On the Infrared Radiation of the Night Sky*

V. I. Krasovskii

In recent years, much attention has been devoted to the investigation and interpretation of the spectra of the infrared radiation of the night sky. Additional observations have been carried out, extensive laboratory and theoretical studies have been performed, and the old material has been more thoroughly examined and used. The infrared spectrum of the night sky from 7000 to 20,000 Å has proved to be not only intense, but also rich in very interesting details. Although the nature of some of the emissions remains unclear, nevertheless even the single reliable detection of bands of molecular oxygen and hydroxyl is so significant that it is difficult to overestimate its great scientific and practical importance.

The infrared radiation of the night sky is of interest not only from an abstract scientific point of view; strange as it may seem at first glance, its study may shed light on obscure questions of great practical importance. The content of hydrogen in the upper atmosphere is very interesting from the standpoint of the cosmogonic problem concerning the exchange of terrestrial and interplanetary matter. The upper atmosphere has now also come to play an essential role in the everyday activity of humankind. Located in it are the ionospheric layers, produced by the Sun’s hard electromagnetic radiation, which provide vitally necessary radio communication and, in a number of cases, interfere with it. The physics and chemistry of the upper atmosphere and the optics of the solar ultraviolet are therefore very important from a practical point of view. Consider, for example, the question of hydrogen—the principal constituent part of water vapor. As is known, at present it is assumed that the principal ionization at heights from 50 to 100 km

*) Works connected with the first stage of research on the infrared radiation of the night sky on the basis of a new methodology are cited under numbers 1, 23–34. References to them and critical remarks may be found in 2–22.

in layer \(D\) is produced by monochromatic radiation of solar hydrogen in the \(L_{\alpha}\) line (1217 Å). Despite the insignificant relative concentration, water vapor plays a large role in the absorption of \(L_{\alpha}\) by the Earth’s atmosphere. If the ionization is indeed due to \(L_{\alpha}\) radiation, then the lower boundary of the \(D\) layer, which is very important for practice and as yet little studied, will naturally depend also on the concentration of water vapor.

Table I

\[ I=I_{0}\cdot e^{-al}, \]
where \(I_{0}\) is the intensity of the incident radiation, \(I\) is the intensity of the radiation that has passed through the medium, and \(l\) is the path traversed in cm.

Table I gives the absorption coefficients for \(L_{\alpha}\) radiation by individual constituents of the Earth’s atmosphere \(^{59,61}\). From these data it is not difficult to conclude that even at a relative concentration of water vapor of about \(10^{-4}\), absorption in it will be equivalent to absorption in the rest of the dry atmosphere. During the last 20–30 years it has repeatedly been suggested that, if the Earth’s atmosphere is mixed, then the maximum limit of the relative concentration of water vapor is determined by the saturated-vapor pressure in the region of the temperature minimum at an altitude of about 20 km, where pearly clouds do in fact sometimes appear, indicating saturated water vapor. At an altitude of 20 km there is usually a total pressure of 42 mm Hg and a temperature of 213° K, at which the pressure of saturated water vapor is \(7\cdot10^{-3}\) mm Hg \(^{7,56}\). Thus, the relative concentration of water vapor may be estimated as \(1/6000\). Thus, for example, \(^{57}\) Humphreys, as early as 1933, assumed that the relative concentration of water vapor in the upper atmosphere was on average \(1/4000\). It was precisely this assumption, in the course of explaining noctilucent clouds by the condensation of saturated water vapor into ice crystals, that enabled him to suppo-

to assume that at an altitude of about 80 km, where the atmospheric pressure is \(10^{-2}\) mm Hg, the temperature is close to \(160^\circ\) K. From the investigations of Barnett, Herndon, and Caster\(^{58}\), carried out with the aid of balloon sondes, it follows that the relative concentration of water vapor at an altitude of 32 km sometimes reaches the value \(1.5 \cdot 10^{-4}\).

However, Byram, Chubb, Friedman, and Gailar\(^{61}\), who studied the absorption of \(L_\alpha\) radiation with the aid of apparatus carried on a rocket, established that already at an altitude of 75 km the excess absorption of the atmosphere, if all of it is attributed only to water vapor, indicates that its relative concentration is no greater than \(1.6 \cdot 10^{-5}\). Rather, it is considerably smaller, if in the upper atmosphere there are significant concentrations of nitric oxide, assumed in order to explain the ionization of layer \(D\). The result of the first few investigations by Byram, Chubb, Friedman, and Gailar cannot yet serve as an indication of the regularity of low humidity. It is possible either that there is great variability in the concentration of water vapor, or that it is completely chemically destroyed at altitudes above 75 km. At present no definite answer can yet be given to this question. If below 75 km the concentration of water vapor is large, then ionization of the lower part of layer \(D\) by \(L_\alpha\) radiation will be impeded. In this case one may suspect, for example, X-radiation as the ionizing agent. On the other hand, a change in the concentration of water vapor may substantially change the intensity of ionization in the upper part of layer \(D\) as well. Finally, it is also of interest to note that the ionization of layer \(D\) is difficult to attribute to NO molecules if the excess absorption at an altitude of 75 km proves to be mainly connected with water vapor.

As will be shown below, the infrared radiation of the night sky is connected with hydrogen in the upper atmosphere. However, for full use of the material on this radiation it is first necessary to establish the altitude at which it arises. Table II compares various data for this quantity for hydroxyl radiation. The altitude of the emitting layer is determined from the change in brightness of the night-sky radiation at different zenith distances. The large scatter is explained not only by the fundamental imperfection of this method, but also by the use of broad spectral regions with radiation of different nature (including diffuse radiation), by the nonuniform surface brightness of the sky, usually with a characteristic maximum or ridge of brightness that changes and moves across the celestial dome during the night, and, finally, by the absorption by water vapor increasing toward the horizon. If narrow spectral regions are not used, then the unaccounted infrared background from cosmic objects and scattered sunlight will lead to an apparently substantial change in the altitude

...of the luminous layer. Berthier^44, to whom the newest study belongs, asserts that hydroxyl radiation originates at an altitude exceeding 100 km. However, it can hardly be considered that Berthier’s work has definitively resolved the question of the height of the layer, although it was the first to use narrow spectral intervals. A cautious approach to the evaluation of Berthier’s result is more appropriate. At present one can confidently object only to the extreme values of the heights of the emitting layers indicated in Table II.

Table II

First of all, regarding the altitude of 70 km obtained by Roach, Pettit, and Williams^46. The study by these authors was usually considered flawless, since it was assumed that they had very accurately taken into account the absorption by water vapor at large zenith distances. However, Roach, Pettit, and Williams, using a photoelement whose sensitivity fell sharply toward long wavelengths, measured only the total photocurrent from the radiation of the night sky in the interval from 9500 to 11 000 Å, and decoded the distribution of the intensities of the hydroxyl bands 9–4, 4–1, and 5–2 on the basis of the hypothesis that the true intensities of these bands are related to one another as 1:1/3:2, respectively. According to our observational data^32,33, however, this ratio, in the best case, is rather characterized as 1:1:2. Thus, Roach, Pettit, and Williams exaggerated the relative role of the 5–2 band near 10 800 Å in the region of intense absorption by water vapor and underestimated the contribution from the 4–1 band at 10 300 Å outside this region. The American authors also took into account neither the cloud structure, nor the diffuse radiation in the spectral interval under study, nor the intense bands of as yet unknown origin at 10 000 Å. Allowance for a more significant relative

intensity of the 4–1 band would undoubtedly have led to an increase in the height of the luminous layer. However, we were deprived of the possibility of making corrections, since Roach, Pettit, and Williams, in their externally circumstantial article, did not report the specific data they used on the absorption of water vapor at different zenith distances. Thus, the extremely categorical indications by many authors as to the reliability of the result of Roach, Pettit, and Williams are not especially well-founded. One can hardly follow the path taken several years ago by Barbier \(^{45}\), who proposed regarding only one height, 70 km, as correct because it best harmonizes with the ozone–hydrogen hypothesis of excitation of hydroxyl radiation, and not on the basis of any other, more convincing criterion.

As a result of the works of Shklovsky \(^{17,18}\) and Heaps and Herzberg \(^{54}\), it may be considered that in a column of the atmosphere with a base of \(1 \text{ cm}^{2}\) no fewer than \(10^{11} \div 10^{12}\) excited hydroxyl molecules arise per second. At an altitude exceeding 110–120 km, where oxygen is completely dissociated, there are no ways for radiation of such intensity to arise. One must categorically reject any connection of this radiation with the penetration into the atmosphere of corpuscular streams of extraterrestrial origin. The level of excitation of hydroxyl observed in the infrared radiation of the night sky is small. If its source were corpuscles, their energy should not be very high; otherwise, states with a higher excitation threshold would be observed in the radiation of the night sky. Even assuming the lightest corpuscle (a hydrogen atom) and the optimal probability of excitation (one radiating hydroxyl molecule per corpuscle), it cannot be admitted that more than \(10^{-12} \text{ g}\) of extraterrestrial matter falls per second on \(1 \text{ cm}^{2}\) of the earth’s surface. At altitudes exceeding 110–120 km, excited molecules can appear only as a result of recombination into a molecule of atoms in triple collisions. There are fewer than \(10^{11}\) such collisions per second in a column of the atmosphere with a base of \(1 \text{ cm}^{2}\) above an altitude of 120 km. Moreover, far from all triple collisions can lead to the appearance of excited hydroxyl molecules.

Thus, the height of the layer in which hydroxyl radiation arises cannot exceed 110–120 km. Even the value 130 km indicated by Bertier is apparently too high. Beitz’s remark \(^{9}\) should be considered correct: hydroxyl radiation cannot originate at a height of several hundred kilometers. At first, when much attention was not paid to the total number of newly formed excited hydroxyl molecules, we tried to find a mechanism allowing for the height of several hundred kilometers indicated at one time by Rodionov \(^{37,38}\). In doing so, even then it was possible to construct only a qualitative explanation—

of a qualitative character. But now this must be completely abandoned, since the invalidity of these data is obvious.

A great uncertainty also exists in the question of the height of the layer emitting the band of molecular oxygen O$_2$ (${}^1\Sigma-{}^3\Sigma$) at 8600 Å. Thus, Meinel$^{42}$ indicates 80 km, while Berthe$^{44}$ gives about 140–150 km.

The molecular bands O$_2$ and OH have a simple structure, readily detectable even with small resolving power. Therefore it is possible easily and accurately to determine the rotational temperature of the emitting molecules. This circumstance is very important primarily for determining the temperature of the upper atmosphere in the various emitting layers. Using the standard temperature distribution, one can estimate from the rotational temperature the approximate height of the emitting layer; this was done, for example, by Shklovsky$^{18}$ and Meinel$^{5}$.

For this purpose, however, one must be fully certain that the rotational temperature of the emitting molecules reflects the temperature of the surrounding medium. The case of the O$_2$ molecule (${}^1\Sigma$) raises no doubt. The mean lifetime of the O$_2$ molecule (${}^1\Sigma$) is about 7 sec$^{42}$. If such a molecule can be destroyed as a result of a chemical reaction with any atoms of the Earth’s atmosphere, then even at heights up to 150–200 km it inevitably undergoes, before disappearing, several collisions with molecules of the medium that do not react with it. So many collisions will be quite sufficient for the rotational temperature of the molecule to correspond to the ambient temperature. The question is more uncertain for hydroxyl. Shklovsky believes that the mean lifetime of OH in the vibrationally excited state, with deactivation by radiation, is considerably less than $10^{-3}$ sec$^{17,18}$. However, Heaps and Herzberg are inclined toward the value $10^{-2}$ sec$^{54}$. At present it is impossible to state what the exact value is of the mean lifetime of hydroxyl in the vibrationally excited state under deactivation by radiation. If the data of Heaps and Herzberg are correct, then the rotational temperature of hydroxyl will correspond to the temperature of the surrounding medium up to heights reaching 110–120 km, since during its existence in the excited state the molecule undergoes at least several collisions. At these heights each molecule undergoes from 3000 to 1000 double collisions per second. If, however, Shklovsky’s data are correct, then the rotational temperature of the emitting hydroxyl molecules will correspond to the temperature of the surrounding medium only at heights below 100 km. At this height and below, each molecule undergoes about 5000 or more double collisions per second$^{7}$.

The observed rotational temperature of the O$_2$ (${}^1\Sigma-{}^3\Sigma$) bands varies within very wide limits$^{5}$ from 130° to 200°K. Inter-

ON THE INFRARED RADIATION OF THE NIGHT SKY

... since in the upper atmosphere a temperature below \(160^\circ \mathrm{K}\) (Humphreys\(^{57}\)) was not even assumed. Such a low temperature as \(130^\circ \mathrm{K}\) can be associated only with the region of the temperature minimum at an altitude somewhere around 80 km. It is difficult to assume such a temperature anywhere else. On the other hand, not only a temperature of \(130^\circ \mathrm{K}\), but even \(200^\circ \mathrm{K}\), is difficult to reconcile with other known data on the upper atmosphere. Thus, according to NACA\(^{7}\), the minimum temperature of \(240^\circ \mathrm{K}\) occurs at an altitude of 80 km, but this value is too large. Average data obtained with instruments carried on rockets\(^{66}\) also indicate a temperature minimum at an altitude of about 80 km. But the mean value of the temperature at this level, according to the cited data, is \(205^\circ \mathrm{K}\). During some flights at an altitude of 80 km the temperature was \(200^\circ \mathrm{K}\). The lowest temperature, \(185^\circ \mathrm{K}\), was obtained at an altitude of about 80 km only in a few rocket experiments. Such contradictions in data obtained by different methods require further study. It is possible that this is explained by the fact that the temperature in the region of the temperature minimum is subject to substantial fluctuations, and the data compared here are not simultaneous. A characteristic feature of the \(O_2(^{1}\Sigma - {}^{3}\Sigma)\) band is its extremely variable intensity. Quite often it is so weak that it is not even distinguishable against the background of the rest of the night-sky radiation\(^{34}\); its intensity varies by hundreds and even thousands of times\(^{31}\). Meinel\(^{1}\) reports that the higher the rotational temperature of the band, the more intense it is\(^{15,42}\). All this unambiguously indicates that the reaction leading to the appearance of excited \(O_2(^{1}\Sigma)\) molecules depends strongly on the temperature of the medium. There is also a very brief report by Meinel on an enhancement of the \(O_2(^{1}\Sigma - {}^{3}\Sigma)\) radiation in the spectrum of the early dawn\(^{5}\).

According to Meinel’s observations, the rotational temperature of hydroxyl was \(260^\circ \mathrm{K}\)\(^{40,41}\). For some time it was regarded as constant, which, as already noted above, was even used for an approximate determination of the height of the emitting layer. However, quite recently Chamberlain, Norman, and Oliver\(^{47}\), who carried out investigations in more northerly regions and had a spectrograph of greater dispersion than Meinel’s, established that the rotational temperature of hydroxyl lies in the range \(300\text{--}350^\circ \mathrm{K}\). These investigators suppose that the upper atmosphere over the polar regions is warmer than over low latitudes. But it seems to us that another possibility is also possible. In polar regions, because of the absence in wintertime of ultraviolet radiation dissociating oxygen, the zone of the most effective dissociation and recombination of oxygen may be located somewhat higher than over low latitudes. Since at great heights the number of binary collisions is small, the rotational temperature

hydroxyl under these conditions may not correspond to the temperature of the surrounding medium and will be determined mainly only by the excitation conditions during the formation of new molecules. It must also be assumed that, since the zone of effective dissociation and recombination of molecules will move higher, i.e. into a medium of lower density, the number of newly formed excited molecules in triple collisions, and the intensity of radiation associated with this process, will become smaller than in lower layers. Bagaryatskii^35,36 observed in the polar regions both a certain strengthening and a certain weakening of hydroxyl radiation. Unfortunately, his material does not make it possible to judge the rotational temperature of hydroxyl. At present we still do not have sufficient observational material for the indicated conclusions. The rotational temperature of hydroxyl observed by Meinel, 260° K, corresponds to the standard NACA^7 temperature at altitudes of 40, 75, and 90 km, while the temperature of 300–350° K observed by Chamberlain, Norman, and Oliver corresponds approximately to 45, 65, and 105 km. Further, a temperature of 260° K according to mean rocket data^56 occurs at altitudes of 40, 60, and 105 km, and 300–350° K approximately at 120 km. The intensity of hydroxyl radiation over middle latitudes is not subject to such large changes as the intensity of the radiation of \(\mathrm{O}_2\) \(({}^{1}\Sigma-{}^{3}\Sigma)\), although changes by several times appear to be real. Such stability of the radiation possibly indicates that the reaction as a result of which excited hydroxyl is formed is not in strong dependence on the temperature of the surrounding medium.

The intensity of hydroxyl radiation is not correlated with the intensity of the radiation of \(\mathrm{O}_2\) \(({}^{1}\Sigma-{}^{3}\Sigma)\). As one of many examples in this direction, one may point at least to the recent work of Berthier^43. It should also be noted that there were cases when intense hydroxyl radiation occurred, but there were no signs whatever of radiation of \(\mathrm{O}_2\) \(({}^{1}\Sigma-{}^{3}\Sigma)\)^34. All this apparently permits the conclusion that the final chemical reactions leading to the appearance of excited OH and \(\mathrm{O}_2\) molecules are different.

In recent years several reactions have been indicated that are suspected of being responsible for the appearance of excited oxygen molecules in the \({}^{1}\Sigma\) state. First of all we shall point to the reaction of excited oxygen atoms in the \({}^{1}D\) state with unexcited oxygen molecules^1. This reaction leads to the formation of an unexcited oxygen atom and an excited oxygen molecule \({}^{1}\Sigma\). Here there may occur both a simple transfer of excitation during collision and, equally, replacement in the molecule of one of the atoms by an extraneous one. In the zone of most effective dissociation and recombination of oxygen at

at an altitude of about 100 km in the daytime an enormous number of atoms \(\mathrm{O}({}^{1}D)\) is formed. Indeed, any photodissociating oxygen molecule breaks up into an unexcited atom and an atom excited to this state. This reaction, however, is capable of providing intense bands \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) only in the illuminated zone, for example in the twilight spectrum. It is possible that this is precisely what explains the strengthening of such a band in the spectrum of early twilight, of which we know from Meinel’s brief indication\({}^{5}\). Meanwhile, an intense band \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) at 8600 Å has also been observed at night\({}^{34}\). Therefore it seems to us that the reaction being analyzed cannot be responsible for the radiation of \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) at night. With deactivation by radiation, the mean lifetime of \(\mathrm{O}({}^{1}D)\) is too short (96 sec.) and cannot ensure the preservation until night of the \(\mathrm{O}({}^{1}D)\) atoms formed during the day. Excitation of the \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) bands by \(\mathrm{O}({}^{1}D)\) atoms appearing at night in addition to the dissociation process must also be excluded, since the 6300 Å emission is always observed, whereas the \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) band is quite often absent. The low rotational temperature of the \(\mathrm{O}_{2}({}^{1}\Sigma-{}^{3}\Sigma)\) bands also does not favor the reaction under consideration. In the zone of intense nocturnal radiation of the red line \(\lambda 6300\) Å, apparently arising above 100 km, there is nevertheless a higher temperature than, for example, \(130^\circ\mathrm{K}\).

It might have been supposed that \(\mathrm{O}_{2}({}^{1}\Sigma)\) molecules appear as a result of the reaction of ozone with oxygen. Ozone is formed intensively wherever oxygen molecules and atoms are present simultaneously. The reaction of ozone formation in triple collisions is the most probable and effective reaction in the terrestrial atmosphere and, in particular, in the zone of maximum intensity of the processes of dissociation and recombination of oxygen. However, the long existence of ozone molecules is possible only when the content of atomic oxygen is small. Under these conditions the newly formed vibrationally excited ozone molecules, before colliding with oxygen atoms, are deactivated in double collisions with unexcited \(\mathrm{O}_{2}\) and \(\mathrm{N}_{2}\) molecules, i.e., lose the energy necessary for activation and become chemically less active from the point of view of reaction with atomic oxygen. At altitudes greater than 90–100 km, where there are very many oxygen atoms, excited ozone, possessing a reserve of activation energy, enters into reaction with them at the very first collisions with oxygen atoms. As a result of each such elementary reaction two excited oxygen molecules are formed\(*\). This process, however, is more or less

\(*\) Let us recall that excited oxygen and nitrogen molecules in the ground state cannot radiate and are capable of being deactivated only in inelastic collisions with other atoms or molecules.

uniform both by day and by night. It cannot be substantially influenced by the ambient temperature, since, let us repeat once more, the ozone molecules that are newly formed are themselves excited and, consequently, possess the store of activation energy necessary for reaction with oxygen atoms. If the emission of the bands \(O_2({}^1\Sigma-{}^3\Sigma)\) were connected with such a reaction with internal activation energy, ensuring independence from the temperature of the external medium, then it would be difficult to expect a correlation of the intensity of the emission \(O_2({}^1\Sigma-{}^3\Sigma)\) with the rotational temperature of the emitting molecules. All this prompts us to refrain from assigning any essential role to this reaction in the excitation of the emission \(O_2({}^1\Sigma-{}^3\Sigma)\). Somewhat further on it will be shown that the hydroxyl emission is most probably produced precisely at the expense of the energy released in this reaction. Consequently, if it were at the same time the cause of both the hydroxyl and the oxygen emissions, then the intensities of these emissions would have to correlate with one another, which, however, is not in fact observed.

Finally, there remains the reaction of ozone with atomic hydrogen\(^{17,18,53}\). Initially it was assumed that as a result of such a reaction hydroxyl molecules appear in the ground state with maximum vibrational excitation, while the accompanying oxygen molecules are not excited. Later, however, Shklovskii began to suppose that the oxygen molecules may be excited to the electronic states \({}^1\Delta\) and \({}^3\Sigma\). Kaplan, studying under laboratory conditions the emission spectrum of a mixture of ozone with hydrogen, confirms precisely this character of excitation of the final products\(^{52}\). Thus, the appearance of \(O_2({}^1\Sigma)\) molecules and, consequently, of the bands \(O_2({}^1\Sigma-{}^3\Sigma)\), will be accompanied by the appearance of hydroxyl molecules excited to the second vibrational level. However, emission in transitions from this state falls in the longer-wavelength infrared region of the spectrum, which so far remains inaccessible to the means of modern research. On the other hand, the appearance of \(O_2({}^1\Delta)\) molecules will be accompanied by the appearance of hydroxyl molecules excited to the fourth vibrational level. In transitions from this level the readily observed hydroxyl band \(4-1\) at \(10\,300\ \text{\AA}\) is formed. An intense band \(O_2({}^1\Delta-{}^3\Sigma)\) is assumed at \(12\,500\ \text{\AA}\)\(^{1}\). However, its reality has not yet been proved. If among the newly formed oxygen molecules a large number of \({}^1\Delta\) states appears, then one may expect a certain correlation between the intensities of the hydroxyl bands \(4-1\) and \(O_2({}^1\Sigma-{}^3\Sigma)\), provided only that there is a constant ratio between the number of newly formed \({}^1\Delta\) and \({}^1\Sigma\) states of \(O_2\). We have already reported in print that the intensity of the hydroxyl band \(4-1\) is anomalously high in comparison with the intensity

of hydroxyl band 5–2 from the higher fifth level32,33. However, our material does not allow us to draw any conclusions about a correlation with the molecular oxygen band \(({}^1\Sigma — {}^3\Sigma)\), for which no simultaneous observations were made. Moreover, we are not certain of the accuracy of the estimates of the intensity of the hydroxyl bands.

The rate of the ozone–hydrogen reaction is determined by the expression

\[ [\mathrm{O}_3]\cdot[\mathrm{H}]\cdot k_1 = n_1, \tag{1} \]

where \([\mathrm{O}_3]\) is the number of \(\mathrm{O}_3\) molecules in \(1\ \mathrm{cm}^3\), \([\mathrm{H}]\) is the number of H atoms in \(1\ \mathrm{cm}^3\), and \(n_1\) is the number of pairs of final products formed per second in \(1\ \mathrm{cm}^3\), with

\[ k_1 = 1.5\cdot 10^{-11}\cdot T^{1/2}\cdot \exp(-E/KT), \tag{2} \]

where \(R\) is the gas constant, \(T\) is the temperature of the medium, and \(E\) is the activation energy. Bates and Nicolet53 take \(E = 3\ \mathrm{kcal}\) per mole. In that case, at \(T = 130^\circ\ \mathrm{K}\), \(k_1 = 1.7\cdot 10^{-15}\ \mathrm{cm}^3\mathrm{sec}^{-1}\), while at \(T = 200^\circ\ \mathrm{K}\), \(k_1 = 1.2\cdot 10^{-13}\ \mathrm{cm}^3\mathrm{sec}^{-1}\). Thus, a change in temperature alone from 130 to \(200^\circ\ \mathrm{K}\) can cause a seventyfold change in the intensity of the radiation. Consequently, this reaction is capable of explaining well the dependence of the band intensities on their rotational temperature. However, an additional change in their intensity may also be caused by a change in the concentrations of \(\mathrm{O}_3\) and H.

Let us try to establish the approximate productivity of the ozone–hydrogen reaction. As was already indicated1, the maximum intensity of the \(\mathrm{O}_2({}^1\Sigma — {}^3\Sigma)\) bands may exceed by a factor of one thousand the intensity of the red line of night-sky emission, estimated as \(\sim 10^8\) quanta per second in an atmospheric column with a base of \(1\ \mathrm{cm}^2\). Consequently, in an atmospheric column with a base of \(1\ \mathrm{cm}^2\), no fewer than \(10^{11}\) oxygen molecules in the \({}^1\Sigma\) state must appear per second. Taking into account the possibility of deactivation without radiation and of excitation of other \(\mathrm{O}_2\) states besides \({}^1\Sigma\), this value must be increased. As an example, let us confine ourselves to a value greater by one order of magnitude. Further, suppose that a layer \(1\ \mathrm{km}\) thick emits. Then in \(1\ \mathrm{cm}^3\) about \(10^7\) new \(\mathrm{O}_2\) molecules will appear. Since on average the rotational temperature of the \(\mathrm{O}_2({}^1\Sigma — {}^3\Sigma)\) band at \(8600\ \text{Å}\) is equal to \(150^\circ\ \mathrm{K}\) \((k_1 = 8.3\cdot 10^{-15}\ \mathrm{cm}^3\mathrm{sec}^{-1})\), expression (1) makes it possible to determine the product of the concentrations \([\mathrm{O}_3]\cdot[\mathrm{H}]\). It turns out to be \(1.2\cdot 10^{21}\ \mathrm{cm}^{-6}\). Let us assume that the concentrations \([\mathrm{O}_3]\) and \([\mathrm{H}]\) are equal. Then in the zone of emission of the oxygen bands there will be contained, in \(1\ \mathrm{cm}^3\), \(3.5\cdot 10^{10}\) oxygen molecules and hydrogen atoms. Since the consumption of the initial products is very great \((10^7\ \mathrm{sec}^{-1}\mathrm{cm}^{-3})\), the indicated concentrations will be sufficient to maintain the emission at a constant level for no more than one hour. This is the most favorable case.

If, however, the concentration of one of the initial products is smaller, then the reaction time constant will also be reduced in comparison with what it was at equal concentrations of O$_3$ and H. Thus, the ozone–hydrogen reaction is in principle capable of explaining the flash-like character of the O$_2$ (${}^1\Sigma-{}^3\Sigma$) radiation by changes in the concentrations of O$_3$ and H. The concentrations \([{\rm O}_3]\) and \([{\rm H}]\sim 3.5\cdot 10^{10}\ {\rm cm}^{-3}\), cited as an example, do not appear impossible. Since at an altitude of 80 km about \(5\cdot 10^{14}\) molecules·cm\(^{-3}\) are contained, the concentration \([{\rm H}]\sim 3.5\cdot 10^{10}\ {\rm cm}^{-3}\) corresponds to a relative concentration of about \(10^{-4}\). If atomic hydrogen is formed in the upper atmosphere by the destruction of water vapor, the relative concentration of which, on the basis of the fact of the existence of noctilucent clouds, is estimated as being of the order of \(10^{-4}\), then one may suppose that practically complete destruction of water vapor takes place already at an altitude of 80 km. Such a conclusion, in itself, is so far based on very arbitrary estimates and, apparently, is of some interest only in the light of the results of the work of Byram, Chubb, Friedman, and Taylor\(^ {61}\), who found that at an altitude of 75 km the relative concentration of water vapor does not exceed \(10^{-5}\).

The ozone–hydrogen reaction, proposed at first only to explain the excitation of hydroxyl, proved more acceptable for the radiation of O$_2$ (${}^1\Sigma-{}^3\Sigma$). However, the absence of a correlation between the radiation of O$_2$ (${}^1\Sigma-{}^3\Sigma$) and OH, as already noted, excludes this reaction as the cause of the principal excitation of OH. The radiation of O$_2$ (${}^1\Sigma-{}^3\Sigma$) can serve as a sensitive indicator of changes in the product \([{\rm O}_3]\cdot[{\rm H}]\) in the zone of the temperature minimum at an altitude of about 80 km, provided only that such radiation is due to the ozone–hydrogen reaction. The result will be unambiguous, since the temperature of the medium is known from the indicated band structure. It seems probable that at the beginning of a flash of O$_2$ (${}^1\Sigma-{}^3\Sigma$) radiation the effective reaction zone embraces a wide range of altitudes around the temperature minimum. Therefore, at the beginning of the flash a higher rotational temperature is observed. Subsequently, however, as the concentrations \([{\rm O}_3]\) and \([{\rm H}]\) rapidly decrease at the hotter boundaries of the effective reaction zone, a substantial value of \([{\rm O}_3]\cdot[{\rm H}]\) is preserved only near the temperature minimum at an altitude of about 80 km. At these moments the effective reaction zone covers only the regions of the temperature minimum, and the O$_2$ (${}^1\Sigma-{}^3\Sigma$) bands therefore reveal a low rotational temperature.

When the ozone–hydrogen reaction was proposed to explain the maximum excitation of hydroxyl only up to the 9th vibrational level, a significant error was allowed,

namely: the activation energy of this reaction was overlooked, taking it into account would make the excitation of the 10th vibrational level of hydroxyl inevitable[^31]. Indeed, the excitation energies of the 9th and 10th vibrational levels of hydroxyl in the ground state are equal to 26,187 and 28,238 cm\(^{-1}\), respectively. The heat effect of the ozone–hydrogen reaction according to Bates and Nicolet is 27,307 cm\(^{-1}\), and its activation energy is 1,050 cm\(^{-1}\). Thus, the limiting excitation of hydroxyl is 28,357 cm\(^{-1}\), instead of 27,307 cm\(^{-1}\) if the activation energy is not taken into account. Shklovskii[^18], apparently seeking to remove this misunderstanding, arbitrarily took the energy of the ozone–hydrogen reaction to be 805 cm\(^{-1}\) (2.3 kilocalories per mole, or 0.1 ev) instead of 1,050 cm\(^{-1}\) (3 kilocalories per mole, or 0.13 ev) as in Bates and Nicolet[^53]. However, Shklovskii does not deny the arbitrariness of this assumption. He himself writes on p. 57 in [^18]: “This value is apparently too low.” It is enough to increase the activation energy by only 126 cm\(^{-1}\), compared with the value adopted by Shklovskii, in order to reach excitation of the 10th vibrational level.

The discovery in the emission spectrum of the night sky of rotational-vibrational bands of hydroxyl stimulated the reproduction of this radiation under laboratory conditions. The aim of such studies was also to obtain more accurate molecular constants of hydroxyl. The spectrum of an oxyacetylene flame in oxygen, first described by Dieke, Crosswhite, and Hager[^50], proved to be the most successful. In this flame the masking action of water vapor turned out to be insignificant, and at the boundary of the visible region it was possible to observe the fine structure of the rotational-vibrational bands of hydroxyl. Hornbeck and Herman[^51] established that laboratory spectra of hydroxyl contain very interesting features. It turned out that in them, just as in the radiation of the night sky, there are no hydroxyl bands that would indicate excitation exceeding the 9th vibrational level. Such excitation cannot be associated with the ozone–hydrogen reaction. In an oxyacetylene flame in oxygen one should not expect any appreciable concentrations of ozone. Another cause is needed to explain such limiting excitation of hydroxyl up to the 9th vibrational level. We pointed out earlier[^1],[^33] that hydroxyl excited to levels exceeding the 9th becomes very chemically active in reaction with unexcited O\(_2\) and N\(_2\) molecules. The final products of such reactions will be hydrogen atoms and O\(_3\) and N\(_2\)O molecules. We suppose that hydroxyl with excitation exceeding the 9th vibrational level is rapidly destroyed, without having time to be deactivated by radiation, which explains the absence of bands from transitions from higher levels. The estimate we made shows that the destruction of strongly excited

deactivated hydroxyl molecules can be effective only at altitudes below 100 km and, undoubtedly, in a laboratory oxy-acetylene flame. With increasing altitude such deactivation will decrease because of the reduction in the total number of collisions with O₂ and N₂.

To explain hydroxyl radiation, in addition to the ozone–hydrogen reaction, we have also proposed the reaction of vibrationally excited oxygen molecules in the ground state with hydrogen atoms¹, ²⁹, ³³. As already reported above, vibrationally excited oxygen molecules are continuously formed in the zone of the most effective dissociation and recombination of oxygen. This process must be more or less constant and regular. O₂ molecules in the ground state, whose vibrational excitation falls on the 28th vibrational level or exceeds it (the excitation energy of this level is about 95 kilocalories per mole), become chemically active with respect to O₂ and N₂.*) As a result, such O₂ molecules cannot exist for a long time, whereas the same molecules with excitation at the 27th vibrational level and below are chemically stable with respect to unexcited O₂ and N₂. Consequently, the lifetime and concentration of such excited O₂ molecules increase, and the less the vibrational excitation, the greater the increase. An O₂ molecule in the ground state at the 27th vibrational level, upon collision with a hydrogen atom, enters into reaction with it. As a result, excited hydroxyl is formed in the ground state at the 9th vibrational level and an unexcited oxygen atom. Less excited oxygen molecules produce excited hydroxyl molecules on vibrational levels below the ninth. Thus the actually observed¹⁷, ¹⁸, ²⁹ increase in the population of the vibrational levels of hydroxyl with decreasing quantum number is well explained. In this way, the limiting excitation of hydroxyl up to the 9th vibrational level, used originally as a very effective argument in favor of the ozone–hydrogen hypothesis of hydroxyl excitation, is now not a convincing or unambiguous proof of this process.

The absence of exact values of the constants of elementary chemical processes and of deactivation of excited states introduces

*) Previously we somewhat underestimated the activation energy in the case of N₂ and made a distinction between O₂ and N₂. A more accurate value of the activation energy for reaction with N₂ (14–28 kilocalories per mole⁶⁴) gives no grounds for distinguishing between O₂ and N₂, as was indicated¹, ²⁹, ³³. Such an assumption of two limiting thresholds of hydroxyl excitation was based on Meinel’s apparently inaccurate borrowed values of the intensities of individual bands.

some uncertainty and makes an exhaustive quantitative interpretation of the infrared radiation of the upper atmosphere difficult. Nevertheless, some qualitative estimates appear quite possible.

For simplicity we shall assume that vibrationally excited oxygen molecules in the ground state are practically deactivated only as a result of ordinary inelastic collisions, in which the vibrational energy is converted mainly into heat. As is known, vibrationally excited molecules in the ground state, in each single effective collision, lose practically only that energy which corresponds to the transition to the next lower vibrational level. The probability of such deactivation may be represented (see, for example, ^39) as

\[ \gamma_{v\to v-1}\sim \gamma_{1\to 0}\cdot v, \tag{3} \]

where \(\gamma\) is the probability of the transition indicated by the subscript, in which \(v\) is the quantum number of the vibrational level. This dependence has been verified experimentally only for small \(v\). Its application to large \(v\) is an approximation requiring further careful theoretical and experimental verification. The total number of collisions \(\nu\) necessary for the transition of a vibrationally excited molecule from a higher level \(v'\) to a lower level \(v''\) is determined by the expression

\[ \nu \sim \frac{1}{\gamma_{1\to 0}\cdot v'} +\frac{1}{\gamma_{1\to 0}\cdot (v'-1)} +\cdots+ \frac{1}{\gamma_{1\to 0}\cdot (v''+2)} +\frac{1}{\gamma_{1\to 0}\cdot (v''+1)} = \]
\[ = \frac{1}{\gamma_{1\to 0}}\cdot \left( \frac{1}{v'} +\frac{1}{v'-1} +\cdots+ \frac{1}{v''+2} +\frac{1}{v''+1} \right). \tag{4} \]

The time \(\tau\) necessary for the transition of a molecule from level \(v'\) to level \(v''\) will be equal to

\[ \tau=\frac{\nu}{\nu_2}, \tag{5} \]

where \(\nu_2\) is the total number of binary collisions per second undergone by each molecule in the zone of the process under consideration. The total concentration of excited molecules \([\mathrm{O}'_2]\) on levels from \(v'\) through \(v''+1\), inclusive, may therefore in this case be represented as

\[ [\mathrm{O}'_2]=n\cdot \tau = \]

\[ = \frac{n}{\nu_2\cdot \gamma_{1\to 0}}\cdot \left( \frac{1}{v'} +\frac{1}{v''-1} +\cdots+ \frac{1}{v''+2} +\frac{1}{v''+1} \right), \tag{6} \]

where \(n\) is the number of newly formed excited \(O_2\) molecules at the level \(v'\) per second in \(1\ \mathrm{cm}^3\).

The deactivation just described of vibrationally excited molecules, however, becomes secondary for strongly vibrationally excited \(O_2'\) molecules, since, as already indicated above, they readily enter into reaction with unexcited \(O_2\) and \(N_2\) molecules. Apparently, the probability of such reactions is such that approximately no less than every tenth collision with \(O_2\) or \(N_2\) proves effective. Therefore one may assume that the order of magnitude of the mean lifetime of a vibrationally excited oxygen molecule at any level, beginning with the 28th and higher, will be equal to

\[ \tau_{v \geq 28} \sim \frac{10}{\nu_2}. \tag{7} \]

If it is assumed that the newly formed oxygen molecules originate in equal numbers at all vibrational levels, then the ratio of the concentration of molecules at the 27th level to the concentration of molecules at the 28th level will be equal to

\[ \frac{[O'_2(v=27)]}{[O'_2(v=28)]} = \frac{\tau_{v=27}}{\tau_{v=28}} \sim \frac{1}{270\cdot \gamma_{1\to 0}} . \tag{8} \]

This means that, in the case \(\gamma_{1\to 0}\sim 10^{-4}\), the population of the 27th level exceeds the population of the 28th level by several tens of times. It is possible that \(\gamma_{1\to 0}\sim 10^{-4}\) is not very different from the true value of this quantity for molecular oxygen in the ground state. We do not know experimental data concerning this quantity for \(O_2\). But Kondrat’ev and El’yashevich \(^{39,70}\) find for \(N_2\) that \(\gamma_{1\to 0}\sim 10^{-4}\). The application of dependence (3) to vibrational levels with larger \(v\) is more risky.

At the present time we are not yet able to indicate the distribution of the initial vibrational excitations over the various levels of newly formed oxygen molecules in the upper atmosphere. For the possibility of some estimates, let us assume that practically all excited oxygen molecules originate at the 27th vibrational level of the ground state, while the lower levels are populated as a result of deactivation by collisions*). Thus, on the basis of expression (6) one may attempt

*) The same assumption was made by Shklovskii, Hyps, and Herzberg when they analyzed the distribution of excitations as a result of the ozone–hydrogen reaction \(^{13,54}\). If new molecules appear with a smaller vibrational excitation, then the population of the lower levels will turn out to be greater than in the case considered here.

estimate the approximate concentration of excited molecules \(O'_2\), beginning with the 27th and ending with the fourth vibrational level inclusive. Because of insufficient energy, oxygen molecules excited at the third and lower levels cannot enter into reaction with hydrogen atoms*). Let us take, for example, the altitude \(102\ \mathrm{km}\) [55], at which solar ultraviolet radiation dissociates more than \(10^7\) molecules of \(O_2\) per second in \(1\ \mathrm{cm}^3\). Under equilibrium conditions, there too, on the average, no fewer than \(10^7\) new molecules of \(O_2\) must be formed per second in \(1\ \mathrm{cm}^3\). Then the concentration of \(O'_2\) molecules vibrationally excited from the 27th through the fourth level inclusive will prove to be equal to \(5\cdot 10^7\ \mathrm{cm}^{-3}\)**). This indicates that their relative concentration at an altitude of \(102\ \mathrm{km}\) is close to \(4\cdot 10^{-6}\).

The rate of reaction of excited oxygen molecules with hydrogen atoms is determined by the expression

\[ [O'_2]\cdot [H]\cdot k_2 = n_2, \tag{9} \]

where \([O'_2]\) and \([H]\) are the concentrations of excited \(O'_2\) molecules and hydrogen atoms, respectively. Owing to the absence of any need for extraneous activation energy [48],

\[ k_2 = 4.25\cdot 10^{-12}\cdot T^{1/2}, \tag{10} \]

where \(T\) is the temperature of the medium in which the reaction proceeds; since the \(O'_2\) molecules are strongly excited, the exponential factor is absent (it is equal to unity). Finally, \(n_2\) is the number of \(O'_2\) molecules in the excited state newly formed per second in \(1\ \mathrm{cm}^3\).

Let us suppose, for example, that at an altitude of \(102\ \mathrm{km}\), \(T=260^\circ\mathrm{K}\) (as in Meinel), and \([H]=3\cdot 10^8\ \mathrm{cm}^{-3}\), which corresponds approximately to a relative concentration of \(H\) equal to \(2\cdot 10^{-5}\). It seems to us that such a relative concentration of atomic hydrogen at an altitude of \(102\ \mathrm{km}\) is not absurd. If atomic hydrogen is regarded as the product of complete dissociation of water vapor penetrating upward from the lower layers, then a similar relative concentration of water vapor would occur at the moment of formation of silvery clouds from saturated water vapor in the temperature-minimum zone at an altitude of about \(80\ \mathrm{km}\), at a temperature of about \(150^\circ\mathrm{K}\). Indeed, a saturated-water-vapor pressure equal to \(10^{-7}\ \mathrm{mm}\) Hg corresponds to

*) The thermal effect of the endothermic reaction \(O_2+H\to O+OH\) is equal to \(-17\) kilocalories per mole.

**) We assume that at an altitude of \(102\ \mathrm{km}\), \(\nu_2\sim 4\cdot 10^3\ \mathrm{sec}^{-1}\), and the total concentration of molecules is \(\sim 1.3\cdot 10^{13}\ \mathrm{cm}^{-3}\); \(\gamma_1\to 0\) is taken equal to \(10^{-4}\).

at a temperature of \(153^\circ\) K. Consequently, at an altitude of about 80 km, where the atmospheric pressure is \(10^{-2}\) mm Hg, the relative concentration of \(\mathrm{H_2O}\) is \(10^{-5}\). Thus we find for \(n_2\) the value \(10^6\ \mathrm{sec}^{-1}\ \mathrm{cm}^{-3}\). Even in the least favorable case, when the thickness of the reaction zone does not exceed 1 km, about \(10^{11}\) excited hydroxyl molecules per second will appear in an atmospheric column with a base of \(1\ \mathrm{cm}^2\). Such an estimate is in good agreement with what is observed in reality and, therefore, does not exclude the hypothesis we have proposed for the origin of hydroxyl radiation.

In one of his papers, Bates\(^9\), quite correctly pointing out the impossibility of the appearance of hydroxyl radiation at the altitude of several hundred kilometers indicated by Rodionov, passes over in complete silence our remark that the ozone–hydrogen reaction, taking into account the activation energy reported by Bates and Nicolet, ensures limiting excitation of hydroxyl to the 10th, and not to the 9th, vibrational level. Continuing to regard the limiting excitation of hydroxyl to the 9th vibrational level as unconditional proof of the ozone–hydrogen reaction, Bates at the same time substantially distorts the new hypothesis of hydroxyl excitation proposed by us. Thus, Bates asserts that we proposed the following series of reactions to explain hydroxyl radiation:

\[ \begin{aligned} &1)\quad \mathrm{N_2 + O + M \to N_2O + M},\\ &2)\quad \mathrm{N_2O + O \to N_2 + O_2'}, \end{aligned} \]

and

\[ 3)\quad \mathrm{O_2' + H \to O + OH'} \]

(the prime denotes an excited state). However, in our work\(^ {29}\) on p. 336 it is written: “According to the views generally accepted up to the present time, atomic oxygen could recombine into molecules only as a result of triple collisions with the participation of a pair of atoms or of an atom and an oxygen molecule. As a result of the latter combination ozone is formed, which in a bimolecular reaction with ozone or atmospheric oxygen gives rise to molecular oxygen. One may indicate still another path for the recombination of atomic oxygen into molecules. At first, in triple collisions, nitrous oxide is formed. Then it enters into a bimolecular reaction with nitrous oxide itself or with atomic oxygen, the products of which are neutral nitrogen molecules and excited oxygen molecules. In the final analysis it is immaterial by what path the oxygen molecules appear. What is essential is only that they are produced in excited states....” “We explained the excitation of hydroxyl by the quenching of metastable states of molecular and atomic oxygen.”

Here are possible quenching reactions, considerably more effective than the ozone–hydrogen reaction:

\[ \text{1) } \mathrm{H_2 + O' \rightleftarrows H + OH'}, \quad \text{2) } \mathrm{O'_2 + H \rightleftarrows O + OH'}. \]

The activation energy of these reactions is small, since excited molecules and atoms take part in them, whose chemical activity is an elementary generally known fact. By stressing only nitrogen oxides and ignoring the excited states, Bates in effect proposes an entirely new sequence of reactions, arbitrarily ascribing it to me. One could have ended the remarks on Bates’s paper here, were it not for the fact that the note to it mentions yet another circumstance of substantial importance for our entire hypothesis. It goes without saying that deactivation by collisions of vibrational excitations of molecules in the ground state cannot at present be characterized by precise data. Thus, for example, Hips and Herzberg suppose that a vibrationally excited OH \(({}^2\Pi)\) molecule is capable of withstanding up to \(10^6\) collisions without being deactivated \(^{54}\). For the \(\mathrm{O_2}({}^3\Sigma)\) molecule we use smaller, but nevertheless somewhat arbitrary, values for the numbers of deactivating collisions (from \(10^2\) to \(10^4\)); this is undoubtedly, for the time being, the weakest point of our hypothesis. Bates, however, considers deactivation by collisions to be so rapid that he attaches no practical significance to it. He supports his opinion by a reference to Massey’s work (p. 248). But the reference to this work is a misunderstanding, since it deals with deactivation of vibrational excitation at higher electronic states of the molecule, and not in their ground state. The effectiveness of such deactivation is generally known (it is, for example, taken into account in \(^{39}\)), and it cannot be transferred to the ground states of molecules, where deactivation by collisions is usually considered less effective.

The sequence of reactions ascribed to me by Bates is mentioned, incidentally, with an indication of my authorship, also by Mitra in \(^{7}\) on p. 540. However, as a categorical objection to these reactions, Mitra advances the inconsistency of these reactions with the altitude of \(70\) km at which, allegedly, the layer of emitting hydroxyl is located. As was reported above, such an altitude, indicated by Roach, Pettit, and Williams, now raises serious doubts. Somewhat earlier the same sequence of reactions was ascribed to me in the abstract in \(^{63}\), which distorts our work.

The small (several-fold) observed fluctuation in the intensity of hydroxyl radiation \(^{1,40,41}\) is difficult to explain by a change in the temperature of the medium. The rate of reaction (9) is proportional only to the square root of this temperature, which substantially

the change of which has not yet been observed. The processes of dissociation, recombination of oxygen, and excitation of hydroxyl actually observed do not allow one to suppose substantial changes in the concentrations of the initial products. Thus, for example, at an altitude of 102 km there are\(^7\) about \(10^{12}\) oxygen atoms in \(1\ \mathrm{cm}^3\), while their consumption was estimated above as a quantity close to \(10^7\ \mathrm{sec}^{-1}\ \mathrm{cm}^{-3}\). A substantial change in the concentration of the initial products, for example under night-time conditions in the absence of photodissociation of oxygen, could occur only in \(10^5\ \mathrm{sec}\), which is much longer than the residence time of the upper atmosphere in the unilluminated state. Fluctuations in the intensity of hydroxyl radiation may be explained either by a change in the density of the atmosphere as a result of a change in pressure, or by a fluctuation in the concentration of atomic hydrogen, or by some other impurities effectively quenching the excited states of molecular oxygen.

If all the constants of the elementary processes were known exactly, then it would not be difficult, from the radiation of hydroxyl in the infrared region of the spectrum, to estimate the presumed changes in the composition and density of the upper atmosphere.

According to Bates and Nicolet\(^{53,67}\), water photodissociates into hydroxyl and atomic hydrogen under radiation shorter than 1300 Å. This radiation, according to data obtained with rockets, penetrates downward to an altitude of 95 km\(^{66}\). However, for certain narrower parts of the spectrum the depth of penetration is considerably greater. Thus, for example, \(L_\alpha(\lambda\,1217\text{ Å})\) radiation penetrates to an altitude of 70–75 km\(^{61,67}\). Later we indicated\(^{31}\) that water may be destroyed by excited oxygen atoms \(\mathrm{O}({}^{1}D)\), always formed in the direct photodissociation of molecular oxygen, beginning from an altitude of 80 km\(^{60}\) and above. If the concentration of atomic oxygen is greater than the concentration of atomic hydrogen, then hydroxyl will enter mainly into reaction with atomic oxygen, the final products of which are newly regenerated atomic hydrogen and molecular oxygen. Thus, hydroxyl exists as an intermediate product in the chain of reactions leading to the combination of oxygen atoms into molecules. Reactions involving atomic hydrogen are certain partial branchings of the process of oxygen recombination in the upper atmosphere. It is quite possible that the upper boundary of noctilucent clouds, located at an altitude of about 80 km, is connected not with an increase of temperature upward, beginning from this level, but with the complete chemical destruction of water vapor in the upper atmosphere, as has already been mentioned here several times. Dissociation of water vapor by itself can create conditions for an increase of temperature above the level of about 80 km. The temperature of the upper atmosphere is determined by the balance of absorp-

ON THE INFRARED RADIATION OF THE NIGHT SKY

... and the emitted energy. The main components of the atmosphere (atoms and molecules of oxygen and nitrogen) play absolutely no role in the removal of heat. Impurities of polyatomic or asymmetric molecules can be active radiators. In the process of the destruction of water, one of the very effective coolers of the upper atmosphere is eliminated. The zone of atomic hydrogen may extend even below the temperature minimum at an altitude of 80 km. In this case noctilucent clouds would be explicable as the result of mixing of the atmosphere near the temperature minimum, or somewhat below it. With such mixing, large masses of water vapor, sufficient for the formation of icy noctilucent clouds, could penetrate into the region of low temperatures. It is very interesting that noctilucent clouds are always in a state of very intense turbulent motion. It is also quite possible that, as a result of mixing in the upper atmosphere, large masses of ozone and atomic hydrogen may come into contact, and this cannot fail to affect the intensity of the radiation, quite accessible to observation, of molecular oxygen in the state \({}^{1}\Sigma\), presumed to be the result of the ozone–hydrogen reaction.

For exact quantitative judgments about the elementary processes of the upper atmosphere it is necessary to know the total intensity of the radiation of hydroxyl and molecular oxygen over the entire infrared region of the spectrum. Shklovskii \(^{17,18}\), and later Heaps and Herzberg \(^{54}\), attempted to calculate theoretically the intensities of the infrared hydroxyl bands not yet amenable to observation, from the intensities of bands in the near infrared region of the spectrum. However, the results of Shklovskii and of Heaps and Herzberg do not agree. Apparently, at present it is possible to predict only the order of magnitude of the quantity of interest to us.

Therefore a direct investigation of the whole infrared region of the spectrum is of great interest. Jones and Gush \(^{14}\) published data on the emission spectrum of the night sky up to \(2\mu\). A lead-sulfide photoconductive cell was used as the receiver of the infrared radiation. It was shown that intense infrared radiation of the night sky is indeed present in those places where hydroxyl bands are presumed. A satisfactory agreement is indicated between the intensities, both observed and calculated by Heaps and Herzberg. The resolving power (300 Å per mm) did not make it possible to discern precisely the structure and intensity of the separate elements of the radiation and to establish their purely hydroxyl nature. It seems to us that the published spectrogram does not exclude, for example, the existence near 12 500 Å, in addition to hydroxyl radiation, of radiation of \(\mathrm{O}_2({}^{1}\Delta - {}^{3}\Sigma)\). Jones and Gush also report a discrepancy of their spectrum with ours in the region near 10 000 Å and explain this by the low res-

resolving power of their apparatus, as well as by the superposition of infrared radiation of auroras. However, according to Bagaryatsky’s data, in the spectra of auroras near 10,000 Å there is rather a weakening than an increase in intensity, and in any case there is no indication of large intensities not associated with hydroxyl radiation. The work of Jones and Gush has not yet added anything new to our knowledge of hydroxyl radiation, but it is very important in that it shows new technical possibilities for investigating the long-wave infrared radiation of the night sky. An increase in the resolving power of the apparatus will make it possible to obtain all the necessary information about this phenomenon.

Attempts have been made to detect the electronic spectrum of hydroxyl in the ultraviolet region of the radiation of the night sky. Goetz and Nicolet^49 apparently found, near 3000 Å, very weak hydroxyl bands \(^{2}\Sigma \to {}^{2}\Pi\). The electronic level \(^{2}\Sigma\) has a higher excitation energy than the vibrational levels of the ground state associated with the infrared radiation of hydroxyl. The transition \(^{2}\Sigma \to {}^{2}\Pi\) is not forbidden; therefore the very weak observed radiation \(^{2}\Sigma \to {}^{2}\Pi\) testifies to a negligible population of the \(^{2}\Sigma\) level and, consequently, to the energetic insignificance of the processes leading to such excitation of hydroxyl. Gunaert and Nicolet in^68 attempt to substantiate the presence of weak rotational-vibrational OH bands in the blue and ultraviolet regions of the radiation of the night sky; in this connection they even indicate bands from transitions from the 10th, 11th, and 12th vibrational levels. All this, however, requires more careful verification.

Thus, in conclusion of our review, one may express the supposition that the principal radiation of the upper atmosphere, called the infrared radiation of the night sky, is connected with recombination into molecules of oxygen atoms formed during the day as a result of the process of photodissociation. New unexcited oxygen molecules appear as the result of a series of elementary intermediate reactions, in which the hydrogen atom plays an essential role. Some excited intermediate and final products are deactivated by radiation that is easily observed in the near infrared region of the spectrum. Thus the most effectively deactivated intermediate product is hydroxyl. To a lesser degree this also applies to the final product—molecular oxygen in the \(^{1}\Sigma\) state.

The role of atomic hydrogen in the cooling of the upper atmosphere in the zone of effective dissociation and recombination of oxygen cannot be overestimated. As has already been reported in print^1, the power of the hydroxyl radiation of the atmosphere may reach 10 ergs per sec—

a second in a column with a base of \(1\ \mathrm{cm}^2\), while the total reserve of thermal energy above the level of \(100\ \mathrm{km}\) in the same volume is about \(5\cdot 10^4\ \mathrm{erg}\). Without atomic hydrogen there would be no hydroxyl emission of the upper atmosphere. In that case neither the radiation regime nor the thermal regime in the atmospheric region responsible for this emission could change.

Of considerable interest is the possibility that this region actually coincides with layer \(E\) of the ionosphere. If we are allowed, in conclusion, to define somewhat the expected, but as yet lacking, exact factual material on the height of the emitting layer, then one may tentatively suppose that ionization in layer \(E\) is due to some extent also to vibrationally excited oxygen molecules, whose ionization energy is lower than that of unexcited molecules. We estimated above the relative concentration of excited oxygen molecules as a value close to \(5\cdot 10^{-6}\). Thus, in the region of layer \(E\) a concentration \([\mathrm{O}'_2]\sim 5\cdot 10^7\ \mathrm{cm}^{-3}\) is possible.

In conclusion it should be noted that much still remains unknown, and among what is known there are many ambiguities and inaccuracies. Therefore further study of the infrared radiation of the night sky is necessary. At present there are very effective means for its investigation. However, everything cannot be exhausted merely by accumulating observational material. At the same time it is necessary to develop a theory of the elementary processes, and also to determine their constants under laboratory conditions. The successful development of studies of the radiation of the night sky gives grounds to hope that in the very near future more extensive and accurate material will be obtained, which will make it possible to eliminate the existing uncertainties and to obtain much additional valuable information about the upper atmosphere.

ADDENDUM

In one of the recent reviews of work on the study of the night sky, the question of selective excitation of the zero vibrational level of the state \({}^{1}\Sigma\) of the oxygen molecule is again discussed\(^ {3a}\). The high population of this level is usually attributed to deactivation of the vibrational excitation \(\mathrm{O}_2({}^{1}\Sigma)\) in collisions. As is known, this deactivation is especially intense for excited electronic states, as is indicated, for example, in \(3^{34,62}\) (work \(^{63}\) is the source to which Bates refers in \(^{9}\)). Meinel doubts this mechanism, since such deactivation would supposedly be inefficient in the case of \(\mathrm{N}_2(A\,{}^{3}\Sigma)\) and \(\mathrm{O}_2(F\,{}^{3}\Sigma)\). In \(^{1}\) such a contradiction was explained by high chemical activity, leading to the very rapid disappearance of \(\mathrm{N}_2(A\,{}^{3}\Sigma)\) and \(\mathrm{O}_2(F\,{}^{3}\Sigma)\) in reactions with O and N atoms respectively, of which there are apparently very many at great heights in the region

of the origin of the Vegard–Kaplan and Herzberg bands. It is assumed here that excited molecules \(O_2(^{1}\Sigma)\) appear in lower layers, where the concentration of H, N, or O atoms is insignificant and where, as a consequence, the \(O_2(^{1}\Sigma)\) molecules with high vibrational excitation can effectively lose this excitation in collisions with oxygen and nitrogen molecules. Such indications of the low altitude of the region of origin of the \(O_2(^{1}\Sigma - {}^{3}\Sigma)\) bands are an additional argument in favor of the assumption that this radiation appears as a result of the ozone–hydrogen reaction.

References Cited

1. V. I. Krasovskii, UFN 47, No. 4, 493 (1952).
2. J. Dufay, Compt. Rend. 231, No. 25, 1531 (1950).
3. J. Dufay, Ann. de géophys. 7, No. 1, 1 (1951); Pub. Observ. Lyon 5, ser. 1 (1951).
   3a. W. W. Kellogg, Trans. Am. Geoph. Un. 34, No. 1, 115 (1953).
4. J. Cabannes et J. Dufay, Journ. phys. et rad. 12, No. 3, 95 (1951).
5. A. B. Meinel, Phys. Soc. Rep. Prog. Phys. 14, 127 (1951).
6. Penndorf, Landolt-Börnstein, Zahlenwerte und Funktionen 3, 777 (1952).
7. S. K. Mitra, The Upper Atmosphere, 2nd ed., pp. 509 and 540, 1952.
8. A. B. Meinel and P. Swings, The Atmospheres of the Earth and Planets, Revised Edition, G. P. Kuiper, p. 209, 1952.
9. D. R. Bates, Ann. de géophys. 8, No. 2, 202 (1952).
10. J. Dufay, J.-H. Bigay et Berthier, Compt. Rend. 235, No. 2, 120 (1952).
11. J. Texereau, L’Astronomie 66, 349 (1952).
12. J. Dufay, Sky and Telescope 12, No. 2, 41 (XII, 1952); Journ. Brit. Astron. Assoc. 63, No. 4, 158 (1953).
13. R. Pastiels, Ciel et Terre 59, No. 1–2, 30 (1953); 69, No. 11–12, 289 (1953).
14. A. V. Jones and H. Guch, Nature 172, No. 4376, 496 (1953).
15. I. S. Shklovskii (editor of the Russian translation), Atmospheres of the Earth and Planets. Collection of articles edited by D. P. Kuiper. IL, 1951.
16. I. S. Shklovskii, DAN 75, 371 (1950).
17. I. S. Shklovskii, DAN 75, 789 (1950).
18. I. S. Shklovskii, Izv. Crimean Astrophys. Obs. 7, 34 (1951).
19. Bulletin of the Academy of Sciences of the USSR, No. 4, 114 (1952).
20. S. I. Isaev, Auroras, Publishing House of the Chief Directorate of the Northern Sea Route (1952).
21. P. Kulikovskii, AZh 29, No. 6, 630 and 742 (1952).
22. P. Kulikovskii, AZh 30, No. 4, 469 (1953).
23. V. I. Krasovskii, DAN 66, No. 1 (1949).
24. A. A. Kalinyak, V. I. Krasovskii and V. B. Nikonov, DAN 66, No. 1 (1949).

1. V. I. Krasovskii, DAN 70, No. 6 (1950).
2. V. I. Krasovskii, Izv. Krymsk. astrofiz. obs. 5, 100 (1950).
3. V. I. Krasovskii, DAN 73, No. 4 (1950).
4. A. A. Kalinyak, V. I. Krasovskii and V. B. Nikonov, Izv. Krymsk. astrofiz. obs. 6, 119 (1950).
5. V. I. Krasovskii, DAN 77, No. 3 (1951).
6. V. I. Krasovskii, DAN 78, No. 3 (1951).
7. V. I. Krasovskii, DAN 78, No. 4 (1951).
8. V. G. Lukashenya and V. I. Krasovskii, DAN 79, No. 2 (1951).
9. V. I. Krasovskii and V. G. Lukashenya, DAN 80, No. 5 (1951).
10. V. G. Lukashenya and V. I. Krasovskii, DAN 81, No. 5 (1951).
11. B. A. Bagaryatskii and M. I. Mordukhovich, DAN 82, No. 1 (1952).
12. B. A. Bagaryatskii, V. I. Krasovskii and M. I. Mordukhovich, DAN 82, No. 4 (1952).
13. S. F. Rodionov et al., DAN 73, No. 1 (1950).
14. S. F. Rodionov, Izv. AN SSSR 14, No. 3, 247 (1950).
15. V. N. Kondrat’ev, Elementary Chemical Processes (1936).
16. A. B. Meinel, Astrophys. Journ. 111, 555 (1950).
17. A. B. Meinel, Astrophys. Journ. 112, 120 (1950).
18. A. B. Meinel, Astrophys. Journ. 112, 464 (1950).
19. P. Berthier, Compt. Rend. 236, No. 18, 1808 (1953).
20. P. Berthier, Compt. Rend. 237, No. 16, 928 (1953).
21. D. Barbier, Mém. Soc. Roy. Sci. Liège 12, I–II, 43 (1952).
22. F. E. Roach, H. Pettit and D. R. Williams, Journ. Geophys. Res. 55, No. 2, 183 (1950).
23. J. W. Chamberlain and N. J. Oliver, Phys. Rev. 90, No. 6, 118 (1953).
24. A. B. Nalbandyan and V. V. Voevodskii, The Mechanism of Oxidation and Combustion of Hydrogen, Publishing House of the Academy of Sciences of the USSR (1949).
25. P. Götz and M. Nicolet, Journ. Geophys. Res. 56, No. 4, 557 (1951).
26. G. Déjardin et R. Falgon, Compt. Rend. 234, No. 2, 200 (1952).
27. G. A. Hornbeck and R. C. Herman, Journ. Chem. Phys. 19, No. 4, 512 (1951).
28. J. Kaplan, Mém. Soc. Roy. Sci. Liège 12, I–II, 295 (1952).
29. R. Bates and M. Nicolet, Journ. Geophys. Res. 55, No. 3, 301 (1950).
30. H. S. Heaps and G. Herzberg, Zeits. f. Physik 133, No. 1/2, 48 (1952).
31. R. Penndorf, Journ. of Meteorology 7, No. 3, 243 (1950).
32. Rocket Panel, Phys. Rev. 88, No. 5, 1027 (1952).
33. W. J. Humphreys, Monthly Weather Review 61, No. 8, 228 (1933).
34. E. W. Barnett, L. R. Herndon and H. L. Caster, Tellus. 2, 302 (1950).
35. W. M. Preston, Phys. Rev. 57, No. 10, 887 (1940).
36. H. Friedman, S. W. Lichtman and E. T. Byram, Phys. Rev. 83, No. 5, 1025 (1951).

1. E. T. Byram, T. Chubb, H. Friedman and N. Geilar, Phys. Rev. 91, No. 5, 1278 (1953).
2. H. S. W. Massey, Phys. Soc. Rep. Progr. Phys. 12, 248 (1948).
3. Phys. Abstr. 54A, 6644 (1951).
4. H. S. Johnston, Journ. Chem. Phys. 19, No. 6, 663 (1951).
5. Masaaki Huruhata, Report of Ionosphere Research in Japan 4, No. 3, 137 (1950).
6. H. Newell and J. Siry, Journ. of the Amer. Rocket Soc. 23, No. 1, 7 (1953).
7. D. R. Bates and M. Nicolet, PASP 62, No. 366 (1950).
8. J. Hunaerts et M. Nicolet, Ciel et Terre 66, No. 9—10, 213 (1950).
9. V. N. Kondrat’ev, UFN 14, No. 8, 982 (1934).
10. M. A. El’yashevich, ZhETF 2, No. 1, 59 (1932).