You are on page 1of 37

Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Contents lists available at ScienceDirect

Advanced Drug Delivery Reviews


journal homepage: www.elsevier.com/locate/addr

Microfluidic and lab-on-a-chip preparation routes for organic


nanoparticles and vesicular systems for nanomedicine applications☆
Lorenzo Capretto a,⁎,1, Dario Carugo b,c,1, Stefania Mazzitelli d, Claudio Nastruzzi d, Xunli Zhang b,e
a
Department of Pharmaceutics, UCL School of Pharmacy, 29-39 Brunswick Square, London WC1N 1AX, United Kingdom
b
Bioengineering Sciences Research Group, Faculty of Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, United Kingdom
c
Electro-Mechanical Engineering Group, Faculty of Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, United Kingdom
d
Department of Life Sciences and Biotechnology, University of Ferrara, Via F. Mortara 17/19, 44121 Ferrara, Italy
e
Institute for Life Sciences, University of Southampton, Southampton, SO17 1BJ, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: In recent years, advancements in the fields of microfluidic and lab-on-a-chip technologies have provided unique
Accepted 1 August 2013 opportunities for the implementation of nanomaterial production processes owing to the miniaturisation of the
Available online 8 August 2013 fluidic environment. It has been demonstrated that microfluidic reactors offer a range of advantages compared to
conventional batch reactors, including improved controllability and uniformity of nanomaterial characteristics. In
Keywords:
addition, the fast mixing achieved within microchannels, and the predictability of the laminar flow conditions,
Microfluidic
Lab-on-a-chip
can be leveraged to investigate the nanomaterial formation dynamics. In this article recent developments in
Nanomedicine the field of microfluidic production of nanomaterials for drug delivery applications are reviewed. The features
Nanomaterial that make microfluidic reactors a suitable technological platform are discussed in terms of controllability of
Polymeric micelles nanomaterials production. An overview of the various strategies developed for the production of organic
Liposomes nanoparticles and colloidal assemblies is presented, focusing on those nanomaterials that could have an impact
Polymersomes on nanomedicine field such as drug nanoparticles, polymeric micelles, liposomes, polymersomes, polyplexes and
hybrid nanoparticles. The effect of microfluidic environment on nanomaterials formation dynamics, as well as the
use of microdevices as tools for nanomaterial investigation is also discussed.
© 2013 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1497
2. Overview of synthetic approaches for organic nanoparticles and vesicular systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1498
2.1. Methods of preparation for vesicular systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1498
2.1.1. Mechanism of vesicle formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1498
2.1.2. Classical production techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1499
2.2. Methods of preparation for organic nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1499
2.2.1. Top-down approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1499
2.2.2. Bottom-up approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1499
2.2.3. Nanoprecipitation mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1499
3. Microfluidic synthesis of nanomaterials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1500
3.1. The microfluidic environment and mixing principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1500
3.1.1. Reynolds number and diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1500
3.1.2. Mixing in microfluidic devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1501
3.2. Features of microfluidic reactors for the production of nanomaterials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1501
3.2.1. Micromixing and reactor design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1502
3.2.2. Temperature control and heat transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1503
3.2.3. Kinetic control and investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1503
3.2.4. On-chip analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
3.2.5. High throughput production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504

☆ This review is part of the Advanced Drug Delivery Reviews theme issue on "Design production and characterization of drug delivery systems by Lab-on-a-Chip technology".
⁎ Corresponding author. Tel.: +44 207 753 5847.
E-mail address: l.capretto@soton.ac.uk (L. Capretto).
1
The authors have equally contributed to the work.

0169-409X/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.addr.2013.08.002
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1497

3.3. Materials used in the fabrication of microfluidic chip for the production of nanomaterials . . . . . . . . . . . . . . . . . . . . . . . . 1505
4. Microfluidic production of nanoparticle and vesicular systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.1. Organic nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.2. Nanomaterial from self-assembly of amphiphilic molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.2.1. Lipid vesicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1506
4.2.2. Niosomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1511
4.2.3. Polymersomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1511
4.2.4. Polymeric micelles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1515
4.2.5. Polymer nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1517
4.2.6. Hybrid nanoparticles and complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1517
4.2.7. Nanomaterial produced through complexation mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1518
5. Microfluidic-based manipulation, immobilisation and in situ analysis of nanoparticles and vesicular systems . . . . . . . . . . . . . . . . . . 1519
5.1. Nanoparticles and vesicular systems manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1520
5.2. Nanoparticles and vesicular systems immobilisation and in situ analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1521
6. Effect of microfluidic environment on dynamics of nanomaterial formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1521
6.1. Computational fluid dynamic (CFD) models for indirect investigation of the nanomaterial formation dynamics . . . . . . . . . . . . . . 1522
6.2. Methods for direct investigation of nanomaterial formation dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1522
6.3. Effect of shear stress on nanomaterial characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1523
7. Assessing the biological performance of nanomaterials produced by microfluidic technology: in-vitro and in-vivo tests . . . . . . . . . . . . . 1523
8. Conclusion and outlooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1527

1. Introduction thermodynamic solubility and dissolution rate. For small particles, the
NP apparent solubility (Sapp) is increased due to the high curvature of
Over the past two decades, research in the field of nanoparticles the particle at the surface according to Kelvin's equation [28]
(NPs), that are particles with dimension in the range of about a few
tens to a few hundreds of nanometres, has experienced a tremendous Sapp 2γV
¼ e RTr ð1Þ
growth. The importance of NPs has been demonstrated and NPs have So
been used in a wide range of applications including microelectronics,
chemistry, energy, life sciences and pharmaceutics [1–4]. Until recently, where S0 is the equilibrium solubility for the bulk material, γ is the
nanotechnology has experienced an intense research on inorganic NPs, interfacial free energy, V is the molar volume, R is the gas constant, T
which is reflected in the plethora of synthetic methods available for is the absolute temperature, and r is the particle's radius. An increase
such NPs and established applications in various fields such as medical in Sapp also results in increased dissolution rate dM=dt , according to
imaging, catalysis and electronics [5,6]. In comparison, only few studies the Noyes Whitney equation [29] (2)
have addressed the production of organic NPs. This is surprising,
considering the high potential of organic NPs in the fields of dM  
pharmaceutics and nanomedicine [7,8], food technology [9] as well ¼ kA Sapp −C ðt Þ ð2Þ
dt
as chemo- and bio-sensing [10,11]. In the next sections, we will
mainly focus on the preparation of nanosized dosage forms, which where A is the solid surface area, C(t) is the solute concentration at time
represent the most widely, explored area in the field of organic t, and k is the constant of the dissolution process. In addition, it should
nanotechnology. be noted that A increases with decreasing the particle dimension,
Many different types of nanosized dosage forms have been explored which synergistically contributes to the increased dissolution rate.
for their possible use in drug delivery and pharmaceutics. These mainly Furthermore, an additional increase of solubility can derive from an
include nanoemulsions [12–14], organic NPs [15,16], macromolecules increase in γ due to the production of nanoparticle constituted of
such as dendrimers [17], and supramolecular aggregates such as micelles thermodynamically unstable polymorphs or amorphous phase [3,30].
[18], liposomes [19] and polyplexes [20]. The reason for the use of In addition to enhancing the solubility of poorly soluble drugs, the
nanosized dosage forms in pharmaceutics stems from the beneficial effect possibility to realise a time controlled release [31], and a passive (through
that a reduction in size, down to the nanoscale, has on bioavailability and EPR effect [32]) or active targeted delivery [33], has also contributed to
targetability of a pharmaceutical active [21,22]. In this review emphasis is drive the research in the field of nanomedicine. Finally, in view of the
placed on those nanosized dosage forms that have stimulated interest in almost infinite range of chemical substances from which nanoparticles
the design of microfluidic reactors for their preparation. These include can be produced, their immense potential in different areas of future
both polymer and non-polymer based “hard” organic nanoparticles applications appears evident.
(i.e. kinetically stable after preparation) and “soft” vesicular systems Despite significant advancements in the field that have been
(i.e. in dynamic equilibrium after preparation) including liposomes, achieved so far, challenges still remain in synthesising and
niosomes, polymersomes, and micelles. It should be noted that formulating organic compound-based nanostructured materials,
“frozen” core micelles can be included within the category of “hard” while the synthesis of inorganic nanomaterials has been extensively
nanoparticles, given the associated slow rate of unimer exchange [23]. studied over decades with optimised control over both particle
Similar to their inorganic counterparts, organic NPs display size- shape and size being reported [34–36]. Consequently, special
dependent characteristics. However, the characteristics of organic NPs preparation techniques are required in order to disperse solid
result from both (i) different electronic properties of the molecules at organic materials into water, maintain the dispersion for a certain
the NP surface [24] and (ii) a largely increased surface-to-volume ratio period of time and functionalise organic NPs.
(SVR) [25,26], rather than depending on the quantum confinement Microfluidics is a new and emerging science and technology field
effect as in the case of metal NPs [27]. dealing with miniaturised systems that process, or manipulate, small
It is well known that the bioavailability of an active formulated as NP volumes of fluids (10− 9 to 10− 18 L), by means of channels with
suspension can be greatly increased by synergistically enhancing both dimensions ranging from tens to hundreds of micrometres [37].
1498 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Microfluidic and lab-on-a-chip technologies rely on the concept of on the relative composition of hydrophobic moieties (or tectons)
“shrinking” traditional bench chemical systems down to the size of and hydrophilic moieties [50]. Conversely, lipid molecules generally
a few centimetres squares. The main feature of such systems is the form lamellar bilayered structures and ― in dilute solutions ― they
miniaturisation of the fluidic environment, with a channel width of arrange into vesicles, referred to as liposomes [51]. Vesicles formed
about 100 μm (about the diameter of a human hair), where chemicals by self-assembly of polymers and polypeptides are instead referred
are brought together using a variety of pumping techniques for to as polymersomes [52] and peptosomes [53], respectively. In the
synthesis, separation or analysis purposes. present paragraph classical methods of vesicular systems production
Recent years have seen the implementation of microfluidic-based are described, together with essential information about their
and lab-on-chip technologies as tools in pharmaceutical research. For mechanisms of formation and related governing parameters.
instance, lab-on-a-chip devices have been used to support complex
chemical processes [38,39], to screen these processes [40–42], to study 2.1.1. Mechanism of vesicle formation
cell-formulation interactions [43] and the mechanisms of action of The process of vesicular systems formation can be categorised
drug devices combination products [44], and to produce microsized as a two-step self-assembly [50]. In this process, the amphiphile
particles or droplets with potential applications in drug delivery [45] forms a bilayer in the first step, which subsequently closes to form
and cell encapsulation [46–49]. a vesicle in the second step. The size of the hydrophobic moiety
Relevant to the formulation of organic nanomaterial, a series of relative to the hydrophilic moiety determines the shape of the
features make microfluidic and lab-on-a-chip devices appealing self-assembled amphiphilic structure, by influencing the curvature of
technological platforms not only to produce particles with a well- the hydrophobic–hydrophilic interface. The latter can be expressed in
controlled size distribution, but also to investigate the nanomaterial terms of mean curvature (H) and Gaussian curvature (KG), which are
formation process by integration with a number of measurement related to the surfactant packing parameter (V am =a  L) by the following
systems into the microreactor [35]. equation [54]:
However, despite their potential, at present few reports have
comprehensively discussed microfluidic techniques for the formation V am K  L3
of organic nanomaterials. Hence, this review aims to discuss the ¼1þHLþ G ð3Þ
aL 3
advantages of microfluidic and lab-on-a-chip technology for the
development and understanding of organic nanomaterial production, where Vam is the amphiphile hydrophobic volume, a is the interfacial
and its current or potential application in drug discovery and drug area, and L is the chain length normal to the interface. With this respect,
delivery. packing parameters of ~0.33, 0.5 and 1 correspond to the formation of
In this article, the most recent research and developments about spherical, cylindrical and bilayered structures, respectively [50]. Based
the production of organic nanomaterials for drug delivery applications on these purely geometrical considerations, it thus appears evident
using microfluidic technology are reviewed, with the aim of elucidating that it is possible to actively modulate the architecture of the obtained
the importance of developing novel strategies for achieving improved structures by controlling the governing parameters (i.e. hydrophilic–
nanoparticle formation. The fluid dynamic characteristics of the hydrophobic block ratio, in case of block copolymer structures).
microfluidic environment and the advantageous features of the However, a more comprehensive analysis of vesicular systems
microfluidic devices are discussed in the context of organic nanomaterial formation must take into account the relation between structures
formation, followed by a summary of the various strategies developed for architecture and free energy. With this respect, in the case of
the microfluidic and lab-on-a-chip production of nanoparticles and amphiphilic molecules, the free energy depends on both the
vesicular systems. Subsequently the use of microdevices in the interfacial energy of the hydrophobic–hydrophilic interface and
manipulation and analysis of nanomaterials are discussed providing the loss of entropy associated with the process of surfactants or
examples on how microfluidic devices could be employed as useful polymer chains aggregation into microdomains [50]. Notably,
tools in the investigation of nanomaterial formation processes. The when the interfacial energy is large and the entropy loss is relatively
biological performance of nanomaterials produced bymicrofluidic small, the thermodynamics of the process is dominated by the
technology is also reviewed. Further, a perspective for future research minimisation of the interfacial area. In this specific scenario,
is provided highlighting the issues that still hinder the widespread hydrophobic and hydrophilic phases are highly segregated and this
success of microfluidic devices for nanomaterial production. state is usually referred to as superstrong segregation limit (SSSL)
[55]. Polymers (or tectons) characterised by low conformational
2. Overview of synthetic approaches for organic nanoparticles and entropy are generally stiff molecules, having low internal degrees
vesicular systems of freedom. In this case, amphiphiles will form structures which
minimise the interfacial area per unit volume, corresponding to
One of the major challenges within nanotechnology field has planar interfaces.
been the development of synthetic methods for nanosized dosage The formation of vesicular systems relies on the principle that
forms production. This interest is mostly driven by the need for tectons energetically prefer a parallel molecular spatial arrangement
developing methodologies capable of producing nanomaterials [50]. Based on this consideration, sheet-like micelles should represent
with tuneable characteristics (i.e. size, physico-chemical properties the optimal architecture. However, at low tecton concentration levels
and concentration) and with high homogeneity, in order to possibly when the sheet-like structures are too large, the energy loss due to
reduce post-production steps while obtaining high standard products. surface tension effects forces the structure to close forming a vesicular
The following paragraphs provide an overview of the synthetic system [51]. This process gives rise to a line energy depending on the
approaches developed for either organic NPs or vesicular-like line tension (γL); and the bending of the bilayer during closure requires
systems in rapid dynamic equilibrium, together with their possible an additional energy depending on the bilayer bending modulus (κ).
mechanisms of formation. The balance between line tension and bending energy determines the
minimal vesicular size [56]. Notably, vesicle formation is generally
2.1. Methods of preparation for vesicular systems favoured when the bending elasticity of the bilayer is low, and the
surface tension is high.
Surfactants and amphiphilic block copolymers at low concentration However, commonly employed lipids and polymeric structures are
levels usually self-assemble forming micelles, with physical and characterised by a low molecular solubility and thus the architectural
architectural properties depending on the relative hydrophobicity and properties of the formed vesicles do not depend predominately on
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1499

exchange and equilibrium processes, but they are mainly governed Lithographic methods such as “Particle replication In Non-wetting
by the preparation conditions. For this reason, these vesicular Templates” (PRINT) [64,65] and “Step and Flash Imprint Lithography”
systems are usually referred to as ‘non-equilibrium vesicles’ [50]. (S-FIL) [66], have also been conveniently employed for the production
Thus, in this case, the vesicle size depends less significantly on of organic NPs with effective control over NP size and shape. PRINT
thermodynamic factors but mostly on non-equilibrium aspects of utilises a non-wetting perfluoropolyether (PFPE) as a mould to create
the formation process. Notably, the shape of the obtained vesicle is monodisperse NPs with various shapes in mild conditions that
the one associated with the minimisation of the bending energy for allow for the use of a range of different active substances including
a given difference in the number of amphiphiles in the inner and those more subjected to degradation. Similarly, S-FIL relies on a
outer monolayers, which depend on the production technique [51]. nanoimprinting process to produce NPs but differs from PRINT for the
Consequently, vesicle diameter and shape can be controlled by method of release of NPs after preparation, which is realised by the
relying on different production methods [57]. solubilisation of a sacrificial layer.

2.1.2. Classical production techniques 2.2.2. Bottom-up approaches


In conventional vesicle preparation routes, the contact between Bottom-up approaches involve the conversion of molecular matter
the aqueous phase and amphiphiles can be achieved by (i) phase dispersed in liquid or gas phase into NPs by a condensation process.
inversion and (ii) organic solvent-free methods [57]. In phase They are far more utilised in research setting since they require less
inversion approaches, the amphiphile is initially dissolved in an expensive equipment and smaller amount of reagents, and often grant
organic solvent. Subsequently, the solvent is exchanged by water, better control over NP characteristics in terms of composition and
via evaporation or dialysis [58,59]. These methods rely on the size. The bottom-up approaches can be further classified in emulsion-
variation of the packing parameters and, thus, the achievement of based and nanoprecipitation methods. In the emulsion-based methods
an optimal surface area per molecule, as mentioned previously. the emulsion generally works as intermediate step to confine the
However, while effective control over vesicle size and architecture organic phase in small droplets, which act as a template for the organic
can be achieved with these techniques, the formed membrane NPs. In a typical setting, the organic compound is firstly dissolved in an
structure contains residual organic solvent, which can potentially organic solvent. This solution is then emulsified with controlling the
compromise the functioning of bioactive molecules encapsulated droplet size by homogenisation method. The NP production is achieved
within the bilayer for biomedical purposes. by separating the solvent via evaporation [67] or diffusion process
On the other hand, in solvent-free approaches, the amphiphile is [68,69].
generally deposited onto a substrate and subsequently hydrated Instead, nanoprecipitation procedures rely on the creation of a
(referred to as hydration method). This is usually performed in the supersaturated solution, by means of temperature change [70] or
presence of an external energy source (i.e. mechanical or electrical) solvent shifting [71], which ultimately leads to the formation of a
[60]. The formation of vesicles from polymeric or lipid films upon precipitate [72]. The mechanism and theory at the basis of the
application of alternating electric field is generally referred to as nanoprecipitation approaches will be discussed in the following section.
electroformation, and it has been observed to result in the formation Based on this theory, it is clear that crucial for the preparation of organic
of micrometre-sized vesicles. NP is the creation of a rapid and spatially uniform supersaturation. In
However, batch methods of production are generally associated order to improve mixing performance, various modifications to the
with high vesicle size polydispersity, requiring post-processing steps basic methodology of nanoprecipitation have been developed. NPs
(i.e. extrusion or sonication) to improve size uniformity. Further details have been produced using bath reaction vessel where the solutions
about these methods and their translation into microscale systems will were mixed by means of magnetic stirring [73] or ultrasound [74].
be provided in the next sections. Moving from conventional bulk systems, membrane systems have
also been used [75] while Johnson and Prud'homme [9–11] have
2.2. Methods of preparation for organic nanoparticles developed mesoscale continuous-flow confined impinging jets (CIJ)
and multiple inlets vortex mixers (MIVM). With the latter they have
Methods of preparation of organic NPs are typically categorised in two further demonstrated that a fast mixing is crucial in order to achieve
groups: (i) top-down and (ii) bottom-up approaches. In the following a kinetically controlled nanoprecipitation, and consequently a fine
paragraphs the methodologies developed for the two approaches are control over the size and polydispersity of NPs [12,13].
briefly discussed. Subsequently, the mechanism of nanoprecipitation
exploited in the bottom-up approaches is described more in detail. 2.2.3. Nanoprecipitation mechanism
The framework of classical crystallization theory provides a useful
2.2.1. Top-down approaches model to describe the formation of NPs via bottom-up approaches,
Top-down approaches rely on reduction of size of macroscopic and for understanding the role that this process exerts in controlling
material down to nanometre scale. Various procedures have been the characteristics of the produced NPs [3,76]. Starting from a
developed with mechanical milling being the most obvious. However, multicomponent, single phase system, the precipitation (or phase
although milling processes have been widely used in organic NP separation) occurs at the onset of supersaturation when the system is
formation [5], they are generally considered unsuitable for pharma- perturbed by modifying a boundary condition such as temperature or
ceutical actives. This applies particularly to the production of NPs with concentration, such that the phase separation becomes favourable to
narrow distribution and small size, because mechanical energy in the reduce system free energy.
form of shearing and impact forces which is applied during particle Fig. 1 shows a simplified representation of the nanoprecipitation
milling, make it difficult to reduce the size of the particles without process. After a system perturbation, the solute concentration
simultaneously inducing particle agglomeration [3]. In addition, increases above the saturation limit (i.e. solubility) and reaches a
abrasion during the milling process can cause contamination of the critical concentration at which the precipitation process is triggered.
end product which is difficult to separate, especially in active compound NP formation model involves an initial nucleation phase in which
formulations, while it may be difficult to produce nanosized organic seed particles, referred to as nuclei, precipitate spontaneously and
particles consistently when using solid-particle milling [61]. A similar a subsequent growth phase in which the initial seeds capture the
approach is represented by the “high-pressure homogenization” remaining dissolved solute.
technique [62,63], where cavitation forces are exploited for the The nucleation phase occurs when the concentration of the solute
reduction of particle size. reaches a critical nucleation concentration at which a shower of nuclei
1500 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

processes further complicate screening and optimisation of the


synthesis conditions and therefore the scale-up to batch-scale stirred
flask synthesis for larger production [4].

3. Microfluidic synthesis of nanomaterials

Size, shape and crystal-structure characteristics strongly affect


NPs' physical and chemical properties. Therefore, it is not surprising
that in the production of NPs a superior control over the process
parameters, in order to produce material of required features, is
desirable [35,36,81,82]. In this respect, microfluidic reactors offer a
series of potential advantages for the production of NPs since they enable
accurate control and manipulation of fluids and fluid interfaces. In
microfluidic devices, reactions are carried out within small reaction
channels, with diameter between a few tens and a few hundreds of
microns. The small dimension, and the resulting large SVR, enables
rapid and uniform heat and mass transfer that can dramatically improve
NP yield and size distribution while reducing undesirable by-product
Fig. 1. Schematic representation of solute concentration trend during the nanoprecipitation formation. In addition, the possibility of solvent recycling, and integrated
process (adapted with permission from [3]. Copyright 2001; John Wiley and Sons). separation techniques, are likely to provide cost effective and
environment friendly technology for the production of NPs [34].
is formed. At this stage, the nuclei form by condensation of molecules
until they reach a critical size at which they are stable against 3.1. The microfluidic environment and mixing principles
dissolution. The nucleation phase proceeds until the concentration of
the solute falls below the critical nucleation threshold, where new In this section, the fundamental theory of fluid and mass transfer,
nuclei cannot be any longer formed [25]. However, this concentration and the implication of using microfluidic devices for mixing are firstly
is still sufficient to allow the growth of the nuclei already formed. The discussed. In principle, fluid flow in microfluidic environments is
growth phase proceeds until the concentration of the still-dissolved governed by the same laws that apply to the flow at the macroscale.
material has fallen to the equilibrium concentration (i.e. bulk solubility). However, the miniaturisation confers additional characteristics that
A precise control over the nucleation phase is often difficult to achieve can be exploited in order to perform processes not otherwise possible
therefore, heterogeneous nanoprecipitation, by means of preformed at the macroscale. Notably, microfluidic devices are not simply a
seed crystals [77] or additives [78] can be used. miniaturised version of their macroscale counterparts. This is because
To obtain NP batches with narrow size distribution it is important to many physical characteristics, including SVR and diffusion-based mass
tune the process in order to ensure that the nucleation occurs in a short transfer, do not scale linearly from macro- to micro-domains. Another
period compared to the time required for the growth phase. In the peculiarity of microfluidic systems is represented by the omnipresence
nucleation phase, nucleation and growth occur concurrently therefore, of laminar flow, due to the predominant role exerted by viscous forces.
the relative kinetics of the two processes greatly affects the quality of These factors thus become significant at the microscale, and their effects
the produced NPs. In this regard, it should be noted that the nucleation should be considered while designing and developing lab-on-a-chip
process is far more dependent on the supersaturation level than on the devices.
growth process; hence, high supersaturation is beneficial to favour the In other words, a microfluidic mixer is not simply a miniaturised
nucleation over the growth [25]. version of a macroscale mixing device; it should be designed in such a
In addition, since nucleation and growth rates depend on way to leverage the physical characteristics of mass and fluid transfer
the supersaturation level, in order to obtain a monodisperse batch, in a micro-confined domain.
a homogenous environment is required in terms of temperature
and species concentration during both nucleation and growth phases 3.1.1. Reynolds number and diffusion
(i.e. temperature variations affect the solubility of the species and thus Fluid flow is generally categorised into two regimes: laminar and
the supersaturation level) [79]. Batch synthesis of NPs is typically turbulent. In laminar regime the fluid flows in parallel layers with no
carried out in stirred flasks, where stirring is conventionally used to cross currents perpendicular to the main flow direction, whereas
rapidly mix reactants and keep growing particles in suspension. turbulent flow is generally characterised by the formation of vortices
Generally, during the synthesis process, the controlled addition of and flow fluctuations in space and time. From a physical perspective,
secondary reactants is required. In such cases, the addition rate and the two regimes differ in terms of the relative importance of viscous
mixing speed often determine the presence or absence of secondary forces (i.e. the friction imparted by the channel walls) and inertial forces
particle nucleation, homogeneity and state of aggregation of the final (i.e. fluid momentum) which is measured by the Reynolds number (Re)
mixture. As mentioned earlier, the homogeneity of concentration and [83]:
temperature is also crucial for monodisperse particle size distributions
to be obtained, due to the sensitivity of colloidal nucleation and ρuDh uDh
Re ¼ ¼ ð4Þ
morphology to the local temperature and composition [3]. However, μ v
macroscale synthesis often grants poor control over the spatial and
temporal distribution of the governing physical and chemical where ρ and μ are the fluid density and dynamic viscosity, respectively;
parameters (i.e. temperature and species' concentration) [80], as ν is the fluid kinematic viscosity; u is the mean fluid velocity and Dh is
well as over species additions and mixing rates. These two factors the hydraulic diameter of the channel. The latter is a characteristic
lead to non-homogenous rate of nucleation and growth, resulting number which depends on the cross-sectional geometry of the channel,
in polydisperse NP output. and is given by [83]
Furthermore, batch synthesis tends to suffer from poor repro-
ducibility of size and quality of the NPs from batch to batch. In addition, 4A
Dh ¼ ð5Þ
the relatively large amount of material used and the time-consuming P wet
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1501

where A and Pwet are channel cross-sectional area and wetted perimeter, of the small length scale, which dramatically increases the effect of
respectively. diffusion and advection, realising efficient and fast mixing [91].
At low Re, viscous effects overcome inertial effects and a fully Micromixer architecture is generally designed in such a way
laminar flow occurs. In this situation, fluid streams flow parallel to to decrease the mixing path and increase the contact surface area.
each other and the velocity at any location within the fluid stream Micromixers are frequently classified as being passive or active, depending
is time invariant when boundary conditions are stationary. Therefore on the mechanism exploited to generate mixing at the microscale.
mass transfer by advection can occur only in the direction of the fluid Active micromixers use external energy sources, as well as fluid
flow, and mixing between streams is dominated by molecular pumping energy, in order to generate time-dependent perturbations
diffusion [84]. Conversely, at the high Re associated with turbulent of the flow field and accelerate the mixing process [92]. Based on
regime the flow is dominated by inertial effects and exhibits motion the type of external force employed, micromixers can be categorised
that is random in both space and time. In this situation there are as pressure field-driven [93], acoustic (ultrasonic)-driven [94],
advective mass transport phenomena in all spatial directions [85]. temperature-induced [95] or magneto-hydrodynamic [96]. Active
A transitional flow regime usually occurs between laminar micromixers have generally higher mixing efficiency compared to
and turbulent regimes. The exact transitional Reynolds number is a passive micromixers [97]. However, the need for system integration
function of many parameters, including channel geometry and with peripheral devices (i.e. actuators for external power source)
surface roughness. The transitional Re is generally in the range and the laborious and expensive fabrication processes, have limited
1500–2500 for the large majority of fluid dynamic problems [86]. the implementation of such devices in practical applications. In
In microfluidic systems, Re is typically lower than 100 and the flow addition, the use of exogenous energy sources (i.e. in the form of
is thus regarded as laminar. This characteristic of fluid flow has a ultrasonic waves) may result in the generation of high temperature
direct effect on mixing within microfluidic devices. gradients which can potentially damage the bio-actives. For this reason,
Notably, in a fluidic domain where the flow is laminar, mass transfer active mixers are not a popular choice when applying microfluidics to
is dominated by passive molecular diffusion and advection. Diffusion is pharmaceutical, chemical and biological applications [97].
defined as the mechanism of molecule transport from a domain of Passive mixers instead rely entirely on fluid pumping energy, and
higher concentration to a domain of lower concentration by Brownian make use of specific channel architectures to restructure the fluid flow
motion, resulting in a gradual mixing of material. Diffusion-based in such a way to reduce the diffusion length and maximize the contact
mass transfer is described mathematically using Fick's laws [87] surface area. They were the first microfluidic devices reported due
to the associated cost-effectiveness and convenient fabrication
dφ routes compared to active micromixers. Furthermore, they can be
j ¼ −D ð6Þ
dx easily integrated with more complex lab-on-a-chip devices. The
reduction of mixing time can be achieved by different means, such
where φ is the species concentration, x is a spatial coordinate, and D is as splitting the fluid stream by means of serial or parallel lamination
the diffusion coefficient. For spherical particles, D can be derived from [98,99], hydrodynamically focusing fluid streams [100], introducing
the Stokes–Einstein equation [88]: bubbles of gas (slug) or liquid (droplet) into the flow [101,102], or
enhancing chaotic advection using ribs and groves designed on the
kT channel walls [103,104]. An extensive overview about the different
D¼ ð7Þ
6πμr types and designs of micromixers has been provided elsewhere
[105].
where k is Boltzmann's constant; T is the absolute temperature; r is
the radius of the particles (or molecules) and μ is the dynamic viscosity
of the fluid medium. The diffusion coefficient is usually expressed 3.2. Features of microfluidic reactors for the production of nanomaterials
in m2s−1 (or cm2s−1), and for a small molecule in water at room
temperature it assumes the typical value of 10−9 m2s−1 [89]. A series of features make microfluidic reactors particularly appealing
Diffusion is a non-linear process in which the time t required for a in the production and investigation of nanomaterials as compared with
species to diffuse scales quadratically with the distance x covered. conventional macroscale reactors. Most of these features derive from
Thus a simple, one-dimensional diffusion process can be modelled the unique characteristics of the flow in microchannels and in general
using the following equation [90]: of the microfluidic environment, such as the omnipresence of the
highly predictable laminar flow and a large SVR. The implications and
2
x ¼ 2D  t ð8Þ advantages of these characteristics on nanomaterial production are
described in the following paragraphs. Briefly, these characteristics
where t is the average diffusion time over the distance x. In a include:
microfluidic channel, x usually represents the stream width of the
fluid to be mixed along the microfluidic channel. On a microscale, • Efficient and controllable mixing under continuous flow conditions
the diffusion distance can be extremely small, especially if fluid resulting in a homogeneous reaction environment.
streams are hydrodynamically focused. From the moment that t • Improved and efficient temperature control and heat transfer.
scales with the square power of x (see Eq. (8)), a decrease in • In situ monitoring of the progress of NP formation through residence
diffusion distance has the effect of dramatically reducing the time time based resolution.
required for complete mixing to be achieved. Therefore, diffusion • Temporal control of reactions by adding reagent at precise time
becomes a viable method to transport particles and mix chemical intervals during the reaction process.
species in microfluidic systems. • Control over characteristics of NP by controlling the kinetics of the
process.
3.1.2. Mixing in microfluidic devices • High-throughput screening of various formulations by on-line
At the macroscale, mixing is generally achieved by generation of variation of the process parameters.
turbulent flow, making it possible to segregate fluid in small domains, • Opportunity to integrate post-synthesis processes and measurement
thereby leading to an increase of the contact surface and a decrease of systems on a single technology platform.
the mixing path. As discussed in the previous paragraph, due to the • Possible scale-up of the process by increasing the number of the
low Re, mixing in microfluidic devices is achieved by taking advantage microreactors.
1502 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

The aforementioned set of characteristics reveals the significant mixing performance, have attracted attention on the use of micromixers
potential of microfluidics to transform current classical batch technology for nanoparticle production. Different types of micromixers have been
for the production of NPs into continuous microfluidic process. Different conveniently applied.
research groups have worked towards this aim and achieved promising A basic “Y-shaped” channel configuration was used to produce
advancements, but clearly, this science is still in its infancy and nanoparticle of hydrocortisone, a poorly water soluble drug [117].
much remains to be done in order to demonstrate the superiority In this study, the authors demonstrated the production of nano-
of microfluidic process over the conventional batch process. sized dispersions, and the possibility to control the mean particle size in
the range of 80–450 nm by modifying the experimental parameters and
3.2.1. Micromixing and reactor design design of microreactors. Using a similar setup, in a subsequent research
The traditional batch methods lack precise control over the [117], the same group employed artificial neural networks to identify
mixing and supersaturation level, leading to uncontrolled nucleation relationships between variables affecting drug nanoprecipitation,
and growth processes, resulting in poor control over final particle demonstrating the critical role of antisolvent flow rate. A “T” type
characteristics [106]. A rapid heat/mass transfer can greatly improve micromixer has been also employed to prepare nanocrystals of
the controllability of these characteristics that in turn determine barium sulphate and boehmite [110]. The authors, leveraging the
the physico-chemical properties of the produced nanomaterials. impingement of the two streams, enhanced the mixing of the reactants,
In this respect, microscale reactors can provide homogeneous, fast and exerted a control over nanoprecipitation output modifying the
(in the order of fraction to hundreds of milliseconds) and reproducible velocity of the impinging streams. The simple “T” and “Y” configurations
mixing conditions, thus potentially representing a method to obtain have also been integrated with on-line measurement systems to
nanomaterials with excellent dimensional characteristics. investigate the fundamental reaction processes [118], and to provide
Microscale mixers/reactors are characterised by a larger SVR, information for an on-line algorithm-based control of the process
offering the possibility to enhance mass and heat transfer as compared parameters, which can drive the system towards a desired goal
to conventional bulk mixing systems [91,107]. This leads to a massively [119].
reduced mixing time that becomes comparable with the nucleation and Wilde et al. [120] introduced a radial interdigitate microreactor
growth kinetics, opening the possibility to exert a control on them [4]. based on the principle of parallel lamination for the production of
With respect to the mixing approach and device characteristics, thiol functionalised gold NPs. The design was constituted of a 16 inputs
microreactors for nanomaterial production can be grouped in two radial interdigitate mixer, which represented a modification of the
categories: continuous flow and segmented flow microreactors. classical straight design for parallel lamination micromixer. The size
standard deviations of the NPs produced with this micromixer ranged
3.2.1.1. Continuous flow microreactors. Continuous flow microreactors, between 0.6 and 0.9 nm, in comparison with a range between 1.3 and
compared to segmented flow reactors, are generally characterised by 2.1 nm for the bulk synthesis. Results also indicated a control over the
higher productivity and the possibility to continuously vary the reactant final mean size diameter by controlling the flow rate and, in turn, the
composition along the reaction channel [34]. In this way, it is possible to residence time and the growth phase. Notably, the device was able to
realise multi-step processes by connecting in series different reactors mix the reactants with high volumetric throughput up to 4500 μL/min.
[108,109]. Additionally, due to the simplicity of the flow pattern, process Parallel lamination micromixers were also used for the production of
scaling-up can be achieved by simply increasing the flow rate [110]. CdS NPs [121,122]. Based on the measurement of the adsorption
Continuous flow microreactors can be further categorised in three spectra, it was observed that the NPs produced in the microfluidic
main sub-groups depending on the architecture of the microchannel device were more homogeneous in terms of size, than those produced
network. These comprise capillary tube, coaxial flow and micromixer. with conventional bulk methods. Moreover, a tendency toward the
Capillary tube microreactors represent the simplest device NP uniformity with the increase of the flow rate ratio was reported.
configuration. They are constituted of silica [111], steel [112] or Increasing the flow rate reduced the likelihood of nanoparticle
polymer [113] capillary tubes with channel lumen dimension in coalescence since the NPs were extracted from the reaction volume
the order of microns. They have been generally applied for the more quickly.
production of metal nanomaterials, where an improved NP output The use of a split and recombine (SAR) micromixer has been
was realised through a fast and accurate temperature control. Their ease presented for the synthesis of metal NPs [123–125], demonstrating
of manufacture and operation, together with the possibility to use that the mixing conditions depend strongly on flow rate and local
robust materials able to withstand the high temperature requirements, geometry. However, due to the complicated 3D architecture, the
has attracted the attention on capillary device for the production of fabrication of this type of mixers might be non-trivial.
metal and semiconductor nanocrystals. However, capillary devices In contrast, a simple planar flow focusing architecture does not
suffer from the risk of lumen blockage, adhesion of chemicals to the require long fabrication process, and provides the possibility to control
channel surface, and relatively high product polydispersity [114]. the dimension of the produced nanomaterials by simply varying the
In order to address these issues, coaxial flow microreactors have volumetric flow rates of the three inlet streams. The control over
been proposed [109,114–116]. The ensheathing flow prevents direct the dimensional characteristics is particularly important in the case of
contact of the reaction mixture with the channel walls thus minimising nanomaterials for drug delivery application, since the dimension
clogging and adhesion to channel walls. In addition, the ensheathing strongly influences the spatial and temporal release of the actives.
provides a way to minimise the polydispersity of the formed NP by Therefore, planar flow focusing architectures have been extensively
constraining the flow containing the precipitating species at the centre applied for the production of different drug delivery carriers such as
of the channel, where a more homogeneous fluid velocity field subsists. liposomes [126], solid lipid NPs (SLN) [127] and polymeric micelles
This in turn results in more homogeneous residence time distribution [128]. The various microfluidic reactors employed for the production
(RTD) of the growing NPs within the microreactor; hence, different of nanomaterial for drug delivery application are discussed more in
growing NPs experience a similar growth process duration, with detail in paragraph 4.
beneficial effect on nanoparticle polydispersity. The implication of RTD When high molecular weight hydrophobic species (i.e., N45 kDa),
on microreactor-based nanoparticle synthesis is discussed later in this such as poly-lactic-glycolic acid polymer (PLGA) are nanoprecipitated
section (see paragraph 3.2.1.2). using planar focusing devices fabricated in PDMS, they tend to
A broad RTD also characterises laminar flow micromixers. However, aggregate on the channel inner surface, potentially resulting in
the possibility to integrate on-chip characterisation systems and realise clogging of the device. The aggregation is caused by the adsorption
tailored channel architectures, together with the potential for tuning of the hydrophobic species on the hydrophobic walls of the device,
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1503

resulting in substantially reduced robustness of operation and, the wall, but are rather surrounded by the carrier fluid. An example of
eventually, irreversible device failure. A 3D focusing, in both the droplet-based microreactor was presented for the synthesis of CdSe
horizontal and the vertical dimensions, has been proposed to nanocrystals in droplets flowing in a perfluorinated carrier fluid. The
overcome this issue [129]. The 3D focusing device did not require droplets were formed by utilising a flow-focusing nanojet structure
complicated fabrication process since it was composed of a with a steep increase in channel height [131].
monolithic single layer characterised by three sequential inlets for The possibility to create stable droplet-flow not only provided a
vertical focusing, followed by a conventional cross-junction. However, way to control the residence time but also provided the opportunity
since the vertical focusing was achieved through two additional for running multiphase processes. Wang et al. [134] developed a
streams of solvent, it caused a dilution of the initial polymeric stream method for the formation of TiO2 NPs using a microreactor that created
resulting in a decrease of the attained supersaturation level. a stable interface between two insoluble liquids. The two insoluble
A complete reactor system has been fabricated by integrating liquid systems used in this study were: 1-hexanol/formamide and
different units with the mixing device [130]. It integrated pumps, cyclohexane/water. The interface between the two phases represented
mixers, valves, micro-heaters and temperature controllers. The authors the site of the hydrolysis reaction between titanium tetraisopropoxide
reported a significant reduction of the reaction time for the formation of (TTIP) and water that led to the production of the titania NPs. Droplet-
hexagonal gold nanoparticles that passed from 30 min to 5 min for based microreactors can also be used to implement multi-step synthesis
traditional bulk and microfluidic syntheses, respectively. [102] by adding additional reagent to droplets through a droplets fusion
process. However, the velocity of the flows must be carefully adjusted to
3.2.1.2. Segmented flow microreactors. Alongside the laminar flow permit the fusion of the droplets limiting the process versatility [132].
microfluidic reactors, various authors presented alternative approaches
to the chemical synthesis of NPs based on segmented flow microreactors.
3.2.2. Temperature control and heat transfer
These microreactors can be further categorised as slug flow (gas–liquid)
Temperature plays a critical role in controlling solubility, super-
or multiphase flow (liquid–liquid) microfluidic reactors [131,132].
saturation level and process kinetics during nanomaterial formation;
In chemical synthesis of NPs, an important parameter which strongly
therefore, efficient temperature control can improve nanomaterial
affects the monodispersity of the produced samples is represented by the
synthesis. Typically, microfluidic devices contain channels with
residence time distribution, which reflects the mean time that each
characteristic dimension in the range 10–400 μm. This small dimension
particle spends inside the reactor. In laminar flow microfluidic reactors
results in increased SVR (in the range 10,000–50,000m2·m−3), compared
the parabolic flow profile (fluid moving slower near the channel wall
with 100–2000 m2·m−3 of their macroscale counterparts [135].
than in the centre) and the associated axial dispersion, cause a variation
Typical microfluidic devices exhibit high thermal transfer
in residence time, which in turn leads to a wider size distribution of
efficiencies, and this allows performing exothermic and/or high
the produced NPs [133]. This problem with laminar flow reactors can
temperature reactions in an efficient and controllable (isothermal)
be overcome with using segmented flow microreactors, which lead to
manner [136,137]. Microfluidic environments have been shown to
a better control of nanoparticle size distribution. This is because the
provide an efficient temperature control in continuous flow reactors
droplets (liquid–liquid) or slugs (gas–liquid) can work as a microsized
chemical synthesis [135,138].
reactor that flows along the channel with a time that is determined
The large SVR also offers the possibility to accelerate heat exchange
only by the flow rate. Mixing in these microdevices is achieved leveraging
for nanoparticle synthesis. In this respect, it has been shown that only
the microstreaming created within the slug or droplet while it is flowing
0.4 s were needed for a 200 μm diameter channel to heat a liquid from
along the straight and winding channels.
20 to 300 °C [139]. Nakamura et al. reported the use of capillary tubing
It has been experimentally demonstrated that segmentation is
in hot oil bath for the production of quantum dots. Efficient and
beneficial in controlling the size distribution of the synthesised NPs.
fine temperature control of the microreactor environment was
Kahn et al. [133] studied the effect of the reactor design on the synthesis
beneficial for controlling particle diameter and achieving reproducible
of silica NPs. Two reactor designs were compared: (i) single phase
preparation of CdSe nanocrystals [139,140]. By varying the temperature
laminar flow reactor (LFR) and (ii) segmented (gas–liquid) flow reactor
and the flow rate ratio, NPs with different size were produced. In
(SFR). The size of the particles produced with LFR was found to be
addition, the author reported a high reproducibility of particle size
correlated with the residence time. Mean particle sizes increased with
distribution that was considered to be caused by increased accuracy
residence time in the reactor, as expected, because of the larger growth
in temperature control. A method has also been reported that
time available for the particles. Notably, the LFR produced wider particle
utilises a three-step process in order to form CdSe–ZnS composite
size distribution at higher linear velocity due to the steeper velocity
NPs. The method involves immersion of capillaries in different
profile, which caused higher axial dispersion of the growing colloidal
heated oil bathes, combined with a microfluidic mixer intermediate
particles as they flowed through the reactor. In agreement with the
step [141].
theory stated above, they found that the use of a SFR could minimise
the axial dispersion effect, and lead to a narrower size distribution of
the produced NPs. Additional advantage of slug-flow reactors is the 3.2.3. Kinetic control and investigation
simple separation of the gas from the final reaction mixture. Therefore, In addition to providing efficient control over the dimensional
they do not require post processing purification steps. However, careful characteristics of the nanomaterials produced, microreactors also
operation and low throughput are usually required to achieve the offer the possibility to investigate and control the fundamental
stability of the multiphase gas-flow pattern [110]. reaction processes of the NP formation. Microfluidic devices provide
Shestopalov et al. [132] reported the use of a droplet-flow (liquid– a platform for in situ monitoring of the temporal evolution of NP
liquid) reactor, where the multiphase flow was generated from two formation through the ability to spatially resolve nucleation and
immiscible fluids rather than from a gas–liquid flow. They further growth phases in the reactor during synthesis [142]. Investigation
demonstrated the utility of this droplet-based microfluidic method of the mechanisms of nanocrystal nucleation and growth is critical
to perform a multi-step synthesis of CdS/CdSe core–shell particles. to understand how to control nanostructure synthesis on a
Slug-flow reactors present the problem of the physical contact of production scale. Different techniques such as spatially resolved
the particle with the wall of the microchannel that could result in photoluminescence imaging and spectroscopy [118], and small
cross contamination and channel clogging. In contrast, microfluidic angle X-ray scattering (SAXS) [143] were used to investigate the
droplet-based reactors represent a possible solution to overcome this kinetics and mechanisms of nanoparticle nucleation and growth
problem, since the reaction droplets are not in physical contact with during synthesis in a microfluidic reaction channel, demonstrating
1504 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

the capability of microreactors as investigation tools on the The potential of on-line characterisation to offer a fast and effective
fundamentals of nanoprecipitation. way to optimise reactions based on real-time readout of particle
Nanomaterials synthesis in microfluidic devices takes also dimensions is obvious, however, very little research has been conducted
advantage of the unique ability of microfluidic channels to operate on on-chip characterisation of NPs, and the available information does
at continuous flow regime, allowing for spatial and temporal control not demonstrate comparable results with traditional off-chip methods.
of reactions by adding reagents at precise time intervals during the In particular, both on-chip DLS and CCS provide the opportunity for
reaction process. Such a feature endows microfluidic reactors with on-line measurement of nanoparticle size as small as 10 nm; however,
the ability to carry out pre– and post-treatments and multi-step this limits their application when NPs with size comparable with the
syntheses in the same reactor. lower detectability limit are produced. In this respect, 100 nm limit is
Shestopalov et al. [132] reported a droplet-based microfluidic considered as the larger optimal size for NPs intended for cancer
method for performing multi-step synthesis of quantum dots with therapy applications [149].
millisecond time-control. The system is constituted of two separate Notably, DeMello's group proposed an improved system constituted
and independent reaction zones that allow for the control of two by a microfluidic reactor to carry out the synthesis, and an on-line
different reactions or stages of the same reaction. For the production spectrometer to monitor the emission spectra of the emergent
of CdS quantum dots, a mixture of CdCl2 and mercaptopropionic acid particles [119]. The data collected were fed into a control algorithm
(MPA), a solution of NaOH and a solution of Na2S were infused into that automatically updated, without any human intervention, the
the left, centre and right inlets of the device, respectively. Droplets reaction conditions (flow rates and temperature) driving the system
were formed when these aqueous streams were infused in the oil to the desired goal.
stream. The reaction was allowed to take place for 75 ms and then
quenched with a stream containing MPA. A comparison of the results 3.2.5. High throughput production
obtained for the reactions with and without the quenching step The use of microreactors for industrial-scale production of
indicated a decrease in monodispersity when the quenching step nanomaterials requires extensive reactor parallelization, since each
was eliminated. They were also able to produce CdS/CdSe core shell individual unit provides only a small volume dedicated to reaction.
NPs infusing Na2Se instead of MPA, from the quench channel. A system Scaling up microfluidic device size signifies a marked increase of
for growth of shell coatings with tuneable thickness and without the volumetric flow rate of the liquid that is processed through the
complications due to secondary nucleation has been also presented by microdevice, according to the following equation [34]
Jensen et al. [144], for coating colloidal silica core particles with titania
layers through controlled hydrolysis of titanium tetraethoxide (TEOT). Q ¼ U·A ð9Þ
By controlling flow rates, reaction times and quenching procedure
Kumar et al. [36] were able to prepare Co NPs with three different where Q is the volumetric flow rate, U is the average fluid velocity,
crystal structures. Results from this study seem to indicate that the and A is the cumulative cross-sectional area of the microchannels.
difference in flow rate ratio and residence time influences the From the moment that a significant increase in fluid average velocity
nucleation phase in a critical way, leading to the production of NPs generally results in undesired increase of the pressure drop across
with different crystalline structures. This study suggests the possible the microchannel, a scale-up strategy usually involves increasing
application of microfluidic device to obtain NPs with different the cumulative flow through the cross-section of microchannels,
structure by controlling the process parameters, hence the kinetics of by increasing the number of microreactors. Generally, there are
reaction. It is well known that synthesis of NPs via metal salt reduction three levels of scale-up:
is often kinetically ― rather than thermodynamically ― controlled,
• Increasing the number of channels arraying identical channels in the
leading to the possibility of generating particle in metastable phases.
same lamina.
Control of the crystal structure is one of the key issues in nanoparticle
• Increasing the number of layers adding multi-layers with channel
synthesis since physical and chemical characteristics also depend
arrays.
directly on crystal structure [145].
• Increasing the number of devices using identical devices connected in
parallel.
3.2.4. On-chip analysis
Miniaturised and integrated on-chip analysis of nanomaterial can In order to achieve an efficient scale up of NP production in
play an important role in high throughput characterisation of various microfluidic channel, the same flow rate must be conveyed into all the
drug delivery formulations, accelerating the understanding of the effect arrayed microchannels. In fact, as explained in the previous section,
and the optimisation of the process parameters. On-chip detection the flow rate across the channel is important in controlling the reaction
devices are situated further downstream the nanoprecipitation zone to condition such as heat and mass transfer, and species residence time
allow in situ measurements. Positioning the detection unit at different [150]. The use of pumps for each channel can meet the requirement of
positions along the channel also enables the collection of spatial and uniform distribution of the flow but is not practical from the standpoint
temporal information about the reaction process. Unfortunately, few of the cost. A more reasonable strategy consisted of distributing the flow
methods exist for on-line size characterisation of the particles in the from a common reservoir through the microchannels to a common
channel. In traditional nanoparticle sizing, Dynamic Light Scattering product reservoir. However, the distribution of flow rate uniformly in
(DLS) is a powerful tool for in situ size analysis of nanometre structures. each channel is not an easy task, and so far, only few authors have
In addition, DLS is ideally suited to be incorporated into microfluidic conducted investigation in this direction.
devices because it can be adapted to measure very small fluid samples. Amador et al. [150] studied, using a method based on electrical
Chastek and colleagues reported a microfluidic chip coupled with a resistance networks, the flow distribution in arrayed microchannel for
DLS apparatus through fibre optic probes that provided a quantitative two different manifold structures called “consecutive” and “bifurcation”.
measurement of micelles as small as 10nm in diameter [146,147]. Kuiper They found that in the absence of channel dimensional variations
et al. [148] described the use of an apparatus in which confocal the bifurcation structure always produced flow equipartitioning as long
correlation spectroscopy (CCS) was employed for sizing both fluorescent as the length of the straight channel after each channel bend was
and non-fluorescent NPs on-chip. CCS compared to other sizing sufficient for a symmetrical velocity profile to develop. Commenge et al.
techniques, such as light scattering, has the ability to measure dilute [151] analysed the influence of the geometrical dimensions of the
samples in small volumes. With such a device, they measured particles reactor microstructure on the velocity distribution between channels,
ranging in diameter from 11 to 300 nm. and optimised the flow distribution in a multichannel system with
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1505

a consecutive distribution. The results obtained were used to design 4.1. Organic nanoparticles
geometries resulting in uniform velocity distributions between the
channels for single-phase flow. A summary of the reviewed studies reporting microfluidic-based
production of organic nanoparticles is provided in Table 1, together
3.3. Materials used in the fabrication of microfluidic chip for the production with the type of nanoparticles produced and the microfluidic strategy
of nanomaterials adopted.
The group of Yao reported a novel method of generation of solid lipid
The fabrication techniques of microfluidic reactors are derived nanoparticles (SLN) using a flow focusing strategy [127,162]. Solid lipid
from microelectronic technology, where silicon is extensively used nanoparticles are a drug delivery system constituted of biodegradable
[136,152,153]. The use of monocrystalline silicon is justified because physiological lipids or lipid substance, and represent an alternative
of its ready availability, the possible integration with electronic drug delivery system to liposomes and polymeric nanoparticles [163].
circuits and its physicochemical characteristics that are compatible The microfluidic method presented involves the precipitation of
with a large number of applications. nanoparticles from a solution of lipid in water-miscible solvent when
Besides silicon, glass has also found application in microreactors it is brought in contact with two lateral aqueous streams. Using this
fabrication. Glass is a particularly favourable substrate material for strategy, the authors were able to produce SLN, with an almost spherical
microfluidic reactors because it has good optical properties, efficient shape, in a range of mean diameters between 100 and 200nm by simply
heat dissipation and a high resistance to mechanical and chemical stress varying the flow rate ratio of the two phases.
[154]. The most commonly used glass types in microfluidics are Ali and colleagues [117] recently reported a microfluidic strategy
borosilicate, quartz and crown white glass (because of their excellent based on and Y-shaped micromixer for the preparation of hydrocortisone
optical properties) and soda-lime glass. In addition, their low cost nanosuspension through a nanoprecipitation technique. A relatively
meets the economic requirement of disposable devices. stable aqueous hydrocortisone nanosuspension was obtained, with
However, the main disadvantage in using glass as a substrate is its nanoparticle dimension being controlled by flow rate and design of
amorphous structure. This can generate non-parallel walls when it is microfluidic reactor. A similar technique was adopted by Su et al. [164]
processed using an isotropic HF wet etching technique. Because the to produce 2,2-dipyridylamine (DPA) nanoparticle. Interestingly they
etching process occurs on the exposed glass surface, as the channel used a commercial T-junction micromixer connected to 0.5mm ID tubing
etches deeper the side walls are also etched leading to low aspect as microreactors instead of a more conventional planar microfluidic
ratio channel geometries. To obtain a very deep channel, a dry etching device. Genot et al. [165] proposed the use of a simple 3D microfluidic
technique such as deep reactive ion etching (DRIE) is used, but this device in order to solve the problem of fouling and clogging within the
requires expensive instrumentation to be performed [152]. main channel due to the deposition of the precipitating species. They
The long cycle times and sophisticated equipment associated with constructed a simple PDMS Y-junction device in which they later inserted
silicon and glass microfabrication require more laborious and less a glass microcapillary positioned into the intersection between the
accessible fabrication methods. Polymers have assumed the leading channels. Without any particular laborious microfabrication process
role as substrate materials for microfluidic devices in recent years they were able to produce a stable 3D focused stream and estimated
[155], of which poly(dimethylsiloxane) (PDMS) has become the mixing efficiency and residence time for the production of rubrene
preferred material for the construction of microfluidic devices because nanoparticles as model organic nanomaterial.
it can be easily moulded and patterned into channels, it can reproduce Finally, Dev et al. [166] reported the use of an unusual mixer for the
micrometre-sized features with high fidelity, it is optically transparent solvent-free production of crystalline meloxicam nanoparticles. The
and it has low permeability to water. However, the major disadvantage procedure involved the mixing of an alkaline solution of the drug with
of PDMS in organic nanoparticles synthesis is its organic solvent a citric acid solution within a rotating tube processor (RTP) [167]. The
resistance; it can swell in contact with organic solvents, such as aliphatic RTP consisted of a 6 × 30 cm (diameter × length) aluminium tube which
and aromatic hydrocarbons, and even dissolve in amine and strong rotated at a speed between 100 and 2000 rpm. The two solutions were
acids (e.g. sulphuric and trifluoroacetic acid) [156]. Alternative polymer fed at one end of the tube. When the solution came in contact with the
materials (i.e. modified PDMS, acrylates, PEEK and COC) have recently inner wall of the tube, it spread on its surface and the rotating movement
been considered in a rapid prototyping technique to fabricate more intensified the mixing process. Authors reported that the fast mixing
solvent-resistant microfluidic reactors [157–159]. kinetics within the RTP caused the formation of thermodynamically
Procedures of microfluidic devices fabrication are not addressed unstable semicrystalline nanoparticles which slowly turned to needle-
here, since they fall out of the scope of this chapter. The reader is like particles after 2 h from preparation.
therefore directed to a number of excellent, recent review articles and
books on this specific subject [152,155,160,161].
4.2. Nanomaterial from self-assembly of amphiphilic molecules
4. Microfluidic production of nanoparticle and vesicular systems
The synthesis of drug carriers, such as liposomes [168], niosomes
In the past decade, there has been a tremendous interest in the [169] and polymeric micelles (PMs) [1,128,170–172], produced
development of microfluidic methods for chemical synthesis of inorganic through the self-assembly of amphiphilic molecules (phospholipids,
nanoparticles for the production of nanocrystalline semiconductors with
tuneable size and uniform distribution of size [35]. However, there has
been much less research on microfluidic methods for the production of Table 1
organic nanoparticles and colloidal systems. Summary of the studies investigating microfluidic-based production of organic
The reason for the investigation of such type of materials mainly nanoparticles.
stems from their possible use as innovative pharmaceutical formulations Nanoparticles type Microfluidic strategy References
able to increase, in different ways, the efficacy of poorly soluble drugs. In
Solid lipid nanoparticles (SLN) Planar flow focusing mixer Zhang et al. [127],
the following section the various strategies applied for the microfluidic- Yun et al. [162]
based production of nanomaterials and vesicular systems for drug Hydrocortisone nanosuspension Y-shaped mixer Ali et al. [117]
delivery applications will be reviewed. For the sake of clarity, we have 2,2-dipyridylamine (DPA) T-shaped mixer; multiphase Su et al. [164]
presented the various strategies dividing them in terms of type of Rubrene 3D flow focusing mixer Genot et al. [165]
Meloxicam Rotating tube processor (RTP) Dev et al. [166]
nanomaterial produced.
1506 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

non-ionic surfactants and block-copolymers, respectively) has also associated with heterogeneous and poorly-controlled chemical and
been investigated. The aggregation of amphiphilic molecules mechanical conditions during lipid vesicles formation, and result in
to form nanomaterial slightly differs from traditional nucleation non-uniform vesicles in both size and lamellarity.
and crystal growth that govern the nanoprecipitation of non- On the other hand, a range of different methods have been
amphiphilic molecules (i.e. inorganic material and hydrocortisone) implemented for the production, manipulation and analysis of lipid
[173]. In the nanoprecipitation of non-amphiphilic molecules, the vesicles ― such as liposomes ― within microscale environments
size of the nanoparticles is predominately controlled by the level of [177,178]. In recent years, considerable attention has been devoted
supersaturation of the solute to be precipitated, and by the duration particularly to the implementation of microfluidic-based strategies for
of the precipitation process. These two parameters can be finely the production of lipid vesicles, with the aim of overcoming the above
adjusted in microfluidic devices by varying the relative flow rate mentioned limitations associated with conventional batch systems.
ratio of the solvent and non-solvent streams and the residence In this regard, it is possible to distinguish between methodologies
time distribution respectively, as demonstrated by other researchers which mostly rely on the scaling-down of analogous macroscale
[117,132,144]. In contrast, the size of nanoparticles obtained from techniques and methodologies which have been designed specifically
amphiphilic molecules is dictated by the magnitude of the steric for application at the microscale. In the present paragraph, an overview
and electrostatic interaction between the amphiphilic molecules, about recent advancements in the field of microscale production of lipid
which is controlled by the polarity of the surrounding solvent vesicles is provided.
environment (i.e. solvent vs. non-solvent ratio) [173–176]. As the
mixing between solvent and non-solvent proceeds, the polarity of
the environment surrounding the amphiphilic molecules increases 4.2.1.1. Electroformation. Electroformation has been widely-adopted for
(i.e. increase in water content), and forces the closure of the growing producing lipid unilamellar vesicles [179]. The process has been firstly
assembly to minimise the exposure of the hydrophobic portion to described in 1986 by Angelova and Dimitrov [60] and involves the
the aqueous environment. As a result, an insertion barrier is created application of an electric field, typically alternating, across a lipid film
that prevents new amphiphilic molecules from entering inside the which has been previously deposited on the surface of an electrode
closed assembly, resulting in kinetically quenching of the growth and immersed in an aqueous solution. In this way, lipids interact with
phase [173–175]. the aqueous solution and the applied electric field by peeling off the
For example, HFF-based reactors provide the possibility of tuning the surface of the electrode in layers and self-assembling into giant vesicles.
mixing time by changing the width of the focused stream. This gives It is believed that the electric field generates fluctuations in the lipid
opportunity to control the rate of solvent shift, and exert a control over bilayers which are responsible for the film detachment [179].
the length of the aggregation time before the formation of kinetically This method has been successfully translated from the macroscale to
quenched nanoparticles. This ultimately results in control over the the microscale, firstly by Kuribayashi et al. [180,181]. In this works,
assembly process, resulting in tuneable characteristics of the produced authors produced various types of giant liposomes encapsulating nano/
nanomaterial. It is important to note that when drug and amphiphilic micro-functional materials (i.e. fluorescent polystyrene beads) within
molecules are co-precipitated the two mechanisms of aggregation, for squared cross-section microfluidic channels. Channels were sandwiched
amphiphilic and non-amphiphilic molecules, act in concert to determine between glass layers which were previously coated with indium tin
the nanoparticle characteristics [1]. oxide (ITO) electrodes. An electric field was applied across the electrode,
A summary of the reviewed studies reporting microfluidic-based initiating the vesicle formation process. The method was compared with
production of NPs formed by self-assembly of amphiphilic molecules other techniques (i.e. gentle hydration) showing increased number of
is reported in Table 2. giant liposomes and reduced enclosing of smaller liposomes within larger
ones [181]. However, liposomes obtained with this technique were
polydisperse, implying that further technological advancements were
4.2.1. Lipid vesicles required to improve control over vesicles dimensional characteristics.
The bulk hydration of lipids in aqueous buffer generally yields large, In order to address this issue, a development of this set-up was later
polydisperse and multilamellar lipid vesicles (i.e. constituted of multiple presented [182], which differentiated from systems previously available
lipid bilayers concentrically organised as an onion-like structure). This given that the lipid film was micro-patterned onto the electrode surface.
method and other traditional bulk production methods, such as freeze– This was achieved by deposition of a phospholipid solution onto a film
thaw cycling, film hydration and reversed phase evaporation, are often of Parylene, which was previously patterned by photolithography and

Table 2
Summary of the studies investigating microfluidic-based production of nanomaterials from self-assembly of amphiphilic molecules.

Self-assembly type Microfluidic strategy References

Liposome Electroformation Kuribayashi et al. [180–183], Le Berre et al. [184], Diguet et al. [185]
Liposome Hydration Lin et al. [189,190], Osaki et al. [192]
Liposome Extrusion Dittrich et al. [193]
Liposome Planar flow focusing mixer Jahn et al. [126,195,199], Wi et al. [196], Zook et al. [197]
Liposome 3D flow focusing mixer Kennedy et al. [200], Paphal et al. [202]
Liposome Ultrasonic mixing Huang et al. [203], Yamashita et al. [204]
Liposome Pulsed jetting Funakoshi et al. [209], Stachoviak et al. [211,213], Richmond et al. [212]
Liposome Transient membrane ejection Ota et al. [214], Kurakazu et al. [215]
Liposome Double emulsion template Shum et al. [210], Tan et al. [217], Teh et al. [218], Davies et al. [221]
Liposome Droplet emulsion transfer Matosevic et al. [228]
Lipid-stabilised microbubbles Planar flow focusing mixer Hettiarachchi et al. [205,207], Talu et al. [206]
Niosomes Planar flow focusing mixer Lo et al. [169]
Polymersomes Emulsion templates Shum et al. [230,231], Duncanson et al. [229], Seth et al. [232],
Kim et al. [233,235,236,239], Perro et al. [234], Thiele et al. [238]
Polymersomes Mixing-based Thiele et al. [241], Brown et al. [242], Thiermann et al. [243]
Polymer-stabilised microbubbles Emulsion templates Lee et al. [240]
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1507

etched with O2 plasma. Thus, by removing the Parylene film, authors control over film peeling, vesicles obtained with this method are generally
obtained a micro-patterned lipid film. With this technique, it has polydisperse and multilamellar. Additionally, the performance is
been demonstrated that different liposome shapes could be obtained extremely sensitive to the physical properties of the fluid medium,
depending on the spatial arrangement of the patterns. Notably, including pH and temperature. In an attempt to overcome these
densely-spaced patterns resulted in the production of balloon-like limitations, hydration has been performed within microfluidic-based
liposomes, while sparsely-spaced patterns resulted in the production devices where individual physical variables can be controlled more
of hemi-spherical dome-shaped liposomes. precisely. The translation of this method in a microscale system has
An analogous microfluidic-based device was employed to produce been firstly described by Lin et al. [189,190] (Fig. 2c). Authors fabricated
solvent-free lipid membranes by electroformation [183]. For this a squared cross-section microchannel within a PDMS layer by soft
purpose, a stack of lipid-coated Parylene masks was aligned onto a lithography technique, which was then bonded to a glass layer. Two
micro-patterned Cr/Au aperture array. Circular lipid patterns aligned cavities were created for buffer pumping and lipid film hydration,
with the micro-apertures were created by removing successive layers of respectively. Lipid microtubes (1–10 μm in diameter) and vesicles were
Parylene. A second electrode was then placed above the micropatterned formed by flowing a buffered aqueous solution over the lipid film
surface and an electric potential was applied across these plates, resulting (Fig. 2c). With this technique it was possible to control the type of
in the formation of solvent-free lipid membranes. This approach allowed obtained lipid structures by regulating the governing physical parameters
overcoming limitations associated with the difficulty of removing (i.e. fluid flow rate). However, the formed liposomes were neither
solvents from the formed lipid vesicles, which can potentially limit their unilamellar nor monodisperse.
application as in-vitro models of biological cells. Notably, the correct In order to achieve better control over vesicles size, a technique
functioning of membrane proteins may be significantly compromised combining micropatterning and gentle hydration was later developed
by the presence of residual organic solvent within the lipid bilayer. by Osaki et al. [191]. Authors performed selective lipid patterning by
Significant steps towards increased control over liposome size were electrospray deposition, following a protocol previously described
further demonstrated by Le Berre et al. [184] (Fig. 2a–b) and Diguet by the same group [192]. The patterned substrate was fabricated
et al. [185] by combining surface micro-patterning and electroformation. by standard photolithography and wet-etching methods. Hydration of
In these works the authors deposited lipid films onto a silicone (Si) the lipid film resulted in formation of giant liposomes uniform in size,
substrate having controlled surface properties (Fig. 2a). By changing the which diameter could be controlled by regulating the size of the
chemical and physical properties of the Si surface, it was demonstrated patterned features. Lipid vesicles produced with this method were
that the properties of the final vesicles (i.e. size and size distribution) arrayed directly on the substrate, which makes this technique suitable
could be finely controlled (Fig. 2b). It should be highlighted that these for on-time in situ bio-chemical analyses.
techniques were capable of producing monodisperse liposomes, thus However, one drawback of hydration methods is the need for lipid
representing a significant improvement over standard electroformation film deposition prior to each process iteration (i.e. vesicle production
methods. step), which limits its potential application in large-scale vesicles
To conclude, it must be mentioned that the effect of the applied production systems due to the laborious and non-continuous nature
electric field on the integrity of membrane proteins or other bio-active of the procedure [177].
molecules has not yet been fully investigated [177], hence representing
a subject of future studies, although incorporation of functional proteins 4.2.1.3. Extrusion. An alternative method to produce unilamellar
within lipid vesicles membranes produced by electroformation has liposomes with controlled size is represented by extrusion [179]. In
been already reported [186,187]. this technique, vesicle dispersion is forced through a membrane with
pores of controlled size, during which lipid membrane rupture and
4.2.1.2. Hydration. Hydration is another widely-adopted method to resealing occur. Vesicles are generally extruded multiple times, until
produce lipid vesicles, in the form of either multilamellar and unilamellar the desired size is achieved. Notably, the latter can be controlled by
giant liposomes [179,188]. In this method, an osmotic pressure gradient regulating the size of the pores. This technique has been performed in
drives water from an aqueous buffered solution into a dried lipid film, a microfluidic device by Dittrich et al. [193], with further modifications
causing the lipid layers to separate [179]. Furthermore, by imposing a with respect to standard macroscale approaches. In this study, a 200 nm
fluid shear stress over the lipid-coated surface results in lipid film rupture thick micropatterned membrane of silicon nitride was fabricated and
and self-assembly into liposomes [177]. However, due to the poor sandwiched between two PDMS layers with channels for fluid delivery

Fig. 2. Microfluidic-based production of lipid vesicles. (a) Schematic of a microfluidic device employed to form liposomes by electroformation. Phase-contrast microscope images of
liposomes obtained by using this device and collected by gravity are reported in (b), for either ITO (left) or Si (right) substrate (reprinted with permission from [184]. Copyright 2008
American Chemical Society). (c) Microscope image illustrating the formation of a single lipid microtube in a PDMS microchannel, by hydration method upon application of a fluid flow
over a thin lipid film (reprinted from [190], Copyright 2006, with permission from Elsevier). (d) Schematic of the process of liposomes formation by self-assembly in a microfluidic
HFF device. Contours of isopropanol/water ratio are reported, illustrating diffusion-governed mass transfer along the microchannel. Furthermore, a three-dimensional contour map of
the fluorescence intensity of a membrane intercalating agent (DiIC18) at the focusing region is illustrated in (e), providing information on the spatial evolution of liposomes formation
(reprinted with permission from [126], Copyright 2004 American Chemical Society).
1508 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

and release. 3 μm diameter apertures were created by photolithography A development of this device was later presented [168], which was
technique on the silicone membrane, which was later coated with characterised by a multiple inlet channel configuration allowing for
an ~ 50 μm multilamellar lipid film. By applying a fluid flow either in water-soluble substances to be encapsulated to the immediate vicinity
the top or in the bottom channels, the lipid layer was extruded through of the alcohol stream where the formation of liposomes occurred, thus
the apertures and micrometre-sized lipid vesicles and tubes were reducing sample consumption. Furthermore, channels were deeper and
formed. Importantly, the dimensional properties of the obtained with a higher aspect ratio, resulting in a more homogeneous velocity
vesicles and tubes were observed to depend on both the size of the profile across the channel height, reducing surface effects in close
apertures and the pressure difference between the top and bottom proximity to the top and bottom walls of the channels. In this study,
channels. A strong pressure difference favoured the formation of lipid authors investigated the effect of total flow rate and flow rate ratio on
tubes, while a low pressure difference resulted in the formation of liposome characteristics. Contrary to previous studies [126], it was
vesicles. Despite the formed vesicles were polydisperse, a considerable reported that by varying the total flow rate, at a constant flow rate
advantage of this technique was that vesicles could potentially be ratio, had no significant impact on liposome size and size distribution,
generated with high flow speed, making it potentially suitable for which suggested that the absolute magnitude of the shear forces between
large-scale production applications. However, set-up implementation the streams had no significant impact on liposomes dimensional
(i.e. membrane coating) is non-trivial [177], which poses some properties. Furthermore, an inverse relationship between flow rate ratio
limitations to its widespread usage. Furthermore, the lack of isolation and liposome mean diameter was determined, which was coherent
between the solution to be encapsulated and that in which the with previous studies and was attributed to changes in the dynamics of
produced vesicles are stored, likely results in poor encapsulation species mass transfer across the streams, rather than to the effect of
efficiency [194]. These limitations may represent the subject of future shear forces at the interface between streams as postulated in previous
investigations. studies [126].
In a further study by Jahn et al. [176] the effect of device design
4.2.1.4. Hydrodynamic flow focusing. As mentioned earlier, the most on liposome characteristics was investigated. For this purpose,
popular way of producing lipid vesicles in a microfluidic system is different intersection angles between the lateral and central
represented by hydrodynamic flow focusing (HFF) [195]. In this inlet channels (i.e. 45° and 90°) and different channel cross-sectional
method, a central flow stream of a lipid-containing alcohol solution is dimensions were considered. Results revealed that microfluidic device
intersected and sheathed by two lateral streams of aqueous solution size and geometry did not affect significantly the properties of the
(Fig. 2d–e). The lipid stream is hydrodynamically focused into a narrow produced vesicles; however, they may have distinct and important
sheet, which size can be varied by regulating the volumetric flow rate technological implications, since larger devices may be easier to fabricate
ratio between central and lateral streams. Either lateral (i.e. two- and operate, and are generally associated with higher throughput
dimensional) or co-axial (i.e. three-dimensional) focusing of the central performance compared with smaller devices (i.e. due to higher flow
stream have been demonstrated [105]. rates achievable). These studies clearly showed that nanometre-sized
It is believed that, with this technique, the formation of liposomes is liposomes with a very narrow size distribution could be produced using
governed mainly by the diffusive mass transfer of species at the miscible continuous-flow microfluidics and opened the way for potential
liquids interface. Notably, the alcohol diffuses into the aqueous solution, applications for on-demand liposome-mediated delivery of point-of-
until it reaches a critical concentration (i.e. below the solubility limit of care personalised therapeutics.
lipids) which triggers the formation of lipid vesicles by self-assembly, Towards this goal, Wi et al. [196] recently developed a HFF
leading to the formation of small, monodisperse and unilamellar microdevice for producing liposomes, which were used as delivery
vesicles. It is technically complex to obtain experimental evidence of vehicles of genes into a porcine kidney cell line. Liposomes were produced
the dynamics of vesicle formation in situ, however, it is believed that using PDMS microchannels, fabricated by soft photolithography
the reduction of lipid solubility associated with alcohol and water technique. At the outlet, a 45° channel intersection was designed
diffusion across the two streams causes thermodynamic instabilities at to separate the smaller liposomes (diameter at 25 °C:~200–300 nm) ―
the edges of bilayer phospholipid fragments (BPF), which bend and which diffused laterally ― from the larger liposomes and lipid molecules
close upon themselves forming the vesicles [195]. (diameter at 25 °C: ~400–600 nm). The effect of processing temperature
The main advantages of this technique include: (i) the viscous- and flow rate ratio on liposomes size distribution and zeta potential was
dominated flow regime (i.e. Re ≪ 1) allowing to finely control the evaluated in this study. Liposomes size was observed to be virtually
interfacial region between the fluid streams; (ii) the small scale which invariant with changing the flow rate ratio in the range 2–7, while
makes the diffusive mass transfer an effective mixing method; (iii) the temperature had a significant effect on liposomes size. With this respect,
possibility to precisely regulate the boundary conditions, which permits liposomes produced at the higher temperatures (i.e. 60 °C) were smaller
to adjust on-demand the governing physical parameters (i.e. residence than liposomes produced at environmental temperature (i.e. 25 °C). This
time of solutes); (iv) the continuous-flow format, which can be exploited was attributed to the reduced interfacial tension between lipid and buffer
for large-scale production of lipid vesicles; and (v) no further post- streams at the higher temperatures, which favoured species mixing
processing steps to homogenise vesicles size are generally required. between the two streams. Furthermore, zeta potential was quantified as
The first work to have demonstrated production of liposomes using an index of colloidal stability and it was found that temperature had no
this method was reported by Jahn and coworkers [126] (Fig. 2d), in significant effect on it, while increasing the flow rate ratio resulted in a
which a trapezoidal cross-section microfluidic device with an angle of reduction of zeta potential, likely due to the reduced vesicles size [196].
90° between the central and lateral inlet channels (Fig. 2d–e) was It must be noted that one drawback of HFF devices produced using
employed to produce monodisperse liposomes in the size range deformable PDMS is the reduced control over the hydrodynamic field
100–250 nm. Importantly, it was demonstrated that by varying the flow due to undesired changes in channels cross-sectional area, particularly
rate ratio (i.e. ratio between the flow rate in the side inlet channels and at the higher flow rates required for large scale production [169].
the flow rate in the central inlet channel) it was possible to control the Further investigations about the role played by the operating
mean size and the size distribution of the obtained liposomes. Notably, temperature on the properties of liposomes formed by HFF have been
higher flow rate ratio was associated with lower vesicle size, which performed by Zook et al. [197], by using a microfluidic device fabricated
was attributed to the increased shear stress exerted on the liposomes by ion etching of Si layers. Authors demonstrated that a kinetic theory
during self-assembly. Furthermore, continuous-flow encapsulation of could be applied to describe lipid vesicles formation [198], which
high concentration of a fluorescent probe (carboxy-fluorescein, CF) was provided a suitable hypothesis to explain ― at least qualitatively ― the
demonstrated with this technique. results obtained by various liposome formulation strategies currently
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1509

employed. According to this theory, the liposome size mainly depended micro-convective mixing, rather than on diffusive mixing only
on the effect that the operating temperature exerted on the line tension (conforming to Pe≪1). With this respect, two hydrodynamic regimes
and the elasticity modulus of the lipid bilayer. In particular, it has been were identified, one characterised by hydrodynamic instabilities induced
postulated that, at temperatures lower than the gel-to-liquid crystal by viscous stratification (at low Re), leading to gentle mixing and lager
phase transition, the lipid membrane is stiffer and the planar bilayer liposomes. In the second regime, mixing between the streams did not
discs continue to grow until they reach an energetically favourable occur within the microchannels, but streams mixed as droplets formed
size at which they bend and form spherical liposomes. It was further at the outlet when collected in a reservoir. This vigorous mixing led to
demonstrated that, at temperatures not too far from the phase transition, the formation of smaller liposomes. Further experimental and theoretical
lower flow rate ratios resulted in the formation of large lipid aggregates investigations may be required to support this fascinating hypothesis. In
within the focusing region, while at the higher flow rate ratios it was this respect, the technique recently developed by Jahn et al. [199] may
possible to form medium-sized liposomes (i.e. radius b 50 μm). Results be employed to further support these experimental findings.
from this study suggested that considerable attention must be paid to It is worth mentioning that ultrasonic waves have been employed
the transition temperature when producing liposomes using microfluidic to enhance species mass transfer within microfluidic environments
systems [198]. However, further investigations are required in order to [92,94], and few studies have been reported regarding the combined
get an all-encompassing understanding of the dynamics of lipid vesicles use of microfluidics and ultrasound (US) for liposomes production. For
formation, and the role played by the local physical properties of the example, in a study by Huang et al. [203] a HFF microfluidic device has
micro-environment on liposome characteristics. been immersed in an ultrasonic chamber and statistically-significant
In order to facilitate the understanding of liposome formation reduction of mean liposomes size by sonication was demonstrated.
mechanisms within HFF microdevices, a recent study by Jahn et al. This could be attributed to the increased mixing, resulting from
[199] reported a relevant technological advancement allowing for the generation of cavitational events within the acoustic chamber. In a
instantaneous immobilisation of the focused stream by rapid freezing study by Yamashita et al. [204] ultrasound exposure of a previously-
with liquid propane, and subsequent cryo-ultramicrotomy and cryo- formed liposome solution was observed to markedly reduce vesicle size
SEM imaging of formed liposomes and disc-like intermediate lipid distribution if performed in a microscale domain (i.e. microcapillary),
structures. Interestingly, randomly distributed clusters characterised rather than in a macroscale system. These preliminary experimental
by high number densities of spherical and flattened lipid vesicles findings suggest that US may be employed as an external physical
were observed, which formation was attributed to the occurrence of stimulus to trigger or control liposomes formation, thus opening the
hydrodynamic instabilities between liquids of different dynamic way for innovative technological advancements in this research field.
viscosity. Using this method, the characteristics of lipid vesicles could Towards this, a more pervasive understanding of the mechanisms
be investigated without altering the physical properties of the governing liposomes formation and US–liposome bilayer interaction is
molecules of interest. Notably, this technique could be extended to the required.
production of other nanoparticles, including polymeric micelles and To conclude, it is worth mentioning that the use of microfluidic-
organic nanoparticles. based HFF has been also reported for the production of lipid stabilised
While studies mentioned above mostly relied on a two-dimensional microbubbles which are used as ultrasound contrast agents or delivery
focusing of the lipid stream; co-axial or three-dimensional focusing has vehicles of bioactive molecules (i.e. anticancer drugs) for US-enhanced
been recently reported, in which the central stream is sheathed in targeted drug delivery [205–207]. Notably, the performance of lipid-
a cylindrical core at the centre of the channel. This approach allows stabilised microbubbles exposed to ultrasonic waves, and associated
overcoming the hydrodynamic effects resulting from the interaction biological effects have been recently investigated by using microfluidic
between the central stream and channel walls, and it is also associated devices [208].
with a more uniform velocity profile in the focused stream region.
Recently, in a study from Kennedy et al. [200] the interplay between
fluid dynamics and mass transfer within a PDMS-based co-axial 4.2.1.5. Pulsed jetting and transient membrane ejection. The formation of
flow focusing microdevice has been thoroughly investigated (Fig. 3a), liposomes by pulsed jetting has been firstly reported by Funakoshi
together with its effect on the properties of the produced liposomes. et al. [209]. In this method vesicles are formed by mechanical
In this device, a lipid solution was focused at the centre of a deformation of a planar lipid bilayer established between two aqueous
125 μm × 125 μm squared cross-section channel [201]. Authors layers (i.e. two water droplets surrounded by an organic solvent).
performed CFD simulations and confocal microscopy acquisitions Membrane deformation is generated by the action of a finely-
to characterise species advection–diffusion along the channel, and controlled pulsed liquid jet against the lipid bilayer (Fig. 3c). Micro-
computational results showed agreement with experiments (Fig. 3b) nozzles or micro-pipettes are generally employed for this purpose.
regarding the shape of the concentration distribution along the The main advantage of this method is that materials contained in the
microchannel. Liposomes in the size range 100–300 nm were produced fluid jet are directly encapsulated in the formed vesicles, with high
with this technique and the size of the produced vesicles was observed efficiency (Fig. 3d–e) [177]. For example, the encapsulation of Jurkat
to increase with increasing the ionic strength, which was attributed cells and chromosomes [209] (Fig. 3d–e), nanoparticles and membrane
to competition for water solvation of lipids by salts. Furthermore, proteins [211], and signalling lipids [212] have been achieved. With this
increasing the concentration of phospholipids in the core solution technique, mostly unilamellar, monodisperse giant vesicles have
resulted in the production of smaller liposomes, which was coherent been produced; however, under certain operating conditions, smaller
with previous studies [197]. It must be noted that liposomes produced satellite vesicles can be formed. The average vesicle diameter of both
with this technique were larger than the ones produced in previous giant and satellite vesicles has been observed to increase with
studies, at comparable sheath-to-core ratios. This is likely due to increasing the dispense time [209].
differences in microchannel size and architecture, and in the molecular In a study by Stachoviak et al. [211], a microfluidic-based pulsed-
weight of the lipids employed. jetting system was developed for increased control over the boundary
In a recent study, Phapal et al. [202] further investigated the conditions and the micro-environment. Interestingly, high-speed video
mechanisms of liposomes formation within 3D HFF microfluidic devices. microscopy was employed to reveal the dynamics of fluid–membrane
For this purpose, an easily reproducible technique was developed interaction and vesicle formation. Monodisperse giant vesicles (diameter:
for fabricating microchannels within a PDMS block. With using this ~200μm) were obtained in this study. Furthermore, it was suggested that
technique, authors demonstrated that at Peclet numbers, Pe≫1, the a combination of low-volume, high-velocity pulses from a small nozzle is
size of the formed liposomes depended significantly on the nature of required for reducing the size of the formed vesicles.
1510 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Fig. 3. Microfluidic-based production of lipid vesicles. (a) Schematic depiction of a microfluidic system for liposomes formation based on a three-dimensional HFF. Concentration contours
of a fluorescent solution reconstructed from confocal microscopy acquisitions illustrate device focusing performance (b) (reprinted with permission from [200]. Copyright 2012, American
Institute of Physics). (c) Formation of lipid vesicles by pulsed jetting method. With this technique, cells can be efficiently encapsulated within the vesicle, as demonstrated from
microscope acquisitions (d–e) (reprinted with permission from [209]. Copyright 2007 American Chemical Society). (f) Mechanism of lipid vesicle formation by W/O/W double emulsion
templates. (g) Images of the formation of phospholipid-stabilised W/O/W double emulsions in a glass microcapillary device (top). Optical micrograph of the collected double emulsions.
Drops are characterised by an aqueous core surrounded by a solvent shell containing phospholipid (bottom) (reprinted with permission from [210]. Copyright 2008 American
Chemical Society).

A development of this apparatus was later presented [212,213], 4.2.1.6. Double emulsion templates. Water in oil in water (W/O/W)
in which an inkjet nozzle was employed to generated fluid jets with double emulsification has been employed as a highly-efficient and
finely regulated characteristics. With this method, control over vesicle convenient route for producing monodisperse giant unilamellar vesicles
membrane composition, asymmetry and content was demonstrated; [177]. Briefly, W/O/W templates are converted into liposomes by
thus making it suitable for highly efficient formation of in-vitro models solvent extraction. Generally, evaporation of the oil phase is performed
of biological cells and high-throughput encapsulation of bio-active by addition of an organic solvent, forcing the lipid monolayers located
molecules for drug delivery applications. at the water–oil interfaces into a lipid bilayer membrane, which
However, despite the aforementioned advantages, the formation of arrange on the double emulsion templates, thereby forming lipid
liposomes by pulsed jetting represents a complex technological challenge, vesicles (Fig. 3f). A study by Shum et al. [210] firstly demonstrated
requiring advanced engineering tools and skills, and application-specific microfluidic-based production of liposomes from double emulsion
optimisation of the experimental protocols depending on the physical templates (Fig. 3g). A glass microcapillary device combining a co-flow
properties of the fluids and the micro-environment [177]. Furthermore, and a flow focusing geometry was employed to produce phospholipid-
it has not been clarified yet if the high shear stress levels associated with stabilised W/O/W double emulsions [216] (Fig. 3g), with a droplet
fluid jetting may potentially compromise the integrity of encapsulated generation frequency of approximately 500 Hz. With this method, the
bio-molecules or cells [177], thus opening the way for further dimensional properties of the shell of the double emulsions could be
investigations in the field. controlled by regulating the flow rates of each fluid phase and capillary
An analogous technique, referred to as transient membrane ejection diameter. Droplets produced with this method were uniform in size,
[177], has been developed by Ota et al. [214] for continuous generation making them suitable templates for obtaining monodisperse vesicles.
of monodisperse unilamellar vesicles. In this study, a lipid film was Furthermore, authors implemented experimental strategies to improve
formed in a T-junction, by sequentially infusing water, oil, and water the stability of fragile phospholipid vesicles during solvent extraction,
into the PDMS-based microfluidic device. The cross flow at the junction which would be otherwise non-trivial to produce.
acted on the lipid membrane, which was continuously thinned, sheared, A similar method was earlier described by Tan and coworkers
and squeezed. This resulted in the membrane releasing multiple vesicles, [217] to produce lipid vesicles with associated high encapsulation
which self-assembled encapsulating water droplets with uniform size. efficiency. In this method, the aqueous phase containing the
Notably, fluid displacement within the microchannels was generated by species to be encapsulated was emulsified in the lipid phase within
laser-controlled expansion of a gas bubble in the aqueous solution. a PDMS microfluidic channel, to produce stable lipid emulsions.
Liposomes produced with this technique were monodisperse (mean The lipid phase consisted of phospholipids dissolved in oleic acid.
diameter: ~16.5 μm), cell-sized and unilamellar. Furthermore, efficient The lipid solution emulsion was then injected into a mixture of water
encapsulation of bioactive molecules, such as membrane proteins and and organic solvent, causing the phospholipids to disassociate from
gene-expression systems, has been demonstrated. liquid lipids and arrange around the emulsions into lipid vesicles.
A development of this method was later presented by Kurakazu et al. Using this method, nano-sized proteins, 4.12 μm diameter fluorescent
[215], where pneumatically actuated valves were employed to drive the beads and fluorescently labelled HeLa cells have been encapsulated
fluids within the microfluidic channels, resulting in an automated lab- inside the formed phospholipid vesicles.
on-a-chip platform to be potentially employed for the production of More recently, Teh et al. [218] described the production of
unilamellar and monodisperse vesicles. monodisperse 20–110 μm diameter lipid vesicles by means of
However, as for pulsed jetting techniques, this method of vesicle double emulsion templates. An HFF device was fabricated in
production required laborious experimental steps to be implemented, PDMS, for producing emulsion droplets uniform in size. Precise
limiting its widespread usability. control over droplets size was achieved by regulation of the flow
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1511

rate ratio. Additionally, authors performed selective hydrophilic of the produced lipid vesicles and thus limit the range of potential
treatment of the PDMS channels surface by coating with PVA, applications.
following an approach described in previous studies [219,220].
With this respect, formation of water-in-oil droplets occurred 4.2.2. Niosomes
within untreated hydrophobic regions, while formation of oil-in- A synthetic, non-ionic, counterpart of liposomes is represented by
water droplets occurred within treated hydrophilic regions. membrane vesicles formed by self-assembly of non-ionic surfactants,
Compared to other hydrophilic surface treatments, the method the so-called niosomes. Hydrophilic molecules can be encapsulated
developed here did not require laborious mask alignment steps or within their aqueous core, while hydrophobic molecules can be
the use of hazardous chemicals. With the developed microfluidic partitioned into the membrane bilayer. For this reason, they are
device, authors demonstrated continuous-flow encapsulation of generally employed as carriers of bioactive molecules (i.e. anticancer
plasmids and long-term expression of GFP within the formed or anti-inflammatory drugs) and contrast agents for clinical imaging,
liposomes. However, no quantification of the encapsulation and their surface can be functionalised for targeted drug delivery
efficiency has been provided. It must be noted that liposomes applications. Conventional bulk methods for producing niosomes
produced with this platform had longer stability (N 3 months) suffer from the poor control over the chemical and mechanical
compared to studies previously reported [217]. environment, and thus generally result in large and polydisperse
Three-dimensional HFF was instead employed in a recent study vesicles. For this reason, post-processing steps (i.e. extrusion or
by Davies et al. [221] for generating double emulsion templates. In sonication) are required to reduce and homogenise vesicles size.
this work, authors developed a novel surface patterning technique With this respect, microfluidic-based production methods can be
using corona discharge. In this way, wettability of PDMS surfaces employed to achieve a more precise control over the physical and
could be spatially controlled, making this technique suitable for chemical properties of the fluidic environment. However, to the
selective surface treatment. Combining surface functionalisation best of our knowledge, only one study by Lo et al. [169] has been
with axisymmetric flow focusing, the developed microfluidic device reported, demonstrating the production of niosomes in a microfluidic
was capable of generating double emulsions in water and oleic acid, system. A microfluidic HFF device has been employed for this purpose,
which were then used as templates for liposome formation. Importantly, following a previous design reported by Jahn et al. [126] (Fig. 2d), in
surfactants were not required to lower the interfacial tension and create which a similar device configuration was employed for producing
the conditions for emulsification [222], making the fluidic environment liposomes.
safe for lipid bilayers membrane formation. Authors further investigated In this study, a surfactant mixture (consisting of a sorbitan ester,
the effect of the fluidic conditions (i.e. Capillary number) on droplets cholesterol and dicetyl phosphate) in isopropyl alcohol was injected in
characteristics and identified the optimal flow regime for the formation the central stream, while an aqueous buffer was injected into the two
of single–core–shell templates, suitable as precursors of liposomes. side streams. Either silicon metal or PDMS-glass hybrid microdevices
Liposomes produced with this technique were structurally integer, were employed. Authors investigated the effect of the fluidic conditions,
as supported by fluorescence recovery after photobleaching (FRAP) chemical structure of the surfactant and device materials on the
experiments. This study demonstrated the potential of combining characteristics of the formed niosomes. It was demonstrated that
different techniques to achieve formation of unilamellar vesicles to increasing the flow rate ratio resulted in a reduction of vesicle
be employed for a range of potential applications; opening the way mean diameter, due to the reduced species mixing time. This finding
for new advancements in this research field. was coherent with previous studies on liposome formation by HFF
[126,176]. On the contrary, wider channel geometry led to the
formation of larger niosomes, due to the increase in species mixing
4.2.1.7. Droplet emulsion transfer. In this method, water-in-oil emulsions time. With respect to the surfactant used, sorbitan ester with the
stabilised by a lipid monolayer are initially produced [223]. Subsequently, shorter and saturated alkyl chain assembled into the smallest
droplets are transferred to an aqueous medium and a second lipid vesicles. Additionally, both silicon metal and hybrid glass-PDMS
layer is deposited around the droplet while it crosses the oil–water microdevices were capable of producing niosomes by HFF. However,
interface. In this way, the lipid bilayer is assembled one leaflet at a as mentioned previously, attention must be paid on the effect that PDMS
time [223]. Importantly, the isolation between vesicle content and distensibility can have on the hydrodynamic focusing performance. By
the outer aqueous phase allows for efficient encapsulation. However, comparing niosomes produced by bulk method with those produced
when performed in macroscale batch systems, it results in the by the microfluidic method, it emerged that HFF led to the formation of
formation of generally polydisperse vesicles. smaller vesicles with a narrower size distribution.
In order to overcome this limitation, production of droplet Results from this study demonstrated the potential of microfluidics
emulsions could be accomplished in a microfluidic-based system, to produce synthetic membrane vesicles by self-assembly of non-ionic
in order to improve control over droplet size and size distribution surfactants. Future investigations should focus on the encapsulation of
[224–227]. In a study by Matosevic et al. [228], the production of molecules within the formed vesicles, thus opening the way for a robust
liposomes by droplet emulsion transfer was performed fully within microfluidic-based alternative to conventional bulk techniques.
a Y-shaped microfluidic device. Lipid-stabilised droplets were
generated and conveyed from one inlet channel, while an aqueous 4.2.3. Polymersomes
solution was delivered into the other inlet channel. In this way, after Polymersomes are synthetic vesicles constituted of a bilayer of
channels merged together, a lipid-stabilised oil–water interface was amphiphilic molecules (i.e. block copolymers) enclosing an inner
formed in close proximity to the droplets flow. A triangular post was liquid compartment (i.e. aqueous solution). Given the tuneable
fabricated in the centre of the channel, which diverted the droplets into permeability of polymersome membrane and the higher mechanical
the aqueous phase and forced them to cross the oil–water interface. In stability compared to phospholipid vesicles [229], they have been
this way, completely on-chip formation of the lipid bilayer was achieved. employed as carriers for drugs, enzymes, proteins, and genetic
Furthermore, authors demonstrated continuous flow encapsulation of material for controlled release applications [230].
fluorescent probe molecules having different size. However, droplets Analogous to liposomes production, conventional macroscale
displacement stage caused a significant percentage of the droplets to be methods for polymersomes synthesis include hydration and
disrupted and a gentler alternative may be evaluated in future studies. electroformation [231]. However, vesicles produced with these
A limitation of droplet emulsion transfer methods is the difficulty to techniques are generally large in size, polydisperse and poorly
obtain oil-free lipid bilayers, which may limit the functional behaviour efficient in encapsulating bioactive molecules.
1512 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

On the other hand, a range of different microfluidic-based devices this microdevice architecture has been employed in a range of studies
have been reported for the production of monodisperse polymersomes. to produce polymersomes with different characteristics. For example,
Here, they have been classified into two main categories, (i) double Seth et al. [232] produced monodisperse polymersomes encapsulating
emulsion templates and (ii) mixing-based techniques. magnetic nanoparticles, with potential application in MRI detection
and magnetic hyperthermia. Kim et al. [233] instead reported the
4.2.3.1. Double emulsion templates. W/O/W double emulsion templates formation of polymersomes with a hydrogel core. For this purpose,
have been widely employed to form polymersomes. This method the formed polymersomes were irradiated with UV causing cross-
closely resembles the one used to produce liposomes, and relies on linking of the prepolymers encapsulated within the vesicle. The
the evaporation of the solvent of a double emulsion stabilised by a hydrogel network was demonstrated (i) to facilitate sustained
copolymer [229]. As the solvent evaporates, the polymer self- release of encapsulated actives, (ii) to provide structural support to
assembles into a vesicular structure using the W/O/W emulsion as the bilayer membrane preventing its deformation under exposure
a template. Thus, monodisperse templates lead to the formation of to external forces, and (iii) to increase vesicle stability.
monodisperse polymersomes. With using the same microfluidic device, Shum et al. [230] reported
A microfluidic strategy to form polymersomes has been developed the formation of multi-compartmental polymersomes, which could be
by Shum et al. [231] (Fig. 4a–b). In this study, a glass capillary device potentially used to carry multiple actives in a single vesicle (Fig. 4c–d).
based on a two steps co-flow and flow-focusing architecture [216] Importantly, isolation between different compartments potentially
was used to form W/O/W emulsions stabilised by PEG(5000)-b- prevents the cross-contamination between incompatible actives and
PLA(5000) (Fig. 4a). Monodisperse vesicles were obtained by off- could be also exploited to encapsulate reactants and subsequently
chip solvent evaporation and were characterised by a mean diameter trigger the reaction by external physico-chemical stimulation (reaction
of approximately 24.5 μm. Furthermore, authors demonstrated on-demand) [230]. In this technique, during solvent evaporation the
highly-efficient encapsulation of hydrophilic probes (FITC-Dextran) copolymers at the inner water–oil (W/O) and at the oil–outer water
and the possibility of tuning on-demand the properties of the vesicle (O/W) interfaces are attracted towards each other and form the
wall (i.e. thickness and permeability) by changing the block ratio of membranes. As a consequence of this, adhesion between neighbouring
the unimer. Notably, release of the encapsulated probe could be inner droplets occurs and multi-compartmental vesicles are formed
triggered by osmotic shock, thus suggesting a simple mechanism for [229] (Fig. 4c). Authors demonstrated control over the number of
controlled release of bioactive molecules from vesicles. Furthermore, inner droplets within the double emulsion by varying the flow rate of

Fig. 4. Microfluidic-based production of polymersomes. (a) Schematic of a glass microcapillary device for generating W/O/W double emulsions. The injection tube and the collection tube
are tapered from glass capillary tubes having an outer diameter of 1 mm, and an inner diameter of 0.58 mm. Final diameters correspond to 10–50 μm (injection tube) and 40–100 μm
(collection tube). Bright field microscope images of formed double emulsion drops following the evaporation of organic solvent are reported in (b). The emulsion drop consists of an
aqueous core surrounded by a diblock copolymer shell, which is subjected to thinning during solvent evaporation (i.e. from left to right). Images are taken at time intervals of 1 h (reprinted
with permission from [231]. Copyright 2008 American Chemical Society). (c) Schematic illustrating the process of formation of multi-compartmental polymersomes from double emulsion
droplets with multiple inner droplets. A photograph of the operating glass microcapillary device is reported in (d), together with a microscope image of dewetted double emulsion drops
with two inner droplets (adapted with permission from [230]. Copyright 2011; John Wiley and Sons).
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1513

the three phases independently. By slightly modifying the previous step flow-focusing cross-junction geometry, where a diblock copolymer
microdevice design, and introducing two separate channels within was dissolved in the organic solvent stream at the first junction,
the injection capillary, authors demonstrated highly-efficient co- another organic solvent miscible with the copolymer-loaded solvent
encapsulation of FITC-Dextran and PEG, thus opening the way for the was injected at the second junction, and an aqueous solution was
use of this technique to co-encapsulate a range of bioactive agents for injected at the third junction. This configuration minimised the
therapeutic applications. The production of multi-compartmental precipitation of the diblock on the channel walls, which can affect the
emulsions using a microfluidic-based device was also demonstrated wettability of the PDMS surface and thus potentially impede emulsion
in a study by Perro et al. [234]. A “glue-free” microdevice consisting formation.
of fused silica capillaries and a T-junction was developed for this Another attempt towards high-throughput formation of poly-
purpose, based on co-flow architecture. Authors thoroughly investigated mersomes was recently performed by Kim et al. [239] who demonstrated
the effect of fluids flow rate, surface tension and dynamic viscosity on the development of a microfluidic device to achieve enhanced production
the size of the internal and external droplets. Based on the experimental of W/O/W emulsions used as templates for polymeric vesicles (Fig. 5a–c).
findings, a predictive model was developed to facilitate the selection of The microdevice relied on the parallelisation of the glass capillary
the optimal combination of operating conditions for a given application. configuration described in previous studies [210,216] (Fig. 5a). Droplets
Multi-compartmental poly(dimethylsiloxane)-graft-poly(ethyleneoxide) formation was performed over a wide range of volumetric flow rates,
and poly(butadiene)-block-poly(ethyleneoxide) vesicles were obtained enabling high-throughput and stable generation of polymersomes
with this technique, and the encapsulation of either hydrophilic or (i.e. production rate was ~4 times greater than that of single-capillary
hydrophobic molecules could be achieved. devices). Additionally, encapsulation of hydrophilic molecules in the
However, the achievement of accurate loading of multiple and distinct vesicle aqueous core was performed. However, the size distribution of
components, and their controlled sequential release without cross- the produced vesicles was wider compared to the one of vesicles
contamination, is still an important challenge towards the application of obtained by single-capillary devices, suggesting that more accurate
multi-compartmental vesicles in drug delivery. In order to address this control over the operating conditions and refinement of the device
issue, Kim et al. [235] developed a microfluidic-based strategy to produce architecture would be required.
polymersomes-in-polymersomes. In this technique, single-compartment To conclude, it is noteworthy to mention that emulsion templates
polymersomes were produced with using the glass capillary device have also been employed to produce monodisperse nanoparticle-
mentioned above (first step) [216,231]. The formed polymersomes shelled bubbles with potential application in the field of US-triggered
were collected in a reservoir and subsequently re-injected as the drug delivery. For example, Lee et al. [240] produced air-in-oil-in-
inner aqueous phase in another glass capillary microdevice, leading water (A/O/W) templates by using the glass-capillary microfluidic
to the formation of polymersomes-in-polymersomes (second step). system previously described [216]. Monodisperse and highly uniform
By varying the inner diameter of the collection tube between the A/O/W compound bubbles were generated at a typical frequency of
first and the second step, the number of encapsulated polymersomes several thousand hertz. PVA in the outer aqueous phase was employed
could be controlled. Importantly, authors demonstrated programmable to prevent the coalescence of the generated compound bubbles,
release of encapsulated molecules, in which the temporal evolution of thus leading to the formation of stable nanomaterials for long-term
the release process (i.e. simultaneous or sequential release) could applications.
be controlled by actively changing the stability of the vesicle layer
by means of hydrophobic homopolymers. Notably, higher order 4.2.3.2. Mixing-based techniques. Recent studies have demonstrated that
polymersomes-in-polymersomes structures could be achieved, by a hydrodynamically well-controlled nucleation and growth process
means of additional re-injection steps, leading to the encapsulation can be exploited to control polymersomes size and size distribution.
of multiple actives. This may open new and innovative perspectives In a study by Thiele et al. [241], a PDMS-based HFF device consisting
in the field of controlled release of bioactive molecules. of perpendicularly crossed microchannels was developed to focus an
However, despite polymersome stability being higher compared ethanolic block copolymer solution in a stream of water. Polymersomes
to that of lipid vesicles, the integrity of their membrane can be nucleated at the periphery of the focused stream, where the ethanol
compromised when exposed to external mechanical or chemical stimuli, concentration decreased to a critical level, below which the copolymer
and undesired release of the load may occur in this situation. This can became insoluble. Authors investigated the effect of flow rate ratio
seriously limit the range of their potential applications. In order to on polymersome characteristics. While at lower flow rate ratios
increase their stability and lower the permeability of vesicles membrane, polymersomes size was almost proportional to the flow rate ratio,
a novel strategy has been reported recently by Kim et al. [236] based on a implying good control over vesicles size distribution, a non-linear
two-step emulsification method achieved by modifying the design of the relation between polymersome size and flow rate ratio was instead
single-capillary device. In this way, water-in-oil-in-water-in-oil-in-water determined at the higher flow rate ratios. This may imply the
(W/O/W/O/W) quadruple emulsions were obtained and employed existence of different vesicle growth mechanisms, which merit further
to form polymersomes-in-polymersomes. Interestingly, the inner investigations. Overall, with this technique, the mean diameter of
vesicles were characterised by protrusions which extended through polymersomes could be controlled over a range from 40 nm to 2 μm,
the membrane bilayer of the outer vesicles, leading to the formation with narrow size distributions and high reproducibility.
of double bilayers by budding process. Authors demonstrated that In a study by Brown et al. [242], a PDMS-based HFF microdevice was
polymersomes having double bilayers were more stable and less developed for producing polymersomes from the pH-sensitive block
permeable compared to polymersomes which membrane was copolymer poly(2-(methacryloyloxy)ethyl phosphorylcholine)-poly(2-
composed of a single bilayer. (diisoprophylamino)ethyl methacrylate) (PMPC-b-PDPA). At the lower
Another challenge in the field of microfluidic-based production of pH (i.e. pH b 6) the polymer chain is completely dissolved in water,
polymersomes ― and vesicular systems in general ― is the capability while above a critical pH value (i.e. pH N 7.2) it becomes insoluble and
of scaling-up the production process, which is a necessary prerequisite self-assembles forming stable polymersomes. Thus, by exploiting the
towards the practical applicability of the developed microdevices. diffusion-governed species mass transfer between fluid streams in
With this respect, the use of devices fabricated from PDMS by soft a HFF microdevice, one can finely regulate the local pH levels within
lithography allows for a significant reduction of the fabrication costs the microenvironment and induce self-assembly of the copolymer.
and time, while preserving accuracy and flexibility [237]. In a study by Authors thoroughly investigated the combined effect of flow rate
Thiele et al. [238] a PDMS microdevice was developed for production ratio and pH of the aqueous buffer solution (lateral streams) on the pH
of double emulsion templates. The microdevice consisted of a multi- of the block copolymer solution (central stream) and the dimensional
1514 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Fig. 5. Microfluidic-based production of polymersomes. (a) Schematic depiction (left) and photograph (right) of a microfluidic device based on the parallelisation of glass microcapillary
devices, for high-throughput production of polymersomes from W/O/W emulsion templates. Formation of polydisperse emulsion droplets from a widening jet (a, top) and monodisperse
emulsion droplets from a narrowing jet (a, bottom) has been demonstrated with this microdevice. Confocal microscope image of monodisperse polymersomes obtained with this
microfluidic system is reported in (b) (taken from [239], with kind permission from Springer Science and Business Media). (c) Schematic illustration of polymersomes composed of a
bilayer of PEG-b-PLA and PNIPAM-b-PLGA diblock copolymers, where gold nanoparticles are doped into the hydrophobic part of the bilayer. (d–e) Microscope images showing distinct
release behaviours for polymersomes composed of a bilayer of PEG-b-PLA and PNIPAM-b-PLGA at 40 °C (e), and polymersomes having a gold nanoparticle-doped bilayer irradiated
with lasers at room temperature. The white arrow in (f) indicates the rupturing edge (taken with permission from [244]. Copyright 2012; John Wiley and Sons).

characteristics of the produced vesicles. Furthermore, continuous 4.2.3.3. Methods of actives release from polymersomes. A range of studies
flow encapsulation of proteins (bovine serum albumin, BSA) was have been devoted to the production of stimuli-sensitive polymersomes
demonstrated, with encapsulation efficiency comparable to conventional and the development of protocols to achieve programmable and
macroscale production methods. Despite it has lower encapsulation controlled release of actives encapsulated within the vesicles [245,246].
efficiency compared to other microfluidic-based methods (i.e. W/O/W However, only few studies have been devoted to the investigation of
emulsion templates), this technique has the advantage of not requiring actives controlled release from stimuli-sensitive polymersomes produced
organic solvents, which may be retained by the vesicle membrane after using microfluidic-based methods.
production and affect its performance. With this respect, a recent study by Amstad et al. [244]
The use of commercially-available micromixers to form poly- demonstrated microfluidic-based formation of polymersomes which
mersomes was recently reported in a study by Thiermann et al. membrane bilayer consisted of a mixture of inert and thermosensitive
[243]. Microdevices included stainless steel caterpillar (CPMM), amphiphiles (Fig. 5e–f) Furthermore, photothermal gold nanoparticles
superfocus (SFIMM) and interdigital slit (SIMM) mixers. Control were encapsulated in the hydrophobic portion of the shell. In this
over polymersome size and size distribution was achieved with way, the release of actives from the vesicle could be induced upon
using these devices, and differences between mixer architecture temperature increase above the lower critical solution temperature
resulted in a different performance (i.e. size of the obtained vesicles). (LCST) of the thermosensitive amphiphiles (Fig. 5e), or upon laser
In addition, due to the similar mixing times, CPMM and SIMM displayed illumination causing nanoparticle heating (Fig. 5f). Either reversible or
an analogous dependence of vesicles mean diameter on the fluidic irreversible membrane poration was demonstrated with this technique,
conditions. SFIMM system instead displayed a more significant suggesting the possibility to achieve control over the kinetics of the
variation of the size of the produced vesicles with varying the flow release process.
regime, and vesicles were generally smaller compared with those A study by Martino et al. [247] instead demonstrated the expression
obtained with CPMM and SIMM. This study demonstrated that of a membrane-related bacterial protein (MreB) within polymersomes
easy-to-use and commercially-available micromixers could be produced by single-capillary microfluidic technology [231], and the
employed to form polymersomes with a relatively narrow size subsequent triggered release of the protein by osmotic shock. For
distribution, opening the way for the implementation of microfluidic- this purpose, polymersomes were transferred into an ipo-osmotic
based continuous-flow techniques for the large-scale production of environment causing an osmotic pressure difference between the
polymersomes. inner core of the vesicle and the outer fluid, and thus a net flux of
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1515

fluid directed into the vesicle. For this reason, the membrane bilayer flow rate ratio. This was in agreement with previous studies [128]
was subjected to significant strain levels, causing the formation of and it was attributed to the reduced mixing time required to achieve
transmembrane pores through which the actives could be released complete mixing by diffusion, at the lower flow rate ratios. By
by diffusion. The same release method was adopted in a previous changing microreactor architecture, it was found that PM mean
study to investigate controlled release of fluorescent probes from diameter and polydispersity decreased with reducing the microchannel
within the vesicle [231]. dimension. This was further attributed to the improved mixing
Mechanically-induced rupture of polymersomes bilayer membrane efficiency within the microfluidic channels. Additionally, authors
was demonstrated by Kim et al. [235], by forcing the vesicles to flow observed that by carefully changing the flow rate ratio, two distinct
through an orifice with diameter smaller than the vesicle diameter. microreactors were capable of producing PMs with similar dimensional
With this technique, authors demonstrated sequential release of properties; however, at a fixed width of the focused stream the smaller
molecules from polymersomes-in-polymersomes structures. Notably, microchannels led to the formation of larger and more polydisperse
by accurate selection of the operating conditions, one can achieve PMs. These experimental findings thus demonstrated that microchannel
reversible or irreversible membrane permeabilisation. In the same dimensions, flow rate ratio and fluid physical properties contributed in
study, controlled release from polymersomes was also demonstrated concert to influence the nanoprecipitation process. In order to obtain
by means of a chemical route, simply transferring the vesicle within a further insights into the complex interplay between the process
non-aqueous fluidic environment which caused the dissociation of the governing parameters, computational fluid dynamics (CFD) simulations
lipid bilayer. were performed (Fig. 6a). Numerical results revealed the relation
between PM characteristics and the dynamic variations in fluid viscosity
and diffusivity, which affected the mass transfer between solvent and
4.2.4. Polymeric micelles non-solvent streams. Notably, by comparison with conventional batch
While the production of lipid-based nanoparticles and vesicles methods, this technique provided better control over PM size and size
in microscale devices represents a relatively established field of distribution (Fig. 6b).
investigation, only a limited number of studies have been reported Using a similar microfluidic strategy, the same group reported the
regarding the formation of polymeric micelles or copolymer nanoparticles encapsulation of the DNA-binding drug, mithramycin, in polymeric
(i.e. frozen core polymeric micelles) by using microfluidic-based micelles (PM-MTH) based on Pluronic® block copolymers [171]
technology. The large majority of these studies rely on diffusion- (Fig. 6c). The effect of a range of experimental parameters on PM-MTH
governed mixing processes to induce the self-assembly of block characteristics was systematically investigated, by means of a changing-
copolymers. A summary of the reviewed studies reporting microfluidic- one-separate-factor-a-time (COST) approach, followed by a more
based production of polymeric micelles is reported in Table 3. statistically sound Design of Experiment (DoE) approach. The factors
A study by Karnink and Gu [128] reported a microfluidic-based considered in the present study included: (i) size of the microchannels,
production of PLGA–PEG block copolymer nanoparticles (or PMs) by (ii) total flow rate, (iii) flow rate ratio between polymeric and aqueous
enhanced diffusion process in a PDMS-based HFF microdevice. According solutions, (iv) concentration of polymer, and (v) concentration of drug.
to the mechanism of self-assembly of block copolymer nanoparticles Results showed that empty PMs produced by microfluidic reactor
[173,248], the results showed a decrease in nanoparticle size and an had a mean radius of 52–61 nm, and that the flow rate ratio positively
increase in homogeneity with a decrease in mixing time. Notably, the correlated with PM size. Furthermore, different polymer initial
nanoparticles showed a decrease in size as the relative flow rate of the concentrations (in the range 7.5–5.0 mM) resulted in empty PMs with
polymer stream decreased and no break point was detected, even a similar dependence on the flow rate ratio; however, slightly larger
when the mixing time was considerably lower than the aggregation PMs were formed at higher polymer concentration (i.e. 15.0 mM). This
time to produce kinetically frozen polymeric micelles. The authors was attributed to the increase of the polymeric solution viscosity,
hypothesised that the results might be related to the time scale of self- which resulted in a slowing down of the mixing process. In contrast,
assembly of the nanoparticles. no significant change in PM polydispersity was observed at different
Microfluidic-based formation of PMs, constituted of Pluronic® tri- polymer concentrations. The effect of MTH encapsulation on PM
block copolymer, was also reported by Capretto et al. [170,249] characteristics was further investigated. In this regard, different
(Fig. 6a–b). For this purpose, a glass HFF continuous-flow microreactor concentrations of MTH (in the rage 10.0–55.0 μM) were considered
was fabricated by pholithography/wet etching procedure [250]. The (Fig. 6d). Due to MTH molecules amphiphilic properties, MTH
flow focusing configuration enabled controllable and fast mixing intercalated within the polymer chains into the core/shell interface,
process, which assisted the formation of PMs through nanoprecipitation leading to larger micelles compared to the empty counterparts.
governed by solvent exchange. Authors investigated the effect of However, the effect of a non-precipitating amphiphilic molecule
(i) process operational parameters, (ii) concentration of polymer, on PM precipitation process is still far from being fully understood,
and (iii) microchannel architecture, on the size and size distribution and it may represent the subject of future investigations. Relevant
of the produced PMs. For fixed microchannel architecture, PM size to application, it was shown that MTH preferentially partitioned in
and polydispersity were observed to decrease with decreasing the the micellar phase.

Table 3
Summary of the studies investigating microfluidic-based production of polymeric micelles. PLGA–PEG: poly(lactic-co-glycolic acid)–polyoxyethylene glycol block
copolymer. Poloxamer: Polyoxyethylene–polyoxypropylene block copolymer.

Self-assembly type Microfluidic strategy References

PLGA–PEG Planar flow focusing mixer Karnink and Gu et al. [128]


Poloxamer Planar flow focusing mixer Capretto et al. [170,251]
Poloxamer + mithramycin Planar flow focusing mixer Capretto et al. [171]
Poloxamer + β-carotene Planar flow focusing mixer Capretto et al. [1]
Poloxamer + Dexamethasone + ascorbyl-palmitate Planar flow focusing mixer Capretto et al. [172]
Polybenzimidazole Planar flow focusing mixer Hasani-Sadrabadi et al. [252]
PLGA–PEG 3D flow focusing mixer Rhee et al. [129]
1516 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Fig. 6. Microfluidic-based production of polymeric micelles and nanoparticles. (a) CFD simulation of species mass transfer in a HFF glass microfluidic device employed to produce block
copolymer micelles. Computational mass fraction contours are reported at three longitudinal positions along the microfluidic device, corresponding to the channels junction, middle
and end of the main mixing channel. PMs size distribution at a polymer concentration of 7.5 × 10−3 M is reported in (b), where the performance of batch reactor is compared with the
one of different HFF microfluidic devices having channel cross-sectional area of 5684 μm2 (Microreactor #1), 1959 μm2 (Microreactor #2) and 844 μm2 (Microreactor #3) (reprinted
from [170], Copyright 2011, with permission from Elsevier). (c) Schematic representation of the preparation procedure for mythramycin encapsulating polymeric micelles by HFF. The
central stream of Pluronic® F127 unimers and mithramycin dissolved in dimethyl sulfoxide is focused by two lateral streams of water. The effect of preparation procedure (i.e. batch
vs. microfluidics) and mithramycin concentration on micelles mean size is reported in (d) (republished with permission of DOVE Medical Press, from [171], Copyright 2012; permission
conveyed through Copyright Clearance Center, Inc.). (e) Atomic Force Microscopy (AFM) images of either empty or β-carotene encapsulating block copolymer micelles produced by HFF.
Size distribution of hybrid NPs (dashed line, solid circles) and empty PMs (dashed line, open circles) at MR = 1.41 is reported (f). The grey solid-line curve represents the sum of the values
of the calculated hybrid NPs and unfilled PMs, while the open squares correspond to the experimental data (reprinted with permission from [1]). (g–i) 3D HFF microfluidic device designed
for the production of polymeric NPs. Device consisted of three sequential inlets for vertical focusing, and a separate inlet for lateral focusing (g). CFD simulations were performed to
optimise device functioning and achieve the desired focusing performance (h). TEM images of produced NPs are shown in (i), at different polymer molecular weight and concentration
(scale bar: 100 nm) (reprinted with permission from [129]. Copyright 2011; John Wiley and Sons).

The same group investigated the production of organic NPs by PMs and drug loaded stabilised nanoparticles. Understanding the
HFF, where core–shell structured hybrid NPs containing synthetic implication of this parameter on the nanoprecipitation dynamics may
β-carotene were encapsulated by amphiphilic block copolymer help the selection of stabiliser type, concentration, and process
[1] (Fig. 6e–f). A glass microfluidic device, with minor adaptations parameters to obtain stable NPs with high percentage of loading while
with respect to previously mentioned studies [170], was employed reducing the formation of “empty” (i.e. unloaded) PMs. To conclude,
for this purpose to reduce drug deposition on channel walls. NPs size results from this study could be potentially translated to investigate a
and the kinetics of the nanoprecipitation process were investigated variety of similar drug/polymer combinations.
under a range of different fluidic conditions, β-carotene and polymer The same group reported co-encapsulation of dexamethasone
concentrations, and β-carotene/polymer molecular ratios (MRs). (Dex) and ascorbyl-palmitate (AP) in polymeric micelles based on
Authors demonstrated that competitive reactions were responsible for Pluronic® block-copolymers [251], by using the microfluidic-based
the formation of two distinct types of NP, either with or without loading HFF configuration previously mentioned [170]. The effect of the
organic β-carotene (Fig. 6f). The corresponding peak area of NPs size amount of AP and flow rate ratio on the characteristics of the formed
distribution provided a quantitative indication of the amount of each PMs was investigated. It was shown that increasing the flow rate
type of NP, as a function of β-carotene/polymer MR. Furthermore, it ratio from 0.03 to 0.13, micelle mean diameter increased from
has been demonstrated that increasing β-carotene/polymer MR ~ 207 nm to ~ 1484 nm, which was coherent with results from
resulted in higher mean particle diameter and wider size distribution. previous studies [128,170]. CFD simulations were implemented to
In order to reveal the effect of the complex interplay between molecular further investigate the effect of the fluidic conditions (i.e. fluid
and fluid dynamic phenomena on the nanoprecipitation process, CFD viscosity, shape of the focused stream) on species mass transfer.
simulations were performed and integrated with an in-house Importantly, as for previous investigations, the microfluidic approach
developed code that modelled the effect of NP formation on the resulted in PMs with superior dimensional quality compared to those
kinetics of NP diffusion and advection. Based on the numerical produced by batch reactor. Furthermore, by increasing the concentration
simulations, a new parameter (nanoprecipitation mismatching) of AP (i.e. from 0.125 mM to 1.000 mM) resulted in a severe increase of
representing the mismatching between the aggregations of the two PM mean diameter (i.e. from 6 nm to 207 nm). A non-linear dependence
precipitant species was introduced. The variation of this parameter of PM size on AP concentration was identified, which was attributed to
has been shown to affect the relative amount of “empty” (i.e. unloaded) the existence of a maximum micellar solubilisation capacity above
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1517

which the excess of drug cannot be solubilised in PM core. The drug in of the molecular weight of the polymer products, which cannot be
excess is thus exposed to a solvent which is prone to precipitate in the quantified via DLS.
form of crystalline suspension causing the observed increase in PM size.
In addition to the aforementioned applications of polymeric 4.2.5. Polymer nanoparticles
nanoparticles as carriers of bioactive molecules, microfluidic- A summary of the reviewed studies reporting microfluidic-based
based production of NPs for fuel cell technology has been recently production of polymeric nanoparticles is reported in Table 4.
demonstrated by Hasani-Sadrabadi et al. [252]. In this study, an An innovative approach to the production of polymer NPs in a
HFF microdevice was employed to achieve control over the size microfluidic system has been recently reported by Anton et al. [253],
and size distribution of organic NPs based on polybenzimidazole. based on an impact-jet micromixer [254]. The micromixer was
As for the formation of lipid vesicles, the vast majority of the HFF constituted of two main components, (i) the mixing plate, in which
microfluidic devices rely on a two-dimensional focusing of the microchannels converged towards a central hole, and (ii) the outlet
central fluid stream. However, production of PLGA–PEG and PLGA plate, which contained an outlet channel where the nanoprecipitation
polymeric nanoparticles by 3D hydrodynamic flow focusing has occurred. With this technique, NP characteristics can be controlled by
been recently reported by Rhee et al. [129] (Fig. 6g–i). The main varying (i) the flow rates of solvent and non-solvent, (ii) the polymer
advantage of three-dimensional focusing is the hydrodynamically- concentration in the organic solvent, and (iii) the presence of
induced isolation of the precipitating polymer from the channel surfactants as stabilising agents. Polymeric (PMMA) NPs, with a mean
inner walls, in both the horizontal and the vertical dimensions, diameter of ~100 nm and a narrow size distribution, were produced
thereby minimising the occurrence of fouling and channel clogging. with this method, and loaded with a lipophilic drug (ketoprofen).
However, the fabrication of 3D HFF devices is generally non-trivial Notably, this setup may open new perspectives for large-scale
and suffers from limited reproducibility. In this study, a simple design production of NPs, given (i) the possibility for online modification of
for a 3D HFF device was described, consisting of three sequential inlets the NP size, (ii) control of the NP loading and monitoring of the
for vertical focusing and a further inlet for lateral focusing (Fig. 6g). formulation, and (iii) higher polymer solution flow rates (i.e. up to
Authors fabricated a 2D HFF device for direct comparison between the 1 mL/min) compared to other microfluidic-based techniques.
two techniques. Results showed that NPs produced by 3D HFF had
smaller sizes and improved monodispersity compared to 2D HFF or 4.2.6. Hybrid nanoparticles and complexes
bulk production. The monodispersity of NPs produced by 3D HFF A summary of the reviewed studies reporting microfluidic-based
enabled to finely tune NPs size, by varying the polymer block size and production of hybrid NPs and complexes is reported in Table 4.
concentration. Mathematical and numerical simulations, together A range of studies have recently focused on the production of hybrid
with confocal microscopy acquisitions (Fig. 6i), were performed in nanoparticles by means of microfluidic-based strategies. The physico-
order to optimise device design and identify the optimal operating chemical and structural peculiarities of these formulations make them
conditions. This simple design may open the way for an easy-to- suitable for a range of biomedical applications, including drug and
fabricate continuous flow microfluidic device, to be potentially gene delivery, and medical imaging.
employed for large-scale production of polymeric NPs and capable Lipid-polymeric hybrid NPs combine the unique characteristics of
of overcoming limitations associated with 2D HFF devices. polymeric NPs and liposomes, such as high encapsulation yield of
Remarkable efforts towards the design of a robust platform for high- bioactive molecules, high stability in blood and slow release kinetics.
throughput production of PMs and in situ monitoring of the process of Moreover, temporally controlled release of different agents from
nanoparticle formation were made by Chastek and Iida [147]. The these NPs has been achieved, by encapsulating one agent in the
microfluidic channels had square cross-section and were machined lipid layer and the other agent within the polymeric core [255].
into an aluminium plate and then sealed with using a thin film of This type of NPs generally comprises three functional components
transparent Kapton. Micelles were obtained by dilution of the block [256]: (i) a hydrophobic polymer core for the encapsulation of
copolymers with a selective solvent, which caused the copolymer to poorly water-soluble molecules; (ii) a lipid layer surrounding the
self-assemble. A micro-well with a miniaturised stir bar was fabricated polymer core, providing a biocompatible shell and promoting retention
to achieve good mixing of the reagents. A dynamic light scattering of encapsulated molecules; and (iii) a hydrophobic polymer sheath
(DLS) unit was integrated with the microfluidic platform, with laser layer to increase NP stability and lifetime in the haematic circulation.
and detector fibre optic probes in direct contact with the fluid samples. Conventional batch methods for producing hybrid NPs are complex
With this technique, authors investigated the temporal evolution and they usually require a two-step formulation process, consisting of
and the mechanisms of PM formation. Notably, they were able to production of polymeric NPs first, followed by encapsulation of
discriminate between the formation of discrete micelles or micelle polymeric NPs within liposomes. Intermediate steps, such as heating,
aggregates, depending on the length of the corona block. A future vortexing and long incubation periods introduce significant variability
development of this platform may integrate also on-line measurements in the characteristics of the formed NPs. In order to overcome this

Table 4
Summary of the studies investigating microfluidic-based production of polymer NPs, hybrid NPs and complexes. PMMA: Poly(methyl methacrylate); LNPs: lipid
nanoparticles; SHM: staggered herringbone micromixer; ODN-LNPs: oligodeoxynucleotide lipid nanoparticles.

Nanomaterial type Microfluidic strategy References

Polymer nanoparticles
PMMA + ketoprofen Impact-jet micromixer Anton et al. [253]
Hybrid nanoparticles and complexes
Lipid–polymeric hybrid nanoparticles Tesla mixer Valencia et al. [255]
Lipid–polymeric hybrid nanoparticles Microvortice mixer Kim et al. [257]
DNA polyplexes T-shaped mixer Ho et al. [264,265]
pDNA polyplexes Various micromixer Debus et al. [261]
siRNA-LNPs SHM mixer Chen et al. [258]
siRNA-LNPs SHM mixer Belliveau et al. [259]
siRNA-LNPs SHM mixer Leung et al. [267]
ODN-LNPs 2-steps planar flow focusing mixer Koh et al. [268]
1518 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

limitation, Valencia et al. [255] developed a PDMS-based microfluidic three-inlet architecture was employed and microvortices were formed
mixer consisting of Tesla structures [260] (Fig. 7a), and operating at at the intersection between the inlets (Fig. 7b), resulting in rapid mixing
Re N 20. NPs produced with this technique were monodisperse and between the polymer (PLGA) and the lipid (lechitin + DSPE-PEG)
with a mean diameter of 40 nm. The rapid convective mixing obtained (Fig. 7c). Importantly, the flow pattern downstream from the intersection
within this microdevice allowed for uniform lipid coverage around the displayed three-dimensional focusing features (Fig. 7c), preventing
polymeric core to be achieved, and the self-assembly of the polymeric polymer aggregation near the channel walls. NPs produced with this
core occurred independently from the lipid component, which acted method were generally monodisperse and small in size (i.e. diameter
as a stabiliser of the overall structure providing long-term stability. range: 55–80 nm), particularly at the higher Reynolds numbers
Conventional production methods (i.e. diffusion-based) instead (i.e. Re N 30) at which rapid and effective convecting mixing occurred,
result in the formation of a combination of liposomes, hybrid NPs as demonstrated by CFD simulations. Furthermore, NP size changes
and polymeric NPs; as a consequence of the slow-mixing kinetics. were more sensitive to the fluidic conditions (i.e. increasing Re resulted
With using the same microfluidic device, authors demonstrated the in a decrease of NP size) than the polymer-lipid composition, which was
production of hybrid lipid-quantum dot (QD) nanoparticles, to be indicative of the role played by the mixing velocity (Fig. 7c). Relevant
potentially employed for biomedical imaging applications (Fig. 7a). to applications, this technique allows for high NP productivity, which
Notably, QDs are semi-conductive nanocrystals with excellent optical is significantly greater than previous methods based on diffusive or
properties; however, hydrophobicity and low colloidal stability make convective mixing [128,255].
them unsuitable candidates for clinical usage. To address this issue,
lipid coating of QDs has been achieved in this study, in order to improve 4.2.7. Nanomaterial produced through complexation mechanism
their stability and biocompatibility. Hybrid lipid-QD NPs produced with Polymer complexes, referred to as polyplexes, have been also
this technique were homogeneous and monodisperse (mean diameter: produced by microfluidic technology. They are formed by electrostatic
~60 nm) (Fig. 7a), thus opening the way for their potential application interaction between negatively charged pDNA and positively charged
in the field of nanomedicine. However, despite the production polymers (i.e. PEI) [261], and have been established as a non-viral
capabilities of this method are significantly greater than those of vector for intra-cellular delivery of a range of different nucleic acids.
strategies relying on diffusive mass transfer only, the fluid flow Bulk production of polyplexes generally relies on pipetting, which
rates employed here (i.e. ~ 50 μL/min) may still limit its translation leads to poor control over NP characteristics [262] and is generally
into large-scale production systems. restricted to relatively low concentration of polyplexes in solution,
Recently, a novel microfluidic-based strategy has been developed limiting their application in the clinic [263]. Furthermore, depending
by Kim et al. [257] for efficient producing hybrid lipid-polymer NPs on the type of polymer employed and its molecular weight, the
(Fig. 7b–c). The method does not rely on diffusive mixing between characteristics of the formed NPs can vary significantly. In order to
species, but on the formation of microvortices (Fig. 7b) at relatively achieve a more accurate control over NPs properties, formation of
high Reynolds number (Re ~ 75). Notably, diffusive mixing does not polyplexes within microfluidic systems has been investigated in
allow for the formation of NPs requiring the assembly of precursors recent studies. In this regard, a study by Ho et al. [264,265] reported
in the aqueous phase (i.e. lipids) with precursors in the organic phase the formation of DNA polyplexes by using a PDMS-based microdevice
(i.e. polymer), due to the associated slow kinetics. In this study, a consisting of a T-junction. By combining quantum-dot-mediated

Fig. 7. Microfluidic-based production of hybrid NPs and complexes. (a) Schematic of a microdevice employed for producing hybrid lipid–polymeric and lipid–quantum dot NPs. Rapid
mixing is achieved by means of Tesla structures within the microfluidic domain. Representative TEM images of the obtained lipid–polymer (top) and lipid–quantum dots (bottom)
NPs are reported (reprinted with permission from [255]. Copyright 2010 American Chemical Society). (b–c) Schematic and cross-section views of a three-inlet microfluidic device
generating two symmetric microvortices and a 3D HFF pattern, employed for rapid and high-throughput production of lipid–polymer hybrid NPs. CFD simulations in (c) illustrate the
mass transfer process within the microfluidic environment. TEM images of NPs produced at Re = 75 (top) and Re = 150 (bottom) demonstrate control over NPs size. Scale bars =
100 nm (reprinted with permission from [257]. Copyright 2012 American Chemical Society). (d) Three-inlets microfluidic platform for producing siRNA–lipid NPs. Rapid mixing is
achieved within a microfluidic channel due to groove structures located on the channel bottom wall (e) (reprinted with permission from [258]. Copyright 2012 American Chemical
Society). (f) Schematic depiction of a siRNA–lipid NPs formulation strategy based on a staggered herringbone micromixer (SHM) which induces chaotic advection of the laminar inlet
streams. Effective mixing is achieved only few milliseconds after solutions are injected within the device (reprinted by permission from Macmillan Publishers Ltd on behalf of Cancer
Research UK: [259]. Copyright 2012).
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1519

fluorescence resonance energy transfer (QD-FRET) and microfluidic value of 0.2 mL/min from all inlets. Based on this observation, authors
technology, authors were able to characterise the kinetics of chitosan/ investigated simultaneous manual injection of the solutions (at an
DNA polyplexes assembly, with high temporal resolution. For this estimated flow rate N0.3 mL/min) and were capable of producing
purpose, a FRET pair of 605QD (donor) and Cy5 (acceptor) were reproducible particle sizes. In this way, a reliable “equipment-free”
labelled to pDNA and chitosan, respectively; and their molecular method of particle formation was established, representing a user-
interaction was monitored on-time via FRET signals. Notably, two friendly and cost-effective alternative to other laborious production
different stages of interaction were detected during the self-assembly strategies.
process: (i) a ‘fast cationic binding’ occurring in the millisecond range Rapid mixing within a microfluidic device has been exploited by
and dominated by electrostatic interactions, and (ii) a ‘nano-assembly Belliveau et al. [259] and Leung et al. [267] to form monodisperse
flocculation’ stage (whose effect was detected at 1 s), which is a typical siRNA-LNPs (Fig. 7f). In these studies, a staggered herringbone
feature of diffusion–reaction limited processes. This study represents micromixer was developed which allowed for effective mixing to
the first attempt to monitor the kinetics of DNA polyplexes in situ by be achieved, at relatively high fluid flow rates suitable for large-
using a lab-on-a-chip platform; and the developed technique could be scale production (i.e. ~ 2 mL/min). Authors employed a variety of
potentially employed to investigate the formation mechanisms of a techniques and bio-physical assays to characterise the structure of
wide range of lypoplex and polyplex systems. formed lypoplexes, including cryo-TEM, 31P NMR, membrane fusion
Recent studies have been devoted to the development of more assays, density gradient measurements and molecular modelling
elaborated microfluidic-based platforms for the production of [267]. Results from this study revealed that siRNA-LNPs had an
pDNA polyplexes. For example, in a work by Debus et al. [261] a electron-dense inner lipid core which contained siRNA duplexes
range of different chips with distinct architectures and mixing complexed to internalised cationic lipids, resulting in siRNA being
lengths were fitted onto a baseboard. Polyplexes with PEI and fully protected from external RNase. Furthermore, the interior of
pDNA were produced within the microfluidic environment, under these NPs exhibited a periodic structure of aqueous compartments,
well-controlled mixing conditions, and compared with those some of which contained siRNA duplexes. Results from this study
produced by traditional pipetting procedure. Results from this study have provided new insights into the mechanisms of formation of
further demonstrated the possibility of using microfluidic technology siRNA-LNPs, to be exploited for the development of novel and more
for producing DNA polyplexes. Importantly, it was demonstrated that sophisticated complexes.
the ratio of PEI to DNA had a significant influence on the properties of Another type of lypoplex formulation which has been
the formed NPs, while other operating conditions were less important. produced by means of microfluidic technology is represented by
Relevant to application, the size of polyplexes prepared with this oligodeoxyribonucleotide (ODN)-LNP systems. The use of ODNs to
technique was observed to be almost constant, in the range down-regulate oncogenes has been identified as a potentially useful
140–160 nm, over a relatively wide range of complex concentrations. anticancer treatment; however, its application in the clinic has been
In contrast, the size of polyplexes prepared by bulk method was limited by its rapid clearance in the haematic circulation and the low
observed to be more variable. Notably, authors expanded the range permeability through the cell membrane bilayer. Thus, complexation
of potential applications by employing the developed lab-on-a-chip of ODNs with cationic liposomes has been recently proposed as an
platform to form siRNA and mRNA polyplexes, thus opening the alternative strategy to overcome these limitations. In a study by Koh
way for a range of further potential applications of microfluidics in et al. [268], the production of ODN-LNPs using a microfluidic device
gene therapy. has been reported for the first time. For this purpose, a five-inlet flow
Another type of complexation NPs produced by microfluidic focusing architecture was designed and fabricated by micromilling of
technology is represented by lipid nanoparticles–nucleic acids PMMA layers. A two-step HFF was achieved within the microdevice,
complexes (or lypoplexes) (LNP). Recently, considerable attention where the first step consisted of an ODN solution focused by two
has been devoted to the formulation of LNPs encapsulating small streams of protamine solution. The obtained flow stream was then
interfering RNA (siRNA), given the therapeutic potential of these focused by two streams of lipid solution in the second step. Using this
formulations [258,266,267]. Conventional methods for producing technique authors demonstrated continuous-flow production of ODN-
LNPs ― such as extrusion ― generally suffer from the need for LNPs with a lower size and a narrower size distribution compared
labour-intensive multiple steps, which are sometime associated with to bulk methods (i.e. vortexing). Furthermore, zeta-potential of NPs
significant loss of expensive chemicals (i.e. siRNA). Other methods, produced by microfluidics was slightly lower compared to the ones
such as sonication, may instead damage the structure of nucleic acids, obtained by bulk method. ODN encapsulation efficiency between
thus compromising the therapeutic efficacy of the product. In order to the two methods was comparable (N70%), with the microfluidic-
achieve precise control over the operating conditions, microfluidic- based strategy resulting in slightly better encapsulation performance
based strategies have been recently implemented. For example, in a compared to bulk production. Regarding NP morphology, it was
recent study by Chen et al. [258], a PDMS-based microfluidic mixer revealed that complexes produced by microfluidics had more multi-
was developed for producing siRNA-LNPs (Fig. 7d–e). The microfluidic lamellar vesicles and onion-like structures and a reduced number
device possessed three inlets for sequential injection of (i) siRNA of large aggregates, compared to bulk technique. However, limitations
aqueous solution, (ii) alcohol solution of cationic lipids and (iii) buffer of this method include the need for off-chip procedures to be performed
solution to prevent inter-particle aggregation (Fig. 7d). Effective and (i.e. dialysis and filtration), and the wider NP size distribution compared
rapid chaotic mixing was achieved thanks to periodic trenches and to methods based on faster mixing [258,259].
ridges which were asymmetrically arranged on the channel bottom
surface [104]. CryoTEM acquisitions revealed that LNPs produced by
microfluidics had a narrow size distribution, with a diameter in the 5. Microfluidic-based manipulation, immobilisation and in situ
range 60–90 nm (at a flow rate of 0.3 mL/min); while LNPs produced analysis of nanoparticles and vesicular systems
by bulk methods (i.e. pipet mixing) were polydisperse, with a diameter
N100 nm. Importantly, microfluidic-produced LNPs were capable of In situ analysis of nanomaterial and vesicle characteristics is of
encapsulating siRNA with an efficiency of ~80%. Increasing the flow crucial importance to evaluate the efficacy and reliability of a given
rate ratio resulted in the formation of smaller NPs, which is coherent production technique (i.e. quality control). However, manipulation
with the theory of vesicle formation and species mixing behaviour. and immobilisation of nanoscale materials are non-trivial, and
However, either NP diameter and encapsulation efficiency remained may require the development of advanced and laborious-to-
approximately unchanged with increasing the flow rate above a critical operate technological platforms.
1520 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

With the emerging of microfluidic-based technology and the related bilayer membrane for investigating protein reconstitution mechanisms.
potential to operate tiny fluid volumes, together with the possibility Furthermore, in a study by Vijayan et al. [283], the manipulation of
to integrate different functions on a single lab-on-a-chip platform, worm micelles using electric fields generated within a microsystem
advances have been made towards the in situ manipulation and analysis has been reported. The easy-to-fabricate experimental chamber
of nanomaterials. Notably, high temporal and spatial resolution analysis was formed between a glass slide and a coverslip, with sealing at
of nanomaterial properties may provide innovative insights into the the four edges. Electrodes were drawn onto the glass slide to
mechanisms of nanoparticles and vesicle formation and, thus, allow generate oscillating electric fields. Controlled displacement and
for the optimisation of current production techniques. The lack of stiffening of the suspended micelles by means of the electric field
quality controls on the produced nanoparticles is indeed limiting the was demonstrated. Recently, lab-on-a-chip systems were developed
translation of production techniques into the clinical or industrial in order to achieve high-throughput nanoparticles manipulation by
routine [269]. DEPs, by means of an array of microelectrodes integrated with a
In the present paragraph, a brief overview on microfluidic-based microfluidic environment [284]. These developments represent an
methodologies for nanoparticles and vesicular systems manipulation, important step towards the miniaturisation of automated biochemical
immobilisation and analysis are reported. analyses, to be potentially employed in diagnostic applications.
Despite DEPs representing an effective strategy for manipulating
5.1. Nanoparticles and vesicular systems manipulation nanomaterials, attention must be devoted to the effect of the applied
electric field on vesicle/nanoparticle integrity, particularly for applications
The use of electric fields as a means to manipulate, sort or separate which require the encapsulation of bio-actives. Notably, local heating and
micro- and nano-materials has been extensively studied, and is usually fluid convection may originate from electric field application, potentially
referred to as dielectrophoresis (DEP) [270]. With this technique, causing undesired alterations of nanoparticle properties.
dielectric nanoparticles suspended in a fluid medium are exposed to a In recent years, the potential of integrating optical and microfluidic
non-uniform electric field. Thus, a net force is generated on the particle, elements into a single platform has been investigated, particularly with
due the interaction of the particle's dipole with the spatial gradient of the aim of improving fluid and particle manipulation [285] (Fig. 8c).
the electric field [271]. Differences in the magnitude and direction The latter can be achieved by means of optical tweezers, or by indirect
of this force can be exploited for a range of applications, including optically-induced microfluidic effects. In a study by Kulin et al. [286] the
NP separation, concentration and size-fractionation [272]. Furthermore, potential of optical tweezers to manipulate liposomes has been reported,
Green and Morgan [273] demonstrated that DEP behaviour of sub- with the aim of facilitating liposomes fusion by UV exposure (Fig. 8d).
micrometric particles is significantly influenced by the surface Importantly, liposomes were observed not to be disrupted by the strain
properties of the particle, thus allowing for detection and quantification levels achieved during optical trapping and manipulation. However,
of changes in particle surface composition. Notably, miniaturised while it is particularly suitable for single-vesicle analysis applications,
electrodes can be incorporated in microfluidic platforms for the this technique suffers from the low number of processed vesicles
generation of electric fields within tiny fluid volumes [274,275]; compared to other continuous-flow strategies, thus limiting its
opening the way for microfluidic-based DEP [271]. This technique spectrum of potential applications. A continuous-flow optofluidic
has been demonstrated to be capable of effectively manipulating strategy was instead developed by Yin et al. [287]. The microdevice
nanoscale model particles (i.e. latex beads) [275] ― down to a diameter relied on planar fluidic and optical pathways for nanoparticle
of 40nm [276] ― carbon nanotubes [277,278], DNA nanoparticles [279] manipulation and detection, and it could be potentially interfaced
(Fig. 8a–b), silica nanoparticles [280] and vesicles [281]. With this with other well-established analytical techniques. Furthermore, due to
respect, sorting of liposomes using microfluidic-DEP has been its planar geometry, this platform could be scaled up to a larger number
demonstrated by Cheng et al. [282]. In this work, a broadly distributed of intersecting waveguides.
liposome sample was sorted into different size ranges, at a relatively The use of magneto-hydrodynamic (MHD) forces to manipulate
high fluid velocity, by means of a DEP microsystem. In addition, phospholiposomes within microfluidic channels has been reported
the negative DEP force employed in this study prevented undesired in a study by Son et al. [288]. Soft lithography was employed to
attachment of nanoparticles to the channel walls. Accurate manipulation fabricate a network of PDMS microchannels, which was bonded to
of liposome position by DEP was achieved by Osaki et al. [281], in a a glass slide by O2 plasma treatment. Authors demonstrated active
PMMA–glass microfluidic system. With this method, authors were control of liposomes trajectory by application of an external magnetic
capable of accurately directing the liposomes towards a planar field (i.e. possibility to direct the vesicles in selected outlet channels).

Fig. 8. On-chip nanoparticles manipulation. (a) Magnified view of a microelectrode array device employed to manipulate NPs by dielectrophoresis (DEP). Upon electric field application,
200 nm diameter NPs can be concentrated into the positive DEP high field regions, as illustrated in (b) (adapted with permission from [279]. Copyright 2008; John Wiley and Sons).
(c) Schematic illustration of the mechanism of transport of two different sizes of NPs by optical manipulation in a slot waveguide. The force Fprop corresponds to the radiation pressure
force responsible for optofluidic transport, while Ftrap is the trapping force holding NPs within the slot region (reprinted by permission from Macmillan Publishers Ltd on behalf of Cancer
Research UK: [285]. Copyright 2012). (d) Optically-driven fusion of two liposomes, one containing fluo-3 dye and the other containing calcium ions. Bright field (left) and fluorescence
(right) microscope images have been recorded. After fusion, the fluorescence intensity increases due to the reaction in which fluo-3 chelates the calcium ions (reprinted with permission
from [286]. Copyright 2003 American Chemical Society).
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1521

Furthermore, they identified two factors influencing liposomes flow, achieve higher control over nanoparticles properties. However, the
namely, the charge density on liposome surface and the Lorentz force large majority of liposomes formulations are still performed using
generated by the transverse field. However, only nanoparticles with conventional macroscale methods, resulting in generally poor control
sufficiently high charge could be effectively manipulated with this over the product quality. In order to facilitate product development, a
technique, posing significant limitations to its widespread applicability. novel microfluidic-based device has been developed by Birnbaumer
On-chip mechanical methods to fractionate nanoparticles based on et al. [269], which contained embedded dielectric microsensors capable
their physical properties (i.e. size) have been implemented in a study of quantifying liposomes formulation stability and composition from
by Nomura et al. [289]. The microfluidic device consisted of a series of measurement of their electric properties. With using this platform,
micropillars which were capable of trapping vesicles based on their authors demonstrated that different liposomes membrane modifications,
diameter. Compared to the methodologies previously mentioned, this including membrane-bound surface proteins, lipid-bilayer soluble drugs,
technique did not require the application of external physical stimuli, protein and dye encapsulation, could be detected and discriminated. The
and it can be regarded as a gentle manipulation approach. developed lab-on-a-chip device represents a significant advancement
toward efficient screening of nanoparticles quality, thus opening the
5.2. Nanoparticles and vesicular systems immobilisation and in situ way for its potential application in industrial production systems.
analysis
6. Effect of microfluidic environment on dynamics of
Recently, few studies have been reported about the implementation nanomaterial formation
of microfluidic strategies for trapping or immobilising nanomaterials,
in order to facilitate the analysis of their physical and chemical As already explained in the previous sections, a controlled formation
characteristics in situ, by coupling with microscope-based and analytical of nanomaterial generally relies on an efficient solvent-displacement
instrumentation. Among these, a hydrodynamic trapping technique has process. This is realised by a fast solvent shifting, from organic to non-
been developed by Nuss et al. [290], by using a PDMS microfluidic device. solvent that triggers the precipitation process.
The microdevice consisted of a series of hydrodynamic traps made of An estimation of the velocity with which this process proceeds can
identical junctions, allowing for highly-efficient trapping performance be obtained considering the mixing time, τmix. It has been recognised
of vesicles with mean diameter in the range 2–40 μm. In situ real-time by different authors as a key parameter for the formation of NPs through
analysis of vesicles behaviour and physical properties (i.e. permeability) precipitation, since it directly affects the dimensional properties of the
in response to changes of the fluidic environment characteristics could produced nanomaterials [173,294]. A shorter mixing time can in fact
be performed with this technique. ensure a spatially homogeneous environment during the precipitation
Recently, Khun et al. [291] presented a novel chemical method to process which in turn leads to uniform reaction condition and particle
achieve long-term immobilisation of liposomes on a substrate. In this formation kinetics.
technique, cholesterol–PEG–biotin was used as a linker, which Despite the processes for the production of organic nanomaterial
binded via avidin to biotinylated BSA (bBSA) previously absorbed are well established, there is still a general lack of fundamental
on a glass surface. The liposomes solution was flowed over the understanding of the mechanism of nanomaterial formation
functionalised substrate within a microfluidic channel. It has been [118,143,176,295–297]. This particularly applies to the production of
demonstrated successful immobilisation of liposomes having nanomaterials within microscale reactors, where the microfluidic
different size, which remained stable for prolonged time and at environment and associated characteristics (i.e. flow rate ratio, channel
relatively high volumetric flow rates. This experimental method is architecture and species residence time) have been demonstrated to
particularly suited for long-term investigations where fluids need to be significantly affect nanomaterial output [176,255].
rapidly changed, for measuring fast kinetics and to study the effect of An effective way to achieve controllable, predictable and fast mixing
fluid shear stress on liposome characteristics. Notably, the immobilised is the use of microreactors in which the architecture of the fluidic
vesicles could be released on-demand after the analysis. By coupling domain allows the generation of a hydrodynamic focusing flow pattern.
with microscope-based technology, this method may allow for in situ The control over the mixing process by varying the width of the focused
analysis of the temporal dynamics of membrane physical properties, stream, wf, provided a direct means to control NPs characteristics, as
including the spatial distribution of lipid domains (i.e. cholesterol-rich demonstrated previously [126,169]. The theoretical mixing time for
lipid rafts), membrane deformation and fluidity. This may be of help to flow focusing reactors (τmix) was estimated as
further explore the potential of lipid vesicles to replicate certain
characteristics of cell membrane dynamic behaviour under biomimetic w2f
conditions [292,293]. τmix ≈ ð10Þ
4D
Advances in the field of nanoparticle manipulation and analysis on-
a-chip have been recently reported by Issadore et al. [284]. The authors where D represents the diffusion coefficient for the solvent. However,
developed a hybrid integrated circuit/microfluidic chip capable of this estimation represents only a largely simplified model; and many
performing programmed manipulation, trapping and poration of lipid other factors should be taken into account to understand the effect of
vesicles, by means of electric fields. An array of micro-electrodes, driven microfluidic environment on nanomaterial synthesis.
by a radio frequency (RF) voltage, allowed for a range of operations to For instance, Capretto et al. [170] have demonstrated that the
be performed efficiently. In particular, the chip could operate in two dimensional properties of polymeric micelles are determined not only
distinct modes, (i) voltages at MHz frequencies were used to trap and by the flow rate ratio, as largely understood, but also by microrector
deform vesicles by DEP, and (ii) frequencies b 1 kHz were used to geometry and fluid physical properties (i.e. viscosity), all of which act
electroporate or fuse the lipid bilayers. By making use of low-cost in concert to govern the mixing process and thus nanomaterial output.
integrated circuit technology, these microdevices could be employed Similar findings were reported by Jahn et al. [176] for the production of
in the field of point-of-care diagnostics and biochemical detections of lipid vesicles. Authors examined the role played by the total flow rate in
analytes. the fine tuning of vesicle dimension under certain focusing regimes.
One of the greatest limitations to the commercialisation of lipid Karnik et al. [128] have demonstrated a consistent decrease in NPs
vesicles is the lack of formulation stability, since small size variations or size as the mixing time decreased far below the estimated aggregation
changes in the surface chemistry can significantly compromise time for block copolymer NPs, which was putatively associated with
their performance in vivo. As described previously, microfluidic-based rearrangements of block copolymer during the growth process. These
production strategies have been developed in recent years in order to observations were in contrast with previous results by Johnson et al.
1522 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

[173] with CIJ mixers, where a minimal critical NP size was associated the mixing interface. In this regard, Jahn et al. demonstrated with the
with a breakpoint mixing time. Furthermore, they showed an effect of aid of a CFD model that total flow rate could modulate liposome size
species mixing time on the surface properties of NPs, as a result of the distribution [176]. Experimental results and CFD models [176,300]
smaller portion of hydrophilic block buried in nascent NPs. These have shown that two different regions can be identified in a focused
observations indicated that the mechanism of nanomaterial production stream, each characterised by different mixing performance. The first
in a microfluidic environment might differ from that postulated for CIJ region is located at the channel junction, in the arrowhead-shaped
and MVIM mixers, and that more investigations are required to open focusing region and is characterised by rapid convective-diffusive
the way for industrial applications of microfluidic-based production of mass transport. The second region is located downstream the channel
organic NPs. junction, in the focused stream region, and is characterised by mixing
In reason of what stated above, this section aims to summarize dominated by molecular diffusion in the direction normal to the flow
the current literature on how the microfluidic environment influences streamlines. Through the variation of the total flow rate Jahn et al.
the dynamics of nanomaterial production leading to somewhat peculiar were able to influence the relative amount of liposomes assembling in
results, as compared with conventional preparation methods. This the two regions. This resulted in a fine modulation of mixing time
influence can be attributed to two main factors being the mixing which in turn resulted in the variation of liposome size [176]. Capretto
dynamics and the fluid shear stress. et al. [172] expanded this view for polymeric micelles self–assembly,
and developed a mathematical tool to rule out the effect of velocity
6.1. Computational fluid dynamic (CFD) models for indirect investigation of field from CFD results, making it possible to investigate the role of
the nanomaterial formation dynamics mixing process and residual solvent content during self-assembly in a
non-convoluted manner with fluid velocity field.
Microfluidic reactors are generally characterised by laminar Taken together these findings have demonstrated that hydro-
flow conditions which generate predictable and quantifiable species dynamics and diffusive mixing in microchannels are highly coupled
concentration distributions. This characteristic can be leveraged by phenomena. Viscosity, flow rate ratio, total flow rate, diffusivity and
indirectly investigating, through computational fluid dynamic fluid velocity profile contribute in concert to determine the mixing
models, the dynamics of nanomaterial formation (i.e. self-assembly process. An additional level of complexity is introduced when hybrid
or nanoprecipitation) and its dependence on the mixing dynamics. nanoparticles are produced. In this case two species undergo the
In a similar way CFD models could also be used to investigate how process of precipitation/self-assembly influencing each other.
various system parameters (such as device geometry, flow rate By using Pluronic F127 and β-carotene as model stabiliser and
ratio, total flow rate, etc.) affect the mixing dynamics with the hydrophobic drug, respectively, it was demonstrated the existence of
ultimate goal of correlating device/system operational conditions competitive reactions resulting in the formation of two types of
to nanomaterial output [170]. nanoparticles, i.e., either with or without loading β-carotene in the
Many researchers have therefore tried to couple computational and core–shell structure [1]. The relative amount of the two types of
experimental results in an effort to unveil the mechanism through particles could be varied by varying the process parameters such as
which microconfinement affects nanomaterial characteristics. Capretto concentration of the two species and fluidic conditions [1]. The authors
et al. [170] used a CFD model that took into consideration the variation explained the results with the aid of a CFD model that simulated the
of the diffusion coefficient of species according to the Stokes–Einstein effect of NP formation on their diffusion and advection kinetics and by
equation, which describes the relation between viscosity of the medium introducing a new parameter representing the mismatching between
and the diffusion coefficient of the species, and investigated its effect on the aggregations of the two precipitant species which affected the
the microfluidic mixing process and polymeric micelles size. They relative amount of the two particles produced. Interestingly, Valencia
demonstrated that variation in viscosity and diffusivity during solvent et al. [255] have demonstrated that in the case of the production of
exchange affects mixing efficiency by modulating mass transport lipid–polymer hybrid NPs, the stabiliser self-assembly (i.e. lipids) does
between solvent and non-solvent streams. The extent of this effect not affect the precipitation of the polymer core; hence, the formation
also depends on channel size and flow conditions. They therefore of the hybrid NPs could be seen as a two separate stages process. Despite
concluded that the interfacial phenomena and fluid properties the contrasting results obtained, the two groups similarly pointed out
significantly affect the nanoprecipitation process, alongside the effect the importance of controlling the amount of shell formation molecules
of reactor geometry and fluidic control. Similar findings which correlate (i.e. Pluronic F127 or lipid) to avoid undesired formation of empty
the effect of viscosity of solvent phase and nanoparticle dimension polymeric micelles or liposomes.
were also recently reported for the production of hyaluronic acid In an effort to correlate mixing and precipitation dynamics, different
nanoparticles [298]. authors have used physical parameters such as cloud point [296] and
It should be also considered that the high viscosity difference critical water concentration [1] as mixing criterion to quantify mixing
between the central focused stream and the two sheath fluids efficiency and the triggering of nanoprecipitation due to solvent-
represents a crucial feature to take into consideration when looking at displacement. This strategy has been found particularly useful in
the mixing process. Multi-fluid flows have been characterised by the understanding the mixing capability of a complex mixer in relation to
tendency for the less viscous fluid (i.e. water) to place itself in regions the variation of the process parameters [296].
of higher shear stress (i.e. near the channel walls) thus enveloping the
high viscosity fluid [299]. In addition, species diffusion across the fluid 6.2. Methods for direct investigation of nanomaterial formation dynamics
streams generates viscosity gradients which affect non-trivially both
fluid velocity field and mixing performance [299]. In this regard, it has Beside the indirect analysis performed with CFD models, direct
been demonstrated that reducing the flow rate ratio, the focused stream techniques have also been used to investigate the nanomaterial
becomes progressively more ensheathed by the non-solvent. This formation dynamics. Schutze et al. [301] have monitored the formation
causes a gradual increase of the surface-to-volume ratio of the interface of block copolymer NPs by using a fluorescent block copolymer and a
between the two phases, which likely accelerates the mixing process careful selection of chromospheres with an adsorption spectrum that
resulting in smaller and less polydisperse nanomaterial output [172]. overlapped with the emission spectrum of the fluorescent polymer. In
It should be noted that the analytical models (i.e. Eq. (10)) are this way the authors were able to spatially resolve particle formation
mostly based on mass flow balance and two-dimensional diffusion as well as incorporation of small guest molecules by microscopic
mass transfer, hence they do not take into account the contribution of fluorescence spectroscopy analysis. Authors showed that mixing
additional influencing factors such as the velocity profile and SVR of process and particle formation occurred simultaneously, with the first
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1523

being the rate determining step. These results suggest that there is not revealed that NP aspect ratio is of crucial importance to obtain a
distinguishable lag time before the creation of a supersaturate solution functional ion exchange nanocomposite. Thanks to the fine control
and the triggering of the aggregation. In addition, they demonstrated over NP characteristics, the fuel cell performance was observed to
that the payload find his way to the core of the particle gradually during improve significantly.
the particle formation process, rather than partitioning when the core– Wang et al. reported a procedure where a segmented gas–liquid
shell structure is already formed. device was used to produce block copolymer micelles of various
Jahn et al. [302] have recently described a method that relies on morphologies [304] and vesicles (Fig. 10c–h) [303]. The authors
instantaneous immobilisation by means of ultra-rapid cooling with postulated that the unique shear stress field produced in the device,
propane jet-freezing (Fig. 9a). With this method the authors were able characterised by shear hotspots at the plug corners, caused a shear-
to study the dynamic of formation of liposome by microscopic analysis induced coalescence of the just formed micelles, which led to the creation
of the slices of frozen fluid domain taken at different distances from of larger micellar structures; the latter, as results of the intramicellar
channels junction (Fig. 9b–f). Different distances correspond to different relaxation rearrangements, gave rise to various non-spherical micelles,
residence time and hence represent different stages along the assembly which became kinetically frozen when collected off-chip in a large excess
process. The method was able to reveal the formation of transient of water. These non-spherical micelles were mainly represented by
intermediate species and it was found a correlation between computed cylinders, however, due to the statistical nature of collision-coalescence
concentration profile and distribution of liposomes across the cross events at the hotspots, various shapes were created including linear
section. Of particular importance is that this method could be applied and looped circles, bilayer sheets and networks. Interestingly, the same
without altering the chemical characteristics of the molecules of interest authors also showed that by changing the solvent environment it was
as opposed to the fluorescent polymer method previously describe in possible to direct the rearrangement step toward the formation of
this section. vesicles rather than cylinders [303].

6.3. Effect of shear stress on nanomaterial characteristics 7. Assessing the biological performance of nanomaterials produced
by microfluidic technology: in-vitro and in-vivo tests
In microfluidic environment, due to the small channel dimension
and to the large surface-to-volume ratio, the wall effect becomes Evaluating the interaction between microfluidic-produced nano-
relevant in terms of the relative amount of fluid volume that it is materials and biological systems (i.e. cells and tissues) is a necessary
influencing. In other words, due to the small dimension, the shear pre-requisite to facilitate their potential translation into the clinical
field extends its effect across the whole cross section of the channel environment. However, only few studies have been performed in this
and impacts a relatively large amount of fluid volume. Recently few sense ― both in-vitro and in-vivo ― and a brief summary of recent
reports have shown how shear induced effects have an impact on findings is reported in this paragraph.
nanostructures formed in microfluidic environments. Using a classical In a study by Capretto et al. [171] a DNA-binding drug (mithramycin)
flow focusing device Hasani-Sadrabadi et al. [252] were able to exert a has been encapsulated within Pluronic® block copolymer micelles (PM-
control over the size and shape of polymeric NPs by tuning the flow MTH) produced by HFF (Fig. 11a–b). Authors investigated the effect of
rate ratio (Fig. 10a–b). The authors have suggested that the formation PM-MTH on erythroid differentiation of both human erythroleukemia
of elliptical NPs could be attributed to the shear field at the focusing and human erythroid precursor cells in-vitro. A comparative inves-
region that organizes the polymer chain in extended wormlike tigation between micelles produced by microfluidic technology and
structures. The rapid mixing then freezes this structures giving rise micelles produced by conventional bulk procedures was carried
to elongated particles. The particles became more elongated as the out (Fig. 11b). Notably, PM-MTH showed a slightly lower toxicity
flow rate ratio increased as a result of a longer mixing time that allows and a greater differentiative activity, when compared with the free
the stretched wormlike structures to be formed. Notably, authors drug. Furthermore, PM-MTH produced by microfluidic technology

Fig. 9. Schematic representation of the instantaneous immobilisation device which performs ultra-rapid cooling of the microfluidic domain with propane jet-freezing (a). Cryo-SEM
images of the slices of frozen fluidic domain: (b) low magnification image showing a clearly distinguishable focused stream at the centre of the fluidic domain; (b–f) high magnification
images showing the radial distribution of lipid vesicles (adapted with permission from [199]. Copyright 2013 American Chemical Society).
1524 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

Fig. 10. Effect of shear stress on nanomaterial characteristics. (a–b) Aspect ratio of nanoparticles was controlled by varying the flow rate ratio (adapted with permission from [252].
Copyright 2012 American Chemical Society). (c–e) Schematic representation of the gas-flow reactor used for the production of block copolymer micelles of various morphologies.
(f) Shear stress field caused a shear-induced coalescence of the micelles, which led to the creation of larger micellar structures which later rearranged to give rise to a (g–h) mixture of
spherical and non-spherical micelles (adapted with permission from [303]. Copyright 2012 American Chemical Society).

were capable of upregulating preferentially the production of γ- clinical symptoms associated with β-thalassemia. Results from this
globin messenger ribonucleic acid, and increasing the accumulation study represent an important step toward a potential clinical translation
of foetal haemoglobin (HbF). The percentage of HbF-containing of the produced polymeric micelles, since an increase in HbF could
cells and their HbF content were also increased without stimulating reduce or alleviate the symptoms of β-thalassemia and sickle cell
the expression of α-globin, which is a known responsible for the anaemia on the treated patients. In conclusion, observations from this

Fig. 11. Biological response induced by nanomaterials produced via microfluidic technology. (a–b) Effect of mithramycin on erythroid differentiation of K562 cells. Microscope images of
cells after benzidine staining (a), upon cell treatment with mithramycin encapsulated in polymeric micelles (i.e. produced by HFF), at varying mithramycin concentrations (mM). Scale
bar = 25 μm. The percentage of benzidine-positive cells is reported in (b) for control (absence of drug), MTH-PMs produced in batch, MTH-PMs produced by HFF, and free drug. Data
represent the average of three independent experiments (run in triplicate) ± standard deviation (microfluidics vs free mithramycin; P b 0.05) (republished with permission of DOVE
Medical Press, from [171], Copyright 2012; permission conveyed through Copyright Clearance Center, Inc.). (c–d) Osteogenic differentiation of hPDLSCs after exposure to PMs produced
by HFF. Differentiation potential of hPDLSCs was determined by evaluating, up to 21 days, the mineralisation of the cell cultures in the presence of control medium, osteogenic traditional
medium, PMs-batch and PMs-microfluidics. Alizarin Red staining for extracellular calcium deposition is reported. Furthermore, viability is determined by double staining assay with
Calcein-AM and propidium iodide (PI). Fluorescence photomicrographs show the presence of green fluorescence (Calcein-AM)-labelled live cells and the absence of red fluorescence,
(PI)-labelled dead cells (reprinted from [251], Copyright 2013, with permission from Elsevier). (e–g) Intracellular delivery of pDNA/PEI complexes produced by microfluidic HFF. pGFP
expression in NIH 3T3 cells at 4 days post transfection is reported for NPs production by batch method (e), HFF (f), and positive control (Lipofectamine) (g). (Scale bars = 500 μm).
(h–i) Confocal microscope images showing the intracellular distribution of PEI/pDNA complexes produced by microfluidic HFF, at 4 days post-transfection. pDNA is labelled with
YOYO-1 dye (green), and PEI is conjugated with rhodamine (red). Scale bars = 10 μm. (reprinted with permission from [305]. Copyright 2009 American Chemical Society).
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1525

study suggest that microfluidic approaches may be exploited for in this study that complexes produced by HFF resulted in higher
producing PM-MTH, to be potentially employed as a novel therapeutic percentage of cells undergoing late stages of apoptosis (i.e. programmed
protocol for β-thalassemia. cell death in the cell's life cycle) compared to macroscale routes,
The same group later reported in-vitro co-delivery of dexamethasone suggesting that different methods of lypoplexes production may
(Dex) and ascorbyl-palmitate (AP) to human periodontal ligament significantly influence cell death mechanisms.
mesenchymal stem cells (hPDLSCs), for induction of osteogenic Recently, Belliveau et al. [259] investigated the performance of
differentiation [251] (Fig. 11c–d). Drugs were co-encapsulated in lipid nanoparticles (SLN)–siRNA systems produced via a staggered
block copolymer micelles produced by a microfluidic HFF device. herringbone micromixer. The gene silencing potency of SLN–siRNA
Both microfluidic-based and bulk production techniques were systems was studied in-vivo, by using a mice model. Importantly,
investigated for comparison. Notably, drugs encapsulated into PMs results from this study showed that LNP–siRNA systems produced by
resulted in a lower in-vitro cytotoxicity compared to the free drug. In microfluidic-based mixing had equivalent or improved performance,
addition, PMs produced by HFF had a stronger osteogenic differentiative when compared with the ones produced by traditional bulk techniques
activity, if compared with PMs prepared by bulk method (Fig. 11c–d). (T-tube synthesis) using cationic lipids with optimised cationic lipid
This was attributed to the smaller size of PMs produced by microfluidic proportions and siRNA/lipid ratios. Furthermore, it was demonstrated
technology, which resulted in a more efficient cellular internalisation of that increasing the cationic lipid content (i.e. from 40 to 60 mol%)
the bio-actives. Results from this study suggested that PMs produced by at a fixed siRNA/total lipid ratio (i.e. 0.06 wt/wt), resulted in a
microfluidics may be employed as drug carriers for lipophilic drugs significant reduction of the siRNA dose level required to achieve
acting as inducers of osteoblastic differentiation and bone tissue 50% of gene silencing, ED50 (i.e. from 0.2 to ~ 0.03 mg siRNA/kg).
regeneration, with potential application in tissue engineering and Further improvement in potency could be achieved by decreasing
repair. the siRNA-to-lipid ratio (ED50 ~ 0.03 mg siRNA/kg, at siRNA/lipid
Koh et al. [305] investigated the use of polyplexes produced by ratio of 0.01). Results from this study suggest that microfluidic mixing
microfluidic technology as non-viral vectors for intra-cellular gene allows for routine production of LNP–siRNA systems, characterised by
delivery (Fig. 11e–g). Polyethylenimine (PEI) and plasmid DNA high siRNA encapsulation efficiency and equivalent or greater gene
(pDNA) complexes (PEI/pDNA) were produced by means of an HFF silencing potency, compared to previous formulation techniques.
microdevice, which was constituted of microchannels milled into layers In a recent study by Wi et al. [196] delivery of gene using liposomes
of PMMA. Both transfection efficiency and cell viability were quantified has been investigated. Lipid vesicles encapsulating recombinant
in-vitro, by using NIH 3T3 fibroblast cells and mouse embryonic stem expression plasmid pORF2 were produced by HFF, and their delivery
(mES) cells. The performance of polyplexes produced by microfluidic into PK-15 cells was studied in-vitro. The expression of Cap
technology was compared with that of bulk methods (i.e. vortexing). protein was monitored by fluorescence microscopy as an indication
Results showed that the microfluidic method allowed for improved of successful transfection. Results demonstrated the capability of
transfection efficiency compared to bulk methods, particularly at liposomes produced by microfluidic technology to allow for effective
the higher flow rate ratios (i.e. FRR = 6.7), which was assessed by transfection of mammalian cells in-vitro; however, no quantitative
either pGFP expression and alkaline phosphatase (SEAP) activity, at 2 analysis of transfection efficiency has been reported in this study,
and 4 days post-transfection. This was attributed to the more uniform which could represent the subject of future investigations.
NPs size distribution and the efficient complexation achieved It is also worth of notice that many of the proposed advancements of
with HFF. Confocal laser scanning microscopy (CLSM) revealed the nanomaterial produced with microfluidic approaches, such as long
spatial localisation of complexes within the cell (Fig. 11h–i). Notably, circulation and targeting properties, resulting from the small particle
complexes prepared by HFF accumulated at the perinuclear region of size and size homogeneity, could be addressed only in in-vivo setting.
the cells at 2 days post-transfection. However, at 4 days post- Considering the interesting results obtained so far with in-vitro test, it
transfection, the large majority of complexes disassociated and plasmids is envisaged that more in-vivo tests will be performed in the near future.
entered the cell nucleus for transfection (Fig. 11i). Importantly, cells
treated with complexes produced by microfluidic technology had the 8. Conclusion and outlooks
highest viability, likely due to more efficient complexation resulting in a
lower amount of free PEI. Results from this study thus demonstrate the Advances in nanomedicine have dictated the need for developing
potential of microfluidic-based HFF methods for producing gene carriers more robust and controllable procedures for the production of
with high transfection performance and biological tolerability. However, nanoparticles [306,307]. Classical methods largely rely on bulk mixing
an accurate design of the optimal production technique must take and tend to suffer from poor reproducibility from batch to batch and
into consideration the effect of fluid shear stress on the integrity of difficulties to implement fast screening and optimise nanomaterial
the encapsulated molecules. In this study, authors showed that the properties. In this respect, continuous flow microfluidic reactors have
hydrodynamic focused sample was characterised by increased number shown the capability to produce micro- and nanoparticles in a
of structural changes of the pDNA molecule (i.e. open circle structures) controllable and reproducible manner offering a possible solution to
compared to macroscale methods. Therefore, an optimal balance the aforementioned issues [106].
between complexation efficiency, volume production and preservation Compared to the intense research at the intersection between
of actives' molecular integrity must be identified. inorganic nanoparticles and microfluidics, relatively little has been
In a further study by Koh et al. [268], a multi-inlet microfluidic HFF done for the production of organic nanomaterials. Considering its high
device has been employed to produce lipopolyplex (LP) containing potential in nanomedicine, the use of microfluidic-based reactors for
Bcl-2 antisense deoxyoligonucleotide (ODN). Modulation of oncogene the production of various nanostructured drug delivery systems has
expression by means of ODNs has long been pursued as a therapeutic increasingly attracted attention from different fields.
route in cancer therapy; however, the use of free ODNs has been limited The fluid dynamic environment within microfluidic domains has been
by rapid clearance from the haematic circulation, opening the way for observed to have important implications on nanomaterials production.
the use of liposomes as drug vehicles. Results from this study showed Notably, unique features make microfluidics an appealing technique for
that complexes produced by HFF exerted greater downregulation effect nanomaterial production, including: (i) an efficient and controllable
on Bcl-2 in K562 leukaemia cells in-vitro, compared to bulk methods. mixing which results in a homogeneous reaction environment; (ii) an
This is likely due nanoparticles smaller size, narrower size distribution, efficient temperature control and heat transfer; (iii) the possibility of
lower zeta potential, and more uniform structure than those produced in situ monitoring the nanomaterial production process; (iv) the
by conventional bulk methods. Furthermore, it has been demonstrated possibility of tuning the characteristics of nanomaterials by controlling
1526 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

the kinetics of the process; and (v) the opportunity to integrate post these limitations, promising advancements have been reported
synthesis processes and measurement systems on a single technology by Tarabella et al. [310] in a recent study. Authors developed a
platform. Those features have been leveraged by various researchers microfluidic-based organic electrochemical transistor (OET) capable of
to produce different categories of nanomaterial, including drug sensing and real-time monitoring of liposome-based structures. The
nanoparticles, polymeric micelles, liposomes, niosomes, polymersomes microdevice demonstrated to be an efficient and sensitive detection
and hybrid nanoparticles. However, despite the breathing activity and system, with a lowest detection limit of 10−7 mg/mL, which falls within
the very encouraging results so far, the field of microfluidic production the range required for drug loading/drug delivery applications.
of nanomaterials is still at its early stage. Microfluidic reactors have thus One of the major challenges in the field of NPs analysis and physico-
the potential to become a standard production tool for the formulation chemical characterisation is also represented by the difficulties
of nanomedicine delivery systems with unprecedented degree of encountered in manipulating nanometre-sized bodies. In this review we
controllability and homogeneity. In our opinion, various aspects should discussed the use of different physical forces (i.e. electro-magnetic, optical
be addressed in the future to advance this technology. and mechanical) for manipulating, trapping and sorting NPs. However,
One of the most relevant shortcomings of microfluidic devices is challenges still remain to achieve high-throughput manipulation and
the relatively small production rate of nanomaterials. This is due to efficient interfacing with analytical and sensor systems. Given to its
the small dimension of the operating channels and the associated contact-less nature, electrophoresis represents the most widely adopted
technological restriction of mass flow rate attainable, due to the increase technique in this field, and recent studies have demonstrated its
in the pressure drop across the microchannels with increasing the flow capability to manipulate NPs in a continuous-flow format [311].
rate. Recently, various strategies to tackle this issue have been proposed Contact-less manipulation methods are usually preferred to mechanical
which have demonstrated nanoparticles production as high as 3 g/h. manipulation, which can potentially result in compromised NP
They relies on new micromixer designs where either microvortices integrity. However, the latter is usually simpler to implement and
[257], jet-impact [253] or multi-lamination strategies [308] have does not require integration with external energy sources. Recent
been employed to attain high production rate without compromising advancements in the field of mechanical manipulation of NPs have
the controllability of mixing. It is envisaged that in the future more been reported in a study by Wang et al. [312]. Authors developed
researchers will try to address this challenge, because it represents ciliated micropillars for multi-scale “gentle” filtration of a liposomes
one of the most important issues which hinder the use of microfluidic suspension. Cilia consisted of porous silicon nanowires which trapped
devices in industrial setting. Enhanced device throughput is often the suspended liposomes and could be subsequently dissolved in PBS
achieved by parallelization of individual functional units [239]. for recovery of the trapped sample. Notably, liposomes were integer
However, it is often difficult to ensure identical physical conditions after device operation.
within each unit, thus sometime resulting in lower control over NPs Another important feature of microfluidic reactors is the limited
properties [239]. It is thus important that increased production requirement in terms of reagent consumption. This, together with the
throughput is accompanied by high quality products. In this respect, possibility of in process variation of reaction conditions, including
soft lithography techniques may be employed to produce microchannels reactant concentration and content, might provide the possibility to
architectures with finely controlled dimensional characteristics, allowing create platform for rapid optimisation and high-throughput screening
for enhanced microchannel parallelization with accurate prediction of of nanomaterial's biological activity. A step towards this paradigm has
the flow field within each segment of the microfluidic system. Recently, been recently presented [313] utilising a continuous flow system.
Kang and coworkers [309] developed a multi-layered device comprising However, considering the fast growing interest and technological
of three PDMS layers and a glass slide, for high-throughput production of advancements in droplet microfluidics for screening applications
polymeric NPs. Device architecture consisted of a tree-like structure [226,314], it may be possible to see in the future the coupling of
with N 100 output channels in which hydrodynamic focusing of a microreactor technology and droplet-based analysis systems.
solution of MPEG–PLGA in acetonitrile was achieved. Furthermore, the Relevant to the assessment of NP biological performance is the
microfluidic device was designed in such a way that NP self-assembly development of devices capable of mimicking the physiological flow
occurred outside the PDMS channel, thus avoiding potential fouling on dynamic environment. Recently, Kusunose et al. [315] reported a PDMS-
the channel wall. NPs with average size in the range 50–200 nm were based micro-flow chamber for investigating the binding of liposomes to
obtained, characterised by significantly narrower size distribution a monolayer of endothelial cells, exposed to physiologically-relevant
compared to batch production. Importantly, NP production rate was fluid shear stress in the range 2.4–8.6 dyn/cm2. This platform could be
improved compared to previous studies (polymer flow rate in the employed as an alternative to animal models for testing the performance
range 0.5–2.0 mL/h) and represents a promising step forward towards of NPs for drug delivery applications. Notably, the mechanisms of
mass production of NPs. interaction between NPs and cells, and the cellular processes that govern
Microfluidic reactors are also suitable for introducing in line intracellular drug delivery remain largely unclear. A recent investigation
measurement systems which provide on time information on the by Sahay et al. [316] has opened new fascinating insights into the
characteristics of the produced nanomaterials [143]. Such information mechanisms of cellular uptake of short interfering RNA (siRNA) loaded
could be then employed to monitor the operation of the device and in lipid NPs produced by microfluidic technology. Authors demonstrated
hence quickly act to prevent or reduce the impact of any undesired that the uptake efficiency can be improved by designing delivery vehicles
deviation on the nanomaterial batch. As already demonstrated [119], which can escape the recycling pathways (i.e. endocytosis).
this information could be fed into a control algorithm to generate Recently various authors have shown that beside nanoparticle size, NP
internal feedback which will automatically tune the reaction conditions shape is also able to affect the process of cell internalisation [317–319]. In
to maintain the nanomaterial critical characteristics. In addition, in line this regard, it is important to remark that the fast mixing achievable in
measurement systems would provide a prompt information on the microfluidics provide a way to “lock” the formed NP in non-equilibrium
characteristics of the nanomaterial produced reducing the time needed morphology and crystal structures whit the aim to produce non-
during optimisation and screening processes. However, despite the conventional nanomaterial shapes. Investigations towards this direction
interesting applications of in line measurement systems, only limited have been recently published with promising results [252] but still with
research has been devoted to this area and it is therefore possible that limited evidence on the underlying governing mechanisms. In particular,
in the near future new studies will address this topic. In particular, Wang and coworkers [320] recently investigated the role played by
the developed techniques sometime suffer from NP-specificity and either chemical and fluidic conditions on the morphology of copolymer
poor sensitivity which makes them unsuitable for robust and large- self-assemblies produced using a two-phase gas–liquid segmented
scale NP characterisation and quality assay. In order to overcome microreactor. Authors highlighted peculiar characteristics of microfluidic
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1527

production compared to macroscale synthesis. This represents a very [13] L. Wang, J. Dong, J. Chen, J. Eastoe, X. Li, Design and optimization of a new
self-nanoemulsifying drug delivery system, J. Colloid Interface Sci. 330 (2009)
fascinating area of research which could open a highway of possibility 443–448.
not only limited to nanomedicine [252]. [14] D.K. Sarker, Engineering of nanoemulsions for drug delivery, Curr. Drug Deliv. 2
Furthermore, it must be highlighted that only few studies have (2005) 297–310.
[15] J. Panyam, V. Labhasetwar, Biodegradable nanoparticles for drug and gene delivery
investigated the encapsulation of actives within NPs in a microfluidic to cells and tissue, Adv. Drug Deliv. Rev. 64 (2012) 61–71.
system or the related governing mechanisms. As a matter of fact, this [16] J. Kipp, The role of solid nanoparticle technology in the parenteral delivery of
is crucial to design novel therapeutic strategies and it is envisaged that poorly water-soluble drugs, Int. J. Pharm. 284 (2004) 109–122.
[17] E.R. Gillies, J.M.J. Frechet, Dendrimers and dendritic polymers in drug delivery,
it will represent the subject of future investigations, as manifested by Drug Discov. Today 10 (2005) 35–43.
recent published studies. With this respect, Majedi et al. [321] employed [18] K. Kataoka, A. Harada, Y. Nagasaki, Block copolymer micelles for drug delivery:
a T-shape 2D HFF device to produce hydrophobically-modified chitosan design, characterization and biological significance, Adv. Drug Deliv. Rev. 64
(2012) 37–48.
NPs loaded with hydrophobic anticancer drugs (i.e. paclitaxel). Authors
[19] T. Lian, R.J.Y. Ho, Trends and developments in liposome drug delivery systems,
investigated the effect of operating parameters on loading efficiency J. Pharm. Sci. 90 (2001) 667–680.
and studied the kinetics of drug release from NPs. [20] N. Nishiyama, Y. Bae, K. Miyata, S. Fukushima, K. Kataoka, Smart polymeric micelles
Another important issue that should be addressed in the future is for gene and drug delivery, Drug Discov. Today Technol. 2 (2005) 21–26.
[21] F. Caruso, T. Hyeon, V.M. Rotello, Nanomedicine, Chem. Soc. Rev. 41 (2012)
represented by the necessity of performing additional post processing 2537–2538.
steps, after the microfluidic nanomaterial production, to obtain samples [22] R.K. Jain, T. Stylianopoulos, Delivering nanomedicine to solid tumors, Nat. Rev. Clin.
suitable for in-vivo and in-vitro tests. These include for instance, the Oncol. 7 (2010) 653–664.
[23] T. Nicolai, O. Colombani, C. Chassenieux, Dynamic polymeric micelles versus frozen
purification of the reaction mixture from toxic solvent through dialysis nanoparticles formed by block copolymers, Soft matter 6 (2010) 3111–3118.
[322], and the separation of the particles from the reaction mixture [323]. [24] A. Patra, N. Hebalkar, B. Sreedhar, M. Sarkar, A. Samanta, T. Radhakrishnan, Tuning
A fascinating area of future research is also represented by the design the size and optical properties in molecular nano/microcrystals: manifestation of
hierarchical interactions, Small 2 (2006) 650–659.
and production of vesicles to be employed as models of biological [25] S.M. D'Addio, R.K. Prud'homme, Controlling drug nanoparticle formation by rapid
systems. Microfluidics could represent a powerful tool to generate precipitation, Adv. Drug Deliv. Rev. 63 (2011) 417–426.
dimensionally controlled cellular models and incorporate sub-cellular [26] M. El-Shabouri, Nanoparticles for improving the dissolution and oral bioavailability
of spironolactone, a poorly-soluble drug, STP Pharm. Sci. 12 (2002) 97–101.
structures. In a recent study by Mijajlovic et al. [324], synthesis of 1- [27] Y. Wang, N. Herron, Nanometer-sized semiconductor clusters: materials
palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) vesicles was synthesis, quantum size effects, and photophysical properties, J. Phys. Chem.
performed with using 2D HFF device. From the moment the POPC is 95 (1991) 525–532.
[28] L. Knapp, The solubility of small particles and the stability of colloids, Trans.
one of the fundamental components of cell membrane, the produced
Faraday Soc. 17 (1922) 457–465.
vesicles could find application in the modelling of cells or organelles. A [29] A. Dokoumetzidis, P. Macheras, A century of dissolution research: from Noyes and
study by Kamiya et al. [325] has recently reported the production of Whitney to the biopharmaceutics classification system, Int. J. Pharm. 321
cell-sized vesicles encapsulating smaller vesicles by multi-step pulsed (2006) 1–11.
[30] B.C. Hancock, M. Parks, What is the true solubility advantage for amorphous
jetting. The produced “vesicles in a vesicle” could represent an artificial pharmaceuticals? Pharm. Res. 17 (2000) 397–404.
cell model to be employed in the investigation of biological processes, [31] G. Huang, J. Gao, Z. Hu, J.V. St John, B.C. Ponder, D. Moro, Controlled drug release
such as the intra-cellular traffic of vesicles. from hydrogel nanoparticle networks, J. Control. Release 94 (2004) 303–311.
[32] H. Maeda, The enhanced permeability and retention (EPR) effect in tumor
In conclusion, despite the fact that the use of microfluidic reactor for vasculature: the key role of tumor-selective macromolecular drug targeting, Adv.
nanomaterial synthesis is not yet a mature technology, research carried Enzym. Regul. 41 (2001) 189–207.
out in the last decades has established that microfluidics could facilitate [33] R. Singh, J.W. Lillard Jr., Nanoparticle-based targeted drug delivery, Exp. Mol.
Pathol. 86 (2009) 215–223.
the production of nanomaterial for drug delivery purposes, offering a [34] C. Chang, B. Paul, V. Remcho, S. Atre, J. Hutchison, Synthesis and post-processing
novel method to improve and tune their properties. Considering the of nanomaterials using microreaction technology, J. Nanoparticle Res. 10 (2008)
rapid expansion of both the fields of nanomedicine and microfluidics, 965–980.
[35] Y. Song, J. Hormes, C. Kumar, Microfluidic synthesis of nanomaterials, Small 4
the expansion of the knowledge at their intersection is likely to provide (2008) 698–711.
new exciting avenues in the future for tailoring the properties of [36] Y. Song, H. Modrow, L. Henry, C. Saw, E. Doomes, V. Palshin, J. Hormes, C. Kumar,
nanomaterial through controlled microfluidic precipitation. Microfluidic synthesis of cobalt nanoparticles, Chem. Mater. 18 (2006) 2817–2827.
[37] G. Whitesides, The origins and the future of microfluidics, Nature 442 (2006)
368–373.
[38] P. Watts, S.J. Haswell, Microfluidic combinatorial chemistry, Curr. Opin. Chem. Biol.
References 7 (2003) 380–387.
[39] W.Y. Lin, Y. Wang, S. Wang, H.R. Tseng, Integrated microfluidic reactors, Nano
[1] L. Capretto, W. Cheng, D. Carugo, O.L. Katsamenis, M. Hill, X. Zhang, Mechanism of Today 4 (2009) 470–481.
co-nanoprecipitation of organic actives and block copolymers in a microfluidic [40] J.M. Prot, O. Videau, C. Brochot, C. Legallais, H. Bénech, E. Leclerc, A cocktail of
environment, Nanotechnology 23 (2012) 375602. metabolic probes demonstrates the relevance of primary human hepatocyte cultures
[2] W. Cheng, L. Capretto, M. Hill, X. Zhang, Organic Nanoparticles Using Microfluidic in a microfluidic biochip for pharmaceutical drug screening, Int. J. Pharm. 408 (2011)
Technology for Drug Delivery Applications, in: Nanotechnologies for the Life 67–75.
Sciences, Wiley-VCH Verlag GmbH & Co., 2011 [41] M.R. Thorson, S. Goyal, B.R. Schudel, C.F. Zukoski, G.G.Z. Zhang, Y. Gong, P.J.A. Kenis,
[3] D. Horn, J. Rieger, Organic nanoparticles in the aqueous phase — theory, A microfluidic platform for pharmaceutical salt screening, Lab Chip 11 (2011)
experiment, and use, Angew. Chem. Int. Ed. 40 (2001) 4330–4361. 3829–3837.
[4] C.X. Zhao, L. He, S. Qiao, A.P.J. Middelberg, Nanoparticle synthesis in microreactors, [42] K. Riehemann, S.W. Schneider, T.A. Luger, B. Godin, M. Ferrari, H. Fuchs,
Chem. Eng. Sci. 66 (2010) 1463–1479. Nanomedicine-challenge and perspectives, Angew. Chem. Int. Ed. Engl. 48 (2009)
[5] S. Kostler, V. Ribitsch, Synthetic Approaches to Organic Nanoparticles, in: 872–897.
Nanotechnologies for the Life Sciences, Wiley-VCH Verlag GmbH & Co., 2011 [43] A. Tosi, S. Mazzitelli, L. Capretto, R. Guerrieri, C. Nastruzzi, Optimization of
[6] G. Schmid, Nanoparticles: From Theory to Application, Wiley-VCH, 2011. lipospheres production by factorial design and their performances on a
[7] V. Wagner, A. Dullaart, A. Bock, A. Zweck, The emerging nanomedicine landscape, dielectrophoretic lab-on-a-chip platform, Colloids Surf. A 340 (2009) 77–85.
Nat. Biotechnol. 24 (2006) 1211–1218. [44] D. Carugo, L. Capretto, S. Willis, A.L. Lewis, D. Grey, M. Hill, X. Zhang, A microfluidic
[8] E.M. Merisko-Liversidge, G.G. Liversidge, Drug nanoparticles: formulating poorly device for the characterisation of embolisation with polyvinyl alcohol beads
water-soluble compounds, Toxicol. Pathol. 36 (2008) 43–48. through biomimetic bifurcations, Biomed. Microdevices (2012) 1–11.
[9] K.P. Velikov, E. Pelan, Colloidal delivery systems for micronutrients and [45] J.I. Park, A. Saffari, S. Kumar, A. Günther, E. Kumacheva, Microfluidic synthesis
nutraceuticals, Soft matter 4 (2008) 1964–1980. of polymer and inorganic particulate materials, Annu. Rev. Mater. Res. 40 (2010)
[10] Z. Tian, W. Wu, A.D.Q. Li, Photoswitchable fluorescent nanoparticles: preparation, 415–443.
properties and applications, ChemPhysChem 10 (2009) 2577–2591. [46] L. Capretto, S. Mazzitelli, G. Luca, C. Nastruzzi, Preparation and characterization of
[11] S.M. Borisov, T. Mayr, G. Mistlberger, K. Waich, K. Koren, P. Chojnacki, I. Klimant, polysaccharidic microbeads by a microfluidic technique: application to the
Precipitation as a simple and versatile method for preparation of optical encapsulation of Sertoli cells, Acta Biomater. 6 (2010) 429–435.
nanochemosensors, Talanta 79 (2009) 1322–1330. [47] S. Koster, F.E. Angile, H. Duan, J.J. Agresti, A. Wintner, C. Schmitz, A.C. Rowat, C.A.
[12] S. Rao, J. Shao, Self-nanoemulsifying drug delivery systems (SNEDDS) for oral Merten, D. Pisignano, A.D. Griffiths, Drop-based microfluidic devices for encapsulation
delivery of protein drugs: I. Formulation development, Int. J. Pharm. 362 (2008) 2. of single cells, Lab Chip 8 (2008) 1110–1115.
1528 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

[48] S. Mazzitelli, L. Capretto, D. Carugo, X. Zhang, R. Piva, C. Nastruzzi, Optimised [81] S. Kumar, T. Nann, Shape control of II–VI semiconductor nanomaterials, Small 2
production of multifunctional microfibres by microfluidic chip technology for (2006) 316–329.
tissue engineering applications, Lab Chip 11 (2011) 1776–1785. [82] D. Talapin, A. Rogach, A. Kornowski, M. Haase, H. Weller, Highly luminescent
[49] L. Capretto, S. Mazzitelli, C. Balestra, A. Tosi, C. Nastruzzi, Effect of the gelation monodisperse CdSe and CdSe/ZnS nanocrystals synthesized in a hexadecylamine–
process on the production of alginate microbeads by microfluidic chip technology, trioctylphosphine oxide–trioctylphospine mixture, Nano Lett. 1 (2001) 207–211.
Lab Chip 8 (2008) 617–621. [83] G.K. Batchelor, An Introduction to Fluid Dynamics, Cambridge Univ Pr, 2000.
[50] M. Antonietti, S. Förster, Vesicles and liposomes: a self‐assembly principle beyond [84] D.J. Beebe, G.A. Mensing, G.M. Walker, Physics and application of microfluidic in
lipids, Adv. Mater. 15 (2003) 1323–1333. biology, Annu. Rev. Biomed. Eng. 4 (2002) 261–286.
[51] D.D. Lasic, The mechanism of vesicle formation, Biochem. J. 256 (1988) 1. [85] B. Weigl, R. Bardell, C. Cabrera, Lab-on-a-chip for drug development, Adv. Drug
[52] B.M. Discher, Y.-Y. Won, D.S. Ege, J.C. Lee, F.S. Bates, D.E. Discher, D.A. Hammer, Deliv. Rev. 55 (2003) 349–377.
Polymersomes: tough vesicles made from diblock copolymers, Science 284 (1999) [86] N. Nguyen, S. Wereley, Fundamentals and Applications of Microfluidics, Artech
1143–1146. House Publishers, 2002.
[53] H. Kukula, H. Schlaad, M. Antonietti, S. Förster, The formation of polymer vesicles [87] M.E. Aulton, J.W. Cooper, Pharmaceutics: The Science of Dosage Form Design,
or “peptosomes” by polybutadiene-block-poly(l-glutamate)s in dilute aqueous Churchill Livingstone, London, 2002.
solution, J. Am. Chem. Soc. 124 (2002) 1658–1663. [88] J.T. Edward, Molecular volumes and the Stokes–Einstein equation, J. Chem. Educ.
[54] S. Hyde, Curvature and the global structure of interfaces in surfactant–water 47 (1970) 261.
systems, J. Phys. Colloques 51 (1990) C7-209–C7-228. [89] A. Einstein, R. Fürth, Investigations on the Theory of the Brownian Movement,
[55] P.G. Khalatur, A.R. Khokhlov, I.A. Nyrkova, A.N. Semenov, Aggregation Dover Pubns, 1956.
processes in self‐associating polymer systems: computer simulation study of [90] Z. Zhang, P. Zhao, G. Xiao, M. Lin, X. Cao, Focusing-enhanced mixing in microfluidic
micelles in the superstrong segregation regime, Macromol. Theory Simul. 5 channels, Biomicrofluidics 2 (2008) 014101.
(1996) 713–747. [91] V. Kumar, M. Paraschivoiu, K. Nigam, Single-phase fluid flow and mixing in
[56] Y. Barenholz, D. Gibbes, B. Litman, J. Goll, T. Thompson, F. Carlson, A simple microchannels, Chem. Eng. Sci. 66 (7) (2010) 1329–1373.
method for the preparation of homogeneous phospholipid vesicles, Biochemistry [92] G. Yaralioglu, I. Wygant, T. Marentis, B. Khuri-Yakub, Ultrasonic mixing in microfluidic
16 (1977) 2806–2810. channels using integrated transducers, Anal. Chem. 76 (2004) 3694–3698.
[57] G. Battaglia, A.J. Ryan, Pathways of polymeric vesicle formation, J. Phys. Chem. B [93] I. Glasgow, N. Aubry, Enhancement of microfluidic mixing using time pulsing, Lab
110 (2006) 10272–10279. Chip 3 (2003) 114–120.
[58] A. Moscho, O. Orwar, D.T. Chiu, B.P. Modi, R.N. Zare, Rapid preparation of giant [94] Z. Yang, S. Matsumoto, H. Goto, M. Matsumoto, R. Maeda, Ultrasonic micromixer
unilamellar vesicles, Proc. Natl. Acad. Sci. 93 (1996) 11443–11447. for microfluidic systems, Sensors Actuators A Phys. 93 (2001) 266–272.
[59] L. Luo, A. Eisenberg, Thermodynamic size control of block copolymer vesicles in [95] H. Tsai Jr., L. Lin, Active microfluidic mixer and gas bubble filter driven by thermal
solution, Langmuir 17 (2001) 6804–6811. bubble micropump, Sensors Actuators A Phys. 97 (2002) 665–671.
[60] M.I. Angelova, D.S. Dimitrov, Liposome electroformation, Faraday Discuss. Chem. [96] H. Bau, J. Zhong, M. Yi, A minute magneto hydro dynamic (MHD) mixer, Sensors
Soc. 81 (1986) 303–311. Actuators B Chem. 79 (2001) 207–215.
[61] M. Inkyo, T. Tahara, T. Iwaki, F. Iskandar, C.J. Hogan, K. Okuyama, Experimental [97] Z. Wu, N. Nguyen, Convective–diffusive transport in parallel lamination micromixers,
investigation of nanoparticle dispersion by beads milling with centrifugal bead Microfluid. Nanofluid. 1 (2005) 208–217.
separation, J. Colloid Interface Sci. 304 (2006) 535–540. [98] A. Kamholz, P. Yager, Molecular diffusive scaling laws in pressure-driven
[62] C. Qian, D.J. McClements, Formation of nanoemulsions stabilized by model microfluidic channels: deviation from one-dimensional Einstein approximations,
food-grade emulsifiers using high-pressure homogenization: factors affecting Sensors Actuators B Chem. 82 (2002) 117–121.
particle size, Food Hydrocoll. 25 (2011) 1000–1008. [99] N. Schwesinger, T. Frank, H. Wurmus, A modular microfluid system with an
[63] Y. Dong, S.S. Feng, Poly(d, l-lactide-co-glycolide) (PLGA) nanoparticles prepared by integrated micromixer, J. Micromech. Microeng. 6 (1996) 99.
high pressure homogenization for paclitaxel chemotherapy, Int. J. Pharm. 342 [100] J. Knight, A. Vishwanath, J. Brody, R. Austin, Hydrodynamic focusing on a silicon
(2007) 208–214. chip: mixing nanoliters in microseconds, Phys. Rev. Lett. 80 (1998) 3863–3866.
[64] J.P. Rolland, B.W. Maynor, L.E. Euliss, A.E. Exner, G.M. Denison, J.M. DeSimone, Direct [101] A. Günther, M. Jhunjhunwala, M. Thalmann, M. Schmidt, K. Jensen,
fabrication and harvesting of monodisperse, shape-specific nanobiomaterials, J. Am. Micromixing of miscible liquids in segmented gas–liquid flow, Langmuir 21
Chem. Soc. 127 (2005) 10096–10100. (2005) 1547–1555.
[65] J.Y. Kelly, J.M. DeSimone, Shape-specific, monodisperse nano-molding of protein [102] H. Song, J. Tice, R. Ismagilov, A microfluidic system for controlling reaction networks
particles, J. Am. Chem. Soc. 130 (2008) 5438–5439. in time, Angew. Chem. 115 (2003) 792–796.
[66] L.C. Glangchai, M. Caldorera-Moore, L. Shi, K. Roy, Nanoimprint lithography [103] T. Johnson, D. Ross, L. Locascio, Rapid microfluidic mixing, Anal. Chem. 74 (2002)
based fabrication of shape-specific, enzymatically-triggered smart nanoparticles, 45–51.
J. Control. Release 125 (2008) 263–272. [104] A. Stroock, S. Dertinger, A. Ajdari, I. Mezic, H. Stone, G. Whitesides, Chaotic mixer
[67] C.E. Astete, C.S.S.R. Kumar, C.M. Sabliov, Size control of poly (d, l-lactide- biN cob/iN - for microchannels, Science 295 (2002) 647.
glycolide) and poly (d, l-lactide- biN cob/iN -glycolide)-magnetite nanoparticles [105] L. Capretto, W. Cheng, M. Hill, X. Zhang, Micromixing within microfluidic devices,
synthesized by emulsion evaporation technique, Colloids Surf A 299 (2007) 209–216. Microfluidics — Technologies and Applications, Springer, Berlin Heidelberg, 2011,
[68] M. Trotta, M. Gallarate, F. Pattarino, S. Morel, Emulsions containing partially pp. 27–68.
water-miscible solvents for the preparation of drug nanosuspensions, J. Control. [106] S. Marre, K.F. Jensen, Synthesis of micro and nanostructures in microfluidic
Release 76 (2001) 119–128. systems, Chem. Soc. Rev. 39 (2010) 1183–1202.
[69] D. Quintanar-Guerrero, A. Ganem-Quintanar, E. Allemann, H. Fessi, E. Doelker, [107] C.Y. Lee, C.L. Chang, Y.N. Wang, L.M. Fu, Microfluidic mixing: a review, Int. J. Mol.
Influence of the stabilizer coating layer on the purification and freeze-drying of Sci. 12 (2011) 3263–3287.
poly (D, L-lactic acid) nanoparticles prepared by an emulsion–diffusion technique, [108] H.R. Sahoo, J.G. Kralj, K.F. Jensen, Multistep continuous flow microchemical
J. Microencapsul. 15 (1998) 107–119. synthesis involving multiple reactions and separations, Angew. Chem. 119 (2007)
[70] J.W. Chew, N. Simon, P.S. Chow, R.B.H. Tan, Comparison between open-loop 5806–5810.
temperature control and closed-loop supersaturation control for cooling [109] A. Abou Hassan, R. Bazzi, V. Cabuil, Multistep continuous-flow microsynthesis of
crystallization of glycine, Ind. Eng. Chem. Res. 46 (2007) 830–838. magnetic and fluorescent γ-Fe2O3@SiO2 core/shell nanoparticles, Angew. Chem.
[71] M.E. Matteucci, M.A. Hotze, K.P. Johnston, R.O. Williams III, Drug nanoparticles by Int. Ed. 48 (2009) 7180–7183.
antisolvent precipitation: mixing energy versus surfactant stabilization, Langmuir [110] Y. Ying, G. Chen, Y. Zhao, S. Li, Q. Yuan, A high throughput methodology for
22 (2006) 8951–8959. continuous preparation of monodispersed nanocrystals in microfluidic reactors,
[72] M.C. Brick, H.J. Palmer, T.H. Whitesides, Formation of colloidal dispersions of Chem. Eng. J. 135 (2008) 209–215.
organic materials in aqueous media by solvent shifting, Langmuir 19 (2003) [111] H. Nakamura, Y. Yamaguchi, M. Miyazaki, H. Maeda, M. Uehara, P. Mulvaney,
6367–6380. Preparation of CdSe nanocrystals in a micro-flow-reactor, Chem. Commun. (2002)
[73] T. Govender, S. Stolnik, M.C. Garnett, L. Illum, S.S. Davis, PLGA nanoparticles 2844–2845.
prepared by nanoprecipitation: drug loading and release studies of a water soluble [112] J. Ju, C. Zeng, L. Zhang, N. Xu, Continuous synthesis of zeolite NaA in a microchannel
drug, J. Control. Release 57 (1999) 171–185. reactor, Chem. Eng. J. 116 (2006) 115–121.
[74] P. Kang, C. Chen, L. Hao, C. Zhu, Y. Hu, Z. Chen, A novel sonication route to prepare [113] J. Boleininger, A. Kurz, V. Reuss, C. Sonnichsen, Microfluidic continuous flow
anthracene nanoparticles, Mater. Res. Bull. 39 (2004) 545–551. synthesis of rod-shaped gold and silver nanocrystals, Phys. Chem. Chem. Phys. 8
[75] Z. Jia, Y. Ma, W. Yang, Z. Zhou, J. Yao, Z. Liu, Preparation of pyrazoline nanoparticles (2006) 3824–3827.
with a membrane mixer, Colloids Surf A 276 (2006) 22–27. [114] M. Takagi, T. Maki, M. Miyahara, K. Mae, Production of titania nanoparticles by
[76] V. LaMer, R. Dinegar, Theory, production and mechanism of formation of using a new microreactor assembled with same axle dual pipe, Chem. Eng. J. 101
monodispersed hydrosols, J. Am. Chem. Soc. 72 (1950) 4847–4854. (2004) 269–276.
[77] G. Majano, A. Darwiche, S. Mintova, V. Valtchev, Seed-induced crystallization of [115] A. Abou Hassan, O. Sandre, S. Neveu, V. Cabuil, Synthesis of goethite by separation
nanosized Na-ZSM-5 crystals, Ind. Eng. Chem. Res. 48 (2009) 7084–7091. of the nucleation and growth processes of ferrihydrite nanoparticles using
[78] D. Oliveira, K. Baba, J. Mori, Y. Miyashita, H. Kasai, H. Oikawa, H. Nakanishi, Using an microfluidics, Angew. Chem. Int. Ed. Engl. 48 (2009) 2342–2345.
organic additive to manipulate sizes of perylene nanoparticles, J. Cryst. Growth 312 [116] A.A. Hassan, O. Sandre, V. Cabuil, P. Tabeling, Synthesis of iron oxide nanoparticles
(2010) 431–436. in a microfluidic device: preliminary results in a coaxial flow millichannel, Chem.
[79] J. DeMello, A. DeMello, Microscale reactors: nanoscale products, Lab Chip 4 (2004) Commun. (2008) 1783–1785.
11N. [117] H. Ali, P. York, N. Blagden, Preparation of hydrocortisone nanosuspension through
[80] A.J. Mahajan, D.J. Kirwan, Nucleation and growth kinetics of biochemicals a bottom-up nanoprecipitation technique using microfluidic reactors, Int. J. Pharm.
measured at high supersaturations, J. Cryst. Growth 144 (1994) 281–290. 375 (2009) 107–113.
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1529

[118] T. Sounart, P. Safier, J. Voigt, J. Hoyt, D. Tallant, C. Matzke, T. Michalske, Spatially- [149] M.E. Davis, Nanoparticle therapeutics: an emerging treatment modality for cancer,
resolved analysis of nanoparticle nucleation and growth in a microfluidic reactor, Nat. Rev. Drug Discov. 7 (2008) 771–782.
Lab Chip 7 (2007) 908–915. [150] C. Amador, A. Gavriilidis, P. Angeli, Flow distribution in different microreactor
[119] S. Krishnadasan, R. Brown, A. Demello, J. Demello, Intelligent routes to the scale-out geometries and the effect of manufacturing tolerances and channel
controlled synthesis of nanoparticles, Lab Chip 7 (2007) 1434–1441. blockage, Chem. Eng. J. 101 (2004) 379–390.
[120] D. Shalom, R. Wootton, R. Winkle, B. Cottam, R. Vilar, A. Demello, C. Wilde, Synthesis [151] J. Commenge, L. Falk, J. Corriou, M. Matlosz, Optimal design for flow uniformity in
of thiol functionalized gold nanoparticles using a continuous flow microfluidic microchannel reactors, AIChE J 48 (2002) 345–358.
reactor, Mater. Lett. 61 (2007) 1146–1150. [152] M. Madou, Fundamentals of Microfabrication, CRC Press, 1997.
[121] F. Bessoth, A. deMello, A. Manz, Microstructure for efficient continuous flow [153] D. Janasek, J. Franzke, A. Manz, Scaling and the design of miniaturized chemical-
mixing, Anal. Commun. 36 (1999) 213–215. analysis systems, Nature 442 (2006) 374–380.
[122] J. Edel, R. Fortt, J. demello, A. Demello, Microfluidic routes to the controlled [154] Q. Chen, G. Li, Q. Jin, J. Zhao, Q. Ren, Y. Xu, A rapid and low-cost procedure
production of nanoparticles, Chem. Commun. 2002 (2002) 1136–1137. for fabrication of glass microfluidic devices, J. Microelectromech. Syst. 16 (2007)
[123] J. Wagner, J. Köhler, Continuous synthesis of gold nanoparticles in a microreactor, 1193–1200.
Nano Lett. 5 (2005) 685. [155] H. Becker, C. Gärtner, Polymer microfabrication methods for microfluidic analytical
[124] J. Wagner, T. Tshikhudo, J. Koehler, Microfluidic generation of metal nanoparticles applications, Electrophoresis 21 (2000) 12–26.
by borohydride reduction, Chem. Eng. J. 135 (2008) S104–S109. [156] J. Lee, C. Park, G. Whitesides, Solvent compatibility of poly (dimethylsiloxane)-
[125] J.M. Köhler, H. Romanus, U. Hübner, J. Wagner, Formation of star-like and core– based microfluidic devices, Anal. Chem. 75 (2003) 6544–6554.
shell AuAg nanoparticles during two-and three-step preparation in batch and in [157] B. Kim, L. Hong, Y. Chung, D. Kim, C. Lee, Solvent resistant PDMS microfluidic
microfluidic systems, J. Nanomater. 2007 (2007) 1–7. (Articles ID 98134). devices with hybrid inorganic/organic polymer coatings, Adv. Funct. Mater. 19
[126] A. Jahn, W. Vreeland, M. Gaitan, L. Locascio, Controlled vesicle self-assembly in (2009) 3796–3803.
microfluidic channels with hydrodynamic focusing, J. Am. Chem. Soc. 126 (2004) [158] R. Delille, M. Urdaneta, S. Moseley, E. Smela, Benchtop polymer MEMS,
2674–2675. J. Microelectromech. Syst. 15 (2006) 1108–1120.
[127] S. Zhang, J. Yun, S. Shen, Z. Chen, K. Yao, J. Chen, B. Chen, Formation of solid lipid [159] J. Greener, W. Li, J. Ren, D. Voicu, V. Pakharenko, T. Tang, E. Kumacheva, Rapid,
nanoparticles in a microchannel system with a cross-shaped junction, Chem. cost-efficient fabrication of microfluidic reactors in thermoplastic polymers by
Eng. Sci. 63 (2008) 5600–5605. combining photolithography and hot embossing, Lab Chip 10 (2009) 522–524.
[128] R. Karnik, F. Gu, P. Basto, C. Cannizzaro, L. Dean, W. Kyei-Manu, R. Langer, O. [160] M.J. Jackson, Microfabrication and Nanomanufacturing, CRC, 2006.
Farokhzad, Microfluidic platform for controlled synthesis of polymeric nanoparticles, [161] T. McCreedy, Rapid prototyping of glass and PDMS microstructures for micro total
Nano Lett. 8 (2008) 2906–2912. analytical systems and micro chemical reactors by microfabrication in the general
[129] M. Rhee, P.M. Valencia, M.I. Rodriguez, R. Langer, O.C. Farokhzad, R. Karnik, laboratory, Anal. Chim. Acta 427 (2001) 39–43.
Synthesis of size tunable polymeric nanoparticles enabled by 3D hydro- [162] J. Yun, S. Zhang, S. Shen, Z. Chen, K. Yao, J. Chen, Continuous production of
dynamic flow focusing in single layer microchannels, Adv. Mater. 23 (2011) solid lipid nanoparticles by liquid flow-focusing and gas displacing method in
H79–H83. microchannels, Chem. Eng. Sci. 64 (2009) 4115–4122.
[130] C.H. Weng, C.C. Huang, C.S. Yeh, H.Y. Lei, G.B. Lee, Synthesis of hexagonal gold [163] A. Almeida, E. Souto, Solid lipid nanoparticles as a drug delivery system for
nanoparticles using a microfluidic reaction system, J. Micromech. Microeng. 18 peptides and proteins, Adv. Drug Deliv. Rev. 59 (2007) 478–490.
(2008) 035019. [164] Y.-F. Su, H. Kim, S. Kovenklioglu, W. Lee, Continuous nanoparticle production
[131] E. Chan, A. Alivisatos, R. Mathies, High-temperature microfluidic synthesis by microfluidic-based emulsion, mixing and crystallization, J. Solid State Chem.
of CdSe nanocrystals in nanoliter droplets, J. Am. Chem. Soc. 127 (2005) 180 (2007) 2625–2629.
13854–13861. [165] V. Génot, S. Desportes, C. Croushore, J.-P. Lefèvre, R.B. Pansu, J.A. Delaire, P.R. von
[132] I. Shestopalov, J. Tice, R. Ismagilov, Multi-step synthesis of nanoparticles performed Rohr, Synthesis of organic nanoparticles in a 3D flow focusing microreactor,
on millisecond time scale in a microfluidic droplet-based system, Lab Chip 4 (2004) Chem. Eng. J. 161 (2010) 234–239.
316–321. [166] S. Dev, J. Toster, S.V. Prasanna, M. Fitzgerald, K.S. Iyer, C.L. Raston, Suppressing
[133] S. Khan, A. Gunther, M. Schmidt, K. Jensen, Microfluidic synthesis of colloidal silica, regrowth of microfluidic generated drug nanocrystals using polyelectrolyte
Langmuir 20 (2004) 8604–8611. coatings, RSC Adv. 3 (2013) 695–698.
[134] H. Wang, H. Nakamura, M. Uehara, M. Miyazaki, H. Maeda, Preparation of titania [167] S. Dev, P. Prabhakaran, L. Filgueira, K.S. Iyer, C.L. Raston, Microfluidic fabrication of
particles utilizing the insoluble phase interface in a microchannel reactor, Chem. cationic curcumin nanoparticles as an anti-cancer agent, Nanoscale 4 (2012)
Commun. 2002 (2002) 1462–1463. 2575–2579.
[135] D. Wilms, J. Klos, H. Frey, Microstructured reactors for polymer synthesis: a [168] A. Jahn, W.N. Vreeland, D.L. DeVoe, L.E. Locascio, M. Gaitan, Microfluidic directed
renaissance of continuous flow processes for tailor-made macromolecules? formation of liposomes of controlled size, Langmuir 23 (2007) 6289–6293.
Macromol. Chem. Phys. 209 (2008) 343–356. [169] C. Lo, A. Jahn, L. Locascio, W. Vreeland, Controlled self-assembly of monodisperse
[136] A. Demello, Control and detection of chemical reactions in microfluidic systems, niosomes by microfluidic hydrodynamic focusing, Langmuir 26 (2010) 8559–8566.
Nature 442 (2006) 394–402. [170] L. Capretto, D. Carugo, W. Cheng, M. Hill, X. Zhang, Continuous-flow production
[137] A. Chow, Lab on a chip: opportunities for chemical engineering, AIChE J 48 (2002) of polymeric micelles in microreactors: Experimental and computational analysis,
1590–1595. J. Colloid Interface Sci. 357 (2011) 243–251.
[138] T. Kawaguchi, H. Miyata, K. Ataka, K. Mae, J. Yoshida, Room temperature swern [171] L. Capretto, S. Mazzitelli, E. Brognara, I. Lampronti, D. Carugo, M. Hill, X. Zhang, R.
oxidations by using a microscale flow system, Angew. Chem. Int. Ed. Engl. 44 (2005) Gambari, C. Nastruzzi, Mithramycin encapsulated in polymeric micelles by
2413–2416. microfluidic technology as novel therapeutic protocol for beta-thalassemia, Int. J.
[139] H. Nakamura, Y. Yamaguchi, M. Miyazaki, H. Maeda, M. Uehara, P. Mulvaney, Nanomedicine 7 (2012) 307.
Preparation of CdSe nanocrystals in a micro-flow-reactor, Chem. Commun. 2002 [172] L. Capretto, S. Mazzitelli, G. Colombo, R. Piva, L. Penolazzi, R. Vecchiatini, X. Zhang,
(2002) 2844–2845. C. Nastruzzi, Production of polymeric micelles by microfluidic technology for
[140] H. Nakamura, A. Tashiro, Y. Yamaguchi, M. Miyazaki, T. Watari, H. Shimizu, H. combined drug delivery: application to osteogenic differentiation of human
Maeda, Application of a microfluidic reaction system for CdSe nanocrystal periodontal ligament mesenchymal stem cells (hPDLSC), Int. J. Pharm. 440
preparation: their growth kinetics and photoluminescence analysis, Lab Chip 4 (2012) 195–206.
(2004) 237–240. [173] B.K. Johnson, R.K. Prud'homme, Mechanism for rapid self-assembly of block
[141] H. Wang, X. Li, M. Uehara, Y. Yamaguchi, H. Nakamura, M. Miyazaki, H. Shimizu, H. copolymer nanoparticles, Phys. Rev. Lett. 91 (2003) 118302.
Maeda, Continuous synthesis of CdSe–ZnS composite nanoparticles in a [174] H. Cui, Z. Chen, S. Zhong, K. Wooley, D. Pochan, Block copolymer assembly via
microfluidic reactor, Chem. Commun. 2004 (2004) 48–49. kinetic control, Science 317 (2007) 647.
[142] S. Zinoveva, R. De Silva, R.D. Louis, P. Datta, K. SSR, The wet chemical synthesis of [175] B. Ratner, S. Bryant, Biomaterials: where we have been and where we are going,
Co nanoparticles in a microreactor system: a time-resolved investigation by X-ray Annu. Rev. Biomed. Eng. 6 (2004) 41–75.
absorption spectroscopy, Nucl. Inst. Methods Phys. Res. A 582 (2007) 239–241. [176] A. Jahn, S.M. Stavis, J.S. Hong, W.N. Vreeland, D.L. DeVoe, M. Gaitan, Microfluidic
[143] J. Polte, R. Erler, A.F. Thunemann, S. Sokolov, T.T. Ahner, K. Rademann, F. mixing and the formation of nanoscale lipid vesicles, ACS Nano 4 (2010) 2077–2087.
Emmerling, R. Kraehnert, Nucleation and growth of gold nanoparticles studied [177] D. van Swaay, A. DeMello, Microfluidic methods for forming liposomes, Lab Chip
via in situ small angle x-ray scattering at millisecond time resolution, ACS Nano 13 (2013) 752–767.
4 (2010) 1076–1082. [178] S. Matosevic, Synthesizing artificial cells from giant unilamellar vesicles: state-of-the
[144] S. Khan, K. Jensen, Microfluidic synthesis of titania shells on colloidal silica, Adv. art in the development of microfluidic technology, BioEssays 34 (2012) 992–1001.
Mater. 19 (2007) 2556–2560. [179] A. Jesorka, O. Orwar, Liposomes: technologies and analytical applications, Annu.
[145] Y. Song, T. Zhang, W. Yang, S. Albin, L. Henry, Fine crystal structure transition of Rev. Anal. Chem. 1 (2008) 801–832.
cobalt nanoparticles formed in a microfluidic reactor, Cryst. Growth Des. 8 (2008) [180] K. Kuribayashi, G. Tresset, P. Coquet, H. Fujita, S. Takeuchi, Electroformation of
3766–3772. giant liposomes in microfluidic channels, in: TRANSDUCERS'05 (Ed.), The 13th
[146] T. Chastek, K. Beers, E. Amis, Miniaturized dynamic light scattering instrumentation International Conference on Solid-State Sensors, Actuators and Microsystems,
for use in microfluidic applications, Rev. Sci. Instrum. 78 (2007) 072201. IEEE, 2005, pp. 1159–1162.
[147] T. Chastek, K. Iida, E. Amis, M. Fasolka, K. Beers, A microfluidic platform for [181] K. Kuribayashi, G. Tresset, P. Coquet, H. Fujita, S. Takeuchi, Electroformation of giant
integrated synthesis and dynamic light scattering measurement of block liposomes in microfluidic channels, Meas. Sci. Technol. 17 (2006) 3121.
copolymer micelles, Lab Chip 8 (2008) 950–957. [182] K. Kuribayashi, A. Utada, S. Takeuchi, Electrofomation of giant liposomes from
[148] C. Kuyper, K. Budzinski, R. Lorenz, D. Chiu, Real-time sizing of nanoparticles in densely micro-patterned lipid films, Twelfth International Conference on
microfluidic channels using confocal correlation spectroscopy, J. Am. Chem. Soc. Miniaturized Systems for Chemistry and Life Sciences, RSC, San Diego, California,
128 (2006) 730–731. USA, 2008, pp. 1120–1122.
1530 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

[183] K. Kuribayashi, S. Takeuchi, Electroformation of solvent-free lipid membranes over [212] D.L. Richmond, E.M. Schmid, S. Martens, J.C. Stachowiak, N. Liska, D.A. Fletcher,
microaperture array, 21st International Conference on Micro Electro Mechanical Forming giant vesicles with controlled membrane composition, asymmetry, and
Systems, MEMS, IEEE, 2008, pp. 296–299. contents, Proc. Natl. Acad. Sci. 108 (2011) 9431–9436.
[184] M. Le Berre, A. Yamada, L. Reck, Y. Chen, D. Baigl, Electroformation of giant [213] J.C. Stachowiak, D.L. Richmond, T.H. Li, F. Brochard-Wyart, D.A. Fletcher, Inkjet
phospholipid vesicles on a silicon substrate: advantages of controllable surface formation of unilamellar lipid vesicles for cell-like encapsulation, Lab Chip 9
properties, Langmuir 24 (2008) 2643–2649. (2009) 2003–2009.
[185] A. Diguet, M. Le Berre, Y. Chen, D. Baigl, Preparation of phospholipid multilayer [214] S. Ota, S. Yoshizawa, S. Takeuchi, Microfluidic formation of monodisperse,
patterns of controlled size and thickness by capillary assembly on a microstructured cell‐sized, and unilamellar vesicles, Angew. Chem. Int. Ed. 48 (2009)
substrate, Small 5 (2009) 1661–1666. 6533–6537.
[186] P. Girard, J. Pécréaux, G. Lenoir, P. Falson, J.L. Rigaud, P. Bassereau, A new method [215] T. Kurakazu, S. Takeuchi, Generation of lipid vesicles using microfluidic T-junctions
for the reconstitution of membrane proteins into giant unilamellar vesicles, with pneumatic valves, 23rd International Conference on Micro Electro Mechanical
Biophys. J. 87 (2004) 419–429. Systems (MEMS), IEEE, 2010, pp. 1115–1118.
[187] S. Aimon, J. Manzi, D. Schmidt, J.A.P. Larrosa, P. Bassereau, G.E.S. Toombes, [216] A. Utada, E. Lorenceau, D. Link, P. Kaplan, H. Stone, D. Weitz, Monodisperse double
Functional reconstitution of a voltage-gated potassium channel in giant unilamellar emulsions generated from a microcapillary device, Science 308 (2005) 537–541.
vesicles, PLoS One 6 (2011) e25529. [217] Y.-C. Tan, K. Hettiarachchi, M. Siu, Y.-R. Pan, A.P. Lee, Controlled microfluidic
[188] M. Criado, B.U. Keller, A membrane fusion strategy for single-channel recordings of encapsulation of cells, proteins, and microbeads in lipid vesicles, J. Am. Chem.
membranes usually non-accessible to patch-clamp pipette electrodes, FEBS lett. Soc. 128 (2006) 5656–5658.
224 (1987) 172–176. [218] S.-Y. Teh, R. Khnouf, H. Fan, A.P. Lee, Stable, biocompatible lipid vesicle generation by
[189] Y.C. Lin, M. Li, Y.T. Wang, T.H. Lai, J.T. Chiang, K.S. Huang, A new method for solvent extraction-based droplet microfluidics, Biomicrofluidics 5 (2011) 044113.
the preparation of self-assembled phospholipid microtubes using [219] D. Wu, Y. Luo, X. Zhou, Z. Dai, B. Lin, Multilayer poly (vinyl alcohol)‐adsorbed
microfluidic technology, in: TRANSDUCERS'05 (Ed.), The 13th International coating on poly (dimethylsiloxane) microfluidic chips for biopolymer separation,
Conference on Solid-State Sensors, Actuators and Microsystems, IEEE, Electrophoresis 26 (2005) 211–218.
2005, pp. 1592–1595. [220] S.-Y. Teh, A.P. Lee, Microfluidic double emulsions for the formation of lipid vesicles
[190] Y.C. Lin, K.S. Huang, J.T. Chiang, C.H. Yang, T.H. Lai, Manipulating self-assembled and the controlled encapsulation of cells, 13th Int. Conf. Miniaturized Systems for
phospholipid microtubes using microfluidic technology, Sensors Actuators B Chemistry and Life Sciences, Jeju, Korea, 2009, pp. 1353–1355.
Chem. 117 (2006) 464–471. [221] R.T. Davies, D. Kim, J. Park, Formation of liposomes using a 3D flow focusing
[191] T. Osaki, K. Kuribayashi, R. Kawano, H. Sasaki, K. Kamiya, S. Takeuchi, Selective microfluidic device with spatially patterned wettability by corona discharge,
lipid-patterning for heterologous giant liposome array, 15th International J. Micromech. Microeng. 22 (2012) 055003.
Conference on Miniaturized Systems for Chemistry and Life Sciences, RSC, Seattle, [222] M. Seo, C. Paquet, Z. Nie, S. Xu, E. Kumacheva, Microfluidic consecutive
Washington, USA, 2011, pp. 1137–1139. flow-focusing droplet generators, Soft Matter 3 (2007) 986–992.
[192] T. Osaki, K. Kuribayashi-Shigetomi, R. Kawano, H. Sasaki, S. Takeuchi, Uniformly-sized [223] S. Pautot, B.J. Frisken, D. Weitz, Production of unilamellar vesicles using an inverted
giant liposome formation with gentle hydration, 24th International Conference on emulsion, Langmuir 19 (2003) 2870–2879.
Micro Electro Mechanical Systems (MEMS), IEEE, 2011, pp. 103–106. [224] K. Nishimura, H. Suzuki, T. Toyota, T. Yomo, Size control of giant unilamellar
[193] P.S. Dittrich, P. Renaud, A. Manz, On-chip extrusion of lipid vesicles and tubes vesicles prepared from inverted emulsion droplets, J. Colloid Interface Sci. 376
through microsized apertures, Lab Chip 6 (2006) 488–493. (2012) 119–125.
[194] O. Rossier, D. Cuvelier, N. Borghi, P. Puech, I. Derényi, A. Buguin, P. Nassoy, F. [225] T. Hamada, Y. Miura, Y. Komatsu, Y. Kishimoto, M.d. Vestergaard, M. Takagi,
Brochard-Wyart, Giant vesicles under flows: extrusion and retraction of tubes, Construction of asymmetric cell-sized lipid vesicles from lipid-coated water-in-oil
Langmuir 19 (2003) 575–584. microdroplets, J. Phys. Chem. B 112 (2008) 14678–14681.
[195] A. Jahn, J.E. Reiner, W.N. Vreeland, D.L. DeVoe, L.E. Locascio, M. Gaitan, Preparation [226] S.-Y. Teh, R. Lin, L.-H. Hung, A.P. Lee, Droplet microfluidics, Lab Chip 8 (2008)
of nanoparticles by continuous-flow microfluidics, J. Nanoparticle Res. 10 (2008) 198–220.
925–934. [227] P.C. Hu, S. Li, N. Malmstadt, Microfluidic fabrication of asymmetric giant lipid
[196] R. Wi, Y. Oh, C. Chae, D.H. Kim, Formation of liposome by microfluidic flow focusing vesicles, ACS Appl. Mater. Interfaces 3 (2011) 1434.
and its application in gene delivery, Kor. Aust. Rheol. J. 24 (2012) 129–135. [228] S. Matosevic, B.M. Paegel, Stepwise synthesis of giant unilamellar vesicles on a
[197] J.M. Zook, W.N. Vreeland, Effects of temperature, acyl chain length, and flow-rate microfluidic assembly line, J. Am. Chem. Soc. 133 (2011) 2798.
ratio on liposome formation and size in a microfluidic hydrodynamic focusing [229] W.J. Duncanson, T. Lin, A.R. Abate, S. Seiffert, R.K. Shah, D.A. Weitz, Microfluidic
device, Soft Matter 6 (2010) 1352–1360. synthesis of advanced microparticles for encapsulation and controlled release,
[198] J. Leng, S.U. Egelhaaf, M.E. Cates, Kinetics of the micelle-to-vesicle transition: Lab Chip 12 (2012) 2135–2145.
aqueous lecithin–bile salt mixtures, Biophys. J. 85 (2003) 1624–1646. [230] H.C. Shum, Y.J. Zhao, S.H. Kim, D.A. Weitz, Multicompartment polymersomes from
[199] A. Jahn, F. Lucas, R.W. Wepf, P.S. Dittrich, Freezing continuous-flow self-assembly in double emulsions, Angew. Chem. 123 (2011) 1686–1689.
a microfluidic device: towards imaging of liposome formation, Langmuir 29 (2013) [231] H.C. Shum, J.-W. Kim, D.A. Weitz, Microfluidic fabrication of monodisperse
1717–1723. biocompatible and biodegradable polymersomes with controlled permeability,
[200] M.J. Kennedy, H.D. Ladouceur, T. Moeller, D. Kirui, C.A. Batt, Analysis of a laminar-flow J. Am. Chem. Soc. 130 (2008) 9543–9549.
diffusional mixer for directed self-assembly of liposomes, Biomicrofluidics 6 (2012) [232] A. Seth, G. Béalle, E. Santanach‐Carreras, A. Abou‐Hassan, C. Ménager, Design of
044119. vesicles using capillary microfluidic devices: from magnetic to multifunctional
[201] M.J. Kennedy, S.J. Stelick, S.L. Perkins, L. Cao, C.A. Batt, Hydrodynamic focusing with vesicles, Adv. Mater. 24 (2012) 3544–3548.
a microlithographic manifold: controlling the vertical position of a focused sample, [233] S.H. Kim, J.W. Kim, D.H. Kim, S.H. Han, D.A. Weitz, Polymersomes containing
Microfluid. Nanofluid. 7 (2009) 569–578. a hydrogel network for high stability and controlled release, Small 9 (2012)
[202] S. Phapal, P. Sunthar, Influence of micro-mixing on the size of liposomes 124–131.
self-assembled from miscible liquid phases, Chem. Phys. Lipids 172–173 (2013) [234] A. Perro, C.l. Nicolet, J. Angly, S.b. Lecommandoux, J.-F.o. Le Meins, A. Colin,
20–30. Mastering a double emulsion in a simple co-flow microfluidic to generate complex
[203] X. Huang, R. Caddell, B. Yu, S. Xu, B. Theobald, L.J. Lee, R.J. Lee, Ultrasound-enhanced polymersomes, Langmuir 27 (2010) 9034–9042.
microfluidic synthesis of liposomes, Anticancer Res. 30 (2010) 463–466. [235] S.-H. Kim, H.C. Shum, J.W. Kim, J.-C. Cho, D.A. Weitz, Multiple polymersomes
[204] K. Yamashita, M.P.B. Nagata, M. Miyazaki, H. Nakamura, H. Maeda, Homogeneous for programmed release of multiple components, J. Am. Chem. Soc. 133 (2011)
and reproducible liposome preparation relying on reassembly in microchannel 15165–15171.
laminar flow, Chem. Eng. J. 165 (2010) 324–327. [236] S.-H. Kim, J. Nam, J.W. Kim, D.-H. Kim, S.H. Han, D.A. Weitz, Formation of
[205] K. Hettiarachchi, E. Talu, M.L. Longo, P.A. Dayton, A.P. Lee, On-chip generation polymersomes with double bilayers templated by quadruple emulsions, Lab Chip
of microbubbles as a practical technology for manufacturing contrast agents for 13 (2013) 1351–1356.
ultrasonic imaging, Lab Chip 7 (2007) 463–468. [237] X. Zhang, S.J. Haswell, Materials matter in microfluidic devices, MRS Bull. 31 (2006)
[206] E. Talu, K. Hettiarachchi, R.L. Powell, A.P. Lee, P.A. Dayton, M.L. Longo, Maintaining 95–99.
monodispersity in a microbubble population formed by flow-focusing, Langmuir [238] J. Thiele, A.R. Abate, H.C. Shum, S. Bachtler, S. Förster, D.A. Weitz, Fabrication of
24 (2008) 1745–1749. polymersomes using double‐emulsion templates in glass‐coated stamped
[207] K. Hettiarachchi, S. Zhang, S. Feingold, A.P. Lee, P.A. Dayton, Controllable microfluidic devices, Small 6 (2010) 1723–1727.
microfluidic synthesis of multiphase drug‐carrying lipospheres for site‐targeted [239] S.-H. Kim, J.W. Kim, D.-H. Kim, S.-H. Han, D.A. Weitz, Enhanced-throughput
therapy, Biotechnol. Prog. 25 (2009) 938–945. production of polymersomes using a parallelized capillary microfluidic device,
[208] D. Carugo, D.N. Ankrett, P. Glynne-Jones, L. Capretto, R.J. Boltryk, X. Zhang, P.A. Microfluid. Nanofluid. (2012) 1–6.
Townsend, M. Hill, Contrast agent-free sonoporation: the use of an ultrasonic [240] M.H. Lee, V. Prasad, D. Lee, Microfluidic fabrication of stable nanoparticle-shelled
standing wave microfluidic system for the delivery of pharmaceutical agents, bubbles, Langmuir 26 (2009) 2227–2230.
Biomicrofluidics 5 (2011) 044108. [241] J. Thiele, D. Steinhauser, T. Pfohl, S. Förster, Preparation of monodisperse
[209] K. Funakoshi, H. Suzuki, S. Takeuchi, Formation of giant lipid vesiclelike block copolymer vesicles via flow focusing in microfluidics, Langmuir 26 (2010)
compartments from a planar lipid membrane by a pulsed jet flow, J. Am. Chem. 6860–6863.
Soc. 129 (2007) 12608–12609. [242] L. Brown, S.L. McArthur, P.C. Wright, A. Lewis, G. Battaglia, Polymersome production
[210] H.C. Shum, D. Lee, I. Yoon, T. Kodger, D.A. Weitz, Double emulsion templated on a microfluidic platform using pH sensitive block copolymers, Lab Chip 10 (2010)
monodisperse phospholipid vesicles, Langmuir 24 (2008) 7651–7653. 1922–1928.
[211] J.C. Stachowiak, D.L. Richmond, T.H. Li, A.P. Liu, S.H. Parekh, D.A. Fletcher, [243] R. Thiermann, W. Mueller, A. Montesinos, D. Metzke, P. Löb, V. Hessel, M. Maskos,
Unilamellar vesicle formation and encapsulation by microfluidic jetting, Proc. Size controlled polymersomes by continuous self-assembly in micromixers,
Natl. Acad. Sci. 105 (2008) 4697–4702. Polymer 53 (2012) 2205–2210.
L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532 1531

[244] E. Amstad, S.H. Kim, D.A. Weitz, Photo‐and thermoresponsive polymersomes for [273] N.G. Green, H. Morgan, Dielectrophoresis of submicrometer latex spheres. 1.
triggered release, Angew. Chem. 124 (2012) 12667–12671. Experimental results, J. Phys. Chem. B 103 (1999) 41–50.
[245] F. Meng, Z. Zhong, J. Feijen, Stimuli-responsive polymersomes for programmed [274] C. Zhang, K. Khoshmanesh, A. Mitchell, K. Kalantar-Zadeh, Dielectrophoresis for
drug delivery, Biomacromolecules 10 (2009) 197–209. manipulation of micro/nano particles in microfluidic systems, Anal. Bioanal.
[246] M.-H. Li, P. Keller, Stimuli-responsive polymer vesicles, Soft Matter 5 (2009) Chem. 396 (2010) 401–420.
927–937. [275] M. Dürr, J. Kentsch, T. Müller, T. Schnelle, M. Stelzle, Microdevices for manipulation
[247] C. Martino, S.H. Kim, L. Horsfall, A. Abbaspourrad, S.J. Rosser, J. Cooper, D.A. Weitz, and accumulation of micro‐and nanoparticles by dielectrophoresis, Electrophoresis
Protein expression, aggregation, and triggered release from polymersomes as 24 (2003) 722–731.
artificial cell‐like structures, Angew. Chem. Int. Ed. 51 (2012) 6416–6420. [276] H. Morgan, D. Holmes, N.G. Green, 3D focusing of nanoparticles in microfluidic
[248] B. Johnson, R. Prud'homme, Flash nanoprecipitation of organic actives and block channels, IEEE Proc. Nanobiotechnol. 150 (2003) 76–81.
copolymers using a confined impinging jets mixer, Aust. J. Chem. 56 (2003) [277] C. Zhang, K. Khoshmanesh, F. Tovar-Lopez, A. Mitchell, W. Wlodarski, K.
1021–1024. Klantar-Zadeh, Dielectrophoretic separation of carbon nanotubes and polystyrene
[249] L. Capretto, W. Cheng, D. Carugo, M. Hill, X. Zhang, Microfluidic reactors for microparticles, Microfluid. Nanofluid. 7 (2009) 633–645.
controlled synthesis of polymeric micelles, J. Control. Release 148 (2010) e25–e26. [278] D.H. Shin, J.-E. Kim, H.C. Shim, J.-W. Song, J.-H. Yoon, J. Kim, S. Jeong, J. Kang, S. Baik,
[250] P.D. Fletcher, S.J. Haswell, X. Zhang, Electrokinetic control of a chemical reaction in C.-S. Han, Continuous extraction of highly pure metallic single-walled carbon
a lab-on-a-chip micro-reactor: measurement and quantitative modelling, Lab Chip nanotubes in a microfluidic channel, Nano Lett. 8 (2008) 4380–4385.
2 (2002) 102–112. [279] R. Krishnan, B.D. Sullivan, R.L. Mifflin, S.C. Esener, M.J. Heller, Alternating current
[251] L. Capretto, S. Mazzitelli, G. Colombo, R. Piva, L. Penolazzi, R. Vecchiatini, X. Zhang, electrokinetic separation and detection of DNA nanoparticles in high‐conductance
C. Nastruzzi, Production of polymeric micelles by microfluidic technology solutions, Electrophoresis 29 (2008) 1765–1774.
for combined drug delivery: application to osteogenic differentiation of [280] A.A. Kayani, A.F. Chrimes, K. Khoshmanesh, V. Sivan, E. Zeller, K.
human periodontal ligament mesenchymal stem cells (hPDLSC), Int. J. Pharm. Kalantar-zadeh, A. Mitchell, Interaction of guided light in rib polymer waveguides
440 (2) (2012) 195–206. with dielectrophoretically controlled nanoparticles, Microfluid. Nanofluid. 11
[252] M.M. Hasani-Sadrabadi, F.S. Majedi, J.J. VanDersarl, E. Dashtimoghadam, S.R. (2011) 93–104.
Ghaffarian, A. Bertsch, H. Moaddel, P. Renaud, Morphological tuning of polymeric [281] T. Osaki, S. Takeuchi, Dielectrophoresis-based liposome delivery to a planar lipid
nanoparticles via microfluidic platform for fuel cell applications, J. Am. Chem. membrane for efficient membrane protein reconstitution, 23rd International
Soc. 134 (2012) 18904–18907. Conference on Micro Electro Mechanical Systems (MEMS), IEEE, 2010, pp. 931–934.
[253] N. Anton, F. Bally, C.A. Serra, A. Ali, Y. Arntz, Y. Mely, M. Zhao, E. Marchioni, A. [282] I.-F. Cheng, V.E. Froude, Y. Zhu, H.-C. Chang, H.-C. Chang, A continuous
Jakhmola, T.F. Vandamme, A new microfluidic setup for precise control of the high-throughput bioparticle sorter based on 3D traveling-wave dielectrophoresis,
polymer nanoprecipitation process and lipophilic drug encapsulation, Soft Matter Lab Chip 9 (2009) 3193–3201.
(2012) 10628–10635. [283] K. Vijayan, Y. Geng, D. Discher, Electric field manipulation of charged copolymer
[254] H. Nagasawa, N. Aoki, K. Mae, Design of a new micromixer for instant mixing based worm micelles, J. Phys. Chem. B 110 (2006) 3831–3834.
on the collision of micro segments, Chem. Eng. Technol. 28 (2005) 324–330. [284] D. Issadore, T. Franke, K.A. Brown, R.M. Westervelt, A microfluidic
[255] P.M. Valencia, P.A. Basto, L. Zhang, M. Rhee, R. Langer, O.C. Farokhzad, R. Karnik, microprocessor: controlling biomimetic containers and cells using hybrid
Single-step assembly of homogenous lipid–polymeric and lipid–quantum integrated circuit/microfluidic chips, Lab Chip 10 (2010) 2937–2943.
dot nanoparticles enabled by microfluidic rapid mixing, ACS Nano 4 (2010) [285] A.H. Yang, S.D. Moore, B.S. Schmidt, M. Klug, M. Lipson, D. Erickson, Optical
1671–1679. manipulation of nanoparticles and biomolecules in sub-wavelength slot
[256] R.H. Fang, S. Aryal, C.-M.J. Hu, L. Zhang, Quick synthesis of lipid–polymer hybrid waveguides, Nature 457 (2009) 71–75.
nanoparticles with low polydispersity using a single-step sonication method, [286] S. Kulin, R. Kishore, K. Helmerson, L. Locascio, Optical manipulation and fusion of
Langmuir 26 (2010) 16958. liposomes as microreactors, Langmuir 19 (2003) 8206–8210.
[257] Y. Kim, B. Lee Chung, M. Ma, W.J. Mulder, Z.A. Fayad, O.C. Farokhzad, R. Langer, [287] D. Yin, E.J. Lunt, M.I. Rudenko, D.W. Deamer, A.R. Hawkins, H. Schmidt, Planar
Mass production and size control of lipid–polymer hybrid nanoparticles through optofluidic chip for single particle detection, manipulation, and analysis, Lab Chip
controlled microvortices, Nano Lett. 12 (2012) 3587–3591. 7 (2007) 1171–1175.
[258] D. Chen, K.T. Love, Y. Chen, A.A. Eltoukhy, C. Kastrup, G. Sahay, A. Jeon, Y. Dong, K.A. [288] Manipulation of phospholiposome in microfluidic channel using Lorentz force, in:
Whitehead, D.G. Anderson, Rapid discovery of potent siRNA-containing lipid NEMS'07 (Ed.), 2nd International Conference on Nano/Micro Engineered and
nanoparticles enabled by controlled microfluidic formulation, J. Am. Chem. Soc. Molecular Systems, IEEE, 2007, pp. 552–555.
134 (2012) 6948–6951. [289] S. Nomura, L. Liu, Y. Chen, H. Maruyama, F. Arai, Giant liposome sorting/collection
[259] N.M. Belliveau, J. Huft, P.J. Lin, S. Chen, A.K. Leung, T.J. Leaver, A.W. Wild, J.B. device: for individual analysis of artificial cell-models, in: MHS (Ed.), International
Lee, R.J. Taylor, Y.K. Tam, Microfluidic synthesis of highly potent limit-size Symposium on Micro-NanoMechatronics and Human Science, IEEE, 2009,
lipid nanoparticles for in vivo delivery of siRNA, Mol. Ther. Nucleic Acids 1 pp. 620–622.
(2012) e37. [290] H. Nuss, C. Chevallard, P. Guenoun, F. Malloggi, Microfluidic trap-and-release system
[260] C.-C. Hong, J.-W. Choi, C.H. Ahn, A novel in-plane passive microfluidic mixer with for lab-on-a-chip-based studies on giant vesicles, Lab Chip 12 (2012) 5257–5261.
modified Tesla structures, Lab Chip 4 (2004) 109–113. [291] P. Kuhn, K. Eyer, T. Robinson, F.I. Schmidt, J. Mercer, P.S. Dittrich, A facile protocol
[261] H. Debus, M. Beck-Broichsitter, T. Kissel, Optimized preparation of pDNA/poly for the immobilisation of vesicles, virus particles, bacteria, and yeast cells, Integr.
(ethylene imine) polyplexes using a microfluidic system, Lab Chip 12 (2012) Biol. 4 (2012) 1550–1555.
2498–2506. [292] T. Das, T.K. Maiti, S. Chakraborty, Nanodomain stabilization dynamics in plasma
[262] O. Boussif, F. . Lezoualc'h, M.A. Zanta, M.D. Mergny, D. Scherman, B. Demeneix, J.-P. membranes of biological cells, Phys. Rev. E. 83 (2011) 021909.
Behr, A versatile vector for gene and oligonucleotide transfer into cells in culture [293] T. Das, T. K Maiti, S. Chakraborty, Flow shear induced changes in membrane fluidity:
and in vivo: polyethylenimine, Proc. Natl. Acad. Sci. 92 (1995) 7297–7301. dependence on cell-substrate adhesion strength, Curr. Anal. Chem. 9 (2013) 9–15.
[263] J.-Y. Cherng, H. Talsma, R. Verrijk, D. Crommelin, W. Hennink, The effect of [294] G. Kwon, Polymeric micelles for delivery of poorly water-soluble compounds, Crit.
formulation parameters on the size of poly ((2-dimethylamino) ethyl Rev. Ther. Drug Carrier Syst. 20 (2003) 357–403.
methacrylate)-plasmid complexes, Eur. J. Pharm. Biopharm. 47 (1999) 215–224. [295] R. Lund, L. Willner, M. Monkenbusch, P. Panine, T. Narayanan, J. Colmenero, D.
[264] Y.-P. Ho, H.H. Chen, K.W. Leong, T.-H. Wang, The convergence of quantum-dot- Richter, Structural observation and kinetic pathway in the formation of polymeric
mediated fluorescence resonance energy transfer and microfluidics for micelles, Phys. Rev. Lett. 102 (2009) 188301.
monitoring DNA polyplex self-assembly in real time, Nanotechnology 20 [296] F. Bally, D.K. Garg, C.A. Serra, Y. Hoarau, N. Anton, C. Brochon, D. Parida, T. Vandamme,
(2009) 095103. G. Hadziioannou, Improved size-tunable preparation of polymeric nanoparticles by
[265] Y.-P. Ho, H.H. Chen, K.W. Leong, T.-H. Wang, Combining QD-FRET and microfluidics microfluidic nanoprecipitation, Polymer 53 (2012) 5045–5051.
to monitor DNA nanocomplex self-assembly in real-time, J. Vis. Exp. (2009) 1432. [297] C. Mora-Huertas, H. Fessi, A. Elaissari, Influence of process and formulation
[266] B. Yu, J. Zhu, W. Xue, Y. Wu, X. Huang, L.J. Lee, R.J. Lee, Microfluidic assembly of parameters on the formation of submicron particles by solvent displacement and
lipid-based oligonucleotide nanoparticles, Anticancer Res. 31 (2011) 771–776. emulsification–diffusion methods: critical comparison, Adv. Colloid Interfaces Sci.
[267] A.K.K. Leung, I.M. Hafez, S. Baoukina, N.M. Belliveau, I.V. Zhigaltsev, E. 163 (2011) 90–122.
Afshinmanesh, D.P. Tieleman, C.L. Hansen, M.J. Hope, P.R. Cullis, Lipid nanoparticles [298] R.C.S. Bicudo, M.H.A. Santana, Production of hyaluronic acid (Ha) nanoparticles by
containing siRNA synthesized by microfluidic mixing exhibit an electron-dense a continuous process inside microchannels: effects of non-solvents, organic phase
nanostructured core, J. Phys. Chem. C 116 (2012) 18440–18450. flow rate, and Ha concentration, Chem. Eng. Sci. 84 (2012) 134–141.
[268] C.G. Koh, X. Zhang, S. Liu, S. Golan, B. Yu, X. Yang, J. Guan, Y. Jin, Y. Talmon, N. [299] T. Cubaud, T. Mason, Formation of miscible fluid microstructures by hydrodynamic
Muthusamy, Delivery of antisense oligodeoxyribonucleotide lipopolyplex focusing in plane geometries, Phys. Rev. E. 78 (2008) 56308.
nanoparticles assembled by microfluidic hydrodynamic focusing, J. Control. [300] D. Hertzog, X. Michalet, M. Jäger, X. Kong, J. Santiago, S. Weiss, O. Bakajin,
Release 141 (2010) 62–69. Femtomole mixer for microsecond kinetic studies of protein folding, Anal. Chem.
[269] G. Birnbaumer, S. Küpcü, C. Jungreuthmayer, L. Richter, K. Vorauer-Uhl, A. Wagner, 76 (2004) 7169–7178.
C. Valenta, U. Sleytr, P. Ertl, Rapid liposome quality assessment using a [301] F. Schutze, B. Stempfle, C. Jungst, D. Woll, A. Zumbusch, S. Mecking, Fluorescent
lab-on-a-chip, Lab Chip 11 (2011) 2753–2762. conjugated block copolymer nanoparticles by controlled mixing, Chem. Commun.
[270] M.P. Hughes, H. Morgan, Dielectrophoretic characterization and separation of 48 (2012) 2104–2106.
antibody-coated submicrometer latex spheres, Anal. Chem. 71 (1999) 3441–3445. [302] A. Jahn, F. Lucas, R.W. Wepf, P.S. Dittrich, Freezing continuous-flow
[271] B. Çetin, D. Li, Dielectrophoresis in microfluidics technology, Electrophoresis 32 self-assembly in a microfluidic device: towards imaging of liposome formation,
(2011) 2410–2427. Langmuir 29 (5) (2013) 1717–1723.
[272] H. Morgan, M.P. Hughes, N.G. Green, Separation of submicron bioparticles by [303] C.-W. Wang, A. Bains, D. Sinton, M.G. Moffitt, Flow-directed assembly of block
dielectrophoresis, Biophys. J. 77 (1999) 516. copolymer vesicles in the lab-on-a-chip, Langmuir 28 (2012) 15756–15761.
1532 L. Capretto et al. / Advanced Drug Delivery Reviews 65 (2013) 1496–1532

[304] C.-W. Wang, D. Sinton, M.G. Moffitt, Flow-directed block copolymer [315] J. Kusunose, H. Zhang, M.K.J. Gagnon, T. Pan, S.I. Simon, K.W. Ferrara, Microfluidic
micelle morphologies via microfluidic self-assembly, J. Am. Chem. Soc. 133 (2011) system for facilitated quantification of nanoparticle accumulation to cells under
18853–18864. laminar flow, Ann. Biomed. Eng. 41 (2013) 89–99.
[305] C.G. Koh, X. Kang, Y. Xie, Z. Fei, J. Guan, B. Yu, X. Zhang, L.J. Lee, Delivery of [316] G. Sahay, W. Querbes, C. Alabi, A. Eltoukhy, S. Sarkar, C. Zurenko, E. Karagiannis, K.
polyethylenimine/DNA complexes assembled in a microfluidics device, Mol. Love, D. Chen, R. Zoncu, Efficiency of siRNA delivery by lipid nanoparticles is
Pharm. 6 (2009) 1333–1342. limited by endocytic recycling, Nat. Biotechnol. 31 (2013) 653–658.
[306] O.C. Farokhzad, R. Langer, Impact of nanotechnology on drug delivery, ACS Nano 3 [317] S.E. Gratton, P.A. Ropp, P.D. Pohlhaus, J.C. Luft, V.J. Madden, M.E. Napier, J.M.
(2009) 16–20. DeSimone, The effect of particle design on cellular internalization pathways,
[307] K. Park, Nanotechnology: what it can do for drug delivery, J. Control. Release 120 Proc. Natl. Acad. Sci. 105 (2008) 11613–11618.
(2007) 1. [318] J.A. Champion, S. Mitragotri, Shape induced inhibition of phagocytosis of polymer
[308] M. Kuang, H. Duan, J. Wang, M. Jiang, Structural factors of rigid-coil polymer pairs particles, Pharm. Res. 26 (2009) 244–249.
influencing their self-assembly in common solvent, J. Phys. Chem. B 108 (2004) [319] R.A. Petros, J.M. DeSimone, Strategies in the design of nanoparticles for therapeutic
16023–16029. applications, Nat. Rev. Drug Discov. 9 (2010) 615–627.
[309] X. Kang, C. Luo, Q. Wei, C. Xiong, Q. Chen, Y. Chen, Q. Ouyang, Mass production of [320] C.-W. Wang, D. Sinton, M.G. Moffitt, Morphological control via chemical and shear
highly monodisperse polymeric nanoparticles by parallel flow focusing system, forces in block copolymer self-assembly in the lab-on-chip, ACS Nano 7 (2013)
Microfluid. Nanofluid. 15 (2013) 337–345. 1424–1436.
[310] G. Tarabella, A.G. Balducci, N. Coppedè, S. Marasso, P. D'Angelo, S. Barbieri, M. [321] F.S. Majedi, M.M. Hasani-Sadrabadi, S.H. Emami, M.A. Shokrgozar, J.J.
Cocuzza, P. Colombo, F. Sonvico, R. Mosca, Liposome sensing and monitoring by VanDersarl, E. Dashtimoghadam, A. Bertsch, P. Renaud, Microfluidic assisted
organic electrochemical transistors integrated in microfluidics, Biochim. Biophys. self-assembly of chitosan based nanoparticles as drug delivery agents, Lab
Acta Gen. Subj. 1830 (2013) 4374–4380. Chip 13 (2013) 204–207.
[311] K. Kato, M. Koido, M. Kobayashi, T. Akagi, T. Ichiki, Statistical fluctuation in zeta [322] Y.-C. Hsieh, J.D. Zahn, Glucose recovery in a microfluidic microdialysis biochip,
potential distribution of nanoliposomes measured by on‐chip microcapillary Sensors Actuators B Chem. 107 (2005) 649–656.
electrophoresis, Electrophoresis 34 (2013) 1212–1218. [323] T. Laurell, F. Petersson, A. Nilsson, Chip integrated strategies for acoustic
[312] Z. Wang, H.-J. Wu, D.H. Fine, J. Schmulen, Y. Hu, B. Godin, J. Zhang, X. Liu, Ciliated separation and manipulation of cells and particles, Chem. Soc. Rev. 36 (2007)
micropillars for the microfluidic-based isolation of nanoscale lipid vesicles, Lab 492–506.
Chip 13 (2013) 2879. [324] M. Mijajlovic, D. Wright, V. Zivkovic, J. Bi, M. Biggs, Microfluidic hydrodynamic
[313] P.M. Valencia, M. Rhee, R. Langer, O. Farokhzad, R. Karnik, Merging ‘micro’ with focusing based synthesis of POPC liposomes for model biological systems, Colloids
‘nano’: on-chip high-throughput synthesis of polymeric nanoparticles for cancer Surf. B: Biointerfaces 104 (2012) 276–281.
therapy, 14th International Conference on Miniaturized Systems for Chemistry [325] K. Kamiya, R. Kawano, T. Osaki, S. Takeuchi, Vesicles in a vesicle: Formation of a
and Life Sciences, Groningen, The Netherlands, 2010. cell-sized vesicle containing small vesicles from two planar lipid bilayers using
[314] K. Churski, P. Korczyk, P. Garstecki, High-throughput automated droplet microfluidic pulsed jet flow, 26th International Conference on Micro Electro Mechanical
system for screening of reaction conditions, Lab Chip 10 (2010) 816–818. Systems (MEMS), IEEE, 2013, pp. 935–936.

You might also like