7/28/2019 Chem rev VOL 1 091_141

1/51

ORGANIC RADICALSM. GOM BERG

Univers i t y of M i ch i gan , Ann Arbor, M i ch i ganI. INTRODUCTION

DeJinition of radical. The concept of radicals, originallyintroduced by Lavoisier and later developed more fully byBerzelius, Dumas, Liebig, and then again by Bunsen, Kolbe,Frankland and others, implied that in the molecules of an or-ganic substance there is at least one immutable cluster of atomswhich remains undestroyed throughout all the chemical trans-formations of that substance. Radicals were thus assumed t oplay the very same function in the organic as the elements doin the inorganic compounds; accordingly, the elements wereconsidered as simple radicals, while the radicals in the organiccompounds were compound radicals. The strife waged longand bitter as to what particular cluster of atoms is to becon-sidered as the radical in any given group of related substances,and also whether radicals are merely creations of our own sub-jective thinking, figments of our imagination, or whether theyhave, or may acquire, an independent existence. Finally, withthe general acceptance in 1858-1860 of the theory of valence ofthe elements, and of the theory concerning the structure of themolecules as based on the valence of atoms, the concept ofradical came to cover any group of atoms which remainsunchanged in the molecule of the substance that has under-gone a chemical reaction. After the universal concurrence inthe belief tha t carbon, in distinction from many other elements,possesses an unvarying valence, namely that of four, the con-cept of radical has acquired a new and more specific meaning:by the term was designated usually a portion of the molecule

* An essay prepared in connection with the dedication of the Sterling Chemis-try Laboratory.91

7/28/2019 Chem rev VOL 1 091_141

2/51

92 M. G O M B E R Gwherein, if separated from the rest of the molecule, there wouldbe one carbon atom functioning in a state of valence not normalto it, that is less than four. Obviously, granting the constancyof the valence of carbon, the independent existence of freeradicals becomes a contradiction and is inconceivable. Thus,after 1865, the interest in the existence of free radicals becamedormant, except for such sporadic speculations as those ofButlerow and of Voii Laar, who tried to account for the phe-nomenon of tautomerism by assuming that the molecules of asubstance dissociate into radicals which then recombine so asto form new molecules of a constitution different from t,heformer.The brilliant experimental work of Nef, 1890-1900, focussedanew the attention of chemists to the question, does carbonalways function as tetravalent? Is it not divalent in manyother compounds as well as it is in carbon monoxide? Hesucceeded in marshalling a formidable array of facts in favorof his view, especially in favor of the view that carbon is diva-lent in the isonitriles. While such an explanation fits in ad-mirably with all the manifold behavior of these substances, itcould not, however, be denied that a consistent explanationbased upon the quadrivalence of carbon may serve equally well.We had in the views of Nef a revival of the old discussion as tothe constitution of the so called unsaturated compounds, andwhich ended, in 1871, in the general acceptance of the theoryof double and triple bonds. The new factor of Nefs querieswas,-must we assume that carbon functions in its maximumvalence also in those unsaturated compounds where the un-saturation in the molecule is centered not on two adjoiningcarbon atoms, but on a carbon and a nitrogen atom in unionwith each other? The question, are isonitriles R - N 5 C orR - N =C, is like an echo of the old question,-is acetyleneH -C E C -H or is it H -C -C -H? Kef was so successful inadducing evidence in favor of the divalent nature of carbon inthe isonitriles, that , notwithstanding the feasibility of anotherand a less radical explanation, his views have found acceptancein many authoritative texts.

7/28/2019 Chem rev VOL 1 091_141

3/51

ORGANIC RADICALS 93The old question, therefore whether carbon in any of its

organic compounds ever possesses a valence of less than four,-has received through Nefs work a renewed interest.From the time when Kekule and Fran-chimont, in 1872, first prepared triphenylmethane attemptsto prepare the analogous hydrocarbon tetraphenylmethanewere made by Hemilian, Fridel and Crafts, E. and 0. Fischer,Magati, Schwatz, V. Meyer, Weisse, Waga, and Meisel, butthe results were seemingly without success. In 1897 the writerdescribed a hydrocarbon which was prepared by heating tri-phenylmethaneazobenzene and which was supposed to betetr aphenylmet hane :

Tetraphenylmethane.

(CsHs)sCN:NCoHs= (CoHs)sC(CsHs)f NzThe reaction was more complex in reality than is indicated inthe equation, and the yield of the hydrocarbon was exceedinglysmall, from 2 to 5 per cent. Consequently it seemed desirableto make a comparative study of the physical and chemicalcharacteristics on some similar completely phenylated hydro-carbon as for instance on hexaphenylethane, proceeding on the

supposition that it would prove easier to obtain large amountsof the latter than of tetraphenylmethane. The contemplatedmethod for this preparation was one in accordance with thefollowing general reaction :(CsHs)sc1 (CaHJ sC(CeHdsC1 (CoHs)sC

+Metal = ] +Metal halideHexaphenylethane, triphenylmethyl .Accordingly, triphenylbromomethane in benzene was treated withmetallic sodium but without success. The chloro compound gave nobetter results. Molecular silver was substituted for sodium. Afterseveral hours boiling a white crystalline body began to separate, andon filtering the hot benzene sulution a considerable amount of the samesubstance separated on cooling. It was recrystallized from benzene,gave a constant melting point, 185C.) and contained no halogen. Inits high melting fpoint and in its only slight solubility in the usual

7/28/2019 Chem rev VOL 1 091_141

4/51

7/28/2019 Chem rev VOL 1 091_141

5/51

ORGANIC RADICALS 95there are always two adjoining carbon atoms acting as trivalent. Theunsaturation in such cases has always been indicated by a doublelinking. In triphenylmethyl there is only one carbon atom that isunsaturated. The existence of such a body means that when threevalences of carbon are taken up by three phenyl groups it is difficult,or perhaps even impossible, to introduce as a fourth group such a com-plicated radical as (CeH5)3C-, Only simpler groups, chlorine, bromine,iodine, oxygen, etc., may still combine with such a carbon atom.Whether this be due t o the negative character of the three phenyl groups,or whether it is caused by the fact that these groups take up so muchspace around the carbon atom as t o hinder the introduction of anothercomplicated group, is a question of an entirely different nature and neednot be discussed here. There are, however, numerous reactions whichgo to show that there is a limit to the number of complicated groupswhich can ordinarily be linked to one and the same carbon atom. . . .Now, as a result of the removal of halogen from triphenylchloromethanethe fourth valence of the methane carbon is bound either to take up thecomplicated group (C&)&- or remain as such, with carbon as tri-valent. Apparently the latter is what happens (1).

Objection to the theory of triualent carbon. Following the pub-lication of the above paper a detailed study of the strikingand unusual behavior of triphenylmethyl was made by variousinvestigators. It was also demonstrated that this hydrocarbonis but an example of a large class of similar triarylmethyl com-pounds. However for almost ten years th e existence of thesefree radicals still remained a question of dispute. On the chemi-cal side, the evidence in favor of the trivalency of carbon seemedconvincing; and yet, the molecular weight of the hydrocarbonin various solvents indicated that it posessed the composition[(CsHb)3C]2,as if it were after all, hexapheynlethane.

Thus, the argument which had proven so decisive in elimi-nating from the realm of free radicals even those outstandinghistorical examples-the cyanogen of Gay Lussac, the cacodylof Bunsen and the alkyls of Frankland and Kolbe-this verysame argument was now being invoked against the triarylmethyls.The latter also were all inferred to be dimolecular, for such a tleast appeared to be the case with triphenylmethyl itself. True,at SOC., in naphthalene as the solvent, the molecular weight of

7/28/2019 Chem rev VOL 1 091_141

6/51

96 M. GOMBERGtriphenylmethyl was found to be much less than dimolecular,415 instead of 486. The conclusion was drawn even at thatearly time, first, that at this temperature measurable dissocia-tion must be occurring into monomolecular triphenylmethyls;second, this phenomenon of dissociation, be it ever so slight atroom temperature, furnishes a satisfactory explanation of t heunsaturated behavior of the compound, since we may assumethat there exists a dynamic equilibrium between hexaphenyl-ethane and triphenylmethyl.

Acceptance of the theory. At the present time, it seems strangethat the examination of the molecular sta te of the other tri-arylmethyls, aside from triphenylmethyl, was delayed for solong a period after the discovery of triphenylmethyl. Had thisbeen done early, it would have been recognized at once thattriphenylmethyl, as regards its dimolecular phase, is an excep-tion to rather than a typical example of the class of triaryl-methyls. Soon after Schlenk reported, in 1910, that the threetriarylmethyls of the composition :

(CsHr- Ce")&>CB I - CJ33 (CsHs-C6H4h

(COH&2 (C.5Hs)exist as monomolecular to the extent of 15, 80, and 100 percent respectively, many other triarylmethyls were also foundto be inonomolecular t o a large degree. Moreover, a reexamina-tion of the original data obtained on triphenylmethyl itself,was now interpreted anew as indicating that it, too, is mono-molecular at room temperature to the extent of 5 to 17 per cent.Thus, the one seemingly valid argument, in fact, the only ob-jection that was at any time raised against the constitution ofthe triarylmethyls as free radicals, turned out on closer examina-tion to be but a phantom argument.With the establishment of the actual existence of free triaryl-methyls containing a carbon atom in the abnormal, or ratherunnormal, state of valence, the search was extended to a studyof the behavior of other elements in this respect. We now speakof radicals which contain divalent and tetravalent nitrogen, or

7/28/2019 Chem rev VOL 1 091_141

7/51

ORGANIC RADICALS 97monovalent oxygen, or trivalent lead, trivalent tin, and thefield is still being extended.In the following pages an attempt will be made to sketchretrospectively some of the important features of this newchapter of organic chemistry. The group of compounds whichcontain a trivalent carbon will be described quite fully, and thecompounds with other elements in abnormal st at e of valencewill be described in lesser detail.

11. COMPOUNDS WHICH CONTAIN TRIVALENT CARBON1 . Triary lmethy l s and analogsDuring the first few years the work on freeradicals was limited almost exclusively to the triarylmethylswherein each aryl group was linked independently and directlyto the central methane carbon atom. Subsequently, analogsof triphenylmethyl were prepared, namely, derivatives of an-thraquinone, xanthone, thioxanthone, and acridine.

KO mportant changes have been introduced into the processof preparing triarylmethyls as originally described, althoughnew and interesting methods of formation of these radicals havebeen discovered. From the laboratory point of view, not in-frequently the most difficult part is the preparation of therequisite triarylcarbinols, especially those with complex arylgroups. The conversion of the carbinols into the correspondingcarbinol chlorides has become much simplified since it has beenshown by the writer that the former react readily, when dis-solved in benzene or in ether, with hydrogen chloride or withacetyl chloride in this manner:

Preparation.

RsCOH +HC1 = RICCl +HOHRsCOH +ClCOCHj = RSCC1 +HOOCCHS

In the xanthone series the problem proved more complex. Herethe arylxanthenyl and thioxant(heny1 chlorides, which shouldcorrespond to the triarylmethyl chloride, were assumed as non-C HEM IC A L REVIETVB, Y O L . I, NO. 1

7/28/2019 Chem rev VOL 1 091_141

8/51

98 M. GOMBERGexistent; instead, isomeric colored chlorides of the constitutioninvolving tetravalent oxygen or sulfur were described :

CdHr c a rI I

Ic1 Ic1It was found that this misconception rested on insufficient anderroneously interpreted experimental evidence. By the use ofmore exact methods colorless carbinol chlorides were preparedwhich proved to be entirely analogous to the triarylmethylchlorides, except that they were somewhat less stable than .thelatter. The same is true in a modified sense, of the acridineseries.For the abstraction of halogen from the carbinol halide, onestill depends upon silver, zinc or mercury; but finely dividedcopper, in the form of copper bronze, has also been suggestedby Schlenk and used with success. Unfortunately copperrequires for reaction a somewhat higher temperature than theother metals do, and many radicals suffer decomposition whenheated for long periods; moreover, a large excess of the metalhas to be employed. Taking all circumstances into account,molecular silver seems to us to be the most suitable metal; ittakes out the halogen quite readily at room temperature, i t neednot be in large excess over the calculated amount, it is compactand thus very little of the solution containing the free radicalis retained by the metal.The choice of solvents is very wide: benzene, ether, carbondisulfide, acetone, ethyl acetate, petroleum ether, have all beenused. The choice is governed primarily by the relative solu-bility of the free radical in the particular solvent, and also bythe ease with which the solvent may be rapidly concentratedor removed without injury to the radical.Generally speaking, the essential conditions for the prepara-tion of free radicals are: (1) exclusion of all inoisturc in order

7/28/2019 Chem rev VOL 1 091_141

9/51

OR GA N I C R A D I C A LS 99to prevent partial hydrolysis of the carbinol halide; ( 2 ) absenceof oxygen; (3) avoiding exposure of the solutions to direct s-m-light, since light induces inter-molecular oxidation and reduc-tion of the radicals; (4) voiding even traces of acids. Schmid-lin, Schlenk, and the writer have each described some form ofapparatus which meets all the above requirements.The general procedure in our laboratory as now in use, is asfollows: Three to ten grams of the carbinol chloride is placedin a tube of suitable size, of shape A , figure 1. From 2 to 3tines the calculated amount of molecular silver is introduced,the tube is almost completely filled with the solvent, and stop-pered with a well fitting cork which carries on its under surfacea cap of parchment paper. The tube is now placed on a shakingmachine for several hours. With very finely divided silver, thereaction may be completed in thirty t o forty-five minutes.The liquid is then siphoned over from the tube A into bulb Band is either concentrated under reduced pressure or is removedcompletely and then replaced by a different solvent, as the casemay be. Upon cooling the concentrated solution, the triaryl-methyl usually crystallizes out. The liquid is drawn off, bysuction, through the stopcock C and the crystalline materialis washed and dried in the bulb, with the pump still in actionwhile a slow stream of carbon dioxide is passed through. Theyield of the pure crystalline substance will vary, but is usuallyfrom 50 to 80 per cent of the calculated, depending upon thesolubility of the substance in the solvent from which it crystal-lized. In the preparation of acridyl radicals, a somewhat modi-fied procedure is necessary, but this need not be discussed here.A large number of triarylmethyls, about 65, have been pre-pared and definitely characterized through their oxidation intothe corresponding peroxides; of these about 20 radicals have beenisolated in the solid state for purposes of molecular weight de-termination, but iopkqpbout 6 have been carefully studied withthe same detail a~~~jphenylmethylas been.While triphenylmethyl itself, in the solid state,is quite stable and ,may be kept undecomposed for several years,other triarylmethils differ very widely as regards stability.

Stability.

7/28/2019 Chem rev VOL 1 091_141

10/51

100 M. GOMBERGSome have been found to retain their unsaturated characteronly a few hours, and others become isomerized or decomposedby the time their solution is concentrated to dryness. Appar-

ently radicals which contain in the nubl&hothe groups, OH,OCH,,OCH2, C6H5,OCOCH,, NOz, are ad&@ the least stable.Mostof the radicals tha t have been isolated so fkr, are colorless orColor of the radicals in the solid state agd in solution.

7/28/2019 Chem rev VOL 1 091_141

11/51

OR GA N I C R A D I C A LS 101so nearly colorless in the solid state as to make it appear prob-able that the yellow coloration has been occasioned by slightoxidation of the substance on the surface of the crystals. I nsolution, however, each shows its own characteristic selectiveabsorption, solutions of every color of the spectrum have beendescribed, from yellow through orange, red, green to deepviolet. The complexity of the aryl groups is not the onlyfactor that determines the intensity of the color. Generallyspeaking, naphthyl groups give greater depth of color; a-naph-thyldiphenylmethyl gives a deep reddish brown solution, whilethe analogous p-naphthyl radical gives a solution of pure redcolor. Again, while di-p-chlorotriphenylmethyl gives an or-ange yellow solution, o-p-dichlorotriphenylmethyl gives violet-red, and the di-o-chlorotriphenylmethyl gives a deep violetsolution. Methyl groups intensify the color from yellow to-wards orange; methoxy groups, more towards red. Interestingis the influence of nitro groups : p-nitrotriphenylmethyl givesan intensely green solution, not unlike that of malachite green,while tri-p-nitrophenylmethyl gives a solution which can easilybe mistaken for that of fuchsine. The radicals of the type

C 6 H 4 C-R are all colorless in the solid phase, and also whenC6H4in solution. On being heated the solutions assume a deepblue fluoresence, and at a still higher temperature the solutionsbecome brown. With cooling, the change is reversible. Thexanthenyl radicals give solutions of deeper and more intensecolor than those of the triaryl methyls.From the fact that the triarylmethyls are almost colorlesswhen solid and become intensely colored when dissolved, wemust conclude that, in dissolving they undergo a change inmolecular structure. To Schmidlin belongs the credit for havingestablished by experiment that, even in solution, both rnodifi-cations, the colorless and the colored, exist in equilibrium witheach other. He showed that when triphenylmethyl is firstdissolved the solution remains colorless for a few seconds, thenbegins to turn yellow and very soon reaches a permanent in-tensity. If now air is allowed to come in contact with this

7/28/2019 Chem rev VOL 1 091_141

12/51

102 M . G O M B E R Gsolution a small amount of peroxide is produced and the coloris discharged, only to reappear again in a few seconds. Thiscan be repeated several times. The number of such successivedecolorizations indicates, it is generally assumed, the relativeproportions of the two modifications of the free radical. Thisphenomenon has been explained by the assumption that th ecolorless radical is hexa-arylethane, the colored is the free radicalproper:

RsC- CRs e R,CThe inadequacy of this explanation will be discussed later inthe paper.Conductivity. Triarylmethyl chlorides, and especially thebromides, when dissolved in liquid sulfur dioxide, as has beenshown by Walden and also by the writer, or in liquid hydro-cyanic acid (2), behave like true salts; they conduct the electriccurrent to a degree equal to that of ethylamine hydrochloride,and the molecular conductivity increases with the dilution. Wemust therefore consider the triarylmethyl chlorides as car-bonium salts, which dissociate into R,C+ and C1-. But howshall we intrepret the remarkable fact tha t triarylmethyls, inliquid sulfur dioxide, also conduct the current? Do we havetwo kinds of ions, RaC+ and RsC-, which correspond to twokinds of monomolecular triarylmethyls, or does the loose com-pound formed from the solute and solvent dissociate into ions:((R3C)2-SOz) = 2 R3C+ + (SOz)-? The question must beleft open for the present.It should be mentioned that the absorption spectrum oftriphenylmethyl in benzene is different from that in sulfurdioxide; also the absorption spectrum of triphenylmethyl radi-cal is different from that of the triphenylmethyl ion which isproduced by dissolving the carbinol in sulfuric acid.This has been determined for a numberof free radicals with most painstaking care, as this constant isthe final and the decisive criterion as to what extent the sub-stance is in t he monomolecular state, i.e., to what extent it isactually a free radical. The cryoscopic method, for the most

Molecular weight.

7/28/2019 Chem rev VOL 1 091_141

13/51

OR GA N I C RADICALS 103part, has been used with quite a variety of solvents freezingat temperatures of -22 to +80C. The per cent of dissociationof the hexa-arylethane into the free radicals has been calculatedfrom the formula

where M , is the molecular weight calculated from the depres-sion of the freezing point, and M , is that corresponding to thecalculated for the hexa-arylethane under examination.In our laboratory, the determinations were always made onfreshly prepared samples, the concentrations of the solute beingfrom 1 to 5 per cent. We have aimed to get appreciable de-pression of the freezing point of the solvent, from 0.200' tonearly a degree. In this way we hoped to eliminate, to someextent, the errors inherent in the method.In table 1 are summarized the results obtained in the mo-lecular weight determinations of a number of hexa-arylethanes.Only those data have been incorporate in this table which wereobtained at moderately low temperatures, with benzene or

nitrobenzene at 5"C., and p-bromotoluene at 27"C., as thesolvents. The ethanes are arranged in order of their ap-proximate complexity. Such a table permits ready comparisonof the various hexa-arylethanes with respect to their tendencytowards dissociation, since under these conditions the effectsof temperature and solvent may be disregarded.Factors inJluencing degree of dissociation of hexa-arylethanes.The principal factor is, the nature of the aryl groups; also, forthe same hexa-arylethane the degree of dissociation must alsodepend upon the nature of the solvent, the concentration of thesolute and the temperature at which the observation is made.It is clear from the table that we may have hexa-arylethanesof all possible gradations of dissociation, the differences in thedegree depending upon the nature of the aryl groups. Ourknowledge concerning this influence is still too meager to permitof sweeping generalizations. Only such results can be of usein these comparisons which have been obtained on different

7/28/2019 Chem rev VOL 1 091_141

14/51

104 M. GOMBERG

"C1k ? ?3 0 09H

----.-,------

7/28/2019 Chem rev VOL 1 091_141

15/51

ORGANIC RADICALS 105

Hf!s?"%272R

ehE

-EaY

3

e,EQ0hEacd

f!a"qrl'

e

#

Hf!B?"%

e,hEII

4xc7b5

7/28/2019 Chem rev VOL 1 091_141

16/51

106 M. GOMBER Garylethanes under nearly .the same experimental conditions asto solvent and the concentration of the solute. A comparisonin table 1 of hexa-phenylethane with the 0- and pmethoxy-and p-benzyloxy-derivatives brings out the influence of theoxygen in the nucleus of the aryl groups in enhancing the dis-sociation tendencies of hexa-phenylethanes, And yet, judgingfrom results soon to be published, two methoxy groups on thesame nucleus do not seem t o have any more influence than asingle group. The successive substitution of the three phenylgroups by the heavier biphenyl groups leads to arylethanes withprogressively increasing dissociation tendencies, i.e., 15, 80 and100 per cent.In order to get further information on this question, Gom-berg and Schoepfle (3) made careful molecular weight deter-minations on a selected series of hexa-arylethanes, all thedeterminations being made in the same solvent, namely, innaphthalene, at 80C. The following seven substances wereused :

1. (C6H6)&. ......................... triphenylmethyl2. (C6H6)&.CloH,. ................... a-naphthyldiphenylmethylc 3 4CJ343. c~H~.c( >................. henyl-xanthyl/CsH4\4. CH*CJ&.C(C )............p-tolyl-xanthyl

6H.5. CLC6H4.C0. .............. -ohlorophenyl-xanthyl

nH46. CioEI7.C C,H> ................. -naphthyl-xanthyl sH4CJ34

CioHs7. c&c( >............... henyl-pheno-8-naphtho-xanthylThe results which were obtained are summarized in figure 2.From Schlenks results with biphenyl-hexa-arylethanes onemight infer that the dissociation of the hexa-arylethane intofree radicals is greatly favored by the complexity or the weightof the aryl groups, the dissociation becoming apparently more

7/28/2019 Chem rev VOL 1 091_141

17/51

O RG ANI C RADI CAL S 107manifest also in proportion to the number of such groups. Somesupport in favor of this inference is given by the a-naphthyl-diphenylmethyl, which was found to be monomolecular to the

FIQ.extent of 60 per cent. But the hypothesis that the dissociationof the hexa-arylethanes is proportional to the complexity ofthe aryl groups becomes wholly untenable when one comparestriphenylmethyl with phenyl-xanthyl,

7/28/2019 Chem rev VOL 1 091_141

18/51

108 M. G O M B E R Gwhich is monomolecular to the extent of 75 per cent. Is theunion of the two phenyl groups the paramount influence in thiscase? If so, why should a substance, constitutionally so closelyrelated t o the xanthyls as phenyl-biphenylene-methyl is,

be completely dimolecular? Again, an a-naphthyl group whenreplacing a phenyl group in triphenylmethyl exerts upon thedissociation equilibrium of the compound a very decided in-fluence in favor of the monomolecular phase, a-naphthyldiphenyl-methyl appearing as wholly dissociated. This favorable in-fluence of the naphthyl group is still retained when the group islinked to a xanthone ring, the resulting compound being alsodissociated to the extent of 100 per cent. And yet, when thenaphthyl group enters as a component in the formation of thexanthone ring itself, it depresses very decidedly the dissocia-tion tendency of the compound, as is evident on comparingthe two isomers, a-naphthyl-xanthyl and phenyl-pheno-p-naphtho-xanthyl. This, indeed, is very strange, the highlyfavorable influence of the xanthone ring is almost wholly offsetby the naphthyl group in the ring, which group is otherwiseeven more favorable in its dissociation influence than the xan-thone ring itself. More facts are needed before we can hopeto unravel this confusing interplay of various influences.The factors which influence the equilibrium between themonomolecular and the dimolecular phase of one and the sameradical are, the nature of the solvent, the concentration of thesolute, and the temperature of the solution. As regards theinfluence of the solvent, the evidence shows that it is in someinstances appreciable. A comparison of the molecular stateof substances at nearly the same concentrations in nitrobenzene,benzene and cyclohexane, all of which freeze at about the same

7/28/2019 Chem rev VOL 1 091_141

19/51

ORGAXIC R A D I C A L S 109temperature, shows that, generally speaking, the dissociatinginfluence of these solvents is in the order mentioned. Othersalso have found that, as conpared with the other cryoscopicsolvents, cyclohexane is very associating (4). More experi-mental work in this direction is desirable.The remaining two factors, concentration of the solute andthe temperature of the solution, have been investigated on somany examples that their influence is well established. Thisinfluence is in accord with what one would expect from the dis-sociation in general: the higher the concentration, the less isthe molecular dissociation of the hexa-arylethane; also, thehigher the temperature, the greater the dissociation of theethane into monomolecular free radicals. Figure 2 brings outthe influence of concentration of the solute on the degree ofdissociation of several different triarylmethyls. Figures 3 and 4bring out on two specific examples, each in a variety of solvents,the influence of the same factor, as well as the influence oftemperature. The influence of each of these two factors isquite appreciable and is fairly uniform.Molecular combinations. Triphenylmethyl combines with avariety of substances from most divergent classes of compounds.This property illustrates in a very striking manner the inherentunsaturated character of the free radical. These combinationsare obtained merely by dissolving triphenylmethyl in the re-spective compounds, and when the solution has been cooleddown to room temperature, the molecular combination, con-sisting of triphenylmethyl and solvent, crystallizes out. Molec-ular combinations of this nature were obtained with aliphaticand aromatic ethers, ketones, aldehydes, esters, nitriles; withethylene oxide; with chloroform; with carbon disulphide, witharomatic hydrocarbons, with olefines, and even with saturatedhydrocarbons such as heptane, decane, cyclohexane and methyl-cyclohexane. The combinations which are formed in thismanner are readily dissociated into the component parts at atemperature near the boiling point of the volatile component.The fact that hexane and octane among the hydrocarbons tested,acetone among the ketones, methyl and ethyl formate among

7/28/2019 Chem rev VOL 1 091_141

20/51

110 M. GOMBERGthe esters, fail to give additive compounds with triphenylmethyl,is to be interpeted in the sense that the dissociation temperatureof these additive compounds is probably below ordinary roomtemperature.

FIQ.The composition of the various additive compounds is alwaysone mole of hexa-arylethane to one mole of the other component,except in case of decane where it appears t o be (R3C)+ClnH22.It is this remarkable uniformity of composition of these com-binations that raises doubt in ones mind whether they should

7/28/2019 Chem rev VOL 1 091_141

21/51

7/28/2019 Chem rev VOL 1 091_141

22/51

112 M. GOMBERGtion to result in fairly stable substances. But as a matter offact, the addition compounds are extremely unstable, dissociatingreadily at temperatures of 50 to 100 into the original compo-nents. Indeed the stability of these additive compounds isapparently no greater than tha t of the additive combinationsthat result from triphenylmethyl with ethylene oxide, withethers, with chloroform, all of which contain no double bondof any kind whatsoever.An explanation in terms of the valence hypothesis for theexistence of all the various additive compounds might still behad, if we are willing to assume that the oxygen, the sulfur, orthe chlorine atom, as the case may be, functions as though ithad acquired a higher valence than that which it ordinarilypossess. Sulfur and chlorine are known to possess a variablevalence, and of late it has become customary to consider tha toxygen too, can acquire a higher than its usual valence state.We then have:

R /CRI /CRa R /CRsCllH C C1( o>.=.:Rf>\CR8 CRs R CR,While such an explanation might perhaps account for the addi-tion of the triphenylmethyl to ethers, esters, etc., the questionstill remains unanswered as t o the mechanism of the additionof the free radical to completely saturated hydrocarbons, bothaliphatic and alicyclic. And yet, these latter combinationsare practically of the same order of stability as all the othercombinations. Taking everything into consideration, we pre-fer, as Schmidlin does, to look upon all the additive compoundsas molecular combinations wherein the second component isretained through the residual affinity in the triphenylmethylmolecule. Howshall we account for the fact that while triphenylmethyl formssuch a wide variety of additive compounds, other triarylmethyls,which are monomolecular to a much greater degree than istriphenylmethyl, fail to give rise t o such additive compounds?

Another phase of this question deserves consideration.

7/28/2019 Chem rev VOL 1 091_141

23/51

ORGANIC RADICALS 113a-Naphthyldiphenylmethyl, p-naphthyldiphenylmethyl and p -methoxytriphenylmethyl are monomolecular to the extent of70 , 30 and 30 per cent respectively. They belong distinctlymore to what we consider as the class of free radicals; conse-quently it would seem that they should produce molecularcombinations more readily than triphenylmethyl does; and yetthe reverse is the case. This apparent contradiction, however,is explainable to a large degree by the following considerations:At room temperature the dissociation of hexa-phenylethane isonly about 5 per cent. In other words, the tendency on thepart of triphenylmethyl to become saturated is evidenced throughthe temporary feeble attachment of one free radical to another,or through attachment to substances such as ethers, esters,ketones, hydrocarbons, etc., if these be present. On the addi-tion to this system of slight amounts of energy from the outsideas on gentle heating, the union is disrupted and the free radicalis regenerated. Thus the obvious predilection of triphenyl-methyl to assume the dimolecular state, and on the other handits proneness to enter into feeble combination with all possiblesolvents may be looked upon as consequences of one and thesame cause, namely tha t the trivalence of its carbon atom isof a low order of stability. And now consider from the samepoint of view diphenyl a-naphthylmethyl. It is monomolecularto the extent of 70 per cent even at 6"C., and the dissociationincreases rapidly with the rise of temperature. Evidently, inthis particular triarylmethyl the trivalence of carbon is of sucha high order of stability that the radical lacks the tendency toenter into combination with itself, and presumably for the verySame reason it is devoid of the tendency to unite with othersubstances, such as esters, ethers, ketones, etc. Thus a highdegree of dissociation of the hexa-arylethane is indicative of thefact tha t in the resulting triarylmethyl the affinity of the tri-valent carbon atom is all utilized in internal combinations ofthe radical itself, and consequently there remains very little, ifany, residual affinity for outside combinations of reversiblenature, whether it be radical to radical or radical to somethingelse. The explanation here given should be tested on various.

7/28/2019 Chem rev VOL 1 091_141

24/51

114 M . G O M B E R Ghexa-arylethanes which cover Z wide gradation of dissociationtendencies.

Chemical reactionsThe triarylmethyls react readily with a great variety of sub-stances, the reactions being as a rule irreversible. While manyof the reactions have been studied in detail only on triphenyl-methyl, still there is very little doubt that they are characteristicof the whole class of the triarylmethyls.Efect of light. Even in the solid state, but especially sowhen in solution, the triarylmethyls are extremely susceptible

to actinic rays. The darker the solution, the slower the decom-position, but it is advisable t o avoid as much as possible ex-posure of any triarylmethyl to direct sunlight, or some decom-position may ensue. Like so many photochemical reactions,we have here also intermolecular oxidation and reduction:C J 3 4ca4

3 (C&)sC 2(Cb")aCH +I )c - CsHsThe problem becomes more complex when the three aryl groupsin the triarylmethyl are not all alike; the dehydrogenation andthe consequent coupling of the two aryl groups may then occurin one or more of several possible ways. Only very few exampleshave been studied carefully in this connection.It was thought at first that hydrogen chlorideinduces catalytically the conversion of triphenylmethyl intohexaphenylethane. Chichibabin, however, showed that theproduct is not hexaphenylethane but a stable isomer of it, namelybenehydryl-tetraphenylmethane. It is now, however, well es-tablished that the action of acids upon triarylmethyls is re-solvable into two stages:

Efect of acids.

(a) RIC - CRa +HCI =RICH +RICCI(b ) %C -0 RaCCI =RIC -0 CRa +HCI

Whether the reaction will stop with stage (a) or proceed furtherwith the formation of the benehydryl according to (b), will

7/28/2019 Chem rev VOL 1 091_141

25/51

ORGANIC RADICALS 115depend in each case upon the nature of the aryl groups. Withtriphenylmethyl, only 6 per cent of the total product is retainedas given in equation (a) and the rest is converted into thebenzhydryl; with diphenylmonobiphenylmethyl, according toSchlenk the reaction does not proceed further than the firststage; with diphenyl a-naphthylmethyl, as well as with p -methoxytriphenylmethyl, it is about 80 per cent according to(a) alone, and 20 per cent proceeds further according to (b).The spontaneous oxidation of tri-arylmethyls with the formation of corresponding peroxidesserves as an infallible test for the presence of a free radical.The absorption of oxygen takes place rapidly, as fast as by yellowphosphorus. The peroxides are colorless compounds, onlyslightly soluble in the usual organic solvents; they are fairlystable and possess characteristic decomposition points whichare higher than the melting points of the corresponding freeradicals or the carbinols. The precipitation of the peroxidewhen a solution of the free radical is exposed to air is of courseaccompanied by the decolorization of the solution. This oxida-tion is an exothermic reaction, and in the case of triphenyl-methyl it is:

Format ion of peroxides.

2 (CBH6)aC+01 = (CaHs)aC-O-O-C(CaHs)s +124 Cal.The amount of oxygen absorbed generally corresponds so nearlyto the calculated that this reaction may serve as a basis for thequantitative estimation of the free radical. While the quanti tyof oxygen taken up by the radical is nearly the theoretical, theactual yield of the peroxide, however, varies with differenttriarylmethyls from 75 to 93 per cent, indicating that somedecomposition of the peroxide must occur during its formationfrom the free radical. Indeed, even during the recrystalliza-tion of the peroxide from boiling benzene, slight decompositionis perceptible; in boiling xylene, at 140, he decomposition oftriphenylmethyl peroxide is complete.

Add i t i on of iodine. The triarylmethyls unite with chlorine,bromine and iodine; while with the first two halogens bothsubstitution and addition occurs, with iodine, addition alone

7/28/2019 Chem rev VOL 1 091_141

26/51

116 M. G O M BE RGtakes place. Next t o peroxide formation, combination withiodine constitutes the most reliable identification reaction forthe free radicals. The amount of iodine which is taken up isalways short of the calculated, because these iodides are everso much less stable than the corresponding bromides and chlo-rides, and there is a tendency on the part of the iodide to dis-sociate again into free radical and iodine:

2 RaC +1 2 2 RaCIThis equilibrium varies with the individual members of theseries, but it is usually reached when 60 to 80 per cent of thecalculated amount of halogen has been taken up. The isolationof the iodides often presents difficulties, since they are readilydecomposed by water, alcohol, and even by atmospheric oxygen:

2 RacI +0 2 = RsC-O-O-CR3 +IzIrreversible addition reactions. There are many other reac-tions of triarylmethyls, which result in well defined derivativesof triarylmethane. Here again, most of the work has beendone with triphenylmethyl, but the reactions are unquestion-

ably typical of the whole class. They will be summarizedbriefly in Table 2:

RaC +H (narcent)R3C +H (in presenceof Pt ,)RaC + SR3C +NOR3C +NOClRaC +NOzR3C +0:CaHd:OR3C +CeHbOHRaC +CJL(CHa)zR3C +NHzKHCaHsR3C +CHzNzRaC +Na

TABLE 2PRODUCT0

RaCHR&HPolysulfidesRaCNORaCNO +RaCClRaCKOzRsC-0. CBH4. O-CRsRaC-CsHiOH +RsCHRaC-CsHa(CH3)z +RaCHRaC-CsH4NH.NHCsHs +RsCHRaC-CHz - CRIRaCNa

O B S E R V E R

GombergSchmidlinSchlenkSchlenkSchlenkSchlenkSchmidlinSchmidlinWielandSchlenkSchlenkSchlenk

7/28/2019 Chem rev VOL 1 091_141

27/51

ORGANIC RADICALS 117It might be mentioned here that because of its preeminent

tendency to make addition compounds, triphenylmethyl hasfound application as a reagent for the detection of free radicalsother than triarylmethyls. For instance it reacts, as Wielandfound, with tetraphenyl hydrazine, as follows :(CBHS)Z?$-N(C&)~+2 RaC = 2 (CeH5)zN-CRa

This reaction has been advanced as an argument in support ofthe hypothesis that tetraphenylhydrazine itself is in solutiondissociated to some extent into tvvo half-molecules, in otherwords into free radicals containing divalent nitrogen. A similarargument has been used in support of the assumption t ha tdehydrophenols contain monovalent oxygen.Relation between dissociation and color. Tautomeri sm of

f ree radicalsThe fact that hexa-arylethanes, colorless in the solid state,give colored solutions in organic solvents has been explained,principally, along two different lines :1. Colorless hexa-arylethanes, in solution, dissociate to formthe colored free radicals. When the substance is colored alsoin the solid state it is then a free radical in that phase as well.2. Dissociation occurs and is followed by tautomerization ofthe benzenoid triarylmethyl into the quinonoid monomoleculartautomer; the latter substance is the principal factor responsiblefor color production.

(C6Hb)aC- C(Ca~6)a (CB&)& 4- (C8Ha)aC/=>/""(c H) 2c =\= \As has been previously mentioned in this paper, upon shakinga solution of triphenylmethyl with oxygen, the color vanishes,but reappears after a few seconds, and this phenomenon takesplace several times before the solution is completely and per-manently decolorized. By weighing the amount of peroxideformed in a single decolorization Schmidlin was able to calculate

7/28/2019 Chem rev VOL 1 091_141

28/51

118 M. GOMBERGthe amount of the colored form which was present. His resultsindicated that the ratio of colorless to colored form was about10 to 1. However, since the molecular weight of triphenyl-methyl at room temperature corresponds very nearly to thedimolecular formula, Schmidlin concluded that the color mustbe due to the presence of some tautomeric form of hexaphenyl-ethane and not to dissociation, as for instance:

/ = v C (C HAI(Ct")& - C(CsHA, (CsHdsC \ =A HSchmidlin also observed the variation in the intensity ofcolor of triphenylmethyl solutions at different temperatures,and found that a t the freezing point of chloroform, -63", thecolor entirely disappeared. He conducted colorimetric deter-minations over a range of temperatures from the freezing pointto the boiling point of benzene, and found that the color in-creased sixfold. From these results he concluded th at the in-tensification of color of triphenylmethyl solutions with rise oftemperature must be due to something other than dissociation.Wieland, Schlenk and Main took the very opposite point of view,assuming that hexaphenylethane is dissociated to some extent.Hence, according to them, the increase in color intensity is dueentirely to dissociation of the ethane into free radicals.What seemed to be the most convincing evidence in favor of thedissociation theory is found in the work of Piccard on the increase ofcolor of triphenylmethyl solutions with dilution. This behavior con-stitutes a deviation from Beer's law, which states that the absorption oflight by solutions should remain constant if there were no chemicalchanges on dilution. In the case of hexaphenylethane we are con-

cerned with one of two phenomena: either tautomerism of the ethanewithout change of molecular weight, or dissociation, which leads to anincrease of the number of molecules. If it were tautomerism the equilib-rium would be between two isomers and, Piccard assumes, such anequilibrium would be independent of the dilution; but if it were disso-ciation, then the equilibrium would be between a polymer and the prod-ucts of its dissociation-and such an equilibrium should be influenced bydilution. He conducted a series of experiments on hexaphenylethane

7/28/2019 Chem rev VOL 1 091_141

29/51

O R G A N I C RADICALS 119solutions and found a decided increase in color upon dilution.therefore concluded that the phenomenon must be dissociation (5 ) .

HeThe occurrence of these three characteristic phenomena hasbeen since verified on a large number of hexa-arylethanes. Theywere found to undergo, on oxidation, several successive decolor-izations, to acquire intensification of color on dilution and tosuffer a marked diminution in intensity of color when theirsolutions were cooled. In fact, the deviation by a substancefrom Beers law has become an important, and often the onlyfactor in the decision of designating the substance as a freeWe have, then, these two concomitant phenomena: on theone hand, the dissociation of hexa-arylethanes increases withtheir dilution and decreases with the lowering of the tempera-ture of the solution; on the other hand, their color intensity alsoincreases with their dilution and decreases with the lowering ofthe temperature of the solution. Hence the conclusion thatdissociation and color production are related to each other ascause and effect.Nobody can gainsay the general plausibility of this explana-tion. And yet, the question arises, is the change in color in-tensity which has been produced in the triarylmethyl strictlyproportional t o the increase in the monomolecular phase of thetriarylmethyl? These two phenomena should, of course, bestrictly parallel to each other if they be related as cause andeffect. What little work of quantitative nature has been doneon hexaphenylethane has led to contradictory conclusions, forthe reason that this compound dissociates at room temperaturebut very little.

* radical.

In di-P-naphthyl-tetraphenylethane he writer and Sullivan founda substance of very considerable and accurately measurable dissocia-tion; its color, bright red in concentrated solutions shading to lightyellow in dilute solutions, lends itself readily to colorimetric study; thesubstance is quite stable towards light; it is readily soluble in a varietyof solvents throughout a wide range of temperature. In addition, itshows striking diminution of color within the temperature range of

7/28/2019 Chem rev VOL 1 091_141

30/51

120 M. GOMBERGordinary cryoscopic determinations. On account of this fortuitouscombination of properties, a study of this hexa-arylethane was made inan effort to throw more light on the relation which exists between colorintensity and changes in the concentration or in temperature of thesolutions of free radicals.

The molecular weight of the radical was determined in carbontetrachloride ( -22"C.), ethylene chlorobromide (- 7"), benzene(+5 .3 ) , nitrobenzene (5.7"), cyclohexane (5.8 ') and p-bromo-toluene (27"). Then, by the use of a properly constructed color-imeter, the effect of dilution on the color of the solution of theradical in benzene, nitrobenzene, and cyclohexane was deter-mined; and secondly, the effect of temperature on the solutionsof the radical in carbon tetrachloride and in toluene, covering arange, respectively of -22 to +30 and -40 to +30 was studied.

It was found that solutions of the free radicals in cyclohexane, ben-zene and nitrobenzene are of about the same intensity of color and theintensity of color changes equally in all cases on dilution, in spite of thedifferences in dissociation in these solvents. Assuming the color of a5 per cent solution to be unity, we can calculate the relative intensityof color at dilutions down to 1 per cent in reference to this unit. Plot-ting these color values against the dilution, that is, the number of gramsof the solvent containing 1 gram of the solute, we obtain curves a, b, c(fig. 5 ) , showing the relative change of color intensity with dilution.Similarly, we assume the degree of dissociation at 5 per cent as our unitand calculate the relative degree of dissociation at other dilutions withreference to this unit. These values are also plotted against the dilu-tion, and the resulting curves, a', b', c', show the relative change indissociation on dilution. It is apparent, in nitrobenzene and benzenethe intensity of color almost doubles with the decrease of concentrationfrom 5 to 1 per cent, and yet the change in dissociation throughout thesame range is very slight. On the other hand in cyclohexane, the dis-sociation increases three times as much as the color. This again showsthat color is independent of dissociation, that is, an increase in dissocia-tion may occur without an equal increase in color intensity.It was found that a solution of the free radical in toluene at -40'or in carbon tetrachloride at its freezing point is only slightly coloredalthough it is dissociated to the extent of from 15 to 20 per cent. Ongradually warming these very light colored solutions an increase in color

.

7/28/2019 Chem rev VOL 1 091_141

31/51

O R G A N I C R A D I C A L S 121begins to take place between -30" and -20" and this color increases10 times with a rise of temperature of 50"while the dissociation increasesonly 3 with the same change of temperature. If we assume as the unit

FIG.of color intensity the color of a solution at -20" and calculate therelative color intensity at other temperatures with reference to thisunit, then plot these values against temperature, we obtain curves a, b(fig. 6) showing the relative increase of color with temperature. If wenow take as our unit of dissociation the amount of dissociation at -20"

7/28/2019 Chem rev VOL 1 091_141

32/51

122 M. GOMBERGwe can calculate the relative dissociation at other temperatures withreference to this unit. Plotting these values against the temperaturewe obtain a curve c, showing the increase of dissociation with rise oftemperature. The accompanying curves show the striking differencesbetween the change in the degree of dissociation and the change in color

FIG.intensity which occurs when the temperature of the solution of the freeradical is varied.

Three other triarylniethyls have been subjected to the samestudy by Mr. Forrester recently in this laboratory and theresults obtained completely corroborate those with di-p-naphthyl-tetraphenylethane.

7/28/2019 Chem rev VOL 1 091_141

33/51

ORGANIC RADICALS 123To sum up, the changes in color intensity of the hexa-aryl-

ethanes which result from variations in the concentration or inthe temperature of their solutions proceed in the same directionas the changes of the dissociation do, but quantitatively theyare not parallel to the changes in dissociation which are thusproduced. These facts point to the conclusion that color offree radicals is not due wholly to dissociation of the hexa-aryl-ethane into the tri-arylmethyl. The most satisfactory explana-tion of the facts is the hypothesis that following the dissociationwe have tautomerization of the benzenoid triarylmethyl intothe quinonoid form. The equilibrium between the hexa-aryl-ethane and the triarylmethyl on the one hand and the equi-librium between the monomolecular tautomers on the otherhand, are not equally influenced by changes either in concentra-tion or in temperature. Not the trivalent state of the carbonatom, but rather the propensity towards quinoidation therebybrought about in the radical, is the cause of color-formation ofthe triarylmethyls.It must be admitted that colorimetric measurements give usonly approximately quantitative results, and our findingsshould be verified by spectrometric measurements. Since,however, those who seek to connect directly color causation anddissociation in triarylmethyls depend merely upon qualitativecolorimetric data, our quantitative measurements are certainlyreliable to the extent of showing the inadequancy of that ex-planation. The pertinent statement of G. N . Lewis (6), thatthe recognition of the existence of the quinonoid form is indis-pensable to an understanding of numerous reactions whichcharacterize the hexa-arylethanes, should apply equally wellto the free radical also-the recognition that the latter itselfexists in a monomolecular quinonoid as well as benzenoid taut-omer becomes indispensable.The term triarylmethyl, then, may stand for any one of thethree things: the ion (R3C)+, corresponding to the hydrogencation, Hf ; he benzenoid or quinoid radical R,C or R2C-

corresponding to atomic or nascent hydrogen, and= V I 3

7/28/2019 Chem rev VOL 1 091_141

34/51

124 M. GOMBERGfinally corresponding to molecular hydrogen. It shouldbe added that, according t o H. I. Cole (7), the dissociation ofhexaphenyl ethane from the viewpoint of the octet theory ofvalence, leads to the formation of the positive, colored ion

R2C =(=_x"iand the negative, colorless ion [R3C-], From altogether dif-ferent premises this interpretation has been arrived at by thewriter some years prior.

Chemical evidence of quinoidation. Chemical evidence infavor of a tautomeric relation between a benzenoid and a quino-noid phase in the triarylmethyls has been gotten from the follow-ing considerations: Suppose we take a triarylmethyl in whichat least one of the three aryl nuclei contains a bromine atom inthe para position to the central carbon atom. The monomolec-ular phase of such a compound must exist, in conformity withwhat has been said above, in the following equilibrium state:R"XZ> A - O B . 7 R2C-D r R a C =

7/28/2019 Chem rev VOL 1 091_141

35/51

ORGANIC RADICALS 125the valence which holds X becomes free, as in the correspondingtriarylmethyl, then the bromine is readily removable by silver.Surely such a remarkably striking change in function must haveresulted only from a change in the constitution of the nucleusitself; a change to the quinonoid state appears t o be the mostrational explanation.In conformity with the above explanation, it has proven pos-sible in some instances t o determine just which of the threearyl groups in the free radical assumes the quinonoid state.When the three aryl groups are independently linked t o thecentral carbon atom, then each ring may assume a quinonoidphase, and the result is an equilibrium with three quinonoidtautomers, involving in each a different aryl nucleus. Thishas been shown to be the case with tri-p-brom-triphenylmethyl.But in those triarylmethyls which contain a xanthone ring, thelatter is of paramount importance and is the dominating featurein the triarylmethyl. Consequently, when the process oftautomerism ensues, the quinoidation may not be directed equallytowards all three nuclei. Of the two isomeric triarylmethyls,

Bronly in the latter is the bromine atom removable by silver, ashas been shown by experiment. Thus quinoidation must beoccurring only in the two nuclei that constitute the xanthonering. The same is true of thioxanthone derivatives. Thework with acridines, which is being carried on by Mr. D. L. Tabernin this laboratory at the present time, has given results whichconfirm those previously obtained with xanthenols and thiox-anthenols. Here quinoidation is localized preferentially in theacridine ring.

7/28/2019 Chem rev VOL 1 091_141

36/51

126 M. GOMBERC:Formation of triarylmethy ls through dissociation of compounds

other th an hexa-arylethanesThe existence of triarylmethyls has been explained by theassumption that the three phenyl groups take up so muchaffinity from the central carbon atom tha t the remaining fourthvalence functions with very little affinity. One is inclined toask whether this is an explanation or merely a restatement ofthe facts in somewhat more strictly chemical nomenclature. Bethat as it may, we know now of many cases where a triaryl-methyl group in combination with different groups becomesreadily detached and assumes the existence of a free radical inexactly the same way as if it had come from a hexa-arylethane.In the following table, Table 3, are collected compounds whichare known to undergo such dissociation. The experiments werelargely limited only to triphenylmethyl compounds.

TABLE 3BUBBTANCE-P RO DUCTS OF DIBBO CIATIO N

RCR& + I (at rooin temperature)R8C-K=N-CR1-+21t~C +N2 at 0 " )RsCNR2-R,C +R2N-NR2at 110C)RsCCRZCRs -RsC +R~C-CRSR~C-S-CS-KC~HI~+RSCR~c-s-s-cR~-----+R~C +

O B B E R Y E R

Gomberg (1900)WielandWielandSchlenlr (1922)Blicke (1923)Blicke (1923)

In the following instances, where the heating of the compoundwas done at temperatures at or above the decomposition tem-perature of triphenylmethyl, transitory color phenomena ensuedsuggestive of dissociation, but positive evidence is lacking thattriphenylmethyl has actually been produced.(CaHs)aC-CH(C,&)r = (Ca&)aC(?) +CH (C6Hs)z( C E H ~ ) ~ C - O C E H I O - C ( C E H ~ ) ~(C&),C +0 :CeHr :0 SchmidlinChichibabin2. Trivalent carbon linked to less than three aryl groups

Many attempts have been made by different investigators toprepare compounds with trivalent carbon in the molecule,wherein the carbon atom is linked to less than three aryl groups.

7/28/2019 Chem rev VOL 1 091_141

37/51

ORGANIC RADICALS 127With the exception of the metal-ketyls, which constitute awhole class of such substances, only isolated cases have beendescribed. In no instance has the dissociation been as yet con-firmed by the evidence of molecular weight determinations, andtherefore the degree of dissociation is unknown.IndividuaZ cases. Kohler, in 1908 found that 1, 2, 3-tri-phenylindyl bromide when treated with metals, gives a coloredsolution which becomes decolorized in air and a peroxide isformed. This constitutes a complete analogy in behavior withtriarylmethyl halides. Consequently, a free radical must haveresulted from the action of the metal upon the bromide. Inthat radical, the trivalent carbon atom is linked to two arylgroups, and the third valence, indicated in the formula by aheavy line, is linked not to an aryl group.

Wieland, in 1911, described the decomposition of triphenyl-He formulates theethyl peroxide in boiling xylene, at 138".reaction which occurs as follows:

(C,")&--O-O-C (CeH,)s-+ 2 (CJIHI)aC--O-; (Ca H ~ )C- heatingi -I - COII,)*C-O-ceH(CsH3*C--O--C b H 4

CJIt.0 cooling

The intermediately produced radical with its monovalent oxygenis supposed to undergo instantly an internal rearrangementgiving rise to diphenyl-phenoxymethyl, which, through unionwith a similar group, forms tetraphenyl diphenoxyethane; itis claimed that this latter compound resembles hexaphenyl-ethane. It is colorless, dissolves colorless, but when its solutionis heated to 80 to loo", a color resembling that of triphenyl-methyl is produced; on cooling, the solution becomes colorless.Peroxide formation by this new radical has not been observed;nor does the molecular weight indicate dissociation.

7/28/2019 Chem rev VOL 1 091_141

38/51

128 M. GOMBERGGomberg and Jickling, in 1913, described diphenyl-thienyl-methyl chloride; this, when treated with molecular silver, givesrise to an intensely colored solution which absorbs oxygen fromthe air and in other ways shows the characteristic reactionsobserved on triarylmethyls. These findings have been corrob-orated recently by Mr. W. Minnis in our laboratory, but the freeradical (C6H&C-C4H6S, has proven so unstable that its isola-tion as such seems impossible. The close analogy of thiopheneto benzene makes it appear probable that a radical of the abovecomposition may exist.Cone, in 1912, has studied the N-alkyl iodide addition deriva-

tives of quinoline and pyridine. He found these when treatedwith zinc, give colored solutions which absorb oxygen and behavein general as if they contain corresponding free radicals con-taining a trivalent carbon atom, or perhaps in equilibrium,radicals containing a tetravalent nitrogen atom. In view ofthe recent publication by Emmert and by Dimroth (8) it appearsnow probable that these substances are dipyridyl derivativeswith no abnormal valence either on the carbon or nitrogen.Scholl ( 9 ) , in 1921, described what he considers to be the firstexample of a whole new peculiar class of compounds with tri-valent carbon atom. The substance gives violet-blue solutionswhich become colorless on exposure t o air. Protected fromair, the substance is so stable that it will stand the temperatureof boiling nitrobenzene, and it is monomolecular in that solventunder these conditions. Its constitution is, according to Scholl:

0II

Ziegler and Ochs ( l o ) , n 1922, described an interesting exampleof what may prove to be a free radical wherein the trivalent

7/28/2019 Chem rev VOL 1 091_141

39/51

ORGANIC RADICALS 129carbon atom is linked to two aryl groups and through the thirdvalence to a secondary carbon atom.

H HIc-c-c (Cad2

MetaZ ketyls. R2C- -K. These constitute a large class ofcompounds which are formed through union of ketones withmetallic sodium or potassium; they were first brought forwardas a class of compounds containing a trivalent carbon atom bySchlenk in 1911. They are prepared, in some instances, throughdirect action of metallic sodium upon the diary1 ketone dissolvedin ether, as has been done by Beckman and Paul as far back as1894; but as Schlenk showed, a more general method consistsin adding the previously prepared potassium compound ofphenyl-biphenyl-ketone, which is quite soluble in ether, toan etherial solution of any other ketone. As the majority ofthe metal ketyls are insoluble in ether, the following reactionensues:(CeHa) (CeHa-CsHd) COK+O:CRR' = (CeHs)(CeHs- CsH,)C :0+RmR'C -OKWe thus have a class of compounds which are analogous towhat Wieland considers his phenyoxy-diphenylmethyl to be:R2C- - sH6, with - OK) corresponding to (OC6H5).A large number of such metal ketyls have been prepared bySchlenk, not only from aryl ketones, but also from anthraqui-none, xanthone,ybenzopyrone, dimethyl pyrone, and a varietyof other compounds containing the C = 0 group.It should be noticed, that a number of these metallic deriva-tives possess a composition which does not correspond to thatanticipated from the structure of the ketones; not infrequentlyonly half the calculated amount of metal is taken up by theketone.

CHENICAL REVIEW S, VOL. I, NO. 1

7/28/2019 Chem rev VOL 1 091_141

40/51

130 M. GOMBERGThe metal ketyls are colored blue or green. They are readily

affected by oxygen and by iodine, but the reactions are differentfrom those shown by the tri-arylmethyl free radicals:2 (C6H6)zCOK+02 = 2 (CeHs)zCO+I

7/28/2019 Chem rev VOL 1 091_141

41/51

OR GA N I C RADICALS 131The products which result in this manner, in virtue of the dehy-drogenation of the phenols, are known as dehydrophenols.As a matter of fact, in practice it was found advisable, if notnecessary, to use the more complex phenols, and particularlyphenols that have one or both hydrogen atoms, ortho to thedyroxyl, replaced by some complex group. Pummerer usedseveral naphthol derivatives, while Goldschmidt worked withphenanthraquinoles.The questions under discussion concerning this class of sub-stances are these: (1) What is the constitution of the dehydro-phenols in their dimolecular phase? (2) To what extent, andunder what conditions do the dimolecular dehydrophenols dis-sociate into monomolecular free radicals? (3) What is theconstitution of the monomolecular compounds so produced-are they really radicals containing monovalent oxygen, or dothey rearrange spontaneously into radicals with a trivalentcarbon atom? Pummererrejects the peroxide formula for the dimolecular dehydrophenolsand prefers of the two remaining theoretically possible structures,the second to the first, Le., the quinol ether structure (b ) t o the

1. As regards the first query, agreement is lacking.

ethane structure (a).(

Goldschmidt, on the other hand prefers the peroxide structure,and formulates the dehydro-phenanthraquinol as (c)

7/28/2019 Chem rev VOL 1 091_141

42/51

132 M. GOMBERG2. Whichever be the true constitution of the dimoleculardehydrophenols, evidence is not wanting that, in solution,dissociation occurs. Unfortunately the deductions are basedupon such low concentrations of the solute and consequentlysuch small depression in the freezing point of the solvent, thatthe quantitative interpetation must be made with considerablereserve. Pummerer reports for his dehydrophenols, in concen-trations of 0.25 per cent as high as 50 per cent dissociation.Goldschmidt finds that the dissociation of his phenanthroxylcompound is a time reaction, requiring about three hours be-fore equilibrium is finally reached; at that stage the dimolecular

modification of the methoxy and ethoxy ether derivative havedissociated to the extent of 40 per cent and 60 per centrespectively. In support of the contention that we are dealinghere with the phenomenon of dissociation both Pummerer andGoldschmidt rely very much upon the fact that dehydrophenols,only slightly colored or entirely colorless in the solid phase,give like the hexa-arylethanes colored solutions, and that thesesolution do not follow Beers law: dilution of the solution leadsto an intensification of color.3. Outside of dissociation and of the color effects just men-tioned, these radicals show very litt le resemblance to the triaryl-methyls. They are not readily affected by atmospheric oxygen;they do not unite with iodine; they however combine withtriphenylmethyl. From a variety of considerations, the con-clusion is drawn by Pummerer that these radicals exist in tau-tomeric equilibrium between the aroxyl and ketomethyl forms :

c-o-c=o -+Goldschmidt, on the other hand, considers that the evidence ispreponderatingly in favor of the radical with monovalent oxygenatom in the molecule.Porter and Thurber described recently a most interestinginstance of what seems to be a dehydrophenol of very simplecomposition. By oxidizing mesitol with silver oxide they

7/28/2019 Chem rev VOL 1 091_141

43/51

O R G A N I C RADICALS 133obtained a substance with an odd or unpaired electron, that isan unoccupied valence.

0- 0CH3-,/(rH3 I_+ CHa--()-CH3\/I CH,H3

Further details will be expected with interest, especially in viewof the opinion expressed by Goldschmidt to the effect that theoxidation product of mesitol is not a dehydrophenol.

IV. C O M P O U N D S WHICH C O N T A I N DIVALENT NITROGENTetra-urylhydruzines. The fact that nitrogen exhibits in itsinorganic compounds a variable valence from one to five, fur-nishes sufficient basis for the expectation that organic compoundsalso may be found in which nitrogen may be functioning asdivalent rather than in its usual trivalent or pentavalent state.

H. Wieland, from his extensive and excellent experimentalwork in the field of tetra-arylhydrazines, has concluded thatwe have in these compounds labile substances which are subjectto dissociation similarly to hexa-arylethanes :RzN-NRz RzN +NRz

In these compounds we have remarkably close analogs to theabove mentioned dehydro-phenols. Like the latter, so alsoare the tetra-arylhydrazines prepared by the action of mildoxidizing agents, and we might call them correspondingly de-h ydroamines :

R-OH $02 --+ [2(R-O-] S R-4-0-RRzN-Hf 0 2 + 2(RtN-] RzN-NRt

Of the some 15 tetra-arylhydrazines that have been described,only in one single instance was the dissociation sufficiently highat room temperature to be measured. The determination in

7/28/2019 Chem rev VOL 1 091_141

44/51

134 M. GOMBERGthat case indicates a probable dissociation into the surmised

136 M. GOMBERGThe two most typical reactions of hexa-arylethanes, namelythe formation of peroxides, and the addition of iodine at roomtemperature, are not shared by the tetra-arylhydrazines.It is obvious that the three reactions of the aryl-hydrazines-spontaneous auto-oxidation and reduction, addition of nitricoxide and of triphenylmethyl, are nicely explainable on theassumption of previous dissociation into free diarylnitrogenradicals which possess a t least a very brief period of temporaryexistence. But it must be conceded that this is not the onlypossible explanation of the mechanism of the above reactions.It is conceivable that auto-oxidation and reduction may occurwithin the whole tetra-arylhydrazine molecule and that it

breaks subsequently to that into unequal parts; it is also con-ceivable that the addition of the unsaturated groups, NO orR,C -, occurs, prior to the cleavage of the hydrazine moleculein virtue of the fact tha t the two nitrogen atoms in it, beingtrivalent have each still two potential valences at their disposal:R~N-NRsI I + R2N-NO+ RzN-NOON NO

Thus, the chemical reactions which characterize the tetra-arylhydrazines, although they resemble those of the hexa-arylethanes, may be interpeted without necessarily assumingthe independent existence of diary1 nitrogen radicals. Thehydrazines themselves represent, to begin with, unsaturatedsystems, and therefore the case is somewhat different from thatof the hexa-arylethanes. In the latter, we start with completelysaturated molecules; hence the spontaneous union with iodine,oxygen, hydrogen, phenol, xylene, phenylhydrazine, etc., couldnot possibly occur unless the ethane does dissociate into radicals.

The physical behavior of tetra-arylhydrazines, however, pre-sents more cogent evidence in favor of assuming dissociation intoradicals. First, the tetra-arylhydrazines, although colorless,give rise to colored solutions, and the color is intensified whenthe solution is diluted or when its temperature is raised. Thisbehavior is entirely analogous to that of the hexa-arylethanes.

7/28/2019 Chem rev VOL 1 091_141

45/51

ORGANIC R A D I C A LS 135The chemical reactions of the tetra-arylhydrazines are, insome respects, analogous to those of the hexa-arylethanes.When a solution is allowed to stand for some weeks, or whenheated for a short time, decomposition ensues. The productswhich result suggest, first, that the hydrazine has split into twodiarylnitrogen radicals; second, these interact, one radical be-coming reduced a t the expense of the other, just as under theinfluence of light, one triarylmethyl radical becomes reducedat the expense of the second; third, the dehydrogenated diary1nitrogen radical polymerizes to a stable perazine product:

CsHs4(C&)zN - (CsHa)*NH + 1t2(CsHs)2N=N (C Hs)

1CdHsThe second characteristic property of tetra-arylhydrazines istheir reaction with nitric oxide, some combining with it at room

temperature, others at elevated temperatures, forming fairlystable, but still dissociable, addition products :(CeH5)t N - N = (CsH6)g +2 N O = 2 ( C G H ~ ) ~-N O

Here again, one might infer that the addition of the nitric oxideoccurs in consequence of a prior dissociation of the hydrazineinto the active diphenyl nitrogen radicals, just as one assumestha t the addition of nitric oxide to hexa-arylethane occurs be-cause of the prior dissociation of the latter into triarylmethylradicals.The third characteristic reaction is that tetra-arylhydrazineacombine at room temperature, or at higher temperatures, withtriphenylmethyl :

RsN-KRz +2 CR, = 2 RzK-CRaHere also, dissociation into diaryl-nitrogen may be assumed tohave occurred previously to taking on the triarylmethyl.

7/28/2019 Chem rev VOL 1 091_141

46/51

136 M. GOMBERGThe two most typical reactions of hexa-arylethanes, namelythe formation of peroxides, and the addition of iodine at room

temperature, are not shared by the tetra-arylhydrazines.It is obvious that the three reactions of the aryl-hydrazines-spontaneous auto-oxidation and reduction, addition of nitricoxide and of triphenylmethyl, are nicely explainable on theassumption of previous dissociation into free diarylnitrogenradicals which possess a t least a very brief period of temporaryexistence. But it must be conceded that this is not the onlypossible explanation of the mechanism of the above reactions.It is conceivable that auto-oxidation and reduction may occurwithin the whole tetra-arylhydrazine molecule and that itbreaks subsequently to that into unequal parts; it is also con-ceivable that the addition of the unsaturated groups, NO orRsC -, occurs, prior to the cleavage of the hydrazine moleculein virtue of the fact that the two nitrogen atoms in it, beingtrivalent have each still two potential valences at their disposal:

RzN-NRzI I + RzN-NO+R~N-NOON NOThus, the chemical reactions which characterize the tetra-arylhydrazines, although they resemble those of the hexa-arylethanes, may be interpeted without necessarily assumingthe independent existence of diary1 nitrogen radicals. Thehydrazines themselves represent, to begin with, unsaturatedsystems, and therefore the case is somewhat different from thatof the hexa-arylethanes. In the latter, we star t with completelysaturated molecules; hence the spontaneous union with iodine,oxygen, hydrogen, phenol, xylene, phenylhydrazine, etc., could

not possibly occur unless the ethane does dissociate into radicals.The physical behavior of tetra-arylhydrazines, however, pre-sents more cogent evidence in favor of assuming dissociation intoradicals. First, the tetra-arylhydrazines, although colorless,give rise to colored solutions, and the color is intensified whenthe solution is diluted or when its temperature is raised. Thisbehavior is entirely analogous to that of the hexa-arylethanes.

7/28/2019 Chem rev VOL 1 091_141

47/51

ORGANIC RADICALS 137If we exclude the possibility that the hydrazines tautomerizeinto some kind of quinonoid modification without change ofmolecular weight, then the color phenomena above cited arebest ascribed t o the fact that dissociation into radicals hasoccurred.While there is at present no experimental evidence in favorof the assumption that the diary1 nitrogen radicals tautomerizeinto the quinonoid modification, the possibility of this occurringis not excluded. We would then have another instance of tr i-valent carbon, as in the dehydrophenok.

(CeHdzN * (CeHdN =/=vH\=/\Triaryl hydrazyls. Quite similar to the tetra-arylhydrazinesis the class of the remarkably unstable and very reactive sub-stances known as the tetrazanes. They too, are made bydehydrogenation of complex amines by means of mild oxidizingagents such as lead peroxide, as for instance:

1The tetrazanes so formed, according to Stefan Goldschmidt (14),are dissociable into two radicals; in the instance cited, one mole-cule of hexa-aryltetrazane dissociated into two triphenylhy-drazyls.In their chemical and physical behavior the tetrazanes re-semble the tetra-arylhydrazines. They undergo when dissolved,spontaneous decomposition even at lower temperature thantetraphenylhydrazines. They unite with nitric oxide and withtriphenylmethyl. Colorless in the solid stage, their solutionsare green to blue, the color increasing in intensity when thesolution is further diluted, or when its temperature is raised.Thus here again, the fact that the solutions of these substancesdo not follow Beers law, seems t o be the main argument in

7/28/2019 Chem rev VOL 1 091_141

48/51

138 M. GOMBERGsupport of the assumption th at they are dissociable into radicalswith divalent nitrogen. Molecular weight determinations inbenzene have been made on four tetrazanes. Three are de-scribed as only slightly dissociable because they give only slightlycolored solutions. While the results obtained from the molec-ular weight determinations would perhaps indicate some slightdissociation of these three tetrazanes, the author himself hesi-tates to draw this conclusion, because, as he says, decompositionof the compound might have ensued. Th e fourth tetrazaneproved the most stable of all so far prepared:

(CsHdaN-N N-N(C&)a

In the solid state it looks like potassium permanganate, itssolutions are of violet color, not unlike that of the permanganatesalt. From this as well as from the molecular weight of thesubstance (calculated 394; found, 335 on a concentration of 0.06t o 1.40 per cent, the conclusion is drawn tha t the tetrazane is dis-sociated in solution completely into the monomolecular radicals.And yet, what appears very strange, this particular tetrazaneis the only one that does not respond to two of the most typicalreactions for radicals with divalent nitrogen. It does not com-bine with nitric oxide at all, and the union with triphenylmethylgives a combination not of the usual and the anticipatedconstitution.

From what has been said above, it is obvious that radicalswith divalent nitrogen, such as diary1 nitrogen radicals andtriaryl hydrazyls, probably do exist, bu t none the less more ex-perimental evidence in favor of this view is much to be desired.Evidently we have here organic radicals which are far less stablethan the triarylmethyls, and when they are produced at all,they have such a transitory existence that, from the experi-

7/28/2019 Chem rev VOL 1 091_141

49/51

ORGANIC RADICALS 139mental side, the dii culties in handling and studying theseelusive compounds becomes greatly enhanced.A divalent nitrogen derivative o f carbazole. Branch and Smith(15) have subjected carbazole to oxidation by means of silveroxide, and obtained a substance which closely resembles thediarylnitroger, radicals. It is colorless, gives rise to colored solu-tions, and has a molecular weight which indicates dissociation.

However, colorimetrically, it obeys Beers law.V . ABNORMAL VALENCE O F TIN AND LEAD

Mention must be made here of the interesting compoundsof tin, described by Rugheimer in 1909. Hexa-ethyl tin givesa molecular weight in benzene which indicates partial dissocia-tion into, presumably, radicals containing trivalent tin. Witha concentration of 1.75 per cent of the compound, dissociationoccurs to the extent of 50 per cent. The substance possessesa yellow color, and the solutions are slightly colored.Krause and eo-workers (16), in a series of papers, have quiterecently described the preparation of several hexa-aryl plum-bates : hexaphenyl, 0- and p-hexatolyl, hexaxylyl, and hexa-cyclohexyl plumbate. Some are pale yellow, some pale greenwhen solid, and they form colored solutions in organic solvents.They have been prepared under specific conditions in accordancewith this reaction.

6 R-MgBr +3 PbC1, = (PbRs)2+Pb +3 MgBr2+3 MgCllFrom the molecular weight determinations of these interestingcompounds Krause concludes that while the hexaxylyl compound

7/28/2019 Chem rev VOL 1 091_141

50/51

140 M. G O M B E R G does not dissociate appreciably, if at all, the other compoundsmentioned do, especially in dilute solution :

(PbR3)2 2PbRsUnfortunately, the total depression in the freezing point ofthe solvent occasioned in these experiments was of low order ofmagnitude, varying from 0.01 to 0.05. The results suggestthe occurrence of dissociation into radicals (Pb RJ, bu t they canhardly be depended upon as furnishing a quantitative measureof the amount of dissociation.As regards the chemical characteristics of these so called

triaryl lead compounds, they absorb oxygen very slightlyand give, with iodine, triaryl lead iodide.V I . SULFUR AND ARSENIC WITH, POSSIBLY, ABNORMAL VALENCE

The many attempts of Lecher to prepare such diaryl sul-fides that would dissociate, R-S-S-Rs2 R-S, nd in this waygive rise to radicals with monovalent sulfur, have not as yetproved successful. Although these disulfides actually give solu-tions which develop color on being heated, molecular weightdeterminations fail to indicate that dissociation into half mole-cules ensues.Tetraphenyl-diarsine has been the subject of many investi-gations, in the expectation that this compound, in anology withtetra-arylhydrazines, might perhaps dissociate into diaryl arsine,(C8HJ2= As. The results of the molecular weight determina-tions give no such indication, and the solution remains colorlesseven when heated. And yet, the substance absorbs oxygen,combines with iodine, manifesting an unsaturated character.P. Borgstrom and Margaret Dewar (17), however, draw theconclusion that it may be said tha t the bond between the arsenicatoms of tetra-phenyl diarsine is easily broken. Divalent arsenicof the type (CsH6)2= A s- may be present in solution, but it isdoubtful if it is the stable form. The valence or configurationof the stable form is unknown.

7/28/2019 Chem rev VOL 1 091_141

51/51

ORGANIC R A D I C A L S 141REFERENCES

The main results of the various investigations on organic radicals have becomeincorporated in the standard texts on organic chemistry. Consequently, i t didnot seem necessary to the writer of this paper to cite in detail the sources in thechemical journals. The following extensive reviews contain quite complete ref-erences to the publications up to 1914:SCHMIDLIN,.: Triphenylmethyl, Ferdinand Enke, Stuttgart, 1913, XI1 +WIELAND,H.: Die Hydrazine, Ferdinand Enke, Stutt gard, X +244 pp.GOMBERG,.:

233 pp.The Existence of Free Radicals, J. m. Chem.Soc., 36, 1144-

Mention should also be made of the brief, but very suggestive address byP. Walden on Free Radicals in Rec. Trav . Chim. Pays-Bas, (4) 41,530 (1922).The few references given below are mainly t o some investigations publishedsince 1914.

1170 (1914).

(1) GOMBERG,.: J . Am. Chem. Soc., 22, 757 (1900).(2) GOMBERG, ., AND SULLIVAN, . W., JR.: J . Am. Chem. Soc., 44, 1810(3) GOMBERG, ., AN D SCHOEPFLE,. S.: J.Am. Chem. Soc., 39, 1652 (1917).(4) MASCARELLI,., AND BENATI,G.: Gazz. chim. i ta l . , (2) 39, 642 (1909).(5) GOMBERG,., AND SULLIVAN,. W., JR.: J. m. Chem.Soc., 44,1812 (1922).(6)LEWIS,G. N.: Proc. Nut . A cad . A r t s Sci . , 2, 588 (1916).(7) COLE,H. I. : Philipp. J. c i . , 19, 681 (1922).(8) DIMROTH, . , AN D FRISTER,.: Ber., 66, 1223 (1922); EMMERT, ., AND(9) SCHOLL, . : Ber., 64, 2376 (1921); 66, 918 (1923).(10) ZIEGLER,K., AND OCHS,C. : Ber., 66,2264 (1922).(11) SCHLENK, .: Ber., 66, 2291, 2299 (1922).

(12) PORTER,. W., AND THURBER,. H. : J. m . Chem. Soc., 43, 1194 (1921).(13) GOLDSCHMIDT,.: Ber., 63, 44 (1920); 66, 616, 628 (1922).(14) GOLDSCHMIDT,.: Ber., 66 , 3194, 3197 (1922).(15) BRANCH, . E. K., AN D SMITH,J. F.: J . Am . Chem. Soc., 42 , 2405 (1920).(16) KRAUSE, .: Ber., 62 , 2165 (1919);64, 2060 (1921);66,888 1922).(17) DEWAR,M .: J.Atn. Chem. S O C . , 4, 2915 (1922).

(1922).

WERB,0 . : bid., 1352.