Any contribution on manganese steel by its discoverer necessarily carries with it much weight and is entitled to serious and close consideration. The momentous discovery of that alloy by Sir Robert Hadfield some 30 years ago was the signal for great activity in the field of alloy steels; and the remarkable results obtained in the last two decades probably constitute the most important modern advance in the metallurgy of steel.

The authors of the paper referred to above are unable to reconcile with the allotropic theory the interesting results obtained by them in heat treating manganese steel. When they speak of the allotropic theory, however, I understand that they have more especially in mind the existence of beta iron. It is not my understanding that they question the occurrence of iron in the two allotropic varieties known as gamma and alpha. If I am wrong, I trust they will correct me.

A careful consideration of the experimental results reported and discussed by the authors leads me to believe that on the contrary they can be very satisfactorily explained in terms of allotropy and that, in the light of our present knowledge, they cannot be accounted for in any other way.

The physical properties of non-magnetic manganese steel described by the authors are those pertaining to iron-carbon alloys in which iron is present in the gamma condition, or, in other words, to austenite. The metal is then characterized by great tenacity and ductility and low elastic limit, while its hardness, mineralogically speaking, is not excessive, but of a special kind termed “tough hardness” by the authors. Even when subjected to the restraining influence of a large amount of manganese and of considerable carbon, however, relatively rapid cooling is necessary to produce austenitic manganese steel free from beta and alpha iron. On very slow cooling from a high temperature, as indicated by the authors, manganese steel becomes, to some extent, magnetic, because in accordance with the allotropic theory such slow cooling permits the formation of some beta and alpha iron. The mineralogical hardness of the metal is then, as it should be, somewhat greater. Non-magnetic, and, therefore, austenitic, manganese steel, while very much more stable than hardened, high-carbon steel, is nevertheless in an unstable condition, as evidenced by the fact that relatively quick cooling is required for its production. On reheating (shall we say tempering?) manganese steel, therefore, we may logically anticipate a partial transformation of gamma iron into beta and, possibly, into alpha iron—and, indeed, the authors show that reheating non-magnetic, austenitic manganese steel for a long period of time to some 500° C. or higher causes the return of considerable magnetism and greatly increased hardness. While they consider this occurrence fatal to the allotropic theory, it appears to me, on the contrary, to carry with it a striking confirmation of the correctness of that theory. If there were no hard allotropic modifications of iron between the soft alpha and moderately hard gamma varieties, the transformation taking place in reheating manganese steel could only imply the formation of some alpha iron and, therefore, decreased hardness. The fact that the treated steel becomes magnetic also points to the formation of some alpha, together with beta iron. This is also consistent with the allotropic transformation observed in carbon steel when we find that whenever a notable quantity of beta iron is formed, causing hardness, some alpha iron also forms, imparting magnetism to the metal; hence the hardness and magnetic properties of hardened carbon steel. It does not seem possible, in the presence, of carbon at least, to produce much beta iron to the exclusion of alpha iron. These two allotropic varieties probably form isomorphous crystals which are both hard and magnetic.

The authors seem to have some difficulty in explaining, in accordance with the allotropic theory, the fact that on heat treating non-magnetic manganese steel, both the hardness and the specific magnetism may increase, their argument being that if magnetism is due to the formation of soft alpha iron, then increased magnetism must mean increased softness. This, however, is only an apparent inconsistency, as I have attempted to show graphically in Fig. 1. We may assume that at 400° C. originally non-magnetic manganese steel remains non-magnetic, the iron present being wholly in the gamma condition. After a long heating, say at 450° C., some of the gamma iron is converted into beta and into alpha iron. Of the iron present, AB per cent, may be assumed to occur now as alpha iron, BC per cent, as beta iron, and CD per cent, as gamma iron. Heating to a higher temperature, or possibly a longer heating at the same temperature, causes further transformation, both beta and alpha iron increasing, the metal containing now A’B’ per cent, of alpha and B’C’ per cent, of beta iron; hence its increased magnetism and increased hardness.

Some beta and alpha iron having now been formed, it is evident that on heating the metal to a sufficiently high temperature, a critical point must necessarily be found, caused by the return of alpha and beta iron to the gamma condition—the only one stable at a high temperature. The authors have located this point at about 675° C., as shown in my Fig. 1. On again cooling the metal relatively quickly, no critical point should be detected, since the steel remains non-magnetic, that is, in the gamma condition.

The fact alluded to by the authors that manganese steel can be so softened by suitable heat treatment as to make it machineable is another evidence of the formation of alpha iron.

From the fact that cast manganese steel is much less ductile than the water-toughened metal, although likewise non-magnetic, we not illogically infer that some beta iron may have been formed during the relatively slow cooling of the casting, while the usual lack of ductility of steel castings in general and the possible separation of carbide may be additional reasons for the properties of cast, untreated manganese steel.

The fact that manganese steel does not become magnetic on cooling it in liquid air, while non-magnetic nickel-steel with some 22 per cent, nickel regains its magnetism after such treatment, does not point to different influences being at work. It can be satisfactorily accounted for, as shown in Fig. 2, in which it is reasonably assumed that with 1 to 1.25 per .cent, carbon it requires some 7 per cent, of manganese to lower the critical point from its normal position (some 700° C.) to atmospheric temperature, whereas some 18 per cent, nickel may be required to cause the same depression. Extending below atmospheric temperature the lines indicating the lowering of the critical points, it is seen that with some 12 per cent, manganese the critical transformation may be lowered to some —400° C., while nickel steel with 22 per. cent, nickel would have a critical point at—100° C. and therefore detectable on immersion in liquid air. It will, of course, be understood that no claim is made here in regard to the exact position of the lines or curves indicating the

depression of the magnetic change with increased manganese or nickle. These positions, so far as I know, have never been accurately determined, but, from the results available, the diagram of Fig. 2 is not altogether a fantastic one.

The authors quote at length from the important work of Pierre Weiss, who has developed a new theory of magnetic transformation in metals.

The attitude of a scientific man toward new theories should be one of open-mindedness and receptiveness—even of reverence—when the new thought springs from an exalted source; but it should also be one of caution. He should guard against too hasty and enthusiastic acceptance. He should be mindful of the theoretical scrap heaps lining both sides of the pathway of scientific progress. For a while, at least, a new-theory must be placed on probation, while being subjected to searching tests. It is the present position of Professor Weiss’s theory.

With a view to securing, if possible, additional information in regard to the relation between the heat treatment of manganese steel, and its critical points, structure and physical properties, some experiments were conducted in the Metallographical Laboratory of Harvard University which will be briefly described and illustrated.

Steel Used

The manganese steel used in these experiments was supplied by the Taylor-Wharton Iron and Steel Co. of High Bridge, N. J. in the shape of cast bars ½ by ¾ in. in cross section and about 12 in. long. The steel contained 1.25 per cent, carbon and about 12.50 per cent, manganese.

Cast Untreated Manganese Steel.—The structure of the cast metal is illustrated in Fig. 3. It consists of polyhedric grains of austenite surrounded by a considerable amount of a constituent generally held to be a double carbide of iron and manganese (manganiforous cementite), or a mixture (possibly a solid solution) of the carbide of iron Fe3C and the carbide of manganese, Mn3C. As is well known manganese steel in this condition has a low tensile strength, low elastic limit and little ductility. It is non-magnetic. Its Brinell hardness number was found to be 196. In the light of the allotropic theory the absence of ferromagnetism is to be ascribed to the absence of alpha iron while the lack of ductility of the metal may be accounted for through the formation during the relatively slow cooling of the castings of some beta iron, although, undoubtedly, the presence of segregated carbide surrounding the austenite grains and the usual structural coarseness of cast steel may be also regarded as causes of brittleness.

Heat Treatments

Some of the steel bars were subjected to the following treatments:

Heating to 1,100° C. and quenching in water.
Heating to 1,100° C. and cooling in furnace.

Heating to 1,100° C. and quenching in water followed by:

Heating to 800° for 2 hr. and cooling in furnace,
Heating to 700° for 2 hr. and cooling in furnace,
Heating to 600° for 2 hr. and cooling in furnace,
Heating to 500° for 2 hr. and cooling in furnace,
Heating to 400° for 2 hr. and cooling in furnace,
Heating to 575° for 90 hr. and cooling in furnace.

Before subjecting the treated specimens to microscopical examination and to the Brinell test for hardness the outside, decarburized portion was removed by grinding.

Manganese Steel Bar Heated to 1,100° C. Quenched in Water

The structure of this bar is illustrated in Fig. 4. The steel is now made up of polyhedric grains of austenite, quenching from a high temperature having caused the disappearance of the segregated carbide so conspicuous in the structure of the cast metal (Fig. 3). We naturally infer that at a high temperature this carbide or mixture of carbides dissolves in the austenite and that rapid cooling does not permit its re-precipitation. To the absence of this carbide must be ascribed in part at least the much greater toughness and tenacity of the quenched metal, although it may reasonably be claimed that the marked change of properties may also be due to the now complete absence of beta iron. As is always the case with austenitic steel, the elastic limit remains low. Two bars subjected to water quenching from 1,100° were tested for hardness giving respectively as factors

of hardness 174 and 181. This slightly decreased mineralogical hardness as compared to the hardness of the untreated metal (196) is in harmony with the conception of the formation of some beta iron during the relatively slow cooling of the cast bars and complete absence of it in the quenched bars. The quenched samples because of the absence of alpha iron are of course non-magnetic.

Manganese Steel Bar Heated to 1,100°  Slowly Cooled in the Furnace

The structure of this bar was similar to the structure of the cast untreated metal (Fig. 3). This was to be anticipated. Heating to 1,100° caused the segregated carbide or carbides to dissolve in the austenite but the very slow cooling that followed resulted in its being again precipitated. The metal was brittle and lacking in tenacity. Its hardness number was found to be 212, a slight increase over the hardness of the untreated metal (196). This may be explained on the ground that the furnace-cooled bar cooled more slowly than the cast bars resulting in the formation of a greater amount of beta iron. From the fact that this bar is non-magnetic we infer that although slow, the cooling was not slow enough to permit the formation of any appreciable quantity of alpha iron.

Manganese Steel Bar Heated to 800° for 2 hr.  Slowly Cooled in the Furnace

The structure of this bar was quite similar to that of the cast, untreated bar, but from its greater hardness, which is now 245 as compared to 196, we infer that the long exposure to 800° followed by slow cooling made possible the formation of a greater amount of beta iron. The bar, however, is still non-magnetic which points to the non-formation of any appreciable quantity of alpha iron.

Manganese Steel Bar Heated to 700° for 2 hr. Slowly Cooled in Furnace

The structure of this bar, illustrated in Figs. 5 and 6 under two different magnifications, clearly indicates the formation of martensite and hence of considerable beta iron. From our knowledge of the properties of martensite we shall expect the steel to be ferromagnetic because of the usual occurrence of an appreciable amount of alpha iron in martensite. We shall also expect the steel to be now mineralogically harder than when in an austenitic condition. These expectations are fulfilled. Manganese steel so treated is very appreciably ferromagnetic and its hardness is now 266, or 70 points higher than in the cast condition and some 90 points higher than after quenching from 1,100°.

Manganese Steel Bar Heated to 600° for 2 hr. Slowly Cooled in Furnace

The structure of this bar was similar to that of the preceding bar, that is, decidedly martensitic while the bar is ferromagnetic from which it follows that the heat treatment to which it was subjected resulted in the production of beta and alpha iron. Its hardness number, 335, indicates a very great increase of hardness. Evidently this bar must contain more beta iron than the bar heated for 2 hr. to 700°. This is readily explained on the ground that 700° is very near the range of temperature in which gamma iron is the stable condition.

Manganese Steel Bar Heated to 500° for 2 hr. Slowly Cooled in Furnace

The structure of this bar was also found to be martensitic while the metal was ferromagnetic but to a decidedly less degree than the preceding bar. Clearly, at 500° tempering does not proceed as far as it does at 600 with the result that less beta iron and less alpha iron are formed. This is perfectly consistent with the allotropic theory. The hardness number of the bar was 255.

It seems evident that a temperature of 600° or thereabout is the most effective one to render manganese steel martensitic and, therefore, hard and ferro-magnetic.

Manganese Steel Bar Heated to 400° Slowly Cooled in Furnace

The structure of this bar was practically that of the untreated quenched metal. It was non-magnetic and its hardness number was 163. Clearly, heating to 400° for 2 hr. is not sufficient to cause any appreciable tempering of the quenched metal. There is no apparent explanation for the fact that the steel is now somewhat softer than in the quenched condition.

Manganese Steel Bar Heated to 575° for 90 hr. Slowly Cooled in Furnace

Seeing that magnetism is readily produced by heating in the vicinity of 600°, while the structure of the metal becomes decidedly martensitic, a quenched bar was kept for 90 hr. at a temperature of 575°. The structure of this bar was of the type illustrated in Figs. 5 and 6; its hardness was 356 and the metal was so magnetic that the bar could be readily picked up by a permanent magnet.

Conclusions

From the above data it appeal’s that manganese steel of the Hadfield type may occur under three distinct conditions: (1) as austenitic steel mixed with a considerable amount of free carbide, a condition which is produced by slow cooling from a temperature exceeding 700°; (2) as austenitic steel practically free from segregated carbides, a condition produced by rapid cooling from a high temperature and (3) as martensitic or possibly austenito-martensitic steel, a condition most readily produced by reheating austenitic steel for a sufficient length of time to a temperature exceeding 500° but not 700°.

In the first condition, which may be described as cementito-austenitic, the metal is non-magnetic, weak and lacking in ductility, while its hardness number is in the vicinity of 200. In its austenitic condition

All tests were made under load of 3,000 Kg. with 10-mm. ball pressure held for 30 sec. 1100W signifies heated to 1,100° C. and quenched in water; 800-2-F signifies heated to 800° for 2 hr. and furnace-cooled, etc.

the metal has a high tenacity, great ductility, is non-magnetic and its hardness number is in the vicinity of 180.

The martensitic condition results from the tempering of austenite as previously explained. The metal is now ferro-magnetic because of the presence of some alpha iron, while its hardness varies between 250° and 350, an increase due to the formation of a considerable amount of beta iron.