Passive Diffusion and Membrane Permeability

A single molecule of oxygen slips across a red blood cell membrane in about a microsecond. It does so unassisted, unpaid, and unstoppable. A sodium ion, sitting a few nanometers away in the same bathing plasma, faces a different fate. It could wait an eternity — many minutes to hours — for the same crossing. Even then, it would need a protein channel built specifically to shepherd it through. Both are downhill journeys. Both are driven by the same thermodynamic imperative to erase a concentration gradient. Yet their rates differ by many orders of magnitude. Understanding why is the subject of this article. It covers the physics of passive diffusion across the lipid bilayer. It also covers the permeability coefficients that quantify how the chemical identity of a solute — its size, its polarity, its capacity to shed or keep waters of hydration — dictates how fast it can cross a membrane with no protein help at all. This matters because the plasma membrane is not a uniform wall. It is built from the amphipathic phospholipids described in The Lipid Bilayer and Membrane Asymmetry. Its interior is an oily, nearly anhydrous slab. That slab is sandwiched between two polar headgroup zones. Any solute wanting to get from one aqueous compartment to another faces a challenge. It must, for an instant, abandon the hydrogen-bonded water it is accustomed to. It must dissolve — however briefly — in hydrocarbon. That partitioning step is the single greatest kinetic barrier in the entire crossing. It is what separates gases and small uncharged molecules, which slip through readily, from ions and sugars. The bilayer excludes ions and sugars almost completely. Cells must move these solutes using dedicated protein machines. The big picture, then, is a story of energetics translated into rate. Diffusion itself — net movement down a concentration gradient — is entropically driven. It needs no metabolic energy. It is a spontaneous process with $\Delta G < 0$ as long as a gradient exists. But the rate at which that spontaneous process proceeds across a membrane depends on a kinetic bottleneck. That bottleneck is the free energy of transferring the solute from water into the low-dielectric bilayer interior. This free energy of partitioning is captured experimentally by the oil/water (or octanol/water) partition coefficient. It combines with a diffusion coefficient inside the membrane to give the permeability coefficient $P$ . This is the single number that, multiplied by a concentration difference, tells you the net flux across the membrane in moles per unit area per unit time. This article builds that machinery from scratch. It starts from Fick's first law and the thermodynamics of chemical potential. It derives the solubility-diffusion model of membrane permeability. It shows exactly how partition coefficients, diffusion coefficients, and membrane thickness combine into $P$ . It works through three fully quantitative examples. These are a classic gas exchange calculation, a drug-design application using the Overton correlation, and a red-blood-cell urea-transport puzzle. The puzzle only resolves once you admit facilitated diffusion. It closes by placing passive diffusion in context. That context is the protein-mediated transport processes — Membrane Transporters and Active Transport and Ion Channels and Selective Permeability — that biology invented. Biology invented these precisely because passive diffusion alone cannot move the solutes that life actually needs moved. It cannot do so on the timescales life actually needs.

Key takeaways

Passive diffusion is downhill (spontaneous, $\Delta G<0$ ), needs no ATP, and stops when concentrations equalize.
The rate-limiting step for crossing a bilayer is partitioning from water into the hydrophobic core, not diffusion within the membrane itself.
The permeability coefficient $P = \dfrac{K_p D_{mem}}{\ell}$ combines the partition coefficient, the membrane diffusion coefficient, and thickness into one measurable rate constant.
Flux follows $J = P(C_{out}-C_{in})$ — a first-order, Fick's-law relationship with $P$ in cm/s.
Facilitated diffusion (via channels/carriers) still runs downhill and is thermodynamically passive, but replaces the lipid-partitioning bottleneck with a protein-mediated pathway, giving far higher rates for polar solutes.
Permeability spans nine orders of magnitude across O₂, water, glycerol, and Na⁺ — a direct consequence of the Overton correlation between lipid solubility and crossing rate.
Key idea: how fast a solute crosses a membrane is set not by the size of the concentration gradient but by how happily that solute dissolves in oil.

Where this lives today

The solubility-diffusion framework derived here is not a historical curiosity; it is load-bearing infrastructure in modern pharmacology, physiology, and biotechnology. In drug design, the entire discipline of computing $\log P$ (octanol/water partition coefficient) and applying Lipinski's Rule of Five is a direct, quantitative descendant of the Overton correlation: medicinal chemists tune lipophilicity to hit a sweet spot where a candidate molecule is lipid-soluble enough to cross gut epithelium and the blood–brain barrier passively, yet not so lipophilic that it becomes insoluble in blood plasma or gets sequestered in fat depots. Anesthetic potency itself, historically the very phenomenon Overton studied, correlates with lipid solubility via the Meyer–Overton rule, still cited in modern anesthesiology even as mechanistic understanding has moved toward specific ion channel targets. In clinical physiology, gas exchange in the lung and placenta relies entirely on the enormous permeability of O₂ and CO₂ predicted here — these gases need no transporter, and impaired diffusion (as in pulmonary fibrosis, where the diffusion distance $\ell$ effectively increases) directly reduces $P$ and thus gas flux, exactly as the $P = K_pD_{mem}/\ell$ relationship predicts. In renal and erythrocyte physiology, the urea transporter story worked through above is not a toy example: UT-A and UT-B transporters are essential for the kidney's urine-concentrating countercurrent mechanism, and mutations or pharmacological inhibition of urea transporters are active targets for a new class of diuretics. In biotechnology and drug delivery, liposomal and lipid-nanoparticle formulations (the same technology underlying mRNA vaccines) are engineered around exactly this physics: cargo that must stay trapped inside a vesicle needs a very low $P$ (large, charged, or hydrophilic payload), while helper lipids and fusogenic components are chosen to transiently increase local membrane permeability at the point of cellular delivery. Finally, in basic cell biology, the entire logic of why cells evolved Ion Channels and Selective Permeability and Membrane Transporters and Active Transport rests on the quantitative gulf demonstrated in the worked examples above: passive diffusion alone, however thermodynamically inevitable, is simply too slow by many orders of magnitude to run a nervous system, a kidney, or a beating heart, and evolution built specific protein machines precisely to overcome the partitioning bottleneck this article derives.

Ask the tutor: How would the permeability equation change if the membrane itself carried a fixed surface charge that altered local ion concentrations near the bilayer surface?

The physical picture

Picture the bilayer as three stacked environments: an outer aqueous phase, a thin oily slab roughly 3–4 nm thick, and an inner aqueous phase. A solute molecule executing passive diffusion must undergo three sequential steps — partition into the membrane at one interface, diffuse across the hydrophobic core, and partition back out into water at the other interface. Because the aqueous diffusion layers on either side are usually fast compared to the membrane step (water is not very viscous and the unstirred layers are thin), the overall rate is almost always set by what happens inside the bilayer. The energetic cost of that first step — moving from bulk water into the alkane-like interior — is the crux of the whole story. Transferring a charged ion from water (dielectric constant $\varepsilon \approx 80$ ) into a hydrocarbon environment ( $\varepsilon \approx 2$ ) costs enormous desolvation free energy, on the order of 200–300 kJ/mol for a monovalent ion by simple Born-model electrostatics. Transferring a small nonpolar gas like O₂ or CO₂ costs almost nothing — these molecules are barely different from a small hydrocarbon cavity. Polar but uncharged solutes (water, urea, glycerol) sit in between: they can partially satisfy their hydrogen-bonding needs during the crossing, but not fully, so the barrier is intermediate. This partitioning free energy shows up experimentally as the oil/water partition coefficient $K_p$ , and — remarkably — a plot of the log of membrane permeability against the log of the olive-oil/water (or octanol/water) partition coefficient for dozens of chemically unrelated small molecules falls on nearly a single straight line. This is the Overton correlation (Charles Ernest Overton, 1899), and it is one of the oldest and most robust empirical laws in membrane biophysics: lipid solubility, not molecular size, controls the crossing rate for small nonelectrolytes.

Figure

The three-step partition-diffuse-partition pathway across a lipid bilayer, showing the free-energy profile a solute experiences

The height of that central hump — the free energy of transfer from water into the bilayer core — is what the partition coefficient $K_p$ measures directly, since $K_p = e^{-\Delta G_{transfer}/RT}$ . A molecule with a favorable (small or even negative) $\Delta G_{transfer}$ , like a steroid hormone or a dissolved gas, barely notices the hump. An ion faces a hump so large that essentially zero molecules have enough thermal energy to surmount it spontaneously — hence the near-total impermeability of the bilayer to Na⁺, K⁺, Cl⁻, and Ca²⁺, and the biological necessity of channels and pumps.

Figure

Comparative permeability coefficients for six classes of solute, log scale, cm/s

This nine-order-of-magnitude spread is the single most important quantitative fact in membrane physiology. It explains why cells need no machinery to breathe (gases diffuse freely) but require elaborate channel and pump proteins merely to maintain a resting membrane potential (ions cannot cross unaided at all).

The core ideas

Fick's first law describes diffusive flux down a concentration gradient in a homogeneous medium:

J = -D\frac{dC}{dx}

where:

$J$ = flux, moles crossing unit area per unit time (mol cm⁻² s⁻¹)
$D$ = diffusion coefficient of the solute in that medium (cm² s⁻¹)
$dC/dx$ = concentration gradient along the direction of transport

Reading: flux is proportional to how steep the concentration gradient is, with $D$ as the proportionality constant; the negative sign says flux runs from high to low concentration, exactly as intuition demands. For a thin membrane of thickness $\ell$ separating two well-mixed compartments, the gradient inside the membrane is approximately linear, and Fick's law integrates to the permeability equation:

J = P\,(C_{out} - C_{in})

where:

$P$ = permeability coefficient of the solute for this membrane (cm s⁻¹)
$C_{out}, C_{in}$ = solute concentrations in the two aqueous compartments (mol cm⁻³, or any consistent concentration unit)

Reading: this is Fick's law repackaged for a membrane whose internal structure you cannot easily probe — all the messy internal physics (partitioning, internal diffusion, thickness) is folded into the single empirical constant $P$ , measured in cm/s, which converts a bulk concentration difference directly into a flux. The thermodynamic driving force behind all of this is the chemical potential difference:

\Delta \mu = RT \ln\frac{C_{in}}{C_{out}}

where:

$R$ = gas constant, $8.314\ \text{J mol}^{-1}\text{K}^{-1}$
$T$ = absolute temperature (K)
$C_{in}/C_{out}$ = ratio of concentrations on the two sides

Reading: as long as $C_{in} \neq C_{out}$ , $\Delta\mu \neq 0$ and diffusion is spontaneous in the direction that shrinks $|\Delta\mu|$ ; the process is entropy-driven, requires no metabolic energy input, and runs until $C_{in} = C_{out}$ , at which point $\Delta\mu = 0$ and net flux ceases (dynamic equilibrium, not static — molecules still cross, but equally in both directions).

Derivation AThe solubility–diffusion model: deriving P from Kp, Dmem, and ℓ

Goal. Show explicitly how the single empirical permeability coefficient $P$ decomposes into a partition coefficient, a membrane diffusion coefficient, and a membrane thickness — the mechanistic content behind Fick's law applied to a bilayer.

Step 1 — set up the three-compartment picture. Treat the system as three regions in series along the x-axis: aqueous compartment "out" ( $0 < x$ , concentration $C_{out}$ ), the membrane itself ( $0 \le x \le \ell$ , internal concentration profile $c(x)$ ), and aqueous compartment "in" ( $x > \ell$ , concentration $C_{in}$ ). Assume the aqueous unstirred layers equilibrate fast compared to the membrane crossing (a good approximation for thin bilayers and stirred bulk solution), so we need only solve diffusion inside the membrane at steady state.

Step 2 — apply the partition equilibrium at each interface. At the outer interface, local equilibrium between bulk water and the membrane surface layer means the membrane-phase concentration just inside the bilayer, $c(0^+)$ , relates to the bulk aqueous concentration by the partition coefficient:

c(0^+) = K_p\, C_{out}

Symmetrically at the inner face,

c(\ell^-) = K_p\, C_{in}

Here $K_p$ is the equilibrium constant for solute transfer from water into the membrane phase (operationally, the oil/water or octanol/water partition coefficient), $K_p = c_{membrane}/c_{water}$ at equilibrium.

Step 3 — solve steady-state Fick's law inside the membrane. At steady state, the flux $J$ is uniform across the membrane (no accumulation), so Fick's law with the internal diffusion coefficient $D_{mem}$ gives a linear concentration profile inside the bilayer:

J = -D_{mem}\frac{dc}{dx} = D_{mem}\,\frac{c(0^+) - c(\ell^-)}{\ell}

This is just the steady-state solution of $d^2c/dx^2 = 0$ (Fick's second law with $\partial c/\partial t = 0$ ) subject to the two boundary concentrations from Step 2.

Step 4 — substitute the partitioning boundary conditions.

J = \frac{D_{mem}\,K_p}{\ell}\left(C_{out} - C_{in}\right)

Step 5 — identify the permeability coefficient. Comparing to the defining relation $J = P(C_{out}-C_{in})$ term by term gives the central result:

P = \frac{K_p D_{mem}}{\ell}

Step 6 — sanity-check the limiting behavior. As $K_p \to 0$ (a solute that cannot dissolve in the membrane at all — an ion, essentially), $P \to 0$ : no crossing, regardless of how fast the solute might diffuse once inside. As $\ell \to \infty$ (an unrealistically thick membrane), $P \to 0$ : distance kills flux, exactly as ordinary diffusion demands. And if $K_p = 1$ and $D_{mem} = D_{water}$ (no partitioning penalty, no internal viscosity penalty), $P$ reduces to $D/\ell$ , the permeability of a simple homogeneous slab — recovering plain Fick's law with no membrane-specific correction at all, as it must.

Go deeper · Derivation B · 1 minFrom Fickian flux to first-order kinetics: the exponential approach to equilibrium across a membrane-bound compartment

Derivation BFrom Fickian flux to first-order kinetics: the exponential approach to equilibrium across a membrane-bound compartment

Goal. Many real experiments (dye efflux from vesicles, solute equilibration across erythrocyte membranes) do not hold $C_{out}$ and $C_{in}$ fixed; instead a finite-volume compartment approaches equilibrium with its surroundings. Derive the resulting time-course from the permeability equation.

Step 1 — write the mass-balance ODE for a vesicle of volume $V$ and surface area $A$ . Let $C_{in}(t)$ be the internal solute concentration in a vesicle bathed in a much larger external reservoir held fixed at $C_{out}$ . The rate of change of the amount of solute inside equals flux times area:

V\frac{dC_{in}}{dt} = J \cdot A = P A \left(C_{out} - C_{in}(t)\right)

Step 2 — nondimensionalize and separate variables. Define $\Delta(t) = C_{out} - C_{in}(t)$ , so $d\Delta/dt = -dC_{in}/dt$ . Substituting,

-V\frac{d\Delta}{dt} = PA\,\Delta

\frac{d\Delta}{\Delta} = -\frac{PA}{V}\,dt

Step 3 — integrate from $t=0$ (where $\Delta = \Delta_0 = C_{out}-C_{in}(0)$ ) to time $t$ .

\ln\frac{\Delta(t)}{\Delta_0} = -\frac{PA}{V}\,t

\Delta(t) = \Delta_0\, e^{-t/\tau}, \qquad \tau = \frac{V}{PA}

Step 4 — interpret $\tau$ and extract $P$ from data. $\tau$ is the characteristic equilibration time constant; a plot of $\ln[\Delta(t)/\Delta_0]$ versus $t$ is a straight line of slope $-PA/V$ , and since $A$ and $V$ are known geometric quantities for a vesicle or cell of known radius, $P$ is obtained directly from the slope. This is, in essence, the operational definition used in every stopped-flow light-scattering or osmotic-swelling permeability assay: measure how fast a suspension of vesicles equilibrates after a concentration jump, extract $\tau$ , and back out $P = V/(A\tau)$ .

Step 5 — limiting case. As $\tau \to 0$ ( $P \to \infty$ , an infinitely leaky membrane, or a channel-mediated pathway), $\Delta(t) \to 0$ almost instantly — the compartment equilibrates as fast as the reservoir concentration can be defined. As $\tau \to \infty$ ( $P \to 0$ , an ion crossing a pure bilayer), $\Delta(t) \approx \Delta_0$ for all practically accessible times — the compartment behaves as if sealed, which is exactly the impermeable-membrane limit that makes a resting membrane potential possible in living cells.

Worked examples

Example 1 — Foundational: comparing the crossing rate of a small gas and a small ion of similar size. Molecular oxygen (O₂) has a permeability coefficient across a typical phospholipid bilayer of about $P_{O_2} \approx 3\ \text{cm/s}$ (physiological temperature, thin bilayer). Sodium ion, of comparable ionic radius, has $P_{Na^+} \approx 10^{-12}$ to $10^{-10}$ cm/s across the same class of pure lipid bilayer (in the complete absence of channels or pumps); take the middle estimate $P_{Na^+} \approx 10^{-11}\ \text{cm/s}$ for this comparison. Suppose both species face an identical, modest concentration difference across a red-blood-cell-sized membrane: $C_{out} - C_{in} = 5\times 10^{-6}\ \text{mol/cm}^3$ (5 mM). Flux of O₂:

J_{O_2} = P_{O_2}(C_{out}-C_{in}) = (3\ \text{cm/s})(5\times10^{-6}\ \text{mol/cm}^3) = 1.5\times10^{-5}\ \text{mol cm}^{-2}\text{s}^{-1}

Flux of Na⁺ under the identical gradient:

J_{Na^+} = (10^{-11}\ \text{cm/s})(5\times10^{-6}\ \text{mol/cm}^3) = 5\times10^{-17}\ \text{mol cm}^{-2}\text{s}^{-1}

The ratio of fluxes:

\frac{J_{O_2}}{J_{Na^+}} = \frac{1.5\times10^{-5}}{5\times10^{-17}} = 3\times10^{11}

Oxygen crosses roughly 300 billion times faster than sodium under an identical driving force, even though both molecules are roughly the same physical size and both face the same thermodynamic pull. The entire difference is attributable to $K_p$ : O₂ is essentially as soluble in lipid as in water ( $K_p \sim 1$ –5), while Na⁺'s hydration shell and formal charge make $K_p$ for the bare bilayer vanishingly small (effectively $10^{-9}$ – $10^{-10}$ once Born desolvation energetics are included). This single example is the thermodynamic reason ion channels exist: no amount of concentration gradient can substitute for the four-orders-of-magnitude-per-decade improvement a channel protein provides by replacing lipid-partitioning with a water-lined pore. Limiting check: if the concentration difference were doubled to 10 mM, both fluxes would double proportionally (linear in $\Delta C$ by construction), but the ratio $J_{O_2}/J_{Na^+}$ would remain exactly $3\times10^{11}$ — the permeability ratio is a property of the membrane and solutes, not of the applied gradient, exactly as the derivation $J = P\Delta C$ demands.

Example 2 — Applied: using the Overton correlation to estimate a drug's oral membrane permeability from its octanol/water partition coefficient. Suppose a candidate drug molecule has a measured octanol/water partition coefficient $\log_{10} K_{ow} = 1.8$ (a fairly lipophilic small molecule, similar to many CNS-active drugs), a diffusion coefficient in the membrane interior estimated at $D_{mem} \approx 2\times10^{-7}\ \text{cm}^2/\text{s}$ (typical for a molecule of moderate size moving through the ordered acyl-chain region), and a bilayer thickness $\ell = 4\times10^{-7}\ \text{cm}$ (4 nm). First convert $\log K_{ow}$ to a linear partition coefficient:

K_{ow} = 10^{1.8} = 63.1

Assume, as the Overton correlation licenses for small nonelectrolytes, that $K_p \approx K_{ow}$ (octanol is a reasonable surrogate for the bilayer core). Apply the solubility-diffusion result:

P = \frac{K_p D_{mem}}{\ell} = \frac{(63.1)(2\times10^{-7}\ \text{cm}^2/\text{s})}{4\times10^{-7}\ \text{cm}}

P = \frac{1.262\times10^{-5}}{4\times10^{-7}}\ \text{cm/s} = 31.6\ \text{cm/s}

This is a very high permeability — consistent with the drug being well-absorbed passively across gut epithelium without needing a transporter, exactly the design goal in medicinal chemistry when optimizing $\log P$ for oral bioavailability (the well-known "Rule of Five" territory, $\log P$ roughly 0–5, aims squarely at this regime). Now contrast with a more polar analog with $\log K_{ow} = -0.5$ (same $D_{mem}$ and $\ell$ , for direct comparison):

K_p = 10^{-0.5} = 0.316

P = \frac{(0.316)(2\times10^{-7})}{4\times10^{-7}} = 0.158\ \text{cm/s}

A drop of 2.3 log units in $\log K_{ow}$ (from 1.8 to $-0.5$ ) produces a 200-fold drop in $P$ (31.6 down to 0.158 cm/s) — the hallmark linear log-log Overton relationship, since taking logs of $P = K_p D_{mem}/\ell$ with fixed $D_{mem}, \ell$ gives $\log P = \log K_{ow} + \text{const}$ , a slope of exactly 1 on a log-log plot when $D_{mem}$ is roughly constant across a congeneric series. Limiting check: as $\log K_{ow} \to -\infty$ (an extremely polar or charged species, $K_p \to 0$ ), $P \to 0$ — the molecule cannot cross the bilayer passively at all, which is exactly the regime where real drugs (and biological polar solutes) require transporter proteins, foreshadowing Membrane Transporters and Active Transport.

Example 3 — Exam/competition level: reconciling the "urea paradox" in red blood cells and inferring the presence of a facilitative pathway. Urea is a small, polar, uncharged molecule ( $M_r = 60$ ), and by the Overton correlation alone (using its modest octanol/water $\log K_{ow} \approx -1.6$ ) one predicts a pure-lipid permeability on the order of

K_p = 10^{-1.6} = 0.025

Using representative bilayer parameters $D_{mem} \approx 5\times10^{-7}\ \text{cm}^2/\text{s}$ and $\ell = 4\times10^{-7}\ \text{cm}$ :

P_{predicted} = \frac{(0.025)(5\times10^{-7})}{4\times10^{-7}} = 0.031\ \text{cm/s}

But this "predicted" value is for a synthetic pure-lipid bilayer with no proteins — call it $P_{lipid} \approx 3\times10^{-6}$ cm/s once more carefully corrected experimental partition data for urea in phosphatidylcholine bilayers are used (urea's actual bilayer $K_p$ is far below simple octanol estimates because urea, despite being uncharged, forms extensive hydrogen bonds that octanol only partially mimics; measured pure-lipid urea permeabilities are typically $10^{-6}$ – $10^{-7}$ cm/s). Now the experimental fact: intact human red blood cell membranes show urea permeability of $P_{measured} \approx 4\times10^{-4}\ \text{cm/s}$ — roughly 100- to 1000-fold higher than the pure-lipid prediction, and moreover this measured permeability is almost completely abolished by phloretin, a specific inhibitor, while being essentially unaffected by moderate changes in bulk lipid composition.

The reasoning an exam question wants: compute the ratio,

\frac{P_{measured}}{P_{lipid}} = \frac{4\times10^{-4}}{3\times10^{-6}} \approx 133

A two-order-of-magnitude discrepancy between predicted passive-lipid permeability and measured whole-membrane permeability, abolished by a specific chemical inhibitor, is the classic experimental signature of a facilitated diffusion pathway — in this case the erythrocyte urea transporter UT-B, a member of the aquaporin superfamily's cousin channel family. The transporter provides a water-filled or partially hydrated conduction pathway that bypasses the lipid-partitioning bottleneck entirely: instead of desolvating into an oily interior ( $K_p \ll 1$ ), urea moves through a protein pore lined with polar residues that substitute for the water it would otherwise need to shed. The thermodynamics are unchanged — urea still moves strictly downhill, $\Delta G < 0$ , and no ATP is consumed — but the kinetic pathway, and hence the rate constant $P$ , is entirely different, because the rate-limiting partitioning step of the solubility-diffusion model has been replaced by a rate-limiting protein-conformational or pore-transit step. Limiting check: if UT-B is fully and specifically blocked, $P_{measured} \to P_{lipid} \approx 3\times10^{-6}$ cm/s, recovering the pure bilayer prediction — exactly what phloretin-inhibition experiments observe, confirming that the two pathways (lipid diffusion and protein-facilitated diffusion) are additive, parallel routes for the same net downhill process.

Limiting cases, scope & common traps

Regime	Behavior of $P$	Physical reason
Small nonpolar gas (O₂, CO₂, N₂)	Very high, $P\sim1$ –5 cm/s	$K_p \approx 1$ , negligible desolvation penalty
Small polar uncharged solute (H₂O, urea, glycerol)	Intermediate, $P\sim10^{-3}$ – $10^{-7}$ cm/s	Partial desolvation penalty; some H-bonds broken
Monovalent ion (Na⁺, K⁺, Cl⁻)	Extremely low, $P\sim10^{-12}$ – $10^{-10}$ cm/s	Enormous Born desolvation penalty, $K_p\to0$
Steroid/lipophilic drug	High, $P\sim1$ –30 cm/s	Large $K_p$ (favorable partitioning)
Facilitated (channel/carrier-assisted) solute	Can rival or exceed gas-level $P$	Protein bypasses lipid-partitioning bottleneck entirely
$\ell \to \infty$ (unphysical, thick membrane)	$P \to 0$	Diffusion distance penalty dominates
$C_{out} = C_{in}$	$J = 0$ regardless of $P$	No driving force; dynamic equilibrium, not zero crossing rate per molecule

Common trapCommon misconception #1 — "Passive diffusion means the same thing as unassisted, protein-free diffusion."

Not quite. Passive is a thermodynamic statement: no metabolic energy is consumed, and net transport runs down the electrochemical gradient ( $\Delta G<0$ ). Facilitated diffusion through a channel or carrier protein is still fully passive in this thermodynamic sense — glucose entering a cell through GLUT transporters moves downhill and needs no ATP — even though it is emphatically not "simple diffusion through naked lipid." The distinction that matters is passive vs. active (energy input required), not protein-free vs. protein-mediated. Only when a solute is moved against its electrochemical gradient does the process require coupled free energy input, becoming active transport, the subject of Membrane Transporters and Active Transport.

Common trapCommon misconception #2 — "A bigger concentration gradient always means faster crossing for any solute."

A larger $\Delta C$ always increases flux linearly ( $J = P\Delta C$ ), but it does nothing to change the permeability coefficient $P$ itself, which is fixed by the solute's chemistry and the membrane's composition. Doubling the extracellular Na⁺ concentration does not make sodium cross a pure lipid bilayer meaningfully faster in any biologically relevant sense — $P_{Na^+}$ stays pinned near $10^{-11}$ cm/s regardless, because the rate-limiting partition coefficient $K_p$ is a property of the ion and the membrane, not of the gradient. This is why gradients alone cannot substitute for a channel: no physiologically achievable concentration difference compensates for nine orders of magnitude of unfavorable partitioning.

The chemist behind it

The Overton correlation is named for Ernest Overton, whose systematic study of anesthetic and dye permeation across cell membranes at the turn of the twentieth century established, well before anyone knew what a phospholipid bilayer actually looked like, that lipid solubility — not molecular size, not charge, not any other single property in isolation — was the dominant predictor of how fast a substance crossed a cell membrane. Overton worked mostly with plant and animal cells directly, testing hundreds of small organic compounds for their rate of penetration and correlating it against their solubility in olive oil, a wonderfully simple experimental proxy that turned out to capture the essential physics of the lipid bilayer core decades before Gorter and Grendel's 1925 bilayer model or Singer and Nicolson's 1972 fluid-mosaic picture gave that oily interior a name and a structure. The modern quantitative treatment of membrane permeability as a partition-diffusion problem — the derivation carried through above — synthesizes Overton's empirical correlation with the rigorous diffusion physics of Adolf Fick (1855) and the thermodynamic framework of chemical potential from Josiah Willard Gibbs. It is a beautiful example of biochemistry absorbing physical chemistry wholesale: the same Fick's-law mathematics that describes heat conduction or the spreading of a dye in a beaker, applied without modification to the crossing of a lipid bilayer, once the key insight — that partitioning into the hydrophobic core is itself the rate-determining chemical equilibrium — is properly incorporated as a boundary condition.

Origins · 1 minCharles Ernest Overton and how this was discovered

Charles Ernest Overton

1865–1933 · Swiss-British physiologist and botanist, University of Zürich, whose systematic 1899 monograph on narcosis and cell permeability first correlated membrane crossing rates with olive-oil/water partition coefficients across hundreds of compounds — establishing the lipoid theory of membrane permeability decades before the bilayer's molecular structure was known

History

- 1855 — Adolf Fick formulates the laws of diffusion, providing the mathematical backbone (Fick's first and second laws) later applied to membranes.

- 1899 — Ernest Overton publishes his monograph correlating narcotic potency and membrane permeability with olive-oil solubility, founding the lipoid theory of the cell membrane.

- 1925 — Evert Gorter and François Grendel extract erythrocyte lipids and estimate their spread area, providing the first strong physical evidence for a lipid bilayer as the membrane's structural basis.

- 1943 — Kenneth Cole and others begin quantitative permeability and impedance measurements on giant squid axon membranes, connecting ion permeability to electrical excitability.

- 1972 — Jonathan Singer and Garth Nicolson propose the fluid-mosaic model, embedding Overton's permeability data and the solubility–diffusion mechanism into a coherent picture of membrane structure with mobile proteins in a fluid bilayer.

Is passive diffusion the same as simple diffusion through naked lipid?

No. Passive diffusion is a thermodynamic category: transport that runs down an electrochemical gradient without consuming metabolic energy. Simple diffusion through the lipid bilayer itself is one mechanism that satisfies this criterion, but facilitated diffusion through channels or carrier proteins is equally passive thermodynamically, since it too moves solute downhill with no ATP hydrolysis. The distinction that actually matters biologically is passive versus active transport — whether energy is spent to move something against its gradient — not whether a protein happens to be involved in the pathway.

Why is the partition coefficient, rather than the diffusion coefficient inside the membrane, usually the dominant term controlling permeability?

Diffusion coefficients for small molecules inside a fluid bilayer core vary relatively modestly across different solutes, typically within one to two orders of magnitude, because the bilayer interior has a fairly uniform viscosity for molecules of similar size. Partition coefficients, by contrast, vary across ten or more orders of magnitude between, say, a nonpolar gas and a monovalent ion, because they encode an exponential Boltzmann factor of a free energy difference, $K_p = e^{-\Delta G_{transfer}/RT}$ , and that free energy difference (dominated by electrostatic desolvation) itself spans hundreds of kJ/mol across different solute chemistries. Since $P = K_pD_{mem}/\ell$ and $K_p$ 's range utterly swamps $D_{mem}$ 's range, partitioning is overwhelmingly the rate-controlling term.

Does passive diffusion ever stop completely?

Not in the sense of molecules ceasing to cross; it stops in the sense of net flux going to zero. At equilibrium, $C_{out} = C_{in}$ , so $J = P(C_{out}-C_{in}) = 0$ , but individual molecules continue to cross the membrane in both directions at equal, opposing rates — this is a dynamic equilibrium, not a static one. If you introduced an isotopically labeled tracer at equal bulk concentration on both sides, you would still observe a nonzero unidirectional flux of labeled molecules in each direction, even though net (unlabeled-basis) flux is zero.

Can passive diffusion ever move a solute against its concentration gradient?

No — this is precisely what distinguishes it from active transport. By the chemical potential expression $\Delta\mu = RT\ln(C_{in}/C_{out})$ , any net passive flux must act to reduce $|\Delta\mu|$ , meaning net movement always proceeds from the higher-potential side to the lower-potential side. Moving a solute uphill requires an external free energy source — ATP hydrolysis, a coupled ion gradient, or light — which is the defining feature of active transport mechanisms and the topic developed in Membrane Transporters and Active Transport.

Why do ion channels achieve much higher permeability for ions than the bare lipid bilayer, if both are ultimately "passive"?

An ion channel replaces the physically demanding step of full desolvation into a hydrocarbon environment with a narrow, water-lined, sometimes partially charged pore that allows the ion to retain a substantial fraction of its favorable electrostatic interactions throughout the crossing (interactions with pore-lining carbonyl oxygens or side chains substitute for lost waters of hydration). This dramatically lowers the effective free energy barrier compared to a bare bilayer, even though no metabolic energy is spent — the channel simply provides a lower-energy kinetic pathway for a thermodynamically identical downhill process, which is developed further in Ion Channels and Selective Permeability.

How is the permeability coefficient P actually measured experimentally?

Two common strategies: (1) stopped-flow or osmotic-swelling assays on vesicles or cells, where a rapid concentration jump is imposed and the exponential relaxation of internal concentration (tracked by light scattering, fluorescence, or radiotracer content) is fit to $\Delta(t)=\Delta_0e^{-t/\tau}$ , giving $P=V/(A\tau)$ ; and (2) steady-state flux measurements across planar bilayers or epithelial monolayers mounted in a diffusion (Ussing) chamber, where the flux $J$ is measured directly (often via radiolabeled solute appearing on the far side) for a known, maintained $\Delta C$ , giving $P=J/\Delta C$ directly from the defining relation.

Does temperature affect membrane permeability, and if so how?

Yes, strongly. Raising temperature increases $D_{mem}$ (faster molecular motion, lower effective viscosity) and can also shift the bilayer from an ordered gel phase to a disordered fluid phase, further increasing $D_{mem}$ near the phase transition temperature; it can also modestly change $K_p$ by altering the effective dielectric environment of the acyl-chain region. Because $P=K_pD_{mem}/\ell$ , both effects generally push $P$ upward with increasing temperature, which is why membrane permeability assays are always performed at carefully controlled, physiologically relevant temperatures, and why hyperthermia can transiently increase membrane leakiness in cells.

Passive Diffusion and Membrane Permeability

Where this lives today

The physical picture

The core ideas

Worked examples

Limiting cases, scope & common traps

The chemist behind it

Before this

Where this leads