Dipole relaxation losses in DNA

The electrodynamic response of DNA in the millimeter wave range is investigated. By performing measurements under a wide range of humidity conditions and comparing the response of single-stranded DNA and double-stranded DNA, we show that the appreciable ac conductivity of DNA is not due to photon-assisted hopping between localized states but is instead due to dissipation from dipole motion in the surrounding water helix. Such a result, where the conductivity is due to the constrained motion of overdamped dipoles, reconciles the vanishing dc conductivity of DNA with the considerable ac response. The electrical conductivity of DNA has been a topic of much recent interest and controversy.1 Measurements from different groups have reached a variety of conclusions about the nature of charge transport along the double helix. DNA has been reported to be metallic,2 semiconducting,3 insulating,4,5 and even a proximity-effect-induced superconductor.6 However, questions have been raised in many of these papers with regard to the role played by electrical contacts, length effects, and the manner in which electrostatic damage, mechanical deformation by substrate-molecule interaction, residual salt concentrations, and other contaminants may have affected these results. Some recent measurements, where care was taken to both establish a direct chemical bond between λ-DNA and Au electrodes and also control the excess ion concentration, have given compelling evidence that the dc resistivity of the DNA double helix over long length scales (<10 μm) is very high indeed (F > 106 Ω‚cm).7 These results were consistent with earlier work that found flat I-V characteristics and vanishingly small conductances5 but contrast with other studies that found a substantial dc conductance that was interpreted in terms of small polaron hopping.8 Although we will revisit this subject below, dc measurements that show DNA to be a good insulator are also in apparent contradiction with recent contactless ac measurements that have shown appreciable conductivity at microwave and far-infrared frequencies,9,10 the magnitude of which approaches that of a well-doped semiconductor.11 Previously, the ac conductivity in DNA was found to be well parametrized as a power law in ω.9,10 Such a dependence can be a general hallmark of ac conductivity in disordered systems with photon-assisted hopping between random localized states12 and led to the reasonable interpretation that intrinsic disorder, counterion fluctuations, and possibly other sources created a small number of electronic states on the base pair sequences in which charge conduction could occur. However, such a scenario would lead to thermally activated hopping conduction between localized states and is thus inconsistent with a very low dc conductivity.7 A number of outstanding issues arise: Are there localized regions along the helix where a continuous conducting path is not present but ac hopping between localized states over distances of a few base pairs can still occur? Are there sensitive length dependencies in the DNA strands? Are there differences between the samples of various groups? Are there perhaps different charge-conduction mechanisms that play a role at finite frequency? To resolve some of these matters, we have performed ac conductivity experiments in the millimeter wave range under a wide range of humidity conditions. We show that the appreciable ac conductivity of DNA in the microwave and far-infrared regimes should not be viewed as some sort of hopping between localized states and is instead likely due to dissipation in the dipole response of the water molecules in the surrounding hydration layer. Our data can be well described by a Debye-like relaxation of water molecules in the surrounding water helix. At low humidities, the response is well modeled by considering the rotation of single water molecules in the structural water layer. As the number of water molecules per base pair increases, dissipation due to the collective motion of water dipoles increases until eventually the conductivity resembles that of bulk water. By measuring both single-stranded (ssDNA) and double-stranded DNA (dsDNA) over a wide range of humidities, we are able to show that, at least in principle, all of the ac conductivity of DNA can be assigned to relaxation losses of water dipoles. This finding can be taken to support those measurements that find a vanishingly small dc conductivity and indicate that DNA is a poor candidate for a molecular wire. Double-stranded DNA films were obtained by vacuum drying a 7 mM PBS solution containing 20 mg/mL of sodium * Corresponding author. E-mail: [email protected]. NANO LETTERS 2004 Vol. 4, No. 4 733-736 10.1021/nl049961s CCC: $27.50 © 2004 American Chemical Society Published on Web 03/11/2004 salt DNA extracted from calf thymus and salmon testes (Sigma D1501 and D1626). The results were found to be independent of the use of calf or salmon DNA. Our choice of these concentrations deserves further explanation. It is well known that at a given temperature the double-helical conformation of DNA can exist in solution only within a certain concentration of positive ions. Excess salt cannot be removed by vacuum drying, so large amounts of residual salt in films could introduce significant errors in conductivity due to both the ionic conduction of the salt itself and its additional hydration during humidity changes. Meltingtemperature calculations13,14 for long native pieces of DNA with C-G content around 40% show that a 2-10 mM concentration of sodium cations is enough to stabilize the double helix at room temperature. Films were prepared with differing salt amounts, and it was found that as long as the excess salt mass fraction is kept between 2 and 5% the final results were not significantly affected. To improve the DNA/ salt mass ratio, we used as high a concentration of DNA as possible, but 20 mg/mL appears to be the limit. Higher concentrations make it difficult for DNA fibers to dissolve, and the solution becomes too viscous, which prevents the production of the flat uniform films that are of paramount importance for the quasi-optical resonant technique. Singlestranded DNA films were prepared from the same original solution as the double-stranded ones, with preliminary heating to 95 °C for 30 min and fast cooling to 4 °C. The dry films were 20-30 μm thick and were made on top of 1-mm sapphire windows. Immediately after solution deposition, the sapphire substrates were vacuum centrifuged at 500g. This expels any air trapped inside the viscous solution; otherwise, the evaporation process causes the formation of air bubbles that destroy the film uniformity. In both dsDNA and ssDNA cases, the conformational state was checked by fluorescence microscope measurements. Additionally, some ssDNA films were also made by depositing 20 mg/mL DNA solution in distilled water or PBS buffer at a constant 95 °C and then keeping the film well above the denaturing temperature until dry. Such films should be entirely constituted of ssDNA; no experimental difference was found between these films and those prepared in the alternative