Skip to content

Composition-Partition Assay

Assay — instantiates Controlled Demixing and Domain Formation

Measures how the constituents actually distributed among the separated phases, turning a demixing into purity, recovery, and residual numbers.

A Composition-Partition Assay answers the question a separation exists to answer: where did everything actually go? Once a mixture has split into phases, the assay samples each phase and measures its composition — how much of each constituent ended up in which region — and from those measurements computes the numbers that decide whether the separation succeeded: the purity of each phase, the recovery (yield) of the target into its intended phase, and the residual left behind where it shouldn't be. Its defining focus is compositional, not structural: it does not care what the domains look like, only what they are made of. It converts a visible split into an accounting of constituents against the composition each phase was supposed to reach.

Example

A biomanufacturing team recovers a therapeutic protein using an aqueous two-phase system — a PEG-rich top phase and a salt-rich bottom phase that separate spontaneously and carry different molecules to different sides. Separation is easy to see; the question is whether the protein went where wanted and left the contaminants behind. The composition-partition assay samples both phases and quantifies each: the protein's concentration top versus bottom gives its partition coefficient; the fraction of total protein now in the top phase is the recovery; the host-cell proteins and DNA still riding along in the top phase are the residual impurities setting its purity.

The readout might be: partition coefficient ≈ 8 (protein strongly favors the PEG phase), recovery ≈ 90%, but host-protein clearance only ≈ 70% — purity short of target. That single line reframes the next move. The domains formed fine; the partitioning is the problem, so the fix is a composition or pH adjustment that pushes contaminants to the other phase — not anything about domain size or interface. The assay is what makes "it separated" into "it separated well enough, on these axes."

How it works

The assay treats each phase as a sample to be analyzed and each constituent as an account to be balanced. It measures concentrations in every phase (by whatever analytical method the constituents demand — spectroscopy, chromatography, titration, mass), then closes a mass balance to yield partition coefficients, per-phase purity, target recovery, and the residuals. What distinguishes it from a morphology measurement is that its entire output lives in composition space: the same perfectly-shaped domains score differently depending only on what partitioned into them, and the assay reports against the composition targets each phase was designed to hit.

Tuning parameters

  • Analyte set — which constituents are actually measured. Tracking only the target is cheap but blind to the impurities that define purity; a full panel is complete but costly.
  • Sampling location and timing — where within a phase, and how soon after separation, samples are drawn. Interfacial or too-early samples cross-contaminate and distort the partition numbers.
  • Detection limit — how low a residual the method can see. A loose limit passes trace contamination invisibly; a tight one catches it but demands more sensitive analytics.
  • Mass-balance closure tolerance — how much unaccounted-for material is allowed before the result is rejected. Tight closure catches sampling and measurement errors; loose closure hides them.
  • Target definition — the purity, recovery, and partition thresholds each phase is graded against, which turn raw concentrations into a pass/fail verdict.

When it helps, and when it misleads

Its strength is making the chemical success of a separation explicit and quantitative — distinguishing a demixing that merely looks clean from one that actually put the right constituents in the right phase — and localizing failures to composition rather than structure, which points at composition- and interaction-based fixes.

It misleads mainly through sampling artifacts: draw a sample that carries entrained droplets of the other phase, or sample before partitioning has equilibrated, and the purity and partition numbers are quietly wrong in a flattering or alarming direction.[1] A tidy partition coefficient can also imply an equilibrium the process never actually reached, and an assay that measures only the target constituent will happily report high recovery while missing the impurities that make the phase unusable. The classic misuse is choosing the analyte panel and detection limit after seeing a clean-looking product, so the assay can only confirm success. The discipline is to fix the analyte set and limits from the purity requirement beforehand, close the mass balance, and confirm that partitioning has reached steady state before trusting the numbers.

How it implements the components

  • phase_purity_and_yield_metric — its primary output: the purity of each phase, the recovery of the target, and the residuals, all as measured quantities.
  • phase_composition_targets — it quantifies each phase's realized composition against the target composition it was meant to reach, making the target a checkable spec rather than an aspiration.

It does not measure domain size, shape, or connectivity — that is Domain-Morphology Imaging — nor does it watch the process approach its phase boundary; that is Phase-Boundary Monitor. The assay reports what the phases contain, not what they look like or when they formed.

Notes

Purity and recovery usually trade against each other: pushing recovery by drawing more of a phase drags in more of the boundary layer and its contaminants, lowering purity. The assay's real value is quantifying where on that trade-off a given separation sits, so the target can be set deliberately rather than discovered by disappointment.

References

[1] The partition (distribution) coefficient — the equilibrium ratio of a constituent's concentration between two coexisting phases — is the standard descriptor of how a species divides itself in a two-phase system. It is an equilibrium quantity, so an assay run before partitioning has settled, or on a cross-contaminated sample, measures something other than the coefficient it reports. ↩