摘要:

QUARTERLY REVIEWS MOLECULAR STRUCTURE DETERMINATION BY X-RAY CRYSTAL ANALYSIS MODERN NIETHODS AND THEIR ACCURACY By G. A. JEFFREY Professor of Chemistry University of Pittsburg and D. W. J. CRUICKSHANK Lecturer in Mathematical Chemistry University of Leeds (With a preface by E. G. COX Professor of Inorganic and Structural Chemistry University of Leeds *) IN concluding his Hugo Muller lecture 8 2 J. D. Bernal deplored the fact that (in 1945) structure analysis " is looked on somewhat as a mystery and [that] it is d a c u l t for any chemist brought up on existing lines even to under- stand the meaning of its results ". This may have been unduly pessimistic but it is certainly true that most statements about diffraction methods and their results contained in books on molecular structure and valency written in the last ten years do little to dispel the mystery and certainly give no help to the chemist who wishes to assess the worth of diffraction results in particular cases.There is undoubtedly a perennial need for such help but a review aiming to give it and to survey the current state of accurate X-ray analysis is particularly appropriate a t the present time because the problems of locating atomic centres in molecules are now well understood and largely solved and crystallographers are beginning to turn their attention to new problems such as that of determining the distribution of valency electrons. I n the 'thirties the major weapons with which modern structure analysis is carried out were forged and tested. The heavy atom and isomorphous replacement techniques first used by Cork on the alums,s3 were applied to organic molecules with such success that in 1936 it was found possible to make a complete structure analysis of a centro-symmetrical molecule without the assistance of chemical information ; l9 three-dimensional methods were used first in the extension of the Patterson synthesisYs4 then with the Fourier synthesis to a limited extent e.g.on pentaerythrito1,s and in 1939 combined with the isomorphous replacement technique in the first determination of the structure of an asymmetric substance,s5 again without the help of chemical information. In spite of these successes however progress in general was rather slow and the results obtained while satisfactory enough when only problems of stereochemistry or of general * This review develops much more fully than was then possible the theme which I took for my Tilden Lecture to the Chemical Society in 1947.-E.G. C. Z 335 336 QUARTERLY REVIEWS molecular dimensions were in question coiild not be used with confidence in discussions of valency problems involving variations of a few hundredths of an Angstrom unit since there were no certain criteria for assessing their accuracy. Since the war apart from the increase in tempo which structure analysis has experienced in common with most other sciences the situation has been changed conipletely by a number of factors o€ which the chief probably are (a) the development of many derivative Fourier methods such as difference and differential syntheses ; ( b ) the application of punched-card and more recently electronic digital machines to the calculation of three-dimensional syntheses and of structure factors ; and (c) the recognition of the different classes of errors to which structure analysis is susceptible the clevelopment of methods for correcting systematic errors and the applicntioii of statistical methods to determine as precisely as possible the effectl o€ random errors on the probable accuracy of the results.It will be observed that these three kinds of development are d l con- cerned with the interpretation of the experimental data and ixdeed the emphasis in recent years has been on the better ratilikation of the experi- mental results rather than on raising the accuracy of the latter; in fact until the methods of dealing with errors referred to under ( c ) had been developed there was no very clear means of deciding how extensive or how accurate the experimental resuhs oughtu to be.This is not to say that cxperimental techniques have been neglected ; many improvements have been made but experimental error has not at any stage in recent years been the factor limiting further advance although it might become so in the mar future. As shown in Section 10 (p. 373) in some recent analyses rather less than half the residua'l errors appear to be due to errors in the experi- mental observations. It should be noted however that t,hese remarks apply primarily to organic substances which are rcnsonably stable solids under ordinary laboratory conditions ; a notable fmture oE expr.r.inient2tl advances in recent years has been the devclopment of techniques for the X-ray examination of substances which rcyuire IOW temperatures inert atmospheres or other special conditions for the experiments and the results obtained in such circumstances may not be as exlJensive or as accurate as is necessary for a fully detailed structure analysis.In the case of benzene for example it is unlikely that accurate measurement of t'he bond lengths will he possible without a considerable improvement in experimental tech- nique (involving very low temperatures) became down to - 20" a t least the thermal motion of the molecules is so great flliat perhaps only a quarter of the total number of reflections theoretically pxsible with copper radiation can be observed. Experimental diEculties niny also set a limit to the accuracy witch which compounds containing wry heavy ittoins C R I ~ be analysed ; thus an accurate determination of the positions of the nitrugen atoms in lead azide is difficult bemuse the X-ray scattering €ram the lead atoms " swamps " that from the nitrogen atoms whose contribu1,ion thus becomes comparable with the experimental errors.Nevertheless there are very many cases where no such ciiEculties wise, JEFFREY AND CRUICKSEANIL MOLECULAR STRUCTURE DETERMINATION 337 and the successful prosecution of the analysis then depends only on over- coming the interpretational difficulties. I n the following Review an attempt has been made to give a connected account of modern methods of doing this which we hope may enable structure analysts to take stock and may give others fairly precise indications of the power md limitations of present - day X-ray analysis ; such an account cannot unfortunately be divested entirely of technicalities but we hope that some chemists will find the details interesting and that still more (pace Professor Bernal) will under- stand the conclusions and their implications.Of the major technical advances referred to above the significance of those relating to errors and accuracy will be immediately recognised by anyone interested in correlating observed bond lengths and bond angles with other molecular properties and particularly with the resulls of wave- mechanical calculations ; they will be even more important in the future when as we hope the refinement of structure analyses will be carried far enough to show the distribution of valency electrons in bonds. Errors and questions of accuracy are dealt with fully in later sections of this Review as are also the various newer methods of Fourier synthesis but it is perhaps not out of place to comment briefly 011 mechanical and electrical aids to computing.Punched-card methods operated by professional computers in collaboration with crystallographers were used during the war in the successful structure analysis of penicillin ;86 it was subsequently found 489 49 that it was possible for crystallographers themselves to operate punched- card methods of varying degrees of complexity and during the past few years the analyses of many structures have been greatly speeded up and in- creased in accuracy with their help. In the last year or two electronic digital computers have been applied to crystallographic calculati~ns,~~~ 5 0 9 5 1 3 78 and it already seems likely that they will revolutionise structure analysis.Eiiormous though the help of punched-card machines has been everything which they have done could have been done in principle by means of desk aachines and it is not greatly exaggerating to say that it could in actual fact have been done that way although naturally with much greater expendi- ture of time. At the present time the same calculations can be done on electronic digital machines in a tenth to a twentieth of the time required on punched-card machines ; this ratio however is largely determined by the relatively long time taken to prepare the material to feed it into the machines and to print out the results for the actual calculating speed of electronic machines is enormously greater than that of any other type.The refinement of a crystal structure as will be clear from the account given below is an iterative process and whereas until very recently it has been necessary to take results from the machine a t each stage and use them to prepare new data to put back into the machine for the next stage it will shortly be possible to arrange an electronic digital machine t o make the necessary adjustments to the data between stages iriteriially without wasting time in printing out. Whc~ii such a progranme has been perfected a structure maIysis will then be taken from the approximate correct struc- ture (see Fig. I) to the end of the refinement i n a few hours instead of as at 338 QUARTERLY REVIEWS present (usually) many weeks and because of the very high speed it will probably also be cheaper.This is already much more than could possibly be envisaged with punched-card machines but it is not too much to hope that the solution of t'he phase problem (see p. 342) is attainable by means of electronic machines. If this should prove to be so interpretative techniques would seem to have very little further to go for some time and the pendulum of development would swing to the experimental side where among other things strong efforts would no doubt be made to raise the speed of intensity measurements although physical laws would limit what could be done without loss of accuracy in this direction. 1. Introduction An X-ray single-crystal structure analysis can be separated into a series of operations of which the difficulties vary considerably from one crystal structure to another.There is however sufficient uniformity in the method to make it possible to outline a general plan. This is shown in Fig. 1 as a block diagram which fornis the framework of our subsequent discussion. As we are mainly concerned with the bearing of recent advances in the technique of refinement and in tlhe estimation of accuracy on the study of molecular structures the initial stages of the method are reviewed briefly and are discussed in det'ail only when relevant to the question of reliability of the final results. It is convenient for the purpose of this article to have in mind a particular problem of average difficulty ; for example the structure anaIysis of a non- planar organic molecule with about ten atoms (carbon nitrogen or oxygen) apart from the hydrogens.Much of this account will apply siniilarly to far simpler arid to much more complex structures though the emphasis at the various stages would then be different in each case. In dimethyltriacc4ylene,l for example the structure is so simple that there is essentially no phase problem whereas in ribonuclease the phase problem is so complicated that it is difficult at present to visualise a detailed refinement. Before discussing the scheme in Fig. 1 in detail a brief account will be given of the method as a whole in order to show the relationships of the various sections to each other and of the results at each stage to the final electlron distribution in the crystal. The preliminary X-ray examination of the crystal aims to find the unit- cell dimensions the number of molecules in the repeating unit and the symmetry relations between niolecules ions or atomic groups which are conveniently expressed as the space group.Other relevant physical observa- tions which may help the structure determination are the cleavage and measurements of refractive indices and pyro- and piezo-electricity. This part of the research may occupy anything from a day to a month depending upon the nature and structure of the cryst'als. It is preliminary t o the structure analysis proper because it is upon its results that the crystal- 1 Jeffroy and Rollctt Proc. Roy. SOC. 1952 A 213 86. Carlisle and Seouloudi ibid. 1951 A 207 406. JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 339 lographer bases his judgment as to whether it is worth while proceeding to a complete analysis which will be a much more extensive research.A good estimate of the time for the more straightforward part of the work can then be made but as only indirect methods are available for solving the phase problem it is not generally possible to predict the whole course of the analysis. If the objective is information about a particular group of atoms which occiirs (with so far as is known no significant variation) in a number of different crystalline substances the wisest procedure is evidently to deter- mine the space groups of as many as possible of them in order to select the most promising for the detailed analysis. If an analysis is to be carried to thc limit of present-day technique as discussed in this Review a complete experimental record must be made. This consists of the measurements of the intensities of all the observable diffraction spectra under the optimum conditions for accuracy.From these measurements the amplitudes of the structure factors can be calculated but not their phases and it is the indirect deduction of these structure phases which constitutes the phase problem. The detailed methods a t present available for solving this fundamental problem are of great technical interest t'o the crystallographer and have recently been discussed elsewhere ; 3 a description of them is outside the scope of t/his Review although it is relevant later to give a brief enumeration of those which are particularIy effective. The attempts at solution of the phase problem lead to triaZ structures and when a correct one has been found the refinerncnt is theoretically a straightforward if somewhat formidable iterative process.In practice as shown in Fig. 1 the refinement is usually broken down into two stages the so-called two-dimensional and three-dimensional analysis. There is a good practical reason for this distinction with a structure of average complexity ; the calculations of the first part can be dealt with by means of a desk calculator while the second part usually requires large-scale computing equipment of the Hollerith or electronic type. The two-dimensional refine- ment of the projections of the molecules in the unit cell is therefore pursued as far as possible before the structure is examined in all three dimensions simultaneously. Before 1945 this was with a few exceptions the end of the analysis. With corrections for certain systematic errors which would now be applied it is still a point at which the analysis can be concluded if the objective is solely to determine configuration or some broad feature of stereochemistry .The three-dimensional refinement of atomic co-ordinates may proceed cither by Fourier or by least-squares methods. There is a formal equivalence of the two methods although they differ in application and the results are not usually exactly coincident owing to different relative weightings of the experimental observations. This Review is mainly written in the terms of the Fourier methods but the least -squares treatment is discussed more briefly in a separate section. Until very recently fhe determination of the atomic co-ordinates from " Computing Methods and the Phase Problem in X-ray Analysis " Edit.Pepinsky The Pennsylvania State College U.S.A. 1952. 340 QUA HTJ(:H LY RE VIEWS the thr~-dimension;tl refinenleiit corrected for syhi,c'iii;it,ic (~r~wrs was the end of thc analysis. It is still the point of conchsion if thc object of tihe research is to elucidate a problem connected with the bond lengths or to represent the molecule in terms of bond character and valence-bond diagrams. Although the maxima of the atomic peaks are accurately known a t this stage other features of the electron density are niuch less certain. More information can be obtained by further refinmient if 3s is often the case the experimental errors arc less than the discrepancies betm een 1 lit. observed structure amplitudes and those cdculatcd from the final atomic co-ordinatcs.This additional information may bc about the location of the hydrogen electrons or about the asyniinctry of the electron distribution around the atomic centres which can arise froiii thermal motion and froin the participation of atomic electrons in the bonds between atoms. For most structure analyses thc computations associated with these last stages of refinement are cxtrernely laborious and it is only in a few selected examples that results of this natlure have SO far been obtained. 1 Preliminary X-ray and optical oxarninution I I k+;tce group ccll dimensions molecular syinmctry I 1 Experimental measi&mwnt of intensities I I The structure amplitudes I f I I The phase problem I The trial 'structure Ferification and two-dimensional refinement of trial structure 1 1 I The approximate correct structure (s.d.of bond lengths - 0.03 A) I I Three-dimcnsional refinement of atomic co-ordinates I I Accurate bond lengths valcncy angles and intcrmolecular distances (d. of bond lengths - 0.01 A) I ~~ I Detailed refinement of the clcctron distributioii I TIiornial vibration parame ters hydrogen a tom positions valency-electron distribution Calculation of accuracy of atomic positions molecular dimensions electron density values Significance comparisons with theoretical calculations and with other experimental results I I FIG. 1 Schematic X-ray structure. analysis. JEFFREY AND CRUICKSHANK MOLECULAX STRUCTURE DETER;LIINATION 3 1 2. Information f r ~ m Optical and Preliminary Crystal Data If crystais of 0-02--0-10 cin. can be grown the optical and preliminary X-ray examination presents 110 difficulty and even if all solvents yield only minute rnal-formed or twinned specimens insurmountable experimental difficulties are seldom encountered.These measurements give much im- port ant information relevant to the crystal-strncture analysis vix. the unit - cc~il dimensions the number of niolecules in the unit cell the spa~e group or a choice of alternativc space groups the refrac;tive indices and the 'Yhe deterinination o€ t81itl imit-cell diinen sioiis froin the measurement of the directions of the diffiactcd spectra can be accurate to 0.001 under good experinitntal conditions. Together with the observed density these data provide the well-known X-ray method of determining molecular weighhs the dorninatiiig factor in the accuracy of which is the difficulty of nieasuring the densities of small crystals.Of the 230 possible space groups only GO are uniquely characterised by lehe X-ray diffraction spectra. Fortunately these include P2,/c which if optically active substances are exchided is by far the commonest for organic crystal structures. The ambiguities among the other 170 arise from the illability of the X-ray diEractioii to rerewl the asymmetry of a structure. A11 crystal stmctures diffract as if they were centro-symmetrical,* and it is only when two elements of translational sylrimetry in combination require a centIre that the space group assignment is unique. Examination by optical or pyro- and piezo-electric methods does not entirely resolve the problem for the external form of the crystals is never absolutely reliable and the electrical measurements cannot detect weakly polar structures.This is illustrated by the historic case of pentaerythritol where in spite of intensive study of the symmetry (Vol. ]I of the " Strukturbericht" PB29 gives 15 references) investigators could not distinguish between two space groups. One of these required the valencies of the central carbon atom to be pyramidal while the other allowed the tetlrahedral arrangement later established by a detailsd ansly~is.~ An X-ray method based on the statis- tical distribut'ion of the intensities of the diffracted spectra has been devel- oped siiice 1945.6 It is iiiost effectivc but suffers from the inconvenience that it gives reliable decisions only with a large number of X-ray intensities whereas it is of-iien desirable that the space group be known before under- iaking the bulk of the experimental intensity measurements owing to the considerable difference in the work involved in an analysis of a ceiitro- symmetrical and a non-cenbrosymmetrical structure.The information about the molecular arrangenient in the crystal which can be deduced from the preliminary data depim€s very much upon the type 4 Peerdeman van Bornmel and Bijvoet PTOC. I<. Akad. Amsterdam 1949 52 313. 6Llewellyn Cox and Goodwin J. 1937 853. Wilson Acfn Cryst. 1949 2 318 ; Rogers ibitl. 1950 3 455 ; Howells Phillips and Rogers ibid. p. 210. * Under very spec i d coidtions not applicable in ordinary structure analysis this is not true and absolute configurations of asyinnietric substames have been determined.4 clc-~age.342 QUARTERLY REVIEWS of molecule and the crystal class. I n layer structures and those with large planar molecules the dimensions of the unit cell together with the refractive indices often reveal the broad features of the molecular packing and put a limit to the range of the possible molecular orientations. I n the crystal structure of a roughly spherical molecule such as a methyl-substituted cycZo- hexane or a hexose on the other hand it is unlikely that the shape and symmetry of the crystal lattice will give any clues to the solution of the structure analysis; for in these the molecular packing will conform to borne three-dimensional system of intermolecular forces and there will be no chain or layer aggregation which can be revealed by the physical proper- ties of the crystals.I n diamond,' which is a particular example of a giant molecule where each carbon atom occupies a special position in a highly symmetrical cubic lattice the interatomic separation is determined directly by measurement of the cell dimension. This parameter has been measured by the very accurate divergent-beam technique * to 0.00005 8. The cell sides of individual diamonds vary over a range of a = 3.55970 & 20EX a t 18" which corresponds to a C-C separation of 1-54447 0.00009 8. There are inorganic covalent or ionic structures which resemble diamond in the simplicity of their structures and many in which a proportion of the atoms can be exactly located from the cell dimensions and space group (as a group the minerals provide the most examples). The most favourable cases in organic molecular structures are those in which tlhe symmetry of the crystal structure corresponds with the symmetry of the molecule and thereby reduces the number of unknown parameters in the analysis.There are niaiiy examples of this for molecules with centres of symmetry and fewer for those with mirror planes or two-fold axes. It is not the general rule however and is exceptional with molecules of higher symmetry. For example although several methane derivatives such as tetranitromethane 9 crystallise with cubic symmetry the six-fold axes of hexaniethylbenzene 10 and coronene11 are not reproduced in the crystal symmetry. 3. The Solution of the Phase Problem From a known or postulated electron distribution in a crystal it is possible to calculate the directions and intensities of all the diffracted bpectra produced under any given experimental conditions.If the reverse were true X-ray structure analysis would be a straightforward procedure leading directly from the experimental measurements to the calculation of the final result. I n fact there is an intermediate step which requires deduc- tive methods and is generally called the phase problem. The structure factor of the diffracted spectrum is denoted by F;;;. where hEZ is the order of the spectrum ; Fgi,". is a complex number giving Structure Reports for 1947-4 Vol. 11 Vosthoek Utrecht. Lonsdale Phil. Trans. 1947 A 340 219. Oda Iida and Nitta J . Chern. SOC. Japan 1943 64 616. lo Brockway and Robertson J. 1939 1324. l1 Robertaon and White J. 1946 607. JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 343 the amplitude and phase of the diffracted wave relative to that from an isolated electron at the origin of co-ordinates under the same experimental conditions.The intensity I h k l which is measured by photographic blacken- ing or some equivalent method is related to Fit;. by where Lhk2 contains Lorentz polarisation and other known factors and I The phases Mhkl cannot be measured experi- mentally and must be deduced during the course of the analysis if use is to be made of the refinement methods discussed later. An obvious method of solving a structurc is to postulate a succession of atomic arrangements until one is found from which the calculated amplitudes I Fg?. 1 agree with the experimental values. This trial structure even if only approximately correct may then serve to give s a c i e n t know- ledge of the phase angles to permit the completion of the andysis by means of Fourier methods.This trial and error method of deduction is very laborious for problems with more than a few independent parameters and important advances have been made in the last few years in the use of Fourier transforms in conjunction with optical analogue mcthods l2 to replace the numerical calculations. There are also two direct approaches to t,hc probleiu of structure solution one discovered in 193.5 by Patterson l3 and the ot)hcr in 1945 by Harker and 1Ca~per.l~ Patterson's method depends upon the fact that while the F:g;- relate to the distribution of atoms in the crystal the 1 Fgk. I 2 relate to the distribution of interatomic distances. Since I Fj?& l 2 is known from (I) this distribution is known and is convcnicntly evaluated by means of the Fourier synthesis expressed in equation (2) and generally called thc Patterson synthesis ; I is the amplitude.This method has had many successful applications and is especially useful if the molecules contain a relatively few atoms of atomic number much greater than the majority. It becomes ineffective for organic struc- tures containing many carbon nitrogen or oxygen atoms because of the multiplicity of interatomic vcctors of equal weight which cannot be resolved. The more recent method is only useful for ccntrosymmetrical structures where the phase angles are 0 or n. It uses the conditions that the electron density is nowhere negative and is represented approximately by a set of spherical atoms in conjunction with the mathematical relationships known as the Schwarz and Cauchy inequalities.A number of structures has been solved with the partial help of this method and a smaller number by this method alone. It has the limitation that it requires several of the X-ray intensities to be exceptlionally strong and tlhc chance of this decreases as l 2 Lipson ant1 Taylor Acta C~gst. 1951 4 458. l 3 Patterson 2. Krist. 1935 90 517 543. Harker and Ihsper. Acta Cryst. 1948 1 70. 344 QUARTERLY REVIE\VY the riuiiibcr of :tt,oni s in the nsynnnietric unit iiicreases.15 A reaerit ,-.tatlist i can1 application appears to extend its scope.16 Other related rniethods of (lire(% sign determination have been proposed which have had a 113orc !imitcd s11ccess.17 1* There are also methods which depend upon special characteri5tics of the structure under investigation ; the most familiar of these are the Iieavj- atom method 19 and the isoinorphous-replacement niethod.l9 By whatever method the phase problem is solved with the possible exception of the last two it is seldom immediately apparent that a correct; solution Iins bee11 foulid ant1 the real test gencrslly lies in the progress of thc subsequent s tmcture refinenlent.4. Verification of the Trial Solution The proof that a set of co-ordinates or of phases constitutes (2 correct solution to the deductive part of the analysis is provided only when it is shown to lead to a structure which refines satisfactorily. The procedure of solution and refinement can be regarded as an attempt to minimise the discrepancies between the observed and cal culttted amplitudes by adjuxsbinq t'hc atomic positions in the unit c ~ l l and the ;ttlomic scattering factor: i.e.i,o ininirnise S O ~ ~ I C wcjghtd fimction of the obw-vcd mid cnlculatmd amplitjndcs ; e.g. where whkl is the weight attached to the observation with indices hkl fi is the scattering factor of thc ith atom whose atomic co-ordinates are xi yi x i in fractions of the unit cell edges ; is summed over all observations ; and The precise forin of R depends upon the method of' refinement but whatever it is 22 is a function of 3N variables which are the co-ordinates in the unit ec4 of the atonis (3N- 1 in a non-ceiitrosyminetrical structure). The aim of the refinement process is to find the lowest miiiimmn in this 3N-dimensional function.In a technically perfect analysis the residual discrepancies would then be due to experimental errors alone. The possibilities of what may happen in refinement can be discussed by considering three forins for the variation of R with change of parameters. Case 1 Jl has a single ininimum only. A form of R in a one-parameter problem might be as shown in Pig. 2 . For this case the trial structure can be any spatial combination of the atoms in the unit cell (with approxima2tcly correct scattering factors) and the correct stniclure within the limits of l6 Hughes Acta Ciyst. 1949 2 34. l6 Zachariason ibid. 1952 5 65. 1' Sayre ibid. p. 60. l9 Robertson J . 1936 1195. hld is suinrned over the N atoms in the unit cell. N l8 Cochran ibid. p. 65. the experinreiital errors will be the tiutoniatic result of an iterative rcfine- inent calculation.It is only in this case that the general method of steepest clesccnts 2o will lead to a solution for any trial set of co-ordinates. For inulti-parameter structures this is a very special case. A one-parameter problem might bc as in Fig. 3. This is the Puttemon ambiguity,21 and is so called because both (or all) the structures which correspond to the minima in Ii have exactly the same distribution of interatomic vector distances. They are therefore equivalent interpretations of the Pattemoii synthesis (equation 2) and of the experimental observations and can in no way be distinguished by X-ray data alone. This is not aJtogether a hypothetical case and a well-known example i 5 the onc-parameter problem of locating There is really no phase problem to solve.Case 2 B has two (or more) identical minima. I J v Q FIG. 2 FIU. 3 Q "i the metal ions in bixbyite (Mn,Fe)203.22 I n this structure there are two physically distiiict arrangements which give the same X-ray intensities. This however is only because the diffraction by the oxygen atoms is so small as to be negligible compared with that of the metal ions. In principle and probably in practice the two structures could be distinguished with very accurate intensity measurements. This was not necessary because onc structure could be rejected as it required the oxygen atoms to occupy chemically implausible positions. If a Patterson ambiguity should occur in the analysis of an organic molecular structure it is even less likely to be unresolved than in the com- pact arrangements of inorganic crystals as the criteria of chemical sense are then more exacting.A mathematical study of the ambiguities in a general one-dimensional array of atoms has been made by Patterson.21 Case 3 .- R has one minimum which is the best fit between the observed nnd calculated sniplitudes and inany others for which the agreement is 3o Booth Nature 1947 160 1 9 6 J. Chem. Phys. 1947 15 415. 21 Patterson Phys. Review 1944 65 195. 22 Pauling and Shappell 8. Krist. 1930 75 128. 346 QUARTERLY REVIEWS more or less inferior. This is the general case which includes all multi- parameter analyses with few exceptions. In a one-parameter problem the form of R might be as in Fig. 4. For an iterative refinement to the true solution to be possible it is essential that the trial structure lies on the slope of the trough that leads to that solution.A trial structure at A will refine to B whereas one at C can only proceed to D. The oidy way of deciding whether a trial structure is at a position like A or C is to proceed some way along the refinement process. This is why tlhe approxiniate true structure is represented in the block riiagrani in Fig. 1 as kiiowii for certain only after the first stage of refinement of the analysis. The distirictioii between a good trial structure e.g. A aid a bad one e.g. G although not immediately apparent is usually clearly defined in practice a t a'n early stage of this first refinement. With a good trial structure the agreement between observed and calculated amplitudes improves progressively at each reitera- tion of refinement until the minimum is reached at which two conditions are fulfilled (i) the agreement is ' satisfactory ' and (ii) the atomic para- meters correspond to a molecule which is chemically sensible.In contrast with a bad trial structure successive refinement does not improve the agreement beyond an unsatisfactory stage and the parameters do not conform to a sensible arrangement of atoms. A qualitative agreement between the observed and calciilated amplitudes is not always sufficient to confirm a correct choice of trial structure and it is useful to have a numerical test of agreement. One that has been very generally adopted is the sum of the differences between the observed and calculated amplitudes divided by the sum of the observed amplitudes for all measured reflexions c hkl I I JEFFREY AND CRUICIfSHANK MOLECULAR STRUCTURE DIIlTZRMINATION 347 Expressed as above where it is necessarily less than 1 it is called the agreement index (or disagreement index !) R ; as a percentage lOOR is the percentage discrepancy.An intercstliiig example of a false trial structure is provided by the analysis 23 of 2 Y-di-(l 3-dioxacycZopenty-1) (I). Chemical evidence sup- ported the cis-1 4 5 S-tetraoxadecalin configuration (11) and a trial structure based on this refined from R = 0.55 to 0.38 but no further. When this was rejected a trial structure based on (I) refined from R = 0.55 to 0.13 and provided the correct solutlion. The Fourier projection corre- I / HZC-0 \ I O-CH (1) (11) sponding to the incorrect trial (Fig. 5 ) was good enough to be inisleading although it implied a C-0 bond length of 1-80 A.It has much in common with the correct electron-density map (Pig. 6). 'The agreement index R in equation (4) does not correspond exactly to the function minimised by either the Fourier or the least-squares method of refinement,24 but it is easily computed and is often used as an indication of the merit of an analysis. I n no way does it replace the proper quantitative calculations of accuracy described later being necessarily a rough assess- ment for quick inspection. As a guide to this diagnostic use it is helpful to consider the values of R fur (i) a random array of the atoms with no relationship to the true structure (ii) a promising trial structure and (iii) a confirmed true solution. For (i) Wilson 25 has shown that the average value is 0.528 for a centrosymnietrical structure and 0.556 for a noii-centro- symmehrical structure.Clearly any trial structure which gives values greater than this is not worth pursuing. A theoretical discussion which has bearing on (ii) has recently been given by Luzzati,26 who has derived a relationship between R and I A r I. I shkl I where I A r I is the root-mean-square of the errors in the atomic positions and I Shkl I = 2 sin Ohkl/a is the reciprocal of the spacing of the crystallo- graphic planes. This relation is shown graphically in Fig. 7 where the full and the broken lines refer to centrosymmetrical and non-centrosymmetrical structures respectively. As would be expected a particular root-mean- square error in the atomic co-ordinates results in a more serious discrepancy between observed and calculated amplitudes for a plane of small spacing than for one of large spacing.The R values shown in Pig. 7 exclude the experimental errors ; they may therefore be 0.05-0.20 less than actual values and the relation is not relevant to the accuracy of a fully refined analysis. 23 Furburg and Hassel Acta Chem. Xcand. 1950 4 1584. 2 4 Vand Acta Cryst. 1951 4 285. 25 Wilson ibid. 1950 3 308. a6 Luzzati ibid. 1952 5 802. 348 QUARTERLY REVIEWS From general experiencc it is found that with a centre of symmetry an agreement index better than 0.50 corresponds to a reasonable trial structure. With a correct solution R should improve on successive refinement to 0.20 FIG. 5 The electron-density m a p corresponding to the limit of reJinement with the incorrect trial structure based o n (11).F I G . 6 T h e electron-density mup corresponding to the limit of refinement with the correct trial structure based on (1). (Figs. 5 and 6 reproduced by permission from Furberg and Hassel Scfu Ch~m. Xcand. 1050 4 1584.) ilr 1. 5 8U“) or better ; further irriproverneiit by tEw inclu4on of hydrogen scattmiiqj and anisotropic scattering factors is dis<:usscd later. For a non-centro- syninietrical structure the values of R should he about 8 of those for centro- symmetrical struct tires other tliirigs being qua1 . JEFFREY AND CRTTICRSIIANK MOLECULAR STRUCTURE DETERMINATION 349 5. Fourier Refinement of Atomic Co-ordinates Apart from providing an elegant representation of the experimental data in terms of an electron-density map,2' the Fourier synthesis is a means of progressing from a crude towards an exact analysis of the structure.A very approximate eiectron-densiby map based on a few known phases can be extended in favourable circumstances to include all the experimental amplitudes by thc well-known method of successive Fozwier rejinement. 28 This was the first systematic method of structure refinement and is still the most efficient mcaiis of carrying out this stage of the analysis. As the refinement of a single structure may require six or more successive Fourier computations this method has etimmmlated the design of a number of special machines the most outstanding of which is the X-RAC,3 where the electron- density contour map is presented on the screen of a cathode-ray tube. The successive Fourier refinement in the centrosymmetrical structure has converged completely when all the observed amplitudes have been assigned phases and there are no changes of phase from the previous step (the non-centrosymmetrical case which has not been properly understood until recently is more complicated and will be discussed later).As is apparent from the discussion in the previous section the method will only lend to the true answer if the first set of co-ordinates correspond t o a correct trial structure. The cis-1 4 5 8-tetraoxadecalin example (p. 347) shows how successive refinement on an incorrect trial structure failed to converge to the true structure. The actual appearance of the electron-density map is often used to judge the progress of the refinement and to makc the decisior whether to pursue or reject a trial structure.The two features which should irnpruvch with ench successive refifieniclnt arc-= (1) the rhapc of the contours near thc :itom ccntrcs which should be almost circular and (2) the electron densihy away from the atoms which should nowhere h a w an appreciable negative value (in the use of X-RAC this cordition ic; the basis of a technique of phase determination for rapid refinement without cslcdation). The Fourier synthesis may be used to represent the distribution of the elcctron density in the unit cell of a crystal in the three different ways expressed in the equations (5)) (6) and (7) where a A and V are the l t q t h of an axis the area of cz projection and the volume of the cell respectively - (5) l a p ( ~ > = a 2 I Fhoo I COB ( 2 d x - cr-noo) .-02 27 Bragg Phil. Tmiis. 1915 215 283. zB Robertson Rep. Progr. Phys. 1935 4 332. 350 QUARTERLY REVIEWS These correspond to the one- two- and three-dimensional Fourier syntheses respectively. I n the one-dimensional synthesis the data for the orders of a principal reflection (- 10 observations) are used to calculate the projection of the electron density iii the whole cell on to ;I principal crystallographic axis. Except in the simplest of structures the eleclron density of inany of the atoms in the cell becomes superimposed and it is generally impossible to obtain any useful information. The two-dimensional synthesis which employs the data from a principal zone of reflections (- 102 observations) is much more commonly used. This gives the projection of the electron density on to a face of the unit cell ; if each atomic peak is resolved in at least two of the three projecfions tiie three co-ordinates of each atom can be deduced.The three-dimensional synthesis uses all the available experimental data (- lo3 observations) and gives the atomic co-ordinates directly. Since atoms in crystals are always more than 1 apart there is no ambiguity due to the overlapping and non-resolution of atomic peaks which is often a source of difficulty in the interpretation of the projections in complicated structures. In every respect bnt one the three-dimensional synthesis is superior to the two-dimensional. This important disadvantage is that the computation of the electron density even at comparatively large intervals (- 0.2 8) throughout the unit cell is a major task ; it is the chief reason why most structure analyses before 1945 were two-dimensional.Now that the requisite large-scale computing mettho& have been developed three- diniensional syntheses can be tackled as a matter of routine and hhey are generally applied to complicated organic structures and to those where the desired standard of accuracy can only be achieved by making use of all available X-ray intensity data. To some extent t h e recent developments in refinement technique and in particular the method of diufference syntheses have overcome the difficultics which restricted the usefulness of the two-dimensional synthesis and these still play an important role in the modern analysis. The three-dimensional refinement is more cumbersome even with large-scale computing equipment and it is commonsense practice to extract as much information as possible from the principal zone reflections before undertaking the full (hkl) coni- putations.The improvements in Fourier refinement technique introduced since ]I 946 although closely iiiter-related are conveniently discussed under three separate headings ( a ) correction for terrnination of series errors ( b ) the use of difference syntheses and (c) the refiririricrit of uuresolvcd projections. ( a ) applies equally to two- and three-cliiiiensioiial syntheses ; ( b ) is also important for both but has so far been used mainly in projectlions. ( a ) Termination o j Series Errors.--Since the first use of Fourier mcthods it has been realised 29 that systematic errors mill occur in the electron- deiisity distribution as a consequence of the use of a finite number of I Fgk;.I in a summation which theoretically requires an infinite series [equations (5) (G) and ( 7 ) ] . These are variously called diffraction errors finite series ewors z9 Bragg mid West Phil. Mag. 1930 10 823. JEFFREY AND CRUICESHANK MOLECULAR STRUCTURE DETERMINATION 351 termination of series errors or simply series errors. The effect for a single peak is illustrated in Pig. 8. The diffraction ripples from one atomic peak shift the maxima of others and in combination they introduce spurious features on the electron-density FIG. 8 (Top) (Bottom) A single peak from a finite Fourier series. A single peak from an infinite Fourier series. map which can obscure and distort fine detail concerning lighter atoms or valency electrons.Another consequence of the finite series is a reduction in the height and curvature of the maxima. A similar effect quite separate from the diffraction errors arises from the spread of the electron cloud by the thermal motion which results in overlupping of the peaks. A consequence of this temperature effect is that n ,-\ FIG. 0 (Top) Two peaks from finite Fourier series with small thermal motion. (Bottom) Two peaks from finite Fourier series with large thermal motion. the positions of the maxima of the electron density a t the experimental temperature are not necessarily the same as those at absolute zero. As an example Cruickshank 30 has shown that with the thermal motion appro- priate to dibenzyl at room temperature two carbon atoms placed 1.386 A 3O Cruickshank Ada Cryst.1919 2 65. A A 352 QUARTERLY REVIEWS apart will appear 0.009 closer in the infinite series electron density; at 1.540 A the contraction is 0.005 A. As explained below the corrections for both the diffraction and the overlapping errors are usually made simultaneously the who1 e correction being called the Jinite series correction although strictly this is ;1 misnomer. Before 1945 attempts were made to eliminate the diffraction errors by applying an arti$ciaE temperature factor to the observed structure ampli- tudes ; this had the effect of flattening and spreading the peaks as illustrated in Fig. 9 thereby increasing the overlapping errors. A simple and effective way of correcting for both diffraction and overlapping errors was proposed by Booth.31 In the h a 1 stages of refinement of a structure analysis there will be close agreement between the observed and calculated F’s and the two Fourier syntheses computed with Fobs.and P a l c . as coefficients will be subject to approximately the same systematic errors provided that both summations are over exactly corresponding sets of terms. The effects of series errors on the maxima in the FCalc. synthesis can be known for they are the differences between the co-ordinates for which the P’s were calculated and those of the corresponding maxima in the synthesis. These differences can then be applied with reversed signs as the corrections to the co-ordinates of the maxima in the Fobs. synthesis. Because the sequence in a structure analysis is to calculate the 27’s from the co-ordinates xobs. of the maxima in the observed synthesis and then to compute an Fc:l*c+ syntthesis to obtain the sealc- the true co-ordinates xt are given by and hence this is sometimes called the back-shift 32 method of correction.In non-centrosymnietrical structures any change in co-ordinates brings about a change in the phases which can have any value between 0 and 2n ; this in turn results in new co-ordinates and the successive Pourier refinement theoretically never ends. Cruickshank 33 has shown that in the limit this infinite process together with the termination of series corrections is equiva- lent to a resultant co-ordinate shift of twice that appropriate for a centro- symmetrical structure. This double-shift rule applies to all co-ordinates in the space group P1 and in other non-centrosymmetrical space groups to those whose directions have no components normal to symmetry axes i.e.those parameters determined only by Y’s with general phases. For corrections in directions norival to symmetry axes e.g. x x in P21 an n-shift-rde 32 is applicable where n has a value between 1 and 2 depending upon the proportions of real and complex P’s. The factor n can coiivenieiitly be introduced into the calculation of the series errors by giving the general asymmetric P’s twice the weight of those whose phase-a’ugles are integral multiples of n.34 The n-shift rule can only be applied where no one atom predominantly determines the phases ; if this is not the case a more elaborate treatment is necessary as e.g. in the analysis of NaNO,.35 1 - - (8) *t = xobs. - (*talc. I p b s . s1 Booth Proc.Rmj. SOC. 1946 A 188 77. 32 Shoemaker Donohue Schomaker and Corey J . Amer. Chcrn. SOC. 1950 72 2328. 3s Cruickshank Acta Cryst. 1950 3 10. Idem ibid. 1952 5 511. 35 Truter ibid. in tJhe press. JEFFREY ANT) CRUICKSHANK MOLECTJLAR STRUCTTJRE DETERMINATION 353 The magnitudes of series corrections to atomic positions are not incon- siderable particularly if the structure contains light and heavy atoms ; this is illtxstratec'l by the analysis of /?-isoprene sulphone. 36 Bond lengths in IB-isoprene sulphone without series correction with series correction H3C Y4 C===CH Theac corrections although never larger than 0.06 A were in such direc- tions as to alter profoundly the interpretation of the results. In the analysis of thi~phthen,~' which also contains sulphur and carbon atoms the greatest series correction resulted in a C-C bond length change of 0.035 A.In the dibenzyl3O analysis the biggest change was 0.037 A ; in naphthalene and anthracene the effects were smaller 38 (Tables 1 and 2). TABLE 1. Bond lengths (A) in naphthalene \Vith serit's corrpct,im 1.369 1.426 1.424 1-362 1.404 1.393 1.. s d * 0.010 0.007 0.009 0-008 0.009 0.0 10 Without w i e i r urrectioii 1.363 1.42 1 1420 1.354 1.395 1-395 TABLE 2. Bond lengths (if) in anthracene 1.366 1.428 1.393 1.398 1.418 1.376 1.408 1.436 1 4 d 0.009 0.007 0.007 0.006 0.008 0.007 0.010 0.007 * E.s.d = estimat,ed standard deviation (see p. 366). Without series correction 1-361 1.426 1.387 1.395 1.412 1-366 1.390 1.440 36 Jeffrey Acta Cryst. 1951 4 55. 38 Ahmed and Cruickshank zhid. 1952. 5 852. 3' Cox Gillot and Jeffrey ibid.1960 3 243. 354 QUARTERLY REVIEWS In all these examples the experimental data were obtained with Cu-K radiation. If the use of a shorter-wave-length radiation increases the number of terms in the Fourier synthesis the finite series corrections are reduced. For example in the resolved projection of oxalic acid &hydrate the mean shifts 3~3 were 0.017 A with the data 4O from Cu-K radiation and 0.002 A with those from Mo-K r a d i a t i ~ n . ~ ~ In anthra- cene the contours in the observed synthesis vary from 7.56 eA-3 to 9.76 eAi-3 but 1.16 eAA3 of this variation has been shown 38 to be due to finite series. In dimethyltriacetylenel a variation of 0.35 e,4-3 from the mean in the uncorrected synthesis was reduced to 0.06 eA-3 by correction. ( b ) Use of Difference Synthe~es.~2! 43-The difference synthesis is a Fourier synthesis with coefficients (3::;- - Fgf.) summed only for those planes for which there is a corresponding term in the Fobs.synthesis [ie. if Fobs. = 0 (Fobs- - P a l c . ) = 01. The series corrections calculated from an Fobs. and an (Fobs- - Pealc.) synthesis will be identical with those from two separate Fobs. and Fcalc. syntheses ; but apart from the computational advantage that the coefficients of difference syntheses are small numbers other uses are made of the fact that they express in graphical form the discrepancies between the observed and calculated electron densities. The criterion for refinement by difference syntheses is to adjust the co-ordinates of the atoms (from which the P a l c . are calculated) until the gradients in the difference synthesis at the atomic positions are zero ; this implies that the gradients in the Fobs.and lircalc. syntheses are equal The finite series errors also affect the relative peak heights. where (pobs. - pealc.) is the electron density on the difference synthesis. The aim is to flatten out the difference map and the appropriate adjust- ments to the atomic co-ordinates or to the scattering factors (fi in equation 3) can be deduced either by direct calculation or by qualitatively interpreting the appearance of the difference map and trying out the effect of successive parameter changes. As all sources of discrepancy between Fobs. and P a l c . superimpose their effects it is always necessary to confirm that a parameter change has resulted in an overall improvement of the synthesis and the trial method has much to recommend it in complicated cases.The characteristic feature of a difference map associated with a co- ordinate shift is illustrated in Fig. 10 in profile and contour. This arises when the observed and the calculated maxima do not become superimposed. Fig. 11 shows an actual example from the analysis of adenine hydr~chloride.~~ The formuh for the calculation of the co-ordinate shift from the values of the difeerence gradient and Fobs. curvature at the ' observed ' co-ordinates are given by Cochran 44 and Cr~ickshank.~~ 39 Ahmed and Cruickshank Aeta Cryst. 1953 6 385. 40 Robertson and Woodward J . 1936 1817. 41 Brill Hermann and Peters Anw. Physik 1942 42 357. 4 2 Booth Nature 1948 161 765. 41 Iakm ibid. p. 81. 4 3 Cochran Acta Cryst.1951 4 408. JEFFXEY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 355 The difference synthesis is also used to extend the mformation which can be obtained from projection data to locate hydrogen atoms and to study the asymmetry of electron density about atom centres these applica- tions are discussed in the next and later sections /' I ' / I \\ / I \ I \ I 1 - I / / \ / \ \ /- \ \ I I \ I / f f '\ I ,I / FIG. 10 full lznes posztzve broken lznes negatzve Theoretzcal difference map feature correspondzng to a co ordznate shzft O n lower dzagram a (0) (b) (c) FIG 11 Example of three successwe stages zn a difference map refinement by co ordznate shzfts of four atoms Contours at zntervak of 0 25 d-= zero contour wuhcated by 0 negatwe contoura broken (Reproduced by permission from Cochran Acta Cryst 1951 4 81 ) 356 QUARTERLY REVIEWS (c) The ReJinement of Unresolved Projections.- A common restriction to the usefulness of Fourier projections is the non-resolution of electron-density peaks which arises when atoms in the same or adjacent molecules overlap in the projections of the unit cell.This is one reason why many of the successful earlier analyses of organic structures mere achieved on planar molecules where by assuming planarity ambiguities could be resolved. The difficulty is well illustrated by the comparison of the two-dimensional projection 45 and three-dimensional sectional representation 46 of the electron-density m a p of single molecules of dibenzyl (Fig. 12). I FIG. 12 (Left-hand side) A two-dimensional pro 'ection of the electron density in chbenzyl.Contour intervals 1.0 23-2. (Reproduced by permission from Robertson Proc Roy Soc 1935 A 150 348) (Right -hand side) T h e three-dimensional sections of the electron density in dibenzyl. Contour intervals are 1.0 o A - ~ zero omitted and one-electron line dotted. (Reproduced by permission from Jeffiey Pruc Roy Soc 1047 A 188 222) I n some analyses the method of bounded projections 47 can be very useful especially for resolving overlapping molecules. It has the disadvantage that it frequently requires computations of the same magnitude as for three- dimensional syntheses without a comparable increase in accuracy. The criterion for applying diflerence syntheses to find the positions of unresolved atoms in projections is the same 34 as for the correction of single- atom co-ordinates i.e.equation (9). The equations given by Cruickshank 34 for the calculation of the shifts from the assumed t o the true atomic positions 4 5 Robertson PTOC. Roy. SOC. 1935 A 150 348. 4 6 Jeffrey ibid. 1947 A 188 222. 47 Booth Trans. Faraday Xoc. 1945 41 434. JEFFRXY AND CRUICKSH4NK MOLECULAR STRUCTURE DETERMINATION 357 involve the gradients in the difference synthesis and some second differential quantities which have to be computed separately. An example of the use of this technique is the two-dimensional refinement of the structure of oxalic acid d i h ~ d r a t e . ~ ~ In practice the (pobs - pcalc.) maps are more FIG. 13 T h e o r e t h l daffereiace nzap feature corresponding to uiwesolved utoirLs placed too close together in profile._... .._ * .. . . FIG. 14 Theoretacal dzfleerence map feature cmrespondzng to unresolved atoms ; a t o w placed too close together full lines posztive broken lines negative ; utorns placed too fur apart vice v0rsa. complicated than for resolved atoms and the technique of flattening them by trial shifts of the co-ordinates is usually the more efficient. The qualita- tive features of difference syntheses which correspond to assumed positions too close together and too far apart are shown in Figs. 13 and 14. 6. The Computational Problem and the Use of Differential Syntheses The refinement of a crystal structure analysis is necessarily an iterative process with the alternate calculation of Fourier syntheses and structure factors. Even in the two-dimensional analysis these calculations are suffi- ciently laborious to inspire the numerous " aids " to crystal analysis which 358 QUARTERLY REVIEWS range from simple slide rules or graphical methods to costly electronic special-purpose machines.For three-dimensional analysis large-scale com- putation equipment is a necessity. The B.T.M.*s and I.B.M.4g punched- card machines have been successfully applied to these calculations and more recently use has been made of the electronic digital computers at Cambridge,5O Man~hester,~l and elsewhere. The speed and capacity of the electronic machines is such that with the Manchester machine programmes have now been devised which make it possible to proceed from the correct 3-7 j /o .$ 2 6 /O /4 /8 22 26 30 ~(60th o f c-aws) ~(60th of r- axis) FIG. 15 Comparison of Fourier and differential Fourier syntheses along the molecular axis in dimethy ltriacety h e .(Reproduced by permission from Jeffrey and Rollett Proc. Roy. SOC. 1952 A 213 86.) trial structure to the refined atomic co-ord- inates by computations carried out cyclically within the machine. For the refinenlent from the confirmed correct trial structure to the accurate atomic co-ordinates the usual form of the three- dimensional Fourier synthesis computation given by equation (7) is very inefficient for the electron density is computed at many unnecessary points in the unit cell. This in- efficiency is enhanced by the fact that the printing-out of the results from an electronic computer is at present some hundreds of times slower than the internal arithmetical opera- tions ; to a lesser degree the same is true of punched-card equipment.The diffpential Fourier synthesis 52 is more suitable for large-scale computing since the values of the gradients and curvatures at a point close to a maximum provide a sensitive method of calculating its exact co-ordinates (cf. Fig. 15). In place of the summation given in equation ( 7 ) ) the method in general requires surnmations for ap/ax ap/ay ap/& of the type shown in of the type shown in equation (11). equation (lo) and for a z p w azp/ag a 2 p w azppx ay azPpx a x azPpy ax where x y z is a point close to the maximurn of the ith atom. The 48 Cox and Jeffrey Acta Cryst. 1949 2 341 ; Cox Gross and Joffrey ibid. p. 351 ; Greenhalgh and Jeffrey ibid. 1950 311 ; Cox Gross and Jeffrey Proc. Leeds Phil. SOC. 1947 5 1 ; Greenhdgh ibid.1950 5 301 ; Truter ibid. in the press. 49 Shaffer Schomaker and Pauling J. Chem. Phys. 1946 14 648 659 ; Donohue and Schomaker Acta Crpt. 1949 2 344; Grems and Kasper ib,id. p. 347. 50 Bennett and Kendrew ibid. 1952 5 109. s1 Ahmed and Cruickshank ibid. 1953 6 765. 6 a Booth Trans. Faraduay Soc. 1946 42 444. JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 359 shifts E ~ E ~ eZ are given by the three equations of the type 7'. The Location of Hydrogen Atoms Since the contribution to the diffracted intensity for each atom depends upon the square of the numbers of electrons the effects of hydrogen atoms are small and their positions can only be detected if a t all in the h a 1 stages of an accurate analysis. In a number of investigations e.g. of DL-alanine 53 and N-acetylglycine 5* the agreement index was significantly improved when contributions were included for the diffraction by the hydrogen atoms.In the analysis of methylammonium chloride,55 in which the parameters were refined by the least-squares method an improved agreement resulted from the inclusion of the hydrogen contribution of a rotating methyl group. I n these and similar examples the hydrogen positions were assumed from the stereochemistry of the adjacent carbon atoms. More recently hydrogen atoms have been seen directly on the Fobs. Fourier maps as ill-defined inflexions or small peaks close to the peaks of the heavier atoms; e.g. anthracene 56 (Fig. IS) he~amethylenediamine,~' ~,-threonine,~~ deca- borane,58 pentab0rane.6~ A much clearer definition of the hydrogen atom electron density is obtained from diflerence syntheses with Fourier coefficients (Fobs.- P a l c . ) in which hydrogen atoms are excluded from Pcalc.. This both removes the peaks of the heavier atoms and eliminates the spurious background effects due to series errors. I n the two-dimensional analysis of adenine hydro- chloride,44 which was based on very accurate Geiger counter intensity measurements a difference synthesis refinement was carried through in con- siderable detail to reveal the hydrogen electrons without any previous assumptions about their positions (Fig. 17). the hydrogen electron density of a rotating methyl group is comparatively well defined in the difference syntheses although barely apparent in the syntheses with Fobs. ; Fig. 18 shows the dis- tribution of electron density above 0-3 eA-3 about a circle coaxial with the linear molecule ; Fig.18 is a section which includes the axis of the molecule and is perpendicular to that shown in Fig. 19. The effect on the agreement index of including hydrogen contributions to the scattering is generally to produce a small but definite improve- ment and in favourable cases values of R lower than 0.10 have been reported. I n dimethyltriacetylene 53 Levy and Corey J . Amer. Chem. SOC. 1941 83 2095. 5 4 Carpenter and Donohue ibid. 1950 72 2315. 55 Hughes and Lipscomb ibid. 1946 68 1970. 5 6 Sinclair Robertson and Mathieson Acta Cryst. 1950 3 251. 57 Binnie and Robertson ibid. p. 424. 58 Kasper Lucht and Harker ibid. p. 436. 59 Dulmage and Lipscomb &id. 1952 5 260. 360 QUARTERLY REVIEWS Three-dimensional Fourier section in the plane of the molecule of anthraccne.Contour intervals are 0.5 eA-a the half-electron line being dotted. (Reproduced by permission from Sinclair Robertson and Mathieson Beta Cryst. 1950 3 245.) FIG. 17 T h e hydrogen electron density in adenine hydrochloride. T h e difference between the electron density projected on (010) and that calculated for isolated C1 0 N and C a t m whose centres are indicated by dots ; contour interuab 0.2 eA-3 negative contours broken. (Reproduced by permission from Cochran Acta C'ryrt. 1951 4 81.) FIG. 18 The hydrogen electron density in dimetkyl- triacetylene. Section of three-dimensional difference synthesis normal to the mole- cular axis. Contour intervals are 0.1 eA-3 ; f u l l lines positive broken lines negative and - * - - - zero leuel.(Figs. 18 and 19 reproduced by permission from Jeffrey and Rollett Proc. Roy. Soc. 1952 A 213 86.) JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 361 The hydrogel5 difference t h e e -dimensional in Fig. 18. 8. The Thermal Motion From the first stages of a crystal analysis allowance must be made for the effect of the thermal motion of the atoms on their X-ray scattering power.60 For isotropic thermal motion the scattering factors for a particular atom fi are calculated from the expression (13) where fio is the theoretical scattering factor calculated and tabulated for the electron distribution of the atom at rest and the exponential term allows for the increased spread of the electron cloud. In this expression B = Sn22 where 2 is the mean-square displacement of the atom in the crystal lattice.The average value of B for all the atoms in the crystal under the experimental conditions can be determined by a statistical method from the slope of a plot of log (1 Fobs. I / I P a l c . I) against log S + B sin2 ell2 where X is the scale factor for converting the observed structure amplitudes 1 Fobs. 1 into absolute values and the I FCalc. I are the calculated amplitudes for a true structure with the atoms at rest. It is usually necessary to adjust the constants X and B several times during a structure refinement. The value of B for diamond a t room temperature calculated from specific- heat data is 0-147 of which 0.127 is the effect of zero-point In organic structures values for B calculated by the method given above vary from close to zero to the order of 10 for molecular structures of low melting point e.g.0.6 A-2 in a-anhydrous oxalic 10.0 in penta- borane 59 a t - 115" which correspond to mean displacements of 0.087 A and 0.36 A respectively. These are not however absolute values for B is a measure of the apparent spread of the electron cloud and will include the 6o James " The Optical Principles of the Diffraction of X-Rays "? 1948 Bell 61 Cox Dougill and Jeffrey J . 1952 4854. fi = fio exp ( - B sin2 6/A2) . London. 362 QUARTERLY REVIEWS effect of all small and random displacements of the atoms from their mean positions. Any detailed study of these atomic thermal motions should also include corrections for certain systematic factors in intensity measurement and in crystal absorption which are often justifiably ignored in an analysis where the object is molecular-structure determination.The absorption corrections for the equatorial reflections from a cylindrical specimen,62 for example lie on a smooth curve which rises with increase of sin O / A ; the omission of these from the Lhkl in equation (1) acts in the opposite sense to the effect of increased thermal motion in that it will sharpen the atomic peaks in an Fobs. Fourier synthesis and so operate in the structure refinement as it weighting factor in favour of the high-order reflections. In layer structures it may be necessary to introduce a t a comparatively early stage in the refinement an anisotropic thermal factor in the form where A and ( A + B) are constants for reflecting planes parallel and per- pendicular to the direction of maximum vibration of the molecule as a whole and 4 is the angle between the normal to the reflecting plane and the direc- tion of maximum ~ i b r a t i o n .~ ~ In the general case the thermal motion is best studied by means of difference syntheses and a refinement of the thermal parameters (if necessary different for each atom) can be carried out simultaneously with the co-ordin- fi = fi* exp { - ( A + B cos2 4) sina O/A2} . - (14) -___- 1 .___#' '. FIG. 20 Theoretical diflercnce map feature cmres- ponding to anisotropic thermal motion. ate refinement .439 44 A discrepancy between Fobs. and P a l c . due to aniso- tropic motion results in characteristic difference map features and the semi- qualitative trial method discussed previously for the adjustment of co- ordinates can also be used for deter- mining the thermal constants.Fig. 20 shows the idealised effect of aniso- tropic motion on a difference map ; in practice there is often an associated scale error so that the pobs-calc- is not zero at the atom centre and the map appears as in Fig. 21.44 The inclusion of the thermal para- meters in FcaIc- generally brings about small changes in the back-shift correc- tions to the atomic co-ordinates which can be assessed directly from the difference synthesis. Ideally the refine- ment of both thermal and co-ordina$e parameters should be pursued until the difference map appears to relate only to the random errors of the experi- mental measurements and shows no systematic features that can be associated with the structure. The detailed procedure will vary from one o a " Internationale Tabellen zur Bestimmung von Kristallstructuren " Vol.11 1935 as Hughes J . Amer Chern. SOC. 1941 63 1737. Borntraeger Berlin. JEFFREY AND CRUICKSHANH MOLECULAR STRUCTURE DETERMINATION 363 structure analysis to another; an example is the refinement of adenine hydrochloride by C ~ c h r a n . ~ ~ In a complete analysis each atom will be defined by three co-ordinates and six thermal parameters all of which have been conjointly adjusted to give the best agreement between Fobs. and lircalc.. There is however a danger in this procedure of attempting to ascribe to thermal motion all the anisotropy of the electron densities about the atoms as this may partly or (0) ( b ) FIU. 21 Examples of three successive stages in a difference reftnement for co-ordinates and anisotropic thermal motion of a chlorine a t m in adenine hydrochloride.Contours at a n interval of 0.5 eA-2; zero contour indicated by 0 negative contours broken. (Reproduced by permission from Cochran Ada Crynt. 1951 4 81.) FIG. 22 The contributions to the atomic scattering factor for carbon from K and L electrons. wholly obscure that due to the distribution of the valency electrons. From a single electron-density distribution it is difficult to distinguish the effects of thermal motion and bonding-electron distribution (cf. the structure analysis of dimethyltriacetylenel). The ambiguity can be resolved in two ways (1) by two complete analyses at very different temperatures ; and (2) by determining the thermal motion parameters from consideration of the high-order reflections only for which the major part of the X-ray scatter- ing is from the inner electrons.Fig. 22 shows the dependence upon sin tl/A of the contributions to the 3 M QUARTERLY REVIEWS atomic scattering power of the electrons in the K and L shells of a carbon atom from the calculations of &I~Weeny.~* Beyond the limiting value of sin O/A for Cu-K radiation the inner K electrons are responsible for more than 85% of the scattering; the outer L electron distribution becomes important only for reflections with sin O/3L < 0.5. Thus if the high-order reflections are adequate to define the thermal motions the difference synthesis method applied to the low-order reflections will reveal the aniso- tropy of the electron distribution due to chemical bonds. At room tempera- ture in hydrogen-bonded and high-melting structures the experimental measurements can be adequately obtained with Mo-K radiation but with crystals melting below 100" it will be necessary to obtain the experimental results at low temperatures so as ko reduce the thermal motion and extend the range of observations to sufficiently high sin 0/A values.I n this discussion the thermal motion has been regarded merely as an added complication to the problem of obtaining accurate inolecular dimen- sions and electron-density distributions. In some structures the thermal behaviour of the molecules is of considerable interest in the study of inter- molecular forces particularly in those cases in which rigid-body angular vibrations take place ; examples of this are geranylamine hydrochloride 65 and antbracene.56 38 More detailed studies of these and other examples can be expected in the future.9. The Least-squares Method of Structure Analysis In the structure analysis of melamine 63 the difficulty that the atoins overlapped in the projections was overcome by refining the parameters by the method of least squares. This was the first application to X-ray analysis of this well-known method,66 in which parameters are determined by making a function of the observed and calculated intensities a minimum. The two functions commonly used in X-ray analysis are or (15) where w and 20'' are weights. The parameters are refined by the stanciard least -squares procedure of successive solution of n simultaneous linear equations in the n unknown parameters (the normal equations). The least- squares method gives a solution even if there are only as inmy observations as parameters but to achieve the best accuracy it is umal to employ all the available data.The computational problem is then of the same order of magnitude its in a Fourier refinement. The relation between the Fourier and least-squares methods was found by Cochran 67 in 1948 who showed for resolved peaks that the co-ordinates 6 4 McWeeny Ada Cryst. 1951 4 512. 6 5 .Jeffrey PTOC. Roy. SOC. 1945 A 183 388. 6 6 Whittaker and Robinson " The Calculus of Observations " 1944 Bluckie London and Glasgow. 67 Cochran Acta Cqst. 1948 1 138. JEFFREY AND CRUICKSRIANK MOLECTJLAR STRUCTURE DETERMMATION 365 found by least squares from the function Bl with w = l/f are the same as those found in the Fourier method when the back-corrections for h i t e series are used.This connection has been examined in detail by Cr~ickshank,~3? 34 who has shown the similarities of the full set of normal equations with the equations used in refining diflerence syntheses when all the peaks are assumed to overlap. When the peaks do not overlap the. non-diagonal trerms between different atoms in the normal equations can be neglected and the use of the least-squares method is then similar to refinement by diflerential syn- theses with back-corrections. Except for the different weighting there is essentially no difference between the use of the least-squares method and the analytical refinement of difference syntheses. The use of B2 is related to the use of the difference Patterson syntheses. Analyses done with the aid of least squares thus do not call for any further discussion and their results are automatically corrected for the finite-series effect.The accuracy of co-ordinates found by the least-squares method is given by standard formulze,66 and the relation of these error estimates to the Fourier estimates to be discussed shortly is similar to that for the refine- ment equations.68 With correct weighting the least-squares method gives slightly more accurate co-ordinates than does the ordinary Fourier method though in several analyses in which co-ordinates have been found by both methods the least-squares errors were estimated to be the larger ; PJJ. in threonine 32 the mean co-ordinate estimated standard deviations were 0.0061 A by the Fourier method and 0.0063 A by the least-squares method.This reversal is probably due to the least-squares weights being inappropriate to the overall errors. In the threonine analysis the root-rnean-square difference of the co-ordinates determined by the two methods was 0.0069 A which is an indication of the effect of the different weighting of the data in the two methods. The use of the least-squares method is not restricted to the finding of atomic co-ordinates and i t may be applied to the adjustment of atomic scale and temperature Cochran has related this to the correspond- ing difference synthesis procedures. 43 The least-squares method requires that the electron density be repre- sented by a finite number of parameters of known type. In this respect it is inferior to the Fourier synthesis method for it does not give the overall representation of the electron density which is not only an aim of the analysis but is also an invaluable guide in steering the coiirse of the refine- 1nent.10. The Accuracy of the Crystal Structure Analysis Up to about 1945 the accuracies of X-ray analyses were judged by the consistency of the results or by the effect of variation of the parameters on the agreement index. In favourable cases bond-length errors were estimated at -& 0.02 to & 0-08 A. No theoretical justification for this had then been given arid it was sometimes even suggested th<at the accuracy of atomic 68 Cruickshank Acta Cryst. 1949 2 154. ** Cruickshank and A. P. Robertson ibid. 1953 6 698. 366 QUARTERLY REVIEWS positions was determined by the resolving power 0-6A/2 sin Omax. This situation was largely cleared up in 1945-1947 by the efforts of a number of workers,7*~ 55 but particularly by Booth’s theoretical work 319 71 on diffraction ripples and the effects of experimental errors which confirmed the possibility of obtaining atomic positions with the order of accuracy claimed from consistency.His treatment of diffraction ripples led to the back-shift method of finite-series correction already discussed and his methods of estimating the other errors have been developed by Cruickshank 3O into the general quantitative method of estimating the standard deviations of atomic positions discussed below. For conciseness the estimated standard deviations will be referred to as e.s.d. (a) Consistency Tests.-A preliminary remark is necessary on the mean- ing of such phrases as “ This bond length is accurate to & 0.05 A ” which sometimes occur in descriptions of structures for which no direct error estimate has been made.The usual meaning seems to be that the errors are unlikely to exceed 0.05 A and not as is usual in statistics that the standard deviation (root-mean-square error) is 0.05 A. This confusion is avoided in the quantitative error method by finding the e.s.d. and then discussing the likely limits of error by statistical significance tests. The use and limitations of consistency estimates of the errors can be illustrated by considering what may happen when a detailed analysis of benzene is carried out. The only symmetry element provided by the crystal space group for each molecule is a centre of symmetry so that there will be crystallographically independent determinations of three of the six carbon atoms.If the molecule is assumed planar one estimate of the errors will be given by the deviation of one carbon atom from the plane through the other two and the centre of symmetry. If the ring is assumed to be a regular hexagon another estimate will be given from the variation of the three C-C bond lengths. The first limitation of the consistency method is thus that it assumes what may require to be proved. In the structure analysis of ethylenethiourea 7 2 (tetrahydro-2-thioglyoxaline) (ITI) the S atom is displaced 0.03 A from the plane of the five-membered ring. An H2C-NH HZC-NH (111) accuracy estimate based on consistency might have taken this as the order of magnitude of the errors but the application of the quantitative inethod showed that the displacement was highly significant.The second limitation of the consistency method is that its estimates of error are based on very few determinations of a parameter so that its 7O Van Reijen Physica 1942 5 461. 7 1 Booth Proc. Roy. SOC. 1947 A 100 489. 7 2 Wheatley Acta Cryst. 1953 6 369. JEFFREY AND CRUICKSHANK IkfOL’ElCULL4R STRITCTTTRE DETHRMINATION 367 reliability is low even when the assumptions are justified. The third limita- tion is that systematic errors are not revealed. In benzene the effect of finite series and peak overlapping will tend to alter all the bond lengths equally without distorting the planarity so that an analysis uncorrected for finite series will tend to show the same deviations as a corrected analysis. Nevertheless if the limitations are borne in mind consistency estimates of the errors can be a useful guide to the progress of an analysis particularly in the early stages when any real departures from say planarity are likely to be small in comparison with the errors.Other rough consistency checks on the progress of an analysis are given by comparisons of similar chemical groupings in different structures or by the comparison of bond lengths with the sums of covalent radii. (b) The Sources of Errors.-The errors in the final results of a structure analysis arise from two entirely different sources the errors in the experi- mental measurements and those in the interpretation of the measurements. The experimental errors are due to (i) errors in the determination of the cell dimensions and (ii) errors in the I Fii;.I . The interpretative errors are due to (iii) approximation errors in the computations and (iv) imperfections in the model used for the Fc:llC.’s causing phase-angle errors and inaccurate finite-series corrections. The overall accuracy of an analysis is fourid by estimating the resultant of these effects ; ordinarily (ii) and (iv) are the most important. (i) By an appropriate experimental technique the cell side errors can be reduced to the order of 0-001 A and they can then be ignored in com- parison with the other errors. However cell dimensions found from the layer line spacings have e.s.d.’s of the order of 0.01 A and then their errors must not be entirely neglected. The fractional atomic co-ordinates given by the refinement process are of coursc independent o€ the cell side errors.(ii) In the majority of tlie crystal-structure analyses mentioned in this Review the X-ray reflections were recorded photographically and the intensities estimated visually. Photometry is a slower and more tedious method and in general when several hundred values have to be measured greater accuracy is obtained in a given time by visual estimation. In d i b e n ~ y l ~ ~ for 94 reflections the s.d. of the difference between inde- pendent measureineiits by phoioriieter and by visual estiniation was 5.0 yo of tlie mean I Fobs. I . Such a cornparison is riot generally possible but the risual practice is to record and visually to estimate each intensity more than once. In t h r e ~ n i i i e ~ ~ for example wherc a careful study of the accuracy was nittde of the 620 possible reflections 65 were too weak to be observed 74 were estiinated once only 303 twice and 175 three times ; the final s.d.of the amplitudes were estimated at 3.50/ for the majority of reflections and somewhat more for the weak and the very strong. Recently Geiger counters have been used for nieasuring intensities giving a considerable improvement in accuracy and in the detection of weak reflections. In a comparison of the relative merits of Geiger-counter and B B 368 QUARTERLY REVIEWS photographic methods Cochran reports 7 3 e.s.d. of 1-2% of for the former against 12-18% of Ihkl for the latter. Apart from the random errors of experimental measurement there nre systematic errors due to absorption and extinction.60* $ * Absorption is well understood and corrections can be made according to the shape of the crystal although this may involve tedious numerical integrations for the (hlcl) intensities.Frequently the problem is avoided by using specimens of such dimensions that the absorption errors are of the same magnitude as the random experimental errors. For example in dimethyltriacetylene the errors of both absorption and visual estimation had e.s.d. 0.05 I I giving a combined value 0.07 I q;;:. I . It is much more difficult to avoid the errors due to extinction. This effect is a departure from the ideal relation of equation (l) which results in low values for the structure ampli- tudes of strong reflections with small sin Ohk2. Sometimes extinction can be reduced by immersing the crystals in liquid air.5 74 Usually extinction affects only a few reflections but if it is suspected that a large number are concerned a lengthy empirical study is necessary.There is no satisfactory theoretical method which can be applied. Extinction is a much more serious obstacle to the study of the positions of hydrogen electrons and valency electrons than to the determination of atomic co-ordinates as the accuracy of the latter depends on quantities such as (2 h2(AP)2}2 the high A P values of the extinguished low-order planes being compensated by their low indices. (iii) Computational rounding off errors are usually small although inadequate methods have sometimes been used. The accuracy of the popular Beevers-Lipson strips for Fourier synthesis has been examined by Cochran 88 and Cruickshank,30 and they have been shown to be adequate for most work on projections.In tlie three-dimensional Fourier analysis of dibenzyl the co-ordinate e.s.d. arising froin their iwe were shown to be only 0.004 8. 30 If co-ordinates are determined from difference syntheses with scaled-up coefficients the strips are adequate for almost any purpose. Of perhaps greater importance is the manlier in which the positions of the maxima are inferred from tlie arrays of numbers given by the Fourier syntheses. Graphical methods or a combination of graphical and analytical methods are very convenient but may not be sufficiently precise for the later stages of an accurate analysis. Strange to say hardly any analyses have been reported using proper two- or three-dimensional interpolation formulx such as the finite-difference form of the equations (12).A sound but laborious analytical method employing more t'hltn the minimum of data is the least squares fitting by a Gaussian function of the 27 densities in a 3 x 3 x 3 block enclosing the i i i a x i r n ~ r n . ~ ~ The use of differential syntheses 5 2 for finding the maxima is more economical and in its full form gives correct positions if the refinements are small. However there may be slight errors if assumptions are made about the peak shapes to save computing the cross derivatives of the type 73 Cochran A d a Cryst. 1960 3 268. 7 4 Lonsdale Mill. M u g . 1947 28 14. ,JEFFREY AND CRUICRSHANK MOLECULAR STRUCTURE DETERMINATION 369 a2p/ax i3y.52 For one of the atonis in ethylenethiourea 72 a difference of 0405 A was obtained for one co-ordinate between the determination using the exact values of the six second derivatives and that using only one inde- pendent second derivative a spherical peak being assumed.The most important computational problem is to ensure that no mistakes have been made. The difficulties arise not from the big mistakes which are almost sure to be noticed but from the smaller ones. A good check is given if the shifts in the last stages of refinement become successively smaller ; ideally the computations should be repeated by a different though equivalent method. (iv) True phase angles and finite-series corrections can only be obtained from the unknown true eleclron density so that those obtained from the idealised model froin which the Fc:llc.’s are derived are necessarily imperfect and some allowance for this must be made.Fortunately it is not usually necessary to estimate this error separately as a simple method is available for estimating the combined effects of this and the experimental errors. (c) Quantitative Accuracy Theory.-The electron density (7) may be written where the inner summations are over those planes having the same I Fhkl I by symmetry e.g. in P2,/a I PA, ] = I Phil I = 1 Fh I = I Fjk. I . Accord- ingly if o(Fhkl) is the standard deviation of Fhkl the standard deviation of tlie electron density is 7 5 For my general (x y x ) position this has the approximate value 30 At special positions in tlie unit cell such as (0,0,0) or (0,0,x) a more general formula has to be used.7 For example in dirnethyltriacetylene space group R3m the estimated error at (0,0,0) was three times that at the general (x,y,s) p0sitions.7~ The values of co-ordinat1e standard deviations depend on the peak cnrvnturcs and on the errors in the slopes of thc c~lt.ctlron density.The simplest formula occurs for n well-resolved spherical peak ill a rcntao- s p n i ~ t r i cal s t m c % i i r e with orthogonal axes ; d i m a(x) = .( 2) / ~ 2 ’ . where for a general position As for the density a more general formula than (21) is required if an atom 7 5 Cruiclmhank and 12ollett Acta Cryst. 1‘35.3 6 706. 370 QUARTERLY REVIEWS is at a special position in the unit ~e11.7~ If the peaks are not spherical or the axes not orthogonal the formula? given by Cruickshank must be used.30 To estimate the co-ordinate errors of peaks in unresolved projections the altogether more general formulz given by Cruickshank and A.P. Robertson are needed.69 In non-centrosymmetric structures the errors in the phase angles have to be allowed for. In the simpler cases the errors are n times those estimated by the centrosymmetrical formulze n being the same con- stant as in the n-shift refinement rule discussed in Section 5(a). These formule can be used to estimate the density or co-ordinate errors if estimates of the (~(3’) can be made. A simple way of estimating the combined effects of the experimental errors (ii) and the model imperfec- tions (iv) is to use AF = 1 I F O ~ S . I - I ~ c a l c . I 1 as an estimate of a ( ~ ) . 3 0 The merit of this method is that the degrees of accuracy of the izleasure- ments and of the precision of the calculated model are reflected in the AP values.It is open to the criticism that it treats the errors (iv) as random whereas they are systematic in the sense that the errors in different Fcalc.’s are correlated but provided the difference map shows no strong features attributable to the calculated model the method gives a satisfactory overall estimate of the errors. An equivalent method of estimating c(ap/ax) suggested by C ~ c h r a n ~ ~ is to average i3(pobs. - pcalc-)/i3x at a number of points where pobs. is expected to be small. If it is of interest to find the errors produced by the errors in the I Fobs. I only estimates of u(F) of the kind discussed in Section 10(b) (ii) may be substituted in (IS) or (21). ( d ) Examples of Co-ordinate Errors.-The most accurate method at present practicable of determining co-ordinates would be to 11 y e Geiger- counter-measured intensities in a three-dimensional analysis.No such analysis has so far been reported for an organic structure. The best results to date have been obtained from three-dimensional analyses using photo- graphic intensities from Cu-K radiation. The lowest e.s.d. yet reported for carbon co-ordinates estimated by using AF for u(F) is U.GO58 A in dirnethyltriacetylene,l a structure with agreement index R = 0.080. 0t)her low estimates are 0.0052 (R = 0*145) a mean of 0-0053 A in anthracene 38 (R = 0-182) a mean of 0.0058 11 in naphthalene 38 (R = 0468) and 04067 A in threonine 32 (R = 0.212 ion- centrosynumetric). All these analyses were corrccted for. terniinatiou of series and in dimethyltriacetyl~.ne slid thrconine 32 the cffects of hydrogen stonis were a1 lowed for.In hetero-atom structures the co-ordiiiates of the heavier atoms are more accurate because of their greater peak curvature ; thus the e.s.d. of the oxygens were 0.00414 A in ar-oxttlic acid and 0.0050 H in threonine. On the other hand the presence of the heavier atoni is detrimental to the accuraey for the lighter atoms because of the lower absolute accuracy of the 1 Fobs 1 ; thus in @-isoprene sulphone 36 the carbon e.s.d. are 0.0167 8 although the sulphur e.s.d. is only 0.0037 A. In ethylenethioures 7 2 the carbon e.s.d. are 0.008 A and those of sulphur 0.002 A (R = 0.13). in ol-anhydrous oxalic acid JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 37 1 The accuracy of the results obtained from two-dimensional analyses depends primarily on whether or not the peaks are resolved.Direct com- parisons between the accuracy of a three-dimensional analysis and two- dimensional resolved projections are available for dibenzyl 30 where the carbon e.s.d. are 0-0074 and 0.0170 A respectively (R = 0.15) and dimethyl- triacetylene,l 0.005 and 0.010 A. Good carbon e.s.d. obtained in resolved projections with photographic intensities from Cu-K radiation are 0.008 8 in ammonium oxalate hydrate 76 (R = O - l l f i ) and 0.013 A in oxalic acid dihydrate 39 (R = 0.146). In spite of the effects of the heavier chlorine atom still better results have been obtained in adenine hydrochloride 44 by using Geiger-counter measurements of intensities from copper and (in part) molybdenum radiation together with finite series corrections from a model with carefully chosen anisotropic thermal parameters giving carbon e.s.d.of 0.007 8 (R = 0.061). The use of the shorter-wave-length Mo-K radiation increases the number of intensities which can be measured though it is more difficult to measure them accurately. I n the resolved projection of oxalic acid dihydrate 39 the e.s.d. of the carbon co-ordinates are 0-0065 A (R = 0.144) as compared with 0.013 A from Cu-K intensities. An example of the accuracy of co-ordinates in unresolved projections is given by the yz projection of the same structure in which the carbon and the carbonyl-oxygen atom overlap seriously their centres being 0.60 A apart. After co-ordinates had been found by the analytical treatment of hhe difference synthesis mentioned above the carbon y co-ordinstc c.s.d.was estimated at 0-031 8 for the Mo data and 0.049 8 for the Cn data only. The errors would have been considerably larger if the co-ordinates in tlhe unresolved peak had not been found by difference syntheses or some equivalent method. (e) Errors in Molecular Dimensions.-From the chemical point of view the interest lies in the errors of the molecular dimensions the bond lengths and the bond angles. The 8.d. a(Z) of a bond between two atoms whose positions are determined independently is where a(A) and a(B) are the atomic co-ordinate s.d. For two independent atoms of the same kind o(Z) = .t/2a(A). For a bond across a centre of symmetry o(2) = 2a(A). Thus apart from the special cases of diamond and graphite the most accurate determinations of C-C bonds yet made have e.s.d. of 0407 8 or 0.010 A if across a centre of symmetry (co-ordinate e.s.d.0.005 8). The s.d. a(@ of an angle 8 between three independent atoms A B C with co-ordinate s.d. a(A) o(B) o(C) is given b y 6 9 a(Z) = [a2(A) + a2(B)]4 02(*) + a2(B) (m2 1 - 2 COS8 aye) = - AB2 For a(A) = a(B) = a(C) = 0-005 A and AB = BC = 1-40 A a(@ = 0.40" for 8 = 120". 7 6 Jeffrey and Parry J. 1962 4864. 372 QUARTERLY REVIEWS If the molecular symmetry can be assumed to be higher tlian the cryst,allo- graphic symmetry the averaged bond lengths will be more accurate than the individual bond lengths. This can be illustrated by reference to the finite-series-corrected bond lengths of naphthalene and anthracene 38 given together with their e.s.d. in Tables 1 and 2. None of the pairs of chemically equivalent bonds in either molecule differs by more than the e.s.d.of their differences so that there is no evidence for differences in the bond lengths caused by intermolecular forces. The averaged bond lengths are given in Tables 3 and 4 allowance having been made for the negative correlation of BC and CD in naphthalene through their common atom C and siniilarly for CD and DE in anthmccne. The e.s.d. of the averaged bonds are also given in these tables the most accurate bond being CU in anthracene with e.s.d. 0.004 A. First the lower accuracy of bonds further from the centres of the molecules Tables 1 and 2 also illustrate two other features. L c d I 0.009 i 1.393 0.010 Tlwor. 1.406 1.424 TABLE 4. Chemically independent bond lengths (8) in anthracenc Exptl 1.10s 1-436 I I II I 1C.h tl ‘I’liwr. 1 I 0.010 1.410 1 0.007 1.430 I owing to the rigid-body angular oscillation in the thermal rnotioii causing the outer peaks to be less sharp; and secondly that though atoni C is itself the most accurately defined in naphthalene the bond CC’ is the least accurate as it is across the centre of symmetry.(f) Examples of Electron-density Errors.-Equation (18) shows that the error in the electron density increases with the number of planes observed. Electron-density errors thus do not have the saim absolute significance as co-ordinate errors and are necessarily relative tlo the number of terms included in the synthesis. However by way of examples in dibenzyl 30 the e.s.d. is 0.125 eA-3 using AF as an estimate of a(F) ; in the two- dimensional projection of adenine hydrochloride 44 the e.s.d.is 0.1 eAW2 ; and in dimethyltriacetylene the e.s.d. due t o the random errors of intensity estimation only is 0.02 eAU3 at the general positions in the unit cell. Especially in metallic structures there is considerable interest in the number of electrons associated with each atom ; the errors in counting electrons in peaks have been discussed by Douglas 7 7 and by C o ~ h r a n . ~ ~ 77 Douglas Acta Cryst. 1950 3 19. JEFFREY AND CRUICKSHANK MOLECULAR STRUCTURE DETERMINATION 373 (9) Possible Improvements in Accuracy.-From the general connections between the least-squares and the Fourier method mentioned earlier it can be shown 68 that the most accurate co-ordinates from a given set of data are obtained either from the correctly weighted least-squares equations or by using a weighted Fourier series where w’ is a weight and f is the scattering factor.It is estimated for dibenzyl 68 that this would improve tlhe co-ordinates e.s.d. by about 25% from 0.0074 A to 04056 8. Accordingly large improvements in accuracy are not to be found by any new method for handling the data. Such im- provements will depend on more and better experimental data and on the use of more elaborate models for the FCalc.’s in the refinement process. More data can be obtained by the use of low temperatures with sufficiently short wave-length radiation. The problems of obtaining more accurate data have already been discussed in Section 10(b) (ii) but it is important to realise that reduction of the random errors of intensity estimation is only part of the obstacle to increased accuracy.In the two-dimensional projections of dibenzyl30 the effect of the random errors of the photographic estimation is estimated as a component of 0.006 A out of the total of 0.017 A. In dinzethyltriacetylene,l where a more refined calculated model was used the photographic random intensity errors were estimated at 0.003 out of the total of 0.005 A leaving 0-004 A due to other causes. These instances show that the photographic method with visual intensity estimates gives good results when used with care and that there is no point in making accurate intensity measurements unless corresponding trouble is taken both with the absorption and extinction errors and with the refinement of the calculated model. A study of Cochran’s account of the refinement of adenine hydro- chloride 44 where Geiger-counter intensities were used will show the import - ance of the latter effects.The use of more elaborate models for the Fca1c.7s is primarily a difficulty of the labour of computation as the principles involved in allowing for bonding electrons and anisotropic thermal motion now seem to be well understood. 11. Comparison of Experimental and Theoretical Molecular Dimensions The probability distribution for the errors in the bond lengths and bond angles approximately follows the Gaussian error law so that one can never be certain that the actual error is less than any given amount. Accordingly it has been suggested 309 69 that statistical significance tests should be used in comparing experimental and theoretical results and in comparing two sets of experimental results.The problem of comparing experimental and theoretical results is usually a matter of testing the hypothesis that the theoretical values are the true ones. For example if a bond length determined experimentally with e.s.d. cry differs by aZ from a theoretical value the question to be decided 374 QUARTERLY REVIEWS is whether the difference might easily occur by chance due to the experi- mental errors or whether the difference is real and the hypothesis that the theoretical value is the true value is false. A standard statistical procedure for making such decisions is as follows. On the hypothesis that the theoretical value is correct let P be the probability that a difference I &’ I or greater can be observed. If P is very small the hypothesis is regarded as doubtful. On the other hand if P is not small the experimental values are regarded as not being inconsistent with the theoretical values’ being correct though of course they cannot prove this.When P is so small as to cast doubt on the hypothesis the differericc is said to be significant ; just how small P has to be for this is arbitrary but most purposes are served by taking P < 0.01 as significant. Values between 0.05 and 0.01 are sometimes taken to be possibly significant and those c< O-COl to be highly significant. For these values of P the corresponding values of t = al/o for the Gaussian distribution are P = 0.05 t = 1.960 ; P = 0.01 t =I 2.576 ; P = 0.001 t = 3.291 ; thus for practical purposes a difference of more t,han three times the e.s.d. may be taken as real. In testing the hypothesis that two experimental bond lengths refer to the same true value 0 is taken as the e.s.d.of their difference. The use of significance testis can be illustrated 38 for the naphthalene a i d anthracene results already discussed. These are prticularly important examples as both molecules have been the subject of detailed theoretical study by Coulson Daudel and Robertson.78 Tables 3 and 4 show the theoretical bond lengths found by the molecular-orbital method. For naphthalene the differences on the bonds RC and AE’ are not significant. For the bonds AB and CC‘ 0.01 > P > 0.001 so that the differences between theory and experiment are significant. For anthracene the only notable difference is for CD with t = 2-5 giving 0-05 > P > 0.01 which is possibly significant.Comparisons of this sort taking the bond lengths one at a time give very useful valuations of a theory but they do not give any overall figure of merit for a molecule as a whole. To overcome this difficulty Cruickshank and Robertson 69 have proposed the use of multivariate significance tests in which by taking account of the mutual correlations of error a simultaneous comparison of a number of parameters can be made. The simultaneous comparison 69 for the four chemically non-equivalent bond lengths of naphthalene gave P = 04001. For the five bonds of anthracene P = 0.05. Thus treating the molecules as wholes the difference between theory and experiment is highly significant for naphthalene but only possibly significant for snthracene. These comparisons neglected the admitted imperfections of the theory,78 which are estimated to produce corrections up to about 0.015 ,& per bond but their importance is that they show that it is hardly necessary to postulate any errors in the theory for anthracene but that the errors for naphthalene though small are very significant.They support 7B Coulson Uauclel and Robertson Proc. Roy. SOC. 1951 A 207 306. JEFFREY AND CRUICKSHANK MOLECULAR STRUCTTJRE DETERMINATION 375 the view 79 that the inolecular-orbital theory becomes progressively better as the size of the molecule increases. 12. The Relation between Experimental and Theoretical Electron- density Distributions Hitherto the discussion of the results of molecular-structure analyses has been based on the comparison of bond lengths and valency angles either from different experimental sources or from corresponding experimental measurements and theoretical predictions.The remarkable degree of cor- relation which can be obtained for the latter under favourable conditions is illustrated by the naphfhnlene and anthracene examples discussed in the last section. Not nearly so much attention has been given to those aspects of the electron-density maps which are not directly concerned with the measure- ment of molecular dimensions. Although in isolated examples 80 44 1 some partial success has been attained in trying t o interpret electron densities in terms of bonding electrons,* it is also very important to try to extend the comparison between experiment and theory to include the overall electron density distribution. Such a comparison would in fact be more direct than that relating t o bond lengths for there would be no necessity for the inter- mediakc use of empirical bond order-lengt,h relations.*l There is however an important gap betwccn the experimental and theoretical descriptions of the electron densities which must be bridged before any nseful comparisons can be made. First the experimental elsctron-density map calculated by Fourier synthesis is not the true distribution in the crystal for it is distorted by the finite series errors ; secondly the theoretical calculations refer to the isoZated molecule at rest. It would be unsound to deduce a sharpened infinite-series electron-density map from the more diffuse observed map for trivial irregularities due to experimental errors might be magnified into apparently significant detail.The procedure must therefore be reversed and the theoretical electron distribution must be given the experimentally determined thermal parameters and then converted into the corresponding series-terminated map. A direct comparison could then be made by means 79 Coulson J. Phys. Chem. 1952 56 311. 8o Brill Acta Cqsf. 1950 3 333. 81 Cox and Jeffrey Proc. Roy. SOC. 1951 A 207 110. 8 2 Bernal J. 1946 643. 83 Cork Phil. Mag. 1'327 4 688. 8 p Harker J. Chem. Phys. 1936 4 381. 8 5 Cox and Jeffrey Nature 1939 143 894. 8 6 Crowfoot Bunn Rogers-Low and Turner-Jones " The Chomistry of Penicillin " Ox€ord Univ. Press 1949 p. 310. 13' Cochran Acfa Cryst. 1953 6 260. 88 Cochran ibid. 1948 1 61. * A recent analysis 137 of salicylic acid carries the interpretation of the electron density further than in any previous work and interesting deductions are made con- cerning the bonding-electrons distribution and the anisotropic3 thermal motion of the atoms. 376 QUARTERLY REVIEWS of an (I$;;. - F $ y ) difference synthesis which would show in coiivenicnt graphical form the overall differences between the electron distribution in the theoretical molecule and in the true molecule in the crystal environment. The Reviewers express their thanks to Dr. Maryon W. Dougill for assistance with the Figures.

ISSN:0009-2681    

DOI:10.1039/QR9530700335

出版商:RSC    

年代:1953

数据来源: RSC

摘要:

By 1%. S. NYHOLM M.SC.(SYD.) PH.D. D.SC.(LOND.) A.R.I.C. (THE NEW SOUTH WALES UNIVERSITY OF TECHNOLOGY) THE magnetic properties of inorganic compounds have been studied for more than oiic hundred years but the value of magnetic measurements in inorganic chemistry has increased remarkably during the last two decades. This development is largely the outcome of work on valency theory which now enables us to correlate more confidently the valency bond type and stereo- chemistry of an atom with its effective magnetic moment. Before dis- cussing the modern views on this subject the fundamentals of magnetism will be sunimarised without experimental details. The application of magnetic susceptibility measurements to the above kinds of problem in inorganic chemistry will then be examined ; particular regard will be paid to the more recent work.Attention will be confined almost entirely to the results of measurements on paramagnetic compounds because these provide more useful information than do studies on diamagnetic substances. Finally some developments in crystalline field work and paramagnetic resonance will be reviewed. Theoretical Principles ( a ) Fundamental Definitions.-Tlieoretical aspccts of this subject are dealt with at length in standard works ; a summary of fundamentals will suffice for our purposes. Cheiiiical substances may be classified on the basis of their behaviour when placed in a non-uniform magnetic field. Those substances which tend to move from a stronger to a weaker parb of the field arc called " diamagnetic '') whilst those which tend to move from a weaker to a stronger part of the field are usually called " paramagnetic ".(Further sub-divisions will emerge as discussion proceeds.) The extent to which a material becomes inagnetised is measured by its magnetic moment per unit volume usually known as intensity of magnetisation I . Inside the inaterial the total number of lines of force per unit area (the magnetic inductlion B) is given by the expression B = H + 4nI where 11 is the strength of the magnetic field in vacuo. On dividing throughout by H we obtIain B / H = 1 + 4nI/H. The ratio of the iiumbcr of lines of force per unit area inside the niaterial to thc number of lines of force per unit area J. H. Van Vleck " Theory of Magnetic and Electric Susceptibilities " Oxford Univ. Press London 1932. a E. C. Stoner " Magnetism and Matt,er " Methuen London 1934.S. S. Bhatnagar and I<. N. Mathur " Physical Principles and Applications of P. W. Xelwood " Magnetochemistry " Interscience Publ. New York 1943. L. F. Bates " Modern Magnetism " Cambridge Univ. Press 2nd Edn. London I\lagnetochemistry " Macmillan London 1935 1948. W. Klemm " Magnetochemie " Akadeinische Verlagsges. Leipzig 1936 377 378 Q WARTERLY REVIEWS in vacuo i e . B/H is called the " magnetic permeability * " (!I,) of the material. The ratio I / H is known as the " volume susceptibility '' of the material and is given the symbol k. This is what one usually nieasures in experimental determinations but of greater interest is the gram-suscepti- bility x related to the volume susceptibility by the expression x = k/p where p is the density of the substance.For chemical compounds the molar susceptibility xnl is more important ; this is obtained by multiplying x by the molecular weight. Similarly the gram-atom xA and gram-ion xA+ susceptibilities are obtained by multiplying x for an atom or ion by the atomic or ionic weight respectively. For most chemical substances x is independent of tlhe field strength but for ferromagnetic substances it first increases rapidly as H increases and then gradually decreases as saturation is approached. We shall not be concerned with ferromagnetism here except insofar RS incipient ferro- magnetism is observed in some paramagnetic compounds e.g. ferric fluoride. Wherever x is found to be dependent on field strength magnetic moments become meaningless. A convenient way of determining x is by the use of Gouy's met,hod.s The simple t'heory of this illustrates several general principles.A cylindrical column of the material of cross-sectional area A and length 2 (immlly in a suitable glass tube) is suspended from a seiisit,ive bdnnce so that one end is placed between tthe poles of an electro-magnet capable of producing a stlrong inagiletic field HI generally of the order of 3000-10,000 oersteds. The length of the specimen must be such that H, the field strength a t the other end is negligible compared with HI. The difference in pull on the balance dw with the magnet on and with the magnet off is then measured. If the volume susceptibilities of the surrounding medium (usually air) and the specimen are respectively E and E l then the force P acting on the specimen owing to the magnetic field is given by P = dw x g = i(kl - k,) (HI2 - HZ2)A.When H is very small compared with H I this reduces to dw x g = +(kl - k,)H12A whence Replacing E by x x density and taking v and W as the volutiie itlid iiiass of the specimen respectively we get This equation contains the unknowns v Hl and A which need not be measured (except in absolute determinations of x) if one is prepared to calibrate the tube with a material of known susceptibility e.g. nickel chloride s o l ~ t i o n . ~ The expression reduces to x = (a + bdw)/TV where a and b are constants for any particular tube and field strength. 6w is gener- ally measured in milligrams. Values of x vary from - 3 to - 0.1 x 10+ for diamagnetic substances and may be as high as + 100 x for para- ' H. Bizette and B.Tsai Compt. rend. 1939 209 205. s Ibid. 1889 109 935. * Not to be confused with the " magnetic moment " discussed presently for which kI = E + %W x g/AH12 x == ( k ~ -+ 2gv dw/AHlz)/W. H. R. Nettleton and S. Sugden Proc. Roy. Soc. 1939 A 173 313. the same symbol is used. NYHOLM MAGNETISM AND MORQANIC CHEMISTRY 379 magnetic substances like ferric salts. An excellent apparatus for measuring both dia- and para-magnetic susceptibilities by the Gouy method is des- cribed by Baddar Hilal and Sugden.10 ( b ) Diamagnetism.-An electron in an orbital produces a magnetic field equivalent to that of an electron moving in a circle. On application of an external magnetic field a precession occurs (the Larmor precession) which gives rise to a magnetic field in the opposite direction to that of the imposed field.(i) Atomic Diawmgnetism.-For an atom containing n extranuclear electrons it can be shown by a classical mechanical treatment that This results in diamagnetic susceptibilities being negative. X A = - 2.832 z? x lo1' 9& where 3 is the mean of the squares of the radii of the projections of the orbits perpendicular to the field. Diamagnetism is characteristic of aZZ atoms though it is often swamped entirely by the much greater para- magnetism arising from spin or orbital motion. A quantum-mechanical treatment gives essentially the same result but the expression obtained enables one to calculate xA more readily. This involves the use of effective nuclear charge ZeR. x e instead of 2 x e. Even so for the apparently simple inert gases only fair agreement has been obtained between experi- mental and calculated results.ll It follows from the above formula that atomic diamagnetic susceptibilities (a) are independent of temperature and (B) increase with increasing size of the atom.(ii ) Ionic Diamagnetism.-Theoretical calculation of this is carried out in the same way as for atoms the effect of a negative charge being to increase the value as compared with that of the free atom. For most practical purposes it is obtained experimentally by assuming that in a solution or ionic lattice xLlf = Xcntion + xAnion. As with similar problems the alloca- tion of the total xM between cation and anion is somewhat arbitrary and variations of the order of & 5% (or more) are found between the values recommended by various worker~.~ 5 12 l3 The diamagnetism of ions has been reviewed by MyersJ14 who emphasises that no simple method for obtaining ionic diamagnetic susceptibilities has yet been found.(iii) MoZecuZar Diamagnetism.-This is of special interest to the inorganic chemist siiice it must be corrected for in calculations of effective magnetic moments. The classical theory of diamagnetism is inapplicable to mole- cules but for hydrogen the susceptibility has been calculated by using yuantuni mechanics ; good agreement between theory and experiment is observed.15 For other molecules the calculation is not practicable and use is made of the expression xlcr = nAXA + A where nA and xA are the n lo J. 1949 132. l2 V. C. Trow Trans. Furuday Xoc. 1936 32 1658; 1941 37 476. l 3 W. Klemrn 2. anorg. Chem. 1941 246 347. l4 W. R. Myers Rev.Mod. Physics 1952 24 15. la G. Stehsholt Phil Mag. 1947 38 748 and refs. therein. l1 P. W. Selwood ref. 4 p. 34. 380 QUARTERLY REVIEWS number and atoinic susceptibilities of the different kinds of atom present and 3 is the “ constitutive correction ” correcting for all incalculable effects such as double bonds stereochemical isomerism etc. I n effect all depar- tures from additivity are embodied in 1. Tables of commonly used values of xA and 3 are available ; l6 they are usually known as Pascal’s constants after the first important worker in this field. Molar susceptibilities cal- culated by using these are at best approximate but since the diamagnetic correction is usually small compared with the paramagnetic susceptibility the uncertainty of atom and ionic susceptibilities is not serious.For mixtures of diamagnetic substances additivity is generally a~sumed,~7 but wheii marked solvatioii occurs as when a salt is dissolved in water con- siderable deviations from additivity are observed.14 The disagreement between experimental and estimated diamagnetic susceptibilities of mole- cules has severely limited the value of magnetic measurements for the con- firmation of structure of diamagnetic compounds.l* Gray and Cruick- shank l9 put forward a method to overcome the lack of additivity of Pascal’s constants which takes into account resonance structures residual charges based on dipole moments and the effect of bond formation as well as more usual features. I n general agreement between calculated and experimental susceptibilities is better by tlheir method than if one uses Pascal’s constants alone but considerable disagreement is often observed.2oy 21 Although frequently successful the mcthod is largely empirical and lacks a sound theoretical basis. The distribution of charge using electric dipole moments is undoubtedly incorrect in many instances because no allowance is made for the iinportance of the lone-pair moment . 2 2 ( c ) Paramagnetism.-The theory of paramagnetic and electric suscepti- bilities has been fully described by Van Vleck,l the essential principles being the same. If a subst)ance with a permanent niagiietic moment / i is placed in a magnetic field with the niolecnlar magnets free tlo orient theinselves they will be subjected to two opposing effects (i) the magnetic field of strength H which tends to align all molecular magnets in the same direction and (ii) the thermal effect of vibration rotation and translation (the kT effect) which tends to make the directions of the molecular magnets entirely random.It can be shown that xal = N2p2/3RT where N = Avagadro’s number and ti = magnetic moment. Prom this expression it follows that ,LL = ~ / ~ R T x / N . In practice we are iiitcrested in the magne-tic riioiiicnt of a particular utorn in ;L comp~mnd rather than that of the molecule 8 s a whole and hence we replace xLII by the susceptibi1it)y of the atorii xL4 with which we are concerned. must be added a correction for the diamagnetic susceptibilities of all atoms present in the inoleculc including that of the paraningiietic atom itself though this last correction is often Therefore to l6 P.W. Selwood ref. 4 p. 52. W. R. Angus Ann. Reports 1941 38 27. lB Trans. Paraclay Soc. 1935 31 1491. 2o S. K. Sidtlhanta J . Indian Chem. Soc. 1947 24 21. 21 S. K. SiclcUianta and P. Ray ibid. 1943 20 359. 2 2 C. A. Coulson “Valence ” Oxford Univ. Press London 1952 p. 210. Idem ref. 4 p. 5 7 . NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 381 ignored. On substituting ;dLll. = (;dM + diamagnetic correction) in the place of xJf in the above expression we obtain the " effective magnetic moment " ,ueff. of the paramagnetic atom i.e. peff. = 2-8392/x3, x T. Since in nearly all work peff* rather than ,u is used the subscript is often dropped. The effect of making the necessary diamagnetic correction is illustrated by Table 1. Magnetic moments are now expressed in Bohr magnetons (B.M.) ; this is the natural unit of magnetism and equals the magnetic moment of an electron assumed to be " spinning " on its own axis.It is given by the expression ehj4nmc and has the numerical value 9.273 x 10W31 erg gauss-'. -__ TABLE 1. EfSect Compound [Ni(NH,),]C12 . . [Ni( dipy),]12,6H,0 . of diamagnetic correction on the magnetic moment 17.10 3962 3.06 133 4095 ~ 3.11 488 1 4156 ' 3.13 4.16 I 3698 I 2.95 I The value of ,u obtained from the above formula is a constant only when xIM is proportional to 1/T ; this is true for a large number of magnetically dilute substances.* The compound then obeys Curie's law but more gener- ally the Curie-Weiss Law xl+x oc 1/(T + O ) represents the variation of xJf with absolute temperature. The value of 8 is constant over a smaller or greater temperature range ; several factors contribute to its origin particu- larly the effect of external fields on the molecular magnet ism?^ 4 9 When 8 is known p is usually calculated by using (T + 0) instead of T in the above formula.Paramagnetism is found mainly in the following classes of compound (a) compounds of the transition elements (particularly the first series from Ti to Cu) owing to the presence of unpaired d electrons ; ( b ) compounds of the rare-earth and actinide series owing to the presence of 4f and 5f electrons respectively ; ( c ) a few compounds of the first period which contain unpaired p elect'rons e . g . O, NO ; ( d ) a temperature- independent paramagnetism is shown by a few compounds which have no unpaired electrons e.g. KMnO, K,CrO,. Except in rare instances e.g.(d) above paramagnetism arises from the presence of unpaired electrons in an atom. Unpaired electrons have both spin angular momentum s and orbital angular momentum I ; a combination of these two gives rise t o the magnetic momcmt but the way in which they are coupled is most important. The usual way in which coupling occurs in an atom containing more than one unpaired electron is according to the Russell-Saunders scheme. In this the spin moments (s) are first coupled to give a resultant spin moinentum X ; if n is the number of unpaired elec- trons in an atom then X -= T L / ~ . The various I values of the different * A magnetically dilute compound c.y. [Ni(NH,),][ClO,], is one in which the number of paramagnetic atoms is small compared with the total numbor of atoms present.Hence the distance between pawmagnetic atoms is large compared with their usual bond radii ; this ensures that the interaction between them is negligible. 382 QUARTERLY REVIEWS electrons are then separately coupled to give a resultant orbital angular momentum L. The spectroscopic rules governing this are discussed by Herzberg; 23 Hund’s rules are of special importance.* The resultant S and L vectors are then coupled together but ofken less strongly than the way in which the various I and s vectors are coupled to one another. The significance of this looser coupling becomes evident in the moments of the compounds of the first transition series ; often the L component is almost entirely quenched leaving only the X component contributing to the mag- netic moment. The coupling of L and X gives the inner quantum number J from which the magnetic moment can be calculated; J measures the resultant total angular momentum.J can have the possible values (L + X) ( L + X - 1) . . . (L - X + l) (L - S ) positive values only being real. It can be shown that when L > X the number of possible J values for a given value of L is 2 8 + 1 and when L < 8 the number of possible J values for a given value of X is 2L + 1. In calculating the magnetic moment it is the value of J corresponding to the lowest energy (J,,) which is usually most important (see Rule 3 7 ) . The energy separation between levels corresponding to successive J values relative to the size of kT deter- mines the extent to which levels other than the lowest are occupied. In the ideal case when the multiplet intervals which correspond to the adjacent values of J are large compared with kT only the lowest energy level (Jo) will be occupied.On application of a magnetic field this level will be split into 2J0 + 1 energy levels the separation between which is gpH where H is the magnetic field #I is the Bohr magneton and g is called the Land4 splitting factor which is given by the expression The magnetic moment (in B.M.) is then given by the expression peff. = gz/J(J + 1). It is noteworthy that for atoms in S states (ie. when L = 0 and J = X) g = 2. If it is found experimentally that g lies between 0 and 2 both spin and orbital momenta are involved in the magnetic moment but when g = 2 the magnetic moment arises from spin alone. This will be referred to again in discussing paramagnetic resonance.As an example the Ce+++ ion has one unpaired electron the term symbol for the ground state being 2F5,z.T Here s = 4 I = 3 and since the 4f shell is less than half full it follows that j = E - s = 24. g can be shown to be 6/7 hence /ceff. (calc.) = 2.54 ; the experimental value is 2-51 B.M. For precise rcsults allowance must be made for the temperature-independent contribution of the high-frequency elements of the magnetic moment of the diamagnetic part (the a correction) ; this is discussed by Van Vleck. 23 “ Atomic Spectra and Atomic Structure ” Dover Publ. New York 1944 p. 128. * Namely ( 1 ) S has the highest value allowed by the Pauli principle ; (2) the value of Ti is the highest then allowed ; (3) J = L - S if the shell is less than half full and J = L + S if it is more than half full.The letters X P D and P represent L values of 0 1 2 and-3 respoctively. The prefix gives the spin multiplicity (28 + l) and the suffix the value of J . ___I_ t In a term symbol the letter (here F ) gives the value of L. NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 383 Modification of the above formula is necessary when certain of the con- ditions e.g. wide multiplets absence of field effects etc. are not satisfied. It is most convenient to discuss the various kinds of magnetic moments observed in paramagnctic coilzpounds under seven headings. We shall then use these as the basis for a survey of recent magnetochemistry of the elements of the Periodic Table. Class I. Spin and orbital momenta free and hv(J + J1) > kT. (Rare-earth type.) Here the separation between the ground state and the next higher level is large (- lo4 cm.-l) compared with kT(- 300 cm-l).The a correction being ignored p is given by gz/J(J i- I ) . Curie's law is obeyed closely by this class of compound. With the vxceptioii of Srn(J11) and Eu(m) this formula agrees well with the observed moments of the compounds of the rare earths. Spin and orbital momenta €ree and hv(J -+ J1) < kT. This is the opposite extreme to Class I; when these conditions obtain the moment is given by 2/4s(S + 1) + L(L + 1). In this expression L(L + 1) gives the orbital and 4X(X + 1) the spin contribution. In practice some quenching of the orbital momenta occurs but reference t o Table 3 shows that in certain circumstances moments of the Feff and Co+f ions agree better with this rather than the " spin-only " formula (Class IV).The Curie law holds for this class of compounds also. Spin and orbital momenta free a,nd hv(J -+ J,) iwuparable witth kT. The expression for calculating p in these circunist:trices is coinplicated owing to the nwe4ty for taking into atcwunl a Bo1r;zmann distribution between various energy 1evels.l Good illustra- tioils of' this class are nitric oxide and the Sm f 1 f arid Eu i-1 + 25 Unlike Classes I and 11 large departures frorri Curie's law are observed. With NO for example the inonient gradually increases with rising tempera- ture towards a limiting value but above 250" the increase with tempera- ture is very small. Spin momenta free but orbital momenta almost entirely quenched. The formula given in I and I1 reduce to 1/4S(S + 1) when the orbital contribution is neglected entirely.It is characteristic of an atom in an X state i.e. when L = 0 ; the best examples of this in paramagnetic compounds are obtained when a sub-shell is half filled-containing three p five d or seven f electrons. Both Pe+++ and Mn+-+ satisfy these conditions (see Table 3). The quenching of the orbital contribution arises from the effect of crystalline fields and is discussed on p. 403. " Quenching of the orbital momenta ' ) means that when an external magnetic field is applied the orbital momenta remain fixed under the influ- ence of the crystalline fields in the lattice or complex ion. This formula is often simply expressed as p = dn(n -+- 2 ) ) where n = number of unpaired electrons because X = n/2. Cbss V. Spin coupling owing to strong covalent bond formation.Class 11. ~ Class 111. (Nitric oxide type.) Class IV. (Iron-group elements.) 2 4 W. Albertson Phys. Review 1935 47 370. 25 P. W. Selwood ref. 4 p. 84. c c 384 QUARTERLY REVIEWS (Covalent-bond type.) The metal ions of many o€ the first transition ele- ments show a large decrease in moment when the ion is co-ordinated with certain ligands ; e.g. the Peii + ion in hexahydrates contains five unpaired electrons but in the complex cyanide K,Fe(CN) the moment falls to 2.35 B.M. indicating one unpaired electron only. Pauling 26 explains this by the hypothesis that certain groups notably the cyanide ion have such a marked capacity for covalent bond formation that unpaired electrons are forced to pair off to provide 3d orbitals for bond formation. This is shown in Fig.1. The implications of this theory are discussed on p. 387 where the alternative Van Vleck explanation is givcn. Ion or compound Fe++(Froe ion) . Fe+++(FrPe ion) Fe(11) (Octahedral 4s4p34dL bonds) Fe(r1) (Octahedral 3d %4p bonds) Pe(111) (Octahedral 4s4p34d2 bonds) Fe (111) (Octahedral Unpaired elections Electronic configuration 3d 4s 4P 4d I * Calciilated on the spin-onIy formula. C2ass VI. Hund’s rules inapplicable. (Heavy-at,oni type.) Magnetic monient * (I3 N ) 4.90 5 92 a (30 Diamag. 5 92 1.73 In dis- cussing the metals of the first transition series it is freely assumed that the arrangement of electrons will be in accord wibh Hund’s rules the fir.;t of which requires that in filling up a sub-shell e . g . the 3d the maximum nuniber of orbitals will be singly filled before electron pairing occurs.Thus Fe k++ has five unpaired electrons as in Fig. 1. However we cannot be sure that Hund’s rules are necessarily applicable to the metals of the second nncl third transition series. It has been suggested 2 7 9 28 that even in the simplc salts of these metals maximum elecfrori pairing may occur became the lowest energy state is not an S one i.e. maximum multiplicity is not obeyed. The 2 6 L. Pauling ‘‘ Nature of the Chemical Boiid ” Cornell Univ. Press N e w York 27 R. B. Janes J. -4mer. Chem. Xoc. 1935 57 471. 28 J. H. Van Vleck ref. 1 p. 312. 2nd Edn. 1945 p. 112 and refs. therein. NYHOLX MAGNETISM AND INORGANIC CHEMISTRY 385 fact that salts of the platinum metals with oxy-anions did not have the high niagnetic monieiits expectxed for free ions had been regarded as some evidence in support of this view.As against this since these metals have a high capacity for covalent bond formation the explanation could be that covalent bonds are formed. In solution the platinum metals are hydrated and it is known Ghat even the [Co(H,O),]+++ ion is diamagnetic showing that even CO(III) can form strong covalent bonds with water molecules. With CO(III) one mnst prepare the complex fluoride K,CoF in order to observe the iiiaxiiiiuiti multiplicit,y of four unpaired electrons. 39 Again the fact K,PtF is diamagnetic 3Q might be regarded as evidence that the Yt+ I f 1 ion disobeys Hiind': rules. Here again t8he bonds might be rcgacrded as covalent as in 1<&iF6 which is :dso diarnngneti~.~~ 32 How- ('ver the sriinll moment o€ K,OsCJ (1.4 B.&L33) suggests that here a brcak- down of Hund's rules has occurred since the expected /& for octahedral covalent binding is 2.83 B.M.[cf. Fe(1v) in Table 31. Nevertheless the evidence for differentiating between Classes V and VI is still meagrc. Class VIT. Small temperature-independent paramagnetism. (Pernian- gmttte type.) Several compounds e.g. KMnO, K,Cr,O, some cuprous sn!t,s and certain cobaltammiiies have a small magnetic moment (ca. 1 B.M.) nfticr allowance has bsen made for the diamagnetism of the rest of the rnole- c.111 e. 'I'hi. ihcceptr d structures for these molecules involve no unpaired e~lt~ctroils snd lienct the origin of the small moment is puzzling. The paraina8giictirsm is tciiipemturc independent. Van Vleck 34 attributes this t o a n unbalanced orbital contribution which may be regarded as arising from the different rates a t which L arid X precess about J .Datar arid Datar 35 have questioned this explanation and suggest that spin coupling may be imperfect. This partial freedom of electron spin is conpidered to account for t'he observed paramagiietism of Cr(v1) in chromates. Against this must be weighed the temperature independence of this paramagnetism. For our purposes it is however uniinportant because it is of practically no diagnostic value in inorganic chemistry. ( d ) Other Knds of Magnetic Behaviour.-Eerrornagnetism is observed in cz lattice of magnctic particles with loose interatomic binding and with parallel ,cpins. 'This is chnrncterisd by a large value of 2 field strength dcpendence of x and the existence of a Curie temperature.The last may be regarded as that temperatlure above which ferromngnetisin changes over LO paramagnetism. The term anti-ferroniagnetisrn is iised to describe a 1 attice similar to ihnt of ferronzagnetie wbstances except that the spins are anti-parallel . Unlike ysramagnctic hubstarices the susceptibility of nn~i-ferro1iiagnttic.s (~ecrcast's as the temperature is lowered. Mil0 ant1 29 J. T. Grey J. Amer. Chem. SOC. 1946 68 605. 3u K. S. Nyholm and A. G. Sharpe J. 1952 3579. 31 FV. Klemm personal comniunicatioii to Kjliolm and Sharp quoted i n ref. 30 ; 32 R. TIoppe L 4 ~ ? g ~ ~ y . C'JL~WZ. 1950 62 33'3. 3 3 W. P. Groves Thesis London 1941. 3 3 Pl~ys. Keuiew 1928 31 687. 35 A\rature 1946 158 518 ; I r i d i u ~ J . Phys. 1949 23 153. cf. ref. 66. Cf. ref'. 5 p. 43.386 QUARTERLY REVIEWS MnSe are examples of compounds which display anti-ferromagnetisni. Kleinin 36 has sumrnarised the various important kinds of magnetism with a table of thzir properties. Applications of Paramagnetic Susceptibilities In inorganic chemistry these measurements are useful for three main purposes. Provided that the number of unpaired eEectrons can be deduced from tlie magnetic monienb it is possible to infer from the former (i) the cakency (ii) the bond type and (iii) the stereochemistry of the metal atom. Uw can be made of the size of the orbital contribution to the moment in the (YAW of first transition elements to obtain information concerning the stereochemistry of the metal atom. (i) Valency Problems.--This is the best known application of magnetic susce1pt)ibility mcasurernents.Having inferred the number of unpaired electrons we can then distinguish different possible valency states provided that they contain different niimbers of unpaired electrons. The decision is specially easy with first trarisition elements because Hund's rules are obeyed for the simple ions. Thus Fei + I - contains 5 unpaired electrons whilst Pe++ has only four (cf. Fig. 1) ; these correspond to moments of 5-9 and 4.9 B.M. respectively. Similarly CU(I) and CU(II) may be distin- guiFhed beeawe the former is diamagnetic ; it is even possiblc to t:stiniate the relative amounts of h ( r ) and CU(II) in ;L compound i f allowance for diiianmg~~ctism is When electroii pairing occurs velericy states may be distinguis-hed if one involves diamagnetism and the other contains vile iiiipirtd electxoii.Thrw conditions obtain for t)hose conipuunds of the first transittion series for which rnaximuni electron pairing occurs e.y . K3Fe(CN)6 and K,Fe(CN), a i d for practically all of the small number of paramagnetic compounds of tlie secolid and third trailsition series e . g . Pd(ri1) iii PdF which contains one and not three unpaired electrons ; 30 both €'d(r~) and Pd(rv) are diamagnetic. A few compouiids fall in between these two extremes e.g. Itu(rv) and certain C'r(xr) complexes which contain two unpaired electrons rather than four or nil. The reason for this will be discussed under bond types. The magnetic criterion of valency is so reliable today tliat unlcss paramagnetism is observed in a compound for which the valcncy state should have an odd number of electrons then the ascribcd vdency must be regarded with some suspicion.The valency of an element when spin coupling between two like atonis omurs is ofteii debatable e.g. as in the Hg,+ ion. Instances may be quoted where spin coupling occurs in the solid state to give in effect it higher vslency state than in solution or in the gas phase. This will be referred to in discussing the compounds of empirical formulae K,Ni(CN), K,Co(CN), and (310,. The certainty with which one can decide the number of unpaired elec- trons in a compound depends upon the class of paramagnetism which it displays. Classes I to IV are usually straightforward provided that the compound is magnetically dilute. If the distance between two paramagnetic 36 2. Elektrochem. 1945 51 14. 37 R. S. Nyholm J .1951 1767. NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 387 atoms is comparable with atomic diameters partial spin coupling becomes feasible. Cupric acetate monohydrate exhibits this effect the moment 38 being only 1-45 B.M. a t 32" as compared with the usual value for CU(II) compounds of about 1-9 B.M. Bleaney and Bowers 39 conclude from para- magnetic resonance studies that metal ion-metal ion interaction occurs. Support for this hypothesis has been obtained by Niekerk and Schoening,40 who have shown by X-rays that the Cu-Cu distance is only 2.64 A little larger than the Cu-Cu distance in metallic copper. The two copper atonis are connected by bridging acetate groups. (ii) Type of Bond in Metal Complexes.-Nagnetic moments can bc nscd t o distinguish between two typcs of binding in metal complexes particulurly those of the first transition series.This division has long been recognised 011 chemical grounds alone ; a freely dissociable CO(II) ammine is known as a '. normalkomplexe " while the more robust CO(III) ammines which do not undergo rcverdible thermal dissociation are examples of " durchdrin- gungskoinplexe " (penetration complexes). In the case of Fe( 111) (see Fig. 1 1 the first clatss (e.g. ferric trisasetylacetone) have moments of 5.9 B.M. indicating five unpaired 3d electrons ; the sceond class (e.g. potacsium ferricysnide) have moments of 2-3-2.4 B.M. a little greater than the spin- only value for one unpaired electron. Two main theories have been advanced t o explain this difference in moment. Pauling 26 suggested that the bonds in the first class are essentially electrostatic (" ionic bonds ") whilst in the latter 3d orbitals are involved in the binding ( 3 d 2 4 s 4 p 3 " covalent bonds ").The two 3d orbitals required for bond formation are supposed to be madc available by electron pairing leaving only one un- paired 3d electron. In the first case the five 3d electrons are unaffected by complex formation. Van Vlcck 41 showed that strong cryafalline forccs * could also bring about electron pairing ; in the ferricyanides for example strong forces were believed to upset the Russell-Saunderrl coupling giving the minimum number of unpaired spins. If the attached ligancis are not held strongly to the Fe atom the spin multiplicity is unaffected. Pauling 26 pointed out that on Van Vleck's theory one might expect spin pairing in K3FeF6 because strong electrical fields should be present.Van Vleclc agreed that in this iiistance at least the Pauling and the molecular-orbital approach are both more satisfactory on empirical grounds. Orgel ,42 after reexamining the relation between the Pauling and the Van Vleck theory liar shown that these are really only two different ways of approaching the problem ; the qualitative similarity is demonstrated and it is pointed out that both theories should in all cases lead to the same conclusion. J n com- paring the effect of a CN- as against a P- ion the polarisability of the ion rather than the intensity of the field at its surface is of paramount import- ance in determining the resulbant crystalline field. This weakens Pauling's 38 B. C. Guha Nature 1945 155 364; Proc.Roy. Soc. 1951 A 206 353. 30 Ibid. 1952 A 214 451. 41 J . Chem. Phys. 1935 3 807. * Electrical fields which arise from the regular arrangement of charged ligands *O Nature 1952 171 36. 4 2 J. 1952 4756. around the paramagnetic atom. 388 QUARTERLY REVIEWS argument against Van Vieck’s picturc. It is srttisfactory to know that there is really no conflict between these two theories because Pauling’s approach is easier for most purposes and has found general acceptance. It will be used widely in the following discussion of stereochemistry. Recent work has done much to remove certain objections (based as much as any- thing on terminology) to Pauling’s original theory. It had been pointed out e.g. by S ~ g d e n ~ ~ that the physical propcrticts of compounds like ferric trisacetylacetone were incompatible with the view thnt the bonds were ‘.ionic ”. . ionic ” was not to be taken too literally and cnvisaged a tppc of ~ovaic~iit binding in the above compound in which four 4s i p 3 bonds rc-somtctl mioiig h i x posittioris in an ionic octnhedral cwinplc>s. wr (mii4dwcY’d to I w too unstable for bond formation Pauling 4 5 rejcctrd Hnggins’s suggestion 46 that 4d orbitals niight be used in conibiriations such as -4s4p33t12 for octa- hedral binding. Huggins’s proposals have been re-examined by ,;everal workers. Taube 47 refers to the two types of complex as “ inner d orbital type ” (3d244s-4p3 bonds) and “ outer d orbital type” (4s4p34d2 bonds). A similar proposal has been put forward by Burstall and Nyholm 4Y in dis- cussing the magnetic behaviour of dipyridyl and diicrtia,ry rirsine coinplexcs of the tranhi tion metals.Overlap calculations provide thcor&ie:xl support for the view that 4d orbitals may be used for o-bond formation in so-called “ ionic ” cornpiexe~.~~ 50 The fact that 4d orbitals are well elongntecl means that good overlap occurs at relatively greater dist8nncc1s from the iiiefal atom than occurs when 3d orbitals are u x d . Hence ihc electron cloiid of the bonding electron pair lies further over towmds thtl l i g i ~ ~ ~ d i.c. t>he boxid is more polar. Thus fhc difference bctwecln the terms ‘ * ionic ” or ‘. liighcr level covalent bonds ” is not as great as it seems at first sight. However the possibility of exchange owing to the size of the overlap integral is explic4 in the latter but iiot in the former.The above ideas enable one to uiiderstand why some atoms are inore effective than others in causing clectron pairing. The more electronegntive atoms (F 0) favour ‘‘ higher level covalent binding ” because they tend to concentrate the electron pair closer to the ligand resulting in the use of 4d orbitals. On the other hand the groups of low electronegativity (e.q. P As) utilise more readily the 3d orbitals and hence give rise to “ lower level covalent bonds ”. In the binding of groups of low electronegativity such as the CN NO, and AsR ligmds it is probable that double (n) binding makes a significant contribution to the strength of the bond. These z bonds could increase the bond strength for three reasons they increase the number of bonds between the metal and ligand; they provide a mechanism whereby the improbably high negative charge on the metal atom may be de-localised ; finally partly as an outcome of the latter they Laltler Pituhg 41 cmphac,iscd that ishe tI-i-m Bccaiise t J h y 43 J .1943 325. 4 4 J . 1948 1641. 45 Op. cit. p. 115. 4fi J . Chem. Phys. 1937 5 527. 17 Chcm. Reviews 1952 50 69. J. 1952 3.570. 40 L. E. Orgel quoted by R. li7ilkins Nature 1951 167 43-1. D. P Craig A. Maccoll R. S. Nyholm b. IF,. Orgel and Le F. Sutton. J. in the press, NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 389 provide a method for strengthening the i~ bond. Detailed references to work on this subject are given by Burstall and Nyh~lm.~* When it is not possible to free 3d orbitals for bond formation by pairing 3d electrons Pauling 45 suggests that “ promotion ” of electrons to higher levels above the bonding orbitals occurs.Such a promotion is postulated in octahedral covalent CO(II) complexes €or which the binding is apparently 3d24s4p3 ; this is shown in Fig. 2 . It can be seen that no such promotion FIG. 2 Elcctroibic coiLJiguration of cobaltous complP.1 es Type of compound Cod t (Free ion) CO(II) (Octahedral 3d%s4p3 bonds) Co(11) (Square 3d494p2 bonds) Elcctronic arraugemcut 3d 4s 4P 5s Unpaired electrons Magnetic moment * 3.88 1-73 1-73 * Calculated on the spin-only formula. is necessary for square complexes. In the octahedral complex the pro- moted etcctron is assumed to occupy an exposcd 6s orbital and hence should be lost easily by oxidation to yield a diamagnetic Co(1r1) complex. This corollary has proved both a sfrengtJh and a weakness of the theory.The chemical behaviour of some complexes e.g. the [CO(NO,),]~- ion supports the assumption but in certain other cases where promotion has been assumed the evidence is unconvincing e.g. CU(II) (see p. 392). (iii) Stereochemistry.-The stereochemistry of an atom follows logically from a knowledge of the bond type and the co-ordination number. The relationship between stereochemistry and different combinations of atomic orbitals is discussed elsewhere ; 22 5l Table 2 lists only the common shapes TABLE 2. Some common stereochem ical arrangements and orbital combinations Linear Angular Square planar Tetrahedral Triangular plane Pyramidal Bip yramid Square pyramid dap2 sp3 d3s nsnp 3nd (n - l)dnsnp3 I 6 ~ Octahedral ~ d2sp3 61 G. E. Kimball J Chem. Phys. 1940 8 194. 390 QUARTERLY REVIEWS with which we are coiicerned in this article.Bor the effect of lone-pair contributions to stereochemistry see Lennard-Jones and Pople.52 Pauling's theory of the stereochemistry of the first transition series (also applicable with limitations to the second and third) may be summarised as follows (i) If the magnetic moment of a complex is essentially the same as that of the free ion * (i.e. no spin coupling occurs) the bonds do not involve 3d orbitals ; however the moment is not diagnostic of 4s4p3 as against 4s4p24d bonds. (ii) Spin coupling with a reduction of either two or four in the number of unpaired electrons is caused by the forinatlion of strong covalent bonds using one or more 3d orbitals ; this necessitates the transfer of unpRired electron (5) originally occupying orbitals now used for bond formation usudly these displaced electron(s) pair off with other 3d electrons.(iii) If no vacant 3d orbital(s) are available to accommodat'c clectron(s) displaced from 3d orbitals required for bond fornzation thew electrons may be promoted to vacant orbitals above those used for bond formation ; such promoted electrons will occupy the orbitals of lowest energy. (iv) The stereochemistry of a complex ion is decided by the orbitals used for bond forniation. Table 3 summarises calculated and observed moments for various types of complex of the first transition series. It may also be used to obtain calculated inoinents for the third and second transition wries. Survey of the Periodic Table Only the iiiore recent developments in valency and sicreochcinistry are discussed the first transition series receiving moat attention.Group 1.-The Alkali Netah. I n their chemical compounds them are invariably univalent and hence diamagnetic ; certain higher oxides e q . KO2 are par~rnagnetic,~~ 54 but the unpaired electron is 011 the 0,- ion. Coppr fliher und Gold. Each of these metals has one s electron inow than a coinplet'ed d sub-shell and hence univalent compounds are diainag- netic (see Table 3 and Fig. 3 ) . The univalent compounds are either two- or four-covalent using linear sp or tetrahedral sp3 bonds. A gradual change in the preferrctd co-ordinatioii number from four t o two occurs in passing from copper through silver to gold. The bivalent compounds contain a1 1 odd number of electrons and hence are paramagnetic.Bot<h CU(II) and Ag(r1) compounds contain one unpaired electron 559 56 but no Au(r1) corn- poiinds are known. Thc apparently bivalent gold complex with the empirical formula (C,H,) ,S,AuBr (benzyl sulphicle-dibromogold) has been 52 PIOC. ROY. JYOC. 1931 A 210 190. 53 E. W. Neumann J . Chem. Phys. 1934 2 31. 54 W. Klemm and H. Sodomann Z . anorg. Chem. 1935 225 273. 5 5 A. A. Noyes K. S. Pitzer and C. L. Dunn J . Amer. Chem. SOC. 1036 57 1229. 6 6 S . Sugden J . 1932 161. * Provided that the free ion contains more than three 3d electrons. This proviso is necessary because magnetic moments do not distinguish between 4s4p34d2 and 3d2494p3 octatiedrel binding when only three unpaired electrons are present e.g. as in Cr(n1) complexes. T~BLE 3 Magnehc propertzes of jirst transxtxon serzes (p in B M ) J o $I\ J ( J 1) No electron pairing I d L ( 5 1) + 4S(S + 1) t'4s(z,+ 1) Electron pairing 2 Covalent bond orbitals I 1 63 4 49 2 8 3 2 8 - 3 1 I Specti scop1c noritial state so 5ree ion 01 if usiiig highw bonding orbitals (4s4p3 4s4p34dz) 0 s Calc from Clement aiid valenq Square Octahedral 3d4s4pz 3da4s4pa S O I * I 0 I O I Didmag 0 Diamag 0 ?/2 I 155 I 3 01 1 7 3 17-1 8 1 7 3 I 1 7 - l S 1 7 3 6 1 2 3 2 8 3 I 2 8 - 3 1 2 83 4F3 3 3 3 2 .3/2 I 0 7 0 I 3 11 I 3 8 7 1385-39 3 87 387 1 3 8 5 - 3 9 "Do 2 2 4 90 2 8 3 1 2 9 - 3 1 0 5 9.2 1 5 9 2 1585-595 1 7 3 I 21-f!4 3 87 "D'I 2 2 5 :lo I -190 I 50-56 O I =lamug 2 83 3 3/1 5 21 I 3 8 7 I 4 3 - 5 3 1 7 3 1 7 3 1 9 - 2 4 I B M Diamag 1 4 6 D 9 4 49 2 8 3 2 9 - 7 4 ~ 5/2 3 SlJ 3 01 1 7 3 19-21 3 2 (NP),* cun cu' Zll" 0 0 0 I Diamag 0 * Moments corresponding to valencj states shown in parentheses have not yet been observed t If all three promoted electrons are unpaired 392 QUARTERLY REVIEWS FIG.3 Electronic cot&umtion and magnetic moments of copper complexes 1 Ion or compound I Cu+ (Free ion) . Cu(r) (Linear 4s4p bonds) . . . . CU(I) (Tetrahedral 4s4p3 bonds) . . 1 Cu++ (Free ion) CU(II) (Squaro 3d4r4p2 bonds) . CU(II) (Squaro 4s4p24d bonds) . CU(II) (Octahedral 4s4p34d2 bonds) . Cu(i11) (Square 3d4s4p2 bond5) . Cu(r11) (Octahedral 4s4p34rZ2 honds) . Xlcctronic coilfiguration 3d 49 4P 4d Unpsirec election! 0 0 0 1 1 1 1 0 L Dinmag Diamab hama) 1-73 1.53 1.53 1-78 Dianiag 2 53 * CalcuIated on tho spin-only formula. shown by Brain Gibson Jarvis Phillips Powell and Tyabji 5' to contain equal numbers of Au(I) and AU(III) atoms hence the observed diamagnetism.The location of the unpaired electron in CU(II) and Ag(I1) compounds has been the subject of much speculation. Since all CU(II) complexes whose structures have been determined are square planar PauIing 58 assumed that the bonds are 3d4s4p2 (see Fig. 3 ) one unpaired electron being promoted to a 4p orbital but there are serious objectioiis to this hypothesis. First promotion of the electron to a 4p orbital should result in facile oxidation of square Cu(rr) complexes to the tervalent state-which is not observed in practice. Also theoretical work 49r 5O leads to the conclusion that fairly electronegative groups like H,O and NH [which do give square CU(II) complexes] are more likely to use 4d than 3d bond orbitals.In the case of Ni(n) for example where the diamagnetism is diagnostic of the square arrangement groups of low electronegativity are required to form 3d4s4p2 square bonds. Huggins's 46 suggestion that the bonding orbitals are 4s4p24d without electron promotion has been re-examined by Ray and 1952 3686 58 L. Pauling op. cit. p . 121. NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 393 who obtained the susceptibilities of a large number of Cu(I1) complexes at various temperatures and calculated /c on the Curie-Weiss law. The nionieiits fell roughly into two classes some between 1.8 and 1.9 B.M. and others between 1.9 and 2.2 B.M. but the division was by no means sharp. The first group are believed to involve 3d4s4p2 bonds with electron promotion aiid the second are considered to be either 4s4p3 tetra- hedral or 4s4p24d square planar.The smaller rnoinent of compounds in the first group is attributed to the greater quenching effect of the crystalline field upon the exposed electron ; however there is no evidence to support the claim that elcctroii proinotion really occurs. Finally the existence of ochhcdral Cu(1r) co~npl~xes e.g. LCn(N€-I,),]X,,6° is more easily explained by t,he formation of 4s4p34d2 bonds without electron promotion than by t,he improbable 3d %4p3 combinntion rrqiiiring the promotion of three clectrons. The former combination follows logicaily from a square 4s1p24d complex by coiripleting the octahedron. The well-known Au( 111) complexes are square planar and the diamagnetism is consistent with the use of 5d6s6p2 bonds.61 Both CU(III) and Ag(rI1) are also known.Malatesta 6 2 t cf. 63 has described a CU(III) periodate of the formula KnH(7-fi,Cu( 10,) z,nH20 mliich is diamagnetic. Also Scholder and Voelskow 64 have prepared the compound Bs(CuO,),,H,O ; as it is diamag- netic itl is concluded that it is probably polymeric ilwo oxygm atoms form- ing a bridge bet ween successive square-co-ordinated Cu(rr1) atoms. The corresponding potassium salt KCnO, has been described by Wahl and Klemm ; 6 5 this also is diamagnetic. In these compounds the electro- iiegaiivity of the oxygm atoins appears to haw becii reduced sufficiently by bridge formation to permit the CU(III) atom to make use of' 3d orbitals. Quite a different result is obtained when CU(II) fluoride is oxidiaed with fluorine. Hoppe 32 cf- 6 6 has prepared K,CuP as a green solid but unlike the oxy-anions above this substance is paramagnetic with a moment corres- pmding to two unpaired electrons.This leads to the conclusion that the bonds are octahedral 4s4p34d2 the two unpaired electrons being in the 3cl shell. Silver also forms compounds of the type K,H~7-n,Ag111(I0,)2 which are diamagnetic.62 6 3 Other diamagnetic and hence presumably square A~(III) complexes have been prepared by using organic ligand~.~' These elements are invariably bivaient aiid hence diamagnetic in their bimplc salts and complex compounds. The paramsgnctism of certain higher oxides arises from the presence of unpaired electrons in the anion. The sub-group elements (Zn Cd and Hg) are diamagnetic in the common bivalent state. The appareiit univalency 6o G. Peyronel Gnzzetta 1941 71 363.The tervalent state of these elements ir of considerable interest. Group II.-Thc Alkaline-earth Metals. 59 J . I n d i a n Chern. SOC. 1948 25 473. 61 R. S. Nyholrn N a t u r e 1951 168 705. 6 2 L. Malatesta Gazzetta 1941 71 467 580. F 3 A. Malaprade Compt. rend. 1940 210 504. b c W. Klemrn and E. HUBS ibid. 1949 258 221. ST P. Ray and K. Chakravarty J . I n d i a n Chem. SOG. 1944 21 47 ; !!)FiO 27 619 j 2. anorg. Chem. 1951 266 256. 6 5 Ibid. 1952 270 69. Nature 1943 151 643. 394 QUARTERLY RE VIEWS of mercury arises from the presence of the [-IIgl1- -Hglr-]++ cation which is diamagnetjc. This kind of metal-metal bond is probably more common than is usually supposed.68 Group 111.-In the A sub-group (B Al Sc Y and the lanthanons) t'ervalency is the rule and in this valency state diamagnetism is observed €or all compounds of all elements except the l~~ntlianons.Xcments of the latter have been discussed. Claims for univaleiicy particularly of A1691 7% 5 1 have been advanced but magnetic data are of no valuc because both uni- :md ter-valent aluminium are diamagnetic. I n the B sub-group (Ua In TI) uni- and ter-valcncy are well known and both are diamagnetic. Conipouiids in which these elements appear to be bivalent c q . Tl,H 72 and TlC12,T3 are diamagnetic and hence contain both uni- and ter-valent tliallium . Group IV.-Compounds of the main group elements (C Si GP 811 and Pb) in the bi- and the quadri-valcnt state are invariably tiiaiiiagnctic. However paramagnetism is displayed by free radicals of thc hype Yh,C* which contain one unpaired e l e ~ t r o n .' ~ In the sub-group (Ti Zr Hf) tlis really interesting behaviour of the first transition series begins becauso titanium is the first transition element to show paramagnetism. In the usual quadrivaleiit state Ti is diamagnetic but Ti(rI1) and Ti( 11) contlain one and two unpaired electrons r e ~ p e c t i v e l y . ~ ~ ~ 7 G This confirms the valency but tells nothing about the stereocheniistry owing to the number of vacant 3d orbitals. Much less is known of the magnetochemistry of hafnium and zirconium ; it has been pointed out that they fall into the most neglected group in the Periodic Table.77 Group V.-Thc main-group elements (N P Ad Sb arid Bi) arc ciiamng- netic in the usual tcr- and quinque-valent sttates and fcw parani-ttgnetic compounds of these clenients are ~~~~~~~11.Nitropi liowevw forms some paramagnetic compounds notably the oxirlcs NO and NO ; t h i h moment of the latter shows unusual temperature depmdence (see Clws 11). Special interest attaches to the magnetic properties of coniplexes contain- ing the NO group. When present as the NO+ cation a$ in [NO]+[NO,]- it is diamagnetic. It also behaves rzs the NO- group e.g. in certain CO(III) complexes. Two types of nitrosopeiit,,2m1uinoco~al~( 111) salts are known,'8? 79 a stable red form having tthe firmula [Co(NH,),NO][NO,], which is diamagnetic and an unstable black form with the formula 68 1,. Pauling Chem. Eng. News 1947 25 2970. 68 W. Klemm and E. Voss 2. anorg. Chem. 1943 251 233. 70 A. S. Russell V. E. Martin and C . N. Cochran J . Amcr. Cheuti. ;>'oc. 1951 '73 7 1 J.P. ILZcGeer J . Chewz. Educ. 1952 29 534. 72A. G. Sharpe J. 1952 2165. 7 3 D. J. Meier and C. S. Garner J. Chem. Phys. 1950 18 237. 7 4 N. V. Sidgwick " The Chemical Elements and Their Compounds " Oxford Univ. 7 5 W. Klemm and L. Grimm 2. anorg. Chent. 1942 249 108. 7 6 C. Starr F. Bitter and A. R. Kaufmann Phys. Review 1940 58 977. 77 P. W. Selwood ref. 4 p. 103. 79 J. L. Milward W. Wardlaw and M'. Way. J. 1038 233. 1466. Press 1950 London p. 534. S. P. Ghosh and P. Ray J . Indian C'hern. Soc. 1943 20 400. NYHoLM MAGNETISM AND IN'ORGANIC CHEMPSTRY 396 [Co(NH,),NO]Cl,. The latter is paramagnetic with a moment a little less than that required for one unpaired electron. In the diamagnetic com- pound the NO- group apparently uses up one of the valencies of the (tervalent) cobalt.If in the second compound the NO were attached by a co-ordinate link to a CO(II) atom four unpaired electrons are expected if the bonds to the Co(r~) atom are " ionic " or two if they are covalent 3d24s4p3 one electron being on the NO in each case. As pointed out by Mellor and Craig,so CO(II) is unlikely to show electron pairing when attached to five NH groups. Ray and Ghosh 78 conclude that the Co atom is ter- vtileiit the compound being a mixture oE two forms-one diamagnetic with the NO in a singlet state and the other paramagnetic with the NO in a triplet state (containing two unpaired electrons). The solution to this problem will now rrrost probably be obtained by using spectroscopic infra- red techniques. The diamagnetism of Sb in K,SbCI is also puzzling since Sb(1v) should contain an odd number of electrons.Pauling considers that equal numbers of diamagnetic [SbIIW1,]- - - and [SbVCI,]- ions are present. The compound is isornorphous with K2PtCl, and Jensen 82 has discussed alternative methods of spin pairing more consistent with this (s bservation. In the sub-group (V Nb and Ta) vanadium has valencies of 2 3 4 and 6. Magizetic irioinents show that in no case does spin pairing occur the spin-only value beiiig observed except in magnetically concentrated coinpoundc.30 The moment is very u:jefd here in the assignment of valenay. Nb and Ta have received iiiiich less attention ; 77 as usual moments arc rmaller than filr V. Thus the nioinents of K,VF', VF, mid TaF :ire respectively 2-79 2.55 and 1-4 B.iK3O Group VI.-1n the mrtiii group (0 S Se and Te) both 0 and S2 are unusual in being paramagnetic.83 The presence of two unpaired electrons is most easily understood in ternis of the molecular orbital theory of valency.** In practically all other compounds (e.g. in valency states of 2 4 and 6) these elements are diamagnetic. The sub-group elements (Cr $10 and W) are interesting owing to variable valency and paramag- netism. Diamagnetism is expected for Cr(v1) but weak temperature- independent paramagnet;jsm is observed. 35 Cr(v) occurs in compounds such as KCrQF, the moment of which indicates one unpaired electron.3* All Cr(m) compounds contain three unpaired electrons but the moment is not diagnostic of 3d 24~4p3 as against 4s4p34d octahedral binding. However the slow rate of exchange of H,O with the [&(H,O),]+++ cation suggests that the former is the more common.85 Cr(I1) in simple salts and in rriost complex compounds has a moment close to the spin-only value for four unpaired electrons ; 86 with the -CN group 8G or with d i ~ y r i d y l ~ ~ 6 o J.P?*oc. Roy. Soc. N.X.W. 1044 78 26. 81 IT. A. Jensen 2. anorg. Chem. 1937 232 193. 8 2 Ibicl. 1944 252 317. 8J C. A. Coulson op. c i t . p. 100. 8 5 R. A. Plane and H. Taube J . Phys. Chem. 1952 56 33. 88 D. N. Humo and H. W. Stone J . Amer. Chem. Soc. 1941 63 1200. 8 3 A. B. Scott J . Amel.. Chem. Xoc. 1949 71 3145. 396 QUARTERLY REVIEWS six-covalent complexes are fcwmed coiitainiiig two uiipsired electrons oiil y and in this case the binding is 3d244s4p3. Unlike either of these compounds chromous acetate has a nionient of loss than 1 R.M. ; s 7 even this may arise from impurity.King and Garner s7 suggest that this indicates 3d34s tetrahedral bonds but no X-ray studies in support of this hypothesis are available. The explanation could be the same as for cupric acetate mono- hydrate (see p. 387). An important elevelopinent is the recent announce- ment by Hein and Herzog 88 of the isolation of a Cr(1) complex having the formula [Cr(dipyridyl),]ClO, obtained by reduction of the bivalent cornplox [Cr(dipyridyl),] [ClO,],. The univalency ascribed to the chroiiiiuin atom is supported by the magnetic moment of 2-1 B.M. iiidicat3iiig only oiie unpaired electron. The unexpected feature of the compound is the high co-ordination number of six proposed €or a univalent metal. The iiioiiient indicates covalent 3d24s4p3 bonds ; further work on this compound should prove of interest.MO(III) has the expected three unpaired electrons in it5 complex compounds but in other valency states both Mo and T/c' form strong covalent bonds aiid arc generally dianiagraetic or contain one un- paired electron according as t'he valency is even or odd. The dat'a are surnmarised by Selwood. p9 Group VII.-The compounds of the halogens are parainagnetic oiily when the valency is even and the compound is monomeric. Chlorine dioxide C102 for example is parainagnetic with one unpaired electron. Chlorine trioxide however in the liquid state is diamagnetic ; this is consistent with the dimeric formula C1,0 attributed to this compound. This iliffercnce iri magnetic behaviour of C10 and Cl,O illustrates how the fornmtion of a dimer can eliminate an unpaired electron.The problem of assigning a formal valency to certain elements is well illustrated by the beliaviour of CI206. The physical properties of this compound leave no doubt that it should be formulated as 03Cl-C10, the chloriiie having a valency of seven ; in the gas phase however the substance is monomeric and here the valency is six.go I n the gas phase the compound should be paramagnetic [cf. K,Co(CN) solid and in solution]. The sub-group elements (Mn T'c and Ite) have valencics from onc to seven. No unpaired electrons are expected in Mn(vrr) but a sinall temp~ratiire-independent paramagnetisin is observed for the same reason RS in chromatles. MT~(vI) has the expected moment corresponding 00 one unpaired el c c t r o ~ i . ~ ~ Mn(rv) oceurd in certain cOlnq?l?X compounds and like the ik;o-electronic Cr(m) contains three unpaired electrons.A good example is K2MnF629 ; Mn(1v) also occurs with three unpaired electrons in complex Iicteropolyinc~yb~~~te~.92 Depending upon the attached groups Nn( 111) coinporincls contain either four or two unpaired electrons ; manganese trisacetylncetoiie illustrates the former and K,&fn(CN) the latter.43 91 M ~ ( I I ) shows eltctrcni pttii*iiig in tlit. c~oniplc~s 87 W. R. King ancl (2. S. Garner J . Chem. Phys. 1050 18 689. sB 2. anorg. Chem. 1952 267 337. 9 1 N. Goldenberg Truns. Furaduy SOC. 1940 36 847. v 2 P. Ray A. Bhadiiri ancl €3. Sarina J . I d a i i C h e m . A'soc. 1918 25 51. Op. ciC. p. 106. N. W. T@w J. Amer. Citeni. Soc. 1926 48 865 ; 1'. Mi. Selmood op. c i t . 11. 123. NYHOLM MAGNETISM AND INORGANIC CHENISTRY 397 cyanide but practically all other compounds contain five unpaired electrons.Mn(1) has been reported in the conzplex cyanide ; 91 the weak paraniagnetism of this is probably due to impurity. Treadwell and Raths g2a have recently prepared pure K,Mn(CN),. Little has been published concerning the magnetochemistry of technetium compounds ; many of the results available for rhenium are anomalous and require further investigati0n.~3 94 Group VII1.-Recent developments in this group are best surveyed wit'h the magnetochemistry of Fe Co and Ni as the background. The behaviour of Pe(11) and Fe(m) has been referred to ; electron pairing occurs rcadily with the CN group and other groups of low electronegati~ity.~~ The mag- netic properties of t'he haematin compounds of iron have been reviewed by H t ~ r t r e e .~ ~ Magnetochemical confirniation of the existence of Fe(v1) in the [FeO,]- - ion has been obtained by H. J. Hrostowski arid A. B. ScottYgs who had the difficult task of disentangling ferro- and para-magnetism in the same specimen because the K,FeO was only 97% pure. The suscepti- bility was measured a t different field strengths and then extrapolated to zero field strength thereby eliminating the ferro-magnetic contribution. As expected the [FeO,] - ion contains two unpaired electrons. For cobalt the valency states of 2 3 and 4 may be confirnied by iiiagnetic nieasure- iiicnts. CO(II) forms both tetrahedral (4s4p3) and octahedral (4s4p34d2) coniplexcs in which no electron pairing occurs and square planar ( 3d4s4p2) and octahedra! ( 3d24.s4p3) complexes which contain only one unpaired chlwtron ; hence magnetic moments enable one to distinguish readily between square arid tetrahedral four-covalent coinple~es.~7~ 98 99 The moment may also be used to decide the type of bond in octahedral corn- plexes.The puzzling problem of the complex cyanide of CO(II) has recently been cleared up. The solid which is diamagnetic is not K,Co(CN) but K,Co(CN),; apparently spin pairing takes place in the solid state. In aqueous solution however the compound is parainagnetic with one un- paired Adanison loo lol considers that the ionic species in solution is [C'o(CN),] - and not the monoaquoprntacyanide. Co(1r1) is diamagnetic in practically all of its coinplexes but has the expected number of unpaired electrons for a Cofff ion in K,COP,.29 The magnetic moment of tlhe CO(IV) complex K,CoF has n u t yet been published.Calvin Eailes arid Wilmarth have studied the magnetic properties of the complexes formed when certain square CO(II) compounds absorb oxygen. They find it to be diamagnetic. lol sra W. D. Treadwdl and W. E. Raths Helv. Chim. Acta 1952 35 2259. 9 3 W. Klemm and H. Schuih 2. unorg. Chem. 1931 220 193. 9 * W. Klemm and G. Prischmukh ihid. 1937 230 220. 9 5 Ann. Reports 1016 43 287. 9 6 J. Chem. Phys. 1950 18 105. 97 D. P. Mellor aid 1). P. Craig J. I'roc. Roy. h'oc. N.S. TT'. 1940 '74 495. s6 L. Pauling op. cit. pp. 97 118. OY P. Ray and S. P. Ghosh J. I n d i m Chem. h'oc. 1943 20 323. loo A. W. Adanison J . Anzer. C!hem. Soc. 1951 73 5710. l01 A. W. Adamson J. P. JTelker and 31. Volpe ibid. 1950 72 4030 lo2 Ibid.1946 68 2254. 358 QUARTERLY BEV3EWS The square compounds contain one unpaired electron but become dia- magnetic on complete oxygenation. If the product contained one mole- cule of oxygen per Co atom tho product would still be paramagnetic because the total number of electrons would still be odd. The diamagnetism is most easily accounted for by the assuinption that an -0-0- bridge between two CO(III) atoms is formed. Paramagnetism 103 corresponding to one unpaired electron is observed in the polynuclear Co complexes of the X,. It has frequently been assumed that the Co atom is quadrivalent but the unpaired electron is probably associated with the oxygen molecule both cobalt atoms being tervnlent . FIG. 4 Electronic configuration and magnetic moments of nickel complexes Zlectronic configuration 3d 48 4P Type of compound N i t + (Free ion) Ni(11) (TetrahedrJ 4s4p bonds) NI(II) (Octahedral 4s4p34d bonds) Ni(11) (Squaw 3d4s4p2 bondr) Ni(11) (Octdhudral 3d24s4p3 bonds) N~(III) (Octahedral 3d24s4p3 bonds) NI(IV) (Octahodral 3d24s4p3 bonds) Ni(0) (Tetrahedral 4s 2 p J bonds) Unpa ii ec electron IIagnetic moment B.N." 9 83 2 33 2 83 L)lalrlag I hamag.1 7 3 Diarnag. Diamag. * Calcdated on t h o spin-only formula. IVFagnetic measurements have proved of considerable value in confiriniiig the newer valency states of nickel. The relationship between the valency stereocheniistry and magnetic behaviour of nickel in its complexes has been reviewed elsewhere lo4 and need only be summarised here. I n short nickel has well-defined valency states of 0 2 3 and 4 and a less certain univalent state.The electronic configurations are given in Fig. 4. Io3 N. V. Sidgwick op. cit. p. 1421. lo4 R. S . Nyholm Chem. Reviews 1963 51 in hhe press. N m O L M MAGNETISM AND INORGANIC CHEMISTRY 399 In the bivalent state the magnetic moment enables one to distinguish between tetrahedral * four-covalent complexes and squaye four-covalent complexes the former being para- arid the latter dia-magnetic. Two kinds of octahedrd Ni(rr) complex are known and the magnetism is here diagnostic. of bond type. The usual type e.g. [Ni(NH,),]Cl, COlltc?Jiils two unpaired 3d electrons and involves 4s4p34d2 bonds ; the second type e.g. [Ni(diar- sine)] [ClO,] is diamagnetic which is interpreted 48 to indicate 3d 24s4p3 binding with the promotion of two 3d electrons to a 5s orbital in which the electrons are paired.Support for this hypothesis is provided by the fact that the complex may be oxidised under suitable condjtions-as should follow if promotion does take place. Octahedral and five-covalent Ni(m) complexes have been described e.g. [NiCl,(diarsine),]CI lo5 and [NiBr, 2 (C,H,),P]O.lOG These contain one unpaired electron rather than the three expected by Hund's rules for a Nit+-+ ion. The location of this unpaired electron and the bonding orbitals used in the five-covalent tervalent complex have been discussed e1se~here.l~~~ loG I n the octahedral Ni(m) complexes the unpaired electron appears to be promoted to a 5s orbital; removal of this gives a diamagnetic Ni( rv) complex.1°5 Since diamagnetism is also characteristic of the square planar Ni(rr) arrangement from which most Ni(1rr) and Ni(1v) complexes are prepared it is often doubtful whether the nickel atom or the ligand has undergone oxidation in some reactions.Ni(1) is believed to exist in the cotnplex cyanide of empirical formula K,Ni(CN)3. This compound however is diamagnetic l o 7 even in aqueous solution.108 The diamagnetism is unexpected since Ni(1) should contain one unpaired electron ; Mellor and Craig lo* suggest that a Ni-Ni bond is involved between two Ni(rr) atoms the anion being dimeric i.e. K,[(CN),Ni-Ni(C"),]. Until paramagnetism has been observed the univalency must remain doubtful. Recently Nast and Pfab loSa have shown that the anion in K,Ni(CN) is dimeric two square-co-ordinated nickel atoms being bridged by two CN groups. The carbon atoms of these bridging groups are each three-covalent similar to those in the bridging CO groups in Fe,(CO),.Nast and Rooa lo8& have shown that K,Ni(CN),CO is also diamagnetic. They conclude that the anion is dimeric and suggest that in both [Ni,(CN)6]4- and [Ni(CN),(CO)]24- the diamagnetism arises from some kind of metal- metal interaction as in the diamagnetic carbonyl Co,(CO),. The remaining elements of Group VIII never contain more than two unpaired electrons in their complex componnds ; lo9 as a rule they are diamagnetic if the number of electrons is even and contain one unpaired electron only if the total number of electrons is odd. Compounds of Ru and lo6R. S. Nyholm J. 1950 2061 ; 1951 2602. l06 K. A. Jensen and B. Nygaard Acta Chem. ~9cu>id. 1949 3 474. lo' L.Szego and P. Ostinelli Gannetta 1830 60 946. lo8 D. P. Mellor and D. P. Craig J . Proc. Roy. Xoc. X.S.W. 1942 '76 281. loaa R. Nast and W. Pfab Naturwiss. 1962 89 300. loSb R. Nast and H. ROOS 2. anorg. Chem. 1953 272 242. 109 D. P. Mellor J . Proc. Roy. Soc. 1943 77 145. * Evidence in favour of the tetrahedral rather than a 4s4p24d square arrangornent is discussed eIsewhere.104 D D 400 QUARTERLY REVIEWS 0 s in valency states 2 6 and 8 are diamagnetic. RU(III) has the usual paramagnetism for one unpaired electron as in covalent 3d24s4p3 Fe( 111) complexes. No RU(III) compound containing five unpaired electrons is known. Quadrivalent Ru and 0 s have unusual magnetic properties ; RU(IV) in K2RuC1 log has the expected two 4d unpaired electrons but the hydroxy-pentachloromthenate( IV) of empirical formula K2[RuC15( OH)] is diamagnetic.log It has been shown by Mathieson Mellor and Stephen- son I1O that this compound is really tbe monohydrate of the dimer [(C15Ru-O-RuC15)2]4- some kind of spin coupling taking place.OS(IV) is similar to RU(IV) except that moments are generally smaller ; thus K,OsCl has a moment of 1-4 B.M.109 as conipar-ed wit,h 2.8 B.M. for the Ru compound. A temperature-dependence dndy of the susceptibility is necessary in the former because the simple Curie law is almost certainly not applicable in this case the moment having been calculated on the (probably erroneous) assuniption that the Curie law holds good. All known Rh(m) and Ir(1rr) compounds are diamagnetic but Rh(1v) and Ir(1v) complexes contain the expected unpaired electron.3* log The anomalous diamag- netism log of Rh(rc) and Ir(u) compounds is being investigated ; ll1 it now seems probable that in these compounds Rh(1) and Rh(m) atoms are present in equal numbers.In the usual bi- and quadri-valent states both Pd and Pt are diamagnetic. The tervalent state is less common than is often sup- posed ; the so-called PtCl, for example contains Pt(I1) and Pt(1v) atoms in equal numbers.ll2 However Pd(m) does occur in the fluoride PdF, as the moment indicates the required unpaired electron. 30 The Trans-uranic Elements.-Magnetic properties have helped in the assignment of the electronic configurations of these elements. There has been some doubt as to whether the elements following actinium * (atomic number 89; ls2 . . . 4f1*5s25p65d106s26p66d17s2) are built up by adding electrons to a 5f shell as with the rare earths or to a 6d shell as for the transition elements.Unlike previous applications in which the ground- state configuration is assumed and valency inferred from the moment here the elements are compared in suitable valency states and the way in which ,u varies with atomic number is examined. Howland and Calvin 113 exam- ined the cations of elements 93-95 in aqueous solution in various valency states to obtain ions containing from one to six electrons in excess of the radon (inert gas) core. They found that the susceptibilities changed along the sequence in the same way as do the susceptibilities of the corresponding rare-earth compounds. Using the mean of figures given by Lister,l14 together with Howland and Calvin's figure for Am(m) in aqueous solut'ion one obtains the relationship shown in Pig.5. The similarity with the rare earths is striking and must be taken as confirming the conclusion from spectroscopic work that the 5f shell is being filled. If these elements I1O Acta Cryst. 1952 5 155. 111 B. Figgis and R. S. Nyholm unpublished experiments. 113 J. Chem. Phys. 1950 18 239. * Th (go) Pa (91) U (92) Np (93) Pu (94) Am (95) Cm (96) Bk (97) Cf (98). Y . K. Syrkin and V. I. Belova J . phys. Chern. U.S.S.R. 1949 23 664. 114 Quart. Reviews 1950 4 20. NYBOLM MAGNETISM AND INORGANIC CHEMISTRY 401 behaved like those of the first transition series (see Table 3) the moments should rise to a maximum withjve unpaired electrons and then fall steadily again ; on the other hand similarity with the other two transition series would most likely lead t o moments corresponding to one or nil unpaired electrons according as the number of electrons is odd or even.I n all cases the moments are less than the calculated values with which the rare earths agree cIosely. This deficiency is attributed by Howland and Calvin to partial quenching of the orbital contribution by the surrounding electrical field the 5f electrons being regarded as less effectively screened than in the rare earths ; alternatively Russell-Saunders coupling is regarded as not fully applicable to these elements. For many U(m) compounds the mag- netic moment is sufficiently close to the spin-only value for two unpaired electrons to suggest a 6d2 rather than a 5f2 arrangement. Lister points out however that most of these spin-only values are restricted to magnetically concentrated compounds like UO, for the moments of U ( N ) sulphate and oxalate agree better with the 5f2 arrangement.However Dawson has measured the moment of UF4 (in solid solution in ThF,) and concludes that the U(IV) ion has the 6d2 configuration ; magnetic dilution is obtained by using a solid solution. A similar moment for PU(VI) in sodium plutonyl acetate leads Dawson to suggest that PU(VI) also has a 6d2 configuration. He concludes that for only two unpaired electrons the 6d2 arrangement is more stable than the 5f2. Nevertheless the unusual behaviour of Cr(r1) and CU(II) acetates suggests that the latter result should be interpreted with J. K. Dawson J. 1952 1185; 1952 2705. 402 QUARTERLY REVIEWS caution. It is concluded by Lister that taken as a whole the magnetic data for the trans-uranic elements undoubtedly support the 5f structures.Hutchison and Elliott l l G have reached the same conclusion. Chemical evidence for the elements preceding uranium suggests that thorium is a Group IV and protoactinium a Group V element. These deviations from " rare-earth " behaviour can be understood when it is borne in mind that 5f electrons are less firmly bound than 4f electrons. A serious difference between the rare-earth and the actinide series is observed in comparing PU(III) with Sm(m) and Am(m) with EU(III). The smaller moments of the actinide series are attributed by Howland and Calvin to wide inultiplet splitting in which case the calculat'ed Sm(m) and EU(III) value based on narrow multiplet separation is no longer applicable.To sum up magnetic data support the view that the actinides have electronic configurations similar to those of the rare earths possibly excepting those with only one or two unpaired electrons ; greater variation owing to inagnctic concen- tration and chemical reactivity however is observed. Orbital Contribution and Stereochemistry A study of the factors affecting the size of the orbital contribution to the magnetic moment of compounds of the first transition series promises t o yield information of diagnostic value Tor inorganic stereochemistry. As mentioned earlier the reduction in the magnetic moment of these elements to nearly the spin-only value is attributed to quenching of the orbital contri- bution by the electric field created by the surrounding atoms.This crystalline field causes a separation of the 'degenerate energy levels of the D or F states of these ions (but not of an S state since here L = 0). Con- sider as an example t'he Co++ ion containing three unpaired electrons ground state term 4J'9/2. It was shown by Betlze 117 that a cubic field (as occurs when the Co+f ion is a t the centre of a perfect octahedron of H,O molecules) splits an F state into three energy levels the separation between successive levels being about lo4 cm.-l (see Fig. 6). Usually small depar- tures from cubic symmetry occur owing to slight distortion of the octahedron of H,O molecules ; this may be regarded a-s equivalent to imposing on the cubic field a small coniponent of lower synimetry e.g. tetragonal or rhombic." This rhombic component causes a further splitting of two of the three energy levels into triplets the energy separation between which is of the order of kT (ZOO cm.-l a t room temperature).This gives seven levels in all. In a similar way a D state is split by a cubic field into two levels with an energy difference of about lo4 cm.-l; these two levels are further split by a rhombic component into a doublet and a triplet respectively. The distribution of tthe atoms between the possible energy levels depends upon the energy separation between these levels and its value relativo to kT at the temperature of investigation. This energy level diagram known as a 116 J . Chern. Phys. 1948 16 920. 117 Ann. Plzy~ilc 1929 3 133; Z . Phpik 1930 60 218. * Cubic field three axes at right angles all equal. Tetragonal field three axes Rhombic fiald three axes at right angles all unequal.at right angles two equal. NYHOLM MAGNETISM AND INOROANIC C'FIEXXSTSY 403 '' Stark Pattern " is given for P and D states under certain conditions in Figs. 6 and 7 respectively. The Stark pattern is inverted when the sign of the field constant (FC) changes from positive to negative. The field constant is the coefficient of the cubic field but will not bc discussed further.1 11'9 11* 119 Gorter 12* has shown that FC is positive when COT t- is surrounded by six identical negat'ive The diagrams are not to scale. charges at the corners of an octahedron (e.g. [Co(H,O),] t+) and is negative when the Co++ is at the centre of a tetrahedron of negative charges (e.g. [CoCl,]-). The important difference between 6(a) and 6(8) is that in the Iatter the singlet level lies lowest and there is a large energy difference between it and the next energy level.Since in 6(a) there is a low-lying triplet with energy separations comparable with kT all three levels will be occupied and a large orbital contribution can be expected ; for 6 ( b ) however much closer approximation to the spin-only formula is to be expected. Table 4 gives typical data for octahedral and tetrahedral Co++ comp2excsY TABLE 4. Magnetic moments of '( ionic " cobaltous corplexes.l24 Octahedral complcws [Co(H,O),]CI . . . . 4.94 [CO(H,O),][C~O~],,~H,O . 4-93 [CO(NH,),][C~~~] . . 5.04 [C~(dlpy),][ClO,] . . . 4.86 [co(Py)G][c104] . rn * . 4-87 Tetrahedral complexw (~~K),[COCI,] . . . . 4.74 (pyH),[CoBr,] . . . . 4.67 [CoC1,,2(C,H5),P]* . . . 4.48 Ng[Co(CNS),] .. . . 4.33 I 11* J. €3. Van Vleck Phys. Review 1932 41 208. ll@ W. G. Pennoy and R. Sclilapp ibid. p. 194 ; 1932 42 666 ; 1933 43 486. lZo Ibid. 1932 42 437. 404 QUARTERLY REVIEWS illustrating the effect of the change in stereochemistry. There is fairly wide variation in both classes but the moments of tetrahedral complexes are certainly lower than in octahedral compounds. Nevertheless the theory is obviously over-simplified because the tetrahedral compounds have moments much larger than that expected for the spin-only value of 3.88 B.M. The magnetic moments of the blue and the violet form of the com- pound CoC12,2py illustrate the effect of stereochemistry on magnetic be- haviour. Barkworth and Sugden121 have shown that the moments of these are respectively 4.62 and 5.33 B.M.Formluz such as [CoCi2,3py]0 [Copy,] [CoCl,] and a polymerised structure involving octahedrally co- ordinated Co( 11) with chlorine bridges suggest themselves. Clearly an understanding of the way in which asymmetry affects the moment will assist in the solution of this problem." It has been shown that Ni(I1) should behave in the opposite manner to CO(II) tetrahedral Ni( 11) complexes having the higher moments.119~ 180 In general the moments of tetrahedral Ni(I1) complexes are larger than those of octahedral complexes containing six identical ligands but if the six groups attached to the Ni(I1) atom are not identical the moment is often much larger than the spin-only value. Thus the moments of tetrahedral Ni(n) bisacetylacetone octahedral hexamminonickel(1I) perchlorate and octahedral Ni(n) bisacetylacetone dihydrate are respectively 3-39 3-15 m d 3.42 B.M.lZ4 Hence in distinguishing octahedral Ni( 11) from tctrahedral Ni(11) results are less conclusive probably because the moment is more sensitive to slight departures from cubic symmetry in the crystalline A study of the way in which various factors affect the size of the orbital contribution for bivalent transition elements has been made by Kanekar.124 It is found that (i) changing only the anion in complexes of the type [Co(NH,),]X2 or [Ni(NH,),]X has practically no effect upon the orbital contribution; the influence of the anion upon the crystalline field is negligible compared with the octahedron of closer NH molecules ; (ii) changing the Zigand in octahedral complexes of the above kind e.g.NH -+ H20 -+ ethylenediamine has a small but significant effect ; this seems to be related to the relative electronegativities of the attached groups and hence the field which they create ; (iii) changing the co-ordination number (stereochemistry) has a very marked effect which is in the sense predicted by the crystalline field theory as discussed above ; (iv) changing the halide in complexes of the type [Cox,]- causes a fall in the moment along t'he sequence pc! > pBr > pI > pCNs.As with (ii) this is apparently related to the relative electlronegativities. For ions in D states with the triplet lying lowest (e.g. the [Fe(H,O),]++ field.122 125 121 Nature 1937 139 374. 122 P. L. Mukherjee 2. Krist. 1935 91 504. 123 G. A. Barclay T. Christie and R. S. Nyholm unpublished experiments.124 Thesis London 1953. 125 A. Bose Indian J . Yhys. 1948 22 25 33 57. * A recent investigation 123 has shown t(hat t,ho blue form almost certainly has the formula [C0C1,,2py]~ whilst the violet form is probably R bridged polymer OF octahedral bivalent cobalt. NYHOLM MAGNETISM AND INORGANIC CHEMISTRY 405 ion) as in 7 ( b ) a considerable orbital contribution is to be expected for the same reason as with 6(a). However in the case of octahedrally co-ordinated Cu++ a doublet lies lowest and although one might expect this to give rise to a considerable orbital contribution yet Bethe 117 has pointed out that such a doublet is essentially " non-magnetic " and should not result in any orbital contribution. I n practice it is observed that the magne5c moment in one direction in the CuSO,,SH,O crystal is equal to the spin-only value but a considerable orbital contribution is observed along other axes.This has been explained by a further energy-level splitting owing to the com- plicated crystalline field operating (see next section). No reference has been made to the size of the orbital contribution in complexes in which electron pairing occurs. Thus we find that octahedral covalent CO(II) complexes e.g. KzCaCo(N02)6,120 have moments of the order of 1.9 B.M. whereas square Co(11) complexes haw much larger moments in the range 2.1-2-5 B.M. This is at least consistent with Pauling's theory that the unpaired electron in the first instance is in a 58 orbital where the orbital contribution should be zero. However considerably more informa- tion is necessary before any general statements can be made about covalent compounds.In particular we need much more data abouti the separation of energy levels and knowledge of the way in which the magnetic suscepti- bilify varies with temperature. Finally the measurement of magnetic aniaotropies will be of great value. Valuable data of this kind for salts of Fe Co Ni and Cu have been obtained by G ~ h a . ~ ~ Paramagnetic Resonance Absorption This subject has developed rapidly of late owing to improvements in the technique of generating and studying inicro-waves. It is now one of the most active fields of research in p h y s i ~ s . l ~ ~ - l ~ 3 The measurements are of importance for our purposes because of the information obtained about energy levels and magnetic moments. When a paramagnetic salt is placed in a steady magnetic field of the order of a few thousand oersteds which field may be varied at will and an oscillating electromagnetic field is applied at right angles to the former marked absorption of energy by the specimen occurs at certain values of the field strength.The position of these peaks in terms of the frequency of the oscillating field and of the steady magnetic field provides information concerning both nuclear and electronic spin of the atom ; we are concerned solely with the latter. An absorption peak is observed whenever the separation of the energy levels of the paramagnetic ion is equal to the quantum of energy (hv) corresponding to the radio 1 2 6 P. Ray and H. Sahu J . Indian Chem. Soc. 1946 23 161. 12' R. I;. Cummerow D. Halliday and G. E.Moore Phys. Review 1947 72 1233. lZ8 D. M. S. Bagguley B. Bleaney J. H. E. Griffiths R. P. Penrose and B. I. lZ9 1'. Ting and D. Williams Phys. Review 1951 82 507. 130 C. J. Gorter '' Paramagnetic Relaxation " Elsevier Leiden 1947. 131 C. Kikuchi and R. D. Spencc Amer. J . Physics 1950 18 167. 132 B. Bleaney Byif. J. Appl. Phys. 1952 3 337. 133 D. M. S. Bagguley and J. H. E. Griffit,hs Proc. Rqy. Soc. 1950 A 204 188. Plumpton Proc. Phys. Xoc. 1948 61 542 551. 406 QUARTERLY REVIEWS frequency. Hence a direct comparison of hv and kT is possible. The advan- tages of this method over others for the study of paramagnetic ions have been suminarised by Bleaney slid his co-workers 128 as follows (i) no corrections for diamagnefisin or diamagnetic impurities are needed ; (ii) direel ineasurenicizts of 9 values are possible (leading to magnetic moments) ; (iii) small splittings of the order of 1 cni.-l can be observed ; (iv) direct observations can be made of the different paramagnetic ions in n unit cell ; (v) very small spin-lattice relaxation times of the order of 10-lo see.may be observed. Nccdleds to my the main disadvantage is the technical difficulty of workiiig with niicro-waves. The technique is fully discussed e l s e ~ r h e r e ~ ~ ~ - l ~ l wherein detailed references to other workers are given. The wave-length of the oscillating field is of the order of 3 em. and the magnetic field is varied betmeeii 0 and 15,660 oersteds. If m is the magnetic quantum number transitions of considerable intensity can occur between levels provided that the change in m is -J= 1.The relationship between the frequency of the radiation v the magnetic field H and tIhe Land6 splitting factor g is given by hv = gPH where P is the Bohr magnetctn. Of the large number of compounds so far examined h e w fail to show the phenomenon of absorption maxima at room temperatures. Failure to do so has been attributeu lZ8 to (a) too large Lsln initial splitting of the orbital levels by the crystalline field or ( b ) too brond an absorption line owing to the very short spin relaxation time. The latter difficnlty can be overcoine by working at low temperatures ; thus Ti(m) salts show no absorption until the tempera- ture is reduced to that of liquid helium. Following a preliiuinary survey of more than 100 salts of the iron group,128 inore detailed studies have been made of chrome alurn,l33 manganese various cobaltous m l t ~ ~ ~ 5 138 and some bivalent nickel salts.l37 As an illustration A.Abragam and M. H. L. Pryce 136 have studied in detail the paramagnetic resonance spec- trum of copper Tutton salts (CuS0,,&!1,S04,6H,0) and conclude that the data can be interpreted by assuming thatl a crystalline field of tetragonal syiiirnetry is operating. Pauling’s view of the stIruc0nre (four 3d4s4p bonds) is considered to be incompatible with their findings. Lack of sufficient crystal-structure determinations to enable theories to be checked is a handi- cap but striking advances have been made in the determination of energy levels and the correlation of stereochemistry with magnetic moment. 134 13. Bleaney and D. J. E. Ingram ibid. 1951 A 205 336. laS A. Abragam and M. H. L. Pryce ibid. 1951 A 206 173. 136 Idem ibid. p. 164. 137 J. H. E. Griffiths and 3. Owen ibid. 1952 A 213 451). 13* B. Bleaney and D. J. E. Ingram ibid. 1961 A 208 143.

ISSN:0009-2681    

DOI:10.1039/QR9530700377

出版商:RSC    

年代:1953

数据来源: RSC

摘要:

TEE SYNTHESIS OF ISOTOPICALLY LABELLED ORGANIC coMPouNp)s By S. L. THOMAS PH.D. and H. S. TURXER PI3.D. (C'HEMICL4L RESEARCH LABOEATORY D.d .I.K. 'FEDDINGTON) Introduction MOST tracer studies necessitate the synthesis of isotopically labelled materials and the last few years have seen important developments in this branch of radiochemistry. It therefore seems appropriate to survey tJhe methods used for the synthesis of organic compounds labelled with isotopes of the more important elements viz. hydrogen carbon nitrogen oxygen sulphur phosphorus and the halogens. Froin t'inie to time information on certain aspects of this subject has been summarized lPg but we shall lay special emphasis on recent developments so that this contribution may be regarded as an extensioii to the survey which formed part of Arnstein and Rentley's review.1 Both chemical and biological methods of synthesis are' covered and an account is given of methods of degradation used for the location of labelled atoms in organic compounds.Information concerning the sterling-area availability of the isotopes and their compounds is provided in a catalogue published by the Isotopes Division of the Atomic Energy Research Establishment. This document together with one entitled ' L An Introductory Manual 011 the Control of Health Hazards from Radioactive Materials ,',lo provides much useful information concerning the health hazards and manipulation of t,hcse substances. Isotopic Synthesis. General Considerations.-The selection of the tracer' element where any choice exists is made on the basis of factors sucl; as stability of labelling ease of synthesis and assay isotope half-life concentra- tion of isotope available influence of radiation eEects etc.Similarly the choice of isotope (radioactive or stable) will depend on whether mass spectro- metric assay is obligatory (see below) on the radioactive half-life the dilution 1 Arnstein and Eentley Quart. Reviews 1950 4 172. 2 Calvin " Isotopic Carbon " Chapman & Hall London 1049. Cold Spring tiarbor Symp. Vol. 13. Crompton and Woodruff ATucZeonics 1950 7 (3) 49 ; (4) 44. Grove and Catch Brit. Mcrl. Bull. 1952 8 234. Kamen " Radioactive Tracers in Biology " 2nd Edn. Academic Press Inc. New York 1951. 7 Lawrence and Hamilton " Advances in Biological and Medical Physics " Vol. 1 (1948); Vol. 2 (1951). * Tabern Taylor and Gleaaon Nucleonics 1950 '7 (5) 3 ; (6) 40 ; 8 ( I ) 80.lo Available from The Direcbor The National Physica.1 Laborat,ory Teddington. U700druff and I+'owler ibid. 1950 7 ( 2 ) 26. Middlesex. 407 408 QUARTERLY REVIEWS expected during the experiment and in certain cases (notably with hydrogen isotopes) on differences arising from an isotope effect.11 Ideally an isotopic synthesis should be simple and should combine a high recovery of isotope with low dilution by unlabelled material. A high yield is sought for reasons of economy and the isotope is therefore introduced at as late a stage in the synthesis as is possible yields being improved by careful selection of reaction conditions by the use of unorthodox methods and new reagents and by refinements in experimental technique. The concentration of isotope in the product must be sufficient to permit assay at the end of the tracer experiment and further the gross quantity of labelled compound must not be so large as to disturb the biological or otbcr system under investigation.Hence dilution during synthesis must be prevented or restricted to an extent determined by the isotope used its half-life its initial concentration the nature of the system being investigated and the sensitivity of the method of analysis. In any particular synthesis the loss involved if low yields are obtained must be set against the extra effort required to secure higher yields. Sirni- Iarly if dilution is to be avoided or kept very small it will usually be necessary to work on a small scale (often below 10 millimoles) and to accept the attendant experimental difficulkies .I2 In the manufacture of labelled compounds it is customary to work both for high yield and low dilution l2 and this is made rather easier by the use of larger quantities of isotope than would be handled by individual research workers.In many cases even for much biological work a 10- or 100-fold dilution during synthesis is permissible. It is then advantageous to work on the ordinary small laboratory scale or alternatively to work initially on a smaller scale and to improve both yield and purity at difficult stages by judicious addition of pure carrier. This process of carrier dilution has been applied to the resolution of racemates l3 although it is not always wholly suc~essful.~~ In certain cases a very high dilution and low yield can be accepted particularly in chemical tracer investigations ; thus Loftfield l5 converted 1*CO into %chloro[l 2-1~Cl]cycZohexunoiie for use in a study of the Faworskii reaction via an %stage synthesis over which the dilution was approximately 30,000-fold.The overall yield of isotope from 14C02 to the end product of the degradation of the reaction product was approxim- ately 0403% and could have been as little as 0-2./ of this value without effect on the accuracy of the results. Detailed descriptions of experimental methods e.g. vacuum-manipula- tion of volatile compounds are included in many of the papers cited in this Review ; more comprehensive accounts have been given by Calvin and Catch .l l1 Ropp hTucleonics 1952 10 (lo) 22. l2 Catch " Radio-isotope Techniques " Vol. 11 H.M.S.O. London 1952 p.100. l 3 E.g. Wood and Gutman J. Biol. Chem. 1919 179 535. l4 Arnstein Hunter Muir and Neuborgor J. 1052 1320. l5 " Use of Tracers in Organic Reaction Mechanism Studies " Brookhaven Con- ference Report BK'i-44(C-10) Jan. 1950 p. 59. TFKOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 409 Purity of Labelled Compomds.-The usual criteria of purity of organic compounds may also be applied to labelled compounds but are often inade- quate. Thus as a result for example of carrier dilution during a synthesis a labelled compound that is pure by normal standards may contain a chemical trace of a radioactive impurity with a gross activity of a similar order to that in the principal compound. This situation arises from the very large dilution many radioactive isotopes may undergo before detection becomes impossible it is much less important when stable isotopes are used.Sensitive determination of chemical impurities is sometimes valuable,16 but the most satisfactory tests of purity are those designed to demonstrate the extent to which the isotope is associated with the compound being studied. Thus an isotopic compound must be purified not only to constant melting point boiling point refractive index etc. but also to constant isotopic composition and reliance should not be placed on one method of purification alone. If for example the purity of an acid is determined by Duclaux distillation the acid and isotope content of the fractions should correspond within experimental error. Similarly the same values for partition co- efficients should be obtained by chemical analysis and by isotope assay.Paper chromatography combined with autoradiography is especially useful in the detection of radioactive impurities.17 Keston Udenfriend and Cannan l8 have developed an ingenious method of analysis for amino-acids in which the mixed acids and authentic pure specimens of each expected component are converted into p-iodophenylsulphonyl derivatives one with the [1311]- and the others with the [35S]-compound. Representative samples are then mixed and chromatographed together and the distribution of radio- activity on the chromatogram is examined. The ratio of sulphur- to iodine- activity (and these are of widely differing character) is constant throughout each pure band and its value provides a critical assay for the acid under examination. The positive absence of any specific impurity may be established by adding the non-radioactive impurity as carrier and re-separating on paper.Finally if the compound is subjected to degradation the sum of the isotope conbents of the fragments should agree with that of the whole compound determined directly. Many of the conventional methods of purification (e.g. precise fractional distillation) are quite unsuitable on the relatively small scale of most isotopic syntheses. Losses are involved in all purification procedures and it is there- fore desirable to devise synthetic methods which give very pure products or products containing unexceptionable or easily removable impurities. Thus if carbon dioxide is reduced by lithium aluminium hydride dissolved in diethylene glycol diethyl ether the methanol product is contaminated with ethanol formed by scission of the solvent.Separation is impracticable l6 Hughes Williams and Young J, 1951 1279. l8 Kostcn Udenfriend and Cannan J . Arner. Chem. SOC. 1949 71 249. E.g. Putman and Hassid J. Bid. Chem. 1952 196 749. 410 QUAGTERLY REVIEWS but the use of an alternative solvent gives methanol free from ethanol. Paper chromatography is applicable on the small preparative scale 17 and has been very widely used in the separation and purification of labelled sub- siances found in biological tracer experiments. 2* Multiple Eabelling.-Solne coixpoiarids behave as t>hough labelled in more than one atom in the molecule whether these atom are of the same or different elements. In such compouiids a distinction may be drawn between those tJiat contain rnolccules bearing more than one labelled atom and those that do not.Thus if methyl iodide (containing say 5 atoms yo 14C) is converted into the Grignard reagent and then carboxylated with 14C02 (also containing 5 atonis yo 14C) some 0.25% of the molecules of acetic acid produced will (apart froiii aiiy possible isotope effect) contain two 15C atoms. This acetic acid could be distinguished by mass-spectrometric assay of derived ethylene from a mixture of 1WH,*12C0,H and 12CH,*14C0213 of the same overall isotopic composition at each carbon atom. This type of distinction is occasionally important 21 but for most practical purposes the two samples behave in an identical manner and am quite indistinguishable by radio- active counting methods. At tracer concentrations the difference is beyond the sensitivity of the mass spectrometer.It is therefore satisfactory for most purposes to use mixtures of singly labelled compounds in place of " truly " multiply labelled compounds. The singly labelled compounds niay be prepared separately in higher yield since conditions are more readily designed to give a high yield from only one reactant. Each compound is however diluted by the others. Effective niultiplication of labelling also occurs when a labelled compound has the appropriah degree of symmetry or has been synthesised via. a synmetrical iiiterniediate. Nomenelak.lre.-Tize nomenclature used for isotopically labelled organic compounds in this Review is that proposed jointly by the Editorial Board of the Biocheinieal Society and the Editors to the Chemical Society and has already been applied in part in puhlicatioris of those societies.22 So far as it affects the compouiids named in this Revicw it is as follows.* The symbol for the isotope introduced is placed in square brackets directly attached to the front of the name as in [l4C]urea1.When more than one position in a substance is labelled by means of the same isotope the number of labelled atoms is added as right-hand subscript (cf. ordinary forniulz) as in [L4Cc,]glycollic acid. When isotopes of more than one element are introduced their symbols are arranged in alphabetical order including 2H and 3H for deuterium and tritium respectively. The isotopic prefix precedes that' part of the name to which it refers as in 2- l9 Cox Turner tmd TVarne J . 1850 316'7. 2o E.g. Benson Bassham Calvin Goodale Haas and Stapka J .Anzer. Chenz. Xoc. 21 TToocl J . Bid. Chern. 1962 194 905. 2 8 J . 1951 3516 1962 5061. * The full sclmme may be obtained from tlm Editor The C'hemical Societ:~ with 1950 '72 1710 ; Tl'intsringlilam Nucleor~ics 1952 10 (3) 62. whose co-operation this account of it hps bwn written. THOMAS AK’D TTJRKER PSOT03’IChLLY LABELLED ORGANIC CONPOTTI? DY 41 1 acetamid0-7-[13~I]iodofluorene c~-naphth[~H]oic acid (C10H,-CQ23H) sodium [14C]formate l-amino[l4C]inethylcycZopeiitanol (NH2*14CCH2*C,H,*QH). When not sufficiently distinguished by the foregoing inems the pouitiom of isotopic labelling are indicated by arabic numerals Greek letters or prefixes (as appropriate) placed within the square brackets and before the symbol of the element concerned to which they are attached by a hyphen ; examples are [l- 2Hl]ethanol(CH,*C€~2H*O€€) [l-14C]aniline [c~-~~C]leucine [cccrboxz~-1~C]leucine [Ne-14C]isoleucine [G 7-14C2]xanthopterin [ ~ L P - ~ ~ C ~ J maleic anhydride [1-l4C 2-13C]acetaldehyde [/3y1WC2 34S]~nethionine [p- 14C XP-~H IjNIserine 2 4-diamino[l 2 3-1WJpyrimidine.When the position o€ isotopic labelling is indeterminate the possible positions are specified together with the number of atoms which are labelled as in [ar-14C,]benzaldehyde (one 14C in the benzene ring) [4 6-14Cl]adenine (one 14C a t position 4 or 6) D-[l 6-14Cl]fructose (one 14C at position 1 or 6). (The device illustrated in the last two examples is an extension of the “ editorial ” proposals.) Hydrogen Hydrogen has two useful tracer isotopes the stable deuterium ( 2H or D) and the weak /?-emitter trit2iurn ( 3H or T z9 - 12 years).Both are available as water and in general syntheses aze applicable to both isotopes. Deuterium can be diluted some 10 3-104-fold with normal hydrogen before the accuracy of analysis is seriously reduced but for tritium (- 50 atoms yo 3H - 1 c per ml.) the corresponding figure is of the order of 1010. particularly in biological applications first because many hydrogen atoms readily undergo exchange or replacement and secondly because of the potentially large isotope effect .ll The element received particular attention in Arnstein and Bentley’s review,l an annual bibliography 24 is published and there have been several other 25 Treatment here will therefore be brief. Deuterium.-Methods of assay 26 referred to by Arnstein and Bentley have been augmented by a spectrographic method based on the stretching frequency of the 0-2D Almost all syntheses reported since the last Review fall into one or other of Arnstein and Rentley’s three categories namely hydrogen exchange hydrogen addition or group replacement by hydrogen.Therefore only syntheses showing particular points of chemical interest will be noted. Hydrogen Exchunge.-[ 1- 2H]Etlhai~ol has been prepared from the product However the isotopes should be used with 2 3 Verley Rachele du Vigneaud Eidinoff and Kxmll J . A m e ~ . Ghem. SOC. 1952 2 4 “ A Review of the Properties of Deuterium Conipounds ” U.S. Dept. of Com- 2 6 KimbaII “ Bibliography of Heavy Hydrogen Compounds ” McGraw-Hill New 26 Kirshenbaum ‘‘ Physical Properties and Analysis of Heavy Water ” RfcGraw- 27 Trenner and Walker Perkin-Elmer Instrument News Fall 1052.74 5941. merce N.R.S. (1946 onwards). York 1049 (to 1945). Hill New York 1951. 412 QUARTERLY REVIEWS of a bromine degradation of the silver salt of an a-enriched propionic acid 28 and from diazoethane 29 *H,O R*COS2H MeCHN - Me.C2HN -+ Me*C2H2.0*COR -+ Me-C2H2*OH 111 Et,O [R*COZ2H = 3 5 1-(N02)2C,H,*C022HJ Partial chromic acid oxidation of the alcohol 28 showed further enrichment of the intermediate aldehyde CH,*C " 0 the protium-alcohol being prefer- entially destroyed. to yield CH2:C2H2. An interesting application of a deuterated diazo-compound has been made by Leitch and his collaborators during extensive and careful studies of the synthetic chemistry of deuterium.Diazomethane either prepared from enriched nitromethane or enriched directly was polyiiierised in ethereal solution by copper powder to yield polydideut eromethylene 31 [C2H2In. Other papers in this series deal with the synthesis of various alkyl halides and polyhalides deuteroformaldehyde etc. Replacement of hydrogen in the biologically important ascorbic acid dissolved in heavy water was studied by Weig1.32 Infra-red analysis suggested the lability of the hydrogen atom The enriched propionic acid has also been electrolysed a t c,. -1 7 o-- I O=C-C=GC-C-CH,.OH I I I 1 OH OH H OH + Addition Reactions.-Two methods have been used for the partial reduc- tion of the acetyleiiic bond. By the use of the chronious chloride in 2HC1 Ronzio prepared [ 2H,]ethylene which polyinerised 33 rather more readily than the corresponding protium compound to yield a polydeuteroethylene.An alternative method based on the use of certain deuterised Raney nickels was established by Khan.34 This procedure has been utilised for the syntheses of [l-2H]acetaldehyde 35 Me-CiC-Me + Me*C2H:C2H*Me -+ Me*C2H-C2H*Me -+ 'O/ HOC2HMe-C2HMe.0H -+ 2Me*C2H0 By low-temperature addition of 2HC1 to anethole followed by a bi- molecular dehalogenation with reduced iron powder [2 5- 2H2]hexcestrol 28 Cornforth and Popj&k Nature 1949 164 1053. 29 Curran and Rittenberg. J. Biol. Chem. 1951 190 17. 30 Kruis Naturwiss. 1948 35 155. 31 Leitch Gagnon and Cambron Canad. J. Res. 1950 28 B 256. 32 Weigl Analyt. Chem. 1952 24 1483. 33 Ronzio U.S. At. Energy Comm. Rep. LA-14 78 ; cf. ref. 31. 34 J. Amer. Chem.SOC. 1952 74 3018. 35 Blacet and Brinton ibid. 1950 72 4715. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 413 having a slightly lower physiological activity than that of the protium compound has been prepared.36 The fundamental importance of the equilibrium H + R*CO,H* + HH* + R*CO,H in catalysed hydrogen-addition and -exchange reactions has recently been demonstrated. 37 The deuterated carboxyl group plays the part of an isotopic buffer. Replacement Reactions.-The simple deuterated hydrocarbons are readily prepared by interaction of deuterium oxide with suitable carbides but [2H4]allene 38 represents an unusual by-product fi-oiii a sj-iithesis of C2H4]- mcthylacetylene from Mg ,C 'I'he acetylenic hydrogen readily undergoes exchange. Several workers have studied the preparation of [Z- 2H]propan- 2-01 (Me,C2H*OH) and [a 2- 2W2]propane by catalytic reduction of acetone.39 It appears to be impossible to prevent partial exchange with the cc-hydrogen atoms during preparation of the hydrocarbon in this manner but It is probable that an authentic propanol may be prepared by catalytic or lithium aluminium hydride reduction of the ketone.As is found with acetaldehyde oxidation of the deutero-alcohol is relatively sl0w.40 Schissler Thompson and Turkevich 41 have devised a niethod for introducing one two or three deuterium atoms at a single carbon atom through Zn-Ac02H reduction of suitable halides. Deuterochloroform has been prepared from trichloroacetophenone and from calcium Crichloroacetate by the haloform reaction.42 I n an ingenious synthesis of [l l-2H2]allyl acetate,43 the crystalline adduct of anthracene and acrylic ester was reduced with lithium aluminium deuteride and the product acetylated.The complex was readily decomposed by heat (see npxt paga). By hypophosphorous acid dearxination of diazoniuin salts in 2H,0 Alexander and Burge 44 have introduced specific labels into aromatic com- pounds but the efficiency of the reaction is poor since protium enters the nucleus preferentially. In a study of optical activity due to the presence of deuterium Alex- ander 45 employed the useful lithium aluminium deuteride reduction of a toluene-p-sulphonate to prepare trans-p- [3- 2H]menthane. The material had [a]2,5 = - 0.09" 4 0-01" while the corresponding protiuin compound was quite inactive. s6 Lacassagne Buu-Hoi Chamorro Xuong and H o h Compt.rend. 1950 231 37 Eidinoff Knoll Fukushima and Gallagher Abs. 118th Amer. Chem. SOC. Mtg. 3* Lord and Venkat,eswarlu J. Chem. Phys. 1949 20 1237. 39 Friedman and Turkevich J . Amer. Chem. Xoc. 1952 74 1669 ; Williams Krieger and Day Abs. 122nd Amer. Chem. Soc. Mtg. 1950 22M. 40 Westheimer and Nicolaides J . Amer. Chem. SOC. 1949 71 25. 41U.S. At. Energy Comm. Rep. AECU-1387. 4 2 Boyer Bernstein Brown and Dibeler J . Amer. Chem. SOC. 1951 73 770; 43 Bartlett and Tate ibid. 1953 75 91. 4 4 Alexander and Burge ibid. 1950 72 3100. 46 Alexander ibid. p. 3796. 1384. 1950 p. 6 6 ~ . Earing and Cloke ibid. p. 769. 414 QUARTERLY EEVIEWS I 1 LiAliHfl (860/,) 2 AcCl (96%) + I I CH,-CH CH,=CH I C2H,*OAc I C2H2*OAc Miscellaneous.-It is not possible to describe here the valuable work on the deuterium-labelling of steroids which has been carried out by Gallagher and others .46 These compounds are frequently more readily accessible than the carbon-labelled substances and are therefore useful for biological studies but it is necessary to ascertain the position and stability of the label as well as to devise methods for its introduction at specific positions.There have also been occasional biosyntheses with deuterium but this is not a valuable method owing to the existence of a rapidly changing hydrogen pool in the living cell. Tritium.-Tritium has become available a t very high specific activity as a result of pile synthesis by the reaction 6Li(n,a)3H. It is a very soft P-emitt'er (max. 0418 mev) and must therefore be analysed4' by a gas counting tube ionisation chamber or scinti1lat)ion counter,48 preferably as hydrogen or as a hydrocarbon derived from water of combustion.In general synthetical methods follow those of deuterium but certain specific examples have been reported. Inter.mediates.-Synthesps of lithium tritide and the useful lithium aluminium tritide have been announced. 49 Exchange Reactions.-Tritiated methanol has been prepared (i) by hydrolysis of the ester derived from diazomethane and a-naphth[ 3H]oi~ a ~ i d ~ 3 and (ii) by a replacement reaction involving the catalytic reduction of methyl fornzate. 50 Other compounds prepared by exchange methods 46 Gallagher " Isotopes in Biochemistry " Churchill London 1951 p. 28 ; Nolin 47 Glascock Riochem. J. 1952 52 699. 48 Farmer and Berstein Science 1953 117 279 ; Hayes and Gould ibid.p. 480. 49 Wdzbaoh and Kaplan J. Amer. Chem. SOC. 1050 '72 595. and Jones Canad. J. Chem. 1952 30 727; Bell and Thomson J. 1952 572. Harman Stewart and Ruben ibid. 1942 64 2293. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 415 include phei~ylalanine,~~ stearic acid,62 benzene,53 and a range of steroids 64 including cortisone. 55 Addition Reactions.-Succinic acid 513 and hexcestrol 57 have been pre- pared by this method the addition of tritium apparently being favoured over that of protium in the latter case. Replacement Reactions.-[l-3H]Ethanol has been prepared 49 by lithium aluminium tritide reduction of ethyl acetate and styrene 58 by toluene-p- sulphonic acid dehydration of the hydrogenation product of acetophenone. In an extensive review of sroinatic substitution Melander 53 has prepared a series of tritiahed aromatic compounds mostly via the appropriate Grignard compound.MisceZZaneous.--Tritiated s tilbene 59 is self'-luminous and may be used as a constant light source for the standardisation of photomultiplier tubes. There have been biosyntheses of labelled iiucleic acids.6O Carbon llC a positron emitter (r+ - 20 minutes) is made. in the cyclotron ti by the reac- tion 10B(d,n)llC but although its energetic radiation makes detection very easy it has been little used since the longer-lived 14C became readily avail- able in 1946. 13C the heavy stable isotope is separated from noriiial carbon (containing - 1.1% of l3C) by fractional distillation of carbon inonoxide or by means of isotopic exchange It is available as Ba13C0 and K13CN and is usually nnalyced as carbon dioxide in the mass ~pectrometer.~~ I4C a weak p-emitter (- 0.15 mev ; -r+ - 5600 years) is made in the pile by the rcaction 14N(n,p)14C.Depending on the choice of nitrogenous target material the 14C inay be obtained in a variety of com- pounds,'d G5 but it is normally available as Ba14C0 from which other com- pounds are prepared by chemical or biological synthesis.66 14C is usixdly analysed by counting 8s #carbon dioxide or barium carbonate 67 but Three isotopes of carbon2 have been used in tracer studies. 51 Gurin and Delluva J. Biol. Chem,. 1947 178 545. 6 2 Rosenthal and Kritchevsky Univ. California Radiation Lab. Rep. 1131. 63 Melander Acta Chena. Xcand. 1948 3 96 ; Arkiv Kemi 1950 2 213. 6 4 E.g. Biggs and Kritchevsky A ~ c h .Biochem. 1952 30 430. 6 5 Fukushima Kritchevsky Eidinoff and Gallaghcr J . A m e r . Chenz. h'oc. 1952 5 6 Williams and Ronzio U.S. At. Energy Comm. Rep. AECU-2226. 57 Idenz J . Amer. Clzem. Xoc. 1950 72 5787. 58 Berstein Bennett and Fields ibid. 1952 74 5763. 69 I d e m Nucleonics 1953 11 (2) 64. 6o Eidinoff Riley Knoll and Marrian J. Biol. Chcm. 1952 199 511. 61 London " Mass Spectrometry " Institute of Petroleum London 1952 p. 141. 6 2 Stewart Nucleonics 1917 1 (2) 18. 63 Wilson (Ed.) " Preparation and Mcasurernent of Isotopic Tracers " Edwards 6 4 Ref. 2 p. 6. 6 5 Croatto Giacomello and Xaddoclr R i c . sci. 1951 21 1598. 6 6 More than 80 14C-labelled compounds arc stvsiiablc from the Radiochemical 67 Neville Atomics 1962 3 309 ; Smith ibid. 1953 4 29. 74 487. Ann Arbor Mich.1948. Centre Amersham. EE 416 QUARTERLY REVIEWS occasionally more complex compounds are counted directly either as solids or in solution.6g 7O Chemical Syntheses.-Definite location of isotopic atoms can normally be achieved by chemical synthesis but if there are symmetrical inter- mediates (effective) multiple labelling may result. I n some cases unex- pected molecular rearrangements occur ; thus when Loftfield 7 1 attempted to prepare [ l-14C]cycZopentanecarboxylic acid by the sequence [ 1-14C]cyclo- pentanol + cyczopentyl bromide - cyclopentyl cyanide -+ cyclo- pentanecarboxylic acid degradation showed -20% of the 14C to be in the methylene-carbon atoms of the ring. Syntheses with 13C and 14C are similar and will not be discussed separ- ately in this Review. However the most highly enriched 1% normally available (65-75 atoms % of 13C) can be diluted only - lo3 times with normal carbon before the accuracy of analysis is reduced while the corres- ponding figure for 14C (- 5 atoms % of 14C ; - 3 rnillicuries per niilliatom) is 10~-lOs.One-carbon Compounds and Simple Intermediates.-Some recent develop- ments are outlined in Table 1. Functionally Labelled Carboxylic Acids.-The carboxylation of organo- 68 E.g. Hogness Roth Leifer and Langham J . Amer. Chem. Soc. 1948 70 3840. e9 Schwebel Isbell and Karabinos Sciencp 1951 113 465. 7O Audric and Long Research 1952 5 46. 71 Loftfield J . Amer. Chem. Xoc. 1951 73 4707. 7 2 Ref. 1 p. 180. 7 3 Von Schuching and Barnes J . Amer. Chem. SOC. 1950 72 3817. 7 4 Williams and Ronzio ibid. 1952 74 2407. 7 6 Adamson ibid.1947 69 2564 ; Henneberry and Baker Canad. J . Res. 1950 28 B 345 ; Malmind Tokarev and Shamyakin Doklady Alcad. Nauk X.S.X.R. 1951 81 195 (Chem. Abs. 1952 46 3889) ; Claus Abs. 121st Amer. Chem. Soc. Mtg. 1952 7 6 McCarter J . Amer. Chem. Soc. 1951 73 483. 77 Abrams ibid. 1949 71 3835. $ 78 Spyker and Neish Canad. J . Chern. 1952 30 461. 7g Cox and Warne J. 1951 1895. so Heard Jamieson and Solomon J . Amer. Chem. Xoc. 1951 73 4985. 8 2 Weygand and Schaefer Chem. Ber. 1952 85 310. a 3 Grant and Turner Nature 1950 165 153. 84 Burr Brown and Heller J . Amer. Chenz. SOC. 1950 72 2560. 85 Wagner Stevenson and Otvos ibid. p. 5786. 8 6 Rdams Selff and Tolbert ibid. 1952 74 2416. 87 Arrol and Glascock J. 1948 1534. s9 Arrol and Glascock J. 1949 S335. 91 Kramer and Kistiakovsky ibid.1941 13'7 654. 9 2 Cox and Wariie J. 1951 1893. g3 Kogl Halberstadt and Barendregt Rec. Trav. chim. 1940 68 3 7 . 9 4 Ostwald J . Biol. Chem. 1948 173 207. 9 5 Fields Rothchild and Leaffer J . Amer. Chem. Xoc. 1952 74 2435. g6 Ropp ibid. 1950 72 4459 ; Gal and Schulgin ibid. 1051 73 2938. 97 Bennett ibid. 1952 74 2420. 98 Fields Walz and Rothchild ibid. 1951 73 1000. PBr3 NaCN Murray and Ronzio ibid. 1952 74 2405. Monat Robbins and Ronzio U.S. At. Energy Comm. Rep. AECU-672. Kilmer and du Vigneaud J . Biol. CJLem. 1944 154 247. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC CONPOUNDS 417 TABLE 1. One-carbon compounds and simple intermediates T 2 Compound I Starting material method and yield *CN*NH . . . KN*CO . . . . Na*CN or K*CN . *CH,*NH . . . *CH,N . . . . H**CHO .. . . H.*CO,Na . . . *CH . . . . . *CH,.OH. . . . *CHi*CH . . . *CH,:*CH . . . *CH,CI*CH,Cl . . *CH,**CHO . . *CH,**CH,.O . . - (*Cc),H) . . Cl*CH,*'CO,H Br.CH,*CO,H CH,(CO,H) 1 CN.CH,*CO,H ) CH,(CN)2 J CK**CH*CO ,Et I I KHAC *CNCH(N,Ph).CN . Ba*CO (heated with NH and NaN,) 94% NH,**CO*NH (heated with K,CO,) 70-80% Ba*CO (heated with NaN,) 75-93% K,WO (heated with Zn in NH,) 90% *CO (via carbon) 59-70y0 H**CO,Na (heated with NaNH,) ;- 85% *CH,I (Gabriel reaction) 9Syo Na*CN (catalytic reduction) 85% *CH,*NH (via nitrosomethylurea) 57-68 % (55% *CH,*OH (catalytic oxidation) 77-81 % *CH,*OH (catalytic oxidation) 86% K*CN (alkaline hydrolysis) N loo?/ "CO (reduction with LiEH,) 73:/ *CH,I (Grignard reaction) 86% *CO (reduction by LiAIH,) 89% H**CO,H (hydrogenation of Cd-Ni salt) 85% *CO (reduction with Ba metal) > 90% Ba*CO (reduction with Ba metal) 98% *C,H (reduetion with TiC1,) 96-987L h'a*CN (via CH,**CN CH,.*CH,*NH, and *C,H (catalytic addition of H,O) 7576 *C,H (via HO.*CH,.*CH,Cl) 53-95:/ H*"CO,Na (440"/0.01 mm.) 900; CH,**CO,Na (Cl,-PCI, P, I,) 6770 CH,*CO,Na (Br,-CH,.COCI) 79-84% Conventional syntheses from NaWX or *CH,*CO,Na from BaCO,) *CH,:CH,) 56% CN**CH,*CO ,Et (nitfrosat ion cat alyt'ic reduction) 76% Na*CN (via *CN*CH,*CO*NH,) 46-53y0 Ref.73 74 75 76 77 78 79 80 79 81 82 83 84 85 19 86 87 88 89 90 91 92 93 94 95 96 97 98 97 metallic compounds especially of Grigiiard reagents gives excellent yields of carboxylic acids (up to 98%) and is probably the most useful first step in isotopic syntheses with carbon ; gg the hydrolysis of nitriles prepared from inorganic cyanides is less important 99 and other methods e.g.reduc- tion of keto-acids,lo0 are relatively unimportant. Most simple derivatives can be satisfactorily prepared by standard methods. Volatile esters are conveniently made by the reaction of alkyl sulphates lo1 or phosphates lo2 with salts of carboxylic acids while volatile acid chlorides are best prepared by the reaction of the acids and phthaloyl chloride.103 Non-functionally Labelled Carboxylic Acids.-These may be made by the above methods starting with suitably labelled alkyl halides etc. The 99 Ref. 1 p. 181; Ref. 2 p. 172. 100 E.g. Jorgenson Bassliam Calvin and Tolbert J . Anzer. Chem. Xoc. 1952 '74 101 Sakami Evans and Gurin ibid. 1947 69 1110. 2418. Ropp ibid. 1950 '72 2229.lo8 Cox and Turner J. 1950 3176. 418 Q CJARTERLY REVIEWS acetoacetic ester log and malonic ester Io5 syntheses and the reduction of keto-acids,lo6 may be applied when appropriate. Aldehydes and Ketones.-The Itosenmund reduction of acyl halides is the method of choice for the synthesis of isotopically labelled aldehyde~,~O~ although for benzaldehyde l08 the method of McFadyen and Stevens lo9 gives rather lower but more consistent yields. Several aldehydes e.g. [1-l4C 2-13C]-acetaldehyde 11* and -benzaldehyde ll1 have been prepared by oxidation of the corresponding alcohols. Several ketones including acetone,l12 [1-14C]cycZohexanone,113 and [ 1 -14C]cyclopentanone,71 have been synthesised by the pyrolysis of salts of carboxylic acids. The Friedel-Crafts reaction has been applied to the small- scale synthesis of a number of functionally labelled aralkyl ketones (yields 71-89%),114 including cyclic ketones.l15 The malonic 116 and acetoacetic ester 11' syntheses reaction of acyl halides with cadmium alkylsj116 and re- action of nitriles with Grignard reagents 117 have obviously a wide application.[2-14C]cycZoHexanone has been obtained in 25% yield by application of the Tiffeneau reaction t o l - a r n ~ i i ~ [ ~ ~ ~ ] m e t h y ~ ~ ~ c ~ o p e n t a n o ~ . ~ ~ ~ Arnstein and Bentley 119 have synthesised 1 3-dihydro~y[%~~C]acetone froin nitro[14C]- methane and formaldehyde. AZcohoZs and Amines.-By far the most satisfactory route to labelled primary alcohols lies in the reduction of acids acyl halides and esters with lithium aluminium hydridc,120 yields exceeding 95% being easily obtained in small-scale preparations.l03? 121 122 If high-pressure equipment is avail- able the hydrogenolysis of esters over copper chromite?lz3 or of the cadmium- nickel salts of acids,124 may be used.The alcohol chosen for esterification may be that formed in the reduction 123 or one of inuch higher or lower boil- ing point .I25 Secondary and tertiary alcohols have most frequently been obtained by the Grignard reaction. lo4 E.g. Coon and Abrahamson J . Biol. Chetn. 1962 195 805. lo5 E.g. Coon Abrahamson and Greeno ibid. 1052 199 75. loo Dauben J . Ainer. C'hern. SOC. 1948 78 1376. Io7 Ref. 2 p. 197. lo* Geissmann Univ. California Radiation Lab. Rep. 1233. loB J. 1936 584. 110 Ehrensvaard Reio Sduste and Stjernholm J . Biol. Chem. 1951 189 93. ll1 Douglass U.S.At. Energy Cornm. Rop. ORNL-1206. 112 E.g. Aronoff Haas and Fries Science 1949 118 476. 113 $peer Humphries and Robcrts J . Ainer. Chem. SOC. 1952 74 2443. 114 Spew and Jeans ihid. p. 2443. 115 E.g. Collins ibid. 1951 73 1038. 116 Dauben Reid Yankwich and Calvin ibitb. 1960 72 121. 117 Cerwonka Brown and Anderson ibid. 1953 75 28. 118 Arnold U.S. At,. Energy Comrn. Rep. AECU-575. 120 Brown " Organic Reactions " Vol. VI Wiley New York 1851 p. 469. 121 Pack and Tolbert. Univ. California Radiation Lab. Rep. 1957 ($opt. 1952). lZ2 Turner and Warne J . 1953 789. 123 Tolbert Christenson Chang and Sali J . Org. Chem. 19-19 14 585. 124 Adams Sslff and Tolbert J . Amer. Chem. XOC. 1952 74 2416. J. 1851 2385. Hauptman Adams and Tolbert ibid. p. 2423. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 419 [ 1 -l4C]Glycero1 has been prepared in 60% yield (from Bal4CO3) by the following series of reactions 126 Ba14C0 -3 Na14CN r Several routes to labelled glycerol have been explored.CH,Ph*O-CH,-CHO 1 Ac,O CH,Ph.O-CH,*CH( OH).' 4CN 4 2 HCI-EtOH LiAlH CH,P~*OCH,.CH(OAC).~~CO,E~ ____f Cat. hydrogn. CH,Ph*O*CH,-CH (OH) s 1 4C€I ,*OH * HO.CH,.CH( OH)* l4CH,*OH In a similar synthesis from glycollic aldehyde glycerol has been obtained in about 13% yield (from Nal4CN).lZ7 Both [1-14C]- and [2-14C]-glycerol have been made from the appropriately labelled malonic ester in an overall $eld of 30% (from Ba14C0,) 12* Pb( OAc) LiAlH CH,(CO,Et) + AcO-CH(CO,Et) --+ HO.CH(CH,.OH) Schlenk Lamp and DeHaas 129 have devised a synthesis for glycerides specifically labelled in the glycerol residue.Many alcohols have been converted into allcyE haEides by conventional methods. A number of amines has been prepared by standard methods. Methyl- amine ethylamine,l3O and aniline,95 are obtained in good yield from the appropriate acids by the Schmidt reaction. Other primary amines have been made by reduction of oximes l 3 l and nitrile~,~O and by the Gabriel reaction. 79 Secondary and tertiary ainines have been synthesirjed by alkylntion methods .132 OZeJins have been prepared by dehydration of alcohols 133 and by dehydro- halogenation of alkyl halides 85 but the thermal decomposition of quaternary ammonium hydroxides gives a more certain location of the double bond (e.g. the preparation of [1-14C]prop-l-ene.133 XaCurated hydrocarbons are made by reduction of ~ l e f i n ~ ~ ~ or from alkyl halides by reduction (Zn-Cu couple 134 or lithium aluminium hydride lS5) or by reaction of tho Grignard compound with water.85 In a few cases ketones have been reduced by the Wolff-Kishner method e.g.in the preparation of [14C,]cyclohexane.113 liydroxy-acids.-Some a-hydroxy-acids are conveniently prepared by the cyanohydrin synthesis. Thus [~arboxy-~~C]la,ctic acid is obtained in 94-96% yield from Na13CN and [ap-13C2]lactic acid in 40% yield from 1"Chem. Eng. News 1952 30 1872. 12' Doerschuk J. Amer. Chem. Xoc. 1951 73 821. 12* Gidez and Karnovsky ibid. 1952 74 2413. 1 2 9 Schlenk Lamp and DeHam ibid. p. 2550. 130 Pharos Arch. Biochern. 1951 33 173. 131 Wilson J. Amer. Phurm. Assoc. Sci. Edn. 1950 39 687. 13a E.g. Walz Fields and Gibbs J . Amer.Chem. Soc. 1951 73 2968. ls3 Fries and Calvin ibid. 1945 '90 2235. 134 Gordon and Heimel ibid. 1951 '93 2942. 136 Phillips Trevoy Jaques and Spinks Canad. J . Chem. 1952 30 844. 420 QUARTERLY REVIEWS Ba13C03 (via acetylene and acetaldehyde lol). Another method of wide application is the halogenation and subsequent hydrolysis of carboxylic acids ; glycollic and lactic acids have been made on the small scale in excellent yields in this manner.136 In a few cases it is advantageous to prepare and reduce a keto-acid e.g. inalic acid from oxaloncetic ester.100 Anderson and Rahman 137 have described a useful synthesis of [14C,]glycollic acid in which potassium [14C]carbonyl is prepared from [14C]carbon monoxide in liquid ammonia and then hydrolysed. A yield of 80% €rom Ba14C03 is claimed.Mandelic acid has been made as follows Ph*14CO*CH 4 Ph-l4CO*CHO ___+ Ph*l4CH(OH)*CO,H (7576 yield) It has been demonstrated by degradation that the carbon chain does not suffer rearrangement .13* Asymmetrically labelled citric acid has been prepared by reaction of (-)-p-carboxy-y-chlorobutyric acid with [14C]- SeO NaOH- cyanide followed by hydr01ysis.l~~ It was converted enzymically a-ketoglutaric acid labelled only in the y-carboxyl group.140 TABLE 2 . Arnino-acids into Amino-acid and position of label Glycine ; [ca,rb~x?y-~~C] . DL-Alanine ; [ca~bomy-~~c] ; [a-l4C) ; [%l4cC] . D- and L-Serine ; [/3-14C ~ f i - ~ H ~ 15N] . DI,-Aspartic acid ; [4-14C] . DL-Aspartic acid ; [3-13C 4-14C] . L-Threonine ; [y-llC 15N] . ~-Threonine ; la-l4CC] . ~ ~ - V a l i i i ~ ; [cr-14C] .DL-Valine ; Lfi-lW] . L-Valine ; [y-13C] . nL-Ornithino ; [a-14C] . D- and r>-Glutarnic acid ; [l 2-14C,] . DL-Glutamic acid ; [5-"Cc] . L-Histidine ; [~arboxy-~~C] . DL-Leucine ; [ca~boxy-~~C] ; [C~-~"C] . L-Leucine ; [carbo~y-l~C] . DL-isoLencine ; [Me-14C] . DL-Lysjne ; [cr-14C] . D- and L-Lysine ; [ M - ~ ~ C ] . D- and L-Phenylalanine ; [a-I3C Ph-14C1] . DL-PhenyIalanine ; [cccrbo~y-~~c] . DL-Tyrosine ; [ c ~ r b o s y - ~ ~ c ] . DL-ni-iodotyrosine ; [ c a r b o ~ y - ~ ~ c ] . Iteference and method 1 4 1 ; a 142; a 143; b ( 2 ) 144; h ( 2 ) 110; b ( 2 ) 145 ; B ( 1 ) 146 ; b(1) 147 c 148; cl 149; CL 93 150; c 150; b(2) 151; c 125; u 152; d 153; cl 98 ; b ( 1 ) !I8 ; b(1) 1 4 ; b ( 1 ) 154; c 155; d 156; d 156 Methods a via the a-halogeno-acid ; b ( l ) synthesis using labelled acylamitio- malonate -cyanoacetate or -acetoacetate ; b ( 2 ) ditto other reactant lnbolled ; c hippuric acid synthesis ; d Strecker or hydantoin synthesis ; e reductive amination of keto-acid._ _ _ _ _ _ _ _ _ _ _ _ _ _ ~ - ~ _ _ ~- _ ~ _ _ _ _ _ ~~ - - ~~ . -~ ~ 136 Hughes Ostwald and Tolbert J . Amer. Chem. SOC. 1952 74 3434. 187 Anderson and Rahman Brookhaven National Lab. Rep. 103. 1x3 Brown and Neville quoted in " Isotopic Carbon " (ref. 21 p. 214. 139 Wilcox Heidelberger and Potter J . Amer. Chem. SOC. 1950 72 5019. 1*0 Ref. 1 pp. 191-194. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 421 Amino-acids (see Table 2).-[cc-13C]Glycine has been prepared by an interesting method 157 K13CN + CS(NHPh) _____f N13C*C(NHPh):NPh PbCOs LiAlH Hydrol.NH2*13CH2*C(NHPh):NPh - NH,*13CH2*C02H The valuable method of partial degradation and resynthesis has been Arnstein used to introduce I4C at C(2) of the glyoxaline ring of ~-histidine.l~~ et ~ 1 . ~ ~ have converted 14C02 into [carbo~y-~*C]Iysine in 12% yield (>". 14C02+ 0 1 W 0 2 H A Me ester 0 0 1 HX,; 2 hydrol. HN H02CfCH2],*C13(NH2)*14C02H + NH,*[CH,],*CH(NH2).14C0,H Keto-acids.-A numbar of labelled keto-acids has been prepared usually [j9-14C]Acetoacetic by the application of standard methods (see Table 3). ester has been made as follows 167 C0,Et C0,Et / -3 CH,-l4CO*CH \ Mg / 2 \ CH,*14COC1 -i- A-CH CO,But CO,But C,H,*SO,H in boiling xylcne f CH,*14CO*CH,*C0,Et (7 1 yo) 141 Uloch J . Biol. Chem. 1949 179 1245. 148 Ostwald Adams and Tolbert J . Amer. Chem. Xoc. 1952 74 2425.143 Elwyn and Sprinson J . Biol. Chem. 1950 184 465. 144 Wsng Winnick and Hummel J . Amer. Chem. Soc. 1951 73 2390. 145 Meltzer and Sprinson J . Biol. Chern. 1952 197 461. 146 Krasna Peyser and Sprinson ibid. 1952 198 421. 14' Adams and Tolbert J . Amer. Chem. SOC. 1952 74 6 8 i 2 . laS Anatol Compt. reizd. 1950 230 1471. 149 Fones Waalkes and White Arch. Biochem. 1951 32 89. 150 Speer Roberts Maloney and Mahler J . Amer. Chem. Soc. 1952 74 2444. 151 Borsoolr Deasy Haagen-Smit Keighley aid Lowy J . B i d . Chem. 1952 196 153 Cerisia Jenkins and Degering J . Amer. Phccrm. Assoc. Sci. Erln. 1951 40 341. 154 Lerner J . Biol. Chem. 1949 181 281. 155 Henneberry Oliver arid Baker Camd. J . Chem. 1951. 29 229. 156 Loftfield J . Amsr. Chenz. Soc. 1950 72 2499. 157 Ehrensvaard and Stjernholm a c t o Chem.Scund. 1949 3 971. 158 Borsook Deasy Haagen-Smit Keighley and Lowy J . Bid. C'hem. 1950. 187 839. 159 Anker ibid. 1945 176 1337. 160 Thomas Wang and Christensen J . Amer. Chem. Soc. 1951 73 5914. 161 Curran J. Bio2. Che92. 1951 191 775. 162 Crandall Brady and Gurin ibid. 1949 181 845. 163 Dauben J . Amer. Chem. Xoc. 1948 70 1376. 164 Weinman Chaikoff Dauben Gee and Entenman J . Biol. Chem. 1950 184 735. 165 Weinman Chaikoff Stevens and Dauben ibid. 1951 191 523. le6 Heidelberger and Hurlbert J . Amr. Chem. Soc. 1050 72 4704. 167 Dauben and Rradlow ibid. 1952 74 5204. 669. lS2 I d e m ibid. 1950 184 549. 422 QUARTERLY REVIEWS tert.-Butyl esters have also been used in the preparation of oxaloacetic acid which may be obtained in 90% yield from the tert.-butyl ester by the above technique.166 TABLE 3.Keto-acids Keto-acid and position of label Pyruvic acid [~arboxy-~~C] . Pyruvic acid [wl*C] . Acetoacetic acid [c~rboxy-~~C] ; [P-W] ; [carbo.ry P-lsC,] . Acetoacetic acid [,8-l4C] . CH3*14CH,*COfCH2],*C02Et . n-C,,H2,~14CH2~CO~[CH,I,.C02EI . n-C,H,,.1*CH,.CO*[CH2J~.C~2H . n-C,,H2,.1~CH,.C0.[CH,I,.C0,H 14CO2H.CH2*CO*C0,H . C0,Et.14CH,CO*C02Et . 14C02H.14CO*[CH,],.C0,H . Reference and method (see p. 421) Methods a via acetyl cyanide ; b acetoacotic ester condensation ; c via CH3*C0.CH(14CO*CH3)*C0,Et ; d cadmium alkyl and COCl*[CH,],,*CO,R. Ring-labelled Aroma.tic Compounds.-Benzene derivatives have been prepared in which the isotope (a) is uniformly distributed within the ring or (b) bears a definite orientation to a substituent. Compounds of the &st kind e.g.[~r-~~C,]benzaldehyde,~~~ may be made from benzene itself which has been synthesised directly by two methods. In the first of these,lf3 [ 1 -1W]cychhexanone (4.v.) is reduced to cychhexane and dehydrogenated to [14Cl]benzene (22:& yield from Baf4C0,). The second 122 gives a 75% yield from 14C02 14CC0,H lWH,*QH 14CCH n - 0 LiAlH -...LFtJ-() - HZO Several useful specifically labelled benzene derivatives have been prepared in rather low yields 16s CH OH CH3 --) 6 KMnO4* 6 CO,H [l-14C]Aniline 96 [l-14C]Phenol Q6 [ 1 - 14C]Chlorobenzene 169 BrMg .[ CH,] ,.MgBr CH,-14C02Et -+ 0 + { 24-37 yo from CO A potentially valuable synthesis of [l 2-14C2]benzoic acid via [or/?-14C2]- 188 Fields Leaffer Rothchild and Rohan J . Amer. Ghem. Xoc. 1952 74 5498. 169 Fields Gibbs and Walz Science 1960 112 591.THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 423 maleic anhydride 170 has been outlined in a review article 171 Most polycyclic compounds have been made from [carbo~y-1~C]-aroic acids by intramolecular acylation (e.g. 2-methyl[4-l4C]napbtha-l 4- quinone 172 and [9-14C]anthracene 173) by application of the Wagner re- arrangement (e.g. 1 2-benz[3 4-14C,]anthracene 174 and [5 6-l4C1]chry- sene 175) or by means of the Elbs reaction (e.g. 20-methyl[ll-14C]cholan- threne 176). It has also been found possible to prepare naphthalene and a-naphthol containing 1% by irradiation of quinoline oxalate in a nuclear react0r.~5 Many ring-labelled heterocycEic compounds have been synthesised usually by adaptations of standard methods ; purines and pyrimidines have received most attention.Steroids,1*7-Partial syntheses of several biologically important steroids Some of these are listed in Table 4. 170 Nyst'rom Loo and Leak J . Amer. Chem. SOC. 1952 74 3434. 1 7 1 Nystrom Loo Mann and Allen quoted in ref. 4. 1 7 2 Liang-Li and Elliott J . 97ner. Chem. SOC. 1952 74 4089. 173 Stevens and Holland Science 1950 112 718. 174 Collins Burr and Hess. J . dmer. Chem. SOC. 1951 73 5176. 175 Toffel Jones and Co!lins ibid. 1053 75 307. 176 Mart,in and Baker U.S. At. Energy Conim. File No. NP-3177 177 Bennett J. Amer. Chem. SOC. 1952 74 2432. 178 Mandel and Brown ibid. p. 2439. 179 Abrams and Clark ibid. 1951 73 4609. 180 Weygand and Grossinsky Chem. Ber. 1051 $4 839. l 8 1 Bennett Skipper Mitchell and Sugiura Cancer Res. 1950 10 644.182 Bentley and ATcubarger Biochem. J. 1953 52 694. lS3 Miller Gurin and M7ilson J . Amer. Chem. SOC. 1952 74 2892. 184 Anker and Boehne ibid. p. 2431. 185 Weygand Mann and Simon Chem. Ber. 1952 $5 463. 186 Williams and Ronzio J . Amer. Chem. h'oc. 1952 74 2409. 188 Turner J. Amer. Chem. SOC. 1950 72 579. 189 Fujimoto ibid. 1951 73 1856 ; Heard and Ziegler ibid. p. 4036. Twombly Vitamins and hTormones 1951 9 237. 424 Q'IJARTERLY BEVIEWS TABLE 4. Heterocyclic compounds I Cclnipound ~ [2-14C]Uracil . [ 2-14C]Thymine [6-l4C]0rotic acid . [4 6-l4C,]Adenine . [8J4C]Adenine . 8-Aza[4 6-14C,]adenine . [2-14C]Guanine . [4-14C]Guanine . [8-14C]Guanine . 8-Aza[ 2-1 4C]guaiiine . 8 -Aza[ 4-14C]guanine . 2 6-Diarnino[2-l1C]purino . [6-1*C]Uric acid . .l-Hydroxy[ 4 cnrboxy -l 4C,]glyoxal ine -5- [6 7-14C,]Sant,hopterin .[2J4C]Folic acid . [9-14C]Folic acid . [14C]Thiamine (vitamin B,) . carhox yamide Ref. ; yield from BaCO (see p. 423) 177 (32-40%) ; 178 (600/,") 177 (2@-28%) 166 (38%) 97 (17%) 179 ( 5 5 x b ) 97 (1776) 177 (4@-50%) 180 (70-75%) 97 (38%) 97 (43%) 181 (8%) 177 (15-20%) 182 183 184 (5%) 185 (1*8yo) 82 (3*65:/0c) 186 a From [l*C]urea. From sodium [1*C]formate. From [14C]metfhaiiol. have been devised. One synthesis of ring A-labelled steroids is as follows lS8 2 1 CH,*CO,Ph (111) 10-20 yo (11) 80-90 yo (Not separated) [3-14C]- and [4-14C]-Cholest-4-en-3-one (R = *CHMe*[CH,],*CHMe,) and testosterone (R = OH) have been made in this way. Better yields are obtained by reaction of (I) with [14C]methylmagne~ium iodide followed by cyclisation.189 Alternatively the isotope may be introduced in inethyl brorn~[carboxy-~~C]acetate by Reformatsky reaction with the keto-ester (IV) O W Mc0,C + THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 425 Cholestenone,lss progesterone and deoxycorticosterone acetate l90 have been prepared by this method.Ring-labelled cholesterol may be made by reduction of cholestenone with sodium-borohydride .lgl Herschberg et aZ.lg2 ring D by the following have labelled dehydroisoandrosterone acetate in procedure 1 l3CH,NI ; 2 Amdt-Eistort ; 3 KOH 0 Steroids labelled in the side chain so far prepared include [26-14C]- cholesterol lg3 and [211-14C]progesterone.194 Curbohydrates.-Carbohydrates labelled in the l-position have been made by the Fischer-Kiliani method (e.g. n-galactose,lg5 D-glUCOSe and D-man- nose lg6) and also by the nitromethane method of Sowden and Pischer l 9 7 (e.g.L-arabinose and ~ - r i b o s e l ~ ~ D-glucose and D-mannOSe 199). Sowden 2o0 has prepared ~-[6-~~C]glucose by the following method CEIO H O-CMe H (V) N OH *CH,*OH \ r/;’O\ H NnRH. H //-o\ H OH O-CMe ln0 Fujimoto J . Amer. Chem. Soc. 1950 72 4328. lnl E.g. Dauben and Eastham ibid. 1951 73 4463. lQ2 Herschberg Schwenk and Stahl Arch. Biochem. 1948 19 300. ID3 Ryer Gebert and Murill J . Amer. Chem. SOC. 1950 72 424’7. ID4 Riepel and Prout J. Org. Chem. 1948 13 933. l n 5 Topper and Stetten J . BzoZ. Chem. 1951 193 149. ln6 Isbell Karabinos Frush Kolt and Schwebel J. Res. Nut. Bur. Stand. 1952 lQ7 Sowden and Fischer J. Amer. Chem. SOC. 1947 69 1963. 48 163. Rappoport and Hassid ibid.1951 73 5524. Idem J . Amer. Chem. SOL 1952 74 4377. lgQ Sowden J . Biol. Chem. 1949 180 55. 426 QUARTERLY REVIEWS Both D- and L-[l-14CJascorbic acid have been synthesised by the osoiie method.201 Biological Syntheses.-Biosyrithetic methods are particularly well suited to the preparation of natural products many of which it is difficult or im- possible to synthesise by chemical means. Wholly specific labelling is not easily achieved but for many purposes is unnecessary. The syntheses frequently start from very simple intermediates and optically active com- pounds are obtained in the natural configuration. Dilution of isotope is usually greater than in chemical syntheses but may in favourable cases be kept small or substantially avoided. Yields vary widcly e.g. froin up to 7(!% in the photosynthctic preparation of sucrose 202 to an estimated yield of less than O.OOS~o 203 in the prepa.ration of digitoxin in nigiiglis pwpurea.204 Microbioloyical Methods.-These have been used extensively and fre- quently give high yields. Autotrophic bacteria grown on I4CO2 invariably produce uniformly labelled compounds but there are many partial syntheses brought about by micro-organisms in which control of labelling may be exercised. The autotrophic bacterium Thiobacillus thiooxidans has been employed 205 in the production of bacterial protein and derived amino-acids from WO,. About lfi?( of the 14C was recovered as separated uniformly-labelled amino- acids ; dilupliuii was very slight. Better radiochemical yields have been obtained 2oG by using Rhodospirillum rubrum which assimilates equimolar quantitics of carbonate and ethanol.The three-fold dilution of activity is relatively unimportant but thc non-uniform labelling to be expected is a more scrims disadvantage. Some syntheses by which specifically labelled compounds can be made are listled in Table 5. Among more complex compounds prepared microbiologically may be mentioned streptomycin 213 which has been isolated from cultures of 8. griseus grown on uniformly labelled glucosc. Photosynthetic ~~etliods.-Photosynthesis in whole plants is extremely inefficient for preparative purposes ; nevertheless a number of drugs and complex natural products have of necessity been prepared in this may 201 Hamilton and Smith J . Amer. C ~ P I I . Soc. 1952 74 5162 ; Buriis nricl King ,Ycior?ce 1950 111 257.Z02 Scully Stavely Skok Stanley Dale Craig Rodge Chorney Watanabt and 203 Ref. 6 p. 272. m4 Geiling K~lsey Slclntosh and Gaw Sciencs 1848 108 558. 205 Franlz Feigelman Werner and Smythe J. Biol. Chem. 1952 195 423. 206 Turver Ta,bar?hnik Canellakis Fraser and Barker Arch. Uiockem. 1952 41 1. 207 Sun Pirtro J . Bid. Chem. 1952 198 630. 208 Ref. 2 p. 274. Ref. 2 p. 276. 210 Isbell arid Karabinos J . Res. A7at. Bur. Stund. 1952 48 438. zll Foster Carson Anthony Davis Jefferson and Long Proc. Nat. Accd. Sci. Baldwin ibicl. 1952 116 87. 1849. 35 663. i l j l and Karnen J. Amer. Chem Soc. 1951 73 2349. 213 Baroa Peck Rosenblum mid Woodbury ibid. 1953 7'4 3056. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANlC COMPOUNDS 427 TABLE 5 . ,Microbiological syntheses Compound [a/3-l4C1 15N]Asparl.ic acid n- [ay- 4C,]Butyric acid (ctc .) n-[3-14C]Hexanoic acid (etc.) ~ - [ 1 6-14C,]Fructose [curboxy-14C,]Fumaric acid [ ctp -l 4C J Succinic acid Precursor and organism [orp-14C,]Funiaric acid Escherkhin coli [Me-14C]Acotic acid (etc.) ; B.rettgeri n-[~arbozll-~~C]B~~~yric acid (etc.) ; CI. Kluyveri D -1 1 -14C]Mannitol ; Acetobacter suboxyduns [ l-1.4C]Ethanol ; Rhizopus nzgricans [Me-14C]Acetic acid ; E'sche?*ichicc coli Yield ; dilution 47% ; small 60-70% ; small 80% ; -23 yo 54% ; small 40-60~0 ; -40% 37% ; -10 x 207 308 209 210 21 1 212 including colchicine 214 digitoxin 2r)4 morphine 215 nicotine n6 and pyre- t h r i n ~ . ~ ~ ' Further tobacco mosaic virus has been labelled 218 with 14C by growing infected plants in 14CQ2. Photosynthesis in detached leaves how- ever affords an efficient means for thc preparation of some important carbohydrates.Using the leaves of Canna indica Putrnan and -Elassid l7 have obtained very pure glucose fructose and sucrose of high specific activity in an aggregate yield of - 70%. Bean leaves appear t o be the most suitable for the biosyiithesis of starch ; 219 dilution is negligible and yields of 18-36y0 of purified starch have been obtained.21g 220 AnimuZ Biosyntheses.-In favourable circumstances relatively high yields of particular substances may be obtained afier the assimilation of suitably labelled precursors. Karlsson and Barker 221 have obtained recoveries of 14C in uric acid excreted by pigeons injected with formate [~-1~CC]glycine and [~arboxy-~~C]glycine of 40 61 and 17% respectively. In the first case > 98% of the 14C1 was found in C(2) + of the uric acid and in the last 87% was in C(*).Similarly after feeding of [8-14C 1 3-1WJadenine to rats some 35% of isotope fed was isolated from the tissues in six separated ribonucleotides ; adenylic acids a and b together accounted for 70% of the recovered isotope.222 In most cases however the bulk of the isotope is distributed generally in the animal largely diluted with normal carbon. Thus feeding with glucose or lactatc and simultaneously injecting [14C]- bicarbonate into fasted rats btts the result that the glycogen laid down in the liver may contain up to - 2.5% of the 14C administered ; the glucose 214 Walaszek Kelsey and Geiling Scie?icc 1952 116 225. 215 McIntosh Kelsey and Geiling J . Amer. Phurrn. Assoc. Sri. E d n .1359 39 216 Gmz Kelsey and Qeiling Ho!. Qaz. 1951 113 195. 217 Pellegrihi Millcr ant1 Sharpless J . I k o n . BJntonaoZ. 1952. 45 532. 218 Schonfellinger arid Broda i?~owtsk. 1352 $3 837. 219 Livirrgston and Medes J. Ocn. Picysiol. 1947 31 7.5. 220 Gibbs Dumrose and Achcr U.S. At. Ensrgy Comm. Rep. AKCU-283. 221 .J. Biol. Chem. 1940 177 597. 232 Mmrian Epicor Balk and Brown ibid. 1951 189 533. 512. 428 QUARTERLY REVIEWS obtained by hydrolysis has - 97% of its 14C content in and C(4).223 Glutathione has been prepared from the liver of a rabbit previously injected with [14C]bicarbonate.224 The yield was about 0.1% and it was shown by enzymic degradation that - 53% of the activity was in the carboxyl group of the glutamic acid residue. Other substances synthesised in animals for biological studies include glucuronic acid 225 plasma phospholipids 226 squalene,227 and bufagin.228 Syntheses in isolated animal tissues appear to have many advantages over the use of whole animals but have been comparatively little used.Anfinson has prepared crystalline [14C]ovalbun~in 229 and [14C]ribonu- clease 230 by incubation of hen oviduct minces and bovine pancreas slices respectively with 14C02 and labelled haemin has been prepared by incubation of duck blood with precursors such as acetate and gly~ine.~31 Brady and Gurin 232 have shown that labelled cholesterol may be obtained in up t o 17 yo yield from [Me-14C]acetate in rat-liver slices. Enzymic 8yntheses.-These have had a rather limited application in spite of their great specificity and efficiency. By the use of sucrose phosphorylase sucrose labelled in the fructose or glucose residue has been prepared.233 A similar procedure using maltose phosphorylase which catalyses the reaction Maltose + Jnorganic phosphate + p-D-Glucose-1 phosphate + D -Glucose has been used for preparation of maltme labelled in either the reducing or the non-reducing glucose residue.234 The glucose-I phosphate was itself prepared from labelled starch by use of potato phosphorylase.233 Other compounds prepared eiizymically include L- [~arboxy-~*C]malic acid 235 and oxaloacetic acid. 236 Enzymic methods for the resolution of racemic amino- acids 237 are well suited for use with isotopically labelled compounds. 238 Degradation of Labelled Compounds.-It is often necessary to locate the isotope in an organic compound and degradation plays an important part in many biological and chemical tracer studies.The degradation must be designed so as to permit the isolation of specific fragments of the molecule in a form suitable for isotopic analysis. Most chemical syntheses lead t o unambiguously labelled compounds but in some cases unexpected molecular 223 Shreeve Feil Lorber and ?Vood J . Biol. CAem 1949 177 679. 2 2 4 Krimsky and Racker ibid. 1952 198 721. 226 Packham and Butler ibid. 1952 194 349. 226 Weinmann Chaikoff Enteaman and Dauben ibid. 1950 187 643. 227 Langdon and Bloch J . Anzcr. Chem. SOC. 1952 74 1S69. 228 Doull Dubois and Geiling Arch. int. Phurmucodyn. 1951 86 454. 229 Anfinson and Steinberg J . Biol. Chem. 1951 189 739. 230 Anfinson ibid. 1950 185 827. 23l E.g. Shemin and Wittenberg ibid.1951 192 315. 232 Brady and Gurin ibid. 1951 189 371. 233 Wolochow Putman Doudoroff Hassid and Barker ibid. 1949 180 1237. 2 3 4 Fitting and Putman ibid. 1952 199 573. 235 Kaufman Korkes and del Campillo ibid. 1951 192 301. 236 Lorber Utter Rudney and Cook ibid. 1950 185 689. 237 See e.g. Birnbaum Levintow Kingsley and Greensbein ibid. 1952 194 465. 238 E.g. Hassan and Greenberg Arch. Biochem. 1952 39 129. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 429 rearrangements occur and these are revealed only on degradation. In biological syntheses degradation of the product is essential except for those applications (e.g. isotopic dilution analysis) where the molecule as a whole is to be traced. The methods used must themselves bE proved on unequi- vocally labelled reference compounds or confirmed by alternative methods.Leete et al. 239 converted biosynthetic N-methyltyramine by methylation and Hofmann degradation into p-methoxystyrene which was then con- verted into p-anisic acid C0,H CH=14CH HgO-1 ____ -Et,O '4C0 3- 0 4= 0 OM0 OMe 14CH,*CH0 I4CO,H Direct oxidation revealed a molecular rearrangement which vitiated the original procedure. Differences in reaction rates caused by isotopic substitution (the isotope effect 11) may interfere with degradation procedures. Two types of reactions may be distinguished. I n those where the relevant labelled atom can be present a t the end of the reaction only in a single compound an isotope effect can in any case only be detected during the reaction and is avoided by carrying the reaction to completion.On the other hand when the labelled atom is distributed between two or more compounds an isotope effect if it occurs cannot be avoided in this way but appropriate correc- tions can be made. Reactions of the first type include decarboxylations combustions to carbon dioxide (e.g. acetic acid urea xanthhydrol ureide) and the absorption of carbon dioxide by alkali. Incomplete combustions will also give false results for more obvious reasons The second type is exemplified by the iodoform reaction on [ I-14C]acetone. Intermediates in degradation procedures are commonly isolated and purified as derivatives. Isotope effects have been observed in the reaction of benzophenone with 2 4-dinitrophenylhydrazone 240 and of formaldehyde with dimedone 241 both reactions falling into the first category.Some of the most common end products of degradation procedures are difficult to purify on the very small scale. Iodoform is a very important example and here a specific oxidation procedure has been devised.242 As an example of the scope of degradative methods two recent partial degradations of biosynthetic cholesterol may be mentioned. Cornforth Hunter and PopjAk 243 have isolated all the carbon atoms of ring A 239 Leete Kirkwoocl arid Rfarion Canad. J. Ckem. 1952 30 749. 240 Brown and Holland ibid. p. 438. 241 Dowiies Austral. J . S c i . Res. 1952 5 A 521. 2 4 2 Shreeve Leaver and Siegel J . Amer. Chem. SOC. 1952 '94 2404. 233 Cornforth Hunter and PopjAk Biochcm. J. 1953 54 590. QUARTERLY REXIX%VS 1 HN ; CH3*CH(NH,).[CH,],.C0,H 2 hydrol. NH2.[CH,],I.CH-(CH,).C0,H ----+ { 79 10 1 1 MeI-MeOH ; 2 KOK ; 150-200" ; 3 KOH; 300" 2 3 4 5 C H3*CH,*CH,.CH2C0,1;I + CH,*CO,H CH,*CH,*CH,*CO,H + CH3*CH,*C0,H The four acids were separated by partition chroinatography and further degraded.244 Wiiersch Huang and Bloch 245 converted chdesterol into dihydrocholesteryl acetate which was oxidised to 3~-hyclroxyaZEocholanic acid and acetone (derived from C(zs) C(zs) and C(2,)). Rlepeated application of the Barbier-Wieland degradation procedure to the acid permieted the isolation of all the carbon atoms of the side chain. 4 3 2 1 19 10 6 -!- Ph,CO etc. Nitrogen There is only one isotope of nitrogen which is suitable for tracer studies. This is the stable 15N which is readily available as nitrate (and hence nitrite) or as phthalimide of about 60 atoms % excess of 15N.It is usually pre- pared 62* 246 from ammonia enriched from the natural abundance of 0.38% in some suitable fractionating device 15NN13[3(gas) + 14NHs(so~ution) S l4SH3(gns) + "NH&olution> Stimulated by the biological importance of compounds such as proteins amino-acids nucleic acids etc. the essential synthetic chemistry of the isotope was largely established several years ago and was reported by 2d4 Hunter and Popj$lc Biochem. J . 1951 SO 163. 245 Wiicrsch I h n n g and %loch J . Eiol. Chesn. 1952 195 439. 2 4 6 Wilson " Preparation and Measurement of Isotopic Tracers " Edwards Ann Arboi- Mich. 2046. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 431 Arnstein and Bentley 1 and others.' The following notes supplement these surveys. Intermediates.-Clusius,247 by operating a thermal diffusion column with atomic nitrogen produced by an electric discharge has obtained 994% 15N2 and has converted this gas into nitrous and nitric acids.Hydroxylamine has been obtained 248 by electrolytic reduction of nitric acid. The synthesis of urea from ammonia and diphenyl carbonate has given rise to explosions and alternative procedures to avoid this phenomenon have been developed. 249 Amino-acids.-/l-AZanine 250 and anthranilic acid 251 (and hence trypto- phan) have been prepared from labelled phthalimide the former by condensa- tion with acrylic acid the latter by the Gabriel method which has also been utilised for the synthesis of [o(-15N]- and [6-15N]-ornithine in both L- and DL-forms. 252 Condensation of ornithine with labelled urea yielded citrulline degradation establishing that 15N was present only in the terminal group.Threonine and aZZothreonine have been prepared,l4j? 253 both by direct amination of the corresponding a-bromo-P-methoxybutyric acid and after the following reactions CHZ-COZH Me*C=C-CO Me*CO-CHCO,Et I I I I N H Ac,O- H O N 0 EtOH I NH*COPh H -+ \ / - -&GiL+ C \ co bh bh soc1 I I HA IkHCOPh --+ 0 N Me*CH-CH*CO,Et Me.CH-CH.C 0 2E t Me.CH.CH*CO,H - HI) A€€ \ / C Ph I A synthesis of considerable potential value14 has been applied to the preparation of ~-[or-~5N]lysine. Ph-CO-NH*[CH,],*CH(NHz)*CO,H NOBr D HCl L L Ph°CO~NH*[CHz],~CH(1sNH2)*C0,H + NHz~[CH,],~CH(15NH2)~C0,247 Clusius Helv. Chim. Actu 1950 33 2122 2134; 1952 35 1103. 248 Farago and Roberson Abs. Amer. Chem. SOC. 122nd Mtg. 1952 4 1 ~ .24D Buzard and Bishop J . Amer. Chem. SOC. 1952 74 2925 ; Williams and Ronzio 2Ko Graff and Hobennann J . Biol. Chem. 1950 188 369. 251 Partridge Bonner and Yanofsky ibicE. 1952 194 269. a 5 2 Stetten. ibid. 1961 189 499 ; Hirs and Rittenberg ibid. 1950 186 429. 2s3 Shulgin Lien Gal and Greenberg J . Amer. Chem. Xoc. 1952 '94 2427 ibid. p. 2407. F F 432 QUARTERLY REVIEWS ~~-[l~N]Valine 254 and ~-[~~N]glutamic acid 255 have been synthesised by the Knoop technique of reductive amination and there have been biosyn- theses of aspartic acid 256 (from [2 3-14C2]fumaric acid and ammonia by Escherichia coli) glutamine 257 (from red beets) and tryptophan 251 (from a yeast supplied with [15N]anthranilic acid) and a whole range of labelled amino-acids has been separated on the large scale by ion-exchange chroma- tography.258 Purines Pyrimidines etc.-The majority of recent synthetic work in this field has been concerned with 14C but improved syntheses of cytosine 259 and 2[15N] 4-&amino[l 3-15N2]pyrimidine 260 avoid the formation of undesirable by-products by utilising an alkoxide-catalysed condensation of isotopic urea with cyanoacetaldehyde diethyl acetal NaOBu _I__ CN \ \ / 15NH CH I I CO CH 15N H+ _j.A synthesis of [l 3-15N2]uric acid 261 via 5-nitrobarbituric acid has been carried out and the oxidative degradation of this and other compounds of the group has been studied.262 2[15N] 6-Diamino[l 3J5N2]purine 263 has been prepared from labelled guanidine in excellent yield CN H 264 Behrens et al. J. Biol. Chem. 1948 175 765. 255 Barker Hughes and Young J.1951 3047. 256 Wu and Rittenberg J. Biol. Chenz. 1949 179 847. 257 Hood Lyman and Tatum Arch. Biochern. 1951 30 351. 268 Aqvist Acta Chern. Xcand. 1951 5 1031. 260 Bendich Gsttler and Brown J. Biol. Chem. 1949 177 565. 260 Bendich Germ and Brown ibid. 1950 185 435. 261 Benedict Forsham and Stetten ibid. 1949 181 183. 2 6 2 Cavalieri and Brown J . Amer. Chem. Xoc. 1948 70 1242 ; Cavalieri Tinker 263 Bendich Furst and Brown J. Biol. Chem. 1950 185 423. and Brown ibid. 1949 71 3973. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 433 [l 3-15N2]Hypoxanthine 264 has been synthesised from thiourea by standard methods and by deamination of adenine. Deamination of guanine similarly yielded [l 3J5N2]xanthine. Biosynthesis 265 of several members of this group of compounds has been carried out with bacteria,266 yeasts,267 and tissue,222 and with the intact animal.268 MisceE2aneous.-Studies of reaction mechanisms involving nitrogen have necessitated syntheses of several aromatic derivatives including phenyl- [P-l5N]hydrazine 269 phenyl azide 270 3 5-dinitrobenzazide 27 diazoamino- benzene 271 and all three singly labelled p-dimethylaminoazobenzenes. 273 Nembutal 274 and the carcinogen 2-[15N]acetamidofluorene 275 have been prepared. Biosynthesis of several porphyrins following assimilation of singly and doubly labelled glycine have been carried out 276 and followed by elegant degradations to show the origin of the various atoms. Pr~digiosin,~’~ a tripyrrylmethane pigment of bacterial origin has been similarly studied. Other biosyntheses include those of stercobilin 278 and nicotinamide 279 (from indole by a selected strain of Neurospora).Oxygen Since the radioactive isotopes of oxygen ( 1 4 0 150 190) have very short half-lives they are unsuited to tracer work and it has been necessary to utilise the stable 1 8 0 and 1 7 0 which occur in atmospheric oxygen to the extent of 0.20 and 0.04% respectively.280 Arnstein and Bentley discuss the problem of the mass-spectrometric assay of 1 8 0 and refer t o the latter author’s review of the subject.281 More recently Dole has also contributed a comprehensive review 282 which contains an extensive bibliography. It is not practicable to separate180 by electrolysis of water but pro- cesses of (1) fractional distillation of water or carbon mon- oxide,61 (2) thermal diffusion or (3) chemical exchange 284 have been 264 Gettler Roll Tinker and Brown J .Bwl. Chern. 1949 178 259. 266Ref. 46; Wilson op. cit. p. 152; Brown op. cit. p. 164. 266 Reichart and Estborn J . Biol. Chem. 1952 188 839. 267 Di Carlo Schultz Roll and Brown ibid. 1949 180 329 333. 268 Bendich Brown Phillips and Thiersch ibid. 1950 183 267. 269 Ref 247 p. 2122. 270 Clusius and Weisser Helv. Chim. Acta 1952 35 1548. 272 Bothner-By and Friedman J . Arner. Chern. Soc. 1951 73 5391. 273 Fones and White Arch. Biochern. 1949 20 118. 274Van Dyke Scudi and Tabern J . Pharmacol. 1947 90 364. 275 Argus and Ray Cancer Res. 1951 11 423. 276 Ref. 46 ; Shemin and Wittsnberg op. cit. p. 41. 277 Hubbard and Rimington Biochem. J . 1951 46 220. 278 London J . Biol. Chem. 1950 184 373. 27g Bonner and Wasserman ibid.1950 185 69. 280 Nier Phys. Review 1950 7’7 789. 281 Bentley Nucleonics 1948 2 (2) 18 ; cf. ref. 3. 282 Dole Chem. Reviews 1952 51 263. 28s Dostrovsky Hughes and Llewellyn Bull. Res. Counc. Israel 1951 1 133. 284 Boyd and White I n d . Eng. Chern. 1952 44 2202. 271 Idem ibid. p. 1624. 434 QUARTERLY REVIEWS ljuccessfully employed. Clusius by operating a series of six 14-m. diffusion units for many months prepared 250 ml. of 99% 1802 but' the isotope is usually supplied as H2180 of much lower enrichment from which oxygen may be generated 285 as required. It can be assayed1 281 282 in carbon dioxide or oxygen by the mass spectrometer or by precise density deter- minations on derived water. The main field of application 281 282 has been to the study of reaction mechanisms (e.g.ester 286 and lactone 287 hydroly- sis) but it is also useful for the assay of oxygen in organic particularly in those substances such as fluoro-compounds which do not normally yield accurate results by established methods ; other applications have been to geochemistry 289 (e.g. for the determination of paleo-tempera- tures) and to metabolic 290 and photosynthetic studies although in a recent paper 291 the validity of certain fundamental premises in this work has been called in question. It is also of interest that the 1 8 0 content of atmospheric oxygen is higher than would be expected if it were in equilibrium with the oxygen of natural waters. Where possible the isotope is incorporated into organic compounds by a process of exchange with H,lsO. Tabulated data which have been pre- sented by Bentley and by Dole indicate that carbonyl compounds generally show a ready exchange particularly when this is catalysed by acid or alkali.The carboxylic acids too may undergo exchange although it is suppressed by the presence of alkali when the acids are in anionic form. Labelled alcohols cannot be prepared by such methods unless powerful labilising groups are present in the molecule and it is therefore necessary t o employ the conventional methods of halide hydrolysis. Similarly phenols must be prepared by fusion of the corresponding sulphonic acid with NalsOH. The oxygen of amides peptides urea etc. is quite inert towards H2180 as are those of many inorganic ions of biological significance such as phos- phate sulphate and nitrate. These ions may be labelled by interaction of heavy water with the appropriate anhydride.I n his review Dole lists a number of organic reactions which have been studied by lSO tracer techniques and offer routes for the synthesis of labelled eaters amides ethers etc. More recently Bender 292 has prepared carbonyl- labelled esters via the corresponding imidates. Phosphorus The biological importance of phosphorus coupled with a ready avail- ability of the radioactive isotope 32P stimulated a variety of early tracer Zs5 Bentley Biochem. J. 1950 45 591. 286 Bunton Comyns and Wood Research 1951 4 383. 287 Long and Friedman J. Amer. Chem. SOC. 1950 72 3692. 288 E.g. Kirschenbaum Strong and Grosae AnaZyt. Chem. 1952 24 1361. 2g0 Cohn J . Biol. Chem. 1949 180 771. 2g1 MacKenzie and Milner J. X. Afr. Chem. Inst. 1952 4 (l) 79.383 Bender J . Amer. Chem. SOC. 1951 73 1626. Urey Science 1948 108 489 ; Silverman Cfeochim. Cosmochim. Actcl 1951 2 26. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 435 applications which have been the subject of several excellent reviews. 293 A formidable number of labelled compounds has been isolated during the course of this work but there have been relatively few deliberate syntheses. 32P (7 = 14.3 days) is the longest-lived of the radioactive isotopes and decays with emission of a p-particle of maximum energy 1.7 mev. First produced by the 3lP(n,~)3~P reaction when ordinary phosphorus was bom- barded by slow neutrons it can also be made more satisfactorily and a t higher specific activity by a variety of other reactions using a radium- beryllium neutron source a cyclotron or an atomic pile vix.31P(d p)32P ; 34S(d,~)32P ; 32S(n,p)32P ; 35Cl(n,~)32P. Methods for the extraction 294 of 32P as phosphate at very high specific activity from pile-irradiated sulphur have been described it is essential to ensure that the material is at a uniform oxidation level before use.295 The energetic nature of the radiation may necessitate some care in manipulating appreciable quantities but it also simplifies analytical techniques for which either solid or liquid counting is suitable. 296 The phosphate ion does not normally undergo exchange 297 with the phosphorus of organic compounds or with the other oxy-acids of phosphorus and such compounds must therefore be synthesised by chemical or biological methods. Preparations of various useful intermediates including the phos- phorus oxy-acids and poly-acids 298 phosphorus halides 299 and phos- phorus oxychloride have been reported.This last -named compound is especially valuable and may conveniently be made by Axelrod's modification of Lindberg's method in which phosphorus pentachloride is heated with radioactive phosphate. 300 The inevitable dilution of activity entailed in this method may be avoided by treating carbonyl chloride with ferric phos- phate and fractionating the products.301 Chemical Syntheses.-Esters and similar derivatives of phosphoric acid prepared by standard methods (e.g. H,PO or POCI + alcohol or Ag,P04 + alkyl halide) account for a large number of the recorded chemical syntheses e.g. those of t r i b ~ t y l ~ O ~ tri-~-tolyl,~O~ p-nitrophenyl,300 propane- di01,3O4 glycerol CC- and @- 299a 305 glucose,3o6 chole~teryl~~~7 %amin0ethyl,~99~ and di-(2-amin0ethyl)~~' phosphate.Radioactive vitamin K substitute 307 2Q3 Ref. 6 p. 279 ; Wood Atomics 1951 2 217. 294 Arrol Nucleonics 1953 11 (5) 26. 2 g 5 Thomas and Nicholas Nature 1949 163 719. 297 Gourlay U.S. At,. Energy Comm. Rep. AECU-1763. Zs8 Hull J . Amer. Chem. SOC. 1941 63 1269 ; Vogel and Podall ibid. 1950 72 1420 ; Gotte and Frimmer Angew. Chem. 1952 65 53. 2g9 ( a ) Chargaff J . Amer. Chem. SOC. 1938 60 1700; ( b ) Chargaff and Keston J . Biol. C'hem. 1940 134 515. 300 Axelrod ibid. 1948 176 295. 301 Gardiner and Kilby J. 1950 1769. 302 Baldwin and Higgins J . Amer. Chem. Xoc. 1952 74 2431. 303 Hodge and Sterner J . Pharmcol. 1943 79 225. 304 Lampson and Lardy J . Biol. Chem.1949 181 697. PopjAk and Muir Biochem. J . 1950 46 103. 306 Ref. 304 p. 693. 307 Morrison and CrowIey Univ. California Radiation Lab. Rep. 1769. 2D* Ref. 6 p. 279. 436 QUARTERLY REVIEWS (2 methyl- 1 4-naphthaquinol diphosphate) and phosphoryl choline 308 have also been synthesised. Physiologically active compounds which have been investigated include the " nerve gas " D.F.P.309 (diisopropyl phosphorofluoridate) the insecti- cides '' OMPA " (octamethylpyrophosphoramide) or " Pestox I11 ",3l* " Iso-pestox " (NN'-diisopropylphosphorodiamidic fluoride) tetra-alkyl pyrophosphorates 311 and Parathion (diethyl p-nitrophenylphosphorothion- ate = diethyl p-nitrophenylmonothiophosphate) . 31 The last compound is obtained in rather poor yield from 32Poc13 or 32Pc15 C S-AICI a H,PO + POCl - PCI - PSCl + 1000" 160" (EtO),P( S)C1 + (EtO),P( S) .OC,H,.NO,-p P*S I P + PCI + PSCI p-NO,*CGH4*O*P(S)Cl p-NO,*C,H4*O*PS(OEt) but Hein and McFarland observed 312 that by using pile-irradiated phos- phorus trichloride doubly (P and S) labelled parathion of 94% purity and activity 2 pc/mg.was produced in 26% yield. The activity was due to the reactions 31P(n Y ) ~ ~ P and 35Cl(n,p)35S. The biological behaviour of a number of phosphine oxides and phosphinic and phosphonic acids and their derivatives which were prepared from active phosphorus trichloride has been investigated by Morrison and Crowley.307 Biosyntheses.-It is impracticable to detail the very many phosphorus- containing compounds which have been isolated during the extensive biological experiments with radiophosphorus (for a summary see Kamen 6)) and there have been relatively few deliberate syntheses.There have been enzymic syntheses of adenosine triphosphate,313 phos- phoglyceric acid,314 phosphocreatine,315 glucose-1 and -6 phosphate,305 315 and of the attendant phosph~glucomutase.~~~ Much work has been devoted to the isolation of phospholipids and nucleic acids from cell-free systems from systems involving surviving tissue slices and from intact organisms.317 The labelling of proteins for immuno- logical studies has been well reviewed by WormallY3ls but there have also been numerous instances of phosphorus-containing proteins isolated from rnetabolising systems. 30* Riley J . Amer. Chem. SOC. 1944 66 512. 309 Witten and MilIer ibid. 1948 70 3886 ; Saunders and Worthy J. 1950 1320. 310 Gardiner and Kilby Biochem.J. 1952 51 78. 311 Roan Fernando and Kearns J . Econ. EntomoZ. 1950 43 319. 312 Lockau Ludicke and Weygand Naturwks. 1950 38 350 ; Murray and Spinks Canad. J . Chem. 1952 30 497 ; Hein and McFarland J . ,4mer. Chem. SOC. 1952 74 1856. 313 Kornberg and Price J . Biol. Chem. 1951 191 535. 314 Sutherland Posternak and Cori ibid. 1949 181 153. 315 Meyerhof and Green ibid. 1950 183 377. 316 Jagannathan and Luck ibid. 1949 179 569. 317 Ref. 6 p. 286; ref. 46 p. 152. s18 Wormall Brit. Med. BUZZ. 1952 8 224 ; Francis Mulljgan and Wormall Nature 1951 167 748. 1'HOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 437 Smith secured the incorporation of phosphorus into vitamin B, by growing Xtreptomyces griseus on labelled media 319 although the yields were very poor.Other entities which have been labelled include flie~,~~O bacteria 21 bacteriophages viruses and blood cells. 24 Sulphur The stable isotopes of sulphur find limited application in tracer studies Thode et al. report some interesting variations in the 32S 34S ratio for materials derived from various natural sources 325 and there has also been some application to biochemical studies but interest has centred largely on use of the more convenient radioactive isotope 35S which has been reviewed by Tarver 326 and 0thers.3~7 3,s The isotope (-r$ = 87.1 days) decays with emission of a low-energy (0.169 mev) p-ray similar to that of 14C and is now usually prepared by pile irradiation of potassium ~hloride3~7 35Cl(n,~)3~S. After oxidation it can be extracted from the target material as sulphate possessing a very high specific activity although for most pur- poses it is diluted with inactive carrier.The radiation hazard is slight and techniques suitable for 14C are generally appropriate although memory effects due to surface adsorption are far more pronounced with sulphur. It is usually determined with counters of thin-window or gas-flow type in samples of barium or benzidine sulphate particular care being taken to ensure quantitative oxidation to the sulphate radica1.329 Chemical Syntheses.-A range of useful intermediates has been prepared including sulphur 330 ~ u l p h i d e ~ ~ ~ the oxides and oxy-acids 332 thiocyanate 333 thionyl chloride 334 carbon disulphide 335 thiols 3 3 0 3 336 thiouraa 337 and 31Q Smith Biochem. J. 1952 52 384 357. 320 Radcliff Bushland and Hopkins J.Econ. Entomol. 1952 45 509. 321 Harper and Morton J. CTen. Microbiol. 1952 7 98. s22 Kozloff and Putman J. Biol. Chem. 1950 182 229 243. s23 Graham Canad. J . Res. 1950 28 E 186. 324 Reeve Brit. Med. Bull. 1952 8 181. 325 Szabo Tudge Macnamara and Thode Science 1950 111 464. 326 Cf. ref. 7 Vol. 2 p. 281. 8a7 Ref. 6 p. 300. 328 Erichsen and Muller Angew. Chem. 1952 64 580. Young Edson and McCarter Biochem. J. 1949 44 179 ; Larson Maas Robin- gon and Gordon Amlyt. Chem. 1949 21 1206 ; Rollinson and Creamer Abs. Amer. Chem. SOC. 122nd Mtg. 1952 1 8 ~ . 330 (a,) Seligman Rutenburg and Banks J . Clin. Invest. 1943 22 275 ; (b) Wood Rachele Stevens Carpenter and du Vigneaud J. Amer. Chem. SOC. 1948 70 2547. a31 Cooley Yost and McMillan ibid. 1939 61 2970 ; Henriques and Margnetti I n d .Eng. Chem. Anal. 1946 18 476. 332 Berry and Peterson J . Amer. Chm. Soc. 1951 73 5197 ; Huston ibid. p. 3049 ; Ames and Willard ibid. p. 164 ; Masters and Norris ibzd. 1952 74 2395. 333 Wood and Kingsland J. Biol. Chem. 1950 185 833 ; Eldjarn Acta Chem. Scand. 1953 7 343. 334 Johnson Norris and Huston J . Amer. Chem. Soc. 1951 73 3052. 335 Eldjarn Acta Chem. Scand. 1949 3 644. 336 (a) Wood and van Middlesworth J. Biol. Chem. 1949 119 629 ; (b) Walling a37 Bills and Ronzio ibid. 1950 72 5510. J . A m r . Chem. SOC. 1948 70 2561. 438 QUARTERLY REVIEWS toluenesulphonyl chloride. 338 Keston et al. have synthesised l8 p-iodobenzene- sulphonyl chloride for use in their elegant method for the analysis of amino-acid mixtures. Amino-acids.-Major effort has been devoted to the sulphur-containing amino-acids particularly cystine and methionine for which a number of preparative methods have been devised.Most of these involve a prelimin- ary synthesis of toluene-m-thiol by interaction of a benzyl halide with an alkali sulphide 330 339 or better by interaction of sulphur with a benzyl- magnesium halide.330 336a Kilmer and du V i g n e a ~ d ~ ~ by condensation of sodium hydrogen sulphide with excess of [13C,]ethylene dichloride followed by a phthalimidomalonate synthesis sodium reduction in liquid ammonia and finally X-methylation prepared doubly labelled [Py-l3C 34S]methionine in yields of 5% on the carbon and 11% on the sulphur. However in order to conserve isotopic material it is better to condense the sulphide with a suitable preformed carbon chain with,326 33% 339-342 or protection of the amino-acid end group.More recently isotopic thiourea has proved a useful donor of the methylthio-group,343 and Fry 344 has described two methods based on serine which are suitable for synthesis of cystine. I NHX G P co / \ \ / S h'H,.CNr CS(NH2)Z + + CI.[CH2]n*HC NH I I co HN CH.[CH,],-CI 17 Optically active compounds have been prepared by an application of the isotope dilution technique,13 or by biosyntheses with yeast 345 or bacteria.346 By direct neutron-irradiation of " cold " material Ball Solomon and Cooper 347 succeeded in producing cystine of low specific activity supporting their claim by degradation studies which established the sole site of radio- 338 Ray and Soffer J . Org. Chern. 1950 15 1037. 339 Tarver and Schmidt J .Biol. Chem. 1939 130 67. 340 Idem ibid. 1942 146 69. 3 4 1 Wood and Gutmann ibid. 1949 179 535. 3 4 2 Melchior and Tarver Arch. Biochem. 1947 12 301. s4s Bloch Abs. Amer. Chem. SOC. 118th Mtg. 1950 22c. 3 4 4 Fry J . Org. Chem. 1050 15 438. 1 4 5 Schliissel Maurer Hock and Hummel Biochem. Z . 1951 322 226 ; Williams and Dawson Biochem. J. 1952 52 314 ; Wood and Mills J . Amer. Chem. SOC. 1952 74 2445. 346 Cowio Bolton and Sands Arch. Biochem. 1952 35 140. 347 Ball Solomon and Cooper J . Biol. Chem. 1949 177 81. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 439 activity in the sulphur atom. Lipp and Weigel were however unsuccessful in parallel experiments. 348 The phthalimidomalonate synthesis has also been applied to [ 35S]cysta- thionine (2 -amino -2 - carbox yethyl 3'- amino - 3'- carbox yprop yl sulphide) ,349 and Weiss and Stekol describe a diketopiperazine method starting from homoserine which is suitable for the preparation of any y-alkylthio-a-amino- butyric acid if necessary in optically active form (e.g.cystathionine homolanthionine ethionine). 350 i~oCysteine,~~l taurine,352 and cysta- mine 352 have all been synthesised by chemical means and a microbiological preparation of glutathione has been rep0rted.3~~ Drugs.-Preparation of a wide range of pharmacologically important compounds has been reported. Mustard gas,354 355 and the derived sul- phoxide and sulphone 355 together with several related compounds,33Ob were synthesised by several groups and now find some application in immuno- logical studies.318 Yields are excellent.CH,*CH,-OH CH,*CH,Cl CH,-CH /O\ / / BaSO -+ Bas + H,S -> S - + s \ CH,*CH,-OH CH,*CH,Cl \ I S -+ R*SH + RS*CH,*CH2.0H -+ RS*CH,*CH,Cl BAL (2 3-di[35S]mercaptopropan- 1-01) has been prepared in moderate Radioactive compounds of the sulphonamide group which have been 358 its di-iodo-derivati~e,~~~ sulphanil- yield froin the dibroinopropanol. 356 studied include sulphanilic amide,3S9 sulphapyridine 360 and sulphathiazole.361 Na,*S ---+ Na,*SO NH,*C,H,-*SO,*NHR Exchange + I Ba*S + H,*SO _______+ NHAc*C,H,**SO,H I with H,SO +y 1 Heat Na,*S -+ H,*S NH,Ph,H,*SO + NH,*C,H,**SO,H a40 Lipp and Weigel Natzcrwiss. 1952 39 189. 349 Rachelo Reed Kidwai Ferger and du Vigneaud J . Biol. Chem. 1950 185 350 Weiss and Stekol J . Amer. Chem. SOC. 1951 '73 2497. 3 5 1 Dziewiatkowski and Wingo Proc.SOC. Exp. Biol. 1949 '70 448. 352 Eldjarn Acta Chem. Scand. 1951 5 677. 3 5 3 Woodward J. Franklin Inst. 1951 251 557. 3 5 4 Axelrod and Hamilton Amer. J. Path. 1947 23 389. 355 Boursnell Francis and Wormall Biochem. J. 1946 40 743. 356 Young Science 1946 103 439; Peters Spray Stocken Collie Grace and 357 Ingraham J . Amer. Chem. SOC. 1952 74 2433. 358 Myers Cancer Res. 1950 10 234. 817. Wheatley Biochem. J. 1947 41 370. 440 QUARTERLY REVIEWS Other physiologically active compounds which have been labelled include '' antabuse ' ' (tetraethylthiuram d i s ~ l p h i d e ) ~ ~ ~ 2-toluene-p-sulphonamido- fluorene,362 insulin sulphate,363 methionine s~lphoximine,36~ penicillin,365 pentothal (scdium 5-ethyl-5-l'-methylbutyl-2-[35S~thiobarbiturate],~ pheno- thiazine 366 2 -p-arninophenylthiazole 367 and the insecticide " Parathion ' ' (diethyl p-nit4rophenyl phosphorothionate).368 MisceZlaneou-s.-Chernical syntheses of xanthates 369 alkyl sulphates,37* tetramethylt hiuram disulphide 37l and dibenzothiophen and its 3- acet arnido- derivatives 372 have been carried out and attention has been drawn to a suitable synthesis for dithiz0ne.~'3 Biosmthesis.-IVlention has already been made of the sulphur-containing amino-acids and peptides which have been prepared by biosynthetic methods. There has also been a very considerable effort devoted to the synthesis of penicillin 374 with high specific activity. Various organisms proteins and viruses have been labelled in connection with biological studies. Halogens Fluorine.-The most suitable isotope for tracer work 18F has a half-life of only 112 minutes and it must be made in a cyclotron or similar device.There have been few applications or syntheses. Chlorine.-The isotopes 34Cl and 38Cl have rather short half-lives (about 8 hour) and have therefore found very limited tracer application. More recently 36Cl ( T ~ - lo6 years radiation 0.73 mev) has become avail- able and one or two syntheses have been recorded. The labelled y-isomer of '' benzene hexachloride " required for isotope dilution analysis of the commercial product has been reported. 375 Acetanilide has been chlorinated hydrolysed and treated with carbonyl chloride to yield p-[36Cl]chloro- phenyl isocyanate from which 6-p- [36Cl]chlorophenyl hydantoic acid was prepared by interaction with glycine. 376 Standard methods of zinc-catalysed 359 Klotz and Melchior Arch.Biochem. 1949,21 35 ; Fingl Christian and Edwards 360 Bray Francis Neale and Thorpe Biochem. J. 1950 46 267. 362 Ray and Argus Cancer Res. 1951 11 274. 363 Stadie Haugaard and Vaughan J . Biol. Chem. 1952 199 729. 364 Roth Wase and Reiner Science 1952 115 236. 365 du Vigneaud Wood and Wright " The Chemistry of Penicillin " Princeton 3 6 6 Lazarus and Rogers Nature 1950 166 647. 367 Noll Sorkin and Erlenmeyer Helv. Chim. Acta 1949 32 609. 368 Jensen and Pearce J . Amer. Chem. SOC. 1952 74 3184. 369 Gaudin and Carr Andyt. Chem. 1952 24 887. 370 Croes and Ruyssen Bull. SOC. Chim. biol. 1951 33 1837. 3 7 1 Craig Davidson Juve and Geib J . PoZymer Xci. 1951 6 1. 3 7 2 Brown Iiirkwood Marion Naldrett Brown and Sandin J . Amer. Chem. SOC. 1951 73 465. Irving and Bell Nature 1952 169 756.374 Hawell Thayer a.nd Labaw Science 1948,107,299 ; Maas and Johnson J . Bact. 1949 58 361 ; Smith and Hockenhull J . Appl. Chem. 1952 2 287. 375 Craig and Tryon Abs. Amer. Chem. SOC. 122nd Mtg. 1950 1 2 ~ . 376 Woeber J . Amer. Chern. Soc. 1952 74 1354. J . Amer. Pharm. ASSOC. 1950 39 693. Noll Bang Sorkin and Erlenmeyer Nelv. Chim. Acta 1951 34 340. Uiiiv. Press p. 892. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 441 H36Cl esterification of cis- and trans-3-chloropropenol were utilised by Hatch Morgan and Tweedie for preparation of the 1 3-dichloropro- penes 377 there was no exchange of chlorine. A general method of synthe- sising labelled chlorides by exchange with aluminium chloride has been suggested. 37 Bromine.-Several radioactive isotopes are known of which *2Br (-r4 - 34 hours) is most suitable for tracer studies despite an appreciable hazard presented by the decay process which involves a cascade of 3 y-rays associated with an initial P-disintegration.37D The isotope which is prepared by neutron-irradiation of a suitable bromide either organic or inorganic [8lBr(n y)g2Br] has found application in chemical medical pharmaceutical immunological and entomological studies where the ease of counting with ordinary thick-walled Geiger-Muller tubes is an advantage.380 The short half-life precludes any lengthy syntheses but it has been used for kinetic and preparative esterification studies,381 and certain aromatic bromides have been prepared by direct exchange with the inorganic (lithium aluminium) halides.382 Addition of elementary bromine obtained by oxidation to the appro- priate unsaturated compounds has yielded 7 8-dibromcestrone 383 and a range of aliphatic dibromo-acids. 384 The isotope has also been substituted into the molecules of dyes required for tumour localisation (dibromotrypan- bl~e,3~5 dibromo-Evans-blue). Several groups have synthesised the cestro- gen bromotriphenylethylene. 3s5 Howarth has studied the action of pro- caine by studies with the labelled dibromo-analogue.387 Iodine.-The physiological significance of iodine in minute quantities soon stimulated biological studies with a radioactive isotope a t a time when only lZ8I (ri - 25 min.) was available.388 The development of the cyclotron and atomic pile made available 13*1 (-c4 - 124 hours) and 1311 (7 - 8 days) and today virtually all iodine tracer work is carried out with the last-named isotope.389 This decays by one of two alternative routes which however both yield ,!?-particles and a series of fairly energetic y-rays.It is prepared - B (TA 30 hr.) by reactions 130Te(d,n)1311 or 130Te(n,y)131Te -> 1311 or is separated from fission products and is available as iodine' iodide or iodate or in a range of useful intermediates and physiologically important compounds a77 Morgan Hatch and Tweedie J. Amer. Chem. Soc. 1952 74 1826. 378 Wallace and Willard ibid. 1950 72 5275. 379 Ref. 6 p. 331. 3eo Winteringham Nature 1949 164 183. 3811dem J. 1949 S 416; Baret and Pichat, Bull. SOC. chim. 1950 17 1294. 382 Kieffer and Rumpf ibid. 1951 18 584. 3*3 Twombly McClintock and Engelman Amer.J . Obst. Gynaecol. 1948 56 260. 384 Buu-HoY Berger Daudel Daudel May and Miguet Helv. Chim. Acta 1946 385 Moore and Tobin J . Clin. Invest. 1943 22 155. 386 Paterson Gilbert Gallagher and Hendry Nature 1949 163 801 ; Twombly Schoenewaldt and Meisel Cancer Res. 1951 11 750 ; Apelgut Cheutin Mom and Berger Bull. SOC. chim. 1952 19 533. 29 1334. 3R7 Howarth Nature 1948 161 857 ; 1949 163 679. 38a Ref. 3 p. 139. saQ Ref. 6 p. 336. 442 QUARTERLY REVIEWS required for analytical chemical and medical studies. The separation 390 and assay 391 of the isotope have been reviewed. Five major methods for the introduction of active iodine into organic compounds have been developed vix. esferifi~ation,~~~ halogen exchange direct iodination with iodine or iodine chloride the Sandmeyer reaction and biosynthesis.Compounds prepared by exchange with inorganic halide include both aryl 393 and alkyl 394 halides iodoacetamide iodoacetainido-acids,395 mustard-gas analogues 396 and thyroxine. 397 Elementary iodine prepared from iodide by oxidation with iodate hydrogen peroxide hypochlorite or nitrous acid has been utilised in pre- parations of p-iodoaniline and hence iodotetrazolium salt ,398 chiniofon ____ h' I '0 II b'+ I 8 I 0 I (8 - hydroxy-7 [1311] -iodo- quinoline-5- sulphonic acid) 399 iodinated cestra- di01s,~OO pheniod01,~~l and thyroxine 402 and its anal0gues,397~ 402 and for labelling a wide range of products such as f a t ~ ~ O ~ fibre~,~O~ polystyrene,405 etc. Wormall 318 has discussed the important subject of the labelling of proteins by methods which include iodination or reaction with p-iodophenyl- diazonium chloride.Elementary iodine has also been utilised for the preparation of iodo- triphenylethylene from the corresponding Grignard compound 406 and 390 Arrol Nucl. Xci. Abs. 1952 6 151. 391 Bruner and Perkinson Nucleonics 1952 10 ( l o ) 57. 392 Ludes and Endler Chem. Abs. 1952 46 9145. 393 Kristjanson and Winkler Canad. J . Chern. 1951 29 154. 394 Heydring and Winkler ibid. p. 790. 395 Friedman and Rutenburg J . Amer. Chern. Xoc. 1950 72 3285. 396 Seligman Friedman and Rutenburg Cancer 1950 3 336 342. 397 ( a ) Frieden Lipsett and Winzler Science 1948 107 353; ( b ) Lemmon Tarpey and Scott J . Amer. Chem. SOC. 1950 72 758 ; ( c ) Taurog Briggs and Chaikoff J . Biol. Chem. 1952 194 655. 398 Seligman Gofstein and Rutenburg Cancer Res.1949 9 366. 399Albright Tabern and Gordon Anter. J . Trop. Med. 1947 27 533. 400 Albert Heard Leblond and Saffran J . Biol. Chem. 1940 177 247. 401 Free Page and Woollett Biochem. J . 1950 48 490. 409 Gross and Leblond J . Biol. Chem. 1950 184 489 ; Michel Roche and Tata Bull. Xoc. Chim. biol. 1952 34 366 466. 403 Rutenburg Seligman and Fine J . Clin. Invest. 1949 28 1105. 404 Sankey Mason Allen and Keating Pulp and Paper Mag. Canada 52 136. 405 Tubis and Jacobs Nucleonics 1952 10 (9) 54. 406 Morrison Univ. California Radiation Lab. Rep. 1719. THOMAS AND TURNER ISOTOPICALLY LABELLED ORGANIC COMPOUNDS 443 iodination with l31ICl has served for the preparation of di-iodofluores~ein,~~~ 3' 5'- [I311 Jdi -iodo -A -met hopt erin ,4 3 5 - [I311 ,I&-iodofolic acid [ 3114] - tetraiodophenolphthalein,4 and iodinated peni~illins.~ Inactive iodine chloride in conjunction with active iodide has been employed for the pre- paration of iodinated sulphanilamide and sulphapyridine 408 and for the carcinogen 2- a cet amido-7- [1311]iodofluorene.409 The Sandmeyer reaction is convenient for the preparation of aromatic compounds and has been employed for the preparation of the analytically important p-iodobenzenesulphonyl chloride 18 and for 2- [1311]iodo-3-nitro- benzoic acid,410 iodinated dyes (XYile-bl~e,~~~ trypan-blue 412) a D.D.T. analogue,413 iodo- and iod~so-benzene,~~~ 2 4-dichloro-5-[1311]iodophenoxy- acetic a~id,~15 etc. Biosynthetic methods have frequently been employed during studies of thyroid metabolism416 and have found some application in protein labelling.One novel method for the preparation of the simpler iodides through a modified Szilard-Chalmers reaction has also been described.417 407 Roe Hayes and Bruner J. Amer. Chem. SOC. 1051 73 4453. 408 Bloch and Ray J. Nat. Cancer Inst. 1946 7 61. 40@ Weisburger J. Amer. Chem. Xoc. 1950 72 1758. 410 Mitchell Wood Wolfe and Irving Science 1947 106 395. 411 Sloviter J. Amer. Chem. SOC. 1949 '71 3360. 412 St,evens Lee Stewart Quinlin and Gilson Cancer Res. 1949 9 139. 413 Jensen and Pearce J. Amer. Chem. SOC. 1952 74 2436. 414 Aten and Aten ibid. p. 2411. 416 Wood Wolfe Doukas Klipp Fontaine and Mitchell J. Org. Chem. 1949 14 416 Gross and Leblond J . Biol. Chem. 1950 184 489 ; Gross and Pitt-Rivers *17 Glueckauf and Ray J. 1949 S 330. 900. Biochem. J. 1953 53 645.

ISSN:0009-2681    

DOI:10.1039/QR9530700407

出版商:RSC    

年代:1953

数据来源: RSC