A. N. Dobrovidov and Yu. V. Grdina.

Recrystallization of Iron.

(From the Institute of Applied Physics at T. T. I.).

With 3 drawings and 8 microphotographs.

I. Brief history of the question.

The growth or diminution of the grain of a metal under the influence of heating is called recrystallization. This process proceeds especially well in the case of chemically and technically pure metals, which constitute an assemblage of microscopic crystals with irregular faces (crystallites). In alloys this phenomenon is considerably distorted.

The possibility of controlling the course of recrystallization in the desired direction is extremely important for practice, which deals with cold stamping, accompanied by subsequent annealing, and with other analogous operations. Ignorance of the conditions of recrystallization leads to considerable unnecessary costs in production, owing to rejected articles. In some cases it is not even suspected that folds, nonuniform drawing, etc., during work on a press or drawing bench depend on the coarse-crystalline structure of the metal.

Crystallization that has begun in a liquid metal proceeds from a certain number of centers at a rate depending on the nature of the metal and on the conditions of solidification. The boundaries of the crystals in this process remain plane only as an exception; inevitably they are curved and irregular. In the case of very slow cooling during the solidification of an ingot weighing tens of tons, crystals of several cubic centimeters in volume may form; usually their dimensions do not exceed fractions of a cubic millimeter.

The physical and mechanical properties of a piece of metal are very closely connected with the size of the crystallites of which it consists. The crystallites are in very close contact with one another; they are, as it were, cemented together by an extremely thin layer of binding substance, the nature of which has not yet been determined with sufficient clarity. If one assumes that a pure metal crystallizes out of the molten mass, then this intercrystalline interlayer will contain impurities that were present in the form of emulsions or in suspension. Those fractions of a percent of foreign substances, from which even the purest metal is not free, will give an alloy of very complex composition. According to Tammann, the interlayer constitutes a complex eutectic¹. Usually this interlayer is stronger than the crystallites themselves, and fracture always passes through the crystallites.

Stress above the elastic limit acts upon the regrouping of crystallites in a metallic specimen very differently, depending on the nature of the metal. In the case of iron, zinc, tin, cadmium, and lead, the crystallites first rotate, and only then do simple or twinning slips appear in them. In copper and gold, slips in the crystallites precede their rotation. In aluminum and silver both phenomena are observed simultaneously[^2]. The rotation of crystals and the appearance of cracks in them are easy to notice if a metallic pyramid with mirror-polished faces is compressed. Under the microscope all changes of the surface corresponding to the stress in the given section are very clearly visible.

In view of this duality of deformation, the physical properties of a metal cannot be considered independently of grain size. The numerical value of any modulus has no absolute definiteness unless the grain size of the specimen, or the direction of the crystallographic axes, is specified, if the specimen is a single crystal.

The growth of crystals in a pure metal depends not only on definite temperature conditions, but also on the preceding mechanical working in the cold state. According to Carpenter³), recrystallization will not take place in a metal if it has not been subjected to plastic deformation and has not been heated to a definite temperature. Fraenkel’s⁴) indication is very interesting: cast and very slowly cooled specimens of gold, upon subsequent heating, did not show recrystallization phenomena, whereas the same metal, when cast onto a cold plate, showed grain growth. Apparently, the internal stresses that arose as a result of rapid and nonuniform cooling in the latter case gave an impetus to recrystallization.

It is highly probable that an ingot of cast metal, cooled very slowly, will not be able to change its grain size during subsequent heatings.

Until recently, many investigators studied recrystallization without paying attention to the preliminary mechanical working of the pieces of metal from which the specimens were made. One of the first to call attention to the recrystallization of iron rolled in the cold state was Heyn⁵). However, the influence of the degree of deformation was not precisely taken into account. Specimens of rolled cast iron were heated at temperatures of 417°, 616°, and 960°. The increase in the area of the grain visible on the polished section reached as much as 300% of the initial value. The next work in time concerning the recrystallization of iron was Joistel’s⁶) investigation of the growth of crystals in wire specimens heated in vacuum. The results of these experiments are extremely curious. The curves of the dependence of crystal growth on the duration of heating made it possible to hope that, at certain temperatures, this growth is directly proportional to the heating time and has no limit. But the later work of Benson and Thompson⁷) gave entirely different results. In the latter authors, the curves for the increase in the size of iron crystallites after heating for a known interval of time turned into straight lines parallel to the time axis. The iron, apparently, passed into some normal state and lost the capacity for further recrystallization.

Robin⁸), among others, was interested in the recrystallization of deformed iron. With iron, in comparison with other metals, he obtained variegated results that did not provide a basis for definite conclusions. Chappell⁹) studied the recrystallization of wire with residual deformations imparted by drawing. He considers time and temperature to be the most important factors for grain growth in iron. An increase in the carbon content in iron above 0.1% greatly retards the growth of crystallites, which presents a great danger in the sense of spoiling the article.

This last circumstance compelled many practicing engineers to take up the question of the influence of plastic deformation, temperature, and duration of heating, as well as chemical composition—chiefly of iron articles—on the course of recrystallization in them. Almost simultaneously there appeared purely practical works by Sherry¹⁰) and White¹¹), concerning the causes of grain growth in iron and steel during annealing after cold stamping, and the recrystallization of boiler tubes.

Sherry, studying the phenomenon from the standpoint of a works engineer, gives a number of recommendations on how to avoid dangerous grain growth and the associated lowering of the quality of the metal. Of his conclusions, the most interesting are:

I. Grain growth in steel with a low carbon content takes place within definite limits of deformation and within definite temperature ranges.

II. The temperature range in which grain growth is possible covers 650°–900°.

III. The greatest growth of crystals corresponds to a plastic deformation of 9% (relative compression or elongation).

IV. There is no serious growth of crystals in steel with a carbon content above 0.15%.

Professor A. E. White, in elucidating the causes of changes in boiler tubes of a water-tube boiler with a carbon content of 0.113%, came to the conclusion that the matter involved recrystallization. Among his conclusions the following are noteworthy:

I. In steel the process of recrystallization proceeds extremely rapidly. Above 700° and below the critical point \(A_3\) (910°) it takes place within fractions of a minute.

II. Mild steel that has been subjected to stress above the elastic limit is in stable equilibrium only at ordinary temperature.

Oberhoffer dealt most fully with the question of the influence of the degree of plastic deformation, temperature, and heating time on grain growth in electrolytic and commercial iron, in a series of small but thorough works \(^{12})\ ^{13})\ ^{14})\). He deformed prismatic and cylindrical specimens by compression in a press. The relative decreases in height, in percent of the initial value of the latter, served as an index of the degree of deformation. The deformations lay within the limits from 5 to 75%. On the basis of his first two works Oberhoffer advanced the proposition that 10% deformation corresponds to the maximum growth of crystals. This conclusion was in agreement with later observations by Sherry \(^{15})\) and Pomp \(^{16})\).

Oberhoffer’s latest work \(^{14})\) gives the greatest growth at a deformation of 5% and a heating temperature of 870°. The influence of time on recrystallization, in the author’s opinion, is not great: the entire process is completed within the first half hour.

II. Experiments with Sheet Iron

It was of great interest to us to approach the question of the nature of the intercrystalline interlayer and of the properties of the metal as dependent on the size of the crystals. It became necessary first to study the conditions of recrystallization of the metal. The authors carried out a series of experiments on commercial aluminum, tin, and commercial iron. The account of the last of these constitutes the content of the present work. The study of the recrystallization specifically of iron is important, as was already indicated, also from a purely practical standpoint. The experiments were performed with sheet iron supplied from the Lysva Works, and with iron wire from the Revda Works. We wished to have a material in which recrystallization, which had occurred during annealing as a result of residual deformations from rolling, had proceeded completely. Such a material, according to Carpenter, should not change the grain size upon heating.

Sheets of iron going at the Lysva Works for the manufacture of one-piece drawn ware are pickled in sulfuric acid to remove the scale covering them, washed with water, and annealed. For annealing they are stacked in the wet state in packages weighing about 1000 kg, covered with cast-iron caps, covered with sand, heated in a furnace to 950° and slowly cooled. The operation of annealing and slow cooling lasts more than

[[unclear: beginning of line]] hours. After annealing, the sheets acquire remarkable plasticity and retain the silvery-white color obtained by etching. Recrystallization in sheets treated in this way has apparently proceeded completely.

The analysis of the factory laboratory gives the impurity content: \(C = 0.09\%\), \(P = 0.05\%\), \(Si = 0.10\%\), \(Mn = 0.55\%\). The authors’ analysis for carbon (by the gravimetric method) gave a content of \(C = 0.08\%\).

In view of the presence of mechanical working in the preparation of the specimens (milling), it was necessary to resort to an additional anneal in order to eliminate the influence of the plastic deformations caused by the cutter at the edges of the specimen.

Pure iron has four crystalline states, each of which corresponds to strictly definite temperature limits. Westgren¹, with the aid of the Debye and Scherer method, showed that iron within the so-called \(\alpha\), \(\beta\), and \(\delta\) regions has a space-centered cubic lattice. On the transition of \(\beta\)-iron into \(\gamma\)-iron, the lattice changes to a face-centered one. The transition from \(\alpha\)-iron to \(\beta\) takes place at \(769^\circ\), from \(\beta\) to \(\gamma\) at \(910^\circ\), and from \(\gamma\) to \(\delta\) at \(1425^\circ\). Since the transition of \(\alpha\)-iron into \(\beta\)-iron is not accompanied by a transition of the spatial lattice of the crystallites from one system to another, the crystallites should not be destroyed in this process. It is otherwise with the transformation of \(\beta\)-iron into the \(\gamma\) modification and of \(\gamma\) into \(\delta\): destruction of the crystallites in this case is inevitable. Westgren used in his experiments wire of comparatively small diameter (0.3 mm), and the size of its grain had a great influence on the sharpness of the photograph obtained in the Debye camera. Thus there arose the necessity of destroying the harmful coarse-crystalline structure. Indeed, a rapid transition from the \(\beta\) region into the \(\gamma\)-iron region and from \(\gamma\) into \(\delta\), or conversely (on cooling), destroyed the large crystallites and brought the material into a definite state.

From this follows the impossibility of obtaining very large crystallites of \(\gamma\)- and \(\delta\)-iron; on cooling and passing the critical points these crystallites must be destroyed. The size of the grains of \(\alpha\)- and \(\beta\)-iron formed from the fragments will apparently depend only on the rate of cooling after the transition point of \(\gamma\)-iron into \(\beta\). Therefore all recrystallization phenomena connected with the growth of large crystallites have to be studied within the limits of existence of \(\alpha\)- and \(\beta\)-iron. The transition from \(\alpha\) to \(\beta\) and back does not entail the destruction of grains formed within one of these regions.

The lowest temperature at which grain growth is still observed in iron lies within the limits \(500^\circ\)—\(600^\circ\). It may be thought that the amount of impurities present in the metal has a substantial influence on the level of this temperature.

Recrystallization requires a certain interval of time, the magnitude of which depends on the height of the heating temperature and on the degree of plastic deformation. The influence of time has not been sufficiently clarified, as was indicated in the survey of the works.

Almost all authors dealt with plastic deformation by compression. We decided to stretch specimens of sheet iron. Stretching, with sufficient specimen length, gives a uniform distribution of stresses in the transverse sections of the specimen, which cannot be said of compression. In order to determine at least part of the conditions most favorable for recrystallization, it was decided to heat two series of specimens with deformation from 2 to 20% (over a length of 65 mm) at a temperature practically maximal for \(\alpha\)-iron for 25 hours.

The specimens had the dimensions of the parallel part: length 65 mm, width—\(7.75 \pm 0.05\) mm, and thickness \(0.55 \pm 0.2\) mm. To eliminate traces of working, the specimens were annealed for half an hour at \(890^\circ\) and for two hours at \(740^\circ\); cool—

were placed together with the furnace. Heating was carried out in Heraeus resistance furnaces; the temperature was measured with a previously calibrated Le Chatelier pyrometer. The points \(A^1\), \(A^2\), and \(A^3\) were determined for each grade of iron. There was no great necessity, during annealing, to pass into the region of \(\gamma\)-iron, since the iron had already undergone complete recrystallization at the works, while the mechanical treatment affected only the surface layer in part. Heating of the control (unannealed) specimens and of the annealed but undeformed ones fully confirmed this.

To avoid surface oxidation, the packet of specimens was placed in a porcelain test tube and covered with iron filings from the same iron. The test tube was sealed with clay. This method proved very suitable for all heat-treatment operations to which the specimens were subjected. Their surface did not oxidize at all and after 25 hours of heating retained its former silvery-white color.

After 5 hours of heating, a certain part was cut off from each specimen for preparation of a polished section; this operation was repeated after two intervals of 10 hours each. The polished sections were etched with a 1% solution of nitric acid in water.

The average size of the visible area of the grains on the polished section was determined in the following manner. In a Zeiss projection microscope with arc-lamp illumination, ordinary glass was inserted in place of the ground glass. On transparent paper placed on this glass, an image of the polished section was obtained at the desired magnification. The contour enclosing a certain number of grains was traced by hand with a pencil on the paper. The area of the contour, measured with a planimeter and divided by the number of grains contained in it, evidently gave the area of one grain enlarged by a definite number of times. A similar operation was performed with each polished section 3–5 times. The difference in grain size, except for specimens with very large crystals (of the order of \(3 \times 10^6 \,\mu^2\)), for one and the same polished section did not reach large values. In any case the deviations did not exceed 5–8%.

The results of the experiments are presented in Table 1.

Table 1.

Specimens of sheet iron.

Average grain size in the annealed specimen: \(3800\)—\(4000 \,\mu^2\).

Heating temperature \(740^\circ \pm 5^\circ\).

In photogr. 1, 2, 3, and 4 are shown, enlarged \(4^{1/2}\) times, parts of specimens Nos. 2, 4, 5, and 7.

The specimens were stretched on an Amsler machine, the load being maintained for 1 minute.

Very large crystals grow in the region of plastic deformation at \(6—8\%\). Up to \(12\%\), the appearance of a coarse crystalline structure is possible. A control specimen, cut from the sheet and not subjected to stretching, did not show phenomena of recrystallization even after twenty-five hours of heating. The same may be said also with respect to deformation of \(20\%\).

The influence of time in the region of growth of very large crystals is very great, as is seen in Table 1. Apparently, in this case recrystallization requires very long intervals of time. In the remaining cases it may perhaps be considered that one hour of heating at a favorable temperature is quite sufficient for complete recrystallization.

Control experiments showed that the production of large crystals in this iron is possible in the interval of plastic deformations \(6—7\%\) under the same temperature conditions.

The direct dependence between the degree of plastic deformation and the state of the crystallites of the specimen compels one to think of a close connection between recrystallization and the rotation of crystallites and the subsequent slips within them. It has already been mentioned that iron belongs to the class of metals in which external forces (tension, compression) first cause the protrusion of crystallites on the polished surface of the specimen, and then slips within them. Twinning slips in iron appear only as an exception, whereas, for example, in copper it is impossible to get rid of twins\({}^{18}\).

We made the following attempt to connect the behavior of the metal grains with plastic deformation under the influence of stretching of the specimen. From the sheet iron used for the experiments, a specimen was cut in the form of a trapezoid with corresponding shoulders for gripping by the breaking apparatus. The trapezoidal part was carefully polished and divided by lines parallel to the base and spaced 2 mm apart. To check possible skewing, besides the axis, two more lines parallel to the latter were drawn from the corners of the small base.

After rupture of this specimen, the changed divisions were measured under the microscope and compared with the original ones.

On the polished surface, at a small (about 100) magnification, the crystallites protruding from the surface, the boundary of the rotated grains, and the part that remained unchanged are clearly visible; so are slips inside the crystals. Numerical data are given in Table 2, and microphotographs in photogr. 5, 6, 7, and 8.

Magnification of the microscope: 65 divisions correspond to 2.00 mm.

The first microphotograph represents a place with a clearly expressed, very characteristic bulging of the crystallites; the figure resembles a weakly etched section of iron—the boundaries of individual grains have begun to be outlined. Then they become more clearly revealed. Figure 6 gives an idea of the region of deformations of about \(12\%\). Almost every crystallite has become covered with folds, testifying to strong internal slips. In photograph 7 this phenomenon is presented on a larger scale.

The beginning of individual rotation of the crystallites is first observed at a deformation of \(1—1^{1/2}\%\). Mass rotation of crystallites with sharp marking of their boundaries corresponds to a deformation of \(3—4\%\). Individual random slips are noticeable beginning from \(5\%\). At \(12—14\%\) there is not a single whole crystallite. Thus the circumstance becomes clear that the growth of crystallites must be preceded by a deformation in which there is not yet

Table 2.

slips within the crystallites. Strongly deformed grains will not grow; they will provide the basis for the formation of small crystallites. A polished section of the same iron, etched with an aqueous solution of nitric acid, is given in photograph 8. The clearly expressed network of ferrite grains fully corresponds to the boundaries of the crystallites in photographs 5–6.

III. Experiments with wire and the influence of decarburization.

The experiments with wire had the aim of verifying the indicated propositions for a metal with a somewhat different carbon content. In addition, we wished to relate the heating temperature, the degree of plastic deformation, and the size of the grain obtained after recrystallization; to clarify, as precisely as possible, the conditions for obtaining coarse-crystalline specimens. Then we were interested in the question of the influence of carbon burn-out on the growth of large crystals and on recrystallization in general. Some investigations touching on this question show that the burn-out of impurities, the release of dissolved gases, their diffusion, etc., observed under the influence of heating in the presence of certain favorable circumstances, in itself promotes recrystallization. The combination of these changes with residual deformation may possibly intensify the influence of the latter.

It may be that Joisten ⁶), who heated specimens of iron wire in vacuum, observed recrystallization under the influence of the release of dissolved

in the metal. This circumstance was not taken into account, and very interesting results were obtained, but they were not confirmed by other investigators⁷).

Brearly¹⁹) cites a case of the formation of extraordinarily large ferrite grains from hypoeutectic steel under the influence of decarburization. A steel plate was heated for a very long period of time in an annealing furnace. In this process the surface layer became completely decarburized to a depth of about 20 mm and exhibited an extraordinarily coarse ferritic structure. A similar phenomenon is also observed in the microphotographs in Houdremont’s article, which concerns surface decarburization of steel²⁰).

Austin²¹) encountered the growth of grains under the influence of decarburization of steel in an atmosphere of hydrogen.

A thick interlayer of pearlite, preventing close contact between isolated islands of ferrite, is transformed, during decarburization, into ferrite crystallites. The rearrangement of atoms that occurs in this process leads to the growth of individual grains at the expense of their neighbors.

Wang²²) dissolved various gases in iron and heated specimens treated in this way in the atmosphere of certain gases and in vacuum. He did not obtain definite results, but nevertheless he believes that gas may be present in iron in a free or chemically bound state between the grains and in the form of a solid solution in the crystallites. A change in pressure and a high temperature cause gases to be evolved and to pass between the atoms, thereby facilitating their displacement.

Edwards²³)²⁴) asserts that the burning-out of carbon in steel promotes the growth of crystallites. In a specimen previously decarburized by heating in an atmosphere of hydrogen and then plastically deformed, single crystallites of extraordinarily large size grew. The most suitable relative elongation lay within the limits of 2—4.25% and depended on the initial grain size, which in turn was affected by the method of decarburization and by the duration and temperature of the latter. Edwards’s investigations were intended to clarify the causes of the appearance of a coarse-crystalline structure in the products of one factory of stamped parts.

Our wire specimens had a diameter \(d = 3.2\) mm and a length \(l = 210\) mm. Carbon analysis showed a content \(C = 0.12\%\). During drawing in the cold state, the wire material acquired very considerable residual deformations. The consequences of such considerable cold working could be eliminated only by passing from the \(\beta\) state to the \(\gamma\) state during annealing \((910^\circ)\). The specimens were annealed for half an hour at \(950^\circ\); then the temperature was rapidly lowered to \(750^\circ\) and maintained at this level for about \(1 \tfrac{1}{2}\) hours. After annealing, the wires were stretched on an Amsler machine and heated for 20 hours at a temperature of \(750^\circ\). The measures for protecting the specimens from oxidation were the same as for the flat specimens. The mean grain size was determined on a longitudinal polish. For control, polishes perpendicular to the axis were taken. The data relating to this series of experiments are given in Table 3, drawing 1.

Very large crystallites were obtained at about 4% relative elongation. In the coarse-crystalline specimens the crystallites (3—5) converge on the axis of the wire at a common point. The transverse section of such a wire resembles a cut orange.

Further experiments had as their aim the clarification of the influence of decarburization. For this purpose it was necessary to decarburize specimens from the same wire by heating in a medium that liberates oxygen. The decarburized wires subsequently had to be subjected to normalization, stretching, heat—

Table 3.

heating. We wished to prepare specimens from single crystals for further investigations. A continuation of experiments similar to those presented in Table 3 showed that it is apparently impossible to prepare such specimens from this wire.

The wires were carefully sprinkled with finely ground scale from under the hammer (hammer scale), enclosed in an iron tube, and hermetically sealed. The tube was held for 6 hours at \(1000^\circ\) and cooled slowly; the fall in temperature from \(1000^\circ\) to \(700^\circ\) lasted two hours. After this operation the specimens were covered on the surface with a thin layer of gray oxide. Microscopic examination of sections cut in different directions showed a considerable decrease in the pearlite inclusions. The sections consisted of ferrite that was rare in purity and uniformity. Analysis gave a carbon content of about \(0.09\%\).

Although after such treatment the material could be considered quite normalized, local stresses might have arisen during removal of the scale. Therefore, before further treatment, the specimens were subjected to annealing at \(950^\circ\) and \(750^\circ\), as in the preceding experiments. To obtain a complete picture of the conditions under which grain growth is possible, 4 series of 5 specimens each were prepared. Each series received a residual tensile deformation of 2, 4, 6, 8, and \(10\%\) and was heated at a definite temperature. Four temperatures were chosen, the most interesting for practice: \(600^\circ\), \(700^\circ\), \(800^\circ\), and \(840^\circ\). Below \(600^\circ\) recrystallization proceeds extremely slowly; above \(850^\circ\) there is danger of entering the \(\gamma\)-iron region. The data of these experiments are collected in Table 4 and presented in Fig. 2.

Table 4.

To determine the influence of repeated plastic deformation and heating on already recrystallized material, a series of wires with an elongation of 9% was heated for 10 hours at a temperature of 800°—850°. The specimens were then given a deformation of 3.5—7.0% and were heated a second time under the same temperature conditions. No growth of very large crystals was observed in this case. In the same furnace in which the specimens with 9% deformation were heated, there was a series of wires with an elongation of 4—6½%; some of the latter (5—6%) showed the growth of very large crystals. See Table 5 and Fig. 3.

Table 5.

It is difficult to draw any conclusions from this experiment, although there is a view that a series of deformations alternating with heating is somehow cumulative¹). Experiments of this kind, carried out by the authors on commercial aluminum, showed that complete recrystallization, provided it did not result in the formation of abnormally large crystals (of the order of 10–200 mm²), restores to the metal its former properties. The influence of each subsequent plastic deformation in this case was entirely independent of the preceding mechanical treatment. The time required for complete recrystallization of the specimens varied in this case within the limits of 70–300 hours. It is quite possible that the duration of recrystallization here is of decisive importance.

IV. Conclusions and deductions.

On the basis of the experiments set forth above, as well as of the literature data available to us, it is possible to form some idea of the factors influencing the recrystallization of iron with a small carbon content.

Good soft iron, commercially available, has approximately the following composition: Fe—99.4%, C—0.1%, P—0.05%, S—0.05%, Si—0.1% and Mn—0.3%. It is usually said that the polished section of such iron consists of ferrite grains separated by an interlayer.

In annealed material, carbon at such a low content can occur only in the form of pearlite; in extremely rare cases an isolated existence of pearlite in the form of inclusions in homogeneous ferrite grains is possible. 0.95% carbon gives 100% pearlite containing 14.3% cementite. In our case, 0.1% C will be capable of forming about 10% by weight of pearlite, if, with slight error, its specific gravity is taken as equal to the specific gravity of pure iron. All the pearlite will enter the intercrystalline interlayer.

Phosphorus, entering carbon steel, forms with ferrite and carbide a ternary eutectic.

In the presence of manganese, sulfur gives a compound of the type MnS in the form of separate isolated crystals²⁵).

Manganese forms with iron a continuous series of solid solutions, manganese dissolving in α-, β-, and γ-iron²⁶).

The quaternary alloy Fe, Mn, Si, and C has the same structure as the ternary alloy Fe, Mn, and C. Silicon with iron gives a solid solution.

Thus it may be considered that our iron consists of a solid solution of Mn and Si in $\alpha$-iron. The interlayer is a complex eutectic with inclusions of MnS and an admixture of a certain amount of slags. A large part of the interlayer is occupied by pearlite. Lowering the carbon content by removing it through the decarburization process reduces the amount of interlayer.

Clarification of the nature of this substance, which surrounds and separates the crystallites from one another, is extremely important for the correct explanation of many phenomena. This is especially interesting in the case of chemically pure metals, when the interlayer seems to have exactly the same composition as the main mass of crystallites. As is known, in some cases it is possible to dissolve the interlayer without altering the crystallites (lead) and, conversely, to destroy the crystallites without damaging or altering the interlayer (cadmium). Study of the isolated interlayer may shed some light on its nature. The Debye and Scherer method may be of great assistance here.

If interlayers did not exist at all, then two closely touching crystallites at a high temperature, which produces great mobility of atoms, could not remain in equilibrium. In that case the atoms would begin to pass from one crystal to the other. This transition is possible only in one direction and depends on the magnitude of the surface energy at the surface of separation. The crystal with the greater surface energy at that place will be absorbed.

Plastic deformation probably brings the crystallites into contact, and the center of recrystallization arises where the ratio of the surface energies of the contacting crystallites is most favorable. In Tammann’s words: “a surface more densely occupied by atoms compels the less densely occupied one to assume its position.”

It seems to us that from all the foregoing the following follows:

1. Heating cannot cause recrystallization if the metal has not undergone plastic deformation, no matter of what kind. A metal in which, under the action of heating, no grain growth is observed may be called a “metal in the normal state.” In all experiments on recrystallization, normalization of the specimens is of decisive importance.

2. The recrystallization temperature, the degree of plastic deformation, and the size of the grain formed as a result of complete recrystallization are connected by a definite dependence. The character of the latter is influenced by the initial grain size of the annealed metal, the degree of its purity, the kind of impurities, etc. Thus it is impossible to give an exact recipe for preparing single-crystalline specimens, especially since impurities have a very great significance. In the case of mild steel, the carbon content has the greatest influence on grain growth.

3. The purer the metal, i.e., the thinner the intercrystalline interlayer, the smaller the deformation required for the beginning of crystallite growth.

4. A change in the composition of the recrystallizing specimen during the very process of grain growth influences the latter. In the case of carbon burn-out, the ferrite freed from pearlite takes part in the growth of crystallites. Combined and dissolved gases, being released, promote greater mobility of atoms and grain growth.

5. The time of complete recrystallization varies very greatly and depends on the magnitude of the residual deformation and on the temperature. It is small in the case of deformation far removed from the critical deformation—corresponding to unlimited grain growth—and, conversely, very large (tens of hours) in the region of growth of large crystallites.

6. Mechanical working by pressure (stamping), combined with subsequent annealing, may ultimately produce a coarse-grained structure of the articles.

in the case where the plastic deformation (tension, compression) lies within the range of 2–10%, and the heating temperature is in the region of $\alpha$- and $\beta$-iron. The growth of abnormally large grains can be combated by raising the annealing temperature above the transition point of $\beta$-iron into $\gamma$ (910°) and by rapid subsequent cooling. Steel with a carbon content above 0.2% is less prone to the formation of a coarse-crystalline structure.

1. Very large iron crystals can be grown in material containing less than 0.1 carbon. For this, deformation within the range of 4–6% and heating near the middle part of the $\beta$-iron region are necessary.

The experiments were carried out jointly by both authors in the metallographic laboratory of the Tomsk Technological Institute; the literary treatment belongs to A. N. Dobrovidov.

In conclusion, we consider it a pleasant duty to express our deep gratitude to T. I. Tikhonov and G. V. Trapeznikov for permission to use many instruments of their laboratory.

Literary sources to which references are made in the work.

1. G. Tammann, Lehrbuch d. Metallographie, Leipzig, 1923.
2. B. B. Beinbürg, Journal of the Russian Phys.-Chem. Society, phys. dept., vol. 2, 1913.
3. Prof. Carpenter, The Journ. of the Iron a. St. Inst., vol. CVII, No. 1, 1923.
4. W. Fraenkel, Zeitschr. f. anorg. u. allg. Chemie, s. 126, 1923.
5. F. Heyn, Zeitschr. d. Vereins deutsch. Ing., h. 44, s. 433, 1900.
6. A. Joisten, Metallurgie, s. 156, 1910.
7. L. F. Benson a. F. C. Tompson, The Journ. of the Iron a. St. Inst., vol. CVII, No. 1, 1923.
8. F. Robin, Ferrum, h. 4, s. 120, 1914.
9. C. Chappell, Stahl u. Eisen, h. 20, s. 847, 1914.
10. R. H. Sherry, The Iron Age, p. 76, 1916.
11. A. E. White a. H. F. Wood, The Iron Age, p. 20, 1916.
12. P. Oberhoffer u. W. Oertel, Stahl u. Eisen, h. 37, s. 1061, 1919.
13. P. Oberhoffer u. H. Jungbluth, Stahl u. Eisen, h. 40, s. 1513, 1922.
14. P. Oberhoffer u. W. Oertel, Stahl u. Eisen, h. 20, s. 560, 1924.
15. R. H. Sherry, The Iron Age, p. 76, 1916.
16. Pomp, Stahl u. Eisen, h. 33, s. 1261, 1920.
17. A. Westgren a. Phragmén, The Journ. of the Iron a. St. Inst., vol. CV, No. 1, 1922.
18. Ewing a. Rosenhain, Chem. News, p. 265, 1899.
19. H. Brearly, Die Wärmebehandlung der Werkzeugstähle, Berlin, 1913.
20. E. Houdremont u. H. Kallen, Stahl u. Eisen, h. 16, s. 587, 1925.
21. R. Austin, The Journ. of the Iron a. St. Inst., No. 1, 1922.
22. Wang, The Journ. of the Iron a. St. Inst., No. 2, 1923.
23. C. A. Edwards a. L. B. Pfeil, The Journ. of the Iron a. St. Inst., No. 2, 1923.
24. C. A. Edwards u. L. B. Pfeil, Stahl u. Eisen, h. 38, s. 1151, 1924.
25. M. T. Yevangulov, Alloys, St. Petersburg, 1911.
26. W. Guertler, Metallographie, Berlin, 1913.

Figure 1. (To Table 3).

![Graph showing stress and hardness versus elongation. The left vertical axis is labeled “Stress kg/mm²” with marked values 23, 27, 31, and 35. The right vertical axis is labeled “Hardness Brinell, 1000 kg” with marked values 20, 40, 60, 100, and 120. The horizontal axis is labeled “Elongation” and runs from 1% to 11%.]

Drawing 2. (To Table 4).

Grain size in \(1000\,\mu^2\)

Plastic deformation in percent

Temperature

Drawing 3. (To Table 5).

[Graph labels visible on the page:]

• Stress, kg/mm²
• Grain size (number of crystallites) in mm²
• % elongation.

Photo 1. ×4½

Photo 2.

Photo 3.

Photo 4.

Photo 5. ×100

Photo 6. ×100

Photo 7. ×400.

Photo 8. ×100

A. N. Dobrovidov and Yu. V. Grdina. Recrystallization of Iron.

1. Westgren. ↩↩