pharmaceutics

Article
Printability of External and Internal Structures Based
on Digital Light Processing 3D Printing Technique
Yan Yang, Yanjun Zhou, Xiao Lin, Qingliang Yang and Gengshen Yang *
College of Pharmaceutical Science, Zhejiang University of Technology, Hangzhou 310058, China;
yangyan10@zjut.edu.cn (Y.Y.); zhouyanjun0521@163.com (Y.Z.); 17816875532@163.com (X.L.);
qyang@zjut.edu.cn (Q.Y.)
* Correspondence: yanggs@zjut.edu.cn; Tel.: +86-0571-88871077




Received: 29 January 2020; Accepted: 25 February 2020; Published: 28 February 2020


Abstract: The high printing efficiency and easy availability of desktop digital light processing (DLP)
printers have made DLP 3D printing a promising technique with increasingly broad application
prospects, particularly in personalized medicine. The objective of this study was to fabricate and
evaluate medical samples with external and internal structures using the DLP technique. The influence
of different additives and printing parameters on the printability and functionality of this technique
was thoroughly evaluated. It was observed that the printability and mechanical properties of external
structures were affected by the poly(ethylene glycol) diacrylate (PEGDA) concentration, plasticizers,
layer height, and exposure time. The optimal printing solutions for 3D external and internal structures
were 100% PEGDA and 75% PEGDA with 0.25 mg/mL tartrazine, respectively. And the optimal layer
height for 3D external and internal structures were 0.02 mm and 0.05 mm, respectively. The optimal
sample with external structures had an adequate drug-loading ability, acceptable sustained-release
characteristics, and satisfactory biomechanical properties. In contrast, the printability of internal
structures was affected by the photoabsorber, PEGDA concentration, layer height, and exposure time.
The optimal samples with internal structures had good morphology, integrity and perfusion behavior.
The present study showed that the DLP printing technique was capable of fabricating implants for
drug delivery and physiological channels for in vivo evaluation.

Keywords: digital light processing; external and internal structures; printability; implants;
physiological channels

1. Introduction
Digital light processing (DLP) is a 3D printing technique, in which photopolymer monomers are
crosslinked layer by layer. Unlike the stereolithography apparatus technique, which is based on a
point laser, the DLP technique uses a digital projector as a light source, leading to significantly reduced
printing time [1]. To date, DLP printing has been successfully used in biomedical fields, such as dental
prothesis and tissue engineering [2], revealing its feasibility for developing a digital workflow for
personalized medicine.
It is noteworthy that the DLP technique is capable of fabricating samples with predetermined
specific external structures for personalized implants. For example, it could be used to generate
dental models with high accuracy [3], and cartilage with proper and satisfactory biomechanical
properties [4]. In addition, the DLP technique could also produce samples with predetermined
internal structures for highly complex organs, including the trachea, heart, lung, and vasculatures [5].
Recently, the DLP technique has been used to prepare solid preparations for oral drug administrations.
Paracetamol tablets with controllable drug-release behavior were printed following the adding of
suitable pore-forming agents [1], and theophylline tablets with programed drug-release profiles were

Pharmaceutics 2020, 12, 207; doi:10.3390/pharmaceutics12030207 www.mdpi.com/journal/pharmaceutics

Pharmaceutics 2020, 12, 207 2 of 16

fabricated with varying surface areas [6]. However, samples with external and internal structures
had different requirements on morphology and function (e.g., drug dissolution and biomechanical
properties). Therefore, it is necessary to further study the printability of different structures based on
DLP printing.
Generally, the essential factor affecting the printability of DLP is the photopolymerizable system,
which is usually composed of photopolymer monomer, photoinitiator and other additives (i.e.,
photoabsorbers, pore-forming agents and plasticizers). To date, commonly used biocompatible
photopolymers, such as poly(ethylene glycol) diacrylate (PEGDA) [7], poly(ethylene glycol)
dimethacrylate [8], gelatin methacrylate [9], and poly(propylene fumarate) [10] have been successfully
used in DLP printing. Among them, hydrophilic PEGDA was attractive material owing to its tunable
mechanical and biological properties [11]. Moreover, the concentration of PEGDA tremendously
affected the fidelity, mechanical strength and release properties of the printed samples [6]. Furthermore,
a photoinitiator determined the light wavelength of the photocuring reaction [12] and played an
important role in printing efficiency [13]. In contrast, photoabsorbers were able to delay the photocuring
reaction and improve the printability of internal channels [14]. In addition, pore-forming agents
(e.g., PEG, NaCl and Mannitol) accelerated and adjusted drug release from the highly crosslinked
structure [1], whereas plasticizers (e.g., PEG and glycerol) promoted the flexibility and biomechanical
properties of the printed samples [15]. Considering that external and internal structures have different
requirements on the ingredients of a photopolymerizable system, further studies on their printability
are necessary.
In addition, the printing parameters also exert a great impact on the photocuring process and
printing fidelity. For example, exposure time, voxel depth, and over-curing depth significantly
affected the curing behavior of different photosensitive resins [16]. Although sufficient light intensity
and exposure time were essential for photocuring, excessive exposure time might reduce printing
resolution owing to the overgrowth caused by light scattering [17]. Reduced layer height led to printed
channels with higher circularity and greater geometric fidelity to those of the original model [14].
The UV post-curing process also led to shape distortion and thus reduction of dimension accuracy [18].
Furthermore, the printing parameters greatly impact the mechanical and dissolution properties of
the produced samples. For example, pattern ratio, orientation and waviness of models affected
the mechanical behavior of the printed hydrogels [19], whereas the specific surface area of models
modulated the drug-release behavior of printed tablets [20]. Considering the various morphological
and functional requirements of implants and physiological channels, further studies on the effect of
printing parameters and model design are important.
This study aimed to fabricate 3D models with predesigned external and internal structures by
DLP using PEGDA as the photopolymer monomer. Furthermore, the effects of additives, printing
parameters, and model design on the printability and functionality of external and internal structures
were thoroughly evaluated.

2. Material and Methods

2.1. Material
PEGDA (400, No. 20180605) was provided by Ryoji Chemical Co. (Shanghai, China).
Diphenyl (2,4,6-trimethylbenzoyl) phosphine oxide (DPPO, No. C1809028), tartrazine (No. F1926074)
and PEG (300, No. L1707087) were purchased from Aladdin Biochemical Technology Co., Ltd.
(Shanghai, China). Methylene blue (No. 20130321) was obtained from SSS Reagent Co., Ltd. (Shanghai,
China). Coccine was obtained from Dyestuffs Research Institute Co., Ltd. (Shanghai, China).
Diclofenac sodium (DS, No. 142436317488) was purchased from Sendi Biological Co., Ltd. (Shenzhen,
China). Ibuprofen (No. KH23) was purchased from Dibo Biotechnology Co., Ltd. (Shanghai, China).
Pharmaceutics 2020, 12, 207 3 of 16

2.2. Printability Evaluation of External and Internal Structures

Pharmaceutics 2020, 12, 207 3 of 16
PEGDA and DPPO (1% of PEGDA amount) were mixed to prepare the photocuring solution.
Water
Methylenewas blue,
added to prepare
coccine, printing were
and tartrazine materials
added with PEGDA at concentrations
as photoabsorbers to improve the of printability
25% to 100%. of
Methylene
internal blue, coccine,
structures. and tartrazine
The printing were
materials wereadded
added asinto
photoabsorbers
the resin tank toof
improve the printability
a DLP printer (NOVA3D of
internal
101; structures.Inc.
NovRobotics, TheShenzhen,
printing materials were added
China) equipped withinto
a 405thenm
resin
lasertank of a DLPModels
generator. printerwere(NOVA3D
sliced
101; NovRobotics,
using the CreationInc. Shenzhen,
Workshop China)with
software equipped
a layerwith a 405ofnm
height laser generator.
0.02–0.05 mm. TheModels
exposure were sliced
time for
using
the the Creation
bottom Workshop
three layers software
was 10–90 with
s, and a layer time
exposure heightforofthe
0.02–0.05 mm. The
other layers wasexposure
2–12 s. time for the
bottomAs three
shown layers was 10–90
in Figure s, and and
1a, convex exposure
concavetimecones
for the
withother layers was
different 2–12 s.(Ф = 2.0, 1.75, 1.5,
diameters
As shown in Figure 1a, convex and concave cones with different
1.25, 1.0, and 0.75 mm) were printed to assess the printing accuracy of 3D external diameters (Φ = 2.0, 1.75, 1.5, 1.25,
structures. The
1.0, andof0.75
height themm)
convexwere printed
cone or theto depth
assess of
thethe
printing
concave accuracy
cone was of 3D externalusing
measured structures. The height
a micrometer. Theof
the convex cone or the depth of the concave cone was measured using a micrometer.
red solution was added to distinguish the boundary of the concave cones. As shown in Figure 1b, The red solution
was added
vertical andto distinguish
horizontal the boundary
channels of the concave
with different diameters cones. As shown
(Ф = 1.75, in Figure
1.5, 1.25, 1b,and
1.0, 0.75, vertical and
0.5 mm)
horizontal
were channels
printed with
to assess thedifferent
printingdiameters
accuracy (Φ = 1.75,
of 3D 1.5, 1.25,
internal 1.0, 0.75,
structures. The and
red0.5 mm) were
solution wasprinted
added to
assess the printing
distinguish accuracyof
the boundary of the
3D internal
channels, structures.
and the The red solution
length of channels was added to distinguish
was measured usingthe a
boundary of the
micrometer. Thechannels,
printed and
lengththeto
length
designof channels was measured
length ratio was defined usingas apatency
micrometer.
degree,Theand
printed
the
length to design
minimum diameterlength
of ratio was defined
the channel withas a patency
patency degree,
degree and the minimum
of 100% diameter
was recorded of the
as the channel
minimum
with a patency degree
printable diameter (Dmin). of 100% was recorded as the minimum printable diameter (D min ).

Figure 1. Models
Models for
for printability
printability evaluation
evaluation and tensile test: (a)
(a) convex
convex and
and concave
concave cones;
cones; (b)
(b) vertical
vertical
and horizontal
horizontal channels;
channels; (c)
(c) pentagram
pentagram and
and grid;
grid; (d)
(d) cylinder
cylinder and
and cuboid;
cuboid; and
and (e)
(e) dumbbell-shaped.
dumbbell-shaped.
∆L: inner arc length; ΔA:
ΔL: ∆A: cavity area.

As illustrated
As illustrated inin Figure
Figure 1c,
1c, pentagram
pentagram and and grid
grid were
were printed
printed toto assess
assess the
the printing accuracy of
printing accuracy of
2D structures. AsAs anan indicator
indicator of
of fusion
fusion at
at the
the inner
inner corner,
corner, the
the inner
inner arc length (ΔL) (∆L) of the printed
pentagram was measured using using aa stereomicroscope
stereomicroscope (ZY-HD1400;
(ZY-HD1400; ZongyanWeiye,
ZongyanWeiye, Shenzhen,Shenzhen, China)
China)
S-EYE v1.1
with an S-EYE v1.1measuring
measuringsoftware.
software.As Asananindicator
indicator ofof fusion
fusion at at
thethe edge,
edge, thethe cavity
cavity area
area (ΔA)(∆A)
of
of the printed grid was measured using the Image J software, and the
the printed grid was measured using the Image J software, and the cavity fusion rate ff) cavity fusion rate (R ) was
calculated. As
calculated. Asillustrated
illustratedinin Figure
Figure 1d,1d,
thethe cylinder
cylinder waswas printed
printed to assess
to assess the substrate
the substrate adhesion,
adhesion, and
and the printed height to the designed height ratio of the cylinder was calculated as
the printed height to the designed height ratio of the cylinder was calculated as the printing ratio. the printing ratio.
Whereas cuboid
cuboid was
was printed
printedtotoassess
assessthe
thephotocuring
photocuringdegreedegreeandandintensity,
intensity,andandthethe
radial strength
radial strengthof
thethe
of cuboid
cuboidwas determined
was determined using a hardness
using a hardnesstester of tablets.
tester of tablets.

2.3. Mechanical
2.3. Mechanical Properties
Properties of
of the
the Printed
Printed Material
Material
With the
With the addition
addition of
of PEG300
PEG300 as
as aa plasticizer,
plasticizer, dumbbell-shaped
dumbbell-shaped samples
samples (Figure
(Figure 1e)
1e) were
were printed
printed
at aa layer
at layer height
height of
of 0.05
0.05 mm
mm and
and the
the exposure
exposure time
time of
of 60
60 ss for
for bottom
bottom three
three layers
layers and
and 88 ss for
for other
other
layers. The mechanical properties of the printed samples were evaluated using a previously
described tensile test method [21] at a constant variable force of 0.008 N/s. The stress-strain curve was
drawn and used in compression simulation [22]. The tensile stress at break (σb) and elongation at
break (εb) were recorded to assess the strength and ductility of the material.
Pharmaceutics 2020, 12, 207 4 of 16

layers. The mechanical properties of the printed samples were evaluated using a previously described
tensile test method [21] at a constant variable force of 0.008 N/s. The stress-strain curve was drawn and
used in compression simulation [22]. The tensile stress at break (σb ) and elongation at break (εb ) were
Pharmaceutics
recorded to2020, 12, the
assess 207 strength and ductility of the material. 4 of 16

2.4.
2.4. Preparation and Compression Behavior
Behavior of
of Implants
Implants
Various
Various implants
implants in
in Figure
Figure 2a2a were
were designed
designed toto verify
verify the
the feasibility
feasibility of
of printing
printing samples
samples with
with
different
differentexternal
externalstructures.
structures.Samples
Sampleswere
were printed
printedat aatlayer height
a layer of 0.05
height mm mm
of 0.05 and the
andexposure time
the exposure
of 60 soffor
time 60bottom three layers
s for bottom and 8 sand
three layers for 8other
s forlayers. A constant
other layers. compression
A constant force offorce
compression 10 g was
of 10loaded
g was
on the surface
loaded of the printed
on the surface implants
of the printed to mimic
implants to the biomechanical
mimic behavior.
the biomechanical The appearance
behavior. of the
The appearance
implants before before
of the implants and after
andcompression was compared.
after compression was compared.

Figure 2. Models of (a) various implants and (b) physiological channels.

Figure 2. channels.

Finite
Finite element
element method
method simulation
simulation was
was conducted
conducted to to analyze
analyze the
the stress
stress and
and strain
strain behaviors
behaviors of
of
the
the implants under compression. Geometry models were established and meshed with PLAN82
implants under compression. Geometry models were established and meshed with PLAN82
elements
elements using
usingthe
theANSYS
ANSYS10.010.0software.
software.The
Thestress-strain
stress-straincurve ofof
curve the
theprinted
printedmaterial determined
material in
determined
Section 2.3 2.3
in Section waswas
inputted as nonlinear
inputted material
as nonlinear properties,
material and the
properties, andcompression force was
the compression loaded
force on the
was loaded
surface of the implants. Von Mises stress (σ ) and Von Mises
on the surface of the implants. Von Misesv stress (σv) and Von Mises strain (ε ) distributions were simulated,
v strain (εv) distributions were
and the overall
simulated, and deformation of the implants
the overall deformation was
of the analyzed.
implants was analyzed.

2.5. In Vitro Dissolution of Implants

2.5. In Vitro Dissolution of Implants
DS or ibuprofen (10%, w/w) was added in the printing formulation, and the drug-loaded implants
DS or ibuprofen (10%, w/w) was added in the printing formulation, and the drug-loaded
in Figure 2a were printed at a layer height of 0.05 mm and the exposure time of 60 s for bottom three
implants in Figure 2a were printed at a layer height of 0.05 mm and the exposure time of 60 s for
layers and 8 s for other layers. The specific surface area of implants was calculated by measuring the
bottom three layers and 8 s for other layers. The specific surface area of implants was calculated by
surface and volume. To compare the drug release behavior of implants with different shapes and
measuring the surface and volume. To compare the drug release behavior of implants with different
applications, a simplified in vitro dissolution method was conducted at 37 ± 0.5 ◦ C. According to
shapes and applications, a simplified in vitro dissolution method was conducted at 37 ± 0.5 ℃.
literature, 10 mL deionized water [23] was used as release medium for implants loaded soluble drug
According to literature, 10 mL deionized water [23] was used as release medium for implants loaded
DS, whereas 10 mL [24] phosphate buffer (pH = 7.2) [25] was used for implants loaded slightly soluble
soluble drug DS, whereas 10 mL [24] phosphate buffer (pH = 7.2) [25] was used for implants loaded
drug ibuprofen. At the predetermined time intervals, the dissolution medium was withdrawn and
slightly soluble drug ibuprofen. At the predetermined time intervals, the dissolution medium was
replaced with a fresh medium. The withdrawn sample was filtered through 0.45-µm Millipore filters
withdrawn and replaced with a fresh medium. The withdrawn sample was filtered through 0.45-μm
and analyzed using a spectrophotometer (UV-2450; Shimadzu, Japan) at 274 nm for DS or 263 nm
Millipore filters and analyzed using a spectrophotometer (UV-2450; Shimadzu, Japan) at 274 nm for
for ibuprofen. Cumulative drug release was calculated based on the validated DS calibration curve
DS or 263 nm for ibuprofen. Cumulative drug release was calculated based on the validated DS
(A = 0.0325C − 0.0069, R2 = 0.9993) or the validated ibuprofen calibration curve (A = 0.0018C + 0.0027,
calibration curve (A = 0.0325C – 0.0069, R² = 0.9993) or the validated ibuprofen calibration curve (A =
R2 = 0.9998).
0.0018C + 0.0027, R² = 0.9998).

2.6. Preparation and Evaluation of Physiological Channels

Physiological channels in Figure 2b were designed to verify the feasibility of printing samples
with different internal structures. Samples were printed at a layer height of 0.05 mm and the exposure
time of 60 s for bottom three layers and 8 s for other layers. For the perfusion test, red and green
solutions were added into the channel. Fluid flow behavior during the perfusion process was
Pharmaceutics 2020, 12, 207 5 of 16

2.6. Preparation and Evaluation of Physiological Channels

Physiological channels in Figure 2b were designed to verify the feasibility of printing samples
with different internal structures. Samples were printed at a layer height of 0.05 mm and the exposure
time of 60 s for bottom three layers and 8 s for other layers. For the perfusion test, red and green
Pharmaceutics 2020,added
solutions were 12, 207 into the channel. Fluid flow behavior during the perfusion process was observed
5 of 16

to evaluate the morphology and integrity of the internal channels.

Computational
Computationalfluid fluiddynamics
dynamics(CFD)
(CFD)simulation
simulationwaswascarried
carriedto
toanalyze
analyzethe
theflow
flowof
offluids
fluidsand
and
the
the interactions
interactionsbetween
betweenliquids
liquidsand
and channel
channel surfaces.
surfaces. The
The3D
3Dgeometry
geometrymodel
modelused
usedwas
wasshown
shown in in
Figure 2b, and the 2D cross-section of the channel inside the model (flow field) was established using
Figure 2b, and the 2D cross-section of the channel inside the model (flow field) was established using
the Gambit 2.4 software. The mesh model was established using Tri&Pave elements with a spacing
the Gambit 2.4 software. The mesh model was established using Tri&Pave elements with a spacing of
of 20. The boundary conditions were defined as follows: the left borders were velocity-inlet, the right
20. The boundary conditions were defined as follows: the left borders were velocity-inlet, the right
border was outflow, and the other borders were wall. The water density and viscosity at 25℃ were
border was outflow, and the other borders were wall. The water density and viscosity at 25 ◦ C were
inputted as material properties, whereas the blood flow velocity of adults (0.14 m/s) was inputted as
inputted as material properties, whereas the blood flow velocity of adults (0.14 m/s) was inputted as
inlet velocity. Steady simulation of the flow fields [26] was conducted using the Fluent 6.3 software.
inlet velocity. Steady simulation of the flow fields [26] was conducted using the Fluent 6.3 software.
A multiple reference frame model was used to simulate the axial flow, and a standard κ-ε model was
A multiple reference frame model was used to simulate the axial flow, and a standard κ-ε model was
adopted to describe the turbulence. The PRESTO! algorithm was used to calculate pressure and the
adopted to describe the turbulence. The PRESTO! algorithm was used to calculate pressure and the
SIMPLE algorithm was used to couple pressure and velocity.
SIMPLE algorithm was used to couple pressure and velocity.

3.
3.Results
Resultsand
andDiscussion
Discussion

3.1.
3.1.The
TheEffect
EffectofofPhotoabsorbers
Photoabsorbers
Using
Using PEGDA
PEGDA as as monomer
monomer and and DPPO
DPPO (1%,(1%, w/v)
w/v) as
as photoinitiator,
photoinitiator, the the effect
effect of
of different
different
photoabsorbers
photoabsorbers (0.25(0.25mg/mL,
mg/mL,w/v) w/v) on printability
on the the printability of external
of external and internal
and internal structuresstructures was
was evaluated.
evaluated.
As shown As shown 3a,
in Figure in Figure 3a, the
the height height
of the cone the=cone
of (Φ (Ф = was
1.5 mm) 1.5 mm) was not different
not different after the after the
addition
addition of methylene
of methylene blue or blue or coccine.
coccine. However,
However, the height
the height of cone
of the the cone decreased
decreased withwith
thethe addition
addition of
of tartrazine,
tartrazine, revealing
revealing that
that tartrazine
tartrazine was was able
able to slow
to slow downdownthethe light-induced
light-induced crosslinking
crosslinking reaction
reaction [14].
[14]. Furthermore,
Furthermore, the the depth
depth of the
of the internal
internal vertical
vertical channel(Φ(Ф==1.5
channel 1.5 mm)
mm) increased afterafter different
different
photoabsorbers
photoabsorbers was wasadded,
added,as asphotoabsorbers
photoabsorberswas wasable
abletotominimize
minimizelight
lightscattering
scatteringinto
intothe
thechannel,
channel,
thereby
thereby slowing down the over-curing of the non-printed area and facilitating the printability of
slowing down the over-curing of the non-printed area and facilitating the printability of
internal
internal structures.
structures.

Figure 3.
Figure (a)Effect
3. (a) Effect of
of different
different photoabsorbers
photoabsorbers (i.e.,
(i.e., methylene
methylene blue,
blue, coccine,
coccine, and
and tartrazine)
tartrazine)on
on the
the
printabilityof
printability ofconvex
convexcone
coneand
andvertical
verticalchannel;
channel;(b)
(b)wavelength
wavelengthscan
scanresults
resultsof
ofthe
thephotoinitiator
photoinitiator(i.e.,
(i.e.,
Diphenyl(2,4,6-trimethylbenzoyl)
Diphenyl (2,4,6-trimethylbenzoyl) phosphine
phosphine oxide,
oxide, DPPO)
DPPO) and
and photoabsorbers.
photoabsorbers.

The wavelength scanning results of photoinitiator and photoabsorber solutions (0.01 mg/mL,
w/v) are shown in Figure 3b. At the printing wavelength (405 nm), the order of the absorbance values
was as follows: tartrazine > coccine > methylene blue > DPPO. With the highest absorbance at 405
nm, tartrazine was the most effective photoabsorber in slowing down the photocuring reaction.
Therefore, tartrazine addition had the most obvious reducing effect on cone printability but
improving effect on channel printability (Figure 3a). Owing to its high photoabsorption efficiency
Pharmaceutics 2020, 12, 207 6 of 16

The wavelength scanning results of photoinitiator and photoabsorber solutions (0.01 mg/mL, w/v)
are shown in Figure 3b. At the printing wavelength (405 nm), the order of the absorbance values was
as follows: tartrazine > coccine > methylene blue > DPPO. With the highest absorbance at 405 nm,
tartrazine was the most effective photoabsorber in slowing down the photocuring reaction. Therefore,
tartrazine addition had the most obvious reducing effect on cone printability but improving effect on
channel printability (Figure 3a). Owing to its high photoabsorption efficiency and good biocompatibility,
tartrazine should be added when printing internal structures. However, it was unnecessary for printing
Pharmaceutics 2020, 12, 207 6 of 16
external structures. Besides, DPPO (1%, w/v) was found to be an effective photoinitiator of PEGDA
polymers at the printing
the determined absorbance wavelength
of DPPOofat405 405nm nm[15,27].
was much However,
lower thanthe determined absorbance of(Figure
those of photoabsorbers DPPO
at 405 nm was much lower than those of photoabsorbers (Figure 3b). A similar
3b). A similar phenomenon was found in the literature [14], and it might be related to the different phenomenon was found
in the literature
effects [14], and it
of photoinitiator might
and be related toon
photoabsorber thethe
different effects of
photocuring photoinitiator
process. and photoabsorber
Other photoinitiators with
on the photocuring process. Other photoinitiators
higher absorbance at 405 nm should be studied in further study. with higher absorbance at 405 nm should be studied
in further
The study.
effect of tartrazine concentration as a photoabsorber on the printability of the internal
The effect
vertical channel of was
tartrazine concentration
evaluated. As shown asina photoabsorber
Figure 4a, there onwasthe noprintability
significant of difference
the internal(Pvertical
> 0.05)
channel was evaluated. As shown in Figure 4a, there was no significant
in patency degree of 0.15 and 0.20 mg/mL tartrazine, or 0.25 and 0.30 mg/mL tartrazine. The difference (p >reduction
0.05) in
patency
in averagedegree of 0.15
patency and 0.20
degree mg/mL
of 0.30 mg/mL tartrazine,
tartrazine or 0.25 and 0.30
compared mg/mL
to that tartrazine.
of 0.25 mg/mLThe reduction
tartrazine mightin
average patency degree of 0.30 mg/mL tartrazine compared to that of 0.25
due to the low printability (irregularly cylindrical internal channel) and high detection deviation. mg/mL tartrazine might
due to the low
However, whenprintability
tartrazine (irregularly
concentration cylindrical
increasedinternal
from 0.20 channel)
mg/mL andto high detectionthe
0.25 mg/mL, deviation.
patency
However, when tartrazine concentration increased from 0.20 mg/mL to
degree significantly increased from 46.1% to 64.4% (P < 0.01), suggesting an improvement of printing 0.25 mg/mL, the patency
degree significantly
accuracy. As shownincreased
in Figure from 46.1% towith
4b, a sample 64.4% (p < 0.01),
different suggesting
diameters and aan improvement
height of 6 mm wasof printing
printed
accuracy. As shown
to determinate the D inminFigure 4b, a sample
. Considering that with
the DLPdifferent diameters
printability andlinearly
is not a heightrelated
of 6 mm to was
the D printed
min, the
to determinate the D . Considering that the DLP printability is not
Dmin of every printed sample (n = 3) were listed in Figure 4b instead of the average and standard
min linearly related to the Dmin ,
the D
deviation.
min of every printed sample (n = 3) were listed in Figure 4b instead
When the tartrazine concentration increased from 0.15% to 0.25%, the Dmin of channel of the average and standard
deviation. When the tartrazine of
decreased. This improvement concentration
printing accuracyincreased of from 0.15% tochannel
the internal 0.25%, the wasDmin of channel
related to the
decreased. This improvement of printing accuracy of the internal channel
absorption of light scattered into the channel by tartrazine. Addition of excessive tartrazine was related to the absorption
(0.3
of light scattered
mg/mL) could not intofurther
the channel
improveby tartrazine.
printability, Addition
suggestingof excessive
the limited tartrazine (0.3a mg/mL)
effect of could not
photoabsorber on
further improve printability, suggesting the limited effect of a photoabsorber
printability. Therefore, the optimal tartrazine concentration of 0.25 mg/mL was used in further on printability. Therefore,
the optimal tartrazine concentration of 0.25 mg/mL was used in further studies.
studies.

Figure4.4.Effect
Figure Effectof of
tartrazine concentration
tartrazine on (a)
concentration onpatency degreedegree
(a) patency of the vertical
of the channel
vertical with a diameter
channel with a
of 1.5 mm of
diameter (**1.5
denote
mm (**p <denote
0.01), and (b) the
p < 0.01), minimum
and printable diameter
(b) the minimum printable (D min ) of the
diameter vertical
(Dmin) of thechannel
vertical
with different
channel with diameters, and typical
different diameters, andsample with
typical 0.25%
sample tartrazine
with was illustrated.
0.25% tartrazine was illustrated.
3.2. Effect of PEGDA Concentration
3.2. Effect of PEGDA Concentration
The effect of PEGDA concentration on the printability of 2D samples is shown in Figure 5a,b.
The effect of PEGDA concentration on the printability of 2D samples is shown in Figure 5a-b.
High PEGDA concentrations (i.e., 100% and 75%) led to large ∆L at the top layer of the pentagram
High PEGDA concentrations (i.e., 100% and 75%) led to large ΔL at the top layer of the pentagram
(0.44 mm and 0.43 mm, respectively), and high Rf at the bottom layer of the grid (52.9% and 58.9%,
respectively). The undesirable crosslinking at the non-printed area was caused by the light scattered
from the solidified edge. The exposure time of the bottom three layers (60 s) was much longer than
that of the other layers (8 s). Therefore, more light was scattered at the bottom, leading to more serious
over-curing at the bottom layers. With a low PEGDA concentration of 50%, both the ΔL of the
Pharmaceutics 2020, 12, 207 7 of 16

(0.44 mm and 0.43 mm, respectively), and high Rf at the bottom layer of the grid (52.9% and 58.9%,
respectively). The undesirable crosslinking at the non-printed area was caused by the light scattered
from the solidified edge. The exposure time of the bottom three layers (60 s) was much longer than
that of the other layers (8 s). Therefore, more light was scattered at the bottom, leading to more
serious over-curing at the bottom layers. With a low PEGDA concentration of 50%, both the ∆L of
the pentagram and the Rf of the grid decreased, revealing less fusion at the inner corner and edges.
Further reduction of PEGDA concentration to 25% resulted in printing failure. Therefore, 50% PEGDA
Pharmaceutics
was preferred2020,
for12,2D
207samples requiring high printing accuracy at the bottom. 7 of 16

Figure 5. Effect of poly(ethylene glycol) diacrylate (PEGDA) concentration on the printability: (a) inner
Figure
arc 5. Effect
length of (b)
(∆L) and poly(ethylene glycol)
cavity fusion rate (Rdiacrylate (PEGDA) concentration on the printability: (a)
f ) of 2D samples; and (c) cone height and (d) cone depth of
inner arc
3D samples.length (ΔL) and (b) cavity fusion rate (Rf) of 2D samples; and (c) cone height and (d) cone

depth of 3D samples.
The effect of PEGDA concentration on the printability of 3D samples is shown in Figure 5c,d.
WhenThe theeffect
PEGDA of PEGDA concentration
concentration on the printability
was reduced from 100% of to 3D
50%,samples is shown
the height of thein convex
Figure 5c–d.
cone
When
(Φ themm)
= 1.0 PEGDA concentration
decreased from 2.53 wasmm reduced
to 0.40frommm,100%
as low to 50%,
PEGDA the height of the convex
concentration led tocone (Ф =
reduced
1.0 mm) decreased from 2.53 mm to 0.40 mm, as low PEGDA concentration
crosslinking degree and negatively affected the printability of fine structure. In contrast, the concave led to reduced
crosslinking
cone (Φ = 1.0degree
mm) couldand negatively affected
not be printed usingthePEGDA
printability of fine
solution structure.suggesting
(100–25%), In contrast,thatthe printing
concave
cone (Ф =cone
concave 1.0 mm)
wascould
morenot be printed
difficult than using PEGDA
printing convexsolution
cone. (100%–25%),
When printing suggesting that printing
large concave cone
(Φ = 2.0 mm), a similar effect of PEGDA concentration on printability was observed. Reducedcone
concave cone was more difficult than printing convex cone. When printing large concave (Ф =
PEGDA
2.0 mm), a similar
concentration effect printability,
decreased of PEGDA concentration
and insufficient onPEGDA
printability was observed.
concentration (25%)Reduced PEGDA
caused printing
concentration decreased printability, and insufficient PEGDA concentration
failure. Therefore, 100% PEGDA (Rx-I) was optimized for printing 3D samples with external structures. (25%) caused printing
failure. Therefore,
To improve the100% PEGDA (Rx-I)
low printability was optimized
of internal structures for printing
in Figure 5d, 3D
0.25samples with external
mg/mL tartrazine was
structures.
added and the effect of PEGDA concentration on the printability was further studied. As illustrated
To improve
in Figure the low
6a, reducing printability
PEGDA of internal
concentration structures
facilitated thein Figure 5d,of0.25
printability mg/mL
internal tartrazine
vertical was
channels
added and the effect of PEGDA concentration on the printability was further
(Φ = 1.75, 1.5, 1.25, 1.0, 0.75, and 0.5 mm). When PEGDA concentration decreased from 100% to 87.5%, studied. As illustrated
in Figure
the patency 6a,degree
reducing PEGDA
of the verticalconcentration
channel (Φ =facilitated the printability
1.0 mm) increased of internal
from 33.5% vertical
to 100%, and channels
the Dmin
(Ф = 1.75, 1.5, 1.25, 1.0, 0.75, and 0.5 mm). When PEGDA concentration
of vertical channel decreased from 1.25 mm to 1.0 mm. Further reduction of PEGDA concentrationdecreased from 100% to 87.5%,
the75.0%
to patency degreehad
or 62.5% of thenovertical channel
significant effect(Фon= 1.0
the mm) increased
printability of from 33.5%
vertical to 100%,
channels. Asand the Dmin
illustrated
of vertical channel decreased from 1.25 mm to 1.0 mm. Further reduction
in Figure 6b, reducing PEGDA concentration also improved the printability of internal horizontal of PEGDA concentration to
75.0% or 62.5% had no significant effect on the printability of vertical channels.
channels (Φ = 1.75, 1.5, 1.25, 1.0, 0.75, and 0.5 mm). When PEGDA concentration decreased from 100% As illustrated in
Figure
to 62.5%, 6b,thereducing
patency PEGDAdegree ofconcentration
the horizontalalso improved
channels themm)
(Φ = 1.0 printability
increased offrom
internal
14.1% horizontal
to 100%,
channels (Ф = 1.75, 1.5, 1.25, 1.0, 0.75, and 0.5 mm).
and the Dmin of horizontal channels decreased from 1.5 mm to 1.0 mm.When PEGDA concentration decreased from 100%
to 62.5%, the patency degree of the horizontal channels (Ф = 1.0 mm) increased from 14.1% to 100%,
and the Dmin of horizontal channels decreased from 1.5 mm to 1.0 mm.
Pharmaceutics 2020, 12, 207 8 of 16
Pharmaceutics 2020, 12, 207 8 of 16

Figure 6. Effect of PEGDA concentration on the printability of (a) vertical and (b) horizontal channels.
Figure 6. Effect of PEGDA concentration on the printability of (a) vertical and (b) horizontal channels.
Dmin : minimum printable diameter.
Dmin: minimum printable diameter.
The horizontal channel was perpendicular to the projection direction of light [14]; therefore,
The horizontal channel was perpendicular to the projection direction of light [14]; therefore, a
a large amount of light penetrated through the transparent crosslinked area into the channel, causing
large amount of light penetrated through the transparent crosslinked area into the channel, causing
over-curing inside the channel. In contrast, the vertical channel was parallel to the projection direction
over-curing inside the channel. In contrast, the vertical channel was parallel to the projection direction
of light, leading to a small amount of light scattered into the channel. Therefore, reducing PEGDA
of light, leading to a small amount of light scattered into the channel. Therefore, reducing PEGDA
concentration was more effective in minimizing over-curing inside horizontal channels than inside
concentration was more effective in minimizing over-curing inside horizontal channels than inside
vertical channels. Although 62.5% PEGDA with 0.25 mg/mL tartrazine led to the highest printability of
vertical channels. Although 62.5% PEGDA with 0.25 mg/mL tartrazine led to the highest printability
internal channels, edge deformation and corner defects were observed (green circle) in Figure 6a,b.
of internal channels, edge deformation and corner defects were observed (green circle) in Figure 6a-
Therefore, 75% PEGDA with 0.25 mg/mL tartrazine (Rx-II) was the optimal printing solution for
b. Therefore, 75% PEGDA with 0.25 mg/mL tartrazine (Rx-II) was the optimal printing solution for
samples with internal channels.
samples with internal channels.
3.3. Effect of Layer Height
3.3. Effect of Layer Height
Using Rx-I as a printing solution, the effect of layer height on the printability of samples with
Using
external Rx-I as awas
structures printing solution,and
investigated, the the
effect of layer
results are height on the
illustrated in printability
Figure 7a. ofAssamples with
layer height
external structures was investigated, and the results are illustrated in Figure
increased from 0.02 mm to 0.05 mm, the height of convex cones (Φ = 0.75 mm) decreased from 2.36 mm7a. As layer height
increased
to 1.76 mm, from
and0.02
the mm to 0.05
printing mm,
time wasthedecreased
height offrom
convex
82 cones
min to(Ф 35=min.
0.75 mm) decreased
This was fromwith
consistent 2.36
mmresults
the to 1.76ofmm,
FDM andprinting
the printing
[28], time was decreased
in which fromheight
reduced layer 82 minwas to 35beneficial
min. Thistowas consistent
printing with
accuracy
the results of FDM printing [28], in which reduced layer height was beneficial to printing
but adverse to printing efficiency. Therefore, the optimal layer height for the external structures was accuracy
but adverse to
determined to be
printing efficiency. Therefore, the optimal layer height for the external structures was
0.02 mm.
determined to beas
Using Rx-II 0.02 mm.
a printing solution, the effect of layer height on the printability of samples with
internal structures was investigated (Figure 7b). As layer height increased from 0.02 mm to 0.05 mm,
the patency degree of vertical channels (Φ = 1.0 mm) increased from 77.5% to 100.0%, whereas the
printing time decreased from 81 min to 35 min. The improved printability of vertical channels with
increased layer height was due to shorter exposure time and less amount of light scattered into the
channel. Although the printability of horizontal channels was lower than that of vertical channels,
increasing layer height was more effective in minimizing over-curing inside horizontal channels.
When layer height increased from 0.02 mm to 0.05 mm, the patency degree of the horizontal channel
(Φ = 1.0 mm) increased from 9.7% to 100.0%. Because the samples had the same height, the printing
time of horizontal channels was not different from that of vertical channels. Considering both printing
accuracy and efficiency, 0.05 mm was the optimal layer height for internal channels within the range of
0.02 mm and 0.05 mm.
Pharmaceutics 2020, 12, 207 9 of 16
Pharmaceutics 2020, 12, 207 9 of 16

Figure 7. Effect of layer height on the printing accuracy and efficiency of (a) convex cones and (b)
Figure
internal7.channels.
Effect of layer height on the printing accuracy and efficiency of (a) convex cones and (b)
internal channels.
3.4. Effect of Exposure Time
Using Rx-II as a printing solution, the effect of layer height on the printability of samples with
With a constant exposure time of other layers (8 s), the effect of exposure time of the bottom
internal structures was investigated (Figure 7b). As layer height increased from 0.02 mm to 0.05 mm,
three layers on substrate adhesion was evaluated using the cylinder with a diameter of 15 mm and a
the patency degree of vertical channels (Ф = 1.0 mm) increased from 77.5% to 100.0%, whereas the
height of 6 mm (Figure 1d). As shown in Figure 8a, when the cylinder was printed using Rx-I at an
printing time decreased from 81 min to 35 min. The improved printability of vertical channels with
exposure time of 10 s, substrate adhesion was insufficient to complete the printing and the printed
increased layer height was due to shorter exposure time and less amount of light scattered into the
sample fell off the printing platform with a printing ratio of 77.3%. With prolonged bottom exposure
channel. Although the printability of horizontal channels was lower than that of vertical channels,
time from 30 s to 90 s, substrate adhesion was sufficient for printing, although more force was required
increasing layer height was more effective in minimizing over-curing inside horizontal channels.
to demold the printed samples. A similar tendency was observed when a cylinder was printed using
When layer height increased from 0.02 mm to 0.05 mm, the patency degree of the horizontal channel
Rx-II as a printing solution. When the bottom exposure time was 10 s, the printing ratio of Rx-II (69.1%)
(Ф = 1.0 mm) increased from 9.7% to 100.0%. Because the samples had the same height, the printing
was slightly lower than that of Rx-I (77.3%), indicating that the water and photoabsorber in Rx-II
time of horizontal channels was not different from that of vertical channels. Considering both
decelerated the crosslinking reaction and negatively affected on substrate adhesion. Cylinders with a
printing accuracy and efficiency, 0.05 mm was the optimal layer height for internal channels within
constant height of 6 mm and different diameters (i.e., 5, 10, and 15 mm) were printed to study the effect
the range of 0.02 mm and 0.05 mm.
of diameters on surface adhesion. The results (not provided) revealed that the different diameters had
no significant influence on surface adhesion per interfacial area. However, it was difficult to evaluate
3.4. Effect of Exposure Time
the surface adhesion of the grid with a small height of 1.5 mm (Figure 1c). No grid sample fell off
duringWith a constant
printing at theexposure
exposuretime
time of
of other
10 s. Itlayers (8 s), the
was related effect
to the of exposure
small time
height and the of the bottom
serious cavity
three layers
fusion, which
Pharmaceutics on substrate
greatly
2020, adhesion was evaluated
12, 207 increased the interfacial area. using the cylinder with a diameter of 15 mm and
10 a
of 16
height of 6 mm (Figure 1d). As shown in Figure 8a, when the cylinder was printed using Rx-I at an
exposure time of 10 s, substrate adhesion was insufficient to complete the printing and the printed
sample fell off the printing platform with a printing ratio of 77.3%. With prolonged bottom exposure
time from 30 s to 90 s, substrate adhesion was sufficient for printing, although more force was
required to demold the printed samples. A similar tendency was observed when a cylinder was
printed using Rx-II as a printing solution. When the bottom exposure time was 10 s, the printing ratio
of Rx-II (69.1%) was slightly lower than that of Rx-I (77.3%), indicating that the water and
photoabsorber in Rx-II decelerated the crosslinking reaction and negatively affected on substrate
adhesion. Cylinders with a constant height of 6 mm and different diameters (i.e. 5, 10, and 15 mm)
were printed to study the effect of diameters on surface adhesion. The results (not provided) revealed
that the different diameters had no significant influence on surface adhesion per interfacial area.
However, it was difficult to evaluate the surface adhesion of the grid with a small height of 1.5 mm
(Figure 1c). No grid sample fell off during printing at the exposure time of 10 s. It was related to the
Figure 8. Effect of exposure time of (a) bottom three layers and (b) other layers on printability. Rx-I:
Rx-I:
small height and the serious cavity fusion, which greatly increased the interfacial area.
100% PEGDA; Rx-II: 75% PEGDA
PEGDA with
with 0.25
0.25 mg/mL tartrazine; RRff: cavity fusion rate.
mg/mL tartrazine;

The effect
The effect of
of bottom
bottom exposure
exposuretime
timeon
onthe
thefusion
fusionofofgrid
gridis is illustrated
illustrated in in Figure
Figure 8a.8a. Rf ofRfthe
TheThe of
2
grid (2 ×(22×mm
the grid 2 mm ) increased
2) increased from
from 54.1%
54.1% to to 100%when
100% whenititwas
wasprinted
printedusing
using Rx-I
Rx-I with
with prolonged
prolonged
exposure time from 10 s to 90 s. A similar tendency was observed when the grid was printed using
Rx-II, although the Rf of Rx-II was much lower than that of Rx-I, because of the adverse effect of the
water and photoabsorber in Rx-II on over-curing. In short, the bottom exposure time was optimized
to 60 s to guarantee sufficient substrate adhesion during printing. However, when a structure with
porous structures at the bottom was printed, the bottom exposure time should be reduced.
Pharmaceutics 2020, 12, 207 10 of 16

exposure time from 10 s to 90 s. A similar tendency was observed when the grid was printed using
Rx-II, although the Rf of Rx-II was much lower than that of Rx-I, because of the adverse effect of the
water and photoabsorber in Rx-II on over-curing. In short, the bottom exposure time was optimized
to 60 s to guarantee sufficient substrate adhesion during printing. However, when a structure with
porous structures at the bottom was printed, the bottom exposure time should be reduced.
In addition, the effect of exposure time of other layers on photocuring degree and intensity was
studied at a constant bottom exposure time of 60 s (Figure 8b). When a cuboid structure (Figure 1d)
was printed using Rx-I with increased exposure time from 2 s to 8 s, the weight increased from 33 mg to
166 mg and the hardness increased from 0 N to 107.2 N. Further prolongation of exposure time to 12 s
led to no difference, suggesting that exposure to light for 8 s was sufficient to complete the photocuring
reaction. A similar tendency was observed when a cuboid was printed using Rx-II, although the
weight and hardness decreased, owing to the adverse effect of the water and photoabsorbter in Rx-II
on photocuring. Considering the photocuring degree and printing efficiency, the optimal exposure
time for other layers was 8 s.

3.5. Compression Behavior and Simulation of Implants

To mimic the biomechanical properties of implants, PEG300 was added as a plasticizer to improve
the flexibility of crosslinked PEGDA [15]. The effect of PEG300 concentration on the printability of
convex cone (Φ = 1.0 mm) is shown in Figure 9a. Cone height was the highest after the addition of
20% PEG300, whereas the cone deformation was the most serious after the addition of 40% PEG300.
This finding suggested that a moderate amount of plasticizer was beneficial to printability, whereas
an excessive amount of plasticizer caused poor strength and deformation. The effect of PEG300
concentration on the mechanical properties of printed materials is shown in Figure 9b. Printed samples
without PEG300 were too brittle to be completely demold and evaluated by the tensile test. As PEG300
increased from 10% to 40%, σb decreased from 7.0 MPa to 5.5 MPa, whereas εb increased from 7.9% to
13.6%. The added PEG interspersed within the PEGDA chains, thus reducing the degree of crosslinking
between PEGDA chains [15]. Considering the printing accuracy and mechanical properties, the optimal
printing solution
Pharmaceutics 207implants was PEGDA:PEG300 = 80:20 (Rx-III).
2020, 12,for 11 of 16

Figure 9.9.Effect
Effectof of
PEG300 concentration
PEG300 on (a)onthe(a)
concentration printability of fine structures
the printability and (b) theand
of fine structures mechanical
(b) the
properties
mechanicalofproperties
printed materials.
of printed materials.

The
The optimal
optimal printing
printing material
material (Rx-III)
(Rx-III) showed
showed obvious
obvious plastic
plastic deformation
deformation during
during the
the tensile
tensile
test (ε >10%); thus, the Von Mises criterion was used as a failure criterion during
test (εbb >10%); thus, the Von Mises criterion was used as a failure criterion during compression compression
behavior
behavior simulation. As shown
simulation. As shown inin Figure
Figure 10a-b,
10a,b, the
the printed
printed U-shaped
U-shaped sample
sample had
had good strength and
good strength and
elasticity under a compression force of 10 g, and the deformation was recovered after the
elasticity under a compression force of 10 g, and the deformation was recovered after the compression compression
force
force was
was removed.
removed. A A typical
typical 2D
2D cross-section
cross-section mesh
mesh model
model ofof the
the U-shaped sample is
U-shaped sample is shown
shown in in
Figure 10c; the vertical pressure loaded on the upper surface of the sample was 8.6 kPa. The
Figure 10c; the vertical pressure loaded on the upper surface of the sample was 8.6 kPa. The simulated simulated
σ distribution (Figure
σvv distribution (Figure 10d)
10d) revealed
revealed that
that the
the bottom
bottom of
of the
the sample
sample withstood
withstood the
the most
most stress
stress (labeled
(labeled
as a red region). The ultimate σ (154.2 kPa) was much lower than the σ of the material
as a red region). The ultimate σvv (154.2 kPa) was much lower than the σbb of the material (6.3 MPa) (6.3 MPa)
shown
shown in in Figure
Figure 9b;
9b; therefore,
therefore, the
the integrity
integrity of
of the
the printed sample could
printed sample could bebe guaranteed.
guaranteed. Furthermore,
Furthermore,
the simulated εv distribution (Figure 10e) suggested that the bottom of the sample withstood the most
strain (labeled as a red region), with the ultimate εv of 0.05%. The deformed shape was also compared
with the undeformed shape (marked with white dashed lines), and the overall deformation and
displacement of the sample could be visualized.
behavior simulation. As shown in Figure 10a-b, the printed U-shaped sample had good strength and
elasticity under a compression force of 10 g, and the deformation was recovered after the compression
force was removed. A typical 2D cross-section mesh model of the U-shaped sample is shown in
Figure 10c; the vertical pressure loaded on the upper surface of the sample was 8.6 kPa. The simulated
Pharmaceutics
σ 12, 207 10d) revealed that the bottom of the sample withstood the most stress (labeled
2020,(Figure
v distribution 11 of 16
as a red region). The ultimate σv (154.2 kPa) was much lower than the σb of the material (6.3 MPa)
shown in Figure 9b; therefore, the integrity of the printed sample could be guaranteed. Furthermore,
the simulated εv distribution (Figure 10e) suggested that the bottom of the sample withstood the
the simulated εv distribution (Figure 10e) suggested that the bottom of the sample withstood the most
most strain (labeled as a red region), with the ultimate εv of 0.05%. The deformed shape was also
strain (labeled as a red region), with the ultimate εv of 0.05%. The deformed shape was also compared
compared with the undeformed shape (marked with white dashed lines), and the overall deformation
with the undeformed shape (marked with white dashed lines), and the overall deformation and
and displacement of the sample could be visualized.
displacement of the sample could be visualized.

Figure 10. Typical U-shaped sample: (a) before and (b) after compression; (c) mesh model with
Figure 10. Typical U-shaped sample: (a) before and (b) after compression; (c) mesh model with loaded
loaded pressure; simulated (d) Von Mises stress (σv ) and (e) Von Mises strain (εv ) distribution.
pressure; simulated (d) Von Mises stress (σv) and (e) Von Mises strain (εv) distribution. Typical arc-
Typical arc-shaped sample: (f) before and (g) after compression; (h) mesh model with loaded pressure;
shaped sample: (f) before and (g) after compression; (h) mesh model with loaded pressure; simulated
simulated (i) σ and (j) εv distribution.
(i) σv and (j) εv vdistribution.

As shown in Figure 10f,g, the printed arc-shaped sample had good strength and elasticity under a
As shown in Figure 10f–g, the printed arc-shaped sample had good strength and elasticity under
compression force of 10 g, and the deformation was recovered after the compression force was removed.
a compression force of 10 g, and the deformation was recovered after the compression force was
A typical 3D mesh model of the arc-shaped sample is shown in Figure 10h; the vertical pressure loaded
removed. A typical 3D mesh model of the arc-shaped sample is shown in Figure 10h; the vertical
on the upper surface of the sample was 1.35 kPa. The simulated σv distribution (Figure 10i) revealed
pressure loaded on the upper surface of the sample was 1.35 kPa. The simulated σv distribution
that the upper surface of the sample withstood the most stress (labeled as a red region). The ultimate
(Figure 10i) revealed that the upper surface of the sample withstood the most stress (labeled as a red
σv (38.1 kPa) was much lower than the σb of the material (6.3 MPa) shown in Figure 9b; therefore,
region). The ultimate σv (38.1 kPa) was much lower than the σb of the material (6.3 MPa) shown in
the integrity of the printed samples could be guaranteed. Furthermore, the simulated εv distribution
Figure 9b; therefore, the integrity of the printed samples could be guaranteed. Furthermore, the
(Figure 10j) suggested that the upper surface of the sample withstood the most strain (labeled as a red
simulated εv distribution (Figure 10j) suggested that the upper surface of the sample withstood the
region), with the ultimate εv of 0.89%. The deformed shape was also compared with the undeformed
most strain (labeled as a red region), with the ultimate εv of 0.89%. The deformed shape was also
shape (marked with white border lines), and the overall deformation of the sample could be visualized.

3.6. In Vitro Dissolution of Implants

As typical soluble and insoluble drugs, respectively, DS and ibuprofen were added into Rx-III to
evaluate the drug-loading ability and dissolution behavior. Various implants (Figure 2a) with a length
of 20 mm and a weight of approximately 500 mg were printed. As shown in Figure 11, T-shaped and
ring-shaped samples could be used as intrauterine devices, U-shaped and arc-shaped samples could be
applied as femoral cartilage and contact lens, and the needle-shaped sample had application potential
as a minimally invasive implant. According to the printability of external structures, the smallest
diameter of cone printed by DLP was about 0.5 mm. Therefore, microneedle (Φ < 0.3 mm) could not
be printed using the DLP method in this study.
As shown in Figure 11a, the DS-release behavior of various implants exhibited sustained-release
characteristics. The DS released from the needle-shaped samples had the highest release speed
and degree during 24 h, owing to its highest specific surface area (Table 1). It was agreed with the
literature [20], where the drug release from printed tablets was dependent on the surface area to
volume ratio. As shown in Figure 11b, ibuprofen release from the implants was much slower than DS
release, and the sustained-release behavior might last for several days (except for ibuprofen release
from the needle-shaped samples). The results showed that both soluble and insoluble drugs could
be loaded into Rx-III, and the sustained-released behavior of the printed implants could be adjusted
by changing the specific surface area. However, it was found that the highly soluble PEG300 added
literature [20], where the drug release from printed tablets was dependent on the surface area to
volume ratio. As shown in Figure 11b, ibuprofen release from the implants was much slower than
DS release, and the sustained-release behavior might last for several days (except for ibuprofen
release from the needle-shaped samples). The results showed that both soluble and insoluble drugs
could be loaded
Pharmaceutics 2020, 12,into
207 Rx-III, and the sustained-released behavior of the printed implants could 12 ofbe
16
adjusted by changing the specific surface area. However, it was found that the highly soluble PEG300
added as plasticizer leached out and greatly accelerated the drug release behaviors in Figure 11.
as plasticizer
Therefore, leached
PEG300 out
was notand
thegreatly accelerated
best choice the drugimplants
for biomedical release behaviors in Figure 11.medication,
requiring long-acting Therefore,
PEG300 was not the best choice for biomedical implants requiring long-acting medication, and
and it was used in this study to evaluate the influence of different shapes on dissolution. To accurately it was
used inthe
assess thissustained-release
study to evaluate behavior
the influence of differentwater-insoluble
of implants, shapes on dissolution. To accurately
plasticizers should be assess the
further
sustained-release
studied, and the behavior
in vitro of implants, water-insoluble
dissolution method shouldplasticizers should
be modified be further
to better studied,
mimic and the
the implant
in vitro dissolution
environment in the method
body. should be modified to better mimic the implant environment in the body.

Figure 11.
Figure Appearance and
11. Appearance and dissolution
dissolution behavior
behavior of
of implants
implants loaded
loaded with
with (a)
(a) diclofenac
diclofenac sodium
sodium (DS)
(DS)
and (b) ibuprofen.
and (b) ibuprofen.
Table 1. The specific surface area of various implants.
Table 1. The specific surface area of various implants.
T-Shaped Ring-Shaped U-Shaped Arc-Shaped Needle-Shaped
T-shaped ring-shaped U-shaped arc-shaped needle-shaped
surface
surface area (m 2 ) (m2)
area 5.8 ×5.8 × 10−4 5.85.8
10−4 × 10 −4 −4
× 10 4.7××10
4.7 −4
10−4 7.57.5× × −4−4
1010 2.2 ×2.2
10×−5
10−5
3 −7
volume (m ) 4.2 × 10 4.2 × 10−7 4.6 × 10−7 5.8 × 10−7−7 3.6 × 10 −9
volume (m3 ) 4.2 × 10−7 4.2 × 10−7 3 4.6 × 10−7 5.8 × 10 3.6 × 10−9
specific surface area (m2/m3) 1.4 × 103 1.4 × 10 1.0 × 103 1.3 × 103 6.2 × 103
specific surface area (m2 /m3 ) 1.4 × 103 1.4 × 103 1.0 × 103 1.3 × 103 6.2 × 103

3.7. Perfusion Behavior and Simulation of Physiological Channels

Typical samples (Figure 2b) were printed to mimic blood vessels and trachea using optimal
printing material for internal channels (Rx-II). As shown in Figure 12a, the vascular model was
successfully printed, and the perfusion process showed the good morphology and integrity of the
internal channel. The vascular model could be used as an in vitro platform to evaluate the elution
performance, diffusion behavior and biological perfusion. As shown in Figure 12b, a tracheal model
with good morphology was printed, and the integrity of the sample was proved by a perfusion test.
To mimic the trachea and bronchi, the inlet and outlet of simplified model overlapped at a large channel
(Φ = 3 mm). However, the ductility and gas permeability of this model should be improved before it
could be used as a simplified platform to evaluate the biomechanical properties of the trachea and the
powder deposition process during pulmonary administration.
In addition, a fluid mixing model with good morphology and integrity is illustrated in Figure 12c.
The green and red solutions were separately added into the inlets on the left. When the solutions
flowed along the internal channels, the green and red solutions mixed, separated, remixed, and then
flowed out from the outlet on the right. The mixing performance and hydrodynamic behavior of
multiple fluids could be evaluated using this model. According to the printability of internal structures,
the smallest diameter of channel printed by DLP was about 1.0 mm. Therefore, microfluidic chip
(Φ < 0.1 mm) could not be printed using the DLP method in this study.
flowed along the internal channels, the green and red solutions mixed, separated, remixed, and then
flowed out from the outlet on the right. The mixing performance and hydrodynamic behavior of
multiple fluids could be evaluated using this model. According to the printability of internal
structures, the smallest diameter of channel printed by DLP was about 1.0 mm. Therefore,
Pharmaceutics 2020,
microfluidic chip12,(Ф
207< 0.1 mm) could not be printed using the DLP method in this study. 13 of 16

Figure 12. Printed samples with internal channels and their perfusion process: (a) vascular model,
Figure 12. Printed samples with internal channels and their perfusion process: (a) vascular model, (b)
(b) tracheal model, and (c) fluid mixing model.
tracheal model, and (c) fluid mixing model.
The fluid mixing model in Figure 12c was used for CFD simulation, and its 2D cross-section
flowThe
fieldfluid mixing
is shown inmodel
Figurein13a.
Figure 12c was used
To simplify for CFD simulation,
the simulation, a multipleand its 2D cross-section
reference frame modelflow was
field is shown in Figure 13a. To simplify the simulation, a multiple reference frame
adopted to simulate the steady-state of the fluid field. As illustrated in Figure 13b, the simulated total model was
adopted
pressure to simulatealong
decreased the steady-state of the
the fluid field fluid
from thefield. As illustrated
entrance inthe
(212 Pa) to Figure
exit 13b, the simulated
(1.3 Pa). total
The simulated
pressure
turbulentdecreased along the
intensity (Figure 13c)fluid
wasfield from the
distributed in entrance (212 Pa)
a non-uniform to the and
manner, exit the
(1.3maximum
Pa). The simulated
turbulent
turbulent intensity (Figure 13c) was distributed in a non-uniform manner, and
intensity appeared at the position (marked in red) where the two liquids intersected. The fluid velocity the maximum
turbulent
(Figure 13d) intensity
was also appeared at the position
non-uniformly (marked
distributed, in red)
and the where velocity
maximum the two appeared
liquids intersected. The
at the position
fluid velocity
(marked in red)(Figure
where13d)
thesewas
twoalso non-uniformly
liquids mixed together. distributed, and the
Furthermore, themaximum velocity
partial pathlines appeared
(Figure 13e)
at the position (marked in red) where these two liquids mixed together. Furthermore,
were simulated to visualize the two-liquid mixing process, and the results at different steps agreed the partial
pathlines (Figureof13e)
with the results the were simulated
perfusion processto in
visualize the two-liquid
Figure 12c. mixing process,
The CFD simulation resultsand thebe
could results
used at
to
different
further steps
study
Pharmaceutics
agreed
the
2020,
with
12,influence
207
the results of the perfusion process in Figure 12c. The CFD
of material strength and channel design on the mixing performance.14 of 16 simulation
results could be used to further study the influence of material strength and channel design on the
mixing performance.

Figure 13. Simulated

Figure 13. Simulated 2D
2D cross-section
cross-section flow
flow field
field of
of the
the fluid
fluid mixing
mixing model:
model: (a)
(a) mesh
mesh model;
model; (b)
(b) total
total
pressure contour; (c) turbulent intensity contour; (d) velocity vector; and (e) partial pathlines at
pressure contour; (c) turbulent intensity contour; (d) velocity vector; and (e) partial pathlines at different
steps colored
different stepsby particles.
colored by particles.
4. Conclusion
4. Conclusion
Medical samples with specialized functional structures were successfully fabricated and thoroughly
Medical samples with specialized functional structures were successfully fabricated and
evaluated by utilizing the DLP technique. On one hand, under the optimal printing conditions for
thoroughly evaluated by utilizing the DLP technique. On one hand, under the optimal printing
external structures, various implants were successfully produced, and their potentials as implants
conditions for external structures, various implants were successfully produced, and their potentials
for drug delivery were proven. On the other hand, under the optimal printing conditions for
as implants for drug delivery were proven. On the other hand, under the optimal printing conditions
internal structures, various physiological channels were successfully prepared, and their application in
for internal structures, various physiological channels were successfully prepared, and their
perfusion evaluation was confirmed. Further studies are still necessary to investigate the possibilities
application in perfusion evaluation was confirmed. Further studies are still necessary to investigate
the possibilities of improving the printing accuracy, and the in vivo evaluation methodology should
be established to further illustrate the biomechanical behavior and flow field of the printed implants.

Author Contributions: Conceptualization, Y.Y., Y.Z. and G.Y.; Data curation and formal analysis,
Y.Y., Y.Z. and X.L.; Methodology; Y.Y. and Y.Z.; Writing-original draft, Y.Y. and Y.Z.; Writing-
Review and Editing, Q.Y. and G.Y.; Funding Acquisition, Y.Y., Q.Y. and G.Y.
Pharmaceutics 2020, 12, 207 14 of 16

of improving the printing accuracy, and the in vivo evaluation methodology should be established to
further illustrate the biomechanical behavior and flow field of the printed implants.

Author Contributions: Conceptualization, Y.Y., Y.Z. and G.Y.; Data curation and formal analysis, Y.Y., Y.Z. and
X.L.; Methodology; Y.Y. and Y.Z.; Writing—Original draft, Y.Y. and Y.Z.; Writing—Review and Editing, Q.Y. and
G.Y.; Funding Acquisition, Y.Y., Q.Y. and G.Y. All authors have read and agreed to the published version of
the manuscript.
Acknowledgments: This work was supported by the China Scholarship Council (CSC No. 201808330014),
the National Natural Science Foundation of China (No. 81402873) and Natural Science Foundation of Zhejiang
Provincial (No. LY20B060007).
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature
DLP digital light processing
PEGDA polyethylene glycol diacrylate
DPPO diphenyl(2,4,6-trimethylbenzoyl) phosphine oxide
DS diclofenac sodium
Dmin the minimum printable diameter
∆L inner arc length
∆A cavity area
Rf cavity fusion rate
σb tensile stress at break
εb elongation at break
σv Von Mises stress
εv Von Mises strain
CFD computational fluid dynamics
Rx-I 100% PEGDA
Rx-II 75% PEGDA with 0.25 mg/mL tartrazine
Rx-III PEGDA: PEG300 = 80: 20

References
1. Krkobabic, M.; Medarevic, D.; Cvijic, S.; Grujic, B.; Ibric, S. Hydrophilic excipients in digital light processing
(DLP) printing of sustained release tablets: Impact on internal structure and drug dissolution rate. Int. J.
Pharm. 2019, 572, 118790. [CrossRef]
2. Lin, W.S.; Harris, B.T.; Pellerito, J.; Morton, D. Fabrication of an interim complete removable dental prosthesis
with an in-office digital light processing three-dimensional printer: A proof-of-concept technique. J. Prosthet.
Dent. 2018, 120, 331–334. [CrossRef]
3. Kim, S.Y.; Shin, Y.S.; Jung, H.D.; Hwang, C.J.; Baik, H.S.; Cha, J.Y. Precision and trueness of dental models
manufactured with different 3-dimensional printing techniques. Am. J. Orthod. Dentofac. Orthop. 2018, 153,
144–153. [CrossRef]
4. Hong, H.; Seo, Y.B.; Kim, D.Y.; Lee, J.S.; Lee, Y.J.; Lee, H.; Ajiteru, O.; Sultan, M.T.; Lee, O.J.; Kim, S.H.; et al.
Digital light processing 3D printed silk fibroin hydrogel for cartilage tissue engineering. Biomaterials 2019,
232, 119679. [CrossRef]
5. Kim, S.H.; Yeon, Y.K.; Lee, J.M.; Chao, J.R.; Lee, Y.J.; Seo, Y.B.; Sultan, M.T.; Lee, O.J.; Lee, J.S.; Yoon, S.;
et al. Precisely printable and biocompatible silk fibroin bioink for digital light processing 3D printing.
Nat. Commun. 2018, 9, 1620. [CrossRef] [PubMed]
6. Kadry, H.; Wadnap, S.; Xu, C.; Ahsan, F. Digital light processing (DLP) 3D-printing technology and
photoreactive polymers in fabrication of modified-release tablets. Eur. J. Pharm. Sci. 2019, 135, 60–67.
[CrossRef] [PubMed]
7. Eshel-Green, T.; Eliyahu, S.; Avidan-Shlomovich, S.; Bianco-Peled, H. PEGDA hydrogels as a replacement for
animal tissues in mucoadhesion testing. Int. J. Pharm. 2016, 506, 25–34. [CrossRef]
Pharmaceutics 2020, 12, 207 15 of 16

8. Burke, G.; Barron, V.; Geever, T.; Geever, L.; Devine, D.M.; Higginbotham, C.L. Evaluation of the materials
properties, stability and cell response of a range of PEGDMA hydrogels for tissue engineering applications.
J. Mech. Behav. Biomed. Mater. 2019, 99, 1–10. [CrossRef]
9. Na, K.; Shin, S.; Lee, H.; Shin, D.; Baek, J.; Kwak, H.; Park, M.; Shin, J.; Hyun, J. Effect of solution viscosity
on retardation of cell sedimentation in DLP 3D printing of gelatin methacrylate/silk fibroin bioink. J. Ind.
Eng. Chem. 2018, 61, 340–347. [CrossRef]
10. Mishra, R.; Roux, B.M.; Posukonis, M.; Bodamer, E.; Brey, E.M.; Fisher, J.P.; Dean, D. Effect of prevascularization
on in vivo vascularization of poly(propylene fumarate)/fibrin scaffolds. Biomaterials 2016, 77, 255–266.
[CrossRef] [PubMed]
11. Tseng, H.; Cuchiara, M.L.; Durst, C.A.; Cuchiara, M.P.; Lin, C.J.; West, J.L.; Grande-Allen, K.J. Fabrication
and mechanical evaluation of anatomically-inspired quasilaminate hydrogel structures with layer-specific
formulations. Ann. Biomed. Eng. 2013, 41, 398–407. [CrossRef] [PubMed]
12. Baroli, B. Photopolymerization of biomaterials: Issues and potentialities in drug delivery, tissue engineering,
and cell encapsulation applications. J. Chem. Technol. Biotechnol. 2006, 81, 491–499. [CrossRef]
13. Zhang, Z.F.; Ma, X.; Wang, H.; Ye, F. Influence of polymerization conditions on the refractive index of
poly(ethylene glycol) diacrylate (PEGDA) hydrogels. Appl. Phys. A Mater. Sci. Process. 2018, 124. [CrossRef]
14. Grigoryan, B.; Paulsen, S.J.; Corbett, D.C.; Sazer, D.W.; Fortin, C.L.; Zaita, A.J.; Greenfield, P.T.; Calafat, N.J.;
Gounley, J.P.; Ta, A.H.; et al. Multivascular networks and functional intravascular topologies within
biocompatible hydrogels. Science 2019, 364, 458–464. [CrossRef] [PubMed]
15. Goyanes, A.; Det-Amornrat, U.; Wang, J.; Basit, A.W.; Gaisford, S. 3D scanning and 3D printing as innovative
technologies for fabricating personalized topical drug delivery systems. J. Control. Release 2016, 234, 41–48.
[CrossRef]
16. Guerra, A.J.; Lammel-Lindemann, J.; Katko, A.; Kleinfehn, A.; Rodriguez, C.A.; Catalani, L.H.; Becker, M.L.;
Ciurana, J.; Dean, D. Optimization of photocrosslinkable resin components and 3D printing process
parameters. Acta Biomater. 2019, 97, 154–161. [CrossRef] [PubMed]
17. Mitteramskogler, G.; Gmeiner, R.; Felzmann, R.; Gruber, S.; Hofstetter, C.; Stampfl, J.; Ebert, J.; Wachter, W.;
Laubersheimer, J. Light curing strategies for lithography-based additive manufacturing of customized
ceramics. Addit. Manuf. 2014, 1–4, 110–118. [CrossRef]
18. Wu, D.; Zhao, Z.; Zhang, Q.; Qi, H.J.; Fang, D. Mechanics of shape distortion of DLP 3D printed structures
during UV post-curing. Soft Matter 2019, 15, 6151–6159. [CrossRef]
19. Jin, T.; Stanciulescu, I. Numerical investigation of the influence of pattern topology on the mechanical
behavior of PEGDA hydrogels. Acta Biomater. 2017, 49, 247–259. [CrossRef]
20. Goyanes, A.; Robles, M.P.; Buanz, A.; Basit, A.W.; Gaisford, B.S. Effect of geometry on drug release from 3D
printed tablets. Int. J. Pharm. 2015, 494, 657–663. [CrossRef]
21. Yang, Y.; Shen, L.; Yuan, F.; Fu, H.; Shan, W. Preparation of sustained release capsules by electrostatic dry
powder coating, using traditional dip coating as reference. Int. J. Pharm. 2018, 543, 345–351. [CrossRef]
22. Yang, Y.; Fang, J.; Shen, L.; Shan, W. Simulation and evaluation of rupturable coated capsules by finite
element method. Int. J. Pharm. 2017, 519, 220–229. [CrossRef]
23. Morgan, S.R.; Pilia, N.; Hewitt, M.; Moses, R.L.; Moseley, R.; Lewis, P.N.; Morrison, P.W.J.; Kelly, S.L.;
Parker, J.E.; Whitaker, D.; et al. Controlled In Vitro Delivery of Voriconazole and Diclofenac to the
Corneausing Contact Lenses for the Treatment of Acanthamoeba Keratiti. Int. J. Pharm. 2020, 579, 119102.
[CrossRef] [PubMed]
24. Alvarez-Lorenzo, C.; Grinberg, V.Y.; Burova, T.V.; Concheiro, A. Stimuli-sensitive crosslinked hydrogels as
drug delivery systems: Impact of the drug on the responsiveness. Int. J. Pharm. 2020, 579, 119157. [CrossRef]
[PubMed]
25. Li, L.; Xie, C.; Xiao, X. Polydopamine modified TiO2 nanotube arrays as a local drug delivery system for
ibuprofen. J. Drug Deliv. Sci. Technol. 2020, 56, 101537. [CrossRef]
26. Yang, Y.; Zheng, N.; Zhou, Y.; Shan, W.; Shen, J. Mechanistic study on rapid fabrication of fibrous films via
centrifugal melt spinning. Int. J. Pharm. 2019, 560, 155–165. [CrossRef]
Pharmaceutics 2020, 12, 207 16 of 16

27. Wang, J.; Goyanes, A.; Gaisford, S.; Basit, A.W. Stereolithographic (SLA) 3D printing of oral modified-release
dosage forms. Int. J. Pharm. 2016, 503, 207–212. [CrossRef]
28. Yang, Y.; Wang, H.; Li, H.; Ou, Z.; Yang, G. 3D printed tablets with internal scaffold structure using ethyl
cellulose to achieve sustained ibuprofen release. Eur. J. Pharm. Sci. 2018, 115, 11–18. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).