en
A
al
Arlette Vega-Gonza´lez a, Pascale Subra-Paternault a, Ana M. Lo´pez-Periago b,
by non-biodegradable and biodegradable polymers (polymethylmetacrylate/polycaprolactone). The influence of several
tuted by biopolymers or blends, which induces the natural extracellular matrix that surrounds cells in
the body. Hence, required characteristics include
an interconnected 3D macroporous network neces-
sary for cell proliferation, high microporosity essen-
tial for neovascularization, and high surface area
erved.
* Corresponding author. Tel.: +34 93 5801853.
E-mail address: conchi@icmab.es (C. Domingo).
Available online at www.sciencedirect.com
European Polymer Journal 44 (
EUROPEAN
POLYMER
0014-3057/$ - see front matter � 2008 Elsevier Ltd. All rights res
operating variables (e.g., liquid solution concentration or flow rate and nozzle design) on the polymers morphology
and properties was evaluated. For all studied systems, fibers with a rough textured surface and an extremely high surface
area in the order of 100–400 m2 g�1 were precipitated. Prepared materials have potential applications in tissue engineering,
since they have intrinsic advantages from a biomimetic approach.
� 2008 Elsevier Ltd. All rights reserved.
Keywords: PMMA; L-PLA; PCL; Blends; Fibers; Supercritical CO2
1. Introduction
Tissue engineering is evolving from the use of
implants that repair or replace damaged parts to
the use of three-dimensional (3D) scaffolds consti-
formation of new functional tissues either in vitro
or in vivo [1,2]. The architecture of a scaffold plays
an important role in modulating tissue growth and
response behavior of cultured cells. A correct archi-
tecture of a scaffold should be able of mimicking the
Carlos A. Garcı´a-Gonza´lez b, Concepcio´n Domingo b,*
aLaboratoire d’Inge´nierie des Mate´riaux et des Hautes Pressions, C.N.R.S., Institut Galile´e, Universite´ Paris 13,
99 Avenue Jean Baptiste Cle´ment, 93430 Villetaneuse, France
b Instituto de Ciencia de Materiales de Barcelona (CSIC), Campus de la UAB s/n, E-08193 Bellaterra, Spain
Received 19 November 2007; received in revised form 20 December 2007; accepted 4 January 2008
Available online 12 January 2008
Abstract
A supercritical carbon dioxide (SCCO2) antisolvent technique was used for the precipitation of several biopolymers into
fibers organized in a three-dimensional network. The work was first focused on separately processing either a biodegrad-
able (polycaprolactone or polylactic acid) or a non-biodegradable (polymethylmetacrylate) homopolymer. Second, we
established that the antisolvent supercritical technique can be also used to make fibrous networks of blends constitute
Supercritical CO2 antisolv
networks of L-PLA, PMM
for biomedic
doi:10.1016/j.eurpolymj.2008.01.009
t precipitation of polymer
and PMMA/PCL blends
applications
2008) 1081–1094
www.elsevier.com/locate/europolj
JOURNAL
1082 A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094
required to promote cell adhesion [3]. Currently,
laboratory designed scaffolds adopt forms ranging
from monolithic microcellular structures (sponges)
to networks of fibers [4–6].
Polymers used in tissue engineering applications
are typically homopolymers or copolymers of lactic
(L-PLA) and glycolic (PGA) acids, as well as poly
(e-caprolactone) (PCL), among other biodegradable
products [7–9]. Non-biodegradable materials such as
polymethylmethacrylate (PMMA) are also being
tested as scaffolds in order to achieve long-term
mechanical stability after implantation [10]. Further-
more, blends composed by biostable and biodegrad-
able polymers have recently gained significant
interests, since the final products can be provided
with specific advantages in degradation and mechan-
ical properties.
Polymeric scaffolds can be obtained by numerous
ways, using either conventional techniques such as
solvent casting, foaming, emulsions, physical separa-
tion and freeze drying [5,6,11,12], or by employing
advanced processing methods such as rapid proto-
typing technologies [13]. The main drawback of
using liquid approaches is that they involve organic
solvents and the residual molecules left in the poly-
mer after processing may be harmful to the trans-
planted cells. Technology based on supercritical
carbon dioxide (SCCO2) has been recently estab-
lished as an alternative to overcome some of the
problems associated with the use of traditional
organic solvents for pharmaceuticals and biomateri-
als preparation [14,15]. In this respect, different
approaches using SCCO2 technology for the pro-
cessing of a wide range of biopolymers, with an
excellent control on morphology, surface properties
and purity, have been reported [11,14,16–19]. For
instance, SCCO2 is widely used to produce polymer
foams by pressure-induced phase separation [17–23].
This method exploits the fact that SCCO2 is an excel-
lent non-solvating porogenic agent for amorphous
polymers, while being a poor solvent [20]. In spite
of this, when using SCCO2 as a porogenic agent, fre-
quently, the obtained microcellular foams did not
have a network of pores sufficiently interconnected
to allow an effective cell spreading [14,17,24,25].
Similar problems are also found when using poro-
genic organic liquids for the preparation of scaffolds
by the temperature quenching process [7,26].
Supercritical antisolvent spray processes have
also achieved considerable success in producing
polymers with different morphologies [15,27–31]. It
should be pointed out that most of the investiga-
tions described in the supercritical antisolvent liter-
ature deal with the production of fine particles,
while the production of fibers has been mainly con-
sidered as failed experiments. On the other hand, it
has been recently established that the use of polymer
fibers for tissue engineering applications has some
intrinsic advantages from a biomimetic approach,
since a wide variety of natural biomaterials pos-
sesses fibrous structures (silk, keratin, collagen,
cellulose, chitin, etc.). Moreover, the development
of new and versatile techniques for the production
of clean and template-free polymeric networks of
fibers, as an alternative to porous sponges in tissue
engineering, has reached a significant interest nowa-
days. Hence, the main objective of this work is to
follow a line of research that explore the possibilities
of using the SCCO2 semicontinuous antisolvent
(SAS) process for the preparation of intermingled
fibers of homopolymers (biodegradable L-PLA or
PCL and non-biodegradable PMMA) and their
blends with controlled morphology.
2. Experimental
2.1. Materials
Characteristics of the used polymers (L-PLA,
PCL, and PMMA) are shown in Table 1. Dichloro-
methane (DCM HPLC grade, Prolabo) and CO2
(99.95wt%, Air Liquide) were the used fluids.
2.2. Equipment and methods
Experiments were conducted in a SAS apparatus
shown schematically in Fig. 1, operated in a semi-
continuous mode. The spray chamber consisted on
a high pressure vessel (1, Autoclave Engineers) of
5 cm i.d. � 25 cm long. The vessel temperature was
controlled by heating jackets (2). At the bottom of
the vessel, a membrane filter placed on top of a
stainless steel frit of 2 lm porosity allowed the col-
lection of precipitated solids. The CO2 was first
cooled (3) and then pumped using a reciprocating
pump (4, Lewa EK3). The DCM solution of poly-
mer(s) to be precipitated was compressed using a
reciprocating dual-piston minipump (8, Milton
Roy LDC). The pressure inside the vessel was con-
trolled downstream with a micrometering valve (7).
In a typical experiment, once the vessel had
attained the desired temperature value (313 K), the
CO2 was introduced, keeping valve (6) closed, until
the selected pressure was reached (11 MPa). Then,
t (g m
Semicrystalline v = 50% Tg = 320 Tm = 449
PCL Powder Aldrich 14,000
A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094 1083
PMMA Beads Bonar polymers 300,000
Table 1
Characteristics of employed raw polymers (v = crystallinity)
Polymer Shape Supplier Molecular weigh
L-PLA Pellets Biovalley 100,000
valve (6) was opened and the system was allowed to
equilibrate maintaining the CO2 flow at the set value
of 100 mL min�1. Next, the polymer solution was
injected into the vessel, by opening valve (9),
through a nozzle. Two types of nozzles were used
(Fig. 1): a plain orifice (disk) and a conical spray
(swirl). In addition, two different injection systems
were studied. In the first one, a separate introduc-
tion of the liquid solution through a nozzle and
the CO2 through a 1/8
00 tubing was performed. In
the second one, a coaxial nozzle was used for the
introduction of both fluids concurrently. The simul-
taneous mass transfer of the SCCO2 and of the
liquid solvent from one phase to the other induced
the supersaturation of the polymer that precipi-
tated. Finally, the system was depressurized across
a metering valve (7). The fluid was separated in a
cyclone (10). The precipitated polymer was dried
using the SCCO2 flow.
2.3. Characterization
The morphology of the polymer samples was ana-
lyzed by scanning electron microscopy (SEM, Leica
5440). The composition of the processed solid
Fig. 1. SAS experimental set-up and schematic draws of the d
Semicrystalline v = 63% Tg = 211 Tm = 339
Amorphous Tg = 393
ol�1) Structure Thermal transitions (K)
samples was determined by 1H nuclear magnetic res-
onance (1H NMR Bruker ARX-300 MHz spectrom-
eter) in deuterated chloroform. The semicrystalline
L-PLA was analyzed by X-ray diffraction (XRD)
with a Rigaku Rotaflex RU200 B instrument, using
Cu Ka1 radiation. The glass (Tg) and/or melting
(Tm) temperatures of polymers were measured on a
differential scanning calorimeter (DSC, 822e/400
Mettler Toledo). Thermograms were obtained at a
heating rate of 10 K min�1 in the range 293–573 K
under N2 purge. The specific surface area (Sa) and
pore volume (Pv) of raw and prepared samples were
determined by low-temperature N2 adsorption,
using an ASAP 2000 Micromeritics INC. Previous
to measurements, samples were dried under reduced
pressure (<1 mPa) at 323 K for 48 h. Sa was deter-
mined by the BET-method (Brunauer–Emmett–
Teller) and Pv was estimated using the BJH-method
(Barret–Joyner–Halenda) [32]. The average pore
diameter (Dp) was calculated as 4Pv/Sa.
3. Results and discussion
The SAS process was first used to precipitate
the homopolymers L-PLA, PCL, and PMMA
isk (hand made) and swirl (Lechler, France, SA) nozzles.
separately. Further, two different compositions of
PCL in the PMMA/PCL blend were processed into
fibers (Table 2). This blend composition was chosen
because previous studies have demonstrated that the
combination of biodegradable PCL and biostable
PMMA could provide materials with good mechan-
ical integrity and biodegradation capabilities, being
an alternative to traditional extrudable copolymers
used in tissue regeneration [33].
For the SAS processing of polymers, the
described main parameter influencing product mor-
phology is the concentration of the liquid solution
[34–36]. The uses of diluted, semidiluted and con-
centrated solutions give place to different polymer
6 mL min�1 flow rate (samples PLA(1) and
PLA(2), respectively, in Table 2). It should be
pointed out that L-PLA polymer processed by
supercritical antisolvent techniques has been gener-
ally reported to precipitate as well defined indepen-
dent particles [27–29,37–39], but previous results
were obtained when working at much lower
L-PLA concentration in the liquid solution (0.1–
1 wt%) than the one used in this work (2.8 wt%).
Fig. 2 shows some SEM micrographs of the
obtained samples. The observed macrostructures
are a reflection of the liquid jet behavior (Fig. 2a
and b). A large spreading in the axial direction
was apparent in the micrographs of the L-PLA
on
1084 A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094
chains behavior. In the diluted regime, polymer
chains behave as independent entities and the poly-
mer usually precipitates as a powder. On the other
hand, in the semidiluted and concentrated regions,
the polymer chains are capable of entangling each
other strongly to form a fibrous structure in the pre-
cipitate. Since in this work the goal was to produce
fibers, we worked at relatively high polymer concen-
tration (semidiluted region) in the liquid solution
(Table 2). Spray conditions were chosen as the
important key parameters of the SAS process deter-
mining polymer fibers macro and microstructure.
Thus, the effects of modifying the liquid solution
flow rate and the nozzle design were analyzed for
each studied polymer.
3.1. L-PLA
3.1.1. Influence of liquid solution flow rate
To analyze the influence of the liquid solution
flow rate on polymer morphology, a 2.8 wt% L-
PLA polymer solution was sprayed into SCCO2,
through a plain orifice nozzle, at either 1 or
Table 2
Experimental conditions for the antisolvent experiments
Sample Polymer Polymer concentrati
(wt%)
PLA(1) L-PLA 2.8
PLA(2) L-PLA 2.8
PLA(3) L-PLA 2.8
PLA(4) L-PLA 2.8
PMMA PMMA 1
PCL PCL 0.8
Blend15(1) PMMA/15 wt% PCL 0.8
Blend15(2) PMMA/15 wt% PCL 0.2
Blend15(3) PMMA15 wt% PCL 0.8
Blend15(4) PMMA/15 wt% PCL 0.8
Blend30 PMMA30 wt% PCL 0.8
sample obtained at the low liquid flow rate of
1 mL min�1 giving place to a cobweb-like macro-
structure (Fig. 2a). In contrast, at the high liquid
flow rate of 6 mL min�1 viscous and inertial forces
offer significant resistance to jet break-up. This led
to the formation of a helical oscillation of the jet
around its own axis, which can be clearly observed
in Fig. 2b. These results are in accordance with pre-
vious works, which have shown that, in the semidi-
luted region, fibers and helical threads emerged
from the nozzle, spreading in both axial and radial
directions from the jet core [40]. At low liquid flow
rates, the jet disintegration into the SCCO2 has been
described to be characterized by axially symmetrical
disturbances, while the radial direction is preferred
at higher velocities [41].
A closer observation of the SEM photographs
indicated that the fibers were constituted by coa-
lesced polymer microparticles (Fig. 2c and d). The
presence of discrete particles suggests that the tran-
sition from diluted to semidiluted region occurs
close to the used polymer concentration [38]. Coa-
lescence of L-PLA particles seemed limited to phys-
Liquid flow rate
(mL min�1)
Liquid/CO2
addition mode
Nozzle, diameter
(lm)
1 Separate Disc, 100
6 Separate Disc, 100
3 Separate Swirl, 100
3 Coaxial Swirl, 200
1 Separate Swirl, 100
0.7 Separate Swirl, 100
1 Separate Swirl, 100
1 Separate Swirl, 100
6 Separate Swirl, 100
1 Separate Disc, 100
1 Separate Swirl, 100
Fig. 2. SEM images of L-PLA precipitated using the SAS process, 2.8 wt% of polymer in DCM, injected separately through a disc nozzle
at: (a, c) 1 mL min�1 (sample PLA(1)), and (b, d) 6 mL min�1 (sample PLA(2)).
Fig. 3. (a) XRD, and (b) DSC or raw L-PLA and PLA(1) sample.
A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094 1085
ical interaction. Indeed, the semicrystalline structure
of the L-PLA makes agglomeration from plasticiza-
tion difficult [34]. At 1 mL min�1 of liquid solution
flow rate, particles with a size of 2–3 lm were pre-
cipitated along with 0.5 lm particles (Fig. 2c). The
bimodal size distribution may suggest the formation
of particles in the jet and in the solution. At
6 mL min�1 of liquid solution flow rate, particles
with a size smaller than 1 lm were produced
(Fig. 2d). In this case, a high flow rate resulted on
enhanced turbulence and convective mass transfer
[29,42]. Therefore, smaller liquid droplet sizes were
formed resulting in considerably smaller particles
than in the run with the low liquid solution flow
rate.
Realistic models of fluids mixing have shown the
complexity of the theoretical work [40–44]. Those,
will be the subject of a further work and, therefore,
they are not deeply analyzed in this investigation,
where a more qualitative assessment of the effect
of a whole set of experimental conditions in precip-
itate characteristics was performed.
Fibers of L-PLA obtained at 1 mL min�1 flow
rate (sample PLA(1)) were further characterized
regarding the microstructure. X-ray diffraction
pattern of raw L-PLA displayed the characteristic
narrow diffraction lines at 16.5� and 19�, corre-
sponding to the a crystalline form (Fig. 3a). Despite
the very fast precipitation rate, XRD analysis
showed that the PLA(1) sample was still semicrys-
talline (Fig. 3a) by displaying the more intense dif-
fraction peak at 16.5�. Moreover, peak broadening
indicated small crystal size in the precipitated fibers.
DSC analysis (Fig. 3b) did not evidence significant
changes with respect to the raw material regarding
the melting point or melting enthalpy (compare
values in Tables 1 and 3). In addition, the values
of crystallinity estimated from the DSC data for
both raw and processed L-PLA were similar
Table 3
Melting point, crystallinity (calculated from the melting enthalpy)
and surface area of some representative prepared samples
Sample Tm (K) v (%) Sa (m
2 g�1)
PLA(1) 452 K (L-PLA) 48 60 ± 8
PMMA – – 150 ± 15
Blend15(2) 337 K (PCL) – 400 ± 25
Blend30 338 K (PCL) – 240 ± 30
DCM
1086 A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094
Fig. 4. SEM images of precipitated L-PLA, 2.8 wt% of polymer in
(sample PLA(3)), and (c, d) coaxially (sample PLA(4)).
, injected at 3 mL min�1 through a swirl nozzle: (a, b) separately
(about 50%). BET analysis revealed a specific sur-
face area value of �60 m2 g�1 for PLA(1) sample
(Table 3), which comprised both meso and macro-
pores.
3.1.2. Influence of the injection mode and nozzle
design
In order to examine the effect of the injection
mode in polymer morphology, the 2.8 wt% L-PLA
liquid solution was introduced into the precipitation
chamber through the swirl nozzle, at a flow rate of
3 mL min�1, either separately or co-currently with
the SCCO2 (samples PLA(3) and PLA(4), respec-
tively, in Table 2). The hydrodynamics of the differ-
ent used nozzles have been already described
previously [45]. SEM photographs of the sample
obtained using the separate injection mode
(Fig. 4a and b) showed coalesced L-PLA particles
with a similar macrostructure to that of the system
precipitated using the disk nozzle and a solution
flow rate of 1 mL min�1 (Fig. 2a and c). According
to that, no significant effect of the nozzle design
(disc or swirl) was observed. However, for the sys-
tem where the CO2 was injected co-currently in
the annular region over the liquid solution, smooth
fibers were formed (Fig. 4c and d) with different
morphology than those obtained using the separate
injection mode (Fig. 4a and b). The co-currently
injection mode enhanced the mixing of the system
[29,44], and L-PLA apparently precipitated into
fibers before individual droplets could be formed
in the liquid jet. The SEM images of sample
PLA(4) showed some 0.5 lm beads deposited in
fibers (Fig. 4d).
3.2. PMMA
Under the working experimental conditions of
polymer concentration, pressure and temperature
(Table 2), PMMA polymer always precipitated as
DCM
A. Vega-Gonza´lez et al. / European Polymer Journal 44 (2008) 1081–1094 1087
Fig. 5. SEM images of precipitated PMMA, 1 wt% of polymer in
to different magnifications.
, injected at 1 mL min�1 through a swirl nozzle. (a–c) correspond
a fibrous network with a similar appearance to that
shown in the SEM images of Fig. 5. The observed
macrostructure can be described as an open airy net-
work (Fig. 5a) composed of very small subfibers of
less than 0.5 lm diameter (Fig. 5b). Once more, the
formation of fine particles appears to indicate that
the transition from diluted to semidiluted region
occurs close to the working polymer concentration.
For PMMA the agglomeration and flocculation
was extended in comparison with L-PLA due to the
PMMA amorphous nature and SCCO2 plasticiza-
tion effect. A value of 150 m2 g�1 of specific