We theoretically study the molecular-weight dependence of DNA thermophoresis, which arises from mutual advection of the n repeat units of the molecular chain. As a main result we find that the dominant driving forces, i.e., the thermally induced permittivity gradient and the electrolyte Seebeck effect, result in characteristic hydrodynamic screening. In comparison with recent experimental data on single-stranded DNA (2 ≤ n ≤ 80), our theory provides a good description for the increase of the drift velocity up to n = 30; the slowing-down of longer molecules is well accounted for by a simple model for counterion condensation. It turns out that thermophoresis may change sign as a function of n: For an appropriate choice of the salt-specific Seebeck coefficient, short molecules move to the cold and long ones to the hot; this could be used for separating DNA by molecular weight.
PACS numbers:When applying a temperature gradient on a colloidal dispersion, one observes thermally driven transport towards the hot or the cold [1,2]. In recent years, thermophoresis has been shown to provide a versatile means for manipulating DNA, including translocation through plasmonic nanopores [3], stretching in nanochannels [4,5], separation by molecular weight [6], sequencespecific detection with functionalized nanoparticles [7], and force-free trapping of single molecules [8]. Protein thermophoresis has become a standard technology in biomedical analysis [9], and the accumulation of RNA in hydrothermal pores is discussed as a scenario for biomolecular synthesis in the early evolution of life [10].In the last decade, much progress has been made concerning the physical mechanisms of thermophoresis of charged colloids. It has been shown that, in addition to thermo-osmosis [11,12], the electrolyte Seebeck field [13][14][15][16][17][18] and concentration gradients of salt [16] or nonionic polymers [19,20], play an important role. These companion fields arise from specific solvation enthalpies of salt ions or nonionic solutes, and are at the origin of the "inverse" Soret effect, where the colloids accumulate in hot regions [16,19]. Regarding the size dependence, there is conclusive evidence that the mobility of colloidal beads does not vary with the radius [21,22].In spite of the many experimental studies mentioned above, little is known on the molecular-weight dependence of DNA thermophoresis. If the hydrodynamic slowing-down of Brownian motion is well understood in terms of mutual advection of the repeat units [23], a more complex picture arises for phoretic motion where external forces are absent and which is driven by non-equilibrium surface properties. For short-ranged dispersion forces, hydrodynamic interactions are irrelevant and the thermophoretic velocity is constant [24,25]; deviations observed for very short polymers in organic solvents [26], arise probably from chemically different end groups. For DNA in a weak electrolyte, however, the electrostatic interaction length may attain tens of nanometers, which suggests an incomplet...