Describing the strength of van der Waals forces in mixed liposome/nanoparticle systems in a quantitative manner is challenging due to the geometry and nature of the system components. In the ideal case of a sphere (medium 1: nanoparticle) interacting with a shell (medium 3: liposome) in an aqueous medium (medium 2), the van der Waals energy is given by the equation [55,56]:

where R₁ is the radius of the sphere (particle), R₂ the outer radius of the shell, h the thickness of the shell, d the distance between the surfaces of the two objects and A₁₂₃ the Hamaker constant of the system. An approximate value of A₁₂₃ may be given in terms of the Hamaker constants of the individual media as follows [55,57]:

where A_xx is the Hamaker constant between two semi-infinite planes of medium x in vacuum.

For “symmetrical” systems, the Hamaker constant is always positive, leading to attractive forces, but between dissimilar surfaces, as in the case of mixed systems liposome/nanoparticle, A₁₂₃ can be either positive or negative leading respectively to attractive or repulsive forces. For phospholipid bilayer interacting with oxide particles (SiO₂, TiO₂) the Hamaker constant is typically (3 − 4) × 10⁻²¹ J ≈ 0.75 to 1 kT (for T = 25 °C) [52,58,59].

In general, the electrostatic double-layer interaction energy between two identical planar surfaces decreases exponentially with the distance. However, for two surfaces of different charge densities or potentials, which is the case in mixed system liposome/nanoparticle, this interaction energy can exhibit a minimum or a maximum at some finite distance [55]. The study on these asymmetric systems has led to different approximate equations for the interaction energy of two surfaces of unequal but constant potentials [60–62]. The “Hogg–Healy–Fuerstenau” equation (HHF) [55,60] for two different planar surfaces of low constant potentials ψ₁ and ψ₂ in 1:1 electrolyte is given by:

where D is the distance between the surfaces and κ their Debye length. This equation has recently been successfully implemented to characterize the double layer interaction between lipid bilayer and silica substrates [52].

Nevertheless, when working with nanoparticles and vesicles one often has to take into account the curvature of the interacting surfaces. To this aim, applying the Derjaguin approximation on equation 4 leads to the following formula for the approximate interaction energy between dissimilar double layers on two spherical particles with radii R₁ and R₂ [60]:

It is important to note that the HHF formula is based on the Debye-Hückel linear approximation. This approximation is thus applicable for sufficiently low potentials (commonly ψ < 25 mV). More recent work using the Poisson-Boltzmann expression for the potential has led to more accurate analytical approximations [61].

Alternatively, The Gouy-Chapman Theory has also been used to describe the electrostatic interaction energy between a flat substrate with surface potential ψ_sub and a flat bilayer of surface potential ψ_bil in monovalent salt solution [63,64] leading to the following formula:

where T is the temperature, k the Boltzmann constant and ρ_∞ the ions concentration. This equation also assumes a constant surface potential at any distance D.

In systems containing phospholipid bilayers, van der Waals and double layer interactions, as jointly expressed by the DLVO theory, often fail to precisely describe the phenomena taking place at short distances due to the presence of repulsive hydration forces [65,66], as has been observed experimentally by means of the surface force apparatus. These repulsive forces are believed to arise from the presence of a layer of water molecules, strongly bound to the hydrophilic surfaces and preventing them from approaching any closer than the thickness of two water molecules [55,65,67].

The full nature of this interaction is still an active field of research. However, in the case of hydrophilic surfaces in aqueous solution, the hydration energy is attributed to stable structured water layers hydrogen bonded to the solid surfaces, while the hydration repulsion between two lipid membranes is believed to be dominated by entropic factors (microscopic thermal fluctuations).

Tero et al. [64] were the first to propose an expression for the calculation of the hydration energy between lipid bilayer and hydrophilic surfaces, by assuming that the hydration between lipid membrane and solid substrate (W_hyd) is the average of each hydration energy, giving:

where W_solid is the hydration force between hydrophilic substrates, depending on the nature of the substrate and its medium. Empirically, W_solid has been found to decay exponentially with the distance (D) between the surfaces [55,63,64]:

where λ₀ is on the order of 1 nm and W₀ depends on the hydration of the surfaces but is usually in the range 3–30 mJ·m⁻² ≈ 0.7 to 7 kT·nm⁻² (for T = 25 °C).

On the other hand, W_lipid is believed to arise from thermal fluctuations [68] and thus depends on the three different types of thermal motion that a membrane can experience: protrusion, peristalsis and undulation. The energetic contribution of these different motions can be separately expressed as follows [55,64]:

where Γ is the surface density of protruding head groups, α_p the (hydrophobic) protrusion energy per unit length_, k_a the area expansion modulus and k_b the bending modulus. Typical values for these parameters are: α_p = 2.5 × 10⁻¹¹ J·m⁻¹, k_a = 0.15 J·m⁻² and k_b = 10⁻¹⁹ J [55,64] while Γ depends on the molecular occupying area of the chosen lipid.

It should be noted that undulation forces may be neglected in the case of membranes carrying unscreened surface charges or being under tension. Nevertheless, when considering every contribution, the hydration energy between a bilayer and a flat hydrophilic substrate is given by the expression:

Additionally, it should be mentioned that the geometry of a curved surface tends to smother short range interaction (hydration) and emphasizes longer range van der Waals and electrostatic double layer forces [65,69,70]. This is of course of crucial importance when studying the balance of forces in mixed systems liposomes/nanoparticles where both species often exhibit a curved surface.

Hydrophobic interactions may arise when phospholipid bilayers, subjected to a stretching force or stress, expand laterally and expose areas of their hydrophobic interior to the aqueous solvent.

Hydrophobic attraction plays an important role in the mechanisms of vesicle fusion or particle embedding, when hydrophobic particles are internalized into the hydrophobic interior of the membrane. In the case of vesicle-vesicle interaction, the increase in stress experienced by the membrane would be directly proportional to the increased adhesion force between two vesicles [55]. In the case of particle embedding, hydrophobic attraction can be expressed as the balance between the free energy change to move a hydrophobic sphere from pure water into a hydrophobic membrane (ΔG_emb) and the energy penalty to deform the bilayer (ΔG_def) [71]. Where ΔG_emb is given by [71,72]:

where D is the nanoparticle diameter and γ the liquid-vapor surface tension of water.

Ultimately, the interplay between Van der Waals, double layer, hydration and hydrophobic interactions can be tuned by playing on numerous parameters, among which are pH, ionic strength and temperature, in order to obtain various structures such as supported lipid bilayers, internalized particles and decorated vesicles.

Supported lipid bilayers (SLB) are continuous fluid lipid membranes (T > T_m) adsorbed on a solid substrate, and separated from this substrate by a thin water layer (1–3 nm) [73]. It is important to note the necessity of having a bilayer in the fluid phase (to have a membrane of sufficiently low elasticity [49–51]) and a hydrophilic substrate (to bind the supporting water layer) to achieve the deposition of SLB.

SLB formation has been studied first for the case of planar substrate (glass, silica, mica) as the basis to produce innovative catalytic surfaces or immobilized protein arrays [52,58,73–82], as well as to study bilayer-bilayer interactions, as they are important in biology, under well-defined conditions. Nowadays, this field of research has been extended to the case of nanoparticles, with the intention of designing nanovectors by rendering nanoparticles biocompatible through the deposition of a lipid bilayer onto their surface [83–85]. Such studies on well-defined membranes are important in order to control the interactions and thereby optimize the design of the delivery systems.

The formation of SLB, as has been investigated on both planar surfaces and nanoparticles, can be achieved by depositing vesicles from the solution [52,75,79,83–87] where the liposomes adhere, fuse and spread on the solid surface. In the case of a planar substrate, this process has even largely replaced the Langmuir-Blodget deposition due to the wider availability of more convenient methods for unilamellar liposomes preparation [76]. Despite the considerable work on vesicle adsorption, there is still no complete picture of the forces responsible for the adhesion, fusion and rupture of the liposomes on solid surfaces. However, the SLB formation is believed to be governed by three types of interactions [79,88]:

Note that the latter has been found to be of less importance in this process [88].

Experimentally, one can play with several parameters (ionic strength, addition of specific ions, pH, surface properties of the substrate, nature of the phospholipid(s) used) in order to monitor the strength of these three different interactions.

Changing the ionic strength and/or the pH is a way to alter the electrostatic properties of the substrate as well as the phospholipid net charge leading to different adsorption behaviors depending on the nature of the substrate and on the type of lipid used [52,75,84,85]. More precisely, the use of specific ions (in particular divalent cations) has been found to have a surprisingly strong effect on SLB formation, when bridging negatively charged or zwitterionic lipids and negatively charged substrates (silica-Figure 1) [79,86–89]. This is not surprising given the known affinity of Ca²⁺ and phosphate. The case of zwitterionic phospholipids is particularly interesting, as the cations convert the headgroup from a dipole toward a positive monopole thus strengthening the interaction with negatively charged substrate (Figure 2) [85,88].

However, even if the deposition mechanism is believed to be mainly driven by electrostatic attractions between vesicle membranes and the substrate surfaces, other forces are also non negligible. As an example, SLB formation or vesicle adsorption have been reported in systems where the electrostatic interactions were clearly repulsive [52] or minimized by experimental conditions (pH [75], substrate surface properties [90]) showing the importance of van der Waals attractive interactions as well as of hydration and steric forces in the deposition mechanism.

Previous studies on the formation of SLB on hard nanoparticles can be divided in two types that are depicted in Figure 2.

(1) In the first case mentioned above, the mechanism of SLB formation proceeds via subsequent adsorption, deformation and rupture of liposomes on the nanoparticle surface [83] much like in the case of flat substrates [52,79,80,89]. After adsorption of the liposomes, the contact area between the lipid surface and the substrate increases, due to adhesion forces until the deformation of the vesicle reaches a limit above which rupture occurs, resulting in the appearance of patches on the particle surface. It is to be noted that this process has been found to occur either independently or dependently of the presence of preformed lipid patches in the vicinity of the adsorbed vesicle. As a matter of fact, the edges of bilayer patches are likely to activate the decomposition of adsorbed vesicles [79]. Hence, the formation of a SLB covering an entire particle often involves the coalescence of neighbouring liposome patches via active edge effects [89]. In any case the binding of a nanoparticle to a bilayer will lead to a loss of configurational entropy of the membrane as it suppresses undulations of the bilayers.

However, the process of SLB formation depends highly on the ability of the membrane to curve around the particle and adopt its shape [92,93]. The curvature of the lipid bilayer is depicted in the literature as the result of the interplay between the (favourable) adhesion energy between the membrane and its substrate and the (unfavourable) elastic energy (or bending energy) needed to bend the bilayer around the particle [48,80,89,92–94]. Corresponding theoretical studies have led to structural phase diagrams demarking the regions of fully enveloped, partially wrapped or freely dispersed colloids [95–97]. The investigations on the transition steps between the wrapped and the free state of particles allow the definition of a critical radius for the particle (R_C), meaning that the formation of SLB is energetically favourable on particles having a radius above this critical value. In the case of a tensionless membrane, the value of R_C may be extrapolated from the expression of the energetic balance when the adhesion energy is compensated by the bending energy [48,98], giving:

where R_C depends on the strength of the adhesion energy per unit area (W) as well as on the value of the bilayer mean bending modulus (κ), thus depending on the nature of both particle and lipid bilayer. Consequently, one can modify this critical value by changing experimental conditions such as the nature of the lipids, the properties of the substrate surface [90] as well as the properties of the medium such as pH [84] and ionic strength [85]. For instance the bending modulus (κ) is proportional to the third power of the bilayer thickness [99–101]. Accordingly, the values of R_C determined experimentally have been found to vary with the nature of the system. In the case of silica particles interacting with zwitterionic lipid bilayers, the critical radius has been found to be around 10 nm [42,92,93] corresponding to an adhesive strength W of a few mJ/m²—a value comparable to usual interfacial energies [55].

Interestingly, supported lipid bilayers adsorbed on small particles (R_C < 100 nm) exhibit morphology changes due to the high curvature they experience. Shifts of the T_m value [102] as well as a broadening and/or splitting of the transition peak obtained by differential scanning calorimetry (DSC) [102,103] suggest a curvature-dependent decoupling between the inner and outer leaflet of the bilayer as has also been observed in the case of small SUVs [104,105]. In extreme cases (R_P ≤ 10 nm) the high curvature induces the creation of a large free volume between the acyl-chains of the supported bilayer [102,106] (Figure 3, left). In this case, interdigitation of the acyl-chains occurs in order to optimize hydrophobic interaction while avoiding the exposure of hydrophobic chains to an aqueous environment (Figure 3, right) [107].

(2) In the second case mentioned previously, the SLB formation on smaller particles is accompanied by the internalization of the nanoparticle into the vesicle interior, as has for instance been observed for the case of silica NPs interacting with DOPC vesicles (Figure 4). As described by Le Bihan et al. [42], the proposed invagination mechanism predicts the adsorption of the particle on the liposome outer surface, followed by the spreading of the membrane around the particle finally leading to the engulfing and the full internalization of the nanoparticle. This process of nanoparticle transmigration is similar to that of cellular uptake (or endocytosis) [108,109]. Simulations on similar systems [110] have been found to consistently reproduce the invagination mechanism described above.

The invagination mechanism is governed by the same energy balance as in the previously discussed case of SLB formation [48,49,96,111]. Accordingly, the same expression (Equation 14) arises for the calculation of the critical particle radius value below which the particle remains adsorbed on the liposome outer surface. In this case, the total adhesion energy does not overcome the energy cost associated with the bending of the membrane, thus preventing the invagination mechanism to proceed further. Nevertheless, the experimental findings [42] seem to contradict some of the requirements expressed by theoretical studies such as the necessity of working with large liposomes (R_Ves > 300 nm) [95] as well as the required presence of rafts which are phase separated from the rest of the membrane in order to promote the fission process allowing the detachment of the internalized particle from the membrane [111]. However, related to the latter, it may not be necessary to tune the membrane to obtain phase separated domains since the adsorption of particles has been found, in some cases, to induce local phase separation on its own [112,113].

Vesicle decoration results from the adsorption of hydrophilic NPs onto the surface of liposome membranes in systems where the nanoparticle radius is below the critical value R_C (Equation 14) [40,43,114] (keeping in mind that this value can be controlled by tuning other parameters such as pH or ionic strength of the solution) [43]. In this case, the balance between adhesion and bending energy is therefore shifted to a range where it is unfavourable to the invagination process. Thus, the membrane is “incapable” of internalizing the particles, leaving them on its outer surface (Figure 5A). Due to the attractive interaction between membrane and the nanoparticles there will be indentations by the NPs into the membrane surface which depend on the balance between attractive interaction and the bending energy required for this partial wrapping. This does not only lead to a deformation of the membrane but affects also its local stiffness and will moreover cause an additional interaction force between the different NPs attached. Furthermore it has to be considered, that in some cases, where the particle radius is close to the critical value, the kinetic of the invagination process is slowed down, allowing the observation of decorated vesicle structures which, in these cases, are only intermediate structures [115].

For a number of cases of charged nanoparticles, the decoration of liposomes by particles (zwitterionic lipids/charged nanoparticles) has been found to increase the absolute Zeta potential value of the dispersion by introducing charges on the liposome surface [40,43]. The appearance of these charges on the vesicle surface gives rise to repulsive electrostatic forces between the decorated vesicle structures and thus stabilizes the vesicle dispersion [40,41,43]. This repulsion can be quantified by means of measurements of the ζ-potential and such experiments have shown a decrease of the ζ-potential and concomitant stability increase which becomes larger with increasing nanoparticle sizes [43]. This result is of great interest given the metastable nature of many liposome dispersions [35] which often leads to rather fast phase separation and which represents a major disadvantage for their potential applications. Accordingly, the addition of charged silica NPs is a way to regulate the stability of liposomes.

However, in these systems, the colloidal stabilization is only achieved upon the adsorption of a minimum amount of nanoparticles, which corresponds to a sufficient surface coverage of the liposomes to ensure the repulsion between the decorated vesicle structures (Figure 5B) [40,41].

This has similarly been observed in recent work by us as documented by the change of particle size as a function of time (Figure 6). This study on the stability of unilamellar DPPC vesicles (R ~ 50 nm) below their phase transition temperature by means of dynamic light scattering showed that the addition of small amounts of small silica NPs (R_h ≈ 8 nm) leads to a rapid destabilization of the vesicle dispersion [40]. Only the adsorption of a sufficiently high amount of silica particles onto the liposome surface, here about 14 NPs per vesicle, is able to prevent the fusion of neighbouring liposomes and ensure their colloidal stability for several months. This is an enormous enhancement compared to the situation of the unmodified vesicles and such long-time stability is very important for the storage of liposome dispersions and also for having controlled release properties.

On the contrary, if the NP concentration is too low, nanoparticles would bridge between adjacent liposomes, thereby introducing interactions and thus accelerating their fusion [40,41], as has been observed in nanoparticle-microsphere systems [116]. In order to limit the magnitude of this effect, one may change the sign and magnitude of the particle charge. As has been found for PC liposomes, the use of cationic particles allows colloidal stabilization at lower surface coverage as compared to anionic ones [114]. In fact, cationic particles adsorb weakly on the membrane due to the geometry of the P⁻–N⁺ dipole of the lecithin head group (Figure 7), and consequently, they are less apt to bridge between neighbouring liposomes. These results confirm the predictions made by computer simulations on analogous systems [117,118].

Specific studies on these nanoparticle-stabilized liposomes have led to interesting findings on their properties. First and in opposition to other stabilization methods (PEG coating [119–121]), the adsorption of nanoparticles neither prevents nor reduces the binding between liposome immobilized ligand (biotin) and free receptor proteins (streptavidin) in the bulk: at low surface coverage (but still sufficient to ensure colloidal stabilization), the outer surface of decorated liposomes has been found to be well accessible and biofunctional despite the presence of NPs [122]. Furthermore, the particle decoration does not seem to prevent its permeability; on the contrary, the release of molecules (proteins) loaded in the aqueous interior of the liposome may be more accurately controlled by appropriately engineering the hydrophilic nanoparticle layer on the liposome outer surface [123]. This is important for potential applications, since it allows for the stabilization of liposomes while retaining their other (biological) functionalities. Additionally, decorated vesicles are easily immobilized on a planar substrate by tuning the substrate charges to create an attractive electrostatic interaction between the substrate and the adsorbed nanoparticles [124]. Such immobilized liposomes have an amazing potential as nanocontainers for the study of biomolecules: they allow the prolonged observation of molecules loaded in their inside with lesser surface perturbation than in the bulk.

The incorporation of nanoparticles within the bilayer membrane occurs when using hydrophobic nanoparticles which have a higher affinity to the acyl-chains confined in the hydrophobic interior of the phospholipid bilayer. The most common preparation pathway to form such vesicle-nanoparticle hybrids is to dry a mixture of hydrophobic particles and phospholipid MLVs in chloroform, thus obtaining a NP-lipid film which is hydrated and further extruded or sonicated to obtain the final Vesicle-NP hybrid dispersion—provided the nanoparticles are sufficiently small [125].

There are numerous studies on such loaded vesicles using different hydrophobic particles including Au [71,126–129], Ag [130,131], Si [125], Fe₃O₄ [132,133], and quantum dots (CdSe, ZnS) [128,130,134,135]. In most cases, the incorporation of nanoparticles does not appear to have a significant influence on the vesicle size or stability [125,136] even though there has been mention of a decrease of the liposome size upon particle embedding for which no explanation has yet been provided [128,133]. However, the internalized NPs will induce local disruption of the bilayer structure, reduce lipid ordering and thereby grant a higher fluidity to the membrane [126,131,136] at the same time changing the thermodynamics of the lipid bilayer phase transition (merging of the pretransition–transition from the ordered gel (L_β) to the rippled gel phase (P_β)–with the main phase transition, and a broadening of the melting region and shifts in T_m) [129,130,136,137].

Theoretical studies have defined the mechanism of particle embedding as the result of the dominance of the favourable free energy change caused by moving a hydrophobic sphere from pure water into the hydrophobic membrane over the energy penalty needed to deform the bilayer [138,139]. In that context, it might be noted that the gain in interfacial energy (Δλ·A) is about 1000 kT for a nanoparticle of 4 nm radius (for a change of interfacial tension of 20 mN/m). Nevertheless, the incorporation within the membrane is only energetically favourable if the particle size is inferior to a critical size depending on the type of lipid used and being close to the value of the membrane thickness [71,125,138,139]. Larger particles cannot become accommodated within the lipid bilayer due to the high elastic energy penalty, which then would suppress the membrane fluctuations, and the geometric difficulties of having a membrane surrounding such a large nanoparticle. Yet the interaction of liposomes or phospholipid molecules with large and highly hydrophobic particles (as well as with flat hydrophobic substrates) has been reported to lead to the formation of a phospholipid monolayer on the hydrophobic surfaces [140–143]. The resulting structures may be used as biocompatible particles for drug delivery [141,144,145], as column materials for high-pressure liquid chromatography [146] or else as a way of investigating membrane-like assembly in an immobilized state [140,142,147].

However, in the case of the internalization of small NPs in the hydrophobic interior of the phospholipid bilayer, the phospholipid membrane must “unzip”, i.e., dissociate the acyl-chains of both layers (Figure 8C). The unzipping creates void spaces in the vicinity of the incorporated NP, an energetically unfavourable situation. In order to minimize this void space as well as the free energy of deformation, incorporated NPs have been found to cluster within the bilayer (Figure 8C) [71,129] resulting in the appearance of either a few fully loaded liposomes (vesicle in the center with dark rim) within a “normal” SUV dispersion (Figure 8A) of vesicles without NPs contained in the membrane. In other specific cases, Janus-type vesicle structures are formed where internalized NPs are concentrated on one side of the vesicle (Figure 8B) [71]. Very recent work on this phenomenon has underlined the importance of the bilayer phase behaviour in this process of particle aggregation. Nanoparticle clustering is observed upon lipid melting, and believed to be driven by greater hydrophobic mismatch between embedded particles in the fluid phase coupled with lipid mediated forces driven by lateral capillarity [129]. Nonetheless, although the argument justifying the clustering of particles is quite convincing, the preferential embedding of particles in a few vesicles within a liposome dispersion or in one side of a vesicle (case of Janus-vesicles) has not been reported in other similar works [130,133,135] and therefore is still a topic open for discussion.

Such nanoparticle-vesicle hybrids are promising tools in nanobiotechnology. For example, the use of quantum dot-vesicles allows the controlled vesicle fusion on cell surfaces while at the same time enabling the visualization of the resulting patches on the cell membrane by fluorescence [135]. Also, embedded paramagnetic nanoparticles, when heated using an electromagnetic field, have been found to provide a useful means of controlling the liposome membrane permeability by inducing bilayer phase transition [133,137,148]. One important challenge of this novel technology is to achieve the phase transition locally, without changing the temperature of the medium which, in vivo, may damage the biological structures in the vicinity of the NP loaded liposome (this is also an important challenge in the case of the light actuation of gold NP, see Section 3.2.). To this end, unlike the case of SPIO (super paramagnetic iron oxide) particles [137], the use of individually stabilized iron oxide particles, has been reported to allow local heating of the membrane without major influences on its environment [133]. This system then allows for directed and controlled delivery of active agents from vesicle systems. In addition, of course, this approach allows for a much lower and controlled energy input, which is a desired effect, especially when considering potential medical applications.