is much brighter than at room temperature, but it decays very sharply and immediately, like lightning. This phenomenon cannot be explained in the preceding way, since no change in density occurs. Likewise there remains unexplained the observation that at \(100^\circ\text{C}\) the afterglow, though of lower intensity, also goes out more rapidly than at room temperature, so that, consequently, in this case the total emitted energy is considerably smaller \(^{62}\).

VII. VOLUME AND SURFACE DEACTIVATION; INFLUENCE OF THE VESSEL WALLS

The dependence of the afterglow on temperature becomes more or less intelligible if, together with Strett \(^{62}\), we assume that there exist two deactivating pro-

process—an assumption fully justified: 1) deactivation in a closed gas volume, accompanied by afterglow and accelerated by increasing pressure and lowering temperature; 2) nonluminous deactivation on the walls of vessels, which depends to a high degree on the properties of the walls and increases with rising temperature. At high temperatures nonluminous deactivation at the walls predominates; at low temperatures, luminous volumetric deactivation does. That luminous deactivation takes place chiefly inside the vessel can be shown by photometry of the afterglow at different layer thicknesses, or else by removing the active nitrogen from the observation window by a stream of inactive nitrogen. The results of both experiments^31 speak in favor of S t r e t t’s hypothesis. As immediate direct proof of the deactivating action of the walls, let us cite G e r b e r t’s^14 experiment. He worked with a quartz vessel and an electrodeless apparatus. By thorough heating and pumping out the gas he completely extinguished the afterglow; if traces of H₂ are then admitted, the luminescence reappears with undiminished brightness and remains during repeated pumpings-out and fillings with pure nitrogen; only renewed heating could destroy it again**. A very clean degassed quartz surface therefore favors, in contrast to the gas contained, the process of surface deactivation, so that in the very shortest interval of time the entire quantity of activated gas returns to the normal state, and volumetric deactivation, and consequently luminescence, is not observed. What factors in fact assist the deactivating

* Its temperature coefficient is therefore negative, in contrast to all chemical processes. K a r i o and K a p l a n^7 assume that all this is caused by impurities which at low temperatures condense on the walls and intensify the deactivation of the walls. The great initial intensity is, of course, not explained by this.

** From measurements of the conductivity, K a r r e r and F a z e l^27 attempted to draw conclusions concerning the volumetric surface of deactivation.

*** Similar observations are described by K e n i g and K l i n k m a n n^88, S t r e t t^85, W o n g e f o r and K a m i n s k i^5.

to the action of the walls, and whether, other conditions being equal, their action depends on temperature—whether it decreases with a lowering of the temperature and increases with its increase, as was assumed for explaining the temperature course of the afterglow—we cannot say in the present state of the question.

In view of these circumstances, and taking into account the independence of the luminescence spectrum from the admixtures causing it,^5 we may reduce the action of catalysts (cf. III, 2) to an increase or decrease of the surface deactivation. Undoubtedly, the smallest admixtures that cause the afterglow of pure nitrogen act precisely in this direction. They change the properties of the walls in such a way that their capacity to deactivate the gas is diminished. A certain amount of the added gas must make possible the occurrence of luminescence; consequently the capacity for afterglow depends on the partial pressure of the added gas. In other words, the maximum permissible addition, in percent, must be greater at low pressures than at high ones. In fact, this is confirmed for oxygen admixtures in a number of different observations.^74,5,14 At 4 mm Hg the most favorable partial pressure of oxygen should be equal to 0.1 mm Hg;^5 at very low pressures it is sufficient to produce a yellow afterglow in air.^18,47,14,40,22 The fact that a larger amount of foreign gas destroys the afterglow is probably explained by other, chemical interactions.^60,61,85,36,5,80

VIII. Extinction of the Afterglow

1. The extinction function.

Let us suppose that the process of deactivation, accompanied by afterglow, occurs between active particles and normal nitrogen molecules, and that in each elementary process there participate $\alpha$ active particles and $\beta$ molecules. The number of the former per unit volume $(n)$ is small in comparison with the number of molecules $(N)$. The number of elementary processes during the time element $dt$

proportional to \(n^\alpha N^\beta\,dt\) and at the same time proportional to the decrease in the number of active particles, hence:

\[ -dn=An^\alpha N^\beta dt;\,^{1} \]

whence

\[ n=(\alpha-1)\left(AN^\beta t+C\right)^{\frac{1}{1-\alpha}}. \]

Let us further assume that each act of radiation is connected with the process of deactivation and conversely. Then the intensity of the afterglow must also be proportional to the decrease in the number of active particles with time:

\[ J=-\frac{dn}{dt}\cdot B=ABn^\alpha N^\beta. \]

Eliminating \(n\), we obtain:

\[ J=ABN^\beta\left[(\alpha-1)\left(AN^\beta t+C\right)\right]^{\frac{\alpha}{1-\alpha}}, \]

where \(C\) is the constant of integration, determined by the initial conditions.

From this we obtain:

\[ \text{for }\alpha=1:\quad J=ABN^\beta e^{-AN^\beta t-C};\ \lg J\simeq t; \]

\[ \text{for }\alpha=2:\quad J=ABN^\beta\left(AN^\beta t+C\right)^{-2};\ \frac{1}{\sqrt{J}}\simeq t; \]

\[ \text{for }\alpha=3:\quad J=\frac{1}{\sqrt{8}}ABN^\beta\left(AN^\beta t+C\right)^{-3/2};\ \frac{1}{J^{2/3}}\simeq t. \]

In reality, however, the assumptions made above, as follows from Section VII, are almost never fulfilled: as soon as deactivation on the walls is present, there can be no mutually unambiguous correspondence between radiation and the elementary process of deactivation, and consequently the decay of the afterglow cannot be calculated

according to the formulas given above.* However, if it turns out that $\lg J$ or $\frac{1}{\sqrt{J}}$ is proportional to the time, then it may be considered proven that the deactivation of the walls can be neglected and that $\alpha$ may be put equal to unity, two, etc.

1. Measurement of the decay. The decay function has been measured many times: in a gas at rest—by Angerer^1 with the aid of a photoelectric recording apparatus, by Rüde^55 and Kneser^33—by subjective photometry under repeated excitations; König and Klinkmann^38, as well as Willey^75, photometrically measured (the former photographically, resolving into a spectrum, the latter colorimetrically) the intensity along a tube through which the luminous gas was flowing, the temporal decay being obtained from the calculation of the velocity of flow.

From all these experiments (with only one exception) it turned out that, under certain conditions favorable to afterglow, the expression $\frac{1}{\sqrt{J}}$ depends linearly on time (with small deviations), i.e. that $\alpha=2$ (see curve $a$ in Fig. 7). Conversely, for a weak short afterglow and, consequently, for considerable deactivation by the walls, $\frac{1}{\sqrt{J}}$ is not proportional to $t$ (curve $b$)* in agreement with the assumptions developed above.

Bonhoeffer and Kaminsky^5 compared the intensity of the afterglow at different places in tubes through which gas was passing, as a function of the concentration, which could

* The inverse case of radiation without deactivation cannot be considered for energetic reasons.

** König and Klinkmann, placing a tube with flowing luminous gas in place of the slit in a spectrograph, measured the blackening from the various spectral lines obtained in this way and found a linear decrease of $\log S$. However, the authors do not think that this should be seen as a contradiction of all the above-mentioned measurements.

*** The scale of abscissae in Fig. 7b, compared with 7a, is enlarged by a factor of 10. Cases were also observed in which the decay curve had a point of inflection; this can be explained by variability in the properties of the walls.

be reduced at the second observation point by introducing an inactive gas between observation points 1 and 2. Throughout the whole tube the pressure, and consequently \(N\), remained constant, but at each observation point \(n\) (the number of active particles per unit volume) changed inversely proportionally to the flow velocity, i.e., to the quantity of gas passing per second through the cross section of the tube.

Fig. 7. Decay curve at 0.1-mm pressure in the presence of deactivation on the walls and in its absence, according to Kneser.

gas, which in the first cross section was \(S_1\), and in the second \((S_2+S_1)\). Hence, if the constant factors are denoted by \(K_1\) and \(K_2\), we have:

\[ n_1=\frac{K_1}{S_1}; \qquad n_2=\frac{K_2}{S_1+S_2}. \]

The quantity

\[ \frac{K_2 n_1}{K_1 n_1}=\frac{S_1+S_2}{S_1} \]

can be obtained from measurements of \(S_1\) and \(S_2\). By means of equation (2) we obtain:

\[ \frac{J_1}{J_2}=\left(\frac{n_1}{n_2}\right)^\alpha . \]

The results of the measurements (Fig. 8) show that \(\dfrac{J_1}{J_2}\) depends linearly on \(\left(\dfrac{h_1}{h_2}\right)^\alpha\). This again means that \(\alpha=2\).

From all that has been said above we conclude that in the elementary process of emission and deactivation there always take part two active particles. From the slope \(M\) of the straight lines representing \(\dfrac{1}{\sqrt{J}}\) as a function of time \(t\) for various \(N\), i.e. various pressures, \(\beta\) is determined by means of equation (4):

\[ M=\frac{d}{dt}\frac{1}{\sqrt{J}}=\sqrt{\frac{A}{B}}\cdot N^{\frac{\beta}{2}} . \]

Fig. 8. Measurement of afterglow intensity as a function of concentration, according to Bonhoeffer and Kaminsky\(^5\).

Kneser finds that in the range from 0.05 to 0.8 mm Hg the angle of inclination is approximately proportional to the square root of the pressure, i.e.

\[ M=\text{const}\,N^{\frac{1}{2}},\quad \beta=1. \]

The same result was obtained by Willey at higher pressures, when the deactivation of the walls is reduced.

3. Triple collision. The results of the preceding chapters hardly allow one to doubt that in each elementary act of afterglow emission two active particles and (probably) one neutral molecule participate. Therefore it may be called a triple collision.

Such a conception is confirmed by the following experiment. Since the probability of a triple collision is proportional to the number of neutral molecules \(N_2\), the intensity of the afterglow must increase with the addition of an inactive gas, even when the increase of damping with increasing pressure due to collisions of the seco-

of this kind⁵. The damping curves shown in Fig. 9a reveal this effect*, which is obtained as a result of admitting a neutral gas into a vessel filled with a luminous gas³³. This effect cannot be caused by a local accumulation of active particles as a consequence of a sharp change in pressure, since it follows from the experimental data that complete mixing already occurs after \(1/2\) sec.

Fig. 9. Experiment with a triple impact; increase in the brightness of the afterglow by admixing an inert gas (pressure initially \(1/20\), finally \(1/2\) mm Hg);
\(J\)—in arbitrary units (after Kneser³³).

These observations and measurements of damping prove experimentally the existence of a triple collision. The addition of argon (Fig. 9b) causes an increase in brightness to almost the same degree; evidently, in this case, instead of the nitrogen molecule, the argon atom assumes the role of the third partner in the complex collision. What remains unclear is the circumstance that in a mixture of nitrogen with noble gases the angle of inclination \(M\), which is a function only of \(N\), i.e.

* Kindly communicated to the author before publication.

** According to Bonhoeffer and Kaminsky, no increase whatever in the brightness of the afterglow is observed in these experiments. They deliberately added the inactive gas extremely slowly, as a result of which, possibly, the enhancement of the afterglow in the very short interval of time after the introduction of the inactive gas was compensated by a decrease in intensity due to the rise in pressure.

the number of inactive particles contained in a unit volume depends much more strongly on the total pressure than on the partial pressure of \(N_2\) \(^{33}\).

IX. Afterglow Spectrum

In the molecular spectrum of nitrogen the following groups of bands are distinguished:

The negative group belongs to the singly ionized molecule. The system of bands formerly designated as the third positive group (3,000—2,000 Å), as it turned out, belongs to NO.

Fig. 10. Scheme of the levels of the \(N_2\) molecule, after Sponer \(^{58}\), supplemented.

The indicated 5 groups correspond to the levels of electronic energy schematically represented in Fig. 10. The system of bands corresponding to the transition \(X—A\) is observed neither in the emission spectrum nor in the absorption spectrum. It is very probable that \(A\) is a metastable state. The difference of the levels \(A—X\) is known only from experiments with electron impacts and therefore can be given with an accuracy only up to \(+0.3\ \mathrm{V}\). The same must be said about the ionization energy \((X—X')\), whereas the remaining differences of levels can be obtained from spectroscopic data, and consequently much more

more precisely. The spectroscopic study of the afterglow shows that, out of the entire region of the molecular spectrum from 1200–10500 Å, in the afterglow spectrum there are present

Fig. 11. Scheme of the first positive group of bands.

only individual bands from the first positive group—namely, 3 or 4 neighboring bands each in the green, yellow, and red parts, as has now been established by Cario and Kappland, and in the near infrared, with very great intensity, whereas the other groups and the atomic spectrum are entirely absent.

Since it is precisely the spectrum that can provide very valuable

information on the nature of active nitrogen, then we must acquaint ourselves somewhat more closely with the structure of the first positive group of bands. Fig. 11 depicts the scheme of levels (terms) according to Birge. It may be regarded as a detail supplementing Fig. 10. The vibrational quanta of state \(B\) (the initial state of the first positive group of radiation) are denoted by \(n'\) (equal to from 0 to 19), and state \(A\)—by \(n''\)

Fig. 12. Distribution of intensity in the first positive group of bands: \(a\)—afterglow in pure \(N_2\), \(b\)—in Ne with traces of \(N_2\) (Rayleigh \(^{70}\)), \(c\)—at very low pressure, \(d\)—at the temperature of liquid air (Herzberg \(^{15}\)).

(equal to from 0 to 15). The transitions observed in the normal spectrum are indicated by thin lines; those observed in the afterglow spectrum—by heavy lines with arrows. They originate predominantly from the levels corresponding, according to Kario and Kaplan, to \(n'=11\) and \(n'=6\). From their height above state \(A\), \(18.0\pm0.3\), one may think that the energy required for elementary nitrogen emission varies from \((8.0\pm0.3)+3.47=11.5\pm0.3\ \mathrm{V}\) to \((0.8\pm0.3)+2.38=10.4\pm0.3\ \mathrm{V}\).

The absorption spectrum of the afterglow is unknown. Even

when the thickness of the absorbing layer is \(1\,m\), in the entire wavelength region from \(10500\)—\(1200\,\mathring{\mathrm A}\) there are neither lines nor absorption bands \({}^{5,64,6}\). This is all the more surprising since, after irradiation, molecules in state \(A\), which is assumed to be metastable,* should accumulate in considerable quantity. In particular, the absence in the absorption spectrum of bands corresponding to the transition \(n' \to 0\) was proved by Cario in studying the infrared part of the spectrum \({}^{6}\).

The distribution of intensities in the spectrum is little changed by the presence of impurities \({}^{5}\), whether they act as catalysts in small quantities or as agents altering the properties of the walls. In mixtures of nitrogen with noble gases, in which the vanishingly small quantities of nitrogen can give an intense afterglow \({}^{15}\), a shift of the intensity maximum toward the red side within each separate group is observed \({}^{70}\). Toward the violet end the intensity maxima shift at low pressures (\(0.1\,\mathrm{mm}\,\mathrm{Hg}\)) and low temperatures (liquid air, Fig. 12). Cario and Kaplan explain the latter circumstance by the fact that, with weakening of the molecular thermal motion, the selectivity of excitation becomes more pronounced.

X. Energy of Active Nitrogen

The fundamental question, the answer to which must be the principal point of any hypothesis concerning the nature of active nitrogen, is the question of its energy.

1. Energy of formation. The experimental material concerning the question of what energy is required to obtain active nitrogen is very scanty, since in the usual electrical method of producing it a quantity of energy that cannot be accounted for is spent on all sorts of secondary effects.

* This is confirmed by Kaplan’s observations \({}^{24}\). He excited in active nitrogen, with a weak discharge, the 4th positive group, the energy level of which lies \(6.8\,V\) above the level \(A_n\) and \(14.8\,V\) above the normal state. In inactive nitrogen this could not be accomplished. Resonance fluorescence, which one would expect for metastable molecules, was also sought without success.

Kenti and Turner\(^{28}\) attempted to activate nitrogen by collisions with slow electrons. They judged the formation of active nitrogen by the appearance of a surface layer on a tungsten wire, altering its resistance. Active nitrogen was formed if the gas surrounding the wire was bombarded with electrons of \(10.8 \pm 0.5\ \mathrm{V}\). This value agrees with the excitation voltage of the afterglow bands. Attempts to activate nitrogen thermally were made by Langmuir, but gave negative results. It is possible that the temperatures required for this are considerably higher than those attainable experimentally.

2. The energy of deactivation, measured by the thermal effect. Rideal and Willey\(^{79}\) measured in a thermostat the amount of heat released during the deactivation of active nitrogen in the presence of NO, and at the same time determined the amount of reaction products. Since the reaction mechanism, which must be known in order to calculate the heat of deactivation, still requires special consideration, the results of these experiments (about \(42\) thousand cal./mole \(= 1.8\ \mathrm{V}\)) are not very acceptable. Another method is based on measuring the change in resistance of a wire observed during the deactivation upon it of active nitrogen, which gives up its energy to it. Since the calculation here involves the concentration of active nitrogen, and since the main assumption—that all the heat of deactivation is given to the wire—is satisfied only very approximately, this result too (\(47\) thousand cal./mole \(= 2\ \mathrm{V}\)) is completely unreliable.

3. From dissociation processes. Those chemical reactions of active nitrogen in which intermediate reaction products might be obtained, such as, for example, the formation of \(\mathrm{N_2O_3}\) from NO, or in which the action of photosensitization can be taken into account, as in the formation of HCl from \(\mathrm{H_2}\) and \(\mathrm{Cl_2}\), do not provide suitable material for determining the energies of deactivation. Even for diatomic molecules which, in all probability, dissociate in a single elementary act and whose dissociation energy \(D\) is known, we do not obtain dissociation products owing to the fact that rapid recombination occurs together with it.

In addition, it is known that in some cases dissociation does not occur, although energetically it is entirely possible, i.e., the supplied amount of energy is greater than \(D\).

Below we give separate results of Wile’s experiments[^83], based on the chemical determination of the reaction products. The following dissociate: HJ \((D = 2.9\ \mathrm{V})\); HBr \((D = 3.6\ \mathrm{V})\) and NH\(_3\). The following do not dissociate: H\(_2\) \((D = 4.4\ \mathrm{V})\); CO \((D = 11.2\ \mathrm{V})\); HCl \((D = 4.3\ \mathrm{V})\), and, in addition, CO\(_2\), N\(_2\)O, H\(_2\)O.

NO \((D = 7.3\ \mathrm{V})\) destroys the afterglow; oxygen cannot be detected chemically as a product of dissociation, despite the formation of N\(_2\)O\(_3\), which is difficult to imagine without prior dissociation of NO. Likewise, upon addition of O\(_2\) \((D = 7.0\ \mathrm{V})\), although neither O, nor O\(_3\), nor NO is chemically detected, NO lines are observed in the afterglow spectrum. Therefore it must be assumed that in both these cases dissociation takes place. In view of the contradictory nature of the results, it is not possible to draw from the above data a conclusion about the energy of active nitrogen.

4. On ionization. Ionization under the action of active nitrogen was observed by Konstantinides[^9] in only one case, using the apparatus described in Chapter VI. If iodine vapor is admixed with active nitrogen, a noticeable ionization current is obtained between the electrodes. The ionization potential of iodine reaches \(10.1 \pm 0.5\ \mathrm{V}\).

Similar experiments with H\(_2\) \((18.1\ \mathrm{V})\) and mercury vapor \((10.4\ \mathrm{V})\) gave negative results. It may be thought that the deactivation energy leading to ionization lies between 10.1 and 10.4 V.

5. On the excitation of line spectra. The richest and clearest material on this question is given by the spectra of gases excited by active nitrogen.* With electrodes of Al, Pt, Cu, W, and Ni, and under a large load—

* However, here too there is the possibility, as in the case of active hydrogen, that, owing to chemical reactions of active nitrogen with the substances under investigation, lines are excited whose excitation energy is greater than the energy released upon deactivation. However, the below-mentioned results, free of contradictions, leave no room for doubts of this kind.

...the spectra of the above-mentioned metals are obtained \(^{43,44,66,85}\). Si and Ti were investigated in the form of chlorides \(^{17}\); metals with low melting points were vaporized on heating, and their vapors were added to active nitrogen \(^{68,54,51}\).

The results of numerous works, especially those of Okubo and Hamada \(^{51}\), are mainly as follows: in the vapors of Al, Cd, Zn, Mg, Na, K, Ti, and Ca the arc spectra are obtained completely, whereas the spark spectra are entirely absent*. Of the latter, the resonance lines Mg II \((2800\ \text{Å})\) require the least excitation energy, namely \(E_a = 12.0\ \mathrm{V}\); however, it is not observed. In the Zn I spectrum the term \(9D\) is excited \((E_a = 9.1\ \mathrm{V})\). In exactly the same way the lines \(1850\) and \(2062\ \text{Å}\) of the iodine atom \((E_a \simeq 8.4\ \mathrm{V})\) were observed. As for the lines of the noble gases, which should be observed at the far ultraviolet end, no investigations were carried out**.

The spectrum of mercury gives sharper limits for the energy. Okubo and Hamada succeeded in exciting all terms up to \(4D\), inclusive; higher ones, however, were not obtained by them. Whereas the spectrogram very clearly gives the triplet \(2^3P — 4^3D\) \((2652\ \text{Å},\ E_a = 9.51\ \mathrm{V})\), the triplet \(2^3P — 5^3D\) \((2483\ \text{Å},\ E_a = 9.65\ \mathrm{V})\) was not noticed even after an exposure lasting several hours***. Thus, upon deactivation, an energy of \(9.58 \pm 0.07\ \mathrm{V}\) is released; this energy is expended on the excitation of other atoms. The value obtained

* The fact that Street and Fowler \(^{68}\) did not find in the Mg I spectrum the line \(4703\ \text{Å}\) \((2P — 5D)\) is a question of intensity. Dzhovons \(^{17}\) observed individual lines of the Ti II spectrum, but did not indicate their wavelengths. The excitation of alkali-metal vapors varies strongly with pressure \(^{61,52}\).

** In the presence of oxygen, Kaplan \(^{25}\) observed in the afterglow the green line of the aurora \((E_a \simeq 3\ \mathrm{V})\), but only immediately after the current was broken.

*** Contrary to this, Reark \(^{54}\) and his collaborators consider \(6^3D\) to be the highest term; however, in their work the limits of the spectrum are not as distinct as in Okubo and Hamada.

...is in good agreement with the value found from ionization experiments.

The intensity rule, usually satisfied for multiplets, is not satisfied in the case of excitation by active nitrogen⁵⁴. The resonance lines of the alkali metals appear with quite unusual brightness⁵¹.

1. On the excitation of band spectra. Strett and Fowler observed the excitation of bands in CuCl₂, SnCl₂, J₂, HgJ, J₂, S₂, and CS₂ in various carbon compounds; Jevons¹⁷, ¹⁸—in SiCl₄, TiCl₄, and BCl₃; Ferguson¹²—in AuCl₃.* Knauss³³ investigated, in a vacuum spectrograph for ultraviolet rays (8500–1000 Å), the excitation of molecules whose energy-level scheme is known. He found excitation of the H₂ spectrum (the lowest excitation potential 11 V); in NO the δ-bands were obtained \((E_a = 0.22\ \mathrm{V})\), and in CO bands which originate from the first six vibrational quanta of level \(A\) \((E_a = 8.2—9.0\ \mathrm{V})\). Hence the limiting energy values are 9.0 and 11 V. The excitation potential of the bands of the first positive group of nitrogen emitted in afterglow is, as was already mentioned, 10.4 or \(11.5 \pm 0.3\ \mathrm{V}\).

2. Energy of the non-luminous modification. In active nitrogen, whose glow has been quenched by heating, the Na \(D\)-lines are excited, but not the yellow-green line \((E_a = 3.4\ \mathrm{V})\). Likewise, the β-bands of NO are not observed upon admixture of N₂ \((E_a = 5.5\ \mathrm{V}^{24})\).

This makes it possible to give an approximate estimate of the energy of the non-luminous modification of nitrogen: it lies between 2.1 and 4.3 V.

XI. Summary of experimental data

Active nitrogen participates in many chemical processes, often accompanied by chemiluminescence, in which inactive nitrogen does not react. It consists of

* Excitation by active nitrogen can be used as a very convenient method for studying the spectra of unstable molecules; see K. Mulliken, Phys. Rev. 26, 1, 1925.

of a large percentage of free atoms, probably also contains metastable molecules, but contains no ions at all. Deactivation occurs either, evidently, at once, and in most cases—but not necessarily—is accompanied by afterglow, or without afterglow under the action of the walls of the vessel and of the substances adsorbed on them. In this case deactivation depends to a high degree on the surface properties and temperature. The attenuation function of the afterglow and its dependence on pressure show that, in the elementary process of deactivation connected with radiation, two active and, probably, one inactive particles take part, i.e., that it proceeds in the form of a triple collision.

The spectrum of the afterglow consists exclusively of individual bands of the first positive group of nitrogen, for the excitation of which an energy of from 11.5 to \(10.4 \pm 0.3\) V is required. Substances mixed with active nitrogen are excited by it or are ionized, for which an energy of not more than 9.6 is required. The energy of the non-luminous modification is contained within the limits from 2.1 to 4.3 V.

XII. Hypotheses on the Nature of Active Nitrogen

1. N\(_3\). Proceeding from the fact of the extremely slow decay of active nitrogen and the negative temperature coefficient of this process, Trautz\(^{75}\) came to the conclusion that, although the initial elements in deactivation are nitrogen atoms, nevertheless over a very short interval of time an equilibrium process is established according to the scheme:

\[ \mathrm{N}_2 + \mathrm{N} \to \mathrm{N}, \]

and then the restoration of inactive nitrogen occurs, accompanied by luminescence:

\[ \mathrm{N}_3 + \mathrm{N} \to 2\mathrm{N}_2. \]

In addition, without luminescence there should proceed the reaction

\[ \mathrm{N} + \mathrm{N} \to \mathrm{N}_2, \]

and to a greater extent the higher the temperature. This explains the decrease in the intensity of the afterglow with increasing temperature.

With contemporary data this hypothesis is not in agreement; in particular, the pressure measurements made by Wrede86 contradict it, and it is also not confirmed by careful spectroscopic observations.

2. Metastable molecules. Long before the scheme of electronic energy levels for the nitrogen molecule became known, Shah and Shur57 proposed a hypothesis based on the assumption that the carrier of the activity is a metastable molecule. They considered the initial state of the ultraviolet Lyman–Birge–Hopfield bands (the \(a\)-state) to be metastable, and estimated its energy at \(8\,V\). This energy, according to Shah and Shur, could be transferred by collisions of the second kind to other atoms and excite them. Afterglow is only a secondary accompanying process and is not observed when the molecule, for whatever reason, returns directly to the normal state without emission of visible radiation. Knowledge of the scheme of energy levels enabled Berge* to modify this hypothesis, and he regarded the initial level of the afterglow bands as metastable, its energy being \(11.4\,V\). He gave an explanation of the role of impurities that favor afterglow: they retain free electrons, which otherwise might lead to a direct transition to the normal state. Taking into account the variety of impurity structures, this view must be considered overly bold. Of the objections raised against such a theory, we shall mention the two most essential ones: the initial state of the afterglow bands is regarded as metastable and possesses an anomalously long lifetime, which contradicts experiment; the decay should proceed “monomolecularly,” which also sharply disagrees with experiment.

3. Atomic nitrogen. The hypothesis of a triple collision. The idea that active nitrogen consists of free atoms, now proven experimentally,

* Later Berge applied it to the atomic hypothesis discussed below.

that both chemical phenomena and the phenomenon of luminescence proceed at the expense of the free energy released in the recombination of two atoms had already been proposed by Lewis^42. It was accepted also by Stretton^62, with certain reservations. Modern data on the formation of molecules in collisions do not, however, allow one to suppose that a molecule formed from two atoms receives the dissociation energy in the form of excitation energy. This difficulty was avoided in Shponer’s^55 hypothesis, according to which recombination occurs as the result of a triple collision. This hypothesis was later confirmed by measurements of quenching and by special experiments with triple collisions. The exceptional duration of the afterglow is then explained very simply—by the extraordinary rarity of such an event; and one may apparently, with full justification, knowing that the lifetime of the hydrogen atom is about 10 sec., ascribe to the nitrogen atom a lifetime of a higher order.

From the excitation potential of the afterglow bands, which begin with the eleventh quantum number of the \(B\)-state, Shponer estimated the work of dissociation at \(11.4\ \mathrm{V}\), or somewhat more, which agrees well with energy calculations for the \(X\)- and \(A\)-states. The third participant in the collision is the normal nitrogen molecule or some other atom or molecule, which absorbs the released dissociation energy and gives it up in the form of radiation. However, apparently the newly formed molecule also absorbs energy, because the afterglow does not change noticeably in the presence of an excess of noble-gas atoms, although in this case the third partner is only a foreign atom.

The triple-collision hypothesis is not in direct contradiction with any experimental fact, although it cannot explain certain facts (the existence of a non-luminous modification and the excitation of spectral lines up to \(9.6\ \mathrm{V}\)) without additional assumptions.

Serious difficulties arose only when new spectroscopic data became known, according to ...

tium the work of dissociation has a much smaller value (about 9.5 V), so that afterglow can no longer be explained as simple excitation as a result of recombination of atoms.

4. Combined hypothesis. In view of all that has been said above, Cario and Kaplan\(^{6,7,26}\) proposed a new hypothesis, which we shall call the combined one, because it assumes both the existence of metastable molecules and atoms and three-body collisions.* According to this hypothesis the initial elements are again nitrogen atoms, formed in the discharge and possessing a longer lifetime. Two atoms recombine in a triple collision with a normal nitrogen molecule; in this process the work of dissociation, numerically equal to 9.5 V, is liberated. It is quite sufficient to create the metastable state \(A\) (8 V) or to excite the third participant, a foreign atom, but is insufficient for excitation of afterglow:

\[ \mathrm{N}+\mathrm{N}+\mathrm{N}_{2}\to \mathrm{N}_{2}^{A}+\mathrm{N}_{2}. \]

If now two molecules with a large store of energy collide with one another, then the total energy of 16 V goes to the dissociation of one of them into two excited metastable atoms (\(\mathrm{N}''\)):

\[ \mathrm{N}_{2}^{A}+\mathrm{N}_{2}^{A}\to \mathrm{N}_{2}+\mathrm{N}''+\mathrm{N}''. \]

Each atom thereby receives

\[ \frac{1}{2}(16-9.5)\simeq 3.3\,\mathrm{V}. \]

In fact it turned out that the atom \(\mathrm{N}\) has two metastable states, 2.37 and 3.56 V (\(2D\) and \(2P\)); one may suppose that the latter (\(\mathrm{N}''\)), by collision or emission of energy, can pass into the former (\(\mathrm{N}'\)). The emission of afterglow, according to Cario and Kaplan, occurs in the collision of the metastable molecule \(\mathrm{N}_{2}^{A}\) and atoms (\(\mathrm{N}'\) and \(\mathrm{N}''\)):

\[ \begin{aligned} \mathrm{N}_{2}^{A}+\mathrm{N}''&\to \mathrm{N}_{2}+\mathrm{N}+E_{1},\\ \mathrm{N}_{2}^{A}+\mathrm{N}'&\to \mathrm{N}_{2}+\mathrm{N}+E_{2}. \end{aligned} \]

* Much earlier, Willey\(^{33}\) and B. Lewis\(^{39}\) had expressed the supposition that molecules and atoms in a metastable state may occur in active nitrogen.

For \(E_1\) and \(E_2\) one obtains the above-mentioned values 11.0 and 10.4 V, precisely the amount of energy required to excite the afterglow.

This hypothesis, by virtue of its complexity, embraces almost the whole complex of facts known to us. It does not contradict the measurements of decay, because the emission of \(1h\nu\) accounts, as an exact calculation shows, for the loss of two active particles, namely two atoms. The first process makes understandable the positive result of the experiments with a triple collision. The quenching of the afterglow by a weak discharge is explained by the fact that an amount of energy sufficient to ionize them (9 V) can readily be transferred to molecules, already rich in energy and metastable.

Naturally, in such a case afterglow cannot arise. For the existence of the unstable modification and for the circumstance that it arises when the temperature is raised, Kario and Kaplan also give the following explanation: the metastable molecule (state \(A\)), owing to thermal motion, receives energy of the order of 0.2 V, with which it becomes capable of passing into the neighboring state \(a\), and from it into the normal state. This in itself excludes the possibility of any processes leading to radiation. Although the equidistance of the energy levels in the states \(A\) and \(a\) (Fig. 11) makes such a transition possible, it should nevertheless be noted that the position of state \(A\) is determined inaccurately and that up to now the Werdzha-Gopfield bands have not yet been found in the afterglow spectrum.

The combined hypothesis encounters certain difficulties on the question of the temperature dependence of the afterglow and of the magnitude of the energy as determined by various methods. In particular, it cannot be said with certainty whether one may dispute the fact that the excitation and ionization energy of the foreign gases, equal to 9.6 V, is obtained from the work of dissociation, although the work of dissociation, in all probability, is smaller than this value.

References

1. Angerer E., Physik. Z., 22, 97, 1921.

2. Bay Z. und Steiner W., Z. physik. Chem., B. 3, 149, 1929.

3. Bay Z., Naturwiss., 17, 442, 1929.

4. Birge R. T., Nature (Lond.), 114, 642, 1924.

5. Bonhoeffer K. F. und Kaminsky G., Z. physik. Chem., 127, 385, 1926.

6. Cario G., Physik. Z., 1929.

7. Cario G. und Kaplan J., Physik. Z., 1929.

8. Comte F., Physik. Z., 14, 47, 1913.

9. Constantinides P. A., Physik Rev., 30, 95, 1927.

10. Crew W. H. und Hulburt E. G., Physic. Rev., 30, 124, 1927.

11. Easson L. H. und Armour R. W., Proc. roy. Soc. Edinburgh, 48, I, 1927/28.

12. Ferguson W. F. C., Physic. Rev., 31, 968, 1928.

13. Hagenbach A. und Frey W., Physik Z., 18, 544, 1917.

14. Herzberg G. Z., Physik., 46, 878, 1928.

15. — —, ibid., 49, 612, 1928.

16. — —, Nature, 122, 505, 1929.

17. Jevons W., Proc. roy. Soc., 89, 187, 1913.

18. — —, ibid., 91, 120, 1915.

19. — —, ibid., 142, 407, 1926.

20. Jonson R. C. und Jenkins H. G., Philosophic. Mag., 2, 621, 1926.

21. Kaplan J., Nature, 122, 771, 1928.

22. — —, Proc. nat. Acad. Sc. U. S. A., 14, 258, 1928.

23. — —, Physic. Rev., 31, 1126, 1928.

24. — —, ibid., 33, 189, 1929.

25. — —, ibid., 33, 154, 1929.

26. — — und Cario G., Nature, 121, 906, 1928.

27. Karrer S. C., Ravel S. und Gassen, B. V., Physic. Rev., 23, 297, 1924.

28. Kenty C. und Turner L. A., Physic. Rev., 32, 799, 1928.

29. Kichlu P. K., Proc. Ind. Assoc. Calcutta, 9, 287, 1926.

30. — — — und Acharya, D. P., Nature, 121, 482, 1928.

31. — — — —, Proc. roy. Soc., 123, 168, 1929.

32. Knauss H. P., Physic. Rev., 32, 417, 1928.

33. Kneser H. O., Ann. Physik, 87, 717, 1928.

34. — — —, Ber. dtsch. phys. Ges., 1929.

35. — — —, unpublished.

36. König A. und Elöd E., Physik. Z., 14, 165, 1913.

37. — —, Z. Elektrochem., 21, 267, 1915.

38. — — und Klinkmann G. H., Z. physik. Chem., 157, 335, 1933.

1. Lewis B., J. Amer. Chem. Soc., 50, 27, 1928.
2. —, Nature, 121, 938, 1928.
3. Lewis P., Ann. Physik., 2, 459, 1900.
4. —, Astrophys. J., 18, 258, 1903.
5. —, ibid., 20, 49, 1904.
6. —, Physik. Z., 546, 1904.
7. —, Physik. Rev., 18, 125, 1904.
8. —, Ann. Physik., 12, 1101, 1904.
9. —, Philosophic. Mag., 25, 826, 1913.
10. —, Physik. Rev., 1, 459, 1913.
11. —, Nature (Lond.) 111, 599, 1923.
12. Ludiam E. B. and Easson L. H., Nature (Lond.), 118, 500, 1925.
13. Okubo I. and Hamada H., Philosophic. Mag., 5, 372, 1928.
14. —, ibid., 7, 729, 1929.
15. Pirani M. and Lax E. W. Veröff. Siem.-Konz., 2, 203, 1922.
16. Ruark A. E., Foote P. D., Rudnick P. and Chenault K. L., J. Opt. Soc., 14, 17, 1927.
17. Rudy R., J. Franklin Inst., 202, 376, 1926.
18. —, Physic. Rev., 27, 110, 1926.
19. Saha M. N. and Sur N. K., Philosophic. Mag., 348, 421, 1924.
20. Sponer H., Z. Physik, 34, 622, 1925.
21. —, ibid., 41, 611, 1927.
    60—66. Strutt R. I., Proc. Roy. Soc., A 85, 219, 1911; 86, 56, 1912; 86, 262, 1912; 87, 179, 1912; 89, 539, 1913; 91, 303, 1915; 3, 438, 1916.
    67—69. Strutt R. I. and Fowler A., ibid., 85, 377, 1911; 86, 105, 1912; 93, 254, 1917.
22. Strutt R. I. (Lord Rayleigh), ibid., 102, 453, 1928.
23. Tiede E., Ber. dtsch. chem. Ges., 46, 340, 1913.
24. — and Domke E., ibid., 47, 420, 1914.
25. —, — and Schleede A., Naturwiss., 11, 765, 1923.
26. —, — and Tanneberger H. Diss. Tanneberger, Berlin, 1923.
27. Trautz M., Z. Elektrochem., 25, 297, 1906.
28. Trowbridge C. C., Physik. Rev., 23, 299, 1906.
29. Warberg E., Wied. Ann. (4), 10, 180, 1903.
30. Willey E. J. B., J. Soc. Chem. Industr., 43, 283, 1924.
    79—83. Willey E. J. and Rideal E. K., J. Chem. Soc., Lond. 1804, 1926; 669, 1927; 2188, 1927; 2831, 1927; 1620, 1928; 228, 1929.
31. Wortning A. C., Physik. Rev., 29, 907, 1927.
32. — and Rudy R., ibid., 23, 767, 1924.
33. Wrede E., Physik. Z. 54, 53, 1929.
34. Zenneck J., Physik. Z. 22, 102, 1921.