Katharina
Bressel
* and
Michael
Gradzielski
*
Technische Universität Berlin, Stranski Laboratorium für Physikalische und Theoretische Chemie, Straße des 17. Juni 124, 10623 Berlin, Germany. E-mail: katharina.bressel@tu-berlin.de; michael.gradzielski@tu-berlin.de
First published on 4th February 2015
The formation of stable vesicles with a controlled size and high stability is an important matter due to their wide application in pharmaceutical and detergency formulations and as drug delivery vehicles. One can control the size of spontaneously formed vesicles in mixtures of zwitterionic and anionic surfactants by the admixture of small amounts of an amphiphilic copolymer of the PEO-PPO-PEO type. Of course, this effect should depend largely on the molecular architecture of the copolymer employed which was varied systematically in this work, and the temporal evolution of aggregate size and final structure was followed by means of DLS and three main effects could be observed. First the size of the formed vesicles is the larger the higher the molecular weight (MW) of the polymer and the higher the polymer concentration. Secondly the amount of copolymer required to induce long time stability is inversely proportional to the fraction of PEO in the polymer. Finally the architecture for a given MW and PEO/PPO ratio has no effect on the vesicle structure but their structure is directly controlled by the length of the PPO block of the copolymer. Thereby by appropriate choice of type and amount of PEO-PPO-PEO copolymer one can exert comprehensive control over size and stability of unilamellar vesicles.
(1) |
That leads to the maximum disc radius Rdisc,max. given by eqn (2) beyond which discs will spontaneously close to form vesicles.
(2) |
This process was described theoretically41,44–49 and for the case of mixing anionic surfactants with cationic or zwitterionic surfactants was studied in detail by time-resolved small angle X-ray scattering (SAXS) for mixtures of TDMAO (tetradecyldimethylamine oxide) and various anionic perfluoro surfactants49–53 and for mixtures of TTAOH (tetradecyltrimethylammonium hydroxide) and TexaponN(70)-H (laureth sulfuric acid).54
Since the vesicle size and polydispersity in these systems are determined by the kinetics of the process and the cut-off radius for the disc growth, the idea to modify the vesicle properties by influencing the formation process is obvious43 and corresponding experiments were successfully performed using an amphiphilic polymer of the Pluronic type as admixture.55 Addition of L35 led to a modified mechanism of vesicle formation. The idea is, that the hydrophobic part of the copolymer becomes incorporated into the micellar disc rim, thereby leading to a decrease in the line tension. Highly time-resolved SAXS measurements that followed the mixing process of TDMAO and LiPFOS (lithium perfluorooctylsulfonate) in a stopped-flow device showed that the addition of L35 allowed to have systematic control of the size of the formed vesicles. The higher the polymer concentration in the mixture the lower the line tension, what led to a prolonged disc growth and a delayed vesicle formation at larger disc sizes. Thus the vesicle size can be controlled by the polymer concentration and these experiments also showed enhanced vesicle stability presumably due to steric interactions between the polymers.55 However, in particular the long time stability of unilamellar vesicles is a very important property and should depend to a large extent on the type of amphiphilic copolymer employed.
Accordingly, in this work we studied the effect of a systematic variation of the polymer architecture on the formation process by following the hydrodynamic radius by time resolved dynamic light scattering (DLS). The aim was to determine the function of the hydrophobic and the hydrophilic part of the polymer on the process and to find a recipe that gives the required type and concentration of polymer for the desired vesicle size and stability, as it is of central importance for the general applicability of the concept of polymer controlled vesicles. For that purpose we applied polymers of the Pluronic type (PEO-PPO-PEO) with varying total polymer length and varying ratio between the length of the hydrophobic and the hydrophilic part (thereby changing the hydrophilicity of the amphiphilic copolymer). We also compared L35 to 10R5, which have the same number of EO and PO units but a hydrophilic core with hydrophobic end caps instead of a hydrophobic core with hydrophilic end caps, and thereby investigate the effect of the molecular architecture.
All samples were prepared from stock solutions. The surfactant stock solutions and polymer-surfactant stock solutions were prepared by dissolving an adequate amount of surfactant or surfactant with polymer in water filtered twice by a Millipore water filter.
Samples were prepared from stock solutions by mixing pure LiPFOS surfactant solutions with TDMAO solutions that contained the polymers. Samples for measurements performed in the stopped-flow were prepared directly from stock solutions in the stopped-flow. Samples for measurements performed in the ALV-Goniometer system were prepared by choosing adequate amounts of stock solutions by weight and by mixing both components by vigorous shaking directly before the measurement.
Sample preparation was carried out using a BioLogic SFM-400 with a rectangular Hellma QS quartz cell (FC-15 cell) with an inner diameter of 1.5 mm provided by BioLogic and by sequential mixing of up to four solutions. Before mixing, these solutions were stored in thermostated reservoirs and were then pushed by step motors in controlled ratios and with a controlled flow rate through high efficiency mixers into the observation cell. Then the flow was stopped by a hard stop and the measurement started simultaneously by an electronic trigger. The length of the pathway to the cuvette defines the initial age of the sample and this dead time depends on the filling volume and the flow rate. To ensure efficient mixing the last mixer located directly under the observation cell was a high density mixer, which allows blockage of convection created by temperature and density effects, instead of the conventionally used Berger ball mixer. For all experiments the flow rate was set to 7 mL s−1 and the dead volume in combination with the high density mixer is 51.3 μL leading to a dead time of 7.3 ms.
The sample was illuminated with a frequency doubled diode pumped Nd-Yag laser with a wavelength of 532 nm and an output power of 20 mW (Compass 215M-20, Coherent). The intensity of the scattered light was collimated under 90° by a combination of a fibre collimator (60FC, Schäfter and Kirchhoff) and a single mode glass fibre cable (SMC, Schäfter and Kirchhoff) and recorded by a single photon counting module (Count-250C-FC, Laser Components). A Flex02-1D digital correlator was used to process the signal.
Measurements for longer times covering a time range of 30 s up to 12 days were performed with an ALV/CGS-3 Goniometer system with a 22 mW HeNe-Laser (λ = 622.8 nm) and employing a pseudo cross-correlation under a scattering angle of 90°. The samples were prepared 30 s before the first measurements and thermostated at 25 °C in a toluene bath during the measurement.
Intensity autocorrelation data from DLS measurements were evaluated with monoexponential fits (eqn (3)). With increasing time for samples, where ageing was observed (for low polymer content), a small shoulder could be observed in the correlation functions, that indicates the presence of large aggregates. In that case the data were evaluated with double exponential fits (eqn (4)). An example is given in the ESI material (Fig. S.1†).
g(2)(τ) = 1 + βexp(−Γτ) | (3) |
g(2)(τ) = 1 + (β1exp(−Γ1τ) + β2exp(−Γ2τ))2 | (4) |
The translational diffusion coefficient was calculated by using the relation DT = Γ/q2, with Γ being the relaxation rate of the exponential fit, and q being the magnitude of the scattering vector (q = 4πn/λsin(θ/2) with refractive index n, wavelength of the light λ, and scattering angle θ). The hydrodynamic radius Rh was calculated from the translational diffusion coefficient DT by employing the Stokes-Einstein-relation (eqn (5)) with the Boltzmann constant kB, temperature T, and viscosity of the solvent η.
(5) |
In the analysis the hydrodynamic radii were determined from the faster relaxation rate since the slower relaxation rate only appeared, when ageing took place and represents large vesicles or agglomerates produced by ageing.
Of course, it has to be noted the eqn (5) is originally valid for dilute diffusing particles and the radii obtained by us on the vesicle systems are apparent hydrodynamic radii. However, if we describe the vesicle–vesicle interactions by a hard-sphere potential and following the work by Phillies56 one arrives at a ratio between real (D0) and apparent hydrodynamic radius close to one, as depicted in Fig. S.2.† Similarly even if one considers electrostatic forces for the colloidal interaction between the vesicles the effect on the apparent hydrodynamic radius should be small. In general, the apparent diffusion coefficient is described by:
(6) |
For that purpose a LiPFOS solution was mixed with a TDMAO solution with varying polymer content. The total surfactant concentration in the mixture was 50 mM and the ratio between TDMAO and LiPFOS was 55:45 since this ratio had been proven before to deliver very monodisperse vesicles.51 To study the evolution of aggregates in the mixtures the hydrodynamic radius was measured as a function of time. As the amphiphilic copolymer is the key component for controlling structure and stability in this study we aimed at elucidating the effect of its architecture on this stabilizing property. For that purpose we varied the arrangement of the EO/PO units for constant sum formula, the EO/PO ratio, i.e., the hydrophilicity of the polymeric surfactant, and its total length.
Fig. S.3† shows the time dependent development of the hydrodynamic radii of aggregates in the system TDMAO-LiPFOS + 10R5 for different polymer concentrations. Both systems show an increase of the hydrodynamic radius (when compared after some minutes) with increasing polymer content and for both systems the hydrodynamic radius increases shortly after mixing (within the first 20–50 s). The observed increase is due to the growth process of the disc-like micelles, that precedes the vesicle formation. Afterwards a further increase in the vesicle size due to ageing can be observed. For polymer concentrations ≥0.1375 mM the hydrodynamic radius reaches a plateau after about 100 s and then remains constant for a longer period of time, i.e., the vesicles can be considered as colloidally stable. That means for polymer contents of ≥0.1375 mM the expected vesicles can be efficiently stabilised by the polymer. For L35 the vesicle radius stays constant for a period of at least 2 weeks (our experimental observation time) while for 10R5 after reaching the plateau still a slight increase of the vesicle radius can be observed which means that the 10R5 is somewhat less effective in long term stabilization. This could be associated with the fact that the EO22 chain that is anchored on both ends within the vesicle bilayer is protruding less far out into the aqueous solution than a single EO11 chain, and in addition their number is twice that high for L35.
Fig. 1 shows the development of the hydrodynamic radius Rh for both polymers and they have the same effect on the aggregates. The vesicles reach the same radius for similar polymer concentrations and the process of vesicle formation shows a very similar time-dependent behaviour.
Fig. 1 Time dependent development of the hydrodynamic radius Rh comparing L35 and 10R5 at two different polymer concentrations (mixture: TDMAO:LiPFOS (55:45) 50 mM + polymer, 25 °C); black squares: mixture without polymer, red symbols: mixture with L35, green symbols mixture: with 10R5, diamonds: c(polymer) = 0.01375 mM, circles: c(polymer) = 0.1375 mM, triangles: c(polymer) = 0.55 mM (results from static light scattering are given in Fig. S.6†). |
Fig. 2 shows the hydrodynamic radii of the vesicles in solution for all polymer contents after 100 s, where the plateau region begins for polymer contents of ≥0.1375 mM. In the graph two regions can be observed. In the area for a polymer content of 0–0.055 mM the hydrodynamic radius of the aggregates increases strongly with the polymer content while in the area for a polymer content of ≥0.1375 mM this dependency is less pronounced. The first area is the area where relatively fast ageing can be observed while the vesicles in the second area stay stable for at least two weeks. In total it is found that the L35 and 10R5 show no significant difference in their behaviour, i.e., the order of the EO and PO blocks has no significant influence on the vesicle size and the stability, and what counts is only the total composition of the copolymer.
The overall behaviour of the evolution of the hydrodynamic radii for F38 addition is similar to that observed with L35. Comparing the hydrodynamic radii in both systems 100 s after mixing no significant difference between both polymers can be observed (see Fig. S.4†). This leads to the conclusion, that the decrease in the line tension is the same for L35 and F38. This is not surprising since the hydrophobic parts in both polymers are the same and in the proposed mechanism the hydrophobic part of the polymer is supposed to be responsible for the decrease in the line tension.
However, in the system with L35 stabilisation is only observed for polymer concentrations ≥0.1375 mM while in the system with F38 stabilisation is already achieved for 0.055 mM. Since the hydrophilic chains are longer in F38 than in L35 apparently they are more efficient in covering the vesicle and lead to a more pronounced steric stabilisation, i.e. vesicle stability depends on the length of the EO block.
Fig. 3 and 4 show the time dependent development of the hydrodynamic radii in the systems TDMAO-LiPFOS + F38 (F88, F108). Comparing the hydrodynamic radius 100 s after mixing in the three different systems as a function of the polymer concentration shows a strong increase of the hydrodynamic radius with the polymer length (see Fig. 5). For the same polymer number density (Fig. S.5†) in the mixture the formed vesicles are larger the longer the polymer chain is, which is interesting as the PPO block is quite long (6.4, 16.4, and 20.8 nm stretched length for F38, F88 and F108, respectively) compared to the bilayer thickness. That means a larger polymer leads to a stronger decrease in the line tension of the discs and a stronger anchoring within the vesicle bilayers. Since an increase in the length of the hydrophilic chain did not have an effect on the vesicle radius it can be concluded, that the increase in the vesicle radius is caused by the hydrophobic part of the polymer. An explanation is that the polymers orientate themselves along the disc rim, so that with increasing length of the hydrophobic part of the polymer less molecules are needed to cover the disc rim thereby causing a higher decrease of the line tension per polymer molecule (see Fig. 6 and S.5†). This agrees with the observation that L35, F38 and 10R5 that all possess similarly sized short PPO blocks show an almost identical effect on the size change of the vesicles (Fig. 5), i.e. this effect seems to be controlled by the hydrophobic PPO part.
Fig. 3 Time dependent development of the hydrodynamic radius Rh (mixture: TDMAO:LiPFOS (55:45) 50 mM + F38, 25 °C) (results from static light scattering are given in Fig. S.7†). |
Fig. 4 Time dependent development of the hydrodynamic radius Rh; mixtures: TDMAO:LiPFOS (55:45) 50 mM: black circles; TDMAO:LiPFOS (55:45) 50 mM + F88 (0.0137 mM, 0.0275 mM, 2.75 mM), 25 °C, red crosses; TDMAO:LiPFOS (55:45) 50 mM + F108 (0.0137 mM, 0.055 mM, 0.275 mM), 25 °C, green diamonds (results from static light scattering are given in Fig. S.8†). |
Fig. 5 Hydrodynamic radius 100 s after mixing in the systems TDMAO-LiPFOS + F38, F88, and F108 at 25 °C. |
Fig. 6 Polymers with a larger hydrophobic part are more effective in stabilising the disc, so that less polymer is necessary to achieve higher disc radii. |
On the other hand comparing the time dependent development of the hydrodynamic radius in different systems with polymers of different hydrophilic chain length, it can be seen that the longer the hydrophilic part is the lower is the number concentration of polymer needed to reach the plateau area in the DLS measurements, i.e. to achieve long-time stability (Fig. 7a). It is interesting to note that the total amount of EO units in the mixture is rather similar for all the copolymers studied, irrespective of the length of the EO chains (Fig. 7b), i.e., apparently the most important quantity that controls the stability is the total amount of EO in the stabilizing polymer shell.
Fig. 7 Polymer content where the plateau area in the DLS measurements is reached in dependency of the length of the hydrophilic part in the polymer at 25 °C. |
In the following we want to estimate the polymer concentration at which the vesicle bilayer can be expected to be fully covered. The steric hindrance of PEO chains attached to an vesicle surface can be described using the concept of de Gennes.58 It assumes that at low polymer density on the vesicle surface the PEO chains are in a mushroom-like conformation each PEO chain covering an area AEO-chain according to eqn (7).
AEO-chain = π(Nk3/5lk)2 | (7) |
From the polymer concentration cpolymer and the corresponding hydrodynamic vesicle radius Rh the area A that is available for each polymer molecule on the vesicle surface can be calculated by eqn (8) where ϕ (for calculation see ESI†) is the volume fraction of amphiphilic material, c(pol) is the polymer concentration, and vves,shell ≈ 4πRh2d is the volume of the vesicle shell.
(8) |
From A = AEO the concentration cpol,stab can be calculated, where the whole vesicle surface is covered with polymer coils, when two vesicles approach each other, so that vesicle fusion is hindered and stabilisation can be expected according to eqn (9) with d = 3.2 nm.
(9) |
Fig. 8 shows that the values for cpol,stab determined from measurements and from theoretical considerations are pretty similar except for L35 where Nk = 5 is very low.
Fig. 8 Black squares: polymer content where the plateau area in the DLS measurements is reached in dependency of the length of the hydrophilic part in the polymer at 25 °C; red circles: polymer concentration cpol,stab where vesicle stabilisation can be expected according to eqn (9); green diamonds: concentration of polymer saturation caused by dense packing of PEO mushrooms on top of the vesicle bilayer; blue triangles: concentration of polymer saturation caused by dense packing of PPO in the vesicle bilayer. |
Similar calculations can give the concentration where the vesicle surface is saturated with polymer. Here the effect from the PEO chains and from the PPO chains has to be taken into account.
If the PEO chains remain in the mushroom conformation on the vesicle surface then full coverage of the surface can be assumed for 2cpol,stab. On the other hand when squeezing the chains together they can form brushes so that the concentration of saturation can be much higher.
Compressing the PPO chains that form a self-avoiding random 2-dim-coil in the vesicle surface is not so easily possible. The Flory exponent for a two dimensional geometry is ν = 3/4 instead of ν = 3/5 for a 3-dim-coil so that the area occupied by each PPO chain in the surface is πNk3/2lk2 with Nk = 7, 7, 18, and 23 being the number of Kuhn units in L35, F38, F88, and F108. Analogous calculations to eqn (9) lead to the concentration of polymer saturation in the vesicle as given as blue triangles in Fig. 8.
In the system TDMAO:LiPFOS (55:45, 50 mM) the time dependent evolution of the z-average of the vesicle size RZ was analysed with eqn (10) developed by von Smoluchowski, where the vesicle fusion is governed by diffusion and where the same kinetic constant K is assumed for all vesicle sizes.60,61
(10) |
Fig. 9 shows that the growth process can not be described by one kinetic constant but that the kinetic constant must depend on the vesicle size (dark green line). To introduce the size dependency of K into eqn (10), an empirical approach was used for K as given in eqn (11) that describes an easier fusion for smaller vesicles, which can be ascribed to the fact that with increasing curvature of the membrane its energy per area increases. For m we assumed a constant value of 2 that would correspond to a situation where the activation energy EA depends directly on the bending energy of the fusing vesicles (as given by eqn (12),42 where C is the energy of a transitional state for fusion and Afus is the area involved in the fusion step, and Rv is the vesicle radius). It can be stated that eqn (12) is at best a very rough approximation for the real situation but in agreement with our experimental observation (where it should be noted that the fits are not very sensitive to the choice of m, but, of course, without having a size dependence of the rate constant, m = 0, one would have a largely different behaviour, as seen by the green solid line in Fig. 9).
(11) |
(12) |
Fig. 9 Time dependent evolution of the z-average of the vesicle size Rh for a sample of TDMAO:LiPFOS (55:45, 50 mM) (black open squares), fit with eqn (10) (dark green line), and fit with eqn (10) taking eqn (11) into account (red line). |
Eqn (10) together with eqn (11) does now describe the overall growth process as shown in 9 (red line).
The same analysis with eqn (10) and (11) was carried out for the mixtures with L35 so that for every polymer concentration K was determined as a function of the radius. The additional activation energy EA,pol of the vesicle fusion process caused by the polymer can be calculated from the kinetic rate constant in the systems with polymer Kpol and without polymer Kpure by eqn (13) at the initial vesicle in the system with polymer Rh,ini to ensure comparability between both systems. It has to be noted that the characteristic time for the vesicles to remain stable at the beginning of the vesicle growth process can not be compared for different polymer concentrations since the kinetic rate constant strongly depends on the vesicle radius and since the characteristic time includes the initial vesicle number density that decreases with the vesicle size.
(13) |
E A,pol increases significantly above a polymer concentration of 0.1375 mM from 0kBT to 3kBT. So when the polymer completely covers the vesicle surface the vesicles are stabilised by an additional activation energy for vesicle fusion of 3kBT (Fig. 10).
Fig. 10 Additional activation energy EA,pol for the vesicle fusion caused by the polymer coverage of the vesicles in dependency on the polymer concentration. |
Accordingly in our investigation the length of the hydrophilic part and the hydrophobic part of the polymer were varied and a conventional Pluronic with a hydrophobic core was compared to a Pluronic with a reversed geometry with a hydrophilic core. Time resolved DLS measurements, that allowed to follow the formation process and to determine the long time stability, led to three conclusions.
First comparing L35(EO11PO16EO11), F38(EO46PO16EO46), F88(EO102PO41EO102), and F108(EO132PO52EO132) we could show that, apart from the amount of added copolymer, mainly the length of the hydrophobic part of the polymer is responsible for the vesicle size. It lowers the line tension of the disc-like micelles by accumulating in the disc rim and the higher the amount of hydrophobic material incorporated in the disc rim the lower the line tension. Presumably the PPO chains are oriented along the disc rim, as this location is not only favourable for the relatively long PO blocks but with their two EO blocks they also have a much bulkier head group, which can be well accommodated in the rim. The longer the PO block the more stable the disc and the larger are the vesicles that are formed after spontaneous closure.
In contrast, the hydrophilic part of the polymer is responsible for the stability of the vesicles. The longer the PEO chain the lower the polymer concentration needed to prevent vesicle ageing. However, the scaling is such that the onset of colloidal stability is simply determined by the total concentration of EO units in the stabilizing PEO shell. The point where vesicle stabilization sets in is in good agreement with the polymer concentration estimated for having dense packing in the vesicle bilayer (assuming a mushroom configuration for the PEO blocks protruding into the aqueous surroundings).
Thirdly hardly any difference could be observed between polymers with a hydrophobic core or a hydrophilic core, when they have the same number of EO and PO units. Only the long-term stability of the vesicles is somewhat reduced for the case of the inverse Pluronic, which can be attributed to the lower number of EO chains and their lower extension into the surrounding aqueous phase.
Hence gaining control over the vesicle formation process by addition of amphiphilic polymers of the Pluronic type, offers a new and versatile mechanism to tailor the size and, in particular, the stability of vesicles. Depending on copolymer architecture and concentration one has wide control over the vesicle properties. In order to achieve the demanded vesicle properties only small amounts of these nontoxic, cheap and easily available polymers have to be added and this study gives the information, which polymer type and concentration will lead to the desired vesicle structures.
In summary, the addition of copolymer of the PEO/PPO type constitutes an elegant way of controlling the size and stability of spontaneously forming unilamellar vesicles. This can be achieved by adding rather small amounts of appropriately chosen copolymer, where the controlling ability and stability effects become more pronounced for longer chain copolymers. This is an important finding as it allows to form tailor-made unilamellar vesicles, an ability that is frequently required for their application in colloidal formulations, for pharmacy, cosmetics or detergency.
Footnote |
† Electronic supplementary information (ESI) available: Materials and preparation procedures; instrumentational details about the stopped-flow device and the DLS; analysis of DLS data; additional graphs. See DOI: 10.1039/c4sm02746a |
This journal is © The Royal Society of Chemistry 2015 |