A. N. Dobrovidov and Yu. V. Gryadina.

Recrystallization of Iron Deformed in the Hot State.

With 3 drawings in the article.

I. General considerations.

In the paper “Recrystallization of Iron”¹, on the basis of a number of experiments, we considered it possible to draw certain conclusions which, in one way or another, elucidate the essence of the phenomenon in the recrystallization of metals.

While sharing Tammann’s idea of the dependence of grain growth on the contact of the surfaces of crystals, we nevertheless could not give an exhaustive explanation of the causes by virtue of which one grain begins to grow at the expense of another². We shall try to clarify the cause of grain growth in a metal, proceeding from the following basic, as though firmly established, facts:

1) heating by itself cannot be the cause of recrystallization if the metal has not previously been plastically deformed in the cold state;

2) the recrystallization temperature, the degree of plastic change of shape, and the size of the grain after heating are connected by a quite definite dependence;

3) the purer the metal, the thinner the intercrystalline layer—the smaller the deformation required to cause crystal growth upon subsequent heating.

A crystallite turned as a result of deformation sometimes comes into close contact with its neighbor, without an intercrystalline layer that has been partially destroyed at the place of contact. Of two spatial lattices that have come into contact, one will tend to increase its extent at the expense of the other. Heating, by causing the thermal motion of atoms, thereby gives an impulse to the departure of individual atoms from the lattice, and the surface with the least surface energy will grow at the expense of the surface most sparsely populated by atoms—that is, the surface with the greatest surface energy.

Such a theory explains the phenomena of recrystallization quite satisfactorily from a purely formal point of view, while leaving unanswered the sometimes observed growth of one of the twinned crystals at the expense of its twin and the rise of the recrystallization temperature with a decrease in the degree of plastic deformation. We have been able to observe the growth of twins within a crystal during recrystallization in studying the recrystallization of metals that readily give twinned formations—silver and copper.

Small deformation, causing rotation of the crystallites without shears within them (iron), may also fail to produce destruction of the interlayer, and in such a case, according to Tammann, there should be no grain growth.

If, however, a very small deformation, as must be assumed, causes a disruption of the integrity of the interlayer, then the connection between the degree of deformation and the temperature at which grain growth begins is not clear.

Once crystallites have come into close contact, then in all cases, regardless of the deformation—provided only that it has not partially destroyed them—recrystallization must always be observed at a temperature definite for the given metal. But plastic deformation, producing in the metal the phenomena of work hardening—mechanical hardening—and causing the grains of the metal to rotate in confined space, thereby produces elastic deformations in them. At the same time, at the boundaries of individual crystallites, as a result of their friction against one another during motion, an additional amorphous intercrystalline layer may be formed, consisting of atoms that have become disorganized and have left the lattice.

And it seems to us that the foregoing account of recrystallization must be supplemented by taking these possibilities into consideration.

The potential energy of a crystal is a function of the interatomic distance. Under normal conditions this distance corresponds to some minimum energy of the crystal. In elastic deformation of the spatial lattice of the crystal, the condition of minimum potential energy is violated, and, owing to the tendency of the crystal to arrange its atoms so that the normal position is restored, the state of the crystal after deformation will not be stable; such a state of unstable equilibrium must be observed, to varying degrees, in all the individual grains of the specimen. In some crystallites, owing to their more favorable orientation and to more favorable deformations during the change in the shape of the specimen, these stresses in the lattice will be small, and in isolated cases even equal to zero.

In the specimen there are crystallites whose irregular outlines form faces with the simplest crystallographic indices—with the densest arrangement of atoms. Let us suppose that a grain of this kind, slightly deformed and in contact by such a face with a strongly deformed crystallite and with faces possessing greater surface energy. During etching, the first grain will grow at the expense of the second.

If the grains are in contact through an amorphous interlayer and not directly, then recrystallization is likewise possible, since the crystallite, under the influence of heating, will tend to place into its lattice the disorganized atoms of the interlayer.

The atoms of an elastically deformed lattice, under the action of thermal motion and favorable conditions (the proximity of a nearly undeformed lattice), will take part in such a rearrangement all the sooner, the more strongly the lattice is deformed. That is, the more strongly the specimen is compressed or stretched, the lower the temperature required for the onset of recrystallization. When the most suitable temperature and deformation coincide, recrystallization proceeds most successfully.

If the elastic deformation of the lattice is in some way eliminated, recrystallization may not occur. When such “recovery” of the metal takes place more rapidly than the growth of crystallites at the expense of the amorphous interlayer or neighboring grains, the phenomena of recrystallization will not be detected at all.

If two crystallites are in such close contact that the distance between the contacting surfaces is of the order of \(10^{-8}\) cm, then recrystallization may begin at a very low temperature. For grain growth it is important

the absence, at the place of contact, of an intercrystalline layer, which at elevated temperature may itself enter into the lattice and thereby ensure the close contact, on the order of atomic distances, of two grains.

Thus it seems to us that what is important is not only the preliminary destruction of the layer (in the case of pure metals), but also the elastic deformation of the spatial lattice of the individual crystallites.

The role of the layer, of residual and elastic deformations, had to be clarified by means of corresponding experiments. If specimens of a material with a well-expressed capacity for recrystallization are slowly stretched in the hot state, then it may be assumed that in crystallites which have rotated and come into contact there will be no internal stresses, since under such conditions the “recovery” of the metal can proceed parallel with the deformation. That is, no grain growth will then be observed.

On the other hand, a heated specimen stretched almost to the elastic limit should recrystallize if destruction of the layer is not necessary for grain growth.

II. Arrangement of the experiments.

For the experiments an iron wire of diameter \(d = 2.5\) mm was taken, with a carbon content of about 0.07%. Specimens of this wire, after mechanical treatment, were normalized by heating for 1 hour at \(1000^\circ\) without access of air, followed by slow cooling (approximately, at first, \(100^\circ\) in 10 minutes). The normalized specimens had a very uniform grain size, with mean area \(Q \sim 2100 \mu^2\), and on rupture showed an elongation \(\tau = 33.5\%\) (measured over a length of 150 mm). The best grain growth in this wire was observed, as preliminary experiments showed, at a deformation of 6% (tension) and an annealing temperature of \(750^\circ\).

A specimen of such wire was heated in a specially adapted electric furnace with a platinum resistance (Heraeus) to \(750^\circ\), and at this temperature was loaded with a weight not yet producing residual deformation. Under stress and at the same temperature (\(750^\circ\)) the specimen was held for 2 hours, cooled, ground on the diametral plane, and etched with a 1% solution of nitric acid in water. Then the grain size was determined under the microscope by the usual method. Although a very large number of experiments of this kind was carried out, no noticeable changes in the grain were detected: deviations from the mean size of the crystallites on the polished section (\(2100 \mu^2\)) were observed in either direction by no more than 8%.

The apparatus for carrying out these experiments was as follows (Fig. 1). To a massive metal table there was fastened a frame made of bicycle tubes. The Heraeus furnace (H) was placed on the table so that its axis coincided with the plane of the frame; a specimen, longer than the inner protective tube of the furnace, was attached by means of clamps with pointers (D) to the frame on one side and to a flexible steel cable (N), thrown over a pulley (S), on the other. The cable ended in an aluminum platform (G) with a stop, on which the weights were placed.

The length of the heating tube of the furnace was 300 mm, while the region with uniform heating reached a length of 150 mm. Two marks were made on the specimen with a fine cutter at a distance of 150 mm from one another, and the distance

the distance between them was measured with the aid of a dividing machine. To prevent oxidation of the specimen, its middle part was protected by a porcelain tube freely filled with iron filings. Since the pointers (D) could not give great accuracy, it was necessary to work by gradually selecting loads, and only after measuring, on the dividing machine, the length of the heated part of the specimen did we make sure that the elastic limit corresponding to the given temperature had not yet been exceeded.

The same apparatus also served for the subsequent experiments. Having established that heating during elastic deformation does not produce recrystallization phenomena, we carried out a series of experiments with plastic deformation of the same wire in the hot (750°) state. In order, as far as possible, to obtain all deformations within the given limits, it was decided to work with specimens of variable cross-section.

The wire was ground down on two sides over a distance of 60 mm, so that at the first mark the section represented a full circumference ($d = 2.5$ mm), while at the second it was a circle without two segments with parallel chords (Fig. 2). Before and after the experiment the intervals $ab$ and $bc$, initially equal to approximately 30 mm, were measured. The part with variable section was protected from oxidation in the same way as in the preceding experiment.

The furnace was heated together with the specimen to 750°. Then the platform was loaded gradually, with the calculation that the plastic deformation of the specimen would, uniformly over a definite interval of time ($10 \pm 1$ min.), reach the required magnitude. This manner of conducting the experiment at first presented very great difficulties, but by a series of repeated experiments we learned to give the specimen the required elongation within the prescribed time. After deformation the specimen was held in the furnace for 2 hours. In Table I one series of such observations is given; it should be noted that no appreciable increase or decrease in grain size was observed in any specimen. Even individual grains of any abnormal size were not found. In addition to grinding along the diametral plane, control polished sections of transverse sections were taken.

III. Discussion of Results.

On the basis of what has been set forth, it may be considered quite definite that heating specimens of soft iron stretched to the elastic limit, as well as subsequent heating of specimens slowly deformed in the hot state, does not produce grain growth. Hanemann and Lucke¹) deformed by impact specimens of various metals and, at certain degrees of deformation, obtained sharply expressed recrystallization phenomena.

Their experiments with soft iron ($C—0.06\%$) consisted in the following. Normalized cylindrical specimens were heated in an electric furnace and, with the aid of preheated tongs, were placed on an anvil. A ram weighing 10 kg, falling from a height of 0–100 cm, upset the specimen; although the time of deformation was not determined by the authors, it may be assumed that it was measured in very small fractions of a second. Grain measurements were made in the places of greatest deformation, at the summits of the “slip cones,” which did not appear in the specimen at small deformations.

Deformation at 650° does not produce recrystallization and, at very great reductions in the height of the specimen, leads only to fragmentation of the grains.

¹) N. Hanemann u. F. Lucke, Stahl u. Eisen 1925, S. 1117.

TABLE I.

Fig. 1.

Visible inscriptions and labels: A; 110 v; D; P; C; H; S; G.

To the article by A. I. Probudov and Yu. V. Gordin: “Recrystallization of Deformed Iron.”

Fig. 2.

Visible inscriptions and dimensions:

• deformation: 30%
• \(l \sim 30\)
• \(l_0 \sim 90\)
• \(l \sim 400\)
• diameter: \(\varnothing 8\) [[unclear: unit]]

Fig. 3.

Vertical axis: Grain size, \(\mu^2\)

Horizontal axis: Deformation, %

Visible scale markings:

• vertical: \(1000\), \(10000\), \(20000\)
• horizontal: \(10\), \(20\), \(30\%\)

The higher the heating, the smaller the deformation that produces grain growth, and above \(1100^\circ\) practically zero deformation, followed by slow cooling, already produces growth of crystals. In all cases the soft iron, after upsetting, was cooled in the furnace, so that the fall in temperature by \(600^\circ\) lasted about 1 hour.

Table II presents the change of grain in iron at a deformation temperature of \(750^\circ\), and in Figure 3 the same data are presented graphically. If the data of Table I are plotted on this scale, the curve of grain size is represented by a straight line.

TABLE II.

Hanemann’s data \(^{2}\)

In the conclusions to his work, Hanemann says that, in designing presses for hot stamping, it is necessary to take these experiments into account and not to give the metal being stamped a deformation of less than 10%.

The growth of crystals in cases of very high deformation rate may be explained as follows. The deformation rate is so great that the crystallites, without being deformed, rotate and behave in the same way as at ordinary room temperature. The grains of metal in the heated state may be compared with ordinary pitch: if deformation is caused slowly, by the gradual action of a force, then the pitch flows like a liquid; under an impact, on the contrary, it breaks into pieces like a solid body. It is also quite possible that the “rest” of the crystals, noticeable in low-melting metals (tin) at ordinary temperature, has a certain significance in this case. In order to “rest,” that is, somehow to free itself from elastic deformations by expansion of boundaries, rearrangement of the lattice, etc., a crystal requires a certain interval of time. If the rate of rest is greater than the rate of recrystallization, then the latter will not occur at all. On the basis of the foregoing, we think that the conclusions given in the previous work must be supplemented by the following:

1. Slow deformation of iron in the heated state and subsequent annealing do not produce recrystallization phenomena.

2. A rapid (impact) change of shape causes growth or diminution of grains.

3. A necessary condition for grain growth is partial destruction of the interlayer and, as a consequence, close contact of the crystallites and elastic deformation of the spatial lattice of the crystallites.

1. A. N. Dobrovidov and Yu. V. Gryadina, “Recrystallization of Iron.” Izvestiya SPb. Tekh. In-ta, 1926, vol. 1 (47). ↩

2. The literature on recrystallization is cited in 1, and also in the article by A. N. Dobrovidov, Vestnik Inzhenerov, 1926, Nos. 7–8. ↩