|
1. |
The Faraday Society |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 003-004
Preview
|
PDF (103KB)
|
|
摘要:
The Faraday Society was founded in 1903 to promote the study of Pure and Applied Electrochemistry, Electrometallurgy, Physical Chemistry, and 3fetallo- graphy . The subscription to the Society is &2 a year for Members, and &I a year for Students. Members also pay an Entrance Fee of LI, with the exception of Members of the Chemical Society, the Institute of Metals, the Institution *of Elec- trical Engineers, the Iron and Steel Institute, and the Physical Society, who art. admitted without Entrance Fee and without the usual formal nomination. The Society publishes quarterly or half-yearly Trurtsactiorts, containing in full Papers which have been read with the discussions thereon, and reviews of books, and Pyocecdiizgs, containing Reports and Notices of Meetings, with abstracts of the Papers read during the previous month and of the discussions thereon.Under an arrangement between the Faraday Society and the American Electro- chemical Society- ( I ) The Transactions of the American Electrochemical Society are sent without charge to all Members of the Faraday Society. ( a ) The Transactions of the Faraday Society are sent to all members of the Amerkan Society, so that Papers published by the Faraday Society have a circulation of over 1,500 copies in America, in addition to the British citculation. It is the practice of the Society to print all Papers before they are read at a meeting ; advance proofs are sent to every Member, thus increasing greatly the value of the discussions, and giving absent Members an opportunity of expressing their views by correspondence.One important feature of the work of the Society is the organization from time to time of a General Discussion on some subject of special interest. The Discussion is usually opened with a Paper summarizing' the existing state of knowledge by some one exceptionally conversant with the subject, and every effort is made to secure representative opinions both by verbal contributions at the meeting and by correspondence. Reprints of the General Discussion on THE HARDEXISG OF METALS appearing in the present volume can be obtained from the Secretary, price 7s. 6d. each. Reprints of the collection of Papers on THERMOSTATS which appeared in the June 1912 issue of Transactions can also be obtained, price IS. 6d. each. Reprints of the Discussions held in 1914 on OPTICAL ROTATORY POWER, in 1913 on COLLOIDS AND THEIR VISCOSITY and on THE PASSIVITY OF METALS, and in 1913 on MAGNETIC PROPERTIES OF ALLOYS, are to be had, price 5s. each. Reprints of the Discussions held in 1910 on THE CONSTITUTION OF WATER, and in 1907 nn OSMOTIC PRESSURE and HYDRATES IN SOLUTION, are to be had, price 2s. 6d. each.Notice to Advertisers are The Sole Advertising Agents for the Transactions of the I FARADAY SOCIETY Electrical Press, Limited, Advertising Agents and Publishers, 13, 14, I 5 &? 16, FISHER STREET, Near KINGSWAY, LONDON, W. To whom all applications for Advertisement space and rates should be addressed. Telephone : 2012 Holborn. Telegrams : " Farsighted, Westcent, London." I
ISSN:0014-7672
DOI:10.1039/TF915100X003
出版商:RSC
年代:1915
数据来源: RSC
|
2. |
The anodic behaviour of alkaline borate and perborate solutions |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 177-196
Wilfrid Gustav Polack,
Preview
|
PDF (1389KB)
|
|
摘要:
The Favaday Society is not resportsible f o r opiiiioris cxpvessed before it by Authors or Speakers. OF F O U N D E D 1903. TO PROMOTE THE STUDY OF ELECTROCHEMISTRY, ELECTROMETALLURQV, CHEMICAL PHYSICS, METALLOGRAPHY, AND KINDRED 8UBJECT8. VOL. x. MAY, 1915. PARTS 2 AND 3. THE ANODIC BEHAVIOUR OF ALKALINE BORATE AND PERBORATE SOLUTIONS. BY WILFRID GUSTAV POLACK. ( A Paper contributed to the TRANSACTIONS of the Favaday Society, September, 1914.) Perborates were first prepared in the pure state in 1898 by Melikoff and Pissarjewski,:: and by Tanatar.t The one commonly prepared is the sodium salt NaBO, * 4H,O, this formula having been established by the above authors and confirmed later by Constam and Bennet,: who concluded from conductivity measurements that it is a salt of a monobasic acid, HBO,.It contains 10.38 per cent. of active oxygen, corresponding with one atom of oxygen in the formula, sodium metaborate being formed by its reduction. It is generally prepared by the action of H,O, or Na,O, on borax solutions, and on account of its oxidizing properties it finds increasing commercial application for blcaching and for laundry work. For such purposes an electrochemical method of preparation would be of value if it could be arranged as a continuous process, after the manner of the hypochlorite electrolysers in use, fresh bleaching solution being drawn off as required and* spent liquor returned to the cell, and such an apparatus has been described in a French patent, to which further reference will be made. On this acconnt an investigation of the electrolytic preparation of perborates appeared to offer an interesting field, especially as the only evidence hitherto available regarding the possibility of preparing perborates electrochemically is of a somewhat unsatisfactory nature.Tanatar 5 states that he obtained sodium perborate by electrolysis of a concentrated solution of sodium orthoborate, but gives no details of the process. Later Tanatar 11 states that perborates are decomposed by the passage of the electric current. Constam and Bennet 4 describe an experi- ment in which the anolyte, consisting of 20 grams borax and 4 grams NaOH in 60 C.C. of water, was contained in a Pukall cylinder which dipped into 10 per cent. NaOH as catholyte. They passed 0.5 ampere for 4 hours at a temperature * Rev., 31, 678 (1898).$ %. Aiiorg. Chevi., 25, 265 (1900) 11 2. Plzys. Clzenz., 29, 162 (1899). VOL. X-T7 -f Z . Plcys. Claenz., 26, 132 (1898). $ LOC. cit. 7 LOC. cit. _ -178 THE ANODIC BEHAVIOUR OF ALKALINE of 4” to 6-5”C., but obtained no perborate, or at most a very small quantity (which they called H,O,), the oxygen evolved being practically equivalent to the total current. Like Tanatar, they omit all mention of such important factors as electrode material, current density, etc. Bruhat and Dubois :: state that sodium perborate can be obtained by electrolysis of a solution of sodium orthoborate, but it is not clear whether they mean to imply that they themselves have prepared it by this method. The onlyother evidence is that of a French patent by Pouzenc,+ and two articles thereon by Beltzer 1 in French technical publications.The full details of these were not available when the present work was commenced, but it was known that the apparatus described consists of a vertical electro- lyscr divided into anode and cathode compartments by a vertical diaphragm, the electrodes being of polished platinum. The electrolyte used for making bleaching solutions is said to contain 24 grams borax + 6 grams NaOH per litre, and is cooled in a reservoir and allowed to flow through the cell, the current density being about 30 to 60 amperes per sq. dm. Later on an opportunity occurred of consulting the original patent and the papers referred to, and they will be discussed in the appropriate place below. As regards the mechanism whereby sodium perborate might be produced electrolytically, Tanatar suggested oxidation of the borate by H,O, formed at the anode by discharge of OH’ ions thus :- 2 0 H ’ -++ H,O, + 28.Riesenfeld and Reinhold, $ however, have shown that even under what they consider the most favourable conditioiis for this reaction (high alkali concentration, low temperature. and high current density) no H,O, is obtained at the anode from NaOH, and with KOH at 0°C. the concentration only reached I/I,OOO molar after electrolgsing for 2 hours with a current density at the anode of 33 amperes per sq. dm. Constam and Bennet make the remark- able statement that, since sodium perborate undoubtedly has the simple formula NaBO,, one would not expect it to be formed electrolytically, because it could not be formed like persulphates or percarbonates by the union of discharging anions.Apparently they were unacquainted with the simple process of anodic oxidation, and there appears to be no good reason why, if sodium perborate can be produced electrolytically, it should not be by simple removal of negative charges, thus :- Anode- Cathode- BO,”‘ --+ BO,’ + ze, 2 H 2 0 + 3 2 0 H ’ + H, $ 2 0 ; BO;” + 2H,O+ BO,’ + 2 0 H ’ + H, ; together- or, writing in the sodium- Na,BO, + zH,O -++ NaBO, + 2NaOH + H, as the total result of the electrochemical process. The removal of negative charges might, of course, take place actually through the intermediate action of an “ oxygen-charged ” electrode, and not directly, /I as shown. Besides t French Patent, No.411,258 (April, 1910). Elekfroclzem., 18, 189 * Conzpt. Rcizd., 140, 506 (1905). $ Revue d’Electuochimie, 5, 1 (1911) ; Lc J f o n i t e w Scieratifiqrre, 1911, p. 10. /I Cp. Grube on oxidation of ferro- to ferri-cyanides, Z . B He),., 42, 2977 (1909). (1912).BORATE AND PERBORATE SOLUTIONS I79 the loss of negative charges the BO,"' ion would presumably undergo a struc- tural change thus :- B-O- / O - -+ O=B-O-O- \O- Perborate ions might also conceivably result from anodic oxidation of various borate ions other than BO,"'. Whatever the mechanism, the conditions to be aimed at are those which lead to a high anodic potential, viz., polished platinum anode, at which the overvoltage is known to be higher than at any other, low temperature, and high current density.Low alkali concentration, by diminishing the tendency for discharge of hydroxyl ions, will also favour a high potential. Also, of course, a reasonably high borate concentration is desirable. In practice the simultaneous realisation of these conditions is somewhat difficult. Detection aiad Estimation of Perborate. Sodium perborate in solution can be titrated in presence of I-I,SO, with permanganate, and behaves like H,O, towards this reagent, two molecules of KMnO, corresponding to five of NaBO;qH,O. A sample obtained (as a white powder) from Kahlbaum w'as found on titration with KMnO, to contain only about 87 per cent. of NaBO, - ,+H,O, which, however, on recrystallising, increased to 98 per cent. Numerous titrations with varying quantities of perborate and acid showed the perinanganate method to be sufficiently reliable.In a 2 per cent. solution the perborate was found to decompose spontaneously at the rate of about I per cent. pcr hour at ordinary temperature. Cooling Arrangements. I t is difficult to arrange ail apparatus with the anode and cathode as close together as would be desirable, while still separating them by a suitable diaphragm and accommodating the necessary thermometer, etc., and the resistance due to the comparatively low conductivity of the solutions is further increased by the diaphragm and the gas evolution at the electrodes. It was therefore necessary in practically all the experiments, in order to keep the anolyte at the low temperature desired, to use a somewhat elaborate arrangement. A cooling liquid at - 16" C.to - 1 8 ~ C. was prepared in a large cylindrical metal tank of 24 litres capacity by mixing crushed ice and salt with the necessary amount of water. The metal tank was contained in an outer vessel of earthenware standing on the floor. The liquid was sucked by means of a water-pump through a wire-gauze strainer, and then through the cooling coils in the electrolysis cell into a large glass bottle standing on the bench, with taps so arranged that from an outlet near the foot of the bottle the liquid could, when desired, be run straight back to the tank for further use, without interrupting the flow of cooling liquid through the apparatus. First Form of Apparatus. The first cell employed consisted of an outer glass vessel which' contained ithe catholyte, and in this glass vessel stood a porous pot':: which contained the anolyte.In order that the gas evolved might be collected, the pot was closed with a rubber stopper fitted with a delivery tube, and fixed through the stopper were also a thermometer] the ends of a glass spiral for cooling the anolyte, an inlet for running in lWaOH from a burette, and the anode itself. The last consisted of strips of platinum wire or foil of suitable size to * The porous pots and tubes used were made thinner and more uniform by grinding them on the lathe with emery.180 THE ANODIC BEHAVIOUR OF ALKALINE give the desired current density, and bent outwards so as to lie near the cooling spiral. The cathode consisted of a cylindrical sheet of nickel encircling the pot, leaving only sufficient space between for gas to escape freely.The top of the pot where it projected out of the liquid was waxed to make it airtight. Coinpositioii of the A iiolyte. It appeared desirable in the first instance to try the solution of Pouzenc referred to above, viz., 24 grams borax + 6 grams NaOH per litre. Such a solution, however, was found to have too low a conductivity, the consequent heat evolution being excessive, and solutions were therefore used which contained more NaOH. As pointed out above, however, it is desirable, in order to get a high anodic potential, to keep the alkali concentration as low as possible, but suitable solutions cannot be obtained by merely dissolving larger amounts of borax in water, because IOO grams of water dissolve at oc C.only 2*S3 grams of borax (Na,B,O; 1oH,0) and at IOO C. 4-65 grams. Potassium borate was found to offer no great advantage in this respect, but by adding KOH a solution could be prepared which contained 27-4 grams K,O and 50.7 grams B,O, per litre (against 8.8 grams per litre of B,O, in the Pouzenc solution), and such a solution, slightly diluted, was used in one case. Similarly with borax the concentration can be increased by adding NaOH (cp. Constam and Bennet, above). Thus a solution containing 80 grams borax+20 grams NaOH per litre (29.3 grams per litre of B,O,) does not deposit any solid at oc C., but in the earlier experiments with such solutions less ozone was produced than in other cases, and this was taken as an indication that the anodic potential was lower, and the conditions therefore less favourable, the alkali concentration being apparently too high.Later on it appeared that other factors are probably of greater importance, and that, moreover, the ozone production is a rather uncertain quantity and depends on various things. Of course in one respect an increase in the proportion of alkali might possibly be advantageous by increasing the concentration of BO;" ions. Apparently, however, the number of B03"' ions present in any alkaline borate solution is very small, the chief ion being probably H,BO,'. The composition which would nominally correspond to orthoborate would require, roughly, equal weights of borax and NaOH, but such solu- tions are very strongly alkaline owing to hydrolysis ; in fact, according to Kahlenberg and Schreiner + practically all the excess of NaOH above that corresponding to metaborate is present in the free state, the solution being therefore no richer in BO;" ions than a less alkaline one.The first few experiments showed that it was necessary to add a little SaOH to the anolyte from time to time to compensate for that removed by the electrolysis, otherwise the voltage across the cell rose rapidly. In some cases it was even necessary to add a little at the commencement in order to reduce the high resistance and consequent great heat evolution. Experiments witlz First Fornz of' Cell. Some experiments with the first forin of cell are briefly summarised in Table A. The first two columns give the composition of the anolyte in grams per litre of borax and caustic soda.The symbols used are as follows throughout the paper :- T =Temperature of the anolyte. C =Current through cell (in amperes). I)*= Current density at anode in amperes per sq. din. V = Voltage across cell. * 2. P h ~ s . Clrcrir., 20, 547 (1896).BORATE AND PERBORATE SOLUTIOXS I81 29gm. 18gm. - I 2 3 4 - 40 to 5" C. i 4-0 ' 22 8.8 to 10.4 45 min. -20 C. 4.0 1 73 10.4 35 min. oo to -10 C. 3.3 ~ 60 11.2 to 8.0 2 hr. ~ I hr- - I O to - 2 O C. 3.5 1 106 11.6 to 9'6 TABLE A. I Borax. NaOH. 1 T. V. 1 Time, The catholyte in each case was 4 per cent. NaOH. These and other similar experiments (in one case using potassium borate) gave negative results. The rate of evolution of oxygen was measured occasionally during each experiment and always corresponded as closely as could be ascertained to the total current.After each experiment the aiiolyte was tested with -I 0 FIG. I. KMnO,. In no case was there any oxidising action except in No. 3, when 10 C.C. of anolyte decolourised one single drop of N / 4 KMnO,. The only hope of obtaining more favourable conditions appeared to lie in the employment of a cooled anode. Although the bulk of the liquid in some of the above experiments was at a temperature in the neighbourhood of zero, the portions surrounding the anode might be at a considerably higher temperature owing to the high anodic current density, and it seemed that by using a cooled anode this condition might be more or less reversed, the liquid in contact with the anode being kept at a sufficiently low temperature, while the bulk of the electrolyte need only be kept a t a moderately low tem- perature.Thus Prausnitz:;: was able to obtain very satisfactory yields of * Z. .f. Elcktrochcnl., 19, 676 (1913). .182 THE ANODIC BEHAVIOUR OF ALKALINE hypochlorite by employing a cooled anode, even when at the same time the bulk of the liquid was actually warmed to reduce the potential drop. Moldenhauer,* who gives a theory of heat evolution at,the electrode surface, obtained improved yields in various electrolyses. Cooled anodes have also been used in the preparation of persulphuric acid from sulphuric acid for the manufacture of hydrogen peroxide. The anode used in the apparatus to be described consisted of a platinum tube t 20 cm, long, 3 mm. external diameter and 0-17 mm.in thickness, the area of the portion actually immersed in the electrolyte being approximately 8 sq. cm. Cooled Anode Ap,barattis. 3 x 12 cm., held in place by the rubber washer W. The electrolysis cell is shown in Fig. I in section. P is a porous tube of S is the cooling spiral, FIG. 2 . A the anode, T, a thermometer, and C the cathode, a cylindrical sheet of nickel as before. The outer glass vessel served to contain the catholyte, The anolyte flowed in at I and gas and liquid came out at 0. The vessel V contained the anolyte (generally about 100 to 125 c.c.), the tap regulating its flow to the anode compartment. Gas and liquid collected in K, the former passing on through the delivery tube, while the latter could be drawn off at D and returned to V, a continuous circulation of the anolyte being maintained.A burette containing NaOH was connected at L, a few C.C. being run in as required. The cooling liquid, which entered at the foot * 2. f. Electrocltenz., XI, 307 (1905). t An attempt was made to use an anode of platinised brass tube. The platinising was carried out with the citrate solution of Langbein (vidt~ Schlotter’s GuEvaizos- tcgie, p. 187), the tube being rotated in the liquid. It was found possible by this means to obtain an excellent platinum surface, which could be well polished and was highly resistant mechanically ; but unfortunately when used as anode in alkaline solution at high current density it rapidly disintegrated, cupric hydroxide being ultimately found in the liquid. The general arrangement of the apparatus is shown in Fig.2.BORATE AND PERBORATE SOLUTIONS I S3 as shown by the arrows, was divided between the anode and the glass spiral, the proportion being regulated by the screw clips M. No electrical leakage took place through the salt solution flowing through the anode ; in some cases two ammeters were connected in the circuit, one on each side of the cell, and these agreed perfectly. The thermometers T, and T, give the temperature of the liquid entering and leaving the anode." The mean of these is the nearest indication that can be obtained of the actual temperature of the anode and is given as TA below. The other letters below have the same significance as before. The catholyte was 4 per cent. NaOH in each case. The composition of the anolyte is stated at the Commencement in each case. T is thc temperature of the anolyte, as given by thermometer T,.E.rperimerats with Cooled Anode Apparatus. I. So grams borax -+ 20 grams NaOH per litre.- c = 4'8. = 60. v = I8 to 20. TA = - I I" C. to - 70. T = 4" to 6" (finally 150). Time, 14 hr. Ozone production very slight. 2 . 3s grams borax + 40 grams H,BO, per 1itre.f- c = 4'8. DA = 60. V = 25. T, = - 1 0 ' 5 ~ to - 5". T = 13" to 20'. Time, I& hr. This anolyte contained more B,O, than No. I (37 grams per litre against 29*3), but was much less alkaline ; at the commencement it was almost neutral, but was kept faintly alkaline throughout. Although such a solution contains complex anions, it would presumably be capable of furnishing some BO;" or other ion necessary for perbordte formation.Evolved oxygen smelt very strongly of ozone. 3. 28 grams borax -+ 5 grams NaOH per litre.- C = 2-4. Da = 30. v= I8 to 15. T,= - 13" to -- 100. T = 4" to oo. Time, I+ hr. Ozone production slight. Attempts were also made to employ much higher current densities, but as the current through the cell could not be increased without undue rise of temperature, it was necessary to cover up portions of the anode by means of closely fitting glass tubing, and the uneven current distribution and cooling gave rise to difficulties. In one case, however, it was found possible to carry on the electrolysis for I hour at a current density of 200, Ta being - 7.5" and T about 20". The above and all similar experiments also gave negative results, except that in No.3 the resulting anolyte showed faint oxidizing action, 50 C.C. decolourising I drop of ~ / 4 KMnO,. It appeared therefore desirable to ascertain whether, if sodium perborate were present in the anolyte at the commencement of such an experiment, its concentration would actually diminish. * The difference between these varied with the rate of flow, but never exceeded 3.5" C. t McLaughlan, 2. Ariorg. Chnii.. 37 (1903), 371.184 T H E ANODIC BEHAVIOUR OF ALKALINE The following experiment was therefore carried out. 4. 28 grams borax + 5 grams NaOH per l i h c + some perboyate.- c = 4'8. D, = 60. V = 30. 'rx = - 11-50 to -4.50. T = 15" to IIO. Time, I hr. Samples of the anolyte were titrated with KMnO, solution before and after 10 C.C.of anolyte required :- the electrolysis. Before, 2.7 C.C. of ~ / 4 KMnO,. After, only 2 drops of N / 4 KMnO,. Throughout the experiment no ozone could be detected, except a faint trace at the very end. This experiment was repeated in another form, the perborate being added during the electrolysis instead of before the commencement. 5. 3jgrains borax + 5grams NaOH per litre.- C = 4% DA = 60. V = 35 to 40. At comniencement- After current had been passing for 10 minutes the perborate (0.5 grain dissolved in 35 C.C. of water) was added, T A being - 6.50, T = I@. The evolved oxygen smelt strongly of ozone when the perborate was added, but this gradually dis- appeared, and 10 minutes later no ozone could be detected, although there was no appreciable change in the temperatures, T, being - 7*5", T = 16".Samples of the anolyte were titrated from time to time with KMnO,. TA= -6'. T = 250. Time. Titre (10 c.c.). (From addition of perborate.) (In C.C. of N / 4 KMnO,.) o minutes 2.0 (calculated) ' 5 1 ) 1 '4 45 9 ) 0.7 These experiments suggested that the perborate acts in some way as a depolariser, lowering the anodic potential (which would account for the effect on ozone production) and being itself decomposed. It appeared, how- ever, that probably this is due not to the perborate itsclf, but to H,O, formed by its hydrolysis. Pissarjewski,:;: from extraction experiments with ether, coiicluded that in aqueous solution at 25" C. sodium perborate is almost com- pletely hydrolysed, and he supposes the hydrolysis to take place thus :- Na0,BO + H,O --+- Na0,H + HBO,, the sodium 1iyd;operoxide thus formed being further hydrolysed to NaOH and H,O,.The hydrolysis might take place in various ways, depending on the conditions, and on what borate ions were produced. Thus in alkaline solution, if BOQI" were formed, the reaction could be simply represented thus,+ omitting the sodium- BO,' + 20H' + B0;" + H.O,, and apparently the hydrolysis would be less the lower the alkali conccntra- tion, e.g., in acid solution (although the perboric acid might then decompose otherwise). Whatever may be the exact mechanism, it is certain that per- borates in solution are very largely hydrolysed to H,O,. The reaction with * 2. Phys. Clzenz., 43 (1903), 170. f This equation is purely formal, and is not intended to imply any particular '( mechanism " of H,O, formation.BORATE AND PERBORATE SOLGTIONS permanganate shows this, and a dilute solution of sodium perborate has the characteristic taste of H,O,.If, therefore, in an experiment such as No. 5 above, H,O, be added instead of perborate, one would expect the result to be similar, and this was found to be the case, the ozone production ceasing and the permanganate titre diminishing rapidly. One experiment was continued till the H,O, con- centration was vanishingly small, when ozone commenced to be produced again, increasing gradually in strength. It might be argued that the disappearance of H,O,, and therefore of per- borate, is due to chemical decomposition, and not to an electrochemical effect as suggested above ; but experiment showed that the rate of decom- position of H,O, in a I per cent.solution of NaOH is comparatively small at *ordinary temperatures, although above 40" C. it increases rapidly. More- over, an electrolysis experiment w-ith acid (phosphoric) anolyte, which will be described below, gave precisely similar results, and under these conditions khe chemical decomposition would be very milch slower than in alkaline n I F I G . 3 . solution. The non-production of ozone might also be due to chemical inter- action with the H202,Z: but it appeared probable that it was chiefly due to lower- ing of the anode potential. I n any case it was thought that actual measure- ments of the anodic potentials in a few cases would be of interest, especially as no such measurements in alkaline solutions appear to have been carried out hitherto.A node Poteriiinl Mcusiiiametits. The method adopted was to measure a combination of the type Pt anode I 1 NaOH I S/IO XaOH sl10 "OH 1 HgO 1 Hg the standard electrode employed being the Hg 1 z$:Nao, electrode devised by Donnan and Al1maiid.f The one actually used was made up * SchGne, Lichigs L 4 ~ ~ ~ ~ ~ ~ ~ ~ ~ , 196 (r879), 2.39. t Trails. CIICIII. Soc., gg, 845 (1911).186 THE ANODIC BEHAVIOUR OF ALKALINE precisely according to the directions given by these authors, and was checked by comparison against a similar electrode, its e.m.f. being, taken as EH = + 0.168 volts at the temperature of the laboratory. The e.m.f. of the combination was measured by the potentiometer method, using a standard cadmium cell, ail accurate platinum-iridium bridge wire, and a sensitive galvanometer.The apparatus is shown diagrammatically in Fig. 3, the cell being shown in section. As the chief thing is to maintain a solution of constant com- position round the anode, a large volume of electrolyte was contained in a reservoir (not shown in the diagram) of 800 C.C. capacity, and was allowed to flow fairly rapidly into the anode compartment. The anode was the platinum tube used in the previous experiments. The diaphragm D con- sisted of a sheet of asbestos paper :: wrapped round the close spiral used for cooling the anolyte, and fixed at the foot between the rubber stopper R and the rubber washer W. The liquid entered the anode compartment at the foot and flowed partly through and partly over the asbestos (the anode compartment being open at the top) into the cathode compartment, from which it overflowed by means of the syphon, the side tube being introduced to keep the level constant.The cathode C consisted, as before, of a cylindrical nickel sheet encircling the anode compartment. The advantage of allowing the electrolyte to flow through both anode and cathode com- partments is that its composition (as regards borate and alkali) remains unaltered, and it can be returned again to the reservoir. The standard electrode was connected in the usual way through the various intermediate solutions to the glass tube K, which terminated in a small capillary. This tube was held in place by the rubber stopper S, the end of the capillary being pressed firmly against a small flattened portion of the anode.The cooling spiral and the thermometers are omitted from the diagram for convenience. The potential actually measured is that between the liquid in contact with the anode at L and the top of the anode A. The latter is practically the same as the potential of the anodc at L, because no current flows through the portion AA’ (the current entering the foot of the anode) and very little through the portion A’L. The capillary was placed fairly near the surface t o mininiise the potential drop + between A’ and L. The first attempts to measure the potential with dilute alkaline borate solutions (such as Pouzenc’s) were not successful. The high resistance made the readings difficult, and violent fluctuations occurred.If the specific conductivity of the solution be taken as 0.02 (that of N/IO NaOH) the potential gradient round the anodc with DA = 38 would be 0*38/0*02 = 19 volts per cm. It is therefore evident that a very minute movemelit of the capillary might greatly affect the readings. * This had to be renewed from time to time, as it becomes weakened by the action of the alkaline liquid. t This can be calculated thus : Assuming the current density to be uniform, the current flowing in the anode at a point x cm. below the surface will be - C, C being the total current, and 1 the length of the anode immersed in the liquid. The potential difference between this point and the surface will therefore be 1; C * K - d . ~ (where R is the resistance of unit length of the anode), which becomes 2 1 0.5 millivolts, which is neolioible.the value would be 3.5 n&vglts. I c R . . v _ _ . With the capillary I j mm. below the surface, and C = j amperes, this becomes If, however, the capillary were half-way down,BORATE AND PERBORATE SOLUTIONS 187 Later on it was found possible, by fixing the capillary more rigidly, and shaping it in such a manner as to prevent gas bubbles collecting in it, and by using a more sensitive galvanometer, to obtain satisfactory measurements with such a solution. In the first instance, however, more alkaline solutions were used, and the readings with these were very satisfactory, the relative error of any one measurement being not more than a few millivolts. The exact values, however, in a given case are not reproducible, as they depend on the precise condition of the anode surface at the given moment.The method adopted to find the effect on the potential of adding H,O, to the electrolyte was to wait until the potential was constant, or at least not rising rapidly, then to add the requisite number of C.C. of concentrated H20,':: to the liquid in the reservoir. When it was desired to continue the measure- ments for a considerable time, the solution issuing from the cell was rapidly titrated, H,O, added to restore the original concentration, and the liquid poured back into the reservoir. 2'0 1'2 0 2 4 6 Time in Hours. FIG. 4. 8 The chief measurements carried out are shown graphically in Fig. 4, the The liquid potentials- numbers on the curves corresponding with the numbers below.NaOH conc. C, I NaOH conc. C, were calculated from the formula e = (2x - I ) 0.058 log C,/C,, . neglecting the difference in ionisation at the two concentrations. The liquid potential in such cases as- N a o ~ o ~ ~ ~ c ' '' I NaOH conc. C , was neglected. For the present purpose the absolute values within 10 or 20 * Merck's Pcr-lr~~rlrol (diluted).188 THE ANODIC BEHAVIOUR OF ALKALINE millivolts are of no importance. Slight fluctuations of the potential are probably largely due to variations in the rate of flow of the electrolyte, which may cause variations in the local concentration round the anode. In Nos. I to 4 the anode was not cooled, the anolyte being cooled by means of iced water. In Nos. 5 and 6, both anode and anolyte were cooled with freezing mixture, as in the former experiments.I. N . NaOH + 35 grains borax $er litre.- Temp. I o O C . D, = 38. This curve shows the rapid risc at the commencement, and the effect of altering D,. At- A ( 2 hr. 40 min.) D, was reduced to 19. B (4 hr. o min.) D, was raised again to 38. C (4 hr. 50 min.) H,O, was added to the electrolyte to make 0.01 Molar ('34 gram per litre), but there was no effect on the potential. 2 . N . Na0H.- T= 10'. D, = 38. Anode I x. NaOH N. NaOH I N/IO NaOH Omission of the borax would presumably have no great effect on the potential and would simplify the conditions, as there will only be H,O, and no perborate present. At D (5 hr. 45 min.), H,O, added to make 0.01 M., but there was no effect o n the potential. At E (6 hr.22 min.), H,O, made 0'10 M. Potential fell slightly, viz. from 1.770 to 1.740, then rose again to 1.766, although the concentration was maintained. After this experiment, the anode, which was used without cleaning or polishing in any way after No. I (i.e. some 12 hours a t 38 amperes per sq. dm.) had acquired a brownish colour. This appearance had been noticed before, but never to the same extent. It was carefully rinsed with water (to remove all H,O,) and gently rubbed on a piece of filter paper, moistened with KI and starch, a deep blue stain being produced. This indicated the presence on the anode surface of one of the higher oxides of platinum. The next experiment was carried out at a much lower current density, as it appeared probable that under these conditions the effect of H,O, on the potential might be more marked.K'lo NaoH 1 HgO 1 Hg. 3. N . Na0H.- T = IOO. D.I= 3'8. Freshly polished anode. Anode I N. NaOH 1 N. NaOH I N/IO NaOH 1 r ' l ~ ~ ~ o H I Hg. I The irregularities at the commencement were caused by a defective connection, with consequent fluctuations and momentary stoppage of the current. Potential fell from 1.595 to 1.407, then remained steady till the reservoir was refilled (no further H,O, being added, its Concentration being now 0.059 M. The potential then rose (G 6 hr. o miti.\ to a higher value. At F (5 hr. 37 min.), H,O, added to make 0.077 M.BORATE AND PERBORATE SOLUTIONS 4. Same as No. 3, only adding sodium perborate instead of H.0,. The method adopted was to select a moment when the reservoir was just becoming empty, and fill it up with a solution of the same composition, but containing in addition the necessary quantity of perborate, the concentration of the latter being such that if completely hydrolysed it would give approxi- mately the same H,O, concentration as in No.3 ; i.e. the KMnO, titre was approximately the same. At H (5 hr. 32 min.) the perborate was added (to make 0.070 M.). The potential fell from 1-544 to 1.432. After 6 hr. 10 min. the titre was maintained constant by adding H,O,, but the readings fluctuated somewhat, 5. 8 grams NaOH + 35 grams borax per litye.- Cooled anode. Da= 50 i.e. a solution similar to those used in the electrolysis experiments and under similar conditions. Temperature of anolyte varied between 13.5" and 17".Temperature of anode varied between - 4" and - 6". No readings were obtained at the commencement. At K (2 hr. 12 min.), H,O, added to make 0.95 M. Potential fell from 1.797 to 1.644, then rose slowly until at L (3 hr. 5 min.) the H,O, concen- tration was doubled (0.19 M.). Potential then fell from 1.655 to 1.550, the drop being nearly the same as at K. 6. 4 grams NaOH + 35 grams borax per litre.- Cooled anode (freshly polished). DA = 50. Temperature of anolyte, 15" to 19". Temperature of anode, - 4" to - 8O. The potential rose very rapidly at first, owing presumably to the high current density and freshly polished anode. The high value reached (El, = 1.953) is probably also due to the low alkali concentration. The subsequent slight fall may be due to temperature variation.No H,O, or perborate was added. No conclusions were drawn from these experiments as regards ozone evolution. The form of the apparatus made it more difficult than in the case of the electrolysis experiments to determine whether ozone were present or not, and it was detected with certainty only in the case of No. I , and in this case it vanished on the addition of H,O,, although the potential remained the same. The above measurements, together with the previous experiments on the diminution of the H,O, concentration, show that H.0, placed round a polished platinum anode which is strongly polarised tends to lower the potential, the actual drop depending on the various conditions, while the H,O, itself is destroyed.190 T H E ANODIC BEHAVIOUR OF ALKALINE These facts are in accordance with the theory, which is now generally accepted, that the overvoltage at such a platinum anode is essentially deter- mined by the platinum oxides:: contained in the surface layers.This theory is chiefly due to Grube,f who from measurements of the potentials in H,SO, of electrodes made of various oxides of platinum, con- cluded that when a platinum electrode is anodically polarised the surface layers contain varying amounts of PtO, PtO,, and PtO, in solid solution. The PtO, has a much higher potential than the other two, and the greater its con- centration, the higher will be the potential of the electrode.: It is formed by oxidation of the PtO, by the anodically produced oxygen, but is unstable, and breaks up again into oxygen and a lower, more stable oxide, its final concentration therefore depending on a balance between its rate of produc- tion from PtO, and its rate of spontaneous decomposition ; the greater the former, the higher will be the final PtO, concentration, and the greater the anodic potential.The actually observed formation of one of these higher oxides, referred to above under Experiment 2 , is of interest in this connection. Hydrogen peroxide can act electrochemically in two ways, corresponding to its chemical action as an oxidising and a reducing agent respectivcly. . I. H , O z ' e 2 0 H ' + 2 0 . 2 . H , O , e 0, + 2H' + 28. This was first clearly shown by Haber,§ and was applied by Foerster 11 to the action of H,O, on a platinum anode, on the theory that an oxide PtO, was formed, which in the light of the subsequent work by Grube we may call PtO,.Foerster's equations then become corresponding to I and 2 above. Most of the experiments which have been carried out by other workers on H,O, potentials are concerned with equilibrium conditions. In the present case, however, H,O, was placed round an anode which was strongly polarised, and in which the PtO, content was therefore very high. We should therefore expect action (b) to take place, resulting in a lowering of the potential and disappearance of the H,O,. Grube found that the rate of self-discharge of his PtO, electrodes was greatly accelerated by addition of H,O,, the potential falling rapidly ; in the present case, however, fresh PtO, is continuously bcing produced, and the actual fall of potential may be com- paratively small, as the experiments showed.The final value reached, if the H,O, concentration be maintained, will again represent a state of balance between formation and decomposition of PtO, and will depend on the H,O, concentration, the current density, the temperature, etc. Applying these considerations to the electrolytic preparation of perborates, it appears that, supposing that perborate ions can actualiy be formed at potentials attainable in practice, the H,O, produced by hydrolysis will have a * At very low current densities capillary phenomena also play a part. -f Z . f. Elektrocheni., 16,621 (1910). oxide, probably PtO,, is also present. 5 2. f. Elektroclzcm., 7, 441 (1901). (1 2. Phys. Clzem., 69, 2j4 (1909). When the anodic polarisation is very high, Grube considers that a still higherBORATE AND PERBORATE SOLUTIONS 191 twofold action.Firstly, it may affect the further formation of perborate by lowering the potential. Secondly, it will be itself destroyed, causing further decomposition of the perborate already formed. As both these effects will increase with increase of the H,O, concentration, an equilibrium concentration of perborate will finally be obtained. It is probable that the same effect plays a part in the electrolytic prepara- tion of persulphates and percarbonates. Persulphates, however, are only very slightly hydrolysed at ordinary temperatures ; percarbonates are certainly considerably hydrolysed, but the conditions under which they are produced {generally from a very concentrated solution of K,CO,) are more favourable than in the case of perborates.In the case of persulphuric acid, it is known that the mono-persulphuric acid formed by hydrolysis acts as a depolariser and affects the yield. It is also possible, of course, that persulphates and percarbonates can be produced at lower potentials than perborates. It was suggested above that the hydrolysis might be less in acid solution. I t also seemed probable that higher potentials would be reached in acid solution. An experiment was therefore carried out (No. I below) using the cooled anode apparatus, with an anolyte consisting of boric acid in presence of phosphoric acid. Of course it was not expected that free perboric acid would be formed in quantity, but it appeared that the solution might show some oxidising properties.Further Experiments with Cooled Anode. z. N . H,PO, (33 grarnsllitre) + 20 grams H,BO, per Iitre.- v = 21 to 22. C = 2.4. DA = 30. Time, 70 Inin. T = 3" to 4". T A = - 9". Samples of the anolyte were tested with KMnO, 5 j and 70 min. from the commencement, but there was no trace of decolourisation. Hydrogen peroxide was then added to the anolyte to make 0-1 N, (1-7 grams per l.), the temperatures being maintained about T = 20 TA = - 5". In 50 minutes the KMnO, titre fell from 3.8 to 2'0. This is of interest because in presence of phosphoric acid the chemical decomposi- tion of H,O, is very slight. In this experiment ozone continued to be pro- duced even in presence of the H,O,. It was pointed out above that in the earlier experiments in which the anolyte consisted of excess of borax dissolved in NaOH, the ozone produc- tion was very slight.In the light of the information subsequently gained, however, even if the potential in such cases is lower, other advantages might possibly outweigh this. In the first place it appeared that probably the temperature of the anolyte could be kept lower, which would probably tend to diminish the hydrolysis.':: The increase in the total borate concen- tration would also be an advantage. An experiment was therefore carried out (with cooled anode) using the solution of Constam and Bennet (see above) containing 282 grams borax+56 grams NaOH per litre. Such a solution deposited solid only slowly at 0°C. As this was the only experiment in which perborate was produced in appreciable, although still very small quantity, it will be given somewhat * Pissarjewski, loc.cit. . more fully.I 92 T H E ANODIC BEHAVIOUR OF ALKALINE 2 . 282 grams borax +- 56 grmm NaOH per litre.- Freshly polished anode. Catholyte - 8 per cent. NaOH. Vol. of anolyte - 142 C.C. C = 4.8, DA = 60 during first 20 min. C = 2.4, DA = 30 thereafter, to enable anolyte to be kept at low t e nip era t u re. The course of the experiment is shown in tabular form below. A few C.C. of 16 per cent. NaOH had to be added soon after the com- The titre given is of 10 C.C. of At the lower current mencement to prevent deposition of solid. anolyte against s/4 KMnO,. density V was 13 to 16. V was 20 to '23 at first. - - __ ___ - - I I Time.I Temperatures. - / _ - . - . - - Hr. Min. ! T T* o o I 7;C. -14" 0 I 0 4 - 1 2 - 5 O - - 0 20 o 40 2.5' - 9.5' I 0 O0 - 9'0" 2 20 0" - 9'5" 2 40 I oo - 8.50 3 0 oo - 5.0' I 45 ,-IC - 7'5" Titre. 0 - I drop 2 drops 2 drops 0.26 C.C. 0'30 C.C. 0-35 C.C. 0.38 C.C. Remarks. NaOH added. Ozone dis- tinct C reduced to 2-4 Still ozone Ozone less strong No ozone ,, 9 , ,9 .-__. The final concentration of perborate represents a chemical yield of about 0.15 per cent. and a current efficiency of about 0.48 per cent. It must be remembered, however, that only about one-third of the anolyte is actually in the cell at a given moment, the whole being kept in constant circulation through the cell. The exact significance of these figures is therefore doubtful.Constam and Bennet, as stated above, also obtained some decolourising effect on KMnO, as the result of their experiment, but said that it was due to production of H202. As there will presumably be an equilibrium between H20, and perborate, their reason for making this statement is not clear. One further experiment was performed with the cooled anode apparatus to determine whether under the conditions of the above experiment, if the perborate concentration in the anolyte were comparatively high, it would tend to diminish. 3. .At the commencement the anolyte was identical with that in the last experiment, but only half the quantity was placed in the reservoir, the remaining half, containing perborate, being added only after some perborate had been formed in the cell.By this method it was ensured that the conditions were as favourable for perborate formation as in the previous case. The tem- peratures, etc., were as nearly as possible the same as in No. 2 ; D.4 was 30 from the commencement, the anode being again freshly polished. As before, a little NaOH had to be added to prevent crystallization, but the amount added after the addition of perborate, to compensate for that removed by electrolysis, did not exceed 5 c.c., so no appreciable error is caused in the titrations on thisBORATE AND PERBORATE SOLUTIONS account. tributed throughout the anolyte before any titrations were made. Time was allowed for the perborate to become uniformly dis- Time. ! Temperatures. i Titre. Remarks. I , Hr. Min. I 0 0 ; 0 15 T 7" c.3" - 4" 3" - O0 O0 I" - 11" - - 10' - - 9'5" - 80 - 7.j" - 8" I I - 1 0.1; C.C. i Ozone distinct. NaOH I ' added. 1 - ' Ozone strong. I - Remainder of anolyte + Ozone very faint. perborate added. l - I 4.20 C.C. I - ' 3-64 C.C. No ozone. 3-18 c c. 1 ,, 2.89 C.C. 1 9 1 2-52 C.C. 7 7 The titre given in this case is of 2 C.C. against N/IO KMnO,. From these figures, and those in the previous experiment, it appears that the equilibrium concentration which would finally be obtained would be very low, probably not very much greater than that actually reached in No. 2 (after electrolysing for nearly 3 hours) which represents a very small yield indeed. It is possible that better results might be obtained by the use of still more concentrated solutions (if such can be obtained) and still lower temperatures ; or by the use of solutions containing a larger proportion of alkali, which may possibly contain more BO3'" ions, but the experiments have not so far been extended in these directions, nor does it seem probable that satisfactory yields would be obtained under any conditions. THE POUZENC PATENT.Under these circumstances it is difficult to account for the existence of the French patent by Pouzenc referred to above. Reference to the patent specification revealed the fact that it contains only a brief description of the apparatus, consisting of the vertical electrolyser referred to above. The two papers by Beltzer, however, were found to contain various details, some of which have already been given. Accorditig to these, solid perborate :k can be obtained by filtering the liquid flowing out of the anode compartment of the Pouzenc cell, the filtrate being mixed with the catholyte and used again.Beltzer says that for preparation of solid perborate it is best to use a '' concen- trated solution of sodium orthoborate," while for making bleaching solutions he recommends the solution containing 24 grams borax + 6 grams NaOH per litre, which he ascribes to Pouzenc. He also states that the electrolyte must Aow rapidly through the cell, and that the solid perborate must be removed immediately from the action of the current and the alkaline liquid, otherwise it '' redissolves in the liquid and decomposes." As a matter of fact, from the diagrams of the apparatus it seems that the * According to Melikoff and Pissarjewski IOO grams H,O dissolve 2-55 grains NaBO, 4H,O at 15' C., so if good yields were obtained solid would separate.194 THE ANODIC BEHAVIOUR OF ALKALINE amount of liquid actually in the cell at a given moment is comparatively small ; in other words the " current-concentration " is high.Such a condition is necessary in cases where the product is chemically unstable, e.g. in the case of hyposulphites; but it has been shown that the important factor in the case of perborates is the electrochemical decomposition. As regards temperature, - 5OC. is mentioned by Beltzer, but it is not clear whether this is meant to refer to the electrolyte itself, or the brine used to cool it. In one of the papers a commercial form of the apparatus is described, but it is clearly stated at the commencement that the electrochemical method was not so far (1911) used for the manufacture of perborates, and a careful study of the papers gives the impression that the author has not actually carried out any experiments on the subject.Q A F I G j. 0 Experiments with the Pouzeiic Apparatus. A small cell coiistructed to imitate as nearly as possible the Pouzenc electrolyser on a small scale is shown in Fig. 5. A is a vertical section through the middle of the cell. B is a plan of the top of the cell. It was constructed of four portions of ebonite sheet (numbered in the diagram) which were clamped together by large clamping screws (not shown in the diagram), thin rubber washers being interposed to make the cell watertight. The electrodes of fairly stout and well-polished platinum foil, shown in the diagram by the dark lines E and E', were fixed between I and 2 and 3 and 4, the diaphragm D of asbestos paper being between 2 and 3.The electrolyte entered the anode and cathode compartments, which were identical, at I and 1', and overflowed a little above the tops of the electrodes at 0 and O', the part of the cell above that serving as a settling chamber forBORATE AND PERBORATE SOLUTIONS 195 spray, cement. The inlet and outlet tubes were of glass, fixed in by a suitable The dimensions of the cell were as follows :- Area of electrodes . . . . . . . . . . . . 8 sq. cm. each. Volume of anolyte . . . . . . . . . . . . . Distance apart . . . . . . . . . . . . 12 mm. ,, catholyte . . . . . . .. . ,, 5 C.C. The electrolyte was contained in a glass reservoir in which it was cool6d by glass spirals through which flowed the usual cooling liquid. Separate .glass tubes conveyed it to the anode and cathode chambers, the rate of flow through each being separately regulated by screw clips. The glass tubes were as short as possible, and well lagged, to prevent access of heat. Experiments were first carried out with the Pouzenc solution, contain- ing 24 grams borax + 6 grams NaOH per litre. With 2 amperes passing through the cell (DA = 25) the voltage across the cell was 25, and although the solution was cooled almost to its freezing-point (- 0.89" C.) the anolyte leaving the cell had a temperature of 17°C. It was tested with KMnO,, hut showed no trace of oxidising action.It is difficult to imagine that such a solution could be used commercially, as the heat produced in the cell, and the power absorbed, would be excessive. Determination of its conductivity showed the latter to be, at- 0°C. K = 9.02 x 10-3, 18" G. k = 14.87 x 10-3. This is less than the conductivity of an N/IO solution of NaOH, which has a conductivity of about 20 x TO-3 at 180. Experiments were then carried out with a solution whose composition corresponded to sodium orthoborate. The electrolyte consisted of a large volume (I litre) of a solution containing 95'5 grams borax + IOO grams NaOH per Zitre,:: cooled to about -5°C. It was very strongly alkaline. Experiments were carried out at current densities of 30 and 60 (the values given by Beltzer) and various rates of flow.With DA = 30 (2.4 amperes) V was 6.5, and with a sufficiently rapid flow the temperature of the anolyte leaving the cell was as low as 2" to 3". In order that it might not exceed 10" the rate of flow had to be not less than 10 C.C. per minute through the anode compartment, and at DA = 60 correspondingly greater. When all the electrolyte had run through it was passed through a second time without interrupting the current, to obtain the benefit of any rise of the anode potential, the total time being 30 to 50 minutes. Samples of the anolyte leaving the cell were tested immediately, but in no case did 10 C.C. decolourise a single drop of N/20 KMnO,. This is not surprising when one considers the very slow rate of perborate formation, and very small yields obtained, even with the cooled anode apparatus. In view of these results it is difficult to believe that solid perboratc has ever been prepared by means of the Pouzenc cell. It is, however, possible that impurities in the borax used (stated to be "commercial") might give rise to slight oxidizing properties in the solution, e.g. chloride giving rise to hypochlorite. Throughout the present investigation' the purest materials were used. * A solution of double this concentration deposited the solid at this temperature.196 T H E ANODIC BEHAVIOUR OF ALKALINE BORATE SUhIMARY. The previous work on the electrolytic production of perborates has been briefly reviewed and criticised. Experiments have been carried out in which solutions of alkaline borates of various composition were used as anolyte in suitable electrolytic cells, under various conditions of temperature and current density, a hollow cooled anode being employed in some cases. With dilute and slightly alkaline solutions, such as would be suitable industrially for bleaching purposes, no perborate (or at most a minute trace) was formed in any of the experi- ments, even with the cooled anode apparatus. Even with much more concentrated solutions only a very small quantity was formed. It has been shown that an equilibrium is obtained between the rate of formation of the perborate and its decomposition owing to the H202 formed by its hydrolysis acting as a depolariser, tending to lower the anode potential and being itself destroyed. Measurements have been carried out of the actual potentials obtained at the anode in certain cases, and of the effect of hydrogen peroxide on the potential. The results accorded well with Grube’s theory of overvoltage at a platinum anode and the known electrochemical behaviour of H2OZ, Finally, a French patent for the electrochemical manufacture of perborates has been discussed and investigated, a similar apparatus having been con- structed on a small scale. It was found to give negative results in practice, no trace of perborate being formed in any of the experiments. In conclusion the author wishes to express his sincere thanks to Dr. A. J. Allmand, at whose suggestion the above investigation was carried out, for his invaluable assistance and advice. M USPRATT LABORATORY, T H E UNIVERSITY, LIVERPOOL, May, 1914.
ISSN:0014-7672
DOI:10.1039/TF9151000177
出版商:RSC
年代:1915
数据来源: RSC
|
3. |
The vapour pressure of liquids in presence of gases |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 197-206
F. H. Campbell,
Preview
|
PDF (776KB)
|
|
摘要:
THE VAPOUR PRESSURE O F LIQUIDS IN PRESENCE OF GASES. BY F. H. CAMPBELL, M.Sc. ( A Paper comrnririicated to !he TRANSACTIONS of the Fnraday Society, October, 1914.) I. A NEW STATICAL METHOD FOR T H E DETERhIINATION OF VAPOUR PRESSURES. While methods for the measurement of the freezing-point, conductivity, viscosity, etc., of solutions have reached a high degree of accuracy and convenience, none of those proposed for the determination of vapotir pressure has come into general use. Beckmann has recently found it necessary to -introduce a number of alterations in his well-known boiling-point method,” with a view, among other things, of eliminating the disturbing influence of .alterations in the barometric pressure during the course of an experiment. It does not appear, however, that the increased complexity of the apparatus is compensated for by greater accuracy in the results.With mixtures of two volatile liquids satisfactory results have been obtained by determining the composition of the distillate from a mixture of known composition, at known temperature and pressure. Von Zawidzki t and Wrewsky 1 worked at constant temperature, Rosanoff and Easly § at constant pressure. In all cases the experimental difficulties are considerable. Regnault 11 determined the vapour pressure of water by saturating a known volume of air by passing it through wet sponge and moist silk screens, absorbing the water in sulphuric acid and weighing it. Results for pure water in substantial agreement with those obtained by the barometric and boiling-point methods have been obtained by various methods depending upon the same principle. Permany bubbled air through a series of large Liebig’s bulbs, the last being provided with a manometer.Together with Price, Perman has extended the use of this method to solutions.‘:;” Lincoln and Klein tf- and Krauskopf 3 passed air over the surface of water, or aqueous salutions contained in a large tube, which was kept in violent lateral motion. Derby, Daniels, and Gutsche $ 5 saturated air by passing it through bulbs filled with beads, agitation being effected by rotating the apparatus. Linebarger 11 /I employed Liebig’s bulbs and liquids other than water. His results were not concordant. Ibid., 1912, 81, I. ** Tram. Furad. Soc., 1912, 8, I. Ibid., 1910, 14, 489. 1111 7. Am.Cltcnz. Soc., 1895, 17, 615 ; Chenz. News, 1895, 72, 167. * Zeit. $ / I J J S . Clzcnz., 1912, 78, 725 ; 79, 565. jI AWI. chim. $hys., 1845 (3), I j, 129. t Ibid., 1900, 35, 129. Ibid., 1909, 68, 641. 7J Proc. Roy. Soc., 1903, 72, 72. tt J. Phys. Clzem., 1907, 11, 31% $§ J. Claeni. Soc., 1914, 36, 793. I97THE VAPOUR PRESSURE OF LIQUIDS Walker,::: on the suggestion of Ostwald, bubbled air through the aqueous solutions under examination and through pure water, the vapour pressure of the former being calculated from the loss in weight of solution and solvent respectively. The Earl of Berkeley and Hartley + conclude that among the reasons for the failure of Walker's method to give satisfactory results were the facts that the hydrostatic pressure in the several vessels varied, so rendering complete saturation of the air impossible, and that liquid was carried mechanically from solution to pure solvent (water).They eliminated both sources of error by passing the air over the surface of the liquid. A serious disadvantage of the method is that a determination occupies several days. Will and Bredig 1 and Orndorff and Carrel1 5 used Walker's original method with ethyl alcohol as solvent, but the method does not seem suitable for use with this liquid, and Will and Bredig found it to be even less so with ether. Dieterici 11 employed a differential method, the vapour pressure of a number of aqueous solutions having been determined by the measurement of the movement of a flexible metallic septum, which was compared with that pro- duced by the pressure of the vapour of pure water under the same conditions.Dieterici obtained good results, but the apparatus is complicated and the manipulation difficult. Lescceur T proposed a method whereby the vapour pressure of a solution is calculated from an observation of the temperature at which a dew forms upon a metallic cylinder in contact with the vapour. The method has been modified and improved by Gumming.'::" While apparently suitable for use with aqueous solutions, this method is unlikely to be equally so in cases in which organic solvents are employed. The ordinary static, or barometer, method has not been used with solutions since Raoult tt completed his classic research upon ethereal solutions. The neglect of this apparently simple method is due to the difficulty of obtaining a mixture of known composition in a gas-free condition, and because of the mechanical difficulties in securing efficient stirring.The investigation of the vapour pressures of binary mixtures of liquids has been confined almost exclusively to those cases in which both constituents are volatile. It was desired to examine the effect of admixing a liquid not measurably volatile at the temperature of experiment upon the vapour pressure of a freely volatilc liquid. Since none of the methods hitherto proposed is easily applicable to such cases, that described below was devised. The principle of the method is that of allowing a liquid, previously saturated with hydrogen at a definite temperature and pressure, to evaporate into a space containing the same gas, under the same conditions.A method based on this principle is no more complicated than the barometric, and is free from the sources of error of that method. The arrangement of the measuring instrument is shown in Fig. I. I t consists of a glass tube A, open at the bottom, to which are joined a side- tube B, and the tube E, which is bent parallel to A. E is provided with two bulbs, between which a blue-glass point is sealed. The lower end of E is bent at a right angle, and the manometer tube D, graduated in millimetres, is sealed to it parallel to A and E. The levelling tube G, * Zeif. phys. Chem., 1888, 2, 602. Ber., 1889, 22, 1084. /I W i d . A m . , 1893, 50, 47. ** J . Chem. SOC., 1909, 95, 1772. t t Zcil. phys. Chem., 1888, 2, 353, and Aim.chim. phys., 1890 (6), 20, 297. t P40C. Roy. SOC., A, 77, 156. $ J. Phys. Chcm., 1897, I , 753. 7 A m . ~ h i n z . phys., 1889 (6), 16, 378.IN PRESENCE OF GASES I99 consisting of thick-walled rubber tubing and a glass funnel, is connected to the horizontal portion of E. The end of A is closed by a rubber bung, through which passes the tube C containing the liquid under investigation. The rubber stopper is protected by a layer of mercury. Fig. 2 is a diagram of the apparatus in which the pure liquid or mixture is saturated with hydrogen. It consists of a tube F, to the bottom of which is sealed a narrow tube with a tap H, and with a similar tube G sealed to the side. F is closed when necessary by a ground-glass stopper, through which is sealed a narrow tube C, the ends of which are capillary.Method ofExperiittent.-In the case of a pure liquid a convenient quantity is placed in F, the bottle is put into a bath maintained at the temperature at which the vapour pressure is to be measured, and a slow stream of pure, dry B C [ : FIG. I . FIG. 2. H 1 hydrogen is passed in through H and out through G. When the liquid is saturated with the gas, the tap on G is closed and that on H manipulated so 'that the liquid rises slowly in C ; when it is almost full C is sealed at the upper end, removed from F and sealed at the other end. With mixtures the pro- cedure is the same, except that the apparatus is first filled with hydrogen, the upper end of C sealed and the apparatus weighed ; C is removed, a quantity of the non-volatile constituent run in, generally from a graduated pipette, C is replaced, hydrogen is again passed through and the apparatus weighed ; after introduction of the volatile liquid and saturation of the mixture, the apparatus is again weighed.Finally the mixture is brought back to the experimental temperature, the end of C is broken off, and the mixture is introduced and sealed into it as described above. A measurement is made as follows : The end of C is pushed through the hole in the rubber stopper, deeply scratched about z cm. above the stopper, which is then firmly inserted into the open end of A. Pure dry mercury is200 THE VAPOUR PRESSURE OF LIQUIDS I poured in through G, and a quantity, sufficient to fill it to a mark etched on the side, is introduced into A.The apparatus is placed in the bath, clamped so that the whole of it except the end of B is immersed, and G is adjusted so that a globule of mercury remains in the horizontal part of E. The air is displaced by passing hydrogen in through B and out through D, the globule of mercury acting as a valve preventing back-diff usion of air. More mercury is now introduced, until its surface is a few millimetres below the blue-glass point, the hydrogen stream is stopped and the end of I3 is sealed off. The apparatus is now lowered until B is wholly immersed, and, when the tempera- ture of the bath has been reached, the level of the mercury is adjusted until the point just touches its surface ; the mercury level in the manometer is read to 0.1 mni. and mercury is again added through G, generally until the upper bulb in E is partially filled ; the liquid is introduced into A by break- ing C at the scratch by gentle lateral pressure and displacing it by mercury, which is effected by careful shaking.I n order to promote saturation of the hydrogen the shaking is repeated periodically, the volume being kept constant by addition of mercury through G. When no change in the pressure exerted occurs after agitation, the new level of the mercury in D is read, the TABLE I. ._____ Liquid. ~ Ether . . . . . . . . . Ether . . . . . . . . . Carbon disulphide ... Acetone ... Methyl a1coh;;l' ... Ethyl alcohol ... Water . . . . . . . . . 1 Temp. j "C. 20'0 30.0 30.0 30.0 40.0 60.0 70.0 Pressure in Hydrogen. Pure Vapour. inn1 . 432'3 642.1 431'9 275'8 257'4 347'3 230.8 min.442'4 281.0 260.0 350'2 647'9 434'6 233'8 Difference per Cent. 2'28 0.89 0.6 I 1-85 0'77 0'82 1-28 vapour pressure being given by the difference between the final and initial readings, corrected to o* C. for the expansion of mercury and the glass scale. The bath used had a capacity of about 32 litres ; its temperature was main- tained within & ' 0 2 O by means of a toluene regulator with a large bulb ; two opposite sides were of plate-glass. Readings were facilitated by illuminating it from behind. The thermometers used were graduated to O-IO, and were standard instru- ments from Kew Observatory, or had been recently compared with the standards. Before being used with mixtures the method was tested by determinations of the vapour pressures of a number of carefully purified liquids.Details of the methods of purification are given. The boiling-points, determined with the usual precautions, are recalculated to 760 mni. pressure. Ether.-Merck's pure ether was washed with ten successive small quantities of water, and allowed to stand over fused calcium chloride, phosphorus pent- oxide, and metallic sodium and sodium amalgam in turn. It was kept in the dark in contact with the last-named substances and was distilled as required. It boiled at 34.6".IN PRESENCE OF GASES 201 Methyl AZcohol.-A pure sample from Kahlbaum was boiled for several hours with metallic calcium under a reflux condenser and was fractionated from it. 'Moisture was carefully excluded throughout. The fraction used boiled at 65". Ethyl Alcohol.-Kahlbaum's 99.8 per cent.alcohol was treated in the same way as the methyl alcohol. Carbon Disul$hide.-Mercks " guaranteed pure " material was shaken with copper sulphate solution, dried over fused calcium chloride and fractionated. I t boiled constantly at 46.3". Acetone.-Crude acetone was fractionated ; the portion boiling below 63" was treated with .powdered potassium permanganate in the cold, then with ferrous sulphate, and finally with fused calcium chloride. It was twice fractionated ; the sample used came over between 560 and 56-15". It was probably not quite pure. Water.-Ordinary distilled water was acidified with sulphuric acid and redistilled. In Table I are given the mean values of several successive experiments with each liquid ; in the fourth column aregiven what appear to be the most reliable of the values obtained by other observers.It will be seen that the differences are small, and that the method is sufficiently accurate for most purposes, including that for which it was devised. With a mixture a determination can be completed within two and a half hours, with a pure liquid in rather less than half that time. If two sets of apparatus be used, four experiments can be carried out in one day; the method is therefore reasonabiy rapid. Substances under examination come into contact with glass, mercury, and hydrogen only, so that the method may be applied to a very large number of liquids. The temperature range also is considerable ; any temperature below that at which rubber softens may be uscd.It boiled at 78-50. 11. THE IXFLUENCE OF THE NATURE OF THE GAS UPON THE PARTI.4L PRESSURE OF THE LIQUID EVAPORATING INTO IT. Nernst :x states that "if we bring a simple liquid into a vacuum, evapora- tion takes place, till the pressure of the gas formed has reached a definite maximal value, namely, the corresponding vapour pressure. In presence of another, but an indifferent gas, evaporation takes place till the partial pressure of the resulting vapour is equal to the vapour pressure." It does not appear to be generally recognized that this statement is an inexact assumption ; that this is the case, however, is shown by the author's experiments and by the results of all previous investigations carried out under comparable conditions. In preliminary experiments with the method described, various liquids were allowed to evaporate into air.The results, though concordant, were invariably lower than those obtained by other methods. The air was replaced by a more soluble gas, carbon dioxide, and by a less soluble one, hydrogen. The results of the three series tabulated below show that, in the case of each liquid, the magnitude of the deviation from the most reliable values is greatest in carbon dioxide and least in hydrogen. That the total lowering is not due to the solution of the gas in the liquid is evident ; this is clear from a calcula- tion froiii data given by Just.+ Just found the ratio of the concentration of * Tltcoretical Cl~t.i?zistry, 3rd ed., 1911, 56. t Zcit. Plzys. Clicm., 1901,37, 342.202 THE VAPOUR PRESSURE OF LIQUIDS hydrogen dissolved in ethyl alcohol at 25" to that of gaseous hydrogen at 760 mm. to be 0.089 ; it follows that the dissolved hydrogen is present in quantity sufficient to lower the vapour pressure of the alcohol by 0.05 mrn., or about 0.08 per cent.; the lowering actually observed at 60° is 2.9 mm., or 0.82 per cent. Liquid. Ether ... ... Ether ... ... Carbon disulphide Acetone ... ... Chloroform ... Methyl alcohol ... E thy1 alcohol . . . Water ... ... Temp. O C . - I 20'0 30.0 30.0 30.0 I 40.0 TABLE XI. Vapour Pressure in Presence of H,. 1 Pure Vapour of Liquid. The values given in the last three columns, for the sake of comparison, are taken from Landolt-Bornstein-Meyerhoffer's TabeZZeiz and Castell Evans' Plzysico-Cheinical Tables.They are compiled from results obtained by .(a) Regnault (Mim. de Z'Acad., 1862, 26, 339) ; (b) Ramsay and Young (Phil. Trans., 1886, 177, I, 123 and 1887, 178, A., 57) ; (c) Battelli ; ( d ) Dittmar and Fawsitt (Edin. Traits., 1886-87, 23, 11, 509) ; ( e ) Schmidt (Zeit. Phys. Chem., 1891, 8, 628). No value for chloroform in hydrogen is given, very irregular results having been obtained. I t is supposed that hydrogen reacts with chloroform, an hypothesis first put forward by Just" to account for his inability to establish equilibrium between hydrogen dissolved in chloroform and the gas. Nitrogen appears to be completely indifferent ; the result in presence of this gas is 0'94 below that of Ramsay and Young. Since carbon dioxide probably forms additive compounds with ether, chloroform and the alcohols,+ even at the comparatively high temperatures employed, the great effect of this gas upon the vapour pressures of these liquids was to be expected.It might be thought that the deviations noted are due to some un- discovered source of error peculiar to the method used ; a review of other work done under comparable conditions, however, makes it evident that this is improbable. From the present point of view the most important work is that carried out by Regnau1t.f Having found that the pressures calculated from the weights of water in known volumes of saturated air and nitrogen respectively were invariably from I to 2 per cent. lower than those developed in vacuo, Regnault extended the investigation to other liquids.In addition to that already referred to, Regnault employed two statical methods. The first of these was similar in principle to that used by the author, a small bulb containing the liquid being broken in a flask of 600 to 700 C.C. * Loc. cit. t Cf. Hempel and Seidel, Rev., 1898, 31, 2997. Dolezalek, Zeit. plzys. Chem., 1910, 71, 191. Ann. plzys. chinz., 184j (3), 15, 129, and M c m . Acad. Sc., 1862,26,679.IN PRESENCE OF GASES Temperature. 203 Partial Pressures. Differences. ' capacity to which a mercury manometer was attached. In all cases the gas present was air, the liquids were ether, carbon disulphide and benzene. The conditions were varied in several ways, but low results were always obtained ; with ether the differences varied between I and 2.9 per cent., with carbon disulphide between 0.8 and 1.2 per cent,, while with benzene the limits were 2-9 and 4.8 per cent.No reliable data as to the solubility of gases in ether are available, but it is of interest to note that benzene is a much better solvent for gases than is carbon disulphide. In the second series of experiments small quantities of ether were intro- duced into a graduated tube containing air, hydrogen, or carbon dioxide. The gas pressure corresponding to each volume being known at a constant temperature, a measurement of the total pressure after each successive decrease of volume gave the partial pressure of the ether vapour. I t was found that when the gas pressure was approximately atmospheric, that of ether was 1.3 per cent. too low in air and 1-5 per cent.too low in hydrogen, but that, at about 1,200 mm. gas pressure, the pressure of the ether vapour slightly exceeded that developed in vacuo. Certain of the results obtained in carbon dioxide are given in Table 111. I "C. TABLE 111. I 7'7 ,, ,) ', mm. 1 0'00 535'38 740.38 I 1,165'78 mm. 260'0 234'40 245'80 235'43 - 9.8 6-2 9'4 In his dynamical experiments the gas used by Regnault was air ; he con- firmed his previous results with water, found that alcohol gave very low results, and that the values found with ether were in good agreement with those calculated. Comparatively recently Shaw :!: repeated Regnault's experiments with water, using his original dynamical method. Like Regnault's, Shaw's results were slightly low. Tammann + modified Regnault's method slightly and applied it to the determination of the vapour pressures of water, unsaturated and saturated aqueous solutions, and of hydrated salts.All the experiments were done at temperatures very near 35". The mean value for pure water was somewhat lower than Regnault's ; unsaturated solutions gave irregular results, owing no doubt to the fact that thcy were not stirred ; with saturated solutions and hydrated salts the relative vapour pressures were higher than those measured in other ways. It is suggested that this last fact is due to a diminution in the vapour pressure of water in air, not shown by saturated solutions and hydrated salts, or shown to a much smaller extent, That this explanation is probably correct is shown by the excellent agreement between the values given by Frowein $ and others and those of Tammann when the values given by * Phil.Trans., A., 1888, 179, 73. t Wid. A m . , 1888, 23, 322. Zeit. plzys. Claern., 1887, I , I.204 T H E VAPOUR PRESSURE OF LIQUIDS the latter are recalculated, the pressure developed by water in vaczio being substituted for that determined by him. Justification for the assumption that a saturated solution of sodium chloride develops the same pressure in air and in vacuo is found in the observation of Geffcken + that the solubility of oxygen in sodium chloride solutions at 25" falls off rapidly with increasing concentration of the salt, so that it is approximately one-half that in pure water at a concentration of about 10.5 grams per IOO C.C. ; hence it may be taken that the solubility of oxygen and nitrogen at 35" in a saturated solution, containing approxi- mately 35 grams per IOO c.c., is negligibly small.If PA be the vapour pressure of pure water, as determined by Tammann, Pa that of the sodium chloride solution under the same conditions, and P, that of pure water Z I Z vucuo, we find that 6 - 0.769, or about j per cent. too high, but that, taking 41-85 mm. as the value for P,, 6 - 0.732, 4-8 per cent. lower. It may be assumed that the solubility of gases in the water of crystalliza- tion of hydrated salts is negligible, so that the same considerations apply as in the case of saturated solutions. Tanimann seems to have failed to realize that the air was not passed over any of the salts at a rate sufficiently slow to allow of its saturation, except in those cases in which the vapour pressure is relatively high, A consideration of the cases in which the relative vapour pressure is independent of the rate at which the air was passed shows that the agreement between Tammann's resrlts and those of Frowein is very close when P, is substituted for PA. P; is the vapour pressure of the salt ira vaczio.PA- Pv - TABLE IV. Na,HPO; 12H,O ... ... 0.879 ~ 0.918 0.886 ' 0'039 0.007 ZnSO; 7H,O ... ... ! 0.730 j 0'774 0'739 0.044 0.009 MgSO, - 7H,O ... ... ' 0'594 0'622 ' 0'595 0'028 1 0.001 I -____ ---______ ___ Linebarger f applied a modification of Regnault's method to the deter- mination of the vapour pressures of mixtures of volatile liquids and inci- dentally measured those of a number of them in the pure state. In all cases in which the probable experimental crror is small enough relatively to be neglected, Linebarger's results are somewhat lower than those obtained by other observers.These differences are attributed by him to the volatility of the liquids, but it will be seen that the differences increase with the solvent power of the liquids for gases: and not with the volatility. With the exceptions to be mentioned immediately, the above review includes mention of all the work of this kind hitherto published. I t is, in consequence, of importance that the results deviate, in all cases, in the same direction from those obtained by methods involving the use of gas-free liquids, that is, that they are always low. The exceptional results referred * Zcit.plzys. Clictii., 1904, 49, 257. t LOC. cit. 1 Cf Just, loc. cit.I N PRESENCE OF GASES to are those for water, obtained by Perman, Krauskopf, and Derby, Daniels and Gutsche, whose papers have been cited. It has generally been assumed that failure to get satisfactory results of vapour pressure measurements in presence of a gas, almost always air, has been due to experimental error, and, in cases in which a stream of gas was used, to incomplete saturation. Regnault convinced himself that the differ-, ences which he found were due to actual differences of pressure. He sug- gested that the molecular attraction between the substance of the walls and the vapour particles causes condensation and that equilibrium is never reached, because the rate of evaporation of liquids in presence of gases is slow and because a film of liquid of the thickness requisite to saturate the walls cannot form on account of the force of gravitation.TABLE V 1 1 Vapour Pressure. Temp. , Difference Liquid. ' oc. I Other per Cent. Linebarger. Observers. I - __ ____ - - ~ mm. mm. Carbon disulphide ~ 20.0 1 296-4 298.1 0'57 Benzene ... ... 1:: 1 34.8 145'4 147.2 1'22 Chloroform ... ... I 35.0 290.1 I 301.1 1 3'65 1 It is suggested that an explanation more in consonance with the facts recorded in this paper is that gases form films on the surfaces of liquids, as on those of solids; in other words, that liquids, like solids, adsorb gases. It is known that, of the three gases, carbon dioxide, oxygen and hydrogen, that nearest to its critical temperature under ordinary conditions-carbon dioxide-is the most freely soluble and is adsorbed to the greatest extent by solids, notably by charcoal, and that hydrogen, the farthest removed from its critical temperature, is least soluble and least easily adsorbed by char- coal. The order is the same when we arrange the gases according to the magnitude of the vapour pressure lowering which they cause.This parallelism lends weight to the hypothesis of the existence of a highly concentrated film of gas at the surface of a liquid, and further, certain of Kegnault's observations receive satisfactory explanation upon this assumption. In his first series of statical experiments Regnault found that the greatest pressure at any particular temperature was developed immediately after the mixture of gas and vapour had been raised to that temperature, and that the maximum pressure was greater when the alteration in temperature took place quickly ; in all cases, however, a gradual fall followed.In the second series the highest vapour pressure at any particular partial pressure of gas was developed immediately after a reduction in volume ; as before, a gradual fall followed. It is evident that, in either case, a fresh film of liquid m u s t be formed, which exerts a pressure equal to or approaching that developed in ZICICUU, and that on standing gas is adsorbed and the pressure falls in consequence. In reference to the results obtained by Krauskopf and others, it is note- worthy that they resorted to violent agitation of the liquid, while Regnault and Tammann passed the gas over the surface of the unstirred liquid.I t206 THE VAPOUR PRESSURE OF LIQUIDS would seem that the success of the former was due to the prevention of the formation of a gas film on the liquid surface, not, as they seem to imply, solely to having saturated the gas. The hypothesis does not account equally well for the fact that Regnault found that ether in air and hydrogen gave the lowest partial pressures at the lowest total pressures and that it increased with the latter until, when the gas pressure was approximately 1.5 atmospheres, that of ether was very nearly the same as in vscuo. It is almost certain that at the low pressures the amount of liquid ether was so small that there was no true free surface. Were this the case the difficulty largely disappears, for it has been proved that, even in vacuo, thin liquid films exert a lower pressure than the normal. The somewhat peculiar results obtained with ether in carbon dioxide are explicable upon the above assumption and that of the formation of a com- pound of ether and carbon dioxide.Thus, with increasing pressure of the gas the amount of liquid ether increases, and with it the partial pressure, but, before a true liquid surface has been formed, the increasing concentration of the additive compound has become the chief determining factor and the partial pressure of the ether again falls as the total pressure increases. The matter, which is of importance from the point of view of the deter- mination of the solubilities of gases in liquids and in other ways, is being investigated further in this laboratory. SUMMARY. I. A statical method for the measurement of the vapour pressures of pure and mixed liquids over a considerable range of temperature, iiivolving the use of a very simple form of apparatus, has been devised. 2 . The results of measurements of the vapour pressures of a number of pure liquids are recorded, which prove the method to be capable of yielding results in good agreement with those generally accepted as standard. 3. I t has been proved by experiments carried out in a number of ways, by the author and others, that liquids exert a lower vapour pressure in presence of gases than when in contact with their own saturated vapours only. 4. It has been shown {hat, in the case of any one liquid, the lowering is greater the more soluble the gas ; certain evidence has been adduced which indicates that, with any one gas, the lowering is related to the solvent power of the liquid. 5. It is pointed out that in all cases the amount of gas dissolved is quite insufficient to account for the whole ol the observed lowering and that the differences tend to disappear when the liquid is violently agitated. 6. It is suggested that the concentration of a gas in an unstirred liquid is much greater at the surface than in those parts of the liquid removed from the influence of the forces there operative. This investigation was undertaken at the suggestion of Professor Orme Masson, and the author has great pleasure in acknowledging his continued interest and, especially, his suggestion of the above-mentioned hypothesis. THE UNIVERSITY OF MELBOURNE.
ISSN:0014-7672
DOI:10.1039/TF9151000197
出版商:RSC
年代:1915
数据来源: RSC
|
4. |
Opening address |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 207-211
Robert Hadfield,
Preview
|
PDF (2095KB)
|
|
摘要:
THE HARDENING OF METALS. A GENERAL DISCUSSION. At the meeting of the Faraday Society held on Monday, November 23, 1914, at the Chemical Society, Burlington House, London, W., a General Discussion on The Hardening of Metals took place. Sir Robert Hadfield, F.R.S., President, was in the chair. OPENING ADDRESS BY SIR ROBERT HADFIELD. The Proceedings were opened by the Chairman, who delivered the following Address. In opening the discussion on this interesting topic, “ The Hardening of Metals,” I might perhaps add a few words as to why this subject comes to be discussed here to-night. At the last meeting of the Iron and Steel Institute several valuable papers were presented, but some of these were not so fully discussed as their importance warranted. As all of us know, at the General Meeting of such an Institute, when so much ground has to be covered at one meeting, it is sometimes not then possible to do justice to the subjects brought up for consideration, It was suggested, therefore, that without in any way interfering with the work of the Iron and Steel Institute, there wodd be no objection to our Faraday Society offering facilities for a further and fuller discussion of some of the papers then dealt with, and also further papers relating to the hardening of steel.Such a proposal met with no objections, and so here we are to-night, when I trust an opportunity will be given to each of those present wishing to take part in the discussion and to put forward their views. The topic is such a wide one that I much fear one evening will leave but scant opportunity to do justice to the important subjects which concern all of us.Before steel was introduced, except the handful produced by the crucible or puddling process, the main object of the metallurgist was to keep his metal as soft as possible. The reason for this was that in those days, when heat treatment was practically unknown, an approach t o hardness meant brittleness, and, therefore, the material was useless, 207208 THE HARDENING OF nif ETALS Imagine a wrought iron of hardened or stiffened nature, that is, in some way stiffened up or even case hardened ; a material so produced would have been of little or no valuc. Let me here add, it must not be concluded that this term ‘ ‘ hardened ” always infers the glass-scratching hardness of quenched steel containing high percentages of carbon.As, however, events began to march rapidly, the user soon found himself on the horns of a dilemma. For example, as railroads began to. be introduced, rails were demanded which would not wear out so quickly, and tyres which would not need renewing every few months. In the very nature of things, a new type of material therefore became necessary. It was almost entirely for this reason and cause that modern steel appeared OG the scene, and thereby came the introduction of processes which enabled material to be produced which can be truly termed “ of hardened charac- ter,” that is, hardened by the introduction of carbon as compared with wrought iron in which carbon was practically absent ; in other words, modern steel was invented. Just imagine that bnt little more than fifty years ago there was practically no steel in existence of the grades used on the present enormous scale.When Bessemer’s process was first sprung upon an astonished world in the year 1863, and soon afterwards followed by the great Frenchman, Martin, t o whom the Iron and Steel Institute has just awarded the Bessemer Gold Medal, with his open hearth or regenerative furnace, it was a t once seen that a new era had opened, and thus commenced that marvellous progress first of all on railroads in which rails of what may be truly termed “ the hardened metal steel ” were used, shortly afterwards followed by applications in many other directions. Since that time railroads have opened up this great world of ours, and from this development has sprung every other conceivable one, chiefly helped on by the metallurgist, who bas risen to such new requirements.It was but yesterday, tha.t is, until this unhappy war broke out, we had a world in u hich the enormous quan- tity of some 55,000,ooo tons of steel, or what may be termed “ hardened ” metal, was being used annually and put into service for one purpose or another. Let it be here remarked that the greater the expansion in the use of steel the greater has become the hardness of that metal. There is, of course, still an enormous quantity of material used of mild quality, but except to a very small degree even most of that is much harder than the old wrought iron. The Brinell ball hardness number of wrought iron is certainly under 100, whereas we are now getting up to special steels with no less than 300 Brinell ball hardness nnmber.Yet these special steels can be used with even greater safety than wrought iron, that is, steels which are able to resist severe vibratory stresses, sudden shocks, wear and tear, and all this in a mariner which the toughest and softest wrought iron would not fulfil. As can be imagined, it has required the efforts of many minds t o arrive a t this position of great advance, much of this progress having taken place as a result of the development of alloys of iron with other elements. Prac- tically the first paper on this subject was that which I read in 1887 in connection with my discovery of alloys of iron and manganese known as manganese steel, followed by papers on alloys of iron with silicon, aluminium, nickel, chromium, tungsten, and other elements.It will, therefore, be seen that it may be very truly said the world’s future progress literally depends upon understanding correctly and inTHE HARDENING OF METALS 209 a scientific manner “ The Hardening of Metals,” a subject to which we are going t o give our attention this evening. If we can, therefore, add but a little to our knowledge of this fascinating and important subject, we shall indeed be doing good work at home and abroad. Whilst we are to-night going t o discuss more particularly the rationale of hardening, the reasons why and wherefore rather than practical applica- tions, if we can understand the former, then advantages will be obtained in the latter, for this must assurely take place as night follows the day.We have present with us this evening an authority on this subject, Dr. Beilby, whose fascinating studies on “ The Amorphous Theory of’ the Hardened State ’ ’ entitle his opinion t o special consideration, and we shall be most delighted t o hear him. Dr. Beilby’s theories are now generally accepted for the hardening of metals by cold working. To quote his own words, “ Mechanical deformation produced smaller crystal- line grains, and the increased hardness is due to the crystalline fineness of structure. ’ ’ We are also t o have papers by Professor C. A. Edwards on “ The Hardening of Metals by Quenching ” ; Mr. McCance on “ The Interstrain Theory of Hardening ” ; Mr. Humfrey on “ The Part Played by the Amorphous Phase in the Hardening of Steels ” ; Dr.Desch on ‘ ‘ The Hardness of Solid Solutions,” and “ A Note on Twinnings and the Martensitic Structure ” ; Dr. Martin Lowry and Mi-. Parker on “ Metallic Filings ” ; Professor Howe on “ Hardening with and without Mar- tensitization,” and a communication from Dr. Stead. I would also add that Mr. Heathcote will demonstrate with some apparatus for determining hardness. This question of accurately determining the hardness of steel alloys has become so important, that the Institution of Mechanical En,‘ cwteers has now formed a Special Committee known as the Hardness Test Research Committee, which with its able Chairman, Dr. Unwin, is investigating and discussing all the phases of this important and fascinating subject.In this respect might I also respectfully and with regret refer t o the death of the late Dr. Geheimrath Martens. of Gross-Lichterfelde, whose personal kindness t o me and many other Englishmen who have visited Berlin cannot be forgotten. It was he who, next t o Brinell, Turner, and Shore, helped t o originate and develop correct methods of studying the hardness of metals. Although we are, alas, a t war with his nation, that does not blind or prevent us from appreciating the merits of our enemy. In the same way as we handed back his sword t o Captain von Miiller, so, notwithstanding the unhappy state of war existing between these two great nations, England and Germany, officially as your President of this Faraday Society, whose name in itself indicates English admira- tion and respect for the work of the scientist, we tender our respectful sympathy to the family of Dr.Geheimrath Martens, who added so many interesting contributions t o scientific investigations in matters relating to mechanical and other sciences, including the special one relating t o the hardness of metals. It will be remembered that the metallographic terms *I Martensite ” and ‘‘ Martensitic ” were named in honour of Dr. Martens by the renowned French metallurgist, M. Osmond. Before concluding, I should like t o show a specimen of steel of unusual [ialue and interest, bearing as it does upon the title of our subject, “ The VOT,. Y-T8210 THE HARDENING OF METALS Hardening of Metals.” This specimen, which is illustrated in Plate I, is probably the first t o be exhibited in modern times of an ancient piece of high carbon steel which has been hardened by quenching.The following is the analysis of its composition :- Carbon. Si. S. P. Mn. Cr. Ni. Fe. ‘70 -04 -008 -020 ‘02 trace trace 99.2% It was possible t o obtain a fracture of the specimen, which weighed about 8 oz., was 3 in. in length, zfr in. in breadth, 4 in. in thickness. This showed fine crystalline but rather brittle structure. After removing the scale the Brine11 ball hardness number was found to be 146. On cutting the specimen through with a saw there was found t o be a quit& fair proportion of the original metal still unoxidized. I received this specimen a few months ago from the Superintendent of Archaeology in Western India, ILIr. Bhandarkar.One of the special points is that notwithstanding the large number of specimens of ancient iron and supposed steel I have examined during the last few years, none of them have contained sufficient carbon t o be termed steel in our modern time meaning. This specimen, as will be seen from the above analysis, contains as much as -70 per cent. carbon, which indicates that it can be readily hardened by heating and quenching in wzter. In other words, this material has been in its present condition for probably more than two thousand years, and now, after being heated and quenched, hardens exactly as if it had been made only yesterday, thus showing that in this long interval, and beyond surface oxidation, this specimen has undergone no secular change of structure, or alteration in the well-known capacity of an alloy of iron with carbon t o become suddenly possessed of glass-scratching hardness after being heated and quenched in water or other cooling medium.The photomicrograph of the material (Plate 11) in the original condition “ as received ” shows Pearlitic struc- ture, and consists of elongated and irregularly disposed crystals of sorbitic pearlite upon a ferrite ground mass. The specimen has evidently not been reheated above the Ac, point since it was deformed. The crystallization varies from fine to coarse. In places the structure is blurred, as a result probably of mechanical work. The carbon in the specimen appears t o vary from -30 per cent. t o ‘75 per cent. The area photographed is nearer the low limit.There are seams of slag in certain portions of the specimen, but apart from these the material appears to be of similar type t o ordinary modern carbon steel. The structure of the specimen (Plate 111) after being quenched from 850° C. in water is Martensitic. Although the specimen was exceedingly small-about 7 mm. by 4 mm.-yet the quench- ing has not been successful in preventing a small amount of troostitic structure from developing and somewhat destroying the definition of the Martensitic needles by darkening the ground mass. On the edges of the specimen where no troostitic structure develops the specimen will scratch glass. I therefore show two specimens, one in the original condition as received, the other after having been hardened. As will be seen, after quencbing the steel readily scratches glass.Mr. Bhandarkar assures me there is not the slightest doubt about the antiquity of this specimen from the bars found beneath the stone pillar of Heliodorus at Besnagar, India. The specimen in question wasPLATE 1.-Specimen of Ancient Indian Steel (probably B.C. 125). Full Size. VOL. X-T8--2lOaPLATE 1.-Specimen of Ancient Indian Steel (probably B.C. 125). Full Size. VOL. X-T8--2lOaPLATE lV'.-This Column is known as Khan Haba, in Besn;igar, in the Bhilsa District, Gwalior State (Central India). The inscription on this Pillar states : '* The Pil!,?r WIS erected by a Greek called Heliodorus, son of Dion. who was an amhassaclor dispatchcd by the Greek I t is heliered this en;ibles thc (Me o f the erection of Inimecliately below the lower end of the Pillar, and between i t and the fonntlation stones, t\vo pieces of iron.king Antialki(1;is t o the court of an Indian Prince named BIiagabhadra, who ruled over Central India." the Column to be considered as about 12j ILC.THE HARDENING OF METALS 211 found a t the bottom of the pillar, shown in Plate IV, and dating back to about 125 B.C. Mr. Marshall, the Director-General of Archaeology in India, was present when the base of the column was excavated, and affirms that from all he saw the column could not have been shifted a t a later date, or that the bars found could have been subsequently inserted. On this is an inscription recording that it was a Garuda-dhvaja erected in honour of the god Vasudeva, that is, in front of his temple. Garuda is more or less a fabulous bird, and is believed by the Hindus t o be the vehicle of that god. Garuda-dhvaja means a pillar surmounted by the figure of Garuda, and such pillars are generally put up before temples dedicated to Vasu- deva. The inscription shows that the pillar was erected by a Greek called Heliodorus, son of Dion, who WZS an ambassador despatched by the,Greek king Antialkidas, King of Taxila, t o the court of an Indian Prince named Bhagabhadra, who ruled over Central India. It is, therefore, quite possible that the metal of the bars, which Mr. Bhandarkar thinks are in the form of chisels, may have been forged by Greeks. Probably, however, they were really wedges, and were inserted to steady the column, specially as they were found between the base of the column and the first foundation slab. The specimen examined is folded over and thinned oEt at the edges. The column itself is locally known as Khan Baba.
ISSN:0014-7672
DOI:10.1039/TF9151000207
出版商:RSC
年代:1915
数据来源: RSC
|
5. |
The hardening of metals—introductory paper |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 212-215
G. T. Beilby,
Preview
|
PDF (342KB)
|
|
摘要:
THE HARDENING OF METALS-INTRODUCTORY PAPER. Dr. G. T. Beilby, F.R.S. (Glasgow), then read the following Paper as an Introduction to t h e General Discussion. The subject may be conveniently arranged under the following sections :- I. Hardening of the pure ductile metals results from any form of cold working. 11. Hardening of the pure ductile metals may result from their mixture with each other or with non-metallic elements, e.g. hydrogen, nitrogen, carbon, etc. 111. When these mixtures are ductile they, like the pure metals, may be further hardened by cold working. 11‘. Certain of these mixtures have the property of hardening by chilling. The sudden drop in temperature which results from chilling may act in cwo ways:- ( a ) It may stereotype the form of chemical combination or of structure, crystalline or non-crystalline, which is in the condition of equilibrium at the temperature at which chilling begins.These strains may be wholly or in part either within or beyond the elastic limits. When the strains exceed the elastic limit and permanent deformation occurs, then hardening may result as in 11. As a means of throwing light on the molecular structure of solids the study of the phenomena under Section I has appealed to me very strongly, because of certain great advantages which are inherent to this side of the subject. Chief among these advantages is the fact that in a pure metal we have only one kind of chcmical or ultimate atom, and that in certain ductile metals the chemical molecule is monatomic. It follows that in dealing with a given inass of metal, either ‘( hard ” or ‘( soft,” we are dealing with the same number and kind of atoms or molecules occupying practically the same volume in whichever state the, aggregate is found.In a problem which is essentially a physical one it is no small advantage to know with certainty that chemical questions have been completely excluded. The ease and certainty with which the pure ductile metals, in particular gold, silver and copper, can be passed from one state to the other-soft to hard, or hard to soft-is an important practical advantage in experimental work. Further, these metals recrystallize from the hardened state at temperatures which are so much above the ordinary atmospheric temperature that their permatie~ice in the hardened state is assured, while the fact that their recrystallizing temperatures are more than SooO below the liquefying point at once removes any suspicion ( b ) It may set up contraction strains within the cooled mass.212THE HARDENING OF METALS 213 that the mobility of the atoms or molecules, which is a necessary feature of recrystallization, is of the same degree as the mobility of the iiquid molecules. The problem of the two states in the pure ductile metals can be stated A ceiitimetre cube of metat in the hard state containing x ultimate atoms or molecules has certain well-marked physical qualities. This cube when heated to about 300” loses these characteristic qualities, so that from the physical point of view it might be a different metal. It still contains x ulti- mate atoms or molecules, and the volume they occupy is still practically a centimetre cube.As the number of molecules and the space they occupy are both constant, the iiztrirzsic colzesive .force of the aggregate is also constant, though the efecfive colzesive force, on which the greater or less rigidity of the aggregate depends, will vary ( I ) with changing energy conditions, and (2) with the relative positions of the atoms or molecules to each other $ fhey $osscss polarity. With respect to energy changes in passing from one state to the other, it has been conclusively proved that there is a considerable storage of potential energy in the hardened state and that this storage is associated with a marked reduction in the vibrational freedom of the atoms or molecules.We may picture this as due to the molecules being so pressed together that they curtail each other’s vibrational freedom. It is not, however, quite correct to speak of their being “pressed” together, seeing that the force which enables them to hold each other more closely is their own attraction or cohesion. From the first it has appeared to me most probable that the atoms or molecules do possess polarity, either as an intrinsic property or more probably as a property developed by heat vibration, and that the metal in the hard vitreous condition is truly amorphous, inasmuch as its atoms or molecules are crowded together heterogeneously as regards their polarities. The tempera- ture-elasticity curves, as well as the curves of thermoelectric potential, show the existence of distinct stages in the recovery from strain as the annealing temperature is raised.There is a comparatively quick relief when the metal is heated to IOOO, followed by a slower rate till the crystallizing range is reached. Over the comparatively short Crystallizing range the relief is so rapid that it definitely suggests that a new type of structural change has set in. The microstructure shows that this corresponds with (I) the disappear- ance of the characteristic hardened structure, ( 2 ) the appearance of minute detached crystalline aggregates, and (3) the apportionment of the whole mass among well-developed crystalline grains. It seems natural to picture the earlier steps in the relief of strain as due to the energizing effect of rising temperature on the molecular vibrations.A slight general expansion of the mass will occur, and the potential energy released by this expansion will augment the kinetic energy of the vibrating molecules, but not to the point at which they can free themselves sufficiently to enable them to take their true orientation under the influence of the crystalline nuclei in the mass, which always persist in spite of the most drastic deformation, When the temperature is raised to the beginning of the crystallizing range we may suppose that a proportion of the molecules have attained the necessary amplitude of vibration to set them free to orient, and with rising temperature further progress towards complete orientation and re-orientation takes place over the remainder of the crystalline range till the whole mass is apportioned into crystal grains, each with its own orienta- tion.Re-orieiziation of the freely vibrating molecules must certainly occur as the larger grains impress their orientation on their smaller neighbours or on other partially oriented aggregates. thus :-214 THE HARDENING OF METALS In this type of recrystallization at a temperature which may be as much as 8000 or 1,000~ below the liquefying point, it is not necessary to suppose that there is any transport of molecules from crystal to crystal. Orientation of the molecules only requires that they shall turn in sifu in response to the cumulative vibration of a more powerful crystal. The foregoing illustration shows the remarkably self-contained nature of the problems of Class I , as well as their almost elemcntal simplicity, at least in outline.The qualitative differences between the two states are not disputed. That these are due to the intrinsic properties of the pure metals is proved by the fact that these differences become more sharply defined as metals of greater purity are used, Most of these differences lend themselves to exact measurement, as, for instance, the hardness, rigidity, or mechanical stability, the electrical conductivity, the thermoelectric potential, the energy of solution and the elasticity by the acoustical method. The microstructure can be recorded by photography, though we need to remember that no photographic records, however perfect, can take the place of direct study with the microscope. In every case the thertnally stable crystalline state supplies a fixed standard of reference.No uncertainty attends the passage of the pure metal from the hard to the soft fully crystallized state, neither is there any uncertainty in the reverse change from soft to hard by cold working, but the particular mode by which this working is effected, and the extent to which it is carried, mainly affect the ultimate texture of the hardened metal and therefore also its mechanical qualities. This question of texture has not received the attention which it deserves, yet it absolutely stares us in the face in the sections of drawn or rolled work. It may be so coarse that it is clearly seen by the unaided eye, or so minute that the highest powers of the microscope are needed to resolve it fully, but .whatever the scale of magnitude the texture is always true to its type.I t is the existence of “ texture” in cold-worked materials which makes the value of all density determinations of hardened metal so uncertain. If cold working brings some of the molecules so close together that they interfere with each other’s freedom of vibration, it is clear that if the volume of the hardened mass is not less than that of the corresponding crystalline mass, other groups of molecules must spread themselves out in wider spacing. To return to the centimetre cubes of our illustration, we have seen that the difference between the two must be due to some difference in the arrange- ment of the atoms or molecules in the given space. The greater hardness or rigidity of the hard cube shows unmistakably that the atoms or molecules move over each other with much less facility than they do in the soft cube.There is also an increase of tenacity which points in the same direction, and in addition confirms the suggestion that the cohesion of the atoms or molecules is being used with much greater effect. It must be admitted that these arefacis, explain them how we may. But there are more facts about the hard cube. The elasticity of its atoms or molecules has been seriously impaired. The acoustical test of elasticity which hIr. McCance has used, and the first use of which in this connection he attributes to M. Robin in 1911, had already been extensively used by me and was described in 1907.;: The coincidence of the elasticity curves with the e.m.f.and tenacity curves was also demonstrated and discussed in the same paper. The steps in the release of elastic strain and their correspondence with the course of re- crystallization were also demonstrated and discussed. Two sets of facts have therefore been established ; the mechanical properties show that the atoms or molecules are holding each other more firmly, while the physical * Proc. Roy. Soc., A, vol. 79, ~907, p. 475,THE HARDENING OF METALS tests show that a state of elastic strain exists throughout the aggregate. My interpretation of the facts is that the second set confirm and emphasize the first set in showing that in the molecular structure of the hard cube a proportion of the molecules are either drawn more closely together by their intrinsic cohesion or are turned into positions relatively to each other in which their mutual cohesion can come into more favourable action.Having traced the passage from hard to soft as recrystallization sets in and completes itself over a certain range of temperature, let us now try to follow the reverse operation. If we drop a hammer on the soft cube the change is accomplished at a blow, and no method has yet been devised by which the steps which separate the soft from the hard state can be watched as they occur. The amorphous theory for which I am responsible is that under mechanical disturbance crystalline 1amell;t: or sheets break down momen- tarily into the liquid state and instantly resolidify in a vitreous amorphous condition, thereby forming a hard cementing material which binds the dis- turbed and partially broken-up crystal aggregates together.By carrying out experimental observations on surface flow in metals and other solids it appeared to me that conclusive evidence had been obtained that the inter- mediate liquid state is a necessary and invariable step in the breaking down of the crystal structure by flow. It further appeared that the type of internal structure, the ‘‘ texture ” of cold-worked metals, strongly confirmed the view that the new aggregate shows all the indications of being a concreted inass. This view, that there is an actual liquefaction and resolidification, appears to account in a satisfactory way for the vitreous character of the flowed metal, for it is really the product of a liquid which has been solidified with a rapidity far exceeding that which can be reached by the chilling methods with which we are familiar. This instantaneous cooling is, of course, accompanied by equally sudden contraction, with its accompanying state of elastic strain. It has appeared best that my remarks should be confined to Section I of the division of the subject with which this paper begins. A study of recent papers on hardening has satisfied me that several able workers in the other sections of this field have not quite realized the significance of the well ascer- tained facts in Section I. My aim has therefore been to draw attention to the significance of these facts.
ISSN:0014-7672
DOI:10.1039/TF9151000212
出版商:RSC
年代:1915
数据来源: RSC
|
6. |
The influence of allotropy on the metastability of metals, and its bearing on chemistry, physics, and technics |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 216-239
Ernst Cohen,
Preview
|
PDF (2091KB)
|
|
摘要:
T H E INFLUENCE OF ALLOTROPY ON T H E METASTA- BILITY O F METALS, AND ITS BEARING ON CHEMISTRY, PHYSICS, AND TECHNICS. Professor Ernst Cohen (Utrecht) was to have read the following Paper on “The Influence of Allotropy on the Metastability of Metals, and its bearing on Chemistry, Physics, and Technics,” but as at the last moment he was unable to come over from Utrecht, the Paper was taken as read. Professor Cohen cabled his greet- ings to the meeting, and said he hoped to visit the Society at a 1 a ter date. I. It might seem surprising that a man is invited to take part in a discus- sion on the hardening of metals who never before occupied himself with this important part of melallography. However, I think the Council of the Faraday Society has seen clearly in giving its members the opportunity of becoming acquainted with a number of phenomena unknown up to the present, and which will, in my opinion, modify our conceptions about metals very markedly.My first duty is to thank the Council for inviting me to give here a survey of the investigations I have carried out these last few years on the nietasta- bility of metals in collaboration with my pupils, Messrs. Helderman, Moesveld, van den Bosch, de Bruin and Dr. S. Wolff. 2. One of the starting-points of our work was suggested by the following clauses set forth in a paper by Matthiessen and von Base,::: (‘ On the Influence of Temperature on the Electric Conducting Power of Metals”: “Pure cadmium, when heated to about So0, becomes exceedingly brittle ; in fact, it may be powdered in a hot mortar with great ease.We should not have been able to carry out the determinations if the wires had not been varnished, as the movement of the oil by the stirrer would have caused them to fall to pieces. . . .” 3. From these and other remarks in the paper mentioned we got the impression that we were dealing with allotropic transformations of the metals which had not been observed before. Up to the present we have occupied ourselves with bismuth, cadmium, copper, zinc, antimony, sodium, potassium and lead. As it is impossible to treat the subject here in full, I intend giving a description of the methods by which the general results were obtained, choosing cadmium and lead as examples, these being especially instructive. As for the other metals concerned, I must refer to our papers in the Proceed- ings of’ fhe k’oninklijke Akadeiiiie van Wetciisclzappcii f e Amsterdam + and the 2 eitsch rifl fiir pliysiknliscli e CIi emie.4. Before mentioning the methods used, it may be pointed out that the metals experimented with were of a high degree of purity. Our cadmium (Kahlbaum, Berlin) contained 0.005 per cent. of lead, 0.001 per cent. of iron, and a trace of zinc, whilst the impurities of our lead * Phil. Tram., 152, I (1862). t Procccdiizgs, 16, 485, 565, 628, 632,807 (1913-14) ; 17, 54, 59, 60, 122, 200 (1914). $ Zcitsc1zu.f. plt-itsik. CILciiizc, 85, 419 (1913) ; 87, 409, 419, 426, 431 (1914). 216THE METASTABILITY OF METALS 217 (Kahlbaum, Berlin) were O'OOI per cent. of copper and 0*0006 per cent. of iron.5. The investigation was carried out using the pycnometer, the dilatometer, and also the electric potential method. I. Experirneiats witlt flze Pycnomefer. 6. Our experiments on bismuth had proved that metals may show very great retardation in undergoing molecular changes at temperatures either above or below their transition-points. This reluctance to undergo change is doubtless one of the reasons why the phenomena to be described here remained undiscovered until so late a period. 7. We generally used the pycnometer (Fig. I) described by Johnston and Adams* (volume about 25 c.c.). All determinations were carried out in duplicate (using 16-60 grams of the metals). The difference between two of these FIG. I. never exceeded three units in the third decimal place. Toluene (or water) was used as the liquid in tlie pycnometer, but as a control we sometimes substituted paraffin oil for it.The thermometers in our thermostat (the temperature of which was kept constant within some thousandths of a degree) had been compared with a standard of the Phys.-Techn. Reichsanstalt at Charlottenburg, Berlin. 8. We received the cadmium (bars) i n two consignments, which we shall distinguish by the letters K, and K,. The metal was reduced to turnings on a lathe, washed with dilute hydrochloric acid, water, alcohol, ether, arid dried iu vacuo over sulphuric acid. d 2 8.63j ; 8.632 ; S-633 ; mean 8-633 The density of I<, we found to be : 2 5" 4" in independent determinations, taking fresh quantities each time. * J O Z L Y ~ . Antcric. Clirrii.SOC., 34, 56; (1912).218 THE INFLUENCE OF ALLOTROPY For K, we found : d 2'" 8641 ; S-644 ; 8-642 ; mean 8.643. 4" 9. After heating K, at a temperature of 150" during 95 hours in a current of dry carbon dioxide which was freed from oxygen, we found : d "0 8.630 and 8.633. 4" These figures show that the density of the metal had not been changed by the heating. 10. As there was a possibility that we 'had passed a transition-point, but that cadmium showed similar retardations to those which we had found in the case of bismuth, we heated a certain amount of K, d 2508.643 during 3 days and nights at 100' in contact with a dilute solution of cadmium sulphate. After this time the metal was chilled (at 00) and washed with water, dilute hydrochloric acid (these liquids had been cooled), alcohol, and ether. I t was then dried at 30" in VUCILO over sulphuric acid.Two determinations gave : ( 4" 1 cl 25" 8.633 and 8.633. 4" This experiment showed that by heating at 100" a change had been produced in the metal which lowered its density (measured at 2 5 O ) by 10 units in the third decimal place ; our duplicate determinations prove that- this difference exceeds considerably our experimental errors. 11. In order to determine if a change of density takes place at tempera- tures below 1000, we warmed the metal ( d 3 8.633 and 8.633 again in contact with a solution of cadmium sulphate for 14 hours at a temperature of 60-70". After this the metal was chilled, washed, and dried in the manner described above. 4" ) Its density was now : which proves that there occurs at 60-70" a diminution in the density of 11 units in the third decimal place.12. We repeated the experiment described in $11 with the specimen the density of which was now d s 8.620, keeping it this time for 24 hours at 40.. We found : d 3 S-642 and S.643. Its density (taken at 25") showed an increase of 22 units in the third decimal place. 13. The experiment of $11 was repeated again with the metal that had now a density of d 5 8.642 (8.643). After having kept it for 24 hours at 60-70" in a solution of cadmium sulphate, we found after chilling, washing, etc. : 4" 4" 4" d 25" 8.631 and 8.633. 4" At 60-70" there has been again a dewcase of 10 units in the third decimal place. 14. The experiments described above show that cadmium is able toON THE METASTABILITY OF METALS 219 undergo a reversible transformation.changes we are treating of here are reversible ones. carried out a series of I t may be pointed out here that all the 15. In order to fix the transition temperature of cadmium more closely we 2 . Experiments wiih the Dilatometcr. 16. In order to measure as accurately as possible the changes of volume which the metal undergoes in short periods of time we used 360 grams of our metal K,. We heated it for 24 hours in contact with a solution of cadmium sulphate. I t was then chilled, washed, dried, and transferred into a dilato- meter, which was then filled with paraffin oil. This had been heated for some hours at 200" under reduced pressure in contact with finely divided cadmium, until there was no more evolution of gas bubbles.In order to reduce as far as possible the quantity of this oil, the expansion of which would have made the measurements more troublesome, a quantity of small glass beads was put into the bulb. The dilatometer was now kept at different but constant temperatures by means of an electrically heated thermostat," the temperature of which remained constant within 0.002 degree. (A Beckmann thermometer was used.) By this device the dilatometer f. becomes an instrument of precision. 17. The results are given in Table I. TABLE I. Temperature. 49.60" 62-40" 64-90" 66.90~ 59.60: 60.45 84'40" Duration of the Observations in Hours. 104 5 34 96 4 I 64 6 Rise of the Level of Oil in mm. - 1,500 - 233 - 66 74 + 53 + 267 - 0 Rise in mm.per Hour. - 170 - I9 8 + 3 f 4 4 - 46 I 0 From these measurements one might conclude that there is a transition- point at 64.9". However, our investigations on copper and zinc had shown that such a transition-point may be changed by varying the previous thermal history of the metal. If there were simultaneously present more than two allotropic modifications of the metal this behaviour might be expected. 18. We also found in the case of cadmium that the transition tempera- ture changes in varying the previous thermal history of the metal. The following experiments may be mentioned in this connexion. 19. A certain quantity of cadmium (K4) was divided in two parts [(K,), and (K&j of 500 grams each. (K4)r was reduced into turnings on a lathe and immediately put into a diiatometer.At 69.9" we observed a decrease of volume (469 mm. in 254 hours). 20. (K,), was converted into turnings in the same way and kept for 5 days and nights at 100" in a solution of cadmium sulphate. After having it put into a dilatometer we made the readings shown in Table 11. * Its description has been given in full in the paper, Zeitsclzr. f. plzysik. Clzenzic, t The capillary tube (diameter of bore I mm.) was bent into a horizontal position. 87, 409 (1914).320 T H E INFLUENCE O F ALLOTROPY TABLE 11. Temperature. Duration of the Observations in Hours. 49'6" 60'4" 62-5" 63'1" 63.7" 69.6" Rise of the Level of Oil in mm. - I00 - 125 - I4 + 1s + 83 + 225 j Rise in mm. per Hour. 1 1% - 21 + 45 I + 249 I + 2,700 The transition-point is 62.8".21. The metal was now kept at 100" in contact with a solution of cadmium sulphate for 7 days and nights. After this it was put into a dilatometer, which was heated for 24 hours at 145", then for 24 hours at 270" (that is, only 50" below the melting-point of the metal). We only succeeded in " bringing it into motion " by heating the metal for 48 hours at soo in a solution of the sulphate. We then got the following results : TABLE 111. I ' Duration of Temperature. , Observations Rise of the Level i Rise in inm. per of Oil in mm. I Hour. in Hours. 60.0" 63.00 63.5" 64-00 69.0" - I O j I - 210 - 33 - I1 I - 8 i - 6 + 22 I + 18 I + 58 ; + 348 The transition-point has been changed to 63.4". 22. In this way we carried out a great many experiments on samples with different previous thermal histories.The extreme limits which were found for this (apparent) transition temperature were 69.3" and 61-30. 23. The experiments described above show that it is very difficult, if not impossible, to fix the real transition-point of the pure modifications in this way. We therefore tried to prepare a sharply defined modification of cadmium avoiding high temperatures. 24. For this purpose we electrolysed an ammoniacal solution of cadmium sulphate between an electrode of platinum and one of pure cadmium (40 volts, 20-2 j amperes, surface of the electrodes 26 cm.*). We kept the temperature of the solution below 40", cooling the vessel with ice. The solution was kept homogeneous by a glass stirrer (Witt), which was rotated by a small motor. The cadmium formed at the electrode was washed with dilute sulphuric acid, then with water, alcohol, and ether.After this it was dried at 40". 25. One hundred and seventy grams of this material were put into a dilatometer. As it is very finely divided, great care must be taken in order to remove the air from the apparatus. We used a Gaede pump for theON THE METASTABILITY OF METALS Cd 22 I Solution of cadmium I Cd amalgam sulphate of arbitrary concentration by weight. 12'5 per cent. Cd purpose. The paraffin oil was boiled on this pump with finely divided cadmium. 26. If there had been formed during the electrolysis only one modifica- tion of cadmium we might expect that no transformation would occur in the dilatometer, in consequence of the absence of germs of a second form.From our earlier experiments we know that even if a second modification were present the retardation may be very strongly marked. 27. We found in our first experiment with this cadmium which had been electrolytically deposited that neither at jo", nor at So0, nor at 100" did any change occur. 28. After having removed the paraffin oil we washed the metal with ether and brought it in contact with a solution of cadmium sulphate (12 hours - at 1000, 48 hours results : at 50'). After this the dilatometer gave the following TABLE IV. Temperature. I *O: 94'8 70'5" 70'5" 60.0" 70.0' 65.0" i D ~ ~ ~ ~ ! , $ , ~ e j Rise of the Level of Oil i n mm. in Hours. '- I 1 ~ I Rise in mm. per Hour. - 351 + 132 - 267 + 47 - 138 + 70 - 53 - 468 + 528 - 46 + 6 4 T 46 - 38 There is a claaizge in the direction of motion of the meniscus at a constant tcmperaiure (70.5").29. This change proves that now (viz. after the treatment'of the metal at 100" and 50" with a solution of cadmium sulphate) there are simultaneously present more than two modifications. 30. Our next task consisted in identifying the product which is formed during the electrolysis of solutions of cadmium salts. Tt may be pointed out here that I found in collaboration with Dr. E. Goldschmidt that when a solution of tin salts is electrolysed below 18°C. there is not formed gj'ey tin, as might be expected, but the modification which is metastable at this temperature.222 THE INFLUENCE OF ALLOTROPY 32. By an exhaustive investigation of these cells we were able to prove that some disturbances, observed by Hulett, could be avoided if their forma- tion were carried out in a special way.The e.m.f. is then immediately after construction 0.0503 volt at zs", and generally remains constant. 33. All determinations to be described below were carried out by the Poggendorff compensation method. The resistances used had been checked by the Physikalisch-Technische Reichsanstalt at Charlottenburg, Berlin. Our two standard elements (Weston) were put into a thermostat which was kept at 25". We used as a zero instrument a Deprez-d'Arsonva1 galvanometer. It was mounted on a vibration-free suspension (Julius). The readings were made by means of a telescope and scale ; 0.02 millivolt could easily be measured. 34. Our dilatometric measurements with cadmium which had been electro- lytically deposited gave the result that this material only undergoes trans- formation at temperatures below IOOO if it has been in contact at 50" (100") with a solution of cadmium sulphatc (comp.$28). The probable and obvious conclusion is that by electrolysis we get exclusively y-cadmium, the modification which is stable at high temperatures. If this were the case the y-cadmium would be transformed into @cadmium at IOO", into a-cadmium at 500 in contact with the solution of the sulphate. If now the y modification is really generated by electrolysis (analogous to what happens with solutions of tin salts), the Hulett cells (H.C.) which have been studied until now would contain this material as the negative electrode. If this modification happened to be transformed into the modification which is stable at ordinary temperatures and pressure (I atm.), this would manifest itself by a decrease in the e.m.f.of these cells. 35. On the one hand we are working in this case under extraordinary favourable circumstances for stabilization (change into the a modification), as the material, formed electrolytically, is in a very fine state of division and surrounded by an electrolyte, while the quantity which has to undergo transformation is very small. (The electrode is formed by 20 or 30 mgrs. of the metal deposited on a platinum spiral.) Consequently the transformation, if it occurs, will be finished in a short space of time. 36. On the other hand, and this is to be borne in mind in researches of this kind, the possibility exists that the transformation which has to take place sponfnizeously may be suspended, if the metal deposited by electrolysis forms only one single modification, as the g e r m needed for transformation are then absent.37. That stability generally does not occur is shown by our dilatometric observations as well as by many other facts, i.e. by the experiments of W. Jaeger,::: Ernst Cohen,f Sijl,l and Hulett,$ who all found the same e.m.f. (50 millivolts at 2 5 O ) for cells which were constructed according to the scheme : Solution of Cd amalgam 1 sulphate 1 by weight. Cd electrolytically cadmium 12.5 per cent. deposited How obviously the transformation may be delayed might also be inferred from Hulett s words// : " Many of these cells are still in good order after five years." * Wicd.Ann., 65, 106 (1898). f Zeitsclir. f. pltysik. Chcrnic, 34, 612 (rgoo). $ Ibid., 41, 641 (1902). 9 TYU~S. Americ. Chenr. Soc., 7, 333 (1905). I / Ibid., 15, 433 (1909).ON THE METASTABILITY OF METALS 223 This would be in perfect accordance with our own experiences : Clark cells which contain ZnSO; 6H,O as solid depolarizer preserved their e.m.f. for five years notwithstanding their having been standing at room tem- perature, i.e. 25 degrees below the transition-point of ZnSO, - 6H,O. As in the case of Hulett's cells they had been sealed UF after formation. 38. On account of these observations it might be expected that even under circumstances favourable to a transformation (stabilization) of the negative electrode, only a certain number of H.C.would show the transformation, 39. On December 11, 1913, we prepared three H.C. (Nos. I, 2 , and j) in the following way : We put two platinum spirals into the H-shaped tube B {Fig. 2). Into the right-side tube we put some I per cent. (by weight) cadmium -I I+ FIG. 2. atnalgam. We filled the tubes with a dilute solution of cadmium sulphate (half saturated at IS"). After this a current of I or 2 milliamperes ( I or 2 mgrs. Cd per hour) was passed from the amalgam to the platinum spiral. After having deposited 20 or 30 rngrs. of cadmium on the left-hand spiral, the capillary tube on the right was brought into connexion with a water-pump in order to remove the amalgam. A number of small pieces of 12.5 per cent. amalgam were then substituted for this.40. These cells gave at once an e.tn.f. of o'oj03 volt when they were put into a thermostat at zj°C. After standing for two months at room tem- perature the cells were measured again on February 26, 1914. The e.m.f. of I, 2 , and j had decreased to 0.0475 volt at 23OC., and this value remained unchanged. As might have been expected, the e.m.f. had decreased by stabilization of the cadmium.::: 41. We prepared two new cells (Nos. 6 and 7) in the same way as I, 2 , and j. Immediately after the preparation their e.m.f.s were 0.04847 and * We prepared a large number of these cells : generglly 30 per cent. of them showed stability. This amalgam is a fluid at ordinary temperature.""4 T H E INFLUENCE OF ALLOTROPY 0.04795 volt respectively.Some days later their values became constant 0.04788 and 0-0477s volt. Stabilization had begun already during electrolysis. 42. As our dilatometric measurements had shown that stabilization occurs with great velocity at 500, we'prepared cells (C and 0) at 47.9". The dilute amalgam was then taken out and an 8.5 per cent. (by weight) amalgam was put in, while a fresh solution of cadmium sulphate was used. We substituted an 8.5 per cent. amalgam for a 12.5 per cent. as our intention was to measure these cells also at oo C. At this temperature the 12.5 per cent. amalgam is a monophase system and such a system must not be used.;:: In this way we found at 25" C. : Cell C : 0.04745 volt. Cell 0 : 0'05022 volt. The cadmium in cell C had thus been stabilized at 47.9". 43.In order to check the results found up to this point, we also deter- mined the e.m.f. of oui- stable and metastable cells at oo C. If the differences in e.m.f. at 25" between the different cells were really to be ascribed to the presence of a-cadmium (cell C) and 7-cadmium (cell 0), the difference, which was at 250 C. 2.8 millivolt, ought to increase at o°C., as we are at that tem- perature at a greater distance from the metastable transition-point a-cadmium y-cadmium. The measurements at oo C. gave the following results : Cell C : 0.05225 volt. Cell 0 : 0.05626 volt. While the difference was 2.8 millivolt at 250 C. it has increased, as might be expected, to 4.0 millivolt at oo C. 44. Several phenomena which have /been 'described by Hulett, but which are obscure until now, may find an explanation in the light of our experiments.Hulett says: "A number of cells were made with addition of Cd(OH),, thinking this might make a more uniform cadmium deposit ; also the air was completely removed from three before sealing, and in others the air was removed and the cell saturated with nitrogen and with hydrogen. All these gave very variable results, but in each case only 10 mgrs. of cadmium had been deposited on the spiral, and I have lately learned this is too little cadmium, since some cells prepared as above described, excepting that only 10 mgrs. of cadmium was deposited on each spiral, showed the same irregularities and tendency to constantly decreasing electromotive force. These cells were recently all discharged, and then, reversing the current, about 26 mgrs.of cadmium were deposited on each platinum spiral, and they seem to be all coming together nicely and to the value indicated by the old cells." 45. Our observations agree perfectly with those of Hulett, but we have to add the following restrictions : A number of our cells in which only 10 mgrs. of cadmium were deposited, indicated immediately after formation an e.m.f. of 0.0502 volt at 25" C. which decreased during two days. Then it became constant : 0.047 volt. Transformation into a-cadmium had consequently occurred ; the fact that only a small quantity of cadmium is present causes the e.m.f. to reach very sooii its definite lowest value. The phenomeiion observed by Hulett is therefore the quick stabilization of y-cadmium.46. Professor Hulett has been kind enough to communicate to us the following facts : " Twelve cells which had been sealed after formation * Rijl, 2eifschr.f. plzysik. Ckeiizis, 41, 641 (1902).ON THE METASTABILITY OF METALS 225 remained unchanged from March 18, 1905, to May 7, 1914, i.e. during nine years. The quantity of cadmium on the spirals varies between 3.7 and 13’7 mgrs. of cadmium .” 47. The decrease of e.m.f. which has been observed with cells which contain only 10 mgrs. of cadmium is consequently not to be ascribed to the minute quantity of metal‘:’ deposited on the spirals ; this quantity is much less in the cells which have been constant during nine years. The reason of the decrease in e.m.f. of those cells is the transformation of y-cadmium into a-cadmium.48. In order to check this conclusion we prepared a number of cells (at room temperature) which only contained 5 mgrs. of cadmium on the spirals. Their e.m.f. has been during all this time 0.0505 volt. I A B C D FIG. 3. Some of these remained metastable (0.050 voit), while others were trans- formed into the stable form (0.047 volt) after some days. 49. Up to the present we have only directed attention to the electromotive behaviour of a- and y-cadmium ; the ,.3 modification has not been mentioned hitherto. However, we were struck by the fact that when constructing a large number of Hulett cells we often got ones which had an e.m.f. of 0.048 volt at 25” C. The e.m.f. of cells which originally had an e.m.f. of 0.050 volt at this temperature spontaneously decreased till the value 0.048 volt was reached. After this their e.m.f.remained constant. * Oberbeck found [ Wied. AIZIZ., 31, 337 (1887)l that a layer of metal A of 2 x I O - ~ mm. suffices to give to a metal on which it has been deposited the potential of A. As the surface of the spirals in the H.C. was 0.28 cm.? the layer of cadmium deposited is much thicker.226 THE INFLUENCE O F ALLOTROPY The conclusion was plain that the cells giving 0.048 volt might contain $cadmium, those giving 0.047 volt a-cadmium, whilst those giving 0.050 volt have y-cadmium as a negative electrode. 50. In order to ascertain if the e.m.f of the p cells has a real significance, experiments were carried out on the following lines : At temperatures above the transition-point of the change a-cadmium p-cadmium (which we found in the neighbourhood of 60" by dilatometric measurements) the e.m.f.of a cells must be higher than that of p cells. In cooling the cells below the transition-point mentioned the contrary will occur. 51. Our experiments in this direction we carried out in the following way : We constructed a large number of Hulett cells ; one of these, the e.m.f. of which had been originally 0'050 volt at 25" C., had an e.m.f. of 0.047 volt (at 25O) after having been kept for 4 weeks at 47.5". After this time it remained constant. We combined this cell (No. 7) with another one (No. 2 2 ) , the e.m.f. of which was 0.048 volt at 25". The two cells AB (No. 7) and CD (No. 2 2 ) (Fig. 3) were connected by a siphon H, which contained the same solution of cadmium sulphate as was present in the cells.The lateral tube E of the siphon was closed by a rubber tube F in which was put a glass rod G. The little apparatus was brought into a thermostat which could be kept at will at 250 or 64-50. 5 2 . We measured the e.m.f. between the cadmium which had been electro- lytically deposited 011 the platinum spirals A and C against the common amalgam electrode B (12'5 per cent. by weight). It is absolutely necessary to use a common electrode as the cadmium amalgam of 12.5 per cent. by weight does not form a heterogeneous system at 64.5"; its e.m.f. is thcn a function of its composition. The use of the two amalgam electrodes B and D might give rise to serious mistakes if there were only small differences in their composition.The absolufe e.m.f. of our amalgam electrode against the cadmium in A and C does not play any r6le in our measurements. 53. Table V gives the results of the first series : TABLE V. Temperature 2 jo. Cell 7 . . . . . . . . . . . . . . . . . . 0'047q volt Cell 2 2 . . . . . . . . . . . . . . . . . . 0.04815 ,, E.m.f. Temperature 64.5". Cell 7 . . . . . . . . . . . . . . . . . . 0.04029 volt . . . . . . . . . . . . . . . Cell 22 * - - 0.03979 ,? After having brought the cells to 25" we found : Cell 7 . . . . . . . . . . . . . . . . . . 0.04741 volt Cell 22 . . . . . . . . . . . . . . . . . . 0.04806 .. The table shows that at 64'5" there has taken place an inversion of the A second experiment with two cells (Nos. 4 and 8) newly constructed poles and that the cells regain their original e.m.f.at 250. gave the results shown in Table VT.ON THE METASTABILITY OF METALS 227 TABLE VI. Temperature 2 5 O . E.m.f. Cell 8 . . . . . . . . . . . . . . . . . . 0'04757 volt Cell 4 . . . . . . . . . . . . . . . . . . 0.04839 ,, Temperature 64- 5". Cell 8 . . . . . . . . . . . . . . . . . . 0.04737 volt Cell 4 . . . . . . . . . . . . . . . . . . 0'04633 ,, After having brought the cells to 2 5 O we found : Cell 8 . . . . . . . . . . . . . . . . . . 0.04776 volt Cell 4 . . . . . . . . . . . . . . . . . . 0.04789 ,, From Table VI it may be seen that we are here at the limit of measure- ment obtainable in working with cells of so small an e.m.f., the reproducibility of which is 0-5 millivolt. 54. From the inversion of poles which has been observed we may conclude that the value 0.048 volt at 25" really has significance and is to be attributed to the presence of @-cadmium.FIG. 4. 55. As our investigations led to the conclusion that a piece of cadmium chosen at random which had been produced from the molten metal contains a-, p- and y-cadmium, we carried out some measurements in order to check this result. If such is the case, it might be expected that the potential of such a material against cadmium which has been formed by electrolysis would be zero. 56. We: prepared a certain quantity of electrolytic cadmium (Prep. A) and determined at 40" the potential difference between this material in a solution of cadmium sulphate which was half saturated at 15" C. and (I) cadmium (which we received from Kahlbaum in the form of bars) in a finely divided state (Prep.B) ; ( 2 ) cadmium which we had used in our dilato- metric measurements; in this material the presence of 7-cadmium was presumed (Prep. C).228 THE INFLUENCE OF ALLOTROPY Making use of the small apparatus shown in Fig. 4, we first determined the potential difference between two samples of the same material, sub- sequently that between samples of different preparations. I n this way we found : Emf. of A against A = o.000037 volt ,, B ,, B = O*OOOOI~ ,, ,) c ,, c =o'ooooo ,, Y J A 3 ) =o'oooo37 )> ,, A ,, C =0'000037 ,, 57. From these measurements we see that y-cadmium our preparations, as the dilatometer had shown. is really present in 58. We already pointed out above that the example of lead is also particularly instructive.A year ago we carried out some experiments on the behaviour of lead, but the results were negative. However, some time ago we received a letter from Mr. Hans Heller at Leipsic, in which he described some experiments in this direction which may be summarized as follows : He electrolysed a solution of lead acetate (to which some nitric acid had been added) between lead plates in order to generate a lead tree. After the experiment, the plates having been standing for three weeks in conlact with the solution, they had lost their softness and tensibility ; they formed then a brittle and crumbling mass. Inoculating a plate of ordinary pure lead with the brittle material in contact with the solution mentioned above, Heller observed that the plate fell to pieces.59. Repeating these experiments, we were able to corroborate the statements made by Heller. Fig. 5 shows a plate of pure lead in its original condition. Fig. 6 represents the plate having been in contact with a solution of lead acetate :: during some days (temperature I;") ; Fig. 7 shows it after three weeks in the same conditions. 60. We then carried out an investigation with the pycnometer and the dilatometer in the same way as we did with cadmium and the other metals It may be treated here in a few words, The photographs are natural size. experimented with. The density of the to 11.341, the metal during 3 weeks. At d 3 : 11.313. At 25O (6 4 O metal, which had been originally d 3: I 1.324, increased 4" having been in contact with the solution at 15" 50" (5 x 24 hours) the density decreased and became 2 jo 4" x 24 hours) it increased again and became d - ~ 11'328.61. Our dilatometric measurements gave the following results : At 50.8" the decrease of the level was 700 mm. (34 hours). ,) 74'4" ,, rise ,, >) >) 275 J ? (2$ hours)* 62. From the measurements of 5s 60 and 61 we may conclude that there are more than two allotropic forms simultaneously present. 63. Special attention may be paid to the following phenomenon, which stands in direct relation to the conclusion mentioned: It is generally known that when a bar of any metal which is more electro-negative (resp. electro- positive) than lead is suspended in a solution of a lead salt, the lead is thrown out of solution and a lead tree is formed.We stated that the same phenomenon occurred when our pure lead was placed in the solution mentioned * The composition of the solution was : 400 grams of Iead acetate in 1,000 C.C. of water ; 100 C.C. of nitric acid (density 1.16 at I 5") were added.FIG. 6. FIG. 7. VOL. X-T8-%3aON THE METASTABILITY OF METALS 229 above or in a (neutral) solution of lead nitrate. Both at room temperature or at higher temperatures (500) a lead tree was formed within some days. 64. With lead we are in especially favourable circumstances to observe these phenomena. The electric current which is generated between the stable and metastable modifications which are present simultaneously electrolyses the solution. The metal which is formed electrolytically shows in this case a characteristic form (lead tree), so that the phenomenon is very striking.65. Finally it may be pointed out here that Theophrast as well as Plutarch seems to have been acquainted with the fact that lead is able to undergo a transformation at low temperatures. Theophrast :: (a pupil of Aristotle) says in his book n ~ p i I I v p 6 g : KarrirEpov ycip quai rcai pFL;XipGot~ 4Sq rarijvai iv r+ IIFL;vT~ ~ c i y o v K U ~ XEipGuog Gvrog vmvttco5, X a X l c b 8; $ayijuai. (It is told that tin and lead melted sometimes in the Poiitos when it was very cold during a strong winter and that copper was disintegrated.) Plutarch tells us in his Syniposinca that & K . ~ v ~ L poX6/3bou (pieces of lead) melted sometimes when it was very cold. 66. Summarizing the phenomena observed in our investigations on cadmium, lead (bismuth, copper! zinc, antimony), we must conclude that the pure metals as we have known them until now are metastable systems con- sisting of two (or more) allotropic forms.This is a consequence of the very strongly marked retardation which accompanies the reversible change of these allotropic modifications below and above their transition-points. Employing certain devices (adding an electrolyte, using the metals in a very finely divided state), it is possible to increase the transformation velocity in such a degree that the change of the metastable to the stable form occurs within a short time. As such changes are very often accompanied by marked changes of volume, the material is generally disintegrated.67. As, until now, chemists, physicists, and metallurgists have always dealt with the a, p, y . . . forms together, all the physical and mechanical con- stants of metals which have been determined refer to the complicated metastable systems. These are entirely undefined, as the quantities of the a, p, 7 modifications they contain are not known. 68. Now it is known that a special physical property of any substance at a definite temperature and pressure depends on its allotropic condition. H. F. Weber found the specific heat of carbon (at I o O C . ) to be- 0.1128 in the form of diamond, 0.1604 in the form of graphite, 0.1653 in the form of charcoal. The following constants were found for white and grey tin : I , White Tin. 1 Grey Tin. I---- - -__ Density .. . . . . . . . . . . . . . 7'28 I 5'8 Spec. heat ... ... ... ... ::: ~ 0'05382 j 0.04962 Spec. magn. susceptibility ( x -10~) ... ! -Oa30 ... ~ + 0.025 With the magnetic susceptibility even the sign is changed. 69. The existing data on the physical constants of nietals known until now are thus to be considered as entirely fortuitous values which depend on * Dr. Ch. M. van Deventer at Utrecht has been kindenough to call our attention to this peculiarity.230 T H E INFLUENCE OF ALLOTROPY the previous thermal history of the material used. Those physical constants which refer to a well-defined condition of the metals are so far unknown. I n order to determine these, and only these have a definite signification and are reproducible, we shall have to carry out in the future all measurements for the pure a, /3, y .. . modifications of the metals. 70. Reviewing the earlier literature dealing with the specific heat of metals, I found that it contains already a number of data which prove unequivocally that the specific heat of the metals does indeed depend on their previous thermal history. Le Verrier* stated in a paper published in the year 1892 (“Sur la Chaleur Specifique des Mhtaux”) that the mean specific heat of copper, silver, zinc, lead, and aluminium remains as a rule constant for periods which do not exceed 200-300°, after which it changes abruptly, as Pionchonf also found in the case of iron, nickel, and cobalt. The variation of the total heat (i.e. the quantity of heat required to raise the temperature of I gram of the substance from oo to to C.) with the temperature is con- sequently to be represented by a curve with breaks and not by a continuous one.As a consequence of the retardations in the structural change (change- meizts d’ttat) of the metal, a different vahe of the total heat is found on cool- ing from that on heating. If a certain piece of metal is cooled or heated repeatedly, different values for the total heat are found. If we start from a lower temperature and return to it after having overpassed the break in the curve of total heat, a closed and not a single curve is obtained. Thus, while our investigations proved that there exist more than two modifications of copper and lead, the same fact was noted a long time before by Lc Verrier, using a different method. The same may be said about aluminium.71. That others had never observed the phenomena described by Le Verrier may be explained by the fact that they had not heated their preparations rePeatedly to high temperatures, as he did. We have also observed during our dilatometric measurements that such a transition-point can be overpassed several hundreds of degrees without any effect. If, on the contrary, the metal is repeatedly cooled and heated the transformation is ‘‘ set going.” 72. What has been said about the specific heat holds evidently for every other physical constant. In our paper on the allotropy of bismuth I we pointed out that numerous phenomena which had been observed in the study of density, electric conductivity (also under pressure), conductivity for heat, melting-point, thermoelectric force, the Hall effect, etc., and which had not been explained, may find thcir explanation if the facts recently found are taken into account.73. In this way a new field of research for chemists, as well as for physicists, presents itself. Whilst it will be the task of the chemist to prepare the pure modifications and study their physicochemical properties, the physicist will require to turn his attention to the determiiiation of their physical and mechanical constants. 74. As the phenomena described here have been unknown up to the present, metallurgists have not been able to take them into account when studying the hardening of metals. And yet these reversible transformations, which often go on so very slowly in consequence of the retardations men- tioned above, must play an important r6le when the metals are subjected to changes of temperature.* C. R., 114, 907 (1892). t Ibid., 102, 675, 1454 (rS86) ; 103, 1122 (1886). In full : Arrn. dc Clziin. ef de P l z j ~ . (6), 11, 33 (1887). ; Zcitsclw. f. pltysik. Chcniic, Ss,qrg t1913).ON THE METASTABILITY OF METALS 231 in its stable modifica- tion at to C. This r6le may become fatal if the metals are in contact with electrolytes (water), as these accelerate enormously the transformation velocity. The volume changes which generally accompany these transformations may cause the disintegration of the materials, as we have shown above. Moreover, it is evident that in those cases where corrosion occurs these phenomcna have to be taken into account.M of an arbitrary concentration VAN'T HOFF LABORATORY, UTRECHT, November 1914. APPENDIX, ADDED MARCH 1915. 75. The investigations which have been summarized above have been Some of the most interesting 76. Up to the present, heats of transformation of metals have only been Some months ago Bronsted:? carried out continued during these last few months. results may be mentioned here. determined in one single case. some measurements on the heat of the transformation grey tin + white tin. He found it to be 532 gram-caiories per gram atom of tin at 0°C. 77. For several reasons the calorimetric method used by Bronsted cannot be applied to our case. We therefore carried out our experiments with a lvaitsition cell of the sixth kind which has been described by Ernst Cohen.! This cell is constructed according to the scheme :- Electrode of the metal hl in its melastable modification at lo C.78. Hitherto it was impossible to make a quantitative application of this cell, as no metal, having a transition-point, was known which exists in an electrically sharply defined condition. Our measurements will prove that the transformation a-cadmium C y-cadmium is especially suitable for such an investigation. As we had in view the carrying out of some other measurements with our a- and y-cadmium, we have not brought them together in one single transi- tion cell, but used them as the negative electrodes in cells which were constructed according to the scheme given by Hulett. These cells were studied separately. Consequently our cells were made up as follows :- ~ Unsaturated solution of Cd-a i CdSO of all arbitrary I Cd-amalgam .* * ' 8 per cent. by weight 1 concentration I and- Unsaturated solution af i concentration 1 Cd-amalgam . . . . - . (y-cell) , Cd-y ! CdSo4 Of an arbitrary I 8 per cent, by weight * Zeitsclzr. f. Plzysik. Chcntie, 88. 479 (1914). t Ibid., 30, 623 (1899).232 THE INFLUENCE OF ALLOTROPY 79. On applying the equation of Gibbs-von Helmholtz- E, dE E, = - + T-' 11E d'l' to the a- and y-cell separately, we find- . . , (a-cell) and- dE (Ee)r = ':) + T(J$) . . . . . . (y-cell). (Eela represents the e.m.f. of the a-cell at To ; (Ec)a the quantity of heat which is generated if at To one gram atom of a-cadmium is dissolved in an unlimited quantity of cadmium amalgam (8 per cent.by weight). The signification of (E,)?. and (Ec)r is quite analogous. 80. From our equations we get- The expression on the left represents the heat of transformation which accompanies the change of I gram atom y-cadmium into a-cadmium at To, i.e. the value to be determined. Therefore we have only to measure the e.m.f. as well as the temperature coefficient of the a- and 7-cell at To. 81. As we pointed out above ($ 34), the cells which have been studied by Hulett are our y-cells. From his determinations between oo and 40°C. it follows that- (E,)fP =0'05047 -O0.0002437(t-25) volt . . . . . (2) 82. We constructed our a-cells starting from 7-cells in which the 7-cadmium was transformed into a-cadmium. The way in which these 7-cells were prepared and in which their e.m.f.was determined has been described above. We prepared 11 y-cells. At 25" C. their e.m.f. was 0.0504 volt. After standing for a fortnight at 250 the 7-cadmium was transformed into the p modification, as was shown by the fact that the e.m.f. had decreased to 0.048 volt at 25" C. In order to transform the p modification into a-cadmium we put the cells for a fortnight into a thermostat, which was kept at 47-50 C. We now put a fresh amalgam into the cells, while a fresh solu- tion of cadmium suiphate was also introduced. We found that in 4 cells (out of 11) the p-cadmium had been transformed into the a modification. The e.m.f. of these cells was 0'0474 volt at 2 5 O C. 83. They were systematically investigated at 250, 20°, and 150 C. respec- tively, in order to determine their temperature coefficients.Table VII contains the results. The measurements may be represented by the equation- (E,): = 0'04742 - O*OOO~OO (t - 25) volt . . . . . ( 3 ) 84. The reproducibility of these cells is not less good than that of the 7-cells. Calculating from (3) the e.m.f. of an a-cell at oo C., the value (E,)~0=o*05245 volt is found, while a cell which had formerly been measured at oo C. (comp. $43) gave the value 0.05225 volt. This cell had been prepared at a different time, using different materials.ON THE METASTABILITY OF METALS 333 Date. Jan. 14 15 a m . 15 p.m. 16 Jan. 18 19 a.m. 19 night 19 p:m. 20 21 22 Jan. 23 a.m. 23 p.m. 24 25 Jan. 25 rempe. rature. _ _ 0 25.0 25.0 25.0 25.0 20'0 20'0 20'0 20'0 20'0 20'0 20'0 15.0 I y o 15.0 15.0 25.0 TABLE VII.E.II1.F. of a-Cells (Volt). Cell H,. 0'04751 0'04725 0.04728 0.04848 0.04843 0.04849 0.04832 0.04840 0.04843 0.04833 0.04908 0.04925 0'04959 0'04924 0.04752 0'0472 I 0.04740 0'04797 0.04790 0'04794 0.04837 0.04833. 0.0484 I 0.04836 - - - - 0.04763 0.047 10 0.047 I o 0.0473' - - - - - 0.04850 0.04833 Cell H6. 1 Mean. 0.04758 1 0.04714 1 0.04710 , 0'04742 0.04731 - - 0'04841 - 0.04860 0.04843 0.0493 7 0.04761 1 I 85. In order to calculate the heat of transformation of a-cadmium into y-cadmium at 18°C. we have to introduce the numerical values into our equation ( I ) . From (2) we find- (E,):*u= 0'05217 volt. volt degree' (3) 7 = -0*0002437 ~- From (3) we get- (Ee)F = 0'0488 j volt. dE volt d r a degree' ( ;2yq0= -0'000200 ___ (EC):" - (Ec)~80 = [o*oj217 - 0.04885 - 291 (- 0*0002437 + 0'000200)] 46105 =739 gram calories.If one gram atom of a-cadmium is transformed into y-cadmium, the change is accompanied at 1 8 ~ C. by an absorption of 739 gram calories. 86. It may be pointed out that the temperature at which (Ee)a=(E8)7 represents the metastable transition-point of the reaction a-cadmium q z 2 y-cadmium. If we put ( 2 ) = (3), we find- 0'00305 = 0'0000437 (,! - 25) t = 94'80 c.234 THE INFLUENCE OF ALLOTROPY 87. I t is evident that the heats of reaction of metals with other substances, as they have been determined hitherto, are indefinite. This has been proved in a special paper.* We also pointed out there that the same is the case with the heats of fusion of metals, and that the values found up to the present show, as was to be expected, very large discrepancies, reaching in some cases 80 per cent.Consequently all thermochemical data of the metals must be redetermined with the pure a, p, 7 . . . modifications of these substances. 88. In $ 5 66-73 it was pointed out that the existing data on the physical and mechanical constants of metals known up to the present are to be con- sidered as entirely fortuitous values, since they refer to the indefinite metastable systems which are produced when metals pass from the molten to the solid state. 89. As for the specific heats of metals, we have already mentioned the determinations of Le Verrier, to which only small attention has been given. In those cases where this has been done, they have generally been considered as less accurate.90. The fact that different authors have found (at the same temperature) for a certain metal very different values for the specific heat must partly be attributed to their omitting to take into account the previous heat treatment of the metal experimented with. From the enormous material found in the literature, we only quote here the values for bismuth found at 18°C. by L. Lorenz + to be 0.0303, whilst Jaeger and Diesselhorst give 0.0292. The difference is 3 per cent. But also differences as high as 13 per cent. occur, as might be seen from a paper of Schiibe1.s 91. That these. facts recently have attracted notice on the part of physicists is evident from a paper published by E. H. Griffiths and Ezer Griffiths,[I who say : ‘‘A further possible source of uncertainty is the effect of sudden chilling of a metal when rapidly cooled from a high temperature.” However, they do not mention the reasons which led them to this conclusion.As these investigations, as well as those of Ezer GriffithsV on “The Variation with Temperature of the Specific Heat of Sodium in the Solid and Liquid State,” play an important r6le in our subsequent arguments, we must dwell upon them. 92. Using a special calorimeter in which the metal was electrically heated, they determined the true specific heat of different metals at different tem- peratures. As the substances were raised from a given temperature through very small ranges of temperature (about 1.4” C.) their previous thermal history was not changed by the experiment itself.Evidently this is of high importance, as in the methods in use up to the present (method of mixtures, ice-calorimeter) the substances are heated to considerably higher tempera- tures, by which procedure changes may occur which are uncontrollable. Evidently they were of opinion that the condition of their material was definite. This is most surprising, as they themselves called attention to the fact that previous thermal treatment must be taken into account. It will be proved once more below that the metals experimented with by Messrs. Griffiths did not correspond to definite states, The values they found for the specific heats of Cu, Ag, Zn, Pb, Al, and Cd must therefore be considered as 93. The metals used by the authors (of high purity) were casf.* Proc. Roy. Acrid. Aimtcrdiilit, 17, 926 (1915). t Wicd. AIZYE., 13, 437 (1881). ; d blamdl. Pltys. Techia. Reiclasanstalt, 3, 269 (1900). $ Zeltsckr. f. &norg. Chcniie, 87, 81 (1914). 1 1 Phil. Tr‘zns. Roy. Sac. IdOILdOII, 213 (A), 110 (1914). 7 PVOC. ROJI. SOC. Lo/;dolz, 89 (A), 561 (1914).ON THE METASTABILITY OF METALS 235 fortuitous values. This is the more to be regretted as their measurements were carried out with the greatest care. 94. We shall now consider the experiments on sodium. The high importance of this research with respect to the question which occupies us may be characterized by the following four points- (a) The previous thermal history of the pure metal was strictly defined. ( b ) Every value given for the true specific heat is the mean of 4 or 6 inde- pendent determinations which are carried out under varying conditions (change of the quantity of electric energy supplied).As Tables VIII and IX show, the agreement of the measurements is perfect. (c) The determinations have been carried out with the solid metal from 0°C. up to the melting-point. 0 20' 40- 60. a d l o o Temperature FIG. 8. ( d ) The author has exclusively taken the standpoint of experimental physicists, giving his quantitative results without any commentary. 95. Concerning the item (a) the following may be pointed out. The author had found that the true specific heat of molten sodium can be repro- duced with great accuracy, even if the previous heat treatment of the molten metal is changed within wide limits. The contrary occurs with the soIid metal ; discrepancies of 2 per cent.at the same temperature were found in his early experiments. Referring to this result he says : " The importance of this point was not sufficiently realized in the early determinations, and a large number of otherwise excellent experirnen ts have been rendered worth- less through lack of attention to the precise nature of the previous heat treatment." 96. The metal used in the final experiments was prepared as follows:236 THE INFLUENCE OF ALLOTROPY Temperature. ,Data. __- __ O0 28.820 49'38" 49'27" 49'07" 67'79" 79.15" 8j.65" 90'03~ 95'53" Aug. 28 Aug. 19 Aug. 8 Aug. 16 Aug. 17 Aug. 20 Aug. 21 Aug. 2 2 Aug. 15 Aug. 23 Aug. 24 Spec. Heat. -~ 0'2835 0.2836 0.2820 0.2826 0.29 I I 0.2910 0.2929 0.2893 0.2954 0.2952 0'2955 0.295 I 0'2955 0.2946 0'2949 0.2964 0.2963 0'2945 0.2942 0.3014 0.303 7 0.3018 o*3010 0.30 I 8 0.3083 0.3086 0'3079 0.3 168 0.3181 0.3165 0.3159 0.3 209 0'3208 0.3 208 0-321s 0.3260 0'3254 0.3260 0'295 I 0'3085 0'3 I78 Mean.0.2829 0*2910 0'2953 0'2953 0.2950 0'3019 0'3083 0-317r 0'3209 0.3258ON T H E METASTABILITY OF METALS 237 4 g Arznealed" (Table VIII). The rate of cooling from the liquid state was less than 4" per hour, which was the rate of fall of the bath from IOOO to 86" C. '( Qzbenched" (Table IX). The metal was heated in an oil bath to 130O C., and then rapidly transferred to a vessel of ice-cold water. The metal was enclosed in a case of copper of special form. The determinations with the quenched metal were made starting from the lowest temperature (0" C.) and progressing in steps up to 95°C.If the cooling had taken place very slowly, the values found at each temperature were definite and reproducible. This proves that the values given in Table VIII refer to the stale of equilibrium as the corresponding temperature. The measurements of August Stli, giving the value 0'2953 (at 49-38") having been completed, the metal was taken out of the calorimeter, heated up to 1000 (melted) and allowed to cool rapidly in air. The value then obtained for the specific heat (at 49-38") was 0.3014. Again heating to 100" (melting) and allowing to cool very slowly gave the value 0'2953 (at 49-38"), identical with the value previously obtained for the "annealed " metal. The broken line Apb represents the results of the determinations after the sodium had been quenched, whilst the curve Cpd represents those concerning the annealed metal.Considering the figure more closely, it is evident that sodium is enantiotropic or monotropic; this was not known up to the present.::: 98. We have here for the first time a case where it is possible to obtain a t will, and in nearly quantitative yield, either the stable or the metastable solid modification from a metallic melt. Until now I have not been able to get this result in my investigations on bismuth, copper, zinc, antimony, and lead. In these cases the different modifications were always simultaneously present in the material experimented with. Only with cadmium I have succeeded in preparing the pure a, p and y modifications by preparing electrolytically the y-form, which was transformed afterwards into the 3 or LC modification. 99. From the work carried oiit by myself and my collaborators it follows that a piece of sodium, chosen at random, is at ordinary temperatures in a metastable condition, as there are simultaneously present both the a- and $sodium. 100. This conclusion is proved in a quantitative way by the very exact measurements of Ezer Griffiths. It is to be expected that sodium which has received heat treatment of an intermediate character (between chilling and annealing) will have at a given temperature a specific heat between the values found at the same temperature for the chilled and annealed material respectively. IOJ. The following experiment proves that this is really the case : The metal was melted and allowed to cool freely in air (Samples A and B, Table X). Whilst the specific heat of the annealed metal was found to be 02829 a t 0" C. and that of the chilled 02870, the experiment gave now (at 0" C.) the values of Table X. 97. The diagram (Fig. 8) represents the results graphically. The specific heat values are found now between 0.2829 and 0.2870. Griffiths says : (' Several determinations were made at temperatures between 88" and 94" after a somewhat similar heat treatment, and the same feature is common to all, the values falling between the extremes corre- sponding to the annealed and the quenched states." * Schadler [Lieb. Ann., 20, 2 (1836)] thinks it probable that sodium crystallizes in the regular system, whilst Long ~ u u r i z . Clzem. SUC., 13, 122 (1860)j found that it is also able to crystallize in the quadratic system.THE INFLUENCE OF ALLOTROPY Temperature. O0 40.16" 68.600 68.60" 82-15" 94'02" TABLE IX. N a --"quen clz ed. '' Data. Aug. 29 Aug. 30 Aug. 31 Sept. 2 Sept. 3 Sept. 4 Spec. Heat. 0.2892 0.2874 0.2864 0.2852 0.2973 0-3002 0'2953 0.2992 0.2983 03024 0.3049 0.3073 0'3034 0.3020 0.3038 0.3087 0.3094 0'3095 0.3079 0'3195 0.3213 0.3 192 0-3 200 Mean. 0.2870 0.2981 0'304Q 03089 0.3200 Preparation. TABLE X. Date. April 7 June 4 June 5 June 6 Spec. Heat at oo C. 0.2861 0.2868 0.2866 0.2864 0.2858 0.2844 0.2871 0.2862 0.2855 0.2864 0.2863 Mean. 0.2864 02863OX THE METASTABILITY OF METALS 239 102. These experiments consequently prove in a quantitative way that sodium, as it has been known up to the present, is also a metastable system in consequence of allotropy and that the physical (and mechanical) constants of this metal which have been determined hitherto are entirely fortuitous values. They must be redetermined with the pure (Z and p modifications of the metal. The same may be said about potassium. VAN’T HOFF LABORATORY, UTRECHT, iUarclz 1915.
ISSN:0014-7672
DOI:10.1039/TF9151000216
出版商:RSC
年代:1915
数据来源: RSC
|
7. |
The part played by the amorphous phase in the hardening of steels |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 240-247
J. C. W. Humfrey,
Preview
|
PDF (716KB)
|
|
摘要:
THE PART PLAYED BY THE AMORPHOUS PHASE IN THE HARDENING OF STEELS. Mr. J. C. W. Humfrey, B.A., M.Sc., M.Eng. (Sheffield), read a Paper on “The Part Played by the Amorphous Phase in the Hardening of Steels.” The existence in solid metals of an amorphous or undercooled liquid phase has already been deduced from two distinct points of view, viz. :- I. As being formed in the crystal gliding planes when a metal is severely overstrained . This phenomenon has been particularly studied by Beilby. 2. As normally existing as a thin film or cement between the individual crystals of which a mass of metal is built up. This theory, which was originally independently advanced by several writers, has received strong experimental support from the work of Rosenhain and Ewen, and the present author.’-4 In a paper 4 dealing with the influence of the amorphous intercrystalline cement upon the mechanical properties of metals, the author put forward a suggestion that the recrystallization which occurs in iron and steel at the Ar, change-point does not take place directly, but by the preliminary break- down of the y crystals to an amorphous state, followed by a subsequent formation of the /3 or a crystals from the amorphous.The suggestion was advanced in order to explain the continuous existence of an intercrystalline cement in iron irrespective of its allotropic changes. A somewhat similar idea was advanced many years ago by Barus and is quoted by Roberts-Austen in his Iiztroductioiz to Metallurgy, as follows : ((When iron passes through the temperature of re-calescence, its molecular condition is for an instant almost chaotic.” In the present paper the suggestion is examined further, and from it is deduced an explanation of the hardening of steels.In the idea as originally advanced it was assumed that the amorphous cement which exists during the different allotropic ranges of iron is always of a similar nature, and corresponds in phase with that of the true liquid metal ; further consideration of the matter has, however, led the author to a somewhat different view, and one, he thinks, in which the theory of the formation of an intermediate amorphous phase during recrystallization be- comes more consistent with thermodynamical principles. We know that a substance, in passing from the liquid to the solid (crystalline) form, changes from a state in which the units are irregularly dispersed to one in which they are arranged in some definite and regular manner, which is characteristic of the substance in question Numerous crystallographic researches-chiefly upon crystals having a perfect external form-have resulted in a theory of crystal structure which the author believes to be now generally accepted, and which has indeed received striking 240AMORPHOUS PHASE IN THE HARDENING OF STEELS 241 confirmation from the recent experiments upon the reflection of X-rays from crystal surfaces.The theory in question asserts that the regularity of structure within a crystal is twofold, viz. :-- I. That the centres of gravity of the molecules are arranged together according to one of a series of geometrical devices called " space lattices," in which each point in the lattice is surrounded by a similar distribution of other points.2 . That in each molecule of a crystal the atoms are similarly situated. The author considers that it is the second form of regularity which essentially gives rise to the first ; or in other words, that in crystallization from fusion it is the forces exerted by each molecule upon its neighbours (i.e. forces due to the resultant reactions of its atoms) which bring about the space-lattice structure of the crystal. Now an allotropic change must be considered as being essentially accompanied by a change in the internal structure of the molecules, e.g. a reorganization of the atoms composing them or a change in their number, Thus we have the two gaseous allotropes of oxygen, the common gas in which each molecule contains two atoms, and ozone in which each molecule contains three atoms.When an allotropic change takes place in a crystalline body we must imagine a similar internal rearrangement of the atoms in each molecule ; and, if the new form which the molecules take involves a corre- sponding change in the external forces which they exert upon one another, then the previously existing space lattice may become unstable and a fresh one may be formed characteristic of some other crystal form. But before the reorganization can be completed there must, at least temporarily, be a state of disorder, and it is during this disorder that the author considers that the structure must be considered as amorphous. The intermediate amorphous state may be realized as corresponding to the liquid which would be formed by the fusion of the solid phase stable at the lower temperatures, if the conditions could be so adjusted that the subsequent recrystallization were avoided.Certain cases are known where such a phenomenon can be actually observed. Thus in the case of sulphur the change from the monoclinic to the rhombic and vice versa normally occurs at a temperature of 95.60 C. ; but the rhombic form stable below this temperature can be heated above it, and in this metastable condition melts at IIj" C. ; if, however, the change to monoclinic is allowed to complete itself before this temperature is reached, the monoclinic crystals of sulphur do not melt until 1 2 0 O C.In the case of iron the temperature-tenacity curves obtained by Rosenhain and the present author 3 give strong indications that a similar phenomenon might occur, the rapid drop in the tenacity of both a and ,6 iron before Ac, strongly suggesting that they are rapidly approaching their melting-points ; and that if on heating a sample of iron suitable conditions of pressure could be applied which prevented the recrystallization to the 7, then a true melting- point could be observed in the neighbourhood of 90" C. We know that physical changes of state, such as from liquid to solid or from one allotropic modification to another, do not take place instantaneously throughout the whole mass of a body, but that starting from certain nuclei they proceed gradually outwards.Thus we must conceive that the inter- mediate amorphous phase existing during an allotropic change which occurs normally (Le. by relatively slow heating or cooling) commences to recrystallize almost as soon as it is formed, and is present at any particular instant during the change merely as thin films or layers between the coexisting crystalline phases, the films moving forward as the change progresses. The course of the breakdown of the one crystalline phase to the amorphous would tend to VOL. X-TS - -242 THE PART PLAYED BY THE AMORPHOUS PHASE follow those planes in which the freedom of movement was greatest, viz. the crystal gliding planes and possibly the intercrystalline boundaries. That a second crystalline phase may form after the first has broken down to amorphous necessitates the condition that the mass is not too viscous for the forces of crystallization to overcome the viscosity and to marshal the molecules into their new orientation. We know from Beilby's work that such a condition is not invariably present, since he has shown that any amorphous phase formed by severe overstrain in the cold possesses a definite stability up to certain well-marked temperatures, and only passes back into the crystalline state when these temperatures are exceeded.Thus, if an allotropic change be made, by suitable conditions, to take place at a tempera- ture at which an amorphous phase is stable, then we can conceive that while the breakdown from the one crystalline state may take place, yet the fresh crystallization may be prevented, and that under such conditions the amor- phous phase formed by the breakdown will persist.That the breakdown from a crystalline to an amorphous state may take place in spite of low temperature is clearly shown by the formation of amorphous from the crystalline by overstrain, and also by the changes in certain alloy steels to be referred to later. I n a metal i n wlziclz a n allotropic change noimally takes place at a temfwa- ture well above that at wlziclz the viscosity is suficient to prevent crystallization, abizorinnl condiiioiis-such as rapid cooliizg-may deluy the charge to well below 2 11 is temp era t w e . From the general considerations which have been advanced above we may now pass to the particular case in view, viz.that of steels. Firstly we have the following experimentally determined data :- I. Pure iron undergoes an allotropic change at about 8W C., which results in a complete recrystallization. The latter fact has been clearly proved by '' heat-reliefs," 5 by the microscopical study of strain effects,3 and by other experiments. 2. With increasing percentages of carbon the temperature of the change is gradually lowered until with '9 per cent. of carbon it occurs at a minimum temperature of 680" C. With less quantities of carbon than '9 per cent. the change does not take place at one temperature, but a diminishing amount of iron (containing the carbon in solid solution) is retained down to 680° C., where the change is finally completed. 3. X second change occurs in iron and in steels containing less than '45 per cent.of carbon at about 7800 C., which change is marked by an evolution of heat on cooling, a discontinuity in the magnetic properties, and a smaller though clearly marked discontinuity in the tenacity. As to how far this change represents a truly allotropic one is a matter of some doubt, but we do know at least that it is not accompanied hy any recrystallization, and for this reason does not enter into the present argument. 4. Iron at temperatures above the allotropic change-point referred to in (I) is capable of holding carbon in solid solution, but in the state which normally exists below the change-point it is incapable (or practically so) of holding it. While it seems probable that it is held in solution in the form of carbide, it is not definitely known whether it is as the carbide Fe3C which normally forms when it is thrown out of solution at 6800 C.; certain experiments rather tend to the contrary view. 5. The temperature at which amorphous matter, formed by severe over- strain in the cold, commences to recrystallize freely on subsequent healing, and which conversely may be considered as the minimum temperature at -which crystallization can occur in an undercooled liquid mass on cooling,IN T H E HARDENING OF STEELS 243 bas been determined for mild steel by Goerens7 and found to be in the nieighbourhood of 500° C. While a certain amount of crystallization can proceed below this temperature, it is very small when compared with that which proceeds when it is attained.It is doubtful, however, whether any very exact limits can bc laid down for the maximum temperature below which the amorphous phase is stable. In the cases where it has been formed by overstrain, it is possible that the actual disturbance from the true crystalline orientation may not be the same in all places where an amorphous structure has been formed, and that therefore a lower temperature is needed to bring about sufficient freedoin for the return to the crystalline structure where the disturbance has been slight, than it is where the disturbance has been more severe. The fact that to each temperature to which an overstrained specimen is heated, there corresponds a certain amount of recrystallization (which ceases to increase after a certain length of time at that temperature), rather lends support to such a view.Again, it has been found impsssible entirely to destroy by strain all or even the greater part of the crystalline phase, and therefore we do not know whether crystallization in a purely amorphous mass will begin at the same temperature or with the same freedom as it would in a mass in which a considerable proportion of crystalline entities are still present ; it would appear that a purely amorphous mass would be the more stable of the two. When formed by overstrain in a normally cooled steel, the amorphous matter will be derived chiefly from the a ferrite, since the cementite is practically incapable of plastic deformation. When formed, however, by the breakdown of austenite in a carbon steel the amorphous matter will consist of a mixture of iron and iron carbide, and this mixture may not possess the same viscosity as pure iron.This consideration is referred to below from a somewhat different point of view. Although it is evident from the above that we cannot state a very exact @re for the maximum temperature of stability of an amorphous iron-iron carbide solution, yet it is probable that on heating, the stability must become small in the neighbourhood of 5ooo C., and that conversely an amorphou,s solution undercooled below this temperature will possess a certain stability which will rapidly increase on further cooling. From the experimental data outlined above we see, firstly, that while in the case of pure iron there is a range of about 400" C. between the tenipera- tures at which the breakdown of the y iron occurs and that at which the recrystallization of a iron becomes dificult, yet in the case of a '9 per cent.carbon steel the range is reduced to less than zooo C. The probability of retaining by sudden cooling some of the amorphous phase formed by the breakdown of the y structure is therefore much greater with increasing carbon contents. But there is another factor quite apart from this lowering of the temperature of the change-point which causes the presence of carbon to still further promote the retention. Above the change-point the carbide is in solid solution in the austenite, and when this phase breaks down to amorphous the carbide molecules will still remain closely intermingled with those of the a iron.Kow, before the amorphous can recrystallize, the two different kinds of molecules have to segregate, since the carbide is insoluble in either a or p iron. Such segregation must necessarily be a slow process in an under- cooled viscous mass, and rapid cooling may easily allow the minimum temperature of crystallization to be passed before it has had time to take place. A somewhat similar argument to this last has already been advanced by Rosenhain: but as a reason rather for the retention of austenite than as hindering the crystallization of a iron from amorphous.244 THE PART PLAYED BY T H E AMORPHOUS PHASE It has been shown by Benedicksg that to retain y iron (austenite) in carbon steels pressure, as well as rapid cooling, is necessary ; this fact also follows from the known lowering of the change-points by pressure.Now the change from a crystalline to an amorphous state in metals is, in practically all cases, accompanied by an increase in volume, and in the case under consideration this increase will be still more marked by the fact that crystalline y iron is denser than crystalline a, and therefore much more so than amorphous a. Thus when a certain amount of y iron has broken down to amorphous it will occupy more space in the mass and will tend to compress the y iron that has not already changed and therefore to prevent its doing so, Quenching therefore, besides preventing the crystallization of the amorphous mixture formed by the breakdown of the austenite, may at the same time preserve a certain amount of that austenite unchanged.The more severe the quenching, the more austenite retained, since the rapid increase in rigidity of the amorphous mass as it cools will enable it to exert its pressure more effectively. The theory of the hardening of steels outlined above depends upon two points : (I) that the temperature of the change-point is lowered by quench- ing, and ( 2 ) that during a normal allotropic change there must exist a temporary amorphous state. The influence of thd first point has already been advanced by Grenet,IO the second now shows how this lowering of the change- point may bring about hardness, since it involves the retention of the amorphous state, and such a state is known to be one of particular hardness in practically all metals. We may now pass to the consideration of how the theory agrees with, and is consistent with, the known physical properties and microstructure of carbon steels hardened by quenching, and of certain alloy steels which possess a similar hard structure even when slowly cocled.PIzysical Piv$eriies.--The physical properties of a steel hardened by quenching and of the same material hardened by overstrain are in many ways similar, and strongly point to the idea that a similar form of structure is present, Thus in both cases it has been found that the hard metal possesses a distinctly lower specific gravity than the same samples when annealed. Among other properties in which overstrained and quenched steels vary in a similar manner from the annealed metal may be cited the possession of high retentivity, coercive force, and coefficient of hysteresis.The fact pointed out above, that in overstrained steel the carbide and the iron molecules will remain segregated, while in a quenched steel they will be very intimately mixed in solution, may prevent the similarity from being too close. Il.3icrosfrucfz~re.-The typical micro-constituent of a quenched steel is martensite, which may-according to the carbon contents and the. rate of quenching-vary from an almost structureless mass to a strongly marked . duplex structure of interlacing needles, the latter type being characteristic of high carbon contents and rapid cooling. According to the present theory the duplex structure consists of the two phases y iron (austenite) and an amorphous solution of CL iron and carbide; the form in which the two phases occur together being due, as advanced above, to the tendency of the austenite crystals to break up most readily along their gliding planes, Conditions which produce an almost structureless martensite are those under which, whiie practically all austenite has been able to change to amorphous, yet little or no recrystallization of the amorphous has been possible.Such conditions occur in practice when samples are quenched from not too high a temperature ; in such a state a steel should possess its maximum hardness combined with a freedom from internal stresses, and therefore be in the mostIN T H E HARDENING OF STEELS 245 favourable condition from a practical point of view. Rapid quenching from too high temperatures, which results in the retention of much of the austenite, will produce a less satisfactory result, since besides the fact that austenite is known to be less hard than the amorphous solution, its presence implies internal stresses and liability to hardening cracks and spontaneous fractures.These points are of ,course well known to all who have dealt with the hardening of steels. AZEoy Steels.-The present theory of hardening applies with special aptness to the case of certain alloy steels, which acquire a hard structure even when slowly cooled. Alloy steels, of which the nickel-carbon series may be taken as a typical examples, are found when normally cooled to possess three different types of structure according to the percentage of these elements which they contain, viz. (a) pearlitic, ( b ) niartensitic, or (c) austenitic.The increasing percentages of the added elements produce a progressive lowering of the change-point, and it has been found that the structure of type ( a ) occurs in all steels in which the change-point lies above about 400.C. ; of type ( b ) in all the steels in which the change-point lies between about 400°C. and atmospheric temperature ; and of type (c) in all the steels in which the change-point lies below atmospheric temperature. The last type, however, become martensitic if undercooled to below their change-point. These results show that in all the steels in which a change-point occurs below 400" C. -a temperature which corresponds very approximately to that below which we have seen the amorphous phase is stable-possess a structure and a hard- ness which is similar to that of a quenched carbon steel.The present theory exactly accounts for such a phenomenon, since it states that when the break- down from the y condition takes place at a temperature at which the amorphous phase is stable, a hard amorphous structure will persist, The fact that the partial change to amorphous can, and does, occur in these steels in spite of low temperature is evidence that the breakdown of a crystalline condition is possible in spite of low temperature. Two further points may be emphasized in regard to the bearing of the present theory upon the normal austenitic or martensitic structure of alloy steels. ( I ) Cold work applied to an austenitic alloy causes a new and intensely hard phase to appear along the slip planes of the y crystals,llflZ and this phase has been shown to possess all the characteristics of martensite.The quantity produced by a given amount of deformation is out of all proportion to the amount of amorphous phase formed by an equal amount of deformation in other metals. If we consider, as the present theory suggests, that the crystals in an austenitic steel are in most cases within less than IOOO C. of a temperature at which they spontaneously change to an amorphous state, it is conceivable that when once a certain breakdown occurs along a slip plane due to the deformation, it may proceed with much greater freedom than it would in the case of an ordinary metal in which no such spontaneous change is involved. ( 2 ) The present theory offers an explanation of the irreversible nature of the changes which occur in alloy steel, i.e.that the change from the austenitic to the martensitic structure takes place at a much lower temperature than the reverse change on heating. I t suggests that while the change from the y state to the amorphous on cooling is independent of low temperature, yet the completion of the change, which would be the recrystallization of the amorphous to a, is prevented by it, and that in consequence the amorphous phase persists. The reverse change of amorphous to y crystals on heating will be prevented by precisely the same cause, viz. that crystallization of any246 T H E PART PLAYED BY THE AMORPHOUS PHASE kind cannot take place until the viscosity has sufficiently diminished to allow of the necessary molecular movements.It has been found that about 4000 C. is the minimum temperature at which the change on heating occurs in any of the series of irreversible alloys, and this temperature may therefore be assumed to be the maximum a t which an amorphous solution of iron, nickel, and carbide is stable. It will be remembered that it is the same as the temperature at which the change occurs in the alloys which first show a martensitic structure on normal cooling. The recrystallization either on heating or cooling of an amorphous solution will depend particularly upon its viscosity, and the viscosity itself upon the temperature and the composition of the mixture. In high-speed steels we have the two properties, firstly that of air hardening, and secondly that of the stability of the hard structure to high temperatures.The latter property appears to be given to the steels by elements such as chromium, tungsten, etc., which are known to form definite carbides. It appears probable that these carbides act by greatly increasing the viscosity of an amorphous solu- tion containing them, and so allow the amorphous state to remain stable at much higher temperatures. Temperiizg.-The tempering of hardened steels will, according to the present theory, consist in the progressive crystallization of the amor- phous solutiori formed by the quenching, but the process is possibly rather more complicated than would at first sight appear to be the case, since, from the experiments of Campbell and Haskins (referred to on page 242) it appears to be accompanied by polymorphic changes in the composition of the carbide. The first segregation of an undercooled solution of u iron and iron carbide would, however, agree very well with Beiiedicks theory of troostite as a colloidal system.I t seems possible that the segregation of the two different kinds of molecules might advance some way before an actual crystalline structure, however fine, is formed; in such a condition the structure would correspond to that of a true liquid colloidal solution. We can in this way distinguish between the two phases troostite and sorbite, the former being a partially segregated amorphous mixture of a iron and iron carbide, while the latter is an extremely fine crystalline mixture of the same constituents.It may be noted that the temperature at which a steel loses all the hard- ness which it has acquired by quenching agrees very approximately with that at which it loses any hardness acquired by overstrain. Summary.-The conclusions arrived at in the present paper may be summarized as follows :- The hard structure which can be produced in carbon steels by quenching aiid in certain alloy steels by normal cooling, is due to the presence of a hard amorphous solution of a iron and iron carbide, which solution may be compared to Beilby’s amorphous phase formed by overstrain. To explaiu the formation of this amorphous phase the author advances a theory that the passage of a substance from one allotropic modification to another of different crystalline form involves the temporary formation of an amorphous state, corresponding to the liquid phase of the modification about to be formed.In iron such a change occurs at Ar, ; and if, due to sudden cooling or to the presence of certain alloyed elements, the change-point is lowered to a temperature below that at which crystallization in the viscous mass becomes difficult, then the amorphous form will be retained in a mctastablc form in the cold. Reference may be made at this point to the “high speed” steels.IN THE HARDENING OF STEELS 4 7 The influence of carbon in enabling a hard amorphous structure to be retained by quenching is twofold, viz. :- (I) As above, by its action in lowering the teinperature of the change point; and ( 2 ) I n a more essential manner, by the fact that-owing to their mutual insolubility in the solid state-segregation movements of both the OL iron and the iron carbide molecules must take place in the amorphous solution formed at Ar3, before either can assume a crystalline formation.REFERENCES. I Rosenhain and Ewen : “ Intercrystalline Cohesion in Metals,” Joui-iial of the I n s t . qf illctnls,, No. 2 , 1912, vol. viii. Rosenhain and Ewen : I ‘ Intercrystalline Cohesion of Metals,” second paper, Jozivrtrzl of the Znst. of Mctrils, No. 2, 1913, vol. x. 3 Kosenhain and Humfrey : I‘ The Tenacity, Deformation, and Fracture of Soft Steel at High Temperatures,” Joiirrial of the Iron aiid Stcel Iizst., 1913, No. I. 4 Hurnfrey : *‘ The Influence of Intercrystalline Cohesion upon the Mechanical Properties of Metals,” Iron and Steel Inst., Cnvncgie M e m o i ~ ’ ~ , 1913. 5 Humfrey : “ The Intercrystalline Fracture of Iron and Steel,” Iron and Steel Inst., cavlzcgie ~ ~ ~ c l l L o i l ~ s l 1913. 6 Campbell and Haskins : ‘ I Effects of Heat Treatment on the Colorimetric Test for Carbon in 0.32 per cent. Carbon Steel,” Joirr-rial of tltc Zvon nird Stcel Zmt., 7 Goerens : “ On the Influence of Cold Working on the Properties of Iron and Steel,” Iron and Steel Inst., Carnegic Memoirs. 1911. 8 Rosenhain : I ‘ Recent Advances in the Metallography of Steel,” Staffordshire Iron and Steel Inst., 1910. 9 Benedicks : I ‘ The Cooling Power of Quenching Velocities, etc.,” Jotiriinl qf tlzc Ivoii niid Steel Zizst., 1908, No. 2. I0 Grenet : “The Transformations of Steel within the Limits of the Tempera- tures Employed in Heat Treatment,” Journal of ilie Ir-oii aiid Steel Zrtst., 1911, No. 2. 11 Carpenter, Hadfield, and Longmuir : “ On the Properties of a Series of Irorl- Nickel-Manganese-Carbon Alloys,” Seventh Report of the Alloys Research Com- mittee, Institute of Mechanical Engineers, November 17, 1905. I* Rosenhain : “ Deformation and Fracture of Iron and Steel ” yonr-nal qf flzc Iron Liiid Stcrl Iiist.,. 1906, No. 2. 1913, ?ryTO. 2.
ISSN:0014-7672
DOI:10.1039/TF9151000240
出版商:RSC
年代:1915
数据来源: RSC
|
8. |
The hardening of metals by quenching |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 248-250
C. A. Edwards,
Preview
|
PDF (255KB)
|
|
摘要:
THE HARDENING OF METALS BY QUELL’CHING. Professor C. A. Edwards, D.Sc. (Manchester), read a Paper on ‘‘ The Hardening of Metals by Quenching.” It is well known that steels are not the only metals which are capable of being hardened by the simple operation of quenching from moderately high temperatures. The degree of hardening attained after subjecting steel to this treatment is certainly much greater than i n the case of other alloys, but that fact alone is not sufficient reason for invoking any special allotropic state or condition which is not possessed by other alloys. All that is necessary is to state the simple fact, that steels are the higher members of a group of different alloys which possess the same general characteristics in varying degrees. All alloys which behave in this way have similar constitutions at high temperatures and undergo precisely similar changes as they cool to the tem- perature of the atmosphere. They undergo decomposition at certain critical tcmperatures during which evolutions of heat occur.It is generally admitted that the hardening produced by quenching any of these alloys is in some way related to the effect of the quick cooling upon those critical heat changes. Only when we come to consider in detail what that effect is are we faced with any divergence of opinion. Within certain limits, the effect of varying rates of cooling upon the carbide change and other similar changes is well known. For example, as the rate of cooling is increased the temperature at which those changes occur is progressively lowered, and at the same time the thermal magnitude or intensity is decreased. Therefore, it is certainly in agreemcnt with actual experimental facts to say that as the rate of cooling is increased there is an increased tendency to prevent the inversion and keep the heat of the carbide change inside the cold material. The crucial question is, are the highest rates of cooling we are able to obtain by quenching sufficient to completely suppress the heat change ? The very careful experiments that Professor Benedicks has made certainly indicate that when the highest quenching velocities are attained there are no indica- tions of any heat being evolved as a result of the decomposition of the initial solid solution and subsequent crystallization of fresh constituents.Further, when heat is evolved the specimens contain the constituent known as troostite. There are, therefore, very good reasons for saying that troostite is the first constitutional step in the breaking up of the previous solid solution.Whilst troostite is unquestionably a hard constituent it is not nearly so hard as the constituent from which it is formed. Hence these facts lead one to the view that the hardening is caused by the suppression of the heat of the carbide inversion. In the opinion of some investigators the structures of 2 4sTHE HARDENING OF METALS BY QUENCHING 249 quenched steels do not seem to support this view. It is, therefore, necessary to consider this question very carefully, especially as all other supposed diffi- culties will automatically disappear once this is satisfactorily settled.There can be no doubt that the internal. microstructure of steels and of other similar alloys when at high temperatures is essentially the same as that of pure metals and ordinary solid solutions, i.e. they consist of ordinary crystals with polygonal boundaries. In ordinary steels which have been quenched that kind of structure is never obtained, but an interlacing network of needles which is called martensite is always formed. On that account it has been generally assumed that the original solid solution has undergone at least a partial decomposition and its constituents have become separated into their own particular crystalline form. It ought to be remembered that differences of structure of this kind do not necessarily indicate a real difference in physicochemical constitution, because exactly the same effect can be pro- duced in a perfectly pure ‘metal.When these differences are observed in a pure metal, it is a sign that the crystals with the acicular markings have been strained, and if the straining has been produced under suitable conditions the metal is generally made harder. The view which Professor Carpenter and the writer have advanced is that in the operation of quenching the normal heat of the carbide, or other similar change, is retained in the rapidly cooled material, and when the heat change is suppressed in this way very severe internal stresses are set up, and these cause internal straining of the material. That the material is internally strained is evident from the facts which have been published, namely, the metallic crystals are broken up into an exceedingly large number of twin lam ell^.Further, we believe that the hardness produced by quenching is brought about by crystal twinning and possibly direct slipping, and the formation of amorphous layers as a result of the internal deformation. In disagreement with this theory Dr. Rosenhain says : ‘‘ This view meets with the insuperable difficulty that although quenching does set up severe stresses in steels it does not cause any serious flow or movement. The strain hardening of metals only becomes marked when severe plastic flow has occurred ; it was large strains that were required, that was, the actual linear displacement of particles of matter within the mass had to be large.” The above is no doubt a correct statement of facts when applied to certain cases and under certain defined conditions such as direct tensional straining.It is, however, rather important to remember that the ratio of stresses to strains, and their quantitative effect upon the hardening of metals, varies within exceedingly wide limits even in the same metals, under different conditions. Hence the ternis “ very severe straining ” and ‘( large linear displacement ” do not seem to have any tangible meaning ; their effects upon the hardening of metals are materially influenced by the two factors, time and temperature. There are also good grounds for considering that the effects of compression stresses and strains upon hardening are quite different in degree from tensional stresses and strains.Thus Hanriot has shown that by simply subjecting cylinders of silver to uniform pressures of IO,OOO Iiilogranimes per square centimetre the hard- ness is increased from 1’0 to 1.7. This hardening is brought about without any apparent deformation or change of shape. Then, remembering that steels in the y condition are capable of forming twin crystals, it is not surprising that in the quenching operation large numbers of twin lamellae are formed. This crystal twinning is a definite indication of internal movement. Mr. Humfrey’s theory of hardening is not much unlike the one advancedz j o THE HARDEKING OF METALS BY QUENCHING by Professor Carpenter and the writer. In both theories the hardness is considered to be due to the presence of amorphous material.Hence it is only a question of how that material is brought into existence. Considering that the formation of an amorphous condition from a previous crystalline state must involve an actual absorption of energy, it is not easy to understand how that energy can be supplied from the interior of a slowly cooling mass of material, as Mr. Hunifrey SUF~OSCS. In other words, it does not seem theoretically .possible, nor does it appear necessary to suppose, that, in passing from one allotropic modification to another, crystals must revert to the amorphous state. On the other hand, when crystal twinning occurs as a result of quenching stresses, energy is absorbed and may be stored up in the amorphous layers. Thus it appears to the writer that the amorphous phase cannot be formed without some internal movement.This difficulty is completely overcome if it is admitted that twinning takes place. Turning now to the note at the end of Dr. Desch’s paper, the writer would like to say that the facts brought forward by that author are not opposed to the theory of hardening by twinning and possibly direct slipping. It should be remembered that the Widmanstatten structure seen in slowly cooled carbon steels is not composed of twinned crystals, and therefore the fact that they are soft cannot be regarded as having bearing upon the theory of hardening metals by crystal twinning which is produced by quenching. It may be that the ferrite of steels containing about 0.5 per cent. of carbon is deposited and continues to grow at the twinning or gliding planes, but that is not a proof that actual twinning takes place during slow cooling. Then again, so far as the writer is aware, the Widmanstatten structure has never been observed in a steel of the eutectoid composition. This seems to indi- cate that the Widmanstatten structure which is developed in slowly cooled steels, with lower carbon content, is caused by the gradual growth of the ferrite. I%’hen steels are quenched from high temperatures the crystals are undoubtedly twinned. A severely twinned metal is much harder than the same material in the undisturbed condition ; in fact, this is the foundation upon which ‘l’ammann builds his theory of hardening by the cold working of metals. Of course, the presence of a relatively small number of twin crystals such as esist in annealed brasses does not niaterially affect the hardness of the mass, but exactly the same applies to the direct slipping upon a few gliding planes produced by slight deformation.
ISSN:0014-7672
DOI:10.1039/TF9151000248
出版商:RSC
年代:1915
数据来源: RSC
|
9. |
The hardness of solid solutions |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 251-255
Cecil H. Desch,
Preview
|
PDF (1918KB)
|
|
摘要:
THE HARDNESS OF SOLID SOLUTIONS. Dr. Cecil H. De&h (Glasgow) presented a Paper on “The Hardness of Solid Solutions,” which was communicated by Dr. T. Martin Lowry, F.R.S. Most of the recent contributions to the subject of the hardness of metals have dealt with the hardening produced by quenching, by cold-working, or by surface flow. A complete theory of hardening must, however, also take into account the facts relating to the hardness of solid solutions (mixed crystals). It is well known that the hardctess of the alloys of a pair of completely isomorphous metals, such as gold and silver, is not an additive property, but is always greater than the hardness calculated by the rule of mixtures. For example, in the instance just mentioned an alloy containing equal weights of gold and silver is about twice as hard as either of the com- ponent metals, and alloys of gold and copper, or of copper and nickel, present similar maxima in the curve of hardness.When the two metals form a compound which enters into solid solution with both of its components, the curve of hardness presents a downward-directed cusp at the composition of the compound, with a maximum on each side. This case is very clearly exemplified by the remarkable alloys of cadmium and magnesium,::: the cusp corresponding with the conipound MgCd. The same conditions may be traced, although complicated by other factors, in the alloys of copper with zinc and tin, and of silver with magnesium. The hardness of the elements is a periodic function of the atomic weight, the curve of hardness bearing a close resemblance to the curve of atomic volume.Benedicks found that the hardness is closely proportional to the atomic concentration, that is, the number of atoms in unit volume, expressed by the quotient of the density by the atomic weight, and the rule has been extended to compounds by employing the mean atomic weight (molecular weight divided by the number of atoms in the molecule). A very close agreement is not to be expected, in view of the numerous sources of error in determining the reIative hardness of different solids. It has been shown + that the most considerable deviations are found in substances which exhibit very perfect cleavage, and therefore yield to an abnormal extent when submitted to the usual methods of testing. A closer agreement between theoretical and experimental values for the hardness has been obtained by the application of Van der Waals’ equation to solids.: In the equation- ’ * G.G. Urazoff, Zcitscli. anor<. Chcm., 1911, 73, 31. t J. L. C. Schroeder v. d. Kolk, Pvoc. k. Aknd. Wetcizsclz., Amstevdniii, 1901, 3, 6%. $ I. Traube, Bcr., 1909, 42, 1594 ; Zeitsch. niaovg. Cltcni., 1903, 3, 413 ; 1904, 40, 372 ; C. Benedicks, ibid., rgoj, 47,. 455. See also G. Tatnmann, Bc~iehir~gerr cwtsclrcii rlcri iitiiereia Kriifteil iiiid E~gci~scltnfici~ rlcr Lduiagciz, Leipzig, 1907, p. 31. 25 ITHE HARDNESS OF SOLID SOLUTIONS 2 is termed the internal pressure, and an arrangement of the elements in V2 order of their internal pressure coincides almost completely with the order of hardness.This is illustrated by the following figures, extracted from a table given by Traube, to which the values of the atomic concentration have been added: Element. Potassium . . . . . . . . . Sodium . . . . . . . . . . . . Lead . . . . . . . . . . . . Tin . . . . . . . . . . . . Copper . . . . . . . . . . . . Nickel . . . . . . . . . . . . Carbon (diamond) . . . . . . Atomic Concentration. 0'022 0.042 0'055 0.06 I 0.140 o.1;r 0'290 Internal Pressure (Mega bars). 8,190 18,500 51-500 68,700 236, roo 306.300 5,458,000 On both of these views, the increased hardness of solid solutions is due to an increase of pressure, regarded by Benedicks as osmotic pressure and by Traube as internal pressure in the above sense. Benedicks has calculated the osmotic pressure due to dissolved carbide in a quenched steel containing 0.89 per cent.of carbon to be 137.8 atmospheres, assutning the carbide molecule to be Fe&, or 68.9 atmospheres, assuming it to be Fe6C,. These relations make it clear that the problem of hardness is connected with that of the close packing of the atoms or molecules in a solid. Different views have been taken as to the nature of the packing in the metals. According to Barlow and Pope : those metals which crystallize in the regular system are represented by the cubic closest-packed assemblage of equal spheres. " It must not, however, be concluded that for this reason all the spheres composing the structure are identically treated in the packed assemblage ; some, selected homogeneously, might be differently treated from others and the assemblage yet exhibit holohedral, hemihedral, or tetartohedral symmetry of the cubic system according to the particular type of homo- geneous distribution of the preferentially treated spheres." In the simplest cubic arrangement, each sphere is in contact with twelve others, and compression continued until the intervening spaces are com- pletely obliterated converts each sphere into a regular rhombic dodeca- hedron.On the other hand, Sollas j- assumes comparatively loose packing of the component spheres. The discovery that the reflection of X-rays by crystals bears an intimate relation to the internal structure of the crystal provides a means of investigating this structure. So far only one metal has been examined in this way.Crystals of native copper, when submitted to the action of X-rays, reflect the latter at certain angles, indicating that the unit of the space-lattice for this metal is the face-centred cube.: This structure corresponds with the simple closest- packed cubic arrangement, in which every atom, regarded as a sphere or parallelohedron, is in contact with twelve similar spheres or parallelohedra. Whether closest-packing is absolutely essential to the interpretation of the * Traits. Cltciir. SOL., 190f, 91, 1150. t Proc. Roy. SOL, 1898, 63, 270 ; 1908, %A, 267. +' W. L. Bragg, PIziZ. Mag., I914 [vI], 28,3jj. I t has recently been found possible to apply an experimental test.THE HARDNESS OF SOLID SOLUTIONS 253 experimental results is, however, not quite clear, and the opinion of an expert crystallographer on this point would be of interest.There are many types of crystalline solid solution, and it will be advis- able to consider only the simplest of these at present. The most important pairs of metals which form complete series of solid solutions with one another are : gold-silver, gold-copper, copper-manganese, copper-nickel, gold-plati- num, platinum-iridium, and antimony-bismuth. All but the last of these pairs crystallize in the cubic system. The pair magnesium-cadmium may be omitted on account of the complication introduced by the compound MgCd. This leaves six pairs which are fairly well known, and of which the members crystallize in similar simple forms. In each case, the two corn- ponents have nearly equal atomic volumes ; and these volumes are small, gold, silver, copper, manganese, nickel, and the platinum metals all occurring at or near the minima on the periodic curve of atomic volumes.The following values may be adopted : Au . . . . . . 10'202 Ni . . . . . . 6-64 Ag . . . . . . 10'233 Pt . . . . . . 9.06 . . . . . . Ir . . . . . . 8-61 c u 7.077 Mn 7'44 . . . . . . The case of the alloys of gold and silver may be regarded as typical. The atomic volumes of the two metals are very closely similar. The internal crystalline arrangement of silver and gold has not yet been determined by the method employed for copper, but the external crystallographic characters are almost identical, and it is probable that an exactly similar space-lattice will be found to fulfil the conditions of all three metals.The difference between the atoms of gold and silver is then so small that the two space- lattices should interpenetrate freely without appreciable distortion, and the solid solutions of the two metals should be crystallographically indistinguish- able from their components. The available data are too scanty to decide whether this is actually the case. The native alloy, electrum, is described as similar. It is hoped that an examination of the alloys by the X-ray method may shortly be undertaken. The fact that the freezing-point and solidus curves of the gold-silver system approach so closely to a straight line also suggests that the mixture is of a very simple type, whilst the specific volume is strictly proportional to the concentration, the greatest deviation from the mixture rule being only + 0.2 per cent.Diffusion of silver into gold takes place very readily ; for example, a gold wire, 0.46 mm. in diameter, was coated electrolytically with silver until the compound wire contained 62-7 per cent. Ag. The electrical conductivity was determined, and the wire was then heated at 900". The conductivity gradually fell, and reached a final value after 120 hours. The diameter of the wire was then 0.98 mm., and its conductivity was that of a homogeneous alloy of the same dimensions and composition.:g These facts indicate that the solid solutions of gold and silver are of a very simple character. Nevertheless, the hardness of the alloys is con- siderable, attaining a maximum value which is twice that of the pure metals, whilst the electrical conductivity is correspondingly diminished, passing through a minimum.Of the two explanations which first suggest them- selves, an alteration in the closeness of packing seems improbable in view of the equality of the atomic volumes throughout the alloys. There remains the assumption of a condition of internal strain. On this view, the hardness * G . Bruni and D. Meneghini, Internatioiial3'ournal of ikletallograplzy, 1912, 2, 26.254 THE HARDNESS OF SOLID SOLUTIONS of a solid solution is due to the same cause as that of a cold-worked metal, namely a disturbance of the crystalline structure, giving rise to an amorphous or irregular arrangement of the atoms or molecules, or at least, if the con- ditions be regarded from a kinetic point of view, to a constraint of the atomic or molecular motions.This is an aspect of the problem which has been treated by Dr. Beilby, but has received very little attention from other writers. According to Tammann,::: the increased hardness of solid solutions is due to the fact that “ the attractive force between two different molecules is always greater than the attractive force between two similar molecules in the same mixture.” I t is therefore to be expected that the attraction, and consequently the hardness, should be a maximum in the solid solution containing equal molecular proportions of the two components, and this rule is approximately fulfilled in the case of several of the series investigated by Kurnakoff and Schemtsc1iuschny.f It may be noted that the rule is confirmed by applica- tion to cases in which an intermetallic compound is formed.Thus, the com- pound MgCd forms a continuous series of solid solutions with both of its components. The maximum hardness is not that of the compound, although this is harder than either component, but the curve plotted with atomic percentages presents two maxima, approximately midway between Mg and MgCd and between MgCd and Cd respectively. These are the two com- positions at which there is the greatest admixture of foreign molecules. Direct evidence of a constrained condition of the molecules of a solid solution, or of a disturbance of the internal crystalline structure, is not easy to obtain in the case of metals, but the conditions are more favourable in that of isomorphous salts,.These salts frequently exhibit ‘‘ optical anomalies,” indicating the presence of internal strain, and such a condition is often assumed to be usual. In his very careful work on isomorphism, Retgers has shown that the anomalies do not always occur, and has used their absence as one of the tests of the homogeneity of isomorphous ‘‘ mixed crystals ” required for investigations of density. The most conspicuous examples of crystals exhibiting optical anomalies are those of salts crystallizing from solutions containing dyes. Here there is no question of the interpenetration of space-lattices, and the production of a strain in the crystals is readily comprehensible, but where two salts of almost identical molecular volume are concerned, it seems probable that Retgers’ view is correct, and that optical anomalies are absent when sufficient care is taken to conduct the crystallization slowly and in ,presence of a large excess of the saturated solution.It is usually stated, also, that solid solutions crystallize less readily, or yield less perfect crystals, than pure substances. This is not invariably the case, but the fact has been observed so frequently that it must be regarded as evidence for a certain disturbance of crystalline structure in solid solutions. The heat of formation of solid solutions has hitherto been determined only in the case of isomorphous salts, and then by an indirect method, the heat of solution of the solid solution in water being compared with that of a mechanical mixture of the components in the same proportions. The pairs of salts which have been examined satisfactorily by this method-NaC1-KC1, NaBr-KBr, NaI-KI, KC1-RbCl, KCI-KBr, KBr-KI, KC1-KI, NaN0,-NaNO, I t is, however, by no means necessarily the case.* Leltrbuclt dcr Metallograplzie, Leipzig, 1914, p. 332. t Zeitsclz. anorg. Clzenz., 1908, 60, I . ; Zeitsch. plzysikal. Cltem., 1889, 3, 497.CRYSTAL TWINNING AND MARTENSITIC STRUCTURE 255 -all give results indicating a negative heat of formation for the solid solution.‘:’ For two of these pairs, namely KCI-KBr, and KBr-KI, the hardness curve, as determined by measuring the pressure required to produce flow through an orifice, passes through a maximum not far from the middle of the series.? It will be seen that the data which are available for drawing a conclusion as to the hardness of solid solutions, even in such a simple case as the alloys of gold and silver, are very scanty.The complete proportionality which exists between density and composition by volume (or between specific volume and composition by weight) seems to exclude any important dif- ferences in the closeness of packing. There remains a difference in the attractive forces between the atoms-what is usually called internal pressure, and even here it is not easy to realize why the attractive force between gold and silver should be twice as great as between gold and gold or between silver and silver. A distortion of the space-lattice, producing a more or less irregular arrangement, is only compatible with the conservation of the original density on the assumption of some form of packing which is not the closest possible.It would therefore seem that the solution of the problem must be sought from the crystallographers, whose newly added resources enable them to determine the direction of the planes of closest packing within a crystal, and the actual form of the spzce-lattice. t NOTE ON CRYSTAL TWINNING AND T H E MARTEN- SITIC STRUCTURE. Dr. Cecil H. Desch also contributed a “Note on Crystal Twinning and the Martensitic Structure.” The present note deals with certain points in the paper communicated to the Spring Meeting of the Iron and Steel Institute by Professors Edwards and Carpenter, and with some experiments undertaken as a result of the dis- cussion on that paper. One of the conclusions drawn by the authors on that occasion was that martensite and austenite were constitutionally identical, and differed only in the repeatedly twinned structure of the former.The hardness of martensite was attributed to the formation of amorphous material a t the surfaces of slip on which twinning occurs. The so-called “acicular ” structure of certain quenched aluminium-copper alloys was also regarded as due only to twinning. The structurc of these alloys is distinctly “ marten- sitic,” and the analogy between them and the martensitic steels is undoubtedly a close one. It is difficult to believe that the only difference between austenite and martensite lies in the repeatedly twinned structure of the latter constituent, accompanied by the formation of amorphous material.As Professor Bene- dicks and Dr. Stead, among others, have pointed out, distinct differences are observable in the behaviour of the two constituents when present togcther in the same specimen of steel, and in the change of volume which occurs when austenite is cooled in liquid air. Hannemann has also shown reason for believing that austenite and martensite behave differently when resolved by tempering. The present writer has never accepted the hypothesis of a hard niodifica- tion of iron, and has always regarded the hardness of martensite as being connected with a partial resolution of austenite. That the resolution should * G. Bruni, Mem. R. Accad. Lirzcci, 1912, p. 50. t T. I3. Vrshesnevsky, J . Rim. Plzys. Clzcnz. SOL, 1911, 43, 1364.256 CRYSTAL TWINNING AND MARTENSITIC STRUCTURE proceed along gliding and twinning planes by preference is only in accordance with the usual behaviour of solid solutions, as evidenced by the formation of the Widmanstatten structure when such solid solutions undergo partial resolution,:: and by the fact that corrosion proceeds preferentially along twinning planes.+ Narrow twins (Neumann’s lines) are readily developed in many metals.Professors Edwards and Carpenter have instanced the case of tin. Bismuth is also a convenient material for exhibiting the effect. Fig. I shows a portion of a crystal grain in a small ingot of cast bismuth, every grain of which is traversed by numerous Neumann’s lines, as well as by broader and less regular twin lamellz. In cast zinc the lines may be so closely packed and so narrow as to simulate very closely the structure of martensite.Whether these lines owe their origin to casting strains or to strains produced in cutting and filing the ingot is not decided, and there is no evidence that the abundance of very narrow twin lamellz is accompanied by increased hardness. The presence of a inartensitic structure without increase of hardness is illustrated by Figs. 2, 3, and 4, which represent fields from a piece of over- heated steel. A piece of steel plate, 0.28 per cent. C, was heated to a white heat in a coke fire and allowed to cool down with the furnace. A very coarse structure, like the Widmanstztten structure but with less regular outlines, was developed. The ferrite cell-walls were fringed with ferrite (Fig.3) and the pearlite areas exhibited a remarkable pseudo-martensitic structure (Figs. 2 and 3) which had all the appearance of coarse martensite. Tested with a needle it proved to be quite soft, as might have been expected from the very slow cooling. Examined with an immersion objective the whole mass proved to be made up of pearlite (Fig. 4), closely packed in some parts and more loosely in others. Such a structure may, of course, be regarded as pseudomorphic after martensite, but it is impossible for the martensitic structure to have been pro- duced by anything of the nature of chilling. The writer is inclined to take the view that resolution of the original austenite takes place most readily along gliding and twinning planes, and that the distribution of the pearlite in the slowly cooled specimen records the earlier stages of the resolution. On this view, the martensitic structure of a quenched steel is an accompaniment, and not the cause, of its hardness. The writer is aware that this suggestion does not provide any explanation of the hardness of martensite, but only opposes the view that that hardness is due to twinning. Martensite is so much harder than its components as to suggest the presence of a solid solu- tion, but whether that solid solution be a distinct phase, not yet recorded in the equilibrium diagram, or an unusual type of highly strained solid emulsion, remains doubtful. The investigation of the distribution of pearlite in a slowly cooled steel is being continued. * * See paper by N. T. Belaiew in y. Inst. Melals, 19x4, vol. 12, and discussion. t S. Whyte and C. H. Desch, ibid., 1914, 11, 235.FIG. I.-Twinning in Cast Bismuth. FIG. 2.-Mprleiisitic Structure in slowly cooled Pearlite, VOL. X-T9--256aFIG. 3.--E’errite Fringe and Pearlite showing hfartensitic Structure. FIG. +-Pearlite in same specimen wit11 z inin. Objective of 1.4 NA.
ISSN:0014-7672
DOI:10.1039/TF9151000251
出版商:RSC
年代:1915
数据来源: RSC
|
10. |
Note on crystal twinning and the martensitic structure |
|
Transactions of the Faraday Society,
Volume 10,
Issue 5,
1915,
Page 255-256
Cecil H. Desch,
Preview
|
PDF (1546KB)
|
|
摘要:
CRYSTAL TWINNING AND MARTENSITIC STRUCTURE 255 NOTE ON CRYSTAL TWINNING AND T H E MARTEN- SITIC STRUCTURE. Dr. Cecil H. Desch also contributed a “Note on Crystal Twinning and the Martensitic Structure.” The present note deals with certain points in the paper communicated to the Spring Meeting of the Iron and Steel Institute by Professors Edwards and Carpenter, and with some experiments undertaken as a result of the dis- cussion on that paper. One of the conclusions drawn by the authors on that occasion was that martensite and austenite were constitutionally identical, and differed only in the repeatedly twinned structure of the former. The hardness of martensite was attributed to the formation of aniorphous material at the surfaces of slip on which twinning occurs. The so-called “acicular ” structure of certain quenched aluminium-copper alloys was also regarded as due only to twinning.The structurc of these alloys is distinctly “ marten- sitic,” and the analogy between them and the martensitic steels is undoubtedly a close one. It is difficult to believe that the only difference between austenite and martensite lies in the repeatedly twinned structure of the latter constituent, accompanied by the formation of amorphous material. As Professor Bene- dicks and Dr. Stead, among others, have pointed out, distinct differences are observable in the behaviour of the two constituents when present togcther in the same specimen of steel, and in the change of volume which occurs when austenite is cooled in liquid air. Hannemann has also shown reason for believing that austenite and martensite behave differently when resolved by tempering.The present writer has never accepted the hypothesis of a hard niodifica- tion of iron, and has always regarded the hardness of martensite as being connected with a partial resolution of austenite. That the resolution should * G. Bruni, Mem. R. Accad. Lirzcci, 1912, p. 50. t T. I3. Vrshesnevsky, J . Rim. Plzys. Clzcnz. SOL, 1911, 43, 1364.256 CRYSTAL TWINNING AND MARTENSITIC STRUCTURE proceed along gliding and twinning planes by preference is only in accordance with the usual behaviour of solid solutions, as evidenced by the formation of the Widmanstatten structure when such solid solutions undergo partial resolution,:: and by the fact that corrosion proceeds preferentially along twinning planes.+ Narrow twins (Neumann’s lines) are readily developed in many metals.Professors Edwards and Carpenter have instanced the case of tin. Bismuth is also a convenient material for exhibiting the effect. Fig. I shows a portion of a crystal grain in a small ingot of cast bismuth, every grain of which is traversed by numerous Neumann’s lines, as well as by broader and less regular twin lamellz. In cast zinc the lines may be so closely packed and so narrow as to simulate very closely the structure of martensite. Whether these lines owe their origin to casting strains or to strains produced in cutting and filing the ingot is not decided, and there is no evidence that the abundance of very narrow twin lamellz is accompanied by increased hardness.The presence of a inartensitic structure without increase of hardness is illustrated by Figs. 2, 3, and 4, which represent fields from a piece of over- heated steel. A piece of steel plate, 0.28 per cent. C, was heated to a white heat in a coke fire and allowed to cool down with the furnace. A very coarse structure, like the Widmanstztten structure but with less regular outlines, was developed. The ferrite cell-walls were fringed with ferrite (Fig. 3) and the pearlite areas exhibited a remarkable pseudo-martensitic structure (Figs. 2 and 3) which had all the appearance of coarse martensite. Tested with a needle it proved to be quite soft, as might have been expected from the very slow cooling. Examined with an immersion objective the whole mass proved to be made up of pearlite (Fig.4), closely packed in some parts and more loosely in others. Such a structure may, of course, be regarded as pseudomorphic after martensite, but it is impossible for the martensitic structure to have been pro- duced by anything of the nature of chilling. The writer is inclined to take the view that resolution of the original austenite takes place most readily along gliding and twinning planes, and that the distribution of the pearlite in the slowly cooled specimen records the earlier stages of the resolution. On this view, the martensitic structure of a quenched steel is an accompaniment, and not the cause, of its hardness. The writer is aware that this suggestion does not provide any explanation of the hardness of martensite, but only opposes the view that that hardness is due to twinning. Martensite is so much harder than its components as to suggest the presence of a solid solu- tion, but whether that solid solution be a distinct phase, not yet recorded in the equilibrium diagram, or an unusual type of highly strained solid emulsion, remains doubtful. The investigation of the distribution of pearlite in a slowly cooled steel is being continued. * * See paper by N. T. Belaiew in y. Inst. Melals, 19x4, vol. 12, and discussion. t S. Whyte and C. H. Desch, ibid., 1914, 11, 235.FIG. I.-Twinning in Cast Bismuth. FIG. 2.-Mprleiisitic Structure in slowly cooled Pearlite, VOL. X-T9--256aFIG. 3.--E’errite Fringe and Pearlite showing hfartensitic Structure. FIG. +-Pearlite in same specimen wit11 z inin. Objective of 1.4 NA.
ISSN:0014-7672
DOI:10.1039/TF9151000255
出版商:RSC
年代:1915
数据来源: RSC
|
|