Rocket Studies of Atmospheric Composition at High Altitudes

B. A. Mirtov

1. Introduction

Among the various studies of the upper layers of the atmosphere, the determination of their gaseous composition occupies a special place. It is no accident, therefore, that in addition to persistent “direct” studies (which already go back more than a hundred years) of the composition, a number of “indirect” methods are also being employed, based, for example, on the study of the glow of the night sky, auroras, the ozonosphere, and so forth. In research practice these observations occupy an important place, in accordance with the importance of the problem of the composition of the upper layers, which, besides its independent significance, plays a primary role in solving a whole series of other problems of the upper atmosphere that at first sight are not connected with it.

Let us turn to several examples.

1. The problem of measuring pressure at the present time cannot be accurately solved without knowledge of the composition of the atmosphere, since the readings of manometers used on rockets depend to a high degree on the nature of the gas surrounding them.

2. The problem of calculating temperature at great heights likewise cannot be satisfactorily solved when the gaseous composition of those layers in which it is determined is unknown. Usually, because direct measurements of temperature are impossible, it is calculated from the barometric formula, which uniquely relates temperature to pressure. The initial materials are pressure data, which are not determined with complete accuracy. In addition, the value of the molecular (atomic) weight of the surrounding gas, which remains unknown, again enters into the barometric formula itself.

3. The problem of the ionosphere is still full of blank spots. These blank spots include our knowledge of its composition, i.e., of the nature of the ionized particles. This knowledge is extremely essential, since it may provide real help in solving a whole series of tasks connected both with questions of the origin and existence of the ionosphere and with questions of determining some of its electrical parameters.

4. The problem of finding the level of gravitational separation of gases, i.e., the level below which the atmosphere remains mixed and above which a regular redistribution of the gaseous composition with height occurs (a relative increase in the fraction of light gases with height), is of primary importance for solving a whole series of the most important theoretical questions concerning the air ocean as a whole. It can be exhaustively solved by studying the composition of the air at various heights. The solution of this problem will make possible, for example, the construction

of the general model of the Earth’s atmosphere, about which at present we have rather vague notions.

1. The problem of the dissipation of the atmosphere into outer space (the problem of helium, for example) is also solved by knowledge of the gaseous composition in the upper layers.

2. The problem of the corpuscular radiation of the Sun can be considerably clarified if we possess reliable knowledge of the gaseous composition of the atmosphere in the zone of polar auroras.

3. Along with the indicated problems of a theoretical character, knowledge of the composition of the upper layers is also of purely practical interest. Thus, from the standpoint of the motion of flying vehicles (artificial satellites, rockets, etc.) moving at high speed and for a sufficiently long time in rarefied layers, the question of in what gas this motion takes place—in a heavy or a light one—becomes far from indifferent.

4. For flying vehicles, certain regions of the atmosphere with chemically active molecules may present a direct danger. It is well known, for example, that oxygen at altitudes of about 100 km, under the influence of the ultraviolet radiation of the Sun, is dissociated into atoms. Upon recombination of these atoms on some surface (rocket, satellite), the large quantity of heat released may, under certain conditions, lead to undesirable consequences.

Although the variety of problems enumerated is not exhaustive, it is quite sufficient to make evident the necessity of a detailed and comprehensive investigation of the composition of the atmosphere at great altitudes.

However, despite the importance of the problem of the composition of the upper atmosphere, researchers still do not have reliable information in this field. Theoretical calculations diverge sharply from the experimental data obtained. Thus, for example, the calculations of Morris¹, Epstein², Mitra and Rakshit³ for the level of gravitational separation give altitudes greater than 100 km, whereas the experiments—admittedly few in number—(Paneth⁴) indicate that the level is situated much lower.

It is therefore necessary to obtain additional reliable experimental material that would make it possible to clarify this question not only qualitatively but also quantitatively.

2. DIRECT INVESTIGATIONS

Direct investigations of the composition of the Earth’s atmosphere began very long ago (at the beginning of the nineteenth century), and investigations of the composition of the Earth’s atmosphere at different altitudes have been carried out for almost as long. The impetus for conducting studies of the composition of the atmosphere at different altitudes was Dalton’s establishment, in 1802, of the law of partial pressures. A consequence of this law should have been the fact that, in the gravitational field of the Earth, the percentage content of light gases in the atmosphere must increase with altitude, as a result of which the composition of the atmosphere at various altitudes should be different. The investigations of the last century and a half were devoted to finding the so-called level of gravitational separation of gases in the atmosphere. Initially these investigations were undertaken with the aim of verifying Dalton’s law, and subsequently they acquired an independent, profound geophysical significance.

The methodology for studying the composition of the air at great altitudes, which over 150 years has gone a long way in its improvement, consisted ...

that, with the aid of various lifting devices (free balloons, pilot balloons, stratospheric balloons), special balloons were raised into the high layers of the atmosphere and took air samples. The air was then analyzed. The maximum altitude reached in these investigations was 25–29 km (Regener, 1936). However, in all such ascents the boundary of gravitational separation was not found, i.e., the composition of the atmosphere at all accessible altitudes proved to be one and the same. This indicated complete mixing of the atmosphere up to an altitude of 25–30 km, and in order to find the level of separation it was necessary to ascend further into higher layers of the atmosphere. But the existing methods of ascent had already almost completely exhausted their capabilities: the rapid decrease in air density with height limited investigations of this kind. To solve the posed problem—finding the level of gravitational separation—it was necessary to penetrate to altitudes of the order of 80–100 km or more*). The possibility of carrying out work at such great altitudes appeared only in 1946, when the first scientific research rocket was launched.

Recently, to determine the composition of air at great altitudes, in addition to the “classical” method of taking samples, a method of analyzing the air directly in flight has begun to be used. For this purpose, a radio-frequency mass spectrometer of the Bennett type is employed 5,6. The decisive advantage of this method should be considered the possibility of determining chemically active, unstable constituents of the upper atmosphere (O, OH, NO, N, etc.) that arise under the action of short-wave ultraviolet radiation**). These constituents, absent in the lower layers of the atmosphere, cannot be detected in a sampling balloon, since as a result of recombination and dissociation processes they quickly cease to exist in a closed vessel.

At present, the radio-frequency mass-spectrometer method is still little mastered, and the foreign data from one or two flights of such an instrument are far from beyond reproach. The present article has as its aim to acquaint the reader with the work of Soviet investigators carried out by the “classical” method of taking air samples in balloons, with subsequent analysis of the air obtained in a ground laboratory. It is precisely to this that all the further exposition is devoted.

3. DIFFICULTIES OF THE EXPERIMENT AND THE SPECIFICS OF ROCKET INVESTIGATIONS

In the early investigations of the composition of air, when samples were taken with the aid of free balloons from relatively low altitudes and, consequently, contained a relatively large quantity of gas, the questions of analyzing these samples and the questions of the technique of capture did not present any special difficulties. The matter changed fundamentally when samples began to be obtained from enormous altitudes by means of rockets. In these new experiments a whole series of difficulties at once became apparent, placing the experimenter in a very difficult position. First, with increasing altitude of the investigation, the quantity of gas brought back in the balloons began to fall rapidly in accordance with the decrease in the density of the surrounding medium***). Samples obtained from an altitude of 18 km brought in liter balloons only 50 cm³

*) Theoretical calculations indicated this altitude.

**) On the application of a radio-frequency mass spectrometer to determine the ionic composition of the ionosphere, see the article by B. A. Mirtov and V. G. Istomin, “Investigation of the Ionic Composition of the Ionized Layers of the Atmosphere,” included in the present collection.

***) Thus, from the surface of the Earth to an altitude of 100 km the density decreases by 6–7 orders of magnitude!

gas, samples from an altitude of about 30 km—9 cm³, from an altitude of 60 km—0.2 cm³, and from an altitude of about 100 km—only 2–3 mm³ of gas in all*). Secondly, unlike the earlier experiments, air samples have to be taken on a rocket whose speed may exceed the speed of sound severalfold. In this connection it should be noted that the small part of the rocket’s internal cavity that is set aside for the placement of scientific apparatus is, generally speaking, a rather peculiar and highly inconvenient “laboratory” for an experiment. This “laboratory” imposes such requirements on the instruments placed in it as, for example, low inertia; high vibration resistance while the rocket engine is operating; and great precision in the timing of all automatic mechanisms. Moreover, it should not be forgotten that the taking of a sample occurs during the rocket’s free flight, i.e., at a time when the rocket, with its engine switched off, is flying by inertia in the Earth’s gravitational field. In this case all mechanisms have to operate under conditions of “weightlessness,” which presents the investigators with additional difficulties.

It is quite natural that the conditions mentioned compelled a substantial reconsideration both of the technique of the collection itself and of the technique for analyzing the sample obtained. The problems of “gettering” and of the leak-tightness of the apparatus used, sorption and desorption of gas by the walls of the cylinders used for taking air samples, and the problems of “gettering” by the body of the rocket itself came to the fore. This last question is extremely important for the experiment on taking air samples. Indeed, on entering the rarefied layers of the atmosphere, the rocket begins to release a mass of “parasitic” gases: the air with which it had been filled before launch, products of fuel combustion, vapors of unconsumed propellant, etc. Owing to the large pressure difference outside the rocket and inside it, all these gases tend to burst outward; and since the free paths of molecules at these altitudes are still not sufficiently large**), the “parasitic” gas does not have time to disperse, surrounds the rocket with a kind of rarefied “cloud,” and, unless appropriate precautions are taken, can ruin the entire experiment.

In rocket investigations of the composition of the air, the questions of microanalysis and ultramicroanalysis of gas mixtures also came to the fore, since the portions of gas subjected to analysis do not exceed several cubic millimeters.

The specific character of rocket investigations of the composition of the air at great altitudes divided the whole experiment especially clearly into three parts: a) obtaining the sample, b) storing the sample, c) analyzing the sample. Each of these parts acquired a quite independent significance.

4. OBTAINING THE SAMPLE

Unlike similar foreign work, where the taking of air samples is carried out directly on the rocket itself, in our work a special automatic container is used for these purposes, which at a certain point of the trajectory separates from the rocket and then descends by parachute. Thanks to this we are able to carry out all measurements and take air samples away from the rocket, in a relatively undisturbed atmosphere, which, of course, is a significant—

*) The volume of gas is reduced to normal conditions ($P = 750$ mm Hg, $T = 15^\circ$C).

**) At an altitude of 100 km the free paths of molecules are of the order of 10 cm.

ous advantage over the foreign methodology. In particular, with our methodology the release of the cylinders that take the sample from the gas “cloud” surrounding the rocket is ensured.

Container. The very design of the container (Fig. 1) is made so as, as far as possible, to reduce gas evolution from its internal volumes. The greatest danger in this case is presented by all sorts of “pockets,” into which near-surface air can be carried into the upper layer and there, slowly seeping through slots and small openings, will contaminate the surrounding space. To avoid this, the container is divided into two parts: a reliably sealed part, where the power sources and program-command mechanisms are located, and a part freely washed by the surrounding air, where the cylinders for taking air samples are placed. To reduce gasification in the openwork compartment where the cylinders are housed, apart from metal, glass, and porcelain, only a very slight use of vacuum rubber is permitted, in the form of thin rubber gaskets between the metal and the glass of the cylinders. Before flight, the entire container is carefully washed in order to minimize the contaminating influence of the instrument itself on the surrounding medium.

Fig. 1. Automatic container by means of which the composition of the air at high altitudes is investigated. In the upper, openwork part, four cylinders for taking samples are placed; below is a sealed compartment with control and power-supply units.

Cylinders. For taking air samples, glass cylinders of two volumes are used: \(400\ \mathrm{cm}^3\) and \(3000\ \mathrm{cm}^3\), equipped with special valves (Fig. 2a).

Fig. 2a. General view of the cylinders used for taking air samples: large cylinder of \(3000\ \mathrm{cm}^3\) capacity; small cylinder of \(400\ \mathrm{cm}^3\) capacity.

Glass cylinders were chosen for reasons of the greatest vacuum reliability with respect to degassing, and also for the best storage of such gases as oxygen, nitrogen, and argon*). During prelimina-

*) In foreign works\(^7\) metal cylinders are used. This is due to the fact that only inert gases are analyzed: argon, neon, and helium. The latter is poorly stored in glass. In addition, the saving of glass cylinders presents considerable difficulties.

...of preliminary preparation, the flasks undergo a special vacuum treatment, which consists in the following: the flask, washed with alcohol, is thoroughly pumped out on a high-vacuum apparatus to pressures of the order of \(10^{-6}\) mm Hg, with simultaneous heating of the glass to \(300\text{--}400^\circ\)C. Thanks to this operation, moisture, always present on the surface of the glass, is removed from the inner surface of the flask, as are various occluded gases, which could subsequently distort the results of the sample analysis.

Fig. 26. Schematic representation of a glass flask used for taking air samples. The primary evacuation of flask 3 is carried out through branch 5. For secondary (control) evacuation, the flask is soldered into the apparatus by means of “tab” 4. At location 6, the tab is cut off after the control evacuation. “Tabs” 1 and 7 serve for soldering the flask into the apparatus for analysis of the collected sample; 2 is the electric furnace of the valve.

The prepared flask is usually not sent into flight immediately; it may lie for a considerable time (months) before being used. Therefore, immediately before launch the flask must again be evacuated, since, owing to desorption of gas by the glass, a certain amount of “parasitic” gas may be released into the flask during storage*). Such repeated evacuation is performed immediately before the experiment itself (1–2 days before it), and only after this may the flask be sent into flight. The adopted method of preparing the flask makes it possible to carry out a preliminary check of the tightness of the flask (valve) immediately before the experiment itself: if, during the control opening in the flask mentioned above, an increased pressure is found (\(10^{-4}\) mm Hg or more), then the flask is rejected and is not allowed to be used.

Valve. The most critical part of the flask is the valve, which must, at a strictly defined moment in time, open the flask and then reliably seal the opened inlet.

Taking into account that the portion of gas entering the flask is measured in cubic millimeters, or even in hundredths of these, the valve must meet a number of stringent requirements:

1. Before and after taking the sample, the valve must be completely tight. Naturally, the slightest leak, because of the small size of the sample itself, will nullify the entire experiment.

2. The valve itself and all processes occurring during its operation must not release “parasitic” gases which, like a leak, can nullify the results of the analysis. In addition, the lubricant (if any is present) must not enter into chemical reactions with the gases of the sample.

*) For such repeated (control) evacuation, a spare “tab” is made on the flask, through which the evacuated flask can be soldered to the apparatus and opened without air entering inside (Fig. 26).

1. The valve must have a large through-opening and a short inlet tube. Otherwise, in those regions of the atmosphere where the mean free paths of molecules become larger than the dimensions of the inlet openings (the Knudsen regime), forced separation of gases according to their atomic weights may begin: light gases will penetrate into the cylinder faster than heavy ones. This effect may create a false impression of gas separation in the free atmosphere.

2. The valve must operate automatically, be as simple and compact as possible, and not draw much power from the supply batteries. The use of glass cylinders naturally implied the use of glass valves, since metal plugs could nullify all the advantages of glass *). The creation of such a valve proved to be a rather complicated matter. To solve this problem, many physicotechnical principles were tried; here are some of them:

3. An ordinary vacuum stopcock with vacuum grease.

4. Melting the glass inlet tube (with sealing) by means of an electric furnace.
5. Melting, by an electric furnace, a tube with a ground-glass stopper.
6. A valve using silver chloride.
7. A valve using optical contact.
8. A stopcock using picein (with melting).
9. A mercury valve.
10. A vacuum stopcock using ceresin (with an inlet tube).
11. A vacuum stopcock using ceresin (without an inlet tube).

The listed valves, with the exception of the last three, did not give positive results. All valves involving the melting of glass had to be abandoned, since when glass is melted many “parasitic” gases are released from it (CO₂, O₂, CO, etc.). These gases spoil the sample. The same may be said of silver chloride, picein, and ordinary vacuum grease. The latter is generally unsuitable for long-term storage of a cylinder with a greased stopcock. Valves with a mercury seal, as well as a valve with ceresin, proved to be suitable in principle. The reliability of the air samples obtained depends to a considerable extent on the quality of the valve; and since most of the results were obtained with the aid of the last of the valves mentioned, it seems necessary to dwell at least briefly on a description of its construction.

The valve is an ordinary vacuum stopcock with a through-opening of the order of \(1 \text{ cm}^2\) and with an improved quality of grinding. Ceresin is used as the grease for the stopcock. Ceresin (mountain wax—a mixture of solid high-polymer hydrocarbons of the paraffin series, \(C_nH_{2n+2}\)) in the cold state is a solid body, melting, depending on the fraction, at \(60\text{–}100^\circ\text{C}\). Our work has shown that ceresin distilled in vacuum, both in the solid and in the molten state, does not release “parasitic” gases in such quantities as could affect the result of the analysis of the sample. At the same time, owing to its chemical inertness, it does not enter into any appreciable reaction with the main components of the sample (\(O_2\), Ar, \(N_2\)).

The mechanical properties of ceresin are also extremely valuable: in the solid state it possesses great mechanical strength; in the liquid—

*) In American studies, as we have already indicated, metal cylinders with metal inlet tubes are used. These tubes are first opened with special knives and then sealed hermetically, sealing the inlet into the cylinder. As a result of the use of metal in the cylinders, oxygen is often not detected.⁸

very low viscosity coefficient. Owing to this, the valve plug turns very easily on the molten picein and becomes “seized” as soon as the picein cools. If the plug is well ground, the cooled picein forms an impeccable vacuum seal.

Fig. 3. Valve for a glass flask. 1 — tube with “dash,” 2 — serving for hanging the flask in the installation for analysis; 3 — “stopper-melter”; 4 — “stopper” wire, which burns out when voltage is applied to it; 5 — spiral springs turning the plug; 6 — metal clamp for fastening the springs and “stoppers”; 7 — molybdenum leads for the nichrome electric furnace 13, wound on the valve coupling 8; 9 — glass jacket surrounding the furnace; 11 — passage through the jacket to the valve plug; 10 — inlet opening of the valve; 14 — flask.

The construction of the valve itself is quite simple. It is shown in Fig. 3. The valve is turned by two spiral springs. On the valve coupling (8) there is mounted a nichrome spiral (13)—a furnace, which is insulated from the surrounding space by a glass jacket (9). In addition, the shutter is provided with a “stopper,” which holds the valve in the required position and then releases it*). Before activation the shutter is in the “cocked” state, i.e., the plug closes the inlet opening and sits firmly on the cold picein, and the springs are under tension. When the shutter operates, the furnace melts the picein and frees the plug, which, turning through a quarter of a revolution, opens the inlet opening and is held in this position by the stopper for 10 seconds. During this time the surrounding air enters the evacuated flask and the pressure in the flask becomes equal to that of the surroundings. Then the stopper holding the plug burns out, and the plug, turning further, again closes the inlet opening of the valve, while the cooled picein reliably seals it.

The switching on and off of the heating furnace, as well as of the shutter release mechanisms, is carried out by a special distributor mechanism at a strictly specified time. Owing to this, the height at which the sample is taken is always known**).

5. STORAGE OF THE SAMPLE

In carrying out our experiments on the investigation of the composition of air, an average of 2–5 days elapses between the moment the sample is taken and its analysis. Although this interval is relatively small***), nevertheless, because of the small quantity of gas obtained, there is danger of distortion of the initial composition of the sample. Since the pressure in the flask with the sample is measured in ten-thousandths of a millimeter of mercury, the free paths of the molecules become comparable with the dimensions of the vessel, and the overwhelming majority of the gas molecules undergo continual collisions with the walls of the flask. Under these conditions the phenomena

*) The “stopper” is a short wire which holds the valve plug during its rotation. By melting the wire with an electric current at the required time, the valve plug can be freed for further rotation.

**) The height at which the sample is taken is also recorded by means of a signal lamp located in the instrument (sealed) compartment of the container. The moment of the flash is recorded by a camera.

***) In American work this time varies from 1.5 to 18 months⁹.

sorption are expressed most strongly and can entail a noticeable change in the gas composition of the sample. At the same time, the phenomena of desorption of the “parasitic” gas by the walls of the balloon are facilitated here, owing to which the sample obtained may also change its initial composition.

As is well known, sorption phenomena are determined by the surface surrounding the gas. Therefore, if we strive to preserve the sample in its initial composition, it is necessary to reduce as much as possible the surface with which the gas is in contact. This can be done if, immediately after obtaining the sample, it is transferred from the balloon into a narrow capillary and compressed there, say, to atmospheric pressure. In this case,

Fig. 4. Capillary for storing a sample. 1 — gas sample, compressed by a column of mercury 2; 3 — capillary channel; 4 — constriction, near which a column of mercury remains when the capillary is melted off (in this case the capillary may be in a vertical position); 5 — place where the capillary is soldered to the apparatus.

as is easy to see, the harmful surface of the glass in contact with the gas of the sample can be reduced by more than 10,000 times*). In this case the air sample will be under considerably more favorable conditions than in the balloon. Owing to the immeasurably smaller glass surface and the higher pressure in the capillary, all desorption phenomena will also be considerably reduced. Figure 4 shows a sample of air enclosed in a narrow capillary and compressed to atmospheric pressure by a mercury “plug.” Experiments carried out by us have shown that for 2–3 years a sample kept in such a capillary with a diameter of 0.4 mm does not change its composition (the exception is helium, which diffuses through glass).

6. ANALYSIS

There are many methods for analyzing gas mixtures, but all of them become quite unsuitable when one has to deal with such small quantities of gas as are obtained in samples from great heights. Therefore, for high-altitude investigations of the composition of air (80 km and higher) it was necessary to create a new technique of gas microanalysis capable of operating with fractions of a cubic millimeter of gas mixture**).

At present, abroad, for the indicated purposes, a physicochemical method of analysis is used, which consists in the following: on the one hand, some of the components are removed by chemical reactions; on the other hand, the remaining components (nitrogen, inert gases) are separated by fractional sorption on charcoal cooled with liquid air. The quantity of one gas or another is judged from the residual pressure in the apparatus, which is carefully measured before and after each operation. This method of analysis, proposed as early as 1912 by Lenard10, was developed for the purposes of studying the composition of air by Paneth and Glückauf11 in 1946. Owing to the high accuracy of the measurements (tenths of a percent of the quantity being measured), this method is applicable when several

*) A gas that, at a pressure of \(10^{-3}\) mm Hg, fills a volume of 500 cm³ will occupy, at atmospheric pressure, a volume of 0.5 mm³.

**) Increasing the volume of the balloons will not help to increase the quantity of gas obtained in the samples, since, in order to obtain an appreciable increase in gas, it would be necessary to increase this volume by 1–2 orders of magnitude (from 3 to 30 or 300 liters), which is completely unrealistic.

cubic millimeters of gas and, consequently, provides for the investigation of altitudes of the order of 80–85 km. At the same time, it is hardly suitable for analyzing an air sample from higher altitudes. In our work we took a different path, using a spectral method for the analysis of small portions of gases. The method of quantitative spectral analysis of gas mixtures, developed in the Soviet Union by S. E. Frisch,^12 at the first stage made it possible to analyze 2–3 mm³ of air (under normal conditions).

Fig. 5. Vacuum apparatus for spectral microanalysis of gases. 1 — fore-vacuum cylinder; 2 — Lempior mercury pump; 3 — trap with liquid nitrogen; 4 — V-shaped stopcock; 5 — McLeod manometer; 6 — fork with various gases and standard mixtures; 7 — gas mixer for preparing mixtures; 8 — cylinder with a sample, sealed into the apparatus (see Fig. 3); 9 — lens; 10 — spectrograph; 11 — shortened mercury lifters connected with the fore-vacuum; 12 — mercury stopcocks for admitting gas into the apparatus.

After some improvement of the technique, it became possible to analyze tenths and even hundredths of a cubic millimeter of a gas mixture, which should make it possible to analyze samples from altitudes of the order of 130–140 km. The method of spectral analysis of gases gives, on average, a relative error in measuring the amount of a given gas in a mixture of about 5–6%. It is precisely with this accuracy that one can, for example, use this method to solve one of the central problems of the upper atmosphere—the problem of gravitational separation of gases.

To solve this problem it is necessary to know (at the altitudes under investigation) the quantitative ratio of the masses of at least two gases: a light one and a heavy one. In our experiments, spectral analysis of air makes it possible to analyze the obtained sample for three gases: oxygen, nitrogen, and argon. The separation effect should be most pronounced for the nitrogen–argon pair, since the difference in molecular weights is greatest here. Therefore, the main attention is devoted to the results of the analysis of precisely this pair.* As for oxygen, the small difference in molecular weights of oxygen and nitrogen does not produce noticeable gravitational separation of these gases at altitudes of 80–100 km. In addition, because of the high chemical activity of oxygen, the results of quantitative analysis in samples cannot always correspond to its true values in the free atmosphere.

* Foreign researchers analyze the ratios nitrogen–helium, nitrogen–neon, nitrogen–argon; however, they have been able to do this only up to altitudes of 80–85 km.

Analysis apparatus. For the analysis of air samples we use an ordinary glass vacuum apparatus, shown in Fig. 5. The apparatus contains metallic mercury; therefore, to obtain a vacuum of the order of \(10^{-6}\) mm Hg, traps with liquid nitrogen are used.

The high requirements imposed on the purity of the experiment do not permit the use of greased stopcocks in the high-vacuum part of the apparatus. The necessary shutoffs are accomplished by mercury U-shaped seals.

Where it is necessary to admit small portions of gas into the apparatus (cylinders with various gases), special mercury seals[^13] are installed, as shown in Fig. 6.

Fig. 6. Mercury seal for admitting gas into the apparatus.
\(3\)—capillary; \(4\)—piece of porcelain fused into the capillary; \(5\)—glass cup with an iron core fused into it; \(6\)—opening for the passage of gas; \(7\)—mercury closing the entrance to the capillary; \(1\)—glass tube; \(2\)—glass partition. The seal admits gas only when the porcelain of capillary \(4\) protrudes from the mercury (with cup \(5\) raised).

Fig. 7. Part of the vacuum apparatus directly involved in the analysis of the sample.
\(1\)—capillary analyzer, shown in Fig. 5; \(2\)—external electrodes of the high frequency generator. A discharge is excited between them inside the capillary; \(3\) and \(4\)—firing pin and safety device for breaking off the upper “tip” \(5\); \(6\)—neck of the cylinder; \(7\)—the cylinder itself; \(A\)—mercury level for the initial raising of the mercury before breaking off the lower “tip”; \(B\)—mercury level before breaking off the upper “tip.”

The central part of the apparatus is the capillary analyzer and the cylinder itself, in which the sample is contained; this part of the apparatus is indicated by a dashed line in Fig. 5 and is shown in greater detail in Fig. 7.

The principal difficulty of the analysis, as we have already mentioned, consists in the fact that extremely small quantities of gas have to be analyzed. The gas, enclos—

contained in the cylinder must not be admitted into the multichannel installation, where it would be sorbed on the large surface of the freshly degassed glass. Therefore the system of connecting tubes shown in Fig. 7 is constructed so that it is possible to transfer the gas from the cylinder directly into the capillary analyzer, bypassing all the auxiliary channels. For this purpose the cylinder is sealed into the installation with unbroken “scored seals,” and pumping is carried out in this position. Then the mercury is raised to level \(A\) and the lower “scored seal” of the cylinder is broken. The mercury fills the cylinder, pressing the gas sample toward the upper “scored seal,” which is then broken when the mercury rises to level \(B\). Further pressing introduces the whole gas into the capillary analyzer. When the necessary pressure is reached in the capillary analyzer (approximately \(7\)—\(12\) mm Hg), high-frequency voltage from a high-voltage generator is applied to the capillary, and the gas enclosed in the capillary begins to glow. The spectrum of this glow is photographed by means of a spectrograph, and the mixture is analyzed from the spectrum.

Spectral analysis. To increase the purity of the experiment we, on the advice of S. E. Frisch, used an “electrodeless” discharge in the gas. In this case the electrodes*) to which the high-frequency voltage is applied are outside the capillary and therefore do not come into contact with the gases of the sample. In this way all the major shortcomings inherent in the method of analysis with internal electrodes are eliminated, in which the gas enters into reaction with metallic surfaces. However, difficulties of a general nature remain; in particular, the difficulty of obtaining good reproducibility of results remains.

When a high-frequency discharge in a gas is used, the result of the spectral analysis depends on a whole series of causes. These causes may basically be divided into three groups:

A. Factors determining the electron temperature in the discharge: the voltage at the electrodes of the high-frequency generator, the gas pressure inside the capillary analyzer, the geometry of the electrodes, the size of the discharge gap, the diameter of the capillary, etc.

B. Factors determining the properties of the inner surface of the capillary: the type of capillary glass, the “previous history” of the capillary, the amount of the mixture being analyzed, the time of glow of the discharge, etc.

C. Factors determining the stability of the photographic and photometric processes: plates, exposure time, development, photometry, etc. The influence on the result of the analysis of the last group of factors is well known—this influence is identical in all analogous photographic work. The influence of the first two groups, however, has as yet been far from sufficiently clarified. As a result of the investigations carried out on the microanalysis of gas mixtures, we succeeded in obtaining good reproducibility of the results (relative error \(3\)—\(5\%\)) only under strict observance of constancy both in the values of the indicated parameters and in the sequence of the operations performed**). These conditions must be observed especially rigorously in analyses of very small portions of gas mixtures (less than \(1\ \mathrm{mm}^3\) under normal conditions).

*) The electrodes consist of two thin metal plates fixed immovably at a distance of \(1\) cm from one another. The plane of the plates is set perpendicular to the axis of the capillary. Holes are made in the center of the plates, through which the end of the capillary analyzer is passed; the glass of the capillary does not come into contact with the electrodes.

**) Discussion of these questions lies beyond the scope of the present article and will be carried out elsewhere. These questions are touched upon in part in the book by O. P. Bochkova and E. Ya. Shreder, Spectral Analysis of Gas Mixtures.

The method of the analysis itself is generally known—it is the three-standard method, widely used, for example, for spectral analysis of alloys. Its essence is that the spectrum of the gas mixture under investigation is photographed on a photographic plate, as are also the spectra of three standard mixtures (mixtures of known concentrations). After this the plate is photometered, and from the photometric data a calibration graph is constructed, an example of which is given in Fig. 8: along the ordinate axis is plotted the ratio of the intensities of the lines of the two components being investigated; along the abscissa axis—the logarithm of the concentration.

In our work the following lines are used for photometry: OI = 7772 Å; NI = 7469 Å; Ar = 7504 Å. The indicated nitrogen line is used as the comparison line. Many difficulties in the spectral analysis of gases are introduced by multicomponent mixtures. The work of Leningrad University (S. E. Frish^14, O. P. Bochkova^15, and others), as well as our own work, has successfully solved this problem also in the case of air \((O_2+N_2+Ar)\), making it possible to carry out quantitative analysis of microportions of gas with the required accuracy: for oxygen and nitrogen we obtain a relative error of 5–6%, and for argon, 3%.

Fig. 8. Along the abscissa axis is plotted the logarithm of concentration \((\lg C)\); along the ordinate axis—the relative blackening difference \((\Delta S)\) of the corresponding lines.

7. RESULTS OF THE EXPERIMENTS

During the period from 1951 to 1956 we carried out many investigations to determine the composition of the air at high altitudes. Samples were obtained and analyzed by the methods that we have described above. Not all experiments were successful, and therefore Table I below gives the results only of those analyses that give no cause to doubt the purity of the experiment.

Table I

Summary data on the composition of air at high altitudes

From Table I one may draw the following conclusions:

1. Up to altitudes of 95 km, no noticeable separation between oxygen and nitrogen occurs.

2. The amount of the heaviest gas, argon, at altitudes of 85–95 km is less than at the Earth’s surface.

However, the decrease of argon with altitude is still comparatively small, and it is premature to draw final conclusions about the existence of diffusive separation at altitudes of the order of 100 km, although it is quite tempting. A final decision must be obtained from work at still greater altitudes. For comparison, in Table II we give the results of foreign investigations of the composition of air at great altitudes, carried out likewise by the method of taking samples with their subsequent analysis, but not by a spectral, rather by a physicochemical method.

Table II

Recent foreign data on the composition of air at great altitudes

Oxygen is given as a percentage of the total amount of gas in the sample. For helium, argon, and neon the table gives values obtained in the following way: the ratio of the amount of the gas under investigation (helium, argon, neon) to the amount of nitrogen is considered. For near-surface air this ratio is taken as equal to unity, whereas for a sample, in the case of diffusive separation, this ratio will differ from unity. It can be shown that the numbers obtained for argon by Paneth agree very well with the numbers obtained in our work.

8. DISCUSSION OF RESULTS

Comparing all the data available at present on the analysis of the chemical composition of the upper atmosphere, we obtain a rather variegated picture. The results of the analysis of samples taken in cylinders in our country and abroad, although they agree with one another, nevertheless come into contradiction with theoretical works[^1][^2][^3]. The results obtained with the aid of a mass spectrometer disagree with the results of sample analysis. In addition, in recent times the very essence of the sampling method has been subjected to great doubt. All this naturally compels one to approach the results obtained with great caution and creates a certain uncertainty in the conclusions. Therefore it is necessary, insofar as possible, to examine the available material on its merits.

It should be noted at once that investigations of the composition of the atmosphere by means of a mass spectrometer, which did not show a redistribution of nitrogen and argon up to altitudes of 137 km, should not yet be taken seriously:

the entire experiment is full of ambiguities. Thus, for example, at altitudes of 100 km, where oxygen in its overwhelming mass is certainly in a dissociated state, oxygen was detected mainly in molecular form. It is hardly possible to assert on this basis that the dissociation of oxygen at these altitudes is almost absent. In exactly the same way, on the basis of the experiment that was carried out, it is hardly possible to assert the absence of gravitational separation of gases up to altitudes of 137 km. It is also necessary to take into account the fact that the measurement error of the radio-frequency mass spectrometer is rather large (25–30%), and therefore traces of an already beginning gravitational separation could easily have escaped observation. As for the objections put forward by Martin^16 concerning errors in the sampling method itself, in our opinion they are insufficiently substantiated.

The essence of these objections comes down to the fact that the gravitational separation detected by Paneth and presented by us in Table II is interpreted by Martin as the result of distortions of the composition arising as a result of measurements on a rocket. In particular, it is indicated that the presence of a long inlet tube at the cylinder (25 cm) can lead to forced separation of the gases entering the cylinder in the Knudsen regime. It is also indicated that the high velocities of the rocket (several times exceeding the speed of sound at the time of sampling), causing a mismatch between the internal pressure in the cylinder and the external pressure, aggravate the possibility of forced separation by creating an unsteady-flow regime.

The corresponding calculations, carried out by Martin and taking into account the specific features of the experiment (the capacity of the cylinder, the diameter of the inlet orifice, the length of the inlet tube, the velocity of motion of the rocket at the moment of sampling, etc.), seem to confirm his assumption about forced separation of gases at the moment of sampling. However, our experiments do not confirm Martin’s assumptions. Indeed, the entire specific character of our experiment is quite different from that of the American work. First, our cylinders of small capacity have almost no inlet tubes (the length of the passage opening in the valve does not exceed 1 cm). Second, the velocities at which samples were taken do not exceed 30–50 m/sec (the apex of the trajectory)*. Therefore the pressures inside the cylinder and outside it are in good agreement with one another, which indicates an equilibrium process of pressure establishment in the cylinder. Third, our measurements, carried out on a small, readily purged container, differ substantially in aerodynamic conditions from the American work on the rocket itself.

Thus, Martin’s considerations, expressed by him with respect to the American work, are hardly applicable to the description of the physical processes occurring during the taking of air samples in our experiments, while the results of our work, as has already been said, are in good agreement with Paneth’s work. It seems highly unlikely that the results of analysis in our country and abroad should coincide by chance. The probability of such a coincidence becomes still smaller if one takes into account that the indicated experiments differ not only in the method of sampling but also in the method of the gas analysis itself. Most likely it should be assumed that the experimental data obtained both by us and abroad are sufficiently objective. There remain

* In this respect the advantage of the container over the rocket is once again apparent: when carrying out an experiment on a rocket, for the “blowing” of a cloud of harmful gases it is consistently necessary to work at high velocities. For the “non-gasifying” container this condition falls away.

as-yet-unresolved discrepancies between the experimental data and theoretical considerations concerning the height of the level of separation of gases. In this connection it should be noted that the argument about a supposedly unstable atmosphere in the layer from 50 to 80 km altitude (a layer with a negative temperature gradient) is of little validity. As S. Chapman pointed out in his time, the entire atmosphere (with the exception of the near-ground layer) is a very stable formation. For instability to arise in the atmosphere, the mere fact of a negative temperature gradient is still not sufficient; this gradient must be less than the adiabatic one, i.e., less than \(-6^\circ\mathrm{C}\) per kilometer. In the layer mentioned, a gradient of only \(-3^\circ\mathrm{C}\) per kilometer is observed, which is wholly insufficient to create instability in this region. A final verification of the theoretical work will be made when samples from greater heights have been obtained.

Thus, in summing up all the work in the field of direct studies of the chemical composition of the atmosphere at altitudes of the order of 100 km, it must be acknowledged that up to the present time this question still remains open. The only thing that can be asserted with sufficient reliability (with an accuracy of up to 1%) is that the atmosphere up to an altitude of 100 km should be regarded as an oxygen–nitrogen atmosphere.

References Cited

1. H. V. Maris, Terr. Mag. and Atm. Electr. 33, 233 (1928); 34, 45 (1929).
2. P. S. Epstein, Ger. Beitr. z. Geophys. 35, 153 (1932).
3. S. K. Mitra, H. Rakshit, Ind. J. Phys. 12, 47 (1938).
4. F. A. Paneth, J. of Chem. Soc., 3651 (1952).
5. W. H. Bennett, J. App. Phys. 2, No. 2 (1950).
6. J. W. Townsend, Rev. Scient. Instr. 23, No. 10 (1952).
7. F. A. Paneth et al., Nature 168, 358 (1951).
8. P. R. Reasbeck, B. S. Wilborg, in the collection Rocket exploration of the upper atmosphere, 1954.
9. F. A. Paneth et al., J. Atmos. Terr. phys. 1, 49 (1950).
10. J. Langmuir, J. Amer. Chem. Soc. 34, 1310 (1912).
11. Glückauf, Proc. Roy. Soc. 185, 98 (1946).
12. S. E. Frisch, Izv. AN SSSR, ser. fiz. 13, No. 4, 465 (1949).
13. B. A. Mirtov, A. A. Baykov, Zav. lab. 7, 871 (1955).
14. S. E. Frisch, Vestnik LGU, ser. m.-f.-kh., No. 6 (1950); No. 8 (1954).
15. O. P. Bochkova, E. Ya. Shreider, Spectral Analysis of Gas Mixtures. Moscow, 1955.
16. G. R. Martin, The Composition of the Atmosphere above 60 km — article in the collection Rocket exploration of the upper Atmosphere, Boyd. Seaton, 1954, p. 161.