Trapping force on a finite-sized particle in a dielectrophoretic cage P. Singh and N. Aubry Department of Mechanical Engineering, New Jersey Institute of Technology, Newark, New Jersey 07102, USA Received 24 February 2005; published 5 July 2005 The point dipole PDmodel is routinely used for estimating the dielectrophoretic DEPforce acting on a particle placed in the nonuniform electric fields of dielectrophoresis devices, such as square cages. We show that if the particle size is much smaller than the dielectrophoretic cage size, the PD model accurately approxi- mates the actual DEP force, computed numerically using the Maxwell stress tensor method. However, when the two sizes are comparable, the actual DEP force differs significantly in both magnitude and direction from that given by the PD model. DOI: 10.1103/PhysRevE.72.016602 PACS numbers: 41.20.Cv In recent years, primarily due to improvements in the mi- crofabrication techniques, many new applications of dielec- trophoresis have been developed for the handling of micron and nanosized particles. For example, dielectrophoresis has been used for separating submicron-sized latex spheres and viruses 1,2, separating DNA molecules and proteins 3, characterizing and separating bacterial cells 4,5, removing cancer cells from human blood 6–8, and trapping submi- cron sized particles in cages with dimensions comparable to the particles size 9,10. Our focus in this paper is on the contact less trapping force acting on a particle placed in a dielectrophoretic square 3D cage which in the past has been estimated using the point dipole PDmodel 11–15. The goal of the present study is to compute the actual force in this configuration and determine the validity of the PD model. A simple 3D dielectrophoretic cage can be formed by placing four electrodes in the four side walls of a square shaped channel, as shown in Fig. 1. The voltages of the four electrodes are selected such that the electric field magnitude is locally minimum at the center of the domain where one wishes to attract and hold the particle. This device is of prac- tical interest as it provides a way to trap a particle in a con- tact less fashion, at the center of the cage. The magnitude of the electric field E in the xz plane for the case without the particle is shown in Fig. 2a. Notice that the magnitude is locally minimum at the cage center and increases with in- creasing distance from the domain center. Figure 2bshows that the electric field inside the cage is nonuniform, and its gradient near the domain center is non- zero, except at the center itself where it is zero. The lines of the gradient of the electric field magnitude shown in Fig. 2bemanate approximately radially from the cage center and end at the edges of the electrodes. If a particle is placed in this domain and its dielectric constant is smaller than that of the liquid, it will experience the so-called dielectro- phoretic DEPforce towards the center of the domain, i.e., in the direction opposite to the lines of the gradient of the electric field magnitude. If the dielectric constant of the par- ticle is greater than that of the liquid, the DEP force is in the direction away from the center. This situation will not be considered here, since it has limited practical applications as in this case the particle will not be trapped at the center of the cage in a contact less manner. An estimate of the trapping force can be obtained using the PD model which considers the particle as a point dipole and thus assumes that the gradient of the electric field is approximately constant over the particle. According to the PD model, the dielectrophoretic force acting on a linearly and homogeneously polarizable spherical particle placed in a dc electric field is given by the expression F DEP, PD =4a 3 0 c E · E, where a is the particle radius, 0 = 8.8542 10 -12 F / m is the permittivity of free space, E is the electric field, and = p - c / p +2 c is the Clausius- Mossotti factor, c and p are the dielectric constants of the liquid and particle. We will assume that the particles and liquid are both perfect dielectrics. Our results are also appli- cable to ac electric fields, provided the rms value of the electric field is used, is replaced by the real part of the complex frequency dependent Clausius-Mossotti factor and the force is the time averaged force. We will also present some results where the DEP force is estimated using the quadrupole model 11,15. Clearly, when the size of the particle being trapped is comparable to the cage size, the PD model is expected to be in error because the assumption made on the electric field nonuniformity i.e., the nonuniformity is modest and its scale is large compared to the particle sizeis no longer valid and also because the presence of the particle modifies the overall electric field distribution in the cage, as the distance between the particle and the cage walls is comparable to the particle size 10. To correct this error, the electric field must be obtained by including the particle in the electric field prob- lem, as was done in 10for the two canonical cases of a cylindrical particle in a cylindrical shell and a spherical par- ticle in a spherical shell. It has been noted in the past that the error in the DEP force due to the assumptions made on the electric field nonuniformity can be reduced by incorporating the quadrupole, and if needed, additional higher order terms 11,15. However, the error due to the modification of the electric field cannot be fully corrected by simply adding the higher order multipolar terms, as these terms are evaluated using the electric field computed without the particle. We make this point by presenting some results where the quad- rupole terms are also included for estimating the DEP force. It is worth noting that in the uniform electric field case, the method of image can be used to determine the modified elec- PHYSICAL REVIEW E 72, 016602 2005 1539-3755/2005/721/0166025/$23.00 ©2005 The American Physical Society 016602-1