3D Printed Microfluidic Separators for Solid/Liquid Suspensions

Abstract

This study investigates the fabrication of 3D-printed microfluidic devices for solid/liquid separation, focusing on additive manufacturing technologies. Stereolithography (SLA) and fused filament fabrication (FFF) were used to create microseparators with intricate designs optimized for separation efficiency. Model suspensions containing quartz sand, nano-calcium carbonate, and talc-based baby powder in water were prepared using an electric magnetic stirrer and conveyed into the microseparator via a peristaltic pump. Different flow rates were tested to evaluate their influence on the separation efficiency. The highest separation efficiency for the model systems was observed at a flow rate of 200 mL min⁻¹. This was due to the increased turbulence at higher flow rates, which hindered the secondary flow perpendicular to the primary flow direction. The particle size distribution before and after separation was analyzed using sieve and laser diffraction, and particle morphology was inspected by scanning electron microscopy. The laser diffraction analysis revealed post-separation particle size distributions, indicating that Outlet 1 (external stream) consistently captured larger particles more effectively than Outlet 2 (internal stream). This work highlights the potential of additive manufacturing to produce customized microfluidic devices, enabling rapid prototyping and fine-tuning of complex geometries, thus enhancing separation efficiency across various industrial applications.

1. Introduction

Additive manufacturing, recognized as a pivotal catalyst for the Fourth Industrial Revolution, stands at the forefront of transformative forces in industry, fundamentally reshaping established manufacturing and design paradigms [1,2,3]. Its profound impact is evidenced by its capacity to facilitate the fabrication of intricately detailed structures and objects deemed unattainable by conventional methods. This paradigm shift brings significant dynamism to the manufacturing landscape, impacting various sectors such as chemical engineering, aerospace, automotive, medical, and energy, emphasizing versatility and potential [1,3,4].

At the center of additive manufacturing is the detailed coordination between digital design and fabrication, managed through computer-aided design (CAD) software or the advanced features of 3D scanning technology [1,3,5]. In this digital realm, models are crafted to meet the specific requirements of diverse applications and manufacturing demands. The increasing availability of additive manufacturing software also represents the improving accessibility and user-friendly nature of these fabrication technologies [1,2,3].

Among additive manufacturing technologies, Fused Filament Fabrication (FFF), also known as Fused Deposition Modeling (FDM), stands out for its versatility and widespread use [1,2,3]. FFF involves the layer-by-layer deposition of thermoplastic filaments, which are heated and extruded through a nozzle to create intricate three-dimensional structures. This technology offers a cost-effective and accessible way to produce prototypes, components, and complex geometries across various industries [1,3,5]. Its ability to work with a wide range of materials, most frequently ABS, PLA, and PETG, makes it suitable for rapid prototyping, tooling, and small-batch production. With advances in FFF technology, including improved precision, material options, and support structures, it continues to revolutionize manufacturing processes and expand access to additive manufacturing capabilities [1,3,4].

Another additive manufacturing technology, stereolithography (SLA), stands out for its precision and handling of photopolymer resin materials [1,6]. Through the controlled use of a laser beam, SLA transforms liquid resin into intricate three-dimensional structures, layer by layer [1,7]. Its exceptional accuracy makes it invaluable in industries where precision is crucial, offering unmatched reliability in the production of prototypes and components [1,8].

Solid-liquid separation, a fundamental process in various industries, is crucial for the purification of liquids, the recovery of valuable solids, and the minimization of environmental impact [9,10,11]. This process involves the separation of suspended solids from a liquid phase, typically accomplished through physical or mechanical means [9,10]. Common methods include filtration, sedimentation, and flotation, each tailored to the specific application based on factors such as particle size, shape, density, and desired separation efficiency [9,11]. As industries prioritize sustainability and resource conservation, the development of innovative solid-liquid separation technologies remains a focal point for optimizing processes, reducing waste, and enhancing overall efficiency [9,12].

Continuous separation processes play a pivotal role in modern industrial operations, offering advantages in terms of efficiency, productivity, and automation compared to batch processes. Unlike batch separation, where materials are processed in discrete batches, continuous separation involves an uninterrupted flow of feed material through the separation unit. This continuous flow enables steady-state operation, leading to more consistent product quality and higher throughput. Various techniques, such as continuous filtration, centrifugation, chromatography, and membrane separation, are employed across industries ranging from pharmaceuticals to petrochemicals, contributing to enhanced sustainability and cost-efficiency in industrial operations [11,12].

Microfluidic separation techniques have earned considerable interest due to their potential applications in various fields, including chemical engineering, biotechnology, pharmaceuticals, and environmental monitoring, as outlined in the papers by Wang et al. [10], Amin et al. [13], Heuer et al. [14], and Weisgrab et al. [15]. These techniques offer precise control of fluid flow at the microscale, enabling efficient separation of particles and molecules based on their physical properties. Among the methods employed for microfluidic separation, devices fabricated by additive manufacturing have emerged as a promising avenue for achieving enhanced performance and customization. This is shown in the papers of Duong et al. [16], Garcia et al. [17], Kim et al. [18], and Dong et al. [19], where the authors present various applications of 3D-printed devices for microfluidic purification and separation.

Inertial microfluidics, an emerging field, exploits the behavior of fluids in microscale channels under the influence of inertial forces. In contrast to conventional microfluidic systems, where viscous forces dominate, inertial microfluidics operates in a regime where inertial effects become significant. By harnessing inertial forces, particles can be sorted, focused, or separated based on their size, shape, or deformability, making inertial microfluidics a versatile tool in various applications such as biological sample processing, drug delivery, and diagnostic assays [14,20]. Some authors have presented 3D-printed inertial microfluidic devices for separation. For example, Bazaz et al. [21] present an inertial microfluidic device for cell separation. However, instead of 3D printing it completely in one piece, the authors bonded the 3D-printed part to a transparent PMMA sheet using a double-coated pressure-sensitive adhesive tape. Enders et al. [22] present a high-efficiency 3D printed microfluidic spiral separation device, also for cell retention. No papers were found in the literature describing the separation of water-powder dispersions presented in this study.

Microfluidic devices, which are integral in various fields such as biomedical diagnostics, chemical synthesis, and environmental monitoring, are fabricated using diverse manufacturing techniques beyond additive manufacturing. These methods cater to different requirements in terms of complexity, precision, and scalability, providing customized solutions based on the device complexity and the application requirements [22,23,24,25]. Although the term microfluidics encompasses various definitions, it generally refers to devices designed to precisely manipulate very small volumes of liquid, typically in the nanoliter (nL) to microliter (µL) range, using small-scale geometries. The compact nature of these devices offers several advantages, including minimal sample and reagent consumption, cost-effectiveness, rapid prototyping, reduced power requirements, and enhanced portability. Additionally, the distinctive behavior of fluids within these small-scale devices leads to unique physical phenomena, enabling innovative applications that are not feasible at larger scales [26]. Typical channel dimensions of millifluidics, microfluidics, and nanofluidics devices are on the scale of 1 mm–10 mm, 100 nm–1 mm, and <100 nm, respectively [27]. However, all small-scale devices can be categorized under microfluidics as a general name, as opposed to macrofluidics, as devices that handle small liquid volumes. This is demonstrated in the paper of Waghwani et al. [28], where the authors describe microfluidic chips with channel dimensions in millimeters.

Manufacturing monitoring technologies play a crucial role in flow chemistry, where efficient separation ensures the production of small particles with defined sizes and shapes, thereby optimizing the overall process and minimizing wastage through intelligent control. This study demonstrates the use of 3D printing technologies, stereolithography (SLA) and fused filament fabrication (FFF), to create microfluidic devices with intricate designs. This represents a significant innovation, as conventional manufacturing methods often struggle to achieve the same level of precision and complexity. Conventional manufacturing methods for microfluidic devices are often limited by their complexity and cost. The study addresses these limitations by demonstrating that 3D printing can be used to produce effective and affordable microfluidic separators, making the technology more accessible.

Significant advances in the field of microfluidics and additive manufacturing are presented, providing practical solutions to existing challenges and setting the stage for future improvements. The integration of flexible control systems within flow chemistry setups allows for precise adjustments in flow rates, which directly impact the efficiency of separating particles of different sizes, ensuring that the process remains adaptable and efficient. In particular, this paper presents a comprehensive study of microfluidic separation by additive manufacturing, encompassing device design, powder characterization, and evaluation of separation efficiency. Through systematic experimentation and analysis, we aim to contribute to the advancement of microfluidic separation technologies and pave the way for their wider use in various industrial and scientific applications. By exploring the novel application of 3D printing in microfluidics, this study lays the foundation for further advancements. Intelligent preparation and maintenance strategies are essential in flow chemistry, where efficient separation processes are crucial for the consistent production of particles with specific characteristics, thereby reducing material wastage and enhancing the reliability of the manufacturing process. It opens up new possibilities for rapid prototyping and customization of microfluidic devices that can be tailored to specific industrial needs.

2. Materials and Methods

2.1. Materials

Quartz sand, obtained from the manufacturer Samoborka (Zagreb, Croatia), has a bulk density in the range of 1360 to 1480 g dm⁻³. Nano-calcium carbonate (CaCO₃), supplied by Schaefer Kalk GmbH (Verbandsgemeinde Diez, Germany), has a specific surface area of 8 m² g⁻¹ and a bulk density ranging from 200 to 600 g dm³. Baby powder, which is sold under the name Babymira posip by City Pharmacies Zagreb, consists of talc and zinc oxide.

2.2. Separator Design

When designing the separators using Autodesk Fusion 360 (v2.0.19440) computer-aided design (CAD) software, the dimensions of the separators were precisely defined, including width, length, thickness, spiral channel diameter, and the number of spiral turns. These dimensions are shown in Figure 1. The separator was designed based on previous research [21], with eight spiral turns, a single inlet, and two outlets. The inlet is located at the top of the separator, while the two outlets are positioned on the side. The two outlets are designated as Outlet 1 (external stream) and Outlet 2 (internal stream), as illustrated in Figure 1. The separators are named according to the manufacturing technology and the cross-section design: SLA 1 (square cross section), SLA 2 (circular cross section), and FFF (circular cross section). These technologies and shapes were chosen to assess the effect of different materials, manufacturing technologies, and channel shapes on the separation efficiency.

2.3. 3D Printing

Two 3D printing technologies were used to produce the separators: stereolithography (SLA) and fused filament fabrication (FFF). The SLA 3D printer used was the Formlabs Form 2 (Somerville, MA, USA), which utilizes Clear Photoreactive Resin from Formlabs. This resin is composed of methacrylate oligomer (≥75%–≤90%), methacrylate monomer (≥25%–≤50%), and the photoinitiator diphenyl (2,4,6 trimethylbenzoyl) phosphine oxide (≤1%). The FFF 3D printer used was the Original Prusa i3 MK3s+ (Prusa Research, Prague, Czechia), which uses polyethylene terephthalate glycol (PETG) filament from Devil Design (Mikołów, Poland)—a thermoplastic polymer filament with a diameter of 1.75 mm ± 5%.

The digital model from the CAD software was imported into the slicer software to define the print settings and segment the model into layers before 3D printing. In the PreForm slicer software (v3.0.0) for the SLA technology, the defined settings were a print angle of 45° on the x-axis, a print angle of 45° on the y-axis, a layer thickness of 0.1 mm, and no supports. In the PrusaSlicer software (v2.8.0) for FFF technology, the defined settings were an extruder temperature of 230 °C, a bed temperature of 80 °C, a layer thickness of 0.1 mm, 100% infill, and no supports (Figure 2).

Following SLA printing, the object was cleaned using isopropyl alcohol to remove any residual resin.

2.4. Separation Process

Suspensions containing powder and demineralized water were prepared with a powder mass fraction of 0.33% in water. In the case of the baby powder, where dispersion in the water proved challenging, isopropyl alcohol was added to the suspension at a mass fraction of 10%, which enabled the homogenization of this suspension. The separation apparatus consisted of a suspension prepared using an IKA RCT Digital electric magnetic stirrer (Staufen, Germany), a FlexiPump peristaltic pump from Interscience (Selangor, Malaysia), and a spiral 3D-printed separator.

The suspension from the stirred beaker was conveyed through the peristaltic pump into the microseparator. The use of a peristaltic pump was crucial to prevent contact with the internal components of the pump, considering the coarse dispersion of particles within the suspension. The operating parameters, particularly flow rates, were systematically varied to determine their effect on the separation efficacy. The flow rates tested were 200 mL min⁻¹, 250 mL min⁻¹, and 300 mL min⁻¹.

In order to determine separator efficiency, the samples collected from the separator outlets were transferred into test tubes and subsequently subjected to centrifugation using the Hettich Universal 320 Centrifuge (Tuttlingen, Germany). This procedure accelerated the sedimentation process of the powder from the suspension. The separated powders were then dried at room temperature and analyzed.

2.5. Characterization

The particle size distribution of each powder sample was determined before and after the separation process. Prior to separation, an analysis was performed to determine the particle size range and distribution within the inlet stream. Following separation, methods for particle size characterization of the powder from the outlet streams were used to investigate the correlation between particle size and separation efficiency.

The sieve analysis was used to determine the particle size distribution in different size intervals. Seven wire sieves with aperture sizes of 850 µm, 710 µm, 355 µm, 180 µm, 125 µm, 90 µm, and 63 µm were used. After separating the particles into size intervals, the samples are weighed to obtain the cumulative particle size distribution.

The particle size distributions (PSDs) of baby powder, calcium carbonate and quartz sand were determined in wet mode (the samples were previously suspended in demineralized water) by applying a laser diffraction method (Shimadzu, Kyoto, Japan, model SALD 3101). In the case of the baby powder particles, 10% isopropyl alcohol was used to enhance dispersion. Diffraction measurements were performed five times under identical process conditions. Each sample was scanned five times under identical process conditions in order to test the reproducibility and to gain more representative size distribution data for each tested sample. The net size distribution data were provided by averaging those five size distribution data for each size interval. The mean PSD is expressed on the basis of volume, with characteristic diameters d₅₀ as the median diameter, d_mode as the modus diameter (dominant size in a population), and the diameter mean, d_3,2 and reported via differential distribution function, dQ₃(d).

The morphology of the individual powder sample was analyzed using a Tescan Vega 3 scanning electron microscope (SEM) (Brno, Czechia). Prior to analysis, the samples were coated with a conductive layer of gold and platinum in argon plasma under a high vacuum of 0.1 mbar at 17 mA for 60 s. A secondary electron (SE) detector and a voltage of 10 kV were employed for this analysis.

3. Results and Discussion

3.1. Characterization of Powders before the Separation

Figure 3 presents the particle size distribution of baby powder particles, as determined by the sieving method, and introduced as the differential distribution function dQ₃(d).

The results indicate a predominant particle size range of 0–63 μm, which accounts for the majority of the powder’s composition. This narrow size range suggests a high degree of uniformity. Additionally, the presence of particles in the 180–355 μm range signifies the occurrence of agglomerates. Agglomeration is a common phenomenon in powders where individual particles tend to aggregate together, forming larger agglomerates due to van der Waals forces, electrostatic attraction, or moisture content. These agglomerates are often resistant to breakdown during the sieving process, leading to an apparent increase in the measured particle size.

Figure 3 reflects the volume particle size distribution of baby powder particles, as determined by the laser diffraction method, and introduced as the differential distribution function dQ₃(d).

The laser diffraction analysis reveals that the majority of the particles are of the sub-100 μm sizes, consistent with the intended fine particle size for baby powder. The significant drop in volume percentage beyond this size range indicates an absence of substantially larger particles, highlighting a very fine and consistent particle size distribution.

The SEM micrographs of the baby powder sample, depicted in Figure 4, provide a detailed view of the powder’s morphology at two different magnifications: 200× and 2000×. These images offer critical insights into the structural characteristics and morphology of the baby powder.

At 200× magnification, the micrograph shows a heterogeneous mixture of particles with varying sizes and rough surfaces. Larger particles are prominent, indicating a broad range of particle sizes that can affect the powder’s flowability and packing density. This irregular morphology is typical for talc-based baby powders and impacts their distribution and application properties. At 2000× magnification, the SEM micrograph highlights the plate-like morphology characteristic of talc, with well-defined edges. Finer particles are more evident at higher magnification.

The CaCO₃ particles exhibit a broad size distribution (Figure 5) using sieving analysis. The most significant fractions of particles are found in the 180–355 μm, 355–710 μm and >850 μm intervals, each constituting a substantial proportion of the overall particle volume. Specifically, the 355–710 μm size range shows the highest peak, indicating a 34.8 volume percentage of particles within this interval. Smaller particle size intervals, such as 0–63 μm, 63–90 μm, and 90–125 μm, display relatively minimal volume percentages, suggesting a lesser presence of fine particles within the CaCO₃ sample. The presence of larger particles can be attributed to the agglomeration of CaCO₃ particles.

The laser diffraction analysis (Figure 5) highlights that the CaCO₃ powder sample predominantly consists of very fine particles, mostly of sizes less than 100 μm. The decrease in volume percentage beyond this size range signifies a lack of substantially larger particles, indicating a highly fine and uniform PSD.

At 1000× magnification, Figure 6 reveals a relatively homogenous distribution of calcium carbonate particles with a noticeable degree of agglomeration. The particles appear to form loosely bound clusters, which is typical in fine particulate systems due to the inherent cohesive forces at the nanoscale. At a higher magnification of 5000×, the finer details of the individual calcium carbonate particles become evident. The particles exhibit an irregular, angular, elongated morphology, which is consistent with the crystalline structure of CaCO₃. The presence of agglomerates at both magnifications suggests that the calcium carbonate particles tend to cluster together, which can affect their dispersibility in various media.

The sieve analysis data (Figure 7) indicate that the majority of quartz sand particles fall within the size range of 180 µm to 355 µm. This suggests that a substantial portion of the quartz sand is medium to big in granularity, which aligns with typical applications where such particle sizes are desirable for their mechanical properties and permeability. Additionally, there is a notable presence of finer particles in the range of 63 µm to 180 µm.

The laser diffraction analysis of quartz sand (Figure 7) shows a prominent peak, indicating a dominant size in a population. In this case, the peak occurs at 352 µm, suggesting that the majority of the quartz sand particles are around this size. The steep increase on the left side of the peak implies a higher number of finer particles just below the modus diameter.

Figure 8 shows SEM micrographs of the quartz sand sample. The particles vary in size, indicating a heterogeneous distribution. The particles appear mostly angular to sub-angular. The size variation is evident, with particles ranging from a few millimeters to sub-millimeter sizes, supporting the differential distribution curve’s indication of a wide size range. The higher magnification provides a closer look at individual particles. The angular nature of the particles is more pronounced, with sharp edges and irregular shapes.

Statistical parameters, including the mean diameter and standard deviation (σ) for each particle size distribution plot (

Table 1. PSD statistical parameters for each pre-separated sample.

Sample	Mean Diameter (μm)	σ (μm)	R2	PDI
Baby powder	12.30	2.81	0.980	0.052
Calcium carbonate	2.06	1.96	0.997	0.905
Quartz sand	147.67	7.00	0.955	0.002

), were derived using the particle size distribution editing tool Hyprotech Explorer (HYSYS^® v2004.2), with fitting options. These statistical parameters were then used to calculate the polydispersity index (PDI) for each pre-separated sample.

3.2. Separation Efficiency

When a fluid containing a suspension of particles flows through the microfluidic separator, the design of the channel induces a centrifugal force. This force affects particles based on their mass and shape, leading to distinct movement patterns. Larger particles, possessing greater mass, experience a higher degree of inertia. Consequently, these particles are less influenced by the fluid flow and tend to deviate more from the streamline path, moving outward or settling at different positions within the channel. In contrast, smaller particles, with lower mass, have less inertia and are more likely to follow the fluid’s streamline, remaining closer to the central flow path. This difference in movement patterns due to varying inertia enables the effective separation of particles based on their size and mass within the microfluidic device. Through the application of intelligent preparation and maintenance, flow chemistry can achieve highly efficient separation processes that consistently yield particles of desired sizes, minimizing waste and improving the overall reliability of the production process.

Table 1,

Table 2. Normalized masses of baby powder after separation for all separators.

Flow Rate	200 mL min−1	250 mL min−1	300 mL min−1
SLA 1
Outlet 1	0.2895 g/100 mL	0.1484 g/100 mL	0.1428 g/100 mL
Outlet 2	0.1447 g/100 mL	0.1627 g/100 mL	0.1859 g/100 mL
Δ	0.1449	0.0143	0.0431
SLA 2
Outlet 1	0.0976 g/100 mL	0.1354 g/100 mL	0.2697 g/100 mL
Outlet 2	0.1533 g/100 mL	0.1253 g/100 mL	0.2595 g/100 mL
Δ	0.0557	0.0101	0.0102
FFF
Outlet 1	0.2927 g/100 mL	0.2979 g/100 mL	0.2695 g/100 mL
Outlet 2	0.2924 g/100 mL	0.2507 g/100 mL	0.2687 g/100 mL
Δ	0.0003	0.0472	0.0008

and

Table 3. Normalized masses of calcium carbonate after separation for all separators.

Flow Rate	200 mL min−1	250 mL min−1	300 mL min−1
SLA 1
Outlet 1	0.2547 g/100 mL	0.2775 g/100 mL	0.1988 g/100 mL
Outlet 2	0.1653 g/100 mL	0.1975 g/100 mL	0.2425 g/100 mL
Δ	0.0895	0.0800	0.0437
SLA 2
Outlet 1	0.1656 g/100 mL	0.1857 g/100 mL	0.3015 g/100 mL
Outlet 2	0.2243 g/100 mL	0.2449 g/100 mL	0.2132 g/100 mL
Δ	0.0587	0.0592	0.0883
FFF
Outlet 1	0.2081 g/100 mL	0.2448 g/100 mL	0.2529 g/100 mL
Outlet 2	0.2633 g/100 mL	0.1729 g/100 mL	0.3161 g/100 mL
Δ	0.0552	0.0719	0.0632

summarize the performance of three different separators: SLA 1 (square cross section), SLA 2 (circular cross section), and FFF (circular cross section), tested at varying flow rates of 200 mL min⁻¹, 250 mL min⁻¹, and 300 mL min⁻¹. To account for uneven volumes at the outlets, the masses in the tables were normalized to 100 mL, enabling direct comparison of mass at each outlet for an equal volume of suspension. The efficiency is evaluated by the normalized powder mass difference (Δ), which is calculated as the absolute difference between Outlet 1 (external stream) and Outlet 2 (internal stream). All of the masses in the tables represent average values obtained from five consecutive separation experiments.

Preliminary tests were conducted using separators with smaller diameter channels; however, these tests were hindered by constant clogging, rendering them infeasible. This outcome suggests that smaller channel dimensions are not well-suited for handling these specific powders.

For baby powder (Table 2), the SLA 1 separator shows the highest separation efficiency, at 200 mL min⁻¹, with a Δ of 0.1449, indicating a clear distinction between Outlet 1 and Outlet 2. When fluid flows through a channel with a curvature, a velocity mismatch occurs due to the varying distances the fluid must travel along the inner and outer edges of the curve. This discrepancy in velocity within the curved section of the channel leads to the formation of secondary flows, which are perpendicular to the primary flow direction. These secondary flows can significantly influence the overall fluid dynamics, potentially affecting separation processes within the channel. The results show that efficiency diminishes significantly at higher flow rates, which can be due to the increased turbulence, causing enhanced mixing. The SLA 2 separator achieves moderate separation at 200 mL min⁻¹ with a Δ of 0.0557. As the flow rate increases, the efficiency drops, with very low Δ values at 250 mL min⁻¹ and 300 mL min⁻¹. The FFF separator demonstrates minimal separation efficiency across all flow rates, with Δ values close to zero. The most notable difference occurs at 250 mL min⁻¹ with a Δ of 0.0472, but overall separation performance remains low. The best overall performance is notable for the SLA 1 separator at a flow rate of 200 mL min⁻¹.

For calcium carbonate (Table 3), the SLA 1 separator demonstrates its highest separation efficiency at 200 mL min⁻¹ with a Δ of 0.0895. The efficiency slightly decreases as the flow rate increases to 250 mL min⁻¹ and significantly drops at 300 mL min⁻¹, indicating that lower flow rates are more effective for this separator. The SLA 2 separator shows a relatively consistent separation efficiency across the different flow rates. The Δ values at 200 mL min⁻¹ and 250 mL min⁻¹ are similar, while the highest efficiency is observed at 300 mL min⁻¹ with a Δ of 0.0883. This suggests that the SLA 2 separator has optimal efficiency at the highest tested flow rate. The FFF separator exhibits a moderate separation efficiency at all tested flow rates, with the highest Δ of 0.0719 occurring at 250 mL min⁻¹. The efficiency decreases slightly at 200 mL min⁻¹ and 300 mL min⁻¹. The best overall performance is once again notable for the SLA 1 separator, at a flow rate of 200 mL min⁻¹, probably due to the increased turbulence at higher flow rates, which causes mixing.

For quartz sand (

Table 4. Normalized masses of quartz sand after separation for all separators.

Flow Rate	200 mL min−1	250 mL min−1	300 mL min−1
SLA 1
Outlet 1	0.2598 g/100 mL	-	0.2830 g/100 mL
Outlet 2	0.3854 g/100 mL	-	0.3496 g/100 mL
Δ	0.1255	-	0.0666
SLA 2
Outlet 1	0.2421 g/100 mL	0.2809 g/100 mL	0.4018 g/100 mL
Outlet 2	0.3652 g/100 mL	0.3410 g/100 mL	0.2250 g/100 mL
Δ	0.1230	0.0601	0.1767
FFF
Outlet 1	0.2193 g/100 mL	0.2947 g/100 mL	0.3548 g/100 mL
Outlet 2	0.4456 g/100 mL	0.3471 g/100 mL	0.3011 g/100 mL
Δ	0.2263	0.0523	0.0537

), the SLA 1 separator shows that the separation efficiency is highest at a flow rate of 200 mL min⁻¹ with a Δ of 0.1255. At a higher flow rate of 300 mL min⁻¹, the efficiency drops significantly to a Δ of 0.0666, suggesting that this separator performs better at lower flow rates. Data for the 250 mL min⁻¹ flow rate are unavailable due to the constant clogging of the separator channels. These data emphasize the realistic problems in the application of such microsystems, which is of great importance. The SLA 2 separator demonstrates its highest separation efficiency at the highest tested flow rate of 300 mL min⁻¹, with a Δ of 0.1767. This indicates that this separator is most effective at separating powder at higher flow rates. The FFF separator shows the highest separation efficiency at the lowest flow rate of 200 mL min⁻¹ with a Δ of 0.2263. As the flow rate increases to 250 mL min⁻¹ and 300 mL min⁻¹, the efficiency drops significantly, with Δ values of 0.0523 and 0.0537. This suggests that the FFF separator is most effective at lower flow rates and less effective as the flow rate increases. The best overall performance is notable for the FFF separator at a flow rate of 200 mL min⁻¹, probably due to the increased turbulence at higher flow rates, which causes mixing. The FFF separator might be a better option for this suspension due to the different channel roughness caused by the manufacturing.

Table 5. Reynolds number of the separators for each flow rate.

	SLA 1	SLA 1	FFF
200 mL min−1	1238	1577	1577
250 mL min−1	1550	1973	1973
300 mL min−1	1860	2368	2368

shows calculated Reynolds numbers for the separators at each flow rate used. The values indicate mostly laminar flow (Re < 2000), while circular cross-section separators at a flow rate of 300 mL min⁻¹ show transitional flow (2000 < Re < 4000). Turbulent flow would have Re > 4000.

3.3. Determination of Particle Size Separation Capability

The findings evaluate the effectiveness of a particle separation process based on laser diffraction measurements (Figure 9, Figure 10 and Figure 11), focusing on characteristic particle size metrics before and after separation (

Table 6. PSD parameters introduced as characteristic diameters for baby powder in SLA 1 separator at 200 mL min⁻¹.

	Before Separation	Outlet 1	Outlet 2
dmode (μm)	20	352	315
d50 (μm)	16	349	325
d3,2 (μm)	44	1182	967

,

Table 7. PSD parameters introduced as characteristic diameters for calcium carbonate in SLA 1 separator at 200 mL min⁻¹.

	Before Separation	Outlet 1	Outlet 2
dmode (μm)	3	3	3
d50 (μm)	2	3	2
d3,2 (μm)	5	11	8

and

Table 8. PSD parameters introduced as characteristic diameters for quartz sand in FFF separator at 200 mL min⁻¹.

	Before Separation	Outlet 1	Outlet 2
dmode (μm)	352	393	393
d50 (μm)	280	401	369
d3,2 (μm)	803	890	624

) at two distinct outlets (Outlet 1 and Outlet 2) with a flow rate of 200 mL min⁻¹. As determined by the difference in normalized powder mass, the best separators for each powder were measured. The SLA 1 was 200 mL min⁻¹ for baby powder and calcium carbonate, and the FFF was 200 mL min⁻¹ for quartz sand.

The modus diameter before separation was 20 μm, indicating a predominance of fine particles. Post-separation, Outlet 1 exhibited a dominant size of 352 μm, while Outlet 2 showed a modus diameter of 315 μm. This indicates that Outlet 1 predominantly contains slightly larger particles compared to Outlet 2. The significant difference in modes (352 μm vs. 315 μm) suggests that Outlet 1 is more effective in capturing the bigger particles, while Outlet 2 retains slightly smaller particles on average.

The median (x₅₀) indicates the particle size at which 50% of the sample’s mass is composed of particles smaller than this value. The median particle size before separation was 16 μm. After separation, Outlet 1 had a median of 349 μm, and Outlet 2 had a median of 325 μm. This difference (349 μm vs. 325 μm) further underscores the trend seen with the mode; Outlet 1 tends to collect larger particles more efficiently than Outlet 2. The median values indicate that Outlet 1’s particle distribution leans more towards larger sizes, whereas Outlet 2 has a slightly lower median size.

The Sauter mean diameter (x_3,2) is defined as the diameter of a sphere that has the same volume-to-surface area ratio as the particle distribution. The initial Sauter mean diameter was 44 μm. Post-separation, the Sauter mean diameter increased to 1182 μm for Outlet 1 and 967 μm for Outlet 2. The Sauter mean diameter is sensitive to larger particles, and the significant difference between Outlet 1 and Outlet 2 (1182 μm vs. 967 μm) confirms that Outlet 1 contains a greater proportion of larger particles. This also highlights the efficiency of Outlet 1 in segregating bigger particles compared to Outlet 2, which can be explained by greater inertia and less turbulent flow. In flow chemistry, the ability to modify flow rates through flexible control systems is crucial for optimizing the separation of particles by size, ensuring that the process remains efficient and adaptable to various conditions.

The particle separation process demonstrated a substantial shift in particle size distribution from finer to bigger particles, as evidenced by the increases in x_mod, x₅₀, and x_3,2 at both outlets. The pronounced differences in particle size distributions before and after separation suggest the great impact of the agglomeration during the process. Agglomeration might occur due to the washout of the surface agents in water, which were preventing substantial agglomeration prior to the testing. The consistent trends across these metrics affirm the process’s efficacy in segregating larger particles. Nevertheless, the formation of agglomerates should be addressed in subsequent research to enhance the accuracy of particle size measurements and further optimize the separation technique.

Outlet 1 retains the same PSD modus of 3 μm, indicating that the most frequent particle size has not changed. However, the median size increases to 3 μm, showing a shift towards slightly larger particles. The Sauter mean diameter increases significantly to 11 μm, suggesting that Outlet 1 is more effective in capturing a higher proportion of larger particles. This increase indicates that the separation process at Outlet 1 is skewed towards larger particles, enhancing the average particle size, which further confirms our theory. Outlet 2 also maintains the dominant size at 3 μm, similar to the initial distribution. The median size remains at 2 μm, indicating that PSD is still centered around finer particles. However, the Sauter mean diameter increases to 8 μm, which is higher than the initial value but lower than Outlet 1. While both outlets maintain the same most frequent particle size (modus diameter), they differ in median and Sauter mean diameters. Outlet 1 shows a higher median and Sauter mean diameter, indicating a greater tendency to capture larger particles. Outlet 2, with a lower median and Sauter mean diameter, retains more fine particles relative to Outlet 1. Efficient particle separation, facilitated by manufacturing-monitoring technologies, is central to flow chemistry, as it enables the production of particles with specific dimensions.

Outlet 1 shows an increase in all particle size metrics compared to the pre-separation state. Modus diameter increases to 393 μm, indicating a shift towards larger particle sizes as the most frequent occurrence. The median increases significantly to 401 μm, suggesting that more than half of the particles are now larger than before. The Sauter mean diameter also rises to 890 μm, indicating a higher average particle size and a more effective capture of larger particles. This highlights Outlet 1’s efficiency in segregating bigger particles. Outlet 2 also exhibits an increase in particle size metrics, though to a lesser extent than Outlet 1. The mode remains the same as Outlet 1 at 393 μm, indicating similar most frequent particle sizes. The median increases to 369 μm, which is higher than the initial state but lower than Outlet 1, indicating a moderate shift towards larger particle sizes. The Sauter mean diameter at 624 μm is significantly lower than Outlet 1, suggesting that Outlet 2 is more effective in capturing smaller particles compared to Outlet 1. The comparison indicates that Outlet 1 is more efficient in capturing larger particles, as evidenced by its higher median and Sauter mean diameter. Outlet 2 shows a more balanced distribution with a lower median and Sauter mean diameter. This suggests that Outlet 1 is better suited for applications requiring predominantly larger particles, whereas Outlet 2 is more suitable for applications needing smaller particle sizes.

4. Conclusions

Manufacturing monitoring technologies are key to flow chemistry, where they facilitate the efficient separation of particles, ensuring that production yields particles with specific sizes and shapes, which supports efforts to minimize waste through enhanced process control. In this study, spiral separators were fabricated using both SLA and FFF 3D printing technologies. Three types of microfluidic devices were used for solid/liquid separation: the SLA 1 (square cross section) separator, the SLA 2 (circle cross section) separator, and the FFF (circle cross section) separator, all of which have a spiral internal structure composed of eight turns. Powder separation was conducted at flow rates of 200 mL min⁻¹, 250 mL min⁻¹, and 300 mL min⁻¹.

Subsequent to the separation procedure, an analysis was conducted on the mass of the powders retrieved and dried from the external (Outlet 1) and internal (Outlet 2) outlets of the separators. By normalizing the powder mass at the outlets of the separator, the most significant difference in mass between Outlet 1 and Outlet 2 was observed for baby powder separation through the SLA 1 separator and operated at a flow rate of 200 mL min⁻¹. In the case of calcium carbonate, the most significant difference in mass was also with the SLA 1 separator at a flow rate of 200 mL min⁻¹, while for quartz sand, it was with the FFF separator, again at a flow rate of 200 mL min⁻¹. The only significant anomaly was the clogging of the SLA 1 separator at a flow rate of 250 mL min⁻¹ with quartz sand, which resulted in failed testing.

The differences between Outlet 1 and Outlet 2 indicate that while both outlets effectively separate larger particles from the initial distribution, Outlet 1 is superior at isolating bigger particles. This distinction is critical for applications that require precise control over particle size distribution. For instance, industries relying on bigger particles for filtration or abrasive applications might prefer the output from Outlet 1, whereas Outlet 2 could be more suitable for applications that require slightly smaller particles. This phenomenon can be attributed to the effect of centrifugal force, where particles with greater mass exhibit increased inertia and thus display different movement patterns compared to smaller particles. Flexible control systems are critical in flow chemistry, allowing for the adjustment of flow rates to improve the efficiency of separating particles by size, making the process more adaptable and minimizing waste.

Upon comparison of the values for the powder particles before and after separation, a notable discrepancy was observed in terms of particle size increase. This discrepancy is attributed to the formation of agglomerates.

Although these separators did not achieve complete solid/liquid separation, the methodology shows great promise. We believe that with further research, the efficiency can be significantly enhanced. Advances in additive manufacturing technologies and the development of new separator designs have the potential to lead to substantial improvements in separation efficiency. Effective intelligent preparation and maintenance in flow chemistry are vital for achieving optimal particle separation, ensuring that particles are produced with consistent quality while minimizing waste and enhancing the stability of the manufacturing process.