No. 50, a new grade used chiefly for bacteriologi cal purposes. For Accurate Results Conduct your experiments with Whatman British-made Filter Papers. Extensively used for their reliability and high quality. No. 43, also Folded Filters and Fat Extraction Thimbles, will be STOCKED BY ALL LABORATORY FURNISHERS. In the event of difficulty in obtaining supplies or H. REEVE ANGEL & COMPANY, 15 NEW BRIDGE STREET, LONDON, E.C. HEMICAL NEWS, Estimation of Manganese in High-speed Steels THE CHEMICAL NEWS VOL. CXV., No. 2985. ON THE ESTIMATION OF MANGANESE By C. T. NESBITT, A.R.S.M. 61 (2) Ammonium Persulphate.-If the 500 cc. filtrate from IN the basic acetate separation is boiled down to about 200 cc. filtered if necessary, and then boiled again with 1 grm. ammonium persulphate, the manganese can be satisfactorily precipitated as the hydrated MnO2 with AmHO, if the persulphate is absolutely pure. When first tried the OWING to the difficulty of obtaining some reagents-more results obtained were all about o'I per cent too high. especially bromine on the outbreak of war, the question This was proved to be due to impurities in the persulof a good method for estimating Mn in high-speed steels, phate. Numerous tests gave approximately 3 mgims. without the use of bromine, became of importance. For impurities per 1 grn. of persulphate. This salt was the the last two years various methods and modifications of ordinary reagent as supplied by dealers. At present no methods have been extensively tested, and their advantages ammonium persulphate sufficiently pure to be used satisand disadvantages noted. These methods and modifica-factorily has been obtained; otherwise this method of tions will be explained and discussed in the present paper. oxidation would be eminently cheap and good. They apply to all high-speed steels proper; i.e., those method has been proved practically by estimating the imcontaining high percentages of tungsten and chrome, purities (SO2, Fe2O3, Al2O3, &c.) in the Mn304 residue tungsten-chrome-vanadium, or tungsten-chrome-vanadium- and deducting. molybdenum. They will also apply to the ordinary alloy steels of commerce. There are three main classifications of the various methods: (a) The gravimetric or basic acetate method. (b) The ZnO or CdCO, separation method. (c) The more or less direct bismuthate methods. (a) The Gravimetric Method.-It is unnecessary to go into the standard gravimetric or basic acetate method in any great detail, as it is well explained in many textbooks. Shortly, the process is as follows: Four grms. of steel drillings are dissolved in HCl. When in solution oxidise with 5 cc. concentrated HNO3. boil down to low bulk, take up with dilute HCl (30 cc. HCI and 70 cc. H2O), boil, filter off the tungstic oxide, and wash well with water. Make up filtrate to about 900 cc., and raise to boiling. Neutralise first with strong AmHO, and when near neutralisation finish with dilute AmHO or Am2CO3 till any further addition would bring down a precipitate. Add 50 cc. Am acetate solution, and boil. Make up to 1000 cc., and take the temperature. Filter of 500 cc. through a dry filter, adjusting the temperature of the filtrate. Boil down to low bulk-say, 100 cc.-filter if necessary, cool, oxidise with 2 cc. bromine for fifteen minutes. Add excess AmHO, and heat till MnO2 precipitate is coherent. Filter, wash, burn off at a high temperature, and weigh the residue of Mn304. Calculate to Mn. The (3) Chlorine.-The addition of KCO3 and HCl as a means of producing chlorine for oxidising the Mn is not altogether satisfactory. It appears that the ammonia salts (chloride, acetate, &c.) very often have the property of dissolving up the MnO2 first precipitated by AmHO. In fact, this means of oxidation is only applicable when the solution is moderately pure and does not contain a large amount of ammonia salts - a condition rarely fulfilled in the basic acetate filtrate. It can be used fairly success. fully by evaporating the filtrate to dryness with aqua regia to drive off ammonia salts, taking up with HC), and then oxidising by the addition of 3 grms. KCIO, heated with 20 cc. of 50 per cent HCl till Cl2 begins to come off. On heating, the Mn is oxidised and can be precipitated in the usual way by excess AmHO. It is obvious that this method is either unsafe or much too long. (4) and (5), H2O2 and Na2O2.-These oxidising agents appear to act very much in the same way as chlorine. It is necessary to have a moderately pure solution of Mn before oxidising, otherwise the precipitated MnO2 goes into solution again on heating. A large number of experi ments on basic acetate filtrates failed to find satisfactory conditions for oxidation of Mn by these reagents. Am Summing up, the alternatives to the bromine method of oxidation in basic acetate filtrates, with the exception of ammonium persulphate, are too unreliable for use. monium persulphate appears to be satisfactory, but must be absolutely pure or the Mn304 is contaminated. (b) ZnO or CdCO3 Methods.-These methods are more This standard method works well with every class of generally used in America. The procedure is as follows: steel-high-speed alloy or plain carbon. It is not an easy-Weigh out 4 or 5 grms. of steel drillings and dissolve estimation owing to the neutralisation point requiring a in 60 cc. 25 per cent H2SO4. Oxidise with 5 cc. conconsiderable amount of practice. When that is acquired, centrated HNO3. Boil down to low bulk and take up however, the rest of the procedure is simple and accurate. with 60 cc. 12HNO3. Filter off tungstic oxide and When bromine was almost unprocurable the following wash. Make up to about 400 cc., and add an em lsin of modifications 'vere tried :ZnO or CdCO3 till all the iron and chromium are thrown down. Make sure there is a slight excess of Zn or CdCO3. Make up to 500 cc. and filter off 250 cc. through a dry filter, making the usual temperature correction if necessary. Boil the filtrate down to one-half its volume. Add 10 cc. concentrated HNO3, cool, add sodium bismuthate, filter through asbestos, wash with dilute HNO3, and titrate with FeSO4 and KMnO4. (1) After the basic acetate separation the 500 cc. of filtrate was boi ed down with 10 cc. concentrated H2SO | to fuming. This was to decompose and drive off all ammonia salts, and especially all chlorides. The residue was diluted with 60 cc. 12HNO3, and the manganese estimated by oxidation with bismuthate and titration with FeSO4 and KMnO4-see (c) bismuthate methods. This modification was tested daily for several weeks against the standard gravimetric method and found to give results on the low side. The average is between 0 03 and 0:06 per cent low on a 0:35 per cent manganese steel. The method was discarded owing to the above and because it was found to be impossible to get out results in an ordinary working day, unless the gas (hot plate) was exceptionally good and hot-a consummation devoutly to be wished for, but not obtained under war conditions. It was found that better results were obtained by adding the emulsion in the cold, though every published account of the method advocates a boiling temperature. This is possibly due to the large amount of chromium present, as the method works equally well hot or cold on plain carbon steels. There seems to be little difference in the precipi tation properties of ZnO or CdCO3 emulsion; ZnO is, of course, always used owing to its greater cheapness. This method tends to give slightly lower results than the standard gravimetric. In a series of forty dup'icate analyses, sixteen gave results agreeing with the gravimetric, the remaining twenty-four giving results varying from 0 02 per cent to o o6 per cent low on a 04 per cent Mn. These low results might, of courte, be due partially to experimental error both in the gravimetric and ZnO methods, but all results varying over 04 per cent were again checked, and there were no results by the ZnO method appreciably higher than the gravimetric. Where results are not wanted nearer than 0.05 per cent, and these always on the low side, the method is a satisfactory one, and fairly easy and rapid to work. A somewhat similar method was published by Mr. J. A. Cashmore in the CHEMICAL NEWS, vol. cxiv., No. 2973. but he did not state whether it was applicable to high speed steels. The only obvious modification necessary would be replacing the KC.03 used for oxidation pur poses by HNO3, boiling down and filtering off the tungstic oxide precipitated, before proceeding with the Zao precipitation. Mr. Cashmore, instead of finishing by the bis nuthate method, oxidises with ammonium per sulphate, precipitates his Mn with AmHO, and finally weighs as Mn304. In this connection-see remarks upon ammonium persulphate previously explained. Attempts to oxidise with bromine and chlorine and precipitate the Mn with AmHO, in the filtrate from the ZnO separation were not successful. High results were obtained in every case. It seems to be impossible to wash out all zinc salts from the manganese dioxide precipitate. Again, all chlorides must be driven off, which is a function of the fuming with H2SO4; and, lastly, titration must be taken to the proper end-point. Of course, if preferred, the titrating solutions can be standardised in the actual assay solutions, so that the end point can be varied to suit the operator to a certain extent. However, if N/10 solutions are used, or results calculated to N/10 strength by factor and the end-point is taken to a permanent pink, the results are in absolute agreement with the gravimetric method. At first sight it appears to be somewhat absurd to dissolve up in HCl and convert to a sulphate solution when the steels will dissolve easily in dilute H2SO4. On testing, however, it will be found that taking up in HCI and oxidising and then converting with H2SO4 is far quicker; in fact, can be done in less than half the time it takes to get an original H2SO4 solution oxidised with HNO3 down to fuming. The H2SO4 solutions have a much greater tendency to spit, and must be evaporated exceedingly carefully and slowly. The aqua regia solu tion with H2SO4 added at the right time can be evaporated to fumes with little trouble. One other little point in the manipulation of this method which should be watched is in boiling off SO2; the estimation should not be boiled for more than a minute after all SO2 is off, otherwise there is danger of oxidising some chromium, with consequently high results. This last bismuthate method is by far the quickest of all the methods tested and reviewed. Six estimations can be done easily in under three hours, and leave a large portion of the time free for other work. Practically all the other methods take more than twice this time. ().ismuthate Methods.--Among the many modifications of the bismuthate method, supposed to be adapted for high-speed steels, there appears to be only one which is at all satisfactory, and even in this modification the conditions have to be very carefully observed, any slight deviation usually resulting in extraordinarily high results. Much work on this method has evolved the following ATMOSPHERIC CORROSION OF COMMERCIAL as being probably the best method of procedure: Weigh up II grm. steel drillings and dissolve in 12 cc. concentrated HCI. Add 5 cc. concentrated HNO3 and take down to syrupy consistency but not quite dry. Add exactly 7 cc. concentrated H,SO4, and wash once round the sides of the beaker with water. On shaking, the estimation goes to a thick cream on the bottom. Leave on edge of hot plate for twenty minutes, the beaker may then be placed in the centre of the hot-plate and fumed very strongly for fifteen minutes. Cool, take up with 30 cc. 12 HNO3 and 20 cc. water. Boil and filter off tungstic oxide. This oxide is somewhat slimy and difficult to filter, but if filtered on to pulp and the top surface broken up with a wash-bottle jet it can be washed four times or more in about five minutes. To the filtrate add 15 cc. concentrated HNO3, to bring the strength of the containing acid to roughly 12 sp. gr. Boil and add a pinch of bismuthate, clear with 10 cc. H2SO3. Boil off SO2 and cool the estimation quickly under the tap. Now oxidise with about 0'2 grm. Na bismuthate, shaking for two or three minutes. Filter through asbestos, using a filter-pump to ensure rapidity. Wash with 3 per cent HNO3 and titrate with N/10 FeSO4 and KMnO4. The ferrous sulphate is added till all permanganate colour is discharged, and then the pink colour is brought back again by means of N/10 KMnO4. The end-point must be taken to permanent pink, i.e., a pink that is permanent for at least five minutes. There is a fleeting pink tinge obtained some few points of a cc. before the real end-point. This must not be confused with the real end-point or results will be high. Under the above conditions I cc. N/10 FeSO, equals o 1 per cent Mn. This method is quite satisfactory if great care is taken concerning the following points: Tungsten should be thoroughly oxidised and removed; all chromium must be properly reduced-this is done in the first place by strong fuming with H2SO4, and in the later stages by addition of H2SO3. OBSERVATIONS UPON THE SHEET IRON.* By E. A. RICHARDSON and L. T RICHARDSON. THE Corrosion of iron and steel is a question that is becoming of increasingly great importance. The problem of finding the true nature of and the nitimate causes that underlie the rusting of iron is as yet unsolved. Much work has been done in the past, is being done at present, and before the problem is solved much more will undoubtedly be done in the future. The paper will not burden itself with a lengthy review of previous work that has been done upon the subject, but will confine itself to a brief summary of such work. As regards the relative merits of wrought iron and steel much work has been done, but observations on the two materials seem to be in poor agreement. As far as Of recent years the question of the rust-resisting properties of copper-bearing steels has arisen and remains as yet unsettled, the chief contenders being the makers of copper-bearing steel and those of pure iron. numerical data are concerned, the advocates of copperbearing steel seem to be in the lead; in fact, it seems to the writers that one weakness in the argument of the pure iron advocates is that they give little numerical data to support their claims. To many their arguments would carry much more weight if such data were included. Briefly, up to about 1910, a considerable amount of work had been done upon the subject of copper in steel, but this work was mainly in the form of fragmentary observations on steels that contained small quantities of copper either accidentally or intentionally added, and a Previous few actual tests upon copper bearing material. to this time there appeared to be no acute controversy as to the effect of copper in steel upon corrosion, although Paper presented at the 30th General Meeting of the American Electrochemical Society, New York City, September, 1916. the Chemical Engineer and Manufacturer, xxiv., No. 4. From CHEMICAL NEWS, Atmospheric Corrosion of Commercial Sheet Iron. judging from the literature it did exist at this time to a rather mild extent as to the effect of this metal upon the physical properties of steel. The general impression one would gain from reading these articles, is that copper in steel is a benefit as regards resistance to corrosion. Shortly after 1910 copper-bearing steel was put upon the market as a corrosion-resisting material. The controversy then became acute, the chief objectors being the advocates of pure iron. The result has been that a considerable amount of work has been done and published without settling the question. Most of this work seems seems to have been done by those interested in copper bearing steel in their efforts to prove that this material is the real rust resisting materia they claim it to be. A la ge share of the published work, which contains results from tests closely approaching actual service conditions, has been done by Mr. D. M. Buck-a strong advocate of copper-bearing steel ("Copper in Steel, the Influence on Corrosion," Fourn. Ind. and Eng. Chem., June, 1913; Buck and Handy. "Research on the Corrosion Resistance of Copper Steel," Journ. Ind. and Eng. Chem., March. 1916). The opposing side appears mostly in discussion of these and other papers, and appears to be taking a defensive stand. The impression gained by the writers, from reading the available published evidence, is that copperbearing steels do possess rust-resisting properties and are sup rior to any iron or steel now on the market. Likewise, the influence of mill-scale upon the corrosion of iron is under discussion. Friend, for instance, states that "traces of oxides upon the surface of metals are powerful stimulators of corrosion ("The Corrosion of Iron and Steel," p. 251). A. Sang says:- Black oxide only protects provided it is continuous and firmly anchored to the iron (BowerBarffing, &c.); as mill-scale, which is loose and fissured, it is detrimental, the iron in contact with it and exposed rusts about 50 per cent faster " ("The Corrosion of Iron and Steel," p. 81). 63 mercial quantities. Such tests do not carry the most weight since they do not have the air of disinterestedness that exist in the case of materials purchased upon the open market. In fact, one of the arguments of the pure iron advocates is the objection to "special tests" and the demand of basing comparisons upon service tests only. In view of the fact that no results had been published of a complete test on commercial materials, it was deemed desirable to make a test containing all the types of iron and steel that could be obtained. Some of the kinds of materials that are commonly used for structures, exposed to atmospheric corrosion, are:1. Steel. 2. Puddled pure iron. 3. Commercially pure iron. This paper deals with the first four classes. Accordingly there were obtained upon the open market the following materials in the form of black sheets of 26 gauge : 1. Bessemer steel. 2. Open hearth steel, 4. Commercially pure iron. 7. Copper-bearing steel. 8. Copper-bearing Bessemer steel. Sulphur. Silicon. 0.054 0'020 Man Copper. ganese. Carbon. I. Trace 0.300 Ο ΟΙ 2. Trace 0413 Ο ΟΙ Phosphorus. 0.087 0.079 Aston and Burgess find "that mill-scale had an accelerating effect for an atmospheric test and a retarding action for a fume test." Further they state that "there is no doubt that the protective property of mill-scale is dependent upon its physical properties, continuity, &c." ("The Rate of Rusting of Iron and Steel," Eighth Int| 9. Congrass of Applied Chem., xxvi., 453). G. C. Wnipple and M. C. Whipple tested steel, ingot iron, and wrought iron with mill-scale, and with mill-scale removed, and obtained erratic results, but, generally speaking, the rusting was slightly greater in the case of the metals from which the mill scale had been removed than in the cases of the metals on which the mill-scale had been left. They state, however, "that the best remedy to protect steel from pitting is to remove mill. scale" ("Mill-scale as a Cause of Corrosion," Eighth Int. Congress of Applied Chem.). Buck and Handy in their latest paper ("Research on the Corrosion Resistance of Copper Steel." Fourn. Ind. and Eng. Chem., March, 1916) state "at the time of exposure of the full-sized corrugated sheets careful notes were taken concerning the physical appearance of the sheets as affected by the amount of mill-scale which was present, and as to whether they were outside or inside in the pack. During the progress of the test this feature was carefully watched, and the time of ultimate failure of sheets whose surfaces were comparatively free from mill-scale was com pared with others of the same grade, whose surfaces were well covered with mill-scale. From these observations we concluded that the influence of this original surface oxide is slight, and is lost in the early stages of rusting, for no difference in final failure could be noticed." One thing, however, in regard to many tests that have been made up on the relative corrosion of iron and steel, is that in so far as the ultimate consumer is concerned, they are not of the greatest value, since many times only two or three materials were compared, or the materials were made especially for test, and cannot be obtained in com Ο ΟΙ 0:08 0 004 0 092 Before starting a work of this character, which involved quite some labour, due consideration was given to the methods of testing and in particular to those methods about which there is some difference of opinion. Questions arose concerning the desirability of a previous heat treatment, and the necessity and means of removing mill-scale previous to testing. Opinions differ as to whether samples of iron should be tested just as received, or whether all test-pieces should receive some treatment so that the physical properties will be as nearly identical as possible. One author contends that samples should be tested as purchased, since the material that the samples represent will be used just as purchased. On the other hand, there are those who contend that all samples tested should have as nearly the same treatment as possible. They suggest, therefore, that the test-pieces be annealed at the same temperature and that mill-scale be removed. It would seem that the method used for testing should depend upon the results to which the data is to be put. If the results are intended for the ultimate consumer the material should be tested just as purchased, since this is the condition that it is used by the ultimate consumer. On the other hand, if the data is for scientific purposes, such as determining the effects of various elements in the material, all variables other than the ones under investigation should be removed if it is possible to do so. However, to make the work as broad and as complete as possible, it was decided not to limit the test to one method, but to try out both ways and to determine the relationship between the two. Since part of the specimens were to have the mill-scale removed, the question arose as to what method should be used for removing this scale. In previous published work, very little has been said about the method of cleaning test specimens. When stated, however, it has usually been done by mechanical means, such as filing or grinding, or by chemical means such as acids. Experiments were therefore conducted on the effects of methods of removing mill-scale upon the subsequent corrosion. The results of these experiments have already been published (E. A. Richardson, "The Effect of Pickling on the Corrosion of Iron," Met. and Chem. Eng., xii., 759). It was shown in this paper that methods of cleaning by the use of chemicals had a profound effect upon the rate of corrosion, especially on tests of short duration. It In the conten plated test a clean iron surface was necessary in which the surface iron was in the same physical and chemical state as the iron beneath the surface. was therefore decided that removal of the mill-scale by pickling in sulphuric acid, thorough washing with water and drying, and subsequent removal of the surface iron with fine emery cloth would give a surface that was identical with the material itself. The question also arose as to the method by which corrosion was to be estimated. There is a choice of methods to determine the rate of corrosion. One method com. monly used is to determine the loss in weight after exposure for a given length of time and considering this loss as an index of corrosion. Another method is to allow the test specimens to corrode until failure occurs, and to take the length of time as an index of corrosion. Inas much as there is some question as to the reliability of the "loss in weight" method, it was thought best to continue the test to the failure of the specimens. This made it necessary to decide upon some standard condition at which failure could be considered to have taken place. Evidently, failure should not be taken at a point when the iron has completely rusted away or even when it has almost disintegrated, since any sheet iron in service would be of no value long before such a condition obtained. The point of failure selected was when the sheet iron was in such a condition that it could be seen to be p-rforated when the rust was removed by gentle tapping with a blunt object, such as a file or nail. By using one specimen of each kind of material as a test piece that was tapped and examined at intervals of about two weeks, a close approximation to the time of failure could be obtained. Thus the other specimens were not molested until the latter part of their useful life and the effects of taping were practically eliminated from the results. By this method it is believed that the life of each material was obtained to within an error of three or four weeks. The sheets of 26 gauge iron were accordingly prepared for test by cutting into pieces 5 in. by 8 in. (12.7 cm. by 20 3 cm.) For the test on the material as received ten pieces of each kind of iron, that was found by measurement to be of standard thickness, were used. For the test on the prepared material a number of pieces of each material were annealed at a red heat and then cleaned by the method already described. Great care was taken to give all of these test pieces exactly the same treatment. Ten of the prepared specimens of each material that were found by measurement to be of the same thickness were taken for test. These prepared pieces were of practically the same thickness as the "as received" samples tested, due, no doubt, to the extreme thinness of the mill-scale. On account of the precautions taken as regards uniform thickness of materials, the method of estimating failure, and the method of cleaning surface, it is believed that the test should be comparative. All samples were then placed securely in a wooden rack, and exposed to the weather May 24, 1914. The atmosphere was what might be described as a semicountry one and cannot be taken as at atmospheric acid test. At the very start a marked difference was noted in the character of the rust formed on different materials. Whereas, on the Bessemer and open-hearth steel samples, the rust was of a yellowish red colour and became loose rapidly, the rust on the others was dark red in colour and much more adherent. This adherent state seemed to reach its maximum in the copper steels, where the rust was very dark and fine grained. The life of the different materials tested is given below; each figure being an average of ten test pieces, and represents the time for failure in days : Material as received. - Material annealed and surface cleaned. From visual observations at this time there appeared to be little difference between pure irons with or without additions of copper. At the present writing the three samples of copperbearing steel are still in good condition. The rust is still adherent and the underlying metal is still intact. The life of these steels has been estimated at 1200 days, which figure the writers believe is very conservative. These specimens are still exposed to the weather, and will be until failure occurs. Conclusions.-Taken as a whole the results indicate that copper-bearing steels are, without doubt, decidedly superior to any of the other materials tested. The remaining materials may be divided roughly into two classes, one class including the ordinary steels (Bessemer and open-hearth), and the other class the commercially pure irons and the copper bearing irons. Charcoal iron is classed with the pure irons. We have obtained results in another test which would indicate that the resistance of wrought iron to corrosion is due to the purity of its iron and not to slag inclusions. In addition it was noted during the present test that charcoal iron appeared to corrode in very much the same way as pure iron. In regard to the steel-iron question it is noted that the pure irons (including charcoal iron) are superior to steel. It is believed that this superiority is due to the purity of the iron or to some combined effect of manganese and copper. The results obtained in regard to the effects of millscale do not agree with the results obtained by others. They indicate that with steels which rust rapidly, millscale is a stimulator of corrosion. On the other materials tested the mill-scale has exerted a protective action. We have no explanation for this. The important feature of the results, however, is that it substantiates the claim made in several other publications that the addition of copper to steel in amounts of about o 25 per cent causes a remarkable increase in its ability to resist atmospheric corrosion. In the present test the addition of about 0.25 per cent copper to Bessemer or openhearth steel has resulted in an increased resistance of 300 or 400 per cent. The addition of copper to pure iron also results in an increased resistance to corrosion, but to no such an extent as the addition of a similar amount to steel. The addition of about 0.25 per cent copper to a commercially pure iron results in the useful life being increased by about 20 per cent. This figure was arrived at by comparing the average life of irons Nos. 3, 4, and 5 with No. 6. We have been unable to determine the reason for this effect of copper in reducing corrosion. It is believed, in view of the fact that the coprer exerts a greater influence in steel than in iron, that it must be due to the combined presence of copper and manganese, since the chief dif |