ANALYTICAL CHEMISTRY.1. ANALYSIS OF STEELS.IN recent years the steel analyst has had to devise methods for the deter-mination of the elements found in steel which are not only quick and accuratebut can be easily adapted to the routine examination of a large number ofsamples. When it is remembered that nearly one third of all the elementsare to be found in steels, many of them in a single sample, details of themethods used in steel analysis will be of value to many besides the steelchemist. To a large extent the necessary speed and accuracy have beenachieved by the employment of physicochemical methods, either as individualmethods or in combination with chemical separations.In the following review the basis of accepted knowledge is the methodswhich are to be found in text books such as “ Chemical Analysis of Iron andSteel,” by Lundell, Hoffman, and Bright, and “ Sampling and Analysis ofCarbon and Alloy Steels,” being the methods of the United States SteelCorporation.Reports on the determination of each element are precededby three headings dealing with the general application of the spectrographto steel analysis, the use of the photoelectric absorptiometer, and polaro-graphic and potentiometric methods. To avoid repetition of matter dealtwith under these and other headings appropriate reference numbers arerepeated a t the head of the report on each element.Application of the Spectrograph.-It has been found possible to determinewith fair accuracy elements present in steel in small amount by using anarc and an hternal-standard method.Greater reliability, however, isobtained by the use of a spark,l particularly where the upper electrode ismade of graphite-a method which, when standardised, is applicable tohigher percentages of the alloying elements giving results varying only byO ~ O Z - O - l ~ o from the chemical figures. The graphite electrode may bereplaced by silver, and a counter electrode made of high-purity aluminiumhas given good results.2 Disturbance of the lines of one element due to thepresence of other elements may be a source of error. Such disturbanceshave been recorded by W. Holzmiiller,3 who gives a list of suitable lines foraome twenty elements present in steel. Characteristic lines in the ultra-violet have also been measured.4 The grating spectrograph, using a spark gapirradiated with ultra-violet light, has been adapted to routine use withsome success in the U.S.A.5 Modifications of the old spectroscope (“ steelo-scope ”) and of the modern spectrograph for use in the visible part of the1 F.G . Barker, Iron and Steel Inst., May 1939, No. 1.2 T. Torok, Spectrochim. Acta, 1941, 2, 26.4 L. A. Ignatieva, and N. N. Sobolev, Zavod. Lab., 1938, 7, 949.5 S. Vigo, A.S.T.M. Bull., 1940, No. 107, 7.Z . anal. Chem., 1938, 115, 81.V. K. Prokofiev, Zavod Lab., 1940, 9, 1267GIASKIN : ANALYSIS OF STEELS. 205spectrum (“ steelometer ”) 7 are in everyday use, Results obtained withthe latter instrument have an average deviation from the chemical figuresvarying from 0.03 to O.OS~O.Use of the Photoelectric Absorptiometer.-It has been remarked that thelaboratory is regarded as the bottle neck of production in metallurgicalwork.8 The removal of this bottle neck has been achieved by the applica-tion of the spectrograph and the photoelectric absorptiometer, and the authorquoted above has shown how satisfactorily this has been achieved by usingthe absorptiometer.8* Experienced users are able to obtain accuracy inthe determination of steel constituents as high as that obtainable with chemi-cal methods in only a fraction of the time required by the latter methods.Moreover, with only a slight loss of accuracy the absorptiometer can beeasily used for routine work by relatively inexperienced assistants with properdirection.Polarographic and Potentiometric Methods.-Although some success hasheen achieved with the polarograph the method cannot be said to be suitablefor routine testing,lO and clearly it is more likely to be of use in the deter-mination of micro-additions of alloying elements.The removal of all theiron is difficult, and traces left in solution adversely affect subsequentpolarograms.11 Greater success is possible when elements are determinedby amperometric titration.Potentiometric methods are not uncommon and they will be foundunder later headings. A. M. Zanko l2 has described suitable procedures forsome eight alloying elements.AZurninium.3~ 5-The separation of small quantities of aluminium fromlarge amounts of iron is the chief problem of this determination.P. Klingerl3has compared four methods, vix., ether extraction, electrolysis, oxidationwith sodium peroxide, and precipitation with cupferron. Of these, the firstgives the best results, the electrolysis giving low figures in the presence ofcopper and erratic results in that of titanium and vanadium. Cupferron isunsatisfactory. Sodium hydroxide precipitation of the iron gives a suitablefiltrate for precipitation of the aluminium with ammonia 14 or for its detectionby a colorimetric method using ammonium aurintricarboxylate,15 which willS. S. Rimlyand, Bull. Acad. Sci. U.R.S.S., 1040, 4, 225.E. J. Vaughan, “ The Use of the Spekker Photo-electric Absorptiometer in Metal-* Idem, “ Further Advances in the Use of the Spekker Photo-electric AbsorptiometerMonograph published by the Institute of Chemistry, 1942.lo M.Von Stackelburg, P. Klinger, W. Koch, and E. Krath, I’ech. Mitt. Krupp:l 1 G. Thanheiser and J. Willems, Mitt. Kaiser- Wilhelrn Inst. Eisenj’orsch., 1939, 21,l2 Trudy Vsesoyuz Konferentsii Anal. Khim. Akad. Nauk S.S.S.R., 1939, 1, 303;l3 Arch. Eisenhlittenw., 1939-40, 13, 21.1 4 S. Shinkai and T. Nagata, J . SOC. Chern. Ind., Japan, 1939, 42, 3970.16 L. P. Adamovich and A. J. Zagorulko, Zavod. Lab., 1939, 8, 1318; Khkm. Referat.lurgical Analysis.”in Metallurgical Analysis.”E’orschungsber., 1939, 2, 59.65 ; Arch. Eisenhattenw., 1939-40, 13, 73.Khisn. Referat. Zhur., 1940, No. 2, 64.Monograph pubiished by the Institute of Chemistry, 1941.Zhur., 1940, 6, 70206 ANALYTICAL cHIcMIeTBp.detect 0.01 "/o of aluminium.Determination of aluminium photometricallyis possible by using eriochrome-cyanine.16 Percentages of aluminium from0.1 to 0.001 present in the hydrochloric acid-soluble part of a steel can bedetermined spectrographically by using a strong spark and a long pre-sparking peri0d.l' The aluminium line 396143 is compared with iron linesat 3973066 and 3951.16. G. Hartlief,l* using Al 3961-5 and Fe 3963,describes a method employing the Feusner spark.Antimony3 and A~senic.~-No new methods other than those indicatedhave recently been described.Beryllium.-Percentages of this element between 0.1 and 1.0 have beendetermined without serious interference from nickel and chromium by aspectrographic method.19 Beryllium lines 2494.6, 3321.1, and 2650.6 werecompared with iron lines 2622.9, 3440.6 and 3286.8, and 2644.7 and it waslater found possible to extend the range of beryllium percentages to0 0 1 -2 * 0.Bor~n.~--hTo satisfactory chemical methods for boron present to theextent of 0.003% have been described, but a spectrographic method20using graphite electrodes impregnated with hydrochloric acid has given resulhfor boron percentages between 0-03 and 2.01.Unequal volatility of thesteel components cauaes changes in the intensities of the lines measured,the best pair being B 2497.72, Fe 2533.8.CuZcium.3-Determination of this element in cast iron as oxalate after aniron separation has been described.21Curbon.3-Chemical methods for the determination of carbon dependalmost exclusively upon combustion in oxygen, and variations in methodare mainly concerned with the treatment of the carbon dioxide and improve-ments in the apparatus. In both macro- and micro-chemical methods thegaa is absorbed in baryta.In the former case the carbonate formed isconverted into barium sulphate,22 contamination by atmospheric carbondioxide being avoided; in the latter case the excess baryta is titrated.23For this purpose A. Lassieur z4 recommends potassium hydrogen phthalate.G. Zaffuto25 has described a rapid method in which oxides of sulphur areabsorbed in aqueous sodium chloride and the carbon dioxide in standardsodium hydroxide, followed by titration with oxalic wid. Factors affectingthe accuracy of the combustion method have been discussed,26 and for rapidwork drying of the oxygen with magnesium perchlorate is re~ommended.~7l6 W.Koch, Arch. EisenhWtenw., 1939, 12, 69.1' 0. Schliesamann, &id., 1940, 14, 211., I e 0. Masi, Spectrochim. Acta, 1941, 1, 501.21 A. T. Sveshnikov and T. V. Boretskaya, Zauod. Lab., 1938, 7 , 1428.22 H. Kempf and K. Abresch, Arch. Eisenhtittenw., 1939-40, 13, 136.23 M. H. Kalina and T. L. Joseph, Heat Treat. Porg., 1939,%, 169.21 Compt. rend., 1938, 20'9, 731.25 Atti X Cong. intern. Chim., 1938, 111, 487.26 E. T. Saxer, R. E. Minto, and R. A. Clark, Blast Furnace Steel Plant, 1941, 29,27 Idem, &bid., p. 619.1* Ibid., 1939-40,13,295.2o Idem, ibi&., p. 462,718In the physical field, carbon contents of 06--1*1% have been correlatedwith magnetic saturation,28 and a similar rapid method compases themagnetic permeability of a sample with a known steel.29Chromium.1~ 3* 4, 5s 7, 8 0 9,119 120 16-Little change is to be noted in thechemical method for the determination of chromium. The steel is &a-solved in aulphuric and phosphoric 80 or perchloric acid,31 followed bypersulphate oxidation in the presence of silver nitrate and subsequenttitration of the dichromate with ferrous sulphrtte and permanganate. Vola-tilisation of the chromium as chromyl chloride enabled W.Dietz 32 to deter-mine this metal iodometrically in the condensate. Methods dependentupon the absorptiometer have shown great increases in speed combinedwith accuracy.Originally the coloured compound used was the dichromate.*Thk requires control of the acid ooncentration and the addition of urea toreduce silver complexes to colourless compounds and perchromate acid todichromate. For small amounts of chromium the violet-red compoundformed when diphenylcarbazide is oxidised by chromate in acid solutiongives a greater photo-cell response.9 For this method the chromate solutionis obtained after precipitation of the iron, etc., by sodium hydroxide andperoxide, with subsequent addition of sulphuric acid to the alkaline chromateaolution. Interference from 2 yo of vanadium was sucoessfully eliminatedby the use of spectrum-green filters. Both W. Koch and B. Bagshawe 33have used the diphenylcarbazide colour, the latter employing the Lovibondtintometer for comparison of his colours.V. F. Maltrtev and T. P. Temi-renko 34 used this same colour in the presence of the iron, comparing it withstandard steel solutions to which chromium had been added.Potentiometric methods allow of the determination in one solution ofchromium, manganese, and vanadium. In one of these35 three titrationswith ferrous sulphate with appropriate treatment of the solution in betweengives figures for all three elements. In another,36 the permanganate istitrated with sodium arsenite, the chromium and vanadium with ferroussulphate, and the quadrivalent vanadium with potassium permanganate.Some of the spectrographic methods have already been mentioned.Recently, P. Habitz 37 has discussed the choice of homologous pairs of linesand selects Ck 3128, Fe 3167, and Cr 3147, Fe 3167.In the visible spectrum,steeloscope methods38 for 0.02-@13:6 of chromium me Cr 6208, Fe 522728 B. A. Rogers, K. Wentzel, and J. P. Riott, Trans. Amer. SOC. Netals, 1941, 29,969.H. H. Blosjo, Trans. Amer. Found. ASSOC., 1939, 47, 469.30 E. C. Pigott, I d Chern., 1940, 18, 283.31 L. Silverman and 0. Gates, Ind. Eng. Chem. (Anal.), 1940, 12, 618.32 Angew. Chern., 1940, 53, 409.Zavod. Lab., 1941, 10, 357.36 J. Mummedal, Arch. Math. Naticrvidenskab, 1941, 44, 1 ; C'hem. Zentr., 1941, IT,8 0 A. S. Goralnik, Zauod. Lab., 1941, 10, 267.37 Spectrochim. Acta, 1941, 2, 158.33 J . SOC. Chem. Ind., 1938, 57, 260.641.L. V. VoIkova, Bull. Acad.Sci. U.R.S.S., 1940, 4, 216208 ANALYTICAL CHEMISTRY.for 0.1% or more and Cr 4254, 4274, 4289 respectively with Fe 4247, 4282,4271 for 0.1% or less.CobaZt.31 12-Although a-nitroso- p-naphthol remains a satisfactoryreagent for the determination of cobalt, greater speed with similar accuracyis claimed for an electrometric method 39 in which the cobalt and any man-ganese are titrated with potassium ferricyanide in the presence of ammoniumcitrate. Photometric methods depend upon the cobalt colour in hydro-chloric acid solution. K. Dietrich40 describes a variation of H. Pinsl's41method using a Leifo polarisation photometer with an incandescent lamp anda 668 filter. Pins1 treated the filtrate from a zinc oxide precipitation withstannous chloride and used a Nitra lamp with an S 66 filter.E. Bischof andG. G e ~ e r , ~ ~ using a Pulfrich photometer, measure the absorption due to acobalt ammino-complex. Interference due to manganese can be avoidedby the addition of ammonium chloride, and nickel in excess of 1% requiresthe preparation of an extinction calibration curve.Copper.3~ 5.12, 754opper can be determined electrolytically in thepresence of the iron by a number of methods. Working with a cell designedfor low temperatures, H. A. Frediani and C . H. Hale 43 obtained satisfactoryfigures over a range of copper content of 0.03-6.44y0. Alternatively, theelectrolysis may be carried out at 60-70" from a sulphate solution withthe addition of hydrazine, a pure aluminium rod being used as anode,u orfrom a sulphuric-phosphoric acid solution a t 0-6 amp.without stirring.45A colorimetric method 46 depending upon the photometric evaluation ofcolloidal copper sulphide is satisfactory in the presence of 20% of tungsten,2% of aluminium, 6% of molybdenum and20% of chromium. I n this methodand in another due to K. Quande14' vanadium interferes. Quandel,using a Zeiss Pulfrich photometer with an HG 578 filter, measures the coppercolour with rubeanic acid. After removal of the iron with ammonia, thisauthor also measures the copper colour with thiocyanate, using an HG 436filter with a mercury vapour lamp and correcting for the copper retained bythe iron precipitate. Two organic reagents are recommended, uix., salicyl-aldoxime 48 and dibromohydroxyquinoline 0xalate.4~ I n the latter casecopper is completely precipitated as a complex, the iron remaining in theacid solution.Titration of copper with potassium cyanide, solution of theiron being avoided, is proposed.5030 G. J. Steele and J. J. Phelan, Gen. Elect. Rev., 1939, 42, 218.40 Metallwirts., 1941, 20, 600.In&. Eng. Chem. (Anal.), 1940, 12, 736.I4 E. V. Smekh and A. M. Naigovzen, Zavod. Lab., 1940, 9, 1218. *46 L. Silverman, W. Goodman, rand D. Walter, Ind. Eng. Chem. (Anal.), 1942, 14,I6 G. Bogatzki, Arch. Eisenhuttenw., 1941,14,661.47 Ibid., p. 601.I8 E. Stengel, Tech. Mitt. Krupp: Porschungsber., 1939, 2, 87.'@ A. M. Zanko and A. J. Bursuk, Ber. Inst. physikal. Chem. Akad. Wiss. Ukr., 1938,I1 Arch. Eisefihuttenw., 1940, 13, 333.Angew.Chem., 1941, 54, 238.236.89; Khim. Referat. Zhur., No. 10, 92.P. I. Schportenko and V. F. Gtbren, Zavod. Lab., 1938, 7 , 1199UASKIN : ANALYSIS OF STEELS. 209Lead.4eparation of the lead as sulphide and subsequent determinationas sulphate 51 or electrolytically as dioxide 52 are suitable procedures.Reviewing available methods, E. Gregory and others 53 note that leadsegregates in steel and recommend its separation as sulphate.ikfanganese.l* 31 41 5. 7, 8 . 9 , 10- 12*35*36--Both volumetric and colorimetricmethods depend upon the formation of permanganate. Persulphateoxidation in the presence of silver nitrate as catalyst, followed by reductionwith atandard arsenite solution, appears to have superseded the bismuthatemethod.Osmic acid is also suitable as a catalyst.54 G. I. Rodin 55 suggeststhe use of thiosulphate for the final titration. Corrosion and heat-resistingsteels and carbide-bearing steels are best examined for manganese by per-sulphate oxidation after a zinc oxide separation of the iron and aluminiumfrom perchloric acid solution^.^^ C. M. Johnson 57 has reviewed the use ofperchloric acid in the manganese determination and has discussed the effectof molybdenum, vanadium and cobalt. uses the per-manganate colour for the absorptiometer, eliminating interference due tonickel and chromium by a difference method. A potentiometric titrationwith potassium permanganate 58 can give accuracy equivalent to the volu-metric bismuthate method. In addition to the spectrographic methodsalready noted, 0.MasiS9 suggests adapting for steel a method due toA. Rivas 60 used for determining manganese in pure salts.12* 37-Despite the considerable simpli-fication of the molybdenum method obtained by the use of the a-benzoin-oxime precipitation,61 considerable attention is still devoted to the deter-mination of this element. To avoid the ignition of the a-benzoinoximeprecipitate, W. W. Clarke G2 and C. Sterling and W. P. Spuhr 63 convert thisinto lead molybdate. Pure precipitates of lead molybdate can be obtained bytreating the neutral filtrate from a sodium hydroxide precipitation of theiron with formic acid, ammonium chloride and paper pulp and then precip-itating with lead acetate.64 W. W. Clarke 65 dissolves ignited molybdenumtrisulphide in sodium hydroxide, removes sodium tungstate, and thenprecipitates the lead molybdate.Butyl acetate can be used to extract thecoloured compound formed by the action of stannous chloride and potassiumE. J. VaughanMoZybdenum.l* 39 41 5* 7 p 8* 9 9 lo,6 1 A. E. Pavlish, J. D. Sullivan, and J. Shea, Met. and Alloys, 1939, 10, 150.52 G. E. F. Lundell, Met. Progr., 1939, a, 383.b3 E. Gregory et al., J. Iron Steel Inst., Advance Copy, Mid-April 1942.54 R. P. Forsyth and W. F. Berfoot, Ind. Cibem. Eng. (Anal.), 1939, 11, 626.5 5 Zavod. Lab., 1904, 9, 111.6 6 B. Bagshewe, J. SOC. Chem. Ind., 1939, 58, 106.s8 M. J. Eenis, E. I. Grenberg, A. M. Zanko, and L. N. Novikovca, Zavod. Lab., 1940,6s Met. ItaZ., 1938, 30, 111; Chim.et I d . , 40, 681.6o Berheft : Z . Ver. deut. Chem., No. 29; Angew. Chem., 1937, fio, 903.61 H. E. Knowles, J . Re8. Nut. Bur. Stand., 1932, 9, Paper No. 453.62 Chemist-Analyst, 1940, 29, 83.6s Chemist-Analyst, 1941, 80, 81.Iron Age, 1938, No. 26, 142, 16.9, 1082.In&. Eng. Chem. (Anal.), 1940, 12, 33E. Gregory, R. B. Foulston, and F. W. Gray, Analyst, 1941, 88,444thiooyanate on molybdenum solutions, and this ooloured extract can bewed for the colorimetric determination of the element.66 G. M. Poole 67has adapted this method for as muoh as 6% of molybdenum, and D. H.Heppell finds it satisfactory for rapid work. R. Sped 6v has used anether extracti~n,~O but E. J, Vaughan,& using controlled conditions, mewuresthe fhiocyanate colour in the presence of the iron.9.Klinger 71 has reviewed all the methods available for molybdenumother than the benzoinoxime separation as follows. Greater reliability isobtained by hydrogen sulphide separation under pressure, but this is notabsolutely necessary. Molybdenum may be weighed as trioxide or as leadmolybdate after separation as trisulphide, and the latter method is unaffectedby the presence of copper. Of colorimetric methods, those using thio-cyanate and stannous chloride, and phenylhydrazine are satisfaotory, butthe xmthate colour is not reliable. Photometric methods are useful forrapid work. Potentiometric methods give good agreement with the gravi-metric figures, but the stannous chloride titration requires great care.Spectrographic methods have been mentioned.The steeloscope can beused for rapid determinations of 0*05-0*3% of m0lybdenum.7~ P. Habitz 37selects as suitable homologous pairs Mo 2807, Fe 2783 apd Mo 2775, Fe 2779.NickeZ.1- 8 0 4 * 9 * 1 0 * 12~31, *5-Recent work haa dealt almost exclusivelywith the titration of nickel with cyanide either chemically 311 45 or potentio-metrically,73* 74 the latter method being preferred for rapid work. Separ-ation of the nickel is not necessary provided that all the iron be oxidised.A modification of an earlier method for copper 75 makes it possible to deter-mine nickel polarogrephicdy 76 in the presence of excess ammonia andprecipitated ferric hydroxide. The polarogram is started at -0.76 v., whichia beyond the copper steps at -0.09 v.and -0.34 v. Satisfactory resultshave been obtained in the presence of chromium, tungsten, titanium, andvanadium.Niobizam.4ee Tantalum.Phosphorw.8* 99 16- 53- @-Few gravimetric methods have been studiedreoently, but a satisfactory referee method eliminates arsenic with hydro-bromic acid and converts the precipitate of phosphomolybdate into leadmolybdate. The modern tendency has been to adopt a colorimetric methodin which the ammonium phosphomolybdctte is reduced to molybdenum-blueby stannous chloride.8 This method requires that interference from arsenic,silicon, and vanadium should be eliminated, and that allowance should bemade for the reduction of excess ammonium molybdate, a reaction which isL. H. James, Ind.Eng. Chem. (Anal.), 1932, 4, 89.6 7 Iron Age, 1941, 148, No. 15, 62, 145.69 Chem.-Ztg., 1940, 64, 363.7 1 Arch. EisenhUttenw., 1040, 14, 667.72 J. P. Belkevitch, L. E. Bruk, and N. S. Sventitskii, Zuvod. Lab., 1940, 9, 1279.73 R. Weihrich, Arch. Etkeoa?&Wew., 1940, 14, 55.74 0. Niezoldi, Chem. App., 1938, 26, 389.7 5 G. Thanheism and G. Msassen, Naturcoiss., 1937, 25, 426.76 J. S. Lialikov and J. I. Usatenko, Zavod. Lab., 1938, 7, 1100.6 8 Id. Chem., 1940, 16, 173.70 C. D. Braun, Z. anal. Chem., 1863,2, 36GASKIN: ANALYSIS OF STJEELS. 21 1largely inhibited by the presence of iron. Interference from silicon can beavoided by fuming, complex vanadium molybdates do not reduce to molybedenum-blue, and arsenic can be removed by boiling with hydrobromic mid.J.L. Hague and H. A. Bright 77 remove arsenic in this manner in a methodbased on measuring the trammittanoe of a phosphate solution to whichammonium rnolybdate and hydrazine sulphate have been added. A. J.Burmk,'8 measuring the molybdenum-blue colour, states that 0.03% ofmsenic causes no interference. T. P. Hoar,79 who uses an aliquot part of Bsodium hydroxide solution of a phosphomolybdate precipitate for treatmentwith ammonium molybdate and stannous chloride, fin& that as much 880.1 yo of arsenic does not affect the results.The yellow colour due to phoapho-vanrtdo-molybda;te is used by GIBogatzki 80 in a method where the iron colour is masked with sodium fluoride.H. H. Willard and 33. J. Center 81 use a Coleman spectrophotometer formeasurements with the same coloured compound.Diluted nitric acidsolution containing small amounts of phosphoric acid gives a delioate tur-bidity with strychnine molybdate, it reaction which can be used for a rapiddetermination of phosphorus.fsSiZicOlz.l* 3* 49 5# 7, 9-Separation of silioa from a hydrochloric aoidsolution is the basis of the chemical method, and to increaee the speed of theseparation S. N. Shkotova 82 adds a 0.1% solution of gelatin. This addition.does not affect a subsequent determination of phosphorus in the atrate.If this method is used in the presence of phosphoric acid, zirconium tmdtitanium are co-precipitated.& The volume of silicic acid which separateswhen a hydrochloric acid solution of steel is centrifuged is propurtional tothe amount of silica present.a This method is satisfactory down to a lowerlimit of 0.10/, of silicon, but tungsten interferes.A modified procedure canbe used for silica as low as 0°04y0.85 An unusual method for the separationof the silica depends upon an electrolysis with the sample as the anodein a bath oontaining sodium chloride, potassium bromide, and sodiumcitrate.g6 A high current is used and the silica remains in the insolubleresidue. A photometric method which can be used in the presence of tung-sten depends on the formation of soluble yellow sficomolybdatea in weaklyacid Spectrographically, some difficulties arise with siliconowing to a variation in intensity of the silicon line for the same concentrationof silicon in different alloys.88 In using the steeloscope and the line 3905,7 7 J .Res. Nut. Bur. Stand., 1941, 26, 405; Research paper 1386.78 Zavod. Lab., 1939, 8, 12. 7D Analyst, 1938, 6% 112.8o Arch. Ekenhfittenw., 1938-39, 12, 195.8 1 Ind. Eng. Chem. (AnaZ.), 1941, 13, 81. Zavod. Lab., 1939, 8, 213.K. L. Weiss, Arch. Eisenhuttenw., 1941, 15, 13.R. Ishii, Sci. Papers Inst. Phys. Chem. Res. Tokyo, 1930, 36, 491.Idem, ibid., 1940, 37, 143.A. Skrspski, A. Bielanski, and M. Sobieuki, Hutnuik, 1938, 10, 460; Met. Abetr.,1939, 10, No. 4, 226.87 R. Weihrich and W. Schwartz, Arch. EhnM$tenw., 1941, 14, 601.V. I(. Prokofiev, Compt. rand. A d . 8ci. U.B.B.S., 1941,239,449212 ANALYTICAL CHENZISTRY.and with a carbon steel as a standard electrode, a decrease of current pro-duces a greater decrease of intensity of the silicon line than of the iron wherethe proportion of silicon exceeds 1 yo .89SzcZph~r.~~--Sulphur is still determined by one of the three methods,combustion of the steel in oxygen and absorbing the sulphur gases produced,absorbing the hydrogen sulphide resulting from a hydrochloric acid attack onthe steel, or by dissolving the steel in an oxidising acid liquor with subsequentweighing of barium sulphate.The last method is the basis of a modernreferee meth0d,~3 which gives an accuracy within 0.0015% of the weight ofthe sample. Earlier, P. SchonggO had claimed high accuracy with thismethod in which analytical procedure is specified. By using the evolutionmethod and absorbing the gases in ammoniacal zinc sulphate it is possible totitrate this mixture directly with iodine after acidiiication.91 T.P. Hoarand G. E. S. Eyles 92 treat the steel with hydrochloric acid in an atmosphereof carbon dioxide, and after absorption of the hydrogen sulphide in am-moniacal cadmium chloride add an ice-cold mixture of sulphuric acid andpotassium iodate and iodide and titrate with thiosulphate. An accuracy of0.00270 is claimed. The nature of the absorbent of the sulphur trioxide inthe combustion method is not critical, hydrogen peroxide, iodine, andsilver nitrate being equally satisfactory.93 A combustion method D4 whichinvolved trapping the gases in potassium iodide and iodate has been criticisedby G. I ~ i m a r u , ~ ~ who states that carbon dioxide affects the result.A variantof this 96 is to absorb the oxides of sulphur in a little water to which a littleiodine and starch have been added, further additions of iodine being made asdecolorisation occurs, a method which has recently been recommended byS. M. Gutman and R. V. G~chfeld.~' Y. Kanamorigs obtains an accuracyof &O.OOl% by absorbing the gas in hydrogen peroxide and expellingcarbon dioxide from the absorbent with chlorine-free air. The use of tinas a catalyst during the combustion has been suggested.99Tantalum and Nwbium.3-T'he problem of the separation of these twoelements owing to the absence of selective reagents has not yet been entirelysolved. Spectrographically, 0. Schliessmann 100 has described a proceduresuitable for niobium contents greater than 0.1% and for tantalum greaterthan l.Oyo.The steeloscope has been used for niobium in the presence oftitanium and zirconium.101 Separation even of the mixed oxides is not anI. S. Kirin and N. S. Sventitskii, Zavod. Lab., 1940, 9, 1270.Ghem.-Ztg., 1930, 03, 364.91 J. Zeutzb, Z. anal. Chem., 1939, 116, 102.93 M. K. Chukavin rand M. N. Markelova, Zavod. Lab., 1938, 7, 1455.94 A. Vita, Stuhl und Eisen, 1920, 40, 033.Q6 Nippon Kinzonkii Sakkai-Si, 1939, 3, 60.O 6 I. Kassler, Chern.-Ztg., 1933, 57, 573.9 7 Zavod. Lab., 1938, 7, 399.sQ G . I. Stukanovskaja, Zavod. Lab., 1938, 7, 1455.loo Tech. Mitt. Kmpp : hrSChUngSbeT., 1939, 2, 186.Iol A. Fedorov, Bull. A d . Sci. U.R.S.S., 1940, 4, 212.02 Analyst, 1939, 64, 666.Tetsu-to-Hagane, 1940, 20, 630GASKIN: ANALYSIS OF STEE5S.213easy matter, and T. R. Cunningham lo2 has given details of the use of cup-ferron for this purpose. Photometric methods based on hydrogen peroxidecolours in specified conditions of acidity appear the most hopeful methodsfor individual determinations of these elements, care being taken to allowfor interference due to titanium. I n lOOyo sulphuric acid niobium gives acolour with hydrogen peroxide, titanium gives a colour which is only 30%of its colour in 20% sulphuric acid and tantalum shows no colour. Con-sequently photometric measurements in 20y0 and 100 yo sulphuric acidallow of a determination of niobium and titanium, tantalum being deter-mined by difference from the original weight of the mixed oxides.lo3 G.Thanheiser 104 determines niobium by its colour with hydrogen peroxide insulphuric acid and 40% phosphoric acid (a strength which eliminates thetitanium colour when the titanium is less than 1%) and tantalum by theyellow colour formed with pyrogallol in 3% ammonium oxalate.A correc-tion has to be made for titanium, determined with chromotropic acid in thesame solution.TeZZuriurn.3-Te 2385.76 is compared with Fe 2378.98, an A.C. arc beingused and log ratios of intensities being plotted.105Tin.3-Existing methods have been re-examined,loB and E. T. Saxerand R. E. Minto lo7 propose titration of the reduced tin with potassiumiodate after separation of the iron with ammonia from a hydrochloric acidsolution.Titanium.3* 5* 339 37* lol* 1°3-As has been indicated under tantalum, theperoxide colour is suitable for determinations of titanium in steel.Separ-ation of the titanium is not essential, although cupferron has been used forthis purpose.33 The Zavodskaya laboratory 108 has developed a rapidmethod by adding phosphoric acid and hydrogen peroxide to a perchloratesolution, a method which is unaffected by 1% of chromium. In addition tothe steeloscope method,lo1 L. E. Bruk and N. N. Sorokina lO9 use the linesTi 3088.032 and Fe 3083.747 for quick results with a 5% accuracy. Otherhomologous pairs of lines suitahle for titanium determinations are Ti 3078,Fe 3097 and Ti 3168, Fe 3167.37T~ngsten.~. 4 9 lo* 12* 68-Tungstic oxide can be separated from a varietyof solutions.This oxide and silica are precipitated quantitatively from aperchloric acid solution, and this precipitate can be ignited and weighed,the silica being removed with hydrogen fluoride.110 This method has recentlylo3 Ind. Eng. Chem. (Anal.), 1938, 10, 233.lo3 P. Klinger and W. Koch, Arch. Ebenhuttenw., 1939-40, 13, 127.lo* Mitt. Kaiser- Wilhelrn Inst. Ekenforsch., 1940, 22, 255.lo5 R. E. Nusbaum and J. W. Hackett, J . Opt. SOC. Amer., 1941, 81, 620.lo6 S. Mischonsniky, Congr. Chim. Ind. Compt. rend. 18drne Congr., Nancy, Sept.-lo’ Steel, 109, No. 3, 1941, 66, 91.lo* S. Ri. Gutman, $1. N. Zarogatskaya, apd Z. A. Vyvapaova, Zavod. Lab., 1940, 9,loo Bull. Acad. Sci. U.R.S.S., 1940, 4, 23.110 A.Clauberg and P. Behmenberg, 2. anal. Chem., 1936, 104, 245.Oct. 1938, 438.1012l4 ANALYTICAL CHEMISTRY.been advooated by S. M. Gutman and R. V, Gochfeld.ll1 8-Hydroxy-quinoline can be used to separate tungsten in an acetic acid solution, andthis precipitate can be ignited to tungstic oxide after treatment with oxdicaoid.l12 Vandiurn and molybdenum interfere in this method. BothD. )I. Heppell 68 and D. P. Chatterjee u3 reoommend dissolving precipitatedtungstic oxide in standard sodium hydroxide, the latter mixing the preoip-itate with warm water and neutral glycerol or mannitol to avoid hydrolysis.The time taken in using the hydrochloric acid solution method and subse-quent oxidation with nitric acid can be reduced if the steel is &st heated to1150-1200" for 15 minutes and quenched in water.l14 A photometricmethod, which requires some correction for chromium, depends upon thered colour formed by tungsten with quinol in a solution containing sulphuricand phosphoric a0idS.1~~ It is necessary to reduce both iron and molyb-denum with sfannous chloride.Urccniz~.~.-8-Eydroxyquinoline is B suitable reagent for separatinguranium after removal of the iron by electrolysis and manganese withsodium carbonate.116 Any aluminium in the uranium can be determinedby fusion and precipitation of the hydroxide.pmsibly the most difficult having regard to the amount of manipulationrequired by the ordinary chemical method.Much work has been done todevelop colorimetric methods. Of the colours available, that with hydrogenperoxide is suitable.68 H.Pins1 114 makes the determination in the presenceof the iron, and adds sodium fluoride to eliminate the colour due to titanium--&method which is satisfactory for 0-%23% of vanadium.finds it necessary for titanium-free steels to apply a correction for thecoloured complex due to molybdenum. The orange colour of the phospho-vanado-molybdate complex can be measured accurately for a percentagerange of vanadium 0.01-5~0,118 as much as 10% of molybdenum, titanium,and cobalt and 20% of tungsten causing no interference. A. L. Davydovand Z. M. Vaisberg119 have studied the formation of molybdenum-blueduring the reduction of the phospho-vanado-molybdate complex. Thisreduction can be used to give an aocuracy of 4% for 0*1-2'3/, of vanadium.E.J. Vaughang uses the same complex to separate small amounts of vana-dium, and after treatment with sulphuric acid and oxidation, memums thevanadium colour with stryohnine. A potentiometric titration of vanadiumwith ferrous sulphate, similar to that already has been madeby F. Eisermann.120Although some latitude can be dowed in the conditions for the spectro-?,7ana&um.l* 3 9 49 5, 8 9 9. 10. 111 12. 35, 369 68-The vanadium determination isE. J. Vaughan111 Zavod. Lab., 1938, 7, 698.114 A. A. Fedorov, Zavod. Lab., 1940, 9, 1319.116 G. Bogatzki, 2. awl. Chem., 1938, 114, 170.11* A. M. Dimov and R. 8. Moltschanova, Zavod. Id., 1938, 7, 663.11' Gksserei, 1940, 27, 441.ll0 Zavod. Lab., 1940, 9, 716.112 Z.S. Muchina, ibi&., p. 407.J. Indian Chern. Xoc., 1940, 17, 369.G. Bogatzki, Arch. EiaenhSttenw., 1938-9, 12, 639.180 Arch. Z&enh2tttenw., 193G-9, 12, 245GASKIN: ANALYSIS OF STEPJLS. 215graphic determination of vanadium, high resolution is necessary, and withthis an accuracy of &5% can be obtained by using V 3271.3, Fe 3212.1a1Por similar work, K. A. Suchenko 1% uses V 4379.24 and E'e 4376.0.Zirconium.-The use of propylaraonic acid as a specsc reagent forzirconium, by which means O*lyo of this metal can be determined in thepresence of many other elements, has been proposed.123Oxygen, Nitrogen, and Hydrogen.-The discussion of the methods for thedetermination of these elements given in " Sampling and Analysis of Carbonand Alloy Steel " is pertinent.Work has been done on the vacuum fusionmethod, tin being used as a flux for all three elements.lM1 lz6# lz6 All thehydrogen can be obtained by heating to 600" in a high vacuum, highertemperatures producing more nitrogen and carbon monoxide.lZ7 Thismethod of vacuum heating rather than vacuum fusion is preferred byK. TawctraThe residue from a hydrochloric acid solution of a steel contains some ofthe nitrogen, particularly in titanium steels, and treatment of this residuewith sulphuric acid and potassium and copper sulphates is necessary.129In this method the ammonia is distilled into standard hydrochloric acid,the excess of which is titrated with sodium hydroxide, sodium alizarin-sulphonate being the indicator. Errors may arise in the titration of excesesulphuric acid when this is used to absorb the ammonia, owing to thepresence of carbonate in the alkali and the indicator used.Bromocresol-blueis rec0rnmended.13~ Increased speed in the riitrogen determination may beobtained by dissolution in sulphuric acid, addition of tartaric acid andsodium hydroxide, and distillation of the ammonia.131 Colorimetric deter-mination of the nitrogen can be made by using the colour with Neasler'sreagent and measuring it photoelectrically.132 Alternatively, G. J. Vein-berg adds a 25% solution of thymol in alcohol to a hydrochloric acidsolution of the steel, followed by sodium hypobromite. The colouredcompound is extracted with ether, and the colour matched against standardammonium chloride.Substitution of isopropyl for ethyl ether producescolours which are intense and unchanging.134Reduction of oxides in nitrogen a t 1260", with tin as a flux, followed byconversion of the resulting carbon monoxide into dioxide by passage overlZ1 J. Wilken, Arch. Eisenhilttenw., 193l3-9, 12, 133.laa Zavod. Lab., 1938, 7 , 693.l Z 3 H. H. Geist and G. C. Chandlee, Ind. Eng. Chem. (Anal.), 193i, 9, 169.lZ4 P. S. Lebedev, Zavod. Lab., 1938, 7 , 1378.125 Eight Report, Heterogeneity of Steel Ingots, J . Iron Steel Inst., May 1939;la6 T . Yazima, IPetszc-to-Hagane, 1938, 24, 947.l Z 7 W. C. Newell, J . Iron Steel Inst., 1940, Advance Copy.lZ8 l'etsu-to-Hagane, 1939, 26;, 413.12' T. R. Cunningham and H. L. Hamner, In&.Eng. Chem. (And.), 1939, ll, 303.lS0 I. Wada and R. Ishii, Sci. Papers Inst. Phy8. Chem. Res. Tokyo, 1940, 3'7, 66.lS1 H. Kempf tmd I(. Abresch, Arch. Eisenrnttenw., 1940, 14, 250.132 H. F. Beeghly, I n d . Xng. Chem. (Anal.), 1942, 14, 137.133 Zavod. Lab., 1938, 7 , 1251.who heats 50-100 g. of steel at 800" for 1-2 hours.Third Report of the Oxygen Sub-committee, ibid., Advance Copy, May 1941.lS4 U. J. Veinberg, &id., 1940, 9, 1073216 ANALYTICAL CHEMISTRY.heated copper oxide affords a rapid method for the determination of oxygen. 135Results obtained by this method compare favourably with those obtainedby vacuum fusion methods and, apart from aluminium-killed steels, can beobtained in less than 20 minutes. Solid absorbents are satisfactory for thegases produced by reduction in hydrogen at 1250°.136J.G. N. G.2. FRACTIONAL DISTILLATION.The subject of fractional distillation has of recent years developed greatlyand become of such wide and rapidly increasing application and importancein Analytical Chemistry that no excuse is offered for considering it again sosoon after the Annual Reports of 1940.It is obviously impossible, in the space available, to deal with everyaspect of the subject, and for that reason careful selection has been madeof the Sections which appear in this Report. Hence, no mention is made ofsuch matters as the practical design, lay-out, and operation of columns, thetheoretical design of columns and the calculations involved, the solutionof problems in binary and multicomponent rectification, discussions of batchand continuous distillation, and the importance of reflux ratio.Instead,it was felt that, following the 1940 review, a rather wider consideration ofcertain branches of the subject might now be of use to indicate the manytypes of problem in which it is employed and to give some idea of the efficientapparatus which has recently become available. To that end, a fairlycomprehensive bibliography is provided.The sections on the determination of the number of theoretical platesin a column and on the choice of a column are included because such a mass ofunco-ordinated and apparently contradictory data has been published onthese matters that some confusion is likely to arise in the mind of the non-specialist.It is therefore hoped that these notes may assist in clarifyingthe position.(i) Normal Distillation.Laboratory Colunms and Accessories.-A review, containing manyreferences up to that date, of the literature on the construction, testing, andoperation of laboratory fractionating columns is given by C. C. Wad1He elaborates six essentials for consideration in the design of fractionatingequipment : (i) the still-pot should be of adequate size and of shape suitablefor maintaining a satisfactory evaporating area throughout the distillationand should be efficiently insulated; (ii) the column should have a height atleast 15 times the internal diameter, should be well insulated and providedwith a closely controllable heater jacket to compensate for heat losses;(iii) an effective packing or other means of bringing vapour and liquid intoclose contact is essential; (iv) the packing should have a large surface area,a small hold-up and pressure-drop and be capable of handling a large through-135 L. Singer, Id.Eng. Chem. (Anal.), 1940, 12, 127.lS6 G. J. Veinberg, Zavod. Lab., 1940, 9, No. 1, 23.l U.S. Bur. Mines, 1939, Tech. Paper 600FAY : FRACTIONAL DISTILLATION. 217put ; (v) bubble-plates should have similar properties to packings ; and (vi)the still-head should be capable of close regulation.Efficient laboratory columns, then, may be divided into two broadclasses, ( a ) those employing plates, which may be perforated or of thebubble-cap type, or a combination of both, and ( b ) those containing packings,which may consist of " dumped " or of uniformly arranged material.Ex-amples of the various types were given in 1940, but mention may be madehere of some which have been described recently.A perforated plate column for analytical batch distillations has beendeveloped by C. F. Oldershaw.2 The plates are of 25 mrn. diameter and have42, 44, or 81 perforations. Variation of the diameter of the holes showedthat with a reflux rate of 2.4 1. per hour the plate efficiency varied from68% with diameter 1.35 mm. up to 90% with diameter 0.65 mm. Forvarious reasons the author favours the use of 0.85 mm. perforations and aplate spacing of 26-30 mm. The paper includes the results of efficiencytests and comparisons with other columns, including the Stedman and thehelix-packed type." Dumped " glass helices are the packing material used in an efficientcolumn, designed by A.J. Bailey,3 with a hold-up of less than 0.1 ml. perplate and H.E.T.P. of 5 cm. A still-head suitable for use at reduced pressuresis also described. The same packing is used in an all-glass unit for thedistillation of corrosive liquids, such as chlorosulphonic acid, in the absenceof air and grease.* A continuous apparatus described by R. W. Hufferdand H. A. Krantz 5 employs nickel helices as packing material and has over50 theoretical plates and a throughput of 3 quarts per hour. Data ondistillation of toluene-methycycluhexane mixtures are given. An expandedshale aggregate packing has been investigated by H.G. Thode and F. 0.Walkling,6 who find that with hydrocarbons the efficiency and throughputcompare favourably with those of other packings, whereas with aqueoussystems it is even more effective. The authors suggest that this materialshould not only be applicable industrially but should also be of use in theseparation of isotopes. H. R. Snyder and R. L. Shriner 7 have designed acolumn, intended in particular for the use of students, packed with crystallinecarborundum, and other simple columns suitable for routine analysis havebeen described by L. SmithAn extremely efficient, uniformly packed column has been designed byH. S. Lecky and R. H. Ewel1.l0 A cupped, stainless-steel gauze spiral isfabricated around a central, grooved metal rod, the whole being fittedclosely into a glass tube.The efficiency is as high as 18 plates per foot withand by R. W. Harkness and R. E. Bland.9Ind. Eng. Chem. (Anal.), 1941, 13, 265.A. W. Hixson and A. H. Tenney, ibid., 1942, 14,345.Ind. Eng. Chem., 1941, 33, 1455.Canadian J . Res., 1942, 20, B, 61.Kgl. Pysdograf. Sallskap. Lund, Handl., 47, No. 5, 1.Oil @as J . , 1941, 39, No. 46, 149.a Ibid., p. 487.J . Chem. Educ., 1940, 17, 588.lo Ind. Eng. Chem. (Anal.), 1940, 12, 64421 8 ANALYTICAL CHEMISTRY.hold-up and pressure-drop only 0.4 ml. per plate and 0.1 rnm. of mercury perfoot respectively.R. H. Baker, C. Barkenbus, and C. A. Roswellll have developed thespinning-band type of column, which has a low hold-up : their design,546 om.high, showed 70 theoretical plates, when tested with the heptane-methylcyclohexane mixture, with a throughput of 2.7 ml. per minute andhold-up 0.1 ml. per plate. A description of a new fractionating column forthe temperature range -190" to 300" has been published by W, J. Podbiel-niak.12 The packing may be of two designs, both of the precision-wound,fine-wire type, so arranged that capillary films between the closely spacedturns provide a large liquid surface. Extreme care is taken to ensureadiabatic conditions of operation, for, in addition to highly efficient vacuumjacketing around the flask, column, head, and all connections, a method ofcompensating for residual leakage of heat is provided. The special vacuum-jacketed ground glass joints may be used for temperature ranges from - 190"to 300" while still remaining vacuum-tight.Complete interchangeabilityof columns and flasks allows the use of the apparatus for a wide variety ofpurposes. A low hold-up and good throughput are claimed, and testsshowed 75 plates in 14 in. length, the performance being maintained when thediameter was increased to 1 inch. Details of efficienoy data for varioussizes are given. Two arrangements of apparatus, one for preparative workand the other for analytical use, suitable for fractional distillation at lowtemperatures are also described by H. Koch and F, Hilberath,l3 and theSimons column has been modified for liquids of b. p. -30" to -5" by E. 0.Ramler and J. H. S i m ~ n s . ~ ~Several types of laboratory column for the separation of close-boilinghydrocarbons have been compared by G.R. Schultze and H. Stage,15 andthe same authors, with K. Klein,l6 discuss recent types and give details of aglass column, employing 4-chamber tubes, which contains 10 theoretioalplates in 62 cm. length. Fractional distillation in spirals with an oscillatingand eccentric movement has been accomplished by J, Piazza:' and F. Rosen-dahl l8 has described a column consisting of chambers through which thevapours pass and into which the reflux liquid is sprayed.and by A. R.Richards,ao who emphasise the advantage of vapour take-off systems, inwhich no liquid hold-up is involved. The latter gives constructional detailsof a head of this type and discusses a modification in which an automaticdevice is incorporated for increasing the reflux ratio as the cut-point isThe design of still-heads is reviewed by R.E. Gurovich11 Ind. Eng. Chem. (Anal.), 1940, 12, 468.lS Brennstoff-Chem., 1940, 21, 197.l4 Id. Eng. Chem. (Awl,.), 1942, 14, 430.2. physikal. Chem., 1941, A , 188, 163.1 7 Ind. y Quim., 1940, 3, 22; Anal. Inst. Invest. cient. tecn., 1938-9, 8-9. 56, 78;l2 Ibid., 1941, 13, 639.l5 2. Elektrochem., 1941, 47, 848.Anal, Soe. cient. Argentinu, 1941, 131, 239.Chem. App., 1941, 28, 70.ID Khim. Maahinostroenie, 1940, 9, No. 2, 17.2o I d . Eng. Chem. (Anal.), 1942, 14, 649FAY ; FRACTIONAL DISTILLATION. 21 9approached. A reflux regulator and head for laboratory columns, designedby F. D. Rossini,21 employs a non-lubricated glass valve and provides ameam of measuring throughput and of controlling and estimating the rateof take-off.A magnetically operated valve, which permits take-off of adefinite amount of distillate at regular time intervals, is incorporated in adesign by B. Ferguson, junr.,22 and with the addition of a timing devioe anda recording thermometer a distillation curve oan be automatically obtained.A vapour partition head, described by P. Arthur and C. L. Nickolls,a isclaimed fo remove impurities more efficiently than the normal type. Aneasily cleaned head is described by A. Turk and A. Matuszak 2p and a newtype of dephlegmator by A. A. G a u ~ h i n . ~ ~Among other subjects of interest to constructors of laboratory columns,mention may be made of an automatic device, employing it photoelectriorelay system, to control the heat input to a sti11,26 a glass needle valve, withgrooved stems, suitable for controlling vapour or liquid flow in distillationheads,27 a double spiral condenser for use in downward distillation or reflux-ing,z8 a device for overcoming frothing by superimposing a tube containingglass beads,29 and details of the construction of glass bellows for vacuumja~keta,~O and of glass helices for column packings.31The Determination of the Number of Theoretical Plates in a Column.-The number of theoretical plates oontained in a fraotionating column iscommonly determined by an application of M.R. Fenske's equation 32 €or abinary mixture under total reflux.If n = the number of theoretical plates in the column, xt = the mol.concentration of the more volatile component, x" = that of the leas volatilecomponent, and ct = the relative volatility of the two components, thenwhere the subscripts T and B refer,to the top and the bottom of the packingrespectively.The procedure is to charge the still with a suitable binary mixture,distil under total reflux until the column is in equilibrium, and then takesimultaneous samples of reflux from above and below the packing for analysis.In practice, the samples are usually withdrawn from the condenser and thestill-pot, allowance being made for the fractionation taking place in these zones.The first, a practical point, is that for Two considerations emerge.a1 J .Res. Nat. Bur. Stand., 1939, 23, 509.22 Ind. Eng. Chem. (Anal.), 1942, 14, 493.23 Ibid., 1941, 13, 356.O 6 Khim. Mashinostroenie, 1939, 8, No. 8, 9.26 S. A. Hall and S . Pdkin, Ind. Eng. Chem. (Anal.), 1942, 14, 652.2 7 G. P. Gibson, J . SOC. Chem. Ind., 1939, 58,317.p8 M. T. Bush, lnd. Eng. Chern. (And.), 1941, 13, 592.29 G. W. Harmsen, Chem. Weekblud, 1941, 38, 330.*" D. J. Pompeo and E. Meyer, Rev. Sci. Instr., 1941, 12, 368.81 E.g., R. W. Priae and W. C . McDermott, Ind. Eng. Chena. (And.)., 1938, 11, 289.32 I d . Eng. Chem., 1932, 24,482.aQ Ibid., 1942, 14, 72220 ANALYTICAL CHEMISTRY.accurate determinations of n, care should be taken so to choose the mixturethat very small concentrations of one component do not occur a t either endof the column.The second is that the value chosen for a will affect thedetermined number of plates. In many cases the arithmetic or geometricmean value of a proves satisfactory over a reasonably wide concentrationrange, but a common method of overcoming possible inaccuracies is illus-trated by L. B. Bragg.33 The concentrations of the components in thebenzene-ethylene dichloride system are measured by the refractive index,and a graph is constructed of this property against the number of theoreticalplates, use being made, not of an average value of a, but of the value corres-ponding to the particular conditions under consideration. Such a plothas, of course, no particular zero point, the theoretical plate scale merelyrepresenting differences, but by its use the number of plates in the columncan be obtained immediately from the difference between the values corres-ponding to the samples from the top of the column and from the still.Bragghas used " ideal " values, calculated from the vapour-pressure data of thepure components.Lecky and Ewell lo have concluded that the systemis not an ideal one, and use " apparent " volatility ratios. They show thatthe values obtained are in fair agreement with those obtained by Bragg'smethod, but for the purpose of determining the plates in a column they haveconstructed the corresponding plot for the " ideal '' n-heptane-methylcycl-hexane system.In short, it may be concluded that it is essential to compare columnsunder absolutely identical conditions.This is very strongly emphasisedby a recent publication of Bragg and A. R. Richards.= They have studiedbinary mixtures for the purpose of testing Stedman columns at reducedpressures, and have determined values of a for benzene-ethylene dichlorideand o-dichlorobenzene-diethylbenzene mixtures a t various subatmosphericpressures. Included in the data, however, is a plot for the former system atatmospheric pressure, and the values of n, for a 20-30 plate column, obtainedfrom the middle portion of this new graph are of the order of 50:/, higherthan those obtained from the previous plot.A recent summary by J. GriswoldN" provides a correlation betweenrelative volatility data for the benzene-ethylene dichloride and heptane-methylcycZohexane test mixtures.It follows that by using the datafavoured by this author for the latter system the agreement between testsmade with the two mixtures is good, and the differences found by Leckyand Ewell (see above) are explained.The Choice of a CoEumn.-Although great importance has been attachedin the literature to the H.E.T.P. of laboratory columns, it is not the only, orindeed the most important, factor to be considered in the selection of a suitablepacking for a particular purpose. The larger the H.E.T.P. value, of course,the greater will be the length of column to contain a given number of plates,Even so, the appropriate values of a are not absolutely fixed.Yo I n d . Eng. Chem. (Anal.), 1939, 11, 283. 34 I n d . Eng. Chem., 1942, 34, 1088.s40 Ibid., 1943, 35, 247FAT : FRACTIONAL DISTILLATION.22 1but this factor will normally enter only in the limiting case where there is in-sufficient height available to accommodate the required column. The shortlength of column obtained by using a packing with a low H.E.T.P. gives theadvantages of ease of manipulation and ease of obtaining adiabatic operatingconditions, but these facilities should not be allowed to over-rule the questionof the time taken to carry out the distillation. In many cases it is fairlysimple to overcome the difficulties associated with the H.E.T.P. a t someinitial expense and trouble, but if the apparatus selected is such as to requirea long time in order to carry out a satisfactory separation, the user willalways be faced with a considerable operating expense or loss of time.This question of the time required to carry out a given distillation isbound up with that of the hold-up of liquid in the column while it is operating.The factors which affect the efficiency of the separation obtained from thefractionating column are : the number of theoretical plates, the reflux ratio,the total hold-up in the packing under operating conditions, and the ratioof charge to the hold-up.The effect of the first two of these factors has beenmuch studied, but considerably less attention has been paid to the remainingtwo. The loss of fractionating efficiency due to increase in hold-up has beeninvestigated theoretically by A. Rose, L. M. Welshans, and H. H. L0ng.3~The point may be brought out by the following simple consideration.Suppose we have a column with a large number of theoretical plates andrunning a t a high reflux ratio, separating a mixture of two components,A and B, and the efficiency of the column is such that substantially pure Awill be taken overhead when the concentration of A in the still is 1%.Wewill suppose that the column hold-up is 50 ml. and that the mixture chargedto the still contains 100 ml. of component A and 900 ml. of component B.By the time distillation has proceeded to the point a t which there is 1% of Aremaining in the still, the contents of the latter will be approximately875 ml. of B, and 9 ml. of A. Inside the column itself the average composi-tion of the hold-up will be 50% of each component; and there will be, there-fore, 25 ml.of A held up. The result is that, up to this point, 66 ml. of Awill have been produced as distillate in a pure condition, and from now onthe degree of purity will gradually fall. Now consider a similar case inwhich the same column and mixture are used but the volume of materialcharged is increased ten times. At the critical point the still will containapproximately 9,000 ml. of B and 90 ml. of A, and the hold-up in the columnwill contain 25 ml. of A as before. Consequently, 885 ml. out of the initial1,000 ml. of A will have been produced in a pure state. The result of increas-ing the ratio of charge to hold-up has thus been to increase the recovery ofpure A from 66 to 88-5%, and furthermore the rate a t which the purity willfall off will be lower in the second case.From these considerations it will be seen that two columns containingthe same number of theoretical plates, operated a t the same reflux ratioand with the same ratio of charge to hold-up, will give comparable fraction-ating efficiencies.Thus, the construction of a still and column can be so96 Ind. Eng. Chem., 1940, 32, 668, 673, 675222 ANALYTICAL CHEMISTRY.arranged that any type of packing will give an identical degree of separationwith that of any other column, since the plates and ratio of charge to hold-upmay both be varied independently and therefore appropriate values may beselected; but once these values have been fixed, the time taken to carry outa distillation has also been automatically fixed.The question of the satisfactory ratio of charge to hold-up is one to whichit is difficult to give a specific amwer, but experience has shown that forefficient fractionation, particularly of complex mixtures, the still chargeshould be at least 20 times the hold-up.Having by these considerations chosen a column, a certain size of charge,and a reflux ratio, the time taken to carry out the distillation will have beenfixed, and it is desirable therefore, when considering column packings, tohave some factor which will allow one to see whether this time is going to belong or short.The factor chosen, called the “ hold-up factor,” is convenientlyexpressed as ml. of hold-up per theoretical plate divided by the operatingboil-up rate in ml. per second.s6 The value of this factor is a direct measureof the time taken to carry out any given distillation.In other words, if twocolumns are arranged with identical numbers of plates, charge, etc., one ofthem being packed with material having a hold-up factor of 4 and the otherwith material having a hold-up factor 8, any given distillation will taketwice as long in the second column as it will in the former. Factors of similarform have also been introduced by L. B. Bragg 37 and W. J. Podbielniak.12The determination of hold-up, particularly under actual operation con-ditions, is not easy, and only a few values are available in the literature.Moreover, their usefulness is somewhat impaired by the fact that whenmaking comparisons between columns it is advisable to ensure that the testahave been carried out by exactly the same methods.In Tables I and I1results of tests on a wide variety of column packings are given. In thefirst table the tests have all been made in exactly the same way under con-ditions similar to those which will occur in practice. In the second tablesome further results are given for the sake of interest, although the test datahave not been determined under strict operating conditions, but in separateexperiments in which cold light petroleum has been run down the columncounterourrent to a stream of air (presaturated with petroleum vapour).It is known that this method of determining hold-up gives high values, andtherefore it should be realised that the hold-up factors given in Table I1 &reprobably on the high side.Theefficiency of a packing is often greatly increased by flooding the columnbefore starting distillation. Several of the packings in Table I have beentested with and without this initial flooding, and the effect on the numberof theoretical plates and the hold-up factor will be apparent.There is one further point to be noted when considering the data.-3* Research Department, Anglo-Iranian Oil Co., Ltd.The author is indebted tothis Company for permission to use data included in this Section and would particularlylike to express his appreciation t o Messrs. P. Docksey and J. W. Hyde for so kindlyplacing their experience and advice at his disposal.37 Trans.A w r . Inst. Chem. Eng., 1941, 37, 19Z'mt Data on Column Packings.Data obtained under operating conditions.TABLE I.NO.12345678910Type of packing.Glass helices, 4 in.Steel ,, 5/32in.Stedman, type 114Spiral screen (cuppedPodbielnlkk (wirl spiral)Empty tubeEyelets (No. 2) **Glasskbe, I.D.6mm.,O.D.I , Y ? f in.spiral 4 mm core)7.5 mm., length 8 mm.Dimensions ofcolumn.mm. cm.20 15720 15735 20586.52o 20 86.525.4 6125.4 6146l7 17 46.Diam.,,-.3.7 1284.15 12836 13736 13749 51737.5 11937.5 119Boil-up rate,* ml./hr.Theoreti-At normal cal platesAt flood operating (C,Hs-point. rate. CsH,ClS).2,000 1,800 41-52,000 1,800 3210.00 8.500 22 ~~ 2;700 $200 13.52,700 2,200 131.800 1.400 231;SOO 1;400 1925 66066U 400 - 150 7 - 250 a7,000 6,000 13.57.000 m o o 13.5400 11.,__- ~- -13,500 lZ,k% 427,000 6,000 7.07.000 6.000 7-0H.E.T.P.,em.3.84.91.0.56.46.62.653.21.834.218.31610.112-3171710.1Light petroleum @.p . SO'). ** Tinned brass hohw cylinders, I.D. 4.14 mm.,TABLE 11.Data obtained by countercurrent method with cold lightEfficiencyDimensions ofcolumn.Diam., Length,mm. em.15714 2o 10014 10036 29102 945214415885 102204 24449 8730{ 1:;cBoil-up rate, ml./hr.-7 A t normalA t flood operating Thcoreti- H.E.T.P.,point. rate. cal plates. cm.2,000 1,800 32 4.9 - 960 4.5 22 - 960 3.5 2810000 8,000 1 43 18.113,000 11,OOO 12.9 16.533,000 25 16.821.63347,700 %E 4111,000 70,000 85 10.380:OOO 50,000 42 22.97.4 L 224 ANALYTICAL CHEMISTRY.Applications.-The requirements for columns for the separation ofhydrocarbon mixtures are enumerated by C.K ~ e p p e l , ~ ~ who describessuitable apparatus, expressing a preference for the long-vapour-path typeof column. Various applications are discussed. A comparison of Podbiel-niak and other type columns for this purpose is made by J. J. Savelli,W. D. Seyfried, and B. M. Filbert,39 and H. Macura and H. Grosse-Oetring-haus 4o have made experiments on the fractionation of aromatic and paraffinmixtures in columns containing various types of packings. A column isdescribed, packed with metal spirals, by means of which as little as 1% oftoluene can be detected, in mixtures with benzene and xylene, a t distillationrates of the order of 30 ml.per hour. Hexane can similarly be determinedin mixtures with pentane and heptane, and a modification of the columnincreases the sensitivity of the aromatic analysis to the extent of detecting0.5% of toluene a t a slightly higher distillation rate.This problem of the routine analysis of coal-tar spirits has assumedconsiderable importance during recent years. The matter is complicatedby the fact that the increasing popularity of the vertical retort has resultedin the production of what is known as “ low gravity ” benzole, i.e., materialin which the aromatic constitutents are associated with considerable quan-tities of other hydrocarbons. The use of empirical methods, such as theclassical Colman-Yeoman procedure, is not applicable with accuracy in suchcases, and a distillation test has been developed a t the Government Labora-tory.41 A small, metal-spiral packed column, known as G.L.1 design,containing 7 theoretical plates, with a low hold-up and a fixed reflux head tosimplify routine operation, is used. A charge of 25 ml. is adequate and adistillation rate of 10-20 ml. per hour may be employed. The accuracyis such that under these conditions small quantities of toluene may beestimated within 0.4% or less. By increasing the charge to 100 ml. theaccuracy is increased to O.lyo. To allow for associated non-aromatichydrocarbons, the percentage of toluene present in the toluene fraction isestimated by refractive index, specific gravity, or critical solution tempera-ture with acetic acid, and graphs have been constructed for this purpose.Applications of the G.L.1 column have also been extended to include theestimation of benzole in wash oil. In this test a “ bridge ” of cycbhexanolis added to prevent the column from flooding owing to distillation of washoil after the benzole has distilled over. The ordinary fractionation methodof determining the benzene content of wash oils has been criticised by W.BrOs~e,*~ who describes a new method involving the use of fine fractionatingcolumns. Several times as much benzene is claimed to be recovered withan error normally within 2%. The same system has also been studied byL.K ~ e p p e l . ~ ~ He concludes that phenol is an undesirable constituent of washOel u. Kohle ver. Petroleum, 1940, 36, 194.39 Ind. Eng. Chem. (Anal.), 1941, 13, 868.40 OeZ Kohle Erdoel Teer, 1939, 15, 591.42 Tech. Mitt. Krupp: Forschungsber., 1940, 3, 2.‘s Gas- u. Wasserfach, 1940,03,73 ; Gl.iickauf, 1939,75,465; Chem. Zentr., 1940, I, 487.41 J. W. J. Fay, unpublishedFAY : FRACTTOXAL DISTILLATION. 225oil for gas stripping, and that cracking of wash oil samples occurred duringlaboratory experiments in which they were heated above 190'. A develop-ment of the G.L. 1 column, known as the G.L. 2 design, is now being usedfor the estimation of phenols and other tar acids in mixtures which couldnot previously be analysed accurately and conveniently.The reported separation and identification, by Goldwasser and Taylor,of six isomeric hexenes with an overall b.p. spread of 2.5' has been criticisedby F. C, Whitmore and others4* and by A. Rose.45 The former, using aPodbielniak-Simons-Taylor column containing approximately 15 theoreticalplates, failed to effect separation of similar mixtures having up to 2.7'spread. Rose calculates that more than 400 theoretical plates would berequired for such sharp batch fractiona.tions and that a column with onlya few plates, even with a very small hold-up and with the use of a very highrefiux ratio, can give no appreciable separation.Stedman columns, the advantages of which have been reviewed by Bragg,37have been used for the fractionation of Turner Valley (Canada) crude oilsby R.M. Donald 46 and by L. M. Watson and J. W. T.. Spinks,4' and thephysical constants of many aliphatic hydrocarbons have been determined,after purification by fractionation, by D. B. Brooks, F. L. Howard, and H. C.Crafton, j ~ n r . ~ ~ F. C. Whitmore, L. H. Sutherland, and J. N. Cosby49have used 20-25 plate columns to prepare pure intermediates for thesynthesis and study of substituted docosanes. Among other applica-tions, mention may be made of the use of the Piazza column for the deter-mination of alcohol in foaming liquors 50 and wines; 51 a discussion of thegeneral laboratory applications of this still by R. Rouzaut ;52 studies of theremoval of entrained impurities from distilled water by a glass ring-packedcolumn 53 and of the packed tower collection of phosphoric acid; 54 thedehydration of methyl benzenesulphonate ; 55 and a description of a simyl-taneous chemical reaction and fractional distillation apparatus, applicableto any isomerisation process, in which the reaction vessel is used as thes till-pot of a continuous fractionating column.56A full account, with photographs, of the apparatus and methods used atthe National Bureau of Standards for the analytical separation and purifica-tion of gases by fractional distillation and rectification a t low temperatureis given by M.Shepherd.57 R. L. Geddes 58 has published an interesting4 4 J . Amer. Chem. SOC., 1940, 62, 795.4 6 Canadian J . Res., 1940, 18, B, 12.4 8 J . Res.Nat. Bur. Stand., 1940, 24, 33.49 J . Amer. Chem. SOC., 1942, 64, 1360.51 J. Piazza and R. Rouzaut, Anal. Inst. Invest. cient. tecn., 1938-9, 8-9, 82.62 Anal. SOC. cient. Argentina, 1941, 131, 251.63 Chi-Chuan Shen, J . Chinese Phurm. ASSOC., 1940, 2, 293.54 W. H. Baskervil, Trans. Amer. Inst. Chem. Eng., 1941, 37, 79.6 s A. M. Shuer, Khim. Mashinostroenie, 1939, 8, No. 8, 19.B. Longtin and M. Randall, Ind. Eng. Chem., 1942, 34,292.5 7 J . Res. Nut. BUT. Stand., 1941, 26, 227.5 8 Ind. Eng. Chem., 1941, 33, 795.REP.-VOL. XL. H4 5 Ibid., p. 793.4 7 Ibid., p. 388.Ind. y Quim., 1940, 3, 29226 ANALYTICAL CHEMISTRY.correlation between true b. p. and the standard A.S.T.M. distillation curvesof petrdeum fractions : on the baais of a large number af collected routineresults it is possible to estimate either of these curves if the other has beendetermined.The microfractionation of a single drop of liquid into 30-70 fractions isdescribed by A.A. Morton and J. F. Mah0ney.5~ A vertical tubular capil-lary, packed with glass wool and jacketed to prevent heat loss, is employed.A graph of the b. p. of each fraction shows the existence of one or more com-pounds and the percentage composition can be estimated from it. Detailsme given of the apparatus and procedure, together with results obtained withbenzene-toluene, benzene-xylene, ethyl alcohol-butyl alcohol, ethyl alcohol-methyl ether, and ethyl acetate-butyl acetate mixtures.A small selection of patents whidh have been taken out provides interest-ing evidence of the application of fractionation processes in this sphere.J.R. Bailey facilitates the separation of such materials as n- and iso-pentane by adding to the complex narrow boiling-range paraffin mixture achemically dissimilar and easily separable carrier having a wider boilingrange than the original mixture. By cutting into fractions and separatingoff the carrier, a series of sharply defined cuts of theinitialmixture is obtained.R. R. Dreisbach and J. E. Pierce retard the polymerisation of vinylaromatic compounds during distillation by packing the column with asubstantially insoluble agent effective in inhibiting polymerisation, and tbesame principle is applied by C. E. Barnei3,62 who uses a metal packing toinhibit polymerisation during the purification of methacrylic acid.Theaddition of a volatile hydrocarbon having an initial b. p. not more than1 5 O F. above that of the required pure material is advooated by F. M. Archi-bald and C. A. &hen 63 to assist in the purification by distillation of aliphaticpolyoxygenated compounds such as glycerol.(ii) Axeotropic Distillation.Reviews.-A comprehensive study of the subject has been made by B. J.&lair, A. R. Ghgow, junr., and I?. D. R ~ a s i n i . ~ ~ They present the generaltheory and, in the partioulrtr application to hydrocarbons, find that almostall polar organic compounds produae azeotropes with theae compounds.The b. p. depression, and heme the ease of separation, decreases in the orderparaffins, naphthenes, mono-olehs, diolefins, aromatics.They recommendchoice of an azeotrope-forming compound with b. p. 0-30" below that of theclose-cut hydrocarbon fraction to be treated and emphasisc the desirabilityof choosing a material easily separated from the hydrocarbon, as, e.g., byextraction with water. The review includes a list of hydrocarbons separableby this means from petroleum.The theory of azeotropic mixtures is dealt with by V. A. Kireev,6K who68 Id. Eng. Chem. (Anal.), 1941, 18,494. 6o U.S.P. 2,231,241.61 U.S.P. 2,240,764. 6a U.8.P. 2,241,176. U.$.P. 2,228,431.64 J . Res. Nut. Bur. Stand., 1941, 27, 39; Refiner, 1940, 19, 430.es Actu Physicochim. U.R.S.S., 1941, 14, 371F A Y : FRACTIONAL DISTILLATION. 227develops equations correlating composition and vapour preswre.Deducedvalues are found to agree well with published data for many different mix-tures of organic liquids. G. Schouls 66 has derived thermodynamical dis-tihtion relationships with particular reference to rates ~f vaporisation andexplains the constancy of temperature and pressure in azeotropic distillationby the hypothesis, which is in agreement with experimental data, ofdistillation a t constant rates.The behaviour of ternary mixtures on distillation has been fully dis-cussed by W. Reinders and c. H. de Mir~jer,~? who pay particular attentionto the effect upon the ternary systems of azeotropic points among the binarysystems. A general discussion of the theory and practice of distillationwith steam and other vapours is given by G.A. Fester,68 and, among hrtherpublications reviewing general or particular aspects of azeotropy, mentionmay be made of a comprehensive discussion of the subject and its use8 byT. Hannotte; 69 its applications in the dehydration of acetic acid and inmany industrial processes by D. F. Othmer ; ' 0 a general discussion of third-component, or '' entrainer," distillation by D. B. Keye~,~1 who gives eK-amples in hydrocarbon and aqueous systems; a review of the maximumboiling mixtures of chloroparaffins with donor liquids by R. H. Ewell andL. M. Welch; 72 and a mathematical conaideration of entropy changesduring azeotrope formation by A. K. Zhdan0v.~3Ewperimntul Work-Many investigators have devoted their attentionduring recent years to the study of relevant data.E. M. Baker, R. 0. H.Hubbard, J. H. Huguet, and S. S. Michalowski74 have constructed com-position curves from refraotive index and density determinations for thesyatems ethanol-water, ethanol-cellosolve, and cellosolve-water, and Baker,with R. E. Chaddock, R. A. Lindsay, and R. C. Werneq75 publishes reaultsfor the corresponding ternary system. W. M. Langdon and D. B. Keyes 76have obtained vapour-liquid equilibrium data for the ethanol-water systemand also for the isopropyl alcohol-water mixture, in which they find that theazeotrope contains 68.35 mol.% of the alcohol. The latter system has alsobeen studied by J. E. Schumacher and H. Runty7' who have investigatedthe minimum- boiling azeotropes in the mixture nitromethane&?opropy1alcohol-water. Data are given for the ternary system and for the threecorresponding binary combinations.W. D. Bonner and M. B. Williams 78have worked a t pressures down to 160 mm. of mercury on the separation ofwater and alcohol in presence of beneene. Using refractive index as aBull. SOC. chirn. Belg., 1940, 49, 214.6 7 Rec. Tp.au. chim., 1940, 59, 207.6* Rev. Centro Eatud. Ing. Quim., Univ. Nacl. Litoral (8anta FB, Argentine), 1840,No. 15, 97.Chern. Zentr., 1940, 1, 3578. '* Chem. Met. Eng., 1941,48, No. 6, 91; Ind. Eng. Chem., 1941,33,1106.7 1 Ibid., p. 1019.78 J . Ben. Chern. Russia, 1941, 11, 403.Bid., p. 1383.J. PhyshZ Ohem., 1940, 44, 404.78 J . Arner. Chern. rSoc., 1941, 63, 2476.74 Ind.Eng. Chern., 1939, 81, 1260.76 Ibid., 1942, 34, 938. 'T Ibdct., p. 701228 ANALYTICAL CHEMISTRY.method of analysis, they find that the water is most efficiently removed a tlower pressures.Among other binary systems investigated, interesting data have beencompiled by S. Takagi 79 on the b. p.'s of formic acid-water mixtures. Start-ing from samples of higher and lower concentrations than those of theazeotropic mixtures, and carrying out experiments to coiucidenc6 of constantb. p.'s, he obtains results of accuracy within &lo in temperature and h0.1 yoin composition. K. Tuda, A. Oguri, and S. Hukusima find 43.6% byweight (0.65 mol.) of acetic acid in the azeotrope with a-diethylamino-butan-y-01, and P. I. Lebed 81 reports 85% by weight of ethyl alcohol in itsazeotrope with m-xylene. The vapour-liquid equilibria in the three binarysystems formed by acetone, chloroform, and benzene have been studied byW.Reinders and C. H. de Minjer,82 who find that only the acetone-chloro-form mixture possesses an azeotropic point, the composition being 786% ofchloroform and the maximum b. p. 64.5". The same authorss3 have deter-mined the course of the distillation line in the corresponding ternary mixture.H. J. McDonald ** states that ethyl alcohol separates from the boiling ternarymixture with glycerol and benzene in such a manner that its mo1.-fraction isthe same in the vapour as in the liquid phase.AppZications.-The fundamental work of Mair and others, to whichreference is made above, has been applied to the isolation of a number ofpure substances from petroleum.The normal procedure is to subject aclose-cut (approximately 2") fraction to azeotropic distillation with a suitablepolar organic liquid in a very efficient column containing between 50 and 100theoretical plates. Using entrainer liquids such as diethylene glycol mono-methyl ether, Mair and A. J. Streiff 85 have separated the aromatic hydro-carbons and isolated in a pure state naphthalene and various other COM-pounds from petroleum, and A. R. Glasgow 86 has separated high-boilingparaffins from the same source.An interesting application has been described by D. F. Othmer, J. J.Jacobs, junr., and J. F. Levy.87 The continuous nitration of benzene isaccomplished by using nitric acid without any dehydrating agent, the waterof reaction being removed as i t is produced by azeotrope formation with thebenzene.has made a study of the system nicotine-water and appliesthe results to the separation of nicotine from related alkaloids.R. Negisiand T. Isobe 89 have investigated mixtures of water with n- and iso-butylalcohols. They find that the azeotropes contain 57.6 and 67% of the alcoholsrespectively, the b. p.'s being 92.6" and 90°, and claim practically quantitativeseparation of the alcohols from associated hydrocarbons.C. R. Smith7O Bull. Chem. SOC. Japan, 1939, 14, 508.81 J . Physical Chem. (U.S.S.R.), 1940, 14, 277.S2 Rec. Trau. chirn., 1940, 59, 369. 83 Ibid., p. 392.A4 J . Physical Chern., 1941, 45, 706.8 5 J .Res. Nat. Bur. Stand., 1940, 24, 395; 1941, 27, 343.86 Ibid., 1940, 24, 509. 8 7 Ind. Eng. Chem., 1942, 34, 286.88 Ibid., p. 251.J . Pharm. SOC. Japan, 1941, 61, 74.Bull. Chem. SOC. Japan, 1941, 16, 278FAY : FRACTTONAL DISTILLATIOX. 229The advantages of an anhydrous entrainer liquid over steam for certainclasses of material are reviewed by G. A. Fester and A. Collados,gO who proposethe use of a kerosene cut instead of water for the distillation of pyrogalloland phenols in general.An idea of the wide application of azeotropism can be obtained by abrief survey of some patents recently taken out which involve the use of thisprocess. For example, the drying of acetic and other aliphatic acids isaccomplished by entraining the water impurity with propyl acetate andpropyl a1cohol:l diisobutyl ketone,92 a mixture of diisopropyl ketone withethyl isopropyl ket0ne,~3 or butyl a~etate.~4 Procedures for the separationof aromatic, unsaturated, and paraffin hydrocarbons from mixtures witheach other and other substances include the purification of low-boilingaromatics by azeotropically distilling associated hydrocarbons with methylalcohol,95 methyl acetate ,06 acetone ,97 or crot onaldehyde ,98 the separationof paraffins from 0lefil1~,99 and the purification of indene by distillation ofthe oil with pheno1,l glycols,2 and many other organic compounds containinga carboxyl, hydroxyl, amino- or pyridinic nitrogen radical.3 A small quantityof maleic anhydride is found to facilitate the purification of anthracene bydistillation in presence of a polyhydric alcohol, such as ethylene glycol,having a b.p. in the neighbourhood of 2OOO.4 A similar azeotrope-formingsubstance is utilised in the removal of cyclic ethers from acetoneY5 fromwhich water may be removed by distillation with an aliphatic hydrocarbonMethods for the separation and purification of phenol and its derivativesinclude its removal from cycbhexanone by distillation with a compoundcontaining a t least two alcoholic hydroxyl groups, such as diethylene glycol.'Coloured impurities are removed from phenothiazine by distillation withchlorinated diphenyl ether.8 Other interesting applications include thesimultaneous production and purification of aliphatic hydroxy-substancessuch as glycerol by treating a corresponding halogen compound with waterand removing the halogen acid from the reacting zone as the aqueousaze~trope,~ the dehydration of unsaturated aliphatic cyanides with methyleneof b.p. 28-100°.6Anal. Asoc. Qulm. Argentina, 1942, 30, 36.91 A. W. Bright and J. H. Zeigler, U.S.P. 2,199, 982.g2 D. F. Othmer, U.S.P. 2,275,862.@' D. F. Othmer and R. E. White. U.S.P. 2,275,802.95 H. M. Spiers and H. K. Suttle, B.P. 536,172.96 E. Field, U.S.P. 2,279,194.g8 F. W. Sullivan, jun., U.S.P. 2,265,220.OD Phillips Petroleum Co., B.P. 521,092.93 Idem, U.S.P. 2,269,163.9 7 Idem, U.S.P. 2,212,810.K. H. Engel, U.S.P. 2,279,778.Idem, U.S.P. 2,279,780.J . A. C. Yule, U.S.P. 2,213,755.British Celanese Ltd., B.P.539,487; J. E. Bludworth, U.S.P. 2,273,923.J. F. Eversole and A. C. Plewes, U.S.P. 2,259,951.E. Field, U.S.P. 2,265,939.Idem, U.S.P. 2,279,779.* E. C. Britton, F. B. Smith, and R. L. Brown, U.S.P. 2,284,124.@ H. Dreyfus, B.P. 536,428230 ANALYTZOAL UEEMBTRY.chloride,l* the conoentration of aqueous halohydrin solutions,ll and theseparation of hexamethyleneimine from the diamine by distilling it as anazeotrope with wafer.18(iii) Vmuetm and Mohular Distillation.Revietos.-Many publications on this increasingly important subject havebeen made in the last year or two. The principles involved in ordinary,vacuum and molecular distillations are discussed by M. Furter,l3 and mole-cular, or short-path, distillation has been reviewed by several workers,including D. D.Howat,la who discusses thermal efficiency and the degree offractionation, W. F. Withera,15 and H. I. Waterman and C . van Vlodrop.I*S. 3. Detwiler, junr.,17 has issued a supplement to his bibliography on thissub j ecf .Apparatus.-The performance of laboratory columns packed withRaachig rings, 3erl saddles, and spiral screen gauze at reduced pressure(20 mm. of mercury) for the fractionation of terpenes has been studied byW. D. Stallcup, R. E. Fuguitt, and J. E. Hawkins,l* who conclude that thelast is the most efficient packing yet described for this type of work.E. Kirschbaum l9 has used ethanol-water mixtures to investigate theefficiency of bubble-plate columns a t pressures down to 50 mm. Muchinformation, including pressure-drop data, is presented, together with agraph indicating permissible vapour velocity as a function of other variables.The concentric-tube type of column, which has been critically examined,among other open tube types, by J.H. Westhaver,m has been used in anapparatus described by S. A. Hall and S. Palkin 21 for the efficient fraction-ation of a- and p-pinene at 20 mm. pressure. Two other vacuum stills whichmay be mentioned are an apparatus for the fractional vacuum distillationof small quantites of high-boiling mixtures22 and a 205-1. capacity still,described by the Histology Labomtory of the University of Amsterdam,zasuitable for fractionation, at about 10 mm. pressure, of alcohol, urine, etc.Another development is the modification, by C.E. Watts, J. A. Riddick,and F. Shea?* of the A.S.T.M. standard distillation apparatus to enableboiling ranges to be determined a t reduced pressures.Various devices have been designed to overcome the tendency to frothingduring vacuum distillations. M. Burger, who has also designed a neatlo E. C. Britton and A. R. Sexton, U.S.P. 2,263,436.l1 W. C. B. Smithuysen, U.S.P. 2,188,264; W. Coltof, Can.€'. 394,847.E. I. du Pont de Nemours and Co., B.P. 536,024.Is Milt. Lebensm. Hgg., 1939, 30, 200.Chem. Age, 1941, 45, 309, 323; 1942, 46, 3.l6 J . PTOC. Austral. Chem. Inst., 1942,9,103.l6 Rev. Chim. ind., 1939, 48, 314.Ind. Eng. Chem. (Anal.), 1942, 14, 603.2. Ver. hut. Ing., 1940, No. 3, 69.21 Ind. Eng. Chem. (Anal.), 1942, 14, 807.22 E.Klenk and K. Shuwirth, 2. physwl. Chem., 1941, 207, 260.1 7 Oil and Soap, 1940,17,241.20 Ind. Eng. Chem., 1942, 34, 126.Ohm. WeekbZad, 1941, 38,646. 24 Ind. Eng. Chem. (Anal.), 1942,14,606FAY : FRACTIONAL DISTILLATION. 231continuous vacuum distillation appnratus,z~ dcxribes a valve accessorywhich permits the introduction of a little air through an extra tube.26 I.Levin27 uses the foam level to actuate a relay to achieve the same object,and D. R. Rexford 2* has designed a device in which an air-leak, providedthrough a stop-cook, is operated by EL counterpoised mercury bulb. Thefrequency of this de-foaming cycle may be adjusted between 8 and 20phaaes per minute.Pressure regulating and measiiriirg devices have been described by manywritrerls.An invertedglam filter cell, lacquered 80 that only a wedge-shaped porous surface isexposed, is immersed in mercury to the required depth and connected to alevel-regulating manometer, so that the amount of gas admitted to fheapparatus can be adjusted and controlled. An extremely simple but effec-tive device, constructed from a gaswashing bottle, is described by M. S.Ne~rnan,~O and others who have designed regulators include J. H. Thelin,31who obtains pressure control within O.l.mm. of mercury up t o 200 mm.,B. Ferguson, junr.F2 who employs a magnetically operated capillary leak,C. 13. dc Witt,33 whose apparatus includes a trap and manometer system,and M. J. Cttldwell and H. N. Barham,34 who use a glass valve pressureregulator capable of controlling preasures above or below atmospheric toabout 1 mm.of mercury. E. R. Kline $6 describes a modified McLeod gaugefor the measurement of moderate vacuum (0-05-2 mm. of mercury).A simple receiver, utilising two standard 3-way stop-cocks, which alllowsany required number of fractions to be collected without disturbing thepressure in the still, has been designed by R. S. Tome, E. E. Young, andL. T. Ebyf6 J. B. Cloke3’ has described a vacuum-jacketed receiver,J. W. Patterson and R. W. Van Dolaha8 adaptors for the collection offractions, and A. J. Bailey 39 a still-head, mounted within the flask, for .thelow-preesure distillation of organic mixtures. Miscellaneous publicationsinclude the description of a method of economising in water during prolongeduse of a filter pump,u a trap t o prevent blackflow from suction pumps,4land the preparation, from shellac and pine.tar oil, of a laboratory cement,suitable for vacuum ~ o r k .~ aF. Wittka 43 has dealt in considerable detail with the equipment, describedup t o that time, for laboratory and large-scale molecular distillation. A2B J . Lab. Clin. Med., 1940, 25, 1221. 2o Chemist*Analyst, 1940,&9,20.27 Ibid., p. 89. 28 Ind. Eng. Chm. (Anal.), 1941, 18,95.29 Ber., 1940, 73, 1023.30 Ind. Eng. Chem. (Anal.), 1940, 12, 274.31 Ibid., 1941, 13, 908.33 Chemist-Analyst, 1941, 30, 40.34 Ind. Eng. Chen. (Anal.), 1942, 14, 495.36 Ibid., p. 542.38 Ibid., 1942, 14, 611.40 G. F. Shapley, €‘ham. J., 1942, 148, 71.‘I A. E. Meyer, Ind.Eng. Chem. (Anal.), 1942, 14, 605.42 W. C. Fernehs, “ Inorganic Synthem,” MeGraw-Hi11 Co., No. 67, p. 189.43 Angew. Chem., 1940, 68, 467.An ingenious method is employed by A. Rollett.=3a Ibid., 1942, 14, 104.36 Ibid., 1941, 18, 626. 37 Ibid., 1940, 12, 329.38 Ibid., p. 71232 ANALYTICAL CHEMISTRY.large number of diagrams and an extensive bibliography are provided.Recently, A. J. Bailey44 has described the construction of a simple mole-cular still from two Pyrex micro-belljars and a heated brass plate, theapparatus being suitable for the distillation of such material as lignin.A n improved form of the laboratory molecular still of Main, Shicktanz,and Rose, suitable for the separation of vegetable and animal fats and oils,is given by S. B.Detwiler, junr., and K. S. M a r k l e ~ , ~ ~ a means for the inter-mittent withdrawal of fractions being provided.In publications dealing with accessory equipment, D. D. Howat 14describes low-pressure gauges and compares the relative merits of mercuryand oil pumps, the design, construction, and working of which are alsodiscussed by K. D. Sinelnikov, A. K. Walter, D. N. Ulezko, and A. N. Yam-nitskiLqs s. Eklund4? has made some measurements of ultimate vacuumand the pump speed of molecular pumps and describes a very efficientdesign.Applications.-Certain aspects of this subject have been considered byJ. N. Ray4* in his discussion of recent developments in the study of theconstitution of natural products. Howat 49 gives an extensive review of theliterature on the preparation of vitamins A , D, and E .Vacuum distillation has been applied to the study of Rumanian petroleumby T.Cogciug," who collects various 2' fractions a t 20 112111. pressure; to therecovery of used mineral oils by K. Thomas,51 and to the preparation ofpure lactic acid by G. Genin.52 F. A. Norris, I. I. Rusoff, E. S. Miller, andG. 0. Burr 53 have investigated the effect of vacuum distillation throughpacked columns on the absorption spectra of the methyl esters of highlyunsaturated fatty acids. They conclude that the products obtained aresufficiently representative to be used in isolation and structure work. Useof spiral ~creen and Stedman columns at pressures of the order of 15-20 mm.has been made for the purpose of separating certain constituents of heavycoal-tar naphtha in a pure state.A study of petroleum and tar products by distillation under a cathodicvacuum has been made by several workers.M. Richter 55 has examinedaeroplane-motor fuels and uses the results obtained by distillation in IL200-ml. copper still a t 0*001 mm. pressure to determine the constitutionand assess the performance of lubricants. The absolute identification oflubricating oils by the application of a similar technique is described by R.Petit, Y. Crimail, and R. Duchene.56 Louis 57 uses the method to study4I In&. Eng. Chem. (Anal.), 1942, 14, 177.46 cJ. Tech. Physics (U.S.S.R.), 1941, 11, 879.d 7 Arkiv Mat. Astron. Fysik, 1940, 27, A , No. 21.Proc. Nat. Inst. Sci., India, 1939, 5, 205.Ann. sci.Univ. Jassy, 1940, Sect. 1, 26, 406; 415.4b Ibid., 1940, 12, 348.Chem. Age, 1942, 46, 41, 53.b2 Lait, 1940, 20, No. 197, 412.64 5. W. J. Fay, unpublished.61 2. Ver. deut. Ing., 1941, 86, 33.6s J . Biol. Chem., 1941, 139, 199.6 5 Luftfahrt-Forsch., 1939, 16, 212.O6 Chemie-Induetrie, 1941, 46, 304; Brennstoff-Chem., 1941, 22, 237.6 7 Bull. As8oc. franc. Techn. Pktrole, 1938, No. 45, 31FAY : FRACTIONAL DISTILLATION. 233paraffins and petrolatums. A rather lower pressure to mm.) isemployed by V. A. Korovkina 68 for the investigation of the origin andchemical nature of bituminous products.An interesting application of low-pressure technique, described by C. D.Hurd and R. W. Liggett,59 is the analytical separation of sugars, accom-plished by fractionation of their propionates. Details are given of severalmixtures, and it is concluded that pressures below 0.01 mm. are best. Theaccuracy obtained is about 1-2% for monosaccharides and 2-4y0 fordisaccharides.Far too many pat.cnts have b'een taken out for any attempt to be madehere to give anything like a comprehensivc summary, but a few may bementioned ps having particular points of interest. K. C. D. Hickman hasdeveloped his unobstructed-path still for the distillation under high vacuum(0.001 to mm.) of such materials as hydrocarbons and animal andvegetable oils torecover unsaturated glycerides, sterols, and vitaminfractions.60The same author, with J. G. Baxter,61 has described a method of minimisingoxidation of a fat-soluble oxidisable oil during high-vacuum distillation byaddition of a glyceride oil containing an anti-oxidant. The use of aluminiumpowder as a coating for the vaporising surface of the still, to lessen thedeposition of interfering solid materials, is advocated by J. C. Hecker.62Other stills have been described by G. Burrows 63 and by Distillation Pro-ducts Incorporated.64 One of the latter designs has as an aid to fractiona-tion rotating screens of rods, strips, or wire gauze, and the other utilises thelatent heat of condensation of the distillate for the vaporisation of the fluidoperating the diffusion pump. The same company G5 and other workers 66apply the process to the preparation of hormones, vitamins, and enzymesfrom fish and vegetable oils and extracts. The purification of fatty acids,oils, glycerol, and similar materials is also dealt with in patents by theEastman Kodak Company,67 E. Morlock,68 K. S0ndermann,6~ and othemiOApparatus and processes for the distillation of lubricating oils are describedby J. E. S ~ h u l t z e , ~ ~ R. V. B e ~ k n e l l , ~ ~ and others,73 and a combined oxidationand vacuum distillation apparatus for the production of acetic anhydridefrom acetaldehyde is given by D. C. Hull and C. A. Rlarshall.74 A neatmeans of collecting a number of condensed fractions a t various distances68 Khim. Referat. Zhur., 1940, No. 7, 125.88 J . Amer. C'herrL. SOC., 1941, 63, 2659.6o U.S.P. 2,218,240, 2,221,691, 2,234,166; B.P. 630,367 (with J. C. Hecker).6 1 B.P. 535,100.G3 B.P. 523,754.6 5 B.P. 532,770.6 6 E. W. Fawcett and G . Burrows, U.S.P. 2,156,669.6 7 Fr.P. 845,957; B.P. 524,390; 524,439.6 8 U.S.P. 2,261,939.i o E.g., T. W. Evans, J. R. Scheibli, and G. H. Van de Griendt, U.S.P. 2,234,400.71 U.S.P. 2,217,385, 2,217,386.73 E.g., D.R.-P. 699,900; V. Voorhees, U.S.Y. 2,224,621.7 * U.S.P. 2,283,209.62 U.S.P. 2,269,153.64 B.P. 535,565, 540,603.U.S.P. 2,179,833, 2,224,025.72 U.S.P. 3,2L7,356.H 234 ANALYTICAL CHEMISTRY.from the vapourising surface and providing reflux of the oondensed portionsis the subject bf a patent by C . V. L i t t ~ n . ' ~The separation of o- andp-chlorotoluenes is dealt wibh by P. D. Hammondand R. W. Harri~,7~ who describe an apparatus for the continuous distillationof a 60% o-mixture through an efficient column, u s i n g a high reflux rabio,under an applied pressure not exceeding 200 mm. absolute, and by M. J. P.Bogart and J. S. F. Carter,77 who also describe B method of distilling phenolsand cresols, 78 J. W. J. F.J. W. J. FAY.J. G. N. GASKIN.76 U.S.P. 2,266,063.7 7 U.S.P. 2,240,762.76 U.S.P. 2,240,962.U.S.P. 2,241,110