2596 THOLE: VISCOSlTY AND ASSOCIATION. PARI' I.CCLXV1.-Viscosity a n d Association. Pa7.t 1. Asso-ciation. o f the Phenols.By FERDINAND BERNARD THOLE.A STUDY of the literature on the subject of viscometry shows theexistence of much evidence which indicates that association isaccompanied by a considerable augmentation of viscosity (compareDunstan and Thole, Trans., 1909, 95, 1556).On the other hand, in a paper published by Jones and Veazey(Amer. Chem. J . , 1907,37,405), it was suggested that the depressionof the degree of association of a liquid is accompanied by an increasein its viscosity. The increase in viscosity produced when alcoholsare dissolved in water was explained by these authors as being dueto a diminution of the association of each component, resulting inan increase in the number of non-associated molecules, and thereforeof the available frictional surface.From a study of the viscosity-concentration curves for aqueous solutions of many inorganic salts,they also concluded that ions of large atomic volume, such aspotassium, rubidium, and msium, give rise to the phenomenon ofnegative viscosity.Since the relative degree of association of many phenols and theirderivatives has been investigated by using a variety of physicalmethods, particularly cryoscopy, specific inductivity, and capillarity,it was considered of interest to compare the results obtained byviscometric methods with those determined by other physicalmethods.Pinette (Annalen, 1888, 243, 32) showed that the boiling pointsof the phenols are, in general, higher than those of their methylethers, and that ortho-substituted phenols boil at a lower tem-perature than their meta- and para-isomerides.The current ideasconcerning association and steric hindrance are in accordance withthese results, assuming the boiling points of similarly constitutedcompounds to vary in the same sense as the molecular weights.Astm and Ramsay (Trans., 1894, 65, 168), by measurement ofthe molecular surface energy of phenol, concluded that it was con-siderably associated.Auwers (Ber., 1895, 28, 2878) carried out a very comprehensiveseries of cryoscopic investigations of the molecular weights of variousphenols in naphthalene solution. He showed that whilst phenolis considerably associated, ortho-substituted phenols are practicallyunassociated, the association increasing to a maximum with the para-compounds.Meta-substituted phenols are associated to an interASSOCIATION OF THE PHENOLS. 2597mediate degree, but approximate more nearly to the para-series.I n the orthu-series, the aldehyde group has the greatest effect ininhibiting association ; then, in order, follow the carbethoxyl, nitro-,halogen, and finally alkyl groups.Speranski (Zeitsch. physikal. Chem., 1903,46, 70) determined thevapour pressures of solid solutions of @-naphthol in naphthalene,and ccmcluded that the former compound is associated.Philip and Haynes (Trans., 1905,87, 998) measured the dielectricconstants of phenol, its methyl and ethyl ethers, and of the cresolsin benzene and in m-xylene solutions.Their results indicate that,whilst the ethers are unassociated, phenol and the cresols aredistinctly associated, 0-cresol to the least extent.Hewitt and Winmill (Trans., 1907, 91, 441), using the capillaritymethod of Ramsay and Shields, studied the association of a con-siderable number of substituted phenols and allied substances.Their results are included in table I, and show that the phenols ingeneral are associated, the association being least in the ortho-series.Considerable depression of association is produced by thecarbethoxyl, nitro-, and halogen groups, but alkyl groups only exerta slight influence. Benzyl alcohol wits shown to be associated, butthe introductlon of phenyl groups inhibits this to such an extentthat benzhydrol and triphenylcarbinol are practically unassociated.It appears uncertain whether substitution in the meta- or para-positions has the greater effect, t,he results on this point beingindecisive.It seems to the author, however, that, although thecapillarity method will show satisfactorily whether or no a substanceis associated, little trust can be placed in figures indicating theactual degree of association, for in a mixture one is not entitled toassume that the constitution of the surface layer is typical of thatof the bulk of the liquid, and associated liquids may fairly beregarded as mixtures of molecular aggregates and simple moleculesEXPERIMENTAL.As the apparatus used has been considerably modified with a viewto increased accuracy since it was previously described by the author(this vol., p.1251), a somewhat detailed account of the viscometerand the accompanying fittings may be of interest.The viscometer was of the original Ostwald type, but was providedwith four etched marks instead of two. Guard tubes were alsoattached to the two limbs, and were provided with bulbs, whichcontained cotton wool moistened with the liquid under investigation.By adopting this precaution, volatile liquids, such as acetone andbenzene, may be safely used as solvents. The dimensions of theviscometer may, of course, be varied, according to the quantity andVOL. XCVII. 8 2598 THOLE: VISCOSITY AND ASSOCIATION. PART I.nature of the liquid used.For the present research, the bulbsA and B contained approximately 3 and 5 C.C. of liquid respec-tively. The length of the capillary was 8 cm., and the diameter0.025 cm.Before use the viscometer was cleaned successively with hotchromic acid mixture, hot absolute alcohol, and boiling, filteredcoductivity water. Finally, it was dried in a current of warmdusbfree air. Every precaution is necessary to avoid the intro-duction of dust, as this will inevitably choke the capillary.The clean, dry viscometer was suspended by means of a carrierprovided with spring clips in a, copper thermostat fitted with largewindows front and back. I nthis bath were fixed a stirrerFonnected with a small hot-airmotor, a standard thermometer,a Beckmann thermometer gra-duated to O*0lo, and a Lowryt'hermoregulator of the '' spiral "pattern.The determinationswere carried out in a laboratoryallocated for the purpose, andthe variation of temperature ofthe thermostat did not exceed0'02O.The liquid was filtered or,whenever possible, distilled intothe viscometer, and after theapparatus had remained in thethermostat for ten minutes, theievel of the liquid was adjustedto the marks c and d. Cottonwool contained in the guardtubes cc was moistened with the liquid, the tubes attached, theviscometer accurately adjusted to a vertical position by means ofthree plumb lines, and the time of flow of the liquid from a t o 71measured in the usual way. A stop watch, reading to 0-2 second,was used.Usually about seven observations, differing by not morethan 0.5 second, were made, and the mean was taken.The densities were determined in a Sprengel pyknometer of 5 C.C.capacity, and in the case of phenol and p-chlorophenol in a 10 C.C.specific gravity bottle. Each estimation was repeated until con-cordant results were obtained, and a correction was introduced forthe weight of displaced air.The constants of the instruments were determined from time toASSOCIATION OF THE PHENOLS. 2599time, using conductivity water. The values adopted for the viscosityof water were those determined by Thorpe and Rodger.For the two determinations at 130°, the ordinary thermostat wasreplaced by a large beaker filled with heavy petroleum and heatedby a small burner.The temperature was regulated by hand, themaximum fluctuation being 0’2O. ,The substances used in the research were carefully purified byrepeated fractionation with a rod-and-disk column or by re-crystallisation. In no case did the liquid used for the determinationboil over a greater range than 0’2O. Unless otherwise stated, thefraction used boiled constantly. The source and method of purifi-cation of the materials is indicated below.Phenol.-Kahlbaum’s (‘ synthetic ” phenol. Fractionated threetimes.A nisoZe.-The specimen used was obtained from a large quantityof the carefully dried ether.PhenetoZe.-Purified in the same way as anisole.PAenyZ ,4 cetate.-Prepared from synthetic phenol and aceticCresoZs.-Schuchardt’s purest products, fractionated three times.TolyZ Methyl Ethers.-Carefully dried and fractionated.o-ChZorophenoZ.-Fractionated three times from a specimen sup-plied by Kahlbaum.m-ChZorophenoZ.-Purified by freezing out and subsequent frac-tionation.p-ChZorophenoL-Purified by freezing out and subsequent frac-tionation.o-Nitrophenol.-Three times distilled in a current of steam, andfinally recrystallised from absolute methyl alcohol.m-Nitrophenol.-Twice recrystallised from pure benzene, anddried in a vacuum over paraffin wax.p-NStropheno2.-Freed from traces of the ortheisomeride bymeans of a current of steam, twice recrystallised from dilute hydro-chloric acid, and dried in a vacuum over potassium hydroxide.Eth$ SalicyZate.-Fractionated three times from a quantity ofthe pure ester.Ethyl m-Hydroxybenzoate.-Twice recrystallised from purebenzene, and dried in a vacuum over paraffin wax.E t h y Z p-Hydroxy b enaoat e.-Twice recryst allised from absolutealcohol.SdicyluZdehyde.-Fractionated twice.Bensyl AZcoho2.-Prepared from re-distilled benzaldehyde by theThe specimenanhydride.The specimen was free from phenol and acetic acid.The specimen boiled within 0.2O.Cannizzaro reaction, and fractionated three times.used boiled within 0’2O.8 G 2600 THOLE: VISCOSlTY AND ASSOCIATION. PART I.Benzyl A cetate.-Fractionated three times.Benzyl illethyl Ether.-Prepared from pure benzyl chloride andmethyl-alcoholic sodium methoxide. The specimen was free fromhalogen and from methyl alcohol.BenzhydroL-Prepared from re-distilled benzaldehyde andmagnesium phenyl bromide, and crystallised several times from lightpetroleum.TTiphenylcarbinoZ.-Recrystallised from benzene.a- and P-Naphthols.-Fractionated from Schuchardt’s purestproducts.Amyl Acetate.-A large quantity of Kahlbaum’s purest productwas twice fractionated.It boiled within 0-5O.Ethyl Alcohol.-Kahlbaum’s absolute alcohol was fractionallydistilled over cleaned calcium turnings.As the differences of viscosity are most marked in the case of thepure substances, the viscosities of all those which are liquid at 4 5 Owere measured at this temperature. Since some of the compoundsit was desirable to investigate are solid at this temperature, solutionsof these substances had to be used.Some care had to be exercised in the choice of a solvent whichwould have the least dissociating effect and at the same time wouldreadily dissolve the substances, which would have a low vapourtension (to minimise errors due to evaporation), and could easilybe obtained in a fairly pure state.Amyl acetate was finally chosenas fulfilling these conditions most nearly.The solutions used were of equimolecular strength, 1 / 100th ofa gram-molecular weight of the substance being accurately weighedinto a stoppered weighing bottle, and dissolved in 6 C.C. of amylacetate, which was run in from a suitably graduated pipette.I n two instances (with ethyl p-hydroxybenzoate and triphenyl-carbinol), it was found impossible to prepare solutions of thisstrength, and in these cases the viscosities of several solutions oflower concentration were determined, the value for the strongersolution being obtained by extrapolation. Although the accuracyof these results may not be of the same order as in the directdeterminatibns, the qualitative nature of the results is not affected.It has been shown (Dunstan and Wilson, Trans., 1907, 91, 83;Getman, Arner.Chem. J., 1903, 30, 1077) that the value of theexpression --?- x 106 indicates to some extent the existence ofassociation in a liquid. For each particular series of compounds,this quantity is approximately constant, and for non-associatedliquids does not exceed 60. The values for hydroxylated liquids are,however, much higher, those for water, ethyl alcohol, and ethyleneMol. volASSOCIATION OF THE PHENOLS.2601glycol being respectively 494, 193, and 2750, and the variation of thevalue of this expression is a very sensitive indication of association.The values for each of t,he substances investigated is gven in thetables, and affords, perhaps, the best means of comparing the results.It should be noted, however, that this value is affected to someextent by symmetry as well as by association. This is seen in thecase of the tolyl methyl ethers, and more particularly in the caseof benzyl methyl ether, where the value of 2 x lo6 is slightlyhigher than the mean figure for the unassociated ethers, probablyowing to the comparatively long sidechain.Mol. vol.TABLE I.The Pure Substances at 45O.Substance.Phenol ..................Phenyl acetate .........Anisole..................Phene tole ...............p-Chlorophenol ......m-Chlorophenol ......o-Chlorophenol ......p-Cresol ...............m-Cresol ...............o-Cresol.. ................m-Tolyl methyl ethero-Tolyl methyl etherp-Tolyl methyl ethero-Nitrophenol .........Ethyl salicylate ......Salicylaldehyde ......Benzyl alcohol .........Benzyl methyl ether&Naphthol (at 130")a-Naphthol (at 130")Benzyl acetate .........HewittandWinmill'sTime of flow 9 106 associationDensicy. Viscosity. j ~ o ~ ~ o l m constants. in seconds.837 '2374'2167.0191-41044.8826.9406.71208'81091.3742.7200-6193.7185'8433.5350.5320.1640.5296.5233.6376.0312.61 '0551 '0520'57070,94271 -2601 *2491'2101.0151 -0141 -0270.95460.95890'94971 -1 831.1061'1411 -0271.0330.9624-0 *040360.017990 '0074090 '0082490-060180'047220'022500 -056070-050570.035060 *OO 87 5 30*0084910 w0080640 *023430.017720.016690 '03 0080 '01 3990.01028-TABLE 11.Solutions in AmylTime of flowSubstance. in seconds.Amyl acetate .................201.4Phenol ........................... 267-1Anisole.. ......................... 202.0Phenyl acetate ............... 223.3Phenetole ..................... 209 '6p-Chlorophenol ............... 295 '9m-Chloro henol ............... 290 *3o-ChloropXenol ............... 281 -1Acetate at 25O.453.0136.066 '663'8590'0459.0212'0527'0475.0333.068.566 *762'8199.0118-0156.0286-096 -381 -1--Density.0.86590.89510.88560.90330.88290.92890.92740'9254Viscosity.O.OG80550 -011 050 *O 082650.0093190 -0085500-012700'012440'012021 '3---1 '221.491 *o1 -621'481.12 ---0 '840-91-66-- ---x 106.Mol. vol. -lG5-067.861'961-491 9389-886 *2602 THOLE: VISCOSITY AND ASSOCIATION. PART I.TABLE 11. (continued).Solutions in Amy2 AcetateTime of flowSubstance. in seconds. Density.p-Cresol ........................o-Cresol ........................m-Cresol ......................m-Tolyl methyl ether ......o-Tolyl methyl ether ......p-Tolyl methyl ether .........p-Nitrophenol ..................912-Nitrophenol ...............o-Nitrophenol ..................*Ethyl p-hydroxybenzoate ...Ethyl m-hydroxybenzoate ..Ethyl salicylate ...............Benzyl alcohol ..............Benzhj drol .....................'Triphenylcarbinol ............Benzyl methyl ether .........Benzyl acetate.. ................8-Naphthol .....................a-Naphthol ..................283.4281.3282.9214.9211'7210.6358.5331.1242.9370'9246.3250.6351-4207'8233.6375.537 1 -2--0.89310.89640'89280.88470.88530.88400'94400'94180-93650.9310.92610.91840'89430.92120.970.88320.90230'92100'9031at 25O.Viscosity.0.011 690-01 1650.011670 -0087840.0086590.0086020 015640'014490.010510.01660.015870 -01 04 50.010350-014960.01900'0084790 '0097380.015980.01549x106.hlol. vol.96.796.796-563.762'862-3106'098.270-893.088.657 '685.774.970 -061 '458.6102.097 -1* By extrapolation.TABLE 111.The Chloropltenols in Ethyl Alcohol at 25O.77 x 106. Tirnc of flowSubstance. in seconds. Density. Viscosity. MaxjAlcohol .................. 309 *2 0.7876 0-01135 -p-Chlorophenol ......... 364.0 0.8646 0 *014 54 97'8o-Chlorophenol.. ....... 364 -2 0.8628 0.01452 97'5nz-Chlorophenol . . , . , . 363 -7 0.8631 0 -0 1450 97'4Discussion of R e s d t s .Table I.-The results in column 4 follow very closely thoseobtained by Auwers and by Hewitt and Winmill, with one exception.In two cases (the cresols and hydroxybenzoic esters), the latterauthors found the order of increasing association to be ortho-meta-para,, and in two other cases (the chlortl and nitro-phenols) tobe ortho-par-meta, whereas Auwers in all cases found the orderto be ortho-meta-para, the meta-compound approximating morenearly to the para-compound.The viscosity results are in full agreement with those of the latterauthor, though it must be remembered that V.Meyer found theorder of steric hindrance in the case of the esterification of sub-stituted benzoic acids to be ortho-para-meta (Bey., 1895, 28,1254)ASSOCIATION OF THE PHENOLS. 2603That the order of the viscosity results is not merely due to theposition of the subs€ituent group in the benzene nucleus apart fromits influence on the hydroxyl group is shown by the fact that thevalues for the tolyl methyl ethers do not fall in this.order.It will be noticed that the mean value of the expressionY? x 106for the ethers is 66, these compounds being practically3101. vol.unassociated. The value for benzene is 65 (Dunstan, Zeitsch.physikat. Chem., 1905, 51, 738).In the case of benzyl methyl ether, the slightly higher value maybe due to disturbance of the symmetry of the molecule by the com-paratively long side-chain. Phenyl and benzyl acetates appear to beslightly associated, this being probably due to the slight residualaffinity possessed by the carbonyl oxygen atom in the acetyl group.For benzyl alcohol and all the phenols, A x loG is markedhigher, indicat'ing considerable association.Ortho-substitutedphenols appear to be less associated than phenol itself, presumablyowing to the proximity of the substituent to the source of association,the carbethoxyl, aldehyde, nitro-, and halogen groups exercising veryconsiderable influence. The marked effect of these groups has beenpointed out by Auwers and by Hewitt and Winmill.The results obtained by Auwers indicate that the aldehyde grouphas the greatest inhibiting influence, followed in order by thecarbethoxyl, the nitro-, the halogen, and finally the alkyl groups.The viscosity results, while agreeing in the main with these, invertthe order of the first two groups. No satisfactory explanation ofthis discrepancy is apparent.It is noteworthy that in the first four cases the ortho-substituentcontains an unsaturated nucleus, and it appears probable that thelatent valency of this nucleus in some way attracts part of that ofthe hydroxyl group, thus lessening the tendency for association.This action is, of course, supplementary to the steric hindranceproduced by the ortho-substituent by virtue of its proximity to thehydroxyl group.This explanation is rendered still more probable by the fact thatthe aldehyde and carbethoxyl groups, which depress association moststrongly, are known to be markedly unsaturated, whilst the methylgroup, which exercises only a slight inhibitive influence, is practicallysaturated.An interesting parallel is observed when the molecular refrac-tivities of various substituted benzene derivatives are considered(Smiles, Chemical Constitution and Physical Properties, p.298).Mol. V O ~ 2604 THOLE : VISCOSITY AND ASSOCIATION PART I.Substance. M(a) (obs.). M(=) (calc.Phenylacetylene ............ 34-46 33-53Styrene ........................ 3 5 9 8 35-08Benzaldehyde ............... 31 -77 31'01Nitrobenzene ............... 32-69 32-10Benzonitrile .................. 31 -32 30-75Aniline ....................... 30.27 29'72Acetophenone ............ ._ 36.00 35-58Methyl benzoate ............ 37 *55 37.23Phenol ........................ - -Iodobenzene ..................Bromohenzene ...............Chlorobenzene ..............Benzene.......................- -- -- -- -). A. + 0.93+0*90 + 0.76+0'59 + 0-57 + 0.55+0'42 + 0-32- 0.07- 0-24- 0'31- 0.32- 0.38A due tosubstituent. + 1.31+1*28 + 1.14 + 0.97+0*95 + 0-93 + 0.80+0*70+0'31f 0'14-t 0.07+0*06The degree of disturbance of the benzene system increases withthe increasing degree of unsaturation of the substituent. It hasalso been shown, from measurements of molecular magnetic rotation(Perkin, Trans., 1896, 69, 1152), that the anomaly shown by di-methylaniline is diminished when the residual af6nity of the basicgroup is satisfied by the addition of hydrochloric acid. A similarargument appears to explain the gradual decrease in the reactivityof the carbonyl group in the series acetone-ethyl acetate-aceticacid. In the first compound, where the carbonyl group is adjacentto two saturated methyl groups, it possesses a, sufficient degree ofunsaturation to combine with sodium hydrogen sulphite and withhydrogen cyanide.In ethyl acetate one of the adjacent groups isethoxyl, the oxygen atom of which possesses a certain amount oflatent affinity (compare the combination of ethyl ether with hydrogenchloride and with magnesium alkyl halides). This affinity exerts anattractive influence on part of the latent aifinity of the carbonylgroup, with the result that ethyl acetate will not combine witahsuch a, number of reagents as will acetone.In acetic acid, the oxygen of the hydroxyl group possesses con-siderable residual affinity, part of which is united with that ofthe carbonyl group, which thus loses its characteristic additiveproperties, whilst the remainder produces association of themolecules.Owing to the comparatively high temperature employed inworking with the fused naphthols, accurate density measurementswere not attainable, and only the time of flow (the chief factor indetermining the viscosity) was measured.The results indicateclearly the influence of the second ring of the naphthalene nucleusin hindering the association of a-naphthol. V. Meyer observed asimilar effect when measuring the velocities of esterification of2-chloro-l-naphthoic acid and 3-chloro-2-naphthoic acid, the formercompound esterifying very slowly (Ber., 1895, 28, 182).Table ZZ.-It should be pointed out that the results in column 4have been obtained with solutions of an empirical concentrationASSOCIATION OF THE PHENOLS.2605and axe therefore comparable only among themselves, and not withthe corresponding figures in table I.It will be seen that the results in table I are exactly confirmed,and also that a further range of substances has been studied. I neach case the phenols are associated, the ortho to the least, and thepara to the greatest extent, the meta-compounds, as before,approximating more closely to the partwompounds. The gradualinhibition of the association of benzyl alcohol by the progressivereplacement of the hydrogen atoms of the sidechain by phenyl groupsis also clearly indicated. This, again, is in full agreement with sur-face energy results.Although the figures for triphenylcarbinol areobtained by extrapolation and therefore are not of the usual degreeof accuracy, it is clear that this substance approximates to the ethersin its slight degree of association.Ethyl salicylate and o-nitrophenol, again, show practically nosigns of association.A very striking point is the dissociating influence of the amylacetate. This was specially chosen as a comparatively inert, non-dissociating solvent, but its effect on the associated solutes is mostmarked. The viscosities of the cresols in the pure state differ veryconsiderably, but solution evidently breaks down the complexes toa large extent, and the viscosities of the solutions are almostidentical. The dissociating influence is more marked in this casethan in that of the other phenols, the cresol complexes beingapparently more unstable than those of the other substitutedphenols.If the difference between the values for --L- x lo6Mol. vol.found for ortho- and para-isomerides be taken as a measure of therelative stability of the complexes, this stability decreases in theorder c arb et hoxyl-ni t r c+-halogen-alkyl. The slight signs ofassociation noticeable in the pure phenyl and benzyl acetates dis-appear in th6 amyl acetate solutions, arzd the substances appearto be completely dissociated.Table ZIT.-As it was observed that a comparatively non-dissociating solvent had such a marked disruptive effect on themolecular aggregates, it was considered of interest to determine theviscosities of solutions of a set of three isomerides in a dissociatngsolvent.For this purpose, solutions of the chlorophenols in ethylalcohol were chosen, and the results indicated that practically com-plete dissociation h*ad resulted. Preliminary experiments withsolutions of the chlorophenols in light petroleum (one of the leastdissociating of ordinary solvents) showed that the dissociation pro-duced was less than in the case of amyl acetate, and a detailedstudy of the effect of various solvents on associated substances isnow being carried out2606 THOLE: VISCOSITY AND ASSOCIATION. PART I.Summary of Results.(1) The results obtained by the viscometric method agree veryclosely with those derived from other physical constants, such asvapour pressure, dielectric constant, molecular surf ace energy,molecular refractivity, and molecular weight determined cryo-scopically.(2) Viscosity determinations, using --?-- x lo6 as a criterionof association, show that phenols are associated, the orthecompoundsto the least, and the para-compounds to the greatest extent. Theethers are unassociated, but the acetates show slight association.(3) The carbethoxyl, aldehyde, nitro-, and halogen groups exerta marked inhibitive influence on association. Alkyl groups onlyexert a slight influence. Since the degree of inhibition of associationappears to be intimately connected with the degree of unsaturationof the substituent, it is suggested that the depression of associationis partly due to some kind of attraction between the latent valencyof the hydroxylic oxygen and the unsaturated substituent, the con-sequence being a diminution of the tendency to form complexesthrough the latent valency of the hydroxyl group. The samehypothesis explains the gradual disappearance of the characteristicreactivity of the carbonyl group in the series acetone-ethyl acetate-acetic acid.(4) Solution in even a comparatively inert solvent, such as amylacetate, produces considerable disruption of the molecular aggre-gates. In a dissociating solvent, such as ethyl alcohol, thedissociation is practically complete.Mol. vol.The author desires t o express his sincere thanks t o Dr. J. T. Hewittand to Dr. A. E. Dunstan for the interest they have taken in t,hework, and to the Chemical Society for a grant which has coveredthe expense entailed.EAST LONDON COLLEGE. EAST HAM TECHXIGAL COLLEGE