A Brief Outline of the Development of Emission Spectral Analysis in the USSR

V. G. Koritskii, V. V. Nalimov, V. V. Nedler,
S. M. Raiskii, A. K. Rusanov, L. N. Filimonov

In 1860–1895, a number of new chemical elements were discovered by the method of spectral analysis. These brilliant successes in the field of spectrochemistry stimulated attempts to apply the method also for the control of industrial production. Special note should be made of the work of the founder of modern metallurgy, D. K. Chernov, who in 1876 published an extensive study devoted to the flame spectra of the Bessemer converter. D. K. Chernov succeeded in establishing a number of interesting regularities connecting the character of the flame spectrum with particular stages of the Bessemer process. However, the results of this and of all other attempts to apply the spectroscopic method for production purposes were qualitative in character. At that stage in the development of physics, it was not possible to establish an unambiguous quantitative connection between the content of elements in the light source and the characteristics of the spectrum. For a long time the situation seemed hopeless; there arose a conviction that quantitative spectral analysis had no prospects whatever.

However, the emergence of modern theories of atomic structure and of spectrum excitation made it possible to comprehend the difficulties that stood in the way of quantitative methods of spectral analysis. These difficulties proved to be quite serious and were overcome only in the 1920s, thanks to the application of relative methods for measuring the intensities of spectral lines (Gerlah’s method of homologous pairs). From that time on, methods of quantitative spectral analysis and their practical use began to develop confidently and rapidly.

In the Soviet Union, the first spectroscopic works devoted to solving analytical problems connected with the needs of production appeared at the end of the 1920s of the present century. Among them were the works of A. N. Filimonov (State Optical Institute), S. A. Borovik (Geochemical Laboratory of Academician V. I. Vernadskii), and A. K. Rusanov (All-Union Institute of Mineral Raw Materials). These works were devoted to the analysis of ores and minerals, in which the chief advantage of the spectral method—its high sensitivity—came to the fore, while the imperfection of the first quantitative methods—their relatively low accuracy—remained in the background. The first Soviet works were carried out on foreign apparatus, since in those years there was no domestic production of spectral instruments.

In 1931, G. S. Landsberg turned to the problem of practical spectral analysis. Together with a number of his students (S. L. Mandelstam, S. M. Raiskii, and later A. R. Striganov, P. A. Bazhulin, F. S. Baryshanskaya, and others), G. S. Landsberg began the systematic development of practical applications of emission spectral analysis.

As a result of the first exploratory studies, the main paths for the development of the method were outlined; these paths have now brought the Soviet Union to one of the leading places in the world in the breadth of application of the spectral-analytical method and in the depth of the scientific work connected with it.

When, in the subsequent years, the method won universal recognition and became firmly established in our industry, G. S. Landsberg, as chairman of the Commission on Spectroscopy of the Academy of Sciences of the USSR, directed its further development and dissemination. Assessing the 25-year history of spectral analysis in the USSR, we can say with full justification that, for our achievements in this field, we owe much to G. S. Landsberg, to his initiative, energy, and high scientific and moral authority.

A principal indicator of the development of any discipline is the number of publications devoted to the problems it encompasses. During the period from 1931 to 1950, about 1000 journal articles*) on emission spectral analysis were published in the USSR; by the present time their number has approximately doubled. This number of published scientific articles testifies to an extraordinarily broad front of work.

The main part of these articles appeared in the pages of the journal Zavodskaya laboratoriya and in the proceedings of the All-Union Conferences on Spectroscopy, published in Izvestiya of the USSR Academy of Sciences, Physical Series.

In the following very brief account, relying on this enormous body of material, we shall attempt to identify the main features of the development of spectral analysis in the USSR.

1. APPARATUS FOR EMISSION SPECTRAL ANALYSIS

In the years when G. S. Landsberg and his coworkers were taking the first steps in creating applied spectral analysis, spectral instruments of domestic manufacture did not yet exist. The comparatively small stock of assorted imported instruments also could not serve as a basis for developing generally accessible standard methods of analysis. G. S. Landsberg clearly appreciated this circumstance, and therefore, in his very first spectral-analytical work, devoted to the determination of silicon in malleable cast iron[^1], we find a description of a specially designed spectral instrument.

At first it was a narrowly specialized visual instrument intended for the analysis of a single element.

Subsequently, by introducing a movable dispersion system of three prisms (one of them of the Abbe type), the instrument was made universal. Its dispersion proved sufficient to resolve almost all the lines in the comparatively complex spectrum of iron. A polarization photometer provided adequate comparison of line intensities. Figure 1 shows the external appearance of this instrument, known as the NIIF MGU stylometer or ST-1[^2].

At present the optical industry is producing a new, more compact model of the ST-7 stylometer, equipped with a wedge photometer (Fig. 2)[^3].

In addition to the stylometer, G. S. Landsberg and his coworkers created a styloscope, i.e., a specialized three-prism spectroscope providing the so-called semiquantitative analysis—sorting of steels[^4]. Although the original design of this instrument was somewhat bulky and not entirely convenient in operation, its analytical capabilities proved so broad and so well suited to the needs of production that

) Spectral Analysis. Annotated Index of Soviet Works on Spectral Analysis from 1931 to 1950.* Publishing House of the Academy of Sciences of the USSR, 1955.

that many thousands of such instruments were manufactured by our industry under the names styloscope NIIF MGU, SL-1, and SL-3.

In 1941, on the initiative of G. S. Landsberg, a more compact and convenient model of an autocollimation styloscope was designed,^5 the basic scheme of which formed the basis for the new industrial model of the SL-10 styloscope, which at present is the principal instrument of this type (Fig. 3).

Fig. 1. NIIF MGU stylometer.

Fig. 2. ST-7 stylometer.

The need to sort large-sized parts at the place where they were located stimulated the creation of a portable model of the styloscope,^6 now being produced in a somewhat improved form under the designation SLP-1.

As is known, visual methods of photometry cannot provide the accuracy that can be attained by photographic methods of measuring intensities. Moreover, the use only of the visible region of the spectrum, accessible for visual photometry, greatly narrows the analytical possibilities of the method, limiting the choice of spectral lines. Therefore, soon after the creation of visual instruments, G. S. Landsberg raised the question of producing domestic models of spectrographs.

Fig. 3. SL-10 styloscope.

The first attempts to create domestic ultraviolet instruments led to the production of the USP-1 and ISP-4 spectrographs developed by the GOI, with a sylvine prism and quartz-fluorite objectives.

A sharp turning point in spectral instrument making occurred in 1946, when the plant named after OGPU (LOMZ) began mass production und-

under the designation ISP-22, an original model of a quartz spectrograph of medium dispersion according to a design developed by the GOI[^7]. This spectrograph, with a mirror collimator, a meniscus camera objective, and a Cornu prism (Fig. 4), has exceptionally high quality in the imaging of lines,

Fig. 4. Spectrograph ISP-22.

an ideally flat spectrum 24 cm long (for the wavelength region from 2000 to 6000 Å), and is so simple and reliable in operation that for more than 10 years it has been the principal spectral instrument in the equipment of most of our spectral laboratories. This instrument successfully competes with the world-renowned spectrograph of a similar class, the Zeiss Q-24, and, apparently, is the best quartz instrument of medium dispersion.

In recent years the mechanical design of this spectrograph has been somewhat modernized in order to make it more convenient in operation, and it is now produced under the designation ISP-28.

Fig. 5. Spectrograph KS-55.

Following the medium-dispersion instrument there appeared the autocollimation spectrograph of high dispersion KS-55 (Fig. 5), in which the spectrum, covering the region from 2000 Å to 10,000 Å, is recorded in sections on three photographic plates 9 × 24 (the total length of the spectrum reaches 700 mm).

The availability of interchangeable quartz and glass optics and the carefully developed mechanical part of the instrument, ensuring a sufficiently precise connection of all optical components necessary in passing from one region of the spectrum to another, place this instrument on a level with the corresponding foreign models[^8]. In a somewhat modernized form this instrument is currently produced under the designation KSA-1.

Although, as has already been said, the visible region of the spectrum does not provide a complete solution of analytical problems, nevertheless a number of problems, for example deter-

separation of gases, is handled chiefly by bringing in this part of the spectrum. For this purpose the glass spectrograph ISP-51 is used, which in essence represents an entire series of instruments; on the basis of a three-prism dispersion system (one of the prisms of the Abbe type), by replacing collimators and cameras one can obtain four versions of the instrument, with a dispersion from 105 to 5 Å/mm for \(\lambda=5000\) Å and an aperture ratio from \(1:2.3\) to \(1:20\)⁸. For emission spectral analysis, the intermediate version of the instrument, with a dispersion of 47 Å/mm and an aperture ratio of \(1:5.5\), has found the widest application (Fig. 6).

Fig. 6. Spectrograph ISP-51.

The appearance of domestic spectrographs created a need for microphotometers, as well as for instruments for decoding spectrograms.

At first, while the search was still under way for the most successful design of a microphotometer and the mastery of its production, L. E. Vvedenskii proposed an original method for carrying out quantitative analysis that did not require the use of a microphotometer⁸˒⁹. This method, which received the name “photometric interpolation method,” is based on visual comparison of the blackenings of analytical lines with the aid of a step attenuator; it provides an experienced observer with an accuracy not inferior to the accuracy of the results of objective comparison of intensity with a microphotometer. With the appearance of domestic microphotometers, the use of the photometric interpolation method, owing to its subjectivity, declined somewhat; however, in a number of cases, thanks to its speed, it is still used at the present time.

The first industrial model of a modern photoelectric microphotometer appeared in the Soviet Union under the designation MF-1⁸. This instrument, close in design to the Hilger model, did not provide the necessary convenience in operation. Soon the optical industry switched to the production of a considerably more advanced model, the MF-2, developed on the basis of a Zeiss Schnellphotometer. At present this instrument is used in the overwhelming majority of our spectral-analytical laboratories⁸. The very positive results of its operation served as the basis for creating, on the basis of the MF-2, a new instrument—the recording microphotometer MF-4 (Fig. 7), which has found application not only

Fig. 7. Microphotometer MF-4.

in the field of spectral analysis, but also in other fields of research, in particular in X-ray diffraction analysis[^10].

For the interpretation of spectrograms, at first fairly simple “stands for interpreting spectrograms,” constructed on the basis of a counting ruler, were used; then the MIR-11 measuring microscope appeared (later the analogous instrument MIR-12), and, finally, the entirely modern IZA-2 comparator (Fig. 8), not inferior to a corresponding model of Zeiss.

However, projection-type instruments are currently the most widespread. Their great advantage is that, when the appropriate atlases are available, they greatly facilitate the interpretation process, which in this case is reduced to the simple superposition of the atlas with the image of the spectrum on the instrument screen. The PS-18 spectroprojector, in combination with the Atlas of Spectral Lines of Elements[^11], makes possible the application of this method to the interpretation of spectrograms taken with ISP-22 and ISP-28 spectrographs, and in combination with another atlas[^12]—of spectrograms taken with the ISP-51 spectrograph with a camera of \(F=800\) mm.

Fig. 8. IZA-2 comparator.

Still greater possibilities for qualitative analysis are offered by the DSP-1 double spectroprojector, which enables comparison of spectra by superposing spectrograms taken on instruments with different linear dispersion.

In the practice of spectral analysis, instruments intended for supplying analytical light sources—an electric arc or a condensed spark—are of the greatest importance. The specific purpose of these sources required the development and industrial production of special generators.

S. M. Raiskii developed the so-called stabilized circuit of spark discharge[^13]. In this circuit the voltage on the capacitor, and consequently the discharge energy as well, is stabilized by means of a specially included, so-called discharge-gap-setting interval. The accuracy of analysis results obtained with this circuit is 2–3% and is not inferior to the accuracy obtained with the Feussner circuit, which is widespread abroad. However, the Raiskii circuit has the substantial advantage that it lacks a synchronous motor—one of the vulnerable points of the Feussner circuit.

The generator based on Raiskii’s circuit is currently being produced under the designation IG-2.

N. S. Sventitskii proposed using alternating current at the usual voltage of 120–220 V to supply an electric arc. However, as is known, an alternating-current arc between metallic electrodes cannot burn steadily in this case. N. S. Sventitskii developed a circuit for a special “alternating-current arc generator,” by means of which, in addition to the mains voltage, a high-frequency high voltage is applied to the electrodes, breaking down the interelectrode gap at the appropriate moment and thereby ensuring ignition of the arc after each of its inevitable extinctions[^14].

As was found subsequently, Sventitskii’s generator has much broader capabilities than was initially assumed. With its aid it is also possible to obtain a spark condenser discharge at a capacitor voltage of 220 V, successfully replacing Raiskii’s spark in a number of cases[^15]. The low-power “igniting” spark of the generator can also serve as an independent light source for solving a number of particular analytical problems, for example for local spectral analysis. An alternating-current arc generator DG-1 is manufactured industrially according to N. S. Sventitskii’s circuit.

The desire to obtain still greater stability of vaporization and spectrum excitation is currently leading to research in the development of generators, both arc and spark, in which the discharge parameters would be regulated by means of electronic control methods. In this direction, substantial successes were achieved by I. S. Abramson[^16]. Such generators are currently being manufactured for equipping photoelectric spectral installations.

Fig. 9. Photoelectric stylometer FES-1.

Methods of spectral analysis with photoelectric recording of the spectrum are the most progressive, since they ensure the maximum speed and accuracy of analysis. Unfortunately, it must be stated that in the area of their development we have lagged behind a number of foreign countries.

The first photoelectric instrument suitable for industrial application was created only in 1953[^17]. True, this model, like its subsequent modifications, being based on the use of standard light sources, does not make full use of the advantages of photoelectric recording with respect to the accuracy of the results obtained.

In 1956 our industry began serial production of the FES-1 photoelectric stylometer—an instrument designed to use only the visible part of the spectrum (Fig. 9). Determination of elements on it is carried out by successively measuring the intensity of their spec-

tral lines. The accuracy of the analysis is approximately 1–2%. The time spent on the determination of one element is 1 minute[^18]. Industrial tests are now being carried out on a 36-channel instrument of the quantometer type developed at the State Optical Institute, in which one or another program for the determination of 12 elements can be set.

There is reason to hope that in the near future emission spectral analysis in the USSR will be provided with all modern types of instruments necessary for its implementation.

Turning to the practical applications of spectral analysis, we are compelled to confine ourselves to only a few typical examples. In reality, at present it has been placed at the service of all branches of the national economy without exception.

2. APPLICATION OF SPECTRAL ANALYSIS IN THE MACHINE-BUILDING AND INSTRUMENT-MAKING INDUSTRY

In these branches of the national economy, spectral analysis is used to control the casting of ferrous and nonferrous metals, and also to inspect semifinished products, parts, assemblies, and finished articles. The metal arriving at enterprise warehouses and issued from warehouses to shops is also inspected. Here semiquantitative and other rapid methods are especially effective. A classic example is the first use of the steeloscope (1934), which at the time made it possible, in a short period, to eliminate the mixing of metal grades at the Moscow Likhachev Automobile Plant, a mix-up that had threatened the production process for making automobiles[^4]. The successful solution of the problem, in which millions of determinations were performed on a single installation, played a decisive role in promoting the new method.

Another example of the specific capabilities of spectral analysis is the control of the chemical composition of parts and finished products in situ (without damaging them). Portable spectral installations are widely used for this purpose; they are applied directly in shops, in warehouses for semifinished products, and in finished-product yards. These installations make it possible very quickly to separate and route to its intended destination production made from acceptable metal, and, in products made from substandard materials, to replace parts[^19].

The introduction and effective use of spectral analysis at the “ZiL” automobile plant played a major role in the promotion of the new method and in its wide dissemination in the machine-building industry. Much credit here belongs to V. F. Smirnov, a student of G. S. Landsberg and S. L. Mandelstam.

3. CONTROL OF FERROUS-METALS PRODUCTION

Spectral analysis has found application in solving the principal production problems of ferrous metallurgy: rapid analysis of steel in the course of its production (during the heat); final, so-called “marking” analysis, the results of which are necessary for drawing up the certificate for the finished product; analysis of cast iron; analysis of charge materials, ferroalloys, scrap, ore and agglomerates, and slags. Finally, the analysis of refractory materials with which metallurgical furnaces are lined should be noted; the composition of these materials sometimes has a very substantial effect on the quality of the steels smelted in the furnaces.

Rapid analysis during the heat is the classic problem which, as already noted above, prompted G. S. Landsberg to undertake the development of methods of applied spectralagext?

of analysis in general. This task appears especially attractive for spectral analysis because of its greater speed, in comparison with the methods of analytical chemistry; in the present case speed is of very substantial importance, since, under production conditions, rapid analysis must be carried out within periods not exceeding 15–20 minutes.

The initial solution was found through the use of the stylometer—a visual instrument equipped with a polarization photometer², which makes it possible to determine one element in 5–6 minutes, with an error of 5–8%. This instrument is still used for determining the content of silicon, manganese and, more rarely, certain alloying elements.

The relatively high accuracy of spectral analysis in determining small concentrations of elements led to the wide use, for rapid-analysis purposes, of a simpler instrument—the styloscope. The styloscopic method is comparatively rough; at best it gives an answer with an accuracy of 20%. However, in the range of concentrations of the elements being determined not exceeding 0.2–0.3%, the accuracy of such a determination for most steel components is not inferior to the accuracy of the corresponding chemical methods; the speed of the analysis, lasting 2–3 minutes, is unattainable for chemical methods.

At present, however, the principal method of rapid spectral analysis of steels is the photographic method. The duration of determinations in this case increases substantially (with visual interpolation of line blackening—to 12–15 minutes, with microphotometer measurement—to 20–25 minutes); but the error of the analysis decreases to 2–3%, which considerably expands the analytical capabilities of the method. The duration of analysis by the photographic method, in the case of heats conducted in modern high-capacity furnaces, in which the metal is melted over 10–12 hours, is still within permissible limits, although it often approaches them.

By photographic means, rapid determination is carried out of the content in steel of silicon, manganese, and also of all alloying elements at concentrations not exceeding 1.5–2%.

However, the existing tendency toward a continuous increase in the productivity of the steelmaker’s labor will in the near future inevitably lead to the point where photographic methods of spectral analysis will cease, in terms of their speed, to satisfy production requirements. In connection with this, metallurgical enterprises are awaiting the introduction of photoelectric spectral instruments capable of radically solving the problems of rapid analysis. In addition to increasing the speed and accuracy of analysis, this creates prospects for the automation of production-process control, since the electrical signal produced by a photoelectric light receiver can be used to actuate various mechanisms regulating the composition of the metal being smelted.

The requirements imposed on grade-identification spectral analysis at present differ little from the requirements for rapid analysis. Twenty years ago, before the appearance of such giants of Soviet metallurgy as the Magnitogorsk and Kuznetsk Metallurgical Combines, the process of smelting steel and its subsequent processing—mainly rolling—were carried out with a considerable gap in time (often the smelting and processing of the metal were performed at different enterprises). This made it possible to analyze steel by the most accurate, but, as a rule, lengthy method. At present

metal, in the hot state, goes directly from the open-hearth shop to the rolling shop, and sometimes even less time is allotted for grade analysis than for rapid analysis (with increased requirements for the accuracy of the result). In this connection there is also no essential difference in the procedures for these varieties of analysis.

The experience of applying grade spectral analysis of steels was recently summarized in a large work on the standardization of methods of spectral analysis of steel ^20.

Pig iron (foundry iron) is an intermediate product of metallurgical production and does not require especially high accuracy of analysis. At the same time, the character of blast-furnace production, which possesses a certain inertia, does not compel analysis to be accelerated to the utmost. Therefore both the stylometric and the photographic methods have found wide application here.

Among charge materials, ferroalloys are of substantial importance for the smelting of special steels. Until recently it was believed that only impurities could be determined in them by the spectral method with a sufficiently high degree of accuracy. The principal elements of the alloys, whose content reaches several tens of percent, were not subjected to spectral analysis. However, ways have recently been found for determining these components as well with sufficient accuracy ^21.

Spectral analysis of scrap—metal waste, which sometimes constitutes up to 30% of the charge consumed for smelting steel in an open-hearth furnace—is of great practical importance. Here great accuracy of determinations is not needed, but the number of required determinations is so large that neither chemical nor any other of the analysis methods known at the present time, except the spectral one, is capable of solving this problem. Only the creation of a portable styloscope, enabling one laboratory assistant to carry out up to 8000 determinations per shift, made it possible not only to pose, but also fully to solve, this very urgent problem of metallurgy ^6. Analysis of ores, sinters, and refractories is performed with the use of photographic methods.

Rapid analysis of slags is of special importance in ferrous metallurgy. For proper conduct of the heat it is necessary to follow not so much the change in the composition of the metal as the change in the composition of the slag, under the layer of which steel is usually smelted. Chemical methods of slag analysis are so lengthy that with their aid one can obtain results having only “historical” value. Therefore the very first attempts at spectral analysis of slags ^22 were met by metallurgists with great interest. However, the experience accumulated later showed that, in the analysis of slags, substantial difficulties are encountered, connected with the influence of the mineralogical structure of the sample on the results of the analysis ^23, ^24, ^25. The problem of rapid slag analysis cannot yet be considered solved, although the spectral method yields certain results ^26.

Of very great production importance is the spectral determination of gases contained in steel (oxygen, nitrogen, and hydrogen). However, the spectrum of these elements is difficult to excite, and, moreover, they enter into steels in hundredths and thousandths of a percent. At first this circumstance precluded the application of the spectral method of gas analysis. It was necessary to develop special procedures based on the use of a powerful excitation source—an impulse discharge of a capacitor ^27; as a result, the determination of gases in steel proved possible and is already being adopted by a number of plants.

In ferrous metallurgy, by the present time there remains almost no enterprise that does not make use of the services of spectral analysis. Considerable credit for its introduction belongs to the pioneer in the application of spectral methods in steel production—the Elektrostal plant, where as early as 1938, under the direction of G. S. Landsberg and F. S. Baryshanskaya, the first stylometer was installed. Here, as at other plants, hundreds of thousands of quantitative determinations of elements in various materials are carried out each year.

4. CONTROL OF THE PRODUCTION OF NON-FERROUS METALS

In non-ferrous metallurgy, as in ferrous metallurgy, methods of semiquantitative analysis (sorting by grades) are successfully applied to alloys based on copper, aluminum, magnesium, zinc, nickel, etc.^28 However, production also requires more accurate analyses of non-ferrous metals, and therefore the corresponding methods were intensively developed in a number of laboratories: at the Research Institute of Physics of Moscow State University under the direction of Academician G. S. Landsberg, at the State Optical Institute under the direction of Academician D. S. Rozhdestvenskii and Prof. V. K. Prokofiev, and also in laboratories headed by K. A. Sukhenko, A. R. Striganov, and L. E. Vvedenskii—pioneers in the application of spectral methods in the production of non-ferrous metals. Later many other laboratories joined this work. The extensive experience accumulated by 1940 was generalized and systematized in a special collection.^29 Three simple visual methods of analysis, which we have already discussed, were widely used: photometry with the aid of a polarization photometer (stylometer), photometric interpolation, and the classical method of homologous pairs. As is known, the advantage of the methods listed is their rapidity, while their drawback is the subjectivity of the estimates and the comparatively large error in the results of analysis associated with it. Therefore the next task consisted in developing and introducing more accurate photographic methods based on objective photometry with the use of a microphotometer.

In the very great variety of specific photographic methods for the analysis of non-ferrous metals developed up to the present time, one may distinguish: general-purpose methods intended for determining impurities and alloying components in low-alloy alloys in concentrations from several hundredths of a percent to several percent; methods possessing increased sensitivity (down to \(1 \cdot 10^{-3} — 1 \cdot 10^{-4}\%\)); and, finally, special, particularly sensitive methods (\(1 \cdot 10^{-4} — 1 \cdot 10^{-6}\%\)). The differences between these methods lie in the ways of preparing samples and introducing them into the light sources.

The general-purpose methods, the simplest, are based on the direct introduction into the discharge, as electrodes, of samples of the non-ferrous metal being analyzed; they do not differ essentially from the methods for analyzing solid samples of ferrous metals.

For controlling the industrial production of metals of high purity, methods of increased sensitivity are successfully applied. As a rule, they use arc light sources, most often a direct-current arc. For this type of analysis, the most characteristic is the method of fractional distillation, carried out in an arc between graphite electrodes, and the so-called globule-arc method.

The method of fractional distillation, as follows from its name, makes use of the difference in the elasticities of the vapors of the impurity being determined and of the base metal.

component of the sample. The sample (or reference specimen) is first converted into a chemical compound in which the principal component is least volatile. Then the weighed portion of the sample is placed in the crater of one of the graphite electrodes of the arc. The shape and size of the electrode, the depth of the crater, and the current strength are selected so that the sample is heated to a temperature ensuring the maximum difference in the evaporation rates of the determined impurities, on the one hand, and of the principal component of the sample, on the other. In this way a very substantial enrichment is achieved of the vapor entering the arc flame with respect to the determined impurities. At the same time, the spectrum of the principal component is sharply weakened, which is very important in the analysis of substances with many-line spectra. The arc spectrum is photographed in the interval of time when the spectral lines of the impurities are most intense. The blackening (intensity) of the lines is related to the concentration of the impurity in the sample, as usual, by means of a calibration graph constructed from reference specimens of known composition.

Such a procedure usually makes it possible to determine 10–20 impurities with a sensitivity of the order of \(1 \cdot 10^{-3}\)—\(1 \cdot 10^{-4}\%\); the error of determination is \(\pm 10\)—20% relative. Within the scope of the present article we shall confine ourselves to only a few typical examples of the results of applying this method, given in Table I.

Table I

Lowest concentrations of impurities determined by the method of fractional distillation in the arc (concentrations are given in weight percent)

In the globule method, the arc discharge directly evaporates a fused, preliminarily oxidized sample of small weight placed on a heat-insulating support. We shall confine ourselves to a brief description of a variant recommended for the determination of impurities in high-purity nickel.

The spectra are photographed with the sample included as the anode and then as the cathode. With the first polarity of the electrodes, the spectrum is enriched with the lines of volatile impurities; with the second polarity, with the lines of nonvolatile impurities. The sensitivity of determination increases and reaches the values given in Table II. The nature of the processes occurring in this case is not yet entirely clear.

About ten years ago the evaporation method was developed—the most sensitive method for determining volatile impurities in nonvolatile matrices. Evaporation is carried out in vacuum \(^{35, 36, 37}\) or in an atmosphere of air—

... arc 38. In the evaporator method the evaporation process is separated from the process of exciting the spectrum. This makes it possible to select the most advantageous conditions of analysis.

Table II

Lowest concentrations of impurities quantitatively determined by means of a carbon arc (in percent by weight)

Evaporation of impurities is carried out by placing the sample in a graphite cup clamped between massive graphite brushes. An alternating current of high strength (150—400 a) from a low-voltage source (6—12 v) is supplied through the brushes to the cup. The heat liberated can heat the cup to a very high temperature, which is varied by changing the current strength and is measured with an optical pyrometer.

To trap the vapors issuing from the cup, a receiver (Fig. 10), made of graphite, copper, or another suitable material, is placed above its opening. After the deposition process is completed, the receiver with the condensate is placed in the spectrograph stand and the condensate is burned in an arc or spark discharge.

Fig. 10. Diagram of the evaporator: 1 — graphite water-cooled brushes for supplying current, 2 — graphite cup with the sample, 3 — receiver.

In the analysis of complex-alloyed and structurally inhomogeneous alloys, effects of the influence of third elements 39, 40, 41 and of the structure of the sample 42 appear, which may be a source of considerable systematic errors. In connection with this, a number of works were carried out to elucidate the general causes producing these phenomena. An extensive cycle of works was devoted to the study of the processes occurring on the electrodes and the regularities of the entry of the sample substance into the emitting cloud (spark or arc) 43, 44, 45, 46, 47. It was shown that the complex of processes taking place on the surface of the electrodes (chemical reactions, diffusion, vapor formation, etc.) can lead to a substantial difference between the chemical composition of the original sample and the composition of the vapor cloud formed from this sample.

Table III gives typical examples of the sensitivity of determinations achieved in the evaporator method.

Table III

Lowest concentrations determined by the evaporator method (in weight percent)

The study of the processes accompanying the conversion of a sample into vapor continues to remain the most important task of the theory and practice of quantitative spectral analysis.

At the present time, photographic methods for the analysis of impurities in high-purity metals have been introduced into production: aluminum, magnesium, copper, nickel, cobalt, tungsten, molybdenum, titanium, niobium, zirconium, tantalum, uranium, thorium, vanadium, cadmium, calcium; and in a number of noble, rare, and scattered elements. Methods of spectral analysis have proved to be the only reliable means of controlling the industrial production of high-purity metals used in the most important fields of new technology: electrovacuum production, semiconductor technology, the atomic industry, rocket technology, and others.

Great successes have also been achieved in the analysis of alloys based on the most important nonferrous metals: aluminum, copper, and others. At plants for the processing of nonferrous metals, more than half of all analytical work is carried out by methods of spectral analysis. However, in the field of analysis of nonferrous-metal alloys, and especially the analysis of copper alloys, successes such as those achieved in the analysis of high-purity metals have not yet been attained. Serious fundamental difficulties arise here.

It may be supposed that the widespread introduction of instruments with photoelectric recording of the spectrum, providing higher accuracy and speed of analysis, will make it possible to overcome these difficulties.

5. SPECTRAL ANALYSIS OF MINERAL RAW MATERIALS AND PRODUCTS OF THEIR PROCESSING

The possibility of rapid and practically simultaneous determination of the chemical elements contained in a sample, as well as the high sensitivity of spectral analysis, have led to an extremely broad application of this method in various fields of geological research and for controlling ore-beneficiation processes.

The number of samples subjected to spectral analysis in the USSR over the last twenty-five years has increased by more than 5000 times.

Geological organizations of the Soviet Union (expeditions attached to prospecting parties, central laboratories of geological administrations, scientific research institutes, etc.) have numerous cadres of spectroscopist-analysts and are equipped with domestic apparatus that makes it possible to solve problems of both qualitative and quantitative spectral analysis of ores of complex composition.

The rapid introduction of methods of spectral analysis was facilitated by the creation of specialized manuals on the spectral analysis of mineral raw materials^48 and handbooks^11, 12, 49, 50, 51.

The methods of spectral analysis used in the geological service are specific and at the same time very diverse. Prospecting for useful minerals by means of metallometric surveying, which has been successfully developing since the thirties, in most cases requires mass semiquantitative spectral determinations of elements in samples taken in the area under study from unconsolidated deposits. The content of these elements usually does not exceed thousandths or hundredths of a percent.

During the course of a year, millions of analyses of such samples are carried out in the USSR for the content of 6–12 elements, with the aim of revealing halos of dispersion of elements above outcrops of ore bodies^52. The performance of such a volume of analytical work is possible only thanks to the use of the spectral method.

Metallometric surveying has as its aim the detection of areas with an elevated content of one or another element; therefore the analysis is reduced to the classification of samples by dividing them into groups containing a given element within some interval of concentrations specified in advance. Most often an interval is specified in which the concentration changes threefold (0.001–0.003; 0.003–0.01; 0.01–0.03%, etc.^53).

The decoding of spectrograms is usually carried out visually, using a spectroprojector. The presence of known analytical lines in the spectrum of a sample, or else the ratio of the blackenings of such lines in the spectrograms of samples and standards, makes it possible to establish the concentration of elements in the sample.

The study of the composition of ores, minerals, and rocks in deposits of useful minerals, as well as of dry residues of natural waters, requires semiquantitative determination in a specimen of 45–50 elements, including impurities, a number of elements contained in thousandths of a percent. In solving geochemical problems it is often necessary to determine still lower concentrations of individual elements.

The spectral method of sorting ore samples before sending them for quantitative analysis is reduced to a semiquantitative assessment of the concentration of several elements (the element being determined, as well as elements capable of hindering the performance of the quantitative analysis).

Unlike the analysis of metallic alloys, quantitative spectral analysis of ores is usually carried out only in special cases, when chemical analysis takes much time, is associated with great difficulties, or gives inaccurate results. Such a case is encountered, for example, in the determination of scattered and rare metals. At present, procedures have been developed and are being applied for the quantitative spectral determination of beryllium, indium, thallium, gallium, germanium, cadmium, rubidium, cesium, lithium, and the rare earths.

Quantitative spectral analysis of mineral samples, which makes it possible to determine the content of elements with an average arithmetic error of $\pm 5\text{--}10\%$, is based on the usual procedure. With the aid of standards, the dependence of the relative intensity of an analytical pair—consisting of the spectral line of the element being determined and that of the comparison element—on concentration is established empirically. The resulting dependence is then used to determine the content of the given element in the samples being analyzed.

Methods of quantitative determination that do not provide for preliminary treatment of the sample prove reliable only when the mineralogical composition of the samples and standards is sufficiently close. For this reason, along with methods for the direct analysis of samples, methods are also used that provide for the preliminary chemical separation of the elements to be determined together with a group of other elements of some constant composition. Thus scandium, hafnium, and niobium are determined. In order to eliminate the influence of the bulk composition of the sample, dissolution of the sample being analyzed is also used. This procedure is employed in the determination of rubidium, cesium, lithium, indium, thallium, gallium, and beryllium.

Methods of direct quantitative spectral analysis have found wide application in the control of ore-beneficiation processes. These include quantitative determinations of the content of useful components in the starting material, industrial products, and tailings. The results of the analyses have sufficient accuracy, and the analyses are carried out with sufficient speed $^{54,55,56}$.

Despite the great variety of problems solved by the spectroanalytical method, and the many modifications of this method, there are fundamental features that distinguish the process of analyzing samples of mineral origin from the process of analyzing metallic alloys. Ore samples, first, are nonconductors or poor conductors of electricity and, second, arrive for analysis in the form of a mechanical mixture of powders of various minerals. Therefore, when spectral analysis of mineral samples is performed, they cannot directly serve as the electrodes of a spark or arc. It becomes necessary to use special methods for evaporating the sample and exciting the spectrum of its vapors.

The most widespread method consists in evaporating the sample from the channel of the lower electrode of a direct-current or alternating-current arc. Depending on the temperature to which the sample must be heated, the carbon electrode is given one form or another in the analysis.

The processes of evaporation of the sample and excitation of the spectrum of the vapors, which underlie spectral analysis, have during the last twenty-five years been subjected to systematic investigation $^{57,58,59,60,61,62}$. The results obtained showed that, in the course of evaporation of the sample, a fractionated entry of elements into the zone of the arc discharge occurs.

An undoubted connection has been established between the duration and rate of entry of an element’s vapors into the discharge and the evaporation temperature of the compound in which this element is present in the sample. The dependence

changes in the intensity of spectral lines with the time elapsed from the moment of switching on the arc and the beginning of the process of evaporation of the sample is illustrated by Fig. 11. Systematization of the extensive factual material on the study of the phenomenon of fractional evaporation of elements from the electrode of a carbon arc made it possible to arrange the elements and their compounds in series according to volatility^48.

Fractional entry of the components of the sample into the arc discharge leads to a change in the composition of the arc cloud. This in turn is reflected in the excitation conditions of the spectra of the elements being determined. The question was studied theoretically and experimentally^57, 62, 63.

Fig. 11. Change in the intensity of spectral lines of certain elements during evaporation of an ore sample.

Fig. 12. Dependence of the temperature of the carbon-arc flame on the ionization potentials of the evaporated elements.

Figure 12 gives the dependence between the temperature of the gas cloud of the carbon arc and the values of the ionization potentials during evaporation of various compounds from the channel of the carbon electrode.

The phenomenon of fractional evaporation leads to the fact that extremely simple and rapid analyses by the method of complete evaporation give results that depend on the bulk composition of the samples and, in addition, are burdened by large random errors caused by local heating of the electrodes. Various procedures were proposed for increasing the accuracy and correctness of ore analyses^64, 65, 66, 67.

Considerable success was achieved by applying the method of pouring samples through an arc discharge^67. In a new variant of the method, the powder is blown through the discharge by an air stream^68. The principle of operation of the corresponding device is shown in Fig. 13; the air stream, carrying the particles of the sample into the flame of the alternating-current arc, is produced by a small fan.

Fig. 13. Diagram of operation of an apparatus for analysis of ores using an air blast.

For the determination in ores of elements with low excitation potentials of the spectrum (alkali and alkaline-earth elements: indium, thallium, gallium, etc.), an acetylene-air or acetylene-oxygen flame is in a number of cases successfully employed^69, 70, 71, 72. The spectrum is recorded by a photoelectric device connected to a monochromator or to an instrument with interference

with light filters. The method has high accuracy; the relative error is ±2–3%.

A special problem is posed by the spectral analysis of precious-metal ores, which are characterized by inhomogeneity that is not eliminated by grinding and mixing. Usually the “bead” method of analysis is used, in which the sample is a grain of alloy extracted from the precious-metal ore, with silver or copper^73. The results of the analysis make it possible to monitor the technological process of ore treatment with sufficient accuracy. Sometimes, for this purpose, solutions are analyzed^74.

The present brief review does not claim to be complete. Many very interesting and important applications of the method have not been considered by us, or have been mentioned only very briefly. We refer readers to the monographs by S. L. Mandelstam^57, V. K. Prokof’ev^7, A. K. Rusanov^48, N. S. Sventitskii^28, G. F. Borovik-Romanova^51, and also to the proceedings of the I–X All-Union Conferences on Spectroscopy^75.

REFERENCES CITED

1. G. S. Landsberg, S. L. Mandelstam, and S. M. Raiskii, Zav. lab. 3, No. 5, 771–779 (1933).
2. S. L. Mandelstam, V. F. Smirnov, and V. V. Tseiten, Zav. lab. 5, No. 4, 447–453 (1936).
3. E. L. Berman, Izv. AN SSSR, Ser. fiz. 19, No. 1, 38–40 (1955).
4. G. S. Landsberg, S. L. Mandelstam, S. V. Tulyankin, and V. V. Tseiten, Zav. lab. 4, No. 10, 1220–1277 (1935).
5. L. M. Ivantsov, Izv. AN SSSR, Ser. fiz. 9, No. 6, 733–738 (1945).
6. A. P. Lyubimov, S. L. Mandelstam, L. N. Filimonov, and N. A. Fokin, Zav. lab. 11, No. 2–3, 174–181 (1945).
7. V. K. Prokof’ev, Photographic Methods of Spectral Analysis of Metals and Alloys, Gostekhizdat, 1951.
8. L. E. Vvedenskii, Author’s Certificate No. 45441, 1935.
9. L. E. Vvedenskii and S. L. Mandelstam, Zav. lab. 5, No. 8, 972–980 (1936).
10. G. Ya. Kossov, Izv. AN SSSR, Ser. fiz. 19, No. 1, 56–57 (1955).
11. S. K. Kalinin, A. A. Yavnel’, A. I. Alekseeva, and L. E. Neimark, Atlas of Spectral Lines, Gostekhizdat, 1952.
12. S. K. Kalinin, L. E. Neimark, V. L. Marguzanova, K. I. Ismagulova, Atlas of Spectral Lines for a Glass Spectrograph, Gosgeoltekhizdat, 1956.
13. S. M. Raiskii, ZhTF 9, No. 19, 1719–1726 (1939).
14. N. S. Sventitskii, Zav. lab. 7, No. 12, 1371–1378 (1938).
15. N. S. Sventitskii, DAN SSSR 37, No. 7–8, 233–236 (1942).
16. I. S. Abramson, Zav. lab. 17, No. 9, 1081–1088 (1951).
17. M. M. Averbukh, L. M. Ivantsov, and A. V. Kandinov, Zav. lab. 20, No. 1, 57–67 (1954).
18. L. P. Malyavkin and I. S. Abramson, Izv. AN SSSR, Ser. fiz. 19, No. 1, 40–42 (1955).
19. V. F. Smirnov, Izv. AN SSSR, Ser. fiz., No. 2, 175–176 (1937).
20. Methods of Spectral Analysis of Steel and Converted Cast Iron, publ. by the Ministry of Ferrous Metallurgy of the USSR.
21. L. I. Topalov, in the collection Modern Methods of Analysis in Metallurgy, Metallurgizdat, 1955.
22. I. M. Veselovskaya, Zav. lab. 13, No. 2, 219–221 (1947).
23. V. V. Nalimov and K. I. Ionova, Zav. lab. 18, No. 3, 301–305 (1952).
24. V. V. Nalimov, N. I. Kambulatov, and K. I. Ionova, Zav. lab. 18, No. 11, 1354–1358 (1952).
25. I. V. Granovskii and N. P. Kuz’mina, Zav. lab. 20, No. 4, 436–440 (1954).
26. O. I. Nikitina, “Results of work on the development of the spectrochemical method of analysis of open-hearth slags,” Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University (in press).
27. S. L. Mandelstam and O. B. Fal’kova, Zav. lab. 16, No. 4, 430–438 (1950).
28. N. S. Sventitskii, The Styloscope and Its Application, Gostekhizdat, Moscow–Leningrad, 1948.

1. L. E. Vvedenskii, S. L. Mandel’shtam, S. M. Raiskii, V. F. Smirnov, A. R. Striganov, K. A. Sukhenko, Methods of spectral analysis of metals, Goskhimizdat, Moscow—Leningrad, 1940.

2. L. N. Filimonov, N. A. Makulov, Z. A. Zakharov, Spectral-analytical determination of impurities in tungsten preparations, Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University (in press).

3. M. E. Britske, A. N. Bronshtein, E. B. Gerke, L. M. Ivanov, Quantitative determination of impurities in molybdenum, Materials of the II All-Union Conference of Spectroscopist-Analysts of Nonferrous Metallurgy, 1957.

4. N. M. Gegechkori, D. M. Shvarts, Zav. lab. 19, No. 5, 580—584 (1953).

5. L. N. Filimonov, Zav. lab. 16, No. 10, 1200—1207 (1950); 19, No. 5, 569—579 (1953).

6. D. M. Shvarts, I. S. Nilova, Collection of materials of technical information, No. 3—4, 1956, p. 3, publ. by the Gipronikel Institute, Leningrad; also Izv. AN SSSR, ser. fiz. 18, No. 2, 280—281 (1954).

7. A. N. Zaidel’, N. I. Kaliteevskii, L. V. Lipis, M. P. Chaika, Yu. I. Belyaev, ZhAKh 11, No. 4, 21—29 (1956).

8. A. N. Zaidel’, N. I. Kaliteevskii, L. V. Lipis, M. P. Chaika, ZhAKh 12, No. 1, 17—22 (1957).

9. Yu. I. Belyaev, A. N. Zaidel’, ZhAKh 12, No. 1, 30—40 (1957).

10. S. L. Mandel’shtam, N. N. Semenov, Z. M. Turovtseva, ZhAKh 11, No. 1, 9—20 (1956).

11. N. M. Gegechkori, T. B. Edneral, S. L. Mandel’shtam, Zav. lab. 12, No. 6, 559—568 (1946).

12. V. K. Prokof’ev, Izv. AN SSSR, ser. fiz. 9, No. 6, 607—617 (1945).

13. E. I. Vorontsov, Zav. lab. 13, No. 4, 438—441 (1947).

14. A. R. Striganov, DAN SSSR, new series 31, No. 5, 437—440 (1941).

15. L. N. Filimonov, Izv. AN SSSR, ser. fiz. 12, No. 4, 402—411 (1948).

16. A. B. Shaevich, On the mechanism of the influence of alloy structure on the results of analysis, Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University.

17. L. N. Filimonov, On the question of the relation between the composition of the sample and that of the emitting vapor cloud in spectral analysis, Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University (in press).

18. N. A. Makulov, Investigation of the relation between the composition of the sample and that of the emitting cloud, Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University (in press).

19. A. B. Shaevich, Zav. lab. 19, No. 10, 1189—1196 (1953).

20. A. K. Rusanov, Spectral analysis of ores and minerals, Gosgeolizdat, 1949.

21. A. K. Rusanov and N. V. Il’yasov, Atlas of flame, arc, and spark spectra of elements, Gostekhizdat (in press).

22. L. N. Il’ichenko, Deciphering spectrograms of ores and minerals, Gosgeolizdat, 1951.

23. T. F. Borovik-Romanova, Spectral-analytical determination of alkali and alkaline-earth elements, Publishing House of the Academy of Sciences of the USSR, 1956.

24. B. A. Sergeev, Geofizika, No. 9—10, 1941.

25. A. P. Solovov, Temporary instructions for metallometric surveying, Gosgeolizdat, 1951.

26. Ya. D. Raikhbaum and E. S. Kostyukova, Izv. AN SSSR, ser. fiz. 18, No. 2, 289 (1954).

27. M. I. Belousova, Izv. AN SSSR, ser. fiz. 18, No. 2, 291 (1954).

28. N. I. Gruzdeva, V. N. Doronina and M. F. Fedorov, Izv. AN SSSR, ser. fiz. 18, No. 2, 289—290 (1954).

29. S. L. Mandel’shtam, Introduction to spectral analysis, Gostekhizdat, 1946.

30. V. S. Mel’chenko, Proceedings of the Siberian Physico-Technical Institute, No. 3, Tomsk (1950).

31. Ya. D. Raikhbaum, Izv. AN SSSR, ser. fiz. 19, No. 1, 70—72 (1955).

32. Ya. D. Raikhbaum, E. S. Kostyukova, A. I. Chernenko and V. D. Malykh, Measurement of the vaporization rate of elements and their compounds in an electric arc, Proceedings of the X All-Union Conference on Spectroscopy, publ. by L’vov State University (in press).

33. A. K. Rusanov, Izv. AN SSSR, ser. fiz. 11, No. 3, 303—309 (1947).

34. A. N. Prilezhaeva, Proceedings of the Siberian Physico-Technical Institute, issue 32, 15—20, Tomsk (1953).

35. O. K. Semenova, Izv. AN SSSR, ser. fiz. 9, No. 6, 715—718 (1945).

36. M. E. Britske and L. N. Varshavskaya, Izv. AN SSSR, ser. fiz. 12, No. 4, 455—456 (1948).

1. B. A. Sergeev, Zav. lab. 13, No. 2, 231—233 (1947); N. V. Buyanov, Izv. AN SSSR, ser. fiz. 12, No. 4, 439—443 (1948).

2. A. N. Broitstein, Izv. AN SSSR, ser. fiz. 19, No. 2, 159—160 (1955).

3. V. V. Nedler, Zav. lab. 5, No. 12, 1469—1474 (1936).

4. A. K. Rusanov and V. G. Khitrov, Zav. lab. 23, No. 2, 175—184 (1957).

5. S. L. Mandel’shtam, DAN SSSR, new series 22, No. 7, 407—410 (1939).

6. V. G. Alekseeva and S. L. Mandel’shtam, ZhETF 17, No. 8, 759—762 (1947); ZhETF 17, No. 7, 765—780 (1947).

7. D. N. Ivanov, Pochvovedenie, No. 7, 423—426 (1949); No. 1, 61—66 (1953).

8. N. S. Popuzhakov, L. I. Kononenko, M. P. Nikonova, ZhAKh 12, No. 1, 10—16 (1957).

9. V. V. Nedler, ZhTF 6, No. 3, 553—560 (1936).

10. V. V. Nedler, F. M. Effenrdiev, Zav. lab. 10, No. 2, 164—167 (1941).

11. Izv. AN SSSR, ser. fiz., No. 2 (1937); ibid., No. 1 (1940); ibid., No. 2—3 (1941); ibid., No. 3 and 6 (1945); ibid., No. 3 and 4 (1947); ibid., No. 4 and 5 (1948); ibid., Nos. 4, 5, and 6 (1950); ibid., Nos. 5 and 6 (1953) and No. 2 (1954); ibid., No. 6 (1954) and Nos. 1 and 2 (1955). Proceedings of the Tenth All-Union Conference on Spectroscopy, published by Lviv State University.