N. P. CHIZHEVSKY,

Professor of the Tomsk Technological Institute of Emperor Nicholas II.

IRON AND NITROGEN.

AN EXPERIMENTAL STUDY

OF THE AMOUNT OF NITROGEN AND THE REASONS FOR ITS PRESENCE

IN CAST IRON, STEEL, AND IRON.

THE INFLUENCE OF NITROGEN ON THE MECHANICAL PROPERTIES OF IRON.

TOMSK.

Typo-lithography of the Siberian Printing Association, corner of Dvoryanskaya and Yamsk. side st.

1914.

Printed by order of the Director of the Tomsk Technological Institute of EMPEROR NICHOLAS II.

Table of Contents

	Page
Introduction	1
Review of the Literature	1
Determination of the Amount of Nitrogen in Iron, Steel, and Cast Iron	12
Pure Iron and Nitrogen	26
Carbonaceous Iron and Nitrogen	32
Manganese and Nitrogen	35
Silicon and Nitrogen	44
Metallography of Nitrogenous Iron	53
Influence of Aluminum on the Nitrogen Content in Iron	63
Investigation of the Course of the Bessemer and Open-Hearth Processes from the Standpoint of the Combination of Nitrogen with the Metal	65
Influence of Nitrogen on the Mechanical Properties of Iron	73

Introduction

The presence of nitrogen in various grades of iron was discovered as early as the 1830s and was confirmed by many investigators. In the 1840s, as Wedding indicates in his Ausführliches Handbuch der Eisenhüttenkunde¹, there existed the opinion that nitrogen plays a considerable role in the manufacture of iron, and some metallurgists thought that it is one of the important impurities in ordinary grades of steel. The French chemists of that time believed that nitrogen was the chief cause of the hardness and brittleness of steel and iron. This opinion gradually disappeared, since no large percentage content of nitrogen could be found in them.

As to the harmfulness of the influence of nitrogen on the physical properties of iron, there can be no doubt; it may be placed in the same category as other harmful impurities of iron, for example sulfur, phosphorus, and oxygen. Looking at Mendeleev’s table, one may already suspect this. Nitrogen is in the fifth group above phosphorus; then it stands in the second row between carbon and oxygen.

	IV	V	VI
2nd row . . .	C	N	O
2nd row . . .	12,0	14,01	16,00
3rd row . . .	Si	P	S
3rd row . . .	28,2	31,0	32,06

A content of \(0.1\%\) \(O\) makes iron completely unforgeable; \(C\) likewise produces brittleness, but at a considerably higher content: cast iron cannot be forged. \(P\), at \(0.1\%\), gives soft iron noticeable brittleness, and carbon steel at even smaller quantities.

Defects in the quality of steel and iron often occur, being discovered not only during manufacture—rolling, forging, etc.—but also later, after the metal has already been made into articles, at var—

construction structures or machines. After some time it turns out that it does not fulfill its purpose, wears out prematurely, develops cracks, breaks. This often, besides material losses, may also entail human casualties.

Precise knowledge of the influence of various harmful impurities, and constant accounting of their quantities in the production of steel and iron, are of the highest importance. As for the other harmful elements that occur in the products of ironworks production and reduce their quality—for example, \(P\), \(S\), \(O\)—their influence has long since been well clarified; analytical methods for determining them have been developed, and in the laboratories of metallurgical plants regular determinations of their content are made, both in the initial and in the final products of the heat. Standards exist for the maximum content of the usual harmful elements in various grades of iron, steel, and cast iron. Measures have also been developed and are everywhere applied to remove the usual harmful elements during plant heats, in order to obtain production products with their permissible quantity for the corresponding grades of material.

If those quantities of nitrogen that are encountered in various grades of iron noticeably lower their quality, then it is unquestionably necessary to make a constant accounting of its content in the products of plant production. In addition, it will be necessary to develop means for the possible reduction of its quantity in the finished material by a known direction of the course of heats or by the composition of the charges. These means can be given only after a detailed investigation of the conditions under which nitrogen combines with iron alloys, after clarifying the influence of the constituent parts of iron alloys on the reactions of their combination with nitrogen.

The present investigation, as it was developed—its principal parts having been printed in various journals as separate articles—is appearing here in collected and supplemented form.

The literature on nitrogen in iron that appeared before my first article is given in Chapter I. Later literature, works on nitrogenous compounds of elements constituting impurities in plant grades of iron, as well as investigations concerning the influence of nitrogen on the mechanical properties of iron, are given in the corresponding chapters. Braune’s work, to which special importance is attached, is set forth in Chapter I in full. The mechanical investigations from it have not been separated into Chapter X, so as not to disturb the integrity of the impression.

Chapter I.

Review of the Literature.

It has been shown by many investigations that nitrogen combines with iron and gives ferronitride, \(Fe_2N\). Iron, heated in a stream of dry ammonia, as Berthollet found \(^{2}\), takes up from 12% to 13% N and thereby becomes brittle. The fact of the combination of iron with nitrogen was confirmed by Thenard \(^{3}\) and Savart \(^{4}\). The experiment of saturating iron with nitrogen over the course of several hours was carried out in a porcelain tube at red-heat temperature. Nitrogenous iron was white in color, brittle, attracted by a magnet, and readily dissolved in acids, forming with increase in volume upon a lowering of the specific gravity. According to Depretz’s investigation, when iron treated with ammonia at high temperature is dissolved in acids, the nitrogen passes into ammonia. Stahlschmidt \(^{5}\), by passing dry ammonia at red heat through \(FeCl_2\) and \(Fe_2Cl_6\), obtained a product which, by analysis, should correspond to the formula \(Fe_2N(Fe_4N_2)\). Ferronitride is a silvery mass, of specific gravity 5.0, readily giving off, with a mass of sparks, in a gas flame. When heated in a stream of hydrogen, ferronitride is converted into soft iron, rapidly rusts in air, and dissolves in nitric acid with the evolution of nitrogen oxide. Stahlschmidt indicates that, when ferronitride is dissolved in weak sulfuric acid, hydrogen, nitrogen, and sulfuric-acid salts of iron and ammonia are formed. Further investigations showed that all the nitrogen under these conditions passes into ammonia. When nitrogenous iron is dissolved in hydrochloric acid, salts of iron, ammonia, and hydrogen are formed; when it is heated with water vapor, ammonia is obtained. Buff and Fremy engaged in the synthesis of nitrogenous iron, and their works confirmed the earlier investigations. The first quantitative determinations of nitrogen in cast iron and steel were carried out by Schafhäutl \(^{6}\), by burning turnings in a vacuum by Dumas’s method, originally proposed by the latter for organic substances, with a mixture of \(K_2O\) and

\(^{2}\) Wedding. Ausf. Handbuch d. Eisenh. 1891—1896, p. 455; Gilb. Ann. 30, pag. 378.
\(^{3}\) Traité de chim., 1834, 1, p. 434.
\(^{4}\) Ann. de chim. et de phys. 1829, 42, p. 122.
\(^{5}\) Pogg. ann. 125, pag. 37.
\(^{6}\) Wedding, ibid., p. 456.

$BaO$. He determined, in white cast iron from South Wales, $0.7637\%$ N; in specular cast iron—$1.2\%$ N; in English cast steel—$0.1831\%$ N.

In 1850 the work of Marchand appeared². He determined the nitrogen content in many grades of cast iron, steel, and iron; moreover, he found more nitrogen in steel than in cast irons, and none at all in soft iron. The analysis was carried out by him according to Lassaigne’s method, igniting a very finely divided sample of the metal with potassium or sodium in a glass tube at red heat. In this process $K_3FeCy_6$ is formed, and the nitrogen is determined from the amount of Berlin blue.

In Marchand’s work, among other things, the influence of nitrogen from the atmosphere of the air on the results of the analysis is indicated. The nitrogen of the air enters into the reaction in this case. Experiments in heating a mixture of finely divided metal with potassium or sodium in an atmosphere of hydrogen or carbonic acid always gave lower results for the nitrogen content in the metal.

His determinations of nitrogen are of interest because of the method used by Schafhäutl. He repeated the latter’s experiments with all possible care, using sodium lime instead of the mixture of $K_2O$ and $BaO$. Marchand found incomparably less nitrogen in the samples he investigated, namely: in gray cast iron from Rotteberode, in one sample $0.009$, in another $0.003\%$ N; in English tool steel $0.014\%$ N.

In 1861 the work of Bois appeared³. The analysis was carried out in the following manner: in a porcelain tube he placed very fine filings of the metal under investigation and, after heating to redness, passed dry hydrogen through; beyond the metal there was pumice. The gases passed through several cylinders with $CaCl_2$. The ammonia formed was collected in titrated $H_2SO_4$. To shorten the duration of the experiment it was necessary to subject the metal to very fine comminution.

The results of his determinations are as follows:

Material	Nitrogen content
Krupp spring steel	0.085
Same	0.011
Cast steel in sheets	0.059
Same	0.037
Same, filings	0.031
Cast steel from Jackson	0.058
Wootz steel (filings)	0.12
Iron in sheets	0.18
White cast iron in pieces	0.15
Gray cast iron in pieces	0.00

Boussingault ^9) carried out the analysis for nitrogen by two methods. The first method consisted in his heating shavings of the metal under investigation, mixed with cinnabar, in a glass tube in an atmosphere of \(CO_2\), and collecting the gases thereby formed over mercury. After absorption of the carbonic acid by caustic alkali, the quantity of nitrogen was determined. He found in cast steel \(0.057\%\ N\); in iron wire, in one specimen \(0.124\%\), in another—\(0.68\%\ N\). The second method consisted in dissolving the shavings of the given specimen in hydrochloric or sulfuric acid. The solution obtained was neutralized with an excess of caustic potash or lime and subjected to distillation. The ammonia distilled over with the steam was collected in a receiver and determined by titration with sulfuric acid. By this method there was analyzed iron saturated with nitrogen by means of dry ammonia; it proved to contain \(2.655\%\ N\). By the same method he analyzed cast steel, in which \(0.012\%\ N\) was found. Analyses made on different grades showed much smaller quantities of nitrogen, for example, \(0.0075\%\ N\) in piano wire, \(0.007\%\ N\) in another specimen of it, and \(0.0086\%\ N\). From this it may be concluded that the preceding method, giving a higher result, is unreliable, all the more so since the latest investigations give still smaller quantities of nitrogen in similar grades of iron.

Of interest is the investigation by Rammelsberg ^10), who determined the nitrogen content in specular pig iron. He dissolved several kilograms of it in dilute sulfuric acid, crystallized out the greater part of the ferrous sulfate, distilled off ammonia from the mother liquor with an excess of lime, and determined it gravimetrically with the aid of chloroplatinic acid; the nitrogen proved to be \(0.002\%\). Analysis of the insoluble residue for nitrogen gave a negative result. Rammelsberg adduces the extremely interesting observation that the residue from the distillation of ammonia, with alkali, after standing several days in the air and being subjected again to distillation, gives further appreciable quantities of ammonia.

Ullgren ^11) drew attention to the purity of the reagents used in such analyses and, since hydrochloric acid usually contained, although small, in the present case significantly error-producing quantities of ammonia, proposed using neutral \(CuCl_2\) for dissolving the metal. He recommends burning the insoluble residue with \(HgCl_2\). By the method proposed by him, a series of analyses was performed by Rinman ^12), which gave the following results:

White pig iron . . . .	\(C—4.54\%\)	\(N—0.008\%\)
Gray pig iron . . . .	„ \(4.56\)	„ \(0.005\)

^9) Compt. rend. 1861, 53, p. 77.
^10) Wedding ibid; Monatsberichte d. Preus. Akad. d. Wiss. Dec. 1862.
^11) Ann. Chem. Pharm. 1862, 124, p. 70.
^12) Jurn. f. pract. Chem. 1867, p. 32.

Material	C	N
Bessemer steel	2.25%	0.005%
Same	1.27	0.005
Same	0.86	0.005
Same, red-brittle	0.40	0.006
Same, red	0.45	0.008
Same, hardened	1.28	0.011
Cement steel	1.50	0.016
Same, hardened	1.50	0.016

This table is of interest in that it shows that cement steel has a higher nitrogen content than Bessemer steel or cast iron; this is confirmed by the most recent investigations.

In 1880 a work appeared by Allen⁴, who found in thin iron wire 0.0123% N, in armor steel 0.0131% N, in open-hearth steel with 0.22% C—0.0107% N, in crucible steel with 1.3% C—0.0172% N, and in specular cast iron with 3.8% C and 20% Mn—0.0041% N.

The well-known Swedish metallurgist H. Tholander⁵ also studied the nitrogen content in cast iron, steel, and iron, making strength tests of the material and paying attention to the nitrogen content. He found that Bessemer steel is less strong than open-hearth steel and has a whiter fracture than the latter, which he attributes to the presence of nitrogen.

For analysis Tholander took a charge of 5 g, dissolved the shavings in dilute sulfuric acid and, after neutralizing with caustic soda, drove it from the flask with a long neck into a receiver with a \(1/20\) N solution of \(H_2SO_4\), poured in there a 4% solution of \(KJO_3\) with a certain amount of \(KJ\), and titrated the liberated \(J\) from the acid remaining unneutralized by ammonia with a \(1/20\) N solution of \(Na_2S_2O_3\). In this, he considers the accuracy of the determination to be up to 0.001%. He paid great attention to the purity of the reagents. In order to obtain a solution of caustic soda free from ammonium salts, he subjected it to boiling in an iron dish. The water used was distilled over sulfuric acid. The sulfuric acid was also used only after previous distillation. According to Tholander’s determination, the amount of nitrogen in Bessemer iron from Avesta ranged from 0.022 to 0.012%.

In these analyses the carbon content is also indicated, from 0.05 to 0.15%, and silicon from 0.006 to 0.014. Open-hearth sheet iron of the ore process from the Avesta, Finspong, and Hammarby works had nitrogen from 0.005 to 0.013%, with carbon content everywhere about 0.1, and silicon from 0.008 to 0.021%.

Open-hearth cast irons contained nitrogen from 0.005 to 0.011; silicon, moreover, from 0.1 to 1.41%. Tholander’s data are of interest not only in the sense that they show the amount of nitrogen in various products of ironworks manufacture, but also because they give here the contents of silicon and carbon as well. However, the fluctuations in the contents of both elements are such that from them no judgments can be drawn as to the influence of silicon and carbon on the nitrogen content in the metal. For example, two samples of Bessemer iron, which each had 0.05% C, contained 0.015% N, while the silicon was 0.006 in the first and 0.009% in the second. A third sample with 0.01% C and 0.014% Si contained almost the same percentage of N—0.016. The same may be said of open-hearth steel. Cast iron with 1.41% Si contained 0.010% N, and another sample with 0.41% Si—0.011% N. Crucible steel contained less nitrogen than the preceding grades of iron, from 0.006 to 0.008%.

The figures for the percentage nitrogen content according to Tholander’s investigations, in comparison with the most recent data for similar grades of cast iron, steel, and iron, are high. One of the chief reasons for this is that he did not take into account the transfer of alkali during distillation into the receiver, and, moreover, distilled the ammonia without using a dephlegmator.

In 1897 an investigation of Bessemer, open-hearth, and crucible steel was carried out by F. Harbord and T. Twinam⁶. These authors give detailed analyses of each sample, determining, in addition to nitrogen, C, Si, P, Mn, As. In Bessemer rail steel the nitrogen content was 0.007–0.027%. We note that from the table given in this work it appears that in those samples which contain the greatest amount of silicon and manganese, there is also more nitrogen. Open-hearth steel contained from 0.11% N to 0.023% N. Here the same phenomenon is observed. Crucible steel contained from 0.007 to 0.009% N. The influence of the indicated elements cannot be discerned from the three analyses cited.

A more detailed work on nitrogen in iron than the preceding one is the investigation of Hjalmar Braune⁷. In addition to a series of analyses of various ironworks products of manufacture, he carried out mechanical investigations of wire containing different amounts of nitrogen, determined the electrical conductivity and magnetic properties, and also made microscopic investigations. On the basis of his work, Braune derives standards for nitrogen content in grades of iron for various purposes. He performs the nitrogen analysis by the wet method, dissolving the weighed sample in $H_2SO_4$ or $HCl$ and distilling, after saturating with $KOH$. The distillate was titrated in the presence of methyl orange or litmus. On the purity

there are no practical indications for it. He carries out the distillation without a dephlegmator. Under these conditions, together with the water vapor, considerable quantities of alkali pass over into the receiver. With the use of dephlegmators, as my investigations have shown, much less alkali passes over, but nevertheless this quantity too has to be taken into account in analyses; otherwise, owing to the small nitrogen content in the metal, inflated results are obtained. Many compounds, when solutions are evaporated, are carried off with the vapors. Under these circumstances, when such small quantities of the substance being determined are involved, attention must be paid to this point.

Braune also determined nitrogen in graphite from cast iron by burning it in a tube with mercuric sulfate, and in doing so obtained a negative result. He prepared specimens for mechanical investigation by saturating iron wire 7–8 mm in diameter with nitrogen by means of dry ammonia at a temperature from 600° to 800°. Since he noticed that the wire was thereby saturated with nitrogen unevenly through its thickness, he then annealed it in sand (the temperature is not indicated). In this process the surface of the wire must have become strongly oxidized. The uniformity of the distribution of nitrogen throughout the entire thickness of the wire, and also along its length, was not checked by analyses; yet the uniformity of the nitrogen distribution throughout the whole specimen is of extremely great importance here.

For the investigation, wrought iron and open-hearth iron were used. The wrought iron was taken in the form of wire of very soft iron with an insignificant amount of impurities: C—0.06%, Si—0.01%, Mn—0.06%, S—0.005%, P—0.05%. Wires containing more than 0.06% N were not obtained uniformly saturated over the whole cross-section and had a core clearly different from the edges of the cross-section (p. 1431). Braune himself finds the testing of such specimens unreliable, and the data from their investigation, it seems to me, should not have been placed either in the table or in the diagram. Here the question arises whether the other specimens too were uniformly saturated with nitrogen, since neither the rate of penetration of nitrogen into the thickness of the specimen nor the optimum temperature of this process had been investigated. If in the preceding specimens as well there was such a core, perhaps only not noticeable to the eye, differing in composition from the peripheral part, then its influence should have had a strong effect on the mechanical test. The results of the mechanical investigations of wrought iron with different nitrogen contents are given in Table I and in the diagram (Fig. 1).

Table 1.

No.	% N	Yield limit $\sigma_s$ kg/qmm.	Temporary tensile resistance $\beta_B$ kg/qmm.	Contraction %	Elongation $(11{,}8\sqrt{F})$	Quality coefficient $\left(\dfrac{\sigma_B}{10}\delta\right)$	Remarks: surface of the specimen after fracture	Remarks: fracture plane
1	1,015	23,2	33,9	69	34,7	1,18	Did not change at all.	Fibrous, without defects.
2	0,015	24,0	34,3	70	33,7	1,16	”	” ”
3	0,028	26,2	34,1	70	32,0	1,09	”	” small irregularity.
4	0,044	27,0	36,5	66	29,9	1,09	”	” ”
5	0,060	27,3	35,3	71,5	30,3	1,07	”	” without defects.
6	0,060	29,4	37,7	57	28,1	1,06	Numerous cracks on the surface, especially in the middle.	” ”
7	0,075	27,8	36,0	66	28,0	1,01	Small cracks near the fracture site.	” ”
8	0,100	22,4	36,5	48	23,0	0,84	”	” ”
9	0,120	31,4	40,2	54	18,0	0,72	More noticeable cracks near the fracture.	” ”

Fig. 1.

Mechanical properties of wrought iron with different amounts of nitrogen.

I — Temporary tensile resistance, kg/qmm.
II — Elongation in %.
III — Tettmayer quality coefficient.

In addition to wrought iron, open-hearth steel was investigated, of the following composition: C—1.15%, Si—0.20%, Mn—0.45%, S—0.012%, P—0.025%. The results of the mechanical testing of wires 7 mm thick, containing from 0.015 to 0.200% N, are given in Table 2 and diagrammatically in Fig. 2.

Table 2.

No.	%, N	Elastic limit $\alpha$ (kg/qmm.)	Tensile strength $\beta$ (kg/qmm.)	Contraction, %	Elongation $(11.3\sqrt{F})$	Quality coefficient $\left(\dfrac{\sigma_B}{10}\delta\right)$	Remarks: surface of specimen after fracture	Remarks: fracture surface
1	0.015	34.9	58.6	50	18.4	1.08	Without change.	Without pores.
2	0.020	35.0	58.7	3	16.0	0.99	” ”	Fine-grained core.
3	0.040	42.4	52.5	—	1.9	0.10	” ”	Shiny crystalline, fine-grained, without pores.
4	0.050	41.3	55.0	—	—	—	” ”	” ”
5	0.090	44.8	60.2	—	2.3	0.14	” ”	” ”
6	0.102	44.5	64.3	—	2.1	0.14	” ”	” ”
7	0.105	43.1	55.0	—	2.0	0.11	” ”	” ”
8	0.106	40.5	53.6	7	2.9	0.16	” ”	” ”
9	0.195	41.0	49.5	—	1.6	0.08	” ”	” ”
10	0.200	39.0	53.7	12	4.9	0.26	” ”	” ”

Fig. 2.

Mechanical properties of nitrogenous cast iron with different amounts of nitrogen.

I — Tensile strength.
II — Elongation in %.
III — Quality coefficient according to Tetmajer.

The composition of the sample of open-hearth iron differs from the composition of welded iron, and the considerable amount of impurities that are present here may have a noticeable influence on the quality of steel saturated to various degrees with nitrogen. The diagram given by Braune for cast steel differs greatly from the diagram he gives for puddled iron and is altogether unexpected in its character. He himself writes that these results astonish him. And indeed, beginning with a nitrogen content of 0.035%, the curves suddenly drop vertically and then extend as a horizontal line; that is, however much nitrogen we add, the properties of the material do not change at all. If we plot on this diagram the points from his second table and connect them with one another, it turns out that the lines he has drawn, especially the upper one, are somewhat arbitrary. The experiments, as can be seen, did not give sufficiently precise results.

Braune then observes that, for hardened steel, nitrogen has greater significance than for unhardened steel. There are no exact investigations in this respect.

In addition to the mechanical test, the electrical conductivity and magnetic properties of an iron wire containing 0.08% C and 0.027% N, and saturated with nitrogen to 0.217 and to 6.6% N, were also investigated. Three specimens of wire were tested for electrical conductivity by means of a Siemens bridge and a mirror galvanometer. The experiment was carried out at 20°.

Wire not saturated with nitrogen, with 0.027% N,

1st determination \(W_1 = 2.7247\).
2nd determination \(W_2 = 2.7294\).

Wire with 0.207% N.

1st determination \(W_1 = 3.1652\)
2nd determination \(W_2 = 3.1762\).

Wire with 6.6% N.

\[ W = 3.0000. \]

For the study of the magnetic properties, wire with 0.08% C and 0.027% N was taken, which was saturated with nitrogen to 0.267% N. For the experiment, pieces of it 200 mm long were taken and joined in sets of 20. The results are seen from the accompanying diagram, Fig. 3.

On the abscissae are plotted divisions proportional to the magnetizing force of the current; on the ordinates, divisions proportional to the magnetic force produced. From the hysteresis loops it is seen that the residual magnetism of iron with 0.267% N is greater than with 0.027% N. The wire with 6.6% N showed very little magnetic force and had very slight residual magnetism.

Microscopic examination showed that, with an increase in the amount of nitrogen, the size of the crystals decreases. The boundaries between the crystals become strongly pronounced. At \(0.12\%\ \mathrm{N}\) many sharp lines are visible, incorrectly called by him Neumannsche Linien. With this nitrogen content the fracture has an entirely conchoidal appearance. At \(0.04\%\ \mathrm{N}\), when the elongation, as he indicates, has completely disappeared, a new structural element appears; it consists of numerous crater-like circles, distributed uniformly in the ferrite at fairly large distances from one another. Boiling with Prof. V. P. Izhevsky’s reagent, sodium picrate, showed that the spaces inside the rings consist of cementite, since they became colored. The rings themselves, in his opinion, consist of nitrogenous iron, which has separated out from the surrounding ferrite.

Fig. 3.

Magnetic properties of nitrogenous iron.

I — wire not saturated with nitrogen.
II — saturated wire.

Hardened specimens of steel containing nitrogen show the usual martensitic structure.

Braune analyzed many samples of cast iron, steel, and iron, Swedish and from other countries. He found that the maximum nitrogen in them was \(0.062\%\), the minimum \(0.02\%\) (there must be a misprint—\(0.002\)). In coke gray cast iron he found from \(0.007\) to \(0.009\%\ \mathrm{N}\). In Swedish gray cast iron, \(0.002\) to \(0.003\%\ \mathrm{N}\) and up to \(0.02\%\ \mathrm{N}\). In white cast iron (foreign), the nitrogen content varied from \(0.021\)—\(0.04\%\) and more nitrogen. In Swedish white cast iron, \(0.003\)—\(0.02\%\ \mathrm{N}\). In some cases here the amount of nitrogen reached \(0.035\%\), and fell to \(0.005\%\). In Lancashire wrought iron there was from \(0.006\)—\(0.008\%\ \mathrm{N}\). Steel and soft open-hearth iron...

of the basic and Bessemer processes had less nitrogen, its content in basic open-hearth steel ranging from 0.006 to 0.018, while in basic Martin steel it reached as much as 0.035%. In crucible steel, in its best grades, he determined nitrogen from 0.015 to 0.020%.

On the basis of his investigations, Braune gives the following standards for iron and steel of various purposes with respect to nitrogen content:

Beams, ship-plate iron—less . . . 0.03% N.
Railway rails . . . . . . . . . . . . 0.025
Springs for railway
cars, large tools . . . . . . . . . . 0.012
Cannon, parts of guns . . . . . . . . 0.008

With regard to Braune’s work, several comments appeared. Brinell expressed the view that the standards given by Braune were still premature. The same opinion was held by S. Dillner, who, together with Brinell, carried out investigations of the influence of nitrogen on various ironworks products and finds Braune’s conclusions too hasty⁸.

In general, they consider Braune’s conclusion concerning the harmful influence of nitrogen on cast iron to be exaggerated; they think that it is negligible, since there is very little nitrogen in cast iron.

However, they do not deny that nitrogen can noticeably impair the quality of wrought iron.

Chapter II

Determination of the Amount of Nitrogen in Iron, Steel, and Cast Iron

Having become interested in the question of the nitrogen content in the products of ironworks production, I undertook a series of determinations of its content in those samples of cast iron, iron, and steel—chiefly of Russian origin—which I had at hand. From the numerous analyses that I made, I came to the conclusion that they contain far less nitrogen than is indicated in the literature. The method used by Braune gave me unsatisfactory results, and I had to introduce certain changes into it. Distillation without a dephlegmator proved very inconvenient, since with the vapors a considerable quantity of alkali passed into the receiver. At first I used Kjeldahl’s attachment. Then I tested a whole series of dephlegmators for this purpose. The ordinary Kjeldahl attachment, while considerably retarding the passage of bases into the receiver, nevertheless allows too much of them to pass there; increasing the burner flame by 2–3 cm increases the passage of bases into the receiver, and, moreover, owing to the variability of the gas pressure, which is observed in many laboratories, it is easy here to make an error with respect to the correction introduced in the analysis for the transfer of alkali into the receiver. If the tube in the bulb of the Kjeldahl attachment, which is directed downward and to the side (Fig. 4), is bent upward in the form of a hook (Fig. 5), then the drop, a, flowing down from the wall of the bulb, as

Fig. 4.          Fig. 5.

is clear from the drawing, is not at the opening of the tube and is not splashed by the jet of steam into the condenser. Koninck’s attachment, to which I turned—

after this, acts almost in the same way as Kieldahl’s modified attachment. Recently I have begun to use, for distillations, a dephlegmator representing a combination of the Kieldahl and Koninck attachments, having the following form (Fig. 6):

Fig. 6.

This apparatus gives still better results and the least fluctuation, as compared with the preceding ones, with respect to the passage of alkali into the receiver when the magnitude of the burner flame and the concentration of the alkali are changed. Lime passes into the receiver in a smaller amount than alkali. The passage (from a flask of 200 cu. cm water, 6 g CaO, 150 cu. cm water being distilled off) with Kieldahl’s modified attachment corresponds to 1.4–1.5 cu. cm of titrated H₂SO₄ solution with a titre of 0.00011. The same distillation, using the last dephlegmator, under the same conditions, corresponds to 0.25–0.3 cu. cm of the same titrated solution.

For the analyses, a weighed portion of shavings was taken of exactly 6 grams; these were dissolved in 35 cu. cm HCl, sp. gr. 1.19 (Kahlbaum), diluted with 50 cu. cm H₂O. The dissolution was carried out in a flask of 250 cu. cm, with a Bunsen valve; at present I find this latter device not

necessary, i.e., dissolution in a flask covered with a watch glass or a glass bulb gave the same results. Initially the filings were dissolved cold, then with heating on a water bath. After dissolution, all of it was transferred into a flask of about 700 cubic cm.

Fig. 7.

with a long neck. The flask was connected to the distillation apparatus, and into it, through a dropping funnel, was poured a small excess of a 1:1 solution of \(KOH\) (Kalium hydricum purissimum, Merck), in an amount of 40 cubic cm. The addition of \(KOH\) was carried out slowly, drop by drop, in order to avoid heating of the liquid and the formation of acid vapors from the acid \(HCl\), which had not yet been neutralized and was present in excess. In this way analyses were made of the specimens listed in Table 3.

Table 3.

Specimen	Nitrogen, %, determination by titration	Nitrogen, %, colorimetric determination
Foundry cast iron from the Yuzovsky Works	0,00255
Lysvensky (Urals) sheet-rolling roll	0,00215
Bessemer cast iron from the Alexandrovsky Works in Ekaterinoslav	0,001099
Gray cast iron from the Gdansevsky Works	0,002026
Coke cast iron, remelted in a cupola, before blowing in the converter of Tropenas, South-Russian Works in Kiev	0,00212
Ural half-and-half cast iron	0,00326
Gray cast iron from the Kasli Works (Urals)	0,00302
White cast iron, remelted from turnings in the open-hearth furnace of the Greter and Krivanek Works in Kiev	0,00534
Foundry cast iron—coke	0,00203
Hematite cast iron from the Gutehoffnungshütte Works	0,00145
Gray Luxembourg cast iron	0,00132

	% Nitrogen	% Nitrogen
	Determined by titration	Determined colorimetrically
Steel (C = 0.29) from a basic open-hearth furnace of the Greter and Krivanek works in Kiev	0.00532
Open-hearth steel of the Kamensk works	0.00312	0.00218
Open-hearth steel of the Risa works (Saxony)	0.00433	0.00267
Open-hearth steel of the Aleksandrovsk works in Ekaterinoslav	0.00408	0.00310
Puddled steel of the Tagil works	0.0151	0.00922
Bloomery iron of the Nizhne-Turinsk works	0.001665	0.00145
Steel for Tropenas converter of the South-Russian works in Kiev	0.00683	not determined
Small-Bessemer steel of the Nevsky works	0.00752	0.00623
Bessemer steel of the Aleksandrovsk works in Ekaterinoslav	0.0153	not determined
The same, another sample	0.0135	0.0121

Recently, however, in all my work on nitrogen in iron I have carried out the distillation using not KOH, but lime water, which, in the form of milk of lime, was poured into the flask through a dropping funnel. Part of it enters into the reaction, while the excess is visible as a slight cloudy precipitate. The passage of lime into the receiver occurs in an entirely negligible amount, especially when distilling with a dephlegmator (see Fig. 6), and during boiling there are none of those shocks that always occur from caustic alkali at the end of the distillation. A corresponding correction was made for the amount of bases passing into the receiver.

Into the receiver there was first poured, see Fig. 7, 10–20 cubic cm of $H_2SO_4\ \frac{1}{100}N$; into it the liquid was distilled off to 105–115 cubic cm, i.e., to approximately $2/3$ of the initial volume.

After the distillation, which was carried on for about $1\frac{1}{2}$ hours, a solution of $KJO_3 + KJ$ was added to the distillate, and the iodine liberated by the acid not neutralized by ammonia was titrated by means of a $\frac{1}{100}N$ solution of $Na_2S_2O_3$ in the presence of starch.

Initially I used a $\frac{1}{10}N$ solution, but obtained unsatisfactory results and therefore was compelled to turn to titration with more dilute solutions. Each sample was analyzed at least twice.

Some of the analyses were checked colorimetrically as well. The colorimetric method is used for determining small amounts of $NH_3$ in water, since Nessler’s reagent, in very dilute $NH_3$ solutions, gives no precipitate but imparts an orange-yellow coloration to the liquid.

The distillate obtained in the manner described above was poured into a graduated cylinder with a ground-glass stopper, into which 2 cc of Nessler’s reagent was added.

A coloration appeared, proving the presence of ammonium salts.

For comparison and for the quantitative determination of nitrogen in this test, I prepared a solution of $NH_4Cl$ containing, per liter, 0.2247 g of $NH_4Cl$, or 1 cc = 0.0000588 g N.

I took 2 cc of this solution with a pipette, poured it into a graduated cylinder of the same size as that used for the substance being analyzed, diluted it with water, and likewise added 2 cc of Nessler’s reagent, in order to compensate for the slightly yellowish coloration possessed by Nessler’s reagent. Then the liquids in the two cylinders were diluted until an identical coloration was obtained.

I found that it is difficult to bring the two columns of liquid exactly to the same density of coloration, since several cubic centimeters of added water sometimes produced only a barely perceptible weakening of the color. To check the dilution I also used another method, namely, I poured both liquids into two test tubes of exactly the same thickness, with marks at the same height from the bottom, and compared the colors of the liquids by looking from above downward, i.e., through columns of liquid of the same height (120 mm). With this method the experiment comes out somewhat more accurately.

As for the reagents, experience showed me that they and the distilled water contained ammonia which, given the small quantities of N contained in the materials under investigation, greatly impaired the accuracy of the analysis. I purified the distilled water from ammonia by distillation over sulfuric acid.

The figures obtained by me proved to be smaller than those given in the literature, as is evident from the analyses presented in Table 3.

Comparing the figures I give, we see that they are lower than those found in the literature.

Comparing the colorimetric determinations with the data obtained by titration methods, we observe that the latter are greater than the former.

It may be thought that this occurs because the nitrogen of steel, and especially of cast iron, combines with the hydrocarbons formed during solution in acid and yields amines. As we know, sulfur, for example, which during the dissolution of cast iron, steel, and iron is converted into $H_2S$, partly gives thioethers (e.g. $CH_3SCH_3$); therefore, in the analysis for sulfur, this circumstance is usually taken into account, and the gases are passed through a heated tube to decompose such compounds.

Let us turn our attention to the heats of formation of dimethyl sulfide and of certain amines.

The heat of formation of \( \mathrm{C_2H_6S} \), according to Thomson’s determinations, is equal to 7.0 Cal (Die Verbrennungswärmen organischer Verbindungen, Stohmann. Zeitsch. f. Phys. Chem. B. VI, 1890, pp. 334—359).

As for the heats of formation of methylamine, ethylamine, dimethylamine, and trimethylamine, they prove to be even higher than that of dimethyl sulfide, and the reactions by which they are formed must proceed much more easily.

\[ \begin{aligned} \text{Heat of form. } \mathrm{CH_3NH_2} &= 9.6 &&(\text{Müller, ibid})\\ \text{“} \quad \text{“} \quad \text{“} &= 8.2 &&(\text{Th.})\\ \text{“} \quad \text{“} \quad \mathrm{C_2H_5NH_2} &= 19.8 &&(\text{B.})\\ \text{“} \quad \text{“} \quad \text{“} &= 13.8 &&(\text{Th.})\\ \text{“} \quad \text{“} \quad \mathrm{(CH_3)_2NH} &= 9.6 &&(\text{Müll})\\ \text{“} \quad \text{“} \quad \text{“} &= 3.6 &&(\text{Th.})\\ \text{“} \quad \text{“} \quad \mathrm{C_3H_9N} &= 14.9 &&(\text{Müll. Bull. Par. 44; 609})\\ \text{“} \quad \text{“} \quad \text{“} &= 9.9 &&(\text{Th.}) \end{aligned} \]

In view of these figures, I come to the conclusion that, along with the formation of organic sulfur compounds during the dissolution of cast iron, steel, and iron, amines may also be formed.

And in fact, the coloration given by Nessler’s reagent to the distillates from cast iron shows the unquestionable presence of methylamine, since it is much yellower than the coloration given by \( \mathrm{NH_4Cl} \).

A solution of methylamine prepared by me gave, with Nessler’s reagent, a bright, transparent, yellow coloration.

The colors of the cast-iron distillates differed so much in their yellowness and greater transparency from the coloration of an \( \mathrm{NH_4Cl} \) solution that it was impossible to make colorimetric determinations of \( \mathrm{N} \) in cast iron using an \( \mathrm{NH_4Cl} \) solution for comparison.

As regards the presence of other amines of greater molecular weight, under our conditions nothing can be said by means of the color reaction with Nessler’s reagent, since with an increase in the organic groups of the amine the coloration of the mercurammonium compound diminishes, and trimethylamine already gives no coloration; in concentrated solutions, however, a white precipitate of its mercurammonium compound is formed.

Thus, during the dissolution of cast iron, hydrocarbons are formed from chemically combined carbon, and these, with the nitrogen of the cast iron, give amines. During the dissolution of steel hydrocarbons likewise appear, and a certain quantity of amines must likewise be formed. It is by this circumstance that I explain the fact that, colorimetrically, lower figures are obtained for the content of \( \mathrm{N} \) in steel than the figures obtained by titration.

Soon after my work, an article by Grahe¹⁸ appeared, in which he expresses doubt as to whether these insignificant quantities—

¹⁸ Revue de Metallurgie 1908. 353.

nitrogen that are encountered in different grades of iron exert a noticeable influence on their mechanical properties. He briefly cites the principal literature in this direction and dwells in greatest detail on Braune’s work, thanks to which the question of the content of nitrogen and of its influence on the mechanical properties of steel and iron has recently acquired particular urgency. In collaboration with G. Petren, he verifies the analytical method used by Braune, at the same time drawing attention to the purity of the reagents, carries out a series of analyses of various samples of cast iron and steel, and thereby obtains results on the basis of which he comes to the conclusion that the percentage figures for nitrogen content in Braune’s work are exaggerated. Thus, 12 grades of Swedish soft iron, according to the investigations of these authors, contained from 0.0020% to 0.0045% nitrogen. Then, 38 grades of steel obtained by means of various metallurgical processes contained 0.0025–0.0125% N. In 20 grades of coke and charcoal pig iron, according to their determination, there was 0.0010–0.0065% N. Grabe, also referring to my work, the figures for the determination of nitrogen in which agree with his determinations, comes to the conclusion that such insignificant amounts of nitrogen content can exert an exceedingly small influence on the mechanical properties of iron, steel, and cast iron.

The development of analytical methods for the determination of nitrogen is a matter of very great importance, since, before judging the degree of its influence on the qualities of the products of ironworks production, it is necessary to be able to determine precisely those quantities of nitrogen that occur there.

On the question of the determination of nitrogen in iron and steel, at the end of the same year there appeared a work by De Osa⁹, in which he dwells in detail on my work and finds the degree of accuracy of nitrogen determinations by this method. According to his calculation it follows that, when ammonia is titrated with a \(1/100\) N solution of sulfuric acid, the error may fluctuate within the limits of \(\pm 3.3\%\) of the quantity being determined, at the contents of it ordinarily found in iron. In order to find experimentally the degree of accuracy of the method, he takes a solution of ammonium chloride containing 0.02887 g \(NH_4Cl\) per liter; 40 cubic centimeters of this solution contain 0.001155 g \(NH_4Cl\), which corresponds approximately to 0.0003 g nitrogen. Such a quantity of nitrogen is minimal in analyses of iron poor in nitrogen. In these 40 cubic centimeters of \(NH_4Cl\) solution he found by analysis 0.001139 g and 0.001047 g \(NH_4Cl\), on the average 0.001093 g, instead of the 0.001155 g \(NH_4Cl\) actually contained there. Thus, the error of determination will be about \(\pm 4\%\). Further, he points to that inaccuracy which

may occur in determining nitrogen from the presence of ammonium salts that may be contained in the hydrochloric acid used for dissolving the sample. Distilling with lime 130 cubic cm of concentrated hydrochloric acid, diluted with 200 cubic cm of water and previously boiled with lime to remove all ammonia from it, he found from two determinations 0.00008172 and 0.00005777 g N. If the larger quantity of nitrogen in the hydrochloric acid is adopted, and if it is assumed that in the receiver we obtained 0.0003 g of nitrogen, then the error in determining nitrogen in such an iron sample, if the purity of the hydrochloric acid is not taken into account, will amount to 27%. Under the conditions of my experiment, when 35 cubic cm of $HCl$—7.3%—were taken. From this it is clear how necessary it is to pay attention to the purity of the reagents. De Osa assumes that, when iron containing nitrogen is dissolved in hydrochloric acid, besides ammonia, nitrogen oxides are also formed, which, in analyses by the indicated method, escape determination.

To determine nitrogen from the oxidized compounds assumed by him to be formed in dissolving iron containing nitrogen, he carried out a series of experiments. For this purpose, after distilling off a portion of the liquid, he adds zinc to the distillation flask and distills off the next portion of liquid. In the first distillation, of course, there will be almost all the ammonia from the liquid being distilled; in the second portion, in his opinion, there should be ammonia formed chiefly from the reduction of nitrogen oxides. In this he is certain that these oxidized compounds will be completely reduced by zinc in the presence of lime and will give ammonia.

To investigate this, he took about 5 g of iron, dissolved it in 130 cubic cm of strong hydrochloric acid, and added 50 g of lime suspended in water. The total volume of liquid was brought up to 400 cubic cm. The first distillate, in the amount of 90 cubic cm, was titrated; then zinc was added to the distillation flask, and a further 90 cubic cm was distilled off.

In the first distillation (90 cubic cm) scarcely all the ammonia passes over. According to my experiments it is necessary to distill off about $^2/_3$ of the liquid, i.e. in this case not less than 250 cubic cm.

It is very probable that residues of ammonia also passed into the second distillation. The results were as follows:

Charge, 5.0898 g of iron:

first distillation	0.0003077 g nitrogen	= 0.0060% N
second “	0.0000901 “ “
Total	0.0003978 g nitrogen	= 0.0078% N

Charge, 5.2583 g of iron:

first distillation	0.0004712 g nitrogen	= 0.0090% N
second “	0.0000373 “ “
Total	0.0005085 g nitrogen	= 0.0097% N

Sample of 5.0927 g of iron:

\[ \begin{aligned} \text{first distillation} \quad &0.0004222\ \text{g nitrogen} = 0.0083\%\,N,\\ \text{second } \quad &0.0000331\ \text{g } \quad \text{”}\\ \hline \text{Total} \quad &0.0004553\ \text{g nitrogen} = 0.0089\%\,N. \end{aligned} \]

De Osa considers the greatest deviations between determinations to be \(+15\), \(-23\%\), not taking into account the figure from the second distillation. If the figures of the second distillation are added to the results of the first distillation, then the average nitrogen content will be \(0.0088\%\) and the deviations \(+10\) and \(-11\%\).

Further, despite precautions with regard to the uniformity of the distribution of nitrogen in the sample, when he took the average wire samples with the greatest care, he obtained the following results: from four samples of wire, with an average nitrogen content determined by one distillation of \(0.0194\%\), deviations of \(+10\) and \(-14\%\) were found; and when taking into account the additional figures from second distillations with zinc, with an average nitrogen content of \(0.0204\%\), the deviations proved to be \(+11.7\%\) and \(-15.7\%\).

From a series of his experiments he draws the conclusion that, in the method of determining nitrogen, a second distillation with zinc is necessary; otherwise part of the nitrogen escapes determination, and its amount varies, as he finds, from 3 to 29% relative to the amount of nitrogen determined in the first distillation. De Osa tries to use other reagents for the analysis of iron: for example, double potassium-copper cyanide, with which he heats iron filings in a vacuum, then sulfur, iodine under the same conditions, but these methods did not give good results.

The best results were obtained by heating small iron turnings with sulfur in a vacuum to 300—400°. In this process a series of gaseous substances is obtained, and among them nitrogen. A sample of about 5 g of comminuted iron wire gave him a volume of hydrogen and nitrogen gases of 0.1—0.25 cubic cm. In addition, oxygen and carbonic acid amounted to from 0.03 to 8 cubic cm, and hydrochloric acid to from 15—35 cubic cm. It is difficult to measure precisely such insignificant quantities of gas as 0.1—0.25 cubic cm, and it is no surprise that he obtains results with considerable deviations from one another. In one determination he obtained \(0.0018\%\,N\), in another—\(0.0059\%\,N\) in the same wire. In cement steel he found, by one determination, \(0.23\%\,N\), by another—\(0.085\%\). In the most careful analysis of the wire he obtained, in the same sample, \(0.003\)—\(-0.011\%\,N\).

From these figures it is evident that the method proposed and described in detail by him is sharply inferior in accuracy to the generally accepted one at the present time, i.e. the method of dissolving iron in hydrochloric aci-

and titrimetric determination of the nitrogen thereby obtained in the form of ammonia.

De Osa’s article prompted me to undertake a series of investigations concerning the possibility that the nitrogen contained in iron, when the latter is dissolved in hydrochloric acid, passes into oxidized compounds, and to determine what influence zinc may have during the distillation of a solution of nitrogenous iron with an excess of both caustic potash and lime. If part of the nitrogen can pass here into an oxidized compound, then, of course, this should be noticeable in analyses of iron rich in nitrogen. The conditions for the formation of oxidized nitrogen compounds during the dissolution of iron rich in nitrogen should be more favorable than with iron very poor in nitrogen, since in the latter case more hydrogen is evolved, which, in statu nascendi, should create more favorable conditions for the conversion of nitrogen into ammonia.

I proposed that F. A. Belyshev carry out investigations concerning the possibility of the formation of oxidized nitrogen compounds when nitrogenous iron is dissolved in acid, and concerning the effect of adding zinc during the distillations.

For the work, finely reduced iron, saturated with nitrogen by means of ammonia, was taken; analysis showed it to contain 5.882% N.

A charge of this iron, in an amount of 0.5 g, was dissolved in 15 cu. cm of hydrochloric acid. To the solution was added a small excess of caustic soda (Natrium hydricum puriss. e natrio pr. anal. Merk.), the solution of which had first been thoroughly boiled and which, with Nessler’s reagent, gave neither any coloration nor a small amount of lime milk to improve boiling. The total volume of liquid was 200 cu. cm. The water for the experiments was purified by distillation over sulfuric acid with chameleon.

After the ammonia had been distilled off, when Nessler’s reagent no longer showed any noticeable yellowing with small portions of the liquid, a series of distillations was carried out (150 cu. cm each time), with only water added each time to the residue in the flask. In a large amount of distillate, yellowing from Nessler’s reagent was clearly noticeable. Colorimetric determination of nitrogen in the successive series of distillates gave the following results:

Distillate	Nitrogen
in the 1st distillate	0.000064 g N
“ 2 “	0.000048 “
“ 3 “	0.000061 “
“ 4 “	0.0000514 “
“ 5 “	0.000057 “
“ 6 “	0.000034 “
“ 7 “	0.000031 “

Distillate	Nitrogen
in the 8th distillate	0,000028 gr. N
9th “	0,000027 “
10th “	0,000018 “
11th “	0,000016 “
12th “	0,000016 “
13th “	0,000016 “
14th “	0,000013 “

Thus, it proves difficult to obtain a distillate that does not yellow with Nessler’s reagent, although the amount of nitrogen, if ammonia is regarded as the cause of the yellowing, decreases.

To the residue in the flask from the preceding distillation, 5 gr. of granulated zinc was added. A consecutive series of distillates of the same volume of liquid gave the following results:

1) 0,0000146 gr. N.
2) 0,0000146 gr. N.
3) 0,0000138 gr. N.

Experiments with ferrous nitride, 11,1% N, gave similar results. Hence it may be concluded that the influence of zinc, if it exists at all, is quite negligible, and it is most probable that oxygen compounds of nitrogen are not formed when nitrogenous iron is dissolved.

Next, the distillation of water with alkali was investigated, in order to determine the properties of the liquid distilling off.

It turned out that such distillates also give a yellow color with Nessler’s reagent, and, if the amount of nitrogen in these distillates is determined, approximately the same amount of nitrogen is obtained as was present in the first distillates of the preceding experiment.

In order to investigate the influence of the atmosphere, the whole apparatus was filled with hydrogen, and the distillations of the solution of nitrogenous iron were carried out in a stream of hydrogen.

Distillate	Contained nitrogen
first distillate	0,000059 gr. N.
second “	0,000048 gr. N.
third “	0,000048 gr. N.

Exactly the same experiment was carried out with distillation in a stream of air, whereby the following results were obtained:

Distillate	Nitrogen
first distillate	0,000058 gr. N.
second distillate	0,000044 gr. N.

From this we see that the atmosphere of the flask has no noticeable influence on the results of the distillation.

In the literature there is an indication (Ram̃melsberg) that, if the residue in the flask from a distillation in the determination of nitrogen in iron is allowed to stand for several days, then, after adding water to it and distilling it off, we obtain a certain amount of nitrogen, far greater than if we were to carry out the distillation immediately.

immediately after the determination of nitrogen. This phenomenon also occurs when we take only the reagents. Thus, the flask from the preceding experiment was left standing for several days, connected to the condenser. The distillate then showed the presence of 0.000094 g of N, i.e., twice as much as in the preceding determination.

Then I investigated the course of the distillation of water with the use of CaO as the base.

Fig. 8.

For this purpose, 200 cubic cm of water purified in the manner described above and 6 g of CaO by weight were taken. By titration, in a series of successive distillations, 0.000063, 0.000061, 0.000063, 0.000065 g of N were determined. Colorimetric determinations of the same distillates gave the following results: 0.00000887, 0.0000097 g of N. In the distillate for the colorimetric determinations, a fresh portion of CaO had been taken. As can be seen, almost always one and the same amount of titratable substance distils over, in which there is a substance that gives a very pale, but nevertheless noticeable, yellow coloration with Nessler’s reagent, visible especially well in the apparatus that served me for the colorimetric determinations. See Fig. 8.

Distillation with Nessler’s reagent, in an amount of 150 cubic cm poured into one of the tubes, in comparison with a column of distilled water of the same height and with the same quantity of Nessler’s reagent, gave, although very slight, a perfectly distinct yellow coloration admitting of no doubt. No precaution, such as calcining the lime, purifying the water, blowing steam through the refrigerator, or distilling in a stream of hydrogen, destroyed this reaction.

It was then of interest to determine the influence of hydrochloric acid, which served for the analyses (sp. gr. 1.19, Merck, für analytische Zwecke). Adding to the same portion of water and lime 10 cubic cm of acid, I found colorimetrically in the distillate 0.00000804 g N, whence I concluded that the acid contained practically no ammonium salts. To determine whether this acid contained nitrogen in oxidized states, 5 g of zinc were added to the same quantity of water, lime, and acid. The colorimetric determination gave 0.00000832 g of nitrogen, i.e., almost the same amount.

In order to confirm the ability of zinc under these conditions to convert oxidized compounds into ammonia, the following was done: 200 cubic cm of water, 6 g of CaO, and 0.1 g of KNO₃ were taken. The distillate gave a yellowing not differing from the distillation of water alone with lime. When water and 5 g of Zn were added to the same flask, the distillate from Nessler’s reagent gave a large precipitate, which indicates a considerable quantity of ammonia. Consequently, the reduction of nitric acid to ammonia takes place here successfully.

Thus, when iron is dissolved in acid, the nitrogen contained in it does not pass into oxidized compounds; if they are formed at all, it is in such negligible quantities that they cannot be detected by the methods used.

Distillation with lime gives almost ten times less of the substance that gives a yellow color with Nessler’s reagent and that we take for ammonia. In view of the fact that caustic alkali, in however pure a form it may be taken, always gives far more carry-over into the receiver, and since the distillation of ammonia with lime proceeds just as well as with caustic alkali, while the latter is convenient in that it does not cause bumping during boiling, I came to the conclusion that, in order to avoid errors in the corrections, which with caustic alkali are subject to much larger fluctuations than with lime, lime should be recommended for this kind of work.

In view of the fact that a series of distillations of water added to the flask to the residue from the distillation of ammonia from a solution of ferrous nitrate always yields amounts of nitrogen very close to those from a series of distillations of lime alone with water, it must be thought that we carry out the distillation of ammonia sufficiently completely, that it passes off with the vapors practically in its entirety.

Thus, the use of zinc, proposed by De Osa, must be regarded not only as superfluous, but also as harmful in the event that nitric or nitrous acid is present in the reagents. A second distillation with zinc gives figures closely approaching those obtained from a long series of successive distillations.

When determining, in the distillate, the amount of base by titration, De Osa considers it entirely to be ammonia and converts it to nitrogen, taking no account whatever of the fact that the lime is known to pass over during distillation into the receiver. The amount of bases thus passing into the receiver together with the vapors of the distilling water depends both on the rate of distillation and on the system of the dephlegmator, and also on the concentration of the alkali.

For such small quantities of nitrogen as occur in iron, the distillation of ammonia without a dephlegmator for its titrimetric determination is entirely unsuitable.

This is why Braune obtained the elevated results of his determinations. It is inconvenient to take too large weighed portions of the material being analyzed in order to obtain larger quantities of ammonia and thereby reduce the influence of the bases passing with the steam into the receiver.

In this case, moreover, a mass of precipitate forms in the flask, which makes it difficult to distill off the last portions of water; too much liquid remains in the flask, and the distillation of ammonia is unreliable.

Chapter III.

Pure Iron and Nitrogen.

Pure iron with dry nitrogen, according to my investigations, does not enter into reaction at any temperature up to melting¹⁰, the same conclusion also having been reached by I. Zhukov¹¹; whereas dry ammonia at 200° acts upon it, forming nitrous iron. For the work¹² iron was taken which had been prepared in the laboratory by reduction with hydrogen from iron oxide obtained from Fe₂Cl₆ (Eisenchlorid cryst. puriss. pro anal. Merk.).

The hydrogen was removed from it by heating to 400° in a glass tube in a vacuum produced by a mercury pump.

In addition to this iron, Swedish soft wrought iron, containing few impurities, was also used in the work:

0.014%	0.14%	0.08%	0.012%	0.1%
Si	Mn	P	S	C

The nails were converted into filings with a 12-inch flat bastard file.

The apparatus for the work was arranged as follows (see Fig. 9).

Fig. 9.

A—Héroult electric furnace, into which a glass refractory tube B (d = 12–15 m/m) is inserted. From one end a thin porcelain tube C (5 m/m internal diam.) was inserted into this tube; in it was enclosed the thermoelement of the Le Chatelier pyrometer; M—millivoltmeter. From the other end, by means of iron wire, one or two porcelain boats with a weighed portion of iron were introduced. Ammonia was taken from a bomb E with liquid NH₃, and the rate of its flow was regulated by means of a reducing valve. Before the ammonia entered tube B, it passed through two tall drying columns, one with pieces of KOH and the other with BaO. In gasometer P there was nitrogen, which from there was passed through a solution of KOH, through sulfuric acid and heated metallic-copper turnings for the decomposition of nitrogen oxides, and then again through H₂SO₄, and finally entered tube B.

The procedure in saturating the iron with nitrogen was as follows. After heating the furnace to a definite temperature, a current of dry nitrogen alone was admitted from gasometer P; boats with the substance were quickly inserted into the front end of tube B and held there for 3–5 min., and then moved into the furnace. The current of nitrogen was continued until the time when the proper temperature had been established in the furnace. (After the boats were introduced into the furnace, the temperature fell by about 50–70°.) When the required temperature had again been established in the furnace, the current of nitrogen was stopped and ammonia was admitted for an hour at constant temperature. After an hour the current of ammonia was stopped and nitrogen was admitted again; after 3–5 min. the boats were removed from the furnace, cooled in the front end of tube B in a stream of nitrogen, transferred to a desiccator, and the substance was weighed for dissolution.

The dissolution of the nitrogenous iron was carried out in a conical flask covered with a watch glass. For a weighed portion of 0.3–0.7 g, 10 cu. cm of strong hydrochloric acid were taken (Kahlbaum, sp. gr. 1.19). Initially the weighed portion was dissolved in the cold, and toward the end it was heated. After dissolution, the contents were transferred to a flask of 500 cu. cm capacity for distillation.

It should be said here, incidentally, that if any alkali does pass from the flask into the receiver, then in our experiments this is reflected in the results only in hundredths of a percent.

However, a correction for the passage of alkali into the receiver was always made. From there, the receiver contents were transferred to a flask of 250 cu. cm capacity and titrated in portions. For titration a solution of NaOH—$1/100\,N$, prepared from metallic sodium, and H₂SO₄—$1/100\,N$ were used. Methyl orange served as the indicator.

In those cases in which the nitrogen content in the solution did not exceed 0.5%, a \(KJO_3\)—\(KJ\) solution was used for titration (per 250 cc of water—2.2 g \(KJO_3\) and 5.5 g \(KJ\)). The iodine liberated from the acid not neutralized by ammonia was titrated with a \(1/100\,N\) solution of \(Na_2S_2O_3\) in the presence of starch.

When distilling the solution with the maximum nitrogen content (11.1%), the acid poured into the receiver was rapidly neutralized by ammonia and the solution became alkaline. In order to make certain that in this case no loss of ammonia was taking place, the receiver was connected to the same bulb, filled with a solution of \(HgNO_3\), and the solution remained clear, without forming \(NH_2HgNO_3\).

Below, in Tables 4 and 5, we give the figures obtained during the work. Samples of 0.3 g to 0.7 g were taken.

Table 4.

Material	200°	300°	350°	400°	450°	500°	600°	650°	700°	750°	780°	850°
Chemically pure iron	0.11	3.05	—	10.68	11.09	10.37	8.90	—	4.67	0.57	0.30	0.29
Chemically pure iron	0.11	3.07	—	10.73	11.11	10.35	8.91	—	4.85	0.50	0.30	—
Chemically pure iron	0.11	3.15	—	—	11.00	—	—	—	—	—	—	—
Chemically pure iron	0.11	—	—	—	11.14	—	—	—	—	—	—	—
On average	0.11	3.09	—	10.705	11.11	10.36	8.905	—	4.76	0.535	0.30	0.29

Table 5.

Material	200°	300°	350°	400°	450°	500°	600°	650°	700°	750°	780°	850°
Swedish soft nail iron	—	—	—	0.48	—	0.73	5.99	5.55	5.18	1.26	0.75	0.48
Swedish soft nail iron	—	—	—	0.49	—	0.83	5.84	5.84	5.14	1.40	0.69	—
On average	—	—	—	0.485	—	0.78	5.915	5.695	5.16	1.33	0.72	0.48

The data of the table are presented graphically in the diagram of Fig. 10.

Examining these curves, we see that, for chemically pure iron in fine powder, the maximum absorption of nitrogen is equal to 11.1% at 450° C, which corresponds to the formula of ferronitride \((\mathrm{Fe}_2\mathrm{N})\). (Table 4).

Fig. 10.

For Swedish nail iron in the form of filings, the maximum absorption of nitrogen in the same time (1 hour) was already 5.9% at a temperature of 600° C (Table 5).

At this temperature (600°), chemically pure iron combines with 8.9% nitrogen.

The difference in the quantities of nitrogen absorbed by chemically pure iron and by Swedish nail iron, at one and the same temperatures, for example at 600° C, can be explained by the coarseness of the filings of the Swedish iron in comparison with the extreme fineness of the iron reduced by hydrogen. The action of nitrogen on iron in this case may be likened to the action of carbon in the cementation of iron. Nitrogen gradually penetrates into the interior of the piece of iron, and the rate of its penetration depends on the temperature and the size of the piece. Then, the absorption of nitrogen by iron apparently depends on those impurities (Si, Mn, and C) which are present in the Swedish iron, although only in small quantity. The latter supposition, it seems to me, follows from the fact that, beginning at 690° (see the diagrams), Swedish iron absorbs more nitrogen than chemically pure iron. The compound of nitrogen with iron at high temperature is unstable. To clarify the course of decomposition of ferronitride with increasing temperature, a series of experiments was made in which it was heated in an atmosphere of nitrogen. For this purpose, nitrogenous iron was taken, obtained by the action of ammonia on chemically pure iron at 450°; it was heated in a Heraeus furnace at various temperatures, in a current of dry nitrogen. In this way the following results were obtained—see Table 6.

Table 6.

Temperatures	Weights in grams	N content in Fe before heating, in %	N content in Fe after heating, in %	Average loss during heating, in %
600° C	0,4100	11,14	5,53	5,53
600° C	0,3454	11,16	5,53	5,53
740° C	0,27?0	11,10	0,26	0,225
740° C	0,3634	11,10	0,19	0,225
800° C	0,2960	11,10	0,31	0,28
800° C	0,3044	11,11	0,25	0,28

If we take the figures of the last column and compare them (Table 7) with the figures for the nitrogen content in iron that had been subjected to the action of ammonia at the corresponding temperatures, but had not been heated in an atmosphere of nitrogen, we shall observe that the latter nitrogen content is somewhat greater than the former.

Table 7.

Temperatures	N content in Fe after heating in a current of nitrogen, in %	N content in Fe nitrided, but not heated in a current of nitrogen, in %
At 600° C	5,53	8,90
At 740° C	0,225	1,70
At 800° C	0,280	0,330

This difference shows us what influence the surrounding atmosphere has on the degree of dissociation of nitrogenous iron.

Let us also draw attention to the fact that, in the process of saturating iron with nitrogen under our conditions, dissociation of ammonia takes place.

According to the latest work of F. Haber und G. van Oordt (Zeitschrift für anorganische Chemie: “Über die Bildung von Ammoniak aus den Elementen,” 1905, p. 341), at a pressure of 760 mm and at various temperatures, ammonia, decomposing, forms equilibrium systems: Table 8.

Table 8.

Temperatures	NH₃	H₂	N₂
At 270° C	98,51	1,12	0,37
At 327° C	8,72	68,46	22,82
At 627° C	0,21	74,84	24,95
At 927° C	0,024	75,0	25,0
At 1020° C	0,012	75,0	25,0

Thus, the absorption of nitrogen by iron under the conditions of our experiments occurred chiefly in an atmosphere of hydrogen and nitrogen at the moment of their formation and with a small quantity of ammonia.

What influence hydrogen has on the combination of nitrogen with iron—at present nothing definite can be said. Perhaps, in this case, hydrogen plays the role of a catalyst, or else nitrogen, independently of it, combines with iron in statu nascendi.

Passing from the conditions of our experiments to the conditions of metallurgical processes in the manufacture of iron, it may be said that, in the latter case, nitrogen is present mainly from the air, and only the most insignificant quantities from the decomposition of ammonia (from generator gases), hydrogen from hot gases, from the decomposition of water vapor, and very small quantities from ammonia; there is no ammonia at all (it decomposes).

No reaction between iron and pure nitrogen, as indicated above, was observed.

In Héraeus’s furnace, experiments in passing nitrogen over heated iron from 300° to 1350° for one hour gave a negative result.

Iron, melted in Tammann’s furnace in a magnesite test tube and heated to 1600°, was kept in an atmosphere of nitrogen for 2 hours and did not reveal the presence of nitrogen in it.

When nitrogenous iron is heated to a high temperature, we have its complete decomposition even before melting. Nitrogenous iron with a content of 2.1% in an amount of 15 g was heated to 1200° for one hour in a kryptol electric furnace in a carbon crucible with thick magnesite lining, covered with a carefully luted lid. Then, the same specimen of nitrogenous iron was melted under the same conditions.

Neither in the first nor in the second case was it possible to detect nitrogen, except for the amount originating from the reagents.

Thus, iron by itself should not, in plant processes, combine with the nitrogen of furnace gases.

If nitrogenous iron did enter the melt, it would have to decompose completely.

The cause of the nitrogen content in grades of iron of factory origin must be sought in impurities.

The chief impurities of iron are carbon, manganese, and silicon.

Other impurities, such as, for example, sulfur and phosphorus in malleable grades of iron, occur in such small quantities that their influence on the combination of nitrogen can hardly be appreciable; moreover, there is always more nitrogen in them than in cast irons, and especially in the products of the Bessemer process and of the small Bessemer process.

Therefore attention must first of all be directed to carbon, manganese, and silicon.

Chapter IV.

Carbonaceous Iron and Nitrogen.

In order to investigate the influence of carbon on the absorption of nitrogen by iron, Swedish white cast iron, obtained in the cold running of a blast furnace, was chosen as the material to be tested¹³.

The carbon content in it was 3.96%. The quantity of other impurities, comparatively very small, is expressed by the following figures.

Si—0.12%; Mn—0.18%; S—0.08%; P—0.02%; Cu—0.07%.

For the investigation the cast iron was comminuted in the following manner: on a planing machine it was turned into fine shavings, which were then ground in an agate mortar to a powder, which was sifted through a fine sieve.

In this form the cast-iron powder was taken for experiments on saturation with nitrogen at various temperatures by means of dry ammonia in a Heraeus furnace.

A charge in the amount of 0.4–0.5 g was poured in an even layer into a porcelain boat, which, with the aid of an iron wire, was introduced at first into the cold, projecting part of the quartz tube.

After the temperature of the furnace had reached the required level, the air was displaced from the tube by dry nitrogen, and the boat was moved into the middle of the furnace under the pyrometric tube inserted from the other end of the quartz tube.

In order that it should be possible to move the boat into the heated furnace and, at the conclusion of the experiment, to withdraw it into the cold part of the tube projecting from the furnace, without bringing it into contact with air, the device shown in Fig. 11 was made.

Fig. 11.

A—a bent glass tube, into which mercury is poured between the cotton plugs c and d. The iron wire passes through the tube freely, only slightly touching the walls of the tube at the three bent places. The cotton plugs are not inserted tightly.

c—a rubber stopper; f—a disk of asbestos cardboard.

Tube A and the wire were supported by wire hooks suspended from a glass rod fastened to a stand.

Saturation with nitrogen was carried out at various temperatures for one hour, i.e. for the same length of time as in the experiments with iron. The current of ammonia was kept the same. After the saturation with nitrogen was completed, the ammonia was displaced from the tube by hydrogen; the boat was moved into the front, cold part of the tube, which was cooled also from the outside with water or ice.

The reaction product was analyzed as before.

The results obtained are given in Table 9.

Table 9.

Temperature.	300°	400°	500°	600°	700°	800°	900°
Nitrogen content in the samples, in percent.	0.529	1.07	5.12	8.35	7.15	3.20	0.43
Nitrogen content in the samples, in percent.	0.511	1.12	5.18	8.11	7.10	3.33	0.51
Nitrogen content in the samples, in percent.	0.530	1.01	5.02	7.93	7.88	3.08	—
Average . . .	0.523	1.07	5.11	8.25	7.48	3.2	0.47

For comparison of the results of this work with investigations on the combination of nitrogen with filings of soft iron, I give the curve of nitrogen absorption at various temperatures by that and the other material in the following diagram: Fig. 12.

From this diagram it is evident that the optima of combination of nitrogen with carbonaceous iron and with soft iron are very close to one another and lie at about 600° C. Both curves proceed in the same direction as each other. At all temperatures, however, the carbonaceous iron absorbs more nitrogen than the soft iron.

I see the reason for the latter circumstance not in the action of the carbon, but in the degree of comminution of the material.

Swedish white cast iron is brittle and could be ground in an agate mortar into powder, whereas tough iron, filed with a file, could not give such small pieces.

If the carbon content did contribute somewhat to the combination of nitrogen with iron, then this would be of no importance for factory production, since the combination of nitrogen with carbonaceous iron occurs at comparatively low temperatures.

Fig. 12.

I — carbonaceous iron.
II — soft iron.

Thus, from all that has been set forth, it may be concluded that carbon is not responsible for the absorption of nitrogen by iron in the industrial processes by which it is obtained.

Chapter V

Manganese and Nitrogen.

In wrought grades of iron there is usually more manganese in cast iron than in weld iron. In cast iron the amount of manganese is 0.1–1.25%. In pig irons it ranges from tenths of a percent to several percent, and in special pig irons to several tens of percent.

Before investigating the capacity of iron–manganese alloys to combine with nitrogen, it is important to determine how the reaction of combination with nitrogen proceeds for manganese itself.

Metallic manganese, under the action of dry ammonia, gives the compound Mn₃N₂ with 14.50% N; under the action of dry nitrogen, Mn₅N₂ with 9.24% N is obtained, as was found by Prelinger ²⁴).

Wedekind also found another compound of manganese with nitrogen, to which he assigns the formula Mn₇N₂ ²⁵).

In addition, the conditions of interaction between manganese and nitrogen were investigated by I. Zhukov ²⁶). The experiments were carried out in a Héroult electric furnace, into which a porcelain tube was inserted, inside which a boat with finely crushed metal was introduced. From one side, through a tap, definite quantities of nitrogen could be admitted; from the other, by means of a three-way stopcock, it was connected to a mercury pump and a manometer. Air was pumped out of the tube, a definite volume of nitrogen was admitted, and the pressure was observed at various temperatures. The temperature was measured with a Le Chatelier pyrometer. This arrangement made it possible to observe both the temperature and the gas pressure. Zhukov determined the onset of absorption of nitrogen by manganese at a temperature from 850° to 875°. The manganese absorbed up to 12% N, but he did not find a constant dissociation tension at constant temperature; the pressure increased uniformly as a function of the amount of nitrogen absorbed. Zhukov considers that, in this case, a solid solution of nitrogen in the metal is formed, or a solution of a definite compound of manganese with nitrogen in an excess of metal.

²⁴) M. 15, 391.
²⁵) B. d. Deutsch. Chem. G. 1908, 3769.
²⁶) Zh. R. F. Kh. O. 1910, p. 42.

I turn to my own investigations.

For the work, metallic manganese was taken, prepared by the Goldschmidt method and obtained from Merck (Manganum metallicum purum fusum),²⁷).

According to analysis it contained:

\[ \mathrm{Mn}—97.1\%;\quad \mathrm{Fe}—1.52\%;\quad \mathrm{Si}—0.86\%;\quad \mathrm{Al_2O_3}—0.43\%. \]

Purification of such manganese by laboratory means is an extremely difficult matter. In the literature there are references to experiments in this direction carried out at the Aachen Polytechnic,²⁸) but they were unsuccessful.

In view of the foregoing, and also because 97% manganese should be quite capable of showing its characteristic features when saturated with nitrogen at various temperatures, I took Merck’s preparation without attempting to subject it to preliminary purification.

The manganese was crushed and ground in an agate mortar to the finest powder. For each experiment this powder was taken in an amount of about 0.5 g.

Initially, experiments were performed on saturating manganese with nitrogen by means of ammonia.

In order to obtain comparable results, all the experimental conditions were exactly the same as in the work with iron.

The nitrogenous manganese obtained after saturation with nitrogen was analyzed in the same way as nitrogenous iron.

The results of the experiments are given in Table 10.

Table 10.

Action of ammonia on metallic manganese.

Temperature	300°	400°	500°	550°	600°	700°	800°	900°	1000°	1100°	1200°
Nitrogen content in weight percent	0.14	0.58	2.71	9.10	14.40	14.48	14.31	10.94	6.81	4.01	1.79
Nitrogen content in weight percent	0.18	—	2.52	9.31	14.28	14.32	14.39	10.65	7.24	3.96	1.71
Nitrogen content in weight percent	—	—	2.65	9.23	14.41	—	—	10.20	—	4.05	1.68
Nitrogen content in weight percent	—	—	—	—	14.46	—	—	—	—	—	—
Average, %	0.16	0.58	2.63	9.22	14.39	14.36	14.35	10.59	7.02	4.00	1.73

Nitrogenous manganese is a light, almost entirely black powder.

²⁷) N. Chizhevsky. Bulletin of the Society of Technologists, 1911.
²⁸) Stadelër, Metallurgie, 1908, p. 260.

This compound is exothermic and forms with considerable evolution of heat. Thus, at 600°—700°, when the reaction proceeds most easily and completely, after a current of ammonia had been passed into the tube of a Heraeus furnace, where, beneath the junction of the thermoelement, there was a boat with 0.5 g of metallic manganese in an atmosphere of hydrogen, the pyrometer sometimes rose by 10 degrees and even more. Such an evolution of heat was not observed with iron.

In the experiments it was found that often the gain in weight of the substance after saturation with nitrogen was greater than was required according to the analysis. Up to 600° the difference between the nitrogen content inferred from the increase in weight and the actual content was very slight, within the limits of the accuracy of the experiments; from 600° the gain increased sharply, the difference being about 6%; for example, according to analysis it should have been 14.4%, whereas the gain turned out to be 20%. Further on this difference steadily decreased and at 900—1000° disappeared completely.

It proved that neither oxidation nor absorption of hydrogen played any part here, since by analysis the combustion of manganous nitride revealed none of the latter in it; the cause was nitrogen, which, on dissolving in hydrochloric acid, is liberated in the form of gas.

In order to obtain the limiting quantity of nitrogen joining under these conditions, several experiments were made on the saturation of manganese with nitrogen at 600°, which were continued to constant weight, attained in the course of about \(1 \tfrac{1}{2}\) hours with a charge of about 1 gram.

For this purpose another sample of manganese was taken, which proved to be richer. According to analysis it contained Mn—98.83%, Fe—0.82, Si—0.25, Al—0.06%.

After saturation was completed, the boat was drawn out from the heated part of the tube inserted in the Heraeus furnace, by means of a wire passing through a mercury seal into the part of the tube projecting outward and cooled with snow. Here the boat was rapidly cooled in an uninterrupted current of ammonia.

The increase in weight of the substance in four experiments proved to be: 1) 20.07%, 2) 19.97%, 3) 20.11%, 4) 19.99%.

The substance from the two boats with the greatest increase in weight, i.e. the first and third, was subjected to analysis. Dissolution of the charge was carried out in an apparatus (Fig. 13) consisting of a tube in the form of a test tube with a lateral outlet tube \(a\) at the top. From above the tube was closed with a ground-in funnel \(b\) with a stopcock. The Tempe burette, \(c\), and also the capillary from it, which was connected to the outlet tube of the dissolving apparatus, were filled with water.

To determine the volume of air present in the apparatus, a funnel filled with water to the end of its tube was inserted into it. Then the air was transferred into the Tempe burette, while water from the funnel entered the apparatus.

In the investigation of nitrogenous manganese, a weighed portion of about 0.4 gram was placed in a test tube, which was covered with a funnel filled with hydrochloric acid up to the bottom of its tube. The acid was added to the substance little by little. For final dissolution the test tube was heated; the gases from it entered Hempel’s burette, and then the entire atmosphere of the apparatus was transferred into the burette, which, together with the capillary and the capillary tube of the burette, was filled with it. In this way, the burette contained the air of the apparatus and the gas formed during dissolution. In doing this, I did not want to neglect the amount of air that was in the capillary, and could determine very accurately the amount of gas formed during dissolution.

Fig. 13.

The volume of gas was measured after it had acquired room temperature. Then the gas was passed through an incandescent tube with palladium asbestos. If hydrogen had been present in the gases, a diminution of the gas volume would have occurred through the formation of water. However, no change in volume was observed.

Gas analysis showed that the gas liberated during the dissolution of manganese saturated in this way with nitrogen was pure nitrogen.

Analyses of the two indicated samples of nitrogenous manganese gave the following results: see Table 11.

Table 11.

Weighed portion	0.3178	0.4028
Cubic cm of nitrogen	15.7	19.9
Barometer height	752.4 at 20°	751.5 at 20°
Temperature	19°	20°
Elasticity of water vapor	16.3 mm Hg.	17.4 mm Hg.
Volume of gas at 0°C, 760 mm Hg.	14.2 cu. cm.	17.9 cu. cm.
Weight of nitrogen	0.01780 gr.	0.02244 gr.
Percent nitrogen	5.60	5.51

According to the analysis by titration, the nitrogen in these samples, which passed over upon dissolution in ammonia, was: 1) 14.37%, 3) 14.44%.

In the first sample the total nitrogen found by analysis was 19.97%, in the third 20.01%, which comes very close to the figures for the impurity. The manganese in it, by analysis, was 78.34%.

The second and fourth samples were not subjected to analysis, since I considered them insufficiently saturated with nitrogen.

The compound of manganese with nitrogen obtained corresponds to the formula MnN, which theoretically requires 20.301%.

Recalculating the quantity of nitrogen, the average of the two preceding determinations—19.99%—as a percentage with respect to pure manganese, we find 20.33% N. In this connection it must be borne in mind that the impurity of iron also must have combined with a certain quantity of nitrogen.

In contrast to iron, manganese reacts directly with nitrogen, but combines with a smaller quantity of it.

The saturation of manganese with dry nitrogen was carried out in the same apparatus and under the same conditions.

The results of the experiments are given in Table 12.

Table 12.

Action of dry nitrogen on manganese.

Temperature.	400°	500°	600°	700°	800°	900°	1000°	1100°	1200°	1300°
Nitrogen content in weight percent.	0.08	0.33	1.45	3.15	5.50	7.32	8.04	5.23	3.29	1.61
Nitrogen content in weight percent.		0.41	1.36	3.10	5.46	7.41	7.57	5.17	3.17	1.49
Nitrogen content in weight percent.			1.42	3.30	5.50	7.18	—	5.28	3.41	—
Average. . .	0.08	0.37	1.41	3.2	5.48	7.3	7.8	5.23	3.29	1.55

In order to compare the data of Tables 11 and 12, I depict them by a graphical diagram, Fig. 14.

The optimum for the compound of nitrogen with manganese under the action of ammonia will be about 600–800°, and under the action of dry nitrogen about 1000°. The first curve intersects the second at about 980° and then goes below the second, i.e., from this point the absorption of nitrogen from ammonia proceeds not as completely as in the second case. I see the reason for this in the dissociation of ammonia, which is very great here. According to the experiments of Ramsay and Young, at 800° 65% of it decomposes; according to my investigations, at 1100° ammonia dissociates almost completely. The free nitrogen formed here is strongly diluted with hydrogen, and, owing to this dilution, at the same rate of gas flow the absorption of nitrogen by manganese must be slowed down.

It is extremely interesting that the maximum absorption of nitrogen under the action of gaseous nitrogen is found at about \(1000^\circ\), and that here such a large amount of nitrogen combines with Mn—\(7.8\%\). At \(1300^\circ\) there is still an extremely large amount of it—\(1.55\%\).

Fig. 14.

I \( \mathrm{Mn} + \mathrm{NH}_3 \)
II \( \mathrm{Mn} + \mathrm{N}_2 \)

\(1300^\circ\)—this is a temperature very close to those temperatures which occur in the production of cast metal.

Ferronitride at \(1200^\circ\) decomposes completely; carbon, as my investigations show, has no noticeable influence on the stability of the compound of nitrogen with iron.

Thus, manganese, being a constant and not insignificant impurity in all ironworks products, more than likely also serves as one of the causes of the presence in them of those small doses of nitrogen which are always encountered there.

In favor of what has been said may also serve the fact that no reaction of gaseous nitrogen with iron was observed at any temperature, whereas manganese combines with it even at such a temperature as \(1300^\circ\) C.

On the basis of the investigations described, it may be supposed that nitrogen is capable of combining with iron containing manganese not only at the time when the iron is in the molten state, but also at lower temperatures. At those temperatures at which iron, when subjected to the action of ammonia, already ceases to combine with nitrogen, manganese combines with considerable quantities of it under the action of gaseous nitrogen.

For practice it is very important that manganese combines directly with nitrogen, in contrast to iron.

Alloys of iron with manganese should react with nitrogen, and the more readily, of course, the greater the content of manganese in them. Therefore there is always danger, when hardening, for example, manganese steels, of obtaining a brittle surface on them. It was nevertheless interesting to confirm experimentally the ability of manganese alloys to combine with gaseous nitrogen. For the investigation, a rich alloy of iron with manganese was taken, containing very few other impurities. It was prepared from Swedish nail iron, the analysis of which is given above, by melting in a kryptol electric furnace with metallic manganese (\(98.83\%\) Mn), and according to analysis contained \(22.1\%\) Mn. When subjected to the action of gaseous nitrogen for one hour in the form of fine filings,

this alloy at \(900^\circ\) combined with—\(1.79\%\) N and \(1.73\%\) N.
” ” ” \(1130^\circ\) ” \(0.099\%\) N.
” ” ” \(1300^\circ\) ” \(0.045\%\) N.

The quartz boat used for the experiments was lined with magnesite, so as not to lose the experiment in the event of the metal melting, since at this time manganese reduces silicon and pierces through the bottom of the boat. If the nitrogen content at \(1300^\circ\) is recalculated to manganese, assuming that the iron is entirely free of nitrogen, we obtain \(0.204\%\) N. If this alloy is heated further at the same temperature, the reaction of nitrogen absorption continues.

Thus, at \(1300^\circ\), over the course of two hours, filings of this alloy of the same fineness and under the same conditions combined with \(0.0676\%\) nitrogen, or, recalculating to manganese, we obtain \(0.301\%\). That is, in the continuation

during the second hour, more nitrogen joined the manganese by 0.097%. The escape of nitrogen from the surface takes place much more readily than its transfer into the interior of the piece of alloy.

Thus, when manganese iron-alloys are heated, there is a danger of obtaining a brittle surface on the article being annealed, owing to nitrogen. With prolonged annealing the article may acquire brittleness to an ever greater depth.

The phenomenon of the reaction of nitrogen with alloys of iron may be likened to overheating of iron articles, which can be remedied only by remelting with the addition of deoxidizing elements, for example manganese. Nitrogenous compounds formed in such cases can in all probability be completely freed from nitrogen by remelting only when we add to the molten metal substances whose nitrogen compound is not soluble in iron.

As regards the combination of nitrogen with molten iron containing manganese, the question here is also one of the degree of solubility of manganese nitride in iron.

To clarify this, I carried out several experiments on the dissolution in iron of manganese nitride prepared by the method described above. The experiments were arranged as follows: 15 g of iron reduced with hydrogen were taken and mixed with 35 g of manganese nitride containing 14.33% N by titration analysis and 19.96% N by weight gain, i.e., the compound corresponded to the formula MnN. This powder was melted in a Tammann furnace in a magnesia tube. The furnace atmosphere consisted chiefly of carbon monoxide and nitrogen. The temperature was brought to 1550°. The alloy was poured into a mold and subjected to analysis, which showed that it contained 0.0119% N.

This sample was again melted in a magnesia tube and after this contained 0.0113% nitrogen.

From this it is evident that manganese nitride must be soluble in iron, and manganese may therefore promote the transfer of nitrogen into the metal under conditions of works production.

In the hearth of blast furnaces, especially in the production of special pig irons, a higher temperature prevails than in steel-making production. It is of interest how much nitrogen there may be in ferromanganese, since, serving as additions in steel-making, they introduce it with them into the steel.

I investigated the nitrogen content of two samples of ferromanganese: one charcoal-smelted at a lower temperature than the other—coke-smelted, from one of the South Russian works. The results of the analyses are given in Table 13.

Table 13.

	Mn %	Si %	N %
Ural ferromanganese, smelted with charcoal	71,32	0,89	0,061
South Russian ferromanganese, smelted with coke	81,41	0,093	0,058

In such cast irons, nitrogen is found to be 20–30 times greater than in ordinary ones.

There is little silicon in them, especially in the coke sample.

Consequently, the increase in the amount of nitrogen, as compared with ordinary cast irons, which contain more than 2% manganese only in rare cases and usually less, must be attributed to the large amount of manganese. However, I do not here lose sight of the silicon content either, and I give its quantity in the ferromanganeses investigated. Silicon too may serve here as a cause of the nitrogen content. I turn to a detailed investigation of the influence of silicon in this respect on iron alloys in the next chapter.

Chapter VI

Silicon and Nitrogen

I

Survey of work on silicon nitride.

Like manganese, silicon combines with gaseous nitrogen. This property of silicon was discovered by Saint Clair Deville and Wöhler¹⁴ in 1859. Deville and Wöhler found that silicon is capable of entering directly into reaction with nitrogen by heating crystalline silicon with charcoal powder in a fireclay crucible placed in a coke furnace. The substance obtained, when fused with caustic potash, evolved ammonia. Subsequently, nitrogenous compounds of silicon were investigated by Schützenberger¹⁵, who derived for them the formulas SiN and Si₂N₄. His further investigations in collaboration with A. Colson¹⁶ showed that the compound they had in hand contained carbon and corresponded to the formula Si₂C₂N.

By the method described, it was not possible to obtain pure silicon nitride without carbon content.

In pure form, silicon nitride is obtained by the method proposed by Geuther¹⁷, by heating silicon in a stream of nitrogen in a porcelain tube at a white heat.

Schützenberger and Colson¹⁸ obtained, in this way, a compound corresponding to the formula Si₂N₃.

According to the investigation of Vigouroux¹⁹, the reaction between silicon and nitrogen takes place far above 1000°.

Besides these compounds, various authors found several more complex silicon compounds, into whose molecules H, Cl, and S also entered.

The most recent work concerning the compound of silicon with nitrogen is the investigation of L. Weiss and T. Engelhardt²⁰.

Nitrogenous compounds of silicon were obtained by passing gaseous nitrogen through an incandescent tube with a boat placed in it, containing the crystalline silicon, ground beforehand into a fine powder. A Heraeus furnace served for heating the tube; the temperature was measured with a Le Chatelier pyrometer. Saturation with nitrogen was carried out at \(t^\circ = 1300—1400^\circ\). At a lower temperature the reaction proceeded weakly; at \(1220^\circ\) it was not observed at all. The silicon saturated with nitrogen was purified from unreacted silicon and, depending on the method of purification, contained larger or smaller quantities of nitrogen. If the crude reaction product was heated with a caustic-potash solution until gas evolution ceased, and then with dilute hydrofluoric acid, a substance was obtained which, by analysis, corresponded to the formula \(Si_2N_3\). If, however, the crude product was treated with a mixture of hydrofluoric and nitric acids, a compound was obtained corresponding to the formula \(SiN\). By saturation with nitrogen for seven hours, when the gain, determined every hour, was already quite insignificant, they obtained a reaction product corresponding to the formula \(Si_3N_4\). They analyzed the nitrogenous silicon by heating its powder in a Heraeus furnace with a mixture of equal parts of lead chromate, lead oxide, and lead peroxide.

Since silicon reacts with gaseous nitrogen, and moreover at such a high temperature as \(1400^\circ\), one may suspect that it is a not unimportant cause of the nitrogen content in iron, steel, and cast iron. This temperature is not far from those which prevail in factory furnaces in the production of iron. Whether the temperature \(1400^\circ\) is an optimum for silicon compounds with nitrogen, or whether at higher temperatures its combination with nitrogen proceeds still more readily, and at what temperature its dissociation begins, is still an open question. Since operating at such high temperatures and obtaining curves like those obtained with respect to iron and manganese is extremely difficult and scarcely accessible under laboratory working conditions, I decided to pose the investigation of the possibility of the influence on the combination of nitrogen with iron containing silicon in another way.

Of course, silicon, being in solid solution, and especially in liquid solution with iron, must react with gaseous nitrogen, as we have seen with respect to manganese; but here the question is whether nitrogenous silicon is soluble in iron, and if soluble, to what extent. If it is not soluble in iron at all, then, even if a compound of silicon with nitrogen were formed during factory production, the nitrogenous silicon would float up into the slag; and if it did remain in the metal, the harm to its mechanical properties from this foreign inclusion would be negligible.

may. It is another matter if it can be present in solution and remain in solid solution with iron; then its influence on the mechanical properties of the latter assumes an entirely different aspect. To clarify these questions, I decided to prepare a nitrogen-rich compound of silicon and alloy it with pure iron. If nitrogenous silicon proves soluble in iron, then it may be possible, by analysis, to find the presence of nitrogen in the alloy obtained and to determine the degree of solubility of nitrogenous silicon.

II

Preparation of nitrogenous silicon.

The starting material for preparing nitrogenous silicon was crystalline silicon (Silicium metallicum cryst. Merck), which was first subjected to analysis.

According to Wiess and Engelhardt, it is most convenient to analyze silicon by fusing it with a mixture of KNaCO₃ and KNO₃ in a platinum crucible. The fusion must be carried out very cautiously; otherwise the crucible may be badly damaged and even melted through. The reaction proceeds very violently, especially with a large amount of saltpeter. Particles of silicon, in contact with the platinum, form an alloy with it; and to protect the walls of the crucible better, one should first melt in it about two grams of KNaCO₃ and, by rotating it in the flame of a Bunsen burner, form on the walls of the crucible a crust of KNaCO₃. Into the crucible thus prepared was poured a mixture of 0.5 g of Si, ground to a fine powder, with well-dried powders of 8 g KNaCO₃ and 0.5 g KNO₃. The crucible was first heated cautiously on a Bunsen burner until gas evolution had nearly ceased, and then was heated strongly on a large Meker burner until the melt became perfectly transparent. The melt was dissolved in dilute hydrochloric acid. Silicon was determined in the usual way. The hydrochloric-acid filtrate from SiO₂, in order to remove platinum, was saturated while warm with hydrogen sulfide, which was then expelled by boiling. Only negligible traces of platinum were found. Fe, Al, Mn, and Cu were determined in the filtrate. The result of the analysis was as follows: Si—97.12%, Fe—0.41%, Al—0.06%, Cu—0.03%, SiO₂—0.63%.

The impurities Fe, Al, Mn, Cu, and also SiO₂, which is the chief one here, should not be harmful for the subsequent experiments.

For the preparation of nitrogenous silicon, I decided to use not a Heraeus furnace, which, although it can give 1400°, can do so only for a very short time and then with considerable risk of damaging it, but a Tammann furnace, which was set horizontally.

Finely powdered silicon in an amount of 4–5 g was placed in a quartz tube and, by means of a glass rod, was spread out

over the tubes in an even layer, to a length of about 60 mm. In this space the fall in temperature from the middle of the furnace to the edges was slight (40–50°), which was determined by moving the junction of the thermoelement wires through a pyrometric quartz tube inserted into the middle furnace tube. The nitrogen was taken from a cylinder and was first passed through sulfuric acid, through a red-hot glass tube with red-copper turnings, and then also through phosphoric anhydride. The nitrogen was very pure. After 10–11 hours of operation the copper turnings darkened from one end by no more than 5 cm. After the air had thus been expelled from the entire apparatus by nitrogen, with a continuous current of it, the furnace was heated to 1500°, which required about 20 minutes. The temperature of 1500° was maintained for some time and then, for another 10 minutes, was raised to 1550°. In order that the rubber stoppers inserted into the ends of the quartz tube, protruding from the furnace on both sides by 200 mm, should not suffer, three partitions of asbestos board were placed inside the quartz tube at both ends, 1 cm apart. In addition, both ends of the quartz tube, wrapped with filter paper, were cooled by a continuous stream of ice water from two flasks by means of siphons. After cooling in a stream of nitrogen, the substance was removed from the quartz tube. It was a light-ash-colored mass, somewhat compacted but easily separating from the walls. The quartz tube could serve only once, since it crystallized strongly into tridymite and became very brittle.

Thus, in eight experiments, 53.3 grams of substance were prepared. To purify it from unreacted Si and from SiO₂, the presence of which was observed each time, despite the most careful removal of air from the apparatus, the substance was heated in a platinum dish on a water bath with a 20% KOH solution until the evolution of gas bubbles ceased, which occurred after about five hours. The residue was filtered off until the alkaline reaction ceased. From 53.3 g there remained 25.81 g, i.e., 48.3%. This powder, in a dry state, in the amount of 25.78 g, was placed in a large platinum dish and left to stand for twenty-four hours under a cover of hydrofluoric acid. Then the hydrofluoric acid was evaporated on a water bath, and the residue was ignited on an asbestos net with a Bunsen burner to constant weight. There were obtained 17.2 g of residue, or 32.2% relative to the crude product. The analysis of the substance obtained for nitrogen was carried out according to the method developed by Weiss and Engelhardt, by heating and fusing the almost white powder of silicon nitride obtained by purification with a mixture of equal parts of lead chromate, lead peroxide, and lead oxide.

A weighed portion of about 0.1 g of silicon nitride with 2.5 g of the indicated mixture was placed in a porcelain boat, which was inserted into a quartz-

tube located in the Heraeus furnace. Above the middle of the boat was the end of the pyrometric tube with the junction of the thermoelement. Behind the boat was placed a long plug of a mixture of red copper for freeing the gases from oxygen, released from the heating of the mixture of lead compounds. In order that the swelling mass should not touch the pyrometric tube and should not spatter, a thin porcelain plate was laid on the surface of the powder, fitting almost tightly inside the boat.

The atmosphere of the apparatus was displaced with carbon dioxide. The gases released upon heating the substance were directed into a gas burette filled to the top with a 25% solution of KOH. The furnace was heated slowly to about \(800^\circ\), and when the evolution of gases subsided, the temperature was raised to \(1000^\circ\); at this point all evolution of gases ceased. The gas from this burette was transferred into a Hempel burette, the capillary tubes of the burettes and the connecting capillary being filled with water, so that the gas from the first burette was transferred into the Hempel burette entirely and completely without any addition of air, as would have been the case if the atmosphere of the capillaries had been neglected. The gas was analyzed by the usual methods.

The following results were obtained: (see Table 14).

Table 14.

Determination of nitrogen.

Sample, g	0.1136	0.1310	0.0935
Amount of gases evolved	36.60 cubic cm	40.9	37.3
CO	1.2	1.8	7.1
\(N_2\)	35.40	39.1	30.25
Air pressure, mm Hg	748.4 at \(20^\circ\)	748.2	741.5
Temperature	\(19.5^\circ\)	\(19.5^\circ\)	\(19^\circ\)
Water-vapor pressure, mm Hg	16.85	16.85	16.3
Gas volume at \(0^\circ\)C and 760 mm Hg	31.81 cubic cm	35.13 cubic cm	26.98 cubic cm
Weight percent \(N\)	35.11%	34.50%	35.72%

Average . . . \(35.11\%\,N\).

Analysis for Si

Sample	0.2881;	\(SiO_2\)—0.3544;	\(Si\)—0.1645;	% \(Si\)—58.09
Sample	0.1763.	\(SiO_2\)—0.2164	\(Si\)—0.1004	56.94%

Average . . . 57.02%.

Thus, by analysis, the purified silicon nitride contained:

\[ \begin{aligned} \mathrm{Si} &— 57.02\%,\\ \mathrm{N} &— 35.11\\ \mathrm{O} &— 7.87\\ \hline &100.00 \end{aligned} \]

If it is assumed that the oxygen is combined with Si in the form of \(\mathrm{SiO_2}\), it turns out that in the given substance there will be \(\mathrm{N}\)—35.11%, \(\mathrm{Si}\)—50.14%, \(\mathrm{SiO_2}\)—14.75%.

The ratio between Si and N will be as \(1 : 1.428\), or \(2 : 2.85\). For Weiss and Engelhardt this ratio came out as \(2 : 2.92\), \(2 : 2.81\), and \(2 : 3.01\).

Thus, the substance obtained by me comes close to the formula \(\mathrm{Si_2N_3}\).

The nitrogenous compound of silicon is very stable. Reagents at ordinary temperature act upon it only weakly. Dilute acids, with the exception of HF, do not decompose it. Boiling with concentrated \(\mathrm{H_2SO_4}\) decomposes it only partly, with formation of ammonium sulfate.

Thus, under the influence of six hours’ boiling with fuming \(\mathrm{H_2SO_4}\), according to the indications of Weiss and Engelhardt, 44.8% N was liberated from SiN, and 45.8% N from \(\mathrm{Si_3N_4}\). Concentrated HF decomposes \(\mathrm{Si_2N_3}\) on threefold evaporation, liberating 26.3% N; SiN—on twofold evaporation, liberating 25.5% N. The action of concentrated HF is much stronger after preliminary treatment of the nitrogenous silicon compounds with an alkali solution.

In view of such weak and incomplete action of reagents, nitrogen in this substance cannot be determined by Kjeldahl’s method.

On fusing nitrogenous silicon compounds with KOH, their complete decomposition is obtained, with formation of ammonia.

III.

Solubility of silicon nitride in iron.

The obtained preparation of silicon nitride was fused with iron reduced by hydrogen; the mixture of the two was taken in different ratios in order to determine the limit of solubility of silicon nitride in molten iron.

A weighed portion of iron reduced by hydrogen, in the amount of 15 g, was mixed with 1 g of silicon nitride powder, which contained, according to analysis, 0.3511 g of nitrogen. With complete dissolution of the silicon nitride in the alloy, there would be about 2.2% N. The mixture was placed in a magnesite tube for a Tammann furnace. The upper part of the tube was closed with a magnesite stopper. Oxidation during fusion of iron in the Tammann furnace, if it occurs at all, is very ...

weakly. Thus, 25 g of this same sample of fine iron, melted in the same test tube and poured into a mold, gave an ingot which forged perfectly, without producing red-shortness. Consequently, if in this case ferrous oxide had formed as well, capable of acting oxidizingly upon silicon nitride, the alloy would have lost very little nitrogen, and with complete solubility of silicon-nitride scale, the amount of nitrogen in the iron should not have differed greatly from 2.2%. A fusible magnesia tube was placed in Tammann’s furnace and rapidly heated to about 1700°. During melting of the mixture, abundant evolution of gas was observed. After the alloy had quieted down a few minutes later, it was poured into a cast-iron mold.

The button weighed 14.1 g (part of the alloy remained on the walls of the magnesite tube) and had a shiny surface; on top there was a slight shrinkage cavity, beneath which, after sawing the ingot, there proved to be a void about 3 mm deep.

In order to determine whether the ingot contained impurities of extraneous substance not forming a solution with iron, a polished section was prepared from it, which under the microscope, in the unetched state, gave a clean mirror without any foreign inclusions, despite a magnification of 800 times. In the etched state, with picric acid, the polished section likewise revealed no foreign inclusions; only the boundaries between the crystals were clearly visible.

The ingot proved very hard and yielded to the file with difficulty. On the anvil it was broken into small pieces, which had a fibrous appearance in fracture. The resulting sample was subjected to analysis. The scale dissolved in hydrochloric acid, leaving much insoluble residue. The solution was filtered off and analyzed for nitrogen in the usual manner. The insoluble residue was washed on the filter until the presence of hydrochloric acid disappeared and was subjected to analysis for nitrogen, which was carried out by the method used for silicon nitride. In addition, in a separate weighed portion the silicon content of this alloy was determined. By titration analysis it was found to contain 0.0186% N.

In addition, another alloy was prepared from 10 g of iron and 0.5 g of silicon nitride, in which there should have been 1.75% N. By titration analysis, 0.02% N was determined in it. Then an alloy was made from 10 g of iron and 1 g of silicon nitride; it should have contained 3.51% N. By titration analysis, 0.0202% N was found there.

Since silicon nitride is extremely difficult to subject to the action of acids, so that, for example, strong sulfuric acid, when boiled for several hours, decomposes it only by half, the nitrogen split off passing into ammonia, then ne-

The question remains rather open as to how silicon nitride behaves toward acids when it is in solid solution with iron.

If it is also poorly attacked by acid, as it is in the free state, then when iron, steel, or cast iron is dissolved in hydrochloric or weak sulfuric acid, as is usually done in analyses for nitrogen, we may either obtain no decomposition of it at all with formation of ammonia, or it may be decomposed only incompletely. To clarify this, the insoluble residue obtained in determining nitrogen by titration in alloys prepared from iron and silicon nitride was filtered off, and the nitrogen in it was determined in the same way as in silicon nitride, i.e. by heating with a mixture of lead oxides to \(1000^\circ\). It then turned out that from the weighed portion of the insoluble residue (first alloy), about 0.5 g, no more than 0.1 cu. cm of nitrogen was obtained. Thus this residue would contain about \(0.042\%\) N, or, in the whole alloy, \(0.000835\%\) N. In analyzing ordinary grades of soft iron and steel, which contain silicon from hundredths of a percent to \(0.2\)—\(0.3\%\), and rarely somewhat more, assuming that the insoluble residue will have the same amount of nitrogen, we shall make an error, and one all the greater the richer the iron is in silicon.

Under these conditions, in iron with a content of \(0.3\%\) Si, about \(0.0004\%\) N will remain in the insoluble residue and, consequently, escape determination; at \(0.1\%\) Si—about \(0.000013\%\) N.

In the specimens examined, obtained by alloying silicon nitride with iron, there was very much nitrogen in comparison with commercial grades. It may be that silicon in an alloy with iron, especially in the presence of manganese, with which it must share the nitrogen, does not present such an obstacle to the determination of nitrogen in iron.

To investigate this, two steel specimens were taken, of which the first contained Si—\(0.14\%\), Mn—\(0.47\%\), the second—Si—\(0.55\%\), Mn—\(0.84\%\). Both specimens were dissolved in strong hydrochloric acid in amounts of 50 g each. The first steel gave 0.142 g of insoluble residue, of which 0.115 g was taken for nitrogen analysis; on dissolving the second, 0.374 g was obtained, and 0.351 g was taken for nitrogen analysis. Analysis of the first residue did not give positive results, i.e. the amount of nitrogen, if it was present there, could not be determined owing to its very small volume. The second precipitate gave about 0.05 cu. cm of nitrogen.

Consequently, the insoluble residue of the second specimen contained about \(0.00175\%\) nitrogen. In relation to the steel this will be \(0.0000125\%\) N. By titration, \(0.00417\%\) N was determined here.

This, approximately, will be the error if the nitrogen remaining in the insoluble residue is not taken into account.

With an increase in the silicon content, this error increases.

A similar investigation was performed on ferrosilicon containing 14.11% Si. A 5-g weighed portion of it, in the form of a fine powder, dissolved in strong hydrochloric acid over the course of several days. The insoluble residue, of whitish-ash color, was obtained in the amount of 2.0528 g. According to theory, if all the silicon passes into SiO₂, the amount of such residue should be 1.51 g. Consequently, about 1.44 g of the weighed portion did not decompose.

Analysis by titration of the hydrochloric-acid solution showed 0.0136% N. The insoluble residue, analyzed in the same manner as silicon nitride, contained 0.051% N. Thus, the total nitrogen in the ferrosilicon proved to be 0.0646%.

Therefore, in the analysis of iron and steel, if one disregards the amount of nitrogen that does not pass into ammonia and remains chemically combined with the insoluble residue, the error we make is insignificant; analyses for nitrogen of cast irons, especially those containing much silicon, however, must be considered incorrect unless the circumstance indicated above is taken into account.

The enormous nitrogen content in ferrosilicon compels one to treat its addition to steel and iron with caution, since the nitrogen in this case may pass into solid solution with the iron and impart brittleness to it.

Chapter VII

Metallography of Nitrogenous Iron

Pure iron can contain nitrogen to a maximum of 11.12%. In experiments on saturating filings of soft iron with nitrogen, it proved that the reaction proceeds best at about 600°, whereas very finely divided iron, reduced by hydrogen, gives the optimum reaction at 450°. In analyzing the reactions of both samples of iron, the supposition was expressed that this difference depends on the solubility of nitrogenous iron in iron; that at 600° nitrogen penetrates into the depth of the piece of iron. This phenomenon may be likened to the penetration of carbon into the body of an iron object during its cementation. If this is indeed so, then iron should form with nitrogenous iron a solid solution.

What the degree of solubility of the nitrogenous compound of iron with pure iron is, is unknown. If the solubility is unlimited, then under the microscope we shall see a clear field of solid solution; if, however, the solubility is limited by some boundary, then among the crystals of ferrite, which may contain some quantity of nitrogen, we shall see crystals of nitrogenous iron of a definite composition, corresponding to a known formula. If the solubility of nitrogenous iron in iron is small, then at some content of it we can determine its presence under the microscope. The crystals of nitrogenous iron, when the polished section is etched, should react differently than the crystals of iron and consequently stand out clearly among the latter. It may be that also in commercial grades of iron one can observe nitrogenous inclusions under the microscope and, if not in all samples, then at least in those where its amount is greatest, for example in metal of Bessemer and Thomas production, and in malleable steel.

To clarify these questions, I undertook a series of investigations.

For the work, iron containing the smallest possible amount of impurities was taken, in order to clarify the phenomenon of solubility of nitrogenous iron in pure iron²¹. For this purpose I used the same

a sample of Swedish nail iron, which was taken in order to determine the course of the reaction of nitrogen with iron at different temperatures.

The iron contained: C—0.1%, Si—0.014%, Mn—0.14%, P—0.08%, S—0.012%.

Pieces of this iron were placed in a quartz tube inserted into a Heraeus electric furnace.

The end of the quartz insulating tube of the pyrometer was inserted so that it was located at the end of the space occupied by the material under investigation.

By moving the wires of the thermoelement inside the insulating tube to the opposite end of this space, it was possible to verify that the temperature during the experiment was everywhere the same there.

The air from the tube was displaced by ammonia, and the furnace was heated to 450°C, i.e., to the optimum temperature for the formation of ferronitride.

After ammonia had been passed through for 5 hours, the electric current was switched off and the furnace cooled slowly.

The total gain in weight proved to be 0.058%. The color of the surface of the iron was white matte.

When the surface of specimens thus obtained was filed with a steel hacksaw, it was very hard, but as soon as a scratch appeared, the specimen was cut very easily.

A polished section from the cut was etched with picric acid, but its appearance under the microscope proved to be exactly the same as that of iron not subjected to nitrogen saturation; the ferronitride crust was not visible; evidently, being very thin and brittle, it broke off and did not enter the polished section.

Under the same conditions, but for 8 hours, iron was subjected to the action of NH₃ at 600°C.

The gain in weight proved to be 1.51%. The external appearance was the same: light matte. When struck with a hammer, scales somewhat less than 1/2 m/m thick easily flew off.

From these pieces of iron polished sections were prepared, which were etched with picric acid or with a 3% solution of nitric acid in amyl alcohol. A microphotograph is presented in Fig. 15.

At the top of the photograph a part of the very brittle light crust of nitrogenized iron is visible. Beneath this brittle crust, which readily flakes off when struck with a hammer, there is a dark, more easily etched layer, from which inward into the piece of iron there extends a mass of long crystals of the same shade as it. Toward the middle of the piece the number of crystals decreases, and the very center was completely free of them: the nitrogen had not yet succeeded in penetrating there. Hence it is clear that at this temperature nitrogen penetrates into the interior of the piece of iron.

The different behavior toward etching reagents indicates a different chemical composition of the crust and of the dark crystals.

Further, it was of interest to determine the composition of both. For this purpose, from soft iron a plate 12 m/m thick was prepared on a planing machine; it was subjected to the action of ammonia at 600°C for 25 hours.

After this, on the planing machine, an extremely brittle crust was removed from it, which separated very easily and cleanly.

Fig. 15.

×300

Analysis showed it to contain 11.11% N, i.e., it represented pure ferronitride Fe₂N.

The hard layer lying beneath it was also removed. This layer was hard, but not so brittle. The thickness of this layer, measured with the aid of a Leitz microscope, the stage of which has the necessary device for this purpose, was about 0.8 m/m. Only the upper part was planed off, so as not to include iron from the uneven inner surface, as is visible in the photograph (Fig. 15).

Analysis of the shavings of this layer showed a content of 2.02% N. Analyses of filings of such a layer, sawn from pieces of soft iron treated with ammonia at 600° throughout, showed less nitrogen (e.g., 1.37, 1.72%), since it was very difficult, after removing the brittle crust with a hammer, to take with the file only this layer without including iron, which was noticeable every time.

The compound with 2.02% N comes very close to the formula Fe₁₂N, which requires 2.04%. The crystals should also be of the same composition,

extending into the interior of the piece, since they are etched equally well by the dark layer and, in many specimens, appear in the form of offshoots of such a layer into the interior of the specimen. This is especially clearly seen in Fig. 16. This specimen was saturated with nitrogen at 700°. The penetration of nitrogen inward proceeds more rapidly and more readily than at 600°. Here the crystals of nitrogenous iron are thicker, there are more of them, and, as is evident, becoming increasingly concentrated toward the outer part (top of the micrograph) of the section, they pass into a continuous dark field.

Fig. 16.

×300

The specimens described, after saturation with nitrogen, were cooled slowly together with the furnace.

Obviously, in this case a stable system is attained much more easily than in carbon-bearing iron, where the decomposition of pearlite into punctate material and then into the ferrite–cementite system occurs with difficulty, when the specimen is subjected to a series of repeated heatings with subsequent coolings at the corresponding temperatures. In order to ascertain the effect of the rate of cooling and to investigate the possibility of obtaining a eutectic between iron and its nitride, heated specimens of nitrogen-containing iron were subjected to more rapid cooling than those described. A specimen of Swedish iron was exposed to ammonia at 600°. A section from it gave a picture similar to Fig. 15. It was then placed back into the furnace and heated in an atmosphere of nitrogen at 800° for one hour. After this time had elapsed, it was quickly pulled out, by means of a wire attached to it, into the part of the quartz tube projecting from the furnace, and there

immediately cooled. In the photograph, Fig. 17, traces are visible of the former crust of nitrogenous iron, which had strongly decomposed. Inward from this place (the left side of the photograph) many dark constituents are visible among the ferrite crystals, the boundaries between which are clearly expressed. These dark constituents appear striated with fine strokes and are very similar to the pearlite of carbonaceous iron. There is no doubt that this is nothing other than a eutectic. This specimen was again heated to \(700^\circ\) and cooled together with the furnace. Its section, etched with picric acid, gave us a different picture: dark inclusions formed between the crystals, in which no striation was visible even at a magnification of 1000 times.

Fig. 17.

\(\times 300\)

In very many specimens of iron subjected to the action of ammonia beneath a layer, where crystals of nitrogenous iron were visible, a mass of little strokes was found in the ferrite crystals; these were scattered at various angles and appeared like segments of straight lines; see Fig. 18. These little strokes indicate cracks that arose as a result of a change in volume owing to the nitrogen entering into combination with it.

The recording of cooling curves showed, in specimens of nitrogenous iron containing carbon, a certain lowering of the pearlitic point.

The curves were recorded by means of Kurnakov’s self-registering apparatus.

Curves recorded during the heating and cooling of nitrogenous iron not containing carbon gave a strong arrest at \(610^\circ\). For this

there was taken ferronitride prepared from finely divided iron, reduced with hydrogen, with 11.11% N. From it briquettes were made by pressing in a hydraulic press at a pressure of 200 atm. per 1 sq. cm. These briquettes, weighing about 15 g, with a thermocouple inserted inside, were heated in a Heraeus furnace to 1000°. Apart from a halt at 610°, no other halt was observed up to 1000°. I consider this halt to depend on the decomposition of ferronitride. On cooling, two barely perceptible halts were obtained at 880 and 780°, which may depend on the transition of iron from \(\gamma\) to \(\beta\) and then to \(\alpha\). The briquettes sintered in the process. Sections made from them under the microscope appeared as pure ferrite with slag inclusions. According to analysis, 0.02% nitrogen remained there. Heating to lower temperatures, with subsequent cooling, gave no indications of the transition points of nitrogenous iron into solution and of its separation from solution. This must be due to the small thermal effect of these transitions and to the easy decomposability of nitrogenous iron.

Fig. 18.

\(\times 300\)

However, if the structure changes with the rate of cooling, then, in samples heated above 600°, the components of nitrogenous iron dissolve in the iron, and hence arises the assumption that, by means of rapid cooling, it is possible to obtain an undecomposed solid solution, i.e. to harden iron containing nitrogen.

For hardening, a specimen saturated with nitrogen at 600° was taken, which was then heated in an atmosphere of nitrogen in a Heraeus furnace at 750°. It was then quickly thrown from the furnace into ice water.

The hardened specimen thus proved to be so hard that it could be sawn with a steel hacksaw only with very great difficulty. For microscopic examination it had to be deeply cut with an emery disk of triangular section and then broken; the surface was finished on an emery wheel. The section was etched with picric acid. Examination under the microscope showed a structure similar to martensite (Fig. 19).

Fig. 19.

×500

Consequently, nitrogenous iron can also be hardened, just like carbonaceous iron.

The iron specimen saturated with nitrogen and hardened in this way was extremely brittle; it broke easily, without showing any bending. The fracture was bright and finely granular. Testing of many hardened specimens showed that their brittleness is incomparably greater than that of unhardened ones.

It was then of interest to investigate what the iron crystals in the unhardened specimen, lying among the crystals of nitrogenous iron, represent: whether they are pure iron or contain a certain amount of nitrogen.

For the experiments an iron plate 10 mm thick was taken, which was subjected to the action of ammonia for 5 hours at 750°.

Previous investigations showed that under these conditions nitrogen penetrates into the interior of the specimen in the form of crystals of nitrogenous iron to a depth of about 2–3 mm. The plate was cut perpendicular

length and examined under the microscope. Then the hard crust was removed from it, the layer of dark crystals was filed off, and the entire place occupied by crystals of nitrogenous iron. Here microscopic examination revealed no signs of inclusions of this component. Since these crystals are in the immediate vicinity of the removed layers that had nitrogenous inclusions, nitrogen must be contained in them, if only iron is capable of retaining a certain quantity of nitrogen in solid solution, without precipitating nitrides during slow cooling. From this place, with a fine file, a layer about \(1/2\) mm thick was removed and subjected to analysis for nitrogen. Analysis of these filings showed the presence of nitrogen. In several experiments from \(0.05\) to \(0.1\%\) N was found. This circumstance shows that, even if the first smaller figure, i.e. \(0.05\%\) N, is accepted as more or less correct, then in commercial grades of iron containing less than \(0.02\%\) N it should not be visible under the microscope.

Soon after the publication of this work of mine there appeared the investigations of S. H. Andrew²², in which he chiefly attempts to determine whether the removal of gases from iron during repeated heating and cooling in a vacuum has any influence on the position of the critical points of iron. In doing so he drew attention to the removal of nitrogen imparted to the material by preliminary heating in ammonia. He found such arrests during the first heatings which, upon repeated heatings, disappeared, and the usual transformation points for the given grade of iron remained.

On a section of iron, after treatment with ammonia, polygonal crystals were observed against the background of the main mass. At higher magnification three structural elements were found: polygonal crystals, a dark constituent, which he considers to be Neumann lines, and a pale-yellow component. It is clear that the polygonal crystals represent a solid solution of ferronitride in iron, while the yellow component is the structural constituent which, according to my investigations, contains \(2.02\%\) N and to which I assign the formula \(\mathrm{Fe}_{12}\mathrm{N}\), requiring \(2.04\%\) N.

On the basis of the observed changes in the position of the critical points, their number, and the mechanical properties of the metal under the influence of nitrogen, the author concludes that nitrogen is chemically combined with iron, but is not occluded.

This conclusion contains nothing new, since the existence of nitrogenous iron as a chemical compound had long since been proved.

A year after this, another new work by S. Andrew appeared²³, in which he saturated specimens of iron with nitrogen by means of gaseous nitrogen under a pressure of 20 atm. at the melting temperature. He obtained iron with a content of up to \(0.3\%\) N. The material used was pure Swedish iron with \(0.16\%\) C. The content of other impurities, e.g. Si, Mn, is not indicated; meanwhile these latter impurities, it seems to me, must have a substantial influence.

The nitrogen-saturated iron specimens were subjected to thermal investigation. It proved that iron, rapidly heated in vacuum to \(1000^\circ\), gave no arrests on cooling. During this heating in vacuum, a certain quantity of gases was evolved from it, consisting of carbon monoxide, hydrogen, insignificant quantities of carbonic acid, and only traces of nitrogen. After annealing in vacuum at \(1000^\circ\) for one hour, \(3.64\) cubic cm of nitrogen was evolved, and then the cooling curve distinctly marked two arrests at \(770\) and \(855^\circ\). Further annealing at \(1000^\circ\), accompanied by a new evolution of nitrogen, contributed to the strengthening of the arrests, to an increase in their temperature, and to the appearance of new arrests. Thus, the upper arrest (\(\mathrm{Ar}_1\)) occurred, after six hours’ annealing of the iron in vacuum, at \(895^\circ\). The lower one showed not entirely clear fluctuations. Similar results were also obtained with steel of \(0.6\%\) C, containing \(0.25\%\) N.

Melting iron in an atmosphere of hydrogen at 100–200 atm. pressure, with subsequent taking of cooling curves, had no influence on the position of the arrests. An atmosphere of CO and \(CO_2\) during the melting of iron likewise did not exert such an action.

Andrew’s observation concerning the property of nitrogen to lower and destroy arrest points seems to me to be of great interest.

It may be that, in scientific investigations where the authors used an atmosphere of “inert gas,” nitrogen, for melting and prolonged annealing of specimens, they more than once obtained negative results in their work, without having the possibility of obtaining cooling curves of the alloy with its characteristic arrest points.

Microscopic investigations of various iron specimens saturated with nitrogen in this way revealed, among the ferrite, the presence of streaks and dots similar to cementite; in the author’s opinion, these constitute a solid solution of ferronitride in iron. No proof of this supposition is given in the work.

Steel with \(0.6\%\) C, annealed in hydrogen and thereby carburized, had among the ferrite crystals light-yellow components, considered by the author to be crystals of a solid solution of hydrogen

in iron, possibly still containing carbon. These components, however, were not subjected to closer investigation.

On the basis of the fact that removal of the last portions of nitrogen from the metal by heating was accomplished with the greatest difficulty, the conclusion is drawn that this refutes the opinion that nitrogenous iron is readily decomposed at high temperature. Here, however, one must speak not of the decomposition of iron, but of the decomposition of silicon or manganese nitride, present in solid solution with iron, among which the nitrogen was distributed in the corresponding proportion.

Nitrogenous iron, as has already been shown, decomposes readily. I observed that silicon nitride, which is obtained at \(1500^\circ\), i.e. near the melting temperature of iron, when immersed in molten iron, liberates nitrogen vigorously. Iron acts as a decomposing agent on silicon nitride. In my experiments no complete decomposition occurred; part of the silicon nitride remained in solution with the iron. It may be that melting silicon nitride with iron in vacuum would give different results. Andrew’s experiments show that prolonged heating in vacuum entails the removal of all nitrogen. Consequently, the nitrides of Si and Mn decompose completely under these conditions, and iron appears to be a reverse catalyst.

Chapter VIII

The Influence of Aluminum on the Nitrogen Content in Iron

The use of aluminum in steelmaking has hitherto been limited only to adding it to liquid steel for its deoxidation. At the same time it has already been observed that aluminum promotes the production of denser castings. One of the reasons for the reduction in the evolution of gases is the combination of aluminum with nitrogen dissolved in the steel and released during cooling. The amount of nitrogen, according to data in the literature, reaches up to 14 and more percent by volume in comparison with the other gases evolved during the solidification of steel. According to my investigations it turns out that aluminum, reacting with gaseous nitrogen, gives a very stable compound. The reaction begins at about \(400^\circ\) and proceeds up to the investigated temperature of \(1350^\circ\) with ever-increasing speed. In this process two compounds are formed—a light-gray powder and a black one, like coal. This black powder, according to the analysis I made, represents the compound \(AlN\). This compound does not dissociate even at \(1650^\circ\), which was tested by heating it in a Tammann furnace. It oxidizes with difficulty, fuses with difficulty in a crucible with \(KNaCO_3 + KNO_3\), is insoluble in \(HCl\) and \(H_2SO_4\), and is decomposed by the mixture of lead oxides used for the analysis of silicon nitride, even when heated to \(1000^\circ\), only in part. It is clear that aluminum, dissolving in molten iron, must combine with the nitrogen dissolved there. The question is only whether the aluminum nitride thus obtained is soluble in iron and, if soluble, to what extent.

To clarify this question, aluminum saturated with nitrogen at \(1100^\circ\), whereby up to 85% of \(AlN\) insoluble in acids is obtained, was heated in a Tammann furnace in a magnesite tube with pure iron reduced by hydrogen. The first time, 20 g of iron and 4 g of aluminum nitride were taken. The temperature was raised to \(1700^\circ\). The contents of the tube were shaken out onto an iron sheet; they consisted of black aluminum-nitride powder and several beads of metal, in which analysis showed \(0.0165\%\ N\). A second melt was made from 16 g of iron and 0.6 g of aluminum nitride; it was found to contain \(0.028\%\ N\). The temperature during melting was \(\sim 1600^\circ\). In the tube there was also found a considerable quantity

a light black powder of aluminum nitride. The latter alloy was melted a second time. Analysis showed it to contain 0.017% N. Microscopic examination did not reveal any foreign inclusions in the ingot. Consequently, the aluminum nitride was present there in solid solution. On the basis of these experiments one may conclude that aluminum, when added to liquid steel, will combine with the nitrogen dissolved there and, converting it into a solid solution in the form of aluminum nitride, will prevent it from being liberated.

If the gas evolved from steel is recalculated as percentages relative to the weight of the ingots, it turns out that aluminum can retain therein a considerable quantity of nitrogen and, by reducing blowholes, impart brittleness to the steel.

Chapter IX

Investigation of the course of the Bessemer and Martin processes from the point of view of the combination of nitrogen with the metal.

The causes of the presence of nitrogen in iron, steel, and cast iron are exclusively their admixtures. In conversion processes, nitrogenous compounds are introduced into the furnace together with the cast iron itself. The smelting of malleable grades of iron from cast iron is associated with the removal from it of the greater part of its admixtures: carbon, silicon, manganese, phosphorus, and sulfur. The course of the burning out of these admixtures differs according to the method of production.

In the preceding chapters the role of carbon, silicon, and manganese was clarified; of these, the first apparently does not promote the retention of nitrogen by iron in chemical combination at high temperatures, whereas silicon and manganese, under the conditions of factory melts, undoubtedly transfer nitrogen from the atmosphere of the furnace space into the metal.

Since silicon and manganese, during the process of smelting malleable metal from cast iron, are constantly subjected to the oxidizing action of gaseous oxygen or of oxidizing slags, the nitrogenous silicon and manganese present in them may also be subjected to the oxidizing action of the indicated reagents. But this latter reaction apparently proceeds with comparative difficulty.

We know that in the Bessemer process, with the vigorous oxidation by the oxygen of the air of all the constituent elements of cast iron, not excluding iron, which in the form of ferrous oxide dissolves in the metal in considerable quantities, the nitrogenous compounds remain in the metal, and there is always more nitrogen there than in the original products. It may be that in the middle of the course of the Bessemer process the nitrogen content in the metal is greater than at its end, when we have only the residues of nitrogenous compounds dissolved in the metal that has been subjected to refining by the oxygen of the air.

It may be that the opposite phenomenon also exists: from the beginning of the blow to the end of the process, an increase in the amount of nitrogen in the metal is possible.

In the basic Martin and Thomas processes, silicon is removed by the end almost completely; in the Bessemer and acid Martin processes, 0.15–0.2% of it still remains at the end of the process. Manganese is not removed to completion in any of these processes.

Thus, we always have present elements that can combine with nitrogen and transfer it into the metal.

It is known that Bessemer metal is always much richer in nitrogen than iron and steel produced by other metallurgical processes. This, however, was to be expected, because here the contact between iron and nitrogen is the closest. In mild Bessemerizing steel contains less nitrogen, since the blast here passes not through the entire thickness of the metal, but over its surface.

In the open-hearth furnace the metal is enriched with nitrogen to an even lesser degree, although the heat lasts much longer than the Bessemer or mild Bessemer process. Here the metal is covered with a layer of slag, and the gases pass quietly over its surface, without disturbing the slag covering.

How the process of nitrogen combination proceeds during the course of the heats has not yet been elucidated; and, among other things, this is of not only theoretical but also practical interest. If, at the end of a Bessemer heat, when strong oxidation of the iron is already taking place, the amounts of nitrogenous compounds in the metal decrease, then it may be that, by adding a suitable quantity of carbon, we could obtain a metal with a smaller amount of nitrogen and, in mechanical properties, close to open-hearth metal.

In order to clarify the question of the nitrogen content at the various stages of the Bessemer and open-hearth processes, I decided to investigate them with respect to the burning out of silicon and manganese at different stages of the heat, in connection with the nitrogen content.

For this purpose, at one of the southern Russian works, on my instructions, N. Blinov followed several Bessemer and open-hearth heats, during which samples of the metal were taken at definite intervals.

The temperature of the metal in the Bessemer converter at the time the samples were taken was measured with Wanner’s pyrometer. The metal samples taken were analyzed for their contents of silicon, manganese, and nitrogen; special attention was paid to the nitrogen content, the figures for the percentage of which are the average of several analyses made by N. Blinov; some of the analyses were checked by me.

Bessemer heat.

Charge.

Material	Amount
Liquid pig iron	602 poods
Scrap	50 poods
Spiegel pig iron	38 poods
Total	690 poods

Rail steel was melted. The results of the investigation of the course of the melt are given in Table 15.

Table 15.

Time	Course of melting and taking of samples	Sample No.	\(t^\circ\) of gases, measured with Wanner pyrometer	Chemical analysis: % Si	Chemical analysis: % Mn	Chemical analysis: % N	Quantity of N in relation to Si, Mn, in %/%
h. m.
11:17	Before charging the scrap and liquid cast iron, the converter had a temperature . . . .		750°
11:19	Charging of scrap—30 poods . .
11:20	Charging of liquid cast iron . .	1	1325	2.18	1.17	0.0022	0.066
11:45	Start of the blast. Flame short and yellow . . . .
11:48	Stoppage of the blast, turning the converter on its side, and taking a sample of metal .	2	1118°	1.65	0.82	0.0025	0.101
11:49	Start of the blast . . . . . .
11:50			1151°
11:51			1207°
11:52	Stoppage of the blast and taking a sample of metal . . . .	3	1279°	1.29	0.70	0.0034	0.170
11:54	Start of the blast; flame yellowish and longer than at the beginning, by 2 times . . .		1325°
11:57	Stoppage and taking of sample .	4
11:58	Start of the blast . . . . . .
12:00	The flame begins to whiten . .
12:01	Stoppage of the blast and taking a sample of metal . . . .	5	1366°	0.79	0.56	0.0045	0.333
12:02	Start of the blast; flame dazzlingly white and long		1442°
12:04	Stoppage of the blast and taking a sample of metal . . .	6	1503°	0.67	0.56	0.0096	0.780
12:06	Charging of scrap, 30 poods . .
12:09	And start of the blast . . . . .		1476°
12:10			1560°
12:11			1688°
12:12	Stoppage and taking a sample of metal . . . . . .	7	—	0.09	0.33	0.0103	2.452
12:13	Start of the blast . . . . . .		1610°
12:14	Stoppage of the blast and taking a sample of metal . . . .	8		0.05	0.22	0.0105	3.88
12:18	Addition of 38 poods of spiegel cast iron . . . . . .
12:20	Discharge of the metal into the ladle
12:25	Pouring into molds and taking a sample of metal . .	9		0.16	0.93	0.0105	1.046

The data of this table are plotted in the diagram of Fig. 20.

Fig. 20.

For the curve showing the change in the amount of nitrogen, a scale one hundred times larger than for silicon and manganese was adopted.

It is evident from this that the amount of nitrogen increases continuously from the beginning to the end of the process.

It is very interesting that, with the rise in temperature between the fifth and sixth samples from 1366° to 1503°, the amount of nitrogen increased more than twofold.

If we recall that silicon in powder form begins to combine with nitrogen from 1300°, and that up to 1550° the reaction proceeds with increasing speed, it will not be surprising that, even under these conditions, being in solution in iron, it begins to combine so intensively with nitrogen. The subsequent increase in temperature does not have such an effect, although it must be borne in mind that the amount of silicon is continually decreasing rapidly. The decrease in the amount of manganese takes place more slowly. It is interesting that, with this decrease in silicon and manganese, the amount of nitrogen does not fall. It is clear that nitrogenous silicon and manganese are poorly affected by oxygen under these conditions.

Let us now see what results were obtained in the investigation of the open-hearth process. The samples were taken from a basic open-hearth furnace.

Open-hearth heat.

Charge.

Material	Amount
Scrap	684
Liquid cast iron	916
Ore	55
Limestone	100
Addition: spiegel iron	20
Addition: ferrosilicon	50
Addition: ferromanganese	15
Total	1685 poods

Scrap and liquid cast iron: 1600 poods.
Additions: 85 poods.

The weight of the ingots obtained was 1556 poods; consequently, the loss of metal was: 1687 − 1556 = 129 poods.

The investigation of the course of the heat with respect to the contents of silicon, manganese, and nitrogen is given in Table 16.

Table 16.

Time	Course of the melting and taking of samples	Sample No.	% Si	% Mn	% N	Amount of N relative to Si and Mn, %
h. m.
from 10²⁵ to 1¹⁵	Charge of the burden: scrap 684 p. + ore 55 p. + limestone 100 p.
1¹⁵	Charging liquid pig iron	1	1.50	1.01	0.0018	0.071
3¹⁵	Taking a sample of metal	2	0.03	0.34	0.0033	0.890
	The bath boiled weakly.
3¹⁰	Several shovels of sand were added to liquefy the slags.
3¹⁵	Taking a sample of metal; at this time the bath boiled more strongly and uniformly.	3	0.02	0.33	0.0031	0.885
3²⁰	12 shovels (9 poods) of ferromanganese were thrown in.
3²⁹	Taking a sample of metal	4	0.01	0.67	0.0031	0.456
3³⁶	” ” ”	5	0.01	0.61	0.0031	0.500
3⁴³	Added 5–6 poods more of ore.
3⁵⁰	Taking a sample of metal	6	traces	0.61	0.0031	0.508
3⁵⁷	” ” ”	7	”	0.56	0.0030	0.535
4⁰²	” ” ”	8	”	0.56	0.0029	0.518
4⁰⁵	A portion of ferromanganese was thrown in and the bath was stirred.
4¹⁰	Taking a sample of metal and after this 50 poods of 10% ferrosilicon and 15 poods of spiegeleisen were thrown in. The bath was thoroughly stirred.	9	”	0.72	0.0033	0.458
4²⁵	Taking a sample of metal, after which the tapping of the metal into the ladle was begun; 2 poods more of 85% ferrosilicon were added.	10	0.10	0.83	0.0038	0.408

On the basis of this table the diagram in Fig. 21 has been drawn. From it one sees that in the open-hearth process the amount of nitrogen, although it does increase during the melting, does so slowly—much more slowly than in the Bessemer process. This, however, could have been expected in advance. It is interesting to note the increase in the amount of nitrogen after the addition of ferromanganese and ferrosilicon.

Since the elements that promote the combination of the metal with nitrogen are always present during melts and are not completely removed at their end, it is impossible in these processes to obtain metal entirely free from nitrogen. Moreover, Bessemer metal will always be richer in it than open-hearth metal. By vigorous oxidation it may be, and could be, possible to remove a large part of the nitrogen, but owing to the large loss of metal this would hardly be economically acceptable.

By prolonged heating in a vacuum, according to Andrew’s data, nitrogen can be removed from iron, but such long heating is hardly practically feasible, although the vacuum may

Fig. 21.

Visible labels in the figure:

• \( \sigma, \delta \) and \(Mn\)
• \(N\)
• \(Si\)
• \(P\)
• \(Mn\)
• \(N\)
• \(P_{\text{elastic}}\) [[unclear: subscript/word partly illegible]]
• \(1\%,\ 2\%\ P_{\text{elastic}}\) [[unclear: exact notation partly illegible]]

need not be obligatory: it can be replaced by an atmosphere of some suitable gas other than nitrogen.

To combat nitrogen, if the use of silicon and manganese in ironmaking always proves unavoidable, there remains only the hope of finding an element that will form a stable compound with nitrogen that is insoluble in iron.

There are indications in the literature that such an element is vanadium, that it purifies steel of nitrogen; but for ordinary products of open-hearth and Bessemer manufacture it is inaccessible because of its high cost, and little hope can be placed in its cheapening.

Another element, which I pointed out as early as my first work on nitrogen, is titanium; how it behaves toward nitrogenous compounds present in molten metal will be clarified later.

Chapter X

The Influence of Nitrogen on the Mechanical Properties of Iron

I

Mechanical tests of iron alloys obtained by various methods of factory production have been carried out repeatedly. As early as 1888, Tholander²⁴ published his work in this direction. To investigate the brittleness of Bessemer metal, he took boiler plate iron, from which a piece of elliptical shape was cut out, Fig. 22. Along the long axis of the ellipse two rows of holes were punched symmetrically, and between them the elliptical plate was cut into two halves. One of the halves was filed along the cut \(a b\), and the holes were somewhat reamed.

Fig. 22.

Then one of the halves, in the hatched place, was heated strongly, while the remaining part was cooled with snow. After the indicated edge had been heated, the whole half of the ellipse was rapidly cooled with cold water. As a result of the internal stresses that formed, this part of the plate, from a light blow of a hammer, split into two parts and produced several cracks, whereas the other part, which had not been subjected to such an operation, bent under strong hammer blows without producing any cracks at all. The cause of the resulting

Tholander considers nitrogen to be the cause of cracks and fracture of the plate; according to his determination, there is much nitrogen in Bessemer steel, and approximately twice as much as in open-hearth steel (see p. 4). A similar experiment with open-hearth steel was not carried out, although it would have been very desirable for comparing these two grades of steel.

Further, he points to the difference in the character of the fracture of Bessemer and open-hearth iron, which in composition, apart from nitrogen, differed hardly at all from one another. The Bessemer iron had a light-silvery fracture, whereas the open-hearth iron was grayish lead-colored. The color of the fracture, in his opinion, also depends on the difference in nitrogen content in the two grades of iron. Products of Bessemer manufacture are considerably inferior in mechanical properties to open-hearth products. This is still observed at the present time, but the reasons for it may also depend on other factors. The course of the Bessemer process is very rapid, and the composition of the metal is much more difficult to regulate than in open-hearth melting.

Only more detailed investigations, and especially of specimens of one and the same works origin, artificially saturated with nitrogen to different contents, can provide reliable indications regarding the influence of nitrogen on the mechanical properties of the metal. The metal specimens for such a test must be taken from a single heat. They must have absolutely one and the same chemical composition and must be subjected to exactly the same temperature treatment, in order to obtain one and the same structure. If under these conditions a difference appears in the mechanical properties at the insignificant nitrogen contents that occur in the products of iron manufacture, then from the degree of this difference it would be possible to draw a conclusion as to the degree of harmfulness of these nitrogen contents.

Thus Braune set the problem of investigating the influence of nitrogen. His work was described above; see p. 5.

In 1909 C. E. Stromeyer delivered a report at the May meeting of the Iron and Steel Institute²⁵ concerning his investigations of the influence of nitrogen on the mechanical properties of cast iron, brief notices of which had appeared in Stahl und Eisen in 1907.

Stromeyer finds that the strongest influence on the deterioration of the mechanical properties of iron is exerted by phosphorus and nitrogen.

Owing to the difference in the chemical composition of the specimens taken, both those already in operation and new ones, he was for a long time unable to arrive at any positive results regarding the harmfulness of nitrogen and phosphorus. Summing the contents of nitrogen and phosphorus, when comparing the results of bending tests, did not give a clear picture. If, however, the nitrogen and phosphorus contents found are divided by the square

their atomic weights and the numbers obtained are added, then in those cases where this sum exceeds 0.8, the mechanical tests showed unsatisfactory qualities in the given specimens.

Stromeyer attempts to establish a formula for determining the temporary tensile resistance, proceeding from the chemical composition of iron and steel, into which he also introduces the amount of nitrogen with a very large coefficient, in comparison with the coefficients of the other impurities:

\[ T = 17.20 + 35 C + 10 Si + 2.5\{Mn - 1.72 S\} + 30 P + 300 N. \]

He confirms the correctness of this formula by a table, from which it is seen that it gives results fairly close to the experimental ones. This formula cannot have great significance, since, besides impurities, both thermal and mechanical treatment have no small influence on the physical properties of iron and its alloys.

In the following year Stromeyer published a work on nitrogen as a cause of the brittleness of cast iron²⁶. He investigated many specimens of boiler iron, some of which had been rejected by factories as unfit for use, while others were of good quality. Ordinary chemical analysis gave him no indication of the cause of brittleness in some sheets; likewise, by mechanical investigations he could not always find a great difference between sheets that proved to be of good quality in service and those that proved unsatisfactory in practice. Braune’s work prompted him to pay attention to the nitrogen content in the iron specimens he had. Stromeyer gives a table of analyses of iron specimens, in which the contents of carbon, nitrogen, and phosphorus are found. On the basis of mechanical tests of the investigated boiler-iron specimens, he comes to the conclusion that the influence of nitrogen, in comparison with the action of phosphorus on brittleness, should be regarded as 10:1 or, at least, as 5:1. As for the influence of phosphorus, it was usually considered that 0.06% of it renders boiler iron unfit for use.

One specimen of boiler iron, investigated by him and taken from a ship’s boiler that burst during a hydraulic test, contained 0.045% P, but at the same time 0.0123% N, i.e., approximately twice as much nitrogen as in those sheets which proved to be of good quality. Another specimen of boiler iron, also from a boiler that had not withstood a similar test, contained 0.052% P with 0.02% N. In general, he comes to the conclusion that, if in a boiler-iron specimen the sum of the phosphorus content and five times the nitrogen content exceeds 0.08%, then the iron is unfit for use.

The cause of the brittleness of iron may also be hydrogen; however, it is easily removed from iron on heating, and Stromeyer considers it improbable that hydrogen should be present in iron in such an amount as to produce a great deterioration of the material, since the iron was obtained by smelting and was then subjected to heating before rolling. From the table he gives it is evident (Table 17),

Table No. 17.

Property of boiler iron.	Designation of samples.	C %.	N %.	P %.	5 N + P.
Iron of poor quality.	Q	0,127	0,0200	0,052	0,152
Iron of poor quality.	L	0,090	0,0153	0,060	0,136
Iron of poor quality.	H	0,210	0,0145	0,079	0,151
Iron of poor quality.	Cc	0,130	0,0123	0,047	0,108
Iron of poor quality.	M	0,090	0,0090	0,052	0,097
Iron of poor quality.	Y	0,110	0,0066	0,052	0,085
Iron of poor quality.	F	0,097	0,0043	0,210	0,231
Iron of poor quality.	R	0,165	0,0030	0,177	0,192
Iron of poor quality.	K	0,205	0,0041	0,095	0,115
Iron of poor quality.	E	0,165	0,0029	0,076	0,090
Iron of poor quality.	S	0,135	0,0038	0,065	0,084
Iron of poor quality.	T	0,200	0,0023	0,038	0,049
Of good quality.	G	0,215	0,0050	0,051	0,076
Of good quality.	Z	0,195	0,0033	0,039	0,055
Of good quality.	b	0,200	0,0040	0,032	0,052
Of good quality.	BB	0,200	0,0032	0,029	0,045

that all the samples of iron which proved to be of poor quality had much more nitrogen than the good samples, and in some cases almost 10 times as much. However, it is very difficult by this method to investigate precisely how many times more harmful nitrogen is than phosphorus, since all the samples differ considerably from one another in the content of other impurities. Nevertheless, this work is of great interest, since it shows that all the samples of iron containing larger amounts of nitrogen proved to be of unsatisfactory quality.

This work attracted much attention and was reported on in detail in the best metallurgical journals, e.g. Metallurgie, Stahl und Eisen.

All the works concerning the mechanical properties of iron containing different amounts of nitrogen and tending to elucidate its influence on the mechanical properties of iron, with the exception of Braune’s work, have the drawback that the authors used for their investigations works samples differing from one another in composition with respect to other impurities.

Owing to differences in composition, the influence of nitrogen on the mechanical properties of the specimens in question cannot be strictly determined; therefore, in none of these authors, with the exception of Braune, do we find diagrams showing the change in the mechanical properties of specimens with a change in the amount of nitrogen. Moreover, it would have been extremely important to pay attention also to the microstructure of the specimens and to subject them beforehand to entirely identical thermal and mechanical treatment. The size of the crystals of the metal plays, as has already been indicated above, a by no means negligible role in its mechanical properties; and with small fluctuations in the composition of nitrogen, even if the content of other impurities were absolutely identical, the action of nitrogen can often be completely obscured by the influence of differences in the structure of the specimens.

Braune’s idea of taking one and the same material for the investigation of the mechanical properties of iron and saturating it with nitrogen to different degrees of content was entirely correct. Such preparation of specimens by a laboratory method has more than once already rendered indispensable services to science. The use of works material could not have made possible such enormous success in its development. The metallurgy of iron in recent times, with its extraordinary successes, owes not a little to the fact that, as a result of improvements in laboratory furnaces, especially electric furnaces, we are able to smelt, with great purity, all possible alloys of iron.

For elucidating the influence of nitrogen on the mechanical properties of iron, it seems to me that the most rational path will be the preparation of specimens by a laboratory method.

II.

We have already seen that iron can be saturated with nitrogen, which spreads into the interior of a piece of metal, similarly to carbon in cementation.

For the preparation of iron specimens with different nitrogen contents, I took wire of 0.63 mm diameter, made of open-hearth iron, which had the following composition:

\[ \mathrm{C}—0.14\%,\quad \mathrm{Si}—0.18\%,\quad \mathrm{Mn}—0.45\%,\quad \mathrm{S}—0.04,\quad \mathrm{P}—0.03. \]

Saturation with nitrogen was carried out with the aid of ammonia. For this purpose a wire 1 meter long was wound into a spiral about 40 mm in diameter, which was placed in an inserted quartz tube situated in a Heraeus furnace. The spiral had to fall within the region of uniform temperature, the extent of which had been measured beforehand. In order that the spiral should not be crushed when it was pushed into the furnace at high temperature and should not spread apart when withdrawn into the cold part of the tube, both its ends were tied to a thick iron—

of wire passing through the center of the spiral, by means of which its movement was effected. The wire passed outward through a mercury seal of the above-described construction. The quartz tube of the pyrometer, when the spiral was inside the furnace, reached to its opposite end, so that by moving the thermoelement it was possible to place its junction at any point of the spiral and thus, in preliminary experiments, the best results were obtained by saturation with nitrogen at \(900^\circ\) for varying times with subsequent annealing. After the furnace temperature had been brought to \(900^\circ\) and the atmosphere had been displaced with dry ammonia, the spiral was rapidly moved from the cold part of the insertion tube into the middle of the furnace. The ammonia current was rapid; its direction in the furnace was changed several times during each experiment, since otherwise one end of the spiral became saturated with nitrogen somewhat more than the other. The action of the ammonia lasted from 1 min. to half an hour. After a definite time had elapsed, the ammonia atmosphere was replaced by nitrogen. Then the temperature was raised to \(1100^\circ\) for annealing the spirals for 4 hours. All wire specimens saturated with nitrogen were subjected, in order to obtain a uniform and fine-grained structure, to annealing. The annealing was carried out at \(950^\circ\) for \(1/4\) hour in an atmosphere of nitrogen.

For chemical and mechanical investigations only those spirals were taken which showed no signs of tarnish colors. Their surface was silvery-white and, the greater the nitrogen content, the more matte it was.

Saturation with nitrogen by a mixture of ammonia with nitrogen or hydrogen required much more time and gave poorer results. Saturation with nitrogen at rising temperatures, owing to the presence of Si and Mn, likewise gave poorer results than the first method.

After the indicated treatment, the spirals were straightened; for this, one end of them was clamped in a vise and the wire was smoothed by pulling it, gripped in the hand, with a towel. Large turns of the spiral made it possible very easily and without effort to obtain straight wire. Spirals with small turns straightened poorly and required considerable effort for this; moreover, such straightening of thin wire must be carried out very delicately, since, by stretching it and strongly straightening it with the aid of a piece of wood or a rag, it is easy to pass beyond the elastic limit, especially in the soft grades. It is known that in this case the specimen, stretched beyond the elastic limit, after the force ceases to act, retains a certain residual elongation. The elastic properties are somewhat changed in comparison with those which the specimen had before stretching. If this specimen is again subjected to tension, the diagram obtained will be of an entirely different form than in the first test. Consequently, its qualities are somewhat altered. The specimen does not

it loses the ability to resist the action of forces deforming it, but the elastic limit and the temporary resistance to rupture increase. Thus, for example, if, while stretching a specimen, after passing beyond its elastic limit, we reduce the stretching force almost to zero and then again subject it to stretching, then on the diagram we obtain a step. With several such stops we obtain a series of rising steps instead of the smooth curve of the diagram of the deformation process. Such a state of a material that has undergone permanent deformation, called the adapted state, has a similarity to cold-working. In both the first and the second cases we find an increase in the elastic limit and in the breaking load, and a great similarity in the character of the diagram of the deformation process. Careful straightening of the wire can so influence its mechanical properties as completely to conceal the action of small amounts of nitrogen.

As an example I shall cite the result of the mechanical test of the same wire that was used for saturation with nitrogen, annealed at 950°, one piece of which was straightened before annealing and had 19.7% elongation at a 15.3 kg breaking load (over the whole section of the wire, 0.63 mm in diameter). When a piece of the same size, clamped at one end in a vise, was strongly wiped several times, being wrapped in a cloth, it gave already 9.9% elongation at a 14.8 kg breaking load. A new specimen, under the same conditions, gave 11.5% elongation at a 14.7 kg breaking load.

Wires saturated with nitrogen and annealed for the uniform distribution of the nitrogen throughout the entire thickness of the metal were investigated with respect to the uniformity of the distribution of the nitrogen.

In the examination under the microscope of a perpendicular section of the wire after annealing at 1100°, a crust with the microstructure of nitrogenous iron was never observed. The structure of wire specimens containing up to 0.15% nitrogen was completely uniform over the whole section. With a nitrogen content greater than 0.15%, the wire was rough. Cracks were visible under the microscope. Owing to such surface defects, which occurred during saturation with nitrogen, the mechanical properties must have suffered considerable deterioration. Both for this reason and because such specimens, in their nitrogen content, are too far removed from iron of factory production, I shall not present the results of their testing. At 0.3% the wire breaks without giving any bend. At 0.4% its brittleness is equal to the brittleness of glass rods.

In addition to examination under the microscope, chemical investigation was also carried out many times, both of filings from the surface of wire specimens containing different amounts of nitrogen and of their cores remaining after filing. In this investigation the following was found:

Surface of the wire: 0,123 % N. Middle part: 0,12 % N.
” ” 0,0762 % N. ” ” 0,0738 % N.
” ” 0,0288 % N. ” ” 0,0257 % N.

Having obtained such results, I considered the method of preparing specimens for the mechanical testing of wire to be sufficiently good.

For the mechanical testing of the wire, a Schopper-system machine of 30 kg was used. The forces in it are produced by means of a hydraulic cylinder, acting very smoothly. The length of the wire specimen being stretched was 180 mm. The load scale of the machine had previously been checked by hanging weights of 5, 10, and 20 kg, and in this case the divisions corresponded exactly to the load. The load was read on a scale having divisions of \(1/10\) kg. The elongations were read on another scale, which directly indicated the percentage elongation. This scale has divisions of 0,2%, and 0,1% can be read quite readily; smaller fractions of a percent are approximate. For this reason an exact determination of the elastic limit is difficult to accomplish. However, the influence of nitrogen on the elastic limit is so considerable that the change in it with increasing nitrogen content is clearly visible on the diagrams.

All wire specimens were subjected to tension at one and the same speed of the machine and gave the results presented in Tables 18, 19, 20, 21, 22, and 23.

Wire not saturated with nitrogen, annealed at 950°.

Table 18.

Load, kg	Elongation in % — Specimen No. 1	Elongation in % — Specimen No. 2	Elongation in % — Specimen No. 3
5	0,06	0,07	0,05
5,5	0,08	0,09	0,07
6	0,10	0,11	0,09
6,5	0,12	0,13	0,12
7	0,13	0,14	0,13
7,5	0,15	0,16	0,15
8,0	0,16	—	0,20
8,5	0,21	0,20	0,21
9,0	0,23	—	0,30
9,5	0,32	0,31	0,40
10,0	0,40	0,37	0,50
10,5	0,60	—	0,75
11,0	0,80	—	0,95
11,5	1,20	1,1	—
12,0	1,80	—	2,00
12,5	2,60	2,4	2,90

Load in kg	Elongation in %
	Specimen No. 1
13.0	3.60
13.5	4.60
14.0	6.50
14.5	8.90
15.0	13.30
15.1	—
15.2	19.6
15.5	—

The breaking load is given here and in all the remaining tables for the full cross-section of the tested wire, $0.3117$ sq. mm. Per 1 sq. mm it will be: 1) 50.01 kg, 2) 50.4 kg, and 3) 51.4 kg.

The nitrogen in the wire was $0.0038\%$.

In the following tables are given the results of the mechanical testing of wire saturated with nitrogen to a different percentage content of nitrogen. In each table a series of experiments is given on pieces of wire from one and the same coil. If, in such tests, the pieces of the coil gave too great a difference among themselves, the experiment was considered unsuccessful and a new preparation was made.

Table 19.

Load in kg	Elongation in %
	Specimen No. 1
7	0.2
7.5	—
8	—
8.5	—
9	—
9.5	0.28
10	—
10.5	—
11	—
11.5	0.3
12	0.35
12.5	0.4
13.0	0.51
13.5	0.8
14.0	2.0

Fig. 23.

Load per sq. mm.

Elongation, in %.

Load, kg	Elongation in %	Elongation in %	Elongation in %
	Specimen No. 1	Specimen No. 2	Specimen No. 3
14.5	4.2	4.0	1.9
15	7.1	4.6	3.6
15.5	9.4	7.6	6.3
15.9	13.0	—	—
16	—	11.1	8.2
16.2	—	12.6	—
16.4	—	—	13.2

The nitrogen content according to analysis was as follows:

No. 1—0.0255% N,
No. 2—0.0289% N,
No. 3—0.0274% N.

Table 20.

Load, kg	Elongation in %	Elongation in %	Elongation in %
	No. 1	No. 2	No. 3
5	0.1	—	—
5.5	0.12	0.1	0.1
6	0.13	—	—
6.5	0.15	—	—
7	0.18	—	—
7.5	0.19	—	—
8	0.20	—	—
8.5	0.21	0.2	—
9	0.22	—	0.2
9.5	0.23	—	—
10	0.24	—	—
10.5	0.25	—	—
11	0.26	0.25	—
11.5	0.28	—	0.3
12	0.29	0.3	0.32
12.5	0.30	0.33	—
13	0.33	—	0.7
13.5	0.35	0.38	0.42
14	0.4	0.4	0.45
14.5	0.43	0.45	0.5
15	0.6	0.52	0.55
15.5	1.1	0.8	0.9
16	2.0	1.5	2.1

Fig. 24.

Fig. 25.

Load, in kg	Elongation in %/% No. 1	Elongation in %/% No. 2	Elongation in %/% No. 3
16,5	2,8	2,4	3,2
17	3,7	3,3	4,1
17,5	4,8	4,4	5,5
18	6,0	5,8	7,5
18,2	—	—	9,7
18,5	7,9	7,6	—
18,6	—	9,8	—
18,8	10,2	—	—

By analysis the following nitrogen content was determined:

No. 1—0,0393% N,
No. 2—0,0372% N,
No. 3—0,0365% N.

Table 21.

Load, in kg	Elongation in %/% No. 1	Elongation in %/% No. 2	Elongation in %/% No. 3	Elongation in %/% No. 4
5	0,1	—	0,06	—
5,5	—	—	—	0,03
6	—	—	0,1	0,05
6,5	—	—	0,11	0,09
7	—	0,12	0,12	0,1
7,5	—	—	0,14	0,12
8	—	—	0,16	0,13
8,5	0,2	—	0,18	0,15
9	—	0,17	0,19	0,18
9,5	—	—	0,2	0,19
10	—	—	0,21	0,2
10,5	—	—	0,22	0,21
11	0,25	—	0,25	0,22
11,5	—	0,2	0,29	0,24
12	—	—	0,3	0,28
12,5	—	—	0,32	0,3
13	—	—	0,35	0,31
13,5	0,36	0,3	0,37	0,33
14	—	—	0,39	0,36
14,5	—	—	0,4	0,39
15	—	—	0,43	0,4
15,5	0,4	0,4	0,46	0,42

Load in kg	Elongation in %%. No. 1	Elongation in %%. No. 2	Elongation in %%. No. 3	Elongation in %%. No. 4
16	0.41	0.4	0.49	0.45
16.5	0.48	0.43	0.51	0.50
17	0.5	0.5	0.56	0.55
17.5	0.6	0.6	0.6	0.6
18	0.68	0.83	0.67	0.72
18.5	0.9	1.15	0.8	0.9
19	1.6	2.6	1.1	1.9
19.5	2.8	4.0	1.9	3.5
20	4.2	5.7	3.2	5.2
20.5	—	6.9	—	—
20.3	—	—	—	6.2
20.4	7.1	—	—	—
20.8	—	—	7.3	—

According to the analysis, in this series of specimens the following nitrogen content was found:

No. 1—0.0658% N,
No. 2—0.0662% N,
No. 3—0.0762% N,
No. 4—0.0687% N.

Fig. 26.

Table 22.

Load, kg	Elongation in %/%. No. 1	Elongation in %/%. No. 2	Elongation in %/%. No. 3
5	0,08	—	—
5,5	0,09	—	—
6	—	—	—
6,5	—	—	—
7	—	—	0,10
7,5	0,13	0,10	0,11
8	0,15	0,11	0,14
8,5	0,18	0,12	0,16
9	0,20	—	0,18
9,5	0,22	—	0,19
10	—	—	0,20
10,5	0,23	0,18	0,21
11	0,25	0,20	0,22
11,5	0,28	0,21	0,24
12	0,30	0,23	0,25
12,5	0,31	0,25	0,26
13	0,32	0,27	0,27
13,5	0,33	0,28	0,28
14	0,35	0,30	0,30
14,5	0,37	0,31	0,32
15	0,38	0,32	0,33
15,5	0,40	0,33	0,35
16	0,41	0,36	0,38
16,5	0,42	0,39	0,40
17	0,43	0,41	0,45
17,5	0,48	0,43	0,50
18	0,5	0,51	0,58
18,5	0,53	0,58	0,70
19	0,71	0,75	1,1
19,5	0,9	1,1	1,7
20	1,9	1,9	2,5
20,5	2,8	2,8	3,2
21	3,9	3,8	4,1
21,5	5,4	5,1	5,2
21,6	6,8	—	—
21,8	—	—	6,7
22,1	—	5,8	—

According to analysis, in this series of specimens the following nitrogen content was found:

No. 1—0,0801% N,
No. 2—0,0743% N,
No. 3—0,0830% N.

Table 23.

Load, kg	Elongation, % № 1	Elongation, % № 2	Elongation, % № 3
6.5	0.01	0.01	—
7	0.03	0.04	—
7.5	0.05	0.06	—
8	0.08	0.07	—
8.5	0.09	0.08	—
9	0.1	0.09	—
9.5	0.11	0.10	0.1
10	0.12	0.12	—
10.5	0.13	0.13	—
11	0.15	0.16	—
11.5	0.17	0.17	—
12	0.19	0.18	—
12.5	0.20	0.20	—
13	0.21	0.20	0.2
13.5	—	0.21	—
14	0.22	0.22	—
14.5	—	—	—
15	0.23	0.23	—
15.5	0.25	0.24	0.25
16	0.28	0.26	—
16.5	0.30	0.27	—
17	0.31	0.3	0.3
17.5	0.33	0.31	—
18	0.36	0.34	0.4
18.5	0.38	0.36	—
19	0.4	0.39	—
19.5	0.41	0.4	0.42
20	0.43	0.45	0.45
20.5	0.45	0.48	0.5
21	0.48	0.5	0.55
21.5	0.5	0.7	0.61
22	0.6	0.9	0.79
22.5	0.8	1.0	1.42
23	1.3	1.6	2.1
23.5	1.7	1.9	2.8
24	2.4	—	—
24.5	3.1	—	—
24.8	3.8	—	—
24.9	—	—	3.5
25.4	—	3.6	—

Analysis showed in the first specimen: 0.122% N,
    ”    ”    in the second    ”    0.126% N,
    ”    ”    in the third    ”    0.131% N.

The readings at the beginning of the action of the force are not distinguished by great accuracy up to the transition of the specimens beyond the elastic limit, since the divisions of the scale were 0.2%, and hundredths of a division cannot be read exactly. However, I give this reading, since from it one can observe, to some degree, the transition of the specimen beyond the elastic limit.

Fig. 27.

Fig. 28.

On the basis of the figures given in the tables, deformation diagrams have been constructed:

Diagram, Fig. 23, for Table 18.
” ” 24 ” ” 19.
” ” 25 ” ” 20.
” ” 26 ” ” 21.
” ” 27 ” ” 22.
” ” 28 ” ” 23.

From the tables and diagrams it is evident that, with an increase in the amount of nitrogen, the breaking load increases with a large decrease in elongation. From the same diagrams it is evident that the elastic limit falls.

On the basis of the mechanical-test data, summary diagrams have been constructed—Figs. 29, 30, and 31—showing the increase in tensile load, the elastic limit, and the fall in elongation.

Fig. 29.

Fig. 30.

These diagrams clearly show that the harm caused by nitrogen to the mechanical properties of iron is very great. The effect of nitrogen is reflected especially strongly in the elongation, and special attention must be paid to this, since

as, for technology, the amount of elongation that a given material can undergo before rupture may be of great significance. Elongation characterizes the ability of a metal to acquire permanent changes of shape before failure. This is important for many methods in the technology of metals, in which metal sheets or bars are given changes of shape that the metal must withstand.

Fig. 31.

In addition, the ability of metallic structures to change their shape perceptibly before failure is a valuable property, giving an indication when the elastic limit has been exceeded and, consequently, when danger of failure threatens.

Nitrogen, by causing hardness in iron and reducing its plasticity, makes it brittle.

Factory grades of iron and steel contain much less nitrogen than was present in the specimens investigated. However, diagrams constructed on the basis of mechanical tests of these specimens, by the steep slope of the curves, indicate a considerable influence of small quantities of nitrogen on the mechanical properties of iron.

Bessemer metal usually contains almost five times as much nitrogen as open-hearth metal.

It is indicated by many authors that Bessemer metal possesses greater hardness and brittleness in comparison with open-hearth metal. If this is so, then it must be thought that nitrogen here plays no small role.

1. 1891–1896, p. 445. ↩

2. Ann. Chem. Pharm. 1882, 7, 375. ↩

3. Compt. rend. 1861, 52, p. 1195. ↩

4. The journal of the iron and steel institute, 1880. ↩

5. St. u. Eis. 1889, p. 115, abstract of the article by H. Tholander, Dr’om, Leo from Jernkontorets Ann. 1889, VII. ↩

6. T. journ. of the iron and steel institute. 1897, p. 161. ↩

7. Jerikoat. Ann. 1906. No. 5 and 6, and also in St. u Eis. 1906, Nos. 21, 23 and 24. ↩

8. Stahl und Eisen. 1906, p. 1518. ↩

9. Revue de Metallurgie. 1908, 493. ↩

10. Iron and Nitrogen. N. Chizhevskii. Bulletin of the Society of Technologists, 1909, No. 10. ↩

11. J. R. Phys.-Chem. Soc., 1910, p. 54. ↩

12. Iron and Nitrogen. N. Chizhevskii and I. Privalov. Bulletin of the Technology of Chemical and Building Materials, 1910. ↩

13. N. Chizhevsky. Bulletin of the Society of Technologists. 1911. ↩

14. Lieb. Ann. 110, 248. ↩

15. Jahresbericht d. Chem. 1870, 231. ↩

16. ibid. 1881, 202. ↩

17. ibid. 1862, 187. ↩

18. ibid. 1881, 202. ↩

19. Ann. Chim. phys. 1897, 12, 13. ↩

20. Zt. anorg. Chem. 65, 36. ↩

21. N. Chizhevsky. Bulletin of the Society of Technologists, 1911. ↩

22. The Journal of the Iron and Steel Institute, 1911. ↩

23. Abstract in Zh. R. M. O., 1912, No. 5, from The Journal of the Iron and Steel Institute, 1902, II; also Ferrum, 1913, 151. ↩

24. Stahl und Eisen 1889, 119 (ref.). ↩

25. Stahl und Eisen 1909, 1491 (ed.). ↩

26. Iron Age. 1910. p. 858. ↩