Controlled block copolymer micelle formation for ... · Controlled block copolymer micelle formation for encapsulation of hydrophobic ingredients Lebouille, J.G.J.L.; Vleugels, L.F.W.;

Controlled block copolymer micelle formation forencapsulation of hydrophobic ingredientsLebouille, J.G.J.L.; Vleugels, L.F.W.; Dias, A.A.; Leermakers, F.A.M.; Cohen Stuart, M.A.;Tuinier, R.Published in:European Physical Journal E : Soft Matter

DOI:10.1140/epje/i2013-13107-y

Published: 01/09/2013

Document VersionPublisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differencesbetween the submitted version and the official published version of record. People interested in the research are advised to contact theauthor for the final version of the publication, or visit the DOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

Citation for published version (APA):Lebouille, J. G. J. L., Vleugels, L. F. W., Dias, A. A., Leermakers, F. A. M., Cohen Stuart, M. A., & Tuinier, R.(2013). Controlled block copolymer micelle formation for encapsulation of hydrophobic ingredients. EuropeanPhysical Journal E : Soft Matter, 36(9), 1-12. [107]. DOI: 10.1140/epje/i2013-13107-y

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?

Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 12. Jun. 2018

DOI 10.1140/epje/i2013-13107-y

Regular Article

Eur. Phys. J. E (2013) 36: 107 THE EUROPEANPHYSICAL JOURNAL E

Controlled block copolymer micelle formation for encapsulationof hydrophobic ingredients

Jerome G.J.L. Lebouille1,2, Leo F.W. Vleugels2, Aylvin A. Dias1,3, Frans A.M. Leermakers4,Martien A. Cohen Stuart4, and Remco Tuinier2,5,a

1 DSM Biomedical, P.O. Box 18, 6160 MD Geleen, The Netherlands2 DSM ChemTech, Advanced Chemical Engineering Solutions (ACES), P.O. Box 18, 6160 MD Geleen, The Netherlands3 DSM Ahead, P.O.Box 18, 6160 MD Geleen, The Netherlands4 Laboratory of Physical Chemistry and Colloid Science, Wageningen University, Dreijenplein 6, 6307 HB Wageningen, The

Netherlands5 Van ’t Hoff Laboratory for Physical and Colloid Chemistry, Department of Chemistry, Utrecht University, Padualaan 8, 3584

CH Utrecht, The Netherlands

Received 21 May 2013 and Received in final form 7 August 2013Published online: 26 September 2013 – c© EDP Sciences / Societa Italiana di Fisica / Springer-Verlag 2013

Abstract. We report on the formation of polymeric micelles in water using triblock copolymers with apolyethylene glycol middle block and various hydrophobic outer blocks prepared with the precipitationmethod. We form micelles in a reproducible manner with a narrow size distribution. This suggests thatduring the formation of the micelles the system had time to form micelles under close-to-thermodynamiccontrol. This may explain why it is possible to use an equilibrium self-consistent field theory to predictthe hydrodynamic size and the loading capacity of the micelles in accordance with experimental finding.Yet, the micelles are structurally quenched as concluded from the observation of size stability in time.We demonstrate that our approach enables to prepare rather hydrophobic block copolymer micelles withtunable size and loading.

1 Introduction

Encapsulating active compounds in a controlled fashionis of paramount importance for applications in food [1–3] and pharmaceutical technology [4,5]. One can usenanosized micellar structures formed by amphiphilicmolecules in a selective solvent. Here we focus on wateras the (selective) solvent and consider block copolymerswith two apolar and one polar block, the so-called ABAblock copolymers where A is an apolar block and B apolar block. The hydrophobic entities are collected ina compact core, whereas the water soluble compoundremains hydrated and forms a corona. Flavors, vitaminsand drugs are often rather hydrophobic ingredients forwhich polymeric micelles are promising carriers, with po-tential for controlled encapsulation of compounds at highloadings. Additionally, using polymeric micelles offersroutes to control release, stability and bio-distribution ofactive agents in the body.

The bio-distribution mainly depends on the micellesize and corona structure [6–9]. It is known that small-sized micelles, e.g., below 30 nm, distribute freely in thehuman body due to a lack of tissue retention/obstruction.Larger-sized objects, i.e., exceeding 400 nm, may cause

a e-mail: [email protected]

problems in the vascular system, especially in the capil-laries which can easily be obstructed by such particles.Indeed, sizes below 100 nm result in relatively longcirculation times and these objects can accumulatein inflammatory or tumor tissues by the enhancedpermeability and retention (EPR) effect [10–14]. Thisphenomenon can be exploited to give passively targeteddrug delivery systems. There is ample evidence that theaverage size and its size distribution mainly determinethe biological fate, and therefore also the efficiency ofa treatment, when nanoparticles/micelles are used fordrug delivery purposes. The corona composition is alsoof importance for the distribution and tissue uptake ofthe particle. It has been shown that PEGylated entities,sometimes called stealth or “disguise” particles, haveeven longer blood circulation times [15]. Furthermore, thepresence of the cationic surfactant dimethylammoniumbromide on the surface of a particle was shown to improveuptake by arterial tissue [16,17].

There is a broad range of amphiphiles, for instancepoloxamers and PEO-PCL or PEO-PLGA (di- and tri-)block copolymers, that can be used to prepare micelleswith encapsulated hydrophobic compounds [18–28]. Rela-tively polar copolymers and surfactants that readily dis-solve in water form rather dynamic micelles with unimeric

Page 2 of 12 Eur. Phys. J. E (2013) 36: 107

exchange rates up to the microsecond time scale [29,30].Encapsulated compounds in such micelles will also be re-leased rapidly because usage as drug delivery systems is al-ways accompanied with significant dilution. A fast unimerexchange implies a high CMC and therefore a fast release,almost instantaneously upon administration.

We may distinguish thermodynamically stable sys-tems, which have an equilibrium size and typically anarrow size distribution that do not depend on theroute of how the micelles are formed, from kineticallyfrozen aggregates. For the latter the route of formationbecomes important. Kinetically frozen aggregates mayshow Ostwald ripening, a phenomenon that over extendedperiods of time large particles grow at the expense ofsmaller ones [31], similarly as emulsion droplets. Thegrowth of large particles at the expense of smaller ones isfacilitated by the solubility of the constituent moleculesin the solvent. Copolymers with a sufficiently long hy-drophobic block form micelles at very low critical micelleconcentrations. The micelles have a compact hydropobiccore and a hydrated corona. Although the chain partsin the corona continuously change conformations dueto thermal motion, the chain parts that form the coreare much less dynamic. Indeed, very often the core is inthe glassy state. The combination of a low (unimeric)polymer concentration in the bulk (low CMC) andthe slow dynamics in the core (glassy core), results inmarginal Ostwald ripening. Micelles with limited Ostwaldripening have a long shelf life.

A complication is that such “frozen” micelles cannotbe prepared by simply dissolving the copolymer in wa-ter, since water is a non-solvent for the relatively longhydrophobic blocks of the copolymers. We apply the “sol-vent shifting” or nanoprecipitation procedure [32–35] toprepare particles in aqueous solutions composed of other-wise water-insoluble molecules or polymers. In this pro-cedure, the (co)polymers are first dissolved in an organic(good) solvent for both blocks. The solvent should alsohave a reasonable miscibility with water. This solution issubsequently added, rapidly, to an excess of water. Dur-ing the mixing procedure the block copolymers gradualgo from a good solvent to selective solvent conditions, be-cause the organic phase is dispersed in the water phase.As a result, the core-forming blocks aggregate (as theyare insoluble) and form the micellar cores. The solvophilicblocks remain solvated during the solvent exchange pro-cess and accumulate outside the core to form a corona. Itis expected that in the core the organic phase will pref-erentially accumulate. This keeps the micelles mobile forsome time. Depending on the conditions, however, the or-ganic phase may be lost for the micelles and then the poly-meric micelles go into a “frozen” or “dead” state, mean-ing that they no longer can exchange copolymers betweeneach other. One may intuitively expect that the solvent ex-change is very fast and the micelles become very quicklytrapped in a frozen state. However, this is not always thecase and one can, alternatively, imagine that the micelleshave sufficient time to equilibrate their size and possiblyto some extent their size distribution. In such a scenario,it is feasible that the micelle size and micelle size distribu-

tion are dictated by some equilibration process that con-tinued in one way or another until (relatively suddenly)the constituent molecules lose their mobility. In this lineof reasoning it is fair to try to attempt a modeling ef-fort to seek guidance to rationalize the relation betweenmolecular structure and micellar topology.

To this end we performed numerical Scheutjens-Fleer–self-consistent field (SF-SCF) computations. The methodand results are explained in [36]. SF-SCF is known to bevery accurate for densely packed polymer systems includ-ing micellar structures [37–40]. However, the theory pre-assumes that the molecules have reached their thermody-namic equilibrium. Although we are sure that the finalmicelles are kinetically frozen, we envision that it is pos-sible that we can find effective parameters that are rele-vant for the micelles that are being formed transiently andto some extent were under thermodynamic control. Themolecules form flower-like micelles in the dispersions stud-ied which is supported by unpublished cryo-TEM anal-yses and DLS measurements at higher triblock copoly-mer concentrations. This implied that the corona is builtup by looping chains, which arguably have some advan-tage for targeting. The idea for this is that, when a mi-nority amount of the triblocks is replaced by copolymersfor which one hydrophobic block is replaced by a (water-soluble) targeting moiety, one has flower-like micelles in-termixed with polymers that have their targeting moietydangling well outside the corona of the remaining triblockcopolymers. This makes the targets to be better, biologi-cally, accessible.

In the following we will first give information on thepolymeric species. In the results section we will focus onthe characterisation of the micelles and elaborate on theuse of SF-SCF modeling. In our conclusions we argue thatthe micelles formed by the precipitation method assume astructure that resembles equilibrium characteristics thatwere present somewhere in the production process.

2 Experimental aspects

2.1 Copolymers and amphiphiles

The used polymeric surfactants are all amphiphilic copoly-mers with a general composition of A-B-A, where A is thehydrophobic group poly(lactic-co-glycolic)acid, poly(ε-caprolactone) (further referred to as: PLGA and PCL)and B the hydrophilic group polyethylene oxide, polyvinylpyrolidone, polyvinyl alcohol (PEO, PVP, PVA). The (co-)polymer blocks which these triblock copolymers are madeof comply with the following prioritized requirements:non-toxic, biocompatible, biodegradable and excretable.

The hydrophilic part of the copolymer is PEO, exhibit-ing good water solubility and meets the above require-ments. Although its non degradability PEO is non-toxicand can easily be removed from the body by the normalexcretion pathways as long as the molecular weight is be-low 20 kDa [41]. The hydrophobic part consists of knownbiodegradable polymers used in commercially availabledrug delivery applications: PLGA and PCL.

Eur. Phys. J. E (2013) 36: 107 Page 3 of 12

2.2 Stability; dynamic, static or dead/frozen micelles

In order to prepare “frozen” micelles in water, several re-quirements need to be met. Importantly, water should bea selective solvent for the surfactant/copolymer, that is, anon-solvent for one block and a good solvent for the other.It is known that the log CMC ∝ Nt, where Nt is the num-ber of apolar segments in the copolymer/surfactant. Pro-vided the non-solvent block is long enough (Nt � 1), thisresults in extremely low CMC-values, inhibiting Ostwaldripening. Finally, the conditions of the core forming blockshould be such that the mobility of the chains is retarded,that is, preferably there should not be a plasticizer in thesystem. Typically, these requirements preclude using thenormal way of making micelles by dissolving the surfac-tant/copolymers, because of the exceedingly low criticalsolution temperature (LCST), in the surfactant scienceoften referred to as the Krafft temperature [42,43].

To overcome this, we opted for the nanoprecipita-tion method: One first co-dissolves the active ingredi-ent, the stabilizer and the excipient, that is, the surfac-tant/copolymer together with a compound that protectsor tunes the release of the active ingredient, in a suitablewater-miscible common solvent and then precipitates itinto a nanoparticulate form in water, which is a selectivesolvent. The common solvent is used to bring the copoly-mers into a homogeneous molecular solution, from whichthe self-assembly into micelles proceeds when added to aselective solvent: one block avoids the selective solvent andforms the core and the other blocks remain solvated andform the corona. Typically, the macromolecular nature ofthe species involved prevents the molecular dispersion ofthe copolymers. In other words, the bulk concentrationof the copolymers (unimers) is extremely low resulting instatic or frozen micelles, minimizing the possible elutionof active ingredients and or excipients out of the particle.

The micellar shape strongly depends on the copoly-mer composition. Typically the spherical shape is stableas long as the dimension of the (highly solvated) coronaH exceeds that of the (almost solvent free) core Rc. Thecore forming blocks collapse and then occupies a volumeproportional to its length. Thus the size of the core is pro-portional to Rc ∝ N

1/3t . The corona block, on the other

hand, remains solvated and is grafted by the ends onto thecores. Due to the lateral interactions, the corona blocksbecome stretched outward and form a molecular brush.The height of the brush H (equal to the corona size) isproportional to the degree of polymerization Nt, that isH ∝ N1

t . Hence, the stability of the spherical micelle mayoccur already for relatively short corona blocks (Nc < Nt).Of course, in principle it remains possible that the dimen-sion of the core and that of the corona are comparable.Then the cylindrical or lamellar structures become thegeometry of choice. In the current project, however, thecorona block is long (dimensionally big) enough to expectspherical micelles to form.

The molecular weight of the micelle, that is the num-ber of copolymers in one micelle, is also controlled by thecopolymer composition. Basically, the longer the corona

block the smaller the aggregation number, whereas an in-crease in the molecular weight of the core forming blocksincreases the aggregation number. The molecular weight ofthe micelle is also a strong function of the driving force formicellisation. In some mixtures of a common solvent anda selective solvent the driving force is expected to increasewith the increase of the ratio “selective solvent”/“commonsolvent”. Indeed, during the precipitation procedure weexpect the driving force to be an increasing function ofthe time after the addition of the selective solvent. In theSF-SCF modeling [36] we have simplified this process bytaking a simple selective solvent, which presents a moder-ate driving force for micellisation. This leads to predictionsin trends in micelle size and molecular weight which candirectly be tested experimentally.

The micelles that can be generated by the precipita-tion method have ideal sizes for drug delivery formula-tions. These formulations are very stable in time and theelution profiles can be governed by the excipients, speciesonly present in the core of the micelle together with theactive ingredient, and the used copolymer, present on theinterface between micelle core and corona. Most releaseprofiles are governed by diffusion and or desorption, anexcipient can alter the desorption of an active ingredi-ent from micellar core to corona and thus having an ef-fect on release. In other cases an excipient can also actas a preservative for the active ingredient. Butylhydroxy-toluene [44,45] (BHT) or β-carotene [46] are often used asa preservative (antioxidant) to avoid oxidative decline ofthe active ingredient due to oxygen, hydrolysis, salt, pHor other chemical species chemically altering the originalactive ingredient.

2.3 Self-consistent field theory and molecular model

The theoretical toolbox for the study of self-assembly ofcopolymers is not very large. Important for the successis that the molecular structure of the copolymers is rel-atively accurately accounted for both from a structuraland from an¡ interaction point of view. Molecular simula-tions can be used, but effectively need a significant coarsegraining step in order to keep the simulation time withinreasonable bounds. As an output, simulations give a verydetailed picture, that is, the micelle structure presents it-self in full glory. Importantly, the thermodynamic infor-mation of the system is typically lacking and therefore itis hard to estimate the relevance of a particular micellarstructure for a practical system of interest.

In this work we opt for an approximate mean-field approach. More specifically we choose for theScheutjens-Fleer self-consistent field (SF-SCF) method.The important argument in favor of this approach isthat the method starts with a (mean field) free-energyformula and therefore the results are readily analyzedin the thermodynamic context. This means that onecan estimate more easily the relevance of a particularresult for the experimental system. The optimization ofthe mean-field free energy gives structural informationof the micelles, which is still rather detailed. The fact

Page 4 of 12 Eur. Phys. J. E (2013) 36: 107

Table 1. Number of Kuhn segments (NK) of blocks used inthe copolymers studied.

Blocks in copolymers NK

PEO 6k 60

PEO 3k 30

PLGA 7.5k 60

PLGA 3.75k 30

PCL 1.9k 17

β-carotene 5

Rapamycin 9

Table 2. χ-parameters for the monomer-solvent interac-tion used in SF-SCF computations (EG: ethylene glycol,LGA: Lactic-co-glycolic, CL: caprolactone, BC: β-carotene andRapa: rapamycin). The block lengths and the correspondingKuhn lengths are collected in table 1.

Monomer-solvent interaction χ

EG-water 0.4

LGA-water 1.6

CL-water 3.0

LGA-EG 1.0

CL-EG 1.0

BC-water 4.0

BC-EG 1.0

BC-LGA 0.4

Rapa-water 6.0

Rapa-EG 1.0

Rapa-LGA 0.4 or 2.0

that the micelles are composed of copolymers that aredensely packed appears important. For this situation eachmolecule interacts with many neighbors and therefore themean-field approximation is relatively accurate. On topof this the calculation time is extremely short (in com-parison to simulations). Last, but not least, molecularlyrealistic models can be implemented with relatively feweffective interaction parameters. Although, in principlethe interaction parameters can be measured, in practicethey are largely unknown. The same holds true, obviously,for the current systems under investigation.

Polymers are considered to be composed of (so-called)Kuhn segments, see table 1. A Kuhn segment occupiesone grid unit (r) in our calculations which corresponds to0.8 nm. This allows the use of freely jointed chain model.Within this chain model, there exists an efficient proce-dure to compute the partition function and thus full ther-modynamic information can be obtained. In this approachthe architecture of the chain parts in the copolymers is ac-curately accounted for. The interactions are accounted forusing the Bragg-Williams approximation, which ignoreslocal density correlations analogous to the Flory-Hugginstheory for polymer solutions. The Flory-Huggins interac-tion parameters that specify the solvent quality of the seg-ments, as well as the interactions between the segmentsare easily estimated, see table 2.

(a)

(b)

Fig. 1. Equilibrium radial density profiles of water, totalcopolymer, PLGA blocks and PEO block as a function ofthe center from a micelle r. In panel (a) β-carotene is addedto a PLGA30PEO30PLGA30 triblock copolymer micelle. Inpanel (b) rapamycin is added to a PLGA60PEO60PLGA60 tri-block copolymer micelle. In this way we compare via SCF andDLS measurement the influence on the hydrodynamic diame-ter and the difference between the theoretical calculation andthe experiment. The SCF hydrodynamic diameter in panel (a)is calculated as follows: (20 × 0.8)(nm) × 2 = 32(nm) and forpanel (b): (28 × 0.8)(nm) × 2 = 45(nm).

As the accurate value of the interaction parametersdepends on how many details of the polymeric chainsare accounted for, it is non-trivial to tabulate these.Hence, one should calibrate the parameter for each sys-tem under investigation. This means that there shouldbe relevant experimental observables to do so. In prac-tice therefore, one typically selects a particular case (herea particular copolymer system), adjusts the interactionparameters somehow until there is a reasonable matchbetween, e.g., the micelle size predicted by theory andfound experimentally. Subsequently, the set of parametersis fixed and the model is used to predict the structuralfeatures of the micelles for other systems. In fig. 1 equilib-rium density profiles of active ingredient loaded PLGA-based triblock copolymer micelles are shown comparingthe SF-SCF calculated and DLS measured hydrodynamicdiameters. In fig. 1(a) the loaded active ingredient is β-

Eur. Phys. J. E (2013) 36: 107 Page 5 of 12

OHHO +

O

O

O

n

O

+O

O

O

O

T=150ºC, nitrogem atmosphere

Tin(II)-octoate

OO

O

O

O

O

O

O

O

On or m x

H

yx

H

y

OHHO +

n

T=150ºC, nitrogem atmosphere

Tin(II)-octoate

OO

nq

O

O

O

HO

O

OH

q

Fig. 2. Schematic of the ring-opening polymerization of the triblock copolymers. In our case x = y, and x + y is the number ofD,L-Lactide and Glycolide repeating units randomly distributed in the hydrophobic end blocks. In the case of m = 136 ethyleneoxide repeating units x + y is 115, referred to as TBB1, and for n = 68 ethylene oxide repeating units x + y is 58, referred to asTBB2 for the PLGA based triblock copolymers. For the PCL based triblock copolymer, referred to as TBC1, there are n = 68ethylene oxide repeating units with q = 17 caprolactone repeating units.

Table 3. PLGA-based triblock copolymer synthesis weights.

Triblock Initiator; hydrophilic Hydrophobic end blocks CatalystcopolymerID

middle block; mass(grams)

D,L-Lactide(grams)

Glycolide(grams)

Sn2Oct(mg)

TBB1 PEO-6000-diol; 2.8467 3.9131 3.2471 4.4TBB2 PEO-3000-diol; 2.8573 3.9098 3.3233 4.4

Table 4. PCL-based triblock copolymer synthesis weights.

TriblockcopolymerID

Initiator; hydrophilicmiddle block; mass(grams)

Hydrophobicend blocksε-caprolactone(grams)

Catalyst-solution58.10 mg Sn2Oct/5 mL Hexane (mL)

TBC1 PEO-3000-diol; 8.8554 11.1555 1.000

carotene, see fig. 6 for the stability data of the β-caroteneloaded micelles. In fig. 1(b) the loaded active ingredientis rapamycin, see table 16 for the stability data of therapamycin loaded micelles. We refer to ref. [36] for moredetails.

3 Materials and methods

3.1 Materials

All triblock copolymers were synthesized by ring-openingpolymerization of D,L-lactide, glycolide or caprolactoneusing PEO-(3.0 kDa and 6.0 kDa)-diol as an initiator andstannous octoate as a catalyst at 150 ◦C under vacuum.D,L-Lactide and glycolide were purchased from Purac(Goringchem, the Netherlands), polycaprolactone and β-carotene from Sigma (St. Louis, USA). PLGA 20 kDawas purchased from Ingelheim Boehringer (Ingelheim amRhein, Germany). Polycaprolactone 80 kDa was purchasedfrom Solvay (Oudenaarde, Belgium) PEO (3.0 kDa and6.0 kDa) and Sn2Oct were purchased from Aldrich (St.Louis, USA). Acetone was purchased from BASF (Bayern,Germany). Rapamycin was purchased from Oscar Tro-pitzsch (Germany).

3.2 Methods

3.2.1 Ring-opening polymerization method for the triblockcopolymers [47,48]

PLGA-PEO-PLGA triblock copolymers

The PEO was weighed into a two-necked round bottleflask after drying for 24 hours in a vacuum oven at 90 ◦Cand subsequently placed in an oil bath at 150 ◦C. A vac-uum was employed for at least 60 minutes before con-tinuing the synthesis. The addition of lactide and gly-colide (molar ratio of lactide:glycolide = 50:50) was car-ried out by removing the vacuum and at the same timeflushing with nitrogen gas. When a homogenous melt, un-der stirring, was obtained the catalyst, stannous octoate(Sn2Oct), was added in the same way as the addition ofthe monomers.

The reaction conditions were maintained for 20 hourswhereafter the vacuum was replaced by nitrogen gas andthe ring-opening polymerization was completed, see fig. 2for the reaction scheme and table 3 for synthesis weights.The copolymers obtained in this way are listed in table 5.

PCL-PEO-PCL triblock copolymersThe PEO along with ε-caprolactone was charged in a100mL round-bottomed flask. The reaction mixture was

Page 6 of 12 Eur. Phys. J. E (2013) 36: 107

Table 5. Triblock copolymers ID and composition.

TriblockID

Triblock (A-B-A) copolymer composition

PLGA-block (A) PEO-block (B) PLGA-block (A)

TBB1 7.5 kDa 6 kDa 7.5 kDa

TBB2 3.75 kDa 3 kDa 3.75 kDa

PCL-block (A) PEO-block (B) PCL-block (A)

TBC1 1.9 kDa 3 kDa 1.9 kDa

Table 6. Different weights for the reproducibility test on the nanoprecipitation process for “empty” triblock copolymer basedmicelles.

Triblock ID Triblock mass (mg)/mL acetone

Sample ID Group ID

TBB2 59.89 TBB2R1

TBB2RTBB2 60.07 TBB2R2

TBB2 60.13 TBB2R3

TBC1 60.07 TBC1R1

TBC1RTBC1 60.28 TBC1R2

TBC1 60.19 TBC1R3

heated to 100 ◦C and stirred till a homogenous mixturewas formed. A catalyst stock solution of tin(II)octoatewas prepared in hexane. 1mL of the catalyst stock solutionwas added to the reaction mixture at 100 ◦C. The reactionmixture was further heated to 150 ◦C for an additional18 hours (overnight) to allow the reaction to proceed.The following morning the reaction mixture was cooledto room temperature, an off white waxy solid materialwas obtained, table 4 shows the synthesis weights. Thecopolymers obtained in this way are listed in table 5.

3.2.2 Purification of the synthesized triblock copolymers

The triblock copolymer was dissolved in acetone at aweight percentage of 10-20%, filtered over an Acrodiscpremium 25mm Syringe filter, GxF/0.45μm PVDF mem-brane, to remove particulate impurities and dust particles,which can interfere with the nanoprecipitation process,collected into an 500mL PTFE beaker and evaporatingof the solvent over night (10-12 hours) at maximum 40 ◦Cand minimum 300mbar.

3.2.3 Nanoprecipitation/nanoparticle preparation method

Typically, 300mg of copolymer was weighed and dissolvedin 5.000mL of acetone resulting in a clear copolymer so-lution after 30 minutes on an orbital shaker. Prior tothe nanoprecipitation process all solutions (milliQ water,copolymer-, copolymer/active ingredient-, copolymer/andcopolymer/active ingredient/excipient-solutions) were fil-tered over an Acrodisc LC25 mm Syringe filter 0.2 μmPVDF membrane. See fig. 3 for basic nanoprecipitationprocess setup. In order to obtain excipient-loaded micelles,300mg copolymer was weighed and subsequently dissolved

Fig. 3. Schematic of basic nanoprecipitation process setup.

in solvent, the excipient was weighed and dissolved in thecopolymer solution. The excipients were chosen from twodifferent homopolymers. The weight percentage of the ex-cipient in ratio to the copolymer was calculated as follows:weight% = [(excipient mass)/(excipient mass + copoly-mer mass) ×100]. A volume of 0.400mL of the copolymeror copolymer/excipient solution was added to 10.00mL ofaqueous solution with an Eppendorf pipette, the additionwith the pipette was carried out within one second, where-after the suspension was manually homogenized withinfive seconds.

Nanoprecipitation reproducibilityTo check the reproducibility of the nanoprecipitation pro-cess, three different formulations were made, weighed anddissolved, at three different days per triblock copolymertype. After nanoprecipitation the sample was measuredwithin 15 minutes. TBB2 and TBC1 triblock copolymerswere used to check the reproducibility of “empty” micelles.

Eur. Phys. J. E (2013) 36: 107 Page 7 of 12

Table 7. Different weights for the reproducibility test on the nanoprecipitation process for active ingredient, rapamycin, loadedmicelles for PCL-PEO-PCL triblock copolymer based micelles.

Triblock ID Triblock mass (mg)/mL acetone

Mass active ingredient(RAPA; mg)

Sample ID Group ID

TBC1 30.14 1.59 TBC1RE1

TBC1RETBC1 30.05 1.63 TBC1RE2

TBC1 30.11 1.54 TBC1RE3

The TBC1 triblock copolymer was also mixed with an ac-tive ingredient, rapamycin, to check the reproducibility onthe active ingredient loaded nanoprecipitation process.

Table 6 shows the concentrations used of the TBB2and TBC1 copolymer in the nanoprecipitation setup.0.400mL of the copolymer solution was precipitated in10.00mL MilliQ water.

Table 7 shows the weights of the used copolymertriblock together with the active ingredient rapamycin(RAPA) weights in the nanoprecipitation setup. 0.400mLof the copolymer/active ingredient solution was precipi-tated in 10.00mL MilliQ water.

Single excipient, homopolymer, loaded micellesThe interest in copolymer micelles is in part due to theirrelatively large loading capacity, even for relatively highmolecular weight compounds the following experimentswere carried out. All three triblock copolymers were testedtogether with a homopolymer as an excipient to determinethe loading capacity/capability and the relation betweenexcipient weight percentage and size. TBB1 copolymerwas made in a stock solution of 62.68mg TBB1 triblockcopolymer per mL acetone (1.2536 gram/20.00mL ace-tone). Different masses (see table 8) of PLGA 20k wereweighed into a vial. Afterwards 1.000mL of the TBB1triblock copolymer solution was added to all weighed ex-cipients. TBB2 copolymer was made in a stock solutionof 63.45mg TBB2 triblock copolymer per mL acetone(0.6345 gram/10.00mL acetone).

Different masses (see table 9) of PLGA 20k wereweighed into a vial. 1.000mL TBB2 triblock copolymersolution was added to the excipient vials. TBC1 copoly-mer was made in a stock solution of 63.21mg TBC1 tri-block copolymer per mL acetone (0.6321 gram/10.00mLacetone). Different masses (see table 10) of PCL 80k wereweighed into a vial and dissolved as the other copolymerexcipient solutions.

After complete dissolution on an orbital shaker, re-sulting in a clear copolymer/excipient solution readyto be precipitated in MilliQ water (0.400mL of thecopolymer/excipient solution was precipitated in 10.00mLMilliQ water).

Nanoprecipitation of single component loaded micellesTBB1 copolymer and TBB2 copolymer were dissolved inacetone, see table 11 for the triblock copolymer solutions.The excipient, PLGA 20k and the active ingredients ra-pamycin and β-carotene, were dissolved in acetone so-lution, see table 12 for the single component solutions.

Table 8. TBB1 triblock copolymer (PLGA-PEO-PLGA; 7.5-6-7.5) excipient (PLGA 20k) loaded micelles (62.69 mg TBB1triblock copolymer per mL acetone) series; TBB1E.

Mass excipient(mg)

Wt% excipient (%) Sample ID

0.000 0.00 TBB1E1

2.334 3.39 TBB1E2

2.912 4.44 TBB1E3

6.222 9.03 TBB1E4

7.581 10.79 TBB1E5

7.969 11.28 TBB1E6

10.888 14.80 TBB1E7

13.554 17.78 TBB1E8

13.213 17.41 TBB1E9

15.758 20.09 TBB1E10

16.521 20.86 TBB1E11

18.072 22.38 TBB1E12

22.126 26.09 TBB1E13

22.541 26.45 TBB1E14

25.864 29.21 TBB1E15

Table 9. TBB2 triblock copolymer (PLGA-PEO-PLGA; 3.75-3-3.75) excipient (PLGA 20k) loaded micelles (63.45 mg TBB2triblock copolymer per mL acetone) series; TBB2E.

Mass excipient(PLGA 20k) (mg)

Wt% excipient (%) Sample IDTBB2E

0.000 0.00 TBB2E1

0.641 1.00 TBB2E2

3.339 5.00 TBB2E3

7.050 10.00 TBB2E4

11.197 15.00 TBB2E5

21.150 25.00 TBB2E6

27.193 30.00 TBB2E7

0.300mL of the triblock copolymer solution was mixedwith 0.100mL of the single component solution, result-ing in 0.400 ml of copolymer/single component-solution,table 13. 0.400mL of the copolymer/single component so-lution was nanoprecipitated into 10.00mL of MilliQ waterand measured by DLS in time to monitor stability.

Page 8 of 12 Eur. Phys. J. E (2013) 36: 107

Table 10. TBC1 triblock copolymer PCL-PEO-PCL; 1.9-3-1.9) excipient (PCL 80k) loaded micelles (63.21 mg TBC1 tri-block copolymer per mL acetone) series; TBC1E.

Mass excipient (PCL80k) (mg)

Wt% excipient (%) Sample IDTBC1E

0.000 0.00 TBC1E10.703 1.10 TBC1E23.538 5.30 TBC1E37.023 10.00 TBC1E411.242 15.10 TBC1E515.803 20.00 TBC1E6

Table 11. Triblock copolymer solutions.

Copolymer ID Copolymermass (mg)

mL acetone Copolymersolution ID

TBB1 164.1 2.400 TBB1CS1TBB2 1894.47 31.575 TBB2CS1

Table 12. Single component solutions.

component ID Componentmass (mg)

mL acetone Componentsolution ID

PLGA 20k 6.8375 1.000 ES1Rapamycin 0.800 0.800 ES2β-carotene 8.75 1.000 ES3

3.2.4 Particle size analysis

Particle size analyses were performed using three differ-ent techniques. First a cryo-TEM study was performedshowing only the presence of perfectly spherical parti-cles. Secondly, a static multi angle light scattering analysiswas performed. The static light scattering experiment wasdone to validate; the more straight forward Dynamic LightScattering (DLS) measurement we performed. Both staticmulti angle and dynamic light scattering revealed similarsizes. Cryo-TEM and static multi angle light scatteringwere both performed on non-loaded, empty, and loadedmicelles. Cryo-TEM and static multi angle light scatter-ing results are not included. We have limited ourselves toreport DLS results which could be performed on all sam-ples.

The size of the micelles was determined by DynamicLight Scattering (DLS) (Zetasizer Nano ZS, Malvern In-struments Ltd., Malvern, UK) at 25 ◦C at a scatteringangle of 173◦. Ideally the number of photon counts ishigh enough to get a good signal-to-noise ratio and yetsmall enough to prevent multiple scattering effects. Thereported polydispersity index (PdI) is as given by theMalvern Zetasizer Nano ZS, as for the reported hydro-dynamic diameter (Dh) (z-averaged hydrodynamic diam-eter). Polydispersity for this light scattering analysis isused to describe the width of the particle size distribution,derived from the polydispersity index. The polydispersityindex is a parameter calculated from the Cumulants anal-ysis of the DLS measured intensity autocorrelation func-

tion. In the cumulants analysis, a single particle size isassumed and a single exponential fit is applied to the au-tocorrelation function. All samples were measured as pro-cessed, undiluted. Size distributions measured with DLSwere unimodal.

4 Results

4.1 DLS results on reproducibility on empty and activeingredient loaded micelles

To enable a nanoprecipitation reproducibility test, TBB2and TBC1, see table 5, triblock copolymers were madein three separate copolymer solutions in acetone and pre-cipitated in MilliQ to see what the reproducibility of theprocess is. Table 14 shows the results of the reproducibil-ity test of empty and active ingredient loaded micelles.We note that the PdI for all samples in this table arebelow 0.1. TBB2R series is to check the reproducibil-ity of making empty TBB2 copolymer micelles, TBC1Rshows the results on the reproducibility of TBC1 copoly-mer micelles and TBC1RE shows the reproducibility ofrapamycin loaded TBC1 copolymer micelles, see tables 6and 7 sample group ID.

The reproducibility of all three sample groups is excel-lent, showing low standard deviations (stdev) on hydro-dynamic diameter and PdI. The hydrodynamic diametersof the separate sample groups are within 1 nm (range).The PdI of the separate sample groups shows narrowmonomodal particle distributions. Another observation isthe smaller averaged hydrodynamic diameter size of theTBC1RE sample group compared to the TBC1R sam-ple group. Although the TBC1RE sample group is loadedwith rapamycin and the TBC1R sample group only con-sists of empty micelles. Still the TBC1RE group has asmaller hydrodynamic diameter size which can only beexplained by strong (hydrophobic) interactions betweenactive ingredient, rapamycin, and the hydrophobic endblocks in the core leading to a higher packing density andlower water content in the core resulting in slightly smallerparticles.

4.2 DLS results of single excipient, homopolymer,loaded micelles

Inspired by the SF-SCF results in fig. 8 of our recent pa-per [36], we investigated whether a hydrophobic polymericexcipient with a chemical composition similar to the hy-drophobic blocks of the copolymers used can be encap-sulated. In this way we determined the loading capac-ity/capability and the relation between excipient weightpercentage and (hydrodynamic diameter) size. It indeedappears to be possible to fill the micelles with inactiveingredients as follows from the increase of the size of themicelles and it turns out that the amount of excipient al-lows tuning the particle size of the resulting micelles. Wehave collected DLS results of the hydrodynamic diameterand PdI in figs. 4 and 5.

Eur. Phys. J. E (2013) 36: 107 Page 9 of 12

Table 13. Copolymer/single component solution; 0.400 mL precipitated in 10.00 mL Milli Q water.

Copolymer/single component solutionCopolymer solution Single component solution

Copolymersolution ID

mL copolymersolution

componentsolution ID

mL singlecomponentsolution

Micellesuspension ID.

TBB1CS1 0.300 ES1 0.100 TBB1EX

TBB1CS1 0.300 ES2 0.100 TBB1R

TBB2CS1 0.300 ES3 0.100 TBB2BC

Fig. 4. Hydrodynamic diameter as function of the loading wt%TBB1, TBB2 and TBC1 excipient loaded, PLGA 20k for TBB1and TBB2 and PCL 80k for TBC1, micelles, as measured byDLS.

Fig. 5. TBB1, TBB2 and TBC1 excipient loaded, PLGA 20kfor TBB1 and TBB2 and PCL 80k for TBC1, micelles, DLSmeasurement results for the PdI.

These results indicate that we can produce tailor-madenanoparticles for drug delivery, at a given size with a givenloading. See the appendix for a rationale for the lineardependence of d (hydrodynamic diameter) on the amountof excipient.

Table 14. Reproducibility results on separate performed nano-precipitation processes on empty and active ingredient loadedmicelles in terms of the (averaged) hydrodynamic diameter,

D(av)h and the standard deviation, σDh .

DLS results for the size (nm)Samplegroup ID

Sample ID Dh Davh σDh

TBB2RTBB2R1 31.1

31.3 0.3TBB2R2 31.6

TBB2R3 31.3

TBC1RTBC1R1 26.8

26.9 0.2TBC1R2 27.1

TBC1R3 26.9

TBC1RETBC1RE1 26.5

26.4 0.1TBC1RE2 26.4

TBC1RE1 26.4

For most medical applications micellar size is ofparamount importance for the therapeutic efficacy of thetreatment. Some anti-cancer therapies take advantage ofthe EPR-effect where size control between 50 and 80 nmis mandated. Using TBB1, see table 5, copolymer tri-blocks with a 5% weight loading of excipient will rendermicelles with a hydrodynamic diameter of approximately50 nm; if using TBB2, see table 5, copolymer triblockswith the same weight percentage of excipient loading willrender micelles with an approximate hydrodynamic diam-eter of 40 nm. If the drug loading/concentration is impor-tant the size can be tuned using higher or lower molecularweights of the triblock copolymers resulting in, respec-tively, smaller or bigger micelles with the same loading ofactive ingredient, mass of active ingredient per micelle.

Remarkable is the difference in slope comparing TBBtriblock copolymer excipient loaded micelles with TBC1,see table 5, triblock copolymer loaded micelles. In ap-pendix A there is a rationale about the linearity of theslope. From eq. (A.5), see appendix A, it follows thatthe slope is proportional to Γ/ccopol. Since in these ex-periments ccopol is fixed, a higher slope indicates thatthe corona density is higher for the PCL triblock copoly-mers. This actually agrees with our SCF computations,see figs. 6, 7 and 9 in [36].

Page 10 of 12 Eur. Phys. J. E (2013) 36: 107

4.3 Size stability of single component, homopolymerand active ingredient (rapamycin and β-carotene)loaded micelle formulations in time

Using hydrolytically degradable polymers (PLGA andPCL) will have an impact on micellar suspension stabilityin time due to hydrolytic degradation of the (hydropho-bic) blocks in the triblock copolymers in an aqueous en-vironment. In order to assess the real micellar stabilityit has been decided to focus on the stability before hy-drolytic degradation can have an effect on micellar stabil-ity. Arbitrarily we chose 15 days as the time after whichthe hydrolytic degradation of the block copolymers willhave the most prominent effect on the stability [49]. Thelack of change in size, hydrodynamic diameter, and PdIwithin 15 days after preparation will reveal the stabilityof micellar suspensions in time. To avoid continuous DLSmeasurements, the samples were subjected to a daily vi-sual inspection. In this way we could detect instabilitiessuch as agglomerates, change in appearance and/or color.If such a change was detected the sample was measuredby DLS. If no changes were observed the formulation wasmeasured after preparation at day 1 and after 15 days.First homopolymer excipient loaded micelles were testedon stability and subsequently active ingredient loaded mi-celles were tested on stability, all stability testing was atroom temperature.

4.3.1 Size stability of single component,homopolymer-loaded micelle formulations in time

TBB1, see table 5, triblock copolymer excipient loaded,PLGA 20k, micelles were tested on stability in time. Be-tween the first day and the following 14 days no visualchange of the micellar suspension was observed. From thereproducibility data in table 14 it is clear that the resultsof day 1 and day 15 are (table 15) very similar. This im-plies that the particles are stable for 15 days.

4.3.2 Size stability of single component, active ingredient(rapamycin and β-carotene), loaded micelle formulations intime

TBB1 (see table 5) triblock copolymer micelles wereloaded with rapamycin as an active ingredient to test thestability of active ingredient loaded micelles. Since theweight percentage of the active ingredient with respectto the triblock copolymer content is small (approximately0.5% (wt%) in ratio to the used triblock copolymer) thesize of the active ingredient loaded micelle does not changevery much compared to the empty micelles. However, itwas expected that the active ingredient would have someeffect on the size, therefore it was decided to measure thissample without any visual indications also on the secondday after processing to see if something happens with theinitial processed size.

Table 15. TBB1 homopolymer, PLGA 20k, loaded micellestability; TBB1EX-series, DLS results.

Sample ID Time (days) Hydrodynamicdiameter (nm)

PdI

TBB1EXDay 1 48.7 0.19

Day 15 49.1 0.10

Table 16. TBB1 active ingredient, rapamycin, loaded micellestability; TBB1R-series, DLS results

Sample ID Time (days) hydrodynamicdiameter (nm)

PdI

TBB1RDay 1 45.3 0.20

Day 2 43.8 0.20

Day 15 45.3 0.11

Fig. 6. TBB2 triblock copolymer active ingredient (β-carotene) loaded micelles stability, TBB2BC-series, DLS re-sults (the line in the hydrodynamic diameter results is to guidethe eye).

As can been seen in table 16 there was a slight decreasein size within the first two days. Subsequently, howeverthere were no visual indications implying any instability.At day 15 the sample was measured and the hydrodynamicdiameter turned out to be similar the measurement onday 1. The PdI however, seems to decline in time, whichis the same for homopolymer loaded micelles (table 15).The reason for the initial size change between day 1 andday 2 needs more investigation as the drop for the PdI.Overall, it seems that the particle size is fairly constantand the dispersion appears to have a long shelf-life.

TBB2, see table 9, triblock copolymers were loadedwith β-carotene as an active ingredient. For making mi-celles loaded with β-carotene it is known that they suf-fer from Ostwald ripening [31]. If these micellar suspen-sions can resist Ostwald ripening (constant size in time) wecan conclude that active ingredient transport from inner-micelle to bulk is limited. Figure 6 shows the results ofthe TBB2 triblock copolymer β-carotene loaded micellestability test. On day 8 the suspensions color changedfrom orange to yellow, probably due to oxidation of the

Eur. Phys. J. E (2013) 36: 107 Page 11 of 12

β-carotene. From the DLS measurements it appeared thatthe hydrodynamic diameter was increasing slightly whilethe PdI was still more or less stable at day 8. At day14 there was a turn over from the color from yellow towhite resulting in a stable hydrodynamic diameter but anincrease in PdI. In order to see what was further happen-ing we continued the measurement until visual aggregationof the suspension was observed on day 26 were after DLSmeasurements were no longer possible.

5 Conclusions

We have shown that well-defined micelles can be preparedcomposed of PCL-PEO-PCL and PLGA-PEO-PLGA tri-block copolymers using the nanoprecipitation approach.By adding hydrophobic compounds we can load the mi-celle in order to achieve a desired particle size and loading.Ostwald ripening was minimized with this approach. Sta-bilization of micelles by block copolymers prevents parti-cle aggregation, but the stabilizing polymer layer is openenough to allow solute mass transfer. In order to pre-vent/minimize solute transfer it is desired to tune the par-ticle core composition to prevent this mass transfer. Addi-tionally, the solubility of the encapsulated compound canbe decreased by antisolvent addition to the bulk result-ing in a significant slow down of Ostwald ripening. Theextremely low solubility of the used triblock copolymerslimits copolymer exchange between micelle and bulk againminimizing solute mass transfer and slowing down Ost-wald ripening. There is no need to use surfactant in thisprocess, conventional nanoprecipitation processes need anexcess of surfactant, mostly very water soluble with rel-ative high CMCs. Since we incorporated the surfactantfunction in the polymer backbone, no exchange of ad-sorbed and free surfactant is needed for stable suspensions.This also avoids washing the nanoparticle suspension toremove excess of free surfactant used in the process andlimits Ostwald ripening. We were able to synthesize dif-ferent kinds of triblock copolymers allowing simultaneoustuning of the size and loading. When performing the nano-precipitation process there is hardly an influence of tem-perature and triblock copolymer molar mass polydisper-sity. However, using these micelles in electrolytes, e.g. invivo, care must be taken to avoid destabilization of the mi-celles due to electrostatic interactions. Non-reported datashows that it is feasible to perform the nanoprecipitationprocess, using the mentioned triblock copolymers, in dif-ferent electrolytes at different pH’s and that the suspen-sion stays stable in time.

SF-SCF computational predictions that we recentlyperformed provide an accurate prediction of the size ofactive ingredient loaded and unloaded micelles. SF-SCFcomputations enable to predict equilibrium copolymer mi-celles. The hydrodynamic size that follows from these com-putations matches well with the measured particle sizesfrom dynamic light scattering. From the computations itfollows that the size of the nanoparticles is determined

by the number-averaged molar mass of the block copoly-mers; polydispersity hardly affects the size of the micelles.SF-SCF is an ideal tool to unravel the structure-functionrelationship between copolymer composition and micellarsize and morphology. Using theoretical SF-SCF predic-tions will lead to more efficient experimentation.

Appendix A.

The linear dependence of Dh on the amount of loadedcomponent(s) can be rationalized as follows. The mi-celles are stabilized by the copolymers with the PEOparts forming a steric stabilization layer. Consider Np

copolymer particles, each having a diameter d and vol-ume vp = (π/6)d3, in a total volume V . Such a dispersionhas a volume fraction φ of particles:

φ =Npvp

V. (A.1)

The total amount of surface in the volume V is

AT = Npπd2. (A.2)

From eqs. (A.1) + (A.2) it follows that

AT =6φV

d. (A.3)

Imagine all copolymers (acting as surfactants) are at theparticle-solvent interface. Then the total (initial) copoly-mer concentration equals

ccopol =Γ∞AT

V, (A.4)

where Γ∞ is the adsorbed amount of polymers (surfactant)at saturation. For example, for homopolymers this amountis ≈ 1mg/m2. Insertion of (A.3) into (A.4) yields

Dh =6Γ∞φ

ccopol. (A.5)

This means that, for instance, for Γ∞ = 1mg/m2, d =30nm and φ = 0.1, one expects an overall copolymer con-centration of 20 g/L is covering the surfaces. From (A.5)it follows that it is fair to assume that Dh increases lin-early with the wt% of loaded component(s) because it isproportional to the volume fraction of particles.

The authors wish to thank J. Put for supporting this study,M. Boerakker, H. Langermans, L. Bremer and B. Voogt forstimulating discussions, T. Kockelkoren, G. Draaisma and T.Handels for synthetic assistance during copolymer synthesisand N. Woike for technical assistance. A special thanks goesto R.K. Prud’homme for inspiring discussions.

Page 12 of 12 Eur. Phys. J. E (2013) 36: 107

References

1. R. Ravichandran, Int. J. Green Nanotech. 1:2, 72 (2010).2. C.C. Chen, G Wagner, Trans. IChemE, Part A 82, 1432

(2004).3. N.S. Sozer, L. Kokini, Trends Biotechnol. 27, 82 (2009).4. K.S. Soppimath, T.M. Aminabhavi, A.R. Kulkarni, W.E.

Rudzinski, J. Control. Release 70, 1 (2001).5. J. Panyam, V. Labhasetwar, Adv. Drug Deliver. Rev. 55,

329 (2003).6. A. Vila, H. Gill, O. McCallion, M-J. Alonso, J. Control.

Release 98, 231 (2004).7. M.G. Qaddoumi, H. Ueda, J. Yang, J. Davda, V. Labhaset-

war, V.H.L. Lee, Pharm. Res. 21, 641 (2004).8. H. Suh, B. Jeong, F. Liu, S.W. Kim, Pharm. Res. 15, 1495

(1998).9. Y. Matsumura, H. Maeda, Cancer Res. 6, 6387 (1986).

10. L.H. Reddy, J. Pharm. Pharmacol. 57, 1231 (2005).11. H. Maeda, J. Wu, T. Sawa, Y. Matsumura, K. Hori, J.

Control. Release 65, 271 (2000).12. H. Maeda, Y. Matsumura, Crit. Rev. Ther. Drug 6, 193

(1989).13. H. Maeda, J. Fang, Adv. Polym. Sci 193, 103 (2006).14. H. Maeda, J. Daruwalla, Eur. J. Pharm. Biopharm. 71,

409 (2009).15. D.E. Owens, N.A. Peppas, Int. J. Pharm. 307, 93 (2006).16. L. Mei, H. Sun, X. Jin, D. Zhu, R. Sun, M. Zhang, C. Song,

Pharm. Res. 24, 955 (2007).17. C. Song, V. Labhasetwar, X. Cui, T. Underwood, R.J.

Levy, J. Control. Release 54, 201 (1998).18. K. Letchford, H. Burt, Eur. J. Pharm. Biopharm. 65, 259

(2007).19. V.P. Torchilin, J. Control. Release 73, 137 (2001).20. R. Tong, J. Cheng, Pol. Rev. 47, 345 (2007).21. Y. Kakizawa, K. Kataoka, Adv. Drug Del. Rev. 54, 203

(2002).22. M.L. Adams, A. Lavasanifar, G.S. Kwon, J. Pharm Pharm.

Sci. 92, 1343 (2003).23. M.L. Forrest, A. Zhao, C-Y. Won, A.W. Malick, G.S.

Kwon, J. Control. Release 116, 139 (2006).24. Y. Zhao, H. Liang, S. Wang, C. Wu, J. Phys. Chem. B

105, 848 (2001).25. T. Nie, Y. Zhao, Z. Xie, C. Wu, Macromolecules 36, 8825

(2003).26. H.M. Burt, X. Zhang, P. Toleikis, L. Embree, W.L. Hunter,

Colloids Surf. B 16, 161 (1999).

27. Z. Song, R. Feng, M. Sun, C. Guo, Y. Gao, L. Li, G. Zhai,J. Colloid Interface Sci. 354, 116 (2011).

28. B. Jeong, Y.H. Bae, S.W. Kim, Colloids Surf., B 16, 185(1999).

29. R. Lund, L. Willner, J. Stellbrink, A. Radulescu, D.Richter, Phys. B 350, 909 (2004).

30. R. Lund, L. Willner, D. Richter, Macromolecules 39, 4566(2006).

31. Y. Liu, K. Kathan, W. Saad, R.K. Prud’homme, Phys.Rev. Lett. 98, 036102 (2007).

32. T. Niwa, H. Takeuchi, T. Hino, N. Kunou, Y. Kawashima,J. Control. Release 25, 89 (1993).

33. H. Fessi, F. Puisieux, J.Ph. Devissaguet, N. Ammoury, S.Benita, Int. J. Pharm. 55, R1 (1989).

34. H. Lannibois, A. Hasmy, R. Botet, O.A. Chariol, B. Ca-bane, J. Phys. II 7, 318 (1997).

35. D. Horn, J. Rieger, Angew. Chem. Int. Ed. 40, 4339 (2001).36. J.G.J.L. Lebouille, R. Tuinier, L.F.W. Vleugels, M.A.

Cohen Stuart, F.A.M. Leermakers, Soft Matter 9, 7515(2013).

37. G.J. Fleer, M.A. Cohen Stuart, J.M.H.M. Scheutjens, T.Cosgrove, B. Vincent, Polymers at interfaces, first edition(Chapman and Hall, London, 1993).

38. P.N. Hurter, J.M.H.M. Scheutjens, T.A. Hatton, Macro-molecules 26, 5592 (1993).

39. Y. Lauw, F.A.M. Leermakers, M.A. Cohen Stuart, J. Phys.Chem. B 110, 465 (2006).

40. E.B. Zhulina, O.V. Borisov, Macromolecules 45, 4429(2012).

41. K. Knop, R. Hoogenboom, D. Fisher, U.S. Shubert,Angew. Chem. Int. Edit. 49, 6288 (2010).

42. M. Kodama, S. Seki, Adv. Colloid Interface Sci. 35, 1(1991).

43. N. Nishikido, J. Colloid Interface Sci. 136, 401 (1990).44. L.R. Fukumoto, G. Mazza, J. Agric. Food Chem. 48, 3597

(2000).45. A. Sokmen, M. Gulluce, H.A. Akpulat, D. Daferera, B.

Tepe, M. Polissiou, M. Sokmen, F. Sahin, Food Control.15, 627 (2004).

46. G.W. Burton, K.U. Ingold, Science 224, 569 (1984).47. S. Chen, R. Pieper, D.C. Webster, J. Singh, Int. J. Pharm.

288, 207 (2005).48. G.M. Zentner, J. Control. Release 1, 203 (2001).49. R. Peters, J.G.J.L. Lebouille, B. Plum, P. Schoenmakers,

S. van der Wal, Polym. Degrad. Stabil. 96, 1589 (2011).