ABSTRACT:Light scattering measurements have been made on nine polyisobutylene (PIB) fractions in heptane, an intermediate solvent, at 25°C to determine the second virial coefficient A 2 and the third virial coefficient A 3 as functions of weight-average molecular weight Mw ranging from '2.7 x 10 4 to 7.8 x 10 6 . For six of the fractions, z-average mean-square radii of gyration have also been determined. In the molecular weight range studied, A 2 and A 3 vary as Mw -0 -21 and M w o.ss, respectively, and the factor g defined by A 3 / A ,2 M w is about 0.33 almost independent of molecular weight. The data of g and 'l' (the interpenetration function) as functions of the cube of the radius expansion factor Cls, combined with previous data in cyclohexane, a good solvent, show that the two-parameter theory breaks down unless Cl/ is larger than about 2 for g and about 5 for 'l'. It is concluded from the comparison of these g data with the recent theory of Norisuye et al. that the failure of the two-parameter theory for g at small Cls 3 is due primarily to the neglect of three-segment interactions and that the effect of chain stiffness on g in heptane is of minor importance in the M w range studied. On the other hand, the stiffness effect on 'l' for PIB is found to be remarkable even at large Cls 3 • In fact, the different dependences of 'l' on Cls' observed for heptane and cyclohexane solutions are explained semiquantitatively by the Yamakawa theory which takes account of the stiffness effect within the binary cluster approximation.KEY WORDS Second Virial Coefficient / Third Virial Coefficient / Polyisobutylene / Excluded-Volume Effect / Expansion Factor / ThreeSegment Interaction / Light Scattering / Recently, we investigated the third virial coefficient A 3 for polystyrene 1 • 2 and polyisobutylene (PIB) 3 • 4 in both good and theta solvents by light scattering, and drew the following conclusions from experimental findings together with some theoretical calculations. 5 • 6 (1) The effect of three-segment interactions on A 3 is remarkable at and near the theta point e at which the second virial coefficient A 2 for high molecular weight M vanishes, so that the binary cluster approximation to A 3 or the two-parameter theory for it at and near e completely breaks down. (2) In good solvents, the ternary cluster effect on the reduced third virial coefficient g is appreciable only for relatively low M, and thus the binary cluster approximation holds for g unless the radius expansion factor a8 ( =