Flory-type theory of a knotted ring polymer Alexander Yu. Grosberg, 1, * Alexander Feigel, 2 and Yitzhak Rabin 2 1 Department of Physics and Center for Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 2 Physics Department, Bar-Ilan University, Ramat-Gan 52900, Israel Received 11 March 1996 A mean field theory of the effect of knots on the statistical mechanics of ring polymers is presented. We introduce a topological invariant which is related to the primitive path in the ‘‘polymer in the lattice of obstacles’’ model and use it to estimate the entropic contribution to the free energy of a nonphantom ring polymer. The theory predicts that the volume of the maximally knotted ring polymer is independent of solvent quality and that the presence of knots suppresses both the swelling of the ring in a good solvent and its collapse in a poor solvent. The probability distribution of the degree of knotting is estimated and it is shown that the most probable degree of knotting upon random closure of the chain grows dramatically with chain compres- sion. The theory also predicts some unexpected phenomena such as ‘‘knot segregation’’ in a swollen polymer ring, when the bulk of the ring expels all the entanglements and swells freely, with all the knots concentrated in a relatively small and compact part of the polymer. S1063-651X9610012-X PACS numbers: 61.41.+e Modern polymer physics is based on the analogy between a polymer chain and a random walk. While this analogy has been extremely fruitful in predicting the long-wavelength static and dynamic properties of polymer chains radius of gyration, relaxation time, etc., it misses one of the most important attributes of polymer chains, namely, their ability to form permanent entanglements, i.e., knots. Although there is a very successful phenomenological theory the reptation model 1 of the effect of temporary entanglements on the dynamics of polymers which give rise to the viscoelasticity of polymer liquids, there is no comparable theory of perma- nent entanglements which play an important role in biologi- cal systems e.g., DNA rings in bacteria. Our understanding of the latter is limited by the fact that, despite the progress in the mathematical theory of classification of knots and topo- logical invariants 2, there are only few rigorous results on the entropic properties of knots 3,4, and further attempts in this direction encounter severe mathematical difficulties 5. Computer simulations 6–8, however insightful, have some basic limitations. What appears to be missing in the area of polymer topol- ogy is a simple ‘‘physical’’ model which could give some testable physical predictions. The main purpose of the present work is to trade mathematical sophistication for the simplicity and effectiveness of such a physical approach and to formulate a model that goes along the lines of the simple Flory mean field theory of linear polymers and takes into account topological constraints, albeit in a very primitive and incomplete manner. It is also worth mentioning what we do not attempt to accomplish here: as we pursue a scaling-type approach, we cannot capture the subtle differences between simple knots, such as, for instance, trivial knot, trefoil, figure eight, etc.; we only hope to be able to describe general tendencies for very complex knots. As far as biological applications involv- ing DNA molecules are concerned, our model does not apply to current experiments on rather small DNA rings 9,10 such as plasmids, but may be relevant to very large circular DNA e.g., that of bacteria. We also stress that we are speaking of topological constraints on the dsDNA as an en- tire thread, and do not have in mind linking of the two strands in the duplex, that is related to biologically important issue of superhelicity 11. We address two questions: 1What is the equilibrium size of a polymer ring, depending on both solvent character- istics or monomer interactionsand knot topology? A simi- lar question for linear polymers, where topology is not an issue, is discussed in every textbook on polymer physics, and it is known that the Flory theory yields a very good approxi- mation for both swelling in good solvent and collapse in poor solvent. Clearly, almost the same behavior, except for some subtle chain end-related effects, is expected for a phan- tom ring polymer which can cross itself. Our goal is to consider a nonphantom ring, with quenched knot topology. 2What is the probability distribution of various knots obtained upon random closure of a linear polymer, or by random motion of a phantom ring with annealed topology? As we wish to build up a simple theory, we do not attempt to characterize knots by sophisticated polynomial invariants. Instead, we introduce the following construction. Consider a polymer chain in some spatial conformation and denote by L the contour length of the chain. Let us first construct a tube that contains the polymer chain and is sufficiently narrow such that the topology of the tube as a whole is the same as that of the polymer. We now inflate the tube such that its length L is preserved, while its cross section is roughly the same everywhere along the tube we assume that different tube portions cannot penetrate each other. This inflation will eventually end when the inflated tube fills the main part of the volume within its loops. Let us denote by D the diameter of the tube in this maximally inflated state. We state that the aspect ratio of the maximally inflated tube * Permanent address: Institute of Chemical Physics, Russian Acad- emy of Sciences, Moscow 117977, Russia. PHYSICAL REVIEW E DECEMBER 1996 VOLUME 54, NUMBER 6 54 1063-651X/96/546/66185/$10.00 6618 © 1996 The American Physical Society