Unilamellar
One bilayer encloses one aqueous interior.
How can a tiny membrane carry a medicine? Meet the carrier, run a loading experiment, and follow the moment a liposome begins to form.
A typical phospholipid has a hydrophilic (water-interacting) head and two hydrophobic hydrocarbon tails. These different regions help explain how a membrane organizes.
Select the head or tails to see the contrast.
Conceptual animation · not to scale · molecular counts, arrangement and animation time are illustrative. Leaflet colors distinguish the two sides, not different lipid species.
The structure explains the possibilities. Water-soluble molecules can be trapped in aqueous spaces, while other molecules associate with the membrane. How much a liposome carries also depends on the drug, its surroundings, and how it is prepared.
One bilayer encloses one aqueous interior.
Concentric bilayers have aqueous spaces between them.
A larger particle contains multiple nonconcentric aqueous compartments.
These are simplified cross-sections at an illustrative scale. They distinguish unilamellar, multilamellar and multivesicular structures; they do not describe every type of lipid nanoparticle.
Based on Xiaoming Xu and Diane J. Burgess, Liposomes as Carriers for Controlled Drug Delivery, in Long Acting Injections and Implants (2012). Lipid structure and vesicle formation: sections 11.1.1–11.1.2, pp.197–201. Drug locations: section 11.1.4.3, pp.206–208.
For a multivesicular example, Manna et al., Probing the mechanism of bupivacaine drug release from multivesicular liposomes (Journal of Controlled Release, 2019), studied Exparel. Its nonconcentric aqueous chambers are separated by lipid membranes, giving a honeycomb-like interior. MVLs can be much larger than the unilamellar vesicle shown in the loading model; these architecture drawings are not on a common measured size scale.
The bending-and-closure sequence explains a structural principle. Eliminating exposed edges competes with membrane bending costs, and formation pathways depend on materials and preparation conditions. This animation is not a molecular-dynamics simulation or a prediction of formation time, particle size or loading efficiency.
The representative lipid has two tails. Drug positions show typical environments; particular molecules may interact with several membrane regions. The illustration is newly drawn and does not reproduce a book figure.
Start with a hydrophilic drug present during formation. Explore how vesicle size and lipid concentration change the water enclosed, then record formulations to compare your experiments.
Explore ideal passive aqueous encapsulation for spherical, unilamellar vesicles. This geometric prediction is not a universal maximum loading capacity. Active loading, binding and precipitation can change the outcome. Molecular area and bilayer thickness describe lipid geometry; the controls are not named-lipid presets.
READ THE RESEARCHPredicting hydrophilic drug encapsulation inside unilamellar liposomesXu, Khan & Burgess · 2012 · Appendix A and Table 3Open paperCalculated from the published geometry.
Width and all other inputs held fixed. Gaps indicate inputs outside the supported physical or numerical domain.
Molecular schematic · color distinguishes leaflets, not lipid species. Count and membrane thickness are illustrative, not to scale.
Two lipid leaflets enclose an aqueous compartment.
Explore the usual locations. This changes the explanation, not the calculation.
For passive loading without drug–lipid interaction, water-soluble drug is entrapped in the aqueous core. The model links encapsulation to the volume of that water.
The calculator always predicts passive hydrophilic loading under its assumptions. It does not calculate membrane-associated or active loading.
Related research: lipid–protein interactionsCholesterol, PEG chains and drug molecules are not depicted here: composition and molecular interactions are not resolved by this model.
Compare the fraction of vesicles in each size bin with that bin’s share of the internal aqueous volume. Both are calculated from the same distribution and geometry.
Percent per 10 nm bin. The illustration above shows one characteristic vesicle, not a simulated microscopy image.
The implementation reproduces every Table 3 model prediction to the published two decimal places. Reproducing calculations is distinct from agreement with experiments.
Published phosphate-buffer example; this reference stays fixed when you change the controls.
| Case | Published | Calculated | Experiment* |
|---|
*Values and ± spreads transcribed from Table 3; these are not confidence intervals calculated by this tool.
Keep a formulation, change one variable, and see the difference.
No runs yet. Your records stay in this page until you download them.
For each diameter D, the model calculates lipid demand k = 4π[r² + (r − t)²]/a and inner volume v = 4π(r − t)³/3, where r = D/2. Distribution-weighted totals determine how much water a fixed amount of lipid encloses.
EE = 100 × cNₐ × E[v]/E[k], with explicit molar and volume conversions. Sample volume scales total vesicle count and internal volume, but cancels from efficiency. Like the published model, this treats sample volume as internal plus external water, neglecting the volume occupied by lipid. It is a volume-based approximation to passive encapsulation. Estimated particle count = total lipid molecules / distribution-weighted lipids per vesicle. This assumes all lipid forms the modeled unilamellar vesicles; it is not a measured particle count.
Spherical, unilamellar vesicles; passive aqueous loading; equal internal and external drug concentration at formation; no drug–lipid association; a log-normal number distribution. Active loading, multilamellar systems and interacting drugs require other models.
The reconstruction uses a 1–999 nm, 1 nm grid and the legacy width convention. It rejects material tail truncation and impossible occupied volumes. Input control ranges are prototype exploration bounds, not an experimentally validated design space.
Build a pH gradient, watch neutral drug enter and become protonated, then collapse the gradient to see release. The controls, animation and loading history describe the same experiment.
Independent random walks and symbolic tracers illustrate transport. Marker paths and counts are not molecular dynamics; readouts use the mass-balance calculation.
B + H⁺ ⇌ BH⁺. The neutral form crosses; protonation favors retention.
Illustrative total drug: 100 µM across the modeled aqueous volume. Enclosed aqueous fraction: 1.7%; external/internal volume ratio: about 58. All drug starts outside (101.73 µM), with an empty interior.
Start the experiment to follow the transport.
A fixed-pH, well-mixed, finite-reservoir model for a monoprotic weak base. Only the neutral form permeates. Geometry is fixed at 100 nm outer diameter and a 5 nm bilayer. The illustrative whole-sample aqueous concentration is 100 µM, with 1.7% enclosed aqueous volume. Initial external concentration is corrected for its 98.3% share of the modeled aqueous volume (101.73 µM). No molecular weight, named-drug dose or drug-specific fit is assumed. This experiment is separate from the polydisperse passive-loading model.
Neutral fractions follow Henderson–Hasselbalch. The effective membrane area is taken at the bilayer midpoint. Flux is PA(Cneutral,out − Cneutral,in). The conserved drug fraction approaches equilibrium by an exact exponential update. Playback speed changes display pacing, not model time. Relaxation half-time is the time to halve the remaining gap to equilibrium, not the time to load half the dose. The fixed-pH assumption requires sufficient buffering: real gradients can weaken as drug consumes available protons. Buffer depletion, binding, precipitation, solubility limits and electrochemical effects are excluded. The pH controls illustrate ionization, not formulation compatibility or vesicle stability.
The initial pH pair (4 inside, 7 outside), pKa 8 and concentration scale are teaching parameters for a generic monoprotic weak base, not a reproduction of a particular formulation or the cited study. Neutral permeability (10⁻⁷ m/s) remains illustrative; temperature is not modeled. Buffer capacity, ion exchange, binding and drug precipitation can affect actual loading but are outside this calculation.
The stochastic display paths, counts and transit time are visualization aids, not calibrated MD or microscopic diffusion predictions. The kinetic model is an ideal teaching extension, not a reproduction or drug-specific fit from the cited paper.
Traditional workflows use thin-film hydration or ethanol injection, often followed by extrusion to control vesicle size. Continuous manufacturing can make production more reproducible and easier to scale, while enabling real-time monitoring and control.
Here, a coaxial coflow turbulent jet illustrates that direction. Watch the lipid–ethanol region spread into water, then move closer to see how solvent exchange changes the environment around the lipids.
Lipids enter dissolved in ethanol. As the feed moves downstream, ethanol spreads radially into the surrounding aqueous flow and water diffuses inward.
Select a stage to move between the process and molecular views.
Continuous flow · schematic molecular motion and display time. No CFD or molecular-dynamics solution.
The formation sequence is based on Costa et al., online 2015/issue 2016, particularly Figures 1 and 12 and the discussion of pro-liposome growth. Solvent-quality changes, lipid association/precipitation, ethanol exchange and reorganization are separated in this animation to make an overlapping process understandable. It is a proposed mechanism and depends on materials and conditions. CCTJ means coaxial coflow turbulent jet; this is not an opposing-stream impinging jet.
VandenBerg et al. (2025) places mixing technologies within the broader move toward continuous nanomaterial manufacturing. This visualization shows a formation step, not a complete continuous production line. Ethanol injection can also be adapted to continuous operation; traditional preparation methods and continuous processing are not mutually exclusive. The stream envelope, radial spreading and moving streaks are schematic, not a computed velocity, diffusion or concentration field. The molecular view illustrates rapid lipid organization that shelters hydrophobic tails as ethanol redistributes, followed by closed bilayer growth and stabilization. These separated stages explain overlapping processes; they do not establish a unique atomistic pathway. Motion, timing, molecular counts and radii are illustrative, not calibrated MD/CFD or a particle-size prediction.
Liposome formation using a coaxial turbulent jet in co-flow · Learning from the future: continuous manufacturing