Presets
Playback
Parameters
Camera
Layers
Quality
Model check
The science
What is modelled
Nanoparticles are pipetted into the medium of a culture well and the cells at the bottom take them up. The dose the cells really see is set by how fast the particles travel the height H of the medium, not by the amount you pipetted. Each particle drawn is a Brownian walker that is carried down by gravity (Stokes settling), jostled by thermal motion (diffusion), reflected at the free surface and absorbed when it reaches the cell layer. Each walker stands for 1/500 of the dose and is drawn larger than a real nanoparticle. Only the vertical coordinate is simulated; the sideways position is fixed.
Equations
- Stokes settling velocity of a sphere of radius r in a medium of viscosity μ:
v = (2/9) (ρeff − ρf) g r² / μ, with μ = 0.78 mPa s (serum-containing medium at 37 °C) times the slider, ρf = 993 kg/m³. - Stokes-Einstein diffusion coefficient:
D = kB T / (6 π μ r), T = 310.15 K. - Agglomerates: N primary particles form a fractal cluster with outer radius
r = r0 N1/Df(Df = 2.1), solid fractionφ = N1 − 3/Dfand effective densityρeff = ρf + (ρp − ρf) φ; the cluster is treated as a sphere of radius r in the drag law and the diffusion law. - Transport of the concentration C(y,t), depth y measured down from the free surface:
∂C/∂t = D ∂²C/∂y² − v ∂C/∂y, no flux at y = 0, C = 0 at y = H (the cells take up everything that arrives), C = C0 at the start. - The Peclet number
Pe = v H / D = (4π/3) (ρeff − ρf) g r³ H / (kB T)does not depend on the viscosity. Above about 5 settling carries the dose, below 0.2 diffusion does. - Exact solution used as reference (dimensionless τ = D t / H², β = Pe/2):
C = eβ y/H Σn an e−(un² + β²) τ sin(un(1 − y/H))with the roots oftan u = −u/βin ((n − 1/2)π, nπ) and the expansion coefficients an from the orthogonality of the sine modes. It is evaluated for Pe up to 40 (beyond that the exponentials overflow the double-precision sum and the plug-flow limit v t / H is shown). - Walker update:
y → y + v Δt + √(2 D Δt) ξwith a reflection at the surface; a walker also counts as delivered if the Brownian bridge between its old and new position crosses the cell layer, with probabilityexp(−(H − y0)(H − y1) / (D Δt)). The step is limited so that neither the drift nor the diffusive step exceeds H/30 per substep.
Checks run in the page
- Pe from the Stokes and Einstein formulas against the closed form (and unchanged by the viscosity).
- The exact series without settling against the classical sine series.
- Walkers against the exact solution at Pe = 4 for three times.
- Settling-dominated limit at Pe = 2000: half the delivery time gives half the dose.
- Size scaling (r² for settling, 1/r for diffusion) and the fractal cluster density.
Honest simplifications
- One-dimensional column with a perfectly absorbing cell layer: real cells take up particles at a finite rate, so the delivered dose is a maximum, and particles that reach the bottom are assumed to stay there. Only the particles that arrive count as the dose; dissolution, uptake and the fate of particles inside cells are not modelled.
- Stokes drag for a sphere at low Reynolds number; particles are spheres or fractal clusters of constant size. The cluster size is chosen by the selector and does not grow in time (no coagulation kinetics), and a cluster uses its outer radius for drag, which is an approximation for open fractal clusters.
- The medium is treated as a Newtonian liquid of constant density and viscosity; the serum proteins that coat particles and change the cluster size are not modelled. Convection (thermal currents, pipetting mixing, plate vibration) is ignored although it can matter for small particles; the 3 mm column and the 0.78 mPa s viscosity are typical values, not a measured assay.
- The well is drawn wider relative to its height than a real one, and the cells and the pipette are cosmetic. The time axis is accelerated (the playback speed sets how many hours pass per second).
What to try
- Compare 100 nm gold (Pe about 66, the dose arrives in about 7 hours) with 100 nm polystyrene (Pe about 0.2): the light particles take days.
- Put 100 nm gold in clusters of 30: the cluster is far bigger and settles faster even though it is less dense.
- Raise the viscosity: settling and diffusion slow by the same factor, so Pe stays put but everything takes longer.
- Increase the height to 6 mm: the delivery time grows with H for settling and with H² for diffusion.
Loading model...
Loading Blender model...