The remaining T.N.T. was again extracted with hot water, cooled, and titrated; the operation being repeated ten times for each sample. The following results were obtained, the acidity being calculated as sulphuric acid.*

Even pure a-trinitrotoluene (re-crystallised from acetone and then from alcohol) melting at 80.7° C. was found to be not quite neutral, and on extracting it in a manner described above had acidities varying between 0.0025 per cent and 0.0050 per cent, calculated as sulphuric acid.

Representing the results obtained above in a graphic form we get the curves shown in the accompanyiug diagram.

The persistency of the acidity of crude T.N.T. and the general dependence of the acidity upon the purity of the product would suggest that the acidity of well-washed crude T.N.T. is due not to the retaining of any mineral acid, but to the contamination with organic acid or phenolic by-products.

There is more reason to attribute the acidity to phenolic than acid bodies, as the latter are far more soluble in hot water, and are removed to a great extent during the washing. Whilst the acidity of re-crystallised T.N.T. could be attributed to the formation of nitroic acids or dinitro-cresols under the influence of the hot water, the comparatively high acidity of washed crude T.N.T. must be attributed to the by-products formed during the nitration.

References.

1. Atti R. Accad. Lincei, 1914, (v.), xxiii., II., 464. 2. Ibid., 484; Gazetta, 1915. xlv., 345, 352. 3. Journ. Soc. Chem. Ind., 1915, xxxiv., 781.

Royal Society of Arts. -Albert Medal.-In the absence of the President of the Society, H.R.H. the Duke of Connaught and Strathearn, K.G., Dr. Dugald Clerk, F.R.S., Chairman of the Council, at a meeting of the Council on January 24, presented the Society's Albert Medal to Prof. Sir Joseph John Thomson, O.M., D.Sc., LL.D., F.R.S., "for his researches in chemistry and physics, and their application to the advancement of Arts, Manufactures, and Commerce." The Society's Albert Medal was founded in 1863 as a memorial of H.R.H., Prince Consort, and is awarded annually "for distinguished merit in promoting Arts, Manufactures, and Commerce."

the

The gradual decrease of the acidity values will become slightly smaller on the introduction of a correction for the solubility of T.N.T. n water (o 021 per cent at 15° C.; 0164 per cent at 100" C.).

AND STEEL.*

By J. NEWTon friend, D.Sc., Ph.D., F.I.C. (Concluded from p. 35).

Results of Experiment and Experience. FROM the foregoing it is evident that our problem is by no means as simple as might appear at the first glance. The factors involved are too numerous to allow of the subject being summarily dismissed in favour either of iron or of steel as the result of a few isolated laboratory experi ments, and whilst theoretical considerations may give us a general idea as to the type of metal that may be expected to withstand corroding media most successfully, there can be no doubt that the final appeal must always be made to the results of experience and of experiments carried out with a liberal hand on a large scale under actual working con ditions.

More work has been carried out along these lines than the average reader is aware, and the following brief summary of the more important researches may prove useful:

1. One of the earliest researches carried out on a large scale is that of William Parker (Journ. Iron and Steel Inst., 1881, i., 39), who in 1881 exposed plates of seven

different kinds of wrought iron and of four varieties of steel to various corroding influences - namely, London air, seawater, bilge-water, and boiler-water. The results obtained were distinctly in favour of wrought iron, the mean results for the relative losses in weight consequent upon corrosion being

The results, however, are now of historic interest rather than of practical value, for the simple reason that since 1881 the different processes for the manufacture of both iron and steel have been more or less completely revolutionised.

2. The researches of Rudeloff, published in 1902 (Mitteilungen aus dem königlichen technischen Versuchsanstalten, Berlin, 1902, xx., 83) must be placed in a dif ferent category, and are worthy of careful consideration. (The tables are compiled and calculated from tables 48 and 49, p. 176 of Rudeloff's memoir). Unfortunately the

CHEMICAL NEWS, Jan. 28, 1916

analyses of the different metals do not appear to be given, and it is therefore impossible to ascertain to what extent any variation in chemical composition may be responsible for the observed differences in corrodibility. Two sets of experiments were carried out, namely, with plates 0.5 cm. and 0.075 cm. in thickness respectively (see Table, preceding column).

Examination of the above tables reveals very considerable differences in the relative corrodibilities of the four metals in different media. The thinner plates exhibit on the whole a greater uniformity than the thicker ones; nevertheless, to take the extreme case among the former, the Thomas steel was found to corrode in smoke twice as rapidly as the other metals.

It will also be observed that the thinner steels differed among themselves quite as much as from the corresponding wrought iron standard.

Finally, it is interesting to note that there was little to choose between the mean corrodibilities of iron and steel in the case of the thinner plates, but the wrought iron had the advantage in the thicker plates.

By exposing the metals to blast-furnace gases, however, the wrought iron plates showed to an overwhelmingly great advantage. The results were as follows:

39

Relative corrodibilities. 100

77

Here, again, the steel has the decided advantage. Ira Woolson in 1910 (Engineering News, Dec., 1910) reported on eighty-nine samples of pipes from hot-water systems in New York; seventeen of these pipes were of wrought iron, the remainder being steel. No appreciable difference in corrosion was observed.

The succeeding year J. J. Wilson reported on an investigation carried out on the Cresson Coal Field, U.S.A., to the effect that the mean depth of pitting in twenty samples of wrought-iron that had seen good service was 0.094 inch, whilst that observed for an equal number of steel samples was o'093 inch. Here, again, the two metals appeared equally good.

In 1912 W. H. Walker (Journ. New England Water Works Assoc., 1912, xxvi., No. 1) gave the results of a similar study of iron and steel service pipes, which had been in constant use throughout New England. As the pipes had not been weighed previous to insertion in the water systems, their corrodibilities could obviously not be determined by weighing. The scale and rust were removed and measurements made of the depths of pitting. As the result of sixty-four comparisons it was found that

Iron corroded more than steel
Steel corroded more than iron
Steel and iron equally corroded
Corrosion negligible

20 cases 18 9 19

17

On the whole therefore there was no difference between the two metals.

The results of several other other researches have been

iron ..

Wrought

94

117

Weather (8 years). 100

105 133

Steel .. 103 Consideration of the above table shows that there is little difference between the corrodibilities of the iron and steel as exhibited in the tests of two years' duration. It is in teresting to note, however, that the steel shows to less advantage when the tests are prolonged to eight years, the corrosion factor for the steel being then 133, which, curiously enough, is the same as the mean value found by Parker (vide supra).

The authors point out, however, that mere loss in weight is not the only point to be considered in determining the relative corrodibilities of different metals. Of equal importance is the study of the pitting, because "if there is a hole the water will run out, no matter how much the pipe weighs." Plates of wrought iron and steel were therefore exposed to the action of hot aerated artificial seawater, and the depths of the pits produced were carefully measured with the following results::

Mean depths of deepest pits. Wrought iron o'028 inch (mean of 9 plates) Steel.... 0'017 inch (mean of 12 plates)

Relative corrodibilities. 100 61

Here the steel has the distinct advantage, so that from this point of view the relative corrodibilitics of the metals are inverted.

Howe and Stoughton also gave the results of an examination of twenty-nine pipes used in the signal systems of an American railroad, twelve of which pipes were steel and seventeen iron. Nineteen of these were practically destroyed by corrosion and pitting, the average life of the pipes being as follows:

interpretation.

As a matter of historic interest it may be mentioned that in 1881 the engineers-surveyors of Lloyd's Register possessed no fewer than 1100 marine steel boilers in actual service, and it was impossible to say which were the more satisfactory, the iron boilers or the steel ones. Even when a boiler of steel was worked alongside of one of iron in the same vessel, no material difference appeared to exist.

Howe ("The Metallurgy of Steel," 2nd ed., i., 101) mentioned in 1891 that he communicated with the leading British and American shipbuilders, &c., with a view to finding out their experience as to the relative corrodibility of iron and steel. The opinions received were almost equally divided between the two metals, and may be stated 7 firms

This is not necessarily due to any irregularity in the metals themselves, as one might be tempted to imagine, but in the majority of cases is attributabie to the fact that no one metal can be expected to offer an equal resistance to all types of corroding media.

Our task, therefore, resolves itself into a wider problem, not so much as to whether iron is better than steel or vice versa, but rather, Which is the best variety of iron or of steel for any particular purpose? It is conceivable that certain makes of wrought iron will prove most useful in certain circumstances, whilst under other conditions the palm will have to be given to steel.

(a) It may lose both constituents simultaneously and at the same rate. Both copper and zinc are then found in the corrosion product and in the proportions in which they are present in the original alloy. This is called "complete" corrosion.

(b) It may lose one constituent only, either copper or zinc. The corrosion product will then contain copper or zinc-not both. This is called "selective" corrosion.

(c) It may lose both constituents simultaneously but at different rates. The corrosion product will then contain both copper and zinc, but in a ratio which is different from that in which they occur in the alloy.

In addition to these three types of uniform corrosion localised action may occur at various points of the metal surface. This localised action may be "complete" or "selective." If the first, it produces a pit; if the second, it produces spongy copper or an alloy richer in zinc. Very frequently in practice these products of a selective local attack are worn away mechanically, and the final result is, as in the case of "complete" localised corrosion, the formation of a pit.

Some of the many factors which determine the character of a corrosive attack are:

(a) The composition of the alloy.

(b) The temperature.

(c) Aeration of the sea-water.

(d) The concentration of the sea-water.

(e) The catalytic action of oxysalts of zinc and copper. (f) The physical condition of the metal.

(g) Contact with electro negative substances, such as

carbon.

(a) The Composition of the Metal.-The rate of corrosion of copper in sea-water at ordinary temperatures is diminished by alloying it with zinc. This diminution increases rapidly as the proportion of zinc is increased until it reaches a minimum, when the alloy contains an equal number of atoms of copper and zinc. Indeed, the alloy containing 50 atoms per cent of copper and zinc appears to be almost unattacked by sea-water at ordinary temperatures.

When the alloy contains more copper than zinc, the proportion of copper in the corrosion product is rather more than in the alloy-i.e., the copper dissolves more rapidly than the zinc. It would seem that the rate of corrosion of pure zinc is diminished by the addition of copper in a correspondingly regular manner until it reaches a minimum of the alloy containing 50 atoms per cent of zinc. In alloys containing more than this amount of zinc the zinc dissolves rather more rapidly than the copper.

In every case the rate of corrosion falls off with time, and more rapidly at higher temperatures than at low. Consequently, although the initial rate of corrosion is greater at higher temperatures, the actual rate of corrosion after a given time will be smaller.

Analysis of the corrosion product at 15° C. and 50° C. indicates that at these temperatures the attack takes place in three successive stages. First there is a selective attack upon the copper; this is succeeded by a period of complete corrosion (after nine days at 15° C. and twenty days at 50° C.); the third stage is a selective attack upon the zinc. In other words, at the beginning of corrosion the rate of solution of the copper is much greater than that of the zinc. This gradually falls off; at the same time the rate of solution of zinc gradually increases. For a short time they are approximately equal; after that the rate of solution of the zinc is greater than that of the copper.

In stagnant sea water this third stage occurs more quickly at 15° C. than at 50° C. It is probably dependent upon the amount of O and CO2 present in solution in the It is hastened by aeration. It appears to be associated with the formation of a film of copper oxide which protects the copper but permits solution of the

sea-water.

zinc.

(c) Aeration of the Sea-Water.-At temperatures up to 50° C. aeration of the sea-water increases the initial rate of corrosion. At 50° C. its effect is very slight, while at 60° C. it diminishes the rate of corrosion. The effect of aeration appears to be greatest at low temperatures. This is no doubt associated with the well-known fact that rapid localised corrosion occurs just inside the inlet end of the tubes in the coldest part of the condenser. In every case it increases the proportion of zinc in the corrosion product.

The corrosion of pure copper in sea-water is checked by aeration owing to the formation of an oxide film. On the other hand, the corrosion of pure zinc in sea-water is considerably accelerated by aeration.

(d) Dilution of the Sea water. - Diluted sea-water attacks 70/30 brass more slowly than ordinary sea-water. At the same time the proportion of zinc in the corrosion product increases rapidly as the sea-water is made more dilute.

Thus after thirty-four days at 50° C. in aerated solutions of sea-water at the following concentrations, s/1, s/2, 5/4, s/8, s/16 (where s/1 is the concentration of ordinary seawater, s/2 being half that concentration, and so on), the percentage of zinc in the corrosion products was respectively 18.0, 24°2, 33′9, 44′1, and 58'5. In s/1 and 5/2 sea water, therefore, the copper in the brass dissolves more rapidly than the zinc, but in s/4, 5/8, and s/16 seawater the zinc dissolves more rapidly than the copper.

This is in agreement with the general belief that estuary waters are more selectively corrosive than the water of the open sea.

The solubility of pure copper in sea-water is diminished by dilution of the sea-water and is least in distilled water. The solubility of pure zinc in sea-water is increased by dilution of the sea-water, and is greatest in distilled water. appears from this that the dissolved salts are responsible for the solution of copper and the water and the dissolved gases for the solution of the zinc.

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

Dilution of the sea-water, therefore, while retarding the actual rate of corrosion of 70/30 brass, actually promotes dezincification.

(e) The Catalylic Action of Zinc and Copper Oxysalts. Moist zinc chloride at ordinary temperatures does not produce dezincification. When heated to 70° C., however, rapid dezincification is set up, owing to the formation of zinc oxychloride. When zinc chloride is added to seawater at 40° C., in which a piece of 70/30 brass is suspended, it produces a finely divided white salt, zinc oxychloride, which gives rise to rapid dezinclfication of the brass similarly at 50° C.

Aerated sea-water at 60° C. produces white spots on the surface of the 70/30 brass. These white spots are found to

CHEMICAL NEWS, Jan. 28, 1916

41

Sea-water when allowed to dry on a piece of 70/30 brass at ordinary temperatures forms a greenish blue crystalline salt which covers a pit. It does not appear to give rise to dezincification, but promotes rather the solution of copper.

Sea-water when allowed to dry on 70/30 brass at 100° C. forms a white salt which does promote dezincification.

At ordinary temperatures, therefore, the attack is greater upon the copper, and the copper oxychloride is formed. This salt catalytically facilitates the solution of copper. At high temperatures (between 80° C. and 100° C. at any rate) the attack is almost completely upon the zinc, and zinc oxychloride is formed. This salt catalytically facili

tates the solution of zinc.

In connection with this fact it is interesting to note the following experiment:—A tube of 70/30 brass was heated internally by steam. Cold sea-water flowed continuously round the outer surface. At various points of the surface air-bubbles were formed. Where these air bubbles appeared upon the surface of the hot tube rings of zinc oxychloride were formed, presumably by the evaporation of the thin film of sea-water in the presence of the air of the bubble. In some cases a succession of air-bubbles had followed one another at the same point of the surface, each producing a ring of white salt, and so gradually increasing the deposit of white salt. This might be a frequent cause of localised dezincification in condenser tubes.

(f) The Physical Condition of the Metal.-When 70/30 brass is hard drawn it corrodes less rapidly than when it is annealed. At the same time, however, there is a greater proportion of zinc in the corrosion product. Dezincifica. tion, therefore, occurs more readily in a hard-drawn 70/30 tube than in an annealed 70/30 tube.

(g) Contact with Electro-negative Substances.-Carbon or coke resting upon a clean brass surface is generally in good electrical contact with the brass. When good contact exists between the two the rate of solution of the brass is increased fivefold, and the proportion of zinc in the corrosion product rises to 60 per cent. At the same time the surface of the metal becomes covered with a loosely adherent red film of copper oxide, which can be removed by a mere touch of the finger. The effect produced by contact with coke is much greater at 50° C. than at 15° C. The e.m.f. between coke and 70/30 is about o'19 volt. Between coke and zinc there is an e.m.f. of about 1 volt.

Other substances, electrically neutral, such as string or brick, act as loci for the accumulation of oxysalts of Zn and copper, and so lead to accelerated local corrosion. A piece of coke fixed in a condenser tube might accelerate selective corrosion in three ways:—

(a) If contact were good it would exert an electrochemical action.

(b) By obstructing the free passage of the water in the tube it would lead to a greatly increased temperature along that part of the tube between the coke and the inlet end. This would facilitate general dezincification of that portion of the tube.

(c) The water on the outlet side of the coke would only be able to trickle along the tube, and the continuous film evaporation to which it would be subjected would, if the tube were hot enough, produce deposits of zinc oxychloride. This also would facilitate dezincification of the

tube.

Other cases in which corrosion, particularly selective corrosion, is localised at certain points along a condenser tube appear to be due to a lack of homogeneity in the metal itself. Perhaps local e.m.f.'s are set up between areas containing dissolved copper oxide and the remaining portion of the tube. The presence of strained metal in the hard-drawn alloy may be a contributory factor in the early stages of corrosion. It is generally noticed that the white spots of zinc salt, which indicate the occurrence of

upon the filed edges of the test-pieces.

A general consideration of the foregoing facts conveys the impression that a solid solution of zinc and copper is intermediate in character between a physical mixture and a chemical compound. The copper and the zinc retain to a very large extent their own characteristic individuality, but in each case it is modified by the presence of the other constituent. It is as though there were some kind of weak chemical combination (or rather physical combination, since the chemical character of the constituents is not changed entirely but only modified) between the two, whereby the available energy of each is diminished. In 70/30 brass the diminution of the available energy of the zinc is of course greater than in the case of the copper, for the influence which is exerted by a dissolved metal upon the character of the solvent metal is a function of the atomic volume. In the case of zinc and copper the atomic volumes are practically equal and the influence of each constituent upon the other becomes a function of the concentration. When, however, the proportions of zinc and copper are equimolecular the available energy of both is reduced to a minimum, and the rate of corrosion at ordinary temperatures becomes practically zero.

In the alloys containing more than 50 atoms per cent of copper it is the copper which dissolves at first, although zinc is more rapidly soluble in sea-water. This suggests that the chemical activity of the zinc is neutralised in the solid solution.

The electrode potential of zinc in sea-water at 16° C. That of is -0.5100 volt and of copper +0.3896 volt. 70/30 brass is +0.3460 volt, and therefore resembles copper rather than zinc. Similarly, the chemical activity of the copper is reduced considerably by the presence of the zinc in solid solution, and appears to reach a minimum when the atomic concentration of the zinc exceeds 50 per cent.

The strength of any combination between copper and zinc and the influence exerted by it upon the properties of the constituents would depend upon

(a) The composition of the alloy. By which is meant the concentration of zinc and copper in the alloy and the presence of other metals, such as iron and tin, and of nonmetals, such as oxygen.

(b) The physical state of the alloy, i.e., whether it is solid or liquid, amorphous or crystalline.

(c) The physical environment of the alloy, its temperature and pressure.

(d) The chemical environment of the alloy, which in this case refers to the nature and composition of the corroding solution.

Is it possible that a compound or compounds of copper and zinc exist stable only at low temperatures-say below 50° C.? These may dissociate into a system of solid solutions as the temperature rises and the composition is altered.

Such a conception appears to afford a reasonable explanation of the corrosion of 70/30 brass. Its application to other fields of investigation does not come within the scope of this paper.

Institute of Chemistry.-Pass List; January (1916) Examinations.-The results of the Examinations of the Institute of Chemistry recently held in London have now been published. Three candidates passed the Intermediate Examination, viz., H. E. Cox, B.Sc. (Lond.); A. J. Somer; and E. E. Wells, B.Sc. (Lond.). Nine candidates passed the Final (A.I.C.) Examination, viz. :-In the Branch of Mineral Chemistry-R. G. Browning, B.Sc. (Lond.) In the Branch of Organic Chemistry - R. Brightman; R. L. Brown, A.R.C.S.I.; H. S. Foster; and A. Hancock. In the Branch of the Chemistry (and Microscopy) of Food and Drugs, Fertilisers, and Feeding Stuffs, Soils and Water-C. E. Corfield; J. J. Geake; F. A. Pickworth, B.Sc. (Lond.); and Fred Smith.