830 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol. 77 The Stability of Metal Chelates in Relation to their Use in Analysis BY HENRY FREISER SUMMARY The acidic dissociation constants of a number of organic reagents in 60 per cent. aqueous dioxan have been determined and the chelate formation constants measured in the same so1ven.t by the Bjerrum - Calvin technique for metals such as Cu, Ni, Co, Zn, Pb, Mg, Cd, La and Cenl. Comparison of the stability of complexes of diketones with those of 8-hydroxyquinoline or 1 : 2 : 3 :4-tetrahydro-8-hydroxyquinoline, or o-amino- phenol and o-aminobenzenethiol illustrate the influence of aromatic resonance, basicity and the electronegativity of bonding atoms. The relative stability of chelates of 8-hydroxyquinoline and its 2- and 4-methyl derivatives throw light on the steric hindrance caused by 2-substituents, while comparison of 2-( o-hydroxyphenyl) -quinoline and 2-(o-hydroxypheny1)-isoquinoline further illustrate points of steric hindrance and of ring size. 2-(o-Hydroxyphenyl)-benzoxazole forms chelates with many bivalent (but not tervalent) metals, but although the relative stability of many reagents followed the order of the second ionisation potential of the metals concerned, complexes of 2- (0-hydroxypheny1)-benzthiazoline were quite exceptional.The order of stability of metal complexes of dimethylglyoxime showed no abnormalities and the specific behaviour towards nickel must be due to solubility factors. This is confirmed by studies of metal complexes of the corresponding 0-methyl ether which present many striking features.Analytical uses of 2- (0-hydroxypheny1)-benzoxazole, 2-(o-hydroxypheny1)- benzthiazole, and 2-(o-hydroxyphenyl) -benzimidazole are indicated. RBsuwk Les constantes de dissociation acides d’un certain nombre de rkactifs organiques en milieu dioxanique a 50 pour cent. ont 6tb mesurkes, et leur constantes de formation de complexes ont ktk dkterminkes dans le meme milieu selon le proc6dk de Bjerrum et Calvin pour des m6taux tels que Cu, Ni, Co, Zn, Pb, Mg, Cd, La et CeIII. En comparant la stabilit6 des complexes des dicbtones et de la 8-liydroxy- quinolkine, ou de la 1 :2 :3 :4-tetrahydro-8-hydroxyquinolkine, ou de l’o-amino- sulfoph6nol et de 1’0-aminoph&nol, on dkmontre l’influence de la r6sonnance aromatique, de la basicit6 et de l’klectronkgativitk des atomes de liaison.La stabilitb relative des complexes de la 8-hydroxyquinolkine et de ses dbrivks mkthyliques 2- et 4-bclaircit l’empikhement stbrique dQ aux substituants en position 2, tandis que la diff krence entre la 2-(o-hydroxyphknyl) -quinolkine et la 2-(o-hydroxyphbnyl)-isoquinolkinie dkmontre nettement l’effet de l’empechement st6rique et du nombre dl’atomes dans la chaine fermke. Le 2- (0-hydroxyphknyl) -benzoxazole produit des complexes avec plusieurs mktaux bivalents (mais pas trivalents), mais bien que la stabilith relative de plusieurs rbactifs suive l’ortlre du second potentiel d’ionisation des mbtaux en question les complexes de la. 2-(o-hydroxyphbnyl) -benzsulfazoline sont tout A fait exceptionnels. L’ordre de stabilitk des complexes mktalliques de la dim6thylglyoxime est normal, et la faGon spbcifique dont celle-ci se comporte envers le nickel doit s’expliquer par des considbrations de solubilitk.Ceci est confirm6 par une ktude des complexes mbtalliques des kthers 0- mbthyliques, qui prksentent plusieurs traits frappants.Dec., 19521 SECTION 6 : ORGANIC COMPLEXES On indique des emplois analytiques pour les 2-(o-hydroxyphCnyl)- benzoxazole, 2-(o-hydroxyph~nyl)-benzsulfazole et Z-(o-hydroxyphthyl)- benzimidazole. 831 ZUSAMMENFASSUNG Es wurde die Dissoziationskonstante einer Anzahl von organischen Reagenzien in 50 prozentigem, wassrigen Dioxan bestimmt und in dem gleichen Losungsmittel die Komplexbildungskonstante fur Cu, Ni, Co, Zn, Pb, Mg, Cd, La und CeIII mittels der Bjerrum - Calvin-Technik gemessen.Eiii Vergleich der Stabilitat von Komplexen rnit Diketonen mit solchen von 8-Oxychinolin oder 1 : 2 : 3 : 4-Tetrahydro-8-oxychinolin oder o-Amino- phenol und o-Aminothiophenol zeigt den Einfluss der aromatischen Resonanz, der Basizitat und Elektronegativitat der Bindungsatome. Die relative Stabilitat von Komplexen des 8-Oxychinolins und ihrer 2- und 4-Methyl- derivate zeigt die sterische Behinderung die durch an der 2-Stellung erfolgte Substitution verursacht wird. Ferner veranschaulicht ein Vergleich des 2-(o-Oxyphenyl)-chinolins mit 2-(o-Oxyphenyl)-isochinolin Beziehungen zwischen sterischer Behinderung und Ringgrosse. 2- (o-Oxyphenyl) -benzoxazol bildet mit vielen zweiwertigen (jedoch nicht mit dreiwertigen) Metallen Komplexe.Die relative Stabilitat zahlreicher Reagenzien folgt zwar den Regeln des zweiten Ionisationspotentials der betreffenden Metallen, aber die Komplexe von 2-(o-Oxyphenyl)-benzthiazolin bilden Ausnahmen. Die Grosse der Stabilitat von Metalldimethylglyoxim- komplexen bildet keine Ausnahme und des spezifische Verhalten gegenuber Nickel beruht auf Loslichkeitsfaktoren. Dies wird durch Untersuchungen von Metallkomplexen des entsprechenden 0-Methylathers erhartet, die viele charakteristische Merkmale aufweisen. Weiter wird die analytische Anwendung von 2- (o-Oxyphenyl) -benzoxazol, 2- (o-Oxyphenyl) -benzthiazol und 2- (0-Oxyphenyl) -benzimidazol gezeigt. THE subject of organic analytical reagents in inorganic analysis is of especial interest to the modern analytical chemist.The striking properties of metallic chelates, which in many respects resemble those of organic more than inorganic coinpounds, have proved most useful in metal analysis. Metal chelates are usually more different from one another than are the conventional metal compounds, and this permits usually difficult metal separations to be performed readily. Many chelates are very insoluble in aqueous media (rivalling the hydrox- ides and sulphides in this respect) and are soluble in organic solvents. Their solubility characteristics thus make chelates suitable for both precipitation and solvent-extraction techniques. Further, many of these chelates have a distinctive colour, enabling them to be adapted to colorimetric procedures. The organic portion of the chelate is often readily oxidisable with reagents such as potassium permanganate or potassium bromate - potassium bromide , so making possible the application of volumetric oxidimetry in metal analysis. With the development of high-alloy steels and other complex alloys the “less familiar” metals found increasing use, posing ever increasingly difficult separation problems to the analyst.Those classical methods of separation that owed their success to the assumed absence of certain infrequently encountered elements were no longer applicable once these elements found their way into alloys and other modern materials. Therefore, new reagents that would be more selective in their action were sought and applied. In this development, organic reagents played an important role.Further, an increasing awareness of the significance of the effect of trace elements upon both physical and chemical properties of matter has caused the analyst to seek out more sensitive reagents. Trace metals have proved to be of utmost importance in plant and animal metabolism as well as in determining the strength of structural alloys and the fluorescence of natural and synthetic phosphors. The solution of the analytical problems connected with these materials has been considerably aided by the application of organic reagents. I t is interesting to note that among the earliest organic reagents to be introduced were those that were considered the most specific. Alpha-nitroso-beta-naphthol, introduced in 1884 by Ilinski, was considered at the time to be largely specific for cobalt.Tschugaeff in 1905 developed dimethylglyoxime, probably the most familiar organic reagent, as a specific reagent for nickel and palladium. A few years later (1909), Baudisch created quite a stirby introducing the ammonium salt of N-nitrosophenylhydroxylamine, which was hailed as a832 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol. 7'7 copper and iron specific and named cupferron to ;point out this specificity. Indeed, there was hope that patient work would lead to the development of reagents that would be so specific that the problem of analytical separations in metal analysis would be eliminated. However, more work with these and other reagents indicated that the best that could be hoped for was selectivity that was somewhat short of the desired specificity.In this connection it is note- worthy that the two most widely used organic reagents to-day are 8-hydroxyquinoline and diphenylthiocarbazone (dithizone) both of which react with quite a number of metals. The hope of developing specific reagents has not completely faded, but it has dimmed considerably. Whilst the development of reagents of such selectivity that they would react with only one metal each is beyond reach, much has been learned to help in the design of better organic reagents for use in difficult metal separations. (One of the most useful ways to characterise an organic chelating reagent, to study the influence of various structural factors affecting chelating tendency, is by the determination of the stabilities of its metallic chelates.The first quantitative evaluation of chelate stability was performed by Calvin and Wilson1 in 1945 by an adaptation of the method used by Bjerrum2 in 1941 to study the formation con- stants of metal ammines. This method involves following the course of chelate formation by measuring the amount of hydrogen ion released .in the reaction- M" + 2HR + MR, + 2H', where M" refers to a bivalent cation, HR to the acid form of the chelating reagent and MR, to the chelate. A mixture of known amounts of chelating agent, perchloric acid and metal perchlorate in 50 per cent. v/v dioxan - water medium (to keep the chelate in solution and eliminate solubility equilibria from the considerations) is titrated with standard sodium hydroxide, free from carbon dioxide, and the titration curve determined with a pH meter with a glass - calomel electrode pair.A supplementary titration is carried out in the absence of the metal ion to determine the ionisation constant of the reagent. The titration data yield information leading to calculation of the equilibrium constants for the combination of the metal ion and reagent anion (R')J as represented by the stepwise reactions- & M" + R' FA M R & MR' + R' :+ MR, Calvin and Wilson studied the stabilities of the copper chelates of a series of /3-diketones and o-hydroxy aromatic aldehydes. They demonstrated that the forces holding a proton to a chelating reagent are similar to those holding a metal ion thereto. In this way they found that the stability of the copper chelates of a seriies of substituted salicylaldehydes or a series of p-diketones increased linearly with the basicity (pK,) of these reagents.Comparison of the copper chelates of these two types of reagents revealed the operation of a second factor. A j3-diketone of comparable basicity to a salicylaldehyde was found to form a more stable copper chelate than the latter reagent. The difference was attributed to decreased resonance in the active grouping in the salicylaldehyde because the C--C linkage marked (B) is of lower (14) bond order (benzene nucleus resonance) than the corresponding linkage ( A ) in the /I- diketone. Calvin concluded that the enolate resonance played a different and far more important role in the bonding of copper than it did in the bonding of hydrogen and suggested the possibilities that (a) copper, like other transition elements, has low-lying vacant orbitals that are capable of accepting electron pairs to form homopolar bonds.The rearrangement ofDec. , 19521 SECTION 6 : ORGANIC COMPLEXES 833 the formal charges on the bonding oxygens might greatly enhance the energy contribution of the enolate resonance or, (b) carrying this one step further, resonance structures involving the copper atom, such as- c=o \ \ c-0 might contribute significantly in the description of the copper chelate molecule. Such struc- tures raise questions as to the availability and geometry of orbitals of the metal. I t is not yet possible to choose between these two views. Calvin and Bailes3 used a polarographic technique to study metallic chelates of a number of salicylaldehyde anils.Shifts of the polarographic half-wave potentials of related chelate compounds were taken as an indicatiop of their relative stabilities, on the assumption that the more stable a chelate the more difficult it is to reduce (the more negative the half-wave potential). They observed a most interesting effect in the great increase in stability accom- panying the increase in the number of rings in the chelate. For example, disalicylidene- methylamine copper, I, is not as stable as disalicylidene-ethylenediamine copper, 11. Note that the binding together of the two halves of the chelate in I1 through the methyl groups on the nitrogen atoms results in the formation of a third ring. The source of this great increase I: ella = +0.02 v.11: ella = -0.75 v. in stability may lie in the greater increase in entropy in the formation of “closed” chelates. Calvin also turned his attention to the effect of the metal ion in the stability of chelate+ and found that, with chelates of the salicylaldehyde type, the stability of copper, nickel, cobalt and zinc chelates is closely related to the second ionisation potentials of the gaseous metal atoms. With the exception of zinc, the ionisation involves the removal of an electron from a d orbital. This promoted the suggestion that in the transition elements a d orbital is involved in chelation. At the University of Pittsburgh we have initiated an A.E.C.-sponsored programme of study of the structure and stability of chelates of analytical significance.Messrs. Robert G. Charles and W. Dwight Johnston of this laboratory have obtained some interesting r e s ~ l t s ~ ~ 6 ~ 7 ~ 8 ~ ~ ~ ~ ~ (summarised in Table I , listing the ionisation constants of the reagents studied, and in Table 11, giving the log K values of the metal chelates) by the Calvin - Bjerrum titration technique. As 8-hydroxyquinoline (oxine; 111) is a most useful and versatile reagent, we first turned our attention to reagents of this type.’ The reactive grouping in oxine comprises a phenolic hydroxyl and a basic nitrogen in position to form a five-membered ring. The stabilities of the chelates of o-aminophenol, IV, whose reactive grouping is that of oxine, were also studied for purposes of comparison.6 SH R NH, I11 : 8-Hydroxyquinoline IV : o-Aminophenol V : o-Aminobenzenethiol The constants listed in Table I1 indicate a very high order of stability for chelates of 8- hydroxyquinoline.The stability of these chelates is greater (by a factor of 4 to 9 log K , units) than that of /3-diketones and substituted salicylaldehydes studied by Calvin and Wilson. This greater stability can be attributed to the fact that the basic character of the 8-hydroxy- quinoline nitrogen is greater, and the electro-negativity lower, than that of the carbonyl oxygens of the j9-diketones and substituted salicylaldehydes. Also, the phenolic group of834 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol. 77 TABLE I ACID DISSOCIATION CONSTANTS OF REAGENTS IN 50 PER CENT. PKNH* 8-Hydroxyquinoline ; I11 . . . .. . 3.97 8-Hydroxylepidine .. .. . . 4.59 8-Hydroxyquinaldine ; VI * . . . 4-51 1-(o-Hydroxyphenyl) isoquinoline ; VIII 4.3 1 2-(o-Hydroxyphenyl) quinoline; VII . . < 2 1 : 2 : 3 : 4-Tetrahydro-8-hydroxyquinoline . , 4-52 o- Aminophenol ; IV .. .. . . 3-75 o-Aminobenzenethiol; V . , .. .. <2 Dimethylglyoxime . . .. .. .. <2 2- (o-Hydroxyphenyl) -benzoxazole .. <2 O-Monomethyl ether of dimethylglyoxime < 2 2-(o-Hydroxyphenyl)-benzothiazoline . . <2 OF DIOXAN AT 25°C. pKOH* 11.54 11.69 11.55 11.59 12-77 11.73 11.57 7.90 12.19 12-84 11.46 10-18 * pKNH is the negative logarithm of the ionisation constant of the reaction 3NH' s N + H' while Basicity increases with pK. p & ~ is that of the reaction R-0-H f R-0' + H'. 8-hydroxyquinoline is a weaker acid than the (ph)enolic groups of the diketones and hydroxy- aldehydes.As, according to Calvin and Wilson, the forces holding a hydrogen ion to a molecule are similar to the forces that will hold a metal ion to the same molecule, the great stability of chelates of 8-hydrocyquinoline can be explained by the high basicity of its func- tional groups. o-Aminophenol, which has the same reactive grouping and virtually the same basicity as 8-hydroxyquinoline, forms metal chelates that are much less stable than the oxinates. This might be attributed in part to the hydrogen atoms in the amino-group of o-aminophenol, which could exert a steric blocking influence on co-ordination with metals. Also, in accord with Calvin and Wilson, the greater possibilities for resonance interaction in the 8-hydroxyquinoline chelates than those of o-aminophenol would account for their greata stability.The fact that tetrahydro-hydroxyquinoline forms chelates that are much less stable than those of oxine despite the greater basicity of its nitrogen strengthens the view that resonance and steric considerations can override those of basicity differences. In comparable compounds, e.g., o-aminophenol and 1 :2 :3 :4-tetrahydro-S-hydroxyquinoline, the stability of chelates is related to basicity. A comparison of the chelates of o-aminophenol, IV, and o-aminobenzenethiol, V, strengthens the view that the electro-negativity of the bonding atoms in the reagent is of significance in the study of chelate stability. For both zinc and lead, the only metals for which it was possible to determine o-aminobenzenethiol chelate formation constants, complexes were formed of significantly greater stability than the corresponding aminophenol complexes.6 The fact that the phenol has a more basic nitrogen and more basic phenolate would lead one to predict that the aminophenol would form more stable chelates than would the aminothiol.That this is not so is a good indication of the possibility that the strength of the bond of the metal with sulphur is greater than that with oxygen because of the lower electro-negativity of the sulphur and consequent increase in the covalent character of the bond. Merritt and Walker1' found that a methyl group in the %position of 8-hydroxyquinoline would prevent the reagent from reacting with a1uminium"I.t To study the more subtle manifestations of this supposed blocking effect, it was decided to determine the stabilities of a series of metal chelates of 8-hydroxyquinaldine, VI, and compare them with the stabilities of the corresponding chelates of 8-hydroxyq~inoline.~ On the basis of the greater basicity of the functional groups of 8-hydroxyquinaldine than those of oxine, correspondingly greater chelate stabilities might be expected.If, however, the methyl group exerts the type of hindrance to chelate formation for the bivalent metals that was encountered with aluminiumn' the oxinates might be found to be more stable. As can be seen from Table 11, the cobalt'I zincn, manganeseII and lead" chelates of the two reagents are of about equal stability. How- ever, the nickel" and copper1L chelates of 8-hydroxyquinaldine are significantly less stable than the corresponding oxinates.Because the steric blocking effect of the 2-methyl group would be more acutely felt in a planar chelate, our results could be explained by assuming t Irving, Butler and Ring,12 extending the study of Merritt and Walker, found that other 2-substituted 8-hydroxyquinolines failed to react with aluminium as did l-hydroxyacridine and 9-hydroxy-l:2 :3 :4- tetrahydroacridine. With all other metals these reagents behaved similarly to 8-hydroxyquinoline.836 Dec., 19521 SECTION 6 : ORGANIC COMPLEXES TABLE I1 CHELATE FORMATION CONSTANTS I N 60 PER CENT. DIOXAN- WATER SOLUTIONS AT 26°C 8-HydroxyquinolineJ III- Copperu. .. .. .. .. Nickel11 . . .... .. CobaltII . . . . .. .. Zinc11 .. .. .. .. Lead11 .. .. .. .. Manganese11 . . .. .. .. Cadmium11 . . .. .. .. Magnesium= .. .. .. Lanthanum"* . . .. .. CeriumI** . . .. .. .. 1 :2 :3 :4-Tetrahydro-8-hydroxyquinoli~~e- Copper11 . . .. .. .. o-Aminophenol, IV- Leadn .. .. .. .. Nickelu . . .. .. .. Zinc" .. .. .. .. Cobalt= .. .. .. .. Lead11 .. .. .. .. Zinc" .. .. .. .. Copper11 . . .. .. .. Nickel" .. .. .. .. CobaltII . . .. .. .. Zinc11 .. .. .. .. Lead" .. .. .. .. o-Aminobenzenethiol, V- 8-Hydroxyquinnldine, VI- Manganeseu . . .. .. .. Cerium111 . . .. .. .. Copper11 . . .. .. .. Nickel" . . .. .. .. Zinc11 .. .. .. .. Cobalt11 . . .. .. .. Lead11 .. .. .. .. Manganese11 . . * . .. .. 8-Hydroxylepidine- l-(o-H~j;droxyphe~y1yk)-is~ua'nolilze, VIII- Copper11 .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. * . Dimethy lglyoxime- Copper11 . . .. .. .. .. Nickel11 . . .. .. .. .. CobaltII . . .. .. .. .. Zinc11 .. .. .. .. .. Cobalt11 . . .. .. .. .. Copper= .. .. .. .. .. Zinc11 .. .. .. .. .. Nickel11 . . .. .. .. .. Manganese11 . . .. .. .. .. Cadmium11 . . .. .. .. .. Nickel11 . . .. .. .. .. Lead* .. * . .. .. .. Cobalt11 . . .. .. .. .. Iron" .. .. .. .. .. ZincU .. .. .. .. .. Lead11 . I .. .. .. .. LanthanumInt . . .. .. .. Nickel" . . .. .. s. .. Manganese11 . . .. .. .. .. O-Monomethyl ether of dimethylglyoxime- 2-( o-Hydroxyphenyl) -ben.zoxazole- 2-( o- Hydroxyphenyl) -benzothia.zolifie- log Kl 13.49 11.44 10.55 9.96 10.61 8.28 9.43 6.38 8.66 9.15 10.58 6-29 6.10 5.99 5-81 8-41 7.33 13.00 (calculated) 9.50 10.65 10.10 10.35 7.72 7.7 1 13.90 11.58 10.69 10.53 11-15 8.32 10.72 11-44 10-46 9.80 7.92 10.0 9.6 7-45 6.36 6-15 6.47 8.54 7.65 9.76 9.20 7.89 7.57 7.30 6.55 5-80 1% K, 12-73 9-94 9.11 8.90 8-09 7.17 7-68 5.43 7.74 7-98 8-43 4.05 4.87 4-96 4.69 6.96 6.77 11.64 8.50 9.55 9.07 8.25 6.84 (extrapolated) (calculated) - - 10.77 9.53 9-41 13-20 7-25 9.02 (calculated) 10.81 (;:;k 7.28 7.1 6.6 6.85 5.65 5.51 5.13 7.78 6.0 8.39 6.83 5-79 6-09 5-97 5.30 - log Kav 13.11 10.80 9.83 9-36 9-30 7.67 8-50 5.93 - - 9.51 5.17 5.57 5.48 5.26 7.69 7.05 12.32 8.98 10.10 9.54 9.30 7.29 - (14-2 & 0.2) 11.18 10.1 1 9.97 9.68 7-79 9.87 (extrapolated) 11.12 9.48 7.60 8.6 8.1 7-15 6.01 5.83 5.80 8-16 6.8 9.08 8.02 6.84 6.83 5.93 (10.2) - - * K, not determined. t log K, = 5-33.%6 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol. 77 that the copper'I and nickel* chelates are planar compounds while the remainder of the transi- tion metal chelates studied have a tetrahedral configuration. In order to establish firmly the fact that steric hindrance was responsible for the marked decrease in stability of the copper and nickel chelates of 8-hydroxyquinaldine, stability meas- urements were carried out on chelates of 4-methyl-8-hydroxyquinoline (8-hydroxylepidine) .It was felt that the inductive and hyperconjugative effects of the methyl groups in both com- pounds would be nearly equal. This is indicated in the similarity of the acid dissociation constants of these compounds. As can be seen from Table 11, all of the 8-1iydroxylepid.inate.s are more stable than the corresponding 8-hydroxyquinolinates, which is in keeping with the greater basicity of the lepidine reagent.The chelates of 8-hydroxyquinoline and the two methylated analogues are profitably compared by plotting the second ionisation potential of the gaseous metal atom against the logarithm of the stability constant of the chelate. It was demonstrated by Calvin and Mel- chior that, for the transition metals, a linear relationship exists between these two variables. As can be seen from Fig. 1, the curves for 8-hydroxyquinoline and 8-hydroxylepidine are nearly Linear, while for 8-hydroxyquinaldinc the discontinuity with copper and nickel is immediately apparent. In this way, the steric hindrance for copper and nickel of the methyl group in the 2-pasition is clearly illustrated.In order to evaluate the effect of ring size on chelate stability, the chelating tendencies of 2-(.o-hydroxyphenyl)-qi1inoline, VII, and 1-(0-hydroxypheny1)-isoquinoline, VIII (which represent six-membered-ring analogues of 8-hydroxyquinoline) , were studied.' The iormer VI : 8-Hydroxyquinaldine VII : 2-(o-Hydroxyphenyl) -quinoline VIII : 1-(0-Hydroxypheny1)- ssoquinoline 8-Hydroxyquinoline 8-Hydroxyquinaldine / /pDec., 19521 SECTION 6 : ORGANIC COMPLEXES 837 showed no tendency to chelate with metals, and this may be attributable to steric hindrance, as this compound has the benzene portion of its quinoline nucleus in a position to exert adverse steric influence.Lack of activity in structurally related compounds has been ob- served in several instances. Hostel3 reported that no chelation with bivalent metals occurred with 2-pyridyl-2’-quinolyl and other related compounds. This is not true of the corresponding isoquinoline compound, VIII, which forms chelates of comparable stability to those of o-aminophenol, IV. The greater basicity of this isoquinoline reagent than 8-hydroxyquinoline would lead one to expect greater chelate stabilities. These results seem to show that, a t least for oxine analogues, five-membered-ring chelates are more stable than those with six-membered rings. The behaviour of 2-(o-hydroxyphenyl)-benzoxazole, IX, as a chelating OH OH OH H IX: 2-(o-Hydroxyphenyl)- X : 2-(o-Hydroxyphenyl) - XI : 2-(0-Hydroxypheny1)- benzoxazole benzothiazole benzothiazoline reagent provides an interesting contrast to that of 2-(o-hydroxyphenyl)-quinoline, VII. Whilst the latter failed to react with any of the metals tested, the benzoxazole formed well- defined chelates with many bivalent metals, although no chelate formation with any tervalent metal but iron was observed, I t is possible that there is a greater distance between the N and 0 atoms in benzoxazole than in the quinoline compound and hence less steric hindrance in the former.Whilst the low solubilities of the complexes of 2-(o-hydroxyphenyl)-benzoxazole prevented us from obtaining the stability constants of all but two of the metals studied, the data was sufficient to establish the relative stability of these complexes.8 The order was found to be Cu>Co>Ni>Zn>Pb.I t will be noted that the order for nickel and cobalt are reversed from that found with 8-hydroxyquinoline. The interesting results with the oxazole prompted us to investigate the properties of 2-(o-hydroxyphenyl)-benzothiazole, X.9 Although this compound is described several places in the literature, its reactions with metals had never been reported. In each investigation it was prepared from the reaction of o-aminobenzenethiol with salicylaldehyde, presumably with the formation of an intermediate thiazoline, which was oxidised in air to the thiazole. A sample of the substance prepared by this method gave reactions with metals that were so different from those of the oxazole that we questioned whether it was actually the thiazole.Hence an attempt was made to prepare the thiazole by another method. In analogy to the reaction employed for the oxazole, o-amino benzenethiol was allowed to react with salicylamide. The product of this reaction had nearly the same melting point as that of the first sample prepared, but it reacted similarly to the oxazole. Infra-red and ultra-violet spectra compari- sons confirmed our suspicions as to the difference between the two compounds and the presence of an absorption band in the infra-red spectrum at 3.1 mp in the first sample identi- fied it as the thiazoline, XI. The thiazoline reacts with most of the common metal ions including the tervalent ions. This difference in selectivity is difficult to explain. I t is possible that the lack of saturation makes the molecule more flexible, so reducing steric hindrance. Stability measurements with this reagent and a number of metals have been made.As can be seen from Table 11, the general magnitude of these stability constants is about the same as that found for the oxazole chelates. However, the order of metals is quite different. Indeed, this reagent seems to behave differently in this respect from any other studied. Whilst no explanation can be offered at the present time for this puzzling behaviour, it is felt that the reactions of the thiazoline contain a very important clue to the solution of the problem of the nature of the interaction of metals and chelating reagents. Determinations of the stability constants of the true benzothiazole are currently in progress.Although no constants are yet available, preliminary indications are that the complexes are somewhat less stable than those of the oxazole. The analytical significance of this will be discussed below. The question of stability sequence of the metals is very interesting, and although the general order of metals is very nearly the same from one reagent to another, following fairly well the order of the second ionisation potentials of the gaseous metal atoms (Calvin and838 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [VOl. 77 Melchior), there are some differences. For example, lead and zinc in the aminobenzenethiol series, and nickel and cobalt in the 8-hydroxyquinaldine and 2- (0-hydroxypheny1)-benzoxazole series, are not in the usual stability order. The stability order of the 2-(o-hydroxyphenyl)- benzothiazoline was quite unusual.It was with this specificity question in mind that we turned our attention to the deter- mination of stabilities of some of the chelates of dimethylglyoxime, XII, a most “specific” reagent.lO It is noteworthy that the sequence is like that found for the other reagents and that copper forms a more stable complex than nickel. This would seem to indicate that the famed specificity of dimethylglyoxime toward nickel has its basis in some property other than stability, probably in the solubility differences in the chelates. Whilst specificity of a reagent toward a particular metal might be viewed in terms of a change in the place of that metal in the stability order of the reagent, the term “specificity” might be extended to cover examples where the effect is not large enough to displace a metal from its usual place in the order, although the change in stability would be large enough to give a bigger separation of other metals.Stability measurements were extended to the 0-monomethyl ether of dimethylgloyxime. This compound is of particular interest, as the structure of the nickel salt of dimethylglyoxime, XII, is known to involve hydrogen bonding, whilst no such hydrogen bonding is possible with the 0-ether, XIII. It is also probable that hydrogen bonding plays some part in the structure CH3 - L C - CH, II * N II N 0 / ‘b Ni ,’ \ 0 It II CH,-C-------(!!--CH, H CH3- C;------------C- CH, XI1 XI11 of other complexes of dimethylglyoxime, particularly with those that are planar.It is sig- nificant, therefore, that the results obtainable for the 0-ether show a number of striking differences from those obtained for dimethylglyoxime itself. The order of decreasing stability for the complexes of the 0-methyl ether is (Co”, Cu), Zn, Ni, Mn, Cd. This order differs considerably from that reported for dimethylglyoxime, particularly with respect to nickel, which is scarcely more stable than manganese with the 0-ether. Another fact of interest is that the copper and cobalt’I complexes of the 0-ether show an abnormally large spread between the values of log K, and log K,. A final important fact is that the complexes of the 0-ether as a whole are of the order of 2 or 3 log K units less stable than those of dimethylglyoxime, with respect to KaV.At least two factors must be considered in explaining the above facts: (i) the absence of hydrogen bonding in the 0-ether complexes and (ii) steric hindrance associated with the 0-CH, group. Since the lack of hydrogen bonding, as such, would not be expected to lead to an abnormal stability order, steric hindrance may be the more important of the two effects, The highly abnormal position of nickel in the stability order, however, suggests that the lack of hydrogen bonding may be an important factor here. If steric hindrance is responsible for the differences in metal order, one would expect a correlation between the size of the entering ion and log K, for dimethylglyoxime minus log K, for 0-ether for a given metal.Work is in progress to test this hypothesis. It is significant that there was no evidence of complex formation between the 0-ether and leadLx, the largest atom used, under the conditions pre- vailing. The consistently lower stabilities of the 0-ether complexes as compared with those of the complexes of dimethylglyoxime is also probably due to steric hindrance, although the absence of hydrogen bonding may be partly responsible. It should be mentioned that this lowering of stability cannot be attributed to a difference in K, for the two compounds, as these values are nearly the same. From the results found with the 0-monomethyl ether of dimethylglyoxime and with 2-(o-hydroxyphenyl)-benzothiazoline, the two compounds with which unusual stability orders were observed, it would appear that a search for specific reagents would be more fruitfulDec., 19521 SECTION 6: ORGANIC COMPLEXES 839 among those reagents in which steric hindrance can operate and the apparently less stable six-membered ring arrangement occurs.It might be of interest to point out briefly some of the analytical applications of a few of the reagents discussed above. The precipitation reactions of 2-(o-hydroxypheny1)-benzox- azole were carried out with many of the common metal ions under several sets of experimental conditions. In an acetic acid - acetate buffer, copper, nickel and cobalt were the only metals to give well-defined reactions. In the presence of an ammonia buffer, precipitation also occurred with calcium, magnesium, zinc, cadmium, lead and manganese.The use of a sodium hydroxide - sodium tartrate buffer limited the reaction to copper, cobalt and cadmium. As copper and cobalt could be precipitated at lower pH values, this reagent was used by Mr. J. Walter of our laboratory to develop a fairly selective method for determining cadmium.l4 Evidently the relative stabilities of the chelates and the tartrate complexes are such as to permit the precipitation of cadmium, although this chelate is probably less stable than those of zinc and lead, which remain in solution. This interaction between a chelating agent and a soluble complexing agent appears to offer promise of making other interesting separations with other precipitating agents where the differences in stability between the chelates and the soluble complexes are not too great.The use of such “masking” agents is not novel, but much work remains to be done before their use can be adequately systematised. 2-(o-Hydroxy- pheny1)-benzothiazole resembles the oxazole in its reactions with metals, but is even more selective, reacting only with copper in acetic acid - acetate buffer, with copper, cadmium, lead, cobalt and nickel in ammoniacal solution, and apparently only with cadmium in sodium hydroxide - sodium tartrate buffer. These reactions appear to indicate that in some instances decreased chelate stability can result in greater selectivity of an organic precipitating reagent. The precipitation reactions of 2-(o-hydroxypheny1)-benzimidazole with metal ions were even more limited than those of the oxazole and thiazole compounds.15 In an acetic acid - acetate buffer mercuryII was the only ion of those tested that gave a precipitate with the reagent.During the subsequent development of an analytical method for mercury based on this reaction, it was found that precipitation was incomplete in the presence of copper, cobalt, bismuth and aluminium unless enough reagent was added to take care of the interfering metal. This is indicative of the fact that the imidazole forms chelates with other metals besides mercury but that these are soluble. This situation is reminiscent of the reaction of copper with dirnethylglyoxime, where the chelate formed is quite stable but fairly soluble in water. In considering the factors upon which depend the success or failure of an analytical separation of two metals by means of an organic chelating reagent, the stability difference and the solubility difference are seen to be most important.In a precipitation process it is the solubility of the chelate in water that must be considered, whilst in a solvent extraction process not only the aqueous solubility but also the solubility in the organic solvent must be taken into account. In the usual formulation of the solubility equilibria involved in the precipitation of a metal chelate, no attempt is made to differentiate between stability and solubility. Often the expression for the solubility product of the chelate is written in the same manner as for an ionic salt (see, for example, Flagg,l6 in which the work of Treadwell and Ammannl’ on the solubility of some metal chelates is discussed).It would be more appropriate to formulate the equilibria as follows- Kf Chelate formation: M” + 2R’ + MR, Chelate solubility : MR, (solid) + MR, (solution) From this, the solubility product constant, K,.,, can be shown to be- S C - (M”)(R’)2. 3 K8.P. = - Kf The separation of a pair of metals, M, and M,, can be considered in terms of their concen- tration ratio at equilibrium- If the solubilities of the chelates of the two metals are not widely different then the Conversely, stability difference will determine the ease of separation of the metal pair.840 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol. 77 under these conditions, the solubility product constants or, more conveniently, the pH values of precipitation could be used as measures of stability.Merrittl* has used the pH of precipita- tion of metallic 8-hydroxyquinolinates to evaluate their stability. This is not a generally applicable technique, owing to the assumption about the similarity of the soh bilities of the chelates involved. The chelates of dimethylglyoxime provide a most dramatic exception ; although the copper complex of dimethylglyoxime is more stable than the nickel complex, it is the latter that precipitates first. Separations based on selective precipitation are complicated by the occurrence of co- precipitation. It is impossible to predict accurately the extent of precipitate contamination, which will reduce the efficiency of separation. Comparatively little work has been done in the study of co-precipitation phenomena in metal chelate precipitates.Removal of con- taminating cations can be most easily accomplished with multiple precipitations. As with precipitation processes, solvent-extraction methods involving chelates depend on solubility as well as stability differences. In the manner of Furman, Mason and Pekola,19 the distribution of a metal between two phases can be described in terms of D, the ratio of the concentration of metal in the organic solvent to that in the aqueous phase, as follows- where K, is the partition coefficient of the organic reagent, HR, K, is the ionisation constant of HR, K, is the stability constant of the metal chelate, and K, is the partition coefficient of the chelate. A measure of the feasibility of the separation of two metals would be given by the ratio of their D values, or- D’ - KjK; - D - K,K, From this expression it can be observed that a study of the organic solvent used in the extraction, which would affect the K, value and hence possibly the efficiency of the separation, would be of use.We are in the process of carrying out such a study and hope to be in a posi- tion to report our findings in the near future. Despite the occurrence of non-equilibrium conditions in solvent extraction separationsao it is quite likely that more difficult separations could be achieved by this technique than by precipitation methods. The financial assistance of the U.S. Atomic Energy Commission with this work is gratefully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.17. 18. 19. 20. REFERENCES Calvin, M., and Wilson, K. W., J . Amer. Chem. Soc., 1945,67, 2003. Bjerrum, J., “Metal Ammine Formation in Aqueous Solution,” P. Haase & Son, Copenhagen, Calvin, M., and Bailes, R. H., J . Amer. Chem. Soc., 1948, 68, 949. Calvin, M., and Melchior, N. C., Ibid., 1948, 70, 3270. Freiser, H., Charles, R. G., and Johnston, W. D., J . Amer. Chem. Soc., 1952,74, 1383. Charles, R. G., and Freiser, H., Ibid., 1952, 74, 1385. Johnston, W. D., and Freiser, H., Ibid., 1952, (October) Charles, R. G., and Freiser, H., in the press. -- , in the press. Merritt, L. L., jun., and Walker, J. K., Ind. Eng. Chem., Anal. Ed., 1944, 16, 387. Irving, H., Butler, E. J., and Ring, M. F., J . Chem. Soc., 1949, 1489. Hoste, J., Anal. Chim. Acta, 1950, 4, 23.Walter, J., and Freiser, H., Anal. Chem., 1952, 24, 984. -- , in the press. Flagg, J. F., “Organic Reagents,” Interscience Publishers Inc., New York and London, 1948. Treadwell, W. D., and Ammann, A., Helv. Chim. Acta, 1938, 21, 1249. Merritt, L. L., jun., Rec. Chem. Prog., Kresge-Hooker Sci. Lab., 1949, 10, 59. Furman, N. H., Mason, W. B., and Pekola, J. S., Anal. Chem., 1949, 21, 1325. Irving, H., and Williams, R. J. P., J . Chem. Soc., 1949, 1841. 1941. , , in the press. -- UNIVERSITY OF PITTSBURGH PITTSBURGH 13, PENNSYLVANIA, U.S.A. March 28th, 1952Dec., 19521 SECTION 6 : ORGANIC COMPLEXES 841 DISCUSSION DR. H. M. IRVING (Oxford) said he was impressed by the wealth of new data on the stability constants of metal complexes and the way in which the measurements had been discussed and interpreted.There were three main points on which he wished to comment. 1. As he was dealing with very water-insoluble metal chelates, Professor Freiser had had to work in a mixed water - dioxan solvent. This introduced a number of new features and difficulties of inter- pretation. In the first place, the effect of increasing organic content of the solution was to increase stability constants so that data obtained in this way were not directly applicable to the discussion of reactions in aqueous solution. Measurements a t Oxford, where a number of metal complexes had been studied in a range of water - dioxan mixtures, had shown that increasing organic content did not affect successive stability constants, k,, k, . . ., to an equivalent extent.Secondly, the general increase in the magnitude of stability constant with increasing organic content varied from reagent to reagent and was not the same even for chelates of any one reagent with a series of metals. Comparison of data for the stabilities of metal oxinates in water (Albert, Biochern. J., 1950, 47, xxvii), in 50 per cent. dioxan (Freiser, this paper) and in 70 per cent. dioxan (Mellor and Maley, Austr. J . Sci. lies., 1949, 2, 92) supplemented their own (un- published) data in showing that the order of stabilities might change when the solvent was changed. Dr. Irving was therefore reluctant to draw elaborate theoretical deductions from the limited data at present available, and asked Professor Freiser whether he had made any fundamental studies of titrations in dioxan - water mixtures, how he set up his pH standards and how he calibrated his glass-electrodes.He was reluctant to follow Professor Freiser in his interpretation of the inverted order of stability of sterically hindered oxines based upon a measured stability cobalt 2-methyl-8-hydroxyquinolinate > nickel 2-methyl- 8-hydroxyquinolinate. Measurements by his own pupils in 50 per cent. dioxan had given substantially different numerical values. He did not wish to suggest that these were more or less reliable than those of Professor Freiser, but he felt that their most significant feature was that the titration curves for cobalt and nickel were indistinguishable throughout their length. The same behaviour had been previously noted by Calvin and Melchior in titrations of formylnapthols in aqueous alcohol. Such factual discrepancies could probably be resolved when differences of technique and computation had been talked over.Professor Freiser referred in his paper (p. 836) to work by Calvin and Melchior and quoted them as saying that they had established that there was a linear relationship between the logarithm of the stability constant and the second ionisation potential of the gaseous metal atom for the transition metals. Professor Freiser made use of this plot in his Fig. 1 and used it diagnostically. Dr. Irving pointed out that there appeared to have been a misreading of Calvin's paper, for this referred only to complexes of four metals, Co, Ni, Cu and Zn, with one ligand, zliz., salicylaldehyde-5-sulphonic acid.The American authors had plotted the logarithm of the stability constants, the first ionisation potential, and the second ionisation potentials against atomic number and noted that the only similar plot was obtained with the second ionisa- tion potential. There was no suggestion of specific functional relationship between complex stability and ionisation potential or any attempt to 'generalise to other metals and other ligands. Dr. Irving further pointed out that the second ionisation potential used in this paper referred to the process- M++ M++ + e Now in a paper published in 1948 (Nature, 1948, 162, 764), Dr. Williams and he had pointed out that there was an approximately linear relationship between the logarithm of the instability constant of many com- plexes of transition metals from manganese to copper and the second ionisation potential as defined by the reaction- This was equivalent to saying that there was a relationship between the stability of a complex and the electron affinity of the bivalent transition metal ion concerned-which would certainly appear to be more fundamental than the correlation noted by Calvin and Melchior.In a forthcoming paper ( J . Chem. SOL, in the press) the significance of ionisation potential and ionic radii would be discussed in greater detail. To anticipate, it could be said that they no longer considered the second ionisation potential to be of over-riding importance when a general treatment of complexes of all metals was attempted, for here considerations of partial molar free energies might be more relevant.But within the limited group of the first transition series, it certainly could act as a rough and ready guide to behaviour. Professor Freiser had made a most interesting comparison between oxine (VI), his ideally specific (but practically useless !) reagent (VII) and the isomeric isoquinoline derivative, VIII. In addition to the change from a five-membered ring to a six-membered ring, which should weaken complex stability and more than offset the gain expected from the greater basicity of the nitrogen atoms in VIII, it was worth noting that steric hindrance, by preventing coplanarity in the case of VIII, would be a further factor reducing complex stability. On the whole, however, this choice of compounds to bring out an important correlation between ring size and stability was not ideal, for after chelation each molecule of VI would form a system of three rings in a resonating structure, while each molecule of VIII would participate in a structure with four. Increase in the number of aromatic rings involved in resonance appeared invariably to lead to greater stability (cf.ferrous his-dipyridyl with nine rings, log K 3 e 17, and ferrous tris-o-phenanthroline with 12 rings and log K, = 21.3). At Oxford they had measured the stability constants of complexes of a variety of transition metals with a number of simple ligands, e.g., NH,CH,(CH,),NH, and NH,(CH,),COOH (where n i= 1 or 2) and HOOC(CH,),COOH (where n = 1, 2 or 3). Here complications due to resonance were 2. M(so1id) --+ Mt+(gas) $: 213 3.842 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol.77 avoided and the expected decrease of stability with ring size was clearly demonstrated. Further data were provided by the measurements of Schwarzenbach on complexes formed between sexadentate “com- plexones” and alkaline earth metals. A full account of this work had been written up and would shortly be published. PROFESSOR FREISER replied to Dr. Irving’s three comments as follows- 1. The difficulty of making valid comparisons of stability constants obtained under different experi- mental conditions, particularly different solvent media, became insurmountable unless the greatest possible precautions were observed. Appropriate corrections must be applied to the observed pH readings, especially in solutions containing more than 50 per cent.of dioxan, owing to changes in hydrogen-ion activity coefficients, properties of the glass electrode, and so on. At Pittsburgh, such corrections were made by comparing an experimental titration curve of perchloric acid and sodium hydroxide with a theoretical one. This practice was in substantial agreement with that described in the careful study by Van Uitert, Fernelius and Douglas (US. Atomic Energy Commission Report NYO-729). Unfortunately, Maley and Mellor did not seem to have applied such a correction, which, in the medium they used (70 per cent. dioxan), was appreciable. In his work (Biochenz. J., 1950, 47, 631), metal salts whose anions (chlorides and sulphates) have considerable co-ordinating ability were used.Rather than a measure of the equilibrium constant for the formation of the chelate from the hydrated metal ion, the data he obtained represented the equilibrium between the anion-co-ordinated metal and the chelate. Following the practice of Calvin and Wilson, the Pittsburgh group has used metal perchlorates in order to minimise the chances of dealing with anion co-ordination complexes of the metal ion. Professor Freiser would be most interested in the results of the Oxford study of the effect of solvent composition on stability constant values. Calvin and Melchior’s graphical representation, in which the curves relating the second ionisation potential of the gaseous metal atom and the log Kstab.. respectively, and the atomic number of the metal were clearly parallel, was equivalent to saying that the relation between log K and the second ionisation potential was approximately linear.Calvin and Melchior did suggest that the relation between the second ionisation potential and log K indicated that a d-orbital was involved in the chelate binding, while the paper by Irving and Williams, referred to by Dr. Irving, made no suggestion as to functional relationship. As for using the sum of the first and the second ionisation potentials for correlation purposes, this gave poorer correlation than the second ionisation potential alone when zinc was included in the series. However, as the first ionisation potential of the transition metals was about the same, the sum of the two potentials gave results no different than did the second alone when zinc was not included. (Martell and Calvin, “Chemistry of the Metal Chelate Compounds,” Prentice-Hall, New York, 1952, p.189). Professor Irving’s discussion of “complications due to resonance” in connection with our com- parison of VI and VIIl served to strengthen our correlation of ring size and stability, i.e., despite the increased number of aromatic rings in VIII, it formed weaker chelates than did VI. PROFESSOR FREISER, in a written reply, added that as a result of Dr. Irving’s remarks and also as a result of work on the temperature dependence of stability constants (which indicated an error in the cobalt hydroxyquinaldinate value) that had been carried out a t Pittsburgh, the cobalt hydroxyquinaldinate stability was redetermined.Repetition of the titrations with more time allowed for the attainment of equilibrium yielded the following revised figures for the chelate formation constants of cobalt11 and 8- hydroxyquinaldine in 50 per cent. dioxan solutions a t 25” C- in place of the figures given ih Table 11. These values could be compared with that observed by Mrs. Rossotti of Oxford, log Ka, = 8-75 (private communication from Mr. F. J. C. Rossotti). The graphical representation shown in Fig. 1 still had much to recommend it in that the discontinuous curve lent emphasis, by grouping copper and nickel together, to the aggravated effect of steric hindrance on the stability of chelates having planar configuration. However, as Mr. Rossotti pointed out, the data could be better represented by one straight line with only nickel exhibiting a marked deviation.This could be seen in the revised plot in Fig. la. DR. J. F. DUNCAN (Harwell) said the attempts that had been made by previous authors to relate the stability constants obtained with a given chelating agent and a series of cations seemed to him to be quite arbitrary and empirical, and not such as would lead to any clear understanding of the variations obtained. This problem was, however, quite straightforward, as a simple discussion of the thermodynamics of the system would show. The utility of Albert’s data for purposes of comparison was also open to question. 2. 3. log I<, =: 9.68 log K, = 8.92 log Ka, = 9.27 For a ligand reaction of the type- ML + M+ + L- it could be shown that the experimentally observed stability constant, I<, was related to the thermodynamic equilibrium constant, Ka, by the relation- where A could be taken as a constant for a given ligand provided the solutions were dilute and there were no wide changes in solution composition.log K, = log K + A Then- RT log I<, = p”c - p o ~ - POLDec., 19521 SECTION 6 : ORGANIC COMPLEXES 843 where the po's were the standard partial molar free energies of the complex, the metal ion and the ligand ion respectively. Now was constant for a given ligand, and p0c could be shown to vary only slightly. Hence, provided p o ~ could be evaluated, changes in K with metal ion could be predicted. The term p o ~ could be expressed in a number of ways, which were revealing in that they showed the inadequacy of some of the plots that had previously been used.1. Plotting log K would only give a linear graph with the first or second ionisation potential or any other physical parameter if these were proportional to p o ~ . The ionisation potentials, a t least, were known not to be. 2. If p o ~ were expressed in terms of the corresponding heat and entropy and, further, a parallel behaviour were assumed between these two quantities, then the finding of linear relations between log K and either the heat or the entropy might be expected. Various heats for reactions related to the formation of an ion in solution had been used previously with reasonable regularity. He had chosen to express log K as a function of the entropy, as this, in turn, might be expressed as a function of M, the mole weight of the They were as follows- 15- Pb- 8-Hydroxyquinoline 8-H yd roxyquinaldine A0 rL 8-Hydroxylepidine log K,, Fig.la ion, and r, the effective ionic radius. From these considerations i t could be shown that a straight line plot of log K against l/r2 would be expected provided M was effectively constant and also that the heat term was either constant or proportional to the entropy term. In fact, neither of these conditions was valid except for a very restricted series (e.g., the transition metals), which accounted for the very large discrepancies obtained in plots depending merely on the ionic radius. This method of expressing the free energy was therefore quite inadequate. Rather than utilise any approximate evaluation of p o ~ , it was sounder to take these values directly from tabulated data.If this were done, quite good straight line plots between log K for a given ligand and p o ~ were obtained. This regular variation of K with the tabulated free energies was strikingly revealed by the series tabulated for tervalent metals by Dr. Irving and Dr. Williams. Except for chromium, the series fell exactly in line with the order of the free energies of the metal ions, and further, the differences in stability constant quoted by Dr. Irving and Dr. Williams were much as would be expected for the free energy values. By utilising the equations given above, it could be shown that for two different ligands, P and Q, and a given metal ion- where B was a constant, independent of the metal ion.Hence, a plot of log Kp against log KQ should yield a straight line of unit slope. It should be emphasised that such relations would only be obtained (a) when there were no steric effects or other causes that would introduce further free energy terms in the above equations (or a t least, when such terms were the same for both ligand pairs), and (b) in cases where the terms (poc)p and ( ~ O C ) Q could be assJmed to be constanti.c., for ligand pairs of the same ionic type. The importance of such relations was two-fold. First, if the stability constants for a single metal ion with two ligands of the same type in which steric and 3. 4. log (K~/KQ) = B This was in fact obtained for a large number of ligand pairs.844 INTERNATIONAL CONGRESS ON ANALYTICAL CHEMISTRY [Vol.77 allied effects were absent were known, a line of unit slope could be drawn through this point with log Kp and log KQ as axes. This line would then give the stability constants to be expected (in the absence of steric effects) with one ligand, for any ion for which the stability constant was known for the other ligand. Secondly, in cases where steric effects did occur, the quantity RT log KExpt/KTh. (where K E ~ ~ ~ and KTh were the experimentally observed and the theoretically expected stability constants) was a direct measure of the free energy involved in the steric effect. This should be of value in enabling an assessment to be made of the contribution of such factors in determining measured stability constants. PROFESSOR FREISER replied that Dr.Duncan had made a valuable contribution to the problem of the significance of chelate stability data by his thermodynamic approach. Professor Freiser found it a little difficult to agree that p"c, the partial molar free energy of the chelate, did not vary significantly. He and his colleagues shared Dr. Duncan's enthusiasm for the study of thermodynamics of chelate formation and had recently undertaken such a study a t the Wniversity of Pittsburgh. Dr. R. G. Charles, by measuring the temperature coefficient of chelate stability, had been able to calculate AH" and ASo for the formation of copper and nickel dimethylglyoxime complexes (paper in the press). His results indicated that much of the driving force of the formation of these complexes arose from the large entropy change accompanying the reaction. Indeed, in the case of the nickel complex, AH" was slightly positive, being about 1 kcal.per mole. These results might help explain the great difference in the solubility of the copper and nickel dimethylglyoxime complexes in that they indicated the possibility that the nickel complex co-ordinated little or no water while the copper complex might have water of co-ordination. More work of this kind was in progress. DR. DUNCAN added, in reply to Professor Freiser, that he would emphasise that, although it was necessary to use published experimental data to show the regularity of the relations, any deviations obtained in no way invalidated the treatment. If such deviations were not due to experimental inaccuracies (which they had no means of judging) , then it followed that the deviations indicated a real free energy contribution from secondary factors. The next step was clearly an assessment of the values of these, with an attempt to relate them to other parameters of the ligand molecular structure. DR.13. M. IRVING (Oxford), in a written contribution, said that it would be inappropriate to comment here a t length on Dr. Duncan's contribution. A paper by Dr. R. J. P. Williams and himself in this field which they had expected would have been in print before the Congress met, would sufficiently indicate to what extent the treatment was novel and the simplifying assumptions justified. DR. R. J. P. WILLIAMS (Oxford) said that with regard to Dr. Duncan's remarks, he would like to say that in comparisons between the stabilities of complexes of elements whose ions were of very different sizes, steric factors were overridingly important.He would draw attention to this point in connection with the group IIA ions in a paper to be published soon in the Journal of the ChemicaZ Society.* He had constantly referred to an averaged complexity constant, log K,Kz/2. This treatment missed the significance of the difference, log K, - log K, (Alz). A,, was usually of the same ofder for all the metal complexes of the Irving - Williams series-about one logarithm unit-but the small differences in Ala that did arise tended thcmselves to follow the Irving - Williams sequence, e.g., in oxinates, glycinates and ethylene- diamine complexes. Deviations from such a rule would be expected when steric hindrance or changes in orbital configuTation of the central metal ion interfered with normal complex formation. He called attention to the peculiar values in the formation of /3-alanine complexes, o-phenanthroline complexes of iron and in some of the author's data. PROFESSOR FREISER replied that they had hesitated to call attention to Alz values when these were small.DR. JOHN F. FL.4GG (Schenectady, N.Y.) said that in this and the preceding paper the question had been raised concerning the physical state of the dissolved chelate. This was of interest in formulating the solubility product relation. The author felt that a molecular non-dissociated form of the chelate existed in solution a t very low concentration. While no definite proof for either dissociated or undissociated chelate could be cited, the following should be considered in favour of the presence of only ionised chelate: (a) dilution was great; (b) hydrogen ion from dissociation of water was always present; (c) exchange between metal ion in solution and solid chelate, measured with radioactive tracers, was very rapid with certain hydroxyquinolates. It appeared very difficult to divorce the stability and solubility considerations. He also asked whether the author would comment on any results of his studies that might a t this time be of interest in practical analysis, Le., new reagents, increased sensitivity, separation and so on.PROFESSOR FREISER replied that in the study of the course of reactions of the type- M"* nHR +MR% + nH+ under conditions in which precipitation did not occur, the observation of the liberation of hydrogen ions as predicted by the equation was proof of the existence of a non-dissociated chelate species. Professor Freiser's data were interesting in one particular to which he had not called attention. Where these were unusually large, suitable mention was made (see p. 838). * This paper has since appeared in J . Chem. SOC. 1952, 3770.Dec., 19521 SECTION 6 : ORGANIC COMPLEXES 845 With regard to the relation of this work to practical analysis, the author would like to emphasise the points discussed in the latter part of the paper (starting a t the bottom of p. 838). They would look for other reagents having unusual stability orders among compounds in which steric hindrance and the apparently less stable six-membered ring arrangement occurred. Reagents giving rise t o chelates of intermediate stability seemed to be more easily adaptable to selective analytical methods, particularly when used in conjunction with “masking agents.” MRS. H. S. ROSSOTTI (Oxford) asked the author if his stability constants were obtained by the same method as that described in reference 5 of his paper. Recalling that he prepared his titration solutions by mixing equal volumes of standard metal perchlorate solution in aqueous acid and of standard reagent solution in pure dioxan, she mentioned that an appreciable contraction occurred when water and dioxan were mixed. Hence, the concentrations of metal, of acid and of reagent would be rather greater than if there were no volume change. She pointed out that knowledge of the absolute concentrations of these species was needed to obtain the stability constants of the chelate system from the pH titration curve (as the calculation involved the absolute term [HI, which was given by the measured pH) and asked if the volume change had been determined, so that the concentration changes might be calculated. Commenting that Professor Freiser gave his values of logloK to two decimal places (implying an accuracy of more than 0.1 per cent. for values of logloK greater than lo), she suggested that the concentration changes mentioned might have an appreciable effect on results of this accuracy. PROFESSOR FREISER replied that the volume contraction that occurred on mixing dioxan and water was noted, but not corrected for. As it amounted to less than 2 per cent., the effect on the pH was less than 0.01 unit.