the oxides of nitrogen. The common elements exhibiting the most pronounced tendency to combine with nitrogen are calcium (Ca), magnesium (Mg), aluminium (Al), boron (B), &c. The carbides of a large number of metals also exhibit a pronounced tendency to combine with molecular nitrogen when heated.

The great advantage from a theoretical standpoint in utilising these processes is that the reaction with nitrogen may be made more complete as equilibrium may be continually disturbed by withdrawing the compound of nitrogen formed or by presenting to the nitrogen to be combined fresh combining surfaces of the substance, and the action may be caused to proceed practically quantitatively, thus avoiding heating large quantities of materials which are inert to the reaction. A vast field of research is opened by these possibilities, as very few of the equilibrium figures have been determined, and it is almost certain that direct combustion methods may eventually be evolved along these

lines.

The experimental data are so meagre that no theoretical limitations can be placed. The reactions assume unusual importance however on account of a wide application in the arts. This may be best illustrated by considering the formation of calcium carbide, CaC2, in relation to its three reversible reactions, namely,—

3.

I. CaO + 3C → CaC2 + CO 2. CaC Ca + C CaO + Cam Ca + CO There are here six substances, some in solid form, some in liquid, and some gaseous (molecular and atomic), and it is evident that the temperature will have a marked influence on the equilibrium which will exist, and the reaction will be greatly affected by very minute changes, for the partial pressure of the gases will be suddenly changed by such conditions as the carbon released in a gaseous state immediately combining to form amorphous carbon, or the metallic calcium vapour combining with the oxygen released by the CO to form calcium oxide, which immediately precipitates as a solid. The fact that calcium oxide, which is most refractory, can be vaporised at a temperature of 1600° C. to 1800° C. in the sense that the calcium is vaporised and decomposes CO to again precipitate CaO, is one of the actions similar to the fumes in smelting furnaces which accounts for a heavy loss of metal at temperatures not ordinarily capable of producing vapours. When nitrogen is inserted in a reaction of this kind, there are immediately formed complex carbon-nitrogen compounds, but the action of the oxygen present is to disso. ciate these, allowing the nitrogen to combine into the molecular form and the metal to precipitate from the fume as a minute particle of metallic oxide, the carbon precipitating as amorphous carbon in the form of soot, as all of these reactions liberate a large amount of energy. The temperature of these reactions is from 1500° C. to 2000° C., and is therefore well within the range of combustion methods if the combustion could be in contact with the substances as in the blast furnace, but the presence of the oxygen necessary for combustion prevents the formation of nitrogen compounds.

The effect of partial pressures in these reactions is of fundamental importance. The active mass of a solid is constant, and hence at the boundary surface where the solid and gases meet there is a high velocity of reaction. Equilibrium will be produced either by the solid forming a coating of the compound which will place it in equilibrium or by the pressure of the gases generated from the reaction producing a condition of equilibrium. Taking the familiar calcium carbonate reaction as an illustration,

CaCO3 = CaO + CO3.

There being two substances in the solid state and one in the gaseous, the equilibrium constant K will be

where CaO and CaCO3 are the very slight vapour pressures of the solids; that is, the sublimation pressures, which in practice are too small to measure. The pressure of the CO2 will be practically the total pressure, and as this varies, the velocity constant K will vary and hence for any temperature there is but one pressure for equilibrium. the above reaction over a wide range and found a variation Le Chatelier measured the temperatures and pressures of in the temperature necessary to produce the reaction of from 547° C. at 27 mm. pressure to 865° at 1333 mm.; or in other words, equilibrium could be produced through a range in temperature of 60 per cent, and at one point a change of two degrees necessitated a change in pressure of over 10 per cent in order to restore equilibrium. Rothmund found the equilibrium pressure of CO in the carbide of calcium reaction to be 250 mm. at 1620° C. If the CO pressure was raised above the equilibrium figure at this temperature, CO was absorbed and no carbide was formed. By inserting inert gases so the CO was diluted and its partial pressure reduced, the temperature of the formation of carbide was varied over 20 per cent. These results all indicate that for the nitrogen reactions must not only be subject to accurate temperature regulation, but the partial pressures must be controlled, and it is probable that definite zones of reaction must be maintained. present commercial applications such as the Serpek and cyanamid processes prepare the compounds of nitrogen by causing the nitrogen to react largely with the solid masses, and thus avoid many of these complications due to the vapour processes, or variable minute pressures of sublimation. reducing the metals in the presence of nitrogen, thus By substituting the resistance furnace for the arc and forming nitrides or forming carbides, and treating these in the presence of nitrogen, there have been developed a number of processes which have been commercially applied. (To be continued).

The

PERMANGANATE DETERMINATION OF IRON
IN THE PRESENCE OF FLUORIDES.
THE ANALYSIS OF SILICATES AND
CARBONATES FOR THEIR FERROUS IRON
CONTENT.*

By O. L. BARNEBEY. (Concluded from p. 19).

Notes on the Method.

1. SCRUPULOUS care must be exercised to exclude the air when decomposing the silicate with hydrofluoric acid in order to prevent oxidation. However, after the addition of the boric acid the ferrous solution becomes quite stable. 2. A carbon dioxide generator, after filling with carbonate (marble) and acid, should be allowed to generate carbon dioxide for a considerable period of time to remove the air as completely as possible. (Heating magnesite would probably be a still better method for obtaining carbon dioxide). Only recently boiled water should be used in the water-bath and only recently boiled distilled water added to the sample.

3. Until accustomed to the method of analysis the analyst should always check his procedure by heating

measured volumes of standard ferrous iron solution with hydrofluoric acid, and titrating with the same manipulation as in the analysis, in order to ascertain if the air is being excluded effectively.

4. Filtration is desirable when solid organic matter is

Journal of the American Chemical Society, xxxvii., No. 6.

visible after the sample is thoroughly decomposed. Filtration may not remove all the influence of the organic matter, however. The soluble organic matter naturally still has opportunity to give a reducing effect. If the clear solution is slightly coloured by the presence of organic matter it should be largely diluted and titrated to the first distinct pink tinge of the solution. The presence of organic matter makes the end-point much less stable than in its absence. Another procedure can be substituted for filtration. The sample can be transferred to a volumetric flask, diluted to the mark and the residue allowed to settle, after which portions can be withdrawn with a pipette and titrated.

5. Instead of adding solid boric acid the fluoride solution can be poured directly into an excess of a saturated solution of boric acid.

6. A convenient strength of permanganate for titrating samples low in ferrous iron content is 1 cc. = 0.001 grm. FeO. Samples 1-5 inclusive (Table XII.) contained no organic matter and were free from sulphides. The endpoints were fleeting in analyses 2a, 3a, and 4a, especially the last two. Samples 6-8 inclusive contained consider able organic matter. In analysis 6a the result is only approximate, as the end-point was very fugitive because of the organic matter present. In 6b, 6c, 6d, 6e the solutions were filtered, as also were the samples 76 and 7c as well as 86 and 8c. Larger portions could have been used to advantage in Samples 5, 7, and 8. Samples were not analysed according to procedure "a" for 7 and 8 because of the presence of large amounts of organic matter. Naturally filtration of the hydrofluoric acid solution is not advisable on account of the rapidity of oxidation of ferrous iron in such a solution by the air and also because of the action of the solution on glass.

TABLE XII.-Analyses of Silicates for Ferrous Iron. (1 cc. KMnO4= =0'001 grm. FeO). Per cent FeO.

application if considerable siliceous matter is present and the total ferrous iron content of the sample is wanted. The action of these preventives may be classified as follows:-Class (a) removes most if not all of the active hydrofluoric acid by the presence of another acid through the medium of which the oxidation proceeds by virtue of its greater strength as an acid or by mass action. Sulphuric acid by conductivity measurements is stronger than hydrofluoric acid; phosphoric acid is much weaker. Hence sulphuric acid gives fairly good and phosphoric acid poor prevention. The acid salts found to be of worth are valuable mostly for their acidity, i.e., NaHSO4, KHSO4. (b) Certain salts react with the fluorides present, forming undissociated fluorides and salts of other acids than hydrofluoric, through the medium of which the reaction proceeds. Examples of this class of preventives are ferric and magnesium sulphates. Phosphates are not desirable, inasmuch as they yield phosphoric acid by interaction with hydrofluoric acid, and phosphoric acid is to be avoided in the titration unless all the hydrofluoric acid is removed. Of the neutral phosphates only those of the alkalis are soluble in water; hence to add metal phosphates, either the solids or the phosphate dissolved in phosphoric acid must be added, thus forming acid phosphates in most instances and again bringing the phosphoric acid into predominance. Still another difficulty arises from the use of, phosphates; sufficient acidity must be present to prevent ferrous phosphate from precipitating. If the ratio of phosphate to acid does not give sufficient acidity the titration is worthless, being low. Classes (a) and (b) are so closely related that they could properly be included in one class, but in (a) hydrofluoric acid is present and in (b) a Aluoride. (c) Other reagents react with hydrofluoric acid in such a manner as to combine the fluorine in the anion which does not dissociate to yield hydrofluoric acid. Boric and silicic acids are such preventives, forming Aluoboric and fluosilicic acids by interaction with hydrofluoric acid.

Of the preventives tried the author prefers boric acid. This acid can be obtained on the market in a high degree of purity and is a cheap commodity. It gives a solution which is clear. The solubility of boric acid in water peculiarly fits it for this purpose, being sufficiently high to allow a rapid reaction with hydrofluoric acid and yet not so high as to cause waste of the chemical when adding the solid in slight excess. The appearance of solid boric acid in excess likewise indicates that sufficient reagent has been added to react with the hydrofluoric acid present. This is a distinct advantage when titrating solutions containing unknown or variable quantities of hydrofluoric acid. Solid boric acid in excess is advantageous rather than detrimental in obtaining the end-oint. The fluoboric acid

solution of ferrous iron oxidises very slowly in the air,

thus removing an undesirable feature of the titration accompanying the usual methods for determining ferrous iron. It gives the best prevention of the hydrofluoric influence during titration.

Summary.

1. This study confirms earlier work which has shown that the permanganate titration of ferrous iron in the presence of fluorides gives an unstable end-point, the instability increasing with increased concentrations of iron and hydrofluoric acid.

of normal hydrofluoric acid. Phosphoric acid cannot be substituted for sulphuric acid, since the former yields a fugitive end-point. Certain acid sulphates accomplish the

same result as the free acid.

influence.

4. Certain oxides also have prevention tendencies-i.e.,

NEWS

The experiments detailed above show that the amounts of the halogen acids which combine with casein, fibrin, and egg albumin, as measured by the usual indicator titration, are low and not accurately proportional to the molecular weights of the acids. The discrepancies are probably due to the lack of delicacy of the indicator in presence of unchanged protein, on the one hand, and to the more or less complete dissociation of the protein-acid compound on the other. The latter is doubtless the more important factor, since it has been shown that a large fraction of the acid may be washed away from the protein. When the proteins named are treated with the halogen acids of o 2 N concentration in excess, and the mixtures evaporated at a low temperature by standing over sulphuric acid some weeks, followed by similar treatment over solid alkali, and final drying to constant weight at 75°, very constant weights of acid are taken up and held by the protein. These weights of acid are not increased by the excess added, which points to the definite character of the reaction. The amounts are not proportional to the molecular weights of the acids, the combining proportion being relatively greater for HI than for HBr, and greater for the latter than for HCl. But the compounds al appear to be salts of the protein molecule, and contain many times as much acid as is suggested by the titration combinations. These dry salts undergo dissociation readily when mixed with water.

If the acid protein mixtures are evaporated on the waterbath in place of being dried at a low temperature the behaviour of HCl and HBr remains essentially the same. No greater amounts of the acid are taken up by a grm. of protein, and we doubtless reach here a maximum in the combining power of the acid and protein. A salt of a type different from that formed in solution at a low temperature is secured. In the case of the HI, however, there is no such limit to the iodine held, and it is probable that we have here a substitution of the element in the nucleus of the protein molecule as well as an addition of the acid. As much as 75 per cent of the weight of the original pro tein may be so held, and the combination has a brownish ochre colour, with loss of protein reactions.

PHYSICAL AND MECHANICAL FACTORS IN

CORROSION.*

By CECIL H DESCH, D.Sc., Ph D., Graham Young Lecturer in Metallurgical Chemistry in the University of Glasgow.

IT is generally recognised that the corrosion of a metal or alloy is intimately connected with the formation of local electrolytic couples at the surface in contact with the solution or atmosphere. The process of corrosion is always initially one of chemical solution, the dependence of which on electrolytic conditions is established. A highly purified specimen of zinc is almost inactive towards acids, whilst commercial zinc, containing lead, is rapidly attacked, the action of the acid beginning in the immediate neighbourhood of the lead particles. This is one of a large class of similar facts, the bearing of which on the question of corrosion is often overlooked. The object of the present paper is to show how the mechanical heterogeneity of metals and alloys affects the nature and velocity of the process of corrosion.

Laboratory tests of corrodibility are commonly made for the purpose of determining which of several materials will offer the greatest resistance to corrosion when exposed under certain specified conditions. As it is impracticable to reproduce these conditions exactly on account of the long duration of the tests, rapid tests, involving the use on the one hand of active chemical corroding agents, or on the other of an applied electromotive force, are commonly adopted. Such tests are usually of an unsatisfactory character. A number of metals or alloys, arranged in order of their resistance to acid solutions, will present a very different order when exposed to technical conditions, such as contact with town air or sea water.

Moreover, tests which involve the determination of the weight of material removed by the corroding agent, usually including that which is removed by brushing or scraping after corrosion, confound a number of successive changes in such a way that their separate influences cannot be disentangled. For example, experiments which determine

A Contribution to a General Discussion on "The Corrosion of Metals," held before the Faraday Society, December 8, 1915.

the relative loss of weight of specimens immersed in an acid or salt solution for equal periods of time fail to dis tinguish between (a) the material actually dissolved and remaining in solution, (b) the loose, flocculent precipitate of basic salts which is produced in salt solutions, (c) the adherent film of basic salts which in some cases has the properties and effect of a tough, protective varnish, (d) the metallic layers, such as the spongy layer of copper formed in the "dezincification" of brass, and (e) crystals mechanically dislodged from the face of the specimen by

FIG. -Muntz Metal in First Stage of Dezincification. (Multiplied 100 diam.).

tains, as shown by Dr. Beilby, films of amorphous material between the separate crystal grains. The amorphous metal is more electro-positive than the crystalline, as is shown by measuring the difference of potential between two specimens of the same metal in the cold-worked and annealed state respectively (W. Spring, Bull. Acad. Roy. Belg., 1903, 1066), consequently cold-rolled metals almost always corrode in such a way that greater chemical action

FIG. 2.-Transverse section of Dezincified Brass.
(Multiplied 250 diam.).

latter referring to localised attack, the solution of the metal being either entirely limited to or greatly intensified in definite areas. Microscopical examination, however, often shows that what is apparently general corrosion is really pitting on a minute scale. An annealed a brass exhibits true general corrosion, the surface being uniformly attacked, except that neighbouring crystal grains of different orientation may show slight differences of level. On the other hand a light aluminium alloy corrodes by

FIG. 4.-Pitting in Light Aluminium Alloy. (Multiplied 250 diam.).

takes place along certain lines than along others, such lines following the direction of the rolling. This effect is not to be confused with the striation observed on the surface of rolled plates due to the inclusion of impurities. Rolled sheet-iron will corrode in parallel lines even in the absence of cold-working, on account of the presence of particles of scale which are embedded in the surface during rolling and produce a heterogeneous structure. Cold-worked metals become striated during corrosion, however, even when the outer skin has been completely