in Z direction, curing time of each layer, and total thickness of the part are the most critical factors to have a high-quality channel with a great surface fin sh. Various cross-sections, ranging from right-angled or isosceles triangular to hexagonal, were fabricated and the best optimized parameters were identifi d (Fig. S1). To complete the fuidic network, 3D-printed inertial microchannels need to be bonded to a substrate with enough optical transparency and rigidity for subsequent testing. In this work, a variety of scenarios has been evaluated, and upon extensive evaluations and characterizations, permanent bonding of 3D-printed channels to a PMMA sheet via a double-coated adhesive tape was selected as the most promising and reproducible method. A transparent double-coated pressure-sensitive adhesive tape (ARcare, Adhesive Research) having 25.4 µm clear polyester flm coated with AS-110 acrylic medical grade adhesive was cut with a similar size of PMMA sheet (Fig. 1AII). Afer the attachment of one side of the tape to the PMMA sheet, the 3D-printed inertial part was manually placed over the other side of the tape and pressed with a tweezer until no bubble was observed at the interface (Fig. S2).
An important feature of PDMS is its optical transparency, which makes it suitable for a broad range of microscopic applications. Given the fact that commercial DLP/SLA resins are not typically transparent, the attachment of 3D-printed microchannels to PMMA sheets provides enough transparency for the optical and fuorescent microscopy (Fig. 1AIII). What makes this approach attractive for a wide range of communities (e.g., biologists and chemists) is its user-friendliness for people without prior knowledge about microfabrication and soflithography. Te entire process from CAD drawing to printing and then testing takes less than 2 hours, portraying the versatility of this method for inertial microfuidic research. More importantly, devices made using this technique are not prone to the deformation and leakage compared to the PDMS-made devices, making them suitable to study new physics, especially at high Re. Furthermore, by considering the fabrication cost, time, and efforts of a complicated inertial microfuidic device, our suggested method is rapid and utilizes a low-cost raw material which are valuable features, especially in areas where resources are limited. Figure 1B,C depict a fi al device fabricated using this technique. Te internal channels are flled with red food color for the sake of illustration.
In order to investigate the bonding quality, a straight microchannel with dimensions of 50 µm height, 200 µm width, and 4 cm length was fabricated and tested accordingly. We have monitored the device performance for the appearance and growth of Safman-Taylor fi gers until it becomes stable, called “infation stability” (Fig. 2A). Te results are presented in a 2D diagram to identify the channel behavior at a given pressure, as shown in Fig. 2C. Our results revealed that the holding strength of double-coated adhesive tape was able to achieve a leak-proof interface between the 3D-printed part and PMMA sheet, not only at typical operating pressure reported in literature46, but also more than the capability of PDMS-made channels in withstanding high fl w rate conditions. Shear rate distribution across a line parallel to the channel width was also evaluated, and as Fig. 2B revealed, increasing the fl w rate leads to imposing more shear forces at the edges of the channels. Te more the fl w rate, the larger the appearance of Safman-Taylor fi gers (insets of Fig. 2B). Te green area in Fig. 2C shows the safe zone for performing inertial microfuidic experiments where no Safman-Taylor fi gers appear during the operation. We have found that at pressures more than 82.6 psi, Safman-Taylor fi - gers begin to appear; however, this does not impose any detrimental efect on the device performance (i.e., no leakage or bonding collapse). Also, we did not observe any delamination or deformation in channels afer consecutive runs at high pressures (i.e., 120 psi), all of which are common in PDMS-based inertial microfuidic devices (see Figs. S3 and S4 for the pressure drop, velocity profle inside the microchannels for a wide range of operating fl w rates).
The surface characteristic of the double-coated adhesive tape was also investigated using a proflometer. As Fig. 3 illustrates, the roughness of the tape is homogenous and is in the submicron range. Te values of Ra and Sa were about 250 and 240nm, respectively. Also, the roughness of the 3D printed parts was evaluated and value of Sa was less than 300 nm. Tese nanometric rugosities indicate that the roughness of tape does not have any efect on the fow profle and particle focusing. Although optically transparent, the optical characteristics of the PMMA sheet (2-mm-thick) and adhesive tape were evaluated to identify the possibility of accurate fuorescence imaging47. Hence, the UV-visible absorbance spectra for a wide range of wavelengths (i.e., from 200 to 1100nm) were recorded via a spectrophotometer (Cary 60 VU-Vis spectrophotometer, Agilent Technologies). Figure 3B,C reveal that the light loss is negligible for both PMMA and adhesive tape within the visible spectrum, resulting in no trace of autofuorescence residual.
Straight microchannels with rectangular or square cross-sections are arguably the most widely used inertial microfuidic systems. Tanks to their ease of fabrication and the ability for parallelization, a myriad of applications have been developed using these platforms over the past decade48. Te required channel length for inertial particle migration to the equilibrium positions is L H = πμ /ρ α U f f m L 2 2 where fL is estimated in the range of 0.02 to 0.05 for (H/W) from 2 to 0.5, and the corresponding fl w rate for inertial migration is calculated as Q ≈ 2 / πμWH 3ρ αL fL 3 2 48. Channel Re (Re = ρUD/μ) and particle Reynolds number ( ) Re Re p H 2 = 2 α are two dimensionless numbers for the characterization of particle migration in a straight microchannel. When particle Re is much smaller than 1, viscous drag becomes dominant, and particles follow the streamline. Increasing particle Re augments inertial forces, causing inertial particle migration become obvious in the microchannel49,50.
Particle migration within a straight channel strictly depends on its cross-section. In square straight microchannels (with an aspect ratio (AR) (width/height) of 1), particles migrate to four equilibrium positions located at the center of each wall. Changing the cross-section to rectangular disturbs this focusing pattern where in a rectangular straight microchannel with AR of 0.5, focusing positions reduce to two near the center of long walls51. Th s behavior was explained by Zhou and Papautsky where they identifi d two-stage particle migration in rectangular straight microchannels21. Further increase in the AR results in the more unpredicted focusing behavior of particles. Generally, in channels with high AR, stable focusing positions are reduced. However, by exceeding Re from a critical value, the number of stable equilibrium positions increases which is a function of particle size, channel dimensions, and Re. Based on reported experimental results, = . κ κ ≤ ≤ ≤ ≤ − . Re A 697( R A / ) (4 5 / R R 60, 5 e 660) c 0 79 was identifed52. Te abovementioned results elucidate that particle focusing is strongly afected by channel cross-section. However, due to the fabrication limitations, dependency of various cross-sections to channel geometries was not systematically investigated. Recently, triangular and semi-circular cross-sections were fabricated using Si anisotropic etching with potassium hydroxide53, a brass for mold fabrication54, FDM for creation of sacrific al mold40, or unconventional micromilling14. However, critical fabrication limitations do not allow for further investigation on the dependency of triangular angle or type (e.g., right-angled triangular) on focusing patterns of the particles. Here, as a showcase, a straight microchannel with rectangular cross-section and AR of 4 (all channel dimensions are provided in Section S3 and Tables S1–5) was fabricated, and the results are illustrated in Fig. 4A. As the results indicate, for low Re (Fig. 4AI), 20µm particles focus at the center of long walls of the channel cross-sections, shown previously in PDMS-made microchannels. Nonetheless, the focusing pattern for particles at higher Re does not obey a specifc role. As clearly can be seen, increasing fl w rates leads to generation of additional focusing positions within the microchannels where side walls are also added to the equilibrium positions of particles (Fig. 4AII–IV). Furthermore, lateral migration of MDA-MB-231 and DU-145 cells at low fl w rates (10~20ml/hr) (Fig. 4BI–III) illustrates their single-line focusing within the rectangular straight microchannel, which is promising for fl w cytometry applications. Moreover, to showcase the versatility of the proposed method, a triangular straight microchannel was fabricated and the results are shown in Fig. 4D. Te results are completely in line with those reported in the literature where 10µm particles and cells occupy one lateral position in the channel for low fow rates (Fig. 4DI,II). For high
Spiral defi es as a curve winding around a center point with continuous decreasing or increasing manner. When fl w passes the curvature, velocity mismatch occurs in the curve section of the channel, resulting in the generation of secondary fl ws. In inertial microfuidics, spiral microchannel has progressed signifcantly, and nowadays, most of the particle/cell separations are performed using these microchannels64. De is used for the characterization of secondary fows within the channel. Intuitively, smaller channel curvature or larger channel size or Re leads to higher De, thereby imposing stronger secondary fl ws within the channel. For a given De, average transverse Dean velocity (UDe=1.84×10−4 De1.63) and Dean drag force = = πµ α π . × µ α − . ( 3 F U 5 4 10 De ) D De 4 1 63 can be identifi d. However, the exact behavior of particle migration at the downstream of the fuid was not thoroughly investigated, and all results are based on experimental data. Te most appealing feature of spiral inertial microfuidics is its high-throughput where 2100 particles per second can be processed9 . Particle sorting is one of the most signifcant applications of spiral microfuidics. Previously, the potential of a rectangular spiral microchannel for continuous and simultaneous isolation of 10, 15, and 20µm based on soflithography was investigated (Fig. 6AI) 65. Dean fl w dynamics for a low-aspect-ratio rectangular spiral microchannel was also thoroughly explored66. Beyond a simple rectangular spiral microchannel, various geometry modifcations for regulation of Dean forces and performance enhancement of the device have been proposed. Beneftting from micromilling (Fig. 6AII), trapezoidal spiral microchannels illustrate promising results in redistribution of lateral focusing positions of particles appropriate for size-based particle separation. In these channels, smaller particles focus along the outer wall, whereas larger ones migrate toward the inner wall67. Th s superior advantage has been widely investigated by our group, among other groups, for circulating tumor cell (CTC) and circulating fetal trophoblasts (CFT) isolation19,68, blood plasma separation69, isolation of microcarriers from mesenchymal stem cells70,71, microalgae separation72, and synchronizing C. elegans73. Also, multiplexing using stack of attached PDMS layers to boost the throughput is illustrated previously69,74. However, most of the aforementioned applications are just doable by utilizing cleanroom facilities or employing conventional micromachining (e.g., metal machining or laser cutting) for the fabrication of microchannel. Besides, micromachining has its own limitations such as inability to make sharp corners or difculty in making spiral loops close to each other. Tese challenges highlight an unmet need for the fabrication of spiral microchannels using a versatile method which is robust and can surmount aforementioned issues.
As a showcase of the versatility of our proposed method, we have fabricated a spiral microchannel with trapezoidal cross-section with a width of 600 µm and heights of 80 and 130 µm. Tese results are then put aside a PDMS chip with similar dimensions, and the data is provided in ESI (Fig. S5). Despite all progress in spiral inertial microfuidics, there is not any report of a spiral with cross-sections rather than rectangular or trapezoidal. In other words, a huge amount of potential remains intact to study spiral microchannels with diferent cross-sections such as triangular (Fig. 6AIII). For this aim, for the fi st time, we have fabricated a spiral microchannel with right-angled triangular cross-section (as schematically shown in Fig. 6B) where the width and height are 600 and 210µm, respectively. As the results are illustrated in Fig. 6C, there is a tight band focusing for particles larger than 10 µm, which is suitable for high throughput fl w cytometry applications where single line focusing is desired. Also, we observed double-band focusing behavior for 20µm particles at fl w rate of ≥4ml/min. Te dimensions (Fig. 6D) and channel cross-section (Fig. 6E) show the accuracy of the proposed method the for fabrication of right-angled triangular spiral microchannel (check Fig. S6 for contraction-expansion array microchannel results). Tese results illustrate the flex bility of this method where a complex cross-section can be fabricated in less than two hours with high robustness and stability. Our results hold promise for leveraging the potential of additive manufacturing for the fabrication of inertial microfuidic devices, which is more challenging using conventional microfabrication methods (see Section S6 for multiplexing of 3D printed inertial microfuidic devices).
PDMS-made inertial microfuidic devices have been widely used for the cell separation using biological samples such as blood and urine. While PDMS is proven to be a biocompatible material with minimum side efects on cells, we have tested the 3D printed devices using DU145 cells, assessing their viability and functionality post-separation. Te collected cells from the device outlet were cultured back into a petri dish for 5 days, showing similar morphological features to the control group as shown in Fig. 7A,B. Te fl w cytometry tests (Fig. 7C) showed that the viability of the cells was not compromised during the operation using 3D printed devices. Te real-time PCR analysis was utilised to assess the expression of genes related to the general activities and stress responses in both treated and untreated cells (Fig. 7D). Te similar expression level of GAPDH and CDKN2A confi med that neither cellular metabolism nor cell cycle progression were afected afer processing