Electrospinning technique was used to prepare PET/PDMS/collagen nanofiber membranes. The electrospinning device (FNM, Iran) typically consists of two syringe pumps, two high-voltage power sources, and a rotary collector covered with aluminum foil. To construct the proposed nanofibrous membrane, PET was dissolved in DMF/THF (1:1) solvent (22 wt%) at 45°C under continuous stirring. PDMS/curing agent (10:1) was added to DMF/THF (1:1) solvent (10% w/w) on a magnetic stirrer at room temperature. Then, a 5% (w/w) collagen solution was prepared in HFIP solvent. When the PET solvent reached transparency, it was allowed to cool to room temperature and mixed with PDMS on a magnetic stirrer at room temperature. PET/PDMS and collagen were then loaded into two separate 5 mL syringes with 18G blunt needles and placed into the appropriate pump. The distance between the injection needle and the collection tube was kept at 15 cm on each side, and the injection rate was set at 0.4 mL/h. The syringe containing PET/PDMS was connected to a 22 kV power supply and the other syringe containing collagen was connected to a 20 kV power supply. The speed of the roller receiver is set at 90 rpm. To achieve a thickness of about 30 μm, the total electrospinning time was set at about 4 h. The resulting woven nanofiber mat was placed in a vacuum desiccator for 30 min, receiving glutaraldehyde vapor (25%) to Cross-link its surface. Finally, the mat was washed with PBS and deionized water. To remove all toxic glutaraldehyde residue from the fibers, the washing procedure is repeated at least 3–4 times.

In using Scanning Electron Microscopy (SEM) the nanofiber samples were coated with gold nanoparticles by an ion sputtering coater (Polaris SCM-200, South Korea). Gold-coated nanofibers were then placed in a high-vacuum chamber, and the images were taken at 20 kV (SNE-4500M, Korea). Nanofibers were evaluated for morphology, size, and size distribution by ImageJ software.

Atomic Force Microscopy (AFM), Nanosurf easy, Switzerland, analysis was carried out to determine the surface roughness and morphology of nanofibers.

The porosity of the nanofiber was evaluated by nitrogen gas adsorption and desorption for samples before (BC) and after Cross-linking (AC). Samples were dried in a clean dry tube for 4 h at 77°C in a degas tube station to eliminate the humidity from the samples (BelSorp, Mini2, Japan).

To examine wettability, the contact angle was measured using water droplets as the desired fluid on the surface of nanofibers. For this purpose, a 5 cm² nanofiber membrane was placed under a contact angle instrument (MehrTavNegar, Iran) equipped with a camera at 25°C. The images of the droplet and surface of the nanofiber were taken at the interfaces (AC and BC).

Fourier transform infrared (FT-IR) spectroscopy analysis was conducted by the FT-IR analyzer (WQF-510A, China) to investigate the chemical structure of the prepared PET, PDMS, and collagen nanofiber in the range of 400–4,000 cm⁻¹. KBr was used as a carrier for the sample in the FT-IR analysis.

Human umbilical vein endothelial cells (HUVECs) and rat glioma cell line (C6) were cultured in DMEM-F12 rich medium containing 10% FBS and 1% penicillin/streptomycin and incubated at 37°C and 5% CO₂. The cell viability of HUVECs cultured on top of nanofibers (outside the microfluidic device) was assessed with 10% AlamarBlue solution (Page et al., 1993). The nanofibers were cut into circles of 1 cm in diameter, sterilized with 70% ethanol for 15 min and then exposed to UV irradiation for 30 min. The nanofibers were then placed at the bottom of a 48-well plate and fixed. Then, 5×10³ cells in 250 μL F12-enriched medium with 10% FBS, and 1% penicillin/streptomycin were seeded into each well and on top of the nanofibers. Cell viability was assessed for 1, 3, and 5 days of incubation. To assess viability, the culture medium in each well was replaced by 250 μL of complete culture medium containing 10% AlamarBlue. Then, the cell culture plate was incubated at 37°C and 5% CO₂ for 4 h and UV absorbance was obtained using the ELISA microplate reader (BioTek, United States) at 570 and 630 nm for cell viability calculation.

HUVECs were cultured in a 48-well culture plate and placed in an incubator at 37°C and 5% CO₂ for 48 h, as explained. Then, the culture medium was discarded, and the wells were rinsed with PBS, after which the cells were fixed by adding a 1:1 solution of 2% glutaraldehyde and 2.5% paraformaldehyde and kept at 4°C for 1 h. The nanofibers were then exposed to 60, 70, 90, and 100% ethanol for 5 min to dehydrate. Finally, the nanofibers were coated with gold for SEM imaging and cell attachment analysis.

The numerical simulation section involved the use of COMSOL MULTIPHYSICS 5.5 (Comsol, Inc.), a commercial finite element code, to simulate the microfluidic flow within the PDMS-based device. The COMSOL model was created in a three-dimensional structure with the same dimensions as the experiments. The simulations were conducted in a stationary mode with the assumption that boundary conditions were not slippery. Single-phase laminar flow was also chosen as model physics, with the governing equations within the model being the Navier-Stokes equation (Eq. 1) and the continuity equation (Eq. 2). The parameters considered in the equations included fluid density (ρ), dynamic viscosity (µ), fluid velocity (V), and pressure (P). The simulation included a typical 10 μL/min input flow in laminar input conditions, with the fluid density set to 1.011 g/cm³ and dynamic viscosity to 0.94 mPa·s. Only the flow in the upper microchannel was simulated due to the symmetry existing in the upper and lower channels. ρ V → . ∇   V → = − ∇ P + μ   ∇ 2 V → (1) ∇ . V → = 0 (2)

COMSOL Multiphysics^® software offers a robust collection of integrated functions and operators designed to create geometric surfaces with a desired roughness. These include functions to generate uniform and Gaussian random distributions, as well as a valuable sum operator, which significantly enhances the flexibility and versatility of the software.

To create the bottom surfaces of the microchannel, the parametric surface module of COMSOL was utilized, resulting in an arithmetical mean height of approximately 0.3 µm for the channel. The equation governing the production of roughness was obtained by the equation of spatial frequencies. To represent the rough surface, the following double sum (Eq. 3) was used (Villegas et al., 2018): f x , y = ∑ m = − M M ∑ n = − N N A m n   cos   2 π m x + n y + φ m , n (3) In which, A_mn is as Eq. 4: A m n = 1 m 2 + n 2 β   g m , n (4)

In these expressions, the spatial coordinates are denoted by x and y, the spatial frequencies are represented by M and N, set to 5. The amplitudes are labeled as Amn, with the phase angles denoted by φ (m, n) being determined using a zero mean uniform random function within the interval of -π/2 to π/2 in COMSOL. The spectral exponent, β, was assigned a value of 0.1. The function g (m, n) corresponded to a zero-mean Gaussian random function. Scaling in the z-direction was applied to the function f (x, y) to achieve the desired roughness values. The following equation (Eq. 5) was utilized to select the amplitude parameter, representing the arithmetical mean height (S_a), serving as an indicator of the surface roughness: S a = 1 A ∬ A | f ( x , y ) | d x d y (5) where A represents the mean-plane area.

A simple PDMS-based microfluidic device containing an upper and a lower channel was designed (for each channel the dimension was 1.4 mm×18 mm, height 200 µm) to house the nanofiber membrane (Figure 1). Once the masks were designed (Figure 1A), soft lithography was used to fabricate molds. SU8-2050 (Microchem, United States), a negative photoresist, was coated on the silicon wafer by a spin-coater (Microchem, Newton, MA, United States). Then, UV light was lit through the mask over the wafer by a mask aligner (Danesh Equipping System, LSM5, Iran), allowing the SU8 to Cross-link onto the surface of the wafer. Once the unbonded photoresist was washed away from the surface of the wafer, a negative imprint of the microchannel design was created. Two silicon molds (upper and lower panels) were fabricated. To create the microfluidic device, PDMS was mixed with its curing agent (10:1) and de-bubbled inside a vacuum pump. The mixture was then poured inside the silicon molds and placed on a hot plate at 80°C for 1 h. Once cast, designed inlets and outlets were created using a biopsy punch (Figure 1B). Next, the surface of each casted PDMS was washed with isopropyl alcohol and acetone (Merck, Germany). After drying, the nanofiber membrane was fused between the upper and lower sections using an oxygen plasma treatment apparatus (Harrick Plasma, Ithaca, NY, United States), forming a leak-free microfluidic device. To sterilize, the device was rinsed with 70% ethanol and exposed to UV for 30 min.

First, the laminar flow of cell culture media into the channels removed all air bubbles then the device was placed inside an incubator for 15 min. The number of. 2 × 10⁶ cells (HUVEC) were diluted in 1 mL culture medium and then 100 µL of that solution was carefully injected inside the micropipette tip pinned to the upper channel inlet port, where gravitational and capillary forces allowed the gentle flow of cells inside the channel. Under a light microscope, the flow of cells toward the nanofiber membrane and the outlet port was evaluated. The device was placed in an incubator for 4 h to allow cell sedimentation and proper attachment. Once cells were attached, 1.5 × 10⁶ C6 cells were diluted in 1 mL culture medium and slowly passed into the lower channel. After injection, the device was placed inside an incubator for 4 h. When cells were attached, the fresh culture medium was injected into the upper and lower channels at a rate of 10 μL/min, and the device was kept inside the incubator for 24 h. Fluorescent staining was utilized to determine whether or not cells were present on the nanofiber membrane.

The upper microchannel was washed twice with PBS and then 50 µL of DAPI solution dissolved in deionized water (1 μg/mL) was injected into the inlet port for 1 min. Then, the device was placed under a fluorescence microscope (Optika, IM-3, Italy), where DAPI was excited with ultraviolet light (358 nm) and was detected through the blue/cyan filter. First, staining was performed with DAPI dye within the first 4 h after cell injection and later after 24 h.

Acridine Orange (AO) and Propidium Iodide (PI) staining was used to assess live C6 cell counts in the lower channel, 50 μL of a 1:1 mixture of AO and PI with a concentration of 50 μg/mL was injected into the inlet of the lower channel, with the device placed under the fluorescence microscope. AO emits green fluorescence with the maximum wavelength at 526 nm (excitation 502 nm), while PI is excited at 488 nm and, with emission at a maximum wavelength of 617 nm.

The structure of nanofibrous membranes has a significant impact on the performance of microfluidic devices designed for cell culture applications. Irregularities and variations in the nanofiber structure can significantly affect the velocity characteristics and shear stress distribution of the microchannel. Maintaining uniform shear rates and velocities in microfluidic devices ensures a consistent fluid flow pattern, minimizing variation in the cell culture area. This uniformity promotes uniform exposure of cells to nutrients, oxygen and signaling molecules, leading to more reproducible and reliable experimental results. Additionally, it allows the study of cellular responses under specific flow conditions, such as shear stress-induced cell migration and binding.

The behavior of the fluid in the microchannel was studied with roughness values of 0.1, 0.2, 0.3, 0.4, and 0.5. Figure 6B illustrates the velocity distribution for roughness values ​​of 0.1 (a) and 0.5 (b) µm. As shown in the velocity slice representation shown in Figure 6Bb, the microchannel with a roughness value of 0.5 µm exhibits non-uniform fluid velocity along the entire length of the channel. However, reducing the roughness to 0.1 µm provides a much more stable velocity profile across the channel. The difference in roughness also has a notable impact on the cutting speed distribution, as shown in Figure 6C. The stiffer microchannel (Figure 6C (b)) exhibits a non-uniform shear rate distribution, with higher shear rates near the rough surface. In contrast, the smoother microchannel, Figure 6Ca, exhibits a more uniform shear rate distribution without sharp spikes near the bottom surface.

The velocity profile across the width of the microchannel was investigated for five different surface roughness values (0.1, 0.2, 0.3, 0.4, and 0.5 μm) at a height of 0.1 mm above the bottom of the microchannel (Figure 7A). The graph shows how the velocity profile becomes non-uniform as the surface roughness of the microchannel increases. The effect of surface roughness on the shear rate distribution was studied at a height of 0.01 mm above the bottom of the microchannel. As shown in Figure 7B, the surface roughness has a significant effect on the shear rate. The rougher microchannels exhibited non uniform shear rate distributions. In contrast, smoother microchannels exhibited a more uniform shear rate distribution without sharp peaks near the channel surface.

As mentioned previously, the fluid flow rate entering the microchannel is a critical factor in the performance of microfluidic devices in cell culture applications. This directly affects the velocity profile and shear stress on the nanofibrous membrane, which serves as a substrate for cell adhesion and proliferation. Uniform shear rate and velocity directly impact cell adhesion, proliferation, differentiation, and overall cellular response. Careful control of shear rate and velocity can mimic physiological conditions and provide cells with an environment that closely resembles their natural microenvironment. The effect of inlet flow rate on the velocity profile across the width of the microchannel is shown in Figure 7C for a surface roughness of 0.3 μm. Velocity profiles were examined at a height of 0.1 mm above the rough microchannel bed, accounting for the effect of varying the inlet flow rate into the microchannel from 5 to 15 μL/min. As observed, as the inlet flow rate increased, the velocity magnitude also increased. Additionally, we found that the influence of microchannel roughness on the velocity profile variation becomes more pronounced at higher speeds, as shown in Figure 7C. The variation of shear stress along the width of the microchannel was also studied for different inlet flow rates into the microchannel with a surface roughness of 0.3 μm (Figure 7D). Inlet flow rates ranged from 5 to 15 μL/min. The results showed that the shear stress values showed a significant increase as the input flow rate increased. The presence of microchannel roughness increased the shear stress values, resulting in larger shear stress fluctuations in the region near the rough surface. This effect increased as the input flow rate to the microchannel increased, as shown in Figure 7D (Cukierman et al., 2001).