Column Chromatography: A Race Decided by Partition Coefficient

A race decided by where a molecule would rather be

Chromatography separates a mixture by exploiting the fact that different compounds have different preferences for two phases in contact: a stationary phase packed into the column (silica, alumina, a bonded organic film) and a mobile phase flowing through it (a solvent, or a carrier gas in GC). Every molecule constantly equilibrates between sticking to the stationary phase and travelling with the mobile phase. A compound that prefers the stationary phase spends more time stuck and travels slowly; a compound that prefers the mobile phase gets swept along quickly. Run the mixture long enough and initially identical starting positions separate into distinct, spatially resolved bands.

The partition coefficient

The preference is quantified by the partition coefficient K, the equilibrium ratio of a compound's concentration in the stationary phase to its concentration in the mobile phase:

K = C_stationary / C_mobile

A high K means the compound is retained strongly and elutes late; a low K means it barely interacts with the stationary phase and elutes early, close to the void time t0 (the time it takes the unretained mobile phase itself to pass through). The retention factor k' (sometimes written k, unrelated to K despite the similar letter) is what is actually measured from a chromatogram — the extra time a compound spends relative to t0:

k' = (tR - t0) / t0     // tR: the compound's observed retention time

Why bands sharpen into peaks: plate theory

A column is modelled, for the purposes of predicting peak shape, as a stack of many discrete theoretical plates — imaginary equilibrium stages, borrowed directly from distillation theory. At each plate the compound re-equilibrates between the two phases; more plates means finer, more repeated equilibration steps, which sharpens the eluting peak and improves the column's ability to tell two similar compounds apart. The van Deemter equation relates plate height H (a smaller H means a more efficient column) to the mobile-phase flow velocity u:

H = A + B/u + C*u
// A: eddy diffusion (packing irregularity, flow path variability)
// B: longitudinal molecular diffusion along the column axis
// C: mass-transfer resistance between the two phases
// H has a minimum at an intermediate u — too slow lets diffusion blur bands,
// too fast doesn't give the compound time to equilibrate

Resolution: how far apart is far enough

Two peaks that are close in retention time but broad will overlap into one blob; two peaks far apart but sharp will separate cleanly. Resolution Rs combines both effects — the gap between retention times relative to the average peak width — and is the actual metric chromatographers optimise, not retention time alone:

Rs = 2 * (tR2 - tR1) / (w1 + w2)     // w1, w2: baseline peak widths

A resolution of 1.5 is the conventional threshold for "baseline separated" — essentially no overlap between the two peaks. Improving resolution means either pulling retention times further apart (changing the stationary phase's chemistry, or the mobile phase's polarity, to shift K for one compound relative to another) or narrowing the peaks themselves (a longer column, smaller particle size, or an optimal flow rate near the van Deemter minimum).

Elution order is chemistry, not size

A common misconception is that chromatography sorts by molecular size the way a sieve does. Column chromatography sorts by affinity — how strongly a compound's chemistry matches the stationary phase versus the mobile phase. In normal-phase silica chromatography, more polar compounds bind the polar stationary phase more strongly and elute later; in reversed-phase (a nonpolar bonded phase with a polar mobile phase, the default in most modern HPLC), the order flips and more polar compounds elute first. Size matters only indirectly, through how it affects diffusion and interaction surface area — the sorting variable is always the partition coefficient.

Frequently asked questions

Try it live

Everything above runs in your browser — open Column Chromatography and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.