Electroosmotic Flow: The Flat Plug That Drives Capillary Electrophoresis

What EOF Is: A Charged Wall That Pumps the Whole Liquid

Electroosmotic flow is the bulk motion of a liquid past a stationary charged surface under an applied electric field. In fused-silica capillary electrophoresis (CE), that surface is the inner wall of the glass. Silica is a network of siloxane (Si–O–Si) and terminal silanol (Si–OH) groups; the silanols are weak acids (surface pKₐ values spread roughly from 4 to 9, with an operationally significant deprotonation setting in above pH ≈ 3–4). Above about pH 3, a growing fraction ionize to silanolate (Si–O⁻), leaving the wall negatively charged.

That fixed negative charge is screened by mobile counter-ions from the buffer — an excess of cations — arranged in an electrical double layer (EDL). When you apply a longitudinal field E (say, 300–600 V·cm⁻¹), the field pushes on the net cationic charge in the diffuse part of that layer. Those solvated cations drag the surrounding water with them by viscous coupling, and because momentum is transmitted across the whole bore, the entire column of liquid moves. At high pH the wall is very negative, the mobile layer is cation-rich, and the bulk flow heads toward the cathode.

The key contrast with a syringe pump is where the force is applied. A pump pushes at the ends, so wall friction bends the profile into a parabola. EOF applies its force in a nanometer-thin sheath at the wall, and the rest of the liquid is simply carried along uniformly. The result is a flat, plug-like velocity profile — the single most important property of EOF for separation science.

The Double Layer and the Helmholtz–Smoluchowski Equation

To make this quantitative we need the structure of the EDL. Immediately at the wall is the compact Stern layer of specifically adsorbed and tightly held counter-ions. Beyond it is the diffuse (Gouy–Chapman) layer, where the potential decays roughly exponentially into the bulk. The characteristic thickness of that diffuse layer is the Debye length, κ⁻¹, where κ = √(2·N_A·e²·I / ε·k_B·T). For a 10 mM 1:1 buffer at 25 °C, κ⁻¹ ≈ 3 nm; for 100 mM it shrinks to ~1 nm. Compare that to a 50 µm bore: the double layer is ~10⁴ times thinner than the channel.

The plane where the fluid begins to move relative to the wall — the shear plane — sits near the Stern/diffuse boundary, and the electrostatic potential there is the zeta potential, ζ. Solving the coupled Poisson–Boltzmann and Navier–Stokes equations for the thin-double-layer limit (κ·r ≫ 1) gives the celebrated Helmholtz–Smoluchowski equation:

• μ_eo = ε·ζ / η, so that v_eo = μ_eo·E = (ε·ζ/η)·E

Here ε is the buffer permittivity (ε = ε_r·ε₀, with ε_r ≈ 78 for water), η the viscosity, and ζ the zeta potential (negative for bare silica). Plugging a negative ζ into μ_eo = ε·ζ/η yields a negative value, yet the resulting EOF velocity is directed toward the cathode — the mobile diffuse-layer charge being carried is cationic, opposite to the wall charge. In practice μ_eo for bare silica is quoted as a positive magnitude with flow toward the cathode, so don't read the raw formula sign as the physical direction. The remarkable feature: μ_eo depends on ζ, ε, η and not on the capillary radius, provided the bore is much larger than κ⁻¹. That radius-independence is why EOF pumps a 5 µm and a 100 µm capillary at the same velocity for a given ζ and E — impossible for pressure-driven flow.

Why the Flat Plug Delivers a Million Plates

Separation efficiency in any migration technique is throttled by band broadening. In pressure-driven flow the parabolic profile is a disaster: the center streamline runs twice the mean velocity while fluid at the wall is stationary, smearing a solute band. This is Taylor–Aris dispersion, and its variance grows as σ² ∝ (r²·u²·t)/D — it worsens with the square of the channel radius. It is precisely why open-tubular LC never took off in wide bores.

EOF sidesteps this entirely. Because the driving force lives in the nanometer-thin double layer and the bulk moves as a rigid plug, there is essentially no radial velocity gradient across the analytical cross-section. Remove that term and the only remaining broadening in an idealized CE run is longitudinal molecular diffusion. The plate count then follows the Giddings/Jorgenson result:

• N = μ_app·V / (2·D)

where μ_app is the apparent mobility (electrophoretic + electroosmotic), V the applied voltage, and D the analyte diffusion coefficient. Notice N scales with the total voltage, not the length — a direct license to crank the field. James Jorgenson and Krynn Lukacs demonstrated exactly this in their landmark 1981 Analytical Chemistry paper, achieving efficiencies of several hundred thousand theoretical plates in glass capillaries and effectively founding modern CE. For slowly diffusing macro-ions like DNA or proteins (small D), N can exceed 10⁶.

A Worked Example: Timing an Analyte in a Real CE Run

Take a routine run: a fused-silica capillary of total length L_t = 50 cm, effective length to the detector L_d = 40 cm, inner diameter 50 µm, filled with 20 mM borate buffer at pH 9.3, and V = 25 kV applied. The field is E = V/L_t = 25000/50 = 500 V·cm⁻¹.

Suppose we measure ζ ≈ −60 mV for the wall. Using μ_eo = ε·ζ/η with ε = 78 × 8.85 × 10⁻¹² F·m⁻¹ and η = 1.0 × 10⁻³ Pa·s, we get μ_eo ≈ (6.91 × 10⁻¹⁰ × 0.060)/1.0 × 10⁻³ ≈ 4.1 × 10⁻⁸ m²·V⁻¹·s⁻¹ = 4.1 × 10⁻⁴ cm²·V⁻¹·s⁻¹, a textbook value. The EOF velocity is v_eo = μ_eo·E = 4.1 × 10⁻⁴ × 500 ≈ 0.20 cm·s⁻¹ = 2 mm·s⁻¹.

Now put a small cation with its own electrophoretic mobility μ_ep = +2.0 × 10⁻⁴ cm²·V⁻¹·s⁻¹ into the mix. It rides with the EOF (both head to the cathode), so μ_app = μ_eo + μ_ep = 6.1 × 10⁻⁴, giving v_app ≈ 0.305 cm·s⁻¹ and a migration time t = L_d/v_app = 40/0.305 ≈ 131 s. A neutral marker (μ_ep = 0) rides EOF alone and elutes at t₀ = 40/0.20 = 200 s. Crucially, even an anion whose intrinsic electrophoresis pulls it toward the anode still reaches the cathodic detector — provided |μ_ep| < |μ_eo| — because the EOF overwhelms it. That is how a single injection resolves cations, neutrals, and anions in one window.

Limits, Subtleties, and Controlling EOF

EOF is powerful but temperamental, because ζ depends on everything that touches the wall. Practitioners exploit and fight this in equal measure:

• pH. Below pH ~2–3 the silanols are protonated, ζ → 0, and EOF nearly vanishes; above pH 8 the wall is fully ionized and EOF is maximal. A sigmoidal μ_eo-vs-pH curve is the signature.
• Ionic strength. Raising buffer concentration compresses κ⁻¹ and lowers |ζ|, so μ_eo drops (roughly |ζ| ∝ 1/√I at low potentials in the diffuse-layer regime; at high |ζ| the Gouy–Chapman scaling breaks down as nonlinear Poisson–Boltzmann and Stern-layer effects take over). High ionic strength also raises Joule heating.
• Wall coatings. Neutral coatings (linear polyacrylamide, PEG) suppress EOF to near zero for reproducible protein work; cationic surfactants like cetyltrimethylammonium bromide (CTAB) adsorb, flip the wall charge positive, reverse ζ, and run EOF toward the anode.
• Temperature. η falls ~2–3% per °C, so μ_eo rises with T; uncontrolled Joule heating from the current both raises T non-uniformly and re-introduces radial gradients, quietly destroying the flat profile.

The deepest limitation is irreproducibility: adsorption of proteins or trace metals onto the wall shifts ζ run-to-run, drifting migration times. This is why serious quantitative CE uses internal standards, rigorous rinses (NaOH regenerates fresh silanolate), or covalent coatings. And the flat-plug idealization breaks down when the double layer is not thin relative to the bore — in nanochannels where κ·r ~ 1, the Smoluchowski limit fails and one must use the fuller Poisson–Boltzmann treatment, where overlapping double layers give ion-selective, non-plug behavior.

History and Reach Beyond the Capillary

The phenomenon is older than most people assume. Ferdinand Friedrich Reuss first reported electroosmosis in 1809 in Moscow, watching water migrate through a porous plug (variously recorded as wet clay or sand) under an applied voltage. Hermann von Helmholtz laid the double-layer groundwork in 1879, and Marian Smoluchowski completed the quantitative theory around 1903, giving us the equation that still bears both names. The technique lay largely dormant until Stellan Hjertén demonstrated electrophoresis in narrow tubes in 1967 and Jorgenson and Lukacs delivered high-efficiency CE in fused silica in 1981.

The applications now sprawl far past the analytical bench. CE and its variants — micellar electrokinetic chromatography (MEKC), capillary gel electrophoresis, capillary isoelectric focusing — became workhorses of pharmaceutical QC, chiral analysis, and clinical assays. Capillary array electrophoresis with laser-induced fluorescence sequenced the human genome: the ABI 3700 and MegaBACE instruments that finished the Human Genome Project were parallel CE machines, sizing DNA fragments through sieving polymers driven by exactly this electrokinetic transport.

The same physics runs lab-on-a-chip microfluidics, where EOF is prized as a pump with no moving parts: apply electrodes and the buffer flows, valves and all realized by switching fields. It even appears in the earth sciences and civil engineering — electroosmotic dewatering and soil stabilization pass current through wet clay to move pore water, and electrokinetic remediation drags charged contaminants out of soils. The humble ionized silanol, in other words, quietly pumps everything from a diagnostic assay to a construction site.

Electroosmotic (plug) flow versus pressure-driven (Poiseuille) flow in a capillary

Property	Electroosmotic flow	Pressure-driven flow
Velocity profile	Flat/plug (uniform across bore)	Parabolic (Hagen–Poiseuille)
Driving force	Body force on double-layer charge (bulk)	Pressure gradient at ends
Band broadening	Diffusion-limited; minimal added dispersion	Large — Taylor–Aris dispersion ∝ r²
Scaling with radius	Velocity ~ independent of r (for r ≫ κ⁻¹)	Velocity ∝ r² for fixed ΔP
Control knob	pH, ionic strength, ζ, wall coating, field E	Applied pressure ΔP