Aqueous Two-Phase Extraction (ATPE)

• ATPE is defined by the formation of two immiscible, water-rich phases from mixtures like PEG/dextran or PEG–salt systems, enabling selective separation.
• The method leverages controlled variables such as polymer incompatibility, acid/salt concentration, and temperature to fine-tune partitioning mechanisms and interfacial structure.
• Applications include DNA localization, SWCNT chirality sorting, nanoparticle extraction, and bacterial partitioning, highlighting its versatility in research and industry.

Aqueous two-phase extraction (ATPE) is the use of aqueous two-phase systems as a separation medium: two immiscible water-rich phases, generated by mixtures of mutually immiscible water-soluble species, are brought into contact so that solutes, particles, or cells partition selectively between them. In the systems considered here, ATPE includes polymer–polymer mixtures such as polyethylene glycol (PEG) and dextran (Dex), polymer–salt mixtures such as PEG–trisodium citrate, ionic-liquid-based acidic aqueous biphasic solutions such as tributyltetradecylphosphonium chloride $\mathrm{P}_{44414}\mathrm{Cl}$/acid/water, and surfactant-assisted PEG/dextran separations of single-wall carbon nanotubes (SWCNTs); the corresponding targets include nucleic acids, bacteria, nanoparticles, metallic ions, and nanotube chiral species (Scacchi et al., 2023, Gande et al., 2022, Meyer et al., 2022, Defillet et al., 2022).

1. System classes and phase architecture

ATPE is implemented in several distinct classes of aqueous biphasic media. In polymer–polymer systems, PEG and dextran demix into a PEG-rich phase and a Dex-rich phase, both remaining overwhelmingly aqueous. In polymer–salt systems such as PEG 6000–trisodium citrate, equilibrium tie-line end phases are likewise water-based, with a PEG-rich top phase and a citrate-rich bottom phase. In ionic-liquid acidic aqueous biphasic solutions, mixing $\mathrm{P}_{44414}\mathrm{Cl}$, acid, and water above roughly $60\ \mathrm{wt}\%$ water yields two immiscible aqueous phases: an IL-rich upper phase and an acid-rich or salt-rich lower phase. In each case, extraction is achieved by preferential distribution of the target between two water-based environments rather than between aqueous and organic solvents (Scacchi et al., 2023, Gande et al., 2022, Meyer et al., 2022).

Phase architecture is not limited to planar bulk interfaces. In confined emulsified settings, gelatin-rich and PEG-rich aqueous phases can coexist inside an outer oil phase, producing core–shell or crescent morphologies relevant to structured microparticles. In PEG/dextran systems, one phase can also be made ferrofluidic by loading maghemite nanoparticles, so that a dextran-rich dense phase is magnetic and a PEG-rich light phase is nearly non-magnetic. These variants do not alter the defining feature of ATPE—the coexistence of two aqueous phases—but they change how interfaces are shaped, stabilized, and externally actuated (Hester et al., 2023, Rigoni et al., 2020).

2. Thermodynamic bases of phase separation

In coarse-grained equilibrium theory for polymeric ATPS, phase coexistence is controlled by effective interactions among the phase-forming species. For two polymers $a$ and $b$, the classical density functional treatment uses integrated interaction strengths and a $\chi$-like criterion,

$$\chi = 2\hat{V}_{ab} - (\hat{V}_{aa}+\hat{V}_{bb}),$$

with phase separation at fixed volume when $\chi > 0$. Coexisting phases satisfy equality of chemical potentials and pressure, $\mu_i^{\rm I}=\mu_i^{\rm II}$ and $P^{\rm I}=P^{\rm II}$, together with mass balance. In field-theoretic descriptions of PEG/Dex demixing, the order parameter $\mathrm{P}_{44414}\mathrm{Cl}$0 identifies Dex-rich and PEG-rich compositions and the late-stage domain size follows $\mathrm{P}_{44414}\mathrm{Cl}$1 (Varma et al., 17 Sep 2025, Yang et al., 22 Nov 2025).

Not all ATPE media are composition-driven in the same way. The $\mathrm{P}_{44414}\mathrm{Cl}$2/acid/water acidic aqueous biphasic system exhibits a Lower Solution Critical Temperature (LCST): below the LCST the solution is homogeneous, whereas above the LCST it separates into two aqueous phases. The biphasic region expands as temperature increases, more acid lowers the LCST and broadens the two-phase domain, and low-acid transition temperatures reported for the model systems are about $\mathrm{P}_{44414}\mathrm{Cl}$3 for DCl and $\mathrm{P}_{44414}\mathrm{Cl}$4 for $\mathrm{P}_{44414}\mathrm{Cl}$5. Nitrate has a much stronger phase-separating effect, described as sufficient for “a single drop” to turn the IL/water solution biphasic. In gelatin–PEG compound droplets, thermally induced phase separation is modeled as a post-quench evolution following cooling to $\mathrm{P}_{44414}\mathrm{Cl}$6, with equilibrium morphologies determined by interfacial energies and Young force balance at triple-contact lines (Meyer et al., 2022, Hester et al., 2023).

These results establish that ATPE phase formation can be tuned by total composition, effective polymer incompatibility, and, in selected formulations, by temperature. They also delimit a common oversimplification: the phase diagram of an acidic aqueous biphasic system cannot be rationalized by ionic strength alone, because specific ion interactions, ion size, and polarizability are implicated in the observed LCST behavior (Meyer et al., 2022).

3. Interfacial structure, micellization, and ultralow interfacial tension

In ionic-liquid acidic aqueous biphasic extraction media, the microscopic picture is explicitly colloidal. $\mathrm{P}_{44414}\mathrm{Cl}$7 behaves like an ionic surfactant and forms spherical micelles in water; at $\mathrm{P}_{44414}\mathrm{Cl}$8, for IL contents below $\mathrm{P}_{44414}\mathrm{Cl}$9, the micellar radius is about $60\ \mathrm{wt}\%$0. Addition of acid or NaCl screens electrostatic repulsion, weakens the SANS correlation peak near $60\ \mathrm{wt}\%$1, and requires a sticky-hard-sphere rather than purely repulsive hard-sphere description. With increasing acid concentration and temperature, the system progresses through stronger “stickiness,” coexisting spheres and cylinders, bicontinuous phases, and very long aggregates before flocculation and macroscopic phase separation. Chloride adsorption in the electrical double layer is exothermic, with an adsorption enthalpy of about $60\ \mathrm{wt}\%$2, and the resulting reduction in effective micelle charge explains the drop in the DLVO barrier that permits aggregation upon heating (Meyer et al., 2022).

Across aqueous two-phase systems more generally, the interface is exceptionally soft. In PEG–trisodium citrate flow experiments, the interfacial tension required to match slug sizes was

$60\ \mathrm{wt}\%$3

about two orders of magnitude below typical organic–aqueous systems. In ferrofluidic PEG/dextran systems, sessile-drop estimates ranged from about $60\ \mathrm{wt}\%$4 down to $60\ \mathrm{wt}\%$5 with dilution, and magnetic-instability-based estimates were of the same order, for example $60\ \mathrm{wt}\%$6–$60\ \mathrm{wt}\%$7 depending on geometry and sample. Such ultralow values compress capillary lengths into the tens-to-hundreds of micrometers and make aqueous interfaces highly deformable under confinement, flow, and external fields (Gande et al., 2022, Rigoni et al., 2020).

This interfacial softness is a central physical distinction of ATPE. It underlies the ease of slug formation or suppression in channels, the deformability of interfaces to bacteria and active colloids, and the feasibility of magnetic normal-field instabilities at wavelengths near $60\ \mathrm{wt}\%$8 rather than the millimeter scale common to conventional ferrofluid interfaces (Rigoni et al., 2020).

4. Partitioning mechanisms and selectivity

For neutral or weakly interacting solutes, equilibrium partitioning can be written in the usual ATPE form,

$60\ \mathrm{wt}\%$9

with the two-phase densities obtained from coexistence conditions. In the classical density functional treatment, strong partitioning is promoted by higher total polymer concentration, stronger affinity of the target for one polymer-rich phase, and smaller solute size. In the corresponding Brownian-dynamics coarse-grained model for PEG/dextran with magnetic nanoparticles, tuning the dextran–particle attraction $a$0 moves the system from modest partitioning to strong dextran-phase enrichment; for example, the fraction of nanoparticles in the dextran phase increases from about $a$1 to $a$2 as $a$3 is increased from 1 to 3, while too strong an attraction produces dense clustering and jamming rather than useful extraction (Varma et al., 17 Sep 2025, Scacchi et al., 2023).

For nucleic acids in PEG/Dex ATPS, the physical origin of partitioning is not exhausted by depletion. In PEG6k/Dex500k mixtures, Asakura–Oosawa estimates predict stronger DNA accumulation under PEG1k conditions, yet experimentally replacing $a$4 PEG6k with $a$5 PEG1k decreases DNA accumulation in Dex-rich droplets. The mechanistic explanation advanced is electrostatic: Dex is slightly more negatively charged than PEG, so the Dex-rich phase accumulates cations, and DNA, as a polyanion, is stabilized in this cation-rich phase. The ion partitioning is described in Donnan form,

$a$6

Experimentally, DNA uptake increases with length from $a$7 to $a$8 and then plateaus, while $a$9 NaCl strongly suppresses localization in Dex-rich droplets. The data therefore argue that Donnan-type ion partitioning is central, and depletion alone cannot fully explain the observed behavior (Sakuta et al., 29 Sep 2025).

For SWCNT ATPE in PEG/dextran, selectivity is instead encoded in the surfactant corona. DOC-coated nanotubes partition according to the hydrophilicity of the SWCNT–surfactant complex, and the chirality-dependent transition between phases is fitted by error functions in the varied control parameter. The transition order shows a periodically modulated dependence on diameter with maxima near $b$0 and $b$1, consistent with close packing of 7 and 8 DOC molecules around the nanotube circumference. Cosurfactants such as SDS and SDBS compete with DOC and shift transition points, but the sorting order itself is not altered. The same framework was used to design predictive two-step separations for specific chiral structures (Defillet et al., 2022).

5. Flowing, confined, and active ATPE

ATPE in channels is governed by the same dimensionless groups used in other two-phase flows,

$b$2

but the ultralow interfacial tension of ATPS shifts the accessible regime boundaries. In a $b$3 ID millichannel fed with pre-equilibrated PEG-rich and citrate-rich phases, three patterns were observed: slug flow, transition flow, and core-annular flow. Slug flow occurs for $b$4; transition occupies roughly $b$5; and core-annular flow appears for $b$6. For equal flow rates, slug formation was observed for $b$7, whereas higher $b$8 favored jetting and core-annular morphologies. Film thickness follows a Han–Shikazono-type correlation and increases with total flow rate. Because microchannels drive $b$9 in such low-$\chi$0 systems, they tend toward parallel or core-annular flow, whereas millichannels preserve access to slug flow at practical throughputs (Gande et al., 2022).

Confined ATPS droplets exhibit additional hydrodynamic selection of morphology. In gelatin–PEG droplets suspended in oil, equilibrium shapes are set by the interfacial energies $\chi$1, $\chi$2, and $\chi$3, but the route to those shapes depends on flow. Cahn–Hilliard–Stokes–Boussinesq simulations show that buoyancy breaks the symmetry of metastable core–shell states and drives the system toward lower-energy crescent morphologies, while pressure-driven shear produces much stronger internal recirculation and a tenfold speedup in particle formation. The result is that neglecting fluid dynamics yields incorrect minimum-energy droplet shapes from mixed initial conditions (Hester et al., 2023).

Active agents convert ATPE from a purely equilibrium separation into a nonequilibrium one. In DEX/PEG systems containing Bacillus subtilis, non-motile bacteria partition exclusively into the DEX-rich phase, but motile bacteria cross the soft DEX/PEG interface and show a DEX-phase fraction that rises from $\chi$4 to $\chi$5 as DEX composition is increased. Optical-tweezer measurements give interfacial restoring forces $\chi$6 from about $\chi$7 to $\chi$8, while bacterial propulsion is estimated as $\chi$9; partitioning is therefore set by competition between propulsive and interfacial forces rather than by thermal activation (Cheon et al., 2024). In Pseudomonas aeruginosa–DEX/PEG systems, wetting with contact angle $$\chi = 2\hat{V}_{ab} - (\hat{V}_{aa}+\hat{V}_{bb}),$$0 and pusher hydrodynamics generate a sequence of nonequilibrium morphologies—self-spinning droplets, elongated droplet chains, branched capillary-like clusters, and highly deformed droplets. Activity suppresses coarsening in the droplet regime through hydrodynamic repulsion but accelerates it when the DEX phase is strongly minority and wetting-mediated attraction dominates (Yang et al., 22 Nov 2025).

6. Applications, design principles, and unresolved problems

The documented applications of ATPE are correspondingly diverse. Ionic-liquid acidic aqueous biphasic solutions are used for metal recycling and are described as having very good extraction efficiency for metallic ions; earlier work cited in that context reported Fe(III) loadings up to about $$\chi = 2\hat{V}_{ab} - (\hat{V}_{aa}+\hat{V}_{bb}),$$1 in HCl/$$\chi = 2\hat{V}_{ab} - (\hat{V}_{aa}+\hat{V}_{bb}),$$2 without hydrolysis. PEG/Dex systems localize DNA in Dex-rich droplets and provide biocompatible environments for live-cell partitioning. PEG/dextran ATP extraction sorts SWCNTs by diameter and chirality in two steps. Polymeric ATPS also partition magnetic nanoparticles, and ferrofluidic variants permit magnetic modulation of interface fluctuations and patterning at length scales far below those usually accessed experimentally. Confined ATPS droplets yield structured microparticles relevant to single-cell analysis, targeted drug delivery, and cell scaffolds (Meyer et al., 2022, Sakuta et al., 29 Sep 2025, Defillet et al., 2022, Scacchi et al., 2023, Rigoni et al., 2020, Hester et al., 2023).

Across these examples, the recurring design variables are polymer incompatibility and total polymer concentration; polymer molecular weight, charge, and effective size; solute–phase affinity; acid type and acid or salt concentration; surfactant composition; interfacial tension and channel size; and, in nonequilibrium settings, active forcing or magnetic fields. Several studies also identify persistent misconceptions. Ionic strength alone is not sufficient to rationalize acidic aqueous biphasic phase diagrams, because specific ion interactions are required (Meyer et al., 2022). Depletion alone cannot explain DNA uptake in Dex-rich droplets, because cation accumulation and Donnan partitioning are central (Sakuta et al., 29 Sep 2025). Hydrodynamic regime maps in ATPS are qualitatively similar to those of organic–aqueous systems, with the principal difference arising from lower interfacial tension rather than from new multiphase-flow physics (Gande et al., 2022). Motile cells do not obey a purely equilibrium partition coefficient, because persistent propulsion alters crossing dynamics at soft interfaces (Cheon et al., 2024).

Open problems remain at every scale. In ionic-liquid AcABS, no single structural model spans the full phase diagram, and higher-acid states contain coexisting spheres, cylinders, bicontinuous structures, and long aggregates. For polymeric ATPS, classical density functional theory remains restricted by charge neutrality, spherical particle assumptions, and equilibrium thermodynamics, leaving explicit ions, nonspherical biomolecules, and kinetics outside the present formulation. Particle-based ATPS simulations capture phase separation, interfacial properties, and affinity-driven partitioning, but still coarse-grain polymers as soft particles and omit hydrodynamic interactions. More generally, the literature repeatedly points to the need to couple equilibrium selectivity, interfacial mechanics, and transport kinetics in a single predictive description of practical ATPE (Meyer et al., 2022, Varma et al., 17 Sep 2025, Scacchi et al., 2023).