Thermodynamically Consistent Continuum Theory of Magnetic Particles in High-Gradient Fields

TL;DR

We address the challenge of predicting magnetic-particle transport and capture in high-gradient fields by deriving a thermodynamically consistent continuum theory from a free-energy functional that couples magnetic energy, entropy, and sterics, yielding a concentration-dependent susceptibility via homogenization bounds. The model yields a generalized magnetostatic equation and a magnetochemical potential, enabling self-consistent field distortion and field shielding without empirical shut-off. Simulations reveal canonical HGMS morphologies and three regimes of capture organized by the Mason number: $ \mathrm{Mn} \ll 10^4 $ (thermodynamically controlled), $ \mathrm{Mn} \sim 10^4 $ (transitional), and $ \mathrm{Mn} \gg 10^5 $ (dynamically controlled), with orientation affecting deposition. The framework provides a scalable platform for in silico optimization and digital-twin development for industrial-scale high-gradient magnetic separation.

Abstract

Magnetic particles underpin a broad range of technologies, from water purification and mineral processing to bioseparations and targeted drug delivery. The dynamics of magnetic particles in high-gradient magnetic fields-encompassing both their transport and eventual capture-arise from the coupled interplay of field-driven drift, fluid advection, and particle-field feedback. These processes remain poorly captured by existing models relying on empirical closures or discrete particle tracking. Here, we present a thermodynamically consistent continuum theory for collective magnetic particle transport and capture in high-gradient fields. The framework derives from a free-energy functional that couples magnetic energy, entropic mixing, and steric interactions, yielding a concentration-dependent susceptibility via homogenization theory. The resulting equations unify magnetism, mass transport, and momentum balances without ad hoc shut-off criteria, allowing field shielding, anisotropic deposition, and boundary-layer confinement to emerge naturally. Simulations predict canonical capture morphologies-axially aligned plumes, crescent-shaped deposits, and nonlinear shielding-across field strengths and flow regimes, consistent with trends reported in prior experimental and modeling studies. By organizing captured particle mass data into a dimensionless phase diagram based on the Mason number, we reveal three distinct regimes-thermodynamically controlled, transitional, and dynamically controlled. This perspective provides a predictive platform for in silico optimization and extension to three-dimensional geometries, and informing digital twin development for industrial-scale high-gradient magnetic separation processes.

Figures (5)

• Figure 1: Hierarchical structure of the magnetic nanoparticle (MNP) suspension. (Microscopic) Single MNPs with diameters of 5--10 nm. (Mesoscopic) Aggregates of clustered MNPs with effective diameters of 100--2000 nm, treated as the primary species in the continuum model. (Continuum) Homogenized slurry composed of multiple aggregates dispersed in the suspending fluid.
• Figure 2: Steady-state magnetic particle concentration fields at $t = 200$ s for orthogonal (left) and parallel (right) wire configurations, shown across increasing magnetic field strengths [$B_{\text{0}} = 0.01$, 0.1, 0.25 T] and flow velocities [$v_{\text{0}} = 0.001$, 0.01, 0.1 m s$^{-1}$]. Colormaps show normalized concentration $(\tilde{c} - \tilde{c}_0)/\tilde{c}_{\mathrm{max}}$, scaled to 10% of the peak value for each condition to enhance contrast. Arrows indicate the imposed flow direction $\mathbf{v}$. Spatial axes are in [mm].
• Figure 3: Dimensionless phase diagram for field–induced particle capture: Captured mass versus Mason number for orthogonal (blue) and parallel (orange) field–flow configurations. The Mason number is defined with far–field parameters as $\mathrm{Mn}=\eta_f v_{0}/(\mu_{0}\chi B_{0}^{2} r_w)$, where $r_w$ is the wire (collector) radius and $\chi\equiv\chi_a$. Three regimes are visible: thermodynamically controlled ($\mathrm{Mn}\ll10^{4}$), transition ($\mathrm{Mn}\sim10^{4}$), and dynamically controlled ($\mathrm{Mn}\gg10^{5}$). Solid lines are power–law fits; the fitted exponents are $-0.94$ (orthogonal) and $-0.92$ (parallel) with $R^{2}>0.98$. Points at very low velocity ($v_{0}=10^{-4}\,\text{m\,s}^{-1}$) approach a saturation plateau. The y–axis reports the cumulative inward particle mass per unit depth.
• Figure 4: Temporal evolution of magnetic particle accumulation (left) and magnetic field magnitude (right) in the parallel configuration at $B_{\text{0}} = 0.01$ T and $v_{\text{0}} = 0.001$ m s$^{-1}$. Concentrations are normalized by $c_{\mathrm{max}}$, the random close-packing limit. Magnetic field intensity [T] shows a $\sim$5% decrease from $t = 30$ s to $t = 70$ s, with the high-gradient zone retreating toward the wire surface (within $\sim$0.1 mm).
• Figure S1: Computational domain: magnetizable cylindrical collector (center) in a nonmagnetic fluid. Outer boundaries set the magnetic scalar potential; internal boundaries enforce no‑flux and no‑slip conditions.