1. Introduction
Polyglycolic acid (PGA) nanoparticles have gained significant attention in recent years as promising nanocarriers for various biomedical applications. PGA, a biocompatible and biodegradable polymer, possesses unique properties such as high water solubility, stability, and a high payload capacity, making it an ideal candidate for drug delivery systems. The small size of PGA nanoparticles, typically in the range of tens to hundreds of nanometers, allows for efficient cellular uptake and enhanced bioavailability. Moreover, their surface can be easily modified for targeted drug delivery, enabling specific interactions with diseased tissues or cells. With their favorable physicochemical properties and versatile surface functionalization, PGA nanoparticles hold great potential for revolutionizing drug delivery, imaging agents, and theranostic applications, contributing to advancements in precision medicine and personalized treatments. The formulation of PGA nanoparticles is typically carried out using emulsification solvent evaporation via batch synthesis methods, which involve the mixing of the organic phase, consisting of PGA, and the aqueous phase, followed by the evaporation of organic solvent [
1]. However, batch synthesis methods, which are the regular method to prepare PGA nanoparticles, are associated with several limitations, including low productivity and poor scalability [
2]. These limitations have led to increasing interest in the development of continuous synthesis methods, such as flow chemistry, for the production of PGA nanoparticles.
Flow chemistry, also known as continuous flow chemistry or continuous flow synthesis, involves conducting chemical reactions in a continuous flow of reagents through a reactor. The adoption of flow chemistry has revolutionized pharmaceutical manufacturing, enabling the rapid synthesis of active pharmaceutical ingredients (APIs), pharmaceutical excipient, and formulation, including nanoparticle synthesis. Flow chemistry has emerged as a promising approach for nanoparticle synthesis, offering several distinct advantages. Through precise control over reaction parameters, such as temperature, pressure, and residence time, flow chemistry allows one to achieve better control over the size, shape, and composition of nanoparticles, ensuring reproducibility and reducing batch-to-batch variations. Additionally, the rapid mixing and efficient heat transfer in flow systems improve reaction kinetics, resulting in nanoparticles with narrow size distributions, high purity, and improved crystallinity. Flow chemistry also offers scalability and high throughput, allowing for large-scale production and the rapid screening of reaction conditions. Moreover, the safety and sustainability aspects of flow chemistry, with reduced risk in handling hazardous materials and minimized formation of undesired by-products, further enhance its appeal as a technique for nanoparticle synthesis [
3,
4,
5,
6,
7].
However, the utilization of general reactors, such as tubular reactors or coil reactors, in nanoparticle synthesis though flow chemistry presents several noteworthy limitations that necessitate careful consideration. One such limitation is the challenge of scaling up the synthesis process. As the reactor volume increases, maintaining uniform flow conditions and consistent heat and mass transfer becomes more complex, hindering the large-scale production of nanoparticles. This limitation is closely linked to the design of the tubular reactor. Factors such as reactor diameter, length, and configuration directly impact flow characteristics and residence time distribution, ultimately affecting the mixing efficiency. Furthermore, the flow regime inside the reactor significantly influences the overall mixing efficiency. Laminar flow, characterized by smooth, parallel layers of fluid, generally results in poor mixing, while turbulent flow, with its chaotic fluid motion, promotes better mixing by increasing the contact between reactants. The flow characteristics being examined are dependent on the flow rate utilized in the synthesis process. It is important to note that tubular reactors have a limited range of flow rates in which optimal mixing efficiency can be achieved [
8,
9]. Nevertheless, ongoing research and technological advancements are focused on overcoming these limitations and new reactor designs are needed to address this issue, aiming to enhance the effectiveness and broaden the applicability of flow chemistry reactors. The design of a reactor’s geometry plays a crucial role in determining flow streamlines and, subsequently, mixing behavior. When streamlines are parallel or minimally interact, there is limited mixing between fluid streams, observed in laminar flows. Conversely, convoluted, intersecting, and complex streamline patterns enhance mixing, characteristic of turbulent flows. Turbulence facilitates the exchange of momentum, energy, and mass, promoting effective mixing through flow instabilities, eddies, and vortices. These structures displace and intermingle fluid particles, resulting in a more homogeneous distribution of properties. Accordingly, mixing efficiency is achieved by promoting vortex patterns in the reactor’s design [
10,
11].
The present study focuses on the design and development of a 3D-printed vortex tube reactor to improve the productivity of PGA nanoparticles via flow chemistry. The optimization of the reactor design was achieved through the use of computer-aided design (CAD) and the design of experiments (DOE). CAD software was employed to model the reactor geometry, while DOE methodology, central composite design, was used to systematically study the effect of different design parameters on the performance of the reactor. In the evaluation of the reactor models, computational fluid dynamics (CFD) simulations were utilized to evaluate the mixing performance and the flow characteristics within the reactor. Mixing efficiency is a critical factor in the synthesis process as it affects both the rate and uniformity of the reaction. To quantify the degree of homogeneity in the mixing process, a mixing index was employed as a quantitative measure [
12]. Additionally, the flow characteristics of the reactors were evaluated. One key parameter used to analyze the flow characteristics is the Reynolds number, which quantifies the relative importance of inertial forces to viscous forces in the fluid flow. By calculating the Reynolds number, we were able to assess the flow regime within the reactor and determine whether it was laminar or turbulent [
13,
14]. Incorporating CFD simulations with DoE methodology allowed for a comprehensive evaluation of both the mixing performance and flow characteristics, facilitating the refinement and enhancement of the optimized vortex tube reactor model.
Furthermore, the optimized reactor design was fabricated using fused deposition modeling (FDM), a popular 3D printing technique. The principle of FDM involves melting and extruding thermoplastic materials through a nozzle, layer by layer, to create a three-dimensional object from CAD model. The choice of polypropylene (PP) as the polymer for the fabrication of the vortex tube reactor was based on its favorable properties, including high temperature resistance, high strength, and good chemical resistance [
15]. In assessment of the 3D-printed vortex tube reactor, the dispersion experiments were conducted to study the residence time distribution (RTD) and deviation within the reactor. RTD is a measure of the time that a fluid particle spends within the reactor, and deviation is a measure of the uniformity of the dispersion. These experiments provided crucial information on the performance of the reactor and helped to validate the optimization process [
16]. Finally, the results of this study hold significant implications for the effect of flow parameters, which consist of the total flow rate and the aqueous-to-organic volumetric ratio, in the formulation of PGA nanoparticle production through flow chemistry. The study demonstrates the feasibility of using a 3D-printed vortex tube reactor for this purpose [
17]. By utilizing the 3D-printed vortex tube reactor, the researchers achieved efficient control over the flow parameters, enabling precise manipulation of the total flow rate and aqueous-to-organic volumetric ratio during PGA nanoparticle production. This novel approach not only paves the way for the scalable and reproducible synthesis of PGA nanoparticles but also opens up new avenues for further research in this area, such as exploring the impact of different flow parameter combinations on nanoparticle characteristics and optimizing the reactor design for enhanced performance.
2. Materials and Methods
2.1. Materials
Polyglycolic acid (PGA), cosmetic grade, was purchased from Chemipan Corporation Co., Ltd. (Bangkok, Thailand). Polyvinyl alcohol (PVA), laboratory grade, was purchased from Sigma Aldrich, Inc. (St. Louis, MO, USA). Erythrosine, food grade, was purchased from Adinop Co., Ltd. (Bangkok, Thailand). Dichloromethane, laboratory grade, was purchased from RCI Labscan Ltd. (Bangkok, Thailand). Acetone, laboratory grade, was purchased from QRëC (Chonburi, Thailand). Polypropylene (PP) filament was obtained from B and Brothers Co., Ltd. (Samut Sakhon, Thailand).
2.2. The Vortex Tube Reactor Model Design and Model Development according to Design of Experiment
The design of the vortex tube reactor was carried out using Shapr3D software (version number: 5.450.0.5689 #a0a93e9a (Educational license)). The reactor designed for flow chemistry is composed of two main components, which are a mixer and a vortex tube reactor. The mixer part design of the reactor is shown in
Figure 1. The model is composed of two inlets (4 mm diameter), and the flow streamlines at mixer inlet 2 are interconnected through a fin for mixing with the flow streamline at mixer inlet 1. The fluid flow is moved tangentially within the mixer part, and the whirling fluid flow is then forced by the fin to change direction and form a vortex streamline before exiting the mixer through the outlet. This vortex streamline then passes through the reactor.
The vortex tube reactor in this flow chemistry system is shown in
Figure 2. The model is designed with a focus on mixing performance and flow characteristics. The reactor is designed to create a vortex streamline that enhances mixing and mass transfer, resulting in greater productivity of the streamlines which are separated into 4 parts of the by-pass tube, intended to increase fluid velocity, which creates a suitable environment for vortex mixing of the reagents inside the chamber.
The design of the reactor and the operating parameters were studied and optimized using the experimental design (DoE), the central composite design. As presented in
Table 1, the interest factors in this study were reactor length (X
1), internal tube to chamber wall width (X
2), percentage of internal tube length per reactor length (X
3), and total flow rate (X
4). Moreover, the experimental runs were generated and tested using CFD (Flow simulation, Solidworks2021), and then the following responses, including mixing index (Y
1) and Reynolds number (Y
2), were monitored and recorded.
2.3. Numerical Analysis
The mathematical models and equations were used to conduct the computational fluid dynamics (CFD) simulations. The simulations were run using flow simulation software (Solidworks2021). These equations take into account the effects of viscosity, pressure, and density to predict the flow behavior in a particular system. The governing equations of the three-dimensional, steady, and incompressible flows are the Navier–Stokes equations which can be solved using the finite element method (FEM) or finite volume method (FVM). The Navier–Stokes equations are given in the following:
where
C is the species concentration,
D is the diffusion coefficient, and
V is the fluid velocity.
The species transport model was employed as it is the most commonly used method to model the mixing of miscible fluids, taking into account the mass diffusion coefficient. The code solved the conservation equations that described the sources of convection and diffusion for each species. The local mass fraction of each species was predicted by solving the convection–diffusion equation, with the sum of the mass fractions of all species being equal to one.
The CFD code was used to solve the governing equations of three-dimensional, steady, and incompressible flows (continuity and momentum). The species transport model, which takes into account the mass diffusion coefficient, was also used. The conservation equations for each species were solved by considering convection and diffusion mechanisms. The liquid properties from the engineering database (Solidworks), shown in
Table S1 in the Supplementary Materials, were used for CFD calculation, and the mixing was governed mainly by chaotic advection. The boundary conditions were rigid and non-movable walls, a non-slip velocity condition, and constant uniform velocity at the inlets. The mass fraction ratio of water/ethanol was equal to 1:0 at inlet 1 and 0:1 at inlet 2, while at the outlet, environment pressure and temperature at 25 °C were assumed.
2.3.1. Mixing Index Evaluation
The evaluation of a mixing index (MI) is established based on the deviation of the mass fraction in a cross-section. This method involves dividing the cross-section into several cells, which in this case are the grid elements. The mass fraction is then calculated for each cell. Ideal mixing is characterized by a uniform distribution of particles from the two branches throughout the cross-section, resulting in equal proportions of particles with indices 0 and 1. Optimal mixing should therefore result in an even partition of particles with indices 0 and 1. The calculation of the MI in each cross-section of the micromixer is conducted as follows:
where σ is the standard deviation of the mass fraction in a cross-section, which is determined using the integrated function of the CFD code:
The standard deviation is at its maximum for unmixed fluids and at its minimum for perfectly mixed fluids.
N represents the total number of cells in a cross-section, and
is the average mass fraction. Mixing is considered to have occurred when the mass fractions of the two fluids are equal, meaning that they both reach the value of 0.5. The maximum standard deviation σ
0 over the data range is calculated as follows:
A mixing index of MI = 1 indicates perfect mixing (σ = 0), and MI = 0 indicates an unmixed state (σ = σ
0). A higher value of MI indicates a more homogeneous concentration and better mixing performance [
12].
2.3.2. Reynolds Number Expression
The generalized Reynolds number (
) expression for the vortex tube reactor was calculated using computational fluid dynamics (CFD) simulations, seen in the following equation:
The following parameters were used in the calculation of the Reynolds number expression. Fluid density (
ρ) and dynamic viscosity (
μ) were taken as inputs based on the fluid used in the vortex tube reactor, which was obtained from an engineering database (Solidworks2021) that provides the values for the fluid as a function of temperature, as seen in
Table S1 in the Supplementary Materials. The velocity is characterized by the average fluid inlet velocity (
v) that is calculated using the integrated function of the CFD code. The velocity was determined by measuring the fluid flow rate and dividing it by the cross-sectional area of the reactor, and the entire fluid flow was assumed to exit through the fluid outlet of the vortex tube reactor. The diameter (
D) of the reactor was taken as an input, based on the overall body of the vortex tube reactor [
13,
18,
19].
2.4. Mesh Independency Test
This sub-section presents a detailed methodology for the mesh sensitivity analysis conducted in the vortex tube reactor study. To evaluate the sensitivity of the simulation results, comprehensive mesh sensitivity analysis was conducted for each mesh resolution. The analysis involved systematically varying seven levels of automated meshing while keeping the other simulation parameters constant. There are two types of mesh: the tetrahedral shape is suitable for curved surfaces, and the hexahedral shape is appropriate for flat surfaces. The difference in mesh refinement is shown in
Table S2 in the Supplementary Materials. By comparing the simulation results obtained from different mesh resolutions, the convergence and stability of the results were assessed. The analysis aimed to determine the optimal mesh size and resolution that would ensure accurate and reliable results in the numerical simulations.
2.5. OptiMization of Geometrical Vortex Tube Reactor Model Design
Once the simulation of each sample is completed, the Design-Expert 11 software is used to calculate the relationship patterns using partial least square models and 3D surface plots. Then, the optimized vortex tube reactor is calculated using the setting criteria shown in
Table 2, where Y
1 means that the mixing index must be close to 1 (target to 1), which indicates complete mixing, and Y
2 means that the Reynolds number must be maximized, which indicates that the flow characteristics are closely turbulent. The prediction of the optimized vortex tube reactor design parameters is optimized for optimal mixing efficiency.
2.6. Fabrication of 3D-Printed Vortex-Tube-Reactor-Based FDM
The fabrication process of the 3D-printed vortex tube reactor was carried out using fused deposition modeling (FDM) technology (UP mini 2, Beijing Tiertime Technology Co., Ltd., Beijing, China) with polypropylene (PP) filament. The design of the vortex tube reactor was created using computer-aided design (CAD) software. The STL file was then converted and sliced into G-code using the slicing software (version no.: 2.6.49.627) of UPstudio (Beijing Tiertime Technology Co., Ltd., Beijing, China), and then loaded into the FDM printer using the printing condition shown in
Table 3.
After completing the printing process, the 3D-printed vortex tube reactor was removed from the build platform and cleaned to remove any excess material. In the post-processing stage, dichloromethane was flowed through the reactor for 5 min at a flow rate of 10 mL/min to dissolve any remaining PP and seal any potential leaks. This was followed by a 15 min flow of ethanol at a flow rate of 10 mL/min, which was used to flush out the dichloromethane and ensure that the reactor was free of any remaining residue. The 3D-printed vortex tube reactor was then assembled and tested to ensure its functionality.
2.7. Dispersion Experiment
A dispersion experiment was conducted in the optimized reactor to observe the effect of fluid dynamics on the dispersion behavior of the tracer within the reactor. A total of 0.5 mL of 1% (w/v) erythrosine solution, as a tracer, was injected into the reactor with a flow rate of 10–120 mL/min of DI water. Samples were collected at regular time intervals and the absorbance was measured using a UV-VIS spectrophotometer (2J1-0004, Hitachi, Japan) at a wavelength of 530 nm. The absorbance measurements of tracer concentration () at the time () were used to evaluate the residence time distribution function , the mean residence time (), and the distribution variance ().
The residence time distribution function (RTD) or
function can be calculated from the sample data collected at various time intervals. The RTD is the fraction of total fluid in the reactor that leaves at a specific time.
The mean residence time (
) can then be calculated from the RTD by finding the first moment of the RTD:
The distribution variance can also be calculated from the RTD as the second moment of the RTD:
To facilitate a comparison of the results obtained from different flow rates of the vortex tube reactor configurations, it is necessary to use dimensionless units. The dimensionless concentration (
) can be calculated by dividing the tracer concentration at time (
) by the initial tracer concentration (
):
Similarly, the dimensionless time (
) can be calculated by dividing time (
t) by the mean residence time (
):
Finally, the dimensionless
can be obtained using
and
:
2.8. Synthesis and Evaluation of PGA Nanoparticles
PGA nanoparticles were synthesized using a vortex tube reactor by combining an aqueous solution of 0.5%
w/
v of PVA in DI water and an organic phase, which contained 0.5%
w/
v PGA in acetone. The impact of different system parameters on the size and PDI of the final formulation was studied by adjusting the total flow rate and aqueous-to-organic volumetric ratio of the incoming streams from 68.92 to 140 mL/min and from a ratio of 3:1 to 9:1 (aqueous-to-organic volumetric ratio), respectively. The total flow rate represents the combined flow rate of the organic and aqueous phases that are pumped through the two inlets, while the aqueous-to-organic volumetric ratio refers to the volume ratio of the aqueous and organic phases, whereby the experimental set-up is shown in
Figure 3.
2.9. The Physical Characteristics of PGA Nanoparticles
The particles’ size and polydispersity index were evaluated using a Zetasizer (MAL1070387, Malvern, UK). To perform the measurements, the PGA nanoparticles were diluted to a concentration of 0.1% w/v using deionized water and agitated for 3 min before analysis. The average values and standard error were calculated based on measurements from three batches of samples. Additionally, the surface morphology of the PGA nanoparticles was examined using scanning electron microscopy (SEM) (LEO 1450 VP, Carl Zeiss, Germany). The samples were concentrated via centrifugal ultrafiltration (MWCO 30 kDa, Amicon®Ultra-15 filter, Merck, Germany) at 3000 rpm for 20 min. A small amount of the sample was fixed on an SEM stub using double-sided adhesive tape and coated with gold. After thorough drying, the samples were observed using SEM.
2.10. Statistical Analysis
The correlations of CFD parameters were determined using Design Expert 11 (Stat-Ease, Inc., Minneapolis, MN, USA) for statistical analysis and graphing of the model response surface. The experimental data were reported as the mean ± standard error of three replicates. To identify statistically significant variations (p < 0.05), ANOVA and Levene’s test for homogeneity of variance were conducted using SPSS version 10.0 for Windows (SPSS Inc., New York, NY, USA). For post hoc testing of multiple comparisons (p < 0.05), either the Scheffé or Games–Howell test was employed, depending on the significance of Levene’s test.