N. V. Gutovsky.

Toward the Theory of the System:

Iron–Carbon.

Part I.

Melting and solidification of iron–carbon alloys

(with 23 figures in the text, 6 phototypes and 3 lithographic tables).

Tomsk.

Siberian Partnership of the Printing Business, Dvoryanskaya St. and Yamsk. Lane, own building.

1914.

CONTENTS

Part I

Melting and Solidification of Iron–Carbon Alloys

	Page
Contents	III
From the Author	V
I. INTRODUCTION (processes of solidification and melting)	1
A. Solidification	2
B. Melting	4
II. PHASE DIAGRAM OF IRON–CARBON ALLOYS	7
III. EXPERIMENTAL VERIFICATION OF THE PHASE DIAGRAM OF IRON–CARBON ALLOYS	13
A. Determination of the points of beginning and end of solidification of alloys	13
I. Cooling curve	13
1) Cause of the absence of critical points of the beginning of solidification of hypereutectic alloys	16
2) Absence of a break in the cooling curves for determining the end of solidification of hypoeutectic alloys	18
II. Quenching method	22
1) Quenching experiments	24
2) Macroscopic examination of quenched specimens	29
3) Phase diagram of iron–carbon alloys	31
B. Cementation experiments	37
IV. THEORY OF MELTING AND SOLIDIFICATION OF IRON–CARBON ALLOYS	47
a) Heyn’s theory	47
b) Goerens’s theory	50
c) Experimental investigation of the process of melting of cast irons	52
d) Theory of melting and solidification of cast irons	58
1) Melting of gray cast iron	58
2) Melting of white cast iron	62
3) Solidification processes	63
V. CONCLUSION	65
Table of microphotographs	67
List of literature sources	68

From the Author

The present work was carried out at the Metallurgical Institute of the Aachen Polytechnic in 1908 and 1909. A preliminary communication about it was published in the journal Metallurgie (1909, pp. 731 and 737) at the end of 1909. Owing to circumstances beyond the author’s control, the full Russian text is appearing in print after a considerable interval of time since the publication of the work in Germany. During this time, a number of works devoted to the theory of iron–carbon alloys have appeared in the specialized technical literature. These works partly confirm and partly criticize the author’s data, while also providing abundant new material that sheds light on the complex iron–carbon system. The author considered it more proper to publish the work he had completed in the form in which it was finished in 1909, and to add to it an appendix (Part II), which should contain a review of the new experimental material and a critical evaluation of the new data.

On the Theory of the Iron–Carbon System.

N. V. Gutovsky.

PART I.

Melting and Solidification of Iron–Carbon Alloys.

I. Introduction.

At the present time it may be considered established facts: 1) that iron with carbon forms a chemical compound: iron carbide, \(Fe_3C\), or the so-called cementite, and 2) that the allotropic modification: \(\gamma\)-iron is capable of dissolving this carbide in itself in quantities determined by the temperature, forming a solid solution, or the so-called mixed crystals. These two characteristic features of the iron–carbon system determine the type of the equilibrium diagram of iron–carbon alloys. According to the classification of B. Roozeboom, elements which form with one another a chemical compound, soluble in the liquid phase and completely insoluble in the solid, follow a diagram of type Va\(^1\)), while elements forming an incomplete series of mixed crystals belong to a diagram of type V. The diagram of iron–carbon alloys, combining in itself the characteristics of types V and Va, is naturally a combination of these two types and therefore must have the form shown in Fig. 1.

Fig. 1.

Up to the present time\(^2\) it has not been possible to obtain pure alloys of iron with carbon with a cont—

\(^1\) See II, p. 20.

\(^2\) Refers to 1909. Alloys with a large amount of carbon will be discussed in Part II.

with a carbon content greater than \(6.66\%\); therefore, in considering questions concerning iron–carbon alloys, one in fact has to deal only with a small region of the complete diagram: \(Fe + C\)—the region within the limits from \(0\%\) to \(6.66\%\) carbon.

The process of solidification and melting of a system of two elements, obeying a diagram of the type of Figure 1, is not complicated and proceeds as follows.

A. Solidification. The beginning of solidification of all alloys from \(A\) to \(D\) occurs at temperatures determined by the intersection of the ordinate of the alloy composition with the lines \(AB\) or \(BD\), called the liquidus. The temperature of the end of solidification is determined by the intersection of the same composition ordinates with the lines \(AE\) and \(EK\), called the solidus. Thus, the process of solidification takes place within a definite temperature interval, different for alloys of different composition, and consists in the fact that, from the liquid alloy, as the temperature is lowered, the solidifying element crystallizes out, increasing quantitatively and changing qualitatively, i.e. in composition. For a more detailed elucidation of the solidification process, let us trace its course in four alloys, designated by the Roman numerals I, II, III, IV. The solidification of alloys with a content of element \(D\) from \(0\%\) to \(Oe\) occurs in the following manner: alloy I begins to solidify at point 1, i.e. at temperature \(T_1\). At this temperature, “mixed crystals” of composition \(Oa_1\) begin to separate from it, representing a solid solution of \(Oa_1\%\) of element \(D\) in \((100 - Oa_1)\%\) of element \(A\). As the temperature is lowered, new portions of mixed crystals separate out, the composition of which changes, following the solidus \(AE\). Owing to the separation of these mixed crystals, the composition of the liquid part (phase) of the alloy changes, moving to the right along the liquidus \(AB\); thus, at some temperature \(T_2\) the alloy, under the condition of sufficiently slow cooling, will consist of mixed crystals of composition \(Oa\) and a liquid phase of composition \(Oc\). At this temperature (and atmospheric pressure), according to Gibbs’s phase rule¹, the alloy under consideration represents a univariant system:

\[ F = K - P + 1 = 2 - 2 + 1 = 1, \]

which, for a given composition of the alloy, can be in equilibrium only at one definite temperature. Such equilibrium is attained under the condition of sufficiently slow cooling, since

are mixed crystals, which separated out in the temperature interval \(T_1—T_2\) and differ in composition from the crystals of point 4, containing \(Oa\%\) of element \(D\), must have sufficient time to equalize their composition with the crystals of point 4 at the expense of the liquid part of the alloy. Upon subsequent cooling from point 2 downward, further separation of mixed crystals will occur until the alloy reaches the temperature \(T_3\) (point 3), at which the last drops of liquid solidify and the entire alloy will constitute a homogeneous solid body—a solid solution of \(Ob\%\) of element \(D\) in element \(A\). The quantity of mixed crystals and the quantity of the liquid part of the alloy that are in equilibrium at some given temperature, for example \(T_2\), is determined from the following relation established in the theory of alloys:

\[ \frac{\text{quantity of mixed cryst.}}{\text{quantity of liquid phase}}=\frac{2.5}{4.2}. \]

This relation shows that the quantity of mixed crystals (in 100 parts of alloy), determined by the horizontal segment between the ordinate of the alloy and the liquidus line, increases during the process of solidification from \(0\%\) (point 1) to \(100\%\) (point 3). The quantity of liquid phase correspondingly decreases from \(100\%\) (point 1) to \(0\%\) at point 3.

In an entirely similar manner proceeds the process of solidification also of alloys of composition from \(Oe\%\) to \(Ol\%\) of element \(D\). The difference from the case just considered will take place only at the final moment of solidification, when the alloy consists of mixed crystals of the limiting composition \(Oe\) (point \(E\)) and a liquid phase of eutectic composition \(Ol\) in the ratio:

\[ \frac{\text{mixed crystals}}{\text{eutectic liquid}}=\frac{10\,B}{E\,10} \]

At the eutectic temperature \(T_c\) the liquid eutectic part of the alloy, as a nonvariant system, solidifies at constant temperature, and alloy II, after the completion of the solidification process at temperature \(T_c\), will represent a mixture of solid eutectic and mixed crystals of limiting composition.

The line \(SE\) in Fig. 1 shows that the solubility of element \(D\) in element \(A\) (having reached its maximum at \(T_c\)) decreases as the temperature is lowered from \(T_c\), and therefore the mixed crystals of all alloys of composition from \(0\%\) to \(Oe\%\) will separate out (in the already solidified state) element \(D\), beginning with the temperature that is determined by the point of intersection of the ordinate of their composition with

along the line $SE$; mixed crystals of alloys of composition from $Oe$ to $O'$ will separate out the element $D$, beginning at the temperature $T_e$.

The eutectic alloy III, representing at the temperature $T_e$ a nonvariant system:

\[ F = K - P + 1 = 2 - 3 + 1 = 0, \]

solidifies at the constant temperature $T_e$.

Finally, alloy IV will solidify in the same way as alloy II, with only the difference that, at the temperature at which solidification begins (point 11), the element $D$ will crystallize out from it, and therefore in the solidified state the alloy will represent a mixture of crystals of the element $D$ and the eutectic $B$ in the ratio:

\[ \frac{\text{crystals } D}{\text{eutectic } B} = \frac{\overline{B\,12}}{\overline{12\,K}}. \]

B. Melting. The melting of the system of elements under consideration takes place in the order reverse to that just examined for the process of solidification. Alloys begin to melt when their temperature reaches the lines $AE$ and $EK$. The melting process ends on the lines $AB$ and $BD$. Thus, melting occurs within a definite interval of temperatures. In alloys not containing eutectic, the mixed crystals melt, separating from themselves a liquid phase, the composition of which, and likewise the composition of the crystals themselves, changes with increasing temperature, following the liquidus and solidus lines exactly as was shown for the process of solidification. Thus, for example, alloy I will begin to melt at point 3. At point 2 (at the temperature $T_2$) it will consist of mixed crystals of composition $Oa$ and a liquid phase of composition $Oc$. At point 1 the melting process will end. In alloys containing eutectic (alloys of type II, IV), the eutectic first melts at the constant temperature of the eutectic, and then mixed crystals (or the element $D$) begin to dissolve in it. The melting process takes place over a temperature interval determined for each alloy, with the composition of the liquid phase and of the mixed crystals changing.

From consideration of the processes of solidification and melting it follows that all alloys begin to solidify at different temperatures, depending on the composition of the alloy, lying on the lines $AB$ and $BD$. Solidification ends at different temperatures (solidus $AE$) in alloys not containing eutectic, and at the constant temperature $T_e$ (line $EK$) in alloys containing eutectic. Conversely, the process of melting begins at different tempera-

temperatures (solidus \(AE\)) in alloys without a eutectic and at the constant eutectic temperature \(T_e\) in alloys with a eutectic. The end of melting of the various alloys occurs at different temperatures (the lines \(AB\) and \(BD\)).

The course of the processes of melting and solidification set forth above is ideal and can take place only when: 1) the alloy is heated or cooled sufficiently slowly for the processes of dissolution and diffusion to have time to occur in their full extent; 2) measures have been taken to prevent the phenomenon of supercooling; and 3) the elements entering into the composition of the alloy, such as, for example, the chemical compound \(D\), do not undergo decomposition at the high temperatures of the period of solidification or melting.

I. The process of solidification usually deviates from the ideal course considered, in the sense that the mixed crystals that separate do not have time, owing to insufficiently slow cooling, to equalize their composition with the crystals of subsequent precipitations; therefore: 1) the solidified solid solution is not perfectly homogeneous; and 2) the change in the amount of the liquid phase does not follow the conditions of the equilibrium state of the system, i.e., at a given temperature more of it remains than is required by the theoretical diagram. However, by means of very slow cooling or appropriate annealing, the indicated phenomenon of inhomogeneity can be eliminated to such an extent that the alloy obtained (the solid solution) may practically be regarded as perfectly homogeneous.

II. The phenomena of supercooling, generally speaking, do not obscure the picture of the process of solidification, and they can be eliminated by slow cooling and by inoculation (Impfen) of crystals.

III. The third of the indicated conditions has a very serious role for the correct course of the processes of solidification and melting: the instability of a chemical compound at high temperatures. Such instability lies in the nature of the elements constituting the alloy; under certain conditions it cannot be prevented: the chemical compound decomposes, and the decomposition reaction completely disrupts the theoretical course of solidification considered above. The latter has precisely this place in iron–carbon alloys.

In the \(Fe + C\) system the chemical compound—iron carbide \((Fe_3C)\)—is unstable at the high temperatures of the region of melting and solidification. It decomposes into its constituent elements according to the formula:

\[ Fe_3C = 3\,Fe + C. \]

The liberated carbon is separated in the form of graphite or temper carbon, and thus the process of solidification is accompanied by the process of graphitization. But the iron \(3Fe\) that is liberated (in the form of the modification \(\gamma\)) at the temperature of decomposition of the carbide does not remain intact with respect to the carbon and carbide, but forms with them a solid solution, thereby changing the composition of the total mass of mixed crystals. Both of the indicated reactions—the decomposition of the carbide and the carburization of the iron (cementation)—proceed at a definite rate, and therefore the processes of solidification and melting are complicated still further as a result of the influence exerted on their course by the rate of cooling and heating of the system.

The processes of melting and solidification in the iron–carbon system are so complex, and their experimental investigation presents so many insurmountable difficulties, that to the present time, despite the large number of works devoted to the study of this system, there exists no theory that, with complete clarity and irrefutability, would illuminate and explain all the obscure, unclear, and therefore disputed phenomena observed during the melting and solidification of cast irons. In view of the theoretical and practical importance of understanding the essence of the phenomena occurring during the heat treatment of the iron, steel, and cast iron metals most widespread in technology, the author believes that experimental investigations in this field are necessary; and this prompts him to publish the data from experiments carried out by him with a view to elucidating the question of the melting and solidification of iron–carbon alloys.

II. Phase Diagram of Iron–Carbon Alloys.

The phase diagram of iron–carbon alloys was first published in 1899 by Roberts-Austen ³) *) In 1900 there appeared the famous work of Bakhuis Roozeboom, “Eisen und Stahl vom Standpunkte der Phasenlehre” ⁴), in which B. Roozeboom, on the basis of the theory of solid solutions developed by him, compared the data of Osmond and R. Austen and, having brought the available experimental material into a consistent order, proposed a new form of the diagram, shown in Fig. 2. (Fig. 2 is presented in the form in which it is given in the work of Carpenter and Keeling ⁵).

Fig. 2.

In this diagram, which must be recognized as the foundation for the subsequent development of the theory of iron–carbon alloys, B. Roozeboom drew the lines \(AE\), \(EF\), and \(ot\) on the basis of exclusively theoretical considerations regarding the course of the solidification process. Of these lines, the line of the end of solidification \(Aa\) has still not received experimental confirmation, since the data of Carpenter and Keeling, owing to the absence in their work of cooling curves, cannot be considered entirely reliable. This assumption concerning the unreliability of the data

*) A detailed review of the historical development of the diagram of iron–carbon alloys will be given below, in the second part of the work, in connection with a critical analysis of the data of individual investigators.

of the investigators named, it is all the more probable that, having set themselves the task of experimentally confirming the diagram of B. Roozeboom, they found points belonging to the lines \(aE\) and \(mo\), the absence of which from the diagram was proved with certainty by later investigators.

The line \(EF\), which according to B. Roozeboom’s supposition was to correspond to the formation of carbide by the reaction:

\[ \text{Graphite} + \text{Martensite} = \text{iron carbide }(Fe_3C), \]

was criticized by a whole series of investigators, since the explanation of the formation of iron carbide proposed by B. Roozeboom contradicted the data of practice and of experimental observations. B. Roozeboom himself, in 1904, in a letter concerning the work of Carpenter and Keeling, abandoned the interpretation of the line \(EF\) previously proposed by him, but left this line in his diagram, indicating that it should correspond to the separation of cementite from mixed crystals with \(2\%\ C\). However, the most recent investigations, and in particular the work of Wüst\(^6\), have proved with complete obviousness that the line \(EF\), as well as \(mo\), do not exist in the diagram of iron-carbon alloys. With the disappearance of the line \(EF\), the existence of the segment \(aE\) loses its meaning; according to B. Roozeboom it was to represent the limit of solubility of graphite in mixed crystals. Usually this line, which will be discussed below, in the most recent diagrams runs independently, to the left of \(SE\)—almost parallel to it.

The dotted lines in B. Roozeboom’s diagram refer to the case of supercooling of the system, in which the end of solidification occurs at a lower temperature.

If one introduces into B. Roozeboom’s diagram only the changes just indicated and unquestionably necessary, then it assumes the form proposed by C. Benedicks and shown in Fig. 3.

The meaning of the solid and dotted lines in this diagram is as follows*). In the opinion of B. Roozeboom, supported by Heyn, Benedicks, and others, when iron-carbon alloys are cooled it is necessary to distinguish two cases: slow cooling and rapid cooling (supercooling). With slow cooling, along the line \(AB\) there separate out mixed crystals \(K_g\), representing a solid solution of elementary carbon in \(\gamma\)-iron. Along the line \(BD\) there separates out

\[ \text{*) The meaning of the lines } BD,\ B'D' \text{ and others in Figs. 2 and 3 is the same; only their designations are reversed.} \]

elementary carbon in the form of graphite. Point \(B\) corresponds to the eutectic:

\[ \text{mixed crystals } K_g + \text{ graphite.} \]

Fig. 3.

The line \(Ea\) indicates the limit of solubility of elementary carbon in \(\gamma\)-iron.

Upon rapid cooling, along the line \(AB\) there separate mixed crystals \(K_c\) of a solid solution of iron carbide (cementite) in \(\gamma\)-iron. Along the line \(B'D'\) cementite separates, and at the point \(B'\) a eutectic is formed:

\[ \text{mixed crystals } K_c + \text{ cementite.} \]

The line \(SE'\) gives the limit of saturation of \(\gamma\)-iron with cementite. The product of such rapid cooling is the white cast iron known in technology.

The alloys obtained by rapid cooling are considered unstable—metastable—having a tendency to assume the stable form, in which carbon is present in the elementary state and which is obtained in practice upon slow cooling. In view of the fact that the solubility of the stable form of any element—in the present case, carbon—according to the law of Le Chatelier is less

of solubility of the metastable form, then the line \(Ea\), according to theoretical requirements, should lie to the left of the line \(SE'\). In practice, the position of the line \(Ea\) has not been established.

It must be noted that B. Roozeboom’s diagram has been criticized by a whole series of investigators*). Various additions and changes have been introduced into it, but the main point of criticism—the question of the appearance of graphite—still remains controversial.

On this question the investigators have split into two groups. Some of them**) are supporters of the so-called “double” diagram of the type shown in Fig. 3: they regard graphite as a product of the primary separation of carbon along the line \(BD\) (Fig. 3), and thereby recognize the existence of mixed crystals \(K_g\) and a eutectic: \(K_g +\) graphite, assuming that the separation of cementite along the line \(B'D'\) and the formation of the eutectic: \(K_c +\) cementite take place only upon rapid cooling. Thus they consider it necessary that there be two diagrams: one for the case of slow cooling (the stable system: dashed lines in Fig. 3), and another for the case of rapid cooling (the metastable system—solid lines).

The other group***) recognizes the theory proposed by Goerens, which consists in the fact that for the iron–carbon system there exists only one diagram, shown in Fig. 4: independently of the cooling rate during the solidification of iron–carbon alloys, crystals are always separated along the line \(AB\)

Fig. 4.

*) Among the investigators who have studied the diagram of iron–carbon alloys, the following should be noted: Stansfield\(^{8}\), Heyn\(^{9}\), G. Charpy\(^{10}\), Benedicks\(^{7}\), Goerens\(^{11}\), Ruer\(^{12}\), Howe\(^{13}\), Bumporff\(^{14}\), and others.
**) Representatives of this group are: Heyn, Benedicks, Howe.
***) Its representatives are: Goerens, Wüst.

steel \(K_c\), i.e., crystals of a solid solution of cementite in \(\gamma\)-iron; along the line \(BD\) cementite (primary) is always precipitated. There exists only one eutectic (ledeburite):

\[ \text{mixed crystals } K_c + \text{ cementite}, \]

which solidifies at the constant temperature \(1130^\circ\). But since cementite, not being in a liquid or solid solution, is very unstable at temperatures close to the solidification of the eutectic, it decomposes immediately after its separation into its constituent parts:

\[ Fe_3C = 3Fe + C. \]

Elementary carbon \((C)\) is precipitated in the form of graphite. Both the primary cementite precipitated along the line \(BD\) (Fig. 4) and the cementite of the eutectic undergo decomposition. The decomposition is the more complete, the slower the cooling proceeds. In order to prevent decomposition and obtain white cast iron, it is necessary to cool the alloy rapidly, and indeed so rapidly that the rate of cooling is greater than the rate of the decomposition reaction of cementite. According to Goerens’s theory there is no primary separation of elementary carbon, and therefore there should exist neither mixed crystals \(K_g\) nor a eutectic \(K_g +\) graphite. The appearance of graphite in alloys is a phenomenon of the decomposition of cementite. Hence, as a logical consequence, follows the needlessness of the existence of two separate diagrams for the stable and metastable systems, since the first of them is obtained by the decomposition of the second in the case of slow cooling.

It is self-evident that the explanation of the processes of solidification and melting must differ essentially depending on which of the existing theories is taken as the basis.

Without entering into a detailed critique of the various theories, which will take place in Part II of the present work, the author considers it necessary to indicate briefly those grounds which compel him to be a supporter of Goerens’s theory, and therefore subsequently to expound the processes of melting and solidification on the basis of the theory of “one” diagram.

The theory of “one” diagram does not contradict practical data on the solidification and melting of cast irons; on the contrary, it easily resolves the obscure aspects of these complex processes. Numerous data from experimental investigations agree with this theory. The author’s investigations\(^{15}\), together with Goerens, on graphite formation confirmed the fundamental proposition of the theory concerning the decomposition of carbide.

at the moment of solidification of the eutectic. Conversely, the theory of the double diagram has not yet been confirmed experimentally: the curves of the corresponding stable and metastable course of solidification have not been precisely determined; the eutectic graphite + mixed crystals \(K_g\) has not been established; the difference between the crystals \(K_g\) and \(K_c\) has not been indicated; and the separation of graphite along the line \(BD\) (Fig. 3) has not been proven.

As was mentioned above, in the diagram of iron–carbon alloys the line \(AE\) (Fig. 4) is not confirmed by experimental data. Likewise, the line \(BD\) has also not received experimental confirmation. The other lines, however, were systematically checked only in the work of Carpenter and Keeling\(^5\). Therefore, in investigating the processes of melting and solidification, the author considered it necessary first of all to verify that part of the phase diagram of iron–carbon alloys which is relevant to the processes under consideration.

III. Experimental Verification of the Equilibrium Diagram of Iron–Carbon Alloys.

A. Determination of the Points of the Beginning and End of Solidification of Alloys.

I. Cooling Curves.

To verify the lines \(AB\), \(BD\), \(AE\), and \(EC\) (Fig. 4), cooling curves were taken for alloys with different carbon contents.

The following materials were used for preparing the alloys:

1) Swedish white cast iron:

\(C = 3.746\%\), \(Si = 0.024\%\), \(Mn = 0.180\%\), \(P = 0.024\%\), and \(S = 0.0076\%\).

2) Swedish iron (Nageleisen):

\(C = 0.010\%\), \(Si = 0.023\%\), \(Mn = 0.090\%\), \(P = 0.032\%\), and \(S = 0.005\%\).

3) Pure graphite (Graphit puriss. v. Merck).

The starting materials were melted in a kryptol furnace, the construction of which is shown in Fig. 5*)

Fig. 5.

The furnace consists of an iron box \(ff\), packed with refractory fireclay mass \(E\). Its inner part is covered with a layer of magnesite \(D\). The crucible \(A\) is placed in a cylinder \(P\) of Marquardt mass, which protects the crucible from contact with the kryptol filling the space between \(P\) and \(D\). Current is supplied to the furnace by means of two carbon electrodes, made movable—for better adjustment of their contact with the layer of kryptol. The cylinder \(P\) is closed with a disk \(G\), and the whole furnace

*) Wüst (16), p. 76.

…is provided with a lid \(H\). From below, through the tube \(B\), some gas may be introduced into the furnace, in the atmosphere of which the experiment is being conducted.*)

The curves were taken in the following manner. When the alloy reached a temperature of \(1500^\circ\), the crucible was quickly removed from the furnace and placed in a vessel with trepel preheated to \(500^\circ\). A Le Chatelier thermo-element was introduced into the alloy through an opening in the crucible lid; with the aid of a supporting ring on the protective tube it was set along the axis of the crucible at a distance from the bottom equal to half the height of the button. The weight of the buttons was \(\sim 150\) gr. For taking the curves, a new thermo-element from the firm Heraeus was used, with a calibration certificate, according to whose data the appropriate corrections were introduced into the temperature readings.

For each alloy two curves were taken: one of the Osmond type, with the aid of a Richards chronograph,**) and the other the ordinary one, by observing the rate of cooling second by second.

The Osmond curve, with the chronograph marking every \(10^\circ\), makes it possible to determine the temperature of the critical arrests only with an accuracy of \(10^\circ\); but since these curves distinctly indicate the critical arrests, they were taken for checking the ordinary curves, which give the temperature of the critical points more accurately, but depict them less clearly. The latter curves, nine in number, are presented in Fig. 6.

The composition of the alloys obtained, the temperatures of the critical arrests, and their duration are given in the following Table I. See Table I on p. 15.

From consideration of the cooling curves it follows:

1) that on them are clearly marked:

a) the points of the line \(AB\)—the beginning of solidification of the alloys,

b) the points of the line \(EC\), determining the end of solidification of the eutectic,

and c) the points of the line \(PK\)—the formation of pearlite.

2) that the curves give no indications whatsoever concerning:

a) the temperatures of the beginning of solidification of hypereutectic alloys (line \(BD\)),

and b) the temperatures of the end of solidification of hypoeutectic alloys from \(0\%\) to \(2\%\ C\) (line \(AE\)).

*) In the present experiments the latter device was not used, since the melting was conducted in an atmosphere of air.

**) For a description, see Goerens (17), p. 16.

Table I.

Fig. 6.

Vertical axis: temperature, °C.

Horizontal axis: carbon, %.

Curve labels: I, II, III, IV, V, VI, VII, VIII, IX, X, XI, XII, XIII.

TABLE 1*)

Alloy No.	Total C, %	Graphite, %	Si, %	Mn, %	P, %	S, %	Beginning of solidification, °C	Eutectic (ledeburite), °C	Eutectoid (A1), °C	Duration of eutectic arrest, sec	Micrograph figure No.	Brief description of the structure of the alloys
I	0,577	—	0,028	0,132	0,030	0,005	1465°	—	701°	—	7	Ferrite and pearlite.
II	1,071	—	0,030	0,125	0,032	0,006	1429°	—	703°	—	8	Pearlite.
III	1,632	—	0,041	0,141	0,028	0,006	1381°	—	708°	—	9	Pearlite and secondary cementite.
IV	1,762	—	0,045	0,143	0,026	0,005	1369°	—	710°	—	10	Pearlite with secondary cementite. Weak ledeburite.
V	2,165	0,282	0,045	0,160	0,022	0,006	1340°	1093°	709°	25	11	The amount of ledeburite increases.
VI	2,795	0,611	0,060	0,171	0,024	0,006	1272°	1123°	709°	110	12	White cast iron.
VII	3,677	2,263	0,064	0,172	0,023	0,007	1199°	1130°	710°	250	13	Gray cast iron.
VIII	4,626	3,413	0,090	0,181	0,024	0,007	—	1134°	710°	260	14	Gray cast iron. Coarse graphite and plates.
IX	4,940	3,910	0,091	0,180	0,024	0,007	—	1134°	715°	225	15	Gray cast iron.

*) The structure of alloys I–IX is presented in micrographs (Figs. 7–15).

The absence on the cooling curves of critical arrests for constructing the lines BD and AE is a fact that has long been observed. These arrests were entirely absent in the author’s experiments, despite the careful recording of the curves, with the use of the thinnest quartz protective tubes, under slow cooling, and with a sufficiently large quantity of alloy.

1) The reason for the absence of critical points of the beginning of solidification of hypereutectic alloys may be explained in different ways.

a) If one adheres to Goerens’s theory of graphite formation and takes into account Osmond’s assertion concerning the endothermic character of the reactions of decomposition of iron carbide, then the absence of critical arrests may be explained by the fact that the heat released upon the separation of carbide crystals is compensated by the heat consumed in their decomposition, which takes place immediately after their separation. But Osmond’s assertion concerning the endothermic character of the reaction of carbide decomposition is open to doubt, since the majority of investigators (Moissan, Mylius-Schoene and Förster, Saniter, Frost, Hautfeuille) regard this reaction as exothermic; and therefore the above explanation is valid only insofar as Osmond’s assertion is true.

b) Consideration of the theoretical process of solidification of alloys gives certain indications that make it possible to assert that the points of the beginning of solidification of hypereutectic alloys must be weakly expressed on the cooling curves, and therefore they usually escape thermal analysis.

Indeed, the critical point of the beginning of solidification, or the so-called break in the curve, is caused exclusively by the liberation of the heat of crystallization. It is self-evident that this break will be the clearer, the greater the amount of heat liberated at the initial moment and during the period of crystallization. The quantity of this heat depends on the following factors:

1) on the heat of crystallization of the given substance,
2) on the capacity of the substance for supercooling,
3) on the composition of the alloy and the slope of the liquidus.

With regard to the first of these factors, in relation to iron-carbon alloys, exactly nothing is known. The absence of critical points forces one to suppose that the heat of crystallization of iron carbide is insignificant.

The absence of critical points also indicates that hypereutectic alloys, at the moment of solidification, are not prone to supercooling—

ing. And since supercooling of bodies depends on two causes: on the small magnitude of the rate of spontaneous crystallization (s. c. k.)*) and on the small rate of development of crystallization (s. k.), and also on the relative position of the maximum of the s. c. k. relative to the curve s. k., it must be assumed that the s. c. k. and s. k. of carbides are considerable, and that the position of the maximum of the s. c. k. is at a temperature at which the magnitude of s. k. is sufficiently large that supercooling does not take place.

Fig. 16.

5°. Finally, with regard to the third factor it is necessary to point out the following:

Let us assume**) for simplicity of consideration that the liquidus \(BD\) is a straight line forming an angle \(\alpha\) with the ordinate axis (Fig. 16); let \(bd = dg = 1^\circ\). Then \(ed = tg\,\alpha\), \(gh = 2\,tg\,\alpha\), and the quantities of crystals separating from alloy \(a\), upon cooling it by one and by two degrees, will be expressed as follows:

\[ \frac{ed}{ce}=\frac{tg\,\alpha}{tg\,\alpha+x} \quad\text{and}\quad \frac{hg}{hf}=\frac{2\,tg\,\alpha}{2\,tg\,\alpha+x} \]

and upon cooling by \(n^\circ\):

\[ \frac{n\,tg\,\alpha}{n\,tg\,\alpha+x}. \]

The latter expression shows: 1) that, for a given composition \(x\), the quantity of crystals separating upon cooling the alloy by \(1^\circ\) gradually decreases; 2) that, for a variable composition of the alloy, the quantity of crystals separating at the first moment of solidification is the greater, the smaller \(x\) is, i.e., the closer the alloy is in its composition to the pure component; and 3) that the quantity of crystals separating at the first moment of solidification is the greater, the larger the angle \(\alpha\), i.e., the more gently the liquidus runs.

It is self-evident that the break on the cooling curve is expressed the more clearly, the more crystals separate in the first moments of solidification.

In hypereutectic iron–carbon alloys a break on the cooling curves is not observed, which must be attributed to the fact that the alloys under investigation are located not far from the eutectic [[unclear: continuation obscured]].

*) [[unclear: footnote obscured by stamp]]

**) [[unclear: footnote obscured by stamp]]

point (condition 2) and that, apparently, the liquidus rises steeply upward (condition 3). Although the position of the liquidus has not been precisely determined, in diagrams it is given the appearance of a line rising steeply upward, which indeed must correspond to reality.

It is precisely these two circumstances that account for the insignificant quantity of crystals separated out in the first moments of solidification; the heat evolved by such an insignificant quantity of crystals is insufficient to produce a clear break in the cooling curve, and therefore the method of thermal analysis does not give positive results for determining the line \(BD\).

2) The absence of a break in the cooling curves for determining the end of solidification of hypoeutectic alloys, i.e., for constructing the line \(AE\), finds its explanation in the following considerations.

Let us trace, as before, the theoretical course of solidification of alloys forming solid solutions). Suppose that the liquidus and solidus* are straight lines (Fig. 17). Let \(bd = dg = 1^\circ\); then: \(de = \operatorname{tg}\alpha\), \(gh = 2\operatorname{tg}\alpha\), and the quantities of mixed crystals separated out on cooling by \(1^\circ\) and \(2^\circ\), respectively, are equal to:

Fig. 17.

\[ \frac{de}{ec} = \frac{\operatorname{tg}\alpha_1}{\operatorname{tg}\alpha_1 + (x - x_1)} \quad \text{and} \quad \frac{gh}{fh} = \frac{2\operatorname{tg}\alpha_1}{2\operatorname{tg}\alpha_1 + (x - x_2)} \]

and the quantity of crystals precipitating when the temperature is lowered by \(n^\circ\) from point \(b\) is equal to:

\[ \frac{n\operatorname{tg}\alpha_1}{n\operatorname{tg}\alpha_1 + (x - x_n)} \]

From the drawing in Fig. 17 it follows that:

\[ x - x_1 = (m - 1)\operatorname{tg}\alpha_2;\quad x - x_2 = (m - 2)\operatorname{tg}\alpha_2 \quad \text{and} \quad x - x_n = (m - n)\operatorname{tg}\alpha_2. \]

Substituting the values \(x - x_1\), \(x - x_2\), and \(x - x_n\) into the first three equalities, we obtain:

*) See (14), p. 283.

\[ \frac{de}{ec} = \frac{\tg \alpha_1}{\tg \alpha_2 \left[m-\left(1-\frac{\tg \alpha_1}{\tg \alpha_2}\right)\right]}; \qquad \frac{gh}{fh} = \frac{2\tg \alpha_1}{\tg \alpha_2 \left[m-2\left(1-\frac{\tg \alpha_1}{\tg \alpha_2}\right)\right]} \quad \text{and} \]

\[ \frac{n\tg \alpha_1}{\tg \alpha_2 \left[m-n\left(1-\frac{\tg \alpha_1}{\tg \alpha_2}\right)\right]}. \]

From these expressions it is evident that the number of crystals separating during the process of solidification depends on the angles \(\alpha_1\) and \(\alpha_2\), i.e. on the character and mutual position of the liquidus and solidus.

If \(\alpha_1 > \alpha_2\), then the quantity \(1-\frac{\tg \alpha_1}{\tg \alpha_2}\) is less than unity, and therefore the denominator in the expressions under consideration increases as the temperature decreases; that is, the quantities of crystals precipitating when the temperature of the alloy is lowered by \(1^\circ\) decrease successively.

For \(\alpha_1 < \alpha_2\), the reverse phenomenon is observed: the denominator of the fractions decreases, and therefore the quantities of precipitating crystals increase when the temperature is lowered by \(1^\circ\). And, finally, if \(\alpha_1=\alpha_2\), then the quantity of crystals separating when the temperature is lowered by \(1^\circ\) remains constant throughout the entire process of solidification and is equal to:

\[ \frac{1}{m}. \]

The influence of the angles \(\alpha_1\) and \(\alpha_2\) on the character of the breaks in the cooling curves is represented graphically in Fig. 18, on which a series of theoretical cooling curves is shown for a system of two elements possessing complete solubility in one another both in the liquid and in the solid state.

Fig. 18.

For the first three curves \(\alpha_1 > \alpha_2\), i.e. the branches of the liquidus and solidus diverge; therefore the increase of crystals during the process

of solidification \(bc\) quantitatively decreases. The consequence of this is: a clearly expressed beginning of solidification (point \(b\)), a bend of the curve \((bc)\) convex with respect to the \(z\)-axis, and a scarcely noticeable transition of the bend \(bc\) into the curve \(ck\).

Curve IV corresponds to equality of the angles \(\alpha_1\) and \(\alpha_2\), and therefore the bend \(bc\) appears on it in the form of a straight line.

Finally, curves V—VII correspond to the region of the diagram where the branches of the liquidus and the solidus converge, i.e., the angles \(\alpha_1 < \alpha_2\), and, consequently, the growth of crystals as the temperature decreases increases. The bends \(bc\) on the curves are concave toward the \(z\)-axis. The point of the beginning of solidification \(b\) is indicated indistinctly on the cooling curves; on the contrary, the end of solidification (point \(c\)) is marked more sharply than in the curves of the first category.

From what has been set forth it follows that determination of the solidus by means of cooling curves is possible only in those portions of the state diagram where the liquidus and solidus curves converge. If these curves diverge, then the heat of crystallization in the last moments of solidification is so insignificant that the bend \(bc\) imperceptibly passes into the curve \(ck\), and the point of the end of solidification is not obtained distinctly.

In iron–carbon alloys the liquidus and solidus are diverging lines, and therefore it is entirely understandable that on curves I—IV of Figure 6 there are absolutely no points of the end of solidification.

Prof. Tammann*) proposes, for determining the end of solidification from cooling curves, to use the inflection point of the curve, which is obtained in the transition of the convex part \(bc\) into the concave \(ck\), and which undoubtedly lies close to the theoretical end of solidification. But since this inflection, according to Tammann’s indications, always lies on the curve somewhat below the end of solidification, he proposes to reduce the interval of crystallization observed on the curves (from the beginning of crystallization to the inflection point) by a certain quantity \(\Delta J\), determined from the following formula:

\[ \Delta J=\frac{1}{100}\left[(100-x)J_A+xJ_B\right], \]

in which \(J_A\) and \(J_B\) denote the apparent solidification intervals of the pure components, and \(x\) is the composition of the alloy in weight parts of element \(B\).

However, Tammann’s method is inapplicable to iron–carbon alloys, because for them the solidification interval of iron carbide is unknown.

*) see (12), p. 292

This method is unacceptable, in the author’s opinion, also because, in general, on cooling curves it is impossible to determine the point of inflection accurately: in determining the temperature of the end of solidification from the point of inflection one may easily make an error of tens of degrees. Moreover, the position of the point of inflection on the cooling curve depends, in the author’s opinion,* on the ratio between the mass of the alloy under investigation and the mass of the crucible and furnace in which the experiments are carried out, and this point may change its position depending on the experimental conditions; at the same time, according to the author’s investigation, the point of inflection always lies above the end of solidification of the alloy.

To what has been stated above concerning the inapplicability of thermal analysis for finding the solidus \(AE\) in the diagram of iron–carbon alloys, it must further be added that the curve \(AE\), even if it could be found by means of thermal analysis, would correspond to the end of solidification of the alloys only in the case in which the process of solidification proceeds ideally, i.e., when this process takes place so slowly that the mixed crystals separating out have sufficient time to assume the composition corresponding to the state of equilibrium between them and the remaining liquid phase of the alloy. To establish such equilibrium, especially slow cooling is required. But in taking cooling curves it is impossible to reduce the rate of cooling below a certain definite value; otherwise the critical points cannot be detected on the cooling curves. Thus, ideal cooling conditions, for purely experimental reasons, can never in reality be fulfilled, as a result of which the temperature of the end of crystallization, determined from cooling curves, will be lower than that at which crystallization theoretically should end. The latter is explained as follows.

Fig. 19.

In Fig. 19 the line \(CdD\) represents the liquidus of the system \(C + D\). The dotted line \(Cd_1D\) is the solidus in its theoretical position. Under ideal cooling conditions, some alloy \(S\) should finally solidify at the point \(d_1\), and the composition of the liquid phase

*) The author’s investigation: “Theoretical and Practical Cooling Curves” is being prepared for publication.

**) See (14), p. 287.

the alloy at the moment solidification ends will be determined by point \(e_1\). But, owing to insufficiently slow cooling, the entire quantity of crystals theoretically required will not have time to separate in the interval \(d—d_1\), and the crystals that separated initially will not equalize their composition with the crystals of the last moment of solidification; as a result, at temperature \(T_1\) a certain amount of liquid phase will remain, and the process of solidification will not end at point \(d_1\), but will continue up to some point \(d_2\). In other words, the solidus \(C d_2 D\), obtained experimentally on the basis of cooling curves, will be more convex than the theoretical one and will indicate temperatures of completion of solidification lower than those which should occur under ideal cooling conditions.

All the above considerations regarding the inapplicability of the thermal method for determining the solidus \(AE\) prompted the author to make use, for the latter purpose, of a metallographic method—the quenching method, which, by means of rapid cooling, makes it possible to fix the structure of alloys at high temperatures and to study it under the microscope.

II. The Quenching Method.

The quenching method for determining the end of solidification of alloys forming solid solutions was first successfully applied by Heycock and Neville) in the study of copper and tin alloys. It was also used by Scherpherd) \((Cu + Zn)\), Carpenter and Edwards) \((Cu + Al)\), and Guillet**) (aluminum alloys), since in all these cases the cooling curves did not provide sufficient data for constructing the solidus.

The possibility of determining the position of the solidus by means of the quenching method is based on the following. Very rapid cooling arrests the changes of structure that occur during slow cooling. Thus, by quenching the specimen under investigation at some high temperature, one can obtain that structure which the specimen had at the quenching temperature. In solid solutions, melting occurs within a definite temperature interval. The beginning of melting is characterized either by the melting of crystals along

) Heycock and Neville. Phylos. Trans. 1903 (202), p. 1.
) Scherpherd. Journ. of phys. Chem. VII (1904), p. 421.
) Carpenter and Edwards. 2 Raport au Comité des alliages* 1907, Jan.
) Guillet. Revue de Métall.* 1906, p. 567.

surface, or by the appearance of drops of liquid in the crystals themselves. Therefore, if one carries out a series of successive quenchings (Fig. 20) at a gradually increasing temperature, heating the specimen at each given temperature for a sufficient time for the melting process to be able to begin, then it is always possible to find two such temperatures (points 3 and 4), at which: at point 3 melting has not yet begun, while at point 4 the melting process has clearly revealed itself. Then the temperature of the beginning of melting will lie between points 3 and 4, or, in other words, between these points the solidus line will pass.

Fig. 20.

This method has the advantage over the thermal method in that it does not give a false position of the solidus, as was indicated in the discussion of Fig. 19. Indeed, if the alloy from which the test specimen is taken cooled insufficiently slowly, as a result of which either the mixed crystals obtained were of nonuniform composition, or, in alloys to the left of point \(E\) (Fig. 20), eutectic \(B\) appeared, then with a sufficiently prolonged heating of the specimen during the quenching experiment, the composition of the mixed crystals becomes equalized, the eutectic dissolves, and the structural constituents of the alloy will be, at the quenching temperature, in an equilibrium close to the theoretical one, which is necessary for determining the beginning of melting and for the precise construction of the line \(AE\).

Quenching experiments could also be carried out in another way, bringing them closer to the method of determining the solidus by means of cooling curves, namely: one could heat the test specimen to melting and then very slowly cool it to the quenching temperature.

From a theoretical point of view there is no difference between these two methods, which pursue one and the same aim—the attainment of an equilibrium state of the alloy at the moment of quenching, because it is entirely immaterial how the equilibrium of the system at a given temperature is established: by gradual heating of the specimen to this

temperature, or by slowly cooling it from some higher temperature. But in practice the second method presents insurmountable experimental difficulties: it requires a high initial heating temperature, at which it is difficult to avoid a change in the composition of the specimen due to oxidation. For alloys with an insignificant carbon content, such a high temperature cannot be attained in the horizontal Heraeus furnaces that must be used for quenching experiments, since furnaces of other systems, by their construction, do not permit the quenching experiments to be surrounded by all the precautions necessary to prevent a change in the composition of the specimen and its cooling during quenching. Moreover, it is practically impossible to carry out an ideally slow and, in all experiments, uniform cooling of the specimen down to the quenching temperature; failure to observe this latter condition deprives this method of quenching of all significance.

Conversely, the first method of quenching presents no difficulties for its practical execution: in Heraeus furnaces it is easy to set the temperature accurately and to keep it constant for a long time. When the temperature of the furnace has become established, the specimen under investigation can be quickly introduced into it, held in the furnace at a constant temperature for a definite time, and then rapidly quenched.

1) Quenching experiments were first used by the author in studying the process of solidification of cast irons in connection with graphite precipitation and gave satisfactory results. However, the procedure then adopted for carrying out the experiments*) has been somewhat modified in the present work, with the aim of increasing the accuracy of determining the quenching temperature and preventing accidental oxidation of the specimens.

Fig. 21.

The arrangement of the little boat for the quenching experiments is shown in Fig. 21. In it, the letter A schematically represents part of a horizontal

*) See (20) and (15).

of a Heraeus electric furnace. \(B\) is a tube of glazed porcelain, inserted into the furnace to protect the heating cylinder of the furnace from accidental damage and to make it possible to heat the specimens in an atmosphere of nitrogen. The boat \(ab\), measuring \(12 \times 12 \times 17\ \mathrm{mm}\), was made of platinum; on its right side a platinum tube \(cd\) is soldered into it, the left end \(c\) of which is sealed. Into this tube fits tightly the protective tube of the thermoelement, whose external diameter is \(d_1 = 3\ \mathrm{mm}\), and internal diameter \(d_2 = 1.2\ \mathrm{mm}\). Into it is inserted a Le Chatelier thermoelement, whose platinum wire is placed in an inner insulating tube of dimensions \(d'_1 = 0.8\ \mathrm{mm}\), \(d'_2 = 0.4\ \mathrm{mm}\). The end of the outer protective tube is not sealed, and thus the junction of the thermoelement remains exposed, which greatly increases the accuracy of the temperature measurement. The ends of the thermoelement wires are connected to a Siemens and Halscke millivoltmeter. Onto the protruding right end \(d\) of the platinum tube a thick-walled porcelain tube \(m\) is tightly fitted, serving as a handle for holding the platinum boat; onto the end of tube \(m\) is wound a stopper \(S\) made of asbestos cord, for properly positioning the boat in the middle of the furnace. The boat is coated on the inside with a layer of magnesia.

To prevent oxidation of the specimens (samples), they were heated under cover of a layer of molten salt. Depending on the quenching temperature, the following were used for the experiments:

\[ \begin{aligned} KCl &\quad \text{melting temperature}^*) \quad T = 790^\circ \\ NaCl &\quad \text{”} \quad \text{”} \quad T = 820^\circ \\ BaCl_2 &\quad \text{”} \quad \text{”} \quad T = 960^\circ . \end{aligned} \]

The quenching experiments were carried out as follows: into the furnace, heated to a definite temperature, a boat filled with the corresponding salt was carefully introduced. When the salt had melted and the temperature \(T\) of the furnace had been precisely established, the boat was withdrawn from the furnace, the specimen \(p\) (Fig. 21) was placed in it, and the boat was pushed back in. During this manipulation the temperature of the boat fell by \(100^\circ\)–\(200^\circ\), then in 3–4 minutes reached its former value \(T\). The specimens were held at this constant temperature in the furnace for 7 minutes; then the boat was quickly withdrawn and its contents (specimen and salt) were overturned into a vessel with water (\(6^\circ C^{**}\)), standing directly at the furnace opening. The vessel for the water consisted of

) Physikalisch-Chemische Tabellen* by Landolt-Börnstein, 1907.

**) Temperature of the water in the water supply in winter.

... a cylinder 10 cm in diameter and 1 m high. The general arrangement of the apparatus during the quenching experiments is shown in Fig. 22.

In order to obtain rapid and complete quenching, it is necessary that the specimens be of the smallest possible dimensions, sufficient only for making a polished section, and, moreover, that the specimens have such a form as, for a given volume, presents the greatest cooling surface area*).

The specimens in the present experiments weighed about 1 g. They were obtained from buttons I–VII by cutting from them, on a lathe, two disks, \(ab\) and \(cd\) (Fig. 23), 2 mm thick, and by cutting these disks into 8 parts. The turnings produced when the disks were cut off (the hatched parts in Fig. 23) were used for determining the chemical composition of the buttons.

For checking the suitability and accuracy of the proposed method of measuring the quenching temperature of the specimens, and also for finding the heating time of the specimens necessary for establishing equilibrium, preliminary experiments were carried out.

Fig. 23.

The construction of the boat guaranteed that the temperature of the specimen corresponded to the temperature of the molten salt; the temperature of the salt was measured by a thermoelement, the junction of which was located at the very end of the platinum tube \(cd\), surrounded by the molten salt. The question arose whether the specimen, while the boat was being removed from the furnace (up to the moment of immersion of the specimen in water), might not have time to cool somewhat. Observations of the millivoltmeter needle showed that during this operation, which lasted less than \(1/2\) sec., the needle gave no oscillation. This must serve as evidence that there was no cooling of the entire quenching apparatus, and consequently of the specimen—cooling which could have been registered by the precise millivoltmeter used for the experiments; consequently, the quenching was carried out practically exactly at the temperature measured by the thermoelement.

Owing to the heating of the specimens in a medium of molten salt, and also to the passage of a stream of nitrogen through tube \(B\) (Fig. 21), there was likewise no oxidation, i.e. no change in the composition of the specimens, which was verified

*) The use of large specimens for quenching (the work of Heyn and Bauer and of Goerens) results in partial tempering during the quenching itself, accompanied by the appearance of troostite in the quenched specimens. Troostite, testifying to incomplete quenching, obscures and distorts the structure of the metal which the latter ought to have at the given temperature under conditions of ideally rapid quenching. In the author’s quenching experiments, no troostitic structure was observed.

Fig. 22.

both by analysis of the specimens after quenching and by microscopic examination of the polished surface of several specimens after quenching.

On the basis of the author’s preceding work, the duration of heating of the specimens at the quenching temperature was taken to be 7 minutes.

In order to verify whether seven minutes’ heating is sufficient for establishing structural equilibrium in the specimens, the following experiments were carried out. After the melting temperature \(T_l\) had been established for alloy IV (with 7-minute heating), and the lower temperature \(T_s\) at which the specimen did not begin to melt, a series of quenches was performed at the temperatures \(T_l\) and \(T_s\), with the specimens heated for 1, 3, 5, 7, 10, 20, and 30 minutes. These experiments showed that at the temperature \(T_s\) the onset of melting did not occur even after half an hour of heating, and, conversely, at the temperature \(T_l\) the melting process was detected after three minutes of heating and was fully established after 5 minutes. The adopted heating time of 7 minutes was thus entirely sufficient, since longer heatings introduced no changes into the structure of the specimen.

The other alloys were subjected to a similar check, with a duration of 15 minutes, at the temperatures \(T_l\) and \(T_s\).

For the quenching experiments a new Le Chatelier thermoelement was taken, with a certificate from the Royal Physical Laboratory in Berlin. During the experiments, after every five quenches the thermoelement was checked against gold (melting temperature \(1063^\circ\)). The corresponding corrections were introduced into the temperatures read off, and the temperature of the cold junction of the thermoelement, placed in glass tubes in running water, was also taken into account.

With all the above-mentioned precautions observed, 35 quenching experiments were carried out, the results of which are brought together in the following Table II: see Table II on p. 28.

Note: As the numbering of the specimens shows, the first series of experiments was carried out on alloy IV, the initial quenching temperature having been chosen above the theoretical line \(AE\) (cf. Fig. 4). The results obtained were checked and confirmed by the experiments of series II. Then the experiments of series III and IV were carried out. The experiments of these four series were sufficient to determine the position of the solidus \(AE\), but for a more exact determination of the convexity of the solidus an alloy X with \(0.827\%\) was prepared.

TABLE II.

Series No.	Sample No.	Temperature of the sample at the moment of quenching	Figure No.	Notes
I; Alloy IV, 1.762% C	1	1227	25	Melted on the outside; under the microscope, partial melting.
I; Alloy IV, 1.762% C	2	1204	25	Melted on the outside; under the microscope, partial melting.
I; Alloy IV, 1.762% C	3	1175	25	Melted on the outside; under the microscope, partial melting.
I; Alloy IV, 1.762% C	4	1167	25	Melted on the outside; under the microscope, partial melting.
I; Alloy IV, 1.762% C	5	1153		Partial melting only under the microscope.
I; Alloy IV, 1.762% C	6	1143		Partial melting only under the microscope.
I; Alloy IV, 1.762% C	7	1140		Traces of the melting process.
I; Alloy IV, 1.762% C	8	1130		No melting.
I; Alloy IV, 1.762% C	9	1120	24	No melting.
II; Alloy III, 1.632% C	10	1218	—	Melted on the outside; under the microscope, partial melting.
II; Alloy III, 1.632% C	11	1198	—	Melted on the outside; under the microscope, partial melting.
II; Alloy III, 1.632% C	12	1172	—	Melted on the outside; under the microscope, partial melting.
II; Alloy III, 1.632% C	13	1154	27, 28	Melting under the microscope.
II; Alloy III, 1.632% C	14	1139		Traces of the melting process.
II; Alloy III, 1.632% C	15	1130	26	No melting.
III; Alloy II, 1.071% C	16	1310		Melted on the outside; under the microscope, partial melting.
III; Alloy II, 1.071% C	17	1300		Melted on the outside; under the microscope, partial melting.
III; Alloy II, 1.071% C	18	1285		Melted on the outside; under the microscope, partial melting.
III; Alloy II, 1.071% C	19	1265		Melted on the outside; under the microscope, partial melting.
III; Alloy II, 1.071% C	20	1243	30	Partial melting only under the microscope.
III; Alloy II, 1.071% C	21	1228	30	Partial melting only under the microscope.
III; Alloy II, 1.071% C	22	1207	30	Partial melting only under the microscope.
III; Alloy II, 1.071% C	23	1185		Traces of the melting process.
III; Alloy II, 1.071% C	24	1170		No melting.
III; Alloy II, 1.071% C	25	1150	29	No melting.
III; Alloy II, 1.071% C	26	1130		No melting.
IV; Alloy I, 0.577% C	27	1365	34	Melted on the outside.
IV; Alloy I, 0.577% C	28	1345	34	Melting under the microscope.
IV; Alloy I, 0.577% C	29	1318	31, 32	Traces of the melting process.
IV; Alloy I, 0.577% C	30	1290		No melting.
IV; Alloy I, 0.577% C	31	1270		No melting.
V; Alloy X, 0.327% C	32	1248		Melting under the microscope.
V; Alloy X, 0.327% C	33	1228		Traces of the melting process.
V; Alloy X, 0.327% C	34	1208		No melting.
V; Alloy X, 0.327% C	35	1186		No melting.

C, and four more experiments of series V were carried out with this alloy.

2) Microscopic investigation. In specimens whose quenching temperature lay considerably above the line \(AE\), the melting process could be established from the external appearance of the specimens, since their surface showed perfectly clear traces of external melting (specimens Nos. 1, 10, 16, and 27). In the remaining specimens, external inspection did not permit a conclusion as to the beginning of the melting process; in order to establish it, all specimens were polished, etched with a 4% solution of nitric acid in amyl alcohol, and examined under the microscope. This investigation made it possible to establish precisely the beginning of the melting process, and showed that the melting of mixed crystals takes place both from their surface and within their mass. In the latter case special points or centers of melting arise, which, increasing and growing in a radial direction, form droplets. The melted portions of the specimens present the outlines of a spreading liquid, in most cases have a eutectic structure, and therefore are quite clearly distinguished from the metal that has not yet melted.

Several of the most characteristic examples of the structure of the specimens, before and after the beginning of the melting process, are shown in Figs. 24–34 (photoplates III and IV). Fig. 24 shows the structure of specimen No. 9, quenched at \(1120^\circ\). The polished section consists of large crystals (austenite), permeated by white lanceolate inclusions (martensite). Between the austenite crystals cracks are observed (black bands in Fig. 24), which arose as a result of rapid quenching. The structure of specimen No. 9 is characteristic of steel rapidly quenched at a high temperature. There are no signs of the melting process in Fig. 24. Fig. 25 shows the structure of the same specimen after quenching at \(1167^\circ\). In it the process of melting of the mixed crystals (austenite) is expressed quite clearly. On the left side a vertically arranged band of eutectic appearance is visible. In its structure (eutectic) and in its external outline it undoubtedly represents the melted portion of the polished section. In the middle of Fig. 25 and slightly below the middle, two centers of melting are visible. The lower right one indicates radial propagation of the melting process into the mass of the crystals; the band on the left is melting of the crystals from the surface (intercrystalline melting). Fig. 25 clearly proves that at a temperature of \(1167^\circ\), i.e., at a temperature lying below the theoretical solidus \(AE\), the melting process had already begun.

The structure in Fig. 26 belongs to specimen No. 15, quenched before the beginning of the melting process \((T_{\text{quench}} = 1130^\circ)\), and represents a transitional stage from austenite to martensite. Fig. 27 shows the structure of the same specimen No. 15 after quenching at \(1154^\circ\) (considerably below the theoretical line \(AE\)). In the center of the photomicrograph there is a clearly outlined light inclusion, differing from the structure of the rest of the field of the polished section. Such inclusions are observed throughout the section. The outline of this inclusion is characteristic of liquid metal. At a magnification of 800 times (Fig. 28) the structure of the inclusion is eutectic. This shows that the inclusion is a melted part of the specimen. Thus, Figs. 27 and 28 prove that at a quenching temperature of \(1154^\circ\) the melting process, in the form of surface fusion of the crystals, had already begun.

Fig. 29 shows the structure of specimen No. 25, quenched before the beginning of the melting process at a temperature of \(1150^\circ\). Fig. 30 represents the structure of the same specimen, quenched at \(1228^\circ\). In the latter, four islands of white color are visible, which by their outlines undoubtedly belong to molten metal. The structure of these islands, however, does not appear eutectic. Characteristic in them is the fringed outline of the contour, which is observed both in Fig. 30 and in the following two figures, 31 and 32, showing the structure of specimen No. 29, quenched at \(1318^\circ\). The fringed edges of the island are especially clear in the latter figure, taken at a magnification of 600 times. This phenomenon, in connection with the absence of a eutectic structure in the islands, finds its explanation in the following: the melting process, as was indicated above, consists in the alloy of a definite composition decomposing into crystals with a lower carbon content and a liquid phase with a higher carbon content than that of the alloy.

This decomposition of the solid solution, owing to diffusion, does not take place instantaneously; appearing predominantly from the surface of the crystals, it spreads gradually inward. Therefore the first drops of liquid appear in the cleavage planes of the crystals, while the crystals themselves become poorer in carbon from the surface.

Thus, some alloy with \(a\%\) carbon (Fig. 20) at some temperature \(4\) will separate a liquid phase of composition \(k\%\) carbon. On rapid cooling (during quenching) the liquid phase will not have time to be absorbed by the solid crystals and therefore will solidify as alloy \(K\) would solidify, i.e., it will separate

mixed crystals, itself approaching the eutectic composition \(B\). This process of solidification takes place very rapidly, and so rapidly that, as experiments have shown, even quenching cannot prevent it. The mixed crystals separating out in the liquid phase, and likewise the crystallites of the eutectic, are attracted by the powerful crystals of solid metal surrounding them and become attached to the latter; as a result, the microscopically small drop of liquid loses its eutectic structure, while its edge, bearing traces of the displacement of the crystals, becomes fringe-like.

If this explanation is correct, then the solidified drop must, in its structure, represent the second constituent of the eutectic, i.e. cementite, which, according to the theory of alloys, cannot appear as a structural constituent in alloys with a carbon content lower than \(0.9\%\).

To verify the proposed assumption concerning the nature of the white islands visible in Figs. 30, 31, and 32, all specimens on which such islands were found were etched with a reagent specific for cementite: sodium picrate, which colors cementite chestnut-black. Fig. 33 presents photographs of the polished sections of Figs. 30 and 31 after etching with sodium picrate. It is clearly visible in it that the aforementioned islands have taken on a black coloration.* This is proof that the islands consist of cementite. The appearance of cementite in alloys with a carbon content lower than \(0.9\%\) is the result of a disturbance of the proper course of the solidification process owing to quenching.

Figs. 32 and 34 furnish a clear proof of the theory of enrichment of the liquid phase at the expense of the carbon of the mixed crystals: in Fig. 32, an island of melted metal is surrounded by a light rim of iron depleted in carbon. This is still more clearly visible in Fig. 34, in which such a place on the polished section is photographed where the process of melting is only beginning: the crystals are decarburized at the surface, and between them there form regions of fragmented crystallites, which subsequently melt.

3) Phase diagram of iron–carbon alloys. On the basis of the cooling curves in Fig. 6 and of the results of the quenching experiments, the phase diagram of iron–carbon alloys shown in Fig. 35 has been constructed.

* See (17), p. 119.

** In the right-hand part of Fig. 33 it was possible to photograph exactly the middle of Fig. 31.

The solid lines of the diagram represent the results of the author’s experiments. The dash-dotted lines are drawn according to the data of Prof. Heyn \((^{21})\)

Fig. 35.

and the dotted line with small dashes represents a line that has been experimentally not yet established.

For the melting point of pure iron, the temperature adopted in the diagram is \(1505^\circ C\), in accordance with the data of Carpenter \(^{(22)}\) and Müller \(^{(23)}\).

The position of the eutectic point \(B\) was checked and established at \(4.2\% C\). For the latter purpose the author obtained from Prof. Goerens an alloy with a clearly expressed eutectic structure, shown in Fig. 36.

Analysis of this alloy gave the following results:

\[ \begin{array}{llll} 1. & 4.144 & 4. & 4.174 \\ 2. & 4.240 & 5. & 4.213 \\ 3. & 4.239 & 6. & 4.182 \end{array} \]

on average: \(4.194 = \sim 4.20\% C\).

The temperature of the eutectic line \(EC\), in agreement with the author’s preceding investigations, was obtained as \(1134^\circ\).

The position of point \(E\) was established at \(1.70\% C\), which differs considerably from the position of point \(E\) at \(2\% C\) accepted up to now, and approaches the original data of R. Austen. The following grounds served for shifting point \(E\) to the left.

Microscopic examination of alloy IV with \(1.762\% C\) revealed traces of eutectic (see Figs. 37 and 10), characterized by a peculiar accumulation, for the eutectic, of cementite plates. Quite clear such eutectic accumulations are observed in alloy V with \(2.165\% C\) (Fig. 11), in which, according to the position of point \(E\) at \(2\% C\) accepted until now, there should have been only traces of eutectic. However, the finding of eutectic in these alloys cannot yet be considered a sufficient basis for shifting point \(E\) to the left, since, according to the explanation of B. Roozeboom*), in alloys close to \(2\% C\) one can sometimes observe a eutectic appearing as a result of insufficiently slow cooling of the alloy: such cooling prevents the composition of the mixed crystals from becoming uniform; the liquid phase remains in excess and may be observed in such alloys in which theoretically it should not exist (cf. the explanation on pp. 21 and 22 for Fig. 19).

However, an alloy with such a eutectic is in unstable equilibrium, and it must necessarily be absorbed upon heating the alloy to a certain temperature.

In order to decide whether the eutectic in alloys with \(2\% C\) is a structural constituent or a product of insufficiently slow cooling, the following verification experiment was made: an alloy with \(2\% C\) was prepared and from it was cut

*) See (4), p. 444.

flat section. The structure of the section is shown in Fig. 38 and shows the undoubted presence of eutectic. A chemical analysis was made of one half of this section, which gave a carbon content \(C = 2.006\%\), while the other half was cut into four parts and subjected to the following heat treatment. All four specimens were placed in a Heraeus furnace, heated in a stream of nitrogen to \(1140^\circ\), and held at this temperature for 15 minutes. Then one specimen was quenched, while the others were left to cool slowly. After 5 minutes, at a temperature of \(1135^\circ\), the 2nd specimen was quenched; after another five minutes, at \(1130^\circ\), the 3rd specimen was quenched; the fourth was left to cool slowly together with the furnace. All four specimens showed the presence of eutectic in the same amount as that found in Fig. 38, and this serves as proof that in an alloy with \(2\%\,C\) the eutectic is a structural constituent, i.e. that point \(E\) lies to the left of \(2\%\) carbon.

For a more precise determination of the position of point \(E\), in addition to microscopic investigation, Tammann’s thermal method was also employed\({}^{24}\). Cooling curves were taken for alloys V, VI, and VII and for cast iron with \(3.868\%\,C\). The experiments were carried out in Friedrich’s kryptol furnace.*)

The weight of the buttons was 30 gr. For taking the curves, the finest protective tubes were used, manufactured at Prof. Tammann’s institute. The temperature was read every 10 seconds. The data from these experiments are collected in Table III; the cooling curves are shown in Fig. 39. In determining the durat—

TABLE III.

Alloy	Carbon content, %	Arrests. Temperature °C. Beginning of solidification	Arrests. Temperature °C. Eutectic	Arrests. Temperature °C. Eutectoid (\(Ar_1\))	Duration of the eutectic arrest in seconds
a	3.801	1180	1134	706	110
b	3.596	1199	1130	707	92
c	2.742	1282	1125	708	52
d	1.913	1358	1066	710	13

) See the description in Metallurgie* 1907, p. 781, and 1908, p. 530.

...of the duration of the eutectic arrest, the cooling curves were idealized, as can be seen from Fig. 39.

Fig. 39.

The obtained values of the arrests are plotted as ordinates in Fig. 35. The straight line \(be\), connecting the ends of the ordinates, comes almost exactly to point \(E\) with a carbon content of \(1.7\%\).

A similar verification of the position of the point was also made from the cooling curves of the original alloys V—IX, but in using these curves it is less accurate, since the weights of the alloys were not completely identical; therefore, in order to obtain the duration of the eutectic arrests, it was necessary to introduce the corresponding correction. The two curves \(Eo\) and \(oC\), drawn in Fig. 35 under the eutectic line \(EC\), represent the change in the duration of the eutectic arrests of alloys V—IX. The curve \(Eo\) shows that point \(E\) undoubtedly lies to the left of \(2\%\) carbon.

The results of the experiments just described and the above-mentioned data of microscopic investigation served as the basis for establishing point \(E\) at \(1.7\%\) carbon.

Turning to the diagram in Fig. 35, it must be pointed out that the eutectic line \(EC\) is drawn on it, in accordance with theoretical requirements, horizontally at a temperature of \(1134^\circ\), although the eutectic arrests on curves V, VI, and VII (Fig. 6), as well as on curve \(d\) of Fig. 39, were obtained at lower temperatures.

That these arrests are indeed eutectic is shown with undoubted clarity by the microscopic examination of the alloys belonging to the aforementioned curves: in all alloys with such an arrest, a eutectic was observed: mixed crystals \(+ Fe_3C\) (ledeburite); the lowering of the eutectic arrest must be explained by supercooling,* which is found the more sharply, the lower the carbon content in the alloy. The phenomenon of supercooling in white cast irons is a well-known fact. One may think that such lowered eutectic arrests were taken by Carpenter and Keeling as points belonging to the curves \(aE\) and \(E'R'\) (Fig. 2).

As regards the line \(BD\), a whole series of quenching experiments for determining this line did not give positive results; therefore, in the diagram of Fig. 35 it is drawn with a dotted line, as unknown.** In concluding this description of the quenching experiments, it is necessary to point out that the adopted method of quenching made it possible to determine only the unknown line \(AE\), the position of which had hitherto been established at considerably lower temperatures than has until now been accepted. The accuracy of the determination made, of course, is in direct dependence on the accuracy of measuring the quenching temperature and on the extent to which, by the method employed, it was possible to attain a structurally equilibrium state of the specimens, i.e. complete homogeneity of the mixed crystals.

Absolute accuracy in fulfilling both conditions is impossible to attain. In practice, however, they were fulfilled: the design of the quenching apparatus guaranteed the possible accuracy of measuring the quenching temperature; microscopic examination of the specimens showed their complete homogeneity, and therefore the experimentally determined line of the beginning of melting of alloys from \(0\%\) to \(1.7\%\ C\) must differ from the true position of the solidus \(AE\) only insignificantly, and in any case no more than the other experimentally determined lines of the equilibrium diagram of iron-carbon alloys.

If one takes into account that the existing grades of carbon steel never possess the theoretically necessary homogeneity of structure, then the line \(AE\) obtained by the author practically precisely determines the temperatures of the beginning of melting of steel. The new, lowered position of the line \(AE\) will have great technical significance in questions—

* A similar phenomenon of supercooling was also observed by Goercke in the investigation of the \(Fe + P\) system (Metallurgie 5, 1908, p. 624).

** Further experiments are being carried out by the author to find this line.

from the safe heating of steel, since according to Stansfield’s theory (25) burning of steel is found from the moment when the steel begins to melt, i.e., crosses the line \(AE\) into the region \(AEB\) of the diagram (Fig. 35).

B. Cementation Experiments.

I. In the processes of melting and solidification of iron–carbon alloys, cementation and the decomposition of iron carbide play an important role, since the composition of the mixed crystals that form the principal mass of steel and cast iron depends essentially upon them at high temperatures.

According to the equilibrium diagram of iron–carbon alloys (Fig. 35), the maximum amount of “carbon” that \(\gamma\)-iron is capable of dissolving in the solid state is determined by the line \(SE\). The position of this line has remained unclear up to now, since, on the question of the solubility of “carbon,” the opinions of investigators differ depending on whether they accept two diagrams or one diagram for iron–carbon alloys.

In the first case, owing to the fact that elementary carbon, as the stable form, must have a lower solubility in iron than the metastable form—iron carbide—there must necessarily exist two lines of solubility of carbon in \(\gamma\)-iron: \(S'E'\) and \(SE\): the first indicating the limit of solubility of elementary carbon, and the second that of iron carbide in \(\gamma\)-iron; moreover, the line \(S'E'\), according to what has been stated above, must run throughout its length to the left of the line \(SE\).

In the second case, it is assumed that elementary carbon does not dissolve directly in iron, but forms iron carbide, which then enters into the solid solution. Under this assumption, for determining the solubility of “carbon” in iron, a single line \(SE\) is sufficient, representing the saturation limit of \(\gamma\)-iron with iron carbide, since carburization of iron by elementary carbon, i.e. cementation, within the temperature limits from \(700^\circ\) to \(1134^\circ\), is possible only to the extent to which iron is capable of dissolving the iron carbide formed during cementation.

The explanation of the cementation process based on the theory of “two diagrams” presents uncertainties. It is difficult, for example, to explain why carbon, having dissolved in iron as such, separates out upon subsequent cooling from the solution in the form of iron carbide.

Prof. Howe), an advocate of the theory under consideration, regards it as “natural” that the precipitation of dissolved elemental carbon takes place in the form of the metastable form, cementite. He writes: “the small particles of graphite that first arise, which are formed during its reverse precipitation from the solution, exert a strong pressure on the medium surrounding them, as a result of which*) the further precipitation of carbon already occurs in the form of cementite. The transformation of graphite into cementite means only that part of the elemental carbon dissolves on heating (as such), and on subsequent cooling is again precipitated in the form of cementite.”

If the process of cementite precipitation proposed by Howe is represented graphically, it must appear in the following form (Fig. 40):

Let us suppose that iron, having dissolved in itself a certain amount \(k\%\) of elemental carbon, begins to cool. Then at point \(a\) the precipitation of “small particles of graphite” will occur. Then, owing to the pressure arising from the precipitation of graphite, cementite will begin to precipitate. But the precipitation of cementite can begin only at point \(b\), at which the ordinate of the alloy intersects the line \(SE\). Thus, the process of cementite precipitation proposed by Howe requires that the cemented hypereutectic steel contain graphite (precipitated on the segment \(a,b\)) and cementite. However, numerous observations on cemented steel testify to the absence of graphite in it, which is a refutation of Howe’s proposed explanation of the process of cementite formation.

Fig. 40.

A further difficulty in explaining the process of cementation from the point of view of the theory of two diagrams is the fact that, during cementation, iron absorbs more carbon than corresponds to its saturation limit with elemental carbon, i.e.

*) See (13), p. 106.
**) Howe bases the influence of pressure on the formation of cementite on the well-known law of Le Chatelier (see (13), footnote 16 on p. 79).

line \(S'E'\), and under some conditions even a pure carbide of iron is obtained.

To explain these phenomena, Howe proposes a new theory of the carburizing process, which amounts to the following. Carburization proceeds in a wave-like manner, owing to the slight fluctuations of temperature that occur. When the temperature of austenite saturated with the carburizing substance (graphite) is lowered, cementite precipitates, as a result of which the austenite ceases to be saturated with respect to graphite. With the subsequent rise in temperature, the austenite, now unsaturated with respect to graphite, dissolves it again. This is followed once more by a lowering of the temperature with precipitation of cementite, a rise with dissolution of graphite, etc.

During the periods of temperature rise, the precipitated cementite also undergoes reverse dissolution; but in the layers adjoining the carburizing substance this dissolution is quantitatively greater than in the subsequent layers, and this forces the cementite to diffuse from the surface toward the center at each rise in temperature. The result of such temperature fluctuations and of the associated absorption of graphite by the outer layers and transfer of cementite from the periphery toward the center is a carburization of the iron to a degree greater than would follow from the equilibrium diagram for the stable system.

Regarding the theory set forth by Howe, the following must be said. It is based on two assumptions: 1) that during carburization there occur constant slight fluctuations of temperature, and 2) that during the period of temperature rise the precipitated cementite does not have time to dissolve and therefore the austenite obtains the possibility of dissolving again a certain amount of graphite.

Both of these assumptions have not been confirmed experimentally. Turning to Fig. 40, we see that the outer layer of the object being carburized, saturated with graphite (point \(a\)), can begin to precipitate cementite, according to Howe’s explanation above, only at point \(b\). The temperature interval between points \(a\) and \(b\), given the large distance between the lines \(S'E'\) and \(SE\), as Howe assumes, must be considerable. Significant (constant) fluctuations of temperature in the carburizing process can hardly occur. Moreover, during the rise in temperature from \(b\) to \(a\), even in the event that the precipitated cementite did not have time to dissolve again, the austenite of the surface layer still cannot dissolve graphite, since throughout the entire interval \(b—a\) it is relatively

supersaturated with graphite. Such dissolution can occur only when the temperature is raised above point \(a\), but this is impossible, since the cementation temperature is assumed to be constant.

The second assumption—that during the period of temperature rise the cementite does not have time to dissolve, and that, owing to this, the graphite has time to dissolve in the austenite—contradicts the practical data of quenching and cementation of steel, from which it is known that the dissolution of cementite in iron (the transformation of pearlite into austenite) proceeds much more rapidly than the dissolution (according to Howe) of elementary carbon in iron (the cementation process).

Howe proposed the theory of cementation set forth above chiefly as a way of explaining the possibility of obtaining the carbide \(Fe_3C\) from pure iron by cementation (experiments of Margueritte \((^{26})\), P. Charpy \((^{27})\), Howe and Offerhaus’s*). But it is precisely to these experiments that the theory set forth cannot be applied, since it considers the phenomena of cementation, i.e. carburization by elementary carbon at temperatures above \(700^\circ\), at which the question of the solubility of carbon in iron plays the primary role; whereas in the experiments mentioned, carburization was carried out not with elementary carbon, but with carbon monoxide and potassium cyanide at temperatures below \(700^\circ\), i.e. in that temperature range where cementite is completely insoluble in iron. The formation of cementite under these conditions must be regarded as a definite chemical reaction of formation of iron carbide, having no relation to Prof. Howe’s theory of cementation.

The process of cementation receives a satisfactory explanation if, as was indicated above, one assumes that the dissolution of elementary carbon in iron occurs through the formation of iron carbide. In such a case, cementation of iron at a given temperature between \(700^\circ\) and \(1134^\circ\) can proceed only until the iron is saturated with iron carbide, i.e. the displacement of the figurative point of the composition of the iron alloy to the right will continue until it meets the line \(SE\). This is required by the theory of a single diagram, and is confirmed also by some experimental data.

The proposed explanation of the cementation process is not in contradiction with the circumstance that iron carbide at temperatures above \(700^\circ\) is an unstable compound, i.e. that the reaction:

* Metallurgie, vol. VI (1909), p. 82.

\[ Fe_3C \rightleftarrows 3Fe + C \]

at a temperature above \(700^\circ\) may proceed from left to right. Indeed, free carbide at a temperature above \(700^\circ\) decomposes (on this is based the process of obtaining malleable cast iron), but the carbide present in solution does not undergo decomposition, and therefore the formation of carbide above \(700^\circ\) can occur provided that the carbide formed dissolves in the iron surrounding it. Such a course does not contradict the laws of chemistry. Thus, the movement of the reaction:

\[ Fe_3C \rightleftarrows 3Fe + C \]

from right to left can proceed until all the iron proves to be saturated with carbide. The movement of the reaction from left to right, i.e. the decomposition of cementite, is possible only for free cementite not present in solution*).

II. From the stated proposition concerning the process of cementation it follows that the carburization of iron at a given temperature (between \(700^\circ\) and \(1134^\circ\)) must come to an end at the moment when the iron proves to be saturated with respect to cementite.

On the constitutional diagram (Fig. 35), the line \(SE\), showing the limit of the solubility of cementite in \(\gamma\)-iron, was determined by Prof. Heyn only over the interval indicated by the elongated dotted line. The further course of this line is unknown. If the point \(E\), determined by the author at \(1.7\%\,C\), is joined with the end of Heyn’s line, one obtains the smooth curve \(SE\), which should determine the end of cementation for various temperatures.

To verify the position of the unknown part of the branch of the line \(SE\), experiments on cementation and annealing were carried out, having as their purpose the determination of the maximum quantity of combined carbon that can

*) This decomposition takes place during graphitization, and also during the annealing of white cast iron. It proceeds the more rapidly the higher the temperature, and only up to the point at which free cementite is present. White cast iron after annealing must represent a conglomerate of annealing carbon and a metallic mass containing as much combined carbon as corresponds to the limit of saturation of iron with cementite at the annealing temperature. If in practice there are cases where wrought castings contain considerably less combined carbon than the constitutional diagram requires, or contain none at all, this is explained by the influence of silicon (Wüst’s experiments\(^{28}\)), which promotes the decomposition of carbide by lowering the ability of iron to dissolve it, and also by the direct burning-out of carbide from the solid solution during oxidizing annealing (Hadfield’s theory\(^{29}\)). The decomposition of carbide is apparently favored by a decrease in the pressure of the medium, as indicated by experiments on annealing cast iron in an airless space (Wüst\(^{30}\)).

to form in iron by cementation of it with elemental carbon. In this connection, in order to confirm the propositions put forward, it was of interest to trace the various forms of cementation, namely: external cementation—with the aid of a cementing substance (carbon), and internal cementation—by means of graphite in gray cast iron and by means of temper carbon—in white annealed cast iron,—in the latter case in order to check the limit of decomposition of free cementite.

Therefore, for the cementation experiments the following materials were taken:

1) Swedish iron (N. E.): \(C = 0.041\%\),

2) White Swedish cast iron (S. R.): \(C = 3.746\%\); graph. \(= 0\%\);

3) Gray finely graphitic cast iron, obtained from the preceding white cast iron by remelting:*) \(C = 3.674\%\); graph. \(= 2.263\%\) and bound \(C = 1.411\%\).

The cementation experiments were carried out in a horizontal furnace of Heraeus. The results of the experiments are summarized in the following Table IV: see Table IV on p. 43.

In experiments No. 36–41, cementation (with graphite) was carried out on iron cylinders (\(d = 8\) mm., \(h = 20\) mm), placed in graphite crucibles. These experiments did not give positive results. They only showed that cementation under the conditions of the experiments proceeds very slowly and that it is stronger the higher the cementation temperature.**)

In experiments 42–43 bone charcoal was used, and all three materials were cemented simultaneously. In these two experiments the cementation of the cylinders also proved insignificant, despite 24-hour heating, and therefore the experiments with cylinders were discontinued. Experiment 42 showed that the cementite of white cast iron did not decompose at \(1080^\circ\) even after 3 hours of heating; this, in agreement with the experiments of Tiemanns’ (31), indicates that the decomposition of cementite below \(1100^\circ\) proceeds slowly. In experiment 43, after 24 hours’ heating, the cementite decomposed and, as microscopic examination showed, entirely without residue, the amount of combined carbon being found by analysis to be \(1.607\%\), which, according to the line \(SE\), corresponds to the solubility of cementite in iron at \(1080^\circ\).

*) During remelting the solidification interval was deliberately accelerated and the subsequent cooling slowed, as a result of which the graphite was obtained in the form of thin and small plates.

) The influence of temperature on the rate of cementation was studied in detail by Giolitti and Carnecalli (Giess.-Zeit. 1909, Nos. 1 and 2).

TABLE IV.

Experiment No.	Material	Duration of experiment, hours	Temperature of experiment, \(T^\circ C.\)	Analysis: total C, %	Analysis: graphite, %	Analysis: combined C, %	Method of cooling	Notes
36	N. E.	1	1050	—	—	—	Cooled slowly.	Not cemented.
37	N. E.	3	1040—50	—	—	—	Cooled slowly.	Slightly cemented at the edge.
38	N. E.	17	1040—50	—	—	—	Cooled slowly.	Cemented at the edge.
39	N. E.	1	1100	—	—	—	Cooled slowly.	Slightly cemented at the edge.
40	N. E.	3	1100—05	—	—	—	Cooled slowly.	” ” ”
41	N. E.	24	1100—10	—	—	—	Cooled slowly.	Cementation did not penetrate to the middle.
42	N. E.	3	1080—90	—	—	—	Quenched.	Slightly cemented at the edge.
42	S. R.	3	1080—90	—	—	—	Quenched.	Cementite did not decompose.
42	G. R.	3	1080—90	3,672	2,084	1,592	Quenched.
43	N. E.	24	1080—90	0,559	—	—	Quenched.	Cemented only at the edge.
43	S. R.	24	1080—90	3,726	2,119	1,607	Quenched.	Complete decomposition of cementite.
43	G. R.	24	1080—90	—	—	—	Quenched.	Charcoal at the edges.
44	N. E.	1	1115—20	1,635	—	1,635	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
45	N. E.	3	1115—20	1,659	—	1,659	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
45	G. R.	3	1115—20	3,672	2,021	1,651	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
46	N. E.	5	1110—20	1,661	—	1,661	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
47	N. E.	3	940—50	1,213	—	1,213	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
48	N. E.	8	940—50	1,258	—	1,258	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
49	N. E.	13	940—50	1,257	—	1,257	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
50	N. E.	15	1020—30	1,453	—	1 455	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
51	N. E.	15	1060—70	1,595	—	1.595	Quenched.	In cementation experiments No. 41—51 (with shavings), microscopic examination of the specimens could not be carried out.
63	S. R.	1/4	1135	3,531	1,818	1,713		Cementite decomposed, beginning of melting.

In experiments 44–51, to accelerate the cementation process, a mixture of 60% sugar charcoal and 40% barium carbonate was used, and iron shavings were subjected to cementation*).

These experiments showed that (under the experimental conditions) cementation at \(940^\circ\)—\(950^\circ\) reaches its limit within 8 hours (experiments 46–48), since during cementation for 13 hours the amount of absorbed carbon did not increase. Cementation at a temperature of \(1115^\circ\)—\(1120^\circ\) ends after three hours (experiments 44–46). Experiments 50 and 51, as controls, were carried out for 15 hours, but even under these conditions the amount of absorbed carbon proved to correspond to the line \(SE\).

In experiments 42 \((G. R.)\) and 45 \((G. R.)\), cementation occurred chiefly at the expense of the graphite of the cast iron: the graphite content decreased, and the amount of combined carbon (1.411%) increased: in the first case to 1.592%, and in the second to 1.651%. These two experiments, in comparison with experiments 36–41, showed that internal cementation proceeds considerably faster than external cementation, which is understandable, since the cementing substance—graphite in the first case—is distributed throughout the entire mass of the metal.

Finally, experiments 43 \((S. R.)\) and 63 (from the following series) established: 1) that only free cementite undergoes decomposition: in experiment 43 \((S. R.)\) the amount of combined carbon (cementite) remaining in the cast iron corresponds to the saturation limit of iron with cementite at the given temperature; and 2) that decomposition of cementite occurs much faster at high temperatures \((1135^\circ)\) than at lower ones, for example \((1080^\circ)\).

The results of experiments Nos. 42, 43, 45, 46, 49, 50, 51 are represented graphically in Fig. 41,

Fig. 41.

*) excluding experiment 45 \((G. R.)\), in which, as also in experiment 42 \((G. R.)\), the cast iron was in the form of a cylinder of the indicated dimensions.

from which it is evident that the points representing the amount of carbon absorbed by iron during cementation lay on a curve which, practically exactly, is a continuation of the curve \(Sc\); the latter, according to Heyn’s experiments, represents the line of the limit of dissolution of cementite in \(\gamma\)-iron.

The results of the experiments described, confirming the proposition stated concerning the course of the cementation process, are in agreement with the investigations of Royston \((^{32})\), who, subjecting malleable cast iron to internal cementation at a temperature of \(1030^\circ\), and carrying out annealing of white cast iron at the same temperature, obtained in both cases a content of combined carbon equal to \(1.5\%\). They differ considerably, however, from the results of experiments by other investigators, for example, Charpy and Grenet \((^{33})\), Heyn and Bauer \((^{34})\), Charpy \((^{35})\), Benedicks \((^{36})\), and Hadfield \((^{29})\).

According to the data of the authors listed (except Hadfield), the solubility of elementary carbon is lower, and according to Hadfield’s data higher, than in the author’s experiments.

With respect to the investigations of Charpy and Grenet, and of Heyn and Bauer, it is necessary to point out that the materials which they used contained a considerable quantity of silicon, which, as is known, lowers the solubility of cementite in iron; therefore the data obtained by them cannot serve for determining the line \(SE\).

In Charpy’s experiments \((^{35})\) the conditions of the experiments are not indicated. The necessity of removing, before analysis, in experiments (a), a layer \(2\ \mathrm{mm}\) thick permits the supposition that oxidation (decarburization) occurred during heating. Such oxidation is mentioned in the description of the experiments of series \(b\). Since, in oxidative annealing, according to Hadfield’s theory, cementite burns out directly, it may be supposed that oxidation in the experiments reduced the content of combined carbon in the cast iron, as a consequence of which the “solubility of graphite in iron at \(1000^\circ\)” was found to be only \(1.00\%\,C\).

Benedicks did not determine analytically in his experiments the solubility of graphite in iron. For a temperature of \(910^\circ \pm 10^\circ\) he considers it, from the appearance of the structure, to be equal to from \(0.9\%\) to \(1.2\%\,C\). The latter figure comes close to the author’s data.

Finally, according to Hadfield’s determination, the solubility of cementite in iron at \(960^\circ\) is equal to \(1.63\%\,C\), which considerably exceeds the author’s data. However, this figure is hardly correct: Hadfield cemented steel with \(1.64\%\,C\) in charcoal at a temperature of \(960^\circ\). After the completion of cementation, the content of combined carbon in the steel

had not changed, whence Hadfield concludes that \(1.63\%\,C\) (the analysis datum after cementation) is the limit of saturation of iron with cementite. If, however, one takes into account that steel with \(1.64\%\,C\) at a temperature of \(960^\circ\) may consist of austenite saturated with cementite at this temperature and of excess cementite not entering into solution, then it is natural that the amount of carbon could not have been increased by cementation. The analysis of the steel, however, could not give the true value of the solubility of cementite in austenite, since the carbon content of \(1.63\%\) included an unknown quantity of carbon from cementite remaining in excess.

IV. Theory of the Melting and Solidification of Iron–Carbon Alloys

After all the experiments set forth above, whose aim was to verify the equilibrium diagram of iron–carbon alloys, to find the lines missing from it, and to consider the phenomena of cementation and annealing as processes participating in the melting and solidification of cast iron and steel, it appears possible to proceed to the investigation of the actual processes of melting and solidification.

At the present time two theories of the melting and solidification of cast iron are known in the literature: Heyn’s theory and Goerens’s theory.

These two theories agree with one another in their consideration of the process of melting and solidification of white cast iron. Both of the named authors regard this process as reversible, as a result of which they assume that the melting and solidification temperatures of white cast iron coincide with one another.

With regard to gray cast iron, both theories, although they acknowledge the processes of melting and solidification of gray cast iron to be irreversible, set forth the actual course of these processes in entirely different ways.

a) Heyn’s Theory

Prof. Heyn \((^{37})\) proposed in 1906 an explanation of the processes of melting and solidification of white and gray cast irons, and in 1907, in a work together with Bauer \((^{34})\), developed the positions expressed earlier, supporting them with experimental data.

In the work of 1906 Heyn describes the processes of solidification of cast irons in the following manner.

White cast iron solidifies with supercooling, following the equilibrium diagram, i.e., mixed crystals are first precipitated (the case of hypoeutectic cast iron), then at \(1130^\circ\) the eutectic solidifies: mixed crystals + cementite; further, secondary cementite is precipitated from the mixed crystals and, finally, at \(700^\circ\) pearlite is formed.

Gray cast iron, however, exhibits considerable deviations from the theoretical course of solidification, owing to the formation of graphite: during its solidification mixed crystals are first precipitated; then, at a temperature of \(1130^\circ(t_2)\), the eutectic solidifies: mixed crystals + cementite. In view, however, of the fact that cementite as

free, as well as that which is in solution, does not undergo decomposition only at temperatures \(t_x^*\) higher than \(t_2\); its existence at temperatures below \(t_2\) is possible only as a result of supercooling, which prevents it from decomposing into graphite and iron (white cast irons). This decomposition (i.e., graphitization) may be complete or partial, if the supercooling is destroyed by some factor (Anreiz). Heyn assumes that the destruction of the supercooling, i.e., the onset of graphitization, occurs at a temperature below \(t_2\), i.e., after the solidification of the eutectic, and depicts this moment in drawing \(8d'\) (Fig. 42): in the mass of martensite (\(S\)) two nuclei (Keim) of graphite appear.

Fig. 42.

With a further lowering of temperature these small plates of graphite grow, drawing carbon out of the martensite surrounding them; as a result, around them there forms an area of iron depleted in carbon (drawing \(8d''\)). The growth of the graphite plates continues down to a temperature of \(700^\circ\) (and perhaps even lower).

In his work of 1907, Heyn, confirming the propositions set forth above, dwells in greater detail on the processes of melting of cast irons, illustrating them with the diagram presented in Fig. 43.

Heyn\(^9\) distinguishes two systems:

\[ \text{the stable } b:\ \text{iron} + \text{graphite} \]

and the unstable \(a\): iron + carbide.

Gray cast iron constitutes the stable system, white—the unstable one.

The processes of melting and solidification according to the diagram of Fig. 43 take place as follows:

Liquid cast iron should solidify in the temperature region \(t_b\) in the form of the stable system \(b\). “However, at this tempera—

*) See (37), p. 1389.

at \(t_b\) the conditions are apparently unfavorable for the separation of graphite; therefore cast iron is supercooled and solidifies in the somewhat lower-lying temperature region \(t_a\), in the form of the unstable system \(a\).

If now, at temperatures below the region \(t_a\), no factor appears capable of destroying the supercooling, then the cast iron will solidify in the form of white iron, and the solidification process will be represented in Fig. 43 (\(A\)) by the solid arrow \(1—2\). If such cast iron is heated, and if during heating no transition into the stable system occurs, then the cast iron will melt, following the dotted arrow \(2'—1'\), the processes of solidification and melting occurring at one and the same temperature, i.e. they will be reversible.

Fig. 43.

If, however, after the solidification of cast iron into the unstable system a factor appears that destroys the supercooling, then solidification (with graphite separation) will proceed along the solid arrow \(1—2—3\) (Fig. 43 \(B\)), as a result of which the stable system \(b\)—gray cast iron—will be obtained.

The melting of gray cast iron will proceed along the dotted arrow \(3'—1'\), i.e. entirely within the region of stable equilibrium, since there is no basis for gray cast iron to pass into the region of unstable equilibrium; therefore the melting temperature of gray cast iron (region \(t_b\)) will be higher than the temperature of its solidification (\(t_a\)), which is also in agreement with practical data.

The essential features of the above theory of Heyn are the following:

1) The melting and solidification of white cast iron follow the equilibrium diagram.

2) The melting and solidification of gray cast iron deviates from it as a result of graphite separation.

3) Graphite separation begins after the solidification of the eutectic mixed crystals + cementite, is the consequence of the elimination of undercooling by means of some factor, and ends at about 700°.

4) The plates of graphite originate in the mixed crystals (martensite) and grow at the expense of the carbon of the latter.

5) The eutectic—mixed crystals + cementite—apparently takes no part at all in graphite formation, or else participates in it indirectly, giving up its cementite, through the thickness of the crystals, to the plates of graphite.

The theory set forth, while not illuminating with sufficient clarity the various aspects of the processes of melting and solidification of cast irons, raises, in the author’s opinion, doubts as to the correctness of some of its propositions, since:

1) it has not been experimentally confirmed to a sufficient degree,

2) it does not take into account two essential factors: cementation and the rate of heating (and cooling), which undoubtedly influence the processes of melting and solidification of cast irons,

and 3) because the experimental investigations of Goerens and of the author concerning graphite formation are in contradiction with points 3, 4, and 5, which assert that graphite originates in martensite after the solidification of the eutectic, whose cementite only indirectly participates in the formation of graphite plates.

b) Goerens’s Theory.

Prof. Goerens regards the process of melting and solidification of white cast iron as reversible and as proceeding theoretically correctly. The production of white cast iron (in pure iron–carbon alloys), however, is possible only under the condition of rapid cooling, which prevents the decomposition of cementite into its constituent parts: iron and graphite.

If, on the other hand, cooling proceeds slowly, then the cementite separating during the solidification of the eutectic (as well as the primary cementite separating along the line \(BD\)) undergoes decomposition and, depending on the rate of cooling, either gray or mottled cast iron is obtained.

Thus, according to Goerens’s theory, graphite separation is the decomposition of cementite and, chiefly, of the cementite of the eutectic, since the decomposition of primary cementite gives graphite in

in the form of so-called “speiss.” Graphite separation therefore coincides with the interval of solidification of the eutectic, and the longer this interval—that is, the slower the cooling—the more completely graphitization takes place, and the coarser, i.e., larger in size, are the graphite plates obtained. If the decomposition of cementite during the period of solidification of the eutectic has not occurred completely, despite slow cooling, then its decomposition is possible even after the end of solidification of the eutectic; but then the product of decomposition will be carbon in the form of temper carbon.

For the melting process of gray cast iron, Goerens proposes the following explanation (Fig. 44):

Fig. 44.

The metallic mass of gray cast iron is steel; therefore the melting temperature of cast iron must depend on the amount of combined carbon contained in it.

Let us suppose that there is cast iron with \(1.8\%\) combined carbon. On heating such cast iron, it will begin to melt at point \(c\), i.e., at the point of intersection of the ordinate of the composition of its metallic mass with the line \(AE\). From this moment it enters the region \(ABE\), in which mixed crystals are in equilibrium with a carbon-rich liquid phase. With a slight increase in temperature, for example to \(T^\circ\), the homogeneous solid mass decomposes into a liquid phase with \(a\%\ C\) and mixed crystals with \(d\%\ C\). Since at temperature \(T^\circ\) the liquid phase can dissolve up to \(e\%\ C\), it will tend to dissolve graphite, thereby forming carbide. As a result, the equilibrium established for temperature \(T^\circ\) will be disturbed; it can be restored only if the mixed crystals also change their composition until they come into equilibrium with the liquid phase of the new composition.

If we assume that the liquid phase \(a\) dissolves carbon in itself up to a content of \(f\%\), then only mixed crystals containing \(h\%\ C\) can be in equilibrium with a liquid of such composition, as is clear from the graphical construction made in Fig. 44.

Thus, in order to restore the equilibrium disturbed by the fact that the liquid phase has dissolved some quantity of graphite, the mixed crystals must increase their carbon content by dissolving part of the carbide contained in the liquid phase. As a result of this the liquid phase will become poorer in carbon and will again acquire the possibility of dissolving some quantity of graphite, and so on.

Such a cycle will continue until, at the given temperature, a certain equilibrium state is established.

It is self-evident that the less combined carbon the gray cast iron contains, the higher the point \(c\) will lie, i.e., the higher will be the melting temperature of the given cast iron, which is also in agreement with experimental investigations.

At the basis of the theory set forth concerning the melting of gray cast iron lies the assumption that, in order to establish equilibrium between the mixed crystals and the liquid phase, the temperature of the alloy must now fall, now rise. If one takes into account that the saturation of the liquid phase with carbon from \(a\) to \(f\) cannot, as Goerens assumes, take place at constant temperature, since the liquid alloy in the present case is situated in the region \(ABD\), i.e., in the region of divariant equilibrium, then the amplitudes of the temperature oscillations must be greater than Goerens indicates, and this is hardly in accord with reality. Moreover, in Goerens’ theory the rate of heating is not taken into account, and it cannot fail to be reflected in the course of internal cementation. Both of these circumstances make one doubt that the melting of gray cast iron actually occurs in the way Goerens assumes.

Heyn’s theory and especially Goerens’ theory give many important indications for clarifying the processes of melting and solidification of cast irons. However, in some of their propositions they have not been confirmed experimentally; they put forward assumptions that hardly have a place in reality; and, finally, they do not take into account the influence of the rate of heating in the process of melting gray cast irons. All this, taken together, deprives these theories of their full applicability for explaining certain questions in the processes of melting and solidification of cast irons and calls for further study of the processes mentioned.

For the stated purpose the author carried out the following experiments.

c) Experimental investigation of the process of melting cast irons.

The investigation of changes in the structure and chemical composition of white and gray cast irons during their gradual heating up to mel—

melting was carried out by means of the quenching method. The experimental setup was the same as in determining the line \(AE\); the samples (for performing chemical analysis) weighed 10 gr.

The materials used for the experiments had the following composition:

White Swedish cast iron:

\(C = 3.746\%\); \(Si = 0.032\%\); \(Mn = 0.18\%\); \(P = 0.024\%\) and \(S = 0.0076\%\)

and gray cast iron:

\[ \text{total } C = 3.674\% \text{ and graph. } = 2.263\% \]

obtained by remelting*) white cast iron of the indicated composition.

The duration of heating of the samples at various temperatures was established on the basis of the author’s previous experiments \((^{15})\) in the study of the graphitization process.

All samples after quenching were polished and examined under the microscope, and then the total content of carbon and graphite in them was determined by chemical analysis.

The data and results of the experiments are summarized in Table V: see Table V on p. 54.

The results of the experiments are also shown graphically in Figs. 45 and 46.

The curve of Fig. 45 represents the change in the graphite content in gray cast iron during slow heating up to its melting. This curve undoubtedly indicates that the quantity of graphite gradually decreases, i.e., that during the process of heating and melting of cast iron an internal cementation takes place. From Fig. 45 it is also seen that, at the rate of heating which occurred in the experiments, the metallic mass of the cast iron does not change its composition up to \(1000^\circ\). Beginning with this temperature, a cementing

Fig. 45.

*) See footnote on p. 42.

TABLE V.

Material	Experiment No.	Temperature of the sample at the moment of quenching, \(T^\circ C\)	Duration of heating at \(T^\circ\), minutes	Analysis: total \(C\), %	Analysis: graphite, %	Analysis: combined \(C\), %	Figure No.	Structure of the quenched samples
Gray cast iron	52	1000	30	3.593	2.212	1.381		Martensite and graphite.
Gray cast iron	53	1070	25	—	2.011	—		″ ″
Gray cast iron	54	1120	20	3.498	1.794	1.704		Austenite and graphite.
Gray cast iron	55	1130	15	—	1.643	—	47	X-structure and graphite.
Gray cast iron	56	1140	12	—	1.485	—		X-structure, graphite, and ledeburite.
Gray cast iron	57	1150	12	3.506	0.726	2.780	48	The quantity of ledeburite increased.
Gray cast iron	58	1160	10	—	Traces	—		White cast iron.
Gray cast iron	59	1175	10	3.496	None	3.496		″ ″
White cast iron	60	1050	40	3.621	None	3.621		White cast iron.
White cast iron	61	1120	20	—	Traces	—		″ ″
White cast iron	62	1130	15	—	0.420	—		″ ″
White cast iron	63	1135	15	3.531	1.818	1.713	49	X-structure and elemental carbon.
White cast iron	64	1140	12	—	1.741	—		X-structure, elemental carbon and ledeburite.
White cast iron	65	1150	12	3.560	1.024	2.536	50	The quantity of ledeburite increased.
White cast iron	66	1160	10	—	Traces	—		White cast iron.
White cast iron	67	1175	10	3.561	None	3.561		″ ″

the action of graphite), which manifests itself the more strongly the higher the temperature, reaching its maximum* between \(1120^\circ\)—\(1150^\circ\).

Fig. 47 (experiment 55) represents the structure of gray cast iron before the beginning of melting. The cast iron consists of small plates of graphite lying in a basic mass resembling austenite. The structure of this mass, consisting of homogeneous crystals penetrated by regularly arranged thin brown-colored needles, was called in the preceding work of Goerens and the author \((^{15})\) “purely austenitic.” In view, however, of the fact that this structure differs considerably from the structures of normal austenite recently published by Maurer \((^{38})\), and at the same time resembles none of the already known structural constituents of iron-carbon alloys, it will hereafter be called the \(X\)-structure.

Fig. 48 (experiment 57) shows the structure of gray cast iron during the period of melting (\(T_{\text{quench}} = 1150^\circ\)). The characteristic structure of gray cast iron has disappeared. Instead of it, in Fig. 48 mixed crystals and ledeburite are visible, i.e. the structure of white cast iron.

Analysis of the specimen from experiment 58 gave traces of graphite, which indicates the completion, at a temperature of about \(1160^\circ\), of the internal process of cementation.

Thus, the present series of experiments proves that, during the melting of gray cast iron, an internal process of cementation by graphite takes place. The formation of cementite and its dissolution in \(\gamma\)-iron occurs both in the region of solid solutions, i.e. in the area \(AESGA\) in Fig. 35 (experiments 52—55), and during the melting period, when the cast iron is located in \(\triangle ABE\) (experiments 56—58).

Fig. 46.

The investigation of the melting process of white cast iron showed that, under the conditions of the experiments, the cementite of white cast iron, at a temperature close to the eutectic \((1134^\circ)\), decomposes into its constituent parts. This phenomenon is clearly represented by Fig. 46. It is seen on it that the line,

* It is understandable that cementation must manifest itself the more strongly the more slowly the heating proceeds. It is possible that, with heating slower than in the experiments, the cementing action of graphite begins below \(1000^\circ\).

representing the content of elemental carbon in white cast iron makes a sharp upward jump at \(1120^\circ\), i.e., up to \(1120^\circ\) the white cast iron contained no elemental carbon, while, beginning with this temperature, chemical analysis detected the appearance of elemental carbon.

As the experiments showed, the decomposition of cementite proceeds very rapidly in the interval \(1120^\circ\)—\(1135^\circ\).

At \(1135^\circ\) (experiment 63), i.e. at the temperature of the beginning of melting (eutectic), all the excess cementite had decomposed, since the amount of combined carbon in the cast iron (1.715%) corresponds to the solubility limit of cementite in \(\gamma\)-iron at the eutectic temperature. Thus, before the beginning of melting, white cast iron is a mixture of crystals of the \(X\)-structure and elemental carbon (graphite or temper carbon), as is confirmed by Fig. 49, i.e. it passes into the stage of gray cast iron, after which the melting process already begins. Naturally, in the subsequent course the melting process of white cast iron cannot differ from that of gray cast iron, since the phenomena of internal cementation must be the same in both cases. This is also confirmed by the form of branch \(de\) of the curve (Fig. 46), which is similar to the corresponding part \(cb\) of the curve for gray cast iron in Fig. 45. The structure of the cast iron in Fig. 50 (experiment 65, \(T_{\text{quench}} = 1150^\circ\)), which does not differ in any way from the structure of gray cast iron shown in Fig. 48 (\(T_{\text{quench}} = 1150^\circ\)), also speaks to this.

Thus, the second series of experiments indicates that, during the melting of white cast iron, depending on the rate of heating, decomposition of cementite (partial or complete) may take place. It must be assumed that, with slow heating of cast iron in technical melting, such decomposition of cementite does occur, and therefore white cast iron during melting passes through the stage of gray cast iron and the very process of melting is not reversible, as the theories of Heyn and Goerens suppose.

From the preceding experiments it became clear that, before melting and during the period of melting of cast iron, internal cementation takes place. Since the cementation process requires a definite interval of time for its completion, there can be no doubt that the rate of heating of cast iron must influence the process of melting of cast iron in the sense that melting of cast iron should begin at the lower temperature, the more slowly the heating proceeds.

To verify this proposition, the following experiments were carried out on melting cast iron at different rates.

Table II.

Fig. 51.

Visible labels in the diagram:

• \(T\)
• \(t\)
• \(3\), \(5\), \(9\)
• \(3'\), \(5'\), \(9'\)
• seconds
• large circle
• \(100\) sec.
• \(0, 60, 120, 180\) sec.
• \(500, 600, 700, 800, 900, 1000, 1100, 1200\)

Table III.

Graph labels visible in the figure: \(4\), \(6\), \(6'\), \(10\), \(10'\); seconds (sec.); gray cast iron.

Fig. 52. N. Gutovsky. On the theory of the iron–carbon system.

Tomsk, Typo-lithography of the Siberian Partnership “Printing Business.”

Into the kryptol furnace (described on p. 14), heated to a certain constant temperature, though different in the different experiments, a crucible was placed with the finely crushed white cast iron of the composition indicated above (p. 53). The temperature of the furnace (1200°, 1300°, 1400°, 1500°, and 1600°), which remained unchanged during the experiment of recording the heating curves, determined a definite rate of the melting process. After the melting had ended, the crucible was removed from the furnace, placed in a tripod heated to 500°, and the cooling curve was recorded from the cast iron.

Following this, from the same cast iron, which as a result of remelting becomes gray*), at the same furnace temperature, the heating curve and then the cooling curve were recorded.

In Figs. 51 and 52 are presented the characteristic curves obtained in these experiments, the data for which are placed in Table VI.

TABLE VI.

Curve No.	Heating: rate of heating in the interval 900°—1100° C, in seconds per 100° — white cast iron	Heating: rate of heating in the interval 900°—1100° C, in seconds per 100° — gray cast iron	Heating: temperature of the beginning of melting, °C — white cast iron	Heating: temperature of the beginning of melting, °C — gray cast iron	Heating: duration of the eutectic arrest, in seconds — white cast iron	Heating: duration of the eutectic arrest, in seconds — gray cast iron	Cooling: beginning of solidification, °C — white cast iron	Cooling: beginning of solidification, °C — gray cast iron	Cooling: temperature of the eutectic arrest, °C — white cast iron	Cooling: temperature of the eutectic arrest, °C — gray cast iron
1	45	—	1145	—	76	—	1194	—	1134	—
2	—	57	—	1150	—	158	—	1198	—	1135
3	61	—	1141	—	68	—	1200	—	1133	—
4	—	64	—	1149	—	172	—	1199	—	1134
5	80	—	1139	—	226	—	1198	—	1132	—
6	—	93	—	1144	—	240	—	1200	—	1134
7	90	—	?	—	216	—	1201	—	1134	—
8	—	105	—	1134	—	240	—	1198	—	1136
9	177	—	?	—	410?	—	1200	—	1134	—
10	—	148	—	1134	—	390	—	1199	—	1132

*) This was verified by breaking the bead.

The form of the curves and the data of Table VI confirm the stated supposition concerning the influence of the rate of heating on the melting processes of gray and white cast irons: the temperature at which melting begins lies higher the more rapidly the cast iron is heated. This influence is expressed more sharply in gray cast iron than in white.

As for the cooling curves, in both cast irons they have the same form, irrespective of the rate at which they were heated. This is natural, since the rate of cooling in all the experiments was practically the same, so that white cast iron, at the given rate of cooling, transformed into gray. (The production of white cast iron could take place only with more rapid cooling, in which the cementite would not have the opportunity to decompose during the period of solidification of the eutectic.)*

The present experiments, in connection with the results of the author’s preceding works, make it possible to propose the following explanation of the processes of melting and solidification of cast irons.

d) Theory of the melting and solidification of cast irons.

i) Melting of gray cast iron.

Let us suppose that the cast iron subjected to melting contains \(p^0/0\) carbon (Fig. 53). The quantity \(\alpha\) of chemically bound—

Fig. 53.

* In addition to the rate of cooling, the decomposition of cementite may apparently depend on the discrepancy between the maximum values \(C. K.\) and \(C\text{-}C. K.\), which regulate the process of crystallization [see Tammann (18)].

of bound carbon is equal to \(a\%\)**). The difference \((p-a)\%\) represents the graphite content.

Such cast iron must be regarded as a steel with \(a\%\,C\), permeated by plates of graphite. If graphite did not exert, at high temperatures, a cementing action, then the beginning of melting of the cast iron under consideration would occur at the point of intersection of the ordinate \(a\) with the line \(AE\) (Goerens’s theory). If the graphite were to exert no influence subsequently either, then we would have the case of melting of steel, proceeding theoretically correctly, as was set forth on p. [[unclear: page number]] (case 1). But when gray cast iron is heated, internal cementation in fact takes place; therefore the figurative point of the composition of the metallic mass of the cast iron will move vertically upward only to some point \(b\), starting from which the composition of the metallic mass of the cast iron will change as a consequence of internal cementation, and the figurative point of the composition will follow one of the paths designated by the numbers: \(1, 2 \ldots n\). The position of point \(b\) and of the curves \(1, 2 \ldots n\) depends on the rate of heating. The greater the rate of heating, the higher point \(b\) lies; the steeper the path \(b-n\), the higher the temperature at the beginning of melting of the cast iron, which is the point of intersection of one of the paths \(b-n\) with the line \(AE\).

Whatever the composition of the metallic mass of the gray cast iron may be (or, in general, the composition of the cast iron), one can always suppose such a rate of heating at which the figurative point of composition will move along some path intersecting the line \(AE\) at the point \(E\), i.e. the beginning of melting will coincide with the eutectic temperature \((1134^\circ)\), and therefore the assertion of the previous theories that gray cast iron always melts at a temperature higher than white cast iron is theoretically incorrect.

From the moment when the figurative point passes into the triangle \(ABE\) (let us suppose along the path \(b-2\)), the metallic mass of the cast iron will begin to melt, i.e. it will decompose into a solid phase (mixed crystals) of composition \(f\%\,C\) and a liquid phase of composition \(k\%\,C\). Subsequently the melting process cannot proceed, as in steel, i.e. vertically upward, because of the presence of an excess solid phase—graphite—which manifests its cementing action.

All the heating curves of gray cast iron showed distinctly a halt with a horizontal segment. Such a halt indicates—

**) The amount of bound carbon in cast irons varies. It is often less than \(0.90\%\), which, however, does not affect the essence of the proposed explanation of the melting process.

testify that in the melting of gray cast iron a certain process takes place at constant temperature; the course of this process can be assumed only in the case where the composition of the phases taking part in it remains constant. Consequently, on the horizontal portion of the heating curves the composition of the phases—liquid \(k'\), solid \(f\), and graphite—does not change; only their quantities change.

Graphically, this period of the melting process will be represented by the fact that the figurative point of the composition will move along the horizontal from \(f\) to \(e\), and this motion will continue until one of the phases, namely graphite, is consumed, i.e. until internal cementation is completed. The latter will occur at the point \(e\), corresponding to the initial composition of the cast iron (\(p^0/\!_0\) carbon). Starting from here, the melting process will proceed theoretically correctly upward along the vertical \(eg\). The temperature at which the melting of the cast iron ends is determined by the intersection of the ordinate of its composition with the line \(AB\), i.e. by the point \(g\).

Thus, in the melting of gray cast iron, three periods must be distinguished:

1) cementation of the solid metal,
2) formation of a liquid phase by cementation,
3) melting proper.

In the first period the motion of the figurative point of the composition of the metallic mass proceeds from \(a\) to \(f\), in the second from \(f\) to \(e\), and in the third from \(e\) to \(g\).

The course of the process of internal cementation in the first period determines the temperature at which the melting of the cast iron begins, which is, therefore, for cast iron of one and the same composition, a variable quantity depending on the rate of heating.

The temperature at which the melting of gray cast iron ends depends on the combined course of the second and third periods; and since in the course of these periods certain deviations from the described course are possible, the temperature at which the melting of gray cast iron of a given composition ends is likewise a nonconstant quantity.

In fact, during the second period a liquid phase is formed by cementation. This occurs, it must be supposed, in such a way that a plate of graphite carburizes the surrounding mixed crystal (of composition \(f\)) until it reaches the composition of the liquid phase \(k\), after which the mixed crystal melts. In this process some graphite plates may prove to be

surrounded only by liquid metal; and then it is possible that, owing to its low specific gravity, it will float to the surface and be separated in the form of spiegeleisen.* Each such plate, having floated to the surface, reduces the carbon content in the cast iron; therefore the figurative point of the composition, when moving from \(f\) to the right, will not reach the point \(e\): the third period will begin earlier, and the temperature at the end of melting will lie above the point \(g\), i.e. it will no longer be determined by the point of intersection of the line \(AB\) with the ordinate of the initial composition of the cast iron.

Cases are also possible in which the third period of melting will be absent:

Without disturbance of the course of melting (if the formation of spiegeleisen is not counted), this will occur, for example, for cast iron of composition \(k\)** (and to the right of \(k\)). For this cast iron the second period of melting (motion from the point \(f\) to the right) may end precisely at the point \(k'\), i.e., the end of melting coincides with the end of the second period.

With disturbance of the described course of melting this may occur if the first period (for cast iron of composition \(p\)***), ends at some point \(f'\). Then the end of the second period will occur at the point \(g'\), i.e. to the left of the ordinate of composition \(p\). The third period will be absent. Beginning from the point \(g'\), the cast iron will be in the region of liquid alloys, and therefore the displacement of the figurative point of its composition up to the end of internal cementation at the expense of the remaining graphite will follow some path \(g's'\). The end of internal cementation must be at the point \(s'\), but since on the path \(g's'\) separation of spiegeleisen will undoubtedly take place, the position of the point \(s'\) is indeterminate.

The phenomenon indicated above, the separation of spiegeleisen, may serve as an explanation of the well-known fact of the whitening (decarburization) of gray cast iron upon remelting.

Taking into account all that has been said about the process of melting gray cast iron, it is necessary to state the conclusion that the temperatures of the beginning and end of melting of gray cast iron depend on the amount of combined carbon in the cast iron, on the rate of heating, and on the influence—insufficiently accounted for—of the formation of spiegeleisen. The temperature of the beginning and the end of melting of cast iron of one and the same composition may vary depending on the last two factors.

*) Plates of spiegeleisen may either remain as such, or burn off, which more often occurs.

**) On the assumption that the first period of melting follows the path \(b—f\).

***) The amount of chemically combined carbon is assumed equal to \(a'\).

2) Melting of white cast iron.

The process of melting white cast iron does not differ significantly from the already considered course of melting of gray cast iron. With slow heating, such as occurs in technical melting, there must occur, as experiments have shown, the decomposition of cementite*) before the beginning of melting; i.e., white cast iron assumes the structural forms of gray cast iron.

If white cast iron of composition \(p\%\ C\) (Fig. 53) is melted, then the figurative point of its composition will move, upon heating, along the ordinate \(p\) upward as far as some point \(c\). At this point the decomposition of free cementite will begin, as a result of which the white cast iron will separate into a metallic mass with a certain content of combined carbon (point \(d\)) and elementary carbon. Upon further heating there will begin an internal process of cementation, as a result of which the figurative point of the composition will move along some path drawn in the form of an inclined curve from point \(d\). When this curve intersects the line \(AE\), the process of melting will begin, and its further course will proceed in the same way as in gray cast iron.

Owing to the fact that the rate of heating influences the position of points \(c\) and \(d\) and the course of the internal cementation of the first period, the initial temperature of melting of white cast iron, just as of gray cast iron, depends on the rate of heating, as is also shown by the heating curves in Fig. 51. However, this influence should not be pronounced, since the composition of the metallic mass (the position of point \(d\)) at normal heating rates cannot deviate greatly from point \(E\). With slow heating point \(d\) may coincide with point \(E\), i.e., the melting temperature of white cast iron will coincide with its solidification temperature, which is often observed (Heyn, Goerens).

As for the temperature at which the melting of white cast iron ends, in view of the considerations indicated it will be determined by the point of intersection of the ordinate of the cast-iron composition \((p)\) with the line \(AB\), if, of course, during the second period no disturbance of the melting process occurs (separation of spiegeleisen).

*) The question of whether, at some greater rate of heating, the melting of white cast iron can occur without decomposition of cementite remains open. In the author’s opinion, decomposition of cementite (partial or complete) must always take place, since the heating rate is technically impossible to increase significantly over that which was used in the experiments. Melting without decomposition is possible, however, in the presence of impurities, for example \(Mn\).

From the consideration of the melting process of white cast iron it follows that this process, just like the process of melting gray cast iron, is irreversible.

3) Solidification processes.

As regards the processes of solidification of white and gray cast irons, in explaining them the author wholly adheres to the theory of Prof. Goerens.

If (pure) cast iron solidifies (cools) rapidly, then cementite does not decompose, the process of solidification follows the equilibrium diagram: white cast iron is obtained: \(Fe_c + Fe_3 C\).

If cast iron of the very same composition solidifies slowly, then the free cementite that separates during the period of solidification of ledeburite decomposes—graphitization takes place—and the cast iron obtained is gray \((Fe_c + Fe_3 C) + C\).

In hypoeutectic cast irons, graphite is the product of the decomposition of the cementite of the eutectic (ledeburite): mixed crystals \(+ Fe_3 C\). In view of the fact that the form and arrangement of the graphite plates resemble a eutectic structure, which has prompted some investigators, despite the absence of experimental data for this, to regard such graphite as eutectic, the author proposes calling graphite that has arisen from the decomposition of ledeburite pseudoeutectic graphite.

In hypereutectic cast irons, graphite is formed:

1) As a result of the decomposition of primary cementite, which separates along the line \(BD\). It is proposed to call this graphite pseudoprimary graphite.

2) As a result of the decomposition of the cementite of ledeburite (pseudoeutectic graphite).

Pseudoprimary graphite often appears in the form of spangles. According to Prof. Goerens’s supposition, this occurs in the following way: the comparatively large plates formed as a result of the decomposition of primary cementite are initially surrounded by the product of its decomposition, \(3 Fe\), and thus can float in the liquid metal. But the iron surrounding them gradually becomes carburized and melts; therefore the graphite plates, freed from their shell, rise upward owing to their low specific gravity and separate from the cast iron in the form of spangles.

The process of spangle formation must have as its consequence a change in the composition of the cast iron in the direction of decreasing the total content

carbon. As the author’s experiments for determining the line \(BD\) have shown, when small quantities of hypereutectic gray cast iron are remelted (in a boat), one always obtains cast iron close to the eutectic composition, which confirms what has been said above. The formation of graphite, both during melting and during solidification and its combustion, was clearly observed in the course of these experiments.

The formation of graphite may, on the contrary, also serve as the cause of a local increase in the amount of carbon.

If the solidifying cast iron has become covered with a crust, then the graphite plates do not leave the cast iron, but accumulate at its surface. Such an accumulation of plates (sometimes together with their envelope) takes place also at the walls of the crucible. There occurs, as it were, a phenomenon of liquation of graphite, whose consequence is a local increase in the carbon content in the alloy and a disturbance of its normal structure*).

According to the course of solidification of gray cast iron described above, the amount of combined carbon in it, in the case of complete decomposition of cementite, theoretically should correspond to the limit of saturation of iron with cementite at the eutectic temperature, i.e., quantitatively be equal to \(1.7\%\ C\). Meanwhile, in reality the amount of combined carbon is generally smaller, and in technical gray cast irons it is sometimes less than \(0.9\%\) (these cast irons contain free ferrite in their structure). The latter circumstance finds its explanation in the following.

During the decomposition of cementite the decomposition product \(3Fe\) can become completely saturated with carbon only under the condition of very slow cooling. In practice, however, cooling proceeds faster than is required by the theoretical conditions, and therefore \(3Fe\) does not have time to become saturated to its limit, and the cast iron contains less than \(1.7\%\) combined carbon and more graphite than is required by theory. Incomplete saturation is promoted by certain admixtures and especially by silicon; therefore in graphitic, and consequently also in siliceous, cast irons the amount of combined carbon is insignificant, which also accounts for the formation of free ferrite, which tends to separate around inclusions in general, and in the present case around graphite plates.

In conclusion the author considers it possible to mention two experiments (from the series of experiments for determining the line \(BD\)) which illustrate the process of graphitization according to the theory of one diagram:

) Apparently such a local accumulation of carbon is represented by the cast-iron specimen in Goerens’ work (11), in Fig. 107, which is an example of a pseudoeutectic* structure of gray cast iron.

1) An alloy with \(5.307\%\, C\) was quenched before the beginning of solidification. Its structure is shown in Fig. 54. If graphite were the primary product of separation, it would have to lie directly in the eutectic. Fig. 54, however, clearly shows that each graphite plate is surrounded by a shell of carbon-containing iron, which argues that graphite is a product of the decomposition of cementite.

2) Two alloys were prepared, with \(4.611\%\, C\) and with \(4.513\%\, C\). Of these, the first was cooled slowly, and the second rapidly. In the first (Fig. 55) large plates of graphite are visible; in the second (Fig. 56), crystals of cementite. A comparison of the size and relative position of the cementite crystals and graphite plates in both figures indicates that the latter, apparently, originated from the former.

V. Conclusion.

The results of the present work are as follows:

Fig. 57.

1) By means of thermal analysis, the lines \(AB\), \(EC\), and \(PK\) of the equilibrium diagram of iron-carbon alloys were checked.

2) By means of quenching, the position of the line \(AE\) was determined (at lower temperatures than had hitherto been assumed).

3) The point \(E\), on the basis of thermal and microscopic investigations, was established at \(1.7\%\, C\). The point \(B\)—at \(4.2\%\, C\), and the eutectic temperature—at \(1134^\circ\).

4) By means of cementation experiments, the upper part of the line \(SE\) was established.

5) On the basis of the data of the first four points, the equilibrium diagram of iron-carbon alloys has the form shown in Fig. 57.

On the basis of this diagram the following table (Table VII) of theoretical solidification temperatures of iron-carbon alloys has been compiled.

TABLE VII.

Carbon content, %	Beginning of solidification, °C	End of solidification, °C
Pure iron according to Stansfield	1505	—
0.5	1472	1329
1.0	1433	1187
1.5	1393	1142
2.0	1348	1134
2.5	1303	1134
3.0	1257	1134
3.5	1209	1134
4.0	1156	1134
4.2	1134	1134
4.5	?	1134

6) On the basis of experiments it has been established: 1) that the rate of heating has a perceptible influence on the temperature at the beginning of melting of cast iron; 2) that during the process of melting internal cementation takes place; and 3) that in white cast iron, before the beginning of melting, decomposition of free cementite occurs.

And 7) a new theory of the melting of cast iron has been proposed.

End of Part 1.

Tomsk, 21 May 1912.

TABLE OF MICROPHOTOGRAPHS

Figure	Experiment	Alloy	Sample	Hardening temperature, \(T^\circ C\)	% C	Notes
7	—	I	—	—	0,577	Cooled slowly.
8	—	II	—	—	1,071	Cooled slowly.
9	—	III	—	—	1,632	Cooled slowly.
10	—	IV	—	—	1,762	Cooled slowly.
11	—	V	—	—	2,165	Cooled slowly.
12	—	VI	—	—	2,795	Cooled slowly.
13	—	VII	—	—	3,677	Cooled slowly.
14	—	VIII	—	—	4,626	Cooled slowly.
15	—	IX	—	—	4,940	Cooled slowly.
24	—	—	9	1120	1,762	—
25	—	—	4	1167	1,762	—
26	—	—	15	1130	1,632	—
27	—	—	13	1154	1,632	—
28	—	—	13	1154	1,632	—
29	—	—	25	1150	1,071	—
30	—	—	21	1228	1,071	—
31	—	—	29	1318	0,577	—
32	—	—	29	1318	0,577	—
33	—	—	21 and 29	—	—	Etched with sodium picrate.
34	—	—	27	1365	0,577	—
36	—	—	—	—	—	Along and across the polished section of the eutectic (ledeburite).
37	—	—	—	—	1,762	—
58	—	—	—	—	2,006	—
47	55	—	—	1130	—	—
48	57	—	—	1150	—	—
49	63	—	—	1135	3,531	—
50	65	—	—	1150	3,560	—
54	—	—	—	—	5,307	—
55	—	—	—	—	4,611	Cooled slowly.
56	—	—	—	—	4,513	Hardened.

List of bibliographic sources to which references are made in the work*).

1) Heyn und Bauer. Metallographie, parts I and II. Published by Goeschen (1909).

2) A. Findlay. Einführung in die Phasenlehre (1907). Leipzig.

3) Roberts-Austen. Fifth Report of the alloys Research Committee. The Metallographist II, p. 186.

4) B. Roozeboom. Stahl und Eisen vom Standpunkte der Phasenlehre. Z. f. phys. Chemie, vol. 34, 1900, p. 437.

5) Carpenter und Keeling. The range of solidification and critical range of iron-carbon alloys. Journ. of the Iron and Steel Inst. 1904 I, p. 224.

6) F. Wüst. Beitrag zum Kenntniss der Eisenkohlenstofflegierungen höheren Kohlenstoffgehaltes. Metallurgie 1906, p. 1.

7) C. Benedicks. Ueber das Gleichgewicht und die Erstarrungsstrukturen des systems Eisen-Kohlenstoff. Metall. 1906, p. 393.

8) Stanfield. The present position of the solution theorie of carburised iron. Journ. of the Ir. a. st. Inst. 1900, II, p. 317.

9) Heyn. Labile und metastabile Gleichgewichte der Eisenkohlenstofflegierungen. Z. f. Elektrochemie, 1904, p. 491.

10) G. Charpy. Sur le diagramme d’ équilibre des alliages fer carbon. Comp.-rend. 1905, 141, p. 948.

11) Goerens. Ueber den augenblicklichen Stand unserer Kenntnisse der Erstarrungs-und Erkaltungsprozesse der Eisenkohlenstofflegierungen. Metall. 1906, p. 175 and 1907, pp. 137, 173.

12) Ruer. Metallographie in der elementaren Darstellung (1907). p. 212.

13) Howe. Das Zustandsdiagramm des Eisen-Kohlenstoffs. Metall. 1909, p. 65.

14) Wittorf. Theory of Alloys (1909). St. Petersburg.

15) Goerens und Gatowsky. Experimentelle Studie über den Erstarrungs-und Schmelzvorgang bei Roheisen. Metall. 1908, p. 137.

16) F. Wüst. Beitrag zum Einfluss des Phosphors auf das System Eisen-Kohlenstoff. Metall. 1908, p. 76.

17) Goerens. Einführung in die Metallographie (1906). Halle.

18) Tammann. Kristallisieren und Schmelzen (1903). Leipzig.

19) B. Roozeboom. Heterogene Gleichgewichte, vol. II (1904). Braunschweig.

*) In the text of the work, numbers are placed over the surnames of researchers; under these numbers, the present list gives the title of the researcher’s work and the place where it appeared in print.

20) Gutowsky. Experimentelle Untersuchung über den Erstarrungs- und Schmelzvorgang bei technischem Roheisen. Metall. 1908, p. 463.

21) Heyn. Bericht über die mikroskopische Untersuchung der vom Sonderausschuss für Eisenlegierungen des Vereins zur Beförderung des Gewerbefleisses hergestellten Legierungen. Verh. d. Ver. z. Bef. d. Gewerbefl. 1904, p. 355.

22) Carpenter. Der Erstarrungspunkt des Eisens. Metall. 1908, p. 679.

23) Müller. Ueber die Darstellung des Elektrolyteisens, dessen Zusammensetzung und thermische Eigenschaften. Metall. 1909, p. 145.

24) Tammann. Ueber die Anwendung der thermischen Analyse. Z. f. anorg. Ch. 47 (1905), p. 289.

25) Stanspield. The burning and overheating of steel. Journ. of the Ir. a. St. Inst. 1903, II, p. 433.

26) Margueritte. Cémentation du fer par le charbon et par l’oxyde de carbone. Compt.-rend. 59 (1864), p. 726.

27) P. Charpy. Note of cémentation. Ir. and Steel Magaz. 1904, p. 305.

28) Wüst und Schlosser. Der Einfluss von Kohlenstoff, Silicium, Mangan, Schwefel und Phosphor auf die Bildung der Temperkohle im Eisen. Stahl u. Eisen, 1904, p. 1120.

29) Hadfield. Die physikalisch-chemischen Vorgänge, welche mit der Entkohlung von Eisenkohlenstofflegierungen verknüpft sind. Metall. 1909, p. 358.

30) F. Wüst. Ueber Theorie des Glühfrischens. Metall. 1908, p. 7.

31) Tiemanns. The temperatures limits for the Separation of graphite from martensite in pur cast-iron. Metallogr. X (1901), p. 319.

32) Royston. The relation of carbon to iron at high temperatures. Journ. of the Ir. a. St. Inst. 1897, I, p. 166.

33) Charpy et Grenet. Sur l’équilibre des systèmes fer-carbone. Bull. de la Soc. de l’encom. p. l’ind. nation. 1902, p. 399.

34) Heyn und Bauer. Zur Metallographie des Roheisens. St. u. Eisen, 1907, p. 1568.

35) G. Charpy. Sur la solubilité du Graphit dans le fer. Rev. de Métall., 1908, p. 77.

36) Benedicks. Ueber die Löslichkeit des Graphits im Eisen. Metall. 1908, p. 41.

37) Heyn. Metallographische Untersuchung für Giessereiwesen. St. u. Eisen, 1906, p. 1295.

38) Maurer. Untersuchung über das Härten und Anlassen von Eisen und Stahl. Metall. 1909, p. 33.

ERRATA

Page	Line	Printed	Should read
4	7 from top	likewise	just the same
5	20 from top	should	ought to
7	5 from top	B. Rooseboom	B. Roozeboom
10 and 14	1 and 3 from top	Goereus	Goerens
17	9 from top	” ”	” ”
17	12 from top	hto	what
29	11 from top	exac tly	exactly
31	11 from top	eutectic,	eutectics
37	1 from top	Stansfild’s	Stansfield’s
39	17 from top	to diffuse	to diffuse—
41	6 from bottom	contain	contain
44	6 from top	46—48	47—49
48	6 from top	to make partial	to make partial
50	19 from top	cementation	cementation
50	13 from bottom	theoretically	theoretically,
52	12 from top	investigations	investigations
53	5 from bottom	fig. 45	fig. 46
55	4 from bottom	46	45
55	10 from top	discernibly	considerably
58	6 from top	they	he
58	9 from bottom	melting	melting
59	3 from top	cast iron	cast iron
59	10 from top	p.	p. 4
61	3 from top	resurfaced	surfaced
61	15 from top	period	period
62	6 from top	accept	accepts

Table 1.

N. Gutovsky. On the theory of the system: iron–carbon.

Fig. 7. × 200

Alloy I.
\(C = 0.577\%\)

Fig. 8. × 200

Alloy II.
\(C = 1.071\%\)

Fig. 9. × 200

Alloy III.
\(C = 1.632\%\)

Fig. 10. × 200

Alloy IV.
\(C = 1.762\%\)

Fig. 11. × 200

Table II.

N. Gutovsky. On the theory of the system: iron–carbon.

Fig. 12. × 200

Alloy VI. $C = 2.795\%$

Fig. 13. × 200

Alloy VII. $C = 3.677\%$

Fig. 14. × 200

Alloy VIII. $C = 4.626\%$

Fig. 15. × 200

Alloy IX. $C = 4.940\%$

Fig. 24 × 200

Table III.

N. Gutovsky. On the Theory of the Iron–Carbon System.

Fig. 25. ×200

\(T = 1167^\circ\)

Specimen No. 4.
\(C = 1.762\%\)

Fig. 26. ×150

\(T = 1130^\circ\)

Specimen No. 15
\(C = 1.632\%\)

Fig. 27. ×150

\(T = 1154^\circ\)

Specimen No. 13.
\(C = 1.632\%\)

Fig. 28. ×800

\(T = 1154^\circ\)

Specimen No. 13.
\(C = 1.632\%\)

Fig. 29. ×150

Table IV.

N. Gutovsky. On the theory of the system: iron–carbon.

Fig. 30. ×150
\(T = 1228^\circ\)
Specimen No. 21.
\(C = 1.071\%\)

Fig. 31. ×150
\(T = 1318^\circ\)
Specimen No. 29.
\(C = 0.577\%\)

Fig. 32. ×600
\(T = 1318^\circ\)
Specimen No. 29.
\(C = 0.577\%\)

Fig. 33. ×150
Specimen No. 21.
Specimen No. 29.
Etched with sodium picrate.

Fig. 34. ×200

Table V

N. Gutovsky. On the theory of the iron–carbon system.

Fig. 36. × 60

Longitudinal section
Transverse section

eutectic (ledeburite) with 4.2% C.

Fig. 37 × 60

Alloy I
C = 1.762%

Fig. 38. × 60

Alloy with
2.006% C

Fig. 47. × 200

T = 1130°
Experiment No. 55.

Fig. 48. × 200

Table VI.

N. Gutovsky. On the theory of the iron–carbon system.

Fig. 49 ×150

\(T = 1135^\circ\)
Experiment No. 63

Fig. 50 ×150

\(T = 1150^\circ\)
Experiment No. 65

Fig. 54 ×200

\(C = 5.307\%\)

Fig. 55 ×200

\(\Gamma_{\mathrm{dif.}} = 1.745\%\)

Cooled slowly
\(C = 4.611\%\)

Fig. 56 ×200

1. See p. 11. ↩