Vous êtes sur la page 1sur 28

ANNUAL REVIEWS

Click here for quick links to Annual Reviews content online, including: Other articles in this volume Top cited articles Top downloaded articles Our comprehensive search

Further

Fluid Dynamics of Dissolved Polymer Molecules in Conned Geometries


Michael D. Graham
Department of Chemical and Biological Engineering, University of Wisconsin-Madison, Madison, Wisconsin 53706; email: graham@engr.wisc.edu

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Annu. Rev. Fluid Mech. 2011. 43:27398 First published online as a Review in Advance on August 25, 2010 The Annual Review of Fluid Mechanics is online at uid.annualreviews.org This articles doi: 10.1146/annurev-uid-121108-145523 Copyright c 2011 by Annual Reviews. All rights reserved 0066-4189/11/0115-0273$20.00

Keywords
microuidics, DNA, diffusion, cross-stream migration, nanouidics, multiscale simulation

Abstract
The past decade has seen a renaissance in the study of polymer solutions owing in conned geometries, the renaissance driven in part by advances in visualization of large DNA molecules and the desire to manipulate DNA for genomic applications. This article summarizes the features of the fundamental polymer physics and uid dynamics that are relevant to the ow of conned polymer solutions, then reviews the recent literature on the topic. Experiments have claried and extended prior work showing that diffusion of conned exible polymers is substantially altered by connement and that, during ow, polymers exhibit substantial cross-stream migration. Simulation methods have been developed that have the capability of capturing both polymer and uid motion in conned geometries and yield results that are in semiquantitative agreement with experiments in dilute solutions. Kinetic-theory treatments of simple polymer models have led to analytically tractable models that qualitatively encompass the key phenomena observed in experiment.

273

1. INTRODUCTION
The dynamics of solutions of long-chain polymer molecules during ow in conned geometries is important in elds ranging from enhanced oil recovery to coating processes to analytical and preparatory separation techniques for macromolecules. Since the 1990s, another important application of this topic has arisen: single-molecule approaches to the manipulation and characterization of genomic DNA, using microuidic and, more recently, nanouidic devices (Dimalanta et al. 2004, Jo et al. 2007, Tegenfeldt et al. 2004b). Along with this important application have come new experimental tools for studying polymer solution dynamics at the single-molecule level: Genomic double-stranded (ds)DNA is a large molecule [many micrometers in length, even for the genome of a virus (Shaqfeh 2005)] that is amenable to staining with uorescent dyes, so it can be directly observed under a microscope (Perkins et al. 1995). This feature of DNA has led to dramatic advances in the understanding of polymer dynamics in ow and in other nonequilibrium situations such as electrophoresis, both in bulk solution and under connement. Recent reviews (Larson 2005, Shaqfeh 2005) have addressed polymer dynamics, with an emphasis on DNA, in bulk solution. The present review focuses on ows of conned dilute polymer solutions, in which there is an intriguing and important interplay between polymer conformations and uid dynamics that has only recently begun to be unraveled. Apart from their direct importance in applications, conned polymer solutions serve as an archetype for the dynamics of other conned complex uids such as suspensions and emulsions. With the emergence of microuidic technologies that involve complex and multiphase uids (Squires & Quake 2005), it is important that the fundamental mechanisms that underlie transport in these systems be understood. What issues arise in a conned polymer solution? If the connement has the same length scale as the polymer molecules themselves, then the equilibrium conformations of the chains can be expected to change substantially from their bulk behavior. Such a restrictive level of connement allows for the possibility of velocity gradients that vary on the scale of the molecule, leading to changes in the dynamics and conformations of the chain in ow. The boundary conditions on the solvent restrict its motion in ways that may affect the polymer dynamicsrelaxation time, diffusivity, transport processes in ow. Most generally, the presence of even one conning wall breaks the translational invariance of the system, so that situations like simple shear that would not generate inhomogenous polymer concentration elds in bulk ow may lead to concentration gradients in conned ones. This last issue, and specically the possibility of the migration of polymer chains toward or away from walls during ow, has long been recognized as signicant and has been studied in a variety of contexts. Agarwal et al. (1994) has reviewed much of the classical experimental and theoretical literature on this topic, so only a few key aspects of this work are discussed here. First, in simple shear or pressure-driven ow of a polymer solution in which the ow geometry is much larger than the molecular size, chains migrate away from walls at high-enough shear rates. This fact is manifested macroscopically in two ways: rst in an apparent slip boundary condition for the ow, with a slip length substantially larger than the molecular size, and second in an increased average velocity for polymer chains in capillary ow relative to the mean velocity of the solvent, especially at higher molecular weights. Exemplary studies of the latter phenomenon include Seo et al. (1996) and Sugarman (1988). In the former study, polyacrylamide solutions were pumped through a long capillary, and the polymer concentration of the eluent was measured with a UV/Vis detector. At shear rates lower than 103 s1 , the polymer moved at the same velocity as the solvent, but at higher shear rates, the polymer eluted substantially faster, indicating that it had migrated toward the center of the channel where the uid velocity is higher than the average. The latter study was similar in principle, but the polymer used was dsDNA from the -phage virus, a molecule that has

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

274

Graham

become the hydrogen atom of single-molecule polymer physics. In both cases, the wall shear rate at which migration began correlates well with the inverse of the longest relaxation time of the molecule in the solution: That is, migration arises when the Weissenberg number W i = exceeds about unity. The principles and mechanisms underlying these observations are a primary theme of this review. Diffusion of macromolecules in conned geometries has also long been studied (Happel & Brenner 1965, Teraoka 1996). Until relatively recently, however, the regime of a exible polymer diffusing in a conning geometry that substantially distorts the polymer has been inaccessible. Again, studies of genomic DNA in micro- and nanofabricated channels have clearly shown the features of diffusion in this regime. These studies form the other major theme of this review.

Persistence length: length scale beyond which orientations of polymer backbone segments become uncorrelated

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

2. CONFORMATIONS OF POLYMER CHAINS IN BULK AND CONFINED SOLUTIONS 2.1. Bulk Conformations
In this section, we briey review the equilibrium conformations of polymer chains in bulk solution (Bird et al. 1987, de Gennes 1979, Doi & Edwards 1986) and then, as background for understanding the dynamics of conned solutions, describe key results regarding how these conformations change in conned geometries. To begin, we consider a long linear polymer in bulk solution and dene a unit vector u(s) tangent to the molecules backbone, where s is the position along the backbone. In the limit of long chain contour length L, the autocorrelation function u(s ) u(s +s ) of backbone orientation satises a relation of the general form u(s ) u(s + s ) e s /L p , (1)

where Lp is a length scale called the persistence length of the polymer. For example, dsDNA can be viewed (at scales large compared to its diameter of 2 nm) as an inextensible string with a bending moment A, in which case the persistence length is given by the length scale at which bending and thermal energy balance: kT , (2) A where k is Boltzmanns constant and T is absolute temperature (Doi & Edwards 1986). The experimental value of Lp for dsDNA is approximately 50 nm (Shaqfeh 2005). For comparison, the widely studied drag-reducing polymer polyethylene oxide has L p 0.15 nm (Mark & Flory 1. 1965). This review primarily considers exible polymers, with N p = L/L p The backbone of a long exible polymer chain can in principle collide with itself many times. Under so-called -solvent or ideal chain conditions, or in highly concentrated solutions, these selfintersections have a negligible effect on the equilibrium chain conformations and can be ignored. In this situation, the equilibrium spatial conformation of the chain is well approximated on length scales much larger than the persistence length by a classical random walk. In particular, the bulk 1/2 radius of gyration Rbulk for the chain scales as L p N p . In a poor solvent, the chain collapses to a compact conformation; we do not consider poor solvent conditions here. In a good solvent, the situation of primary interest here, the chain segments exhibit an effective repulsive interaction, as they enthalpically prefer to be surrounded by solvent rather than other chain segments. Here the equilibrium chain conformations are self-avoiding random walks and Rbulk L p N p , where 3/5. Polymer concentration (number density of chains) c is often reported relative to the overlap concentration, at which the chains are concentrated enough to begin to contact one another. Lp =
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 275

ELECTROKINETIC PHENOMENA IN DNA TRANSPORT


DNA is an acid, so it is negatively charged in solution, a fact that has long been used for its manipulation via various forms of electrophoresis (Viovy 2000). The dynamics of a charged polymer in electric and/or ow elds can be complex, as the motion of the polymer is coupled with the motion of the other ions in solution. All the experiments described in this review were performed in solutions with sufciently high salt concentration (100 mM) that any electrostatic interactions between segments of the DNA backbone were screened by the counterions, and the chain backbone motions were sufciently slow that the counterions could very rapidly move to stay with the backbone. In this situation, the DNA can be treated as an uncharged polymer. During electrophoresis of DNA in bulk solution, the counterions move in the direction opposite to the chain backbone, and this electroosmosis leads to screening of hydrodynamic interactions between the chain segments. A nontrivial consequence of this fact is that the electrophoretic mobility of DNA in free solution is approximately independent of molecular weight. In nonuniform electric elds or in cases in which the DNA is not fully free to move under the inuence of the eld, the balance of electrophoresis of the chain and electroosmosis of the counterions is broken, and hydrodynamic interactions are no longer fully screened.

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

4 3 This concentration, denoted c , is dened as the value of concentration at which c = ( 3 Rbulk )1 . Unless otherwise specied, we consider only the dilute regime c c .

2.2. Conformations Under Connement


Turning to the issue of connement, and denoting the connement length scale (e.g., the width of a conning slit or tube) as W, three primary regimes can be identied (Figure 1). W. In this weakly conned regime, the equilibrium conformational statistics of 2.2.1. Rbulk the polymer chain are largely unchanged from bulk values. There will nevertheless be a depletion layer of thickness Ld Rbulk near solid surfaces, as it is entropically unfavorable for a polymer chain to have a conformation such that its center of mass is closer than about Rbulk from a wall. During ow, when the chain is extended by a substantial fraction of its contour length L, the ratio L/W may become important (Hernandez-Ortiz et al. 2006b). W Rbulk . In this moderately conned regime, the polymer chain is conned 2.2.2. L p substantially but still behaves as a exible chain. In ideal solvent conditions, the chain would be compressed in the conned directions, but the chain statistics in the unconned directions would be unchanged (Casassa 1967). In the more usual situation of a good solvent, connement of the chain in one or two directions leads to expansion of the chain in the unconned dimensions. Daoud & de Gennes (1977) (see also Hsieh & Doyle 2008, Teraoka 1996) presented predictions of the scaling of chain expansion in the unconned dimensions by generalizing the concept of a blob that had been previously developed in the context of concentrated solutions or polymer melts (de Gennes 1979)loosely speaking, a blob is a subchain of connected segments with correlated behavior. These authors argued that excluded volume interactions would lead to the chain behaving essentially as a string of blobs, each of diameter W, as illustrated by the translucent spheres in the center image of Figure 1. For scales smaller than W, the chains backbone behaves as a simple three-dimensional (3D) self-avoiding random walk as in the bulk, but on larger scales the connement alters this behavior. A blob with diameter W has g (W /L p )5/3 segments and thus there are N b N p /g N p (L p /W )5/3 blobs in the chain. The overall average chain extension
Graham

Blob: term used in polymer physics to represent characteristic length scales intermediate between Lp and Rbulk that arise in many situations, including good solvents, semidilute solutions, and connement

276

Weak confinement

2Rbulk W y x z Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only. Moderate confinement

W 2R W

Strong confinement

W
Figure 1 Regimes of connement for exible polymer chains in solution. In the center image, the translucent spheres indicate blobs of size W of polymer segments.

is then determined by assuming that the chain undergoes a self-avoiding random walk, with step length W in the unconned direction(s). In the quasi-2D case of a slit, the root-mean-squared 3/4 end-to-end distance or radius of gyration Rslit scales as WN b . Nondimensionalizing this result using the bulk chain size, it becomes Rs lit /Rbulk (W /Rbulk )1/4 . (3)

In the quasi-1D case of a tube, the chain cannot execute a random walk on length scales greater than W because it would have to turn back on itself, an energetically highly unfavorable situation. Therefore, the end-to-end distance in this case, Rtube , will simply scale as WN 1 , or b Rtube /Rbulk (W /Rbulk )2/3 . (4)

These scaling results are in good agreement with simulations (Chen et al. 2004, Jendrejack et al. 2003a,b, Wall et al. 1978) and form a starting point for predicting the scaling of dynamic properties such as diffusivity or relaxation time, as discussed below. Rbulk . In this strongly conned case, arguments based on the conformation 2.2.3. W Lp of exible chains break down completely and must be replaced by the direct consideration of the
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 277

i ri

Figure 2 A bead-spring chain model of a exible polymer molecule.

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

backbone of the chain (Odijk 2008, Reisner et al. 2005, Tegenfeldt et al. 2004a). Flow in this regime has not been extensively studied because of the very large pressure drops required to drive ow at such small scales ( 100 nm in the case of dsDNA). So although this case is important for DNA nanotechnology, in which case electrophoresis can be used as an alternate to pressure-driven ow, it is beyond the scope of this review (see Hsieh & Doyle 2008 for a discussion of statics and dynamics in this case).

3. LOWREYNOLDS NUMBER FLOWS DRIVEN BY PARTICLE MOTIONS


The most common coarse-grained model of a exible polymer in a solvent is the bead-spring chain model (Bird et al. 1987, Larson 2005), illustrated schematically in Figure 2. In this model, collections of polymer segments are treated as spherical beads connected by springs representing the coarse-grained entropic forces associated with deformation of the random walk of the chains backbone. Hydrodynamic drag, Brownian, and excluded volume forces (i.e., the repulsions between polymer segments in a good solvent) complete the force balance on each bead of the chain. The ow driven by the motion of each bead affects the motions of other beads, a phenomenon called hydrodynamic interaction. Connement alters this ow and thus the dynamics of polymer chains in ow. Before discussing the recent literature on the dynamics of conned polymers, it is therefore useful to review some fundamental issues associated with particle motions in ow.

3.1. The Mobility Tensor for a System of Point Particles


If we treat bead i of a bead-spring chain as a sphere of radius a that experiences Stokes drag as it moves through the ow, then the relationship between the bead velocity v i and the force it exerts on the uid is given by f i = (v i v(ri )), where is the Stokes law friction coefcient = 6 a, v(ri ) is the uid velocity at the bead position, and is the uid viscosity. (The nite size of the bead only arises in the friction coefcientfor all other purposes it is considered to be a point particle.) There are three contributions to this uid velocity: (a) the externally imposed uid velocity eld v (e.g., a Poiseuille prole), (b) the motion driven by the forces exerted by the other beads in the system, and (c) the correction to the velocity experienced by the bead due to the presence of conning walls. We consider rst a two-bead chain (a dumbbell) in an unbounded domain. Here the force balance (neglecting particle inertia) shows that the uid velocities v(ri ) experienced by bead i are
Graham

Hydrodynamic interaction: the effect that the uid motion generated by one moving object in uid has on other objects in the uid

278

given by v1 v2
1 ( 8 r

1 v (r1 ) + v (r2 ) G (r2 r1 ) 2

G (r1 r2 ) 1

f1 , f2

(5)
T

+ rr/r ) is the Oseen-Burgers tensor, or Stokeslet. Letting V = (v 1 , v 2 ) , where G (r) = and so on, this can be succinctly rewritten: V = V + M F. (6)

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

The tensor M is called the mobility tensor. In simulations of coarse-grained models of polymer dynamics, the singularity of the Stokeslet must be regularizedthe most common choice is the socalled RPY (Rotne-Prager-Yamakawa) tensor (Bird et al. 1987, Rotne & Prager 1969, Yamakawa 1970). Because the Stokeslet changes in a conned geometry, the mobility tensor changes as well. Without loss of generality, the Stokeslet for a conned geometry can be written ( Jendrejack et al. 2003b) as G(ri , r j ) = G (ri r j ) + GW (ri , r j ), and the pair mobility becomes M= + G W (r1 , r1 ) G (r2 r1 ) + GW (r2 , r1 )
1

Long-ranged interaction: interaction between objects that decays sufciently slowly with distance that the cumulative effect of many distant objects is not negligible

(7)

G (r1 r2 ) + GW (r1 , r2 ) . 1 + G W (r2 , r2 )

(8)

The contributions of GW to the diagonal elements of M reect the change in mobility of each bead individually because of the connement, whereas the changes in the off-diagonal elements reect the effect of connement on the hydrodynamic interactions between beads. The latter are the most important in determining the qualitative changes to polymer dynamics brought on by connement.

3.2. Hydrodynamic Interactions in Unconned Stokes Flow Are Long Ranged


An important issue in understanding hydrodynamic connement effects is the nature of hydrodynamic interactions between particles in Stokes ow. Let us consider a suspension of particles, homogeneously distributed in an unbounded d-dimensional domain, so that on large scales the system can be considered to have a constant concentration c. If the particles interact through a eld that decays with distance as rp , then a naive scaling estimate E of the effect on one test particle of all the others is given by E = lim
L a L

c r p r d 1 dr.

(9)

This integral diverges when d p, in which case the interactions are said to be long ranged. This divergence indicates that the number of particles in the shell of size dr a distance r away from the test particle grows faster with r than the interaction between particles decays with r. Because the Stokes velocity eld driven by a system of point forces or dipoles in three dimensions decays with p = 1 or p = 2, respectively, the hydrodynamic interactions among a set of particles in an unbounded uid are long ranged. In polymer solution dynamics, the long-ranged nature of bulk hydrodynamic interactions has the important consequence that the diffusivity Dbulk of a exible chain in solutions obeys 1 Zimm scaling, Dbulk Rbulk (Bird et al. 1987). Hydrodynamic interactions act cooperatively to couple distant chain segments to one another and thus reduce the resistance to motion relative to uncoupled segments. If the hydrodynamic interactions between chain segments were ignored, then
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 279

Screening: the presence of physical effects that act to make nite the range of nominally long-ranged interactions

1 the diffusivity would satisfy Rouse scaling, Dbulk N p . As shown below, the long-ranged nature of hydrodynamic interactions changes substantially under connement, leading to signicant changes in polymer dynamics.

3.3. Properties of Point Force Flows in Bounded Domains


3.3.1. Half-space. We consider rst a point force at position x = (0, y, 0) above a plane noslip wall y = 0. Blake (1971) showed that the exact solution in this case can be written as a Stokeslet plus a simple sum of images located below the wall at position x = (0, y, 0), namely a Stokeslet corresponding to a force of opposite sign, a source dipole and a Stokeslet doublet. (These images make up GW .) Given the solution for a point force above a plane wall, it is straightforward to determine the solution for a point force dipole. As further described below, this solution is particularly important because it plays a dominant role in the dynamics of a deformable particle or macromolecule suspended in a ow near a wall. 3.3.2. Slit. Now we consider the quasi-2D case of ow driven by a point force in a slit with height W. In this case an exact solution, due to Liron & Mochon (1976), is available, but not in a simple form, so here we simply point out some key aspects of the solution (Diamant 2009). For a force perpendicular to the walls, the velocity eld decays exponentially rapidly. For a force parallel to the walls, the velocity eld has a parabolic form in the wall-normal ( y) direction and decays as 1/r 2 . Specically, the far-eld solution has the form of a 2D (or, more precisely, Hele-Shaw) source dipole. This structure has interesting consequences for pair hydrodynamic interactions of colloidal suspensions conned to a slit. In bulk solution, the motion of one particle pulled in the positive x direction, for example, drives a Stokeslet ow that moves all other particles in the positive x direction as well, regardless of their position relative to the pulled particle. In the slit, this is not truesome particles will actually be moved in the negative x direction, an effect dubbed antidrag (Diamant et al. 2005). The nature of the hydrodynamic interactions for a conned polymer in solution is addressed in detail below. 3.3.3. Pore. In a quasi-1D pore geometry (e.g., a channel with circular or square cross section), the ow is conned in two directions, and the uid motion generated by a point force in any direction does decay exponentially at scales larger than W (Diamant 2009).

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

3.4. Screening of Hydrodynamic Interactions by Connement


In bulk solution, hydrodynamic interactions between particles are long ranged. We turn now to the important issue of how this feature of Stokes ow is altered by connement. The simplest case to consider is the quasi-1D situation of particles in a pore, d = 1, where the Stokeslet ow eld decays exponentially on the length scale W. As the interaction decays faster than any power of r in this case, Equation 9 convergeshydrodynamic interactions in the quasi-1D system are not long ranged and are said to be screened on length scales larger than W. The physical origin of screening in this case is that the momentum imparted to the uid by the point force is rapidly absorbed by the stationary walls. The quasi-2D situation is more complex. For a point force in this geometry, p = 2, d = 2 in Equation 9, leading to a logarithmic divergence of E. Nevertheless, as described below, both experiments and simulations of the diffusion of long exible DNA molecule chains in microuidic slits (Chen et al. 2004, Jendrejack et al. 2003a,b) are consistent with screening of hydrodynamic interactions on the scale W. Resolving this seeming paradox requires moving beyond the naive
280 Graham

scaling estimate E to examine more closely the structure of hydrodynamic interactions in a slit. Alvarez & Soto (2005) and Tlusty (2006) noted that at a given distance, the angular average of the quasi-2D Stokeslet vanishes due to symmetry: Therefore, if a particle is suspended in a homogeneous isotropic suspension of particles, the expected value of the far-eld hydrodynamic force on it due to the motions of the other particles vanishes. This screening due to cancellation in the angular averaging does not arise in three dimensions. Tlusty (2006) calls this result screening by symmetry.

3.5. Pair Interactions and Force Dipoles Near a Wall: The Migration Tensor
We consider a pair of suspended particles of radius a both a distance y a above a plane wall and a distance d a from one another. Imagine that particle 2 exerts a central force f1 on particle 1 and vice versa. (These particles could make up a bead-spring dumbbell.) The force balance on particle 1 requires that it exert a force f1 on the uidthis force will drive a uid motion that will tend to move particle 2 both parallel to the wall and, more importantly, normal to it. Similarly, the uid motion generated by particle 2 will also move particle 1 normal to the wall. If the particles are attracted to one another, each will generate a uid motion that convects the other away from the wall, whereas if they repel one another, each will move toward the wall. This result was directly observed in suspensions of charged colloidal particles (Dufresne et al. 2000) and can be derived (Anekal & Bevan 2005, Dufresne et al. 2000) using the single-wall Greens function of Blake (1971). One can idealize the situation of a pair of interacting particles or indeed any single force- and torque-free deformable particle (polymer chain, droplet, capsule, biological cell) as a symmetric point force dipole D. Using the point force solution of Blake, it is straightforward to determine the velocity v mig at which the dipole is convected relative to the wall due to the uid motion that it generates. The result, rst presented by Smart & Leighton (1991), is v mig = M : D. (10)

Migration tensor: relationship between the force dipole generated by an object in uid and the velocity at which that object is convected by the ow generated by the dipole

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

The third-order tensor M is called the migration tensor; it can be found for any ow domain given the point force dipole solution in that domain. For a half-space domain with a no-slip wall at y = 0, it is given by M= 1 M, y2 (11)

where M is a constant tensor. The 1/y 2 decay of this function arises from the dipolar nature of the particle-wall hydrodynamic interaction.

3.6. Equations of Motion for a Coarse-Grained Flexible Polymer Molecule in Solution


Although it is not the purpose of the present review to describe bead-spring chain models in any detail (e.g., see Bird et al. 1987, Ottinger 1996, Larson 2005), it will be useful to briey present the basic mathematical structure of these models. Following the notation introduced above for the pair mobility, we consider a system of N beads each with position ri and dene R = (r1 , r2 , . . . , r N )T . Similarly, we let F = ( f 1 , f 2 , . . . , f N )T , where fi is the sum of the external (e.g., spring, excluded volume, gravitational) forces exerted on bead i. In the limit of negligible uid and particle inertia, the balance of hydrodynamic, Brownian, and external forces on each bead can be written as the
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 281

following stochastic differential equation (Ottinger 1996):


Brownian dynamics simulation: numerical time integration of a stochastic differential equation (e.g., Equations 12 and 13)

d R = V + M F + kT where

M d t + B d W, R

(12)

B B T = 2kT M,

(13)

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

and each element of dW is an increment of an independent Wiener process. Equation 13 is a statement of the uctuation-dissipation theorem for this system. In most of the simulation results discussed below, these are the equations that are solved via so-called Brownian dynamics (BD) simulation. (In some contexts, the term BD-HI is used to denote Brownian dynamics simulations that incorporate hydrodynamic interactions (HI), i.e., the use of the full mobility tensor, whereas BD is used to denote simulations that neglect hydrodynamic interactions. In the present work, the term Brownian dynamics implies the use of the full mobility tensor.) Some computational efciency issues and alternate methods for simulating the evolution of coarse-grained polymer models are discussed in Section 4.2.

4. DYNAMICS OF CONFINED FLOWING POLYMER SOLUTIONS 4.1. Results from Experiment, Theory, and Computation
In this section, we indicate how the recent literature weaves together the issues of polymer conformations described in Section 2.2 with those of hydrodynamics in conned geometries described in Section 3. Brochard & de Gennes (1977) described scaling results for the diffusivity D and relaxation time of a exible polymer conned to a slit or tube (i.e., quasi-2D or quasi-1D connement) in the Rg W . These results are based on an argument that hydrodynamic interactions are case L p screened at distances greater than W and that under these conditions the chain friction coefcient should scale linearly with molecular weight (Rouse scaling). Using this argument and Equation 4, and dening D = D/Dbulk and W = W /Rbulk , they predict that D W 2/3 . (14)

Although the assumption of screening is correct, as discussed below, the issue is more subtle than indicated by this paper. As mentioned above, in the slit case, screening actually occurs because of the symmetry of the hydrodynamic interactions. With regard to relaxation time, this review does not go into detail, for two reasons. First, the effects of connement on relaxation time are closely related to those on diffusion, for which more detailed measurements exist. Second, although many measurements are available in the literature, these are based on many different, although related, denitions of relaxation time. The reader is referred to Bakajin et al. (1998), Reisner et al. (2005), Hsieh et al. (2007), and Bonthuis et al. (2008) for measurements and discussions of relaxation time. Diffusion in the quasi-1D moderately conned case was further considered by Harden & Doi (1992). They performed a self-consistent mean eld theory calculation to predict the distribution of polymer segments in a tube. Then they combined this prediction with the exact solution for the Greens function due to a point force in a cylindrical tube and used the so-called Kirkwood approximation (Bird et al. 1987) for the chain diffusivity to predict its dependence on the degree of connement. Over the range of their computations, the results were well approximated by D W 0.61 , rather than the exponent of 2/3 found from the above scaling arguments. The authors argue that the discrepancy between their results and the simple scaling arises because of
282 Graham

the depletion of polymer segments in a region of size Lp near the tube walls. This argument is corroborated by simulation results ( Jendrejack et al. 2003b) and experimental results (Balducci et al. 2006, Hsieh et al. 2007), as further discussed below. Brunn and Grisa (Brunn 1984, 1985; Brunn & Grisa 1985) developed a polymer kinetic theory for bead-spring chain models in inhomogeneous ows, including hydrodynamic interactions between beads, but not accounting for the change in hydrodynamic interactions due to the presence of a wall. They note that in inhomogeneous ows, the polymer diffusivity can vary with position, because its degree of stretching varies with position, and that this variation in diffusivity can lead to cross-stream migration. Let us consider Poiseuille ow in the case of weak connement. At the channel center there is no deformation of polymer chains, so they retain their equilibrium diffusivity. With increasing distance from the center, the strain rate and thus the polymer stretching increase. Polymer diffusivity decreases as stretch increases (de Gennes 1979), and Brunns theory predicts therefore that chains should migrate from the region of highest diffusivity toward regions of lower diffusivity, i.e., away from the channel center during ow, in seeming contrast with experimental observations. We return to the issue of migration away from the centerline in pressure-driven ow in the discussion of more recent results below. These kinetic-theory results for Poiseuille ow have been extended to more complex beadspring chain models (Nitsche 1996) and to Brownian rigid rods (Nitsche & Hinch 1997, Schiek & Shaqfeh 1997). These studies again predict migration toward higher gradients (i.e., toward the wall) in Poiseuille ow, for the same reason as Brunns. Woo et al. (2004a,b) modeled the effect of connement on the dynamics of chains by considering how the entropic spring force in a bead-spring chain model would be modied. Chain-wall hydrodynamic interactions were taken into account by introducing a position-dependent mobility of each bead, an approximation that is appropriate in the limit of moderate to strong connement, where hydrodynamic interactions are screened. Chain migration per se was not studied, but predictions of chain conformational dynamics and rheology were made, many of which remain to be tested by direct simulation or experiment. In related work, Jhon & Freed (1985) presented a formalism for studying the migration of a polymer chain in ow near a solid surface that does in principle incorporate hydrodynamic interactions between the chain and a single wall. They predict that a chain migrates away from a wall in shear, but it should be noted that their equation 11a, for the Greens function due to a point force above a plane wall, is incorrect. The migration tensor evaluated using this Greens function would actually predict migration toward the wall. In the context of direct solutions of polymer solution dynamics, wall hydrodynamic effects were rst introduced in a series of papers by Jendrejack et al. (2003a,b, 2004). These papers studied the dynamics in the dilute limit of bead-spring chain models of genomic DNA ( Jendrejack et al. 2002) of contour lengths from 4 to 420 m in square microchannels of various sizes, from W Rg to W Rg . (For -phage DNA stained with uorescent dye, Rbulk 0.7 m and L 21 m.) The Greens function for a point force in this geometry can be calculated exactly as an innite series, but for performing computations, Jendrejack et al. took a different approach. They wrote the Greens function as a free-space piece plus a correction such that the sum of the two pieces satised no slip on the walls and periodicity at the channel entrance and exit. The correction was determined numerically using a standard nite-element method and stored as a look-up table. Regularization of the Stokeslet at short distances was achieved with an ad hoc generalization of the RPY tensor. The Brownian displacement B dW was determined using the polynomial approximation method proposed by Fixman (1986) (see also Jendrejack et al. 2000). Several important sets of results emerged from these simulations. The rst, described in Jendrejack et al. (2003b), is an explicit prediction of the diffusivity of a exible chain as a
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules

-phage DNA: the genomic DNA of the -phage virus; when stained with uorescent dye for visualization, it is approximately 21 m long

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

283

5 0 0

5 0 0

z (m) Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.
5 5

y (m)

z (m)
5 5

y (m)

Figure 3 Predicted steady-state center-of-mass probability (concentration) distribution versus cross-sectional chain position for dimers of -DNA in a square channel with W = 10.6 m (W 10). (a) Equilibrium distribution. (b) Distribution at = 308 s1 (Wi = 92). Figure reprinted with permission from Jendrejack et al. (2003a). Copyright 2003 by the American Physical Society.

function of the degree of connement. The crossover from bulk to conned behavior was found to begin at W 10, and in the region 1/W 1 a power-law behavior D W 0.5 was found. This exponent is different from the value of 2/3 predicted by Brochard & de Gennes (1977). Consistent with the work of Harden & Doi (1992), the authors attribute the deviation to the fact that there is not a wide scale separation between the channel size and the equilibrium distance between beads in the model (approximately 4 persistence lengths). On the other hand, results for the equilibrium stretch closely followed the W 2/3 prediction of Daoud & de Gennes (1977) once W 1 5. Turning from dynamics at equilibrium to dynamics in ow, Jendrejack et al. (2003a, 2004) examined pressure-driven ow of DNA solutions in square microchannels over roughly the same range used for the diffusion simulations. One of the primary results of this study is the prediction in the weakly conned regime of a depletion layer whose size at high Wi is much larger than the chain radius of gyration (Figure 3). The degree of migration increases with increasing Weissenberg number, consistent with experimental observations that DNA (and other polymers) can be separated by pressure-driven ow through a capillary (Seo et al. 1996, Sugarman 1988). A distinctive feature of the predicted concentration proles in the square channel is their volcanolike shape, as seen in Figure 3: The concentration is essentially zero near the walls, rises rapidly toward the center, but then displays a slight dip in the region very near the centerline. The migration away from the wall is driven by the wall contribution to the hydrodynamic interactions, and the slight migration away from the centerline results from the phenomenon predicted by Brunn, in which chains migrate to regions of lower mobility. The volcano shape is also found by Saintillan et al. (2006) in their simulations of chains of freely jointed Brownian rods in a slit. To account for rod-wall hydrodynamic interactions, these authors used a regularized form of Liron & Mochons (1976) Greens function for a slit that was introduced by Staben et al. (2003). This work also revisited the limiting case of a polymer consisting of a single rigid Brownian rod (see Nitsche & Hinch 1997 and Schiek & Shaqfeh 1997, discussed above), demonstrating that the rods actually migrate away from the walls, albeit weakly. Park et al. (2007) performed a similar analysis, arriving at the same result.
284 Graham

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

In the moderately conned regime, substantial migration does not occur, although other interesting phenomena are found. Section 2.2 above discusses the concept of blobs of chain segments that are of size W, independent of molecular weight. In their simulations, Jendrejack et al. (2004) found that in this regime, the wall shear rate at which the polymer chains start to stretch is also independent of molecular weight: Polymer stretching begins when the Weissenberg number for a blob of size W becomes of order unity. Presently there is not an explicit experimental test of this prediction, although, as noted by Hsieh et al. (2007), there are data that are consistent with it. Stein et al. (2006) presented images of -phage DNA during pressure-driven ow in a slit with height 250 nm (approximately Rbulk /3) at a wall shear rate of approximately 90 s1 . In unbounded ow at this shear rate, chains would be stretched, but no stretching was observed in their experiment. Similarly, Larson et al. (2006) reported stretch versus mean velocity for BAC12m9 DNA (which is about four times longer than -phage) during pressure-driven ow in a 1-m-high slit microchannel. From this data one can estimate that there is not substantial chain stretching until a strain rate of approximately 1,000 s1 , again much larger than would be expected for this molecule in unbounded shear ow. Complementary to the simulations just described, Jendrejack et al. (2004) and Ma & Graham (2005) revisited the dilute solution kinetic theory for bead-spring dumbbell models of polymers in solution, incorporating both intrachain and chain-wall hydrodynamic interactions. For the case of a dumbbell (i.e., a two-bead chain) in solution in a semi-innite domain bounded by a no-slip wall, substantial analytical progress can be made. In particular, in the limit where the dumbbell size is much smaller than its distance from the wall, a closed form expression for the ux j of polymer chains can be found. This expression depends on the polymer number density c, bulk uid velocity eld v , polymer conformation tensor = qq (where q = r2 r1 is the dumbbell end-to-end vector and denotes ensemble averaging over q), polymer stress tensor , and (conformation-dependent) Kirkwood diffusity evaluated in an unbounded domain DK ,bulk . The expression is j = c v + c qq : v + M : p 8 c DK , bulk DK , bulk c .

(15)

The rst term in this expression is convection, and the last term is normal Fickian diffusion; the other terms lead to migration. Let us consider rst the term containing the migration tensor and the stress tensor. Each dumbbell induces a force dipole ow in the surrounding solventthe stress tensor is the ensemble average of this dipole. In the presence of a wall, the force dipole induces a uid velocity M : /c at the position of the dumbbell (Smart & Leighton 1991). This term was missing in previous theories of polymer migration. The term containing the divergence of DK ,bulk is the effect noted by Brunn (1984). In nonhomogeneous ow, the term containing v predicts the lag of a macromolecule behind the solvent along the streamline (Aubert & Tirrell 1980) but no cross-streamline migration in rectilinear ows, and unless the velocity nonhomogeneity is so large that it cannot be ignored even on the length scale of the polymer molecule, this term is small. [Nevertheless, it may be important in curvilinear ows that are maintained for very long times (see, e.g., Macdonald & Muller 1996).] It is important to note that substantial wall-induced hydrodynamic migration also arises in suspensions of deformable particles such as droplets or capsules (Leal 1980, Olla 1999). Hudson (2003) has written an expression similar to Equation 15 for the transport of emulsion drops that undergo migration and shear-induced rather than Brownian diffusion. With Equation 15, it is possible to make some explicit predictions in various special cases. The simplest of these is an explicit prediction of the steady-state concentration prole c( y) in simple
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 285

shear ow above a plane wall: c (y) = c b exp (Ld /y), (16) where cb is the bulk concentration and Ld is the depletion-layer thickness. For a general dumbbell model, and neglecting the conformation dependency of diffusivity, 9 N1 N2 h Rbulk , Ld = (17) 128 c kT where h 0.25 is the hydrodynamic interaction parameter for the dumbbell, and N1 and N2 are the rst and second normal stress differences (Bird et al. 1987); these are both linear in c so it cancels out of the expression. For the FENE (nitely extensible, nonlinearly elastic) dumbbell model, N 2 = 0 and N 1 c W i 2/3 , yielding a distinct scaling prediction from this model of Ld W i 2/3 Rg .
Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

(18)

When W i 1, depletion layers can be much larger than the equilibrium molecular size. With this expression, Ma & Graham (2005) also made predictions of the spatial development of the depletion layer (the migration analog of the Graetz-Leveque problem) as well as the temporal development upon start-up of steady shear. These predictions are both in the form of similarity solutions and share two important features. The rst is a transient pile-up of concentration in the boundary layerthe concentration prole is nonmonotonic because rapidly migrating chains near the wall tend to catch up with more slowly migrating chains initially farther from the wall. The second is that the time (meaning residence time in the spatially developing case) for the development of the steady-state depletion layer scales as the time L2 /Dbulk required for chains to d diffuse over a length scale Ld . For the spatially developing case, and using the scaling estimate from the FENE dumbbell model, this implies an entrance length that scales as Wi3 Rg . Many extensions of the framework established in Ma & Graham (2005) have been made. Hernandez-Ortiz et al. (2006a,b) showed that in the single-wall case, Brownian dynamics simulations with dumbbell and chain models were in good agreement with the analytical theory. In a slit geometry, the analytical theory can be extended by using a single-reection approximation, in which the migration effects due to the two walls of the slit are simply summed, ignoring the fact that this summation leads to violation of the no-slip boundary condition. For moderately conned chains at very high Wi, weak migration toward the wall can occur because in this case the hydrodynamic migration effects of the two walls essentially cancel, and the chain, which is stretched in the ow direction, is actually slightly compacted in the wall-normal direction, allowing its center of mass to come closer to the wall than it can at equilibrium (see de Pablo et al. 1992). Butler et al. (2007) generalized Ma & Grahams (2005) theory to allow for the presence of an external force on the dumbbells. A particularly interesting prediction is that the combination of shear ow and an external eld can lead to migration toward the wall, an effect that competes with the wall-induced hydrodynamic migration. This effect arises from the following mechanism: The shear ow causes the polymer to be tilted on average with respect to the ow direction. If the external force on the polymer drives it in the direction opposite to the ow, the anisotropy of the hydrodynamic drag on the stretched chain tends to drive transverse migration of the chain toward the region of lower velocity, i.e., toward the wall. The anisotropy of drag results from the intrachain hydrodynamic interactions. This result is consistent with the experimental observations of Zheng & Yeung (2002, 2003) under combined electric and ow elds. A closely related work is that of Hoda & Kumar (2007), who study the interaction between hydrodynamic migration and electrostatic attraction for a polyelectrolyte owing near an oppositely charged surface. This study also removed the far-eld approximation for the chain-wall hydrodynamic interactions used by Ma & Graham (2005), an important feature when considering adsorption. The authors predicted
286 Graham

that the competition between electrostatic attraction and hydrodynamic repulsion could lead to a steady-state concentration prole with a maximum, although it is unclear whether the parameter regime in which this phenomenon occurs is experimentally accessible. (In most experiments with genomic DNA, for example, the Debye layer thickness is 1 nm, which is much smaller than the bulk equilibrium size of the chain and thus much smaller than the scale at which a dumbbell model is expected to be valid.) Sendner & Netz (2008) consider the dynamics of dumbbell models with nonzero equilibrium length as models of semiexible polymers. In particular, they extend the scaling result given in Equation 18 to the semiexible case and corroborate the scaling prediction with Brownian dynamics simulations. The entrance length prediction of Ma & Graham (2005) implies that a long channel may be required for full development of the hydrodynamic depletion layer in steady ow. Chen et al. (2005) and Jo et al. (2009) circumvented this issue by considering a zero-mean oscillatory pressure-driven ow. In the rst half-cycle of the oscillation, a constant ow rate is imposed, with the direction of ow changing sign in the second half-cycle. In this case there is no net bulk ow, and the development of the depletion layer can be observed at a single position in the microchannel. This system is characterized by both a Weissenberg number W i = | | and the strain in each half-cycle = / f , where f is the inverse period of the oscillations. Substantial migration only occurs at high Wi and high . For genomic DNA with radii of gyration 1 m, depletion-layer thicknesses of 10 m or more were predicted with Brownian dynamics simulations at high Wi and . As shown in Figure 4, the predictions were in reasonable agreement with experimental results, although the simulations slightly overpredicted the depletion-layer thicknesses and the time (2 3 min) required to reach steady state. Both the simulation and experimental studies showed the existence of off-center peaks in the concentration prole. It is unclear whether this feature is more closely related to the pile-up prediction of the boundary layer theory treatment of Ma & Graham (2005) or the steady-state volcano-peak result of Jendrejack et al. (2003a, 2004). Fang et al. (2005) reported uorescence microscopy measurements of the conguration and concentration of DNA molecules in simple shear ow near a surface in a microchannel with W Rbulk . They reported depletion layers as thick as approximately 10 m for -phage DNA (which has Rbulk 700 nm) at large Wi. This study was extended by Fang et al. (2007), who presented more extensive experimental measurements of depletion-layer thicknesses for dilute solutions of -phage DNA, as well as Brownian dynamics simulations of chains near a single wall [the latter are similar to the single-wall simulations of Hernandez-Ortiz et al. (2006b)]. The experimental results displayed qualitative but not quantitative agreement with the dumbbell theory predictions of Ma & Graham (2005); the experimentally observed depletion-layer thickness was substantially thinner than the prediction. Bead-spring chain simulations were in better agreement with the experimental results, only overpredicting the depletion layer thickness by about a factor of two, but overpredicting the timescale for steady state to be reached by an order of magnitude. These authors concluded that more rened polymer models were necessary to obtain more quantitative agreement with data. Finally, Fang & Larson (2007) examined the concentration dependency of the depletion layer in DNA solutions, nding that as the concentration exceeds 0.1 of the overlap concentration c for the chains, the hydrodynamic migration effect begins to diminish, although it does not vanish until the concentration exceeds approximately 3c . This result is qualitatively consistent with the computational results of Hernandez-Ortiz et al. (2007, 2008) (further described below) as well as with the observation that, in bulk solution, hydrodynamic interactions begin to be screened at concentrations above approximately 0.1 c . As described above, both scaling and computational predictions have been made regarding the diffusion of chains in the moderately conned regime. Chen et al. (2004) reported measurements and comparisons to simulations for -phage DNA and two- and threefold concatemers thereof.
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 287

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

20

Concentration/ average concentration


3 2

10

y (m)

1 0

10

20

100

t (s)

200

300

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

20

Concentration/ average concentration


2

10

y (m)

0 1

10

20

0 0 100

t (s)

200

300

Figure 4 (a) Simulation and (b) experimental results for the time evolution of the concentration distribution of T2-DNA molecules in a 40-m square microchannel and subjected to oscillatory pressure-driven ow with f = 0.25 Hz. Experimental results are at W i 60 and computational results at W i = 50. Brightness is proportional to concentration. Experimental results at each time instant are averaged over the axial position in the channel. Computational results are averaged over 100 simulations with random initial positions for the molecules. Figure adapted with permission from Chen et al. (2005). Copyright 2005 American Chemical Society.

The experiments in this work were performed in a slit geometry and were in good agreement with the simulation results and reasonable agreement with the scaling theory. The simulations examined both slit and square cross sections, yielding good agreement with the static scaling predictions of Daoud & de Gennes (1977) for chain extension in both geometries. The simulation results for the slit geometry were closer to the prediction of a 2/3 power-law exponent than were those for the square channel (as discussed above), but some deviation still remained. Rough agreement with the 2/3 exponent for conned diffusion was also experimentally found with DNA by Stein et al. (2006). The experimental aspect of Chen et al.s (2004) work was extended in Balducci et al. (2006) and Hsieh et al. (2007). The former paper reports experimental observations of diffusion of DNA in slit channels in the moderate and strong connement regimes. Figure 5a shows experimental results for diffusion as a function of degree of connement, as well as some of the simulation results presented in Chen et al. (2004)the good agreement between experiment and simulation is apparent. Furthermore, they demonstrated, as shown in Figure 5b, that once the slit height is 1 even slightly smaller than the bulk radius of gyration, diffusion follows Rouse scaling (D N p )
288 Graham

1
8 7 6 5 4

Blob scaling (2/3)

D/Dbulk

M13mp18 DNA -DNA -DNA 2 -DNA Simulation (Chen et al. 2004) Experiment (Chen et al. 2004)

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

0.1
8
8

0.1

Rg,bulk/h

10

b
1

8 6 4

Zimm scaling (0.6)

D (m2 s1)
0.1

8 6 4

W= 545 nm 280 nm 190 nm 100 nm 50 nm


2 3 4 5 6 7

Rouse scaling (1.0)


8 9 2

MW/MW,-DNA
Figure 5

(a) Experimental and simulation results for the diffusion of DNA in a slit microchannel. (b) Diffusivity as a function of molecular weight, relative to -DNA for DNA in slit microchannels of various heights. For all of the conned cases, the diffusivity closely follows Rouse scaling. Figure reprinted with permission from Balducci et al. (2006). Copyright 2006 American Chemical Society.

closely, indicating that hydrodynamic interactions are screened over the coil size of the polymer. On the other hand, as noted above, the experimental results for D/Dbulk versus Rbulk /W deviate from the 2/3 exponent. These results demonstrate that for the DNA molecular weights under consideration here, there is not sufcient scale separation between the chain persistence length and degree of connement to justify the assumption in the scaling theory that the polymer chains W Rbulk is not satised. This work form well-developed blobsthe double condition L p
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 289

also generalizes the observation by Alvarez & Soto (2005) and Tlusty (2006) of screening by symmetry. Hsieh et al. (2007) also reported measurements of rotational relaxation time for DNA in a slit channel and corroborate the observation that, in the experimental regime studied, scale separation is insufcient for predictions based on a blob model to strictly hold.

4.2. Complex Geometries, Finite Concentrations, and Computational Issues


Returning to the issue of simulation approaches for polymer solutions in conned geometries, it is important to consider the computational expense of the Stokes ow-based approaches described above. Direct construction and matrix multiplication of M both cost O(N 2 ) operations per time step. Exact determination of B costs O(N 3 ) operations, although the Chebyshev polynomial approximation introduced by Fixman (1986) reduces this cost so that it scales linearly with the cost of a matrix-vector multiplication. In periodic domains, however, these scalings can be dramatically improved by use of Ewald-sum-based and P3 M (particle-particle-particle-mesh) methods, which use fast Fourier transforms (Deserno & Holm 1998a,b, Hasimoto 1959, Hockney & Eastwood 1988, Smith et al. 1987, Toukmaji & Board 1996), and one can compute these interactions in periodic domains in O(N ln N) operations. The Brownian term can then also be approximated in O(N ln N) operations using the Fixman method. (Strictly speaking, these scalings refer to the number of mesh points used for spatial discretization, but for many situations of interest, this is proportional to N.) In the context of suspensions of rigid Brownian particles, this approach is embodied in the accelerated Stokesian dynamics method of Banchio & Brady (2003) and Sierou & Brady (2001). For studying connement effects, these Fourier-transform-based methods are not directly applicable [although boundaries can be built in articially, as done, for example, by Freund (2007) in a study of wall effects on blood ow]. Nevertheless, the idea used in those methods of splitting a solution to the Stokes equation into singular, short-ranged parts and smooth long-ranged parts remains useful even for nonperiodic domains. This observation is used in the general geometry Ewald-like method of Hernandez-Ortiz et al. (2007), which has O(N) or O(N ln N) scaling. With this approach, Hernandez-Ortiz et al. (2008) studied the effects of concentration on migration during Poiseuille or Couette ow in a slit geometry with W 10Rbulk , nding that as the concentration increases toward the overlap concentration, the depletion-layer thickness decreases, with migration vanishing for c 0.2c . This agrees qualitatively but not quantitatively with the experimental observations of Fang & Larson (2007), although the specics of both the polymer and connement were substantially different in the two cases. This study also described a simulation of a solution at nite concentration in a more complex geometry. Flow over a channel with grooves oriented perpendicular to the ow direction was considered. As shown in Figure 6a, at low concentration and high Wi, the groove was almost completely depleted of chains, and this observation was explained by arguing that once chains diffuse out of the groove into the main ow, they can only move back into the groove by diffusing across the hydrodynamic depletion layer in a time that is less than the time it takes for the chain to cross the mouth of the groove, an unlikely event. Interestingly, at concentrations approaching overlap, the concentration difference between bulk and groove is substantially reduced to only about a factor of two, but only if hydrodynamic interactions are included in the simulation. An argument is made that at higher concentration, for the main ow to drag one chain out of the groove it has to drag many, because of the hydrodynamic coupling between the chains in the groove. Because of this mechanism, and because the mobility of chains in the groove is reduced by the higher degree of connement there, chains are less susceptible to being dragged into the bulk ow. These predictions and the mechanisms used to explain them are experimentally untested,
290 Graham

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Concentration/ maximum concentration

a
x3/(kB T/H)1/2

10

1 0.8 0.6

0 0.4 5 0.2 0

10 20

10

x1/(kB

T/H)1/2

10

20

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Concentration/ maximum concentration

b
x3/(kB T/H)1/2

10

1 0.8 0.6

0 0.4 5 0.2 0 10

10 20

x1/(kB

T/H)1/2

10

20

Figure 6 Predicted concentration proles, normalized with maximum concentration (white), for ow of a polymer solution over a grooved wall. The top wall moves to the right while the bottom (grooved) wall is stationary. (a) Prediction for dilute solution. (b) Prediction for c /c = 0.12. Figure adapted with permission from gures 6 and 8 of Hernandez-Ortiz et al. (2008). Copyright 2008 by the Korea-Australia Rheology Journal.

although they are not inconsistent with experimental observations that above c , concentrations in side channels can actually be higher than in the bulk (Agarwal et al. 1994). All the computational and theoretical studies described above have been based on a Stokes ow treatment of the solvent motion. In recent years a number of other approaches to the treatment of the solvent dynamics have gained substantial attention, either as potentially more efcient alternatives to Stokes ow simulation methods (but see below) or as treatments that capture more of the small-scale physics of the solvent than a continuum approach can. The lattice Boltzmann method (Dunweg & Ladd 2009) discretizes the Boltzmann equation rather than the Stokes or Navier-Stokes equation, nominally resulting in a highly efcient, parallelizable, method. Usta et al. (2005, 2006) used this method to study polymer solutions in the weakly and moderately conned regimes, nding a roughly 2/3 scaling exponent for diffusion as a function of connement, as well as hydrodynamic migration and the characteristic volcano-peak structure described above for Poiseuille ow in a slit. The stochastic rotation dynamics method (Malevanets & Kapral 1999) introduces particles that have articial collision dynamics at short times but whose behavior at long times reproduces uctuating hydrodynamics. Watari et al. (2007) used this method to study bulk and conned polymer solution dynamics, qualitatively reproducing the hydrodynamic
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules 291

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

migration phenomenon. The dissipative particle dynamics (DPD) method (Espanol & Warren 1995, Groot & Warren 1997, Ripoll et al. 2001) explicitly solves the equations of motion for coarse-grained solvent particles that have both conservative and dissipative interactions, to model the dynamics of clusters of solvent molecules rather than individual ones. Early work treating conned polymers with this approach (Fan et al. 2003) was unable to predict the existence of hydrodynamic migration, perhaps because of the nite Reynolds number of those simulations (see Hernandez-Ortiz et al. 2006b for a brief discussion of niteReynolds number effects on migration). More recent work (Fedosov et al. 2008) has examined the issue of migration in DPD simulations in some detail. At Reynolds numbers larger than 1, this study predicts depletion at the centerline but no hydrodynamic migration from the wall. At Reynolds numbers smaller than 1, there is some migration away from the wall but no migration away from the centerline (this simulation was in a very small channel, W = 3Rbulk ). The migration toward the centerline in the higherReynolds number case was attributed to the Segre-Silberberg effect (Segre & Silberberg 1962a,b), in which even rigid particles in pressure-driven ow at nite Reynolds number move to an equilibrium position that is off the channel centerline. The mobility gradient effect described above may also be active, but it is unclear which is dominant in this particular system. Recent work (Millan & Laradji 2009) was able to qualitatively capture both the wall and mobility gradient effects, producing the expected volcano-peak structure. Finally, Khare et al. (2006) and Kohale & Khare (2009) have performed direct molecular dynamics simulations of polymer solutions with an atomistic solvent in a slit. Their results demonstrate the persistence of the hydrodynamic migration and chain mobility gradient contributions to cross-stream transport down to very small scales and also highlight the potential importance of thermal diffusion (the Soret effect) in molecular simulations and in experimental nanoscale systems. Finally, we make a further comment on computation time: For lattice Boltzmann, the computation time scales linearly with the number of lattice pointsthis is the same scaling [to within O(ln N)] as the general geometry Ewald-like method and the other accelerated Stokes ow approaches described above. Stochastic rotation dynamics and DPD scale linearly with the number of solvent particles, a situation analogous to a continuum method that scales linearly with the number of mesh points. Thus at this stage in the development of computational methods for conned polymer solutions or suspensions, all current methods lead to linear scaling with problem size. The more relevant consideration, therefore, is the extent to which a given method can efciently capture the time and length scales of the important phenomena. This is a nontrivial issue even for lattice Boltzmann (Pham et al. 2009), but especially for stochastic rotation dynamics, DPD, and molecular dynamics, which are trying to capture Navier-Stokes dynamics with a method that is inherently limited to time steps characteristic of the relaxation processes of the solvent molecules, which are much shorter than the relaxation or diffusion time of a suspended particle or polymer chain.

5. CONCLUSIONS
The past decade has seen dramatic advances in the understanding of the single-molecule dynamics of conned dilute polymer solutions during owat least in simple unidirectional ow, the primary phenomena at work seem to have been elucidated. Single-molecule and dilute solution experiments with uorescently stained DNA have enabled the direct observation of conned polymer chains, and theoretical and computational methods are now able to make detailed predictions of polymer behavior in conned ow. There are some cases in which quantitative agreement between experiments and models can be obtainedfor example, in the case of diffusion in a slitbut in other cases, such as cross-stream migration, agreement between experiment and simulations is semiquantitative at best.
292 Graham

Many challenges remain in gaining a predictive understanding of transport in conned polymer solutions. Experiments (Del Bonis-ODonnell et al. 2009) and simulations (Hernandez-Ortiz et al. 2008) indicate intriguing and potentially technologically important phenomena during ow of polymer solutions in complex geometries. Chains tethered to surfaces display complex dynamics in ow (Beck & Shaqfeh 2006), and the combination of tethering and connement introduces new dynamics that would be absent otherwise (Zhang et al. 2009). All these cases admit only limited understanding at this time. We also do not understand in any detail how the migration and diffusion phenomena discussed above depend on concentration, particularly in complex geometries, nor how they are altered in mixtures. Finally, an important transport topic, especially for DNA nanotechnology ( Jo et al. 2007), is that of electrokinetic effects; DNA is a charged molecule that resides in an ionic solution. The tools that led in the past decade to the understanding of singlemolecule conned dynamics during ow provide starting points for the coming decades work in addressing these more complex issues.
Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

SUMMARY POINTS 1. The transport of dissolved polymer molecules during ow in conned geometries arises in many applications, from oil recovery to DNA nanotechnology, and displays many complexities including hindered diffusion and cross-stream migration. Direct visualization of uorescently stained genomic DNA has allowed the direct observation of these phenomena. 2. For polymers in dilute solution, the primary focus of this review, three general regimes of connement can be identied, based on the relative length scales of the molecule and the conning geometry. For a polymer chain with persistence length Lp and bulk radius of gyration Rbulk in a domain with characteristic scale W, these regimes are weak W ; moderate connement, where L p W Rbulk ; and connement, where Rbulk Rbulk . strong connement, where L p W 3. Connement changes the uid dynamics of suspended particles or macromolecules in three important ways: (a) Hydrodynamic interactions are long ranged in bulk solutions and are screened in conned geometries, as walls absorb the momentum imparted to uid by a moving object in the ow; (b) walls modify the ow driven by moving objects and generically lead to cross-stream migration of deformable particles or macromolecules; and (c) cross-stream migration can also occur due to Brownian diffusion in inhomogeneous ows in which the conformations and thus diffusivity of polymer molecules vary with position. In general this effect competes with the migration driven by the presence of walls. 4. Moderate or strong connement changes the equilibrium conformations of exible polymer molecules; in the moderately conned regime, relatively simple physical arguments lead to predictions of the scalings of equilibrium polymer shape and size that agree well with experiments and simulations. 5. Combining the scaling arguments for static polymer conformation under connement with the argument that hydrodynamic interactions are screened at scales larger than W leads to predictions for the scaling of polymer diffusivity with molecular weight that agree well with experiments on DNA. Predictions of scaling with the degree of connement at xed molecular weight agree somewhat less well because the separation of scales argument W invoked in the scaling predictions is not satised in the experimental systems. Lp

www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules

293

6. Direct simulations of bead-spring chain models of exible polymers in dilute solution can now efciently incorporate connement effects on ow, predicting the existence of hydrodynamic depletion layers that can be much larger than the equilibrium molecular size. These predictions are in qualitative and, to some extent, quantitative agreement with experiment, although discrepancies still exist. Predictions from simulation of diffusion in conned geometries agree well with experimental results for DNA in the moderately conned regimes. 7. For dilute polymer solutions in the weakly conned regime, where migration is most important, kinetic-theory expressions for polymer ux have been derived for a number of important cases, and in some situations analytical solutions for concentration distributions are available.
Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

8. The transport of polymers in the nondilute regime and/or in complex conned geometries remains poorly understood, although new computational methods are beginning to address these regimes.

DISCLOSURE STATEMENT
The author is not aware of any afliations, memberships, funding, or nancial holdings that might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS
The author gratefully acknowledges his long-time collaborators on the dynamics of conned DNA, Juan J. de Pablo and David C. Schwartz, as well as the graduate students and postdocs, Richard Jendrejack, Hongbo Ma, Yeng-Long Chen, Raj Khare, Juan Hernandez-Ortiz, Eileen Dimalanta, Kyubong Jo, and Yu Zhang, who have worked with him on it. The author has also benetted from interactions with many other researchers, including Eric Shaqfeh, Ron Larson, and Pat Doyle. The authors research in this area has been supported by the National Science Foundation, grants ECS/BES/CTS-0085560 and DMR-0425880 (Nanoscale Science and Engineering Center).

Excellent review of the classical literature on migration of polymers in ow.

LITERATURE CITED
Agarwal US, Dutta A, Mashelkar RA. 1994. Migration of macromolecules under ow: the physical origin and engineering implications. Chem. Eng. Sci. 49:1693717 Alvarez A, Soto R. 2005. Dynamics of a suspension conned in a thin cell. Phys. Fluids 17:093103 Anekal SG, Bevan MA. 2005. Interpretation of conservative forces from Stokesian dynamic simulations of interfacial and conned colloids. J. Chem. Phys. 122:034903 Aubert JH, Tirrell M. 1980. Flows of dilute polymer solutions through packed porous chromatographic columns. Rheol. Acta 19:45261 Bakajin OB, Duke TAJ, Chou CF, Chan SS, Austin RH, Cox EC. 1998. Electrohydrodynamic stretching of DNA in conned environments. Phys. Rev. Lett. 80:273740 Balducci A, Mao P, Han J, Doyle PS. 2006. Double-stranded DNA diffusion in slitlike nanochannels. Macromolecules 39:627381 Banchio AJ, Brady JF. 2003. Accelerated Stokesian dynamics: Brownian motion. J. Chem. Phys. 118:1032332 Beck VA, Shaqfeh ESG. 2006. Ergodicity breaking and conformational hysteresis in the dynamics of a polymer tethered at a surface stagnation point. J. Chem. Phys. 124:094902
Graham

Presents experimental results and comparisons with simulations (see Chen et al. 2004) for diffusion in the moderately conned regime; demonstrates screening of hydrodynamic interactions for Rbulk W.

294

Bird RB, Curtis CF, Armstrong RC, Hassager O. 1987. Dynamics of Polymeric Liquids, Vol. 2. New York: Wiley. 2nd ed. Blake JR. 1971. A note on the image system for a stokeslet in a no-slip boundary. Proc. Camb. Philos. Soc. 70:30310 Bonthuis DJ, Meyer C, Stein D, Dekker C. 2008. Conformation and dynamics of DNA conned in slitlike nanouidic channels. Phys. Rev. Lett. 101:108303 Brochard F, de Gennes PG. 1977. Dynamics of conned polymer chains. J. Chem. Phys. 67:5256 Brunn PO. 1984. Polymer migration phenomena based on the general bead-spring model for exible polymers. J. Chem. Phys. 80:582126 Brunn PO. 1985. Linear polymers in nonhomogeneous ow elds. II. The cross-stream migration velocity. J. Polym. Sci. Polym. Phys. 23:89103 Brunn PO, Grisa S. 1985. Linear polymers in nonhomogeneous ow elds. I. Translational diffusion coefcient. J. Polym. Sci. Polym. Phys. 23:7387 Butler JE, Usta OB, Kekre R, Ladd AJC. 2007. Kinetic theory of a conned polymer driven by an external force and pressure-driven ow. Phys. Fluids 19:113101 Casassa EF. 1967. Equilibrium distribution of exible polymer chains between a macroscopic solution phase and small voids. J. Polym. Sci. B Polym. Lett. 5:77378 Chen YL, Graham MD, de Pablo JJ, Jo K, Schwartz DC. 2005. DNA molecules in microuidic oscillatory ow. Macromolecules 38:668087 Chen YL, Graham MD, de Pablo JJ, Randall GC, Gupta M, Doyle PS. 2004. Conformation and dynamics of single DNA molecules in parallel-plate slit microchannels. Phys. Rev. E 70:060901 Daoud M, de Gennes PG. 1977. Statistics of macromolecular solutions trapped in small pores. J. Phys. Paris 38:8593 de Gennes PG. 1979. Scaling Concepts in Polymer Physics. Ithaca, NY: Cornell Univ. Press de Pablo JJ, Ottinger HC, Rabin Y. 1992. Hydrodynamic changes in the depletion layer of dilute polymer solutions near a wall. AIChE J. 38:27383 Del Bonis-ODonnell JT, Reisner W, Stein D. 2009. Pressure-driven DNA transport across an articial nanotopography. New J. Phys. 11:075032 Deserno M, Holm C. 1998a. How to mesh up Ewald sums. I. A theoretical and numerical comparison of various particle mesh routines. J. Chem. Phys. 109:767893 Deserno M, Holm C. 1998b. How to mesh up Ewald sums. II. An accurate error estimate for the particleparticle-particle-mesh algorithm. J. Chem. Phys. 109:7694701 Diamant H. 2009. Hydrodynamic interaction in conned geometries. J. Phys. Soc. Jpn. 78:041002 Diamant H, Cui B, Lin B, Rice SA. 2005. Hydrodynamic interaction in quasi-two-dimensional suspensions. J. Phys. Condens. Matter 17:S278793 Dimalanta ET, Lim A, Runnheim R, Lamers C, Churas C, et al. 2004. A microuidic system for large DNA molecule arrays. Anal. Chem. 76:5293301 Doi M, Edwards SF. 1986. The Theory of Polymer Dynamics. New York: Oxford Univ. Press Dufresne ER, Squires TM, Brenner MP, Grier DG. 2000. Hydrodynamic coupling of two Brownian spheres to a planar surface. Phys. Rev. Lett. 85:331720 Dunweg B, Ladd AJC. 2009. Lattice Boltzmann simulations of soft matter systems. Adv. Polym. Sci. 221:89166 Espanol P, Warren P. 1995. Statistical mechanics of dissipative particle dynamics. Europhys. Lett. 30:19196 Fan XJ, Phan-Thien N, Yong NT, Wu XH, Xu D. 2003. Microchannel ow of a macromolecular suspension. Phys. Fluids 15:1121 Fang L, Hsieh CC, Larson RG. 2007. Molecular imaging of shear-induced polymer migration in dilute solutions near a surface. Macromolecules 40:849099 Fang L, Hu H, Larson RG. 2005. DNA congurations and concentration in shearing ow near a glass surface in a microchannel. J. Rheol. 49:12738 Fang L, Larson RG. 2007. Concentration dependence of shear-induced polymer migration in DNA solutions near a surface. Macromolecules 40:878487 Fedosov DA, Karniadakis GE, Caswell B. 2008. Dissipative particle dynamics simulation of depletion layer and polymer migration in micro- and nanochannels for dilute polymer solutions. J. Chem. Phys. 128:144903
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules

Derives the analytical solution to Stokes equation for a point force above a plane wall, the foundation for studies of particle migration in ow near walls.

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Provides the rst physical arguments and scaling predictions for diffusion and relaxation time of conned exible polymers.

295

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Shows how particlemesh Ewald-type computational methods for computing hydrodynamic interactions can be generalized from periodic to arbitrary geometries.

Introduces a beadspring chain model for -phage DNA parameterized to match experimental data for bulk static and dynamic properties; this model was later used to predict properties under connement.

Demonstrates through Brownian dynamics simulations and kinetic theory many key features of polymer migration in a conned geometry.

Fixman M. 1986. Construction of Langevin forces in the simulation of hydrodynamic interaction. Macromolecules 19:1204207 Freund JB. 2007. Leukocyte margination in a model microvessel. Phys. Fluids 19:023301 Groot RD, Warren PB. 1997. Dissipative particle dynamics: bridging the gap between atomic and mesoscopic simulation. J. Chem. Phys. 107:442335 Happel J, Brenner H. 1965. Low Reynolds Number Hydrodynamics. Englewood Cliffs, NJ: Prentice Hall Harden JL, Doi M. 1992. Diffusion of macromolecules in narrow capillaries. J. Phys. Chem. 96:404652 Hasimoto H. 1959. On the periodic fundamental solutions of the Stokes equations and their application to viscous ow past a cubic array of spheres. J. Fluid Mech. 5:31728 Hernandez-Ortiz JP, de Pablo JJ, Graham MD. 2006a. N log N method for hydrodynamic interactions of conned polymer systems: Brownian dynamics. J. Chem. Phys. 125:164906 Hernandez-Ortiz JP, de Pablo JJ, Graham MD. 2007. Fast computation of many-particle hydrodynamic and electrostatic interactions in a conned geometry. Phys. Rev. Lett. 98:140602 Hernandez-Ortiz JP, Ma H, de Pablo JJ, Graham MD. 2006b. Cross-stream-line migration in conned owing polymer solutions: theory and simulation. Phys. Fluids 18:123101 Hernandez-Ortiz JP, Ma H, de Pablo JJ, Graham MD. 2008. Concentration distributions during ow of conned owing polymer solutions at nite concentration: slit and grooved channel. Korea-Aust. Rheol. J. 20:14352 Hockney RW, Eastwood JW. 1988. Computer Simulation Using Particles. New York: Taylor & Francis Hoda N, Kumar S. 2007. Kinetic theory of polyelectrolyte adsorption in shear ow. J. Rheol. 51:799820 Hsieh CC, Balducci A, Doyle PS. 2007. An experimental study of DNA rotational relaxation time in nanoslits. Macromolecules 40:5196205 Hsieh CC, Doyle PS. 2008. Studying conned polymers using single-molecule DNA experiments. Korea-Aust. Rheol. J. 20:12742 Hudson SD. 2003. Wall migration and shear-induced diffusion of uid droplets in emulsions. Phys. Fluids 15:110613 Jendrejack RM, de Pablo JJ, Graham MD. 2002. Stochastic simulations of DNA in ow: dynamics and the effects of hydrodynamic interactions. J. Chem. Phys. 116:775259 Jendrejack RM, Dimalanta ET, Schwartz DC, Graham MD, de Pablo JJ. 2003a. DNA dynamics in a microchannel. Phys. Rev. Lett. 91:038102 Jendrejack RM, Graham MD, de Pablo JJ. 2000. Hydrodynamic interactions in long chain polymers: application of the Chebyshev polynomial approximation in stochastic simulations. J. Chem. Phys. 113:2894 900 Jendrejack RM, Schwartz DC, de Pablo JJ, Graham MD. 2004. Shear-induced migration in owing polymer solutions: simulation of long-chain deoxyribose nucleic acid in microchannels. J. Chem. Phys. 120:251329 Jendrejack RM, Schwartz DC, Graham MD, de Pablo JJ. 2003b. Effect of connement on DNA dynamics in microuidic devices. J. Chem. Phys. 119:116573 Jhon MS, Freed KF. 1985. Polymer migration in Newtonian uids. J. Polym. Sci. Polym. Phys. 23:95571 Jo K, Chen YL, de Pablo JJ, Schwartz DC. 2009. Elongation and migration of single DNA molecules in microchannels using oscillatory shear ows. Lab Chip 9:234855 Jo K, Dhingra DM, Odijk T, de Pablo JJ, Graham MD, et al. 2007. A single-molecule barcoding system using nanoslits for DNA analysis. Proc. Natl. Acad. Sci. USA 104:267378 Khare R, Graham MD, de Pablo JJ. 2006. Cross-stream migration of exible molecules in a nanochannel. Phys. Rev. Lett. 96:224505 Kohale SC, Khare R. 2009. Cross stream chain migration in nanouidic channels: effects of chain length, channel height, and chain concentration. J. Chem. Phys. 130:104904 Larson JW, Yantz GR, Zhong Q, Charnas R, DAntoni CM, et al. 2006. Single DNA molecule stretching in sudden mixed shear and elongational microows. Lab Chip 6:118799 Larson RG. 2005. The rheology of dilute solutions of exible polymers: progress and problems. J. Rheol. 49:170 Leal LG. 1980. Particle motions in a viscous uid. Annu. Rev. Fluid Mech. 12:43576
Graham

296

Liron N, Mochon S. 1976. Stokes ow for a stokeslet between 2 parallel at plates. J. Eng. Math. 10:287303 Ma HB, Graham MD. 2005. Theory of shear-induced migration in dilute polymer solutions near solid boundaries. Phys. Fluids 17:083103 Macdonald MJ, Muller SJ. 1996. Experimental study of shear-induced migration of polymers in dilute solutions. J. Rheol. 40:25983 Malevanets A, Kapral R. 1999. Mesoscopic model for solvent dynamics. J. Chem. Phys. 110:860513 Mark JE, Flory PJ. 1965. The conguration of the polyoxyethylene chain. J. Am. Chem. Soc. 87:141523 Millan JA, Laradji M. 2009. Cross-stream migration of driven polymer solutions in nanoscale channels: a numerical study with generalized dissipative particle dynamics. Macromolecules 42:80310 Nitsche LC. 1996. Cross-stream migration of bead-spring polymers in nonrectilinear pore ows. AIChE J. 42:61322 Nitsche LC, Hinch EJ. 1997. Shear-induced lateral migration of Brownian rigid rods in parabolic channel ow. J. Fluid Mech. 332:121 Odijk T. 2008. Scaling theory of DNA conned in nanochannels and nanoslits. Phys. Rev. E 77:060901 Olla P. 1999. Simplied model for red cell dynamics in small blood vessels. Phys. Rev. Lett. 82:45356 Ottinger HC. 1996. Stochastic Processes in Polymeric Fluids. New York: Springer Park J, Bricker JM, Butler JE. 2007. Cross-stream migration in dilute solutions of rigid polymers undergoing rectilinear ow near a wall. Phys. Rev. E 76:040801 Perkins TT, Smith DE, Larson RG, Chu S. 1995. Stretching of a single tethered polymer in a uniform ow. Science 268:8387 Pham TT, Schiller UD, Prakash JR, Dunweg B. 2009. Implicit and explicit solvent models for the simulation of a single polymer chain in solution: lattice Boltzmann versus Brownian dynamics. J. Chem. Phys. 131:164114 Reisner W, Morton KJ, Riehn R, Wang YM, Yu ZN, et al. 2005. Statics and dynamics of single DNA molecules conned in nanochannels. Phys. Rev. Lett. 94:196101 Ripoll M, Ernst MH, Espanol P. 2001. Large scale and mesoscopic hydrodynamics for dissipative particle dynamics. J. Chem. Phys. 115:727184 Rotne J, Prager S. 1969. Variational treatment of hydrodynamic interaction in polymers. J. Chem. Phys. 50:483137 Saintillan D, Shaqfeh ESG, Darve E. 2006. Effect of exibility on the shear-induced migration of short-chain polymers in parabolic channel ow. J. Fluid Mech. 557:297306 Schiek RL, Shaqfeh ESG. 1997. Cross-streamline migration of slender Brownian bres in plane Poiseuille ow. J. Fluid Mech. 332:2339 Segre G, Silberberg A. 1962a. Behaviour of macroscopic rigid spheres in Poiseuille ow, Part 1. Determination of local concentration by statistical analysis of particle passages through crossed light beams. J. Fluid Mech. 14:11535 Segre G, Silberberg A. 1962b. Behaviour of macroscopic rigid spheres in Poiseuille ow, Part 2. Experimental results and interpretation. J. Fluid Mech. 14:13657 Sendner C, Netz RR. 2008. Shear-induced repulsion of a semiexible polymer from a wall. Europhys. Lett. 81:54006 Seo YH, Park OO, Chun MS. 1996. The behavior of velocity enhancement in microcapillary ows of exible water-soluble polymers. J. Chem. Eng. Jpn. 29:61119 Shaqfeh ESG. 2005. The dynamics of single-molecule DNA in ow. J. Non-Newton. Fluid Mech. 130:128 Sierou A, Brady JF. 2001. Accelerated Stokesian dynamics simulations. J. Fluid Mech. 448:11546 Smart JR, Leighton DT. 1991. Measurement of the drift of a droplet due to the presence of a plane. Phys. Fluids A 3:2128 Smith ER, Snook IK, van Megen W. 1987. Hydrodynamic interactions in Brownian dynamics: I. Periodic boundary conditions for computer simulations. Phys. A 143:44167 Squires TM, Quake SR. 2005. Microuidics: uid physics at the nanoliter scale. Rev. Mod. Phys. 77:9771026 Staben M, Zinchenko AZ, Davis RH. 2003. Motion of a particle between two parallel plane walls in lowReynolds-number Poiseuille ow. Phys. Fluids 15:171133 Stein D, van der Heyden FHJ, Koopmans WJA, Dekker C. 2006. Pressure-driven transport of conned DNA polymers in uidic channels. Proc. Natl. Acad. Sci. USA 103:1585358
www.annualreviews.org Fluid Dynamics of Dissolved Polymer Molecules

Derives the analytical solution to Stokes equation for a point force in a slit and demonstrates key features of its far-eld structure.

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Presents analytical theory for migration of bead-spring dumbbells above a no-slip wall, including a general ux expression, steady-state concentration prole, and similarity solutions for spatial and temporal development of depletion layers.

Shows for the rst time that polymer dynamics in ow can be examined at the single-molecule level, via uorescence microscopy of stained DNA molecules.

297

Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Sugarman JH. 1988. Microcapillary chromatography and radial migration of water-soluble polymers. PhD thesis. Princeton Univ. Tegenfeldt JO, Prinz C, Cao H, Chou S, Reisner WW, et al. 2004a. The dynamics of genomic-length DNA molecules in 100-nm channels. Proc. Natl. Acad. Sci. USA 101:1097983 Tegenfeldt JO, Prinz C, Cao H, Huang RL, Austin RH, et al. 2004b. Micro- and nanouidics for DNA analysis. Anal. Bioanal. Chem. 378:167892 Teraoka I. 1996. Polymer solutions in conning geometries. Prog. Polym. Sci. 21:89149 Tlusty T. 2006. Screening by symmetry of long-range hydrodynamic interactions of polymers conned in sheets. Macromolecules 39:392730 Toukmaji AY, Board JA Jr. 1996. Ewald summation techniques in perspective: a survey. Comput. Phys. Commun. 95:7392 Usta OB, Butler JE, Ladd AJC. 2006. Flow-induced migration of polymers in dilute solution. Phys. Fluids 18:031703 Usta OB, Ladd AJC, Butler JE. 2005. Lattice-Boltzmann simulations of the dynamics of polymer solutions in periodic and conned geometries. J. Chem. Phys. 122:094902 Viovy JL. 2000. Electrophoresis of DNA and other polyelectrolytes: physical mechanisms. Rev. Mod. Phys. 72:81372 Wall FT, Seitz WA, Chin JC, de Gennes PG. 1978. Statistics of self-avoiding walks conned to strips and capillaries. Proc. Natl. Acad. Sci. USA 75:206970 Watari N, Makino M, Kikuchi N, Larson RG, Doi M. 2007. Simulation of DNA motion in a microchannel using stochastic rotation dynamics. J. Chem. Phys. 126:094902 Woo NJ, Shaqfeh ESG, Khomami B. 2004a. The effect of connement on dynamics and rheology of dilute deoxyribose nucleic acid solutions. II. Effective rheology and single chain dynamics. J. Rheol. 48:299318 Woo NJ, Shaqfeh ESG, Khomami B. 2004b. Effect of connement on dynamics and rheology of dilute DNA solutions. I. Entropic spring force under connement and a numerical algorithm. J. Rheol. 48:28198 Yamakawa H. 1970. Transport properties of polymer chains in dilute solution: hydrodynamic interaction. J. Chem. Phys. 53:43643 Zhang Y, de Pablo JJ, Graham MD. 2009. Multiple free energy minima in systems of conned tethered polymers-toward soft nanomechanical bistable elements. Soft Matter 5:3694700 Zheng J, Yeung ES. 2002. Anomalous radial migration of single DNA molecules in capillary electrophoresis. Anal. Chem. 74:453647 Zheng J, Young ES. 2003. Mechanism for the separation of large molecules based on radial migration in capillary electrophoresis. Anal. Chem. 75:367580

298

Graham

Contents
Experimental Studies of Transition to Turbulence in a Pipe T. Mullin p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Annual Review of Fluid Mechanics Volume 43, 2011

Fish Swimming and Bird/Insect Flight Theodore Yaotsu Wu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p25 Wave Turbulence Alan C. Newell and Benno Rumpf p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p59 Transition and Stability of High-Speed Boundary Layers Alexander Fedorov p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p79 Fluctuations and Instability in Sedimentation Elisabeth Guazzelli and John Hinch p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p97 Shock-Bubble Interactions Devesh Ranjan, Jason Oakley, and Riccardo Bonazza p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 117 Fluid-Structure Interaction in Internal Physiological Flows Matthias Heil and Andrew L. Hazel p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 141 Numerical Methods for High-Speed Flows Sergio Pirozzoli p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 163 Fluid Mechanics of Papermaking Fredrik Lundell, L. Daniel S derberg, and P. Henrik Alfredsson p p p p p p p p p p p p p p p p p p p p p p p 195 o Lagrangian Dynamics and Models of the Velocity Gradient Tensor in Turbulent Flows Charles Meneveau p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 219 Actuators for Active Flow Control Louis N. Cattafesta III and Mark Sheplak p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 247 Fluid Dynamics of Dissolved Polymer Molecules in Conned Geometries Michael D. Graham p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 273 Discrete Conservation Properties of Unstructured Mesh Schemes J. Blair Perot p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 299 Global Linear Instability Vassilios Theolis p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 319
v

HighReynolds Number Wall Turbulence Alexander J. Smits, Beverley J. McKeon, and Ivan Marusic p p p p p p p p p p p p p p p p p p p p p p p p p p p p 353 Scale Interactions in Magnetohydrodynamic Turbulence Pablo D. Mininni p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 377 Optical Particle Characterization in Flows Cameron Tropea p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 399 Aerodynamic Aspects of Wind Energy Conversion Jens Nrkr Srensen p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 427
Annu. Rev. Fluid Mech. 2011.43:273-298. Downloaded from www.annualreviews.org by Indian Institute of Science- Bangalore on 03/18/12. For personal use only.

Flapping and Bending Bodies Interacting with Fluid Flows Michael J. Shelley and Jun Zhang p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 449 Pulse Wave Propagation in the Arterial Tree Frans N. van de Vosse and Nikos Stergiopulos p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 467 Mammalian Sperm Motility: Observation and Theory E.A. Gaffney, H. Gad lha, D.J. Smith, J.R. Blake, and J.C. Kirkman-Brown p p p p p p p 501 e Shear-Layer Instabilities: Particle Image Velocimetry Measurements and Implications for Acoustics Scott C. Morris p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 529 Rip Currents Robert A. Dalrymple, Jamie H. MacMahan, Ad J.H.M. Reniers, and Varjola Nelko p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 551 Planetary Magnetic Fields and Fluid Dynamos Chris A. Jones p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 583 Surfactant Effects on Bubble Motion and Bubbly Flows Shu Takagi and Yoichiro Matsumoto p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 615 Collective Hydrodynamics of Swimming Microorganisms: Living Fluids Donald L. Koch and Ganesh Subramanian p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 637 Aerobreakup of Newtonian and Viscoelastic Liquids T.G. Theofanous p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 661 Indexes Cumulative Index of Contributing Authors, Volumes 143 p p p p p p p p p p p p p p p p p p p p p p p p p p p p 691 Cumulative Index of Chapter Titles, Volumes 143 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 699 Errata An online log of corrections to Annual Review of Fluid Mechanics articles may be found at http://uid.annualreviews.org/errata.shtml
vi Contents

Vous aimerez peut-être aussi