ANALYTICAL CHEMISTRY.1. INTRODUCTION.THE various branches of analytical chemistry are seen to be inter-relatedwhen it is admitted that any property, physical or chemical, possessed by asubstance may be used to identify that substance, and to determine the pro-portion of that substance in a mixture. These topics are referred to in over2500 abstracts published in Section C of British Abstracts in 1946, and it ismanifestly impossible to report adequately in a few pages on the importantwork which many of these and other publications represent.A coherent impression of the progress made in a branch of analyticalchemistry may be presented either by considering means of identifyingand determining members of groups of substances (such as the constituentsof coal-tar distillates) or by describing analytical techniques and illustratingtheir applications by examples drawn from the various fields of chemistryin which they have been applied : such techniques are mass spectrometryand infra-red absorption spectrophotometry, which are referred to extensivelyfor the first time under Analytical Chemistry.Trends are indicated in theapplication of spectrography, colorimetry, turbidity measurements, polar-ography, and X-ray diffraction to the determination of small quantities, withemphasis on the importance of characterisation of physical state in additionto determinations of composition. J. R. N.2. CONSTITUENTS OF COAL-TAR DISTILLATES.Until comparatively recent years, the analysis of the various fractionsand products distilled from coal-tar has been, like the analysis of petroleumproducts, largely in terms of broad classes of constituents, for example, thetotal response to bromination or sulphonation as a conventional measureof the content of olefins or aromatics.Modern technological methods,however, leading to the isolation on a practical scale of a greatly extendednumber of individual compounds of actual or potential commercial use,',make it possible to think of the coal-tar distillates much more in terms of pureconstituents than was formerly the case. There is thus a corresponding callfor analytical methods for the determination of particular compounds, e . g . ,the individual cresols in mixed isomers, if these are to be used for condens-ation to plastic resins, or indene in distillation cuts intended for the same pur-pose ; indole, in the highly purified state required for perfumery ; or toluenefor nitration. In other cases analytical methods are required for the deter-mination of undesirable constituents in products of industrial use and im-portance, e.g., thiophen in benzene, aniline in dye intermediates, or o-hydroxy-1 0. Kruber, Angew.Chew., 1940, 53, 69; Ber., 1941,74, 1688.2 E. A. Coulson and J. I. Jones, Ind. Chent., 1946, 679308 ANALYTICAL CHEMISTRY.diphenyl in phenol (as a distinction of the synthetic from the natural material).In the present report an attempt has therefore been made to survey methodspublished over a space of about ten years, which appear applicable to thedetermination of individual compounds which occur in the coal-tar distillates.These compounds are, of course, frequently encountered in mixtures ofconsiderable complexity, often with homologues or other compounds of closelysimilar properties. It may therefore be understood as a general rule that suchspecific methods as are available will be applied only after the completestpracticable preliminary separation by solvent extraction, fractional dis-tillation, etc.Improved analytical techniques in fractionation have beendiscussed fairly recently by J. W. J. Fay,3 and a further contribution byW. J. Gooderham may also be noted. Much similar work has been done inthe nearly-related field of petroleum hydrocarbons. Adsorption also, usuallyon silica gel, has been applied to the separation of hydrocarbons.6>7-Filtration of light hydrocarbons through a column of silica gel retains thearomatic constituents but allows the passage of paraffins, olefins, andnaphthenes, which are preferably washed through with a light paraffin suchas pentane.The aromatics are desorbed by means of methyl alcohol, and,when this has been removed by washing with water, are recovered quantit-atively and can be further separated by fractionation. Active carbon, oxidesof magnesium and aluminium, and " Filtrol " have been used as alternativesto silica.A chromatographic concentration of anthracene in anthracene oils is advo-cated by F. R. Cropper and N. Straff~rd,~ adsorption being effected on activealumina, with a 1 : 4 mixture of chlorobenzene and light petroleum as solventand developing medium.Among the hydrocarbons, even after preliminary separation by suchmethods as the foregoing, few compounds exhibit sufficiently characteristicreactivity to allow of specific analysis by chemical means.Considerablework has, however, been directed to the elaboration of the colorimetric re-actions of nitrated aromatic hydrocarbons with a ketone in alkaline condi-tions.lo B. H. Dolin l1 reported that, when the nitrated hydrocarbons aretreated with butanone, colours are given by benzene, toluene, and thexylenes, but only that due to benzene persists on acidification with aceticacid. H. D. Baernstein l2 eliminates toluene from the nitrated mixture byoxidation to dinitrobenzoic acid, which does not react with butanone ; henceAnn.Reports, 1943, 40, 224.B. J. Mair and A. F. Forziati, J . Res. Nut. Bur. Stand., 1944, 32, 161, 165.N. C. Turner, Oil and Gas J . , 1943, 41, 48.J . SOC. Chem. Ind., 1944, 03, 6 5 ~ .5 E.g., H. J . Hepp and D. E. Smith, Ind. Eng. Chem. (Anal.), 1945, 17, 579.* P. Harteck and K. A. Suhr, Chemie, 1943, 56, 120. Certain hydrocarbons, e.y.,n-heptane and toluene, have also been separated on zeolites by the " molecular sieve "technique of R. M. Berrer ( J . SOC. Chem. Ind., 1946, 64, 131).@ Ibid., 1944, 68, 2 6 8 ~ .lo J. Peltzer, Chem.-Ztg., 1933, 57, 162.11 Ind. Eng. O h . (Anal.), 1943,16, 242; cf. Chm. Abs., 1946,40, 6023.I d . Eng. Chem. (Anal.), 1943, 16, 261KELLETT : CONSTITUENTS OF COAL-TAR DISTILLATES.309a combination of this procedure with that of Dolin makes it possible to estimatebenzene, toluene, and xylenes in the presence of each other. M. S. Bykhov-skaya l3 estimates benzene and toluene vapours when present together in air,by nitration in conditions leading to the formation of dinitrobenzene and tri-nitrotoluene. The latter is measured directly by its colour with alcoholicpotash. For the benzene estimation, the trinitrotoluene is eliminated byhydrolysis to phenolic compounds ; these are separated, by partition betweenaqueous alkali and an organic solvent , from the dinitrobenzene, which isthen estimated by the colour developed with potash and acetone. R. P.Marquardt and E. N. Luce l4 apply the reaction t o determine o-xylene inhydrocarbon liquors containing monoalkyl- and alkenyl-benzenes such asstyrene.The olefinic compounds are eliminated by steam-distillation follow-ing mercuration, which renders them non-volatile. The distilled alkyl-benzenes are nitrated and treated with potash and acetone, whereupon themonoalkylbenzenes give a blue colour fading to red, and the xylenes a stablegreen. The fading of the monoalkylbenzene colour is expedited by additionof ethanolamine, after which the intensity of the green element is measuredphoto - electrically.The standard gravimetric nitration method, due t o H. P. Reichel, fordetermination of m-xylene, has been extended l5 by the same author so as topermit of the determination of p-xylene also.Under the specified conditions(mixed nitric and sulphuric acids in glacial acetic acid) m-xylene is nitratedquantitatively to the trinitro-derivative, which is recovered by crystallisationfrom acetone, with an allowance for its small solubility in the cold solvent.p-Xylene is nitrated only in part to the trinitro-derivative, but the yield ofthis is a constant fraction (71.4%) of the theoretical. It may therefore beused as a measure of the p-xylene present, being recovered by evaporationfrom the acetone liquors, and recrystallised from alcohol, with allowance asbefore for its small solubility. The nitration products of o-xylene and ethyl-benzene, being soluble in alcohol, do not interfere. By an analogous pro-cedure, Reichel determines mesitylene, as its acetone-insoluble trinitro-derivative, in the coal-tar spirit fraction of boiling range 160-180".Chemical determinations of naphthalene are for practical purposes limitedto those based on formation of the picrate, after elimination of other picrate-forming substances.The earlier methods of handling the picrate are wellreviewed by A. P. Munch and R. T. Heukers l6 and W. L. Miller.17 A newprocedure has been proposed by A. Bolliger,l* who determines the picratesby extractive titration in chloroform with methylene- blue.For cyclopentadiene and its dimer, a colorimetric method has beendescribed by K. Uhrig, E. Lynch, and H. C. Becker,lg dependent on thel3 Zavodskaya Lab., 1945, 11, 537, through Chem. Abs., 1946, 40, 2419.l4 Ind. Eng.Chem. (Anal.), 1944, 16, 751l 5 Chem.-Ztg., 1941, 65, 446; 1943, 67, 121. l6 Chent. Weekblad, 1935, 32, 411.J . Assoc. 08. Agric. Ghem., 1934, 17, 308.Quart. J . Pharm., 1940, 13, 1 ; Analyst, 1939, 64, 416.I n d . Eng. Chem. (Anal.), 1946, 18, 550310 ANALYTICAL CHEMISTRY.formation of the highly coloured phenylfulvene by condensation of cyclo-pentadiene with benzaldehyde. Dicyclopentadiene does not react, but isdetermined as additional monomer after controlled depolymerisation in thepresence of decahydronaphthalene. It is stated that the materials normallyassociated with cyclopentadiene do not interfere; and though the method isdevised with reference primarily to petroleum oils, it would appear potentiallyapplicable to coal-tar spirits after a sufficient preliminary fractionation.R.Sefton 2o estimates cyclopentadiene in the lightest coal-tar distillates bymeasurement of its heat of condensation with maleic anhydride. The lackof specificity of the reaction is largely palliated by the much greater (almostinstantaneous) speed of reaction with cyclopentadiene, as compared with theother unsaturated hydrocarbons likely to be present.A second application of calorimetric technique is the estimation of smallamounts of benzene in solution by measurement of its heat of nitration.21This method, though obviously limited in its scope, is said to be rapid andconvenient for repeated estimates in special cases.The gravimetric method for indene proposed by M. Weger and A.Billman,22 based on precipitation of its benzylidene compound on treatmentwith benzaldehyde, can be effectively applied to determination of indene incoal-tar fractions such as heavy naphtha, if these are first subjected to steam-di~tillation.~~The Hochst gravimetric procedure for anthracene in anthracene oils(oxidation to anthraquinone) has been adversely criticised by Cropper andStrafford,S who prefer the measurement of anthracene by ultra-violet absorp-tion, after a chromatographic separation in which the anthracene zone on thechromatogram is located by its fluorescence in ultra-violet light.The ultra-violet fluorescence spectrum has been applied also to the evaluation of1 : 2-benzpyrene in anthracene the anthracene being first eliminatedby precipitating, from benzene solution, its complex with maleic anhydride.These procedures are representative of a general tendency to supplementthe limited number of chemical methods available for hydrocarbon analysisby physical, and especially by absorptiometric, techniques.Ultra-violetabsorption has been applied to the analysis of benzene, toluene, and xylenemixtures by A. Luszczak 25 and P. Laurin,26 who estimated the proportionsof these compounds by comparing the intensities of those bands which arecommon to their spectra and those which are not. Characteristic bands, theintensity of which could be used as a measure of concentration, were reportedby C. Weizmann, V. Henri, and E. Bergmann 27 for benzene, toluene, xylene,naphthalene, anthracene, and phenanthrene ; and R.Schnurmann and2o J . SOC. Chem. Ind., 1945, 64, 104; see also A. V. Kirsanov, I. M. Polyakova,and Z . I. Kuznetsova, J . Appl. Chem. U.S.S.R., 1940, 13, 1406 (through Chem. Abs.,1941, 35, 2445) for iodometric estimation of the excess of maleic acid.21 R. L. Bishop and E. L. Wallace, Ind. Eng. Chem. (Anal.), 1943,15, 563.22 Ber., 1903, 36, 640.24 A. Kling and M. Heros, Compt. rend., 1941, 212, 348.25 Wien. Med. Wochenschr., 1936, 86, 91, 150.26 J . Pharm. Chim., 1938, 27, 561.z3 R. D. Haworth, unpublished.27 Nature, 1940, 146, 230KELLETT : CONSTITUENTS OF COAL-TAR DISTILLATES. 31 1S. Whincup 2* have recorded the spectra of ethyl- and propyl-benzenes,styrene, and chrysene as well as the commoner aromatics. A. Berton29finds the bands narrower and more easily identified in the vapour state thanin liquids, and has thus determined as little as 0.01 mg.of benzene per litreof air; similarly 0.1 mg./litre of toluene or styrene, and 0.2 mg./litre ofxylene, anthracene, or phenanthrene.R. R. Gordon and H. Powell 3O measure the optical density and molecularextinction coefficient of hydrocarbon mixtures (benzene and toluene, ethyl-benzene and xylenes), obtaining readings at as many wave-lengths as themixture has components. The total optical density at any wave-length beingthe sum of those of the individual components, the group of readings can beformulated as simultaneous equations and solved for the concentration ofeach component. Ultra-violet spectrophotometry applied to anotherhydrocarbon (diphenyl) is further mentioned bel0w.3~Similar methods based on infra-red absorption have been applied t o mix-tures of benzene, toluene, and xylene with paraffins and naphthenes byB.Mani2~e,~l and although applications to benzenoid hydrocarbons have beenworked out in less detail than those of ultra-violet spectroscopy, the develop-ment of analogous methods for petroleum oils 32 suggests that the infra-redmay prove of similar use in the analysis of coal-tar spirits.Use has also been made of the Raman spectra of the benzene hydrocarbonsfor their estimation in admixture with one another. Although P. Traynard 33and D. H. Rank, R. W. Scott, and M. R. Fenske 34 were able to extend thistechnique only to binary mixtures (benzene and toluene), it has more re-cently 35 been applied to mixtures of from two to eight aromatic components,the proportions of which were determined with an accuracy of about 2%.Spectroscopic methods have been similarly adopted $0 some extent forthe analysis of the phenolic fractions from coal-tar.For instance, thequantitative and qualitative investigation, by infra-red absorption, of cresylicacids containing the three isomeric cresols has been described in papers byH. W. Thompson and D. H. Whiffen 36 and H. W. Thompson ; 37 in the latterpaper, the theoretical bases of the procedure are discussed in some detail.It is claimed that the bands of the infra-red spectra are in general sharperthan those of the ultra-violet. By utilising the sharper bands obtained in thevapour as compared with the liquid state, however, A.Berton 29 has employedultra-violet absorption for the determination of phenol and the three cresols.Mention may also be made of the estimation of o-hydroxydiphenyl, when usedas a fungicide, by means of its ultra-violet absorption in cyclohexane s0lution.~828 Petioleum, 1945, 8, 122.32 E.g., D. L. Fry, R. E. Nusbaum, and H. M. Randall, J . Appl. Physics, 1946, 17,33 Bull. SOC. chim., 1944, 11, 552. 34 Ind. Eng. Chem. (Anal.), 1942, 14, 816.35 D. H. Rank and R. V. Wiegand, J . Opt. SOC. Amer., 1946, 36, 325.36 Chem. and Ind., 1944, 343. 37 Analyst, 1945, 70, 443.3a H. E. Cox, ibid., p. 373. cycloHexane for spectroscopy may be freed fromaromatics by filtration through silica gel (S.A. Ashmore, unpublished).29 Ann. Chirn., 1944, 19, 394.31 Ann. Chim. anal., 1941, 23, 173. J . Inst. Petroleum, 1945, 31, 428.150312 ANALYTICAL CHEMISTRY.Diphenyl, which is generally present with its hydroxy-derivative, has also awell-defined maximum absorption, and can be similarly determined. In thecase of o-hydroxydiphenyl, the use of a spectrophotometer is not indispens-able, since the intensity of its fluorescence in ultra-violet light can be observeddirectly and compared with standards.The chromatography of the phenolic compounds has been studied byW. Bielenberg and L. F i ~ c h e r , ~ ~ who concluded that direct adsorption ofthese constituents gave no promising results. After a preliminary couplingwith diazotised p-nitroaniline, however, adsorption on alumina allowed ofgood separation ; this procedure was utilised for qualitative identificationof phenol, the three cresols, and p-xylenol in the presence of each other, andhas subsequently been extended to all the hydroxy- benzenes boiling below220".Although a number of methods for the chemical determination of phenolsin particular circumstances have been published during the period undersurvey, yet most of these cannot be said to introduce any new chemicalprinciple which might serve as a basis for a more general analytical reaction.Reference may be made to turbidity methods,m341>42 and to gravimetricmethods with iodine-applied also t o the naphthols and guaiaco143 and too-cresol and o-hydroxydiphenyl.a L.Bettelheim 45 separates phenol fromphenol-cresol mixtures and higher homologues by shaking it out with a 33%solution of sodium benzenesulphonate, while J.N. Miller and 0. M. Urbain 46effect its differential oxidation with chromic acid; the total phenols beingdetermined colorimetrically with diazotised sulphanilic acid both before andafter this treatment, phenol may be estimated by difference.The Chapin method for phenol, adopted by the Standardization of TarProducts Tests Committee, has been extended by T. S. Harrison,47 using theSpekker absorptiometer, to the simultaneous estimation of m-cresol ; the o-and p-isomers give no coloration with Millon's reagent, but m-cresol gives ayellow distinct from the red phenol colour, so that both can be determinedin the same alkaline extract.In conjunction with the S.T.P.T.C. methodfor o-cresol (observation of the melting point of its complex with cineole 48),C. E. Sage and H. R. Fleck 49 propose to utilise gravimetrically the resin-forming reaction given by o- and m-cresols, but not by p-cresol, with form-aldehyde in acid solution; from the two analyses, the m-cresol content of38 Biennstoff-Chem., 1940, 21, 236; 1941, 22, 278; 1942, 23, 283.40 W. Seaman, A. R. Norton, and R. T. Foley, Ind. Eng. Chem. (Anal.), 1943, 15,41 R. Paris and J. Vial, Compt. rend., 1946, 222, 324.42 J. Kay and P. J. C . Haywood, Ind. Eng. Chem. (Anal.), 1944, 16, 772.43 M. Franqois and M. Seguin, Bull. SOC. chim., 1933, 53-54, 711.44 W. 0. Emery and H. C . Fuller, Ind. Eng.Chem. (Anal.), 1935, 7 , 248.45 Svensk Kem. Tidslcr., 1942, 54, 194, 219, through Chem. Abs., 1944, 38, 3220,4 6 Ind. Eng. Chem. (Anal.), 1930, 2, 123.4 * F. M. Potter and €I. B. Williams, Analyst, 1932, 57, 267; 1939, 63, 621.4O Ibid., 1932, 57, 567, 773.159.3928.4 7 J . SOC. Chern. Ind., 1943, 62, 119KELLETT : CONSTITUENTS OF COAL-TAR DISTILLATES. 313mixed isomers can be estimated by difference. The resin-forming reactionwith formaldehyde has also been advocated by A. Castiglioni for the esti-mation of the naphthols ; the method is applicable to either a- or p-naphthol,but is not suitable for mixtures of the two. An alternative to the gravimetrictreatment is colorimetric estimation of the a-naphthol by means of the bluecolour given by the a-naphthol resin with sodium hydroxide.The well-known Koppeschaar deternlination of phenols , by brominationand final titration of the iodine liberated from potassium iodide by the excessbromine, has been further studied by W.Bielenberg and E. K ~ h n . ~ l Thecourse of the bromination was followed by a potentiometric method, andmodifications of the standard technique, especially as regards the use ofpotassium iodide, are proposed.Bor qualitative identification of many of the amino-compounds of thecoal-tar distillates, the formation of characteristic diliturates (&nitrobar-biturates) 52 may prove useful. The optical and crystallographic proper tiesof crystals of these derivatives are distinctive even when prepared frommixtures of isomers.Specific quantitative methods available in the amino-compound groupare not very numerous.The estimation of residual free aniline in anilinederivatives is, however, often of importance, and an interesting techniquefor aniline in aminoazobenzene has been described by I?. R. Cropper.53After diazotisation of the material and coupling with H-acid, the red dyeproduced from the aniline present is separated as a chromatogram on filter-paper, and the intensity of the band may be used as a quantitative measureof the aniline content. Aniline in methylanilines can be estimated by thepicryl chloride method in ethyl acetate solution.= The sodium chlorideresulting from neutralisation of the hydrochloric acid liberated is extractedwith water and titrated potentiometrically against silver nitrate.Analternative manipulation in the picryl chloride method is described byG . Spencer and J. E. B r i r n l e ~ . ~ ~A specific method for p-toluidine is based by C. H. Benbrook and R. H.Kienle 56 on the evolution of nitrogen on heating the diazotised material.The p-toluenediazonium derivative is relatively so stable, that o- and m-toluidines, aniline, etc., may be eliminated by three hours’ heating of thediazotised mixture, and any nitrogen subsequently evolved used as a measureof the p-toluidine content.For the polynuclear amino-compounds, no methods new in principle havebeen introduced for a considerable time. The methods available for di-phenylamine, with special reference to its estimation when used as a stabiliserin explosives, have been reviewed and compared by F.Ellingt~n.~’ A colori-50 Z. anal. Chem., 1938, 113, 428.52 13. T. Dewey and E. M. Plein, I n d . En,g. Chew,. (Anal.), 1946, 18, 516.53 Analyst, 1946, 71, 265.54 J. Haslam and F. Sweeney, ibid., 1945, 70, 413.6 5 J . SOC. Chem. Ind., 1945, 64, 53.6 6 I n d . Eng. Chern. (Anal.), 1942, 14, 427.5 1 Ibid., 1943, 126, 88.67 Analyst, 1946, 71, 305314 ANALYTICAL OHEMISTRY.metric method based on oxidation with potassium dichromate has beenproposed by H. Barnes.6*Among the heterocyclic constituents of the coal-tar distillates, preponder-ant importance attaches t o pyridine, and critical surveys of the methodsavailable for its determination have been published by A.Hamer, R. Pomfret,and W. V. S t ~ b b i n g s , ~ ~ R. P. Daroga and A. G. Pollard,60 and C. Belcot.61The method finally adopted by Daroga and Pollard, suitable for smallquantities, is a colorimetric measurement of the blue produced by the actionof reducing agents on the pyridine complex precipitated by silicomolybdicacid, while Hamer, Pomfret, and Stubbings prefer to utilise the clearing-temperature of solutions of pyridine perchlorate.Although a number of colorimetric estimations of indole have beendescribed, these mostly have reference to its production in bacterial cultures ;they are thus appropriate only to small amounts, and are frequently not veryspecific. Mention may, however, be made of the qualitative colour reactionwith xanthhydrol, which is not given by skatole or apparently by any otherindole-ring compound which is substituted in the P-position; 62 and of themost recent reviews of the determination with Ehrlich’s (p-dimethylamino-benzaldehyde) reagent .G3, f~For detection of acridine, J.C. Baro Graf 65 has recommended the pre-paration of highly characteristic crystals of the silicotungstate, easily dis-tinguishable under the microscope from those given by pyridine or quinolinebases, and obtainable a t a dilution of one part of acridine in 70,000. Theprocedure has, however, been criticised by G. Kohn and I. M. Kolthoff.66The technique of extractive titration of picrates and picrolonates againstmethylene-blue, already referred to,18 has been applied also to the determin-ation of the acridine bases.Colorimetric methods for thiophen, intended for its estimation in “ purebenzole ” and therefore adapted to very small quantities of thiophen, havebeen worked out by K.H. V. French.67 The colour reactions utilised arethose with isatin in the presence of ferric sulphate and sulphuric acid, andwith alloxan in the presence of sulphuric acid. The latter is somewhat theless sensitive, but the colour is more stable and gives on the whole betterprecision in working. F. S. Fawcett and H. E. Rasmussen68 have deter-mined the constants of a highly purified sample of thiophen, which may be ofuse in the preparation of standards for the colorimetric procedure.E. G. K.6 8 Analyst, 1944, 69, 344.6o J . Soc.Chern. Ind., 1941, 60, 2071..61 Ann. Chim. anal., 1938, 20, 173.6 2 W. R. Fearon, Analyst, 1944, 69, 122.63 L. H. Chernoff, I n d . Eng. Chena. (Anal.), 1940, 12. 273.O4 M. 73. Jacobs and S . Pinciis, Science, 1945, 102, 204.6 5 A n d Asoc. Qufm. Argentina, 1942, 30, 44, through Chem. A h . , 1942, 36, 5732.6 G J . Biol. Chem., 1943, 148, 711.6 7 J . SOC. Chem. Ind., 1946, 65, 15.60 Ibid., 1946, 71, 419.J . Amer. Chem. SOC., 1945, 67, 1705GRIFPTTRS : MASS SPECTRA. 3153. MASS SPECTRA.At very low pressures a suitable ribbon-shaped stream of positively chargedgaseous ions can be deflected in a carefully chosen electric or magnetic field,or both, so that ions of each value of mass/charge are brought to a separatefocus. If these operations are conducted in a mass spectrograph, the seriesof foci is arranged to fall on a photographic plate and produces a series oflines, each of which corresponds with a different mass provided each ioncarries only one electronic charge.An ion carrying a multiple electroniccharge suffers a larger deflection, and may be focused on the same spot as anion of lower mass carrying a single electronic charge. In a mass spectro-ineter arrangements are made to focus in turn ions of each value of mass/charge on a slit, behind which is a device for collecting and recording electriccharge or ion current.Early work in the field showed that positively charged gaseous ions travel-ling at right angles to an electric field are deflected along the direction of thefield, but in the case of a magnetic field deflection is at right angles to the fieldand to the line of motion.When a pencil of positive ions all of equal mass and electronic charge andtravelling with different velocities is subjected to parallel electric andmagnetic fields a t right angles to the line of motion, the ions fall on a paraboliccurve on a plane perpendicular to the original line of motion.This is theprinciple of the parabola method described by (Sir) J. J. Thomson for thestudy of positive ions. He pointed out that under these conditions ions ofeach different mass yield a different parabola, and therefore it would bepossible to identify ions in terms of their masses, and indicated the value ofthe method in chemical analysis, including the fact that only a very smallamount of material is required.It is important to remember that electricallyneutral atoms and molecules are not deflected in this way, and the preliminarybut essential process of ionisation which the volatile matter must undergousually causes some decomposition thereby altering the composition, and itis only within very recent years that it has been found possible to relate thocomposition of themixture of ions to the composition of the original electricallyneutral gas or vapour mixture, thus establishing a means of determining thecomposition of the latter by positive-ray analysis or mass spectra. Earlierwork was devoted almost entirely to the discovery and identification ofisotopes and the determination of their masses and relative abundance.In the course of this “ analysis of the elements ” various types of instrumentwere developed, based on several different methods devised for focusing thepositive ions.In Aston’s type of instrument the beam of ions is hetero-geneous with respect to velocity, and resolution and focusing are achievedby deflecting the ions electrically and then magnetically in the oppositedirection. But the beam of positive ions can be made homogeneous in“ Rays of Positive Electricity and their Applications to Chemical Analyses ”, 192 1 ,p. 179.2 F. W. Aston, “ Mass Spectra rand Isotopes ”, 1942316 ANALYTICAL CHEMISTRY.respect of one of the possible variables before passing through the focusingfields, thereby permitting the complexity of the latter to be diminished.In K.T. Bainbridge's systemy3 the ions are passed through a " velocityselector " and all ions emerging have the same velocity and are deflected alongsemicircuIar paths and focused in a uniform magnetic field. W. R. Smytheand J. Rlattauch removed all ions save those having certain velocities byapplying suitably spaced alternating electric fields a t right angles to thebeam of positive ions. The ions were then anaIysed by a radial electric fieldalone. Other developments including automatic recording are described byW. Bleakney and others.'A mass spectrograph is used in the accurate determination of atomicmasses since it is possible very accurately to determine the relative positionsof lines made by positive rays on a photographic plate, and relative abundancesof the different atoms may be determined by photometry of these lines.Amass spectrometer is used in the accurate determination of relative abund-ances of ions, and instruments of the form due to A. 0. Nier based on thatof A. J. Dempster are used in this type of analytical work. The rays areformed by the controlled ionisation of a stream of the vapour of the elementor compound or mixture under test, and the ions, of mass m and charge e, aredrawn out of the vapour stream by a small voltage and then acceleratedthrough two slits by a suitable applied potential E , focused magneticallyround the semicircular analyser, and collected on a plate behind the slit, andthe resulting ion current is measured by a valve-amplifying device.The equation of the circular path of radius r traversed by the ions focusedon the slit by the magnetic field H which is a t right angles to the plane of thesemicircle ismle = H2r2/2EIons of different mass may be focused successively on the slit by varyingthe accelerating potential E.The parts of the analyser are enclosed in a Pyrex container which permitseffective baking and out-gasing, a most important advance which permits theelimination of contamination by traces which by their presence would vitiatethe analyses of substances introduced into the apparatus.J.E. Taylor described a Nier type maw spectrometer suitable for routineisotope abundance measurements. He out-gased at 350" for 48 hours beforea determination, but there remained a small background of masses 18 (H20)and 28 (CO,).The abundance ratio 13C/12C was determined with a probableerror of 2% from abundance measurements a t mass 46 (12C1601a0), 45(13C160, and 12C160170)y and 44 (l2C160,).H. G. Thode, P. L. Graham, and J. A. Ziegler describe a mass spectro-meter for rapid determination of isotope abundance ratios with high accuracy,Physical Rev., 1932, 39, 847. Ibid., 1932, 40, 429.]bid., 1918, 11, 316.Canadian J . Res., 1945,533, B, 40.5 Ibid., 1937, 52, 933.7 Ibid., 1932,40, 496; 1934, 45, 761; 1938,53, 531.* Rev. Sci. Instr., 1944, 15, 1GRIFFITES : MASS SPEaTRA. 317I n a rapid recording instrument,lO the mass spectrum is scanned across theexit slit by varying the magnetic field, a procedure favoured by N.D.Coggeshall.ll Instruments are also the subject of patents.12Application.-There are two fields of analytical chemistry in which massspectrometry is an important, if not essential, technique. One is in thedetermination of isotope abundance ratios, and the other is in the analysis ofmixtures of gases or vapours, particularly hydrocarbons. The advantage ofspeed which the technique confers in petroleum refinery analysis and plantcontrol is frequently referred to.Isotope abundance ratios. A substance suitable for the direct determin-ation of the isotope abundance ratio of a constituent atom, by introduction intoa mass spectrometer, must be sufficiently volatile, should be well chosen as tothe masses of the ions which it will yield and must be pure and free fromsubstances which afford ions of masses which interfere.In measuring the rate of isotope exchange reactions between gases,l3ain testing theories of thermal diffusion of gases, and in following changes inabundance ratios consequent upon the operation of processes designed toseparate isotopes, the mass spectrometer has played an essential part.Signi-ficant features of technique already mentioned are illustrated by the work ofA. 0. Nier,13b who showed that a concentration gradient of 12CH, and 13CH4is set up in methane enclosed in a vessel in which a temperature gradientexists. Samples of the methane were burned to carbon dioxide in excess ofpure oxygen, and the carbon dioxide was purified by condensation in a Iiquid-air trap, any excess of oxygen or unburned methane being pumped off.Water vapour was later condensed out at about - 78".The carbon dioxidewas ionised by controlled electron impact and the ion currents due to 13C02(mass 45) and l2C0, (mass 44) were recorded. The relative heights of thepeaks for masses 45 and 44 were corrected for the 7% contribution to the 45peak due to 12C170160. Tests showed that the burning of the methane andthe subsequent manipulation had a slight but insignificant effect on the13C/12C ratio. It was necessary to burn the methane and operate with theresulting carbon dioxide rather than to analyse the methane itself owingto the identity of the masses of the ions 13CH4 and 160H; the latter isderived from the traces of water which could not be eliminated from theapparatus.H. G .Thode,l* in a review of the applications of stable isotopes, pointsout the advantages of using the mass spectrometer in place of density measure-ments on water when determining the abundance ratio of oxygen isotopesin tracer experiments. In order to avoid experimental difficulties which arisewhen l80 is introduced into a mass spectrometer in the form of water orlo J. A. Hipple, D. J. Grove, and W. H. Hickam, Rev. Sci. Instr., 1945, 16, 69.11 J . Chern. Physics., 1944, 12, 19.le H. Hoover, jun., U.S.P. 2,341,551, 15.2.44.13a J. D. Brandner and H. C. Urey, J . Chem. Physic4 1945,13, 351.lab Physical Rev., 1939, 66, 1009.14 Canad. Citem., 1943, 27, 647318 ANALYTICAL CHEMISTRY,oxygen gas, M.Cohn and H. C. Urey l6 converted l80 into C160180 by meansof the exchange reactionC1602 + Hz180 C160180 + H2l60and determined the 180/160 ratio in the carbon dioxide obtained.Such equilibria ashave been investigated extensively, and equilibrium constants can be cd-culated from the small differences in isotope abundance ratios found by massspectrometer measurements on the relevant molecules in equilibrium in thevapour and liquid phases.gThe determination of nickel isotopes after diffusion of the stable isotopesof nickel into copper involved an elaborate series of processes preliminary toconverting the nickel into nickel carbonyl which was analysed in a massspectrometer.16The versatility and fundamental importance of the technique is furtherillustrated by the accurate determination of differences in the abundance oflead isotopes 204, 206, 207, and 208, in common lead derived from mineralsof various geological ages and in radiogenic leads.The metal was convertedinto lead iodide, and the vapour at about 400" and 10-p-10-5 mm. Hg wasionised and analysed in a mass spectrometer.17 From a consideration ofthese results it is concluded that the most probable age of the earth is 3,350million years.In principle, any element or atom in amolecule, with an isotope abundance ratio different from normal, can befollowed through a sequence of processes or reactions by means of appropriateisotope abundance measurements. The analytical problems are very similarto those already indicated.Briefly, it may be necessary to synthesise thesubstance under investigation so that certain atoms have abnormal isotopeabundance ratios and these are determined by converting a few mg. of thesubstance into molecules suitable for examination in a mass spectrometer.After completion of the processes under investigation, the products are iso-lated, purified, and converted into substances for isotope abundance ratiodetermination. For example, methionine was synthesised l9 to have isotopeabundances above normal as indicated : CH3*34S*13CH2*13CH2*CH(NH2)*C02H,and when this was fed to rats, the cystine recovered from their hair had anabundance of 34S above normal, but the 13C/12C ratio was normal.20 TheNon-radioactive tracer elements.l6 J .Amer. Chem. SOC., 1938, 60, 679.16 W. A. Johnson, Amer. Inst. Min., Met. Eng., 1946, Tech. Publ. 2007, 13 pp. ;1 7 A. 0. Nier, J . Amer. Chem. SOC., 1938,60,1571; A. 0. Nier, R. W. Thompson, and18 A. Holmes, Nature, 1947, 159, 127.19 G. W. Kilmer and V. du Vigneaud, J . Bwl. Chem., 1944,154,247.Metals Tech., 13, No. 4.B. F. Murphy, PhysicaZ Rev., 1941,60, 112.V. du Vigneaud, a. W. Kilmer, J. R. Rschele, and M. Cohn, aid., 166,646GRIFFITI3S : MASS SPBUTRA. 319carbon was examined isotopically as carbon dioxide, but it was more con-venient to convert the sulphur into hydrogen sulphide than into mlphurdioxide 21 for examination.Many valuable tracer experiments have been described during recentyears 227 23 which illustrate the application of the above-mentioned principles.The interpretation of the isotope ratios found may be complex matters,24and the possible incidence of isotope exchange reactions during the course ofan investigation has t o be ~onsidered.~~T h e " isotope dilution method " of analysis depends for its success on theprovisos that a compound which has an abnormal isotope content of one ormore elements is inseparable from the compound of normal isotopic com-position by the ordinary laboratory procedures for isolating and purifyingthe compound,26 and that the relevant atoms are found not to undergoexchange reactions during the various processes.27 The method is particularlyvaluable in instances where it is difKcult, if not impossible, to separate fordetermination in a pure form all of a constituent from a mixture ; one maycite the difficulty of determining palmitic acid, for example, in a mixtureof higher fatty acids, and the amino-acids in protein hydrolysates.Theprinciple of the method is as follows.To a weight of a mixture containing an unknown weight y of a substanceY is added a weight x of substance Y containing a concentration Co abovenormal of, say, the heavy isotope of nitrogen, 15N, and after the mixture hasbeen made homogeneous, a proportion of Y (it does not matter how small) isisolated and purified, and the concentration, C, of 15N above normal isdetermined; then y = (Co/C - l)/x.The relative abundance of 15N in organic compounds has been deter-mined 28 by digesting a weight of sample, containing 0.5-2 mg.of nitrogen,in a micro-Kjeldahl, and, from the ammonia produced, the nitrogen isliberated by hypobromite and purified by passage through a trap immersedin liquid nitrogen. The ratio 15N : 14N is determined with a precision of0.003% in 15N in about 0.5 C.C. of the purified nitrogen which is introduced intoa mass spectrometer, in which the ion currents due t o 14N2 (mass 28) and15N14N (mass 29) are compared.The application of the method to the determination of a particular com-pound involves the synthesis of that compound from substances in which theproportion of the rarer isotope of one of the elements has been increased,e.g., 15N in ammonium salts, 13C in sodium cyanide, andW in sodium sulphate,21 A.0. Nier, Physical Rev., 1938, 53, 282.22 D. Rittenberg, J . Appl. Physics, 1942, 13, 561.23 Many authors, J . BioE. Chem., 1939,130, to 1946,166.2p E.g., D. Shernin and D. Rittenberg, ibid., 1946, 166, 621.z6 H. G. Wood, C. H. Werkman, A. Hemingway, and A. 0. Nier, ibid., 1941, 139,377.26 D. Rittenberg and G. L. Foster, ibid., 1940,133, 737.27 A. S. Keston, D. Rittenberg, and R. Schoenheimer ibid., 1939,127, 316.** D. Rittenberg, A. S. Keston, F. Rosebury, and R. Schoenheimer, ibid., 1039,127,201320 ANALYTICAL CHEMISTRY.Synthetic methods may have to be devised to avoid any loss of the valuablerare isotope ~oncentrate.~gComplications which are introduced by the use of synthetic isotope-richracemic compounds with optically active substances in the ordinary way areeliminated by either racemising the optically active substances or by resolvingthe isotope-rich racemic compounds and using the appropriate isomers orby adding the racemic mixture and isolating the natural isomer.A modifiedprocedure is used in attempting to detect d-glutamic acid in the presence ofI-glutarnic acid.30 The accuracy of the method depends upon (1) the purity ofthe isotope-rich compound added, (2) the purity of the compound isolated (apoint which can be checked by isotope ratio determination a t successivestages of the determination), and (3) the accuracy of the isotope determin-ation.The determination of the amino-acid composition of proteins is under-going substantial advances 31 as regards both decrease in quantity of proteinrequired and increased accuracy, and the isotope dilution method is playinga significant part.By the latter method, G. L. Foster 32 has determined theglutamic acid, aspartic acid, lysine, leucine, and glycine in only 7 g. ofx-lactoglobulin, the compounds isolated for purification being benzoylglycine,dibenzoyl-lysine, and the benzenesulphonyl derivative of leucine.AnaZysis of Miztures.-In principle it is possible to identify every elementin a mixture by means of determinations of the masses of the isotopes present,but on the experimental side there are difficulties presented by the necessityof volatilizing the elements and, in some instances, of distinguishing betweenisotopes and molecular ions of equal mass. Moreover, a particular method ofanalysis, however excellent, will only be used or gain acceptance if it hasadvantages over existing methods.The mass spectrometer has been applied to the analysis of traces of gases inmixtures of oxygen, nitrogen, and hydrogen, and of helium in nitrogen.Itis also employed where less than 1 C.C. of gas is available, and where continu-ous indication of composition is desired.33 The sensitivity of mass spectrumtechnique, used in conjunction with methods of concentrating material, intesting for the presence of traces may be illustrated by the demonstration thatstable 3H does not exist to anything like the extent of 1 in 1010 in ordinaryhydrogen.34The composition of a mixture may be deduced from the relative propor-tions of the different elements found by means of a mass spectrometer; e.g.,hydrogen, helium, oxygen, nitrogen, neon, and argon have been determinedin natural gas.50 Interference by other ions was allowed for by selectingsuitable mass peaks for observation.The error due to the last may amount to about 1.5%.29 R.Schoenheimer and S. Ratner, J. Bid. Chem., 1939,127, 301.30 s. Graff, D. Rittenberg, and G. L. Foster, ibid., 1940, 137, 745.31 Ann. Reports, 1945, 42, 247.83 J. A. Hipple, J . Appl. Physics, 1942, 13, 651.84 Lord Rutherford, Nature, 1937, 140, 303; F. W. Aeton, Proc. Roy. SOC., 1937, A,32 J. BioE. Chm., 1945, 159, 431.168, 391QRIPFITHS: MASS SPECTRA. 321Compounds. Apart from a few isolated instances, positive-ray or massspectrum technique has not been applied to the analysis of mixtures of com-pounds until recently.It was thought that the method had shown thepresence of methane, ethane, propane, and butane in the product obtainedby irradiating with ultra-violet light a mixture of ethylene and hydrogencontaining mercury v a p o ~ r , ~ ~ but subsequent experiments 36 with a massspectrometer showed that under electron impact butane yields positive ionscontaining C,, C,, C,, and C, and the propane previously reported might havebeen derived from butane disrupted by electron impact. It was found thatthe number and proportion of ionized fragments obtained by electron bom-bardment of various vapours depended on the nature of the molecule. Thus,approximately SCr-SO% of the positive ions derived from benzene containedc@ lO-20% contained C,, and small proportions only contained C,,C,, C, andC3' Somewhat similar results were obtained by E.Friedlander and H.Kallmann.38 These data were consistent with the proportions of individualions c6H6, C,H,, c6H,, etc., found later 39 by means of mass spectrometersimproved by developments in high-vacuum technique and the incorporationof arrangements for preventing contamination of the ion beam with productsfrom the decomposition of the vapour at the source of electrons. Undersimilar conditions, octane was more extensively disrupted than benzene.40Relative abundances of the positive ions produced by controlled electronbombardment of ammonia (N, NH, NH,, NH,, NH,, and N,), hydrazine(N2H, N2H2, N2H3, N2H4, and N ions),41 methane,42 ethane,43 ethyleneyU andmethyl and ethyl alcohols 45 have been determined under a variety of condi-tions.The proportion of C2H, ions produced in the ionisation of n-butaneis greater than in that of isobutane 46 and this is evidently related to differ-ences in the dissociation probabilities of the different linkings under electronimpact .47As a sequel to and consistent with these results, it is found that withmodern technique each hydrocarbon, methane, ethane, etc., gives its ownparticular abundance ratio of the various ions into which it is broken downby controlled ionisation. These characteristic abundance ratios are obtained36 A. R. OlsonandC. H. Meger, J . Amer. Chem. Soc., 1927, 49, 3131.36 H.R. Stewart and A. R. Olson, ibid., 1931, 53, 1236.37 E. G. Linder, Physical new., 1932, 41, 149.38 2. physikal. Chem., 1932, B, 17, 265.39 P. Kusch, J. T. Tate, and A. Hustrulid, PhysicaZ Rev., 1937, 51, 1007; 1938,40 E. G . Linder, J . Ghem. Physics, 1933, l , , 129.4 1 D. D. Taylor, Physical Rev., 1935, 47, 666.42 J. A. Hipple, jun., and W. Bleakney, ibid., p. 802.43 J. A. Hipple, jun., ibid., 1938, 53, 530.44 P. Kusch, A. Hustrulid, and J. T. Tate, ibid., 1937,452, 843.46 C. S. Cummings sand W. Bleakney, ibid., 1940, 58, 787.4 6 R. F. Baker and J. T. Tate, ibid., 1938, 53, 944.4 7 J. Delfosee and J. A. Hipple, jm., ibid., 54, 1060; M. W. Evans, N. Bauer, and1037.J. Y. Beach, J . Chem. PhpiM, 194% 1% 701.REP.-VOL. XLIII.322 ANALYTICAL CHEMISTRY.irrespective of whether the substance is pure or in a mixture. Further, ifeach of two or more constituents of a mixture yields a common ion, theproportion of this ion measured is the sum of the proportions derived fromeach of the constituents. Isomers, such as n- and iso-butane, yield differentabundance ratios, and as a consequence isomers can be determined inmixtures .47aH. W. Washburn, H. F. Wiley, and S. M. Rock 48 used a Nier type instru-ment ; ions of each mass were caused t o fall successively on the collector andthe resulting successive ion currents were so amplified that the magnitude ofeach ion current, or peak, was recorded at four different amplifications to 1 yoon an oscillograph. The quantitative analysis of such mixtures as propylene,propane, n- and iso-butane, isobutylene, butylene-1, butylene-2, n- and iso-pentane, and pentenes was carried out relatively rapidly.Less than 0.1 C.C.of liquid sample was required and a complete analysis, including the com-putations involved, usually took less than 4 hours compared with severaldays required by other methods. Routine analysis of such mixtures requiredsubstantially less than 4 hours. The error in determining components presentin large proportion was usually less than 1 mol. yo, and in the case of con-stituents present in very small proportions was usually less than 10% of themol. yo actually present. As many as 20 samples, containing as many as 15components, could be analysed in an 8-hour day.As an illustration of themethod of computation employed, in a mixture of n- and iso-butane, propane,ethane, and methane, only the first two contributed to peaks at masses 57and 58, and by means of coefficients derived from calibration experiments withthe pure hydrocarbons, the proportions were calculated, and their contribu-tions to the peak at mass 44 deducted from the observed value, the balancebeing due to propane. The ethane and methane contents were similarlydeduced from values at masses 30 and 16 respectively. The contribution ofeach hydrocarbon to other mass peaks was calculated, and the analysis wasregarded as satisfactory if the residuals were less than 1% of the peaks.Larger discrepancies were attributed to the presence of other substances andled, for example, to the detection and determination of acetone and napthenesin two hydrocarbon mixtures.The technique has been used for the analysisof mixtures of aromatic hydrocarbons and for the determination of smallamounts (0.036-8 yo) of diethylbenzene in ethylbenzene. Six isomericoctanes showed sufficient differences to permit the composition of mixturesto be determined with an error of less than h1.7 mol. %, but extension ofthe method to organic compounds containing oxygen has not met withuniform succe~s.4~The method of analysis has been examined by A. K. Brewer and V. H.Dibeler,% who have identified and determined 10 impurities in 1 : 3-buta-diene of 98% purity and analysed many mixtures of gases and vapours,47aD. P. Stevenson and J.A. Hipple, jun., J . Amer. Chem. SOL, 1942,64, 1688.48 Id. Eng. Chem. Awl., 1943,15, 641.49 H. W. Washburn, H. F. Wiley, S. M. Rock, and C. E. Berry, aid., 1945,17, 74.60 J . Res. Nat. Bur. Stand., 1945,35, 125; 1946,86, 338GRIFFITHS : INFRA-RED ABSORPTION SPECTRA. 323including natural gas and oil-flame fumes with as many as 14 components.Duplicate analyses agreed to O~1-4~O01~o. The vapour at about mm.Hg pressure is bombarded with electrons having not less than 50 volts ofenergy, and the ratios of the fragments into which the molecules are brokenare the same over a wide range of pressure for each molecular species but arenever the same for different species. Errors in the analysis may originate inthe instrument, in the sample, or in the computation.In precision work, themass spectra of the main ingredients of the mixtures should be checked daily.The proportion of ions with more than one electronic charge is small and canbe allowed for. The heights of certain peaks are corrected by deductingcontributions arising from the inclusion of ions containing 13C or D. Thus,the peak of mass 44 will include propyl ions containing either of these heavyisotopes, and corrections are calculated from adjacent lighter peaks by meansof coefficients. Accuracy in thepreparation from pure components of small quantities of mixtures of knowncomposition has been increased. R. C. Taylor and W. S. Young 51 describethe use of valves of sintered glass and mercury in mixing definite quantities ofvolatile liquids and in introducing the mixtures into mass spectrometers.The analysis, by mass spectrometer, of a standard mixture of six isomericoctanes prepared by this means was consistent with the proportions mixed.R, H.Busey, G. L. Barthauer, and A. V. Metler 52 blend low-boiling hydro-carbons by means of small bombs of the pure hydrocarbons connected to asystem of measuring vessels, a manometer, and a stock bomb into which thepure hydrocarbons are successively condensed. Composition is calculatedfrom individual pressure measurements or changes in the weight of thebombs. J. G. A. G.A table of these correction factors is given.4. INFRA-RED ABSORPTION SPECTRA.Until recent years, the analytical applications of infra-red absorption.spectroscopy have been limited mainly to the photographic and overtonevibration regions extending from 0.75 p to 2-5 p approximate1y.l Beyondthe photographic region (0.75-1-3 p) the mapping of spectra was tediousowing to the limitations of instruments, and relatively slow progress wasmade.The discovery of the unique characteristics of the infra-red absorp-tion spectra of molecules has stimulated the development of technique; andthe striking advances recently reported extend the range of application to15 p (the limit of transmission of rock-salt), and in principle to 25 p (thelimit of transmission of potassium bromide). It appears that the funda-mental vibration region, which covers the range 2-5-25 p approximately,is providing data of great utility in analytical chemistry, and the largenumber of publications during the past year justify a short report whichshould be read against the background of last year’s report on “RecentAdvances in Infra-red Spectroscopy ”.z61 Ind.Eng. Chem. And., 1945,17, 811. 62 Ibid., 1946, 18, 407.Ibid., 1945,42, 6. Ann. Reports, 1931, 28, 181 ; 1938, 35, 395324 ANALYTICAL CHEMISTRY.It is convenient to recall that there are two units employed in designatingportions of the infra-red spectrum : a wave-length unit, the micron, p(10,000 A. = 1 p = lo-* cm.), and a frequency unit, the wave-number orcm.-l, related to the wave-length unit by 1 cm.-l = l / l (cm.); for example,4000 cm.-l = 1/2.5 xInvestigations of infia-red absorption spectra depend upon the avail-ability of (1) a source emitting a continuous range of wave-lengths, (2) ameans of focusing and dispersing this radiation into very narrow bands ofdefinite wave-lengths, (3) a means of interposing a sample of suitable thick-ness in the path of the radiation, and (4) means of detecting, measuring,and recording the radiation transmitted.The source commonly employed is a Nernst filament or a silicon carbiderod (Globar) electrically heated, but nichrome and an alloy containing iron,chromium, and aluminium have also been used.Owing to the difficulty ofconstructing achromatic and transparent lenses for focusing infra-red radi-ation, surface-coated mirrors of gold or aluminium are employed. Prismstransparent to the limits indicated (glass 1.5 p, quartz 3 p, lithium fluoride5 p, fluorite 9 p, rock-salt 15 p, and potassium bromide 25 p) are generallyused to disperse the radiation.In a fresh comparison of relative merits,fluorite is preferred to lithium fluoride.6Calibration of a prism spectrometer may be effected by means of 6 or 8points ranging from the sodium D line to the carbon dioxide 14.97 p band.7Although atmospheric moisture causes deterioration of rock-salt surfaces, itis much used as a prism material and as the transparent portions of absorptioncells. Resistance to atmospheric corrosion may be increased by heating at600" for a few hours.* Other precautions, such as a small heating elementunder the prism tables, may be in~tituted.~ Absorption cells for substancesmolten at elevated temperatures may be made by cementing rock-salt platesto " Pyrex " glass with silver chloride, but the cell must be kept above 150"to prevent stresses from cleaving the rock-salt.lOAdvances in the detection, measurement, and recording of infra-redradiation are very striking.The present limit of photography (1-3 p) islikely to be extended to 1-53 p by placing infra-red sensitive phosphor screensin contact with photographic plates.ll A sensitive photo-conductive cell oflead sulphide with maximum sensitivity at 2.5 p and threshold at 3.6 pis mentioned,12 and filters of organic dyes and plastics, transmitting in theregion 1-3 p but passing little visible radiation, have been produced.13cm. = 1/26 p.3 N. Wright and W. Herscher, J .Opt. SOC. Amer., 1946, 36, 195.4 J. Savage, J . Sci. Instr., 1946, 23, 295.F, N. W. Scott, J . Opt. SOC. Amer., 1946, 36, 711.6 R. C. Gore, R. S. Macdonald, V. 2. Williams, and J. U. White, ibid., 1947, 37, 23.D. S. McKinney and R. A. Friedel, ibid., 1946,36, 715.A. Elliott, Nature, 1946, 157, 299. P. J. Kipp, J . Sci. Itaetr., 1946, 23, 246.12 R. J. Cashman, ibid., p. 356.10 G. L. Simmard and J. Steger, Rev. Sci. Instr., 1946, 17, 166,11 F. W. Paul, J . Opt. SOC. Amer., 1946, 36, 175.l3 E. R. Blout, W. F. Amon, jun., R. G. Shepherd, jun., A. Thomas, C. D. West, andE. H. Land, ibid., p. 460GRIFFITHS : TNFRA-RED ABSORPTION SPECTRA. 325The rapid scanning and recording of absorption spectra demands radiationdetectors of small time constant, and comparative studies of the performanceof infka-red receivers have been made.l*In respect of speed of response, thermopiles are somewhat wanting, butthe construction of thermopiles having time constants of only 0-01-4.03sec:15 suggests that the objection has been at least partly removed.Sensi-tive thermopiles of several designs, including vacuum types, have beendescribed.16 In one instrument, two halves may be connected in oppositiongiving a compensated thermopile which is free from drift or the two halvesmay be illuminated with different beams such as may be obtained with acompensated optical system. Another thermopile has four receivers in linefor use with a.double-beam infra-red spectrometer in which each of twodifferent beams fall on one of the two inner receivers, the two outer receiversproviding compensation.In some instruments the time constant has beendiminished to less than 0-05 sec.The construction of sensitive bolometers, instruments in which use is.made of the rapid change of resistance of a metal ribbon or film with temper-ature, has been described.17 The application of a fast superconductinginstrument operating at 14" K. is foreshadowed.ls Thermistor bolometersare made of semi-conductors of which the resistance varies rapidly withtemperature. An instrument with a time constant of 3 millisecs. is described l9,together with a detector system which scans 1 p in 1 minute.20G. F. Lothian 21 has given a survey of modern spectrometers, and, althoughthe relative merits of single- and double-beam instruments are debated,22reference may be made to a mirror double monochromator intended forwork in the infra-red, visible, and ultra-violet regions of the spectrum, andhaving two prisms each of quartz (0.2-2-7 p), flint (04-1-5 p), and rock-salt (0.2-16.0 v).The radiation receiver for infra-red is a compensatedthermopile and galvanometer, deflections of which may be magnified by arelay outfit and secondary gal~anometer.2~ J. U. White 24 described asimple infra-red spectrometer recording optical densities directly, and meansof presenting spectra extending over widths as great as 3 v on the screenof a cathode-ray tube have been reported.25Analysis.-Theory and experiment appear to be in agreement that unlesstwo molecules are identical, or are optical enantiomorphs, they will havel4 E.E. Boll, R. F. Bahl, A. H. Nielson, and H. H. Neilsen, J . Opt. SOC. Amer., 1946,l6 L. Harris, ibid., p. 597.17 F. G. Brockman, J . Opt. Soc. Amer., 1946, 36, 32; B. H. Billings, W. L. Hyde,and E. E. Barr, ibid., p. 354.18 D. H. Andrews, R. M. Milton, and W. DeSorbo, ibid., p. 518.19 W. H. Brattain and J. A. Becker, ibid., p. 354.20 J. A. Becker and H. R. Moore, ibid., p. 354.2l J. Sci. lmtr., 1946, 23, 293.Z3 P. J. Kipp, ibid., p. 246.Z6 E. F. Daly and G. B. B. M. Sutherland, Nature, 1946, 157, 547; J. King, R. € 336, 355.16 E. Schwarz, J . Sci. Instr., 1946, 23, 246.22 W. C. Price, ibid., p. 295.24 J . Opt. SOC. Amer., 1946,36, 362.Temple, and H. W. Thompson, ibid., 158, 196326 ANALYTICAL CHEMISTRY.different arrays of vibration frequencies and correspondingly different infra-red absorption spectra. It follows that each pure substance has its owncharacteristic infka-red absorption spectrum by which it can be identified.In order to operate a system of identification based on these spectra, a largenumber need to be recorded and suitably classified.Spectra of 363 organiccompounds have been indexed 29 and there are many distributed throughoutthe literature. As illustrative of the value of infra-red methods of identi-fication, and foreshadowing applications in the analysis of plastics, it isfound that natural rubber and synthetic rubbers afford different distinctiveabsorption spectra which can be used in the quantitative analysis of mix-tures.26 In an examination of 7 cycbpentanes and 5 cyclohexanes, thespectral differences between four dimethylcyclopentanes were found to bequite marked.27 neoPentane, not found hitherto in crude oil, was identifiedand determined along with the other consfituents of a fraction also containingpropane, n- and iso-butane, and isopentane.2*There are, however, spectral similarities between chemically relatedsubstances, and, as a result of correlation work, it has been found that certaingroups of atoms, and linkings, give rise to absorption bands in characteristicregions of the spectrum.29 These depend in some measure on the massesof the vibrating parts, and in this connection it may be recalled that theexistence of the heavy isotope of hydrogen was confirmed by the discoveryof bands due to 2H35Cl and 2H37Cl in the calculated position about 4.8 p ascompared with 3.46 p for lHC1.W In the high-frequency (3 p) region, thestretching vibrations between hydrogen and other atoms give rise to bandsbetween the limits indicated : 0-H 3700-3500 cm.-l (if hydrogen bondingoccurs, the frequency is lower), N-H 3500-3200 cm.-l, C-H 3200-2800cm.-1, S-H 2500 cm.-l approximately; CiC, CiN, and CiO are related to anabsorptian at close to 2000 cm.-l; C:O in esters, acids, aldehydes, and ketonesaffords absorption at 1750-1650 cm-l, and aliphatic C:C at 1660-1600cm.-l However, absorptions such as that due to C:C may be weak or non-existent if the linking occurs in a symmetrical position in a molecule, sinceinfra-red absorption only occurs if the associated vibration causes a changein dipole moment.Other correlations relating to more complex groupingshave been worked out, and they all play an essential part in analysis. Theunexpected appearance of such a band in the absorption spectrum of asubstance of known spectrum indicates the presence of an impurity.For example, isoborneol in camphor may be detected down to low limits bythe O-H band near 2 ~ 9 p . ~ ~ If the chemical history of the substance is(AnaE.), 1944, 18, 9.R. B. Barnes, V. Z . Williams, A. R. Davis, and P. Giesecke, I d . Eng. Chem.2 7 E. K. Plyler, R. Stair, and C. J. Humphreys, J. Opt. SOC. Amer., 1946,36, 716.L. C. Jones, jun., R. A. Friedel, and G.P. Hinds, jun., Id. Eng. Chem. (Anal.).2Q R. B. Barnes, R, C. Gore, U. Liddel, and V. Z. Williams, " Infra-red Spectro-30 J. D. Hardy, E. F. Barker, and D. M. Dennison, Physical Rev., 1932, 42, 279.31 G. B. B. M. Sutherland, Trans. Faraday SOC., 1945, 41, 206.1945, 17, 349.scopy ", 1944GRIFFITHS INFRA-RED ABSORPTION' SPECTRA. 327knpwn, the alien band may afford an important clue as to the identity ofthe impurity, particularly as the precise values of these frequencies arerelated to the structure of the rest of the molecule. When the identity ofthe impurity is established, then the characteristic band may also be used todetermine the proportion present.Some recent work on DDT [l : 1 : l-trichloro-2 : 2-&-(4-~hIorophenyl)-ethane] brings out the advantages of using infra-red absorption as comparedwith other analytical methods such as the determination of halogen orcolorimetric reactions.32 DDT has strong bands a t 9-1 p and 9-8 p which arecommon to isomers and impurities containing a p-C1-substituted phenylgrouping; isomers and other impurities have bands not possessed by DDTas indicated :1 : 1 : l-trichloro-2-(2-chlorophenyl)-2-(4-chlorophenyl)ethane, 9.6 p1 : 1 : l-trichloro-2-(3-chlorophenyl)-2-(4-chlorophenyl)ethane, 10.9 p ;1 : l-dichloro-2 : 2-di-(4-chlorophenyl)ethylene, 10.2 p ;di-4-chlorophenyl sulphone, 7.5, 8-6, and 7.8 p ;2 : 2 : 2-trichloro-l-(2-chlorophenyl)ethyl4-chlorobenzenesulphonate,The presence of these bands in commercial DDT was presumptive evid-ence of the presence of the corresponding impurities, and their concentrationscould be inferred from absorption measurements a t wave-lengths correspond-ing with these key bands. The fact that the absorption spectrum of pp'-DDD [l : l-dichloro-2 : 2-di-(4-chlorophenyl)ethane] in the range 7-15 IJ.shows only small differences from that of DDT adds a warning note thatinfra-red technique has limitations.The failure to observe certain bands characteristic of impurities does notnecessarily imply their absence.A certain minimum concentration, possiblyseveral units yo, may be necessary before bands from the impurity can bedistinguished from absorption due to the principal substance, or bands dueto the latter may mask those of the impurity. As instances, as little as0.5% of ethyl alcohol can be detected in acetaldehyde by the alcohol band ofwave-number 1052 cm.-l.The O-H band was not used because the aldehydcabsorbs a t 2.9 p.31 Concentrations of a-pinene as low as 2% could be de-tected in a mixture of terpenes by a band a t 787 cm.-l. Illustrative of theversatility of the technique, there have been detected organic phosphites inphosphonates, impurities in ethylidene chloride and tetrachloroethylene,trichlorobenzoic acid in the i&hl~ro-acid,~~ cyclohexane in toluene, and aslittle as 1 p.p.m. of hexane in carbon tetrachloride.34Quantitative Analysis.-Quantitative analysis may be performed by32 J. R. Downing, W. V. Freed, I. F. Walker, and G. D. Patterson, Ind. Eng. Chem.33 D.H. Wiffen, P. Torkington, and H. W. Thompson, Trans. Furuday Xoc., 1945,34 R. C. Gore and J. B. Patberg, Ind. Eng. Chem. (Anal.), 1941, 13, 768.(o-chlorophenyl group) and 13.3 p ;8.4 and 10.1 p.(AnuE.), 1946, 18, 461.41, 200.L 32s ANALYTICAL CHEXISTRY.empirical calibrations using mixtures of kiiowii composition, or by applicatioiiof Beer’s law, but i t must be borne in mind that this law would be expectedto be followed only if determinations of optical density are made with homo-geneous radiation. If a narrow band of wave-lengths is employed, as is usualin infra-red work by reason of the finite slit width which has to be employed,then departures from Beer’s law may be expected whenever the observationsare made at wave-lengths over which the extinction coefficient E changessharply with change of wave-length.I n practice, observations made a t thepeaks of bands usually follow the Lambert-Beer law.On the assumption that these laws are applicable to each of the n eom-ponents of a mixture, and that the optical density, dh a t any wave-length, isthe sum of the optical densities of the components, then, if I is the thicknessof the cell,dh = d1h + dd + - - * + &A= qc,s,x + + . * * + C?LE,IX)Consequently, it is necessary to determine the optical density at each ofn wave-lengths, a t which the values of E have been determined for the purccomponents, in order t o obtain n equations from which the concentrationsC,,C, . . . . C, can be calculated. The calculations inay be facilitated by specialnianipulation of the linear equations and tlie use of a calculating machinc,:35but the accuracy of the concentrations deduced depends upon several factors.For highest accuracy, tlie percentage absorptions should be in the neighbour-hood of 63% ; an error of 1 yo leads to larger errors in the value of C.36 Thethiclmcss of absor*i)tion cells may be made as small as 0.005 inin.in conse-quence of high values of E exhibited by many liquids, arid the accuracy withwhich values of C can be calculated depends on the accuracy with which I isdetermined. Interference methods with infra-red and visible light and theweight of contained liquid have been used.37 J. H. Lee38 has developedi n e t l d s of correcting for (1) scattered energy which reaches the receiver bycircuitous paths, (2) errors resulting from finite slit width and narrow absorp-tion bands, (3) pressure broadening and the effects of admixed molecules, andhas determined the composition of hydrocarbon vapour inixtures containingfive components with an error of less than 1-5 mol.o/,.He records datarelating to bands of 12 hydrocarbons. For instances of binary mixtures towhich Beer’s law does not apply, 1111. Fred. and F. W. Yorsche 39 describe agraphical method in which obscrved optical densities of the mixture detcr-mine tlie location of a point inside a co-ordinate network reading directlyin concentration.J. Lecomte 40 gives criteria for the use of infra-red spectra in determiningthe purity of hydrocarbons and makes the point that in the analysis of35 1,.J. Cornrio, J . S c i . Instr., 1944, 21, 129; J. L. Suuridcrson and H. H. Grossman,J. Opt. Xoc. Amer., 1946, 36, 243.3G A. It. Yhilpottx, I’rams. E’araclay SOC., 1945, 41, 197.37 G. B. B. 31. Sutherland and H. A. Willis, ibid., p. 181 ; A. E. Martin, ibid., p. 181.38 IrLd. Eng. Chent. (Anal.), 1946, 18, 650.3* Ibid., p. 603. Compt. rend., 1946, 222, 648CIASKIN : THE DETERMINATION OF SMALL QUANTITIES. 329hydrocarbon mixtures, such as motor spirits, the spectra are too complicatedunless carefully purified fractions are employed. Preliminary separationsare oft,en an essential feature, as, for example, in the analysis of mixtures ofxyle~iols.~~Infra-red absorption spectroscopy is playing a part in many branchcv ofnnalytical chemistry and recent applications include the accurate spctro-pholornetrjc determination of copper in hydrochloric acid solution by rii(misof measurements a t 0.97 p 41 and routine determination of CIZ'., Fe", Ni, and(lo" may be made by means of thallium sulphitle photo-clements. 12,42Minute amounts of hydrocarbons in soil gases may be tletcrmiiwtl bycombustion to carbon dioxide which is measured by infra-red absorption a t2.5-2-8 p,43 and the water vapour in a vertical column of the atmoqherehas been determined by using a transmission replica grating and an infra-redsensitive photo-cell to compare the radiant flux in the 0.94 p water vapourabsorption band with that at 1-01 p where no absorption O C C U ~ S .~ ~ Thedetermination of leucine and isoleucine in mixtures of the two derived fromthe hydrolysis of proteins is a matter of great difficulty analytically, andG . B.B. 31. Sutherland 45 has found sufficient differences between infra-rcyiabsorption spectra of these amino-acids, and also between their acetylderivatives, to permit determinations of the proportions with an accuracy ofabout 5%. The concentrations of oxyhzmoglo bin, methzmoglobin, aridcarboxyhzmoglobin in samples of blood have been determined from spectro-photometric observations made in the infra-red and the visible region of tliespectrum.46Among the advantages of infra-red absorption spectroscopy as a method ofanalysis, mention may be made of (1) the small quantities of material usuallyrequired for an analysis, (2) the absence of any decomposition and its conse-quences, by tlie radiation, except perhaps in very rare instances, and (3)the simplicity and speed with which an analysis can be performed afterthe preliminary calibrations have been made.Sufficient has been saidto indicate that the technique, in common with other physical methods ofchemical analysis, has its own particular fields of utility, in parts of which itis the method of choice, and in certain circumstances may be the only metliotlby which it is possible to carry out a particular analysis. J. G. A. G.5 . THE DETERMINATION OF SMALL QUANTITIES.It was the intention of the Reporter, under the above title, not only toreview the progress made in the determination of small quantities, but also touse such a review to indicate the present trends in analytical chemistry.l ' e c h ., 8, No. 5.4 1 Y. Giesecke, Amer. Inst. Min. Met. Eng., 1944, Tech. Publ. 1740, 15 pp. ; il.Zin.49 G. Berrae and E. T'irasoro, Anal. Inst. invest. cient. tecn., 1942-1943,12-13, 147.43 W. J. Sweeney, U.S.P. 2,170,435, 22.8.39.44 N. B. Foster and L. W. E'oskett, J . Opt. SOC. Amer., 1945, 35, 601.45 J . Inter. SOC. Leather Trades Chena., 1946, 30, 11.413 I3. L. Horecker and F. S. Brackett, J. Biol. C'hem., 1944, 152, 669330 ANALYTICAL CHEMISTRY.The term " small quantities " was to include, not only analyses made wherethe total material available was small, but also those where the quantitydetermined was small irrespective of the amount of material available. Thereview of work done was therefore to include largely microchemical methods,together with certain applications of quantitative spectrography, colorimetricand turbidity mFasurements, polarography, X-ray diffraction analysis,and some macro-analytical methods.A recent monograph 1 has, however,provided a very complete review of microchemical progress besides drawingattention to other reviews dealing with the same 33 In these cir-cumstances further detailed reference to such work here is unnecessary, and theReporter intends to proceed to the second half of his intended subject afterbriefly reviewing the other methods mentioned above. The references torecent literature in the following text are not necessarily comprehensive ;they have been selected became they indicate certain important trends inanalytical practice.Quantitative 8pectrography.-The large majority of workers in this fieldare broadly concerned with only two aspects of a spectrum, the position in itof a particular elemental line and the relative density of that line whenrecorded on a photographic plate.The accuracy with which the second ofthese measurements is being made is increasing steadily, and in recent workat the expense of the speed in making a determination. As regards the firstmeasurement the spectrograph is established as a powerful tool in qualitativeanalysis. Two recent publications, one dealing with the analysis of highpurity zinc and zinc alloys and the other with metallurgical analysis,6describe clearly the amount of care and research necessary to obtain resultsof maximum accuracy. Much attention must always be paid to generaltechnique, of which the photographic aspect is by no means the leastimportant.',Accurate assessment of the value of the quantitative spectrographicmethod involves three factors.As it is at present used, the spectrographernormally has some previous knowledge of the composition of the material heexamines, and the method has been largely applied to analyses where thisknowledge is largely implicit, e.g., metallurgical analysis. Such previousknowledge may, of course, have been obtained by the use of the spectrographbut more frequently by other examination. Thus it has been observed91 R. Belcher, " Microchemistry and its Applications ", Monograph published by the2 L.T. Hallett, Ind. Eng. Chem. Anal., 1942, 14, 956.G. H. Wyatt,Chem.and Ind., 1942,61,132.6 " Polarographic and Spectrographic Analysis of High Purity Zinc and Zinc Alloysfor Die Casting ", British Standards Institution Panel of the Non-Ferrous IndustryCommittee ; H.M. Stationery Office, 1945.6 " Collected Papers on Metallurgical Analysis by the Spectrograph ", edited byD. M. Smith; British Non-Ferrous Metals Research Association, 1945.7 E. H. Amstein, J . SOC. Chern. Ind., 1943,62, 61.* N. S. Brommelle and H. R. Clayton, ibid., 1944,63, 83.Royal Institute of Chemistry, 1946.H. Roth, Angew. Chem., 1940,53,441.W. Seith, Deut. Tech., 1941, 9, 264; Chem. Zentr., 1941,11, 928UASKIN : THE DETERMINATION OF SMALL QUANTITIES. 331that a combination of spectrographic and chemical methods is better andmore reliable for the determination of impurities in zinc than other methodssuch as the polarographic.Secondly, the spectrographer must be assured of a representative sample.Difficulties connected with this can to some extent be overcome by the facilitywith which many determinations can be made, but the Bureau of Mines inAmerica,lo for example, has found it necessary to develop methods for thepreparation of samples to be used for both spectrographic and X-ray examin-ation in the evaluation of dust hazards.Thirdly, the ability to make manydeterminations in a reasonable time allows a statistical survey of the resultsto be made.Such surveys, where an adequate number of results is available,are of recognised value.The above points have been mentioned to draw attention to the tendencyto use and regard the spectrograph as a testing rather than an analyticalinstrument. This is not surprising, for the natural advantages of themethod-low sample consumption, automatic permanent record of results,speed in making many determinations, etc.-are all of great value wheremuch routine testing has to be done. Nevertheless, the complexity of aspectrogram, the recognised interferences of elements present in the excitingsource, and the by no means negligible influence on a spectrogram of thephysical state of the material under examination all suggest that valuableinformation, additional to the amount of a particular element, might beobtained by further interpretation of a spectrogram.On these lines there islittle progress to report.Colorimetric and Turbidity Measurements.-It has been remarked that theabsorptiometer is primarily of use in the quantitative analysis of certainsolutions the composition of which is already qualitatively and possiblypartly quantitatively known.ll Such a statement applies equally well tomost instruments used for the mechanical measurement of colour and tur-bidities and indicates the value of these instruments in repetitive work. Forit is in this work that matching by means of a photoelectrical measuringdevice is more effectively done than matching by eye.12 It is not surprising,therefore, that, having an instrument which will give the same reading for thesame amount of a coloured substance, a great deal of attention has been paidto the development of measuring techniques13 and to the preparation ofselective organic reagents which produce -highly coloured compounds.14Methods involving turbidity measurements with some form of photo-electric cell have not made such strides.Certain satisfactory determinationshave been recorded, such as the determination of zinc by measurement of thelo J. W. Ballard, H. I. Oshry, and H. H. Shrenk, J . Opt. SOC. Arner., 1943, 33,l1 H. K. Whalley, Chem. and I n d . , 1942, 61, 495.l2 A. Ringbom, Chirn. et Ind., 1941, 45, No. 3 bis 304.l3 Abstract review of lectures delivered at symposium of the Analytical Group ofVerein deutscher Chemiker, Die Chemie, 1942,55, 361.l4 J.G. N. Gaskin, Ann. Reporb, 1945, 42, 256.667fluorescent turbidity of the oxine complex l5 and $he determination of s m damounts of bismuth by measurement of the turbidity produced by theaddition of bromate-bromide mixture.l6 Generally, howdver, acoura,temasurements of turbidities have inoreaaingly revealed the extent to whichsmall variatims in conditions affect the turbidity, e.g., barium alphate 1'and barium carbonate,ls In fact, basium sulphata figures are aften inaccurate.lg It becomes clear, then, that the introduction of the phoh-cellinto turbidity measurements czould provide much valuable knowledge of theformation of precipitates, paJcticularly up to the stage of coagulation.Pot?urugruphy.-The polarograph, as an instrument suitable for theacourate determination of amall amounts of particular elements and Cornrpounds, is becoming more widely appreciated.Reviews of its uses, rangingfrom its elementary applications zo to recent developments,21 hwe beenpublished. A panel of the British Standards Institution has produced recom-mended methods for the polarographic and spectrographic analysis of high-purity zinc and zinc alloys for die-casting, together with an aceount of theexperimental work leading to the recommendations made. More recently,accounts have been given of applications of the polarograph,22 its use in bio-chemical ahaJy~is,~3 in the analysis of aluminium, magnesium, and zincand of high-purity selenium, and compounds of nickel and cobalt.26The authors of these four papers emphasise that the polarograph must beregarded as complementary to, rather than replacing, existing analyticalmethods.They fhd it difl6cult to generalise about polarographic problem$ ;each problem has to be treated on its merits. Some polarographic methodsare considered t o be outstanding, e.g., the determination of cadmium as animpurity in zinc. The polamgraph is most easily adapted to routine testing,and it may be said that its potentialities in other directions have not beensufficiently examined because of this.X-Ray Diflrmtion.-Hitherto in this account the diffecent analytidmethods described can and often do provide the same information, that is;the amount of an element in a given material.It is frequently a matter ofpersonal choice whether mioro- or macro-methods, spectrograph, or polaro-graph are used. The X-ray diffraction camera on the othep hand providesinfbrmation which the other methods (except sometimes indiredly) cannot,and herein lies its importance. Thus the determinations of 0.1 % of calciumoxide in magnesium oxide, and of 0.2% of zinc oxide in zinc sulphide arequite feasible 26 and have been made. Similarly, X-ray diffraction studies ofl6 L. L. Merritt, jun., I d . Eng. Chem. Anal., 1944,16, 758.l7 W. Volmer and F. Frohlich, 2. anal. Chem., 1944,126,401.l* J. G. N. Gaskin, unpublished.l9 E. Canals and A. Charm, Bull. SOC. chim., 1945, 126, 89.2o J. G. N. Gaskin and H.K. Whahy, Chem. and Id., 1943,62,441.21 J. E. Page, Nature, 1944, 154, 199.23 J. E. Page, ibid., p. 52.p6 R. H. Jones, ibid., 1945, 70, 60.A. K. Majumdar, J . Indian CAaem. Soc., 1944,2l, 157.W. CuIe-Davies, Analyst, 1946, 71, 49.A. S. Nickelson, a%id., p. 58.*' H. P. Rooksby, ibid., p. 166GASKM : THE DETERMINATION OF SMALL QUANTITIES. 333psviog asphakj 2' reveal information unobtahmble by other means. and suchmaterids meeting the same specification have been found to differ greatly.A complete review of the applioation of monochromatic X-ram to the analysisof mixtures has been published by M. Patry ;28 S- T. Gross and D. E. Martin 2ghave also described the use of powdar-diffraction methods for the analysis ofcrystallhe mixtures. Attention has been drawn to the value of powder-diffraction analysis supplemented by and frequently preceded by spectro-graphic examinaticnM The preparation of suitable samples for X-ray workhas already been mentioned.Nqcro-methods.-For the purposes of this account two examples of the useof macro-methods in the determination of small amounts are of value.Theseare the published standard methods 31 for the chemical analvsis of high-purityzinc and zi4c alloys for die-casting whereby known standard metals can beprovided for spectrographic and other purposes, and the successful adajtationof the method of H. H. Willard and 0. B. Winter 32 to the determination ofsmall amounts of fluorine in f~ods,~S in coal and tactory dusts, and in aimMModern Outlook.-The great majority of the papers abstracted in theanalyticd sections of the various publications are concerned with eitherqualitative or quantitative examination for elements.Apart from organicmalysis only one of the recently developed methods attempts to make deter-minations of compounds as such in a submitted material. Few papers relatethe determined analytical @ures with the physical state of the materialexamined.Analysis for elements has been brough€ to a considerable state of per-fection, so much so that the successful repetition of quite difficult determin-ations is a commonplace. This achievement, and it is an achievement, hasundoubtedly been made possible by the introduction of the new physicalmethods, the spectrograph, the absorptiometer, and the polarograph.Nevertheless, it must be recognised that this ability to make numerouselemental determinations, as and when desired, and by the particular methodfavoured by the opergtor, does not constitute the full meamng of analysis orits complete purpose. Having made certain of his ability to determine theprimary constituents of his material, the analyst must surely now desire tostudy its structure.It has bees recognised that a combination of chemical and physicalm e t W can yield more information than either indi~idually.~~ The com-27 C. L. Wilfiford, Agric. and Mech. Coll. Texas, Eng. Exp. Stat. Bull., 73, 70 pp.;Road Ahs., 1945, 12, No. 4, 8.28 Chim. et Ind., 1941,45, No. 3, 259; Chem. Zentr., 1941,II, 3221.20 Ind. h'ng. Chern. Anal., 1944, 16, 95.30 L. K. Frevel, ibid., p. 209.31 British Standard Specification 1005, 1942 ; British Standards Institution.32 Ind. Eng. Chern. Anal., 1933, 5, 7.33 " Determination of Fluorine in Foods ", Report ofa Sub-committee of the Analyti-34 J. G. N. Gaskin, unpublished.*5 W. C . Crone, junr., The Proptier, 1945, 8, No. 4, 3/5 and 10/11.cal Methods Committees of the Society of Public Analysts, Analyst, 1944, 69, 243334 ANALYTICAL CHEMISTRY.plete realisation by the analyst that he must combine the information he getsfrom all of the main methods of examination (micro-, spectrograph, etc.)is a step in the right direction. Then, to render substantial assistance is thedevelopment of the X-ray diffraction camera which already yields informationbeyond the powers of existing analytical methods. Finally, certain possibledevelopments of the other physical methods may help. Harnessed to therigid necessity of performing its present qualitative and quantitative work,the spectrograph has to avoid different physical states in the exciting source,whereas such variation might be related to varying physical state. Similarlywith line interferences. How the absorptiometer might assist has alreadybeen indicated. The polarographer usually wishes to suppress unwantedmaxima in his diffusion currents and does so with " surface active " sub-stances. Attempts have already been made to use the suppression of thesemaxima to measure the quantity or indicate the presence of such compounds.Further, the polarograph can be used to prove the presence of and determinethe amounts of compounds where these are reducible a t the droppingelectrode.It must be recognised that this analysis for compounds and the determin-ation of physical state is of fundamental importance. A single example willshow this. Despite the tremendous amount of work which has been done inthe evaluation of dust hazards, the determination in a dust of the kind andamount of silica which causes silicosis is a problem which as yet is not com-pletely solved. The X-ray diffraction camera is providing new information,but for that to succeed preliminary and complementary work using manymethods will be necessary. Here then is the present and future position ofthe analyst. He will provideinformation as to compound constituents and their physical state.It would be wrong in this review not to draw attention to a matter whichhas been frequently discussed in recent times, viz., the necessity of improvinganalytical instruction in this country. It is hoped that the Reporter hasmade it clear what is to be expected of an analyst. Such an analyst wouldrequire a very wide scientific analytical knowledge. Where is he to get i t ?He can provide all elemental information.J. G . N. G.J. G. N. GASKIN.J. G. A. GRIFFITHS.E. G. KELLETT.J. R. NICHOLLS