Nanomechanical resonators with high aspect ratio, such as nanotubes and nanowires are of interest due to their expected high sensitivity. However, a strongly nonlinear response combined with a high thermomechanical noise level limits the useful linear dynamic range of this type of device. We derive the equations governing this behavior and find a strong dependence ͓ϰd ͱ ͑d / L͒ 5 ͔ of the dynamic range on aspect ratio. © 2005 American Institute of Physics. ͓DOI: 10.1063/1.1929098͔The limits of mechanical detection with nanoelectromechanical systems ͑NEMS͒ are being actively pursued for sensing applications, such as the attainment of sub-attonewton force sensing for magnetic resonance force microscopy, 1 sub-attogram mass sensing, 2,3 mechanical single spin detection, 4 or the study of mechanical motion approaching the quantum regime. [5][6][7][8] Applications like these require both high responsivity and ultra-high-frequency operation. 9 Both can be attained simultaneously with small diameter, large aspect ratio doubly clamped resonators. Nanoscale materials such as carbon nanotubes or nanowires are a natural choice for these resonators due to their intrinsic small size. We recently reported a bottom-up nanomechanical resonator, a Pt nanowire, and found that it takes a very low driving power to bring this device into the nonlinear regime. 10 Here, we show how the onset of this nonlinear regime decreases with decreasing diameter, while the thermomechanical noise increases with aspect ratio. We conclude that the useful linear dynamic range of such devices is severely limited, with the result that many applications will involve operation in the nonlinear regime.A typical layout for a doubly clamped nanomechanical resonator is shown in Fig. 1. The resonator can be driven and detected in several ways, e.g., magnetomotively, 11 or optically. 12 The driving force f͑t͒ leads to a time dependent bending profile z͑x , t͒, which can be found by solving the differential equationwith boundary conditions z͑0͒ = z͑L͒ = z x ͑0͒ = z x ͑L͒ = 0. Here, A is the cross-sectional area, E is Young's modulus, is the density, and I is the moment of inertia about the longitudinal axis of the beam. The term in between brackets describes tension in the beam, and is a sum of residual tension T 0 and a bending-induced tension, respectively. Since Eq. ͑1͒ cannot be solved exactly we use the Galerkin discretization procedure, 13 representing the solution to Eq. ͑1͒ in terms of a linearly independent set of basis functions n ͑x͒ where each basis function satisfies the boundary conditions. The error associated with this approximation technique is
͑2͒The Galerkin procedure requires this error to be orthogonal to each basis function, or in other words, the error is a residual that cannot be expressed in terms of the given finite set of basis functions:We are interested in the response of the beam at resonance when the first mode is dominant, so it suffices to consider the case n = 1. For a doubly clamped beam, the simplest function that approximate...