Liquid–Liquid Phase Separation of Two Non-Dissolving Liquids—A Mini Review

Abstract

The separation of immiscible liquids is critical in many industrial processes, such as water treatment, different extraction processes, the petroleum industry, food production, and medicine. This work provides an overview of present research on the separation of liquid mixtures. A brief summary of the thermodynamic basis is provided, covering phase equilibrium, phase diagrams, and thermodynamic properties of phases. Additionally, the fundamentals of dispersion, necessary for discussing liquid–liquid separation, are presented. Subsequently, different liquid–liquid separation methods are discussed, highlighting their advantages and limitations. These methods include decanters, coalescers, centrifugal separators, membranes and electro-coalescers for liquid–liquid separation. Phase properties, dispersion formation, and time and space constraints specify the most efficient separation method. Phase recycling is also briefly discussed as a method to reduce the environmental impact of liquid–liquid extraction with subsequent phase separation. In summary, liquid–liquid separation methods are compared and future perspectives of liquid–liquid separation are discussed.

1. Introduction

Separating a liquid–liquid mixture into two distinct phases is a critical component of numerous chemical processes, including industrial wastewater treatment [1,2,3,4], solvent extraction in hydrometallurgy [5,6,7,8,9], extraction in the field of medicine [10,11,12,13,14], separation of oil and water [15,16,17,18], ATPS (aqueous two-phase systems) [19,20,21,22], and extraction of food components [23,24,25,26]. Phase separation can be achieved as an integral part of the process design, e.g., phase separation achieved by liquid–liquid extraction [27]. However, there are cases where an independent (single) separation device is beneficial or required. Standalone separation devices can be gravity decanters [28,29,30], coalescers [29,31,32,33,34], centrifuges [29,35,36,37], and hydrocyclones [38,39,40,41,42,43]. Besides these widespread separation devices, recently membranes [44,45,46,47,48] and electro-coalescers [49,50,51] are becoming popular for liquid–liquid separation.

Liquid–liquid separation is the process of separating two immiscible liquids from each other. It is a challenging process as the two liquids can have similar densities, viscosities, and surface tensions, making it difficult to separate them effectively. Additionally, the separation process can be complicated by the presence of impurities, such as solids or other liquids that can interfere with the separation process or reduce the efficiency of the separation. There are several methods for liquid–liquid separation that are discussed in this work. However, the effectiveness of these methods varies depending on the specific properties of the liquids being separated and the desired level of separation.

The review is structured into five sections. The initial section discusses the significance, practical uses, and rationale behind liquid separation. The Section 2 summarizes the thermodynamic basis, presenting phase equilibrium and transport properties, as well as phase diagrams. Additionally, this section covers the fundamentals of dispersion and dispersed drop size, which are important for describing the stability and separation process of mixtures [52,53]. Section 3 focuses on exploring various separation equipment, presenting their theoretical fundamentals and specific examples of each individual equipment. Section 4 is dedicated to recycling of phase-forming components, which is becoming more important due to environmental and economic issues. The final section compares and summarizes the discussed separation equipment and gives an outlook on future prospects of liquid separation.

2. Thermodynamic Basis

Liquid–liquid separation depends primarily on the separated components’ thermodynamic equilibrium. Knowledge of the thermodynamic background and relationships between phases is essential, especially when selecting the proper separation device, deciding how many stages the process needs, or evaluating the interphase mass transfer and separation efficiency rate. The following sub-sections discuss the thermodynamic basis for immiscible liquid phases.

2.1. Phase Equilibrium and Activity Coefficient

Two phases are at equilibrium when the total Gibbs energy is at minimum. In other words, two non-reacting phases are at equilibrium when the chemical potential of each component is equal in both phases [54,55]:

µ_i^I = µ_i^II, i = 1,2, …, n,

(1)

where µ_i represent the chemical potential of component i in phases I and II.

Phase separation in liquid–liquid equilibrium is a complex phenomenon and can be affected by different forces and interactions [56,57,58]. Forces and interactions that play a role in phase separation can be intermolecular forces, temperature and pressure, kinetics, entropy, and enthalpy. The intermolecular forces between the molecules in each phase can influence the phase separation process. For example, suppose the intermolecular forces between the molecules in one phase are stronger than those in the other phase. In that case, the system tends to separate into two phases with different compositions [59,60]. Temperature and pressure can affect the intermolecular forces and/or the entropy of the system, leading to changes in the phase behavior [61,62,63]. The kinetics of the phase separation process can also play a role. For example, the rate at which the two phases form can be influenced by factors such as the initial concentration of the components, the size of the system, and the presence of other molecules or particles in the system [64,65]. In liquid–liquid phase separation systems, the enthalpy change (ΔH) and the entropy change (ΔS) can play an essential role in determining the phase behavior of the system [66,67,68]. This change can be either positive or negative depending on the specific properties of the liquids involved [69,70,71]. Le Chatelier’s Principle [72] offers a highly effective approach to comprehend and manage chemical equilibria. For closed systems at constant temperature and pressure, the following correlation is valid:

ΔG = ΔH−TΔS

(2)

If the system is also at equilibrium, ΔG equals zero leading to the following expression:

ΔH = TΔS

(3)

Since T is always positive, the signs of ΔH and ΔS must be the same for a system in equilibrium. The entropy reduces (ΔS < 0) for exothermic phase mixing systems at equilibrium resulting in enhanced phase separation, while the entropy increases (ΔS > 0) for endothermic phase mixing systems hindering phase separation [73,74]. In general, the entropy change is the driving force for liquid–liquid phase separation, while the enthalpy affects the separation process [67]. If the enthalpy change of the liquid mixture is high, it can require more energy to separate the liquids because of the stronger intermolecular forces. On the other hand, if the enthalpy change is low, the separation can be easier since the intermolecular forces are weaker. The specific values of the entropy and enthalpy changes for a given liquid–liquid phase separation depend on the properties of the involved liquids and the separation conditions.

For two phases at the same temperature and pressure, the phase equilibrium for species is typically described using mole fractions and activity coefficients and is given by [75,76]:

y_iγ_i^I = x_iγ_i^II, i = 1,2, …, n,

(4)

where y_i and x_i represent the mole fractions of component i in phases I and II, respectively, and γ_i is the activity coefficient of component i. Mole fraction is a measure of the relative amount of each species in the mixture, expressed as the number of moles of a particular species divided by the total number of moles in the mixture [77,78]. Activity coefficients can be defined as measurement of a deviation of a mixture from ideal behavior [79,80]. Activity coefficients are typically approximated with mathematical models. These models can be divided into correlative and predictive models. The most prominent correlative models are the Wilson [81], the NRTL [82], and the UNIQUAC [83] models. Common predictive models are the UNIFAC [84], the UNIFAC-LBY [85], the UNIFAC-DMD [86], and the NIST modified UNIFAC (UNIFAC-NIST) [87] models.

2.2. Phase Diagrams

Phase equilibria can be presented via phase diagrams. Phase diagrams are characterized by monophasic regions (homogeneous) and multiphasic regions (heterogeneous) and depend on the type of equilibria, e.g., liquid–liquid or solid–liquid. Common diagram types are TP (temperature–pressure), Txy (temperature–fraction), and Pxy (pressure–fraction). The choice of suitable phase diagrams for a given system depends on the degrees of freedom defined by Gibbs’ phase rule. Furthermore, the experimental viability of data and the intended application influence the choice of phase diagram type (TP, Txy, Pxy) for a particular system. For instance, for evaluating the temperature effect on the volatile compounds contained in oils, the Txy phase diagram is most appropriate [88].

Lower critical solution temperature (LCST) is the critical temperature below which the components of a mixture are miscible in all proportions, and is not commonly observed in phase diagrams [89]. LCST behavior is typically displayed for mixtures involving hydrogen-bonding compounds such as amine, ketone, or etheric alcohol in combination with water [90,91,92]. Several varieties of “water+glycol ether” are known to display LCST behavior [92]. In these systems, hydrogen bonding results in complete miscibility below LCST. As the temperature rises, the increasing thermal energy disturbs the hydrogen bonding, and hydrophobic interactions prevail, causing partial miscibility at temperatures above the LCST. Figure 1a shows an example of a system exhibiting only LCST, e.g., “ethylene glycol+triethylamine” [54]. Figure 1b depicts a binary system that demonstrates both an upper critical solution temperature (UCST) and a lower critical solution temperature (LCST). An example for such a system is the “water+2-butoxyethanol” [93] binary system, and it displays partial miscibility at temperatures ranging from LCST to UCST. Typically, as the temperature increases, most mixtures become increasingly soluble in each other, which indicates UCST behavior.

Multiphase systems can have more than two components, e.g., a mixture of water, oil, and surfactants (microemulsion) [96,97], known as ternary systems. The ternary and pseudo-ternary systems can be [75] one of the following:

• type I—one binary pair with restricted miscibility;
• type II—two binary pairs with restricted miscibility.

Figure 1c shows a triangular diagram of a type I system. “Heptane+toluene+sulfolane” is an example for such a system, where the “heptane+sulfolane” binary pair exhibits partial miscibility [96]. The boundary line that distinguishes the area where two liquid phases emerge is known as a binodal curve [88]. Lines ab, cd, and ef (tie lines) connect equilibrium compositions for each phase. The tie line is the line that describes the exact composition of the liquid phases in equilibrium for a given temperature and pressure. At the plait point, the tie lines intersect, and the length of the tie line approaches zero. This point occurs on the binodal curve when both liquid phases reach the same composition.

Figure 1d shows a triangular phase diagram for type II systems. Numerous immiscible liquid systems demonstrate a critical solution temperature at which the system ceases to divide into two liquid phases [95]. Type II systems change to type I systems when exceeding the critical temperature of the solute and extraction-solvent binary system T_BS. Above the critical temperature of binary system T_AS (including the feed solvent and extraction-solvent), the system reached miscibility. These systems can also demonstrate a temperature below which a critical solution point occurs, making the phase soluble.

Merchuk et al. [98] investigated the determination of equilibrium binodal lines, tie lines, and phase inversion points. The Merchuk approach involves plotting the values of the equilibrium tie lines on a graph where the axes are defined by mole fractions of two components. As mentioned, tie line connect the equilibrium points of two coexisting phases at a given temperature and pressure. In order to identify the binodal line, Merchuk et al. plotted the tie lines at a constant temperature and pressure for different compositions. The binodal line is determined by identifying the endpoints of these tie lines, which represent the critical composition at which the two coexisting phases become identical. However, some ambiguity exists where the continuity of the phases is influenced not only by the mixture composition but also by fluid dynamics. To acquire information regarding the phase composition of one system, a set of equations has to be solved following the study of Merchuk et al. [98]. The mass share of a two-phase system (Y-X system) relative to the binodal curves can be correlated with the following equation:

$$\mathrm{Y}=\mathrm{A}·{\mathrm{e}}^{\mathrm{B}·{\mathrm{X}}^{0.5}+\mathrm{C}·{\mathrm{X}}^3}$$

(5)

where A, B, and C are binodal curve parameters determined from the regression of experimental data [99]. Overall, the Merchuk approach is a useful tool to determine the phase behavior of liquid–liquid systems, particularly for systems that exhibit non-ideal behavior and for multi-component systems. Several recent studies have followed the Merchuk approach to determine the binodal curve in their studies [99,100,101,102].

2.3. Liquid Substance (Thermodynamic) Data

The separation of liquid–liquid phases depends on the equilibrium between phases and transport properties. Physical properties having significant impact on transport behavior are interfacial tension, liquid density, and viscosity. These properties impact how the liquid flows and develops, and potential drops coalescence. Therefore, these factors are crucial in determining the performance of liquid–liquid phase separators. Extensive liquid density and viscosity data databases exist in the literature, e.g., the collection of transport properties by Wypych and Flick [103,104]. Estimating density and viscosity is possible [105], but it is best to obtain these values from experimental measurements [106]. Estimating density from the equation of state and viscosity from the molecular theory is achievable by using different correlations such as gas law equation, deviations from ideal-gas law, equation of van der Waals, and other equations. Such ideal-gas law extensions helped develop molecular theory but, more importantly, provided the framework for correlating physical properties of fluid [105]. One common example of an equation of state is the Peng–Robinson equation, which is often used to model the behavior of hydrocarbon mixtures. Given the composition, temperature, and pressure of a fluid, the Peng–Robinson equation can be used to estimate the density of the fluid [107].

Interfacial tension values can be derived from two types of methods. The first type determines interfacial tensions by measuring the shape, contact angle, or volume of a drop suspended in a liquid. These methods include the following:

• the pendant drop method (a heavy liquid drop is suspended from a vertically mounted capillary tube that is submerged in the light liquid) [108,109];
• the sessile drop method (a heavy liquid drop rests on a plate that is submerged in the light liquid) [110,111,112];
• the spinning drop method (one liquid drop is held in suspension within a tube contacting second liquid and rotating) [113].

The sessile drop method is beneficial for measuring interfacial tension in the presence of surfactants or macromolecules. On the other hand, the spinning drop method is ideal for measuring low interfacial tensions [114].

The second method type is based on the adherence of a ring or any other shape to the liquid. Two methods can be distinguished here:

• the du Noüy’s method (measuring the force required to detach a ring of wire from the liquid interface) [115];
• the Wilhelmy method (measuring the force required to detach a plate of glass or platinum foil from the liquid interface) [116].

The Wilhelmy plate method is used to measure interfacial tension when surfactants are present, making this method superior to the du Noüy ring method [117]. In the presence of surfactants, the du Noüy ring method is not suitable due to changes in the wetting characteristics of the ring caused by adsorption of surfactant molecules. As a result, the contact angles on the inside and outside of the ring may differ [118,119].

An approximate interfacial tension value can be obtained based on the maximum drop size that can maintain its shape in a quiescent dispersion without any agitation. The drop size can be determined by examining a photograph of the dispersion under these terms. Using this approach, the balance between interfacial tension and buoyancy forces can be used to approximate the interfacial tension [75]:

σ~d_max²Δρg,

(6)

where d_max is the maximum drop diameter.

Antonov’s rule [120] can also be utilized to estimate the surface tension as it suggests that the interfacial tension between two liquids is roughly equivalent to the difference in their liquid–air surface tensions measured under the same conditions [121].

2.4. Liquid–Liquid Mixture Fundamentals

Emulsion refers to a type of dispersed systems comprising two immiscible liquids, where drops of one liquid (known as the dispersed phase) are spread evenly throughout another liquid (known as the continuous phase) [122]. Droplets can re-coalesce, resulting in simultaneous drop breakup and coalescence processes during emulsion formation [123]. The typical design of liquid–liquid separation devices aims to produce suspended drops of one liquid within the other liquid, rather than creating liquid films. Holdup (or the volume fraction of the dispersed phase) within the device is defined as

ϕ_d = V_d/V_f,

(7)

where V_d is the volume of the dispersed phase, and V_f is the total fluid volume. The total fluid volume in the device is the volume that remains after deducting the volume of any internal components, such as trays, packing, or impellers. The Sauter mean diameter d₃₂ refers to an average diameter based on the ratio of volume to surface area:

$${\mathrm{d}}_{32}=\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{N}}_{\mathrm{i}}{\mathrm{d}}_{\mathrm{i}}^3}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{N}}_{\mathrm{i}}{\mathrm{d}}_{\mathrm{i}}^2},$$

(8)

where N_i is the amount of drops with diameter d_i. The Sauter mean diameter is directly related to holdup and interfacial area (spherical shape is assumed). Therefore, it is often used to characterize liquid–liquid and gas–liquid dispersions [124]. It can be estimated from the total dispersed volume divided by the total interfacial area [125]:

d₃₂ = 6εϕ_d/a,

(9)

where a refers to interfacial area per unit volume, while ε refers to the void fraction within the device, i.e., the fraction of internal volume (not considering any packing, trays, etc.).

Dispersed Phase

It is essential to identify the dispersed phase and the continuous phase to predict the drop size. Selker and Sleicher [126] proposed a set of guidelines that employ volume rations, density, and viscosity to predict the dispersed phase. An expression to determine the dispersed phase is given by [127]:

Χ = (Q_L/Q_H)[(ρ_Lµ_H)/(ρ_Hµ_L)]^0.3,

(10)

The volume is denoted by Q, while phase properties density and viscosity are ρ and μ. L and H represent the light and heavy phases, respectively. The light phase is dispersed if χ < 0.3. The light phase is likely to be dispersed for χ = 0.3–0.5, while both phases can be dispersed for χ = 0.5–2, and phase inversion can occur. The heavy phase is likely to be dispersed for χ = 2–3.3 while the heavy phase will be dispersed for χ > 3.3. For cases where either phase could be dispersed, the added phase is usually dispersed. Phase inversion can be triggered by changes in concentrations, temperature, and physical properties [127].

Liquid–liquid dispersions involve one phase being dispersed as droplets into the other phase. The varying sizes of the dispersed droplets can have varying impacts on the performance of the equipment being used. A narrow droplet size distribution is preferred to ensure optimal dispersion performance [128]. Compared to gas–liquid systems, the drop settling for liquid–liquid systems is slower. This is caused by the continuous phase in liquid–liquid systems being much denser and more viscous [129]. The maximum drop size is determined primarily by the balance between buoyancy and interfacial tension forces acting on the flow. In static devices, disregarding viscosity effects (assuming low viscosity of the dispersed phase), the maximum drop size is directly proportional to the square root of the interfacial tension (σ) divided by the difference in density (∆ρ) [130]:

d_max = c₁(σ/Δρg)^0.5,

(11)

The constant c₁ is approximately 1 [130]. Equation (11) does not indicate the Sauter mean diameter but the maximum stable drop diameter. However, these two diameters are proportionally related, and their values can be close [75,131,132]. The equilibrium between drop breakage and coalescence rates determines the drop size in agitated devices [133]. Breakage occurs due to turbulent stresses caused by the agitator, while coalescence occurs when drops collide with each other [133,134]. The maximum drop size in an agitated liquid–liquid dispersion was investigated by Kolmogorov [135,136] and Hinze [134], who developed the following expression:

d_max = c₂(σ/ρ)^3/5ϵ^2/5,

(12)

where ϵ is the turbulent energy dissipation rate and the constant c₂ is approximately 0.7 [134,137]. Equation (12) assumes low dispersed-phase holdup and neglects viscous forces and breakage, which is a valid assumption for water and typical low-viscous and moderate-viscous organic solvents [75,134]. Wang and Calabrese [138] investigated the effects of viscous resistance on droplet breakage and identified interfacial tension and dispersed-phase viscosity as influencing parameters.

A general equation for the maximum drop size in agitated liquid–liquid dispersion is given by [139]:

d_max/D_i = We^−3/5,

(13)

where We is the Weber number (ratio of inertia to surface tension) [140], and D_i is a characteristic diameter. For applications using rotating impellers, D_i is the impeller diameter, and the appropriate Weber number is We = ρω_i²D_i³/σ, where ω_i is the impeller speed (in rotations per second) [136]. For static mixers, D_i = D_sm and We = ρv_sm²D_sm/σ, where D_sm is the static mixer pipe diameter, and v_sm is the superficial liquid velocity (inlet velocity).

2.5. Shake Test

The dimension of dispersed drops can influence separator selection and design. Prediction of drop diameter is difficult, especially when drops are tiny or sampling is challenging. Moreover, knowing the drop size does not give any information about drop coalescence. As a workaround, the feed can be classified based on the simple shake test [75]. The feed classification can be used to identify suitable separator types. The shake test is performed by hand-shaking the feed in a closed cylinder of around 2.5 cm diameter for 30–60 s. In systems where drops merge rapidly, a sharp boundary emerges between two settling liquid layers [75]. The movement of drops, toward or away from the boundary, will dictate how quickly the layers undergo phase separation. In many other systems, droplets gather at the boundary and create a dispersion band—a stratum of drops that coalesce at a slow pace. In these cases, the phase separation speed is determined by how rapidly the droplets merge.

Table 1. Shake test characterization [75].

Type	Shake Test Observations	Interfacial Tension *	Density Difference *	Viscosity of Each Phase *	Presence of Fine Solids or Surfactants *
I	Dispersion band collapses within 5 min with crystal-clear liquids on top and bottom	Moderate to high, 10−6 N/m or higher	Δρ > 100 kg/m³	µ < 0.005 Ns/m2	Negligible
II	Dispersion band collapses within 10 to 20 min with clear liquids on top and bottom	Moderate, ~10−6 N/m	Δρ > 100 kg/m³	µ < 0.02 Ns/m2	Negligible
III	Dispersion band collapses within 20 min but one or more phases remain cloudy	Low to moderate, (3–10)·10−7 N/m	Δρ > 50 kg/m³	µ < 0.1 Ns/m2	Might be present in low concentration
IVa	Stable dispersion is formed (dispersion band does not collapse within an hour or longer)—high viscosity	Low to high	Δρ > 100 kg/m³	µ > 0.1 Ns/m2 in one of the phases	Negligible
IVb	Stable dispersion is formed—low interfacial tension	<3·10−7 N/m	Δρ > 100 kg/m³	µ < 0.1 Ns/m2	Negligible
IVc	Stable dispersion is formed—low density difference	Low to high	Δρ < 50 kg/m³	µ < 0.1 Ns/m2	Negligible
IVd	Stable dispersion is formed—stabilized by surface-active components or solids	Low	Δρ > 100 kg/m³	µ < 0.1 Ns/m2	Enough surfactant/solids to keep emulsion stable

* A sample is described by the shake test result (second column), not its physical properties. Physical properties are listed only as reference data.

classifies feed materials into four categories based on the shake test results [75]. In addition, common data of interfacial tension, density difference, and viscosity also are provided.

3. Liquid–Liquid Phase Separation Equipment

3.1. Gravity Decanters (Settlers)

Gravity decanters (also known as settlers) are vessels where two phases settle into separate layers over time [30]. Figure 2 indicates the operating principle of decanters. The liquid mixture enters at one end of the vessel (feed) and separates into layers based on density difference between phases. The denser phase settles down faster under the influence of gravity, while lighter phase rises to the top. The separated phases are discharged through respective outlets at the other end of the vessel. The outlets and the top of the vessel are vented to the atmosphere. Based on the shake test characterizations (see Table 1), gravity decanters are appropriate for handling type I and the majority of type II systems. Coalescence is a limiting factor for mixtures of type II; therefore, the configuration and size of decanters can be different than the fast coalescing type I systems.

The typical length-to-diameter ratio for horizontal gravity decanters is greater than 2 [28]. The cross-sectional area (phase boundary between the two settled layers) is maximized with this ratio. This makes the horizontal decanter more effective compared to the vertical design. However, vertical decanters are more practical for low-flow rate applications and space limitations.

3.1.1. Stokes’ Law Design Method

Stokes’ method relies on the assumption that the coalescence rate of drops is fast, and it requires knowledge of the drop size. The drop settling velocity based on Stokes’ law [127,141] is as follows:

u_t = gd²Δρ/18µ,

(14)

where d is a characteristic minimum drop diameter; g is gravitational acceleration; ∆ρ is the density difference between the drop and fluid; and μ is the dynamic viscosity of the fluid. Only the continuous phase viscosity is used in Equation (14). Therefore, defining which phase is dispersed and continuous is essential. According to [141], the settling velocity from Equation (14) is related to small Reynolds numbers (Re < 0.5) where the Reynolds number is given by

Re = u_tdρ/µ,

(15)

Stoke’s law calculations are not reliable for designing systems with slow coalescence. Jeelani and Hartland [142] recommended the following expression for the decanter performance:

1/(Q/A) = 1/(k₁ΔH) + 1/(k₂),

(16)

where ΔH is an average steady-state dispersion band height; Q is total volumetric throughput; and k₁ and k₂ are empirical constants [142].

3.1.2. Vented Decanters

Venting supplies to the decanter have to be included for systems containing gases or vapors. Gases or vapors can occur during azeotropic distillation in vacuum where condensation occurs or during the decanting of liquids from extractors operated at high pressure. Figure 3a shows a typical design for vented decanters. The liquid feed is introduced into the decanter at a level below the liquid surface. Gases have to flow through the liquid before reaching the vapor headspace.

Figure 3b shows an alternative decanter design where the feed is introduced from the top of the vessel in the vapor headspace. Adding a perforated baffle provides a quiescent zone below the top feed nozzle. This baffle separates disturbances caused by the feed stream and a calm separation zone where the two liquid phases can coalesce and disengage before being drawn off.

3.1.3. Coalescing Internals

Decanter performance can be improved by adding coalescing internals. Coalescing internals promote the growth of drops and reduce the length that drops need to rise/fall to a coalescing surface [143,144]. Consequently, the size of decanter handling dispersions with slow coalescence (e.g., type II from Table 1) can be reduced when using internals. Coalescing internals can be wire meshes, knitted wires or fibers, and flat or corrugated plates [29].

3.2. Coalescers

Coalescing internals can be applied to pre-treat decanter feeds in separate coalescer vessels [33,34]. Coalescers are particularly used for type III and IV systems (see Table 1). Benefits of coalescing internals are improved separation efficiency, reduced operating costs, and increased throughput. The configuration of coalescers depends on the dispersed phase. For the light phase being dispersed, the configuration is upflow, and for the heavy phase being dispersed, the configuration is downflow. Coalescers are typically packed with granular materials, metal wire meshes, polymer filaments (or both), or fine fibers in woven or nonwoven composite sheets [29]. Coalescers can tolerate feeds containing fine solids if granules or wire meshes are used. The feed should be free of solids for coalescers containing fine granules or fibers to avoid potential requirement for pre-filtration [31,32,145,146]. A standard design for a coalescer is given in Figure 4. The liquid mixture enters at one end of the vessel (feed). Then, liquids flow (1) through the coalescer elements inside the vessel, separate, and discharge at respective outlets: (2) clean product of light phase, and (3) dispersed heavy phase. Generally, the best coalescer performance is achieved when the dispersed phase wets the packing material. [29].

An essential criterion in the design of coalescers is the pressure drop, which depends on particle size and shape, bed height or filter thickness, and throughput. For near spherical particles, assuming incompressible fluid, the pressure drop due to frictional losses can be estimated from the Ergun equation [147,148]:

ΔP/L = [150(1−φ)²µv_sv/(d_m²φ³)] + [1.75ρv_sv²/d_mφ³]; Re_particle = v_svρd_m/µ ≤ 10,

(17)

where L is the length of the packed section; v_sv is the superficial velocity of the total liquid flow; d_m is effective particle diameter (given by six times the mean ratio of particle volume to particle surface area); and φ is the volume fraction of voids (flow channels) within the bed [147]. The minimum value of φ for a tightly ordered bed of uniform spherical particles is 0.26 [75]. This varies depending on the particle size distribution and particle shape for real media.

3.3. Centrifugal Separators

Centrifugal force can impact the separation of phases significantly. Unlike gravity being constant, centrifugal force can be varied either by the rotational speed or equipment dimensions. Devices for performing centrifugal separation are cyclone separators for gas–gas and gas–solid separation, hydrocyclones for liquid–solid and liquid–liquid separation, and centrifuges for liquid–solid, liquid–liquid, and gas–gas separation [35,37]. Centrifugal devices are beneficial in the following instances [35]:

• the difference in density between the phases is small;
• immediate contact is needed to prevent degradation;
• feed and solvent emulsify easily;
• specific required throughput due to capacity restrictions.

Centrifugal devices can achieve high throughput at low liquid residence times [35]. Residence time is given by the holdup volume of the feed phase divided by the volumetric feed rate, t = V_t/V. Centrifugal separators allow feed rate and rotational speed adjustments, making them flexible and beneficial for operation. However, these devices have some potential disadvantages:

• maintaining high-speed rotating machinery—even though the high acceleration in operation makes a good performance, in some cases, this can promote back-mixing or emulsification;
• presence of solids—some of these devices are designed to handle feeds containing solids, such as whole fermentation broth. However, if solids are present in the feed, there is a potential for plugging, e.g., in separators with close internal clearances.

3.3.1. Centrifugal Force

Rotating objects undergo centripetal and centrifugal forces. Figure 5 illustrates the action of these forces. While the centripetal force keeps an object moving in a circle and is pointed toward the circle center, the centrifugal force is an outward force on an object moving in a circle.

The centrifugal force F_c is given by

F_c = mω²r,

(18)

where m is the mass; r is the distance from the origin of a frame of reference; and ω is the angular velocity.

The centripetal force F_cp is given by

F_cp = mu^2/r,

(19)

where u is the tangential velocity.

In centrifuges, gravity can be neglected because centrifugal force is significantly higher than gravity [36].

3.3.2. Centrifuges

Centrifuges are devices used to separate liquids with small density differences. They can handle high throughputs, e.g., up to 100 m³/h. Centrifuges can also break almost any emulsion [29] and are suitable for type III and type IV systems (see Table 1). When centrifuges are stationary, the free liquid surface is horizontal, and the liquids (and possible particles) settle due to gravity. Once the rotation starts, the suspension is subjected to gravity and centrifugal force, resulting in the suspension taking the position of the axe of rotation (vertical or horizontal free liquid surface). The suspension can consist of several components having different densities. In this case, the lightest component will stratify nearest to the rotation axis and the heaviest farthest to the rotation axis [35]. Common types of centrifuges are bowl (tubular bowl) or disk (disk bowl) centrifuges [29].

The working principle of centrifuges is always the same. The liquid mixture is introduced to the centrifuge bowl, where centrifugal force is created through the drive mechanism rotating the centrifuge bowl or centrifuge internals. As a result of centrifugal force (rotation), phases are being separated and discharged. The phases are separated as follows: the lighter phase is close to the rotation axe where the centrifugal force is lowest, and the heavier phase is close to the centrifuge wall where the centrifugal force is the highest.

Figure 6 shows a schematic tubular bowl. Tubular bowls are high-speed, vertical centrifuges to separate immiscible liquids, such as water and oil (and fine solids). The bowl is typically rotated at around 15,000 rpm (250 Hz), and the generated centrifugal force exceeds 130,000 N [29]. The tubular bowl is designed for low capacities, 0.19 to 1.9 m³/h (50 to 500 gph), and can handle only low solid concentrations [35].

Figure 7 shows a schematic disc bowl. Disc bowls are used to separate liquids and fine solids and classify solids. These devices consist of conical discs that split the liquid flow into multiple thin layers [29].

Disc bowl centrifuges are larger than tubular bowl centrifuges and rotate at slower speeds. These centrifuges can handle up to 114 m³/h (30,000 gph) of feed and moderate quantities of solid particles [35]. Particles migrate radially (toward or away) from the axis of rotation, depending on whether the density of the dispersed particles is greater or lower compared to the continuous phase [35].

3.3.3. Hydrocyclones

Another type of centrifugal separator for mixtures of liquid phases are hydrocyclones. Figure 8 shows a schematic sketch of a hydrocyclone [38,39]. Among the existing cyclones, the gas–solid cyclone, known as cyclone separator, has the longest history of usage [40]. Although the first hydrocyclone patent was granted in 1891 [149], these devices were not used until the early 1940s when numerous research results appeared in the literature, resulting in the expansion of hydrocyclones use for a broad spectrum of applications [40,41,42,43]. Hydrocyclones have traditionally been used for liquid–solid separation. They can also be applied to separate liquids after adjusting their design [35,40,41,42,43,75]. Hydrocyclones generate centrifugal force from fluid rotation [75]. Figure 8 illustrates the working principle of hydrocyclones. The hydrocyclone feed is introduced via a tangential inlet, which initiates the formation of a primary vortex near the hydrocyclone wall containing the heavy phase and a secondary vortex in the center containing the light phase. The heavy phase, or underflow stream, leaves through the cone’s apex via the underflow outlet, while the light phase, or overflow stream, leaves through the vortex finder, a duct that expands from the cylinder top into the inner part of the hydrocyclone.

Due to the presence of a high shear generated by the rotation of liquid in the device and the turbulent flow at the top of the unit, hydrocyclones have a tendency to create foam or emulsify mixtures with low interfacial tension [40]. Nevertheless, hydrocyclones are particularly suitable for separating type I and type II mixtures (Table 1) that contain solids, particularly when only a coarse separation is required. The main variables governing the operation of a cyclone are the feed pressure, feed flow rate, and split ratio, which refers to the proportion of fluid exiting from the top and bottom of the unit. The split ratio can be adjusted by the specifications of the over- and underflow outlets. When construction material for the cone is wetted by the heavy phase, the efficiency of the hydrocyclone can improve [40,75]. The reverse is true for the vortex finder and lighter phase [40]. If preliminary tests demonstrate adequate performance, hydrocyclones can be simple and cost-effective units that do not require any moving parts. Scaling up can be achieved by utilizing multiple smaller units in parallel, as large-diameter units are unable to generate sufficient centrifugal force.

3.4. Other Devices

3.4.1. Ultrafiltration Membranes

Ultrafiltration is a separation process for solid–liquid mixtures where membranes act as filters [44]. Ultrafiltration membranes involve the rejection of macromolecules or particles at the membrane surface. The membrane pore size ranges from 0.001 to 0.1 μm [44]. Membranes are typically made of polymers [150], but other materials, such as zeolite, can also be used for membranes in separation processes [151,152,153]. The feed enters the membrane at low pressure (typically less than 5 bar) [44]. A clarified solution is obtained on the permeate side due to the continuous phase passing through the membrane pores due to a pressure difference. The membrane pores allow liquid passage while the solid particles or macromolecules (bigger than membrane pore size) are rejected at the membrane surface. The rejected particles or molecules continue flowing tangentially along the membrane surface with the remaining continuous phase. The shear at the membrane surface should be high enough to stop particle or molecule aggregation on the membrane’s polymeric surface but sufficiently low to prevent the colloidal particles from breaking [44].

Pores of ultrafiltration membranes are tiny channels that allow only certain molecules or ions to pass through, while blocking others. In industrial applications, feeds may contain suspended particles or impurities that can block membranes by physically obstructing the membrane pores. When the feed containing particles is passed through a membrane, particles can accumulate in the membrane pores, leading to blockage. An example of the feed that contains particles is crude oil extracted from the ground, which can contain a variety of particles such as sand, clay, and other minerals that are mixed with the oil. During processing, oil can also pick up particles from equipment and other sources. Any of these particles within the feed can impede the feed flowing through the membrane and cause membrane fouling. Fouling is the main disadvantage of membranes because it reduces the separation quality [44,45]. Hydrophilic membrane materials or surface treatments with hydrophilic properties can prevent membrane fouling [46,47]. An alternative method for clarifying oil-in-water dispersions is to employ hydrophobic microporous membranes that allows intentional permeation of the oil phase through the membrane [48]. Other limitations of membranes are high capital costs and lower throughput, e.g., compared to hydrocyclones [45].

Ultrafiltration applications for liquid–liquid separation include water treatment, food and beverage processing, biotechnology, oil industry and pharmaceutical production. For example, in the food and beverage industry, ultrafiltration can be used to concentrate milk or fruit juice, while in the biotechnology industry, it can be used to separate proteins or enzymes for use in research or manufacturing processes.

3.4.2. Electrostatic Coalescers

Electrostatic coalescers (electro-coalescers) apply an electric field in liquid phase separation to achieve fast, clean, and efficient separation [49]. These devices can be used in liquid separation, e.g., water and oil, solvent extraction and dispersion, and to fractionate mixed oils [50,51]. An electric field facilitates the merging of small droplets either with each other or with an interface. The larger drops can easily settle by gravity. Therefore, the application of electric fields increases the coalescence rate, enhances the drop migration speed, and facilitates phase separation [49]. Eow et al. [49] reviewed the major mechanisms of electro-coalescence and separation, e.g., chain formation of droplets, dipole coalescence, dielectrophoresis, electrophoresis, random collisions, and film drainage and rupture. The predominant mechanism depends on electrode design and set-up, the type of employed electric fields, and emulsion properties [49]. Phase separation with the use of electrostatic coalescers has been reviewed by Mhatre et al. [154]. The mechanism of electro-coalescence of two drops involves three main stages: drop–drop approach, film drainage, and thin film breakup. Accelerating every stage can make the coalescence faster [154].

4. Recycling of Phase-Forming Components

The recycling of separated liquids depends on the nature of the liquids and the purpose of the separation. If the separated liquids are hazardous or contaminated, they may need to undergo additional treatment or disposal processes before recycling. If the separated liquids are clean and reusable, they can be recycled back into the process or used for other purposes. For example, in a wastewater treatment plant, the separated clean water can be recycled for irrigation or industrial use. In the food and beverage industry, separated fats and oils can be recycled and used for the production of biodiesel or animal feed. The recycling of separated liquids can be an environmentally friendly and cost-effective solution, as it reduces the amount of waste generated and conserves resources. However, it is important to ensure that the recycling process is safe and efficient and complies with environmental regulations [155]. Over the last years, significant research on new separation techniques and recycling of separated phases have been concluded, e.g., recycling of ionic liquids in aqueous two phase system separation [156,157], use of eutectic solvents in different extraction and separation processes [158,159,160,161,162,163,164,165,166], and separation and reuse of rare earths using ionic liquids [167,168,169,170].

5. Conclusions

Over the last decades, with growing interest in environmental and nature resource concerns, the separation of liquid phases has received renewed interest. Various methods for separating liquid phases from mixtures have been in use for decades. New designs or design modifications of existing equipment emerge for novel separation tasks.

Available separation methods for liquid–liquid separation are reviewed. Gravity decanters are the optimal choice for separation processes where time is not a constraint, as they rely solely on gravity to separate phases. Adding coalescing internals, such as granular materials, metal wire meshes, polymer filaments, or fine fibers, can improve the separation process. Coalescers can be integrated into the decanter design or used as a separate device. Centrifugal separators are the most commonly employed separation devices, relying on centrifugal force instead of constant gravity. These devices provide greater control over the separation process, as the centrifugal force can be adjusted using either the rotational speed, in the case of centrifuges, or the equipment dimensions, as with hydrocyclones. Other methods for liquid separation are ultrafiltration and electrostatic coalescence. Membranes separate macromolecules or particles at the membrane surfaces, while electrostatic coalescers use an electric field to facilitate the merging of small droplets.

The centrifugal devices seem to be the most suitable of the investigated devices for liquid–liquid separation. In comparison to the decanters or coalescers, the centrifugal devices can achieve faster separation, and their ability to vary centrifugal force allows better process control. Electrostatic coalescers utilize an electric field to speed up droplet merging and enhance coalescence rates. However, the larger droplets still rely on gravitational settling, making centrifugal devices favorable compared to electrostatic coalescers. In comparison to membranes, centrifugal devices have advantages in terms of being less susceptible to fouling and can achieve higher throughputs.

Open challenges for liquid–liquid separation are optimization of existing separation techniques, exploration of new solvents, development of new separation techniques, and more study of fundamental principles. Work on the optimization of existing techniques would improve separation efficiency and reduce costs. An example would be modifying the membrane structure to improve its performance. The choice of solvent is a critical factor in the success of liquid–liquid extraction. Researchers could explore novel solvents with improved extraction properties or greater environmental sustainability compared to conventional solvents. The development of new techniques could mean innovation that reduces energy consumption or potential miniaturization (advances in microfluidics could lead to smaller and more compact liquid separation systems). Developing sustainable technologies (recycling of components) and automation could also reduce the need for human intervention and provide greater accuracy and efficiency. Understanding the fundamental principles of mixture separation can also lead to new insights. Researchers should investigate the thermodynamics and kinetics of liquid–liquid extraction and other separation techniques to gain a more comprehensive understanding of the separation processes. In general, continued research and enhancement of liquid mixture separation can result in separation techniques that are both more efficient and cost-effective.